567

tion of IgG-containing complexes in the 3 patients withextrarenal manifestations but no proteinuria or harma-turia, in the 3 patients with diffuse proliferative glomer-ulonephritis and extrarenal manifestations of fever,rash, vasculitis, and arthritis, and in the one patientwith nephritis only. All the patients had small amountsof complexes eluting in the dimer IgG position; this maybe IgG antiglobulin associating with IgG. Apart fromthis, the one patient without extrarenal disease had onlymedium-size IgG-containing complexes (1-1.5 x 106M.w.) and the 3 patients with extrarenal manifestationsbut no nephritis had only very large complexes(2.5-4 x106 M.w.); the 3 patients with both renal andextrarenal disease had both medium and very large com-plexes. The relation between nephritis and medium-sized complexes is significant (P=0.029, exact test).

Discussion

We have shown that IgG-containing immune com-plexes in the sera of S.L.E. patients vary in size. All thosestudied had small complexes, but the 3 with no evidenceof nephritis were the only ones without medium-sizedcomplexes, and conversely the patient with no extra-renal manifestations was the only one without largecomplexes. It is therefore possible that the medium-sizedcomplexes are the ones responsible for nephritis. This isconsistent with the animal work of Germuth,1O whofound that nephrotoxicity is related to size of complex.Doubtless the complexes will vary i’" constituents as

well as size, even in such a well-defined soluble-complexdisease as s.L.E. where one of the antigens (D.N.A.) is

recognised. The test used permits analysis of complexesfor their constituents, so greater precision in these dis-criminations may soon be possible. Despite these varia-tions in complexes, the results of the test, which detectsIgG-containing complexes of all sizes, are related to ac-tivity of nephritis within and between patients. We donot yet know whether tests for complexes, with or with-out sizing, can replace renal biopsy as an indication forstarting plasmapheresis, cytotoxic drugs, or methylpred-nisolone pulse therapy, but this simple test does seemuseful for monitoring treatment. A particularly interest-ing observation was that very large doses of intravenousmethylprednisolone reduced the amount of complexesrapidly when large doses of oral prednisolone had failedto do so. The general view is that in such diseases corti-costeroid drugs owe much of their effect to their anti-inflammatory effect; their suppression of macrophagefunction might be expected to reduce rather than in-crease elimination of complexes. Perhaps at these highdoses they have a considerable immunosuppressiveeffect.

There has been little appraisal of the relation of othertests for complexes (mostly more difficult and less pre-cise) to disease activity, degree of renal involvement, oreffect of treatment. Complement components and D.N.A.-antibody levels are not satisfactory clinical guides sincethey do not correlate with disease activity or progressionof renal disease.5 It is noteworthy that, in our patient 2,plasmapheresis was clinically effective even though thereduction of immune complexes was less than that withmethylprednisolone. These results suggest a need forformal comparison of these two treatments for lupuscrisis.As in many other diseases in which soluble complexes

are found, in s.L.E. the complexes are unlikely to be thesole source of injury, although they are clearly impor-

tant in pathogenesis of the nephritis. Simple quantita-tive analytical tests, such as ours, will permit extensivestudy of these aspects of the disease.

We thank Dr M. W. Turner for immunochemical help, Dr R. N.Maini for providing some serum samples, Dr T. M. Barratt for allow-ing us to study his patients, and the National Fund for Research intoCrippling Diseases for support.

Requests for reprints should be addressed to R.J.L., Department ofImmunology, Institute of Child Health, 30 Guilford St., LondonWC1N 1EH.

REFERENCES

1. Bardana, E. J., Harbeck, R. J., Hoffman, A. A., Pinofsky, B., Carr, R. I.Am. J. Med. 1975, 59, 515.

2. Zubler, R. H., Lange, G., Lambert, P. H., Miescher, P. A. J. Immun. 1976,116, 232.

3. Agnello, V., Koffler, D., Eisenberg, J. W., Winchester, R. J., Kunkel, H. G.J. exp. Med. 1971, 134, 228 s.

4. Levinsky, R. J., Soothill, J. F. Unpublished.5. Cameron, J. S., Lessof, M. H., Ogg, C. S., Williams, B. D., Williams, D. G.

Clin. exp. Immun. 1976, 25, 418.6. Yonemasu, K., Stroud, R. M. J. Immun. 1971, 106, 304.7. Fahey, J. L., McKelvey, E. M. ibid. 1965, 94, 84.8. Pincus, T., Schur, P. H., Rose, J. A., Decker, J. L., Talal, N. New Engl. J.

