The risk of developing a non-Hodgkin lymphoma (NHL) in AIDS patients is increased some 60–300-fold, with an even higher risk for high-grade (immunoblastic) and Burkitt's lymphomas (BL). The association between AIDS and NHL is stronger in men, in white (compared with black) patients, and in those of higher socioeconomic status [1–3]. In Africa, there are few epidemiological data on the effect of the HIV epidemic on the frequency of NHL. To date, only two case–control studies have estimated the risk of NHL (all histological subtypes) in HIV-positive compared with HIV-negative individuals [4,5]. In both studies, the relative risk in HIV-positive subjects was considerably less than that observed in Europe and the USA: 12.6 [95% confidence interval (CI) 2.2–54.4] in Rwanda and 4.8 (95% CI 1.5–14.8) in South Africa.
Endemic BL of childhood was common in Equatorial Africa long before the AIDS epidemic, and in some areas comprised more than 90% of NHL in children; the relationship with Epstein–Barr virus (EBV) and malaria is well documented . Hospital and autopsy series do not provide any evidence for an increased frequency of BL cases since the onset of the AIDS epidemic [7–9]. In Uganda, which was one of the first countries in Africa to be affected by the AIDS epidemic in the beginning of the 1980s, registry data show an increase in the incidence of NHL (all types) in adults only since 1995, whereas for BL, there has been a threefold increase in incidence in children (0–14 years) between the 1960s and 1995–1997 .
The objective of the present study, part of a wider investigation of the relationship between HIV infection and cancer in Uganda [11,12], was to investigate the histology, immunology and epidemiology of NHL, with particular emphasis on the role of HIV and EBV infections, and to compare the results with patterns observed in the west.
Materials and methods
Recruitment of subjects
Between August 1994 and April 1998, adults (aged over 15 years) and children with a provisional diagnosis of cancer were systematically recruited and interviewed in the four main hospitals of Kampala, Uganda. Children with benign, non-infectious, surgical or orthopaedic conditions were also recruited from appropriate wards and clinics. All subjects (including the parent or guardian of children) were interviewed using standard questionnaires. These included questions on standard demographic and social variables, and on medical and family history; for adults, information was also sought on occupation, and sexual habits (including reproductive/fertility-associated indicators in women). A sample of blood was taken, and HIV status was determined using the Cambridge Bioscience Recombigen enzyme-linked immunosorbent assay (Cambridge, MA, USA). When blood could not be obtained, a small specimen of saliva was tested for anti-HIV antibodies using the GACELISA test (Wellcozyme, Glaxo-Wellcome Inc., USA). Initially, the confirmation of positive enzyme-linked immunosorbent assay tests was performed with Western blot, but because quality control assays using blinded standards from the United States Centres for Disease Control and Prevention (Atlanta, GA, USA) revealed test sensitivity and specificity of 99%, this procedure was abandoned. Informed consent was obtained for participation in the study (interview and haematological tests) and, separately, for HIV testing.
For cases of lymphomas, paraffin-embedded biopsies were sent to Paris, where conventional histology was performed on material stained with haematoxylin-eosin and Giemsa's methods and independently reviewed by two pathologists (M.R., A.M.). Morphology was classified according to the revised European-American lymphoid classification . Immunophenotypic studies were performed on paraffin-embedded tissue sections using the streptavidine–biotin–peroxidase technique (LSAB kit; DAKO, Trappes, France), polyclonal antibodies directed against CD3 (DAKO) and monoclonal antibodies directed against CD20 (L26) and CD30 (BerH2; DAKO). EBV DNA was detected using in-situ hybridization on paraffin sections with fluorescein isothiocyanate-labelled oligonucleotide complementary to small nuclear EBV-encoded RNA (EBER1; DAKO). The expression of latency gene latent membrane protein 1 was investigated by immunohistochemistry using monoclonal antibody CS1-4 (DAKO). Cases of lymphoma diagnosed without histology (on the basis of clinical signs and symptoms) and cases not confirmed by the review of biopsies in Paris, were excluded.
