Ann Rheum Dis (1982), 41, Supplement p 1 Introduction The choice of, and the decision to apply, disease modifying drugs COLIN G BARNES From The London Hospital, London El, UK Although rheumatoid arthritis (RA) is recognised to be a systemic illness, frequently the only manifesta- tion is an inflammatory polyarthritis in which synovitis causes joint destruction. Extra-articular or systemic features include vasculitis which, when pre- sent, may be the cause of potentially serious or fatal manifestations. The aetiology remains unknown, although immunological disturbances are certainly involved in the perpetuation of the disease. The for- mation of immune complexes within the joint is probably responsible for the release of chemical mediators of inflammation. The natural history of the disease is variable, rang- ing from a brief episode of polyarthritis which resol- ves spontaneously, to a rapidly progressive arthritis with widespread systemic features. The evidence for an immunological disturbance arises from: (1) the presence of serum and synovial fluid autoantibodies-rheumatoid factors; (2) the presence of other serum antibodies, for example antinuclear antibodies; (3) the histology of the synovium which includes infiltration with plasma cells and lympho- cytes; (4) reduced synovial fluid complement levels in actively inflamed joints; (5) the presence of immune complexes in syno- vial fluid; (6) reduced serum complement levels in patients with active systemic disease; (7) amyloid deposition. The first line of treatment always consists of the use of non-steroidal anti-inflammatory analgesics, in full dosage, to achieve both pain relief and some reduc- tion of the inflammatory component of the synovitis. There is, however, no evidence that these drugs influence the natural history of the disease, and therefore further drugs are continually being sought which will modify the disease process and suppress the synovitis and vasculitis in particular. Such drugs which are available at present are potentially toxic and have not been shown predictably to halt the progression of the disease. Thus their use, in a disease which is usually not fatal, depends on the careful assessment of the course of the disease in the indi- vidual patient and on the making of a balanced judgement between the likely effect of the disease, if not suppressed, and the possible toxicity of the treatment. The decision to use a drug which may suppress the disease, drugs which are also known as 'slow-acting', 'long-acting' or 'second-line' agents, depends on a definite diagnosis with evidence that the disease is progressing. Evidence of such progression may be demonstrated by some or all of the following features: (1) persistent active synovitis -prolonged morning stiffness -pain, warmth and tenderness of joints -clinical synovitis; (2) deteriorating functional capacity owing to active inflammatory synovitis; (3) increasing number of affected joints; (4) rheumatoid factor tests -becoming positive -increasing titre; (5) radiological erosions -development or increase in number; (6) development of extra-articular features, for example nodules, vasculitis, etc. The group of suppressive drugs consists of seven different classes of agents: (1) gold salts; (2) antimalarials-chloroquine preparations; (3) 'immunosuppressives' (cytotoxic: anti- metabolic agents); copyright. on April 10, 2021 by guest. Protected by http://ard.bmj.com/ Ann Rheum Dis: first published as 10.1136/ard.41.Suppl_1.1-a on 1 January 1982. Downloaded from
Transcript
Ann Rheum Dis (1982), 41, Supplement p 1
Introduction
The choice of, and the decision to apply, diseasemodifying drugs
COLIN G BARNES
From The London Hospital, London El, UK
Although rheumatoid arthritis (RA) is recognised tobe a systemic illness, frequently the only manifesta-tion is an inflammatory polyarthritis in whichsynovitis causes joint destruction. Extra-articular orsystemic features include vasculitis which, when pre-sent, may be the cause of potentially serious or fatalmanifestations. The aetiology remains unknown,although immunological disturbances are certainlyinvolved in the perpetuation of the disease. The for-mation of immune complexes within the joint isprobably responsible for the release of chemicalmediators of inflammation.The natural history of the disease is variable, rang-
ing from a brief episode of polyarthritis which resol-ves spontaneously, to a rapidly progressive arthritiswith widespread systemic features.The evidence for an immunological disturbance
arises from:(1) the presence of serum and synovial fluid
autoantibodies-rheumatoid factors;(2) the presence of other serum antibodies, for
example antinuclear antibodies;(3) the histology of the synovium which includes
(5) the presence of immune complexes in syno-vial fluid;
(6) reduced serum complement levels in patientswith active systemic disease;
(7) amyloid deposition.The first line of treatment always consists of the use
of non-steroidal anti-inflammatory analgesics, in fulldosage, to achieve both pain relief and some reduc-tion of the inflammatory component of the synovitis.There is, however, no evidence that these drugsinfluence the natural history of the disease, andtherefore further drugs are continually being sought
which will modify the disease process and suppressthe synovitis and vasculitis in particular. Such drugswhich are available at present are potentially toxicand have not been shown predictably to halt theprogression of the disease. Thus their use, in a diseasewhich is usually not fatal, depends on the carefulassessment of the course of the disease in the indi-vidual patient and on the making of a balancedjudgement between the likely effect of the disease, ifnot suppressed, and the possible toxicity of thetreatment.The decision to use a drug which may suppress the
disease, drugs which are also known as 'slow-acting','long-acting' or 'second-line' agents, depends on adefinite diagnosis with evidence that the disease isprogressing. Evidence of such progression may bedemonstrated by some or all of the following features:
(1) persistent active synovitis-prolonged morning stiffness-pain, warmth and tenderness of
Of these, the first four are well established in thetreatment of RA. Levamisole has been shown to beeffective in a majority of patients but is considered bymany to produce too high an incidence of toxic effectsto be considered for routine treatment.98 116 209 229The effectiveness of dapsone and sulphasalazineremains debatable.135-137
Historically gold salts are the most firmly estab-lished form of suppressive treatment of RA, havingbeen introduced over 50 years ago by Forrestier. Itwas not until controlled trials were conducted byFraser70 and the Empire Rheumatism Council63 thatits effectiveness was demonstrated. Nevertheless, itsmode of action remains uncertain, prediction of bothbenefit or toxicity in the individual patient is notpossible and measurement of serum levels has notbeen proved to be a useful method of monitoring goldtherapy.' 188
The chloroquine group of drugs were added next,initially being used empirically but later being shownto be effective in controlled trials,87 139 180 but lesseffective than gold or azathioprine.56The first report of the use of a cytotoxic agent in the
management of RA was in 1951, when Jimenez-Diaz'06 treated nine patients with nitrogen mustardon the basis that it was appropriate to treat prolifera-tive synovitis with an antiproliferative agent. Therefollowed other reports of the use of this agent, but by1964 interest had moved to the so-calledimmunosuppressives, in particular azathioprine,chlorambucil and cyclophosphamide. The rationalefor the use of these agents was that if an immunologi-cal disturbance plays an important part in thepathogenesis of RA, then it would be beneficial forthis immunological overactivity to be suppressed.Thus, by 1972 there were 25 published trials ofazathioprine in 350 patients, nine trials of chloram-bucil in 330 patients, and 12 trials of cyclophos-phamide in 370 patients, and an additional 235patients had been treated in trials with aminopterine,6-mercaptopurine or methotrexate. The majority ofthese trials were uncontrolled.'4 However, there havenow been many controlled trials, mainly of azathiop-rine and cyclophosphamide, which have demon-strated their efficacy and their toxicity, which havecompared them one with the other and also withgold salts, chloroquine and penicillamine, andwhich have compared different dosageschedules.5 18 34 35 38 56 84 89 95 124 143 224 225 237 239 Never-theless, it has not been convincingly demonstrated
that immune responses are suppressed in vivo.Of particular concern is the debate on the poss-ible oncogenic effect of the immunosuppressivecytotoxic agents,5 although to date a large follow-upsurvey has only revealed a small increase in thenumber of non-Hodgkin's lymphomata in patientswho are not organ transplant recipients.'15Most recently penicillamine has been studied, the
results of the first series of 21 patients being pub-lished by Jaffe in 1965.103 Although the mode ofaction of penicillamine remains unknown, controlledtrials have shown it to be effective and it has beensubjected to trials of different dosage regimens and incomparison with other suppressive agents. This drugtoo is potentially toxic.'8 48 97 99 133 155None of the agents available at present provides a
complete answer to the need to suppress RA, in thatapproximately 75 % of patients respond to treatmentand certainly a complete cure is not effected. Themeasurement of efficacy depends on clinicalresponse, improvement in laboratory parameters ofdisease activity and, possibly most importantly, thedemonstration of a reduction in progressive jointdamage. Only the 'immunosuppressive' cytotoxicagents have been shown possibly to reduce the prog-ression of radiological erosive change,34 38 and a his-tological study of the effect of chlorambucil on activesynovitis did not reveal any significant improve-ment.5
This symposium, therefore, aims to re-examine ourpresent state of knowledge of the effectiveness andtoxicity of the drugs currently available which maymodify the rheumatoid process, and of the'immunosuppressive' cytotoxic agents in particular.
Questions which may be posed are:(1) May these drugs be considered to modify the
progression of the disease by:(a) suppressing the activity of the inflam-
matory component of rheumatoid arth-ritis?
(b) reducing or preventing the progressionof erosive joint damage?
(c) suppressing the laboratorymeasurements of inflammation andimmunological overactivity?
(2) What is the long term toxicity of theseagents?
(3) What is the comparative efficacy and toxicityof the different agents?
(4) What are appropriate dosage schedules?(5) What are their modes of action?
We shall consider both clinical and laboratory dataand hope for a wide-ranging discussion between thespeakers and the audience.
Immunosuppression and the rheumatic diseasesA M DENMAN
From the Connective Tissue Diseases Study Group, Clinical Research Centre, Northwick Park Hospital,Watford Road, Harrow, Middlesex, England
SUMMARY Ignorance of the basic nature ofrheumatoid arthritis precludes the introduction ofrational schemes for using cytotoxic drugs. It is stillplausible that the autoimmune and other immunologi-cal abnormalities which accompany this diseaseare the secondary effects of persistent antigen, forexample, related to microbial infections. In this event,cytotoxic drugs may diminish the inflammatoryresponse but their effects on immune responseswould be irrelevant or even undesirable. Shouldrheumatoid arthritis prove to be a primary immuno-proliferative disorder, cytotoxic drugs may prove to beof value not because of their conventionalimmunosuppressive effects but because of theirselective action on the proliferating cells. Indeed,current evidence suggests that these drugs enhancerather than depress conventional immune responses,at least in the doses given to patients with rheumaticdisorders.
* * * * *
Cytotoxic drugs were introduced into rheu-matological practice because of rather simplisticviews about the pathogenesis of these disorders. Thebasic abnormality was considered to be a loss oftolerance to self-antigens, allowing the proliferationof abnormal clones of lymphocytes with self-reactivity.2"0 Cytotoxic drugs, it was thought, mighteliminate these clones thereby terminating autoim-mune diseases such as rheumatoid arthritis andsystemic lupus erythematosus. Furthermore, therewas experimental evidence that drugs such ascyclophosphamide might aid and abet the re-establishment of tolerance. The situation is now rec-ognised to be far more complicated. There is nolonger such confidence that these diseases can beexplained by simple unitarian theories, because auto-immune phenomena can be induced by a variety ofmechanisms and need not reflect the basicpathogenetic process in these diseases. Moreover,cytotoxic drugs and other immunosuppressive agentsdo not have a simple dose-related effect on immuneresponses, a possibility which earlier knowledge ofimmunological mechanisms could not take intoaccount.
It is now accepted that conventional and autoim-mune reactions involve the interactions of several
populations of lymphocytes and accessory cells, bothwith varying sensitivity to immunosuppressiveagents. Above all, there are populations of lympho-cytes which control the nature and magnitude of allimmune responses whether these are mediated bylymphocytes, antibodies, or even by non-specific'natural killer' (NK) cells. Thus immunosuppressivedrugs, which interfere selectively with the control ofimmune responses, may increase the vigour of suchresponses if the cells which effect these responses arethemselves resistant to the agent in question.There are additional difficulties because the causes
of most inflammatory connective tissue diseases areunknown. Until recently techniques for assayingimmune function in clinical practice have been crudeand inexact. Few clinical studies have taken phar-macokinetic factors into consideration when assessingthe effects of cytotoxic drugs on immune reactions. Itis hardly surprising, therefore, that there should belittle reliable information about the crucially impor-tant ways in which immunosuppressive agents affectthe clinical course of these diseases.
Persistent antigens and connective tissue diseases
Logic, rather than practical results, still inspires thebelief that persistent inflammation in the synovialmembrane and other tissues affected by rheumaticdisorders results from persistent antigen. Quite apartfrom the obvious interest in efforts to isolate theprovoking 'agent', this approach has importantimplications for interpreting the immunologicalaberrations which abound in the rheumatic diseases,and thus for the correct line of treatment. If a diseasesuch as rheumatoid arthritis is induced by persistentantigen, the disorder should logically be regarded as aform of immunodeficiency because of the host's in-ability to eradicate the causative agent. The arthritisand autoimmune features which accompany grossimmunodeficiency are often attributable to specificinfections. This arouses the suspicion that morespecific forms of immunodeficiency to preciselydefined organisms might account for similar dis-orders in patients without evidence of generalimmunodeficiency.1OMany of the causes of antigen persistence are also
liable to produce immunological abnormalities.These are of two kinds, the first being depressedimmune responses to the antigen itself, and the sec-ond being indirect in nature. For example, one post-ulated source of persistent antigen is bacterial cellwall products which escape immune elimination orenzymatic degradation."7 Bacterial cell walls, or pep-tidoglycans isolated from this source, induce a varietyof inflammatory lesions in experimental animals, thedistribution of which variously resembles rheumatoidarthritis or diffuse inflammatory connective tissuediseases. In addition, bacterial cell wall peptidogly-cans stimulate a brisk proliferative response by lym-phocytes and this is usually matched by equivalentantibody response."7 Nevertheless, bacterial infec-tions may fail to induce an appropriate antibodyresponse. One notable example is provided byexperimentally infecting mice with living or killedcocci.83 The infected mice show a severely impairedimmune response against the cell wall antigens in theinvading micro-organisms. In addition, the infectedmice display a variety of immunological aberrationswhich closely resemble those observed in patientswith rheumatic disorders. These include a diminishedcapacity to react against other antigens, the produc-tion of autoantibodies resembling rheumatoid factorand the generation of antibodies to bacterial cell wallglycoproteins which cross react with lymphocytemembrane glycoproteins. These lymphocyte reactiveautoantibodies are analogous to lymphocytotoxins,and show a restricted specificity for those lymphocyteantigens which are displayed only during the non-proliferative phase of the cell cycle. A wide spectrumof autoantibodies is also observed in experimentalanimals with chronic parasitic infections such astrypanosomiasis. 9Virus infections of man and experimental animals
also induce a varying range of immunosuppressiveeffects and immunological aberrations.43 A numberof mechanisms contribute to these effects. In man themost important of these is the ability of viruses togrow in sub-populations of those lymphocytes whichare essential for the induction of specific immuneresponses.'73 There is little evidence that virus infec-tions in man are potentiated by the ability of theinvading virus to inactivate the specific host immuneresponse to the virus in question. However, there arehuman diseases in which specific immune defects dopredispose to an unusually severe or atypical infec-tion. A notable example is the wide range ofimmunoproliferative disorders induced by Epstein-Barr (EB) virus infections in patients with pre-existing defects in specific cell mediated immunity tothis virus, or possibly with defects produced by theinfection itself. Thus, immunological aberrationswhich may result from persistent infection include
impaired immune responses, hypergamma-globulinaemia, rheumatoid factor and auto-antibodies of varying specificity, all of which haveoften been considered to be characteristic of primaryautoimmune disease. This conclusion has promptedthe therapeutic assumption that the suppression ofsuch abnormalities will ameliorate the disease. How-ever, if these abnormalities are the result, rather thanthe cause, of the inflammatory lesions in diseasessuch as rheumatoid arthritis, there are differenttherapeutic implications. Obviously, simply correct-ing these defects would have little relevance to theprimary cause of the inflammatory lesion. Equallyobviously, the most appropriate way of correctingthese abnormalities would be to remove the source ofpersistent antigen, but this can rarely be achieved.Nevertheless, it is pertinent to this argument thatimmunological aberrations are observed in patientswith sub-acute bacterial endocarditis, includinghypergammaglobulinaemia and the production ofrheumatoid factor and antinuclear antibodies, butthese abnormalities disappear in patients whose dis-ease responds to antibiotics. Similarly, the range ofautoimmune phenomena provoked by drugs mayequal that observed in patients with systemic lupuserythematosus of unknown cause, and theseimmunological abnormalities also resolve once theoffending drug has been withdrawn.235
Given that the object of immunosuppressivetreatment in most patients with rheumatic diseases isto control the disease process rather than to eliminatethe cause, one can question the extent to which suchtreatment can be expected to provide therapeuticbenefit if the immune aberrations are of secondaryimportance. There are a number of ways in whichcytotoxic drugs, and other immunosuppressiveagents, can blunt the inflammatory changes inducedby antigen persistency. In general terms these involvethe depression of inflammatory reactions which areset in train by specific immune events and are toowell-known to need further description. Theseinvolve humoral mechanisms with detectable cir-culating immune complexes, and a variety of cellularevents mediated by granulocytes, cells of themonocyte-macrophage series, 'natural killer' cellsand K cells. Experimental observations in man testifyto the suppressive effects of these drugs-on inflam-matory reactions and, perhaps the most telling obser-vation of all, in the frequency of opportunistic micro-bial infections and the reactivation of viral infectionsconsequent upon such treatment.
Nevertheless, some fundamental questions remainunanswered. The first concerns the extent to whichcytotoxic drugs really reinforce the effects of steroidson these inflammatory events, or whether they have amore fundamental effect on the immunological reac-
Immunosuppression and the rheumatic diseases Suppl p 5
tions which induce the inflammation. It is in thiscontext that the complexity of immune responsesmakes this a difficult question to answer. A variety ofsuppressor cells have been detected in man that couldinterfere with efficient responses to persistent anti-gens. For example, a sub-population of T lympho-cytes interferes with natural cytotoxicity againstvirus-transformed cell lines normally mediated byNK cells.41 There is good evidence also that varioussuppressor cell populations are activated in humanhosts chronically infected with different microbialinfections.2"7 Thus the inability to eliminate the puta-tive persistent agent in a disease such as rheumatoidarthritis could also result from suppressor cells inter-fering with the generation of an efficient cell-mediated immune response against persistentlyinfected cells. There is, for example, good evidencethat delayed hypersensitivity reactions to solubleproteins such as ovalbumin and bovine gammaglobu-lin are controlled by suppressor T lymphocytes. Cyc-lophosphamide ablates this suppressor mechanism,thereby producing exaggerated delayed hypersen-sitivity responses to these antigens.'68 The effects ofcytotoxic drugs in connective tissue diseases are,therefore, unpredictable. Logically, these drugscould overcome a defective immune response to apersistent antigen and improve responses to unre-lated antigens in addition.
The effects of immunosuppressive agents on
immunoproliferative diseases
The clonal theory of autoimmune diseases has beenstrengthened by recent observations concerning thenormal generation of immune responses. It is nowapparent that immune responses to foreign antigensare initiated by T lymphocytes which see such anti-gens in combination with self-antigens. T lympho-cytes destroy target cells bearing foreign antigensmost efficiently when the target cells and the T lym-phocytes concerned in their destruction have histo-compatability antigens in common. Indeed it is nowproposed that the principal function of histo-compatability antigens is to expedite the surveillancefunction of T lymphocytes.'60 Autoimmunereactions, therefore, are no longer viewed as qualita-tively different totally abnormal responses, com-pared with conventional responses, to exogenousantigens; it is the capacity of autoreactive cells toescape normal regulatory mechanisms which deter-mines whether or not such responses will predomi-nate. This concept has been strengthened byrepeated observations that autoreactive B lympho-cytes are found in normal subjects.2"O Theseregulatory defects were initially thought to involvecontrol mechanisms such as suppressor T lympho-cytes and the capacity of the normal immune system
to generate antibodies with specificity for unique Vregion sequences (idiotypes) in the immunoglobulinmolecule. More recently it has been established thatthe generation of antibody diversity in B lymphocytesinvolves a number of recombination steps betweengenes coding for different portions of the heavy andlight chain molecule. Thus, there is great potential forerrors in the sequence of events by which B lympho-cytes synthesise antibodies of different isotypes andof different specificities. Consequently, more atten-tion has been given to the possibility that auto-reactive B lymphocytes may proliferate in an uncon-trolled manner, because of errors in the differentia-tion of the precursor stem cells for these B cells or intheir maturation.50So far, the clonal origin of lymphoproliferative
diseases has been established only in malignantforms, such as chronic myeloid leukaemia andpolycythaemia vera. To some extent, cytogenetic andimmunochemical analysis of both membrane-boundand secreted immunoglobulins will establish thispoint. However, the point can be made most clearlyin black females who are heterozygous for the X-linked enzyme glucose-6-phosphate dehydrogenaseand who developed myeloproliferative diseases.67The cellular origin of these disorders can beexamined in detail in such subjects. B lymphocytes indiseases such as chronic myeloid leukaemia andpolycythaemia vera are derived from a single abnor-mal malignant clone even though these B lympho-cytes produce immunoglobulins with the full reper-toire of heavy and light chain determinants, aproperty previously considered to be characteristic ofB cells of polyclonal origin.'42 It has also been clearlyestablished that the autoantibodies in some patientswith cold agglutinin disease and haemolytic anaemiaare also of clonal origin. There is an attractive possi-bility, therefore, that the apparently polyclonal anti-bodies in other common autoimmune diseases, suchas rheumatoid arthritis and systemic lupuserythematosus, may also be produced by B lympho-cytes derived from a limited number of progenitorstem cells. By this concept, rheumatic disorderswould be considered as primary lymphoproliferativedisorders, and this would explain the failure to findthe postulated persistent antigens at sites of chronicinflammation. So far there is no firm evidence of asimilar nature in immunoproliferative diseases.Nevertheless, there are many striking similaritiesbetween the immunological features of the rheumaticdiseases and those detected in murine models ofhuman autoimmune disease.50 The immuno-pathological events in these mice have been inten-sively studied and the primary defect involves thedifferentiation and maturation of the B cell progenyof bone marrow stem cells. Moreover NZB mice at
first display a polyclonal hypergammaglobulinaemia,but monoclonal paraproteins subsequently emerge.This process can be correlated with the emergence ofa single abnormal clone, detectable cytogenetically,whose progeny eventually infiltrate the spleen andlymphoid system.There have been many attempts to demonstrate
that there is a fundamental abnormality in the regula-tion of immune responses in patients with rheumaticdiseases. Most attention has been directed at demon-strating a defect of suppressor T lymphocytes andabnormalities have been reported in patients withsystemic lupus erythematosus and their relatives.Until recently, less attention has been given to seek-ing defects in B lymphocyte function. However, it isnow clear that such abnormalities exist. Cultured Blymphocytes from patients with systemic lupuserythematosus spontaneously synthesise largeamounts of immunoglobulin in vitro, but arerefractory to pokeweed mitogen, a mitogen whichnormally induces polyclonal immunoglobulin syn-thesis by human B lymphocytes. Moreover, patientswith this disorder show impaired in vitro and in vivoantibody synthesis in response to immunisation withtetanus toxoid.159 Techniques are now available foranalysing the control of specific antibody synthesis invitro. Recent observations172 have shown thatwhereas the lymphocytes from normal subjects pro-duce antibody after in vitro challenge with influenzaviral antigen, lymphocytes from patients withsystemic lupus erythematosus fail to respond in thismanner. Experiments in which isolated populationsof T and B lymphocytes from lupus patients and fromnormal controls were co-cultured in vitro show thatthe defect resides in the B lymphocyte population.Thus B lymphocytes from lupus patients were able tofunction normally as 'helper' cells in specific antibodysynthesis, but B lymphocytes from lupus patientswere still defective even when combined with 'helper'T cells from normal donors. One can propose thatsystemic lupus erythematosus is an oligoclonal lym-phoproliferative disorder in which B lymphocyteswith memory for previous immunological events ex-press some conventional antibody response, but themajority of immunoglobulin synthesis is pro-grammed to produce autoantibodies. Speculationcontinues about the basic defect which could initiatethis uncontrolled proliferation. There are many,admittedly incomplete, hints about the possiblenature of these initiating insults. Drugs which bindDNA may induce a syndrome resembling spontane-ous systemic lupus erythematosus, and subjects whoacetylate such drugs poorly are particularly prone todevelop this complication.16 There is also a clinicalimpression that intercurrent virus infections exacer-bate inflammatory connective disorders. It is there-
fore possible that viruses and drugs act as pro-mutagens, that is, agents which bind to DNA andinduce mispairing with other bases.122 If not excisedby the appropriate excision and repair enzymes, per-sistence of promutagens could, in proliferating lym-phocytes, produce mutations coding for abnormalimmunoglobulins with autoreactive specificities.There are indications that lymphocytes from patientswith inflammatory connective tissue diseases, includ-ing rheumatoid arthritis, are abnormally susceptibleto promutagens such as alkylating agents and low-dose ultraviolet irradiation, and this sensitivityappears to be related to a deficiency in removingalkylated bases from lymphocyte DNA.88
Should inflammatory connective tissue diseasesprove to result from primary immunoproliferativediseases, this would have clear implications for theways in which immunosuppressive agents mightoperate. Under these circumstances the aim oftreatment would be to eradicate abnormal auto-antibody producing cells in a manner analogous tothe treatment of malignant lymphoproliferative dis-eases. Moreover, drugs such as cyclophosphamideand azathioprine would be more likely to suppressthese cells rather than lymphocytes producing con-ventional antibodies, since such lymphocytes arehighly resistant to cytotoxic drugs. Indeed the resis-tance of conventional secondary antibody responses,but sensitivity of autoantibody production tocytotoxic drugs, points to this conclusion.