Med. 1969, 281, 701.9. Lurhuma, A. Z., Cambiaso, C. L., Masson, P. L., Heremans, J. F. Clin. exp.

Immun. 1976, 25, 212.10. Germuth, F. G. J., Senterfit, L. B., Dreesman, G. R. Johns Hopk. med. J.

1972, 130, 335.

PREVENTION OF FATAL POSTOPERATIVEPULMONARY EMBOLISM BY LOW DOSES OF

HEPARIN

Reappraisal of Results of International MulticentreTrial*

Summary The results of a multicentre trial, de-signed to assess the efficacy of low-dose

heparin in preventing fatal postoperative pulmonaryembolism, were published in July, 1975. In view of in-consistencies which have now become apparent in thedata from one of the twenty-eight centres which tookpart in the trial, the results of the trial have been re-ex-amined, excluding the data from this centre. Of 4031 pa-tients remaining after exclusion of these data, 2033 werein the control group and 1998 in the heparin group. 170(4.2%) patients died during the postoperative period, 94in the control group and 76 in the heparin group; 70.2%of deaths in the control group and 65.7% in the heparingroup had necropsy examination. 15 patients in the con-trol group and none in the heparin group were found atnecropsy to have died due to acute massive fatal pul-monary embolism (p<0.001). Exclusion of data fromthe one centre did not alter already published significantdifferences in the total incidence of deep-vein throm-bosis, nor in the frequency of operative and post-operative bleeding complications observed in the controlgroup and the heparin group.

Introduction

THE efficacy of low-dose heparin in preventing fatalpostoperative pulmonary embolism was investigated in aprospective randomised multicentre trial (M.C.T.), theresults of which were published eighteen months ago.’ 1

*Report by the Co-ordinating Centre: Prof. V. V. KAKKAR, thrombosisresearch unit, department of surgery, King’s College Hospital MedicalSchool, London SE5; Mr T. P. CORRIGAN, Mater Misericord&aelig; Hospi-tal, Dublin, Republic of Ireland; Mr D. P. FOSSARD, Leicester RoyalInfirmary, Leicester; Dr I. SUTHERLAND, Medical Research CouncilStatistical Research and Services Unit, London; and Mr J. THIRWELL,computer centre, St Mary’s Hospital Medical School, London W2.

568

The data in a recent article from one (Basle) of thetwenty-eight participating centres suggest that theM.C.T. results should be re-analysed with the Basle dataexcluded, for the following reasons:

(1( There are inconsistencies (table I) between the data fromBasle which were included in the report of the M.c.T.,l and thosereported by the same workers in separate publications,2 3 yeteach of these three sets of data were obtained from substantiallythe same group of patients. It is now claimed that of 232patients included in the control and heparin groups, 23 pa-tients died and necropsies were performed in 22. Our recordsshow that of 223 patients whose proformas were returned bythe Basle centre, only 16 died and necropsy was performed inonly 13.

(2) The most serious discrepancies refer to the differences inthe incidence of fatal and contributory pulmonary emboli(table I). Analysis of the data reported from Basle in 19753shows that of 105 patients allocated to the heparin group, only1 patient died due to massive pulmonary embolism and inanother 3 emboli found at necropsy were considered to be con-

tributory to the patient’s death. The data sent for inclusion inthe M.C.T. report indicated that there were 3 fatal and 1 con-

tributory emboli among 110 patients allocated to receive

heparin. However, it is now claimed2 that 6 out of 94 patientswho received heparin died due to massive pulmonary embo-lism. Furthermore, before the M.c.T. was published, the Baslecentre had agreed that they had no more than 3 fatal pulmon-ary emboli in their patients entered in the M.C.T.4

(3) Before the results of the M.C.T. were published in July,1975, we had had problems with the data submitted by theBasle centre. The random allocation of patients into heparinand control groups was not followed in this centre during thelater part of the study. Of 223 patients’ proformas sent in fromBasle, 123 (596%) had to be excluded for this reason alone,compared with only 32 (0.54%) of 4248 patients submitted bytwenty-seven other centres. One or several assistants duringthe later part of the study did not pay attention to the concor-dance between the randomised envelope number and the pro-forma number, and proformas were incorrectly chosen to

record the data of patients. 4

(4) The other disturbing feature was the high incidence ofpulmonary embolism reported by this centre; 7 (31%) of 223patients who died were found to have emboli at necropsy. Bycontrast, only 17 (0-4%) out of 4248 patients operated on intwenty-seven other centres developed this complication. Fur-thermore, the only 3 patients in the entire study who receivedheparin and whose deaths were attributed to massive pulmon-

TABLE I-SUMMARY OF DATA FROM BASLE AS SUBMITTED TO

M.C.T. COORDINATING CENTRE AND AS REPORTED IN TWO

SEPARATE PUBLICATIONS2 3

Figures in parentheses refer to patients found at necropsy to have dieddue to contributory emboli.

