Saturday 6 April 1991
No 8745
ORIGINAL ARTICLES AIDS-associated non-Hodgkin lymphoma
Non-Hodgkin lymphoma is associated with HIV infection. We investigated the epidemiology and aetiology of AIDS-related non-Hodgkin lymphoma by analysing data from cases reported to the Centers for Disease Control, Atlanta, USA, up to June 30, 1989. During this period 97 258 AIDS cases were reported, of whom 2824 (2·9%) had non-Hodgkin lymphoma. The condition was about 60 times more common in AIDS patients than in the general US population. 1686 cases were immunoblastic lymphoma, 548 primary lymphoma of the brain, and 590 Burkitt’s lymphoma, a condition which is not normally associated with immunosuppression. The proportion of AIDS patients with immunoblastic lymphoma increased from 0% in those under 1 year old to 3·5% in those aged 50 or more. Primary lymphoma of the brain was constant at 0·6% for all ages. The frequency of Burkitt’s lymphoma increased from zero in infants to a peak at 10-19 years of age (1·8%). Each type of lymphoma was twice as common in whites as in blacks and in men as in women. Lymphoma was most common in patients with haemophilia or clotting disorders and least common in those born in the Caribbean or Africa who had acquired HIV by heterosexual contact.
Epidemiological data suggested that whilst infectious agents (eg, Epstein-Barr virus) may be associated with development of non-Hodgkin lymphomas in AIDS patients there was probably no single cause for all the types of lymphoma. Perhaps the most puzzling question is why Burkitt’s lymphoma is commonly associated with HIV infection but not with other types of
immunosuppression. Introduction An outbreak of non-Hodgkin lymphoma in homosexual men was reported in 1982/-3 soon after the AIDS epidemic was first described. Many cases of non-Hodgkin lymphoma
in association with human immunodeficiency virus (HIV) infection have since been described.4-9 Most HIVassociated lymphomas are high grade and of the large-cell, immunoblastic type. Extranodal involvement is common, patients often presenting with a primary lymphoma of the brain 4-9 True Burkitt’s lymphomas, with characteristic c-myc chromosomal translocations, also occur in patients with AID S.1-10 Little is known about the aetiology of non-Hodgkin lymphoma. However, three features of the epidemiology of the condition are well described. First, congenital or acquired immunosuppression substantially increases the risk of non-Hodgkin lymphoma.ll Most lymphomas in immunosuppressed transplant recipients are of the immunoblastic type; primary lesions of the brain are common and believed to be caused by infectious agents.1U2 Second, Burkitt’s lymphoma can be distinguished from the other non-Hodgkin lymphomas on a pathological and molecular basis, and it has a distinctive geographical distribution, being found particularly in Africa." Third, there is an association between some of these lymphomas and Epstein-Barr virus (EBV).13,14 Lymphomas associated with HIV infection and with other forms of immunosuppression are similar in that they tend to be high grade and involve extranodal sites. However, whereas the EBV genome has been found in the tumour cell DNA of most lymphomas in immunosuppressed transplant recipients, it is found in only about half of HIV-associated lymphomas" Also, Burkitt’s lymphoma is commonly associated with HIV infection but not with other forms of immunosuppression. Previous studies of HIV-associated lymphomas have included fewer than 100 cases and have not investigated risk factors for the malignancy.4-7 By June 30, 1989, 2824 patients with AIDS-associated non-Hodgkin lymphomas had been reported to the Centers for Disease Control (CDC), Atlanta, USA. These data allow us to describe the epidemiology and comment on the possible aetiology of HIV-associated non-Hodgkin lymphomas.
ICRF Cancer Epidemiology Unit, Radcliffe Infirmary, Oxford OX26HE, UK (Prof V. Beral, MRCP); and Centers for Disease Control, Atlanta, Georgia, USA (T. Peterman, MD, MSc, R. Berkelman, MD, H. Jaffe, MD). Correspondence to Prof V. Beral. ADDRESSES:
806
Subjects and methods Data on persons
diagnosed as having AIDS have been collected
by CDC since 1981. Immunoblastic lymphoma and Burkitt’s lymphoma are among the reportable manifestations of AIDS.15 As many as 24 conditions may be reported in the same subject. The diagnosis of immunoblastic lymphoma (large-cell, high-grade, diffuse histocytic or diffuse undifferentiated lymphoma) and Burkitt’s lymphoma (small non-cleaved lymphoma) must be histologically confirmed and meet standard criteria.16 Histological review of the lymphomas was not done. The diagnoses of Burkitt’s lymphoma are likely to be correct, since it has been shown that HIV-associated lymphomas described as Burkitt’s on histological grounds have the chromosomal translocations characteristic of this lymphoma.8-10 All primary lymphomas of the brain are reportable,
irrespective of the histological type. Whilst primary lymphoma of the brain and Burkitt’s lymphoma in AIDS patients have been reportable since 1981, immunoblastic lymphoma has been a reportable condition only since 1985.15 Lymphomas that occur after the initial notification of an AIDS case are often not reported to CDC. Individuals were assigned to HIV transmission groups according CDC-defined criteria. Homosexual and bisexual men were classified as such even if they were intravenous drug users. Patients in the "heterosexual contact" group were classified according to whether they were born in those African and Caribbean countries where heterosexual spread predominates. The expected number of cases of non-Hodgkin lymphoma among persons with AIDS was estimated from the age, sex, and race specific incidence rates of all non-Hodgkin lymphomas in the USA from 1981,85,1’ from the age-specific incidence of primary lymphoma of the brain in the USA from 1973-81,18 and from the age and sex specific incidence of Burkitt’s lymphoma in the USA from 1973-81,19 on the assumption that the average time from onset of immunosuppression to death is 5 years. Differences between risk groups in terms of the percentage with non-Hodgkin lymphoma were examined. When necessary, the effects of age, sex, race, and year of diagnosis of AIDS were studied with log-linear regression methods. to
Results Risk among AIDS patients compared with the general population Up to June 30, 1989, 97 258 cases of AIDS among residents of the USA were notified to CDC. Non-Hodgkin lymphoma was reported in 2824 (2-9%) vs the expected number based on incidence rates in the USA of 47-8. Thus, the risk of non-Hodgkin lymphoma in the USA is about 60 times greater in persons with AIDS than in the general population. The ratio of observed to expected cases decreased substantially with age (table 1). Before the age of 20, non-Hodgkin lymphoma was 360 times more common in AIDS patients than in the general population; by 60 and older it was 20 times more common. The expected numbers of cases of non-Hodgkin lymphoma were estimated separately by sex and race and adjusted for age. The ratio of observed to expected was 59 for males (2696 observed, 45-4 TABLE I-OBSERVED AND EXPECTED INCIDENCE OF NON-HODGKIN LYMPHOMA IN AIDS PATIENTS
*Based
on
the incidence of
non- Hodgkin lymphoma
m
the USA from 1981 to 85
Percentage of AIDS patients with non-Hodgkin lymphoma by age.
