786

10. Gleiberman L. Blood pressure and dietary salt in human populations. Ecol Food Nutr1973; 2: 143-56.

11. Williams RR, Hopkins PN. Salt, hypertension, and genetic-environmentalinteractions Prog Clin Biol 1979; 32: 183-94.

12. Mauny R Tableau geographique de L’Ouest African au moyen age: d’apres les sourcesecrites, la tradition et l’aret l’archeologie. Amsterdam Swets and Zeitlinger, 1965.

13. Sundstrom L The trade of Guinea Lund Hakan Ohlssons Boktrycker, 196514 Alexander JA The salt industries of Africa their significance for European prehistory.

In: de Brisay KW, Evans KA, eds Salt the study of an ancient industry. Colchester:University of Essex, 1975: 81-83.

15. Braude F. The perspective of the world: civilisation and capitalism. New York: Harper& Row, 1984

16. Curtin P Economic change in precolonial Africa: Senegambia in the era of the slavetrade. Madison- University of Wisconsin Press, 1975

17. Tessman G. Die Pangwe: Volkerkundliche Monographie eines West-AfrikanischenNegerstammes. Berlin: Verlegt bei Ernst Wasmuth, 1913.

18. Good MC Salt, trade, and disease: aspects of development in Africa’s northern greatlakes region. Int J Afr Hist Studies 1972; 4: 543-86.

19 Meillassoux C. The development of indigenous trade and markets in West AfricaOxford- Oxford University Press, 1971.

20 Lovejoy P. The Borno salt industry Int Afr Hist Studies 1978; 11: 629-6821. Gouletquer PL. Niger, country of salt. In: de Brisay KW, Evans KA, eds. Salt: the

study of an ancient industry. Colchester. University of Essex, 1975 47-51.22. Sutton IB. The Volta river salt trade: the survival of an indigenous industry. J Afr Hist

1981; 22: 43-61 23. Bunge G Text-book of physiological and pathological chemistry. Philadelphia. P

Blakiston’s, 1902.24. Ree GH. Arterial pressures in a West African (Gambian) rural population. J Trop Med

Hyg 1973; 76: 65—70.

25. Beiser M, Collomb H, Ravel JL, Nafzinger J Systemic blood pressure studies amongthe Serer of Senegal. J Chron Dis 1976; 29: 371-79.

26. Akingubee OO. Arterial pressures in rural and urban populations in Nigeria. Br Med J1969: ii: 222-24

27. Grollman A. A conjecture about the prevalence of essential hypertension and its highincidence in the black Texas Rep Biol Med 1978; 36: 25-32

28. Luft FC, Grim CE, Higgens JT Jr. Differences in response to sodium administration innormotensive white and black subjects. J Lab Clin Med 1977, 90: 555-59

29. Alland A Jr. Human biological diversity in Africa. In: Skinner EP (ed) Peoples andcultures of Africa. Garden City, New York- Natural History Press, 1973 59-70

Hospital Practice

RESULTS OF CHEMOTHERAPY FORUNSELECTED PATIENTS WITH ACUTE

MYELOBLASTIC LEUKAEMIA: EFFECT OFEXCLUSIONS ON INTERPRETATION OF

RESULTS

THE TORONTO LEUKEMIA STUDY GROUP*

Summary In a study of 272 consecutive patients withacute myeloblastic leukaemia admitted to 14

general hospitals in the Toronto region the completeremission rate ranged from 43·8% to 85· 3% depending onthe exclusion applied. The median duration of survival for allpatients was 5 months, and for those who achieved completeremission it was 16·8 months. The first 130 patients receiveddifferent treatments from the other 142, with the secondgroup showing an improved remission rate. Differences inexclusion criteria might have affected considerably the

comparison of these two consecutive groups. To aid in thecomparative evaluation of drug regimens the population basefrom which the patients are drawn must be fully described.This would help haematologists explain prognosis to

unselected and unreferred patients.INTRODUCTION

TREATMENT of acute myeloblastic leukaemia (AML) inadults has improved with the introduction of cytosinearabinoside and daunorubicin. I Combinations with

6-thioguanine have been reported to give complete remissionrates of 80% with improved remission duration andsurviva1.2-4 However, remission induction regimens may stillcarry high morbidity rates and mortality associated with

