Znt. J . Cancer: 29, 397-400 (1982)

FINAL CASE REPORTING FROM THE UGANDAN PROSPECTIVE STUDY OF THE RELATIONSHIP BETWEEN EBV AND BURKITT’S LYMPHOMA A. GESER~, G. D E T H ~ ’ ~ ~ , G . LENOIRI,N.E. DAY’ andE.H. WILLIAMS^^^ International Agency for Research on Cancer, 150 Cours Albert Thomas, 69372 Lyon, France; Kuluva

Hospital, Arua, Uganda (former address); Faculty of Medicine Alexis Carrel, Lyon, France (present address); 45 Northcourt Avenue, Reading, United Kingdom (present address).

A prospective epidemiological study was carried out in the West Nile District of Uganda from 1972 to 1979 in order to investigate the aetiological role of the Epstein- Barr virus (EBV) in Burkitt’s lymphoma (BL). By 1976, fourteen BL cases had been detected among the 42,000 children originally bled in the study area. Testing of sera from BL candidates and neighbourhood controls showed that children who develop BL later have EBVNCA titres several dilutions higher than their age- and sex-matched neighbours. This appearance of a strong EBV activity long before BL development was taken as evidence of a causal role of EBV in B L In order to add to the unique material of pre-bled BL cases, BL detection was continued up to March 1979 when field work became impossible in Ugan- da. Two additional pre-bled BL cases were found during this extension of the study. The serological and virological evaluation of these additional cases showed that the EBVl VCA titres, but not the EA and EBNA titres, were about two dilutions higher in the BL candidates than in the con- trols. Hybridization assays showed that both lymphomas contained EBVIDNA in the tumour cells. These additional resulb thus confirm the findings in the first 14 cases and strengthen the epidemiological evidence for a causal role of the EEV in endemic B L

A prospective study designed to test the hypothesis that the Epstein-Ban virus (EBV) plays a causal role in Burkitt’s lymphoma (BL) was implemented in the West Nile District of Uganda in 1972. From 1972-74, approximately 42,000 sera were collected from the 0- %year old children living in five selected counties of the district, and by the end of 1976 a total of 14 pre- bled BL cases had been found among these children.

Sera from these children were tested for EBV anti- bodies and the results showed that EBVNCA (but not EA nor EBNA) antibody titres were significantly higher in the pre-BL sera than in sera from age- and sex-matched neighbourhood and general population controls as early as 4 years before tumour develop- ment (de The et al., 1978). Testing for antibodies to other herpes-like viruses (HSV, CMV, measles and adenovirus) (de ThC, et al., 1978; Geser et al., 1979) showed that the elevation of antibody titres in the sera from future BL cases was specific for the EBV, and the findings of the prospective study thus support the hypothesis that this virus plays a causal role in BL.

Since this finding - although statistically significant - was based on 14 cases only, the search for pre-bled cases was continued to characterize better the associa- tion. Such a continuation was possible in the West Nile District, since about half the children living in the study area were still at the age of high BL incidence (5-10 years) in 1976, However, by March 1979, when the civil disturbances in Uganda reached the West Nile District, all project activities had to cease and, to- date, there is no prospect of resuming field work in the area. Thus, the pre-bled BL cases included here are

the final ones to be reported from the prospective study.

MATERIAL AND METHODS

The rationale and methods of the prospective study in the West Nile District have been described (Geser and de The, 1972; de ThC etal., 1978). The fundamen- tal issue was to determine the immune status, with respect of EBV, of patients before the tumour de- velops. This question was addressed by comparing the level of EBV antibodies in ‘‘pre” BL sera with levels in children - of the same age and sex - who were bled in the neighbourhood at the same time as the BL can- didate but who did not get BL. For this purpose, four of five control sera were tested with each pre-BL serum. In addition, EBV antibody titres in the BL patients were measured at the time of diagnosis (“post” BL) to evaluate the immunological effect of tumour appearance. The sera were tested under code both at IARC‘S laboratory in Lyon and at Dr Werner Henle’s laboratory in Philadelphia.