53 for females (128 observed, 2-4 expected), 66 (2114 observed, 32 expected), and 39 for blacks (374 observed, 9-6 expected). Expected values could not be calculated for Hispanics. 548 cases of non-Hodgkin lymphoma were primary lesions of the brain and 590 were Burkitt’s lymphoma. The expected numbers of cases were 0-1 for primary brain lymphoma and 0-5 for Burkitt’s lymphoma. Thus, primary lymphoma of the brain and Burkitt’s lymphoma were at least 1000 times more frequent in people with AIDS than in the general population. The remaining 1686 cases were reported as immunoblastic lymphoma; there are no data on
expected), for whites
the incidence of this condition in the USA.
Effects of age, sex, and race The age pattern of the three types of lymphoma differed markedly (X2 for heterogeneity on 12df= 33-71, p < 0-001). The percentage of AIDS patients with primary lymphoma of the brain hardly varied with age (figure); 0-6% of all subjects had the condition. Immunoblastic lymphoma became more common with age, increasing from 0% in those under 1 year old to 3-5% in those over 50 (figure). Burkitt’s lymphoma was most common in the 10-19-yearold age group (1 ,8 %), and the incidence remained relatively constant after the age of 30 at 0-6% (figure). There were no cases of Burkitt’s lymphoma among 969 AIDS patients under 2 years old, 4 cases among those aged between 2 and 4 (out of 570 children), and 3 cases among those between 5 and 14 years old (out of 327 children). The relative risk of lymphoma was estimated separately for each age group with a log-linear model and adjusted for sex, race, and time of diagnosis of AIDS (table II). The pattern by age and type of lymphoma shown in the figure was essentially unchanged after these adjustments. More
807
TABLE II-RELATIVE RISK OF
(NUMBER OF CASES)
AND PERCENTAGE OF PATIENTS WITH NON-HODGKIN LYMPHOMA AMONG ALL AIDS PATIENTS
I
I
I
I
I
I
’
’
.Signlflcantly different from 1 not
0 (p < 0-05). trelative risk with reference to group aged 20-29, 11 relative risk with reference to males, §relatme risk with reference to whites--does include 26 other races—eg, Asian. All relative risks adjusted for age, sex, race, and year of diagnosis of AIDS, as appropriate
males than females had lymphoma, even after adjustment for age, race, and year of diagnosis of AID S, and whites were more likely to have lymphomas reported than blacks or Hispanics. The differences by sex and race were generally similar for each type of lymphoma (table n). HIV transmission group The percentage of AIDS patients who had non-Hodgkin lymphoma ranged from 1 % in those who had acquired HIV by heterosexual contact and were born in the Caribbean or Africa to 5-2% in those with haemophilia or clotting disorders (table ill). Adjusting for age, sex, racial structure of each group, and year of diagnosis of AID S diminished the differences but did not change the overall finding (table in). Risk relative to that in transfusion recipients was calculated because this was the only HIV transmission group in which all ages and races and both sexes were represented. The distribution of the three types of lymphoma did not differ greatly between the HIV transmission groups. Place of residence and birth, time trends, and incubation period There was no systematic variation in lymphoma risk by of residence or place of birth. The highest frequencies of lymphoma among AIDS patients were reported from Alaska 11 % (7 cases), Minnesota 6% (32), Colorado 5-7% (58), and Oregon 4-9% (28). The lowest frequencies were reported from Kansas 1-2% (3), Alabama 1-4% (8), New state
TABLE III-PERCENTAGE OF AIDS PATIENTS WITH NON-HODGKIN LYMPHOMA (NUMBER OF CASES OF PRIMARY BRAIN/BURKITT’S/IMMUNOBLASTIC LYMPHOMA) AND RELATIVE RISK OF LYMPHOMA BY HIV TRANSMISSION GROUP*
*Does not include 79 patients with unknown HIV-transmission group relative risk with reference to transfusion recipients, adjusted for age, sex, race, and year of diagnosis of AIDS iSlgnlflcantly different from 1 0 (p < 005).