*The Toronto Leukemia Study Group: chairman: Dr M. A. Baker; steeringcommittee: Dr M. A. Baker, Dr B. Garvey, Dr A. Keating, Dr K. Shumak,Dr D. Sutton, Dr R. N. Taub; statistical consultant: Dr W. H. Carter Jr,Medical College of Virginia; pharmacology consultants: Dr R. N. Taub,Dr A. Hindenburg; central pathology review: Dr K. Carstairs,Dr D. Pantalony; clinical trials assistant: Virgilio De Leon. Participatingphysicians: Dr D. Amato, Dr M. A. Baker, Dr S. Berger, Dr K. Butler,Dr H. Chiu, Dr J. Crookston, Dr A Cuthbert, Dr M. Davidson, Dr D. Dotten,Dr H. Farquharson, Dr W. Francombe, Dr B. Garvey, Dr A. Keating,Dr M. King, Dr G. Kutas, Dr W. Kwant, Dr P. Poldre, Dr J. Scott,Dr A. Seidenfeld, Dr K. Shumak, Dr H. Silver, Dr D. Sutton, Dr H. Watt,Dr J. Watt, Dr J. Wilson. Participating institutions: Doctors Hospital,Etobicoke General, Humber Memorial Hospital, Mississauga Hospital,Mount Sinai Hospital, Northwestern Hospital, Oshawa General Hospital, StJoseph’s Health Centre, St Michael’s Hospital, Toronto General Hospital,Toronto Western Hospital, Wellesley Hospital, Women’s College Hospital,York Finch Hospital.

treatment complications.5 Studies have indicated thatadvanced age, high white-blood-cell count and pre-existingdisease reduce the likelihood of a favourable treatment

response.6 The toxicity of the induction regimens and theinformation on prognostic indicators may affect decisionsabout treatment or referral to tertiary-care centres. Patients inhigher risk categories may be segregated for study of neweragents or approaches’-9 and therefore not included in studiesof current remission protocols.To determine the effectiveness of chemotherapy regimens

in adult AML, without exclusions, and study the prognosisand clinical spectrum of the disease we followed the clinicalcourses of 272 consecutive patients with AML. We alsostudied the effect of exclusions on the analysis of the results.

METHODS

The patients in this study were all those admitted consecutively to14 general hospitals in the Toronto region with AML. Informationabout patients was obtained from doctors offices, hospital records,and haematology laboratory reports. AML included the FABcategories Ml to M6. Patients with myelodysplastic syndromeswere included only when morphological criteria were met fortransformation to acute leukaemia.

Patients who were treated received chemotherapy according tostandard remission induction and maintenance protocols as follows:the first 130 patients registered were considered for inductiontherapy with cytosine arabinoside (ARA-C) 100 mg/m2intravenously over 24 h by continuous infusion for 7 days, withdaunorubicin 45 mg/m2 intravenously on days 1, 2, and 3.10 Ifnecessary, a second course was given which consisted of ARA-C for5 days and daunorubicin for 2 days at the same daily dose. Patientswho achieved complete remission after one or two courses of

therapy were given remission maintenance chemotherapy as

follows: ARA-C 100 mg/m2 subcutaneously every 12 h for 10 dosesand 6-thioguanine 100 mg/m2 every 12 h for 10 doses-both drugsgiven for the first 5 days of each month. Patients in this group wereasked to take part in a study of immunotherapy for remissionmaintenance, with BCG or BCG and irradiated leukaemia-cellvaccine in addition to maintenance chemotherapy, and form part ofa series of patients previously reported. II iThe next 142 patients registered were considered for induction

therapy with ARA-C 100 mg/m2 intravenously every 12 h for 7days, 6-thioguanine 100 mg/m2 orally every 12 h for 7 days, anddaunorubicin 60 mg/m2 intravenously on days 5, 6, and 7.3 Thiscourse was repeated if necessary on day 21. Patients who achievedcomplete remission were treated with ARA-C 100 mg/m2 2subcutaneously every 12 h for 5 days once monthly for 4 months,and daunorubicin 45 mg/m2 intravenously on day 5 monthly for thefirst 2 months only. Some patients were entered in a randomisedtrial that compared continuing monthly maintenance

chemotherapy with 6-thioguanine and ARA-C with no furthermaintenance therapy. In that trial, which included many patients