RESULTS

Case detection When field work finally had to be abandoned in the

West Nile District in March 1979, a total of 20 addi- tional BL suspects had been detected since December 1976. These 20 cases were diagnosed as possible BL by the medical consultant (Dr E.H. Williams) on the pro- ject in the West Nile District. Pathology specimens from these cases were forwarded to Professor Dennis Wright in Southampton, who confirmed the BL diag- nosis in 12 of them. Among the 20 BL suspects, four had been bled before tumour manifestation. Clinical and demographic data for these four “pre-bled” cases are given in Table I which also gives the results of EBV antibody testing and DNA hybridization assays. Two of the pre-bled cases (Nos. 272 and 285) were confirmed as BL by the pathologist; one (No. 274) was found to be a retinoblastoma and the material ob- tained from the last (No. 282) was inadequate for a final diagnosis to be made. This patient was a seven- year old boy with an eye tumour, clinically compatible with a BL, but no BL cells were found in tumour aspirates by the medical consultant in the West Nile.

One of the two confirmed pre-bled BL cases (No. 285) was a girl with a maxillary tumour which ap- peared when she was six years old. She had been bled twice before getting BL, first in 1972 when she was only 3 months old, in the main serological survey, and next in 1973 when a special re-survey was carried out

Received: November 27, 1981 and in revised form Febru- ary 16, 1982.

4 2 d m m d bl

rn 3

W m % . , v

h

GESER ET AL.

in a small representative sample of all the children who had been bled. The other confirmed BL case (No. 272) was a girl who had been bled in the main survey in 1972, when she was 4 years old, and who developed BL of the liver and jaw when she was 9 years old.

EBV antibody titres Table I gives the results of anti-EBV (VCA, EA-R,

FA-D and EBNA) antibody titres of the sera collected before (“pre”) tumour appearance as well as the sera collected after (“post”) diagnosis. Only the antibody titres obtained at IARC are shown since the results obtained in Dr Henle’s laboratory were similar. The further analysis was confined to the IARC results.

Table I shows that the EBVNCA titres in the two confirmed BL cases were high before the time the tumour appeared and then declined somewhat after BL development, indicating that the high EBVNCA titres which usually characterize BL patients are al- ready present years before tumour appearance. The unconfirmed case (No. 282) had low EBV antibody titres before as well as after tumour manifestation.

The retinoblastoma (No. 274) already had an ele- vated VCA titre (320) in the “pre” serum and at the time the tumour was diagnosed the VCA titre had risen to 2,560 and positive EA antibody titres had developed as shown in Table I. This could indicate reactivation of EBV infection in the retinoblastoma but, in this context, it should be remembered that in the West Nile District, more than 20 % of the children in the age group 5-10 years have EBVNCA titres measuring 320 or more, and that about 7 % of them have EBV VCA titres as high as 2,560 (Geser et af. , 1980). It is thus possible that the high titres observed in the one case of retinoblastoma could be a chance phenomenon.

In Figure 1, the VCA titres of the two new BL cases and their controls are plotted against length of time between “pre”-bleeding and tumour manifestation. For completeness, the results obtained in 1976 on the fourteen pre-bled BL cases, and their controls, are also shown. The two new confirmed BL cases (nos. 285 and 272) have EBVNCA titres which are higher than titres among all their controls, except for one control of No. 272 who had a VCA titre equal to that of the case. Figure 1 further shows that the elevation of the VCA titres occurs already 72 months before BL appearance and as early as 3 months after birth.

A “t” test applied to the 1976 IARC results (de T h B , et al., 1978) showed that the excess of VCA titres in the pre-BL sera over the control sera was statistical- ly significant ( t = 3.00: p = 0.01). When the VCA antibody titres observed in the two new pre-BL sera and their controls are added to the previous results, the significance of the difference in VCA titres is in- creased ( t = 3.63: p < 0.001). Alternatively, using an extension of the Wilcoxon test appropriate for this situation suggested by Cusick (personal communka- tion, 1981), one obtains a normal deviate of 3.57 giv- ing a p value of 0.0002. Both test thus show an in- creased significance level when the new observations are added to the results from 1976. Including the new doubtful case (No. 282) did not alter the significance level materially.

