|Home | About | Journals | Submit | Contact Us | Français|
Reasons for higher incidence of lymphoid neoplasms among men than women are unknown. Because female sex hormones have immunomodulatory effects, reproductive factors and exogenous hormone use may affect risk for lymphoid malignancies. Previous epidemiologic studies on this topic have yielded conflicting results. Within the National Institutes of Health-AARP Diet and Health Study cohort, we prospectively analyzed detailed, questionnaire-derived information on menstrual and reproductive factors and use of oral contraceptives and menopausal hormone therapy among 134,074 US women. Using multivariable proportional hazards regression models, we estimated relative risks (RRs) for 85 plasma cell neoplasms and 417 non-Hodgkin lymphomas (NHLs) identified during follow-up from 1996-2002. We observed no statistically significant associations between plasma cell neoplasms, NHL, or the three most common NHL subtypes and age at menarche, parity, age at first birth, oral contraceptive use, or menopausal status at baseline. For menopausal hormone therapy use, overall associations between NHL and unopposed estrogen and estrogen plus progestin were null, with the potential exception of an inverse association (RR=0.49, 95% CI, 0.25-0.96) between use of unopposed estrogen and diffuse large B-cell lymphoma (DLBCL), the most common NHL subtype, among women with a hysterectomy. These data do not support an important role for reproductive factors or exogenous hormones in modulating lymphomagenesis.
Lymphoid neoplasms, including non-Hodgkin lymphomas (NHL), Hodgkin lymphomas, and plasma cell neoplasms, are the sixth most commonly diagnosed group of cancers worldwide.1 Severe disruption of immune function is a well-established, strong risk factor for lymphomas, but much of lymphoid-neoplasm etiology remains unknown. The potential effects of modest immune perturbation on lymphomagenesis are not well understood.
In humans and animals, menses, pregnancy, and exogenous hormone use affect immune function.as reviewed in 2, 3 This raises the possibility that reproductive factors and exogenous hormones modulate lymphomagenesis in women, which could help explain the lower incidence of almost all lymphoid neoplasm subtypes among women compared with men. Overall, associations between these factors and lymphoma risk are conflicting.4-26
Recent studies reported strong, statistically significant inverse associations between menopausal hormone therapy (MHT) and lymphoid neoplasms.20, 26 Odds ratios (ORs) of 0.6—40% risk reductions compared with never-use—for diffuse large B-cell lymphoma (DLBCL),20, 26 Hodgkin lymphoma,18 and all NHL combined13 could, if true, reveal etiologic or prevention clues. However, the validity of these findings is unclear. The types of available MHT and patterns of use have substantially changed since the 1960s,27 so associations with broadly defined “MHT” (especially from older studies) may reflect exposures that are not comparable across studies conducted during different time periods. Associations with specific subtypes of NHL are biologically plausible28, 29 but warrant cautious interpretation because they could be due to chance and often are difficult to replicate.30 Large, prospective studies with detailed data on use of specific MHT formulations and regimens have helped identify replicable associations between MHT and other cancers. We therefore evaluated MHT and other reproductive exposures as risk factors for lymphoid neoplasms and specific lymphoid neoplasm subtypes in the National Institutes of Health (NIH)-AARP Diet and Health Study.
This study began in 1995-1996, when 3.5 million AARP members, male and female, ages 50-71, residing in six states (CA, FL, LA, NJ, NC, and PA) and two metropolitan areas (Atlanta, GA and Detroit, MI) received a mailed questionnaire.31 The 567,169 AARP members (16.2%) who satisfactorily completed it received a second questionnaire, which collected detailed data on menstrual and reproductive factors and use of oral contraceptives and MHT, as well as other risk factors, in 1996-1997. A total of 334,908 AARP members (59.1%) completed that questionnaire. Overall, participants in the study were more likely to be non-Hispanic Caucasian, were slightly better educated, and had a slightly lower rate of current smoking compared with the general population.31 Participants are followed annually for change of address and linked to state cancer registries and the National Death Index (NDI).
To focus on the detailed MHT data from the second questionnaire, we excluded 188,117 men, 10,383 proxy respondents, and 1550 women who reported a personal history of cancer on either questionnaire (including 119 lymphoid neoplasms), and 784 women for whom all MHT data were missing. Analysis therefore included 134,074 women.