787

who were admitted to hospital after this study, there was nodifference in remission duration or survival in the maintenance andnon-maintenance groups. 20

’

RESULTS

272 patients with AML were registered with the TorontoLeukemia Study Group from 1978 to 1982. The analysis wasconducted in 1985. When all patients are considered in theanalysis, 43 - 807o (119) attained a chemotherapy-inducedcomplete remission (table 1). The median survival for allpatients, with no exclusions, was 5 months from diagnosis. 43patients were not given chemotherapy, in most cases becausethe treatment was judged to be unsuitable by the patient,family, or attending physician. This group includes 15

patients aged 70 and over, 13 patients with precedingmyelodysplastic syndrome (2 over age 70), 7 patients withprevious chemotherapy or coincident disease, and 2 patientswith a white-blood-cell count over 100000/mm3 (table 11).86% of untreated patients had one or more of these features,compared with 4707o for all treated patients. When alluntreated patients were excluded, there were still 119

patients of the remaining 229 who went into completeremission-a rate of 52%. The median survival of the 229

patients in whom treatment was started was 8 months.There were 31 patients who received treatment but did not

complete one full course of chemotherapy, in most casesbecause of severe complications or death. Of these, 6 wereaged 70 or over, 8 had a preceding myelodysplastic disorder(one over age 70), and 4 had previous chemotherapy orcoincident disease. When these patients are also excluded,then 119 of the remaining 198 patients (60-1%) achievedcomplete remission.There were 79 patients aged 70 and over. Of these, 15 were

not treated and 6 received only partial treatment. These 21patients are included in the groups discussed above. 58patients in this age-group received a full course of inductionchemotherapy but only 13 achieved complete remission.TABLE I-EFFECT OF EXCLUSIONS ON COMPLETE REMISSION RATE

AND DURATION OF SURVIVAL

A = patients untreated for any reason; B = patients who received treatment butdid not complete one full course; C = patients aged 70 and over who receivedtreatment; D=patients with a preleukaemic phase who received treatment;E= patients with previous chemotherapy or coincident malignant disease whoreceived treatment.

TABLE II-PROGNOSTIC FEATURES OF UNTREATED AND TREATED

PATIENTS

TABLE III-EFFECT OF EXCLUSIONS ON RESULTS OF THERAPY IN

CONSECUTIVE GROUPS TREATED WITH A 2-DRUG OR 3-DRUGREGIMEN

*Exclusions-see table I and text.DA=daunorubicin, cytosine arabinoside; DAG=daunorubicin, cytosinearabinoside, and 6-thioquanine.

With the use of a proportional hazards model age over 69 wasfound to have a significant negative effect on outcome

(p<0 05). If the 58 treated patients aged 70 and over areexcluded from analysis in addition to those excluded above,then 106 of 140 patients (75 - . 7&deg;70) achieved completeremission. Median survival in this group was 12 - 7 months.AML developed in 34 patients after a myelodysplastic or

preleukaemic phase. Of these 34, 5 were aged 70 or over, 13had received no treatment (including 2 patients over 70), and8 had been only partially treated (including 1 patient over 70).Of the 11 patients in this category who received a full courseof treatment, 2 achieved a complete remission. Analysis witha proportional hazards model confirmed that a precedingmyelodysplastic syndrome has a significant negative effect onprognosis (p<0 05). If these patients are excluded as well,then 104 of the remaining 129 patients (80-60’/o) achievedcomplete remission, with a median duration of survival forthe 129 patients of 13 - 4 months.

15 patients in this series had received previous cytotoxicchemotherapy for malignant disease (Hodgkin’s Disease,multiple myeloma, carcinoma of ovary) or as

immunosuppression (renal transplant). 3 patients hadcoincident malignant disease (carcinoma of cervix, postradiotherapy; carcinoma of lung and colon). 7 of the 18patients received no therapy, 4 had only partial treatment,and 7 were given a full treatment course; but none achievedcomplete remission. The negative prognostic impact ofprevious cytotoxic chemotherapy or coincident malignantdisease was confirmed by means of statistical analysis with aproportional hazards model (p<0’05). If all patients in thiscategory are also excluded then 104 of the remaining 122patients (85 - 3%) achieved complete remission. The medianduration of survival for the 122 patients was 15.7 7 months.The 130 patients treated for the first 3 years of intake

