The Surveillance, Epidemiology and End Results(SEER) program at the National Cancer Institute reported a 50% increase in the incidence of non-Hodgkin's lymphoma (NHL) in the United States in a 14-year period between the early 1970s and the late 1980s(1). Although some of the increase can be attributed to recent changes in the diagnosis, reporting, and classification systems, relatively few risk factors have been identified that would explain the large increase in incidence over the past 20 to 50 years.
NHL was added to the AIDS surveillance case definition by the U.S. Centers for Disease Control in 1985. By 1994, 11% of the deaths of women and 23% of the deaths of men between the ages of 25 and 44 years in the United States were attributable to HIV infection(2). In the 5-year interval from 1987 to 1992, the rate of HIV-related deaths due to NHL increased from 3.9% to 5.7%(3). At the same time, the incidence of specific opportunistic infections decreased significantly. NHL is the second most commonly diagnosed malignancy in AIDS patients, reported for 3% of all presenting AIDS cases(4-7), and the number of HIV-associated NHL cases is expected to continue to increase as survival improves with treatment of HIV-related opportunistic infections(4,8,9). It has been estimated that there will be a 5% to 10% increase per decade in the cumulative incidence of HIV-associated NHL(9). Genetic or induced immunodeficiencies(10,11) and infection with Epstein-Barr virus(EBV) or human T-cell lymphotrophic virus type I(HTLV-I)(12-14) are confirmed risk factors that explain some of the observed increase in new cases. However, even after accounting for the number of new cases attributed to these conditions, much of the disease incidence remains unexplained(15).
San Francisco, CA, has one of the highest incidence rates of AIDS in the United States: 130 cases per 100,000 men in 1995(16). Between 1989 and 1994, there was a 7% increase in the incidence of AIDS-defining opportunistic illnesses among men who have sex with men(16). An earlier San Francisco Bay Area study of never-married men between the ages of 25 and 44 years who lived in census tracts with at least 50 cases per 100,000 never-married men between the ages of 25 and 44 found that the incidence of NHL in this group doubled from 52 cases per 100,000 men in 1984 to 105 cases per 100,000 men in 1986(17). Despite consistent evidence that indicates an increase in NHL among homosexual men, little is known about potential risk factors other than HIV infection that may account for this increase.
Environmental and occupational exposures are possible etiologic factors for NHL. However, results from studies investigating associations between occupation and NHL have been inconsistent. These were summarized by Pearce in 1992(18). We conducted a population-based case-control study of NHL in the San Francisco Bay Area between 1988 and 1995 to determine risk factors for NHL and to determine whether environmental and occupational exposures play a role in the incidence of this disease among homosexual men. Risk factors suggested by previous population-based studies for the general population and other potential risk factors were analyzed for their relation to NHL in this subgroup of the population and to determine potential effects of exposures by HIV status. Results are presented for 312 homosexual men with NHL and 420 homosexual control subjects. Analyses based on demographics, sexual orientation and other lifestyle factors, and medical history are provided in the companion paper in this issue(19).
MATERIAL AND METHODS
Detailed methods of patient recruitment can be found in the companion paper(19). Briefly, a rapid case-finding system operated by the Northern California Cancer Center was used to identify NHL patients within 1 month of diagnosis in hospitals in the six counties of Alameda, Contra Costa. Marin, San Francisco. Santa Clara, and San Mateo in the San Francisco Bay Area. A total of 312 homosexual men with NHL were among the 1593 eligible NHL patients who comprised the final study population.
Control subjects were frequency matched to the patients by gender, county of residence, and age within 5 years. They were identified using random-digit dial telephone methods(20-23). A total of 420 homosexual men who were free of NHL were among the 2515 control subjects who completed interviews, and these 420 men formed the control group for this study.
In-person structured interviews were conducted by trained interviewers in the study subjects' homes or at a place convenient to the subject. No proxy interviews were conducted. Detailed questions were asked about history of chemical and occupational exposures, exposure to animals and other pets, consumption of raw dairy or meat products, and travel to foreign countries. Interview topics also covered demographic characteristics: family income; homosexual experiences: use of tobacco, alcohol, and illicit drugs; family medical history; and medical history of the subject, including common allergies and other disturbances of the immune system, use of therapeutic drugs, vaccinations, and viral infections. Results have been published in the accompanying paper for the factors not covered in this report(19). Most questions pertained to incidence of exposures or activities up to the year before the interview.