The baseline questionnaire ascertained gynecologic surgeries, demographics, reproductive history, oral contraceptive use, menopausal status, smoking, family history of cancer, and basic information (ever-never, recency, and duration) about MHT use. The second questionnaire collected detailed MHT data. With those data, we used self-reported type, dates, and duration of MHT use to classify exposure according to formulation (i.e., unopposed estrogen vs. estrogen plus progestin), regimen [sequential (fewer than 15 days of progestin per cycle) or continuous (at least 15 days of progestin per cycle), for use of estrogen plus progestin], duration, and recency, as previously described.32 Exposure categories included women who reported use of only one MHT formulation (~95% of MHT users); the ~5% of women who reported multiple formulations were classified separately.
Probabilistic linkage to state cancer registries in the study states,33 plus three states to which relocation is common (AZ, NV, and TX) identified incident cancers among participants. Linkage to NDI provided date and cause of death for fatal cancers.
We defined incident lymphoid neoplasms using International Classification of Diseases for Oncology, Second and Third Edition (ICD-O-2 and ICD-O-3)34, 35 histology codes provided by the cancer registries. According to the World Health Organization classification36 and the International Lymphoma Epidemiology Consortium (InterLymph) guidelines,37 we grouped 634 total lymphoid neoplasm cases into plasma cell neoplasms (ICD-O-3 codes 9731-4; N=85), Hodgkin lymphoma (9650-55, 9659, 9661-9667; N=16), NHL (9591, 9670-1, 9673, 9675, 9678-80, 9684, 9687, 9689-91, 9695, 9698-9702, 9705, 9708-9, 9714, 9716-9, 9727-9, 9760-4, 9823, 9826-7, 9831-37, 9940; N=417), and lymphoid neoplasms NOS (9590, 9822; N=116). We also evaluated the three most common NHL subtypes: diffuse large B-cell lymphoma (DLBCL; 9678-80, 9684; N=96), follicular lymphoma (9690-1, 9695, 9698; N=105), and chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL; 9670, 9823; N=84).
Cox proportional hazards regression, with age as the time scale and ties handled by complete enumeration,38 generated hazard ratios to estimate relative risks (RRs). Tests of proportional hazards revealed no departures. Follow-up began when the second questionnaire was received and scanned. Follow-up ended at the earliest of the following dates: diagnosis of lymphoid neoplasm (N=634), movement out of a registry catchment area (N=5219), death (N=4496), diagnosis of a non-lymphoid hematopoietic neoplasm (ICD-O-3 codes 9740-58, 9800-5, 9840-9931, 9945-64, 9975-9989; N=75), or June 30, 2002 (N=123,650). We truncated follow-up in 2002 because we did not update MHT use during follow-up and millions of women changed or stopped their MHT use after the July 2002 publication of Women's Health Initiative study results.39
Models included adjustment for age at baseline, race, education, menopausal status, and use of oral contraceptives. Additional adjustment for calendar time or other potential confounders had little influence of RRs, so these factors were dropped from statistical models. For each exposure, we evaluated potential associations with five outcomes: plasma cell neoplasms, NHL, and the three most common NHL subtypes: DLBCL, follicular lymphoma, and CLL/SLL. We excluded Hodgkin lymphoma from further analysis due to too few cases (N=16). In addition, for MHT, we investigated formulations (i.e., unopposed estrogen vs. estrogen plus progestin) two ways: first, among the entire cohort, and second, after stratifying on hysterectomy status at baseline. For the latter, unopposed estrogen was evaluated only among women with hysterectomy and estrogen plus progestin was evaluated only among women with an intact uterus, to reflect the predominant prescribing patterns over the last 15 years.
The 134,074 women in the NIH-AARP Diet and Health Study accrued 719,035 total person-years during follow-up. The mean age at study entry was 63 years (range 51 to 72 years), and 91% of women were Caucasian. The 417 NHLs diagnosed during follow-up were similar to the number expected among US women of comparable age (based on 12 SEER registries, standardized incidence ratio (SIR) 1.05, 95% CI 0.95-1.16), whereas the 85 plasma cell neoplasms diagnosed during follow-up were fewer than expected (SIR 0.76, 95% CI 0.61-0.94), likely because our study population includes proportionally fewer non-white women than the SEER areas.