received induction therapy with ARA-C and daunorubicin(group 1), and the next 142 patients received ARA-C,daunorubicin, and 6-thioguanine (group 2). Although theseare consecutive groups, they are drawn from the same popu-lation and are comparable in the distribution of adverseprognostic variables (data not shown). Although the completeremission rate improves from 34 - 607o to 52-1% (p=0’004),the median survival of all patients, the median duration ofremission, and the median survival of patients who achievedremission is not significantly different in the two groups(table III). Exclusion of patients has the same effect for each of

788

the two groups as noted above for the whole series. A

difference in exclusion criteria would potentially have had aprofound effect on the relative remission rates of the twogroups.

DISCUSSION

We believe this patient series to be representative of the fullclinical spectrum of AML, and have shown that if all patientsare taken into account, the clinical course may be lesssuccessful than previously thought. 2-4A large number of patients (43/272) were not given

chemotherapy. In most cases, this was because the attendingphysician judged the patient to be unlikely to withstandchemotherapy. These patients are never included intreatment series and yet a discussion of treatment results and

prognosis with new patients should include this information.31 patients were started on chemotherapy but did notcomplete a single course of treatment, usually because ofinfectious complications leading to early death. Treatmentseries will often exclude such patients as "inevaluable", butthey are better regarded as treatment failures.Many treatment series report an average patient age of 45 to

50 years,’ 2-14 whereas the mean age of all AML patients andthe mean age in our series, is 60.15 Protocols may excludepatients over 50, 60, or 70 years,12,16,17 or these patients maynot be referred to treatment centres. Patients with precedingmyelodysplastic syndromes, and patients in whom AMLdevelops secondary to cytotoxic chemotherapy have beenshown to respond poorly to chemotherapy for leukaemia.6Patients with poor prognostic variables have been segregatedin separate series.7,8,18 It is likely, therefore, that these

patients’ are also generally excluded from recent

chemotherapy series. If we exclude patients aged 70 and over,patients with preceding myelodysplastic syndromes, andpatients with previous chemotherapy, analysis shows a highcomplete remission rate (85-307o) as reported in otherseries.2-4Reports of recent improvements in chemotherapy for

AML may reflect inadvertent exclusions and refinements inleukaemia classification rather than true improvement intreatment results.190ur new treatment regimen led to animprovement in complete remission rate but no significantdifference in overall median survival, when all patients aretaken into account. If only the second series were subject tothe exclusions we have discussed, the apparent improvementcould have been considerable. The remission rate for all

patients in the first series (35%) might then be compared withthe remission rate for patients with various exclusions in thesecond series (60 - 90%).

Results of treatment trials in AML that show an overall

improvement in results should include an analysis of patientswho are untreated or partially treated, and the proportion ofpatients with adverse prognostic indicators.

This work was supported by the National Cancer Institute of Canada andGrant CA31761-04A2 from the National Cancer Institute, USA, and theWilliam J. Matheson Foundation.

Correspondence should be addressed to Dr Michael A. Baker, OncologyClinic, Toronto General Hospital, 657 University Avenue, Toronto, OntarioM5G 1L7, Canada.

REFERENCES

1. Bloomfield CD. Treatment of adult acute nonlymphocytic leukemia-1980 Ann IntMed 1980; 93: 133-34.

2. Rees JKH, Sandler RM, Challener J, Hayhoe FGJ. Treatment of acute myeloidleukaemia with a triple cytotoxic regimen: DAT. Br J Cancer 1977; 36: 770-76.

3. Gale RP, Cline MJ, for the UCLA Acute Leukemia Study Group. High remission-induction rate in acute myeloid leukaemia Lancet 1977; i: 497-99.

4. Gale RP. Progress in acute myelogenous leukaemia. Ann Int Med 1984; 101: 702-035. Cassileth PA, Begg CB, Bennett JM, et al. A randomized study of the efficacy of

consolidation therapy in adult acute nonlymphocytic leukemia. Blood 1984, 63:843-47.