Jersey 1-6% (113), Rhode Island 1-6% (4), and New York 1-9% (450). 3% of AIDS patients born in the USA (2457 cases) had non-Hodgkin lymphoma, as did 3-2% of those born in Canada (6), 35% of those born in Mexico (25), and 1-2% of those born in Haiti (19). That the percentage of AIDS patients with a lymphoma increased from 1-6% of those diagnosed in 1984 to 2-9% of those diagnosed in 1985 was entirely due to the addition of immunoblastic lymphoma to the CDC reporting form. Since 1985, about 3% of AIDS patients diagnosed each year have been reported to have a lymphoma. For transfusion recipients with AIDS the incubation period can be estimated from the date of transfusion, if known, to the date of onset of AIDS. There was no difference between mean incubation periods for subjects with Burkitt’s lymphoma (56 months, 9 cases), primary lymphoma of the brain (50 months, 9 cases), immunoblastic lymphoma (49 months, 32 cases), and no lymphoma (50 months, 1405 cases). Discussion We found that non-Hodgkin lymphoma was about 60 times more common in AIDS patients than in the general population of the USA. This increased risk is of a similar order of magnitude to that reported in immunosuppressed transplant recipients." Immunosuppression itself is, therefore, a major determinant of risk in HIV-infected people. It is likely that we have underestimated the percentage of people with AIDS who get lymphomas since conditions which develop after the initial case-report to CDC are often not recorded. This may lead to under reporting of lymphomas that develop late in the course of AID S. 20 However, our results accord with those reported for Italian AIDS patients,21 and under-reporting should not bias comparisons within the study population. The epidemiology of non-Hodgkin lymphoma and of Kaposi’s sarcoma is different in people with AIDS, which suggests different aetiologies for these two malignancies. The risk of Kaposi’s sarcoma is increased in patients who acquire HIV through sexual contact,22 but no such association was evident for non-Hodgkin lymphoma, nor was the risk of lymphoma increased in all groups with HIV infection acquired parenterally. Thus, if infectious agents have a role in the aetiology of lymphomas, their predominant
808
mode of transmission is unlikely to be sexual or parenteral. Differences in the frequency of non-Hodgkin lymphoma by race and HIV transmission group are not great, but suggest that socioeconomic factors may be important in determining risk. Although others have reported similar differences in risk in homosexual men and intravenous drug users,23 it is possible that both these findings were the result of lymphomas being more likely to be recognised and reported in homosexual men than in intravenous drug users. In addition, homosexual men and those with haemophilia may be more likely to receive zidovudine, thus surviving longer and changing the risk of getting lymphoma .20 The classification of lymphoma is complex and has changed over time,24 but certain epidemiological features of the three types of lymphomas reported to CDC are so distinctive that they could not be caused by classification, diagnostic, or reporting biases. In any case, misclassification and diagnostic errors would blur the distinctions between the three types of lymphoma rather than create artificial ones. Burkitt’s lymphoma is strongly associated with HIV infection and was at least 1000 times more common in AIDS patients than in the general population of the USA. One fifth of non-Hodgkin lymphomas in patients with AIDS were reported to CDC as Burkitt’s lymphoma, a proportion similar to that found previously. 4,6,7This contrasts with results in immunosuppressed transplant recipients, in whom fewer than 1% of the non-Hodkin lymphomas were described as Burkitt’s. HIV-associated Burkitt’s lymphoma is indistinguishable at a molecular level from that which arises in the absence of HIV infection.8--10 Epidemiological features are also similar-whether HIV-associated or not, Burkitt’s lymphoma in the USA is more common in males than females, in whites than blacks, and the peak incidence is between the ages of 10 and 20 years19 (table n). This suggests that all forms of Burkitt’s lymphoma have a common trigger, whether associated with HIV infection or not. In Africa Burkitt’s lymphoma is almost always associated with EBV, but EBV DNA has been found in less than one fifth of cases of sporadic Burkitt’s lymphoma in the USA and in about half the cases in AIDS patients. 9,10,13 In contrast with Burkitt’s lymphoma, primary lymphoma of the brain and immunoblastic lymphoma are characteristic of immunosuppressed people. We found that primary lymphoma of the brain was several thousand times more common in AIDS patients than in the general population; comparable figures for immunoblastic lymphoma were not available. Primary lymphoma of the brain in AIDS patients occurred with equal frequency at all ages, and was the only type of lymphoma reported in patients less than 1 year old. This age pattern contrasts with that of the general population of the USA, where the incidence increases with age.18 It is claimed that EBV DNA can be detected in all primary brain lymphomas in people with AIDS.25 Immunoblastic lymphomas that arise in transplant recipients and AIDS patients are believed to be the result of B-cell proliferation secondary to impaired T-cell surveillance.26 The EBV genome is found in immunoblastic lymphomas of virtually all transplant recipients, but in only half of these lymphomas in AIDS patients.9,1O,14 It is not clear why tmmunoblastic lymphoma in AIDS patients is the only of lymphoma whose risk increases with age (figure)./Possibly the age of immunosuppressed individuals the type of lymphoma that they develop,
determines of the irrespective
trigger.
We conclude that immunosuppression is the main determinant of increased risk of non-Hodgkin lymphoma in AIDS patients. Whilst it is likely that infectious agents, and possibly EBV, play a part in development of these lymphomas, there is no single or simple explanation for their occurrence. Epidemiological evidence, inconsistent findings of EBV, and reports that new retroviruses have been isolated from AIDS-associated lymphomas2’ suggest that different agents might be responsible for each type of lymphoma. In searching for causal agents, each type of lymphoma should be studied separately and it should be borne in mind that the route of transmission is unlikely to be sexual or parenteral. Perhaps the most puzzling question is why Burkitt’s lymphoma is common in association with HIV infection but not with other forms of immunosuppression. Much of this work was done while V. B. was a visiting scientist at CDC. We thank Ms Mitzi Mays for help with computing and Ms Sarah Jones for preparing the typescipt.