For children, 334 lymphoma cases were initially recruited, 151 (45%) had a biopsy, and the remaining cases were diagnosed clinically. Fig. 1 shows the process of selection of cases for the analysis. Ninety-two of these biopsies (61%) were sent to Paris for phenotyping and the detection of EBV. Eight (9%) were not confirmed as lymphomas. Phenotyping was not possible in 22 cases (24%), mainly because of insufficient fixation of the samples. For the 59 children whose biopsies were not sent to Paris, 13 (22%) were excluded because of an uncertain diagnosis of lymphoma based on the records. Finally, we ended up with 62 ‘validated’ cases (61 NHL and one Hodgkin's disease; HD) with full phenotyping and detection of EBV, and 68 cases with a histology in Uganda only (61 NHL and seven HD). HD cases were not retained in the analysis.
In adults, 112 cases of lymphomas were initially recruited, 84 (75%) had a biopsy, and the remaining cases were diagnosed on clinical suspicion alone (Fig. 2). Sixty-one of these 84 biopsies (73%) were sent to Paris for phenotyping and the detection of EBV. For nine biopsies (15%), the diagnosis of lymphoma was not confirmed and for several others, the subtype of NHL was revised. Phenotyping was not possible for 16 biopsies. For the 23 cases whose biopsies were not sent to Paris, 10 were excluded because of uncertainty of the diagnosis of lymphoma based on the records. Finally, there were 36 ‘validated’ cases (31 NHL and five HD), with full phenotyping and EBV detection, and 29 additional cases (19 NHL and 10 HD) with histological diagnosis only, for the most part in Uganda. The HD cases were not included in the analysis.
Control subjects for adult lymphoma cases were selected from all other cancer patients without lymphomas, excluding those with cancers known or suspected to be related to HIV infection: cervix cancer, Kaposi's sarcoma, penile cancer and squamous cell carcinoma of the eye. Leukaemias were also excluded because of diagnostic overlap with lymphoma. Only cancer controls confirmed histologically were included. The controls for childhood lymphoma cases had a variety of non-malignant (mainly orthopaedic and surgical) conditions; children admitted with infectious diseases (including malaria and diarrhoea) were excluded. For each lymphoma case, controls were selected at random from all eligible individuals of the same age group (± 2 years) and sex. A maximum of five controls per case were selected, but there were often fewer than this (3.7 controls per case for adults, 3.6 for children).
Odds ratios (OR) were calculated maintaining the matching by age and sex. After examining the risk for individual factors, conditional logistic regression  was used to adjust the main effects for potential confounders. Two groups of cases were considered, ‘validated’ cases, which had had phenotyping performed, and a larger group including cases without phenotyping, but a histological diagnosis performed in Uganda.
Table 1 shows the median and mean age, sex distribution, as well as HIV prevalence and EBV positivity in the 61 ‘validated’ cases of NHL in children, by histological subtype. The vast majority (90%) were BL followed by lymphoblastic (two) and diffuse large B cell (DLBC) lymphomas (two), and most cases were boys. The mean age was 7 years for BL and 6 years for other subtypes (P = 0.30). The detection of EBV was performed for 51 of the 56 BL cases and all were positive. EBV was not detected in the five cases of other lymphoma subtypes. The site was recorded for 23 BL cases: 15 (65%) were located in the face, seven (30%) in the abdomen and only one (4%) in the lymph nodes. HIV prevalence was 4.9% (3/58) in cases, compared with 5.0% (11/194) in controls, corresponding to an OR of 0.8 (95% CI 0.2–3.4) overall, and 1.0 (0.3–3.9) for BL. When the 61 cases with an unverified histological diagnosis were added, the results were unchanged. OR were rather higher for cases living in the urban area of Kampala than in those from elsewhere, but the differences were non-significant (χ2 = 2.0;P = 0.36).
BL and DLBC lymphomas represented the majority (71%) of the 31 validated cases of NHL in adults (Table 1). Other subtypes were infrequent: lymphocytic (four), peripheral T cell (two), grade 1 follicular (one) and T lymphoblastic (two) lymphomas. Men were predominant, and subjects with BL were younger (mean age 31.3 years) than those with other types (45.3 years, P = 0.07). EBV was detected in three-quarters of BL cases and approximately 10% of the other subtypes. Among these ‘validated’ cases, HIV prevalence was 27% (6/22) (Table 1), not significantly different from that in the control subjects (OR 2.1, 95% CI 0.3–6.7). For all 50 cases (with and without validated histology) 28.3% were specified as being in extranodal sites, although none was located in the central nervous system.