Unconditional logistic regression was used to obtain odds ratios as estimates of relative risks adjusted for age alone and later adjusted for appropriate potential confounders in multivariate models. The 95% confidence intervals for odds ratios and two-sided p values were calculated. Effects were reported as somewhat elevated or borderline when p values were between 0.05 and 0.10. Possible associations and interactions between predictors were assessed. They are mentioned only when they had an effect on the factor of interest. Wilcoxon ranksum tests also were performed to compare the distributions of the length of exposure between patients and control subjects. Details regarding the analysis have been published in the companion report(19).
As described in the accompanying paper(19), HIV status was associated with a 20-fold increased risk for NHL. Because previously examined covariates were highly correlated with HIV infection and because some chemical and occupational exposure characteristics and their effect on the risk of NHL differed in men with HIV disease, analyses were stratified by HIV status. Age was categorized into eight groups and all odds ratios were adjusted for age group. Covariates not inherently categorical were divided into quartiles or tertiles using all control subjects unless paucity of measurements or meaningfulness of categories forced a slightly different classification.
For some factors, when the odds ratios were in the same direction, frequencies and odds ratios are presented pooled over the HIV-positive and HIV-negative men. Pooling of similar data allowed estimation of effects at minimal or at elevated levels of exposure when cell frequencies were too small to perform tests separately by HIV status.
Chemical exposure data were obtained as the number of hours exposed and location of exposure, whether during work, leisure, or school. Occupational exposures were studied by appropriate categorization of the amount of time spent in each type of job. Nature of the job(i.e., whether full-time, part-time, seasonal, or sporadic work) was accounted for when calculating the cumulative number of years worked in each occupation. Occupations first were classified into six broad categories, and then a variety of specific occupations were examined individually. In some instances, the specific occupational categories overlapped, and subjects could fall into multiple categories if their job titles warranted. Because there were no strong or consistent associations by occupational or chemical categories, no multivariate models are presented in this report.
Distribution by age, race, and HIV status of the patients and control subjects can be found in the companion manuscript(19). The median age for NHL subjects and control subjects was 40 and 39 years, respectively. The median age of HIV-positive NHL patients was 40 and that of HIV-negative NHL patients was 48; the median age for HIV-positive and HIV-negative control subjects was 39. NHL patients were much more likely to be HIV positive than were control subjects (age-adjusted OR = 20, CI = 14 to 31, p < 0.0001)(19). One NHL patient and two control subjects did not know their HIV status and were ineligible or chose not to have their blood drawn. All HIV-related analyses were based on 311 patients and 418 control subjects.
Age-adjusted odds ratios and 95% confidence intervals for years of employment in six broad occupational categories are presented in Table 1. HIV-positive patients were likely never to have worked in or to have worked for shorter durations in technical, sales, or administrative occupations; in any service occupation; and in precision production, craft, and repair occupations than were HIV-positive control subjects. HIV-negative men with NHL also were likely to have worked for shorter duration in managerial or professional specialty occupations and in technical, sales, or administrative occupations, but they were somewhat more likely to have worked as operators, fabricators, or laborers than the control subjects. HIV-positive NHL patients had worked at fewer jobs for 6 months or longer than had HIV-positive control subjects (Wilcoxon p = 0.01), but there was no such association among HIV-negative subjects.
Odds ratios and confidence intervals for specific occupational groups are presented in Table 2. Men were exposed to many specific occupations that often were of short duration. Most occupations were not associated with incidence of NHL. An elevated odds ratio of 5.1 (CI= 0.65 to 39), based on 14 exposed NHL patients and 1 exposed control subject, was observed among the HIV-positive men ever having worked in health-diagnostic occupations for 6 months or more (data not shown intable ). However, there was no association among HIV-negative men based on 3 exposed NHL patients and 18 exposed control subjects (OR = 1.1). Among the HIV-negative men, an elevated odds ratio of 2.3 was associated with ever having worked as a hairdresser or a cosmetologist for 6 months or more, but this was based only on 2 exposed NHL patients and 7 exposed control subjects and had a wide confidence interval (CI = 0.45 to 12; data not shown in table). When pooled over HIV status, subjects with NHL were somewhat more likely to have worked as motor vehicle operators (including truck drivers and rail operators) for more than 1 year than the control subjects.
Agricultural occupations, fence work, or work with animals were not associated with increased incidence of NHL in this study (see Table 2). But among HIV-positive subjects, there was an interaction between age and fence building or repairing for at least 2 weeks. These activities were associated with a reduced risk for NHL among men younger than 40 years of age (OR = 0.24, CI = 0.09 to 0.63, p < 0.0001) and with an elevated risk in men 40 years or older (OR = 1.5, CI = 0.57 to 3.7). Without stratification to control for the interaction effect, the odds ratio was 0.64 (see Table 2). There was no increased risk for NHL and having owned pets (data not shown). When responses to questions about type of residence at ages 8, 15, 25, 35, 45, and 55 were analyzed, HIV-negative subjects who had lived on a farm at age 8 appeared to be at increased risk for NHL, but they were not at an increased risk if they moved to a farm as an adult.