6. Estey EH, Keating MJ, McCredie KB, Bodey GP, Freireich EJ. Causes of initialremission induction failure in acute myelogenous leukemia. Blood 1982; 60: 309

7. Preisler HD, Early AP, Raza A, et al. Therapy of secondary acute nonlymphocyticleukemia with cytarabine. N Engl J Med 1983; 308: 21.

8. Capizzi RL, Poole M, Cooper MR, et al. Treatment of poor risk acute leukemia withsequential high dose ARA-C and asparaginase. Blood 1984; 63: 694-700.

9. Harousseau JL, Castaigne S, Milpied N, Marty M, Degos L Treatment of acute non-lymphoblastic leukaemia in elderly patients. Lancet 1984; ii: 288

10. Yates J, Wallace J, Ellison RR, Holland JF Cytosine arabinoside (NSC-63878) anddaunorubicin (NSC-83142) therapy in acute nonlymphocytic leukemia. CancerChem Rep 1973; 57: 485-88.

11. Baker MA, Taub RN, Carter WH, and the Toronto Leukemia Study GroupImmunotherapy for remission maintenance in acute myeloblastic leukemia CancerImmunol Immunother 1982; 13: 85-88

12. Sauter C, Fopp M, Imbach P, et al Acute myelogenous leukaemia: maintenancechemotherapy after early consolidation treatment does not prolong survival Lancet1984; i: 379-82.

13. Buchner T, Urbanitz D, Fischer J, et al. Long-term remission in acute myelogenousleukaemia Lancet 1984; i: 571.

14. Champlin R, Gale RP, Elashoff R, et al. Prolonged survival in acute myelogenousleukaemia without maintenance chemotherapy. Lancet 1984, i: 894-96

15 Wintrobe MM, Lee GR, Boggs DR, et al, eds. Classification, pathogenesis and etiologyof neoplastic disease of the hematopoietic system. In Clinical hematologyPhiladelphia Lea and Febiger, 1981: 1455-58.

16. Weinstein HJ, Mayer RJ, Rosenthal DS, Camitta BM, Coral FS, Nathan DG, Frei ETreatment of acute myelogenous leukemia in children and adults N Engl J Med1980, 303: 473-78.

17. Cassileth PA, Begg CB, Bennett JM, et al A randomized study of the efficacy ofconsolidation therapy in adult acute nonlymphocytic leukemia. Blood 1984; 63:843-47.

18. Keating MJ, McCredie KB, Benjamin RS, et al. Treatment of patients over 50 years ofage with acute myelogenous leukemia with a combination of rubidazone andcytosine arabinoside, vincristine and prednisone (ROAP) Blood 1981, 58: 584

19. Feinstein AR, Sosin DM, Wells CK. The Will Rogers phenomenon. Stage migrationand new diagnostic techniques as a source of misleading statistics for survival incancer. N Engl J Med 1985; 312: 1604-08

20 The Toronto Leukemia Study Group. Survival in acute myeloblastic leukemia is notprolonged by remission maintenance or early reinduction chemotherapy Blood

1985; 66 (suppl): 210a.

Trial DesignIS A CONTROLLED TRIAL OF LONG-TERM

ORAL ANTICOAGULANTS IN PATIENTS WITHSTROKE AND NON-RHEUMATIC ATRIAL

FIBRILLATION WORTHWHILE?

PETER SANDERCOCK

JOHN BAMFORDCHARLES WARLOWRICHARD PETO

IAN STARKEY

Walton Hospital, Liverpool; Radcliffe Infirmary, Oxford; andNorthern General Hospital, Sheffield

Summary A controlled randomised trial large enoughto assess the value of anticoagulating stroke

patients in atrial fibrillation would be difficult to conduct inthe UK and the results would be applicable to only a smallproportion of stroke patients. It would be more worthwhile toorganise a trial that also assessed the value of other treatmentsthat are simpler and applicable to all stroke patients. A trialthat assessed the value of aspirin and beta-blockers againstcontrol in all stroke patients would not cost much more thanone restricted to comparing anticoagulants against control inpatients with stroke and atrial fibrillation but would provideinformation of more relevance to the management of patientswith stroke in the UK.

INTRODUCTION

MANY stroke patients in atrial fibrillation (AF) are givenanticoagulants, but there is little direct evidence on whetherthis is wise. It would seem that a large randomised trial of