Table 2 compares all 50 cases of NHL (with and without validated histology) with their matched controls, with respect to various risk factors. An HIV result was available for 38 cases; 13 (34%) were positive. HIV positivity was rather higher (7/16; 44%) among the NHL cases with histological diagnosis only, than among the ‘validated’ cases (6/22; 27%, Table 1), so that the risk of NHL in HIV-positive subjects was approximately double that of HIV-negative individuals (OR 2.2, 95% CI 0.9–5.1). The mean CD4 cell count at diagnosis for HIV-positive NHL cases was significantly lower (P = 0.04), 291/μl (median 180/μl), compared with that for HIV-negative cases, 751/μl (median 795/μl).
There were no significant differences between cases and controls in terms of their current marital status, but rather more of the cases were residents of the urban area of Kampala than were the controls (Table 2). Higher levels of education were associated with an increased risk of NHL, as were occupations of higher social status (service workers and professionals). Cases were more likely to travel away from their home, and to have a history of sexually transmitted diseases; however, there was no difference in the reported number of sexual partners (median five in both groups), in the frequency of condom use, nor (in women) in parity (data not shown).
In a model incorporating the seven variables that were statistically significant, or almost so, in univariate analysis, the same associations were observed, although the OR for HIV positivity (OR = 1.7, 95% CI 0.6–4.7) was rather lower, and residence (urban/rural) was not associated with risk (OR = 1.0) (Table 2).
In Africa, epidemiological data on the effect of the HIV epidemic on the risk of cancer are sparse. The present study is part of a larger project examining the potential role of HIV on the risk of different cancers in the population of Kampala .
The lymphoma cases included in the analysis had all been histologically confirmed. Although lymphomas are not infrequently diagnosed clinically (without histology) in Africa, this is at best an uncertain procedure, and the more so when HIV infection is prevalent, because lymphadenopathy is a common feature of AIDS and the AIDS-related complex [15,16]. In contemporary Africa, a high percentage of individuals undergoing lymph node biopsy are found to be HIV positive, although very few have lymphomas . The diagnosis of lymphomas, and their differentiation from polyclonal lymphoid proliferation, is not easy even with light microscopy, and it is necessary to study a range of immunophenotypic markers to be certain of the diagnosis, determine the nature of the cells, and subtype the lymphoma. Priority in the analysis was given to cases for which the histology had been independently validated, their phenotype determined, and the presence of EBV evaluated .
It is possible that in selecting only lymphoma cases that had been biopsied, there may have been an under-representation of extranodal lymphomas (compared with all incident cases). As AIDS-related lymphomas involve extranodal sites somewhat more often than non-AIDS lymphomas [2,19], such a selection bias would give rise to an underestimation of HIV positivity (and the associated risk) in all lymphoma patients. Nevertheless, the proportion of lymphomas at extranodal sites among the adult cases in our study (28.3%) is similar to that noted in the USA  and in the Kampala Cancer Registry for the period of the study (1993–1998: 30.8%). It is, however, possible that some deep-seated lymphomas may not be detected at all in hospitals where sophisticated diagnostic tools such as computed tomography scans and nuclear magnetic resonance imaging are not available. No intracranial lymphomas were included, for example, whereas these comprise 15% of AIDS-related lymphomas and 1.4% of non-AIDS lymphomas in the USA .