Clinical observations in patients receiving immuno-suppressive agents
Considerable information has accumulated concern-ing the effects of immunosuppressive agents on vari-ous immunological functions. Unfortunately, theresults have been conflicting and confusing. Suchcriticisms particularly concern attempts to assay thein vitro function of lymphocytes isolated frompatients receiving immunosuppressive drugs. Tech-nical problems account for some of the discrepanciesin published results. Most investigators have carriedout tests of lymphocyte function in vitro using a singleconcentration of -cultured cells and a standardmitogen challenge. However, it is now clear that theconditions for detecting peak responses to mitogenicstimuli vary in different diseases and are affected bymany secondary factors.65 Another problem is theunderstandable failure of earlier workers to appreci-ate the complexities of immune responses. It is nowapparent that meaningful assays of immune functionmust include tests for all the populations of inflam-matory cells which contribute to tissue damage andwhich modulate these effects. Two principalstrategies are being introduced. The first relies on the
Immunosuppression and the rheumatic diseases Suppl p 7
ability of monoclonal antibodies to identify suitablemarkers of each function of sub-populations of lym-phocytes, and to score these both in the peripheralblood and in the inflammatory lesions of patientswith rheumatic disorders. The second involves func-tional assays, such as specific antibody production,and, of equal importance, suppressor cell activity. Sofar there is too little evidence to make dogmaticstatements about the extent to which differentimmunosuppressive agents, or regimens, have selec-tive effects on the different cell populations andimmune mechanisms which contribute to the diseaseprocess. Nevertheless, there are already firm indica-tions that theoretical immunosuppressive regimensdo not predictably suppress conventional immuneresponses as judged by current standard techniques.Indeed, most observations of standard immuneresponses in such patients have indicated that theseresponses were as often increased as depressed bycytotoxic drugs. Paradoxically, too, immunosuppres-sive treatment commonly improves bone marrowfunction and host responses to infection in subjectswith bone marrow depression and immunodeficiencyattributable to connective tissue diseases. This pointis emphasised by experience in a comparative trial inwhich patients with systemic lupus erythematosus,polyarteritis, Behqet's syndrome, or dermatomyositishave been treated with steroids alone in conventionaldoses, or with a combination of high dose steroid,antilymphocyte globulin and cytotoxic drugs.44 Con-trary to expectations, bone marrow function andresistance to infection improved more rapidly inpatients receiving the more intensive regimen. Thispoint is illustrated by two clinical examples.The first (fig 1) is a 16-year-old patient with
systemic lupus erythematosus, presenting withgrand-mal fits and skin lesions, with clinical cutaneousvasculitis, hypertension, hepatosplenomegaly andprogressive renal disease. He had strongly posi-tive tests for antinuclear antibody, circulatingimmune complexes in high titre and hypocomple-mentaemia. He failed to respond to steroids in highdosage (up to 100 mg prednisone daily) anddeveloped congestive heart failure and staphylococ-cal septicaemia. Treatment with intensiveimmunosuppression produced a full clinical remis-sion and the disappearance of most of the characteris-tic serological abnormalities. However, it is note-worthy that his haemoglobin and white cell count alsoreturned to normal on treatment. Depressed lym-phocyte function, which had been detected in vitroassociated with his disease, also reverted to normalon allegedly immunosuppressive treatment.The second case is an 18-year-old boy who
developed polyarteritis with a mononeuritis multip-lex, vasculitis leading to digital ischaemic changes
AN 16 years male SLE
16-
Hb (g/1) 14-
10. 16'0
WBC x 10-3/l43t6
80~C3 (mg%) 40J
DNA b(%) 1]50]
IC * + + 44+ +4 - -
ANA 320 320 0 S0 0 320 60400 60 32040 40
Antibiotics s150
Prednisone mgA
1977 1978 1979 1980 1981
FIG 1 Bone marrow recovery in a patient with systemiclupus erythematosus receiving intensiveimmunosuppression. (C3 =complement, DNA =DNAbinding assay, IC=immune complexes, ANA =antinuclearantibody reciprocal titre.)
FIG 2 Response ofpolyarteritis to immunosuppressivetreatment. The three phases ofthe disease are described in thetext. Cytotoxics: azathioprine and cyclophosphatnide;prednisone: 150 mglday maximum; anti-platelet: aspirin,sulphinpyrazone, tolazoline. ALG =antilymphocyteglobulin.
and renal disease (fig 2). The first phase of his illnesswas characterised by severe clinical disease whichresponded to immunosuppression with high dosesteroids (maximally 150 mg prednisone daily),azathioprine, cyclophosphamide and anti-lymphocyte globulin (ALG). His disease went intoremission and all drugs were withdrawn (phase 2).However, the vasculitic lesions and neuropathyreappeared although, at this stage (phase 3), thelesions were characterised histologically by anobliterative endarteritis rather than by inflammatory
vasculitis. He eventually recovered with furthertreatment with high dose steroids and cyclophos-phamide. It was noteworthy that resistance to secon-dary infections accompanied periods of active diseasebut were controlled once adequate immunosuppres-sive regimens were instituted.These findings do not help us to distinguish be-
tween the two major hypotheses explaining the clinicaland immunological abnormalities of the rheuma-tic diseases. Improved conventional immuneresponses and the disappearance of immunologicalanomalies, such as autoantibody production, wouldbe expected if cytotoxic drugs corrected the abnor-malities consequent on antigen persistence, for ex-ample, in association with persistent infection.Equally, if cytotoxic drugs eliminate clones of auto-antibody producing cells which are particularly sensi-tive to these drugs, the proliferation and maturationof B lymphocytes derived from normal stem cellsmight be improved, thereby allowing conventionalimmune responses to recover. The immunodefi-ciency associated with malignant myeloproliferativediseases, such as the leukaemias, is indeed oftenreversed as the result of chemotherapy which hasbeen regarded as a favourable prognostic sign.
ConclusionsDespite many years of treatment with cytotoxicdrugs, no clear picture has emerged of the ways inwhich these drugs influence inflammatory connectivetissue diseases. The issue is therapeutically impor-tant since, if steroids produce their undoubtedbeneficial effect through immunosuppressive mechan-isms, then any drugs which potentiate those effectsshould be clinically beneficial. Ironically, rapidimprovements in our knowledge of the pharmacolog-ical events in inflammation have called in questionthe belief that the therapeutic benefits of steroidsowe anything to their effects on lymphocytes andimmune function. Indeed it is arguable that the latterare undesirable side effects of steroid treatment.Given our ignorance of the pathogenesis of thesediseases and the many plausible ways in whichcytotoxic drugs could produce benefit, it is hardlysurprising that the issues remain unresolved. Thequickest way to resolve the dilemma will be to eluci-date the pathogenesis of these diseases. In the mean-time, technical improvements in assaying responsesin clinical practice at least allow the prospects ofpursuing meaningful studies concerning the effects ofcytoxic drugs on immune function. copyright.
on April 10, 2021 by guest. P
rotected byhttp://ard.bm
j.com/
Ann R
heum D
is: first published as 10.1136/ard.41.Suppl_1.1-a on 1 January 1982. D
From the Wellcome Research Laboratories, Research Triangle Park, NC 27709, USA
SUMMARY Immunosuppressive agents have diverse(although often multiple) sites of action in the cellsequences that are involved in immune responses.
New routes to selectivity are apparent at both thecellular and the biochemical level. Meanwhile, clini-cal work is finding new uses and more selective em-
ployment of the currently available agents.
* * * * *
Immunosuppression had a spectacular debut with thefinding, by Schwartz et al,3t 193 194 that immunologicaltolerance could be produced by the administration ofpurine-6-thiol (6-MP) with, or soon after, a dose ofbovine gammaglobulin to rabbits. The production oftolerance was closely dependent on timing of drugand antigen administration, as shown in fig 1. It also
FIG 1 The effect of timing ofdrug administration withrespect to antigenic stimulus. Data drawn from Schwartz etal.31 193 194 In each case bovine gammaglobulin 2 mglrabbitwas administered on day 0. In panel A the drug (horizontalbar) wasgiven before the antigen, in panelB during antibodyproduction, and in panel C starting with antigenadministration and continuing 7 days.
was dependent on both the dose of antigen and thedose of drug, so that it had characteristics of theimmune paralysis induced by massive doses of anti-gen.9" Moreover, both Schwartz et al and Frisch andDavies73 showed that it could be highly specific.Frisch and Davies, for example, gave human erythro-cytes to mice, waited two days, then gave a singledose of thioguanine and sheep red blood cells. Anti-bodies were produced to the sheep erythrocytes, butnot to the human.The observations of Schwartz and Dameshek.94
were quickly developed in two directions. Theybegan investigations of autoimmune diseases that arestill continuing and are the major focus of today'sworkshop. The finding also provided entry into thefield of organ transplantation, then awaiting only ameans of controlling the rejection of allograftedorgans. This field has made rapid progress; todayrenal allografting is regarded not as an experimental,but as a therapeutic procedure, with a high probabil-ity of long-term survival.58 Almost from its inception,renal transplantation has involved the use ofazathioprine, which had been synthesised originallyas a masked form of 6-MP for anti-cancer studies,59and it has superseded 6-MP, not only in this, but alsoin the autoimmune field.60 92 An overview of thescope of this field is provided in table 1 which simplylists the types and numbers of clinical investigationsknown to be in progress in the USA.Many of the problems and difficulties of interpre-
tation that could be perceived in 196791 are still with
TABLE 1 The types and numbers ofclinical investigations ofautoimmune disorders known to be in progress in the USA,using azathioprine
others. Before attempting to discuss these fields, it isprobably pertinent to mention the numerous addi-tional substances that have immunosuppressive
activities. Most of these are still being drawn from thearmamentarium of anti-tumour agents9' (table 2). Itwas this multiplicity of agents that prompted thequestion "are all immunosuppressants the same?"The short answer, of course, is "no". An attempt willbe made in this paper to analyse the differences fromseveral viewpoints and to point the way towardsfirmer knowledge in several essential, but possiblyancillary, areas.
The last two decades have witnessed remarkableadvances in the recognition of lymphocyte types andfunctions. What in the early 1960's was simply a
distinction between cell-borne and humoral immun-ity soon became a distinction between T and B lym-
phocytes. The importance of the thymus was el-egantly shown by the experiment depicted in fig 2.Meanwhile, the past decade has seen a rapid prolifer-ation in the recognition of various subclasses of Tlymphocytes86 tt' and factors involved in their dif-ferentiation and activities. Much of this material willbe dealt with elsewhere in this workshop; however,an attempt will be made here to present some of theeffects of drugs on specific cell types. A selectionamong such reports is presented in table 3. This is, in
fact, a selection not guaranteed for immutability,since surveys of the literature reveal a dependence ofsuch identifications on the specific conditions of thereporting laboratory. However, one must regard suchidentifications as a step in the understanding of themechanisms of immunosuppression that may lead tothe selective use of particular agents.What may eventually be regarded as the most
important breakthrough of all is the recent discoverythat deficiencies in certain enzymes, concerned withpurine intermediary metabolism, are associated withimmunological defects. he first of these to bereported was congenital deficiency of adenosinedeaminase (ADA) in a patient with severe combinedimmunodeficiency affecting both B and T cells.78Three years later, the same group found another
IABtlE 2 Ilsnnunosuppressive agetits
A SteroidsB Alkylating agentsC Antimetabolites
(I) Purinethiols(2) Inhibitors of dihydrofolate reductase(3) Nucleoside analogues
D IntercalatorsE Cell cycle blockcrsF EnzymesG Antibody
84
76
48
x
40
0
D 32E
-C24
0.E>1 16
8
0
,
0
0
0
0
-U-
0
0
as ~0
00 0~~~~~~~~~~~~~5
~ ~ ~ Smoa
_
(OI
'r1e
Response cf splenic lymphocytes
Fi(G 2 Documenetation of the role of the thYmnus in thedifferemttiation of lymphoctes. The respotnses oJspleniclymphocytes to culturinig itn the presence ofj(oncanavalin Aare shown. (Courtesyv of Life Sciences Research Reports1981; 11: (1). SRI internzational Metilo Park, CA 94025,USA.)
'IABLE 3 Reports of cellular respottses toimmunosuppressive agenits
Agent Effects Reference
Cyclosporin TCGF response 120Dexamethasone TCGF production 1 12(0Azathioprine Inhibits RFC (T cells) 53 228Cyclophosphamide 1B cells, 4suppressor T cells 128 197Melphalan lActivated B cells 76BCNU JActivated and resting B cells 76
deficiency, of purine nucleoside phosphorylasc(PNP),77 that may eventually be even more helpful inunderstanding the biochemical events of the immuneresponse. Most of the explanations of the effects ofADA deficiency revolve around the toxicity ofdeoxyadenosine. However, PNP deficiency is selec-tively toxic to T cells.77 Deficiencies in phos-phoribosyltransferases have only minor effects,although some impairment of B cell activities have
Differences among immunosuppressive agents Suppi p 11
been noted.: The explanation of the selective effectsof PNP deficiency is attributable to the existenceof a deoxycytidine kinase that phosphorylatesd-guanosine (d-Guo), whereas in the ribo-seriesthere is no similar enzyme. The deoxyguanylateeventually reaches the triphosphate stage and exertsa feedback inhibition of ribonucleotide reductase.The phosphorylation is the preferred route ofmetabolism of d-Guo in T cells, but is almost non-
existent in B cells. Thus, d-Guo is highly toxic to Tlymphoblastoid cell (T-LCL) lines but only mildly so
to B-LCL, and not at all to B-PNP deficient.6 Thesefindings are summed up in table 4.
It is apparent that observations of this sort may notonly lead to more precise definitions of immuneresponse, but lay the basis for rational approaches toselective drugs.
Above the line, abnormalities producing major effects on immune responses.Below the line, deficiencies having at most minor effects. ADA=adenosinedeaminase, PNP=purine nucleoside phosphorylase, APRT=adeninephosphoribosyl transferase, HGPRT=hypoxanthine guanine phosphoribosyltransferase, d-Ado=deoxyadenosine, ATP=adenosine triphosphate,dATP=the corresponding deoxyribose derivative, cAMP=cyclic adenylate,d-Guo=deoxyguanosine, d-GTP=deoxyguanosine triphosphate. Arrowssignify increases (up) and decreases (down).
SUMMARY Cytotoxic drugs are effective in diseasesuppression and affect the immune system. The cor-
relation between these is poor, in many patients, atthe dosage used in man. This may partly explain whythe large increased risk of neoplasia, initially pre-
dicted, has not been shown to follow the use ofcytotoxic therapy.The cause of death in patients with rheumatoid
arthritis has been assessed both by surveys based on
clinical populations and by reviews of necropsy
records. There have been few attempts to assess theinfluence of drug therapy on mortality, or on theincidence of neoplasia. In addition, evidence isaccumulating to suggest that lymphoreticulartumours may be associated with autoimmune dis-ease.
There have now been several long-term studies ofpatients with rheumatoid arthritis treated withimmunosuppressives; azathioprine and cyclophos-phamide in particular. These studies are reviewedand it is concluded that the incidence of most of thecommon cancers is not increased.
* * * * *
The use of immunosuppressive drugs in rheumatoidarthritis has, in general, proved to be of benefit.82 143Azathioprine or cyclophosphamide, started early inthe course of rheumatoid arthritis, have been shownto be equally effective and similar in their efficacy togold therapy.38 Since a short course of such treatmentdoes not produce a lasting remission, long-termadministration would be expected in many instances.This focuses attention on the adverse effects of pro-longed administration of these drugs including bonemarrow suppression, susceptibility to infection andthe potential development of malignant tumours.Some of the potential toxic side effects are listed intable 1.
TABLE 1 Toxicity of immunosuppressive drugsCyclophos- Azathioprine Chloram- Methotrexatephamide bucil
The relative contributions to the patient'sincreased susceptibility to infections arising from thedisease itself, and from the frequent concomitantsteroid therapy, remain difficult to evaluate. Infec-tions with organisms such as cytomegalovirus, herpessimplex, mycobacterium tuberculosis and fungi arepossible. The use of cyclophosphamide may beaccompanied by alopecia, haemorrhagic cystitis andbladder fibrosis; it may also lead to ovarian failureand azoospermia. Azathioprine is apparently free ofsuch side effects.
Cytostatic drugs are potentially teratogenic inman, although several women receiving azathioprinehave conceived and delivered healthy babies.Methotrexate, if given repeatedly, can lead to hepaticfibrosis. The most obvious effect of these drugs is onthe haemopoietic system. Since they affect rapidlyproliferating cells they have an effect on bone mar-row precursors, such as stem cells and the myeloidand erythroid, as well as the lymphoid, series.
Induction of malignant disorders
The most difficult general problem to assess, associ-ated with the use of immunosuppressive agents, is thepotential threat of the development of neoplasia. It is
The comparative incidence of malignant disease in rheumatoid arthritics Suppl p 13
theoretically possible that the development of a neo-plasm may occur many years after continuousimmunosuppressive therapy or even after discon-tinuation of such therapy. It took many years torecognise the increased risk of leukaemia that occursin patients who had received radiation therapy forankylosing spondylitis some 20 to 25 years previ-ously.37Over the last decade considerable evidence has
accumulated suggesting that immunosuppressive andcytotoxic drugs cause an increased incidence ofmalignancy, particularly when used for long periods.Most of the evidence is derived from studies ofpatients receiving organ transplants, and more anec-dotal evidence comes from the use of these drugs inrheumatic diseases. Animal experiments show thatimmunosuppression facilitates the transplantation ofmalignant cells, increases the incidence of virus-inducedcancers and accelerates the growth of meta-stases. Perhaps the expected increased risk of neo-plasia, initially predicted, has not materialised becausethere is little correlation between efficacy of diseasesuppression and effect on the immune system in man.Recently a prospective study"15 of 3823 patients
who had received a renal transplant and of 1349other patients, including cases of rheumatoid arth-ritis, receiving cytotoxic drugs (particularly azathiop-rine), showed no clear evidence that immunosup-pressive drugs produce an increased risk of most ofthe common cancers. However, few of the patientshad been under observation for over five years, andan increase may appear later. Follow-up of the renaltransplant recipients showed an almost 60-foldincrease of non-Hodgkin's lymphoma together withan excess of squamous cell skin cancer and mesen-chymal tumours. The patients without transplantsshowed an excess of the same tumours, though to alesser extent.The most striking excess of lymphoid tumours
occurred as non-Hodgkin's lymphomas (mainlydescribed as reticulum cell sarcomas and microg-liomas) in the transplant series, which confirms whathas been reported before.4 This excess of lymphoidtumours was extraordinary in its magnitude, pre-dilection for the brain and very short inductionperiod. Possible hypotheses to explain these findingsinclude disturbance of immunosurveillance or ofgraft-versus-host reaction, chronic antigenic stimula-tion and the effect on oncogenic viruses. For each ofthese there is laboratory support. The short inductionperiod, sometimes within a few months after trans-plantation, suggests a viral origin, since transforma-tion could take place immediately if the virus wasalready present.That patients without transplants also showed an
excess of lymphoid tumours suggests that
immunosuppression is important, though the pres-ence of a graft may contribute to tumour develop-ment, since in transplant recipients the excess wasmuch greater. Other observations support the prim-ary importance of immunosuppression. Firstly, therewas no tendency for lymphoid tumours to be morecommon in recipients who experienced repeatedrejection crises, nor in those who had had multiplegrafts. Secondly, a study of patients treated by renaldialysis, who have severely depressed immunereactions, showed that they also had an increasedincidence of non-Hodgkin's lymphoma, and that inone of them, who had never been given immunosup-pressive drugs, the lymphoma affected the brain.Thirdly, certain rare hereditary disorders character-ised by major immunological impairment, such as theWiskott-Aldrich syndrome, are associated with anincreased risk of lymphomas.A direct mutagenic effect of the chemical agents
appears to be unlikely because of the abrupt increasein incidence, the subsequent constancy of the riskfrom the time of first exposure and the lack of anynoticeable dose effect, all of which are in pronouncedcontrast to the usual findings with known chemicalcarcinogens.
Only four patients without transplants developednon-Hodgkin's lymphoma (three of whom weretreated with azathioprine)."5 A three-fold excess oflymphomas was observed in a large series of patientswith rheumatoid arthritis in Finland"'' and it was notclear whether this was associated with any particularform of treatment. However, cases of lymphomaaffecting the brain in patients without transplants,but receiving immunosuppressive drugs, have beenreported.'27 220 221The risk of non-Hodgkin's lymphoma in the study
of Kinlen et al"5 was smaller in patients withouttransplants than in transplant recipients. Differencesin the type, amount and duration of drug treatmentwere, however, substantial. Azathioprine was giveninitially only to 64'%, of patients in the non-transplantseries and was continued for over two years in lessthan 30'Yo, whereas azathioprine was continued inover 90 % of the transplant recipients throughout theperiod of observation. Similar differences wereobserved in the case of prednisone.Walder et al in 197 121' emphasised the common
occurrence of squamous carcinomata in transplantrecipients. Kinlen et al"5 also reported a ten-foldincrease in mortality from skin cancer in transplantrecipients. In the latter study there were three casesof acute leukaemia, all in transplant recipients. com-pared with the expected figure of less than one, andsix mesenchymal tumours when none was expected;three of these were in the non-transplant group.There were four cases of bladder cancer in patients
without transplants and the risk of this disease isrelated specifically to the use of cyclophosphamide.
In order to try to distinguish between the effects ofimmunosuppressive therapy and the disease processin rheumatoid arthritis, surveys of the cause of deathmust be examined.
Cause of death
The cause of death in patients with rheumatoidarthritis has been assessed both by surveys based onclinic populations and by reviews of necropsyrecords.13 33 55 75 102 151 156 167 219 226 There have beenfew attempts to assess the influence of drug therapyon mortality or on the incidence of neoplasia.Most surveys of the causes of death in rheumatoid
arthritis agree that patients with the disease have ahigher mortality rate than the rest of the population,though they die not so much from specific complica-tions of the disease as from all the usual causes ofdeath, but at an earlier age than usual.
In 1970 Uddin et al219 reviewed 475 patients withrheumatoid arthritis, first seen between 1954 and1966. The cumulative survival rate was determinedin each year for up to ten years, compared with that ofa normal matched population. This was lower thanexpected for each sex at each year of follow-up, par-ticularly in the later years.Cobb et al in 195333 followed 583 patients for a
mean duration of 92 years. The overall mortality ratewas 24 4 per 1000 patients per year in contrast to18 9 for a comparable group matched for age and sex.More recently, Isomaki et al in 1975102 observed1000 patients with rheumatoid arthritis for threeyears. All were over 40 years of age. Deaths werecompared with an age and sex matched control popu-lation. The 122 deaths in the rheumatoid group com-pared with 69 in the controls. Deaths from cancerwere fewer in the rheumatoid group.
It is difficult to compare studies of the causes ofdeath in rheumatoid patients. Most studies rely oncertified causes of death which, on occasion, do notcorrelate well with findings at necropsy. There arealso differences in the periods of follow-up.The improved hospital supervision of rheumatoid
patients and the use of an expanded range oftherapeutic agents in the last decade may invalidatecomparisons with early studies.
Reports of the incidence of neoplasia, independentof therapy, vary between studies. Only Moesman in1969151 found a greater frequency of malignant dis-ease in a small group of elderly patients.
Induction of malignant disorders
In view of reports of malignancy, Hunter et al in
197595 carried out chromosomal studies in patientsreceiving long-term azathioprine. This showed atwo-fold increase in chromosomal abnormalities inpatients, similar to that reported during cyclophos-phamide treatment of rheumatoid arthritis by Tol-chin et al in 1974.216 Since similar changes may occurwith phenylbutazone, and a variety of other drugs, itssignificance will have to await long-term observation.
Evidence is accumulating to suggest that lym-phoreticular tumours may be associated with auto-immune disease. 28 149 211 Patients with Sj6gren'ssyndrome are especially prone to develop extensivelymphocyte infiltrates in extrasalivary organs withpseudolymphoma formation and occasional progres-sion to malignant lymphoma.6 Three series ofpatients with systemic lupus erythematosus haveshown an increased risk of cancer.29 52 126
Azathioprine has been widely used as animmunosuppressive agent in the United Kingdom,whereas cyclophosphamide is used more widely inNorth America.Owing to the influence of de Seze and Kahn,46
chlorambucil has been the most widely usedimmunosuppressive drug in France. This choice wasbased on the use of the drug in chronic lymphaticleukaemia where there were hopes that it would havea greater affinity for lymphocytes. The enthusiasmfor the drug declined during the 1970's on account ofits side effects. Deshayes et al47 summarised theresults concerning side effects of chlorambucil notedby several French authors in a total of 495 patients,and compared these to the effects of azathioprine andcyclophosphamide.Treatment linked mortality probably depends on
the duration of exposure to risk, which is difficult todetermine for a treatment with prolonged after-effects. Its assessment depends on the length of timefor which the patients are followed after treatmenthas begun, and on whether all deaths are included oronly those which can be attributed to the treatment.Most important is the fact that no true control groupexists, and a comparison can only be made with thelife expectancy of the general population of the sameage or that of a random group of rheumatoid patients.