0

CD

C D

EBV AND BURKI??*S LYMPHOMA

I ,

, I

C 3

0

.A I I I 1 I I I , I

72 80 54 51 33 M 18 17 1413 11 8 7 4 t

Time (rnonlhrl beelwean f i rst bloedingand EL occurrence

FIGURE 1 - EBVNCA antibody titres in sera collected from BL cases before tumour manifestation and from controls. Key: 0 BL case, sedage in years or months (m) at time of “pre” bleeding; 0 Control.

The serological results were also analysed using a regression approach for matched case-control studies (Breslow and Day, 1980). If the analysis includes all cases with a histological diagnosis consistent with BL, the relative risk increases multiplicatively by a factor of 5.05 for each dilution by which the VCA titre is above the average in the general population of that age and sex (i.e. for two dilutions the relative risk is 5.052 = 25, and in general for n dilutions the relative risk is 5.05”). If only those cases are considered for which the presence of EBVDNA was demonstrated in the genome, the multiplicative increase in risk is 9.16 for each dilution.

The level of other anti-EBV antibody titres (EA-R, EA-D and EBNA) found in “pre”-sera from the two confirmed BL cases (Table I) evaluated in this study, were also compared to the level of these specificities in the controls but no significant differences were found for any of them. This observation is also in agreement with the results obtained in the previous 14 pre-bled BL cases (de The ef al., 1978) and thus confirms that the elevation of EBV antibody levels before BL mani- festation is limited to the VCA specificity. EBVIDNA in the turnour

The two confirmed BL tumours underwent EBVl DNA hybridization assays both in the laboratory of Dr. G. Klein in Stockholm and of Dr. G. Bornkamm in Freiburg, and were found to contain many genome’s equivalents per BL cell (Table I). The un- confirmed tumour (No. 282) was tested in the Stock- holm laboratory only and did not reveal any presence of EBVDNA.

DISCUSSION

The results of EBV antibody testing in the two addi- tional pre-bled BL cases from the West Nile District confirmed the finding made previously in the first 14 pre-bled cases, viz. that an elevated EBVNCA titre in

a 0-8-year old child increases the risk that the child will develop BL in the future. Both the earlier analysis (de ThC el al., 1978) and the regression analysis given here indicate that a child with an EBVNCA titre two dilu- tions or more above the average for the general child population of the same age and sex has an approxi- mately 30-fold higher risk of getting BL.

The three additional pre-bled tumours (two con- firmed BL and one unconfirmed) included in this re- port were examined by EBVDNA hybridization. EBVDNA was detected in cells from the two con- firmed BL tumours, but not in the third case for which the BL diagnosis was unconfirmed. Failure to find the EBV genome in a tumour does not rule out the diag- nosis of BL since lymphomas clinically and pathologi- cally similar to BL but without demonstrable EBV genome in the tumour tissue occur. It has been found that 93-97 % of the African BL carry the EBV genome (Olweny et of., 1977; de Thk et af., 1978) whereas only about 20% of BL tumours found in the USA have been positive in EBV hybridization test (Gravel1 et af., 1976; Anderson el al., 1976) and it is possible that the entity termed “BL” may in fact include a variety of conditions with different aetiologies in different parts of the world.