The possibility of bias in the selection of the controls influencing the results must also be considered. The age and sex distribution of cases and controls was balanced by matching, and the use of several controls per case (between one and five depending on case age) improved the statistical power of our analysis. The use of ‘other cancers’ as controls for adults had advantages in terms of simplicity and practicality; subjects were similar in their likelihood of seeking medical attention and recall and interviewer bias was minimized [20,21]. The accuracy of the estimate of relative risk depends upon how well the control group represents the population from which the cases were drawn. The mixture of different cancers in the control group should ensure that socioeconomic status was similar to that of the population from which the cases were drawn, and the fact that all subjects had been biopsied implied that cases and controls were receiving equivalent levels of medical attention. With respect to HIV prevalence, this should have been achieved by the exclusion of diseases possibly related to HIV infection, including various cancers [22–24], and for the childhood controls, infectious diseases. The prevalence of HIV infection in our controls (approximately 15%) was similar to that of the general population in Uganda . Finally, it should be noted that HIV status and results for other variables were missing for some 18–25% of cases and controls; this may have biased the risk estimates in Table 2 if the composition of the missing category, with respect to the variables of interest, was very different between cases and controls.
In the USA and Europe, NHL are the most common malignancy in paediatric AIDS patients [3,26] and approximately a third are BL . Within the morphological spectrum of BL, there are three clinical subtypes: endemic (eBL), sporadic (sBL) and AIDS-related BL (AIDS-BL). eBL occurs almost exclusively in Africa, has a peak of incidence at 7 years of age, a preponderance of males over females and is associated with EBV in virtually 100% of cases. The predominant sites of involvement are the jaws and abdomen [28,29]. In developed countries, sBL comprises approximately 20–30% of NHL in children and is rare in adults. It frequently presents as lower abdominal masses [28,30]. AIDS-BL occurs with a peak at 10–19 years of age  and frequently involves lymph nodes and the bone marrow . The EBV genome is present in approximately 30% of cases, similar to the proportion in sBL cases [32,33]. Molecular analyses of translocation breakpoints in the chromosomes of tumour cells suggest that the pattern in most cases of AIDS-BL resembles that observed in sBL in Europe and America, and is quite different from that in endemic, EBV-related, African BL . The precise role of EBV in the pathogenesis of these tumours is not yet completely understood .
In our study, almost all BL in children was of the endemic type and was not associated with HIV. The universal association with EBV, median age of 7 years, predominance of males, and localization in facial and abdominal sites are characteristic of eBL. The lack of association between HIV and BL is consistent with previous studies in Africa. In Uganda, Mbidde et al.  reported the absence of HIV infection among 50 children with BL studied early in the AIDS epidemic. All 17 cases of childhood BL in a series from the Nairobi Hospital, diagnosed in 1995–1996, were HIV negative . The autopsy series of Lucas et al.  in Côte d'Ivoire found no NHL in 78 HIV-positive children. In Zambia, there was a decrease in the number of histological diagnoses of BL in the main teaching hospital between 1980–1982 (pre-HIV epidemic) and 1990–1992 (post-HIV epidemic) . However, the long time series from the Cancer Registry in Kampala indicates a threefold increase in the incidence of BL in children between the 1960s and 1995–1997 . The results of our study do not support the hypothesis that this increase could be due to the HIV epidemic, which began in the 1980s. Other factors such as malaria infection, which is endemic in this country, may be responsible, but we have little information about the evolution of malaria endemicity in recent years. One of the explanations for the absence of association between HIV and BL is the poor survival of children infected perinatally with HIV; only 34% of HIV-infected children in Kampala survive to the age of 3 years .
BL also occurred in young adults, mainly males, with predominantly lymph node involvement (9/11 cases). These are characteristics of BL in AIDS cases in the west; however, only three out of seven cases tested for HIV were positive, whereas threequarters of the tumours contained EBV, a higher proportion than in Europe and the USA . DLBC lymphomas represented the other frequent type of NHL in our series; 80% were not associated with HIV and 90% were not associated with EBV.
Because it was not possible to study each histological subtype separately (small numbers in each category), all NHL were considered together. The risk of NHL in relation to HIV was modest (approximately 2), which is much lower than in the west. This finding is consistent with the two previous case–control studies in Africa, which also found a rather small excess risk [4,5].