REFERENCES 1. Doll DC, List AF. Burkitt’s lymphoma in a homosexual. Lancet 1982; i: 1026-27. 2. US Department of Health and Human Services. Diffuse,
undifferentiated non-Hodgkin’s lymphoma among homosexual males-United States. MMWR 1982; 31: 277-79. 3. Ziegler JL, Drew WL, Miner RC, et al. Outbreak of Burkitt’s-like lymphoma in homosexual men. Lancet 1982; ii: 631-33. 4. Ziegler JL, Beckstead JA, Volberding PA, et al. Non-Hodgkin’s lymphoma in 90 homosexual men. Relation to generalized lymphadenopathy and the acquired immunodeficiency syndrome. N Eng J Med 1984; 311: 565-570. 5. Epstein LG, DiCarlo FJ, Joshi W, et al. Primary lymphoma of the central nervous system in children with acquired immunodeficiency syndrome. Pediatrics 1988; 82: 355-63. 6. Levin AM. Lymphoma in acquired immunodeficiency syndrome. Semin
Oncol 1990; 17: 104-12. Kaplan LD, Abrams DI, Feigal E, et al. AIDS-associated non-Hodgkin’s lymphoma in San Francisco. JAMA 1989; 261: 719-24. 8. Whang-Peng J, Lee EC, Sieverts H, Magrath IT. Burkitt’s lymphoma in AIDS: cytogenetic study. Blood 1984; 63: 818-22. 9. Lenoir GM, Delecluse HJ. Lymphoma and immunocompromised host. In: Revilland JP, Wierzbicki N, eds. Immune disorders and opportunistic infections. Suresness: Foundation Franco-Allemante 7.
1989: 173-83. 10. Thomas JA, Allday
MJ, Crawford DH. Epstein-Barr virus associated lymphoproliferative disorders in immunocompromised individuals. Adv Cancer Res (in press). 11. Kinlen LJ. Immunosuppressive therapy and cancer. Cancer Surv 1982; 1: 567-83. 12.. Penn I. Lymphomas complicating organ transplantation. Transplant Proc 1983; 15: 2790-97. 13. Burkitt’s lymphoma. Lenoir G, O’Connor G, Olweny CLM, eds. Lyon: IARC Scientific Publications No 60, 1985. 14. Editorial. Lymphoma in organ transplant recipients. Lancet 1984; i: 601-03. 15. Centers for Disease Control. Revision of the CDC surveillance case definition for acquired immunodeficiency syndrome. MMWR 1987; 36 (suppl 1S): 1S-15S. 16. World Health Organisation. International classification of diseases, ninth revision. Geneva: WHO, 1985. 17. US Department of Health and Human Services, National Cancer Institute. 1987 annual cancer statistics review, including cancer trends: 1950-1985. Bethesda, Maryland: NCI, 1988. 18. Eby NL, Grufferman S, Flannelly CM, Schold SC, Vogel FS, Burger PC. Increasing incidence of primary brain lymphoma in the US. Cancer 1988; 62: 2461-65. 19. Levine PH, Connelly RR, McKay FW. Burkitt’s lymphoma in the USA: cases reported to the American Burkitt’s lymphoma registry compared with population-based incidence and mortality data. In: Lenoir G, O’Connor G, Olweny CLM, eds. Burkitts lymphoma. Lyon: IARC Scientific Publications no 60, 1985: 217-224. 20. Pluda JM, Yarchoan R, Jaffe LS, et al. Development of non-Hodgkins lymphoma in a cohort of patients with severe human immunodeficiency virus (HIV) infection on long term antiretroviral therapy. Ann Intern Med 1990; 113: 276-82.
809
Cooperation Group for AIDS-related Tumours. Malignant lymphomas in patients with or at risk for AIDS in Italy. J Natl Cancer
21. Italian
Inst 1988; 80: 855-60. 22. Beral V, Peterman TA, Berkelman RL, Jaffe HW. Kaposi’s sarcoma among persons with AIDS: a sexually transmitted infection? Lancet 1990; 335: 123-28. 23. Beckhardt RN, Faraday N, May M, Torres RA, Strauchen JA. Increased incidence of malignant lymphoma in AIDS: a comparison of risk groups and possible etiologic factors. Mt Sinai J Med 1988; 55: 383-89. 24. Editorial. Trends in lymphoma diagnosis. Lancet 1989; i: 249-50.
25. Cremer KJ, Spring SB, Gruber J. Role of human immunodeficiency virus type 1 and other viruses in malignancies associated with acquired immunodeficiency disease syndrome. J Natl Cancer Inst 1990; 82: 1016-24. 26. Editorial. Epstein-Barr virus silver anniversary. Lancet 1989; i: 1171-73. 27. Ng VL, Khayam-Bashi F, Marsh J, McGrath MS. Serologic analysis of proteins associated with and preliminary seroprevalence studies of two novel retroviruses isolated from AIDS-associated lymphomas. IVth International Conference on AIDS, Montreal, Canada, 1989; C777 (abstr): 688.