Odds ratios pertaining to exposure to various chemicals are presented in Table 3. Chemical exposures were infrequent in this group of men, and many were of short duration. Among HIV-positive men, NHL patients had less exposure to tar, pitch. soot, or ash; to welding, soldering, or metal fumes; and to aldehydes, cleaning solvents, and adhesives. Exposure to insecticides also was somewhat less frequent among HIV-positive patients compared with HIV-positive control subjects.
Among the HIV-negative men, exposures to tar, pitch, soot, or ash for 25 hours or more, exposure to herbicides for 10 hours or more, and to radioactivity or x-rays for 10 hours or more were associated with odds ratios of 2.0 or greater for NHL, but only the risk for radioactivity or x-rays had confidence intervals that did not overlap unity. Exposure to plastic, vinyl, rubber, foam, or plastic fumes also was associated with an increased risk of NHL in the same group (OR = 3.9, CI = 1.1 to 14.1, p = 0.03), but the estimate was based on 4 men with NHL and 8 control subjects who were exposed to these substances. The odds ratio was about 1.0 for the HIV-positive men, based on 8 men with NHL and 3 control subjects who were exposed (data are not presented intable ).
Among HIV-positive and HIV-negative men. no association was found between the incidence of NHl and consumption of raw dairy and meat products, history of professional pest extermination, or travel to foreign countries (data not presented), with travel analyzed first by individual country and then by region.
Multivariate models from the companion report(19) were augmented to include factors presented in this report. Results from the models were not different from those of the univariate analyses previously presented, or the estimates of effects were not robust because of small numbers of subjects, and therefore the models are not presented.
Individuals infected with HIV are at increased risk for NHL(9,24,25), and the risk of NHL in individuals with AIDS was computed to be 60 times that of the general population(26). The data from our study showed that HIV infection resulted in a 20-fold increase in the risk of NHL among the homosexual men who were interviewed in this population-based study(19).
Various occupational, environmental, and chemical exposures were found to be risk factors for NHL in earlier studies. The primary factors were occupations such as farmers, metal workers, painters, or dry cleaners and exposure to zoonotic viruses, solvents, radiation, herbicides, and pesticides, as reviewed by Pearce(18) and by Blair(27). We examined the effect of these and other potential risk factors related to chemical and occupational exposures among HIV-positive men who were at increased risk for NHL as a result of their HIV disease. We analyzed the data to detect synergistic effects between HIV status and potential risk factors for NHL and for interaction effects among exposures. Overall, among homosexual men, there were no strong associations between chemical or occupational exposures and the incidence of NHL. Moreover, the risk estimates often were in opposite directions in HIV-positive and HIV-negative men, and they occasionally were based on few subjects and low levels of exposure, warranting cautious interpretation. When medium or high exposure levels were considered, the associations were likely to have resulted from chance, and there often was no consistent trend effect. There also was no evidence of any exposure producing a synergistic effect along with HIV status on the risk for NHL, although effects of exposures on the risk for NHL often differed by HIV status. There were no significant differences for any occupational or chemical exposure categories when distributions of length of exposure were compared between NHL patients and control subjects using the Wilcoxon rank-sum procedure.
Results from investigations of the risk for NHL associated with agricultural occupations have been inconsistent, as reviewed by Smith in 1989, Johnson in 1990, and Blair in 1990(28-30). Studies of farmers in Kansas(31) and Nebraska (32) found significantly increased risk for NHL associated with herbicide exposure, but studies of farmers in Washington state(33) and New Zealand (34) found no such association. In the group of homosexual men from the San Francisco Bay Area, there was no association of risk for NHL and agricultural occupation, although residence on a farm during childhood provided nearly a 2.5-fold increased risk among HIV-negative men. Exposure to herbicides was associated with a 2-fold increased risk among HIV-negative men, although confidence intervals overlapped 1.0. Exposure to petroleum products, aldehydes, cleaning solvents, adhesives; welding and soldering fumes; and tar, pitch, soot, or ash was related to a reduced risk for NHL among HIV-positive men, even after adjustment for education, but all were associated with modest and inconsistently elevated risk estimates among the HIV-negative men. Oils and greases have been shown to increase the risk for NHL(27).