NHL is a relatively late manifestation of HIV infection, compared with Kaposi's sarcoma and some opportunistic infections [33,37]. Although it seems that the period between the onset of HIV seropositivity and clinical AIDS may be shorter in Africa than in western countries , it is still relatively long; in rural Uganda only 22% of incident HIV had progressed to AIDS within 5 years . However, the degree of immune dysfunction at AIDS diagnosis, as measured by CD4 cell counts, was less in Africa than in industrial countries, and median survival times much shorter . As the risk of NHL in AIDS and other immunodeficiency states is related to the degree of immune dysregulation [41–43], it could be that the apparently low risk of lymphoma in HIV-positive individuals in Uganda is the result of competing mortality, particularly from infectious diseases, in AIDS patients with relatively low levels of immunosuppression. The CD4 cell count at diagnosis in our HIV-positive lymphomas was higher than is generally observed in Europe and North America , and Lucas et al.  found that NHL was present undiagnosed at autopsy in 2.8% of HIV-positive individuals (4% of patients with AIDS). This is not very different from the cumulative probability of developing a lymphoma observed in cohorts of AIDS patients; 4.3% in the study of AIDS patients in the USA, for example .
In Uganda, infection with HIV is linked to social status and mobility [44,45]; however, the association between NHL and high socioeconomic status (measured in our study by the level of education, occupation and travel) was independent of HIV status. In general, there is no consistent association between socioeconomic status and the risk of NHL . Although we are not aware of any previous study in Africa, in developed countries high socioeconomic status has been found to be positively associated with AIDS-related NHL [1,2]. The relatively high proportion of BL cases among adult NHL, and the high percentage that are EBV positive may be relevant to our observations; although numbers are small, it does appear that these cases were of higher socioeconomic status than the other NHL. For HD, the clear socioeconomic gradient in risk has been explained in terms of age at infection with EBV . However, at least in western countries, patterns of EBV infection (as reflected by antibody titres) appear to be quite different in HD and NHL .
Childhood lymphomas were predominantly eBL, the risk of which was not modified by HIV infection. In adults, BL and diffuse B cell lymphomas comprised the majority of cases, but the risk in patients with HIV infection was much lower in Uganda than in western countries. This is possibly due to the poor survival of HIV-positive individuals. Future studies will require careful attention to the subtyping of lymphomas, to investigate the possible differences between them, and to take account of other co-factors such as infectious agents, including malaria.
The authors would like to thank P. Okong (Nsambya Hospital), R. Moser (Rubaga Hospital), C. Koheri (Mengo Hospital) and F. Miiro, I. Kakande, R. Mugerwa, E. Ddumba and the staff of the Immune Suppressive Syndrome Clinic (Mulago Hospital), for access to their patients. The following staff in Uganda contributed to the study: R. Biansi, N. Byabazaire, M. Kalinaki, E. Katabira, S. Mbulataiye, the late J. Mugerwa, S. Nambooze, C. Rwatooro, V. Sembajwe and B. Tushimiere.
1. Beral V, Peterman T, Berkelman R, Jaffe H. AIDS-associated non-Hodgkin lymphoma.
Lancet 1991, 337: 805 –809.
2. Biggar RJ, Rabkin CS. The epidemiology of acquired immunodeficiency syndrome-related lymphomas.
Curr Opin Oncol 1992, 4: 883 –893.
3. Franceschi S, del Maso L, La Vecchia C. Advances in the epidemiology of HIV-associated non-Hodgkin's lymphoma and other lymphoid neoplasms.
Int J Cancer 1999, 83: 481 –485.
4. Newton R, Grulich A, Beral V, Sindikubwado B, Ngilimana P-J, Nganyira A, DM Parkin. Cancer and HIV infection in Rwanda.
Lancet 1995, 345: 1378 –1379.
5. Sitas F, Bezwoda WR, Levin V. et al
. Association between human deficiency virus type 1 infection and cancer in the black population of Johannesburg and Soweto, South Africa.
Br J Cancer 1997, 75: 1704 –1707.
6. IARC Monographs on the Evaluation of Carcinogenic Risk to Humans. Epstein–Barr virus and Kaposi's sarcoma herpesvirus/human herpesvirus 8
, Vol. 70. Lyon, France: IARC; 1997.
7. Chintu C, Athala UH, Patil PS. Childhood cancers in Zambia before and after the AIDS epidemic.
Arch Dis Child 1995, 73: 100 –105.
8. Lucas SB, Diomande M, Hounnou A. et al
. HIV-associated lymphoma in Africa: an autopsy study in Côte d'Ivoire.