Farber et al64 compared 126 patients treated withcytotoxic drugs with 126 age and sex matchedrheumatoid arthritic patients in hospital during thesame years but who did not receive cytotoxic therapy.Patients predominantly received nitrogen mustard orcyclophosphamide or both. The patients werefollowed from 1965 to 1974 and no increase inmalignancy was recorded, nor were any lymphomasseen. This view is not shared by others. Parsons andco-workers in 1974'69 carried out a long-termfollow-up of patients with rheumatoid arthritistreated with cyclophosphamide. They found eight
The comparative incidenice of malignanit disease in rheumatoid arthritics Suppi p 15
malignancies including seven of the lymphoprolifera-tive variety, but the incidence of Sjdgren's syndromeis not recorded. Pollock et al in 1973t79 reported twomalignancies in rheumatoid arthritic patients treatedwith cyclophosphamide (both solid tumours), and a
recurrence of a previously treated melanoma was
reported in a patient following cyclophosphamidetherapy.
Renier et al in 1978t84 followed 131 cases ofrheumatoid arthritis for at least four years andrecorded 13 deaths through malignant disease, ofwhich seven were haemopoietic neoplasms. Cancerwas diagnosed in five other patients. Eighteen of the131 patients had received chlorambucil, and thenlater cyclophosphamide, and seven of the cases ofmalignant disorder were observed in these 18patients.Four cases of leukaemia have arisen in association
with rheumatoid arthritis following treatment withazathioprine,32 t:t t96 and in two cases followingtreatment with cyclophosphamide.tttKahn et al in 1 979, 08 using results from a collective
retrospective study, calculated the incidence of acuteleukaemia during and after cytotoxic therapy. Theresults of this study, which included 2006 patientswhose average age was 55 years, are shown in table 2.A total of 1711 patients had received chlorambuciland were followed for one to 13 years. One third ofthe patients were not followed up. Kahn et altos com-
pared these figures with cyclophosphamide treat-ment (three cases of leukaemia in a co-operativestudy of 229 patients) and with azathioprine, whereSeidenfeld et alt96 noted an incidence of 1-3 'Yo. Kahnet al emphasised that no case of leukaemia was seen in
patients who received less than Ig chlorambucil andin whom the duration of treatment was less than sixmonths. The nature of the acute leukaemia withchlorambucil was variable, and the delay from thebeginning of treatment to the appearance ofleukaemia was 5-7+2-9 years.tus 145 The risk ofdeveloping leukaemia seems more pronounced in
children suffering from chronic rheumatoid arthritis;three cases out of 40 infants treated with chloram-bucil compared with 0 out of 160 control infants.`
TABLE 2 Results ofa survey of the incidence of acuteleukaemia during and after therapy with chlorambucil (Kahnet all"5)Reason for therapy Number in study Number ofcases
Study of cause of death with particular reference toazathioprine
In view of the few attempts to assess the influence ofdrug therapy on the incidence of neoplasia, theauthor initiated a survey on a group of closelyfollowed patients with rheumatoid arthritis pre-dominantly treated with azathloprine. "'The patientsin this study were unusual in that treatment withazathioprine was started, in some patients, as long as11 years previously, often in high dosage (5mg/kg/day) and sodium aurothiomalate (Myocrisin)had been given in combination. For purposes of com-parison, an analysis of mortality was performed bycalculation of expected mortality in the patientgroup, based on rates for a similar age and sex distri-buted sample of the general population.A total of 311 patients with classical or definite
rheumatoid arthritis was studied over this 11-yearperiod. Follow-up was complete in all patients. Thedeath certificate was obtained for all the 46 who died,and results of necropsy were available in 19 cases.A total of 214 patients received cytotoxic therapy
(table 3) and 63 received cyclophosphamide (range0-2 to 8 years) or chlorambucil, alone or in combina-tion with azathioprine, at some stage in their disease.The mean duration of treatment in the 151 patientstreated with azathioprine alone was 2-5 (range 1 to11) years, a total of 500 patient years.The tumours diagnosed in both the dead and the
survivors are recorded in table 4. Five patients havehad apparently successful treatment of their neo-plasm (three patients had received cytotoxic therapybefore the discovery of the neoplasia). Ten of the 20neoplasms occurred in patients who were receivingcytotoxic therapy (table 5). There is no evidence tolink rheumatoid arthritis with any particular form oftumour.The one case of lymphosarcoma occurred in a
71-year-old patient, without evidence of Sjogren'ssyndrome, who had not received cytotoxic therapy.
TABLE 3 MalignancY associated witi treattment
Method ofmanagement Number of Number of Incidencepatients malignancies (5%)
The principal causes of death are listed in table 6.Ischaemic heart disease heads the list, followed byneoplasia. Deaths from infections all involved thelung and the incidence is lower than previouslyrecorded in patients with rheumatoid arthritis.24 167
The observed death rate among the rheumatoidarthritis patients, from all causes according to age,has been compared with expected values obtainedfrom the Registrar General's records using a x2 test.The death rate was higher than expected in the 45 to64-year-old age group, but lower in those aged 75years or over. The differences are significant(p<0 0005). Analysis of age related specific causes ofdeath (table 7) showed an increased frequency ofdeaths from neoplasia and ischaemic heart diseaseoccurring in the 45 to 64-year-old age group, but thisis not significant.There was increased evidence of neoplasia in the
patients with rheumatoid arthritis (table 3) but thiswas not associated with immunosuppressive treat-ment, nor was there any difference in duration oftreatment between those developing tumours com-
pared with the rest.One patient developed a transitional cell car-
cinoma of the bladder after eight years' therapy withcyclophosphamide. This is of interest, as Wall and
TABLE 6 Causes ofdeath in rheumatoid arthritis (46patients)
Clausen in 1975532 described five cases of carcinomaof the bladder after prolonged cyclophosphamidetherapy. In each case interstitial haemorrhagic cys-titis preceded the development of the carcinoma. Thepatients all developed fatal and invasive carcinomas.Dale and Smith in 197439 suggested that the tumourcould be small and possibly non-recurrent. Thepatient in this study has had local recurrences whichhave been removed at routine cystoscopy.The presence of haemorrhagic cystitis in a patient
receiving cyclophosphamide should suggest immedi-ate and permanent discontinuation of the drug.However, the cessation of haematuria and with-drawal of the drug does not eliminate the need forcystoscopy, and probably this should be repeatedperiodically, indefinitely thereafter. Previous obser-vations have shown atypical epithelial cells oncytological examination of urine in haemorrhagiccystitis during treatment with cyclophosphamide.6" Anecropsy study showed that ten of 49 patients whohad received long-term cyclophosphamide had signsof urinary bladder fibrosis, the frequency beingrelated to dose and duration of therapy. In about50% of cases the fibrosis did not produce symptoms.These findings suggest caution in the use of cyc-
lophosphamide for non-malignant conditions. It issuggested that protracted periods of treatmentshould be avoided, and that the drug be used inintermittent courses if at all possible.
Louie and Schwartz130 compiled a list of 109treated patients who developed neoplasms and com-pared them with patients who did not receive
The comparative incidence ofmalignant disease in rheumatoid arthritics Suppl p 17
immunosuppressive therapy. The most prominentdifference between the treated and untreatedpatients was the increased incidence of acute myelo-blastic leukaemia in the treated group. The mostimportant association was therapy with an alkylatingagent. Certainly this review of the literature suggeststhat these agents are more likely to produce prob-lems. However, assessment of the true risk is difficult
as we do not know the expected number of neoplasmsin untreated patients.The potential oncogenic effect of immunosuppres-
sive drugs remains one of the main deterrents to theirwide-spread use in the treatment of rheumatoid arth-ritis. Further studies are required to assess the trueinfluence of these drugs on the incidence of tumoursin rheumatoid arthritis.
Long-term effects of azathioprine in rheumatoidarthritis *
M B UROWITZ, H A SMYTHE. T ABLE, C S NORMAN AND C TRAVIS
From the Rheumatic Disease Unit, Wellesley Hospital, 160 Wellesley Street E, Toronto, Ontario, CanadaM4Y IJ3.
SUMMARY Efficacy and safety of azathioprine in'high' and 'low' dose regimens in rheumatoid arthritis(RA), both in short-term studies and in follow-upover 40 months, have previously been shown. In thepresent report, 36 patients with RA treated withazathioprine (group I) and 49 age-matched patientswith RA (group II), were studied to detect potentialearly markers of malignancy.
Chest x-rays were similar in both groups. Onepatient in group I had a positive PAP smear and wassubsequently found to have uterine carcinoma.Alpha-fetoprotein was positive in one patient ingroup I and none in group II. CEA was negative in allpatients in group I, but positive in seven in group II.On chromosomal analysis group I showed a greaterfrequency of breakage. Group I showed lower serumfolates and a highly significant number of megalo-blastic features in marrow aspirates.
In group I seven tumours, three being malignant,occurred while taking azathioprine, and in group IIsix tumours, one malignant, were identified(p=017).The apparent increased risk of malignancy previ-
ously suggested by others warrants further studieswith larger populations and over a continuous longerperiod.
* * * * *
Azathioprine has been shown to be an effectiveagent in the treatment of active rheumatoid arthritis(RA).38 124 143 224 The long-term side-effects of thispurine analogue, especially the potential for theinduction of neoplastic change, have been describedonly sporadically in patients with RA. In 1976, twopatients with seropositive RA, treated with bothazathioprine and alkylating agents and whodeveloped leukaemia, were reported from thisunit.196 This prompted us to examine a group ofpatients with RA on azathioprine, with respect to theincidence of neoplasms, to try to detect any specific
'This work was supported by the Canadian Arthritis Society, 920 YongeStreet, Ste 420, Toronto, Ontario, Canada. Grant Number: 7-195-74.
clinical or laboratory markers in patients with neo-plastic change.
Method
Thirty-six patients with classical or definite RA byAmerican Rheumatism Association (ARA)criteria'86 receiving azathioprine were chosen to takepart in this in-patient study.
Forty-nine age and sex matched patients, withclassic or definite RA, who volunteered for a 'cancercheck-up', were chosen from our in-patient and out-patient population as controls. These patients wereon standard anti-inflammatory and disease suppres-sant (gold, chloroquine, penicillamine) medications;none had ever taken azathioprine or any cytotoxicagents. A medical history was obtained from eachpatient and a complete physical examination was per-formed. Total active joint count, grip strength"' andfunctional class were recorded.Laboratory tests performed included: haemoglo-
bin, white blood cell (WBC) count and differential,platelet count, erythrocyte sedimentation rate (ESR)by the Westergren method, latex fixation test,"0'SGOT, alkaline phosphatase, protein elec-trophoresis, serum iron and total iron bindingcapacity. Serum folate determinations before July1977 were performed using a microbiologicalassay.36204 After July 1977, a radiodilution methodwas adopted (Amersham Folate Radioassay Kit).Serum B,, was performed by a radiodilutionmethod.'83 Alpha-fetoprotein was measured byimmunodiffusion in agar gel.'64 Carcino-embryonicantigen (CEA) was detected by radioimmunoassay(Hoffman-LaRoche Kit). Urine and sputum cyt-ology, when obtainable, was studied on each patient.
Radiological examinations included chest x-rayson all patients, mammography on each female andbarium contrast studies, when indicated. All patientswere examined by a dermatologist with attention toneoplastic changes of the skin. All women were
Long-term effects of azathioprine in rheumatoid arthritis Suppl p 19
examined by a gynaecologist and cervical and vaginalsmears were taken.Bone marrow aspiration was performed on all
patients receiving azathioprine and the samplesexamined by one of us (CSN). Marrow chromosomalanalyses'53 195 215 were performed on 25 patients ofthe 36 in the azathioprine group and in 11 of the 49controls.
Statistical analyses of the results were carried outusing the Fisher's exact test and the x2 test. Signifi-cance testing was carried out at the 5 % level.
Results
Group I (azathioprine treated) consisted of 36patients, eight males and 28 females. Group II (con-trol) consisted of 49 patients, 13 male and 36 female.There was no significant difference between thegroups with respect to sex. The mean age was signifi-cantly higher in group I than in group II (57-8 years vs514 years). Group I had a slightly longer diseaseduration (15 3 years vs 11 9 years) (table 1).Group I had taken azathioprine for a mean dura-
tion of 38 5 months. The mean dose of drug for thegroup at assessment was 1-08 mg/kg/day. This wasless than the mean dose at initiation of azathioprinetreatment (133 mg/kg/day). When the groups werecompared with respect to concomitant medicationswithin the three months before assessment, more group Ipatients had taken chloroquine and for a longer dura-tion than patients in group II. There was a higherfrequency of gold and penicillamine treatment in thecontrol group at the time of assessment. Elevenpatients in group I were taking low-dose prednisone,while only two patients in group II were on pred-nisone. Four patients in the azathioprine treatedgroup had had intra-articular injections of their kneeswith radioactive gold; three patients had had bothknees injected since the diagnosis of RA. One patientin group II had had a knee injected with radioactiveyttrium and one had had a knee injected with radio-active gold. All patients had had intra-articularsteroids at some time in their course. None was takingphenylbutazone.
TABLE I Study population
Group I; Group II; Significanceazathioprine controltreated
No of patients 36 49Male/female 8/28 13/36Age (yr) 578 514 p=005
Disease duration (yr) 15 3 11 9 p=0-22Duration of
azathioprine (mth) 38-5Dose of azathioprine 1 08 mg/kg/day
Age, duration and dose are given as means.
TABLE 2 Clinical featuresAzathioprine Control Significancetreated (36) (49)
Clinically, both groups were comparable as to gripstrength, extra-articular manifestations and latextitre (table 2). They differed significantly in the jointcount, with group II having a greater number ofactive joints (16 2 vs 11 3), and ESR, with group IIhaving a lower ESR (51-3 vs 23 3). There were nomarked differences in fertility, hormone treatmentand history of cigarette smoking between the groups.Both groups were comparable in terms of SGOT,
alkaline phosphatase and protein electrophoresisresults. The groups did not differ with respect to ironstudies, haemoglobin and platelet, neutrophil andlymphocyte counts. The total white blood cell countwas significantly lower in the azathioprine treatedgroup (table 3).
Chest x-rays were similar in both groups. Inflam-matory and fibrotic abnormalities were seen in 15patients in group I and nine patients in group II.There were no suspicious lesions seen in either group.Mammography showed a comparable number ofbenign lesions, that is, fibrosis, generalised calcifica-tions and/or dysplasia. No masses or signs of malig-nancy were seen except for one suspicious lesion in awoman in group I and three in group II. Thesewomen are being followed carefully and, thus far,show no evidence of malignancy.
Cervical smears (Papanicolaou) on all womenwere negative except for one patient in group I whosubsequently was found to have adenocarcinoma ofthe uterus on curettage. Urine cytology was abnor-mal in four patients in group I and one patient ingroup lI, showing dyskaryotic changes. Thosepatients demonstrating abnormal urine cytologyhave been followed and none has shown develop-ment of urinary tract neoplasms.
Alpha-fetoprotein testing was positive in onepatient in group I, with no evidence of associated
malignancy. All patients in group II were negative.CEA levels were negative in 15 patients tested ingroup I. In group II, 35 patients were tested andseven were positive. These seven people have beenfollowed and show no evidence of malignancy. Fam-ily histories revealed a higher incidence of patients ingroup I who had first degree family members withdiagnosed malignancies (15 vs six).The occurrence of tumours in both groups was
examined. In group I, seven patients had eight neo-plasms, one of which, a thyroid adenoma, occurredbefore azathioprine was started. Hence, sevenpatients developed seven neoplasms after commenc-ing azathioprine. Of these seven tumours three weremalignant (table 4). For the seven patients the meanexposure to azathioprine was 1-6 years before diag-nosis of their neoplasms. Four patients had a positivefamily history of malignancy.
In group II, five patients had six neoplasms in all,one of which was malignant. The tumours were:thyroid adenoma, lipoma, neurofibroma, two fib-roids and a diffuse histiocytic lymphoma. None ofthese five patients had a positive family history ofmalignancy.
In terms of malignant and potentially malignantlesions, group I had three such tumours (twoadenocarcinomas of the uterus, one carcinoma-in-situ of the cervix) and group II had one diffuse his-tiocytic lymphoma. This difference did not reachstatistical significance (p=0O1728). The clinical andlaboratory results of those patients with malignanciesin group I did not differ significantly from group I as awhole.
Examination of marrow aspirates revealed a highlysignificant number of megaloblastic features in 28 of36 patients studied in group l, whereas only one outof 19 patients studied in group II showed such achange (table 5). Serum B,2 levels were similar inboth groups, but serum folate levels, by both micro-
biological and radiodilution assay, were significantlylower in group I.Analysis of marrow chromosomes performed in 25
of the 36 patients in group I and 11 of the 49 in groupII showed striking differences in aberration index,percentage of cells with aberrations and aberrationTABLE 5 Laboratory features: erythropoietic factors
Azathioprine Control Significancetreated (36) (49)
o Patients with neoplasmsFIG 1 Marrow chromosomal analysis on 25 patients ingroup I and 11 in group llshowing similar percentage ofcellswith polyploidy but a higher aberration incidence in group I.
TABLE 4 Neoplasms in azathioprine treated patients
Patient Sex Age Azathioprine Duration of Other Tumour Family(yr) dose azathioprine medications history
(mg/kg/day) (mth) at diagnosisofcancer
JA F 44 1-8 34 Prednisone Ca-in-situ +Chloroquine
PMcM* F 54 0-85 34 Prednisone Adeno Cauterus
DP F 47 1.0 12 - Adeno Ca +uterus
JD F 50 1-38 20 - Haemangioma +DD F 52 1-16 11 - Benign liver
cystMT F 70 2 02 12 - Anal polypJH F 47 0 80 12 - Fibroadenoma +
FIG 2 Marrow chromosomal anal vsis ongroup I and 11 in group II shouwing a greg
cells with aberrations and a higher aberragroup 1.
incidence. The azathioprine treatedmore abnormal patterns (figs 1, 2) inIchromosomal variables measured. (Pour laboratory for the aberration inciless.)Four of the patients with neopla:
showed a greater number of chrornmalities than the other 21 studied inof these patients had benign and twoplasms. However, the two patients widiffered from the remainder of groupthe variables, the aberration index.
Discussion
The use of azathioprine in severely ac
tory to standard disease suppressz
shown to be effective in several studiepotential neoplastic effects have rem
concern. There have been several (
malignancy in patients with RA o
alone, or in combination withics,2 32 79 125 176 177 196 with a striking fremyelogenous leukaemia, occurring pI
males.79 196
Azathioprine was first shown to bean increased incidence of malignancikcombination with prednisone in theorgan transplants. 175 176 Although th(dence of this complication may be hi)the possible overlapping incidence
Long-term effects of azathioprine in rheumatoid arthritis Suppi p 21
Aberration index during chronic renal failure,'98 there is no doubt thatorgan transplant patients maintained on chronicimmunosuppressive treatment have a significantlyincreased risk of developing de novo cancers within
o the first few years of transplantation. Almost all such. patients received azathioprine and prednisone, as
o.eeo0 * well as other immunosuppressive medications in sev-
000;" eral cases. This phenomenon has been variously-000" explained by: the chronic presence of, and stimula-
tion by, foreign transplanted antigen; immunosup-pression sufficient to allow survival of spontaneoushost tumours; and increased susceptibility to viraloncogenesis; or direct oncogenic effect by theimmunosuppressive agent.
In more recent years, azathioprine has been used inGroupl Group II many non-malignant conditions characterised as(n=25) (ro11) being of 'immunological' origin. Although Sharon et
p <0 05 al's series of SLE patients'98 and Worthington'spasms observations (cited in Skinner and Schwartz2"2) of25 patients in
only one malignancy per 3000 such patients have25 patients in been used to deny an increased incidence of neo-ater percentage Of plasms in patients treated with immunosuppressivetion inidex iti agents, over 30 cases of malignancies have been
reported in this group treated by azathioprine alone,or in combination with other cytotoxic agents. Many
group I showed of these cases admittedly have a tenuous relationshipthree of the four to the drug in question.' 29 32 61 74 90 118 125 127 1:34 141 146 157Normal value in 171 177 181 187 192 196 199 200 20:3 218 221 2:1S 240
dence is 10% or We have compared 36 patients with RA takinglong-term azathioprine therapy (group I) with 49 age
sims ln groupr and sex matched patients not taking this medicationtosomal abnor- (group II). The groups were similar in clinical andthe group. Two laboratory features except for a higher age and lowermalignant neo- WBC counts in the azathioprine group, as previouslIth malignancies reported.95 225I in only one of
The outstanding finding was the demonstration ofseven tumours in group I while they were takingazathioprine, three of which were malignant or pre-malignant. This was compared to five benign and onemalignant tumour in group II. Of the 36
tive RA, refrac- azathioprine treated patients in this study, onlyants, has been women developed neoplasms. The three malignant,38 124 14:3 224's,3 244;22but lesions were located in the genitourinary tract, aained a serious common site for neoplastic change in the elderlycase reports of female, and the benign lesions were found in the skin,n azathioprine liver, breast, bowel and uterus. There was oneother cytotox- malignant neoplasm identified in control group 11, aquency of acute diffuse histiocytic lymphoma. Previously reportedredominantly in prevalence rates of malignancy in RA have been
either comparable to non-rheumatoid groups,1"' orassociated with unexpectedly low (0-6 ( ).'26 The finding of threees when used in malignancies among 36 azathioprine treated patientsmanagement of as compared with one among the 49 controls is note reported inci- statistically significant (p=0-1728). Furthermore,gher because of statistics from the Ontario Cancer Research Founda-of malignancy tion suggest that the expected incidence of malignan-
cies over a three-year period in age matched femalesshould be less than one case per 28 women (0-75cases/28 women) (Ontario Cancer Treatment andResearch Foundation).'62 This is not significantly lessthan our finding of three cases per 28 women.
Significantly the diagnosis of neoplastic change ingroup I occurred early in the course of their treat-ment, within a mean of 1*6 years from the initiation oftherapy, as compared to a mean duration of 3 5 yearsof azathioprine treatment in group I as a whole. It isof interest that an additional male patient in group Ideveloped a malignant lymphoma while still takingazathioprine, one year after his assessment in thisstudy, and two patients, already reported from thisunit previously, developed acute leukaemia. No con-trols for these patients were available. On the otherhand, the four patients with benign lesions continuedto take azathioprine for a further mean period of 20months and have suffered no malignant transforma-tion nor further neoplasms. Of the three patients withmalignancies, two have elected to continue azathio-prine after treatment of their neoplasms and havesuffered no recurrence nor further primary malig-nancies. This pattern would imply that a premalig-.nant lesion may be present when azathioprine is initi-ated, and with immunosuppression the malignancymay manifest itself relatively early during the courseof treatment. Carcinomas of the cervix are a commonde novo cancer in renal transplant patients occurringin the first years after transplantation.'76 Cervicaldysplasia occurring as soon as 12 days after initiationof azathioprine therapy has been reported.85Marrow chromosomal analysis has been shown to
be a more effective means of determining cytogeniceffects of agents than leucocyte chromosomalanalysis.'04 144 We performed marrow chromosomalanalysis and found an increased number of abnor-malities in the azathioprine treated group despite thefact that the mean dose of azathioprine used was 1 08mg/kg/day, a low dose regimen. Nevertheless,chromosomal changes were significantly abnormal,but the number of abnormalities could be correlatedneither to the dose nor to the duration of azathiop-rine treatment, a finding previously reported.'05 Ourresults confirm previous reports of the clastogenic
effect of azathioprine on chromosomes in non-malignant disease. These changes have beenreported to return to normal on discontinuation ofthe drug.95 105 However, there was a marked degreeof overlap between results in group I and group II. Aswell, this overlap was apparent when comparingresults of those patients in group I with tumours andthe remainder of group I. Although those withtumours had greater chromosomal abnormalities as agroup, other subjects in group I, as well as in group II,had equally abnormal analyses when looked at indi-vidually (figs 1, 2). Thus, the chromosomal analysisalone could not identify those individual patientswith neoplasms. The development of tumours mayreflect an increased susceptibility to the oncogenicpotential of azathioprine in certain patients, a suscep-tibility possibly marked by a positive family historyfor malignancy (table 4).
Originally claimed to be a marker of bowel car-cinoma, the CEA has now proven to be much lessspecific. None of the patients in the azathioprine-treated group had increased levels of CEA. Ofinterest, is the fact that seven of the control group hadsignificant levels without evidence of any neoplasms.This finding of raised CEA levels in seropositive RAhas been previously reported.222 223
Alpha-fetoprotein is another oncofetal antigenpreviously reported to be a potential marker forhepatic carcinoma.4 We detected this antigen in onlyone patient in group I, who proved to have no evi-dence of an associated malignancy. All patients ingroup II were negative for alpha-fetoprotein.The problem of therapy with azathioprine in RA
and the increasing incidence of neoplastic diseaseresulting from such treatment is a serious considera-tion, but to date no registry has been organised togather and analyse such cases in a similar manner tothat performed for organ transplant malignancies.Without such an organisation, meaningful statisticsand conclusions will inevitably be delayed. We con-clude that the apparent increased risk, as demon-strated in this study, warrants further studies withlarger populations and over a continuous longerperiod.