Associations between occupations and the risk for NHL often were different for the HIV-positive men and the HIV-negative men. Among the HIV-positive men, some of the occupational categories were associated with a reduced risk for NHL, and these associations did not diminish significantly after adjustment for the number of different jobs held and for education. Among the HIV-positive men, control subjects were more likely to have worked in technical, sales, or administrative occupations; in service occupations; and in precision production, craft, and repair occupations, but significance among these broad categories does not provide information about the risk associated with occupations with exposure to certain chemicals. Service occupations and precision production, craft, and repair occupations were not related to NHL incidence among the HIV-negative men. HIV-negative men with NHL were more likely to have worked in health-diagnostic occupations, a group that included physicians, dentists, veterinarians, optometrists, nurses, pharmacists, various types of therapists, dietitians, and paramedics. However, this result may be biased, because those who worked in one of these occupations most likely would have sought early treatment and consequently would have been more likely to have been alive for an interview.
Evidence linking radiation exposure to incidence of NHL has been rather weak in earlier studies, as reviewed by Boice in 1992(35). We found an association among HIV-negative homosexual men and not among HIV-positive men. Similar to the findings of a Swedish case-control study(36), exposure to plastic, vinyl, foam, and plastic fumes were found to increase the risk for NHL in our study among HIV-negative men, although the number of exposed subjects was small. HIV-negative NHL patients were somewhat less likely to have held technical, sales, or administrative jobs, similar to results found earlier(37). A study conducted in Sweden linked an increased risk for NHL with all pets except dogs, cats, and birds(36), and we found no association between pets and the risk of NHL among homosexual men.
The findings of this study supports the causal role of HIV infection in the increased incidence of NHL and provides little evidence that occupational factors are related to NHL risk among HIV-positive men. Even though 80% of NHL incidence among homosexual men is estimated to be attributable to HIV infection(19), none of the occupational risk factors studied provide much of an etiologic clue to NHL incidence among those infected with HIV. Among the HIV-positive men, these results support the hypothesis that the dramatic effect on the immune system resulting from HIV infection is the major factor related to the increased incidence of NHL.
Among the HIV-negative homosexual men in the San Francisco Bay Area, only a few occupations and chemical exposures had elevated risk ratios for NHL, and these often were inconsistent with regard to duration of exposure, and the analyses often included few exposed subjects. Lifetime exposure to herbicides and exposure to radioactivity or x-rays for 10 hours or longer were the only factors retained in a multivariate model that included lifestyle, sexual history and medical history characteristics that were part of the model presented in the companion report(19). However, there were only nine HIV-negative NHL cases with herbicide exposure and seven HIV-negative NHL cases with exposure to radioactivity or x-rays. Multivariate models including these variables were not presented.
The findings of this study indicate that factors related to HIV infection, allergies, and immune response play a greater role than that of occupation in the etiology of NHL in this population of homosexual men in the San Francisco Bay Area(19).
Acknowledgment: This work was supported in part by grant numbers R01-CA45614 and U01-CA66529 from the National Cancer Institute, National Institutes of Health, U.S.A.
1. Miller BA, Ries LAG, Hankey BK, Kosary CL, Edwards BK. Cancer statistics review: 1973-1989
. NIH publication number 92-2789. Bethesda, MD: National Cancer Institute, 1992.