Int J Cancer 1994, 59: 20 –24.
9. Mbidde EK, Banura C, Kazura J, Desmond-Hellmann SD, Kizito A, Hellmann N. NHL and HIV infection in Uganda. 5th International Conference in Africa on AIDS
10. Parkin DM, Wabinga H, Nambooze S, Wabwire-Mangen F. AIDS-related cancers in Africa: maturation of the epidemic in Uganda.
AIDS 1999, 13: 2563 –2570.
11. Ziegler JL, Newton R, Katongole-Mbidde E. et al
., for the Uganda Kaposi Sarcoma Study Group. Risk factors for Kaposi's sarcoma in HIV positive subjects in Uganda.
AIDS 1997, 11: 1619 –1626.
12. Newton R, Ziegler J, Beral V. et al
., and the Uganda Kaposi's Sarcoma Study Group. HIV and Cancer in adults and children in Kampala, Uganda [Abstract].
J Acquir Immune Defic Syndr Hum Retrovirol 1999, 21: A10. A10.
13. Harris NL, Jaffe ES, Stein H. et al
. A revised European-American classification of lymphoid neoplasms: a proposal from the International Lymphoma Study Group.
Blood 1994, 84: 1361 –1392.
14. StataCorp. Stata Statistical Software: Release 5.0.
College Station, TX: Stata Corporation; 1997.
15. Ioachim HL, Cronin W, Roy M, Maya M. Persistent lymphadenopathies in people at high risk for HIV infection.
:Clinicopathologic correlations and long-term follow-up in 79 cases.
Am J Clin Pathol 1990, 93: 208 –218.
16. Baroni CD, Uccini S. The lymphadenopathy of HIV infection.
Am J Clin Pathol 1993, 99: 397 –401.
17. Bem C, Patil PS, Bharucha H, Namaambo K, Luo N. Importance of human immunodeficiency virus-associated lymphadenopathy and tuberculous lymphadenitis in patients undergoing lymph node biopsy in Zambia.
Br J Surg 1996, 83: 75 –78.
18. Martin A, Garcia-Giannoli H, Parkin DM. et al
. Malignant lymphomas in Uganda: histological distribution and prevalence of HIV infection [Abstract].
Ann Oncol 1999, 10 (Suppl. 3) : 42. 42.
19. Coté TR, Biggar RJ, Rosenberg PS. Non-Hodgkin's lymphoma among people with AIDS: incidence, presentation and public health burden.
:AIDS/Cancer Study Group.
Int J Cancer 1977, 73: 645 –650.
20. Linet MS, Brookmeyer R. Use of cancer controls in case–control studies.
Am J Epidemiol 1987, 125: 1 –11.
21. Smith AH, Pearce NE, Callas PW. Cancer case–control studies with other cancers as controls.
Int J Cancer 1988, 17: 298 –306.
22. Beral V, Newton R. Overview of the epidemiology of immunodeficiency-associated cancers.
J Natl Cancer Inst Monogr 1998, 23: 1 –6.
23. Goedert JJ, Cote TR, Virgo P. et al
. Spectrum of AIDS-associated malignant disorders.
Lancet 1998, 351: 1833 –1839.
24. Grulich AE, Wan X, Law MG, Coates M, Kaldor JM. Risk of cancer in people with AIDS.
AIDS 1999, 13: 839 –843.
25. WHO/UNAIDS. Epidemiological fact sheet on HIV/AIDS and sexually transmitted diseases.
26. Serraino D, Franceschi S. Kaposi's sarcoma and non-Hodgkin's lymphomas in children and adolescents with AIDS.
AIDS 1996, 10: 643 –647.
27. Mueller BU. Cancers in human immunodeficiency virus-infected children.
J Natl Cancer Inst Monogr 1998, 23: 31 –35.
28. Lenoir G, O'Conor G, Olweny CLM (editors). Burkitt's lymphoma: a human cancer model.
Lyon, France: IARC Scientific Publications no. 60; 1985.
29. Magrath IT. African Burkitt's lymphoma.
Am J Pediatr Hematol/Oncol 1991, 13: 222 –246.