Interim observations on benefit/risk of azathioprineversus D-penicillamine in the treatment ofrheumatoid arthritisJOHN R WARD* AND H JAMES WILLIAMS*
From the School of Medicine, University of Utah, Salt Lake City, Utah, USA
Azathioprine has been shown to be an effective drugfor the treatment of rheumatoid arthritis. Levy et al124have reported improvement in joint counts withazathioprine and Urowitz et al1224 found azathioprineto be superior to placebo in reducing the articularindex and joint counts. Under double-blind condi-tions, azathioprine, cyclophosphamide and gold pro-duced comparable clinical improvement over 18months of observation.3" In a single-blind externalobserver trial, azathioprine 2 5 mg/kg/day producedsimilar efficacy and toxicity when compared withD-penicillamine 1 g/day." In a parallel uncontrolledopen trial, azathioprine, gold and chloroquine were
essentially equivalent in the control of disease activ-ity in patients with early rheumatoid arthritis.5' Addi-tional studies support the beneficial effects ofazathioprine in rheumatoid arthritis.25 143 223
The selection of slow-acting antirheumatic drugs(SAARDs) for treatment still remains controversial.Our group considers gold to be the SAARD of firstchoice. However, for patients who fail to respond togold, or develop toxicity, cytotoxic drugs or
D-penicillamine represent the next therapeuticoption. Thus, the current study was developed tocompare the efficacy and toxicity of azathioprine andD-penicillamine in a prospective randomised control-led double-blind 24-week trial. The study will evalu-ate long-term efficacy and safety during a five-yearnon-blinded follow-up period, during which patientsmay continue to take one of the study drugs as long as
it is clinically indicated.
Materials and methods
PATI ENTSA total of 200 patients will be entered. Patient selec-tion criteria include the following.
(1) Definite or classical rheumatoid arthritis(ARA criteria) of greater than six months'duration.
*For the Cooperating Clinics, Cooperative Systematic Studies of theRheumatic Diseases. Supported by NIAMDD Contract NOI-AM-6-2218and a grant from Burroughs Wellcome Company.
(2) Onset after the age of 16 years.(3) Active disease defined as having six or more
swollen joints that are considered responsiveto therapy and at least two of the following:(a) Nine or more responsive joints tender
on pressure;(b) 45 minutes or more of morning stiffness;(c) Westergren erythrocyte sedimentation
rate of 28 mm/h or greater.(4) Failure to respond to at least 750 mg paren-
teral gold or toxic reaction to gold.(5) Inadequate control of arthritis with thera-
peutic doses of aspirin or other non-steroidalanti-inflammatory drugs (NSAIDs).
(6) Absence of childbearing potential in females.(7) Absence of contraindication: hepatic disease,
(8) No previous treatment with cytotoxic drugs.(9) Constant optimal dose of aspirin, NSAID,
and/or prednisone (10 mg or less/day).(10) Informed consent.
TREATMENTA randomised schedule assigned 12 to 24 patients toeach of 11 participating clinics, with equal numbersto receive azathioprine in doses of 1-25 to 1-5mg/kg/day or D-penicillamine in doses of 10 to 12mg/kg/day with a five-year follow-up. Each patientreceives azathioprine tablets (25 mg) and placebocapsules, or D-penicillamine capsules (125 mg) andplacebo tablets. The azathioprine is initiated to pro-vide the full dose from start of trial, whereasD-penicillamine is increased by 250 mg every fourweeks until the full dose is reached and maintainedfor the 24 week duration of the double-blind phase.
MEASUREMENT OF DISEASE ACTIVITYFor measurement of efficacy the following variableswere measured at entry and every six weeks.
(1) Functional assessment.(2) Duration of morning stiffness.
(3) Onset of fatigue.(4) Overall assessment of disease activity as
estimated by both physician and patient.(5) Joint counts for number of tender joints and
tenderness score for joints (O=none, 1 = mild,2=moderate, 3= severe).
(6) Joint counts for number of swollen joints anda swelling score.
Additional criteria for response include*:(1) Remission.
(2) Significant clinical improvement.(3) Overall meaningful clinical improvement.(4) Subjective improvement.(5) No response.(6) Progressive disease.Extra-articular manifestations, including sub-
cutaneous nodules, skin ulcers, vasculitis, etc, will beevaluated.
OTHER ASSESSMENTS
Patients are seen every two weeks for evaluation ofside-effects, monitoring of drug compliance and labora-tory studies, which include complete blood count,
'Definitions will be provided in a report when the trial has been completed.
TABLE 1 Comparison ofdescriptive variable at entry forpatients completing the trial by treatment group
Variable Measure Treatment group
A (n=38) B (n=31)
Age (years) Mean 50 53Sex Female (%) 71 74Race Caucasian (%) 84 90
Black (%) 16 10Duration (months) Mean 135 124Functional class Mean 2-2 2-1Severity Mean 2-2 2-1
TABLE 2 Median differences for measurement variables bytreatment group between the conclusion and initiation ofthetrial*
Variable Treatment group
A B
Functional class (scale 1=no disability,4=incapacitated) 0-179 0 000
*Median change from entry to completion. A positive value indicatesimprovement.
platelet count and urine analysis. A chemical survey
and Westergren sedimentation rate is done every sixweeks. Rheumatoid factor and antinuclear antibodytests are done at entry and completion.
Results
PATIENTS STUDIED
To date 148 of the 200 expected patients have beenenrolled and 69 have completed the double-blindphase. Because the study is incomplete and still inprogress, only blind comparisons of the two treat-ments are presented. The patients in each studygroup were comparable (table 1).
RESPONSE TO THERAPY
The results for selected measurements of diseaseactivity are shown in table 2. There was improvementin duration of morning stiffness, overall assessmentof disease activity as judged by both physician andpatient, number of tender joints and joint tendernessscore and number of swollen joints and joint swellingscore. The differences were similar but seemed tofavour treatment B. No formal statistical testing hasbeen done because of the ongoing nature of the trial.
WITHDRAWALSOf the 148 patients entered, 16 patients withdrew forthe reasons shown in table 3. More detailed reasonsleading to withdrawal are shown in table 4. Toxicity
TABLE 3 Reason for withdrawal ofpatients who haveentered the trial to dateTreatment No ofpatients No ofpatients withdrawn
enteredADR NR UD
A 77 6 0 2B 71 5 1 2
ADR=adverse drug reactionNR=non-responsivenessUD=unrelated to drug (moved, lack of co-operation, intercurrent illness, etc)
TABLE 4 Reason for withdrawal from study by treatmentgroup
Reason for withdrawal No ofpatients
Treatment A Treatment B
Thrombocytopenia 1Proteinuria 1Leucopenia 2Nausea 2 1Abnormal liver function tests 1Itching and rash 1Oral ulcers 1Herpes zoster 1Inadequate response 1Unrelated disease 1Lack of co-operation 2 1
Interim observations on benefit/risk ofazathioprine versus D-penicillamine treatment Suppl p 25
as determined by withdrawals was similar althoughfrom different causes for the two groups.
Discussion
Preliminary results of this study to compare theefficacy, safety and tolerance of selected doses ofazathioprine and D-penicillamine have indicatedimprovement in rheumatoid arthritis with bothtreatments. However, the data to date would suggestbetter efficacy for treatment B. Until completion ofthe trial, blinding will be maintained. The nullhypothesis states that one group is equal to the othergroup. Failure to reject the hypothesis when a realdifference exists is termed a type II error, while rejec-tion of the hypothesis when no real difference ispresent is a type I error. No formal statistical com-parisons are presented because of the small samplesize and thus the risk of committing a type II error.The trial was designed to detect a 30% differencebetween the treatment groups with a=0 05 and,/=0O1, and allows for a dropout rate of 30%. Until
the 200 patients have completed the trial, only astriking difference in efficacy or toxicity will termi-nate the trial!An important component of this study is to
examine for a subset of unique responders. Thus, apriori definitions of remission and varying degrees ofclinical response and worsening were developed.Analysis of these subsets will also await completionof the formal trial.The long-term follow-up of patients, who continue
on their treatment for extended periods of time, isrequired to help the physician in the selection oftreatment. Short-term efficacy is not always equat-able to value of therapy, as tolerance and continuedbenefit are important.
It would have been far more useful to have beenable to present not only the results of the double-blindcomparison, but the long-term follow-up. While thisis not possible, the ultimate results of this trial shouldhelp clinicians select which drug they judge would behelpful in the treatment of 'unresponsive'rheumatoid arthritis.
An overview of benefit/risk of disease modifyingtreatment of rheumatoid arthritis as of today*HAROLD E PAULUS
From the Department of Medicine, UCLA School of Medicine, Los Angeles, California 90024, USA
SUMMARY In chronic rheumatoid arthritis (RA),disease modifying drugs are used in an attempt tosuppress the progressive damage to tissues and jointsthat is associated with active disease. Their success inachieving this goal is variable; responses vary fromcomplete suppression of all signs and symptoms ofRA to continued active disease, with progressive dis-ability, despite prolonged therapy. Because diseaseactivity almost always recurs after the therapy isstopped, early interruption of an effective therapy forany reason will make its benefit insignificant in alifelong disease such as RA. Similarly, short-termsequential use of multiple disease modifyingtherapies is unlikely to be beneficial. The immediateproblems with these therapies are substantial. Ingeneral, fewer than 50 'Y% of patients are able to con-tinue a particular drug for more than one year. Sinceit takes three to 12 months or longer to achievemaximum effects, those patients who are unable tocontinue the drug receive little benefit from it.Inevitable delayed side-effects, such as those associ-ated with chronic corticosteroid therapy, may makethe benefit/risk ratio unacceptable. Potential latelethal adverse effects, such as malignancy, weight thebenefit/risk ratio to varying extents for individualpatients, depending on the relative probability thatthe adverse effect will occur during the remainder ofthe patient's anticipated life span, and are of greaterimportance in younger patients. In that minority ofpatients who achieve remission or near remission andare able to tolerate a disease modifying treatment formany years, it is of truly significant benefit. We arestill searching for a therapy that will reliably achievethis goal for most patients with RA.
* * * * *
The purpose of disease modifying antirheumaticdrug (DMARD) therapy of rheumatoid arthritis(RA) is to:
(1) reduce, suppress, or eliminate disease activitynow; and
(2) thereby prevent the extended and perhapspermanent effects of disease activity.42*Supported in part by USPHS Grant NO1-AM6-2218.
Currently available DMARDs include organicgold compounds, antimalarial drugs (chloroquineand hydroxychloroquine), D-penicillamine andthe immunosuppressants (azathioprine,6-mercaptopurine, cyclophosphamide, chloram-bucil and methotrexate). Investigational diseasemodifying therapies include the T cell stimulantslevamisole and thymopoietin, lymphocyte deple-tion or suppression by thoracic duct drainage,lymphopheresis, antithymocyte globulin, or totallymphoid irradiation and removal of plasmaconstituents by plasmapheresis.These interventions are remarkably diverse in
their chemical and pharmacological properties andeffects. Yet, they appear to be remarkably similar intheir effects in patients with RA. Their administra-tion does not result in any observable immediateamelioration of the signs or symptoms of RA, such asone expects to see with corticosteroids or non-steroidal anti-inflammatory drugs. However, afterweeks or months of treatment, a subtle and some-what questionable decrease in the severity of symp-toms may occur. With continued therapy, the averagepatient demonstrates overall moderate suppressionof disease activity, but this average includes somepatients who have no improvement at all, and otherswho eventually achieve complete, or nearly com-plete, remission of disease manifestations, oftenaccompanied by normalisation of laboratory abnor-malities that are associated with disease activity, andstabilisation ofx-ray evidence of joint damage. How-ever, these beneficial effects are not permanent.When the disease modifying treatment is stoppedbenefit usually persists for weeks or months, but thendisease manifestations gradually recur in the samedelayed and subtle fashion as they had disappearedwhen the therapy was started.How well do we succeed in our goal of preventing
the progressive damage to tissues and joints that isassociated with disease activity using the DMARDsthat are available today? What price do our patientspay for these benefits?
Chronic RA is usually a life-long disease. If weexpect to restore the patient to normal, productive
An overview of benefitlrisk ofdisease modifying treatment ofRA as of today Suppl p 27
life with a DMARD induced remission, that remis-sion must be sustained for many years. Since itgenerally takes three to 12 months or longer toachieve maximum benefit with DMARD therapy,patients who are unable to take the drug for morethan one year receive little benefit from it.
Benefits
In evaluating the long term benefits of DMARDs onthe life quality of patients with RA, the classic pub-lished placebo-controlled double-blind studies thathave established the use of these drugs are of littlevalue because, in order to avoid unnecessarily pro-longed administration of placebo, they are generallyof the shortest possible duration, usually no morethan four to 12 months.34 63 71 140 155 224 However,there are some published reports on the long-termusefulness of some of the DMARDs.
GOLD COMPOUNDSSrinivasan et alt0' reported our findings in a cohortstudy of 111 patients who were followed for an aver-age of three years after starting gold therapy. Fifty-eight (52%) patients stopped gold within the first 18months. Of those who continued gold for more than18 months, 83 % achieved remission, although five ofthese subsequently flared while continuing goldtherapy. Thus, of the 111 patients starting gold,about 35 % had major sustained improvement anaverage of three years later.Rothermich and associates'89 started 97 patients in
a prospective study of gold therapy. After an aver-age of 12 years, 41 % were having sufficient clinicalbenefit to continue the drug. However, four to sixyears after the study was started, only 14 patientswere continuing gold therapy, and two more haddiscontinued because of remission. Most stoppedgold because of loss of benefit, or toxicity, or bothsimultaneously.
In the same report, Rothermich et al'89 examined171 patients who started D-penicillamine after fail-ing gold therapy and 53 % had stopped gold withinthe first year, more for toxicity than for lack ofbenefit. At the end of three years, only 20% continuedgold, somewhat fewer than Srinivasan's finding of35 %.205 At the end of five years, only 9% were stilltaking gold. Of course, these figures are biased by therequirement that only patients who had stopped goldwere counted; the analysis is completely retrospec-tive and the population may not be representative.Nevertheless, the majority of these patients stoppedgold for loss of benefit.
D-PENICILLAMINESeveral similar studies of D-penicillamine therapy
have been reported recently. Kean et al"2 found that47 % of 101 patients continued D-penicillamine afterone year and 38 % after two years. The major reasonfor stopping therapy in the first year was toxicity.After two years, two patients were in remission, fivein partial remission, and 31 had 75 % improvement.Webley and Coomes"3 reported that 46 % of 1 14
patients continued penicillamine after an average of101 months' treatment. Of the 62 withdrawals, 40were for toxicity and 12 for lack of response. Therewere no differences in toxicity or improvement ratesin patients receiving more or less than 600 mg daily,or in patients who had or had not had previous goldtherapy.
COMPARATIVE STUDIESHusain and Runge96 evaluated the risks of termina-tion of treatment with gold, penicillamine, hydroxy-chloroquine and levamisole using a life tableanalysis. The risks for termination of gold and penicil-lamine were similar to those reported above, whilethat for hydroxychloroquine (25 % in 12 months) waslower, and that for levamisole (57% in 12 months)higher than for the other two drugs. However, inanother paper, the same group reported that themedian termination time was 60 months for gold and13 months for antimalarial compounds. Lack of effi-cacy was responsible for treatment termination in41 % of the antimalarial group in this study in whichtreatment was frequently terminated for lack ofresponse after three to six months of treatment.'83Dwosh et al"6 randomly assigned patients with class
II RA to azathioprine, gold or chloroquine. After sixmonths of treatment, all three agents were compar-able in terms of toxicity and clinical improvement.However, the chloroquine group did not maintaincontinued improvement beyond six months, while inthe gold and azathioprine groups benefit continuedfor up to 44 months.Hoh et al93 reported the incidence of remissions of
RA in patients treated with gold, penicillamine, orchloroquine for at least one year. Prolonged remissionsoccurred in only 10% of the 230 drug courses, while59% had modest partial responses and the remainderhad no response.
Currey et alt8 compared azathioprine, cyclophos-phamide and gold in a well-designed 18-monthdouble-blind randomised trial of 121 patients. Theyconcluded that the two immunosuppressives producedclinical improvement comparable to that with gold.Cyclophosphamide was perhaps marginally moreeffective. Drug management was easiest withazathioprine. However, only 25 % of patients onazathioprine, 36'Yo on cyclophosphamide and 24%Y,on gold completed 18 months of treatment. Ofcourse, withdrawals for toxicity were increased by the
double-blind nature of the trial and the inclusion ofgold, which necessitated withdrawing sevenazathioprine and 11 cyclophosphamide patients formarrow depression.On the other hand, Cade et al25 reported that 14 of 16
patients treated with azathioprine for two to five yearsachieved 100% work ability, although it took up to32 months of treatment to achieve that maximumresponse. Furthermore' serial x-rays showed no pro-gression of destruction after completion of the sec-ond year of therapy, even when followed for as longas eight years.
Risks
Only major risks will be evaluated here. It is obviousthat patients who are forced to stop treatment becauseof temporary reversible rash, proteinuria, marrowsuppression, drug induced hepatitis, etc, do notbenefit from the drug, but these adverse reactionsrarely shorten their life span or produce permanentdamage.
DEATHDavis4" has recently reviewed the undesirable effectsof gold therapy. With current dosages, marrowaplasia is the most serious adverse effect. Over aseven-year period in Great Britain, 16 deaths related togold therapy were founid, an estimated incidence of1 6 deaths per 1(0 000 prescriptions. This figure sug-gests to Davis that gold is ten times more toxic thanany other therapy used in Great Britain. Davisreviewed four studies, published in 1962, 1968,1976, and 1977, reporting a total of 50 patients withgold induced marrow aplasia, which was fatal in 66 %.
Similarly, Kay... collected information on tenpatients in Great Britain with pancytopenia associ-ated with penicillamine therapy, six of whom died.The dose range was 250 to 1000 mg/day with anaverage of 615 mg. The duration of therapy was fromthree to 60 months with a mean of 16 months. Shereports that the Committee of Safety of Medicineshad had reported to them 18 deaths fromD-penicillamine from 1964 to 1977 (three of whomwere included in the six deaths reported by Kay).Fourteen of the 18 deaths were due to blood dys-crasias.
Similarly, deaths have occurred in patients whohave developed agranulocytosis while takinglevamisole. '47Lewis et al'25 reported on the causes of death in 31 1
patients with RA who were observed over an 11-yearperiod. A total of 46 patients died, 13 of them frommalignancies. The incidence of deaths from neoplasiawas not greater in 214 patients treated with azathio-prine and/or chlorambucil or cyclophosphamide
(2 8%) than in 97 patients who did not receive acytotoxic drug (7 %).
Isomaki et al"'2 10 reported two large studies ofcauses of death in patients with RA in Finland. In thefirst, an age and sex matched comparison of 1000patients with RA and 1000 control subjects, malig-nant neoplasms were less frequent in RA (9% ofdeaths; 11 % of the population) than in the controls(30% of deaths; 2 1% of the population). In thesecond study of more than 45 000 patients with RA,the risk of death from lymphoma, leukaemia ormyeloma was twice as great in RA (130 deaths) asthat expected in the general population (59 deaths).No data are given regarding the drugs used by thesepatients.
OTHER SERIOUS TOXICITIESMaculopathy occurred in 10% of 222 patientstreated with 800 mg/day hydroxychloroquine or 500mg/day chloroquine by Dubois for systemic lupuserythematosus, but after 10 years of follow-up, visualacuity was adequate in all but two of the patients.'However, with 400 mg daily for more than one yearin 99 patients, no loss of vision occurred."'
Sterility occurs with cyclophosphamide andchlorambucil, haemorrhagic cystitis with cyclophos-phamide and hepatic fibrosis with methotrexate. Allof the immunosuppressive drugs are thought toincrease susceptibility to infection, particularly whencombined with large doses of corticosteroids.
Induction of autoimmune diseases has beenreported with penicillamine therapy for RA, and alsofor Wilson's disease. In addition, the renal lesionsassociated with gold induced proteinuria have thecharacteristics of an immune complex mediatednephropathy, and the agranulocytosis of levamisoleappears to be immune mediated. We have also seennew autoimmune manifestations develop during theapparently successful treatment of RA with thoracicduct lymphocyte depletion. These observations sug-gest that, while correcting some disease relatedabnormalities of the immune system, we may perturbthe system in such a way that other manifestations ofautoimmune disease emerge in some patients.
Conclusions
For those patients who achieve remission with aDMARD and are able to tolerate it for many years,the benefit clearly is greater than the risk. Unfortu-nately, because most patients do not achieve optimalbenefits, or must stop treatment for annoying side-effects, our goal of favourably altering the course oftheir RA is not achieved. Except for the anti-malarials, there appears to be a real, although small,risk of drug associated death with all of the available
An overview of benefit/risk ofdisease modifying treatment ofRA as oftoday Suppl p 29
DMARDs. However, the available data are not suffi- that participate in immune responses or their pro-cient to produce a satisfactory rank order of this risk ducts. With increasing knowledge about this systemwith the various drugs. and its methods of communication, and increasingWe have no reason to be satisfied with our present attention to the antigens associated with autoimmune
accomplishments with DMARD therapy. However, diseases, perhaps future DMARDs will be moreit seems clear that DMARDs affect either the cells effective.
EULAR register of patients on immunosuppressivedrugsA KAY*
From the ARC Epidemiology Research Unit (Clinical Section), Department ofRheumatology, Guy'sHospital, London SE]
A collaborative United Kingdom/Australasian studyof cancer in 5172 patients treated with immunosup-pressive drugs115 showed an excess of non-Hodgkin'slymphoma, squamous cell skin cancer, and possiblymesenchymal tumours, in a series of patients with andwithout transplants. The 1349 non-transplantpatients in the series, which included 492 withrheumatoid arthritis or ankylosing spondylitis,showed less marked excesses of the same tumours. Inother studies, leukaemias have been reported innon-transplant patients exposed to cyclophos-phamide.'07The purpose of the European League Against
Rheumatism (EULAR) Study is to register andfollow-up a substantial number of patients with con-
nective tissue diseases who are, or have been, treatedwith immunosuppressive drugs, in order to deter-mine the subsequent malignancy rate.
Seventy-seven rheumatologistst in 13 membercountries registered 716 patients treated withimmunosuppressive drugs between January 1979and December 1980: registration of patients con-
tinues. The 478 patients who started immunosup-pressive treatment on or after 1 January 1979 form a
complete prospective sample. The 238 patientsalready taking immunosuppressives on 1 January1979 are an incomplete sample but include patientswith a longer drug exposure who may provide useful'early-warning' information.Of the 716 patients, 220 are male and 494 are
female (two sex not stated). The mean age of thepatients at the time of registration was 55 years(range six to 85). The conditions for whichimmunosuppressive drugs were given are shown intable 1. Azathioprine was the most commonly useddrug (table 2) except in the treatment of polyarteritisnodosa, for which cyclophosphamide was more fre-quently used (table 3).
Seven malignancies have been reported so far(table 4) but it is not possible yet to estimate themalignancy rate. Follow-up of the patients continues
*On behalf of EULAR Standing Committee on International Clinical Studiesand Therapeutic Trials.
and the first full analysis of the findings will be under-taken in 1983.
TABLE 1 Conditions for which immunosuppressive drugswere given
TABLE 4 Type ofmalignancy reported in relation todiagnosis, sex, and immunosuppressive drugsDiagnosis Sex Drug Tumour
RA F Cyclophosphamide Acute myeloid leukaemiaRA M Azathioprine Reticulum cell sarcomaRA M Azathioprine Ca bronchus (oat cell)RA M Azathioprine Ca bronchusRA F Azathioprine Ca cervixRA F Azathioprine Ca head of pancreas
The incidence of malignant disease in patientsreceiving cytotoxic therapy for rheumatoid arthritis
ALLAN B KIRSNER, STEPHEN J FARBER, ROBERT P SHEONAND ROBERT I FINKEL
From the Division ofRheumatology, The Toledo Clinic, and the Medical College of Ohio, Toledo, Ohio,USA
SUMMARY One hundred and twenty-six patientswith definite or classical rheumatoid arthritis admit-ted to hospital between 1965 and 1974 for cytotoxictherapy were studied for the presence and type ofmalignant disease. Each of the cytotoxic treatedpatients was age and sex matched to a rheumatoidarthritis patient admitted to hospital during the sameyears but who did not receive cytotoxic therapy.There was no increase in malignancy in the cytotoxictreated group over the control group.
* * * * *
The introduction of cytotoxic drugs in the treat-ment of rheumatoid arthritis (RA) has raised concernabout haematological and other malignancy in thesepatients. Since our initial report32 of acute leukaemiaoccurring in a patient with rheumatoid arthritis whowas treated with cytotoxic therapy (CT), severalpapers have been published documenting theoccurrence of malignancy in cytotoxic treatedpatients.20 27 30 127 177 178 196 236
In this paper, we present a retrospective age, sex
and duration of disease matched study of the inci-dence of malignancy in patients with rheumatoidarthritis receiving cytotoxic therapy.