Angioscopic coronary macromorphology in patients with acute coronary disorders
To investigate the pathogenesis of acute coronary disorders and to clarify what type of plaque precedes these disorders, percutaneous transluminal coronary angioscopy, by means of a new angioscope, was carried out during catheterisation in 100 consecutive patients anatomically suitable for such investigations. The quality of the angioscopic image was good enough for analysis in 84 patients (14 with acute myocardial infarction [within 8 h of onset], 16 with recent myocardial infarction [3 days-2 months since onset], 24 with old myocardial infarctions, 10 with unstable angina, and 20 with stable angina). Thrombi were observed in most patients with acute coronary disorders (all 14 with acute myocardial infarction, 9 of 10 with unstable angina). Occlusive thrombi were more common in patients with acute myocardial infarction than in those with unstable angina (11 [79%] vs 1 [10%]; p
No 8745
ORIGINAL ARTICLES AIDS-associated non-Hodgkin lymphoma
Non-Hodgkin lymphoma is associated with HIV infection. We investigated the epidemiology and aetiology of AIDS-related non-Hodgkin lymphoma by analysing data from cases reported to the Centers for Disease Control, Atlanta, USA, up to June 30, 1989. During this period 97 258 AIDS cases were reported, of whom 2824 (2·9%) had non-Hodgkin lymphoma. The condition was about 60 times more common in AIDS patients than in the general US population. 1686 cases were immunoblastic lymphoma, 548 primary lymphoma of the brain, and 590 Burkitt’s lymphoma, a condition which is not normally associated with immunosuppression. The proportion of AIDS patients with immunoblastic lymphoma increased from 0% in those under 1 year old to 3·5% in those aged 50 or more. Primary lymphoma of the brain was constant at 0·6% for all ages. The frequency of Burkitt’s lymphoma increased from zero in infants to a peak at 10-19 years of age (1·8%). Each type of lymphoma was twice as common in whites as in blacks and in men as in women. Lymphoma was most common in patients with haemophilia or clotting disorders and least common in those born in the Caribbean or Africa who had acquired HIV by heterosexual contact.
Epidemiological data suggested that whilst infectious agents (eg, Epstein-Barr virus) may be associated with development of non-Hodgkin lymphomas in AIDS patients there was probably no single cause for all the types of lymphoma. Perhaps the most puzzling question is why Burkitt’s lymphoma is commonly associated with HIV infection but not with other types of
immunosuppression. Introduction An outbreak of non-Hodgkin lymphoma in homosexual men was reported in 1982/-3 soon after the AIDS epidemic was first described. Many cases of non-Hodgkin lymphoma
in association with human immunodeficiency virus (HIV) infection have since been described.4-9 Most HIVassociated lymphomas are high grade and of the large-cell, immunoblastic type. Extranodal involvement is common, patients often presenting with a primary lymphoma of the brain 4-9 True Burkitt’s lymphomas, with characteristic c-myc chromosomal translocations, also occur in patients with AID S.1-10 Little is known about the aetiology of non-Hodgkin lymphoma. However, three features of the epidemiology of the condition are well described. First, congenital or acquired immunosuppression substantially increases the risk of non-Hodgkin lymphoma.ll Most lymphomas in immunosuppressed transplant recipients are of the immunoblastic type; primary lesions of the brain are common and believed to be caused by infectious agents.1U2 Second, Burkitt’s lymphoma can be distinguished from the other non-Hodgkin lymphomas on a pathological and molecular basis, and it has a distinctive geographical distribution, being found particularly in Africa." Third, there is an association between some of these lymphomas and Epstein-Barr virus (EBV).13,14 Lymphomas associated with HIV infection and with other forms of immunosuppression are similar in that they tend to be high grade and involve extranodal sites. However, whereas the EBV genome has been found in the tumour cell DNA of most lymphomas in immunosuppressed transplant recipients, it is found in only about half of HIV-associated lymphomas" Also, Burkitt’s lymphoma is commonly associated with HIV infection but not with other forms of immunosuppression. Previous studies of HIV-associated lymphomas have included fewer than 100 cases and have not investigated risk factors for the malignancy.4-7 By June 30, 1989, 2824 patients with AIDS-associated non-Hodgkin lymphomas had been reported to the Centers for Disease Control (CDC), Atlanta, USA. These data allow us to describe the epidemiology and comment on the possible aetiology of HIV-associated non-Hodgkin lymphomas.
ICRF Cancer Epidemiology Unit, Radcliffe Infirmary, Oxford OX26HE, UK (Prof V. Beral, MRCP); and Centers for Disease Control, Atlanta, Georgia, USA (T. Peterman, MD, MSc, R. Berkelman, MD, H. Jaffe, MD). Correspondence to Prof V. Beral. ADDRESSES:
806
Subjects and methods Data on persons
diagnosed as having AIDS have been collected
by CDC since 1981. Immunoblastic lymphoma and Burkitt’s lymphoma are among the reportable manifestations of AIDS.15 As many as 24 conditions may be reported in the same subject. The diagnosis of immunoblastic lymphoma (large-cell, high-grade, diffuse histocytic or diffuse undifferentiated lymphoma) and Burkitt’s lymphoma (small non-cleaved lymphoma) must be histologically confirmed and meet standard criteria.16 Histological review of the lymphomas was not done. The diagnoses of Burkitt’s lymphoma are likely to be correct, since it has been shown that HIV-associated lymphomas described as Burkitt’s on histological grounds have the chromosomal translocations characteristic of this lymphoma.8-10 All primary lymphomas of the brain are reportable,
irrespective of the histological type. Whilst primary lymphoma of the brain and Burkitt’s lymphoma in AIDS patients have been reportable since 1981, immunoblastic lymphoma has been a reportable condition only since 1985.15 Lymphomas that occur after the initial notification of an AIDS case are often not reported to CDC. Individuals were assigned to HIV transmission groups according CDC-defined criteria. Homosexual and bisexual men were classified as such even if they were intravenous drug users. Patients in the "heterosexual contact" group were classified according to whether they were born in those African and Caribbean countries where heterosexual spread predominates. The expected number of cases of non-Hodgkin lymphoma among persons with AIDS was estimated from the age, sex, and race specific incidence rates of all non-Hodgkin lymphomas in the USA from 1981,85,1’ from the age-specific incidence of primary lymphoma of the brain in the USA from 1973-81,18 and from the age and sex specific incidence of Burkitt’s lymphoma in the USA from 1973-81,19 on the assumption that the average time from onset of immunosuppression to death is 5 years. Differences between risk groups in terms of the percentage with non-Hodgkin lymphoma were examined. When necessary, the effects of age, sex, race, and year of diagnosis of AIDS were studied with log-linear regression methods. to
Results Risk among AIDS patients compared with the general population Up to June 30, 1989, 97 258 cases of AIDS among residents of the USA were notified to CDC. Non-Hodgkin lymphoma was reported in 2824 (2-9%) vs the expected number based on incidence rates in the USA of 47-8. Thus, the risk of non-Hodgkin lymphoma in the USA is about 60 times greater in persons with AIDS than in the general population. The ratio of observed to expected cases decreased substantially with age (table 1). Before the age of 20, non-Hodgkin lymphoma was 360 times more common in AIDS patients than in the general population; by 60 and older it was 20 times more common. The expected numbers of cases of non-Hodgkin lymphoma were estimated separately by sex and race and adjusted for age. The ratio of observed to expected was 59 for males (2696 observed, 45-4 TABLE I-OBSERVED AND EXPECTED INCIDENCE OF NON-HODGKIN LYMPHOMA IN AIDS PATIENTS
*Based
on
the incidence of
non- Hodgkin lymphoma
m
the USA from 1981 to 85
Percentage of AIDS patients with non-Hodgkin lymphoma by age.