30. Wright D, McKeever P, Carter R. Childhood non-Hodgkin lymphomas in the United Kingdom: findings from the UK Children's Cancer Study Group.
J Clin Pathol 1997, 50: 128 –134.
31. Knowles DM. Etiology and pathogenesis of AIDS-related non-Hodgkin's lymphoma.
Hematol Oncol Clin North Am 1996, 10: 1081 –1109.
32. Hamilton-Dutoit SJ, Raphael M, Audouin J. et al
. In situ demonstration of Epstein–Barr virus small RNAs (EBER 1) in acquired immunodeficiency syndrome-related lymphomas: correlation with tumor morphology and primary site.
Blood 1993, 82: 619 –624.
33. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Human immunodeficiency viruses and human T-cell lymphotropic viruses
, Vol. 67. Lyon, France: IARC; 1996.
34. Roithman S, Andrieu J-M. Clinical and biological characteristics of malignant lymphomas in HIV-infected patients.
Eur J Cancer 1992, 28A: 1501 –1508.
35. Lazzi S, Ferrari F, Nyongo A. et al
. HIV-associated malignant lymphomas in Kenya (Equatorial Africa).
Hum Pathol 1998, 29: 1285 –1289.
36. Marum LH, Tindyebwa D, Gibb B. Care of children with HIV infection and AIDS in Africa.
AIDS 1997, 11 (Suppl. B) : S125 –S134.
37. Biggar RJ, Rosenberg PS, Cote T. Kaposi's sarcoma and non-Hodgkin's lymphoma following the diagnosis of AIDS.
:Multistate AIDS/Cancer Match Study Group.
Int J Cancer 1996, 68: 754 –758.
38. Grant AD, Djomand G, De Cock KM. Natural history and spectrum of disease in adults with HIV/AIDS in Africa.
AIDS 1997, 11 (Suppl. B) : S43 –S54.
39. Morgan D, Maude GH, Malamba SS, Okongo MJ, Wagner HU, Mulder DW, Whitworth JA. HIV-1 disease progression and AIDS-defining disorders in rural Uganda.
Lancet 1997, 26: 245 –250.
40. Boerma JT, Nunn AJ, Whitworth JA. Mortality impact of the AIDS epidemic: evidence from community studies in less developed countries.
AIDS 1998, 12 (Suppl. 1) : S3 –S14.
41. Pluda JM, Yarchoan R, Jaffe ES. et al
. Development of non-Hodgkin lymphoma in a cohort of patients with severe human immunodeficiency virus (HIV) infection on long-term antiretroviral therapy.
Ann Intern Med 1990, 113: 276 –282.
42. Moore RD, Kessler H, Richman DD, Fexner C, Chaisson RE. Non-Hodgkin's lymphoma in patients with HIV infection treated with zidovidine.
JAMA 1991, 25: 2208 –2211.
43. Rabkin CS, Hilgartner MW, Hedberg KW. et al
. Incidence of lymphomas and other cancers in HIV-infected and HIV-uninfected patients with hemophilia.
JAMA 1992, 26: 1090 –1094.
44. Berkley SF, Widy-Wirski R, Okware SI, Downing R, Linnan MJ, White KE, Sempala S. Risk factors associated with HIV infection in Uganda.
J Infect Dis 1989, 160: 22 –30.
45. Kirunga CT, Ntozi JP. Socio-economic determinants of HIV serostatus: a study of Rakai District, Uganda.
Health Transit Rev 1997, 7 (Suppl.) : 175 –188.
46. Faggiano F, Partanen T, Kogevinas M, Boffetta P. Socio-economic differences in cancer incidence and mortality
. In:Social inequalities and cancer
. Kogevinas M, et al
. (editors). Lyon, France: IARC Scientific Publication no. 138; 1997.
47. Jarrett AF, Armstrong AA, Alexander E. Epidemiology of EBV and Hodgkin's lymphoma.
Ann Oncol 1996, 7 (Suppl. 4) : 5 –10.
48. Mueller N, Mohar A, Evans A. Epstein–Barr virus antibody patterns preceding the diagnosis of non-Hodgkin's lymphoma.
Int J Cancer 1991, 49: 387 –393.