Materials and methods
One hundred and twenty-six patients with definite or
classical RA, admitted to hospital between 1965 and1974 for cytotoxic therapy, were matched by com-
puter for age, sex and duration of disease to a groupof patients admitted to hospital with definite or
classical RA who were not treated with cytotoxictherapy (table 1). Groups differed in that thecytotoxic treated group had more active aggressivedisease. Rheumatoid nodules were present in 57 ofthe cytotoxic treated patients versus 37 controlpatients (p<0 05). The mean rheumatoid factor titre(latex tube dilution) was 1:282 in the non-cytotoxic
treated patients and 1:970 in the cytotoxic treatedpatients (p<0001). Acute phase indices (erythrocytesedimentation rate and protein bound hexoses) weregreater in the cytotoxic treated group (table 2).Patient follow-up from entry into the study was 4*7
TABLE 1 Study population
No Duration of Age at Age atdisease at onset of entryentry into disease (yr) * intostudy (yr) study (yr)
The incidence ofmalignant disease in patients receiving cytotoxic therapy for RA Suppl p 33
years for the control group and 4 9 years for thecytotoxic group. Among the cytotoxic treatedpatients, 105 received cyclophosphamide and/ornitrogen mustard; 17 received one or both of these incombination with chlorambucil, thiotepa, matho-trexate or azathioprine; and four received eitherchlorambucil, thiotepa or azathioprine alone.
Results
Five control patients and seven cytotoxic treatedpatients had malignancies before entry into the study(table 3). Nine new malignancies developed in thecontrol patients following entry into the study com-pared with nine new malignancies in the cytotoxictreated patients. There was a preponderance of skinand lung carcinomas in both the control and cytotoxictreated groups. Acute leukaemia did not occur in thecytotoxic treated patients, and our initial case reportof acute granulocytic leukaemia32 remains the onlyhaematological malignancy that occurred in ourpatients with rheumatoid arthritis on cytotoxictreatment. No patient developed bladder carcinoma.
Discussion
Cytotoxic drugs are commonly used in the treatmentof connective tissue disease.20 27 30 32 127 177 178 196 236Following our initial report, Seidenfeld et al196described two patients treated with cytotoxic drugsfor rheumatoid arthritis, who subsequentlydeveloped acute leukaemia. Chromosomal abnor-malities were noted in both patients, and a sidero-blastic anaemia developed in both patients before thedevelopment of leukaemia. The author suggestedthat the onset of sideroblastic anaemia, chromosomaldamage and a markedly decreased ability to formcolonies in marrow culture, were indications for thediscontinuance of cytotoxic therapy.
Pinals1" reported 25 patients receiving azathiop-rine in the treatment of chronic polyarthritis. Twopatients with rheumatoid disease were found to havecarcinoma of the breast and lung; both were treatedwith azathioprine. Pinals proposed that the earlyappearance of malignancy suggested that the role ofazathioprine may have been to permit accelerationof tumour growth rather than to induce malignanttransformation.
Boyle et al20 reviewed the incidence of malignancyin 27 patients ten years after cyclophosphamidetherapy for rheumatoid arthritis. Three patients haddied, one of adenocarcinoma of the breast. Threeadditional solid tumours were identified. One patienthad ductal carcinoma of the breast, another hadadenocarcinoma of the stomach and a third had car-cinoma of the uterus. There were three squamous cellcarcinomas. Compared to a population of age and sexmatched controls, there was no increased incidenceof malignancy in their patients. However, they notedthat the two carcinomas of the breast approached astatistically significant increased incidence.
Plotz et al"78 studied 11 patients treated with cyclo-phosphamide for rheumatoid arthritis. Anaplastictransitional cell carcinoma of the bladder was foundin one patient with rheumatoid arthritis five years afterstopping a therapeutically successful and uncom-plicated course of 50 g cyclophosphamide. Theysuggested that the development of cystitis andbladder carcinoma was a strong argument for limiteduse of cyclophosphamide in non-malignantinflammatory rheumatoid conditions.With the exception of our initial case report, none
of our cytotoxic treated patients developedhaematological malignancy or bladder carcinoma.Nine patients in the control group and nine patientsin the CT group developed malignancy. Thus, wehave not been able to document an increasedincidence of malignancy in cytotoxic treated patientswith rheumatoid arthritis.
Excess risk of lymphomas, leukemia and myelomain patients with rheumatoid arthritis*
H A ISOMAKI, T HAKULINEN AND U JOUTSENLAHTI
From the Rheumatism Foundation Hospital, 18120 Heinola 12; the Finnish Cancer Registry, 00170Helsinki 17; and the Social Insurance Institution of Finland, 00250 Helsinki 25, Finland
SUMMARY The incidence of malignant neoplasmsamong 11 483 male and 34 618 female individualswith rheumatoid arthritis was studied using two sep-
arate nationwide data registers covering the wholeFinnish population: the Social Insurance Institution'sPopulation Data Register, which includes informa-tion on medication for certain chronic diseases, andthe Finnish Cancer Registry, with data on all cancer
patients diagnosed in Finland. The follow-up com-
prised a total of 213 911 person years. The totalincidence of all malignant neoplasms was higher inmales and on the level expected in females. Theexpected number of cases of leukemia, lymphomas,Hodgkin's disease and myeloma in both sexes was
59 6 as compared with the 130 cases observed. Thisdifference is statistically highly significant (p<0001).The incidence of cancer of the respiratory organs washigher in males, and the incidence of cancer of therectum and stomach lower than expected inrheumatoid females.
* * * * *
Conflicting evidence has been presented on the riskof malignant neoplasms in patients with rheumatoidarthritis (RA). Moesman'5' found cancer to be twiceas common in rheumatoid patients as in the generalpopulation. On the other hand, the proportion ofmalignant neoplasms as a cause of death in patientswith rheumatoid arthritis was lower than that in con-trols matched for sex and age.'02 Cancer was under-represented in two autopsied series of rheumatoidarthritis.'3 156 Lewis et al'26 found that cancer was lesscommon in hospitalized RA patients than in patientswith hypertension.
Special attention has been paid to the simultaneousoccurrence of lymphoproliferative neoplasms andinflammatory rheumatic diseases such as
RA.2629 149 234 Oleinick'6' studied the literature inEnglish published up to December 1965. He col-lected a series of 951 RA patients with 9346 person
*Reprinted fromJ Chron Dis 1978; 31: 691-6.
years at risk from published case reports. No supportwas found for the hypothesis that there is anincreased risk of leukemia or lymphoma in RApatients. Miller"49 found the same prevalence of RAin patients with solid tumors (042%) or lympho-proliferative neoplasms (038%) as in the generalAmerican population (0 38 %).
It is possible that the immunological abnormalityin RA patients increases the risk of contractingleukemia, lymphoma, or myeloma. The case reportscannot, however, resolve this question. The potentialsignificance of this hypothesis in an understanding ofthe pathogenesis of these malignancies warrantsfurther investigation.161
In this study the incidence of cancer in a large seriesof patients with rheumatoid arthritis was evaluated.
Patients and methods
The incidence of malignant neoplasms in rheumatoidarthritis patients was studied by making use of twocomputerized nationwide data registers, the SocialInsurance Institution's Population Data Register andthe Finnish Cancer Registry.The Social Insurance Institution's Population Data
Register has included information since 1965 onpatients with chronic diseases entitling them to reim-bursable medicines. One such disease is rheumatoidarthritis. The category 'rheumatoid arthritis' in thisregister also includes systemic connective tissue dis-eases (before 1970) and ankylosing spondylitis (from1970 onwards). In 1974, subjects with ankylosingspondylitis accounted for 2-2% of all patients andthose with systemic connective tissue diseases for1 7% of all patients in the register.'70The Finnish Cancer Registry2"3 was started in
1952, and according to established health servicepractice, every cancer case is reported to this registry.Reporting has been obligatory since 1961. The regis-try receives information from multiple sources: fromhospitals (at various stages of the disease), from prac-titioners, and from pathological laboratories.Moreover, the registry makes an annual screening of
Excess risk oflymphomas, leukemia and myeloma in patients with RA Suppl p 35
all death certificates issued in the country, and thus isinformed of the death of cancer patients from bothcancerous and non-cancerous causes. On average,five notifications are received per case.The patients entitled to fully reimbursable medica-
tion because of RA during the period 1967-1973were retrieved from the Social Insurance Institution'sPopulation Data Register. A total of 11 483 maleand 34 618 female RA patients were identified.
This RA patient series was matched against theFinnish Cancer Registry. The series was followed upfor new cancer cases on or after 1 January 1967, from
TABLE 1 All malignant neoplasms in males by age
Age Number of Person years Observed Expectedrheumatoid at riskpatients
the time at which RA medication was prescribed until31 December 1973, or the death of the patient (atotal of 213 911 person years). Malignant diseasescontracted before the diagnosis of RA wereexcluded.The numbers of various types of neoplasms in both
sexes were compared with those expected on thebasis of the national cancer morbidity figures for theyears 1966-1970.213 The significance of the differ-ences was tested on the assumption that the observednumber of cases followed a Poisson distribution.57
A total of 407 neoplasms were observed in males, theexpected number of cases being 3541. This differ-ence is statistically significant (p<001, table 1). Theincidence of leukemia, lymphomas, myeloma, andcancer of the respiratory organs was significantlyhigher in rheumatoid patients than in the generalmale population (table 2).
In females, the incidence of malignant neoplasmswas approximately equal to that expected (table 3).An excess of leukemia was observed, but the differ-ence was not statistically significant. The incidence oflymphomas, myeloma and Hodgkin's disease wassignificantly higher in females too, while the inci-dence of cancer of the stomach and rectum was lowerthan expected (table 4).The expected number of all cancers of the lym-
phatic system (leukemia, lymphomas, Hodgkin's dis-ease, and myeloma) in both sexes was 59 6 comparedwith the 130 cases observed. This difference is statis-tically highly significant (p<0001).
Discussion
Both data registers used in this study cover the wholeFinnish population. This is one major advantage of asmall population (4-7 million). In addition, it is poss-ible to avoid the problems facing investigators whohave used hospital registers on tumors. When astudy is based on hospital patient registers, it isalways possible that some rheumatoid patients havemalignancies that have been treated at other institu-tions, and the frequency of malignant neoplasmstherefore seems low.
It is probable that the drug user data of the SocialInsurance Institution's Population Data Register donot include all Finnish RA patients. According to thisregister, the prevalence of RA among the wholepopulation in 1971 was 0 3% in males and 1 0% infemales. These figures are low compared with theprevalence of RA obtained in epidemiologicalstudies"48 which, however, usually only apply to pre-valence in adult populations. On the other hand, theFinnish Cancer Registry includes practically all diag-nosed cancer cases in the country.213Cancer has been reported to be more common
among patients with systemic lupus erythematosusthan among, for example, those with RA and hyper-tension126 or among the general population.29 Theprevalence of all systemic connective tissue diseases,including SLE, in this population was only 1-7 %. It isprobable that this does not significantly change theincidence of cancer in the total series. Moreover, no
cancers were found among the causes of death of 30SLE patients in the series of Feng et al.66Many malignant neoplasms can cause rheumatic
manifestations. This is especially true in children, inwhom the clinical picture may suggest, e.g. juvenilerheumatoid arthritis.26 Multiple myeloma may con-fusingly resemble RA.234 Carcinoma polyarthritis issometimes clinically indistinguishable from RA.26 Ifthese rheumatic manifestations preceded the overtappearance of malignancy, the patient may havebeen registered first as RA. Such cases are, however,rare and probably have no significant influence onthe results.
In general, the incidence of epithelial cancers wasas expected. The higher than expected incidence ofcancer of the respiratory organs in males and thelower than expected incidence of cancer of therectum and stomach in females may be due to chance.However, Bradley et al" found an association be-tween rheumatoid factor positivity and lung cancer ina British population.The excess risk of lymphomas, leukaemia, and
myeloma in patients with RA seems indisputable.There has been discussion on whether or not con-tinuous immunological stimulation in RA causes pro-liferation and malignant transformation of someclones of immunologically competent cells.2"4Hyperplasia of the lymphoid system is one of thecharacteristics of RA, and plasmocytosis is fre-quently seen in this disease. Miller149 discusses thetheories on the etiologic relationship betweenimmune diseases and malignant lymphomas. In hisseries of 17 patients, the onset of both processesoccurred simultaneously in nine. He therefore con-cludes that the patients are susceptible to both dis-eases.The excess risk of leukemia among RA patients
was greater for males than for females. The seriesincludes male patients with ankylosing spondylitis.This disease is strongly associated with HL-A antigenB27. Lawler et al2" found an elevated frequency ofthis antigen in patients with lymphoblastic leukemia.It remains to be determined whether more cases ofleukemia have accumulated among patients withankylosing spondylitis than among rheumatoids.The low proportion of cancer deaths in rheumatoid
patients3 102 152 does not seem to be due to a lowmorbidity in cancer. Infections, cardiovascular dis-eases and renal diseases are more often the cause ofdeath in rheumatoid than in non-rheumatoid subjectsresulting in a lower than expected proportion ofdeaths due to cancer.33102152 In addition, the defini-tion of the cause of death of an individual patient isalways a matter of debate.
Side-effects of azathioprine treatment in rheumatoidarthritis: analysis of 10 years of experienceF SPEERSTRA, A M TH BOERBOOMS, L B A VAN DE PUTTE, H J VANBEUSEKOM, M W M KRUIJSEN AND J P VANDENBROUCKE
From the Department ofRheumatology, St Radboud Hospital, Nijmegen, and the Department ofEpidemiology, Erasmus University, Rotterdam, The Netherlands
SUMMARY Our experience with azathioprine in thetreatment of rheumatoid arthritis covers ten years,during which 91 rheumatoid patients (66 female and25 male) received this drug, with a median treatmentperiod of 36 months. Total follow-up experience,during and after treatment, was 399 person years.Twelve patients died. The principal causes of deathwere malignant neoplasm (six patients) and cardio-vascular diseases (three patients). The mortalityin our patients was compared to that of the generalDutch population by the Standardised MortalityRatio (SMR). In the male patient group a significantexcess of both total mortality and mortality frommalignancy was observed. The female patientsshowed no differences from the general population.In this follow-up study, no lymphoreticular tumoursoccurred during or after azathioprine therapy.
* * * * *
Several clinical studies have reported improvementof articular symptoms in rheumatoid patients duringazathioprine therapy.89 95 124 143 177 224 225 After about15 years of experience details of treatment and man-agement during therapy are well defined. The ques-tion whether or not the risk of malignant growth inthese patients is increased is still unanswered andremains of major concern in the prescription of thisdrug. The present follow-up study was undertakenbecause of the disquieting observation of a threehundred-fold increase in occurrence of lympho-proliferative tumours, with a particular cerebral local-isation, in organ transplant patients who receiveazathioprine among other immunosuppressivedrugs. 174
Patients and methods
Ninety-one patients (66 female and 25 male) withclassical or definite rheumatoid arthritis, according tothe American Rheumatism Association (ARA)criteria,'86 were studied retrospectively. Thesepatients were seen at the Department of Rheumatol-ogy of the St Radboud Hospital between 1 January
1970 and 31 December 1979. All suffered fromactive polyarthritis for which azathioprine therapywas instituted. From these patients data were col-lected retrospectively in 1980-1981 on age and sex,previous antirheumatic therapy, duration of arthritis,details of azathioprine treatment, survival and even-tual cause of death. These data were available on allpatients. The causes of death were determined eitherby direct communication with the general prac-titioner or from our medical records or from thedeath certificate. The mortality experience was com-pared with the mortality of the general Dutch popula-tion by means of the Standardised Mortality Ratio(SMR).9 Person years of follow-up were calculatedfrom the start of azathioprine therapy up to the com-mon closing date of the follow-up (31 December1979), for males and females separately, and byfive-year age categories. Death rates for the calcula-tion of the expected mortality were obtained as anaverage of the general population death rates in theNetherlands during the follow-up period. For each ofthe sexes we took the average of the five-year agecategory specific death rates of three calendar yearsin the middle of the follow-up, that is, 1973, 1974and 1975.165 The 95 % confidence limits of the SMRwere calculated under the assumption that theobserved and the expected are realisations of a Pois-son variable9 that can be transformed to approximatenormality with constant variance by square roottransformation.'9
Results
Characteristics of this patient population arepresented in table 1. Previous drug therapy to controlthe rheumatic disease had consisted of gold salts (83patients), D-penicillamine (35 patients) or antima-larial agents (13 patients). Four patients had beentreated with another cytostatic drug; cyclophos-phamide (three patients) and chlorambucil (onepatient). At the start of azathioprine therapy allpatients were suffering from polyarthritis and 53
TABLE 1 Characteristics of 91 rheumatoid patients (25male and 66 female) treated with azathioprine. The data areexpressed as median values
Duration of disease (yr) 9Age at start of therapy (yr) 54Duration of treatment (mth) 36Total dose (g) 120Follow-up (mth) 60
were receiving corticosteroids. The median durationof the disease was nine years (range one to 33 years).The daily dosage of azathioprine ranged from 1P5 to 2mg/kg. In most cases the drug was used without inter-ruption. With this regimen, some patients toleratedthe drug for longer than ten years.
Side-effects consisted mainly of dose relatedreversible bone marrow depression, but were severeenough to necessitate discontinuation of therapy innine patients. Intractable nausea occurred in fourpatients. Additional reasons for discontinuingtherapy were severe bacterial infections (fourpatients) and herpes zoster (four patients). In fiveother patients, azathioprine was withdrawn becauseappropriate outpatient clinic supervision was notguaranteed.The clinical response was judged ineffective in
seven patients after at least three months of treat-ment. The majority experienced clinical improve-ment with long-lasting remission during continuousadministration of azathioprine; half of the patientswere on this regimen for more than three years.
Twelve patients died during the follow-up period(table 2). Necropsy was performed in five cases.Malignancies (six patients) and cardiovascular dis-eases (three patients) were the most common causesof death. The observed and the expected numberswere compared using the method already described.In the male group the difference in both 'total mor-tality' and 'all malignancies' was statistically significantat the 5% level (tables 3 and 4). No differences were
TABLE 2 Causes ofdeath in 12 rheumatoid patients treatedwith azathioprine
Sex Age Necropsy
Malignant tumoursBronchial carcinoma M 64 +
M 60 +M 69 +M 69 -
Renal tumour F 72 -
Breast carcinoma F 57 -
Cardiovascular diseaseMyocardial infarction M 67 -
M 51 -M 60 +
Infectious diseasesDiverticulitis F 56 +
Pneumonia F 81Other
Uraemia/amyloidosis F 66
TABLE 3 Comparison ofmortality experience ('totalmortality') (ICD 000-989) in male and female rheumatoidpatients treated with azathioprine, with the values expected inthegeneral Dutch population on the basis ofthe StandardisedMortality Ratio (SMR) *
Follow-up Observed Expected SMR 95% confidenceperson number number limits ofSMRyears ofdeaths ofdeaths
*Death rates for the general Dutch population were obtained from publishedDutch vital statistics for the calendar years 1973, 1974, 1975."'6tA significant increase at the 5% level in total mortality was observed in thegroup of male patients.
TABLE 4 Mortality in male and female rheumatoid patientstreated with azathioprine. Comparison of the malignanciesobserved in the deceased with the values expected for 'Allmalignancies' (ICD 140-209) ofthe Dutch population on thebasis of the Standardised Mortality Ratio (SMR) *
*Death rates for the general Dutch population were obtained from publishedvital statistics for the calendar years 1973, 1974, 1975."'tThe difference in the group of male patients is statistically significant at the 5 %level.
found in the female group. In four male patientssquamous cell bronchial carcinoma was diagnosed.Since malignant tumours of the respiratory tract, aswell as cardiovascular disease, were the only causesof death in the male patients, we reviewed thepatients' smoking habits from our medical records.An overwhelming majority (96%) had a history ofcigarette smoking, compared to 55 % in the generalmale Dutch population.22'
In the interpretation of our results it should benoted that one female patient had strong clinicalevidence of malignant renal neoplasm (haematuriawith massive tumour on intravenous pyelography),but necropsy was not performed. She was classified asa cancer death.No differences with respect to total dose and dura-
tion of azathioprine therapy could be demonstratedbetween the patients who developed malignanciesand those who did not. Seventy-nine patients are stillalive and under medical outpatient care, whichincludes regular physical examination, blood counts,analysis of liver function and urine testing. X-rays ofthe chest and, if indicated, contrast radiologicalexaminations have been performed. At the end of thefollow-up period, no evidence of further malignan-cies had occurred in these patients, and no lympho-proliferative neoplasms have been diagnosed duringor after azathioprine therapy.
Side-effects ofazathioprine treatment in RA: 10 years analysis Suppl p 39
Discussion
For several years the clinical efficacy of azathioprinein the treatment of chronic inflammatory diseases,including rheumatoid arthritis, has been firmly estab-lished. Studies in rheumatoid patients, both in con-trolled124 143 224 225 and in open studies,8995177 haverevealed improvement of articular symptoms in themajority of patients. Moreover, reduction of cor-ticosteroid dosage has been achieved in the first yearof treatment.'43 III Patients included in these studieswere all suffering from long-standing active polyarth-ritis despite previous antirheumatic therapy.
In our study the majority of patients had previouslybeen treated with either gold salts orD-penicillamine. Disease activity was reflected in thenumber of patients receiving corticosteroids (58 %).In accordance with earlier observations124 143 224improvement of articular symptoms was most pro-nounced in the first six months of treatment withazathioprine and could be maintained, to somedegree, during long-term treatment with a dose of 1 5to 2-0 mg/kg daily. Substantial reduction of corticos-teroids (50% or more) was effected in patients whotolerated the drug for at least six months.
Since long-term treatment is anticipated inrheumatoid patients, one might suspect that thecytotoxic and immunosuppressive properties ofazathioprine may increase the risk of oncogenesis.Cytotoxic effects on bone marrow1 105 and theurogenital tract,' as well as an increased number ofchromosomal aberrations, have been reported,95 butcould not be confirmed by others.230 Whether ameasurable degree of immunosuppression isachieved by low to moderate dosage regimens isuncertain.'24 208 The high incidence of lymphomas,especially reticulum cell sarcoma, with intracerebrallocalisation, in patients with organ transplants isunprecedented."' 1'4 Among other considerations,this fact focuses attention on the possible oncogenic
effect of azathioprine in this category of patient.Nevertheless, the literature does not clearly showwhether the occurrence of any particular malignancyin rheumatoid patients is increased in associationwith azathioprine therapy.6' 125 208Two studies report a higher incidence of lym-
phomas, but both series included various chronicinflammatory conditions""5'7 and different cytosta-tic drugs were considered in one.'74 Moreover, therisk of development of this type of tumour seems tobe increased in rheumatoid patients"'0 and this factshould be taken into account.
Case reports dealing with the presumed oncogeniceffect of azathioprine include leukaemia,3279adenocarcinoma of the lung 177 and breast car-cinoma.'77 Two cases are of special interest, becausethese patients developed malignant lymphoma withintracerebral localisation in both.'27 15 Our observa-tions, covering 399 person years, revealed no lym-phoreticular malignancies during or after azathiop-rine therapy. The excess mortality in the 25 matepatients attracts attention, especially because fourpatients developed bronchial carcinoma.
In rheumatoid arthritis, death rates tend to behigher in the younger patients.33 55 125 219 Principalcauses of death are cardiovascular disease, renal fail-ure 102 156 and, in earlier studies, infectious dis-eases.33 55 Two authors reported an increase ofmalignancies in their series,25 151 but found no differ-ence in the type of tumour between the cytostatictreated group and the control patients.' 125 Theincreased number of squamous cell bronchial car-cinomas found in our male patients may well berelated to their smoking habits. Nevertheless, carefulobservation and follow-up of patients withrheumatoid arthritis receiving azathioprine shouldcontinue. Meanwhile this drug should only be givento those patients with severe rheumatoid arthritiswho do not respond to, or cannot tolerate, otherantirheumatic drugs.
Compliance and long-term effect of azathioprine in65 rheumatoid arthritis cases
P VAN WANGHE AND J DEQUEKER
From the Afdeling Reumatologie, Academisch Ziekenhuis, B-3000 Leuven, Belgium
SUMMARY Azathioprine has been used in our unit as
a third line disease modifying drug (DMD) since1969. In 65 patients with severe rheumatoid arthritis(RA), [45 females and 20 males, mean age 55 2 years
(32 to 76), mean duration of disease 14 years (1 to41)], azathioprine was given in an average dose of 1 5mg/kg body weight/day for a mean duration of 33-4months (range 1 to 108 ).The mean follow-up was five years. One hundred
and eighty-four patient years of treatment withazathioprine were observed. After three months'treatment, significant subjective and objectiveimprovement was observed in 65 % of the cases. Thisimprovement remained in 29 cases who received con-
tinuous treatment for two years. In 12 of the 20seropositive RA cases, a reduction of at least threedilutions in the rheumatoid factor titre was noted. Inthe 24 patients who were corticosteroid dependent,the dosage of steroids could be reduced by 35% andin four steroids could be stopped completely. Com-pliance after two years (n= 54) was still 67 %.Azathioprine treatment had to be stopped in 23patients because of ineffectiveness in nine andadverse effects in 14. In three cases (4 6%) a malig-nant tumour occurred: one lymphoma and twoadenocarcinomata. Low dose azathioprine therapywas shown to be useful as a third line disease modify-ing drug in RA without an increase in oncogenic risk.Compliance for azathioprine was found to be very
satisfactory compared to other drugs.