53 for females (128 observed, 2-4 expected), 66 (2114 observed, 32 expected), and 39 for blacks (374 observed, 9-6 expected). Expected values could not be calculated for Hispanics. 548 cases of non-Hodgkin lymphoma were primary lesions of the brain and 590 were Burkitt’s lymphoma. The expected numbers of cases were 0-1 for primary brain lymphoma and 0-5 for Burkitt’s lymphoma. Thus, primary lymphoma of the brain and Burkitt’s lymphoma were at least 1000 times more frequent in people with AIDS than in the general population. The remaining 1686 cases were reported as immunoblastic lymphoma; there are no data on
expected), for whites
the incidence of this condition in the USA.
Effects of age, sex, and race The age pattern of the three types of lymphoma differed markedly (X2 for heterogeneity on 12df= 33-71, p < 0-001). The percentage of AIDS patients with primary lymphoma of the brain hardly varied with age (figure); 0-6% of all subjects had the condition. Immunoblastic lymphoma became more common with age, increasing from 0% in those under 1 year old to 3-5% in those over 50 (figure). Burkitt’s lymphoma was most common in the 10-19-yearold age group (1 ,8 %), and the incidence remained relatively constant after the age of 30 at 0-6% (figure). There were no cases of Burkitt’s lymphoma among 969 AIDS patients under 2 years old, 4 cases among those aged between 2 and 4 (out of 570 children), and 3 cases among those between 5 and 14 years old (out of 327 children). The relative risk of lymphoma was estimated separately for each age group with a log-linear model and adjusted for sex, race, and time of diagnosis of AIDS (table II). The pattern by age and type of lymphoma shown in the figure was essentially unchanged after these adjustments. More
807
TABLE II-RELATIVE RISK OF
(NUMBER OF CASES)
AND PERCENTAGE OF PATIENTS WITH NON-HODGKIN LYMPHOMA AMONG ALL AIDS PATIENTS
I
I
I
I
I
I
’
’
.Signlflcantly different from 1 not
0 (p < 0-05). trelative risk with reference to group aged 20-29, 11 relative risk with reference to males, §relatme risk with reference to whites--does include 26 other races—eg, Asian. All relative risks adjusted for age, sex, race, and year of diagnosis of AIDS, as appropriate
males than females had lymphoma, even after adjustment for age, race, and year of diagnosis of AID S, and whites were more likely to have lymphomas reported than blacks or Hispanics. The differences by sex and race were generally similar for each type of lymphoma (table n). HIV transmission group The percentage of AIDS patients who had non-Hodgkin lymphoma ranged from 1 % in those who had acquired HIV by heterosexual contact and were born in the Caribbean or Africa to 5-2% in those with haemophilia or clotting disorders (table ill). Adjusting for age, sex, racial structure of each group, and year of diagnosis of AID S diminished the differences but did not change the overall finding (table in). Risk relative to that in transfusion recipients was calculated because this was the only HIV transmission group in which all ages and races and both sexes were represented. The distribution of the three types of lymphoma did not differ greatly between the HIV transmission groups. Place of residence and birth, time trends, and incubation period There was no systematic variation in lymphoma risk by of residence or place of birth. The highest frequencies of lymphoma among AIDS patients were reported from Alaska 11 % (7 cases), Minnesota 6% (32), Colorado 5-7% (58), and Oregon 4-9% (28). The lowest frequencies were reported from Kansas 1-2% (3), Alabama 1-4% (8), New state
TABLE III-PERCENTAGE OF AIDS PATIENTS WITH NON-HODGKIN LYMPHOMA (NUMBER OF CASES OF PRIMARY BRAIN/BURKITT’S/IMMUNOBLASTIC LYMPHOMA) AND RELATIVE RISK OF LYMPHOMA BY HIV TRANSMISSION GROUP*
*Does not include 79 patients with unknown HIV-transmission group relative risk with reference to transfusion recipients, adjusted for age, sex, race, and year of diagnosis of AIDS iSlgnlflcantly different from 1 0 (p < 005).