TABLE II-INCIDENCE OF FATAL P.E. IN 27 CENTRES INCLUDED INTHE M.C.T.

ary embolism came from this single centre. However, 89 pa-tients belonging to this centre, in whom the agreed protocolhad been correctly followed, were included in the final analysisof the M.c.T., and 3 patients with pulmonary embolism (p.E.)were in this group. This was done to ensure that the analysisof the M.c.T. results could not be criticised as being biased.We agree with Gruber et al.1 that the definition of

fatal pulmonary embolism is open to debate. However,pathological criteria for a fatal pulmonary embolismwere clearly defined and agreed by each participatingcentre before the trial began. We are surprised that,eighteen months after the publication of results, Gruberet al., having accepted the definition, now criticise thetrial for this reason.

Patients and Methods

The protocol followed in the multicentre heparin trial hasbeen reported in detail previously.1 Patients were allocated toa control or heparin treated group on the basis of a computer-ised series of random allocations. Each patient allocation wassealed in a numbered envelope. Patients were considered tohave been admitted to the trial when the envelope was openedand they were then followed up until discharge or death in hos-pital. Proformas for the tabulation of patient data and sealedallocation envelopes were issued to the centres in batches of 50.The randomisation was such that any group of 50 proformascould be discarded without interfering with the randomisationof the trial as a whole. A record of the allocation sequence wasretained at the control centre. When the trial was completed,the allocation of the patients to the control or treatment groupwas checked to see that the randomisation was correctly fol-lowed.

Results

4471 patients were admitted to the trial, 4248 fromtwenty-seven centres, and 223 from the Basle centre.

Analysis of Data Submitted for 27 CentresOf 4248 patients, submitted by twenty-seven centres,

217 (5.1%) were excluded for various reasons, leaving4031 patients in whom the agreed protocol had beencorrectly followed-2033 in the control group and 1998in the heparin group.

Comparison of GroupsThe two groups were well matched for age, sex,

weight, blood-groups, risk factors possibly predisposingto venous thromboembolism, and types of operationsperformed. The details of distribution of these factorshave already been published,’ and exclusion of datafrom the Basle centre did not alter the results materially.

Total Mortality-rateOf 4031 patients included in the trial, 170 (4.2%)

died-94 in the control group and 76 in the heparingroup. 66 (70.2%) in the control group and 50 (65.7%)in the heparin group had a necropsy examination.

Incidence of Pulmonary EmbolismOn the basis of criteria already defined,’ 15 (0.7%) of

66 patients in the control group died from acute massivepulmonary embolism. None of the 50 patients in theheparin group examined at necropsy died due to this

complication (table rt). The difference in the frequencyof fatal pulmonary embolism between the two groups ishighly significant (Fisher’s exact test: two-tail,p<0.001). These results differ from our previous report’since all the fatal pulmonary emboli in the heparingroup were reported from the Basle centre.

569

The distribution of the fatal pulmonary emboli in thecontrols was pulmonary trunk (5), main pulmonaryartery (9), lobar artery (9), and segmental artery (6).

In addition, in 5 patients from the control group and2 from the heparin group, emboli found at necropsywere considered either contributory to death or an inci-dental finding, since death in these patients was attri-buted to other causes. Taking all pulmonary embolitogether, 20 in the control group and 2 in the heparingroup, the findings are again statistically significant.

lncidenee of D. V. T.

D.V.T. was diagnosed clinically by the radioactive-

fibrinogen-uptake test and at necropsy. When the 89 pa-tients from the Basle centre, initially included in theM.C.T. analysis, were withdrawn, the significant differ-ences observed between the control and heparin groupsremained.

Treatment of D. V. T. and Pulmonary EmbolismExclusion of the Basle data did not change the highly

significant differences which were observed in thenumber of patients requiring treatment for D.v.T. and/orpulmonary embolism in the two groups.

Operative and Postoperative BleedingExclusion of the Basle data did not influence the pre-

viously reported results’ relating to the incidence ofoperative and postoperative bleeding.