* * * * *
Since the introduction in 1951106 of immunosup-pressants in the treatment of rheumatoid arthritis,several studies have shown that immunosuppressivedrugs can induce a measurable improvement in theinflammatory process."1 34 38 56 69 95 109 113 177 214 225 237
Although the mechanism of action is still unclear,these agents have been used extensively in recentyears. In our unit, azathioprine has been in use since1969, and is considered as the third line 'diseasemodifying drug' after gold salts and D-penicillamine.In this retrospective study, the long-term effect,compliance and adverse effects of azathioprine have
been evaluated in 65 patients who received the drugin the past decade.
Patients and methods
Sixty-five patients with classical rheumatoid arthritis,according to the American Rheumatism Association(ARA) criteria, were entered into the study. Thecharacteristics of the patients are shown in table 1.There were 20 men and 45 women, with ages rang-
ing from 32 to 76 years (mean 55-2 years), and theaverage duration of rheumatoid arthritis was 14 years(range 1 to 41 years). The mean starting dosage ofazathioprine was 15 mg/kg/day. Dosage wasadjusted during the course of the disease according toclinical response. This study covers 184 patienttreatment years, with an average duration of 33 4months (range 1 to 108) of azathioprine therapy.Treatment before azathioprine is listed in table 2.
Most of the patients had had a course of gold, anti-malarials or D-penicillamine treatment beforeazathioprine was started. Gold, antimalarials orD-penicillamine were discontinued before startingazathioprine. Fewer patients were treated withD-penicillamine since this drug was introduced sev-eral years after azathioprine in our unit. In patientsreceiving steroids and/or depot-ACTH, an attemptwas made to reduce the dose as much as possible.TABLE 1 Characteristics of65 patients treated withazathioprineMale/female 20/45Mean age 55 2 yr (32-76)Seronegative/seropositive 17/48Mean duration of disease 14 yr (1-41)Mean dosage of azathioprine 1-5 mg/kg/dayMean duration of therapy 33 4 mth (1-108)Mean duration of follow-up 5 yrTotal patient treatment years 184
Compliance and long-term effect of azathioprine in 65 RA cases Suppl p 41
All patients continued the treatment with salicy-lates or non-steroidal anti-inflammatory drugs(NSAID), which they were taking when azathioprinewas started.The indications for azathioprine therapy are shown
in table 3. Azathioprine was started, in most cases,because of failure of, or adverse effects to, 'classictherapy': gold, chloroquine or D-penicillamine. In14 patients treatment was given in order to reducecorticosteroid dependency, and in six to controlsystemic involvement such as pericarditis or vas-culitis. Each patient was seen at regular intervals atthe outpatient clinic where the following assessmentswere made: subjective status (better, unchanged,worse), morning stiffness (hours) and clinicalexamination with special attention to the joint status.Laboratory tests performed at each visit includedblood counts, sedimentation rate (ESR), C-reactiveprotein (CRP), rheumatoid factor, serum proteinelectrophoresis, liver function tests and urineanalysis. The activity of the rheumatoid arthritis, asdescribed in the charts, was defined by the followingclinical activity scores, which were designed forretrospective studies.205Grade 0 (remission)-no morning stiffness, no per-
sistent joint pain, tenderness or swelling.Grade I (mild RA)-persistent joint pain withoutmorning stiffness, joint tenderness or swelling.
Grade II (moderately active RA)-presence of
TABLE 3 Indications for azathioprine therapy
Failure of 'classic' therapy 30Adverse reactions to 'classic' therapy 15Corticosteroid dependency 14Systemic involvement 6
joint pain, tenderness and swelling and morningstiffness < 120 minutes.
Grade III (severe RA}-persistent moderate or
marked articular tenderness and swelling,associated with warmth and/or erythema andmorning stiffness >120 minutes.
When charts were incomplete because of failure toattend during the last year, or death, information on
the patient's condition, or cause of death, was soughtfrom the family physician.
Results
THERAPEUTIC EFFECTS
After three months of therapy 42 (65%) of the 65patients had improved subjectively, morning stiffnesshad decreased in 42 (65 %) and the number ofswollen joints had reduced in 46 (70%).To evaluate the long-term effect, a group of 29
patients who were treated for at least 24 months were sel-ected for study. The results for this group who tolerated thedrug are shown in table 4. Subjective improvementoccurred in 21 (72 %) after three months. This effectlasted for 24 months and was confirmed byimprovement in the clinical findings; morningstiffness, number of swollen joints and ESRdecreased significantly. The titre of the latex testdecreased by three or more tube dilutions in 12 of 20seropositive patients after 24 months. There was nosignificant change in haemoglobin levels, CRP or
albumin/globulin ratio.In 24 cases who were dependent on
corticosteroids, it was possible to reduce the dosageby approximately 35 %, and in four steroids could bestopped completely. The response of systemicinvolvement (two vasculitis, two pericarditis, one
TABLE 4 Clinical outcome of29 RA patients receiving azathioprine for 24 months
pleuritis, one Felty's syndrome) was good in all sixcases.
COMPLIANCEIn order to assess the long-term effects and toleranceof azathioprine, compliance to treatment withazathioprine was evaluated in the 54 patients whostarted their treatment two years before completion ofthis study. A compliance study may give importantinformation on therapeutic benefit since a drug whichdoes not adequately relieve pain, or which causesintolerable side-effects, will be discarded by thepatient. After two years, 36 (67 I%) patients werestill continuing azathioprine therapy.The figure compares the drop-out curve presented
as life table survival curve for azathioprine with thoseof gold, D-penicillamine, hydroxychloroquine andlevamisole, as published by Husain and Runge.96Compared to the compliance for other diseasemodifying drugs, azathioprine scores very well.
WITHDRAWALS AND ADVERSE EFFECTSTwenty-three (35 %) of the 65 patients werewithdrawn, nine because of inadequate therapeuticresponse and 14 because of untoward effects whichwere: gastrointestinal disturbance in six cases, bonemarrow suppression (leucocytes <3500,thrombocytes <105/mm3) in seven cases and liverfunction disturbance in one case.
Thirty-eight (58%) of our patients experiencedside-effects mainly in the initial three months oftherapy (table 5).
In three cases a malignant tumour occurred: anundifferentiated lymphoma of the stomach 3-5 years
after start of treatment in a 54-year-old man withFelty's syndrome (the patient is still alive and wellseven years later); one adenocarcinoma of thestomach in a 65-year-old female 6 5 years afterstopping 15 years' treatment with azathioprine(patient is still alive and well ten years later); and onecystadenocarcinoma of the appendix in a 66-year-oldfemale two years after stopping seven years ofazathioprine treatment.
Sixteen patients died and the causes of death areshown in table 6. The average age of death was 64 8years (range 52 to 73).Discussion
The results of our study are in accordance with thoseof others.' 3 5695 109 177 225 After three months ofazathioprine treatment, with an average starting doseof 1-5 mg/kg/day, the majority of our patients,unresponsive to conventional therapy, showedevidence of improvement by subjective, clinical andlaboratory criteria. In a group of patients who
TABLE 6 Cause and age of death of 16 patients who diedduring follow-up period
No Age Sex Cause ofdeath Duration of Interval betweentherapy termination of(yr) therapy and
death (yr)
69 M Postoperative 5 2complication
60 M Postoperative 9 04 complication
67 M Postoperative 3 2 0complication
64 F Postoperative 4 5 4 5complication
71 F Cardiovascular 0 75 2 5disease
69 F Cardiovascular 15 64 disease
67 M Cardiovascular 6-5 0disease
64 F Cardiovascular 3 0disease
2 { 73 F Cachexia 4-5 063 M Cachexia 1 0
2 68 F Sepsis 2-5 076 F Sepsis 3 7
1 64 F Malignancy 7 21 64 F Diabetes 3 01 52 M Suicide 3 5 01 F Unknown 0-5
0 3 6 9 12 18 24Months
* Data from Husoin Z, Runge L J Rheurotol 1980; 7 825-30
Compliance and long-term effect ofazathioprine in 65 RA cases Suppl p 43
tolerated the drug for a long period, thisimprovement continued for at least two years. Exceptfor sedimentation rate and rheumatoid factor, noimportant changes in biochemical parameters ofinflammation were observed.A moderate corticosteroid sparing effect was also
observed. The clinical response observed in thisretrospective study is similar to other studies.Urowitz et al22. treated their patients for 24 weeksand found that, after 12 and 24 weeks, a significantclinical improvement occurred in more than 80% oftheir patients, independent of the dose of the drugtaken. Laboratory evaluation revealed no importantchanges. Hunter et al95 followed their patients up to40 months; the initial improvement at 16 monthscontinued until 40 months. Biochemically they founda decrease in the latex titre, ESR andimmunoglobulin level. Pinals"' treated his patientsfor at least 20 weeks and observed clinicalimprovement in 56 % of his cases. Sedimentation ratedid not change significantly, but the dose ofcorticosteroids could be reduced in seven of the 17patients.Dwosh et al5f compared azathioprine with gold and
chloroquine and found significant improvement in allthree groups after 12 and 24 weeks. The latex titrefell significantly in the azathioprine and gold treatedgroups, but ESR rose in the azathioprine group after12 weeks' treatment, despite continued clinicalimprovement. Berry et al"8 compared azathioprinewith D-penicillamine in a one-year study and foundequal efficacy of both drugs (70%). ESR and latextitre fell in both groups. Currey et al38 comparedazathioprine with gold and cyclophosphamide andobserved a comparable improvement in the threetreatment regimens. The ESR fell in the threegroups, rheumatoid factor did not change andsteroid requirements tended to fall.The higher 'compliance' with azathioprine
compared to other disease modifying drugs96 may beexplained by three facts: as a measure of efficacy it isin accordance with the observed therapeuticresponse; since it is a toxic drug, the follow-up ofpatients is generally more intensive, which has a
positive effect on patients' motivation; patients knowthat azathioprine is, more or less, the 'last chance'drug in controlling their disease. Nevertheless, 35%of our patients were withdrawn from azathioprinetherapy, this being higher than the 18 to 32 %reported in the literature. Adverse reactions werefrequent (58 %), but led to discontinuation oftreatment in only 21 %. Most important weregastrointestinal disturbance and bone marrowsuppression, the latter being reversible in all cases.The frequency of untoward events in our study iscomparable with the mean frequency in the studiesmentioned above.18 38 56 95109 177 225 Only haematuria,observed in six cases, has never been associated withazathioprine and is probably the result of otherfactors (NSAID, infection, cyclophosphamide).
Malignancy occurred in three patients (4-6%).Pinals17" reported an incidence of 8% and Lewis etal"25 4%, which was lower than in patients not treatedwith immunosuppressants (10%), suggesting that analtered immune status of RA patients8" mightfacilitate oncogenesis. Immunosuppressants mayplay a role in tumour induction by chromosomaldamage95 or other mechanisms. According topublished reports, alkylating agents are much moresuspect than antimetabolites.'08 154 184
In our study 16 patients died during follow-up,mainly because of cardiovascular disease andpostoperative complications (table 6). We assumethat the high mortality rate is the result of the lowerlife expectancy of rheumatoid arthritis patients ofthis age group (52 to 73 years, mean 64-8years).33 102 125 182 219 In only two cases is it possiblethat there was a causative relation between treatmentand death (that is, infection and neoplasia).One can confirm that azathioprine has an
important antirheumatic activity but, considering itstoxicity and potential oncogenic risk, its use shouldbe reserved for the specific indications of:
(1) failure of conventional therapy in activerheumatoid arthritis;
Rheumatoid arthritis: treatment with azathioprine(IMURAN (R)). Clinical side-effects and laboratoryabnormalitiesJ K WHISNANT AND J PELKEY
From the Immunology Section, Department of Clinical Investigation, Burroughs Wellcome Co, USA
SUMMARY A retrospective review of the literaturehas been carried out to determine laboratory abnor-malities occurring in patients with rheumatoid arth-ritis (RA) treated with azathioprine, in order toestablish a profile for this agent in the treatment ofthis disease.A total of 542 patients in 24 studies, reported in
the literature, were given a range of doses ofazathioprine for up to four years. Fifteen percent ofpatients were withdrawn because of toxicity. The twomajor toxic effects were gastrointestinal symptomsand alteration in blood counts. Clinically significantleucopenia (less than 2500/mm3) occurred in 14 ofthe total of 93 patients reported to have developedleucopenia. Some adverse reactions, which wouldhave been expected from the use of azathioprine inother diseases, were uncommon, namely significantinfections, hepatotoxicity and pancreatitis. Adverseexperience with azathioprine in rheumatoid arthritiscompares well with other slow-acting, or diseasemodifying, drugs.
* * * * *
Immunosuppressive therapy of rheumatoid arthritis(RA) with the thiopurine azathioprine (IMURAN(R)) has been reported since 1965. The initialexperiences with this drug stimulated someenthusiasm and there were 11 publications in theperiod 1965-1970.5 22 45 51 72 129 143 150 206 207 212 In thenext five years (1971-1975), the problems ofsignificant haematological toxicity and the possibilityof malignancy in renal transplantation patientstreated with azathioprine caused concern, so that lessinvestigation and interest was evident during theseyears. Major studies were reported again in1976-1977 and some of these, especially those byBalken,12 Berry et al, 18 Dwosh et al5" and Goebel etal,8° stimulated re-evaluation of the place ofazathioprine in the treatment of rheumatoid arthritis.A total of 24 studies published in the English litera-
ture, or available in translation, over an 11-yearperiod is now available for review to assess the com-posite adverse experience with azathioprine in thisnon-transplant disease category.
Methods of review
A composite azathioprine literature listing, includingcomputerised reference banks, was first used toestablish a primary literature data base. There are, asalways, multiple citations which mention azathio-prine forRA but which give no clinical data or patient
TABLE 1 Twenty-four articles reviewed with year ofpublication, number ofpatients treated with azathioprine andconcomitant therapy
Reference Year Azathio- Other therapyprine
Apostoloff et al' 1974 9Balken" 1976 41 Steroids, analgesics,
antirheumatic agentsBarnikol and Vorlaender" 1967 9 Antirheumatic*Berry et al' 1976 33 Steroids, anti-
inflammatory agentsBruckner et alP 1969 6 Steroids*Cade et al" 1976 18 Salicylates, steroids (1)*Currey et al" 1974 44 Corticosteroids, aspirin,
paracetamolDenman et al" 1970 5 Prednisone*Dixon et al" 1971 14Dodson and Bennett" 1969 23*Dwoshet al" 1977 11 Steroids (1)Fricke and Deicher72 1969 8 6-methyl-prednisolone,
phenylbutazone,indomethacin
*Goebel et al" 1976 34 Aspirin, indomethacinKhanna and Woodbury"' 1973 21Levy et al'3 1975 49 Salicylates, steroidsLorenzen et al'29 1969 9Mason et al"' 1969 27 Steroids, paracetamolMoens and Brocteurl' 1965 11 Steroids, 6-methyl-
deltahydrocortisone,actinomycin-C
Pinals'77 1976 21 Corticosteroids (17)Swannell and Coomes206 1969 9 Steroids (9), ACTH (1)Swannell and Kersley207 1969 26 PrednisoloneTausch et al"' 1970 73 Steroids (46)*Urowitz et at2" 1973 17 Salicylates*Urowitz et at2" 1974 24 Salicylates
RA: treatment with azathioprine. Clinical side-effects Suppl p 45
experiences, which were eliminated. A smaller, butsignificant, number of publications report fewer thanfive patients without comparison or control patientgroup information; these, too, were eliminated. Theselected published studies included in this analysisare listed in table 1, with the numbers of patients andyear of each publication. Some of the 24 studies werepublished more than once, but duplicate reporting ofthe same patient groups was consolidated in abstract-ing information. The authors' statements regardingdiagnosis, or diagnostic criteria, were accepted.Patients included in these azathioprine studies wereusually those with severe, disabling arthritis, unre-sponsive to conventional therapy. Careful attentionwas directed to eliminating the patients with diseasesother than RA sometimes included in these reports.
All clinical toxicities and laboratory abnormalitiesincluded in each publication were tabulated. Sum-mary profile sheets on each study were prepared, andthese are available from the authors. It was not poss-ible to impose a uniform grading system for severityof toxicities, except where specific haematological orother laboratory data were given. Results are given interms of total percentage of patients with a specifictoxicity. Particular attention was given to casesrequiring withdrawal from treatment, to cases with
e45
-4
4, -
0 20
11
life-threatening or fatal reactions and to toxicitiesoccurring in control groups, where such wereincluded (designated by asterisk in table 1). Adetailed review of original patient records was con-ducted on four of the 24 studies: Urowitz et al,224Urowitz et al,225 Levy et al"22 and Cade et al.25
Results
From this review of the literature, 542 patients couldbe studied (table 1). The most important side-effectof this antimetabolite thiopurine is its effect on whitecell production manifest by peripheral leucopenia.The total incidence of haematological toxicity is illus-trated in fig 1. Individual patient side-effects mayhave been multiple, and the percentage representedis the total for that individual toxicity. There were142 haematological toxicity events reported in a totalof 542 patients. Had all of these events occurred inseparate patients, a total of 26% of patients mighthave been affected.
Leucopenia of any degree was mentioned in 93(17%) of the 542 patients. Of these 93 patients,actual leucocyte values were reported in 67; these aresummarised in fig 1. Fourteen patients were reportedto have a leucopenia low enough to cause clinical risk
34
M/234
,,
-afis
16
14
._-
do00
z
12 I
10 k8
6
4
2
0 I77
13
19
7g
4P f5: 1mm3
FIG 1 Incidence ofhaematological adverse effects reported in 24 published reports of542 patients receiving azathioprine(left). No ofcases with degree of leucopenia as reported in patients on whom data are available (right).
(less than 2500/mm3). The majority of patients hadwhite blood cell counts between 2500 and 3500/mm3,and an additional approximately 25 % had countsabove 3500/mm3. There were an additional fourcases mentioned to have developed 'neutropenia',but the degree of neutropenia was not specified. Dis-continuation of therapy or withdrawal from a studywas uncommon as a result of haematological toxicity.A review of the six studies published in
1976-197712 18 25 56 80 177 shows an incidence ofleucopenia of only 10%.The haematological data from 108 patients in the
four studies reviewed in depth were similar to those
tabulated from the literature. The occurrence of mul-tiple blood count abnormalities in one patient wasconfirmed in these reviews. The study by Levyetal,523in which a higher mean dose of azathioprine (2 9mg/kg/day) was used, included a relatively high inci-dence of leucopenia. A patient's disease activitymeasurements did not correlate with the degree ofleucopenia.
Gastrointestinal complaints were reported 103times in the 542 patients, an incidence of 19% (fig 2).Approximately 10% of patients had nausea and /orvomiting, and gastrointestinal symptoms, not other-wise specified, were reported in 6% of patients.Other symptomatology was reported in less than 1 %of the total for each of the following: anorexia six
cases, ulceration four cases, diarrhoea two cases andulcer two cases. Two gastrointestinal toxicities,reported in other investigational uses of azathio-prine, were not found in a significant percentage inthis review. There were only two reports ofhepatotoxicity, a significant problem in the use ofazathioprine in renal transplantation. There was onlyone report of idiosyncratic, or 'allergic', pancreatitiswhich has been reported in inflammatory bowel dis-ease treated with azathioprine.The review paid particular attention to patient
withdrawal and to prohibitive or life-threateningreactions reported by the authors. Interruption ofazathioprine treatment, or withdrawal of patients,was mentioned in relation to both haematologicaland gastrointestinal toxicity. Eighty-one patientswere withdrawn because of drug-related adversereactions (15%).
Reports of infection (24%) in the context of thisimmunosuppressive therapy were lower than mighthave been expected, with most of these being bacter-ial infection of a non-threatening nature. Fivepatients, in three reports published before 1970, died
TABLE 2 Deaths or neoplasia reported in seven patients
Reference Diagnosis Sex Age Azathioprine Duration Cause ofdeath Comment(mglday)
Denman et al" Rheumatoid F 66 100 12 weeks Bone marrow Chlorambucilarthritis suppressionneuropathy
Lorenzen et al129 Rheumatoid Few weeks Pancytopenia Prednisonearthritis - - -
RA: treatment with azathioprine. Clinical side-effects Suppl p 47
during treatment.45 129 212 These five patients, plustwo who developed malignancy, are summarised intable 2. The three patients reported by Tausch et al2"2had renal dysfunction before immunosuppressivetherapy, and it is doubtful that their deaths duringtreatment were caused by the drug.The special concern about the induction of neop-
lasia in patients receiving this immunosuppressivethiopurine is the subject of other presentations inthese proceedings. While we are aware of 18 indi-vidual case reports of malignancy in RA patients whohave received azathioprine, the total number ofpatients treated is unknown. This review of 542patients from 24 studies identified only two cases ofcarcinoma, an incidence of 0 4 %. This may be com-pared with estimates of 0 6 to 2 6% malignancy in allpatients with RA.
Discussion
The total frequency of adverse reactions in patientswith RA treated with azathioprine compares favour-
ably to the incidence of toxicity with other slow-acting, remission inducing agents. Physicians reportunacceptable side-effects, or significant toxicitiesrequiring withdrawal, in a significant percentage ofpatients and 1% of patients may develop life-threatening toxicity.
The six studies published in 1976 to197712 18 25 56 80 177 show a lower incidence ofleucopenia of 10%, and new treatment programmesare designed to follow recommendations for usinglower doses of azathioprine (1 0 to a maximum of 2 5mg/kg/day). Investigators have suggested temporaryinterruption of treatment, until white blood cellcounts return to normal, and perhaps restarting thedrug at half dosage. Dose modifications for control-ling white counts, in patients taking 6-mer-captopurine or 6-thioguanine, are well known.However, in contrast to the treatment of malignantdiseases, intentional suppression of peripheralleucocyte counts by progressive increase in doseis not considered to be the critical determinant oftherapeutic response in RA.
DR J L DECKER (NIH-USA)We have had a very edifying morning, hearing a verythorough review of the latest available information,which I think most of us are trying to apply practicallyto our patients. I solicit your questions, or your com-ments, on anything that you have heard today.
DR BEGUIN (PARIS, FRANCE)I would like to ask, in view of the time during whichthese drugs have been used, and we now have five orsix years' experience, which are nevertheless verydangerous, why are experiments continued inhumans? We know that there is a certain percentageof deaths and a higher percentage of complications,which may be serious, and which leave the patientmuch more worried by these complications than bytheir polyarthritis. I wish to know how we may pressfor these experiments to be limited; they are also veryexpensive for the laboratories.
DR DECKERAs I understand it, the commentator categorisesmuch of what you have heard this morning as exper-iments and asks if they should be continued. Myresponse would be that rheumatoid arthritis is a ter-rible disease and patients need something. Accord-ingly, I suppose that most of us in the room feel thatthe risks that have been described today, the expense,the laboratory studies, etc, are justified in an attemptto make their lives more bearable. You have to diesomehow ...
DR B CORRIGAN (SYDNEY, AUSTRALIA)Could we hear something of the effects of time ontreatment. In other words, is there an optimal treat-ment time, or time after which azathioprine shouldnot be used at all? Are there any data on the effects oftime on the side-effects? We had one particularinstance of malignancies tending to occur early. Itstrikes me that everybody is looking now for tumoursoccurring later. What is the effect of time and dosageon the problem of malignant side-effects?
DR DECKER
The time to begin treatment would presumably be afunction of the disease. I suppose that there are somepeople who have got into enough trouble with gold,or something else, within the first year of their
rheumatoid arthritis, so that it might be appropriateto begin drugs of this category then. When should webegin to look for malignancies? We have heard DrUrowitz suggest that these neoplasms were already inthe incipient phase, in three females with genitalneoplasms, when the drug was begun. As they werefollowed, one might say that the incidence of malig-nancy was dropping. Do I interpret your remarkcorrectly.
DR M B UROWITZ (TORONTO, CANADA)That might be reading more into the data than isthere, but that is what the data show. The malignan-cies that occurred, occurred early, and this is similarto the experience seen in the renal transplant situa-tion, when the malignancies tend to occur within thefirst two years. I mentioned those two women whoinsisted on going back on azathioprine after they hadtheir hysterectomies for treatment of carcinoma ofthe uterus and cervix. They have been followed nowfor almost a further two years and have notdeveloped a recurrence or another malignancy. Fromthat, I drew the inference that, maybe, they had apremalignant lesion there. I don't think that we canever let up our vigilance. If, after 12 or two years,there is no malignancy, you cannot stop looking. Thepurpose of these large, on-going, and oftenretrospective studies, is to see whether the incidenceis at all increased and, if it is increased, is it increasedearly or late? I think only large studies will give usthese answers.
DR DECKER
This to some extent depends on the drug involved aswell. I think that with alkylating agents moreextended concern is warranted, even after the drugsare stopped. I think Dr de Silva has made it very clearthat the dose of chlorambucil over time is also verypertinent to the potential development of malig-nancy.