2. Centers for Disease Control and Prevention. Update: mortality attributable to HIV infection among persons aged 25-44 years—United States, 1994. MMWR Morb Mortal Wkly Rep
3. Selik RM, Chu SY, Ward JW. Trends in infectious diseases and cancers among persons dying of HIV infection in the United States from 1987 to 1992. Ann Intern Med
4. Serraino D, Franceschi S, Tirelli U, Monfardini S. The epidemiology of acquired immunodeficiency syndrome and associated tumours in Europe. Ann Oncol
5. Casabona J, Salas T, Salinas R. Trends and survival in AIDS-associated malignancies. Eur J Cancer
6. Centers for Disease Control and Prevention. U.S. HIV and AIDS cases reported through December 1995. HIV/AIDS Surveill Rep
7. Pedersen C, Barton SE, Chiesi A, et al. HIV-related non-Hodgkin's lymphoma among European AIDS patients: AIDS in Europe Study Group. Eur J Haematol
8. Gail MH, Pluda JM, Rabkin CS, et al. Projections of the incidence of non-Hodgkin's lymphoma related to acquired immunodeficiency syndrome. J Natl Cancer Inst
9. Obrams GI, Grufferman S. Epidemiology of HIV-associated non-Hodgkin's lymphoma. Cancer Surv
10. Filipovich AH, Mathur A, Kamat D, Shapiro RS. Primary immunodeficiencies: genetic risk factors for lymphoma. Cancer Res
11. Kamel OW, van de Rijn M, Hanasono MM, Warnke RA. Immunosuppression-associated lymphoproliferative disorders in rheumatic patients. Leukemia Lymphoma
12. Cleghorn FR, Manns A, Falk R, et al. Effect of human T-lymphotropic virus type I infection on non-Hodgkin's lymphoma incidence. J Natl Cancer Inst
13. Mueller NE, Mohar A, Evans A. Viruses other than HIV and non-Hodgkin's lymphoma. Cancer Res
14. Purtilo DT, Strobach RS, Okano M, Davis JR. Epstein-Barr virus-associated lymphoproliferative disorders. Lab Invest
15. Hartge P, Devesa SS, Fraumeni Jr JF. Hodgkin's and non-Hodgkin's lymphomas. Cancer Surv
16. Centers for Disease Control and Prevention. Update: trends in AIDS among men who have sex with men—United States, 1989-1994. MMWR Morb Mortal Wkly Rep
17. Horn PL, DeLorenze GN, Brown SR, Holly EA, West DW. Re: "Temporal trends in the incidence of non-Hodgkin's lymphoma and selected malignancies in a population with a high incidence of acquired immunodeficiency syndrome (AIDS)." Am J Epidemiol
18. Pearce N, Bethwaite P. Increasing incidence of non-Hodgkin's lymphoma: occupational and environmental factors. Cancer Res
19. Holly EA, Lele C. Non-Hodgkin's lymphoma in homosexual men in the San Francisco Bay area: allergies, prior medication use and sex practices. J Acquir Immune Defic Syndr Hum Retrovirol
20. Waksberg J. Sampling methods for random digit dialing. J Am Stat Assoc
21. Hartge P, Brinton LA, Rosenthal JF, Cahill JI, Hoover RN, Waksberg J. Random digit dialing in selecting a population-based control group. Am J Epidemiol
22. Olson SH, Kelsey JL, Pearson TA, Levin B. Evaluation of random digit dialing as a method of control selection in case-control studies
. Am J Epidemiol
23. Lele C, Holly EA, Roseman DS, Thomas DB. Comparison of control subjects recruited by random digit dialing and area survey. Am J Epidemiol
24. Rabkin CS, Goedert JJ. Risk of non-Hodgkin's lymphoma and Kaposi's sarcoma in homosexual men. Lancet
25. Rabkin CS, Biggar RJ, Horm JW. Increasing incidence of cancers associated with the human immunodeficiency virus epidemic. Int J Cancer
26. Beral V, Peterman T, Berkelman R, Jaffe H. AIDS-associated non-Hodgkin lymphoma. Lancet
27. Blair A, Linos A, Stewart PA, et al. Comments on occupational and environmental factors in the origin of non-Hodgkin's lymphoma. Cancer Res
28. Smith AH, Bates MN. Epidemiological studies of cancer and pesticide exposure. In: Ragsdale NN, Menzer RE, eds. Carcinogenicity and pesticides
. Washington, DC: American Chemical Society, 1989:207-22.
29. Johnson ES. Association between soft tissue sarcomas, malignant lymphomas, and phenoxy herbicides/chlorophenols: evidence from occupational cohort studies. Fundam Appl Toxicol
30. Blair A, Zahm SH. Herbicides and cancer: a review and discussion of methodologic issues. Recent Results Cancer Res
31. Hoar SK, Blair A, Holmes FF, et al. Agricultural herbicide use and risk of lymphoma and soft-tissue sarcoma. JAMA
32. Zahm SH, Weisenburger DD, Babbitt PA, et al. A case-control study of non-Hodgkin's lymphoma and the herbicide 2,4-dichlorophenoxyacetic acid (2,4-D) in Eastern Nebraska. Epidemiology
33. Woods JS, Polissar L. Non-Hodgkin's lymphoma among phenoxy herbicide-exposed farm workers in western Washington state. Chemosphere
34. Pearce NE, Sheppard RA, Smith AH, Teague CA. Non-Hodgkin's lymphoma and farming: an expanded case-control study. Int J Cancer
35. Boice Jr JD. Radiation and non-Hodgkin's lymphoma. Cancer Res
36. Persson B, Dahlander AM, Fredriksson M, Brage HN, Ohlson CG, Axelson O. Malignant lymphomas and occupational exposures. Br J Indust Med
37. Scherr PA, Hutchison GB, Neiman RS. Non-Hodgkin's lymphoma and occupational exposure. Cancer Res