Conclusion

The evidence that low doses of heparin prevent D.v.T.in most postoperative patients is irrefutable. The mainunresolved question was whether this form of prophy-laxis also reduces deaths from pulmonary embolism.This question was answered by the M.c.T.’, and the con-clusion of that trial is not affected by our reappraisal ofthe results: subcutaneous heparin administration beforeand after surgery significantly reduced the incidence offatal pulmonary embolism, detected at necropsy. Therecent article by Gruber et al. casts doubt on the con-clusion of this trial. However, as summarised by Sherry5"... one is forced to the conclusion that there is a veryhigh probability that the differences in the primary end-point (fatal P.E.) between the two groups is real, andthat the study should be used to influence the practicehabits of the profession for preventing fatal pulmonaryembolism following abdominal surgery".

Requests for reprints should be addressed to V.V.K.

REFERENCES

1. International Multicentre Trial Lancet, 1975, ii, 452. Gruber, U. F., Duckert, F., Fridrich, R., Torhorst, J., Rem, J. ibid. 1977,

i, 207.3. Rem, J., Duckert, F., Fridrich, R., Gruber, U. F. Schweiz. med. Wschr.

1975, 105, 827.4. Communication; Rem, J. A. Inaugural dissertation, Faculty of Medicine,

University of Basle, 1975, to the Co-ordinating Centre from Basle. Letterreceived on June 30, 1975, with comments before the publication of theM.C.T. results.

5. Sherry, S. Prophylactic Therapy of Deep Vein Thrombosis and PulmonaryEmbolism. D.H.E.W. publications no. (N.I.H.) 76-866, 1975, p. 229.

Preliminary Communications

ISOLATION AND PARTIAL CHARACTERISATIONOF

A NEW VIRUS CAUSING ACUTEH&AElig;MORRHAGIC FEVER IN ZAIRE

K. M. JOHNSONJ. V. LANGE

P. A. WEBBF. A. MURPHY

Virology Division, Center for Disease Control,Atlanta, Georgia 30333, U.S.A.

AN outbreak of hsemorrhagic fever with an excep-tionally high mortality-rate occurred in southern Sudanand northern Zaire with peak case-rates in September,1976. A W.H.O. International Commission operated inSudan and Zaire from October onward.’ 2 Blood and tis-sue specimens from persons with haemorrhagic diseasewere sent to laboratories in Belgium and England, andfindings from these laboratories. appear in the accom-panying reports.3 4 While these specimens were beingstudied, Mr E. T. W. Bowen (Microbiological ResearchEstablishment, Porton Down) sent an aliquot of anacute blood specimen from a patient in Zaire (no. 718,patient M.E.) to the Center for Disease Control,Atlanta, for additional study.

This specimen, and all subsequent acute specimens,were inoculated into Vero (African green monkey) cells.Three days later a distinct cytopathic change (focalrounding and refractility) was evident, and an aliquot ofsupernatant fluid was removed for negative contrast

electron microscopy.

ELECTRON MICROSCOPY OF CELL CULTURES S

Carbon-coated grids were sequentially floated on droplets of

the cell-culture fluid and then on 2% sodium silicotungstatepH 7. Large numbers of filamentous virus particles were seen(fig. 1). They were approximately 100 nm in diameter and var-ied in length from 300 nm to more than 1500 nm. Many hadterminal blebs. Particles had regular surface projectionsapproximately 10 nm long, and when stained they were seento have internal cross-striations indicative of a helical corestructure (fig. 2). In all details, these particles were indis-tinguishable from Marburg virus particles studied in 1967 (iso-lates from Germany) and 1975 (isolate from South Africa).5-’Two characteristics were more prominent in the 1976 Zaireisolate: there was more branching of the filamentous particles(fig. 1); and more evidence of envelope continuation beyondthe ends of the more rigid internal structure (fig. 1, arrow).

Vero cells infected with the same isolate from Zaire were ex-amined also by thin-section electron microscopy. Filamentousvirus particles were found budding from the plasma membraneof cells (fig. 3), and many of the cells contained inclusionbodies. These intracytoplasmic inclusions were complex anddistinct, and consisted of a finely fibrillar, or granular groundsubstance which condensed into tubular structures. The latterhave been considered to be the internal helical structure ofmature virus particles. These tubules were sectioned randomly,some in cross-section, some linearly. The virus particles in

TABLE !&mdash;COMPARISON OF RECIPROCAL I.F.A. TITRES OF

MARBURG (’67, ’75) AND MARBURG-LIKE (’76) VIRUS DISEASESERA