DR H DEICHER (HANOVER, GERMANY)I have a question relating to the increased incidenceof lymphoreticular tumours which has been referredto by some of the speakers here this morning. I meanthe increased incidence of lymphoreticular tumoursin diseases like lupus and others. Since we have DrDenman with us, who has already provided us with a
number of interesting assumptions, I should like toask him whether he has an explanation of theincreased incidence of this type of tumour in theseso-called autoimmune diseases.
DR DECKER
Dr Denman, before you respond, Dr Barnes wouldlike to add to your troubles.
DR C G BARNES (LONDON, UK)
I wonder if I might also ask Dr Denman what hethinks is the relevance of Sjogren's syndrome inassociation with rheumatoid arthritis in thesepatients. We know that some patients with Sjogren'ssyndrome have an increased propensity for lym-phoreticular problems. I find it difficult that in manyreports of such tumours in patients treated with vari-ous agents the incidence of Sjogren's syndrome isnever stated.
DR A M DENMAN (NORTHWICK PARK UK)I think that it is evident from what has been said thatwe are speculating about the nature of diseases suchas rheumatoid arthritis. Thus, I can only pile specula-tion on speculation. But let us start with two clearobservations. The first point is that the incidence oflymphoreticular tumours in patients who have re-
ceived cytotoxic drugs following renal and othertransplants is much higher than that observed inpatients receiving the same drugs for the treatment ofautoimmune and chronic rheumatic diseases. Thetransplants provoke persistent lymphoproliferativeactivity which, in combination with mutagenic drugs,leads to neoplasia. The second point is that even
potent carcinogens in animals have a long latentperiod before tumours are evident, and even thenonly certain organs are affected. Thus if one takes avery potent carcinogen such as 06 methylnitrosourea(MNU), and gives it to strains of mice which areeither sensitive or resistant to the oncogenic effects ofthis agent, there is a long latent period beforetumours emerge and only certain sites are involved,as Phil Lawley has clearly shown. His studies haveemphasised that two factors predispose, above all, tothe development of cancer in these mice. The first isthe extent of proliferative activity which goes on inthe target cells for neoplastic transformation. Thus,the thymus has a high level of lymphoproliferativeactivity and is particularly susceptible to tumoursinduced by MNU. It is interesting to reflect thatlymphoreticular tumours are more likely to occur inyounger patients and in systemic lupuserythematosus (SLE) rather than rheumatoid arth-ritis, since in the former conditions there is a high
Discussion Suppl p 49
rate of spontaneous lymphoproliferative activity.Thus, cultured B lymphocytes from patients withSLE produce large amounts of immunoglobulinsspontaneously.The second factor is the extent to which different
tissues can repair the damage induced by MNU in theexperimental system and by cytotoxic drugs such asalkylating agents in clinical practice. More attentionhas to be paid than in the past to such repair mechan-isms. What is peculiar about the thymus, whichis cancer-sensitive in mice receiving powerfulmutagens, as opposed to other organs which appearto be resistant to tumour induction, is the poor abilityof the thymus to remove promutagens from the DNAof thymus cells. The thymus in mice which are cancerprone is inefficient at removing alkylating agents,whereas other tissues like the liver are extremelyefficient. What may save human patients from theoncogenic effects of cytotoxic drugs is the muchgreater efficiency of human lymphocytes, comparedwith murine lymphocytes, in removing promutagens.This point has been emphasised by the results ofcollaborative studies with Phil Lawley and GilmourHarris of the Kennedy Institute. Lymphocytes frompatients with autoimmune diseases may showimpoverished ability to remove promutagens com-pared with lymphocytes from normal subjects, butthis deficiency is not absolute. Lymphocytes bearingpotentially malignant mutations may also be killed bycontinuous exposure to cytotoxic drugs. These arethe sort of areas we should explore in the future, butfor the moment what I have said remains mainlyspeculative.
DR DECKER
In connection with Dr Barnes's remarks, you wouldcount Sjogren's syndrome as a disease of lympho-proliferation, in some degree, in its own right, and thatmakes patients more prone, perhaps.
DR DENMAN
One is always worried that an audience like this couldcontain people such as Keith Whaley or NormanTalal, who know far more about the subject than I do.However, it is most people's impression that theemergence of lymphomas in Sjogren's syndrome isthe end stage of a lymphoproliferative process, whichis initially polyclonal and ends up as a B cell lym-phoma originating in a single malignant clone. If onebelieves that there is some association between the Bcell aberrations which produce autoantibodies andthose which produce lymphoma, it is not unexpectedthat one process should evolve into the other in somepatients.
DRWWBUCHANAN (HAMILTON, CANADA)I think one should inject just a word of caution aboutthese drugs. They are described as 'slow-acting'drugs, but they are also, we should remember,weak-acting drugs. When one looks at the clinicaltrials that have been done, for instance of gold, whichwas Dr Ward's standard 'second-line' drug, one findsin fact that the results indicate it limping ahead ofplacebo. The Empire Rheumatism Council trial63showed a difference between the means of thenumber of active joints in the gold treated people ofnine, and in the control group of six, which is hardly adramatic effect. Grip strength increased by 30 mmHgin the gold treated patients and 10 mmHg in theplacebo treated patients. So I would first make thepoint, we are really dealing not only with a slow-acting but a weak-acting drug.The second point is the toxicity of gold which Dr
Paulus reviewed so well: one death in 10 000 pre-scriptions. We should recall that chloramphenicoldoes that at the rate of 1 in 30 000 and it was removedfrom the pharmacopoeia in the UK as a result, andphenylbutazone in fact kills 1 in 80 000. So this, then,from that data, would be a very toxic drug indeed.One should recall that the data that he was basingthat on was voluntary, not obligatory, recording ofdeaths so in fact we may be dealing with a much moretoxic drug than we would like to believe.
DR DECKER
I thought Dr Paulus made the point very well thatthese drugs were, in fact, hardly perfect forrheumatoid arthritis. We are all aware of their fail-ures, Dr Buchanan, and one of the reasons, I think,we try them is because there are small numbers ofpatients out there, on the far extreme away from themean, that do magnificently and one is always hopingthat each patient will do magnificently. Certainly thepopulation is not uniform in behaviour. This is a veryinteresting issue which we have been running into inthe lymphopheresis work. It is quite clear that whatwe describe as active rheumatoid arthritis is not thesame thing in all patients. I mean, the immunologicalevents going on in 15 patients with activeRA-nodules, systemic involvement, synovitis-aredifferent and we are going to have different effects ofthese medications. This is one of the reasons why it isso important to try to look for predictors, so that wecan pick out people who will respond, and not subjectthe others to the risk of whatever form of treatmentyou are dealing with.
DR S ALEXANDER (CALIFORNIA, USA)Pursuing the first question of Dr Corrigan, there was
a report from Dr Isomaki's group from Heinola, Fin-land,'32 indicating that the gold salts were much moreeffective if used in the first two years, that is whenused early, and in fact when used after the diseaseprocess had been present for longer than two yearswere not effective at a statistically significant level. Iwonder if any of the panelists have any information tobear on this, as to whether this would be confirmed;because if it is true, it would seem that it wouldinvalidate the results of most of these studies andperhaps the effectiveness would not be as bleak as DrBuchanan has pointed out.
DR DECKER
Would anybody care to respond to that?
DR H A ISOMAKI (HEINOLA, FINLAND)I would like to comment. As to the effect of gold, wedon't know if it depends on the effect of the drug itselfor if it is that patients come frequently to the doctorwho sees and takes good care of them. I think that atleast part of the therapeutic effect of gold is the resultof good care of patients. The doctor sees them veryoften and, of course, if you start treatment early, atthe beginning of the disease, you must get betterresults than if you start it later. If you start aftererosions occur you cannot prevent them.
DR M D SKEITH (SEATTLE, USA)A practical point, since we're always looking, DrDecker, for the patient who responds very well tothese medications. Is there anyone on the panel whowould like to comment about the duration of timethat one should persist in the use of any of theseindividual drugs-say azathioprine for instance-before one abandons it as being an ineffectivedrug? Would that be three months, six months, ortwo years?
DR DECKER
I suppose we could have a good long discussion onjust that point. Dr Paulus, would you care torespond?
DR H E PAULUS (LOS ANGELES, USA)That is always a difficult decision to make when oneruns along for three months, or six months, or a yearwith a patient on a particular drug and who still hasactive disease. One of the points that I was trying tomake in my talk was that the sequential use of anumber of slow-acting drugs is of little benefit. It maytake a year before you see a major benefit from adrug. If you stop at that time or earlier and then youstart another one, it takes another six months, or
longer, before you can expect to see a response. In atotally non-statistical way, in patients that I follow, Ifrequently see patients on gold, say after a year oftherapy, who continue to have symptomaticallyactive disease and they are not very happy with howthey are doing. Usually I try to carry on, but some-times the patient convinces me to stop. So we stopand switch to something else, and the patient oftengets a lot worse within the next three or six months,before he sometimes gets better with the seconddrug. So my tendency would be to try to carry on a lotlonger than we usually do.
DR WHISNANT (BURROUGHS WELLCOME,USA)In the US prescribing advice, the data allow you totreat for 12 weeks and, if there is no response by thattime, current studies say that you are at the limit ofthe data. Let me just remind you that Dr Ward'sco-operative study was designed to try to extend thatto a six-month treatment period, which we think hasprobably got to be a minimum time to try to assess anindividual patient's response.
DR BARNES
I remind you that one follow-up study in the UnitedKingdom showed progressive improvement up tosometimes as long as one year.
DR M DE SILVA (CAMBRIDGE, UK)We have evidence of the continued effect ofazathioprine after five years of continuous use. We havewithdrawn azathioprine and we have shown that evenafter five to eight years of continuous use, in a doubleblind manner, if you discontinue the drug the diseasereactivates and it responds to the reintroduction ofthe drug. So the answers to the first and subsequentquestions are that azathioprine continues to exert abeneficial effect over a prolonged period-and ourstudy also included combination with gold-and afterwithdrawal and reintroduction it can again beeffective.
DR UROWITZI think that there is a little confusion here with thequestion. One question is, what do you do with thepatient who does not respond? I think Dr Whisnant'scomments are right, that the data would indicate thatby 12 weeks there is definitely some response if thereis going to be a response. If there is not any clinicalresponse by 12 weeks, it is probably not worth pursu-ing azathioprine. On the other hand, if a patient doesget a response, how long should you continue thatdrug? Of the patients who have already begun to
respond, all the studies have now shown that as youfollow them further and further down the line theymaintain and even improve on that response. Sothere are two answers: if they respond, continue; ifthey have not responded by 12 weeks, you are prob-ably at the end of the trial.
DR GONZALEZ (CANARY ISLANDS, SPAIN)When a patient with rheumatoid arthritis does wellwith gold, do you stop or do you maintain the goldtherapy? How long do you maintain the goldtherapy?
DR DECKER
The question is, with a good response to gold therapy,how long does one continue it? I suppose thateveryone in this room would have a differentresponse to that question. For myself, when I havegot a solid remission, I have continued it for life at therate of 100 mg sodium aurothiomalate (Myocrisin)every three months. It may be a placebo, but I wouldcontinue gold until I got a complete relapse thatdemanded a change to go to something else or I gotinto toxicity-which, as a rule, you don't get afteryears like that, although you can. You did notexamine any x-rays in that AB study, Dr Ward?
DR WARD (SALT LAKE CITY, USA)No!
DR A J GOLDBERG (LONDON, UK)I wonder if I could ask a question of the panel aboutsecond courses of treatment with the same drug. It isour experience that a second course of gold is veryoften not as successful as the first. The questionwould first of all be, why does the panel think thatmight be so, and secondly what is the panel's experi-ence with second courses of other disease modifyingagents as far as success is concerned?
DR UROWITZ
In terms of the cytotoxics, I think that I've probablyanswered, as if I haven't had a response after 12weeks, I give up a cytotoxic.
DR GOLDBERGBut what if a patient had been doing well and wantedto go back on it?
DR UROWITZ
Oh yes, I would have no hesitation in restarting. Andthen your question is the specific side-effects it wasstopped for?
No, my question is, what is the likelihood of a success-ful response to a second course of therapy if the firstone was successful and was stopped for side-effects,but the patient, for example, wanted to start again.We know that with gold, the second course is oftennot very successful. What is the experience with othercytotoxics-other disease modifiers?
DR UROWITZ
For instance, with azathioprine, patients who havebeen stopped for leucopenia and have flared, orpatients who have stopped because we were afraid tocarry them on beyond one or two years in our earlystudies, and then flared, have been restarted. Theyhave responded and we have actually reported that.So that the patients who stopped, because of our fear,have been recaptured.
DR PAULUS
I would agree. I think that a fair number of patientswith gold will have a second response if they are
restarted carefully following the resolution of atoxicity.
DR DECKER
I think that we are reaching the end of our comments.Dr Hitchings.
DR G H HITCHINGS (BW & CO, USA)Much of what we have heard today, including empiri-cal studies of a clinical nature, shows that even aftertwo or three decades we have not come to definitivetreatments. We are beginning to see glimmers bysorting out cell types-cell types that are responsiveto specific drugs. We are beginning to find glimmersof what kind of biochemistry goes on in theseresponding cells. I would like to predict that the nextdecade is going to see some revolutionary changes inthis field.
The problem of adequately suppressing disease activ-ity in rheumatoid arthritis remains with us, and it isacknowledged that currently available drugs are onlyeffective in a proportion of patients. These drugs donot eradicate the disease, their effect in some patientsmay only be modest and all are potentially toxic.The immunological basis for the use of the
immunosuppressive agents is well known and therationale for their use has been reviewed in theseproceedings by Dr Hitchings and Dr Denman, bothof whom envisage substantial developments in thisform of treatment in the future.
Again, the clinical efficacy of gold and azathio-prine has been reviewed, indicating long-term effect ofthese drugs in suppressing rheumatoid inflammatoryactivity. The point has been stressed, however, that tohope to obtain lasting benefit for the patient, sup-pressive medication must be continued for manymonths or years. Increasing benefit to the patientmay continue for one or more years after the drug hasbeen started, and thus, in a chronic progressive dis-ease, short courses of drugs which may modify thedisease process are unlikely to prove effective inmodifying the long-term progress of the disease. It isdifficult to achieve suitable radiological comparisongroups over extended time periods, but azathioprineover 18 months has been associated with less x-raydeterioration than that observed with gold.38
In view of the potential toxicity of this form oftreatment, and the need to continue treatment long-term, much of the discussion has concentrated ontoxic effects. Of these, gastrointestinal intolerance tooral medications and bone marrow suppression, thelatter virtually always reversed by dose reduction, are
well known. These have again been reported in asubstantial proportion of cases. However, the poss-ible potential for the 'immunosuppressive agents' toinduce neoplasia has been reviewed in as much detailas is presently available. The results remain inconclu-sive with sporadic reports of commonly occurringsolid tumours developing in patients while receivingazathioprine or other antimitotic agents. Uncertain-ties as to the incidence of cancer in rheumatoidpatients who have not received cytotoxic drugs aresufficiently great, and the incidence in azathioprinetreated rheumatoids is so low, as to necessitatefurther controlled analysis of the problem. There hasnot been any substantial incidence of lymphoreticu-lar tumours reported as is known to occur in patientsreceiving these drugs after renal transplantation.
Thus, this workshop has served to review the cur-rent knowledge on the effect and toxicity of drugsaimed at suppressing rheumatoid disease, with theconclusion that all of these drugs-gold salts, penicil-lamine and immunosuppressives (azathioprine andcyclophosphamide in particular)-may be partiallyeffective but toxic. Although the immunosuppres-sives still require further evaluation regardingoncogenesis and-studies are continuing, there do notseem to be any notable differences in frequency orseverity of toxic reactions when one drug is comparedwith another. The alkylating agents may present aslightly increased risk. Until more effective and lesstoxic drugs are developed to suppress rheumatoiddisease in our patients, we continue to have to useavailable agents to improve their clinical situation,realising the need for careful supervision duringlong-term treatment.
1 Abel T, Urowitz MB, Smythe HA, Keystone EC, Nor-man CS. Longterm effects of azathioprine inrheumatoid arthritis. Arthritis Rheum 1978; 21: 539.
2 Alexson E, Brandt KD. Acute leukemia after azathio-prine treatment of connective tissue disease.Am J MedSci 1977; 273: 335-40.Allison AC, Hove T, Watts RWE, Webster ADB.Immunological observations on patients with Lesch-Nyhan syndrome, and on the role of de novo purinesynthesis in lymphocyte transformation. Lancet 1975;ii: 1178-82
4 Alpert ME, Uriel J, deNechaud B. Alpha-l-fetoglobulin in the diagnosis of human hepatoma. NEngl J Med 1968; 278: 984.Amor B, Mery C. Chlorambucil in rheumatoid arth-ritis. Clin Rheum Dis 1980; 6: 567-84.
6 Anderson LG, Talal N. The spectrum of benign tomalignant lymphoproliferation in Sjogren's syndrome.Clin Exp Immunol 1972; 10: 199-221.Apostoloff E, Reitzig P, Jendrusch C, Apostoloff G.Experience with long-term immunosuppressive treat-ment of diseases of the connective tissue. Z GesamteInn Med 1974; 29: 26-30 (translation).Aptekar RG, Steinberg AD, Decker JL. Complicationsof cytotoxic agents in systemic lupus erythematosus andrheumatoid arthritis. Arthritis Rheum 1973; 16: 533(abst).Armitage P. Statistical methods in medical research.Oxford: Blackwell Scientific Publications, 1971:388-91.Asherson GL, Webster ADB. Clinical immunodefi-ciency diseases. Oxford: Blackwells, 1980.
12 Balken WJ. Therapy of progressive chronic polyarth-ritis. Clinical studies of the effectiveness of azathiop-rine. Fortschr Med 1976; 94: 1504-6 (translation).
13 Ball J. Postmortem findings and articular pathology inrheumatoid arthritis. In: Duthie JJR, AlexanderWRM, eds. Rheumatic diseases. Pfizer Medical Mono-graph No 3. Edinburgh: Edinburgh University Press,1968: 124-6.
14 Barnes CG. Immunosuppressive therapy inrheumatoid arthritis. In: Miescher PA, ed.Immunosuppressive therapy. Proceedings of the Wies-baden Symposium 1972. Basel, Stuttgart: Schwabe,1980: 57-68.Barnikol HU, Vorlaender KO. Prerequisites and initialresults with the therapeutic administration of azathio-prine (Imuran) in chronic-progressive articularrheumatism and the so-called collagenoses. Med Welt1967; 3: 160-9 (translation).
16 Batchelor JR, Welsh KI, Tinoco RM, et al. Hydral-lazine induced systemic lupus erythematosus: the influ-ence of HLA-DR and sex upon susceptibility. Lancet1980; 1:1107-9.
17 Bennett JC. The infectious aetiology of rheumatoidarthritis. Arthritis Rheum 1978; 21: 531-8.
1# Berry H, Liyanage SP, Durance RA, Barnes CG,
Berger LA, Evans S. Azathioprine and penicillamine intreatment of rheumatoid arthritis: a controlled trial. BrMedJ 1976; i: 1052-4.Box GEP, Hunter WG, Hunter JS. Statistics forexperimenters. New York: Wiley Scientific Publica-tions, 1978: 144-5.Boyle DJ, Day JF, Kassan SS, Thomas MR, RobinsonWA, Steigerwald JC. Incidence of malignancy 10 yearsfollowing cyclophosphamide use for rheumatoid arth-ritis. Paper presented before the American Rheuma-tism Association, Boston, Massachusetts, June 3, 1981.Brinkley JR Jr, Dubois EL, Ryan SJ. Long-term courseof chloroquine retinopathy after cessation of medica-tion. Am J Ophthalmol 1979; 88: 1-11.
22 Bruckner I, Stroescu 0, Stroescu V, et al. Investigationson the effect of azathioprine (Imuran) in rheumatoidarthritis and lupus erythematosus disseminatus. RevRoum Med Intern 1969; 1: 19-30.
23 Buriot D, Prieur AM, Lebranchu Y, Messerschmitt J,Griscelli C. Leucemie aigue chez trois enfants atteintsd'arthrite chronique juvenile traites par chlorambucil.Arch Fr Pediatr 1979; 36: 592-8.
24 Bywaters E, Curwen M, Dresner E, Dixon AStJ. 10year follow up study of rheumatoid arthritis AnnRheum Dis 1961; 20: 198.
25 Cade R, Stein G, Pickering M, Schlein E, Spooner G.Low-dose, long-term treatment of rheumatoid arthritiswith azathioprine. South Med J 1976; 69: 388-92.
26 Calabro JJ. Cancer and arthritis. Arthritis Rheum 1967;10: 553-67.
31 Chanmougan D, Schwartz RS. Enhancement of anti-body synthesis by 6-mercaptopurine. J Exp Med 1966;124: 363-78.
32 Cobau CD, Sheon RP, Kirsner AB. Immunosuppres-sive drugs and acute leukemia. Ann Intern Med 1973;79: 131-2.
3 Cobb S, Anderson F, Bauer W. Length of life and causeof death in rheumatoid arthritis. N Engl J Med 1953;249: 553-6.
34 Cooperating Clinics Committee of the AmericanRheumatism Association. A controlled trial of cyc-lophosphamide in rheumatoid arthritis. N Engl J Med1970; 283: 883-9.Cooperating Clinics Committee of the AmericanRheumatism Association. A controlled trial of high andlow dose of cyclophosphamide in 82 patients withrheumatoid arthritis. Arthritis Rheum 1972; 15: 434.
36 Cooperman JM. Microbiological assay of serum andwholeblood folic acid activity.Am J Clin Nutr 1967; 20:1015.
37 Court-Brown WM, Doll R. Leukaemia and aplasticanaemia in patients irradiated for ankylosing spon-
dylitis. MRC Special Report Series No 295. London:HMSO, 1957.
3 Currey HLF, Harris J, Mason RM, et al. Comparison ofazathioprine, cyclophosphamide and gold in treatmentof rheumatoid arthritis. Br Med J 1974; iii: 763-6.
3 Dale GA, Smith RB. Transitional cell carcinoma of thebladder associated with cyclophosphamide. J Urol1974; 112: 603-4.
4 Davis P. Undesirable effects of gold salts. J Rheumatol1979; 6 (suppl 5): 18-24.
41 DeBoer KP, Kleinman R, Teodorescu M. Identifica-tion and separation by bacterial adherence of humanlymphocytes that suppress natural cytotoxicity. JImmunol 1981; 126: 276-81.
42 Decker JL. Chairman's summary of session V. Evalua-tion of DMARD's. In: Paulus HE, Ehrlich GM, eds.Symposium Medica Hoechst. Controversies in the clini-cal evaluation of analgesic/anti-inflammatory/anti-rheumatic drugs. Stuttgart, New York: Schattaur Ver-lag, 1981: 378-9.
4 Denman AM. Viruses and autoimmune diseases. In:Holborow EJ, ed. Autoimmunity. Clinics in Immunol-ogy and Allergy 1. London: Saunders, 1981: 17-39.
i4 Denman AM, Hollingworth P, Tyndall AdeV, et al.Intensive immunosuppression versus prednisone in thetreatment of connective diseases. Ann Rheum Dis (inpress).
4' Denman EJ, Denman AM, Greenwood BM, Gall D,Heath RB. Failure of cytotoxic drugs to suppressimmune responses of patients with rheumatoid arth-ritis. Ann Rheum Dis 1970; 29: 220-31.
46 de Seze S, Kahn M. L'association de produits a I'actioncytolitique aux corticosteroides dans le traitement desrhumatismes inflammatoires chroniques. In:Rhumatismes inflammatoires chroniques. Lym Tre-voux, 1965.
47 Deshayes P, Renier JC, Bregeon C, Hondent G. Inci-dents et accidents des therapeutiques a vis6eimmunodepressive dans la polyarthrite rhumatoide.Rev Rhum Mal Osteoartic 1971; 38: 797.
48 Dixon AStJ, Davies J, Dormandy TL, et al. Syntheticd(-) penicillamine in rheumatoid arthritis: double-blind controlled study of a high and low dosage regi-men. Ann Rheum Dis 1975; 34: 416-21.
49 Dixon AStJ, Lindsay DJ, Collins EM. Immunosuppres-sive drugs in rheumatoid arthritis. (Letter.) Br Med J1971; i: 460.
5 Dixon FJ, Theofilopoulos AN, Izul S, McConahey PJ.Murine SLE-aetiology and pathogenesis. In:Fougereau M, Dausset J, eds. Immunology 80. Pro-gress in Immunology IV. London: Academic Press,1980: 959-95.
51 Dodson WH, Bennett JC. Possible usefulness ofazathioprine (Imuran) in severe rheumatoid arthritis. JClin Pharmacol 1969; 9: 251-8.
52 Dubois EL. Lupus erythematosus: a review of the cur-rent states of discoid and systemic lupus erythematosusand their variants. New York: McGraw-Hill, 1966.
5 Duclos H, Maillot MC, Kreis H, Fries D, Galanaud P.T suppressor cell function impairment in peripheralblood lymphocytes from transplant recipients under
azathioprine and corticosteroids. Transplantation1979; 28: 437-8.
5 Duclos H, Maillot MC, Kreis H, Fries D, Galanaud P.Depressed T suppressor cell function in PBL of trans-planted patients under azathioprine and corticos-teroids. Transplant Proc 1979; 11: 1246.