Jersey 1-6% (113), Rhode Island 1-6% (4), and New York 1-9% (450). 3% of AIDS patients born in the USA (2457 cases) had non-Hodgkin lymphoma, as did 3-2% of those born in Canada (6), 35% of those born in Mexico (25), and 1-2% of those born in Haiti (19). That the percentage of AIDS patients with a lymphoma increased from 1-6% of those diagnosed in 1984 to 2-9% of those diagnosed in 1985 was entirely due to the addition of immunoblastic lymphoma to the CDC reporting form. Since 1985, about 3% of AIDS patients diagnosed each year have been reported to have a lymphoma. For transfusion recipients with AIDS the incubation period can be estimated from the date of transfusion, if known, to the date of onset of AIDS. There was no difference between mean incubation periods for subjects with Burkitt’s lymphoma (56 months, 9 cases), primary lymphoma of the brain (50 months, 9 cases), immunoblastic lymphoma (49 months, 32 cases), and no lymphoma (50 months, 1405 cases). Discussion We found that non-Hodgkin lymphoma was about 60 times more common in AIDS patients than in the general population of the USA. This increased risk is of a similar order of magnitude to that reported in immunosuppressed transplant recipients." Immunosuppression itself is, therefore, a major determinant of risk in HIV-infected people. It is likely that we have underestimated the percentage of people with AIDS who get lymphomas since conditions which develop after the initial case-report to CDC are often not recorded. This may lead to under reporting of lymphomas that develop late in the course of AID S. 20 However, our results accord with those reported for Italian AIDS patients,21 and under-reporting should not bias comparisons within the study population. The epidemiology of non-Hodgkin lymphoma and of Kaposi’s sarcoma is different in people with AIDS, which suggests different aetiologies for these two malignancies. The risk of Kaposi’s sarcoma is increased in patients who acquire HIV through sexual contact,22 but no such association was evident for non-Hodgkin lymphoma, nor was the risk of lymphoma increased in all groups with HIV infection acquired parenterally. Thus, if infectious agents have a role in the aetiology of lymphomas, their predominant
808
mode of transmission is unlikely to be sexual or parenteral. Differences in the frequency of non-Hodgkin lymphoma by race and HIV transmission group are not great, but suggest that socioeconomic factors may be important in determining risk. Although others have reported similar differences in risk in homosexual men and intravenous drug users,23 it is possible that both these findings were the result of lymphomas being more likely to be recognised and reported in homosexual men than in intravenous drug users. In addition, homosexual men and those with haemophilia may be more likely to receive zidovudine, thus surviving longer and changing the risk of getting lymphoma .20 The classification of lymphoma is complex and has changed over time,24 but certain epidemiological features of the three types of lymphomas reported to CDC are so distinctive that they could not be caused by classification, diagnostic, or reporting biases. In any case, misclassification and diagnostic errors would blur the distinctions between the three types of lymphoma rather than create artificial ones. Burkitt’s lymphoma is strongly associated with HIV infection and was at least 1000 times more common in AIDS patients than in the general population of the USA. One fifth of non-Hodgkin lymphomas in patients with AIDS were reported to CDC as Burkitt’s lymphoma, a proportion similar to that found previously. 4,6,7This contrasts with results in immunosuppressed transplant recipients, in whom fewer than 1% of the non-Hodkin lymphomas were described as Burkitt’s. HIV-associated Burkitt’s lymphoma is indistinguishable at a molecular level from that which arises in the absence of HIV infection.8--10 Epidemiological features are also similar-whether HIV-associated or not, Burkitt’s lymphoma in the USA is more common in males than females, in whites than blacks, and the peak incidence is between the ages of 10 and 20 years19 (table n). This suggests that all forms of Burkitt’s lymphoma have a common trigger, whether associated with HIV infection or not. In Africa Burkitt’s lymphoma is almost always associated with EBV, but EBV DNA has been found in less than one fifth of cases of sporadic Burkitt’s lymphoma in the USA and in about half the cases in AIDS patients. 9,10,13 In contrast with Burkitt’s lymphoma, primary lymphoma of the brain and immunoblastic lymphoma are characteristic of immunosuppressed people. We found that primary lymphoma of the brain was several thousand times more common in AIDS patients than in the general population; comparable figures for immunoblastic lymphoma were not available. Primary lymphoma of the brain in AIDS patients occurred with equal frequency at all ages, and was the only type of lymphoma reported in patients less than 1 year old. This age pattern contrasts with that of the general population of the USA, where the incidence increases with age.18 It is claimed that EBV DNA can be detected in all primary brain lymphomas in people with AIDS.25 Immunoblastic lymphomas that arise in transplant recipients and AIDS patients are believed to be the result of B-cell proliferation secondary to impaired T-cell surveillance.26 The EBV genome is found in immunoblastic lymphomas of virtually all transplant recipients, but in only half of these lymphomas in AIDS patients.9,1O,14 It is not clear why tmmunoblastic lymphoma in AIDS patients is the only of lymphoma whose risk increases with age (figure)./Possibly the age of immunosuppressed individuals the type of lymphoma that they develop,
determines of the irrespective
trigger.
We conclude that immunosuppression is the main determinant of increased risk of non-Hodgkin lymphoma in AIDS patients. Whilst it is likely that infectious agents, and possibly EBV, play a part in development of these lymphomas, there is no single or simple explanation for their occurrence. Epidemiological evidence, inconsistent findings of EBV, and reports that new retroviruses have been isolated from AIDS-associated lymphomas2’ suggest that different agents might be responsible for each type of lymphoma. In searching for causal agents, each type of lymphoma should be studied separately and it should be borne in mind that the route of transmission is unlikely to be sexual or parenteral. Perhaps the most puzzling question is why Burkitt’s lymphoma is common in association with HIV infection but not with other forms of immunosuppression. Much of this work was done while V. B. was a visiting scientist at CDC. We thank Ms Mitzi Mays for help with computing and Ms Sarah Jones for preparing the typescipt.