As would be expected, the additional cases detected during the present extended period of the prospective study had been bled longer before BL manifestation and at an earlier age than the previous cases. As men- tioned, one of the cases (No. 285) was bled at the age of only 3 months, 6 years before the BL appeared. Although the serum Erom this child did not contain EBV IgM antibodies (when tested at IARC), it is most likely that its high VCA level (1,280) signified active EBV infection since the mother’s serum had a low VCA titre (80). Usually, primary (“active”) EBV infection in Africa does not occur until 5-10 months after birth (Biggar et al., 1978), and it is believed that

400 GESER ET AL.

the high VCA titre in this BL patient is the earliest example of EBV primary infection which has been reported. This finding supports the theory put forward by de ThC (1977) that it may be very early (perinatal) EBV infection which acts as an initiator for BL.

ACKNOWLEDGEMENTS

The expert help of Dr. Werner Henle in EBV anti- body testing, of Drs. G. Klein and G.W. Bornkamm

in the hybridization assays and Dr. D.H. Wright in evaluating the pathology of the tumours included in this study are gratefully acknowledged. We are inde- pted to Ms. M.-F. LavouB for carrying out serological testing and to Ms. C. Bonnardel for technical assis- tance.

This study was supported by contract No. NO1 CP 43296 between the National Cancer Institute (USA) and the International Agency for Research on Cancer.

REFERENCES

ANDERSON, M., KLEIN, G., ZIEGLER, J.L., and HENLE, W., Association of Epstein-Barr Viral Genomes with American Bur- kitt’s Lymphoma. Nature (Lond.), 260, 357-359 (1976). BIGGAR, R.J., HENLE, W., FLEISHER, G., BOECKER, J,, LENNET’E,

GESER, A. FEORINO, P.M., and SOHIER, R., Further Studies of Antibody Levels to Herpes Simplex Virus, Cytomegalo Virus, Measles Virus and Adenovirus Type 5 in Burkitt’s Lymphoma Patients. Med. Microbiol. Zmmunol., 167, 175-180 (1979).

E.T., and HENLE, G., Primary Epstein-Barr Virus Infection in African Infants. I. Decline of Maternal Antibodies and Time of Infection. Inl. J . Cancer, 22, 239-243 (1978). BESLOW, N,E,, and DAY, N,E., Statistical Methods in Cancer Research. IARC Scientific Publications No. 32, Lyon, 1 (1980).

GESER, A., BRUBAKER, G . , and OLWm, G.W.. The Frequency of Epstein-Barr virus infection and Burkitt’s lymphoma at high and low altitudes in East Africa. Rev. Epidkm. et Santk Publ., 28, 307-321 (1980).

DE T&, G., Is Burkitt’s Lymphoma Related to Perinatal Infec- tion by Epstein-Barr Virus?, The Lancer, I, 335-338 (1977). DE T&, G . , GESER, A., DAY, N.E., TURKEI, P.M., WILLIAMS, E.H., BERI, D.P., SMITH, P.G., DEAN, A.G., BORNKAMM, G.W., FEORINO, P., and HENLE, W., Epidemiological Evidence for Causal Relationship between Epstein-Barr Virus and Bur- kitt’s Lymphoma from Ugandan Prospective Study. Nature (Lond.), 274, (5673), 756-761 (1978). GESER, A, , and DE T&, G., Does the Epstein-Barr Virus Play an Aetiological Role in Burkitt’s Lymphoma? In: Oncogenesis and Herpes Virus, IARC Scientific Publications No. 2, Lyon (1972).

GRAVELL, M., LEVINE, P.H., MCINTYRE, R. F., LAND, V. J., and PAGANO, J.S., Epstein-Barr Virus in an American Patient with Burkitt’s Lymphoma. Detection of Viral Genome in Tumor Tissue and Establishment of a Tumor-derived Cell Line (NAB). J . Nat. Cancer Inst., 56, 701-704 (1976).

OLWENY, C.L.M., ATINE, I., KADDU-MUKASA, A, , OWOR, R., ANDERSON-ANVRET, M., KLEIN, G., HENLE, W., and DE T&, G., Epstein-Barr Virus Genome Studies in Burkitt’s Lymphoma and Non-Burkitt’s Lymphomas in Uganda. J. Not. Cancer Inst., 58 (9, 1191-1196 (1977).