5 Duthie JJR, Brown PE, Truelove LH, Naragar FD,Lawrie AJ. Course and prognosis in rheumatoid arth-ritis. A further report. Ann Rheum Dis 1964; 23: 193.
56 Dwosh IL, Stein HB, Hunter T, Urowitz MB, SmytheHA, Ogryzlo MA. Azathioprine in early rheumatoidarthritis: a comparison study with gold and chloro-quine. Arthritis Rheum 1977; 20: 685-92.
5 Ederer F, Mantel N. Confidence limits on the ratio oftwo Poisson variables. Am J Epidemiol 1974; 100:165-7.
5 Editorial. Renal transplantation in the 1980s. Br MedJ1980; 280: 503-4.
5 Elion GB, Callahan SW, Bieber S, Hitchings GH, Run-dles RW. A summary of investigations with 6-[(1-methyl-4 nitro-5-imidazolyl)] thiopurine. CancerChemother Rep 1961; 14: 93-8.Elion GB, Hitchings GH. Azathioprine. In: SartorelliAC, Johns DG, eds. Antineoplastic and immunosup-pressive agents II. Berlin, Heidelberg, New York:Springer Verlag, 1975: 404-25.
61 Elliot GB, Silverberg DS, Dossetor JB, Muir CS.Latent carcinoma of the prostate in a 24 year old manreceiving cyclophosphamide and azathioprine. CanMed Assoc J 1977; 116: 651-2.
62 Empire Rheumatism Council. Gold therapy inrheumatoid arthritis: report of a multicentre controlledtrial. Ann Rheum Dis 1960; 19: 95-119.
63 Empire Rheumatism Council. Gold therapy inrheumatoid arthritis: final report of a multicentre con-trolled trial. Ann Rheum Dis 1961; 20: 315-34.
68 Forni AM, Koss LG, Geller W. Cytological study of theeffect of cyclophosphamide on the epithelium of theurinary bladder in man. Cancer 1964; 17: 1348-55.
69 Fosdick WM, Parsons JL, Hill DF. Longterm cyc-lophosphamide therapy in rheumatoid arthritis. Arth-ritis Rheum 1968; 11: 151-61.
72 Fricke R, Deicher H. Studies on therapy of rheumatoidarthritis (progressive chronic polyarthritis) with cytos-tatic drugs. Z Rheumaforschung 1969; 28: 20-9.
7 Frisch AW, Davies GH. Inhibition of hemagglutininformation by thioguanine: dose-time relationships.Proc Soc Exp Biol Med 1962; 110: 444-9.
74 Fusco F, Mattioli F, Bertocchi I. Development ofmalignancy after immunosuppressive treatment(azathioprine). First Meeting ofthe European Divisionofthe International Society ofHematology, Milan, Sep-tember 1971.
7 Gardner DL. Causes of death. In: The pathology ofrheumatoid arthritis. London: Arnold, 1972: 183-7.
76 Ghaffar A, Lichter W, Wellham LL, Paul RD, SigelMM. Effects of various alkylating agents onT-dependent and T-independent immune responses.Proceedings of the 11th International Congress onChemotherapy 1980; 2: 1716-7.
7 Giblett ER, Ammann AJ, Wata DW, Sandman R,Diamond LK. Nucleoside-phosphorylase deficiency ina child with severely defective T-cell immunity andnormal B-cell immunity. Lancet 1975; i: 1010-3.
78 Giblett ER, Anderson JE, Cohen F, Pollara B,Meuwissen HJ. Adenosine-deaminase deficiency intwo patients with severely impaired cellular immunity.Lancet 1972; ii: 1067-9.
7 Gilmore IT, Holden GT, Rodan KS. Acute leukemiaduring azathioprine therapy. Postgrad Med J 1977; 53:173-4.
80 Goebel KM, Janzen R, Joseph K, Borngen U. Disparitybetween clinical and immune responses in a controlledtrial of azathioprine in rheumatoid arthritis. Eur J ClinPharmacol 1976; 9: 405-10.
81 Goldenberg GJ, Paraskevas F, Israels LG. The associa-tion of rheumatoid arthritis with plasma cell and lym-phocytic neoplasms. Arthritis Rheum 1969; 12:569-78.
8 Gordon DA, Smythe HA, Pruzanski MA, OgryzloMA. Azathioprine in rheumatoid arthritis. A double-blind cross over study. Arthritis Rheum 1973; 16:411-8.
83 Grooten J, DeBaetselier P, Vercauteren E, Hamers R.Anti-micrococcus antibodies recognise an antigenicmarker of confluent mouse lymphoid cell lines. Nature1980; 285: 401-3.
84 Gumpel JM, Hall A, Ansell B. A double-blind com-parative trial of cyclophosphamide and gold inrheumatoid arthritis. Ann Rheum Dis 1974; 33: 574.Gupta PK, Pinn VM, Taft PD. Cervical dysplasiaassociated with azathioprine (Imuran) therapy. ActaCytol (Baltimore) 1969; 13: 373.
86 Gupta S, Good RA. Markers of human lymphocytesubpopulations in primary immunodeficiency and lym-phoproliferative disorders. Semin Hematol 1980; 17:1-29.
87 Hamilton EBD, Scott JT. Hydroxychloroquine sulfate('Plaquenil') in treatment of rheumatoid arthritis. Arth-ritis Rheum 1962; 5: 502-12.
88 Harris G, Lawley PD, Hylton W, Denman AM,Asbery LJ. Blood mononuclear cells in auto-immunediseases are deficient in removal of 06-methylguaninefrom DNA. In preparation.
89 Harris J, Jessop JD, Chaput de Saintonge DM. Furtherexperience with azathioprine in rheumatoid arthritis.Br Med J 1971; 4: 463-4.
9 Hicks ND. An unusual form of leukemia associatedwith immunosuppressive therapy. Pathology 1973; 5:77.
92 Hitchings GH, Elion GB. The role of antimetabolites inimmunosuppression and transplantation. AccountsChem Res 1969; 2: 202-9.
9 Hoh KF, Dwosh IL, Ford PM, Anastassiades TP, KellyHG. Drug-related disease remissions in rheumatoidarthritis. Arthritis Rheum 1981; 24: suppl 72.
9 Hoover R, Fraumeni JF Jr. Risk of cancer in renaltransplant recipients. Lancet 1973; ii: 55.
9 Hunter T, Urowitz MB, Gordon DA, Smythe HA,Ogryzlo MA. Azathioprine in rheumatoid arthritis: a
long term follow up study. Arthritis Rheum 1975; 18:15-20.
96 Husain Z, Runge LA. Treatment complications ofrheumatoid arthritis with gold, hydroxychloroquine,d-penicillamine, and levamisole. J Rheumatol 1980; 7:825-30.
97 Huskisson EC, Gibson TJ, Balme HW, et al. Trial com-paring d-penicillamine and gold in rheumatoid arthritis.Ann Rheum Dis 1974; 33: 532-5.
98 International symposium on levamisole in rheumatoidarthritis.J Rheumatol 1978; 5, suppl 4: 1-153.
9 International symposium on penicillamine. JRheumatol 1981; 8, suppl 7: 1-181.International workshop and symposium on gold salts inthe rheumatic diseases. J Rheumatol 1979; 6, suppl 5:1-164.Isomaki HA, Hakulinen T, Joutsenlahti U. Excess risksof lymphomas, leukemia and myeloma in patients withrheumatoid arthritis. J Chron Dis 1978; 31: 691-6.
182 Isomaki HA, Mutru 0, Koota K. Death rate and causesof death in patients with rheumatoid arthritis. Scand JRheumatol 1975; 4: 205-8.
103 Jaffe IA. The effect of penicillamine on the laboratoryparameters in rheumatoid arthritis. Arthritis Rheum1965; 8:1064-79.
104 Jensen MK. Chromosome studies in patients treatedwith azathioprine and amethopterin. Acta Med Scand1967; 182: 445.
185 Jensen MK. Effect of azathioprine on the chromosomecomplement of human bone marrow cells. Int J Cancer1970; 5: 147-51.
186 Jimenez-Diaz C, Lopez Garcia E, Merchante A,Perianes J. Treatment of rheumatoid arthritis with ni-trogen mustard. Preliminary report. JAMA 1951; 147:1418-9.
17 Kahn MF, Arlet J, Bloch-Michel H, Caroit M, ChaouatY, Renier JC. Leuce'mies aigues apres traitement paragents cytotoxiques en rhumatologie: 19 observationschez 2006 patients. Nouv Presse Med 1979; 8/17:1393-7.
188 Kahn MF, Arlet J, Bloch-Michel H, Caroit M, ChaouatY, Renier JC. Le risque de leucose aigue apres traite-ment des rheumatismes inflammatoires chroniques et
des connectivities par les cytotoxiques a vis6eimmunosuppressive. Rev Rhum 1979; 46: 163-7.Katona G, Larazza 0, Isai M. Objective evaluation ofthe anti-inflammatory effect of azathioprine inrheumatoid arthritis. Am J Rheum and Infl 1978; 1:232-6.Katz DH. Lymphocyte differentiation, recognition andregulation. New York, San Francisco, London:Academic Press, 1977.Kay AGL. Myelotoxicity of d-penicillamine. AnnRheum Dis 1979; 38: 232-6.
112 Kean MF, Dwosh IL, Anastassiades TP, Ford PM,Kelly HG. The toxicity pattern of d-penicillaminetherapy. Arthritis Rheum 1980; 23: 158-64.
113 Keitel W. Ergebnisse und Erfahrungen zumimmunosuppressiver Therapie der Rheumatoide Arth-ritis. Allerg Immunol (Leipz) 1976; 12: 303-5.
114 Khanna V, Woodbury JFL. Usefulness of azathioprinein rheumatoid arthritis. Ann R Coll Phys Surg Canada1973; 6: 60 (abst).
1 Kinlen LJ, Sheil AGR, Peto J, Doll R. CollaborativeUnited Kingdom-Australasian study of cancer inpatients treated with immunosuppressive drugs. BrMedJ 1979; ii: 1461-6.
117 Klein J, Juretic A, Baxevanis CN, Nagy ZA. The tradi-tional and a new version of the mouse H-Z complex.Nature 1981; 291: 455-60.
110 Klein MB, Pereira FA, Kantor I. Kaposi sarcoma com-plicating systemic lupus erythematosus treated withimmunosuppression.Arch Dermatol 1974; 110: 602-4.Kobayakawa T, Louis J, Izui S, Lambert PH. Autoim-mune response to DNA, red blood cells and thymocyteantigens in association with polyclonal antibody synth-esis during experimental African trypanosomiasis. JImmunol 1979; 122: 296-330.
120 Larson EL. Cyclosporin A and dexamethasone sup-press T cell responses by selectively acting at distinctsites of the triggering process. J Immunol 1980; 124:2828.
121 Lawler SD, Klouda PT, Hardisty RM, Till M. The HLAsystem in lymphoblastic leukaemia. Br J Haematol1971; 21: 595-605.
132 Lawley PD. DNA as a target of alkylating carcinogens.Br Med Bull 1980; 36: 19-24.
123 Levy J, Paulus HE, Bangert R. Comparison ofazathioprine (AZ) and cyclophosphamide (CPA) in thetreatment of rheumatoid arthritis (RA). ArthritisRheum 1975; 18: 412-3 (abst).
124 Levy J, Paulus HE, Bamett EV, Sokoloff M, BangertR, Pearson CM. A double-blind evaluation of azathio-prine treatment in rheumatoid arthritis and psoriaticarthritis. Arthritis Rheum 1972; 15: 116-7.
125 Lewis P, Hazleman BL, Hanka R, Roberts S. Cause ofdeath in patients with rheumatoid arthritis with particu-lar reference to azathioprine. Ann Rheum Dis 1980; 39:457-61.
126 Lewis RB, Castor CW, Kinsley RE, Bole G. Frequencyof neoplasia in systemic lupus erythematosus and
127 Lipsemeyer EA. Development of malignant cerebrallymphoma in a patient with systemic lupuserythematosus treated with immunosuppression. Arth-ritis Rheum 1972; 15: 183-6.
128 Lopez DM, Paul R, Herbert L, Wellham L, Lichter W.Selective effects of cyclophosphamide on differentparameters of the immune response. Proceedings ofthe11th International Congress on Chemotherapy 1980; 2:1726-7.
129 Lorenzen I, Brun C, Videbaek A. Treatment ofimmunologic diseases with cytostatics. Acta Med Scand1969; 185: 501-6.
130 Louie S, Schwartz RS. Immunodeficiency in thepathogenesis of lymphoma and leukaemia. SeminHematol 1978; 15: 117.
121 Love RR, Sowa JM. Myelomonocytic leukaemia fol-lowing cyclophosphamide therapy of rheumatoid dis-ease. Ann Rheum Dis 1975; 34: 534-5.
132 Luukkainen R. Chrysotherapy in rheumatoid arthritis:with particular emphasis on the effect of chrysotherapyon radiographical changes and on the optimal time ofinitiation of therapy. Scand J Rheumatol 1980; suppl34: 1-56.
134 McAdam L, Paulus HE, Peter JE. Adenocarcinoma ofthe lung during azathioprine treatment. ArthritisRheum 1974; 17: 92-4.
135 McConkey B, Amos RS, Durham S, Forster PJG,Hubball S, Walsh L. Sulphasalazine in rheumatoidarthritis. Br Med J 1980; i: 442-4.
136 McConkey B, Davies P, Crockson RA, Crockson AP,Butler M, Constable TJ. Dapsone in rheumatoid arth-ritis. Rheumatol Rehabil 1976; 15: 230-4.
37 McConkey B, Davies P, Crockson RA, Crockson AP,Butler M, Constable TJ, Amos RS. Effect of gold,dapsone and prednisone on serum C-reactive proteinand haptoglobin and the erythrocyte sedimentationrate in rheumatoid arthritis. Ann Rheum Dis 1979; 38:141-4.
138 McCracken JD. Additional risk from immunosuppres-sive therapy for rheumatoid arthritis. Ann Intern Med1973; 79: 611-2.
139 Mackenzie AH, Scherbel AL. Chloroquine and hyd-roxychloroquine in rheumatological therapy. ClinRheum Dis 1980; 6: 545-66.
meiotic chromosome damage following in vivo expos-ure to methotrexate. Clin Genet 1971; 2: 28.Menkes CJ, Levy JP, Weill B, Delrien F, Mathiot C,Delbarre F. Leucemie aigue a megacaryoblasts sur-venue apres traitement immunodepresseur d'unepolyarthrite rhumatoide. Nouv Press Med 1975; 4:2869.
146 Michlmayr G, Gunther R, Lederer B, Huber H.Malignes lymphom nach zweijahriger immunosuppres-siver therapie einer periarteritis nodosa. Med Klin1973; 68: 180.
147 Mielent H, Veys EM. A study of hemotologic sideeffects of levamisole in rheumatoid arthritis withrecommendations. J Rheumatol 1978; 5: 4.
148 Mikkelsen WM. The epidemiology of rheumatic dis-eases. In: Hollander JL, McCarty DJ Jr, eds. Arthritisand allied conditions. Philadelphia: Kimpton, 1972:211.
149 Miller DG. The association of immune disease andmalignant lymphoma. Ann Intern Med 1967; 66:507-21.
150 Moens C, Brocteur J. Treatment of rheumatoid arth-ritis with immunosuppressive drugs. I. Clinical study.Acta Rheum Scand 1965; 11: 212-20.
151 Moesman G. Malignancy and mortality in subacuterheumatoid arthritis in old age. Acta Rheum Scand1969; 15: 193-9.
152 Monson RR, Hall AP. Mortality among arthritics. JChron Dis 1976; 29: 459-67.
153 Moorhead PS, Nowell PC, Mellman WJ. Chromosomepreparation of leukocytes cultured from humanperipheral blood. Exp Cell Res 1960; 20: 613.
154 Mornington B. Nebenwirkungen von azathioprinebeim Menschen. Vortrag des 10 e Europaischen Kon-gresses fur Allergologie und Klinische Immunologie,September 1977, Prag.
1 Multicentre Trial Group. Controlled trial ofd-penicillamine in severe rheumatoid arthritis. Lancet1973; i: 275-80.
156 Mutru 0, Koota K, Isomaki H. Causes of death inautopsied RA patients. Scand J Rheumatol 1976; 5:239-40.Neuberg R. Immunosuppression and plasmocytoma ofthe cervix. J Obstet Gynaecol British Commonwealth1974; 81: 165-7.Neuhaus K, Thornhorst J, Bertel 0, Speck S. Mehr-fachtumoren unter Azathioprin Therapie beiprogredient-chronischer Polyarthritis. Praxis 1976; 65:213-21.
165 Overledenen naar doodsoorzaak, leeftijd en geslacht inhet jaar 1973-1974-1975. CBS HoofdafdelingGezondheidsstatistieken.
166 Owen DS, Waller M, Toone E. Rheumatoid diseaseand malignancy. Arthritis Rheum 1967; 10: 302.
167 Owen DS Jr, Waller M, Toone EC Jr. Rheumatoidarthritis and malignancy. Med Coll Virginia Q 1970; 6:8-10.
168 Parker D, Dwyer JM, Turk JL. The effect of cyclophos-phamide and the role of suppressor cells in the desensit-ization of delayed hypersensitivity. Immunology 1981;43: 191-6.
169 Parsons JL, Strong JS, Fosdick WM. The causes ofdeath in patients with rheumatoid arthritis treated withcytotoxic agents. J Rheumatol 1974; 1, (suppl 1): 75.
170 Patiala H, Isomaki H, Laurinkari J. Rheumatoid arth-ritis and totally reimbursable medicines (in Finnish).Sosiaalilaaket Aikak 1975; 12: 584-7.
195 Seabright M. A rapid banding technique for humanchromosomes. Lancet 1971; ii: 971-2.Seidenfeld AM, Smythe HA, Ogryzlo MA, UrowitzMB, Dotton DA. Acute leukaemia in rheumatoid arth-ritis treated with cytotoxic agents.J Rheumatol 1976; 3:295-304.
197 Shand FL. The immunopharmacology of cyclophos-phamide. ntJ Immunopharmacol 1979; 1: 165-71.
198 Sharon E, Greenwalk R, Solish G, Diamond H, KaplanD. Adverse effects of azathioprine (Az) in patients withsystemic lupus erythematosus (SLE). Clin PharmacolTher 1972; 13: 152.Sharpstone P, Cameron JS. Nephrotic syndrome due toprimary renal disease in adults. II. A controlled trial ofprednisone and azathioprine. Br Med J 1969; ii: 535.
200 Silvergleid AJ, Schrier SL. Acute myelogenousleukemia in two patients treated with azathioprine fornonmalignant diseases. Am J Med 1974; 57: 885.
201 Singer JM, Plotz CM. The latex fixation test. Applica-tion to the serologic diagnosis of rheumatoid arthritis.Am J Med 1956; 21: 888.
202 Skinner MD, Schwartz RS. Immunosuppressivetherapy. N Engl J Med 1972; 287: 221.
203 Slater A, Whittaker JA, Fisher OJH. Reticulum cellsarcoma and Sjogren's syndrome in a patient treatedwith azathioprine. New Engl J Med 1976; 294: 51.
References Suppl p 59
Spray GH. Estimation of red cell folate activity. J ClinPathol 1969; 22: 212.
2 Srinivasan R, Miller BL, Paulus HE. Longtermchrysotherapy in RA. Arthritis Rheum 1979; 22:105-10.
6 Swannell AJ, Coomes EN. Preliminary results ofazathioprine treatment in severe rheumatic disease.Ann Phys Med 1969; 10: 112-20.Swannell AJ, Kersley GD. Preliminary results withazathioprine treatment in rheumatoid arthritis. AnnPhys Med 1969; 10: 163-6.Symington GR, Mackay IR, Lambert RP. Cancer andteratogenesis: infrequent occurrence after medical useof immunosuppressive drugs. Aust NZ J Med 1977; 7:368-72.
210 Talal N, ed. Autoimmunity. London: Academic Press,1980.
211 Talal N, Bunim JJ. The development of malignant lym-phomata in the course of Sj6gren's syndrome. Am JMed 1964; 36: 529-40.
212 Tausch G, Eberl R, Siegmeth W, Sochor H. Long-termtherapy of chronic progressive polyarthritis withazathioprine. Z Rheumaforsch 1970; 29: 162-7 (trans-lation).
212 Teppo L, Hakama M, Hakulinen T, et al. Cancer inFinland 1953-1970: incidence, mortality, prevalence.Acta Path Microbiol Scand 1975; suppl 252.
214 Thorpe P, Hasall JE, York JR. Rheumatoid arthritistreated with chlorambucil. Med J Aust 1976; 2: 197-9.
215 Tjio JH, Whang J. Chromosome preparations of bonemarrow cells without prior in vitro culture or in vivocolchicine administration. Stain Techniques 1962; 37:17.
216 Tolchin SF, Winkelstein A, Rodnan GP, Pan SF, Nan-kin HR. Chromosome abnormalities from cyclophos-phamide therapy in rheumatoid arthritis and progress-ive systemic sclerosis (scleroderma). Arthritis Rheum1974; 17: 375-82.
217 Turk JL. Immunology of infections. In: Dick G, ed.Immunological aspects of infectious diseases. Lancas-ter: MTP Press, 1979: 421-52.
210 Tumbull A, Almeyda J. Idiopathic thrombocytopenicpurpura and Kaposi's sarcoma. Proc R Soc Med 1970;63: 603.
219 Uddin J, Kraus AS, Kelly HG. Survivorship and deathin rheumatoid arthritis. Arthritis Rheum 1970; 13:125-31.
220 Uhl GS, Williams JE, Arnett FC. Intracerebral lym-phomas in a patient with central nervous system lupuson cyclophosphamide. J Rheumatol 1974; 1: 282.
221 Ulrich J, Wuthrich R. Multiple sclerosis: reticulum cellsarcoma of the nervous system in patient treated withimmunosuppressive drugs. Eur Neurol 1974; 12: 65-8.
222 Unger A, Panayi GS, Lessof MH. Carcinoembryonicantigen in rheumatoid arthritis. Lancet 1974; 1: 781.
223 Unger A, Panayi GS, Lessof MH. Carcinoembryonicantigen in rheumatic diseases. Rheumatol Rehabil1975; 14:19.
214 Urowitz MB, Gordon DA, Smythe HA, Pruzanski W,
225 Urowitz MB, Hunter T, Bookman AAM, Gordon DA,Smythe HA, Ogryzlo MA. Azathioprine in rheumatoidarthritis: a double-blind study comparing full dose tohalf dose.J Rheumatol 1974; 1: 274-81.
226 Van Dam G, Lezwijn A, Bos JG. Death rate of patientswith rheumatoid arthritis. In: Atti del Congresso del laLega Internationale Contro il Rheumatismo Roma,Melazioni, Torino. Minerva Medica 1961: 2.
227 Vandenbroucke JP, Kok FJ, Matroos A, Dekker E.Rookgewoonten van Nederlandse huisartsen ver-geleken met die van de bevolking. Ned TijdschrGeneeskd 1981; 1: 4-7.
228 Verhaegen H, DeCock W, DeCree J. The effects ofazathioprine and levamisole on rosette-forming cells ofhealthy subjects and cancer patients. Clin ExpImmunol 1977; 29: 311-5.
230 von Vormittag W. Zytostatische immunodepressiveTherapie, chromosomale Aberrationen und kar-zinogene Wirkung. Wien Klin Wochenschr 1974; 3:69-75.
231 Walder BK, Robertson MR, Jeremy D. Skin cancer andimmunosuppression. Lancet 1971; ii: 1282-3.
232 Wall RL, Clausen KP. Carcinoma of the urinary blad-
der in patients receiving cyclophosphamide. N Engl JMed 1975; 293: 271-3.
233 Webley M, Coomes EN. An assessment of penicil-lamine therapy in rheumatoid arthritis and the influ-ence of previous gold therapy. J Rheumatol 1979; 6:20-4.
234 Wegelius 0, Skrifvars B, Andersson L. Rheumatoidarthritis terminating in plasmocytoma. Acta Med Scand1970; 187: 133-8.
235 Weinstein A. Drug-induced systemic lupuserythematosus. In: Schwartz RS, ed. Progress in clinicalimmunology. Vol 4. New York: Grune and Stratton,1980: 1-22.
236 Wheeler GE. Cyclophosphamide-associated leukemiain Wegener's granulomatosis. Ann Intern Med 1981;94: 361.
237 Williams HJ, Reading JC, Ward JR, O'Brien WM.Comparison of high and low dose cyclophosphamidetherapy in rheumatoid arthritis. Arthritis Rheum 1980;23: 521-7.
238 Wishart J. Reticulosarcoma of vulva complicatingazathioprine-treated dermatomyositis. Arch Dermatol1973; 108: 563.
239 Woodland J, Chaput de Saintonge DM, Evans SJW,Currey HLF. Azathioprine in rheumatoid arthritis:double-blind study of full vs half dosage vs placebo.Ann Rheum Dis 1980; 39: 605-6.
240 Worlledge SM, Brain MC, Cooper AC, Hobbs JR,Dacie JV. Immunosuppressive drugs in the treatmentof autoimmune haemolytic anaemia. Proc R Soc Med1968; 61: 1312.