REFERENCES 1. Doll DC, List AF. Burkitt’s lymphoma in a homosexual. Lancet 1982; i: 1026-27. 2. US Department of Health and Human Services. Diffuse,
undifferentiated non-Hodgkin’s lymphoma among homosexual males-United States. MMWR 1982; 31: 277-79. 3. Ziegler JL, Drew WL, Miner RC, et al. Outbreak of Burkitt’s-like lymphoma in homosexual men. Lancet 1982; ii: 631-33. 4. Ziegler JL, Beckstead JA, Volberding PA, et al. Non-Hodgkin’s lymphoma in 90 homosexual men. Relation to generalized lymphadenopathy and the acquired immunodeficiency syndrome. N Eng J Med 1984; 311: 565-570. 5. Epstein LG, DiCarlo FJ, Joshi W, et al. Primary lymphoma of the central nervous system in children with acquired immunodeficiency syndrome. Pediatrics 1988; 82: 355-63. 6. Levin AM. Lymphoma in acquired immunodeficiency syndrome. Semin
Oncol 1990; 17: 104-12. Kaplan LD, Abrams DI, Feigal E, et al. AIDS-associated non-Hodgkin’s lymphoma in San Francisco. JAMA 1989; 261: 719-24. 8. Whang-Peng J, Lee EC, Sieverts H, Magrath IT. Burkitt’s lymphoma in AIDS: cytogenetic study. Blood 1984; 63: 818-22. 9. Lenoir GM, Delecluse HJ. Lymphoma and immunocompromised host. In: Revilland JP, Wierzbicki N, eds. Immune disorders and opportunistic infections. Suresness: Foundation Franco-Allemante 7.
1989: 173-83. 10. Thomas JA, Allday
MJ, Crawford DH. Epstein-Barr virus associated lymphoproliferative disorders in immunocompromised individuals. Adv Cancer Res (in press). 11. Kinlen LJ. Immunosuppressive therapy and cancer. Cancer Surv 1982; 1: 567-83. 12.. Penn I. Lymphomas complicating organ transplantation. Transplant Proc 1983; 15: 2790-97. 13. Burkitt’s lymphoma. Lenoir G, O’Connor G, Olweny CLM, eds. Lyon: IARC Scientific Publications No 60, 1985. 14. Editorial. Lymphoma in organ transplant recipients. Lancet 1984; i: 601-03. 15. Centers for Disease Control. Revision of the CDC surveillance case definition for acquired immunodeficiency syndrome. MMWR 1987; 36 (suppl 1S): 1S-15S. 16. World Health Organisation. International classification of diseases, ninth revision. Geneva: WHO, 1985. 17. US Department of Health and Human Services, National Cancer Institute. 1987 annual cancer statistics review, including cancer trends: 1950-1985. Bethesda, Maryland: NCI, 1988. 18. Eby NL, Grufferman S, Flannelly CM, Schold SC, Vogel FS, Burger PC. Increasing incidence of primary brain lymphoma in the US. Cancer 1988; 62: 2461-65. 19. Levine PH, Connelly RR, McKay FW. Burkitt’s lymphoma in the USA: cases reported to the American Burkitt’s lymphoma registry compared with population-based incidence and mortality data. In: Lenoir G, O’Connor G, Olweny CLM, eds. Burkitts lymphoma. Lyon: IARC Scientific Publications no 60, 1985: 217-224. 20. Pluda JM, Yarchoan R, Jaffe LS, et al. Development of non-Hodgkins lymphoma in a cohort of patients with severe human immunodeficiency virus (HIV) infection on long term antiretroviral therapy. Ann Intern Med 1990; 113: 276-82.
809
Cooperation Group for AIDS-related Tumours. Malignant lymphomas in patients with or at risk for AIDS in Italy. J Natl Cancer
21. Italian
Inst 1988; 80: 855-60. 22. Beral V, Peterman TA, Berkelman RL, Jaffe HW. Kaposi’s sarcoma among persons with AIDS: a sexually transmitted infection? Lancet 1990; 335: 123-28. 23. Beckhardt RN, Faraday N, May M, Torres RA, Strauchen JA. Increased incidence of malignant lymphoma in AIDS: a comparison of risk groups and possible etiologic factors. Mt Sinai J Med 1988; 55: 383-89. 24. Editorial. Trends in lymphoma diagnosis. Lancet 1989; i: 249-50.
25. Cremer KJ, Spring SB, Gruber J. Role of human immunodeficiency virus type 1 and other viruses in malignancies associated with acquired immunodeficiency disease syndrome. J Natl Cancer Inst 1990; 82: 1016-24. 26. Editorial. Epstein-Barr virus silver anniversary. Lancet 1989; i: 1171-73. 27. Ng VL, Khayam-Bashi F, Marsh J, McGrath MS. Serologic analysis of proteins associated with and preliminary seroprevalence studies of two novel retroviruses isolated from AIDS-associated lymphomas. IVth International Conference on AIDS, Montreal, Canada, 1989; C777 (abstr): 688.
Angioscopic coronary macromorphology in patients with acute coronary disorders
To investigate the pathogenesis of acute coronary disorders and to clarify what type of plaque precedes these disorders, percutaneous transluminal coronary angioscopy, by means of a new angioscope, was carried out during catheterisation in 100 consecutive patients anatomically suitable for such investigations. The quality of the angioscopic image was good enough for analysis in 84 patients (14 with acute myocardial infarction [within 8 h of onset], 16 with recent myocardial infarction [3 days-2 months since onset], 24 with old myocardial infarctions, 10 with unstable angina, and 20 with stable angina). Thrombi were observed in most patients with acute coronary disorders (all 14 with acute myocardial infarction, 9 of 10 with unstable angina). Occlusive thrombi were more common in patients with acute myocardial infarction than in those with unstable angina (11 [79%] vs 1 [10%]; p