We observed no statistically significant associations between any of the outcomes and age at menarche, parity, age at first birth, oral contraceptive use, hysterectomy, or menopausal status at baseline (Table 1). No clear patterns of association emerged for age at first birth or parity when analyses were restricted to parous women (data not shown). Use of MHT, regardless of formulation, was also not significantly associated with plasma cell neoplasms, NHL overall, or NHL subtypes (Table 2). Risk estimates were unchanged after adjustment for race, menopausal status, oral contraceptive use, education, and smoking.
We evaluated associations with MHT formulations after stratifying on hysterectomy status (Tables (Tables33 and and4).4). We observed no statistically significant associations with plasma cell neoplasms, NHL overall, follicular lymphoma, or CLL/SLL. Ever-use and current use of unopposed estrogen were statistically significantly associated with DLBCL (RR=0.49, 95% CI, 0.25-0.96 and RR=0.47, 95% CI, 0.22-0.98, respectively). For estrogen plus progestin, ever-use, duration of use, recency of use, and regimen used generated null associations with plasma cell neoplasms, NHL overall, and NHL subtypes.
This large cohort of US women with detailed exposure information revealed null associations between lymphoid neoplasms and most reproductive factors, including menarche, parity, oral contraceptives, and menopausal status. Substantial changes in the types of available MHT and patterns of use over the last several decades27 complicate the comparisons with older studies of MHT and lymphoid neoplasm risk. Almost all MHT use consisted of unopposed estrogen therapy until the 1970s, when the discovery of increased endometrial cancer risks among women taking estrogen therapy led to sharp declines in MHT use.27 The combined estrogen-plus-progestin formulations emerged in the 1980s, and use of unopposed estrogen (among women with hysterectomy) and estrogen plus progestin (among women with an intact uterus) steadily increased until the early 2000s, when clinical trial data documented unfavorable risk-benefit profiles for MHT use among postmenopausal women.27 Few previous studies have considered MHT formulation or accounted for hysterectomy status, a strong predictor of unopposed estrogen use.13-21, 23, 26
Although our overall findings for MHT use were null, our risk estimates were generally less than one for all lymphoid neoplasm types, which is consistent with most of the previous literature on this topic. In analyses restricted to women with hysterectomy, we observed inverse associations between unopposed estrogen and NHL (RR=0.63, 95% CI, 0.29-1.37) and DLBCL (RR=0.49, 95% CI, 0.25-0.96). These results are most comparable with four previous studies that evaluated risk by MHT formulation20 or collected data when essentially all reported MHT use was unopposed estrogen (i.e., 1980s through the early 1990s).13, 15, 26 Of those studies, three also reported inverse associations with NHL,13, 20, 26 and the two that considered NHL subtypes also found significant associations for DLBCL.20, 26 In contrast, we and others did not observe the positive association between unopposed estrogen use and follicular lymphoma that was reported by Cerhan et al.15 Although some previous studies have observed somewhat stronger associations with current or long-duration use,15, 16, 18, 20, 21, 24 the absence of consistent statistical significance and dose-response by duration or recency argue against a true association or a simple, uncomplicated relationship between MHT and lymphoma. Nevertheless, the potential inverse association between unopposed estrogen use and DLBCL—which could also be due to chance—warrants further study. Our data on unopposed estrogen and the other main subtypes, plasma cell neoplasms, CLL/SLL, and Hodgkin's lymphoma, were too imprecise to draw firm conclusions about these potential associations.
In contrast to the unopposed estrogen data, our analysis of estrogen plus progestin use revealed null associations with plasma cell neoplasms, NHL overall, and specific NHL subtypes. These findings could be due to low statistical power, especially for the NHL subtypes. For the entire cohort of 134,074 women and the outcome of NHL overall, we had approximately 86% statistical power to detect a RR=0.7 for ever-use of unopposed estrogen and 57% power to detect a RR=0.7 for ever-use of estrogen plus progestin. For women with hysterectomy and NHL overall as the outcome, we had 56% power to detect a RR=0.7 for ever-use of unopposed estrogen. For women with an intact uterus and NHL overall as the outcome, we had 51% power to detect a RR=0.7 for ever-use of estrogen plus progestin. If our null associations with estrogen plus progestin are confirmed in future studies—to date, none of the other studies of NHL and MHT have reported data on estrogen plus progestin use among women with an intact uterus—then the potential contrast between inverse associations with unopposed estrogen and null associations with estrogen plus progestin would suggest possible distinct roles for estrogen and progestin in lymphomagenesis. This suggestion is particularly intriguing in light of evidence suggesting that estrogen, but not progestin, influences B-cell development and antibody production.2, 40
Previous data on menstrual and reproductive factors and risk of lymphoid neoplasms are inconsistent.4, 10, 11, 13, 14, 21, 22, 24, 25 Study design, participant demographics, and exposure ascertainment do not appear to explain the inconsistencies, as both significant and null associations have been reported in older and newer studies, among younger and older women, and in case-control and cohort designs. Although decreased risk of Hodgkin lymphoma has been associated somewhat more consistently with reproductive factors such as higher parity and earlier age at first birth,5-9, 11, 12, 18 we did not have a sufficient number of Hodgkin lymphoma cases in this older adult cohort for analysis. The inverse association we observed between DLBCL and increasing duration of oral contraceptive use might not be real because it was based on small numbers and not statistically significant. Nonetheless, the potential for two different exogenous hormones (i.e., oral contraceptives and ET) to decrease risk of DLBCL warrants further study.
Our study's strengths—e.g., large sample size, prospective design, and subtype-specific analyses—increase the validity of these results. Detailed data on specific formulations and regimens minimized misclassification of exogenous hormone use and allowed us to evaluate contemporary MHT use in detail, instead of relying on broad exposure categories that capture diverse patterns of MHT use over time. We lacked follow-up data on hormone therapy use after baseline, but we assume that many current users in 1996-1997 continued their use at least through mid-2002.32 Selection bias might particularly affect analyses of reproductive factors and lymphoma in previous case-control studies.41 Low response to our mailed questionnaires could theoretically produce similar bias, but this seems unlikely because response was not associated with reproductive factors.32 Some other limitations, such as small numbers of some NHL subtypes, affect most single studies of NHL.
Our primarily null associations between menstrual and reproductive factors, use of oral contraceptives, and specific detailed measures of MHT and precise NHL subtypes add substantial weight to the previous suggestions from other studies that reproductive factors and exogenous hormones are not generally strong risk factors for lymphoid neoplasms. Before the observed inverse associations between exogenous hormones and DLBCL can be considered indicative of hormonal exposures reducing risk for some lymphoid neoplasms, they should be replicated in sufficient detail—both for specific MHT exposures and NHL subtypes—across diverse study designs or in pooled analyses. Other hormone-related risk factors that might contribute to the observed sex differences in susceptibility to lymphoid neoplasms warrant continued consideration.
Cancer incidence data from the Atlanta metropolitan area were collected by the Georgia Center for Cancer Statistics, Department of Epidemiology, Rollins School of Public Health, Emory University. Cancer incidence data from California were collected by the California Department of Health Services, Cancer Surveillance Section. Cancer incidence data from the Detroit metropolitan area were collected by the Michigan Cancer Surveillance Program, Community Health Administration, State of Michigan. The Florida cancer incidence data used in this report were collected by the Florida Cancer Data System under contract to the Department of Health (DOH). The views expressed herein are solely those of the authors and do not necessarily reflect those of the contractor or DOH. Cancer incidence data from Louisiana were collected by the Louisiana Tumor Registry, Louisiana State University Medical Center in New Orleans. Cancer incidence data from New Jersey were collected by the New Jersey State Cancer Registry, Cancer Epidemiology Services, New Jersey State Department of Health and Senior Services. Cancer incidence data from North Carolina were collected by the North Carolina Central Cancer Registry. Cancer incidence data from Pennsylvania were supplied by the Division of Health Statistics and Research, Pennsylvania Department of Health, Harrisburg, Pennsylvania. The Pennsylvania Department of Health specifically disclaims responsibility for any analyses, interpretations or conclusions.
Supported by the Intramural Research Program of the National Cancer Institute, NIH, DHHS.