Maternal, prenatal and perinatal characteristics and first trimester maternal serum hormone concentrations.

In uncomplicated pregnancies, first trimester androgen, oestrogen and prolactin concentrations were higher in nulliparous (n=160) than parous (n=260) mothers. Androgens and estrogens were higher in younger than older mothers. These data are consistent with elevated hormone concentrations mediating the breast cancer protection from a first pregnancy and pregnancies occurring at younger ages.

Diet during adolescence and risk of breast cancer among young women.

BACKGROUND: A variety of breast cancer risk factors pertain to a woman's adolescence and may be related to nutritional influences. We assessed risk of early-onset breast cancer related to diet during adolescence in a case-control study. METHODS: Study participants were accrued from the following three geographical regions covered by cancer registries: Atlanta, GA; Seattle/Puget Sound, WA; and central New Jersey. Case patients (n = 1647) were newly diagnosed with breast cancer, and control subjects (n = 1501) were identified by random-digit-dialing techniques. In an interview, each subject was asked to recall the frequency of consumption and portion size of 29 key food items at ages 12-13 years. Mothers of a subset of respondents completed questionnaires, and food groups were recalculated after removal of foods with poor agreement between mother and daughter. Logistic regression analyses were used to calculate odds ratios and 95% confidence intervals. RESULTS: When high versus low quartiles of consumption were compared, there was a suggestion of a reduced risk associated with high consumption of fruits and vegetables, although this finding was not statistically significant. Slight increases (of borderline statistical significance) in risk of breast cancer were found for intake of chicken or high-fat meat. Intake of animal fat, high-fat foods, high-fat snacks and desserts, or dairy products during adolescence had no apparent influence on breast cancer risk. Removal of foods suspected to be poorly recalled by the daughters did not change any of the risk estimates. CONCLUSION: These data do not provide evidence for a strong influence of dietary intakes during adolescence on risk of early-onset breast cancer.

Case-control study of endogenous steroid hormones and endometrial cancer.

BACKGROUND: It has been suggested that identified risk factors for endometrial cancer operate through a single etiologic pathway, i.e., exposure to relatively high levels of unopposed estrogen (estrogen in the absence of progestins). Only a few studies, however, have addressed this issue directly. PURPOSE: We assessed the risk of developing endometrial cancer among both premenopausal and postmenopausal women in relation to the circulating levels of steroid hormones and sex hormone-binding globulin (SHBG). The independent effect of hormones was assessed after adjustment for other known risk factors. METHODS: The data used in the analysis are from a case-control study conducted in five geographic regions in the United States. Incident cases were newly diagnosed during the period from June 1, 1987, through May 15, 1990. The case patients, aged 20-74 years, were matched to control subjects by age, race, and geographic region. The community control subjects were obtained by random-digit-dialing procedures (for subjects 20-64 years old) and from files of the Health Care Financing Administration (for subjects > or = 65 years old). Additional control subjects who were having a hysterectomy performed for benign conditions were obtained from the participating centers. Women reporting use of exogenous estrogens or oral contraceptives within 6 months of interview were excluded, resulting in 68 case patients and 107 control subjects among premenopausal women and 208 case patients and 209 control subjects among postmenopausal women. The hormone analyses were performed on blood samples obtained from case patients or from hysterectomy control subjects before surgery. The odds ratios (ORs) and 95% confidence intervals (CIs) were estimated by use of an unconditional logistic regression analysis after we controlled for matching variables and potential confounders. All P values were two-sided. RESULTS: High circulating levels of androstenedione were associated with 3.6-fold and 2.8-fold increased risks among premenopausal and postmenopausal women, respectively, after adjustment for other factors (P for trend = .01 and < .001, respectively). Risks related to other hormone fractions varied by menopausal status. Among postmenopausal women, a reduced risk was associated with high SHBG levels and persisted after adjustment was made for obesity and other factors (OR = 0.51; 95% CI = 0.27-0.95). High estrone levels were associated with increased risk (OR = 3.8; 95% CI = 2.2-6.6), although adjustment for other risk factors (particularly body mass index) diminished the effect (OR = 2.2; 95% CI = 1.2-4.4). Albumin-bound estradiol (E2), a marker of the bioavailable fraction, also remained an important risk factor after adjustment was made for other factors (OR = 2.0; 95% CI = 1.0-3.9). In contrast, high concentrations of total, free, and albumin-bound E2 were unrelated to increased risk in premenopausal women. In both premenopausal and postmenopausal groups, risks associated with obesity and fat distribution were not affected by adjustment for hormones. CONCLUSION: High endogenous levels of unopposed estrogen are related to increased risk of endometrial cancer, but their independence from other risk factors is inconsistent with being a common underlying biologic pathway through which all risk factors for endometrial cancer operate. IMPLICATIONS: Further research should focus on alternative endocrinologic mechanisms for risk associated with obesity and body fat distribution and for the biologic relevance of the increased risk associated with androstenedione in both premenopausal and postmenopausal disease.

Reproducibility of laboratory assays for steroid hormones and sex hormone-binding globulin.

The relationship of serum hormones to cancer risk has recently been pursued in epidemiological studies, but few have reported on the reproducibility of laboratory findings. Prior to conducting a study of endogenous hormones and endometrial cancer, we evaluated the reproducibility of measurements for several hormones (estrone, estradiol, free estradiol, albumin-bound estradiol, and androstenedione) and sex hormone-binding globulin. We obtained a single unit of blood from each of six women and prepared aliquots of serum for repeated testing. Three laboratories analyzed multiple samples on consecutive working days from which estimates of intraassay and interassay measurement variability were obtained. For estrone and estradiol, a log transformation of the data produced distributions which were nearly normal and permitted the use of parametric statistical tests. In general, we found measurements for most hormones varied considerably between assays. Moreover, differences were observed in the absolute values of sex hormone-binding globulin and of the hormones, particularly for estrone and estradiol, from one laboratory to the next. Our findings suggest that variability of current laboratory procedures may hamper efforts to study the association between disease and endogenous hormones in epidemiological studies. In addition, validation of hormone assays is essential in order to assure standardized results and enable comparisons of data across studies.

A case-control study of serum folate levels and invasive cervical cancer.

Although small intervention trials have suggested that folate supplementation reduces cervical dysplasia, the association of blood folate concentrations with invasive cervical cancer risk has not been investigated in well-controlled epidemiological studies. A study was conducted with newly diagnosed Stage I and II invasive cervical cancer cases and controls in 4 Latin American countries. Ninety-five% of subjects donated blood samples, resulting in 330 case and 565 control serum samples analyzed for folate concentrations by radioassay. Cases did not differ significantly from controls in mean levels of folate (5.00 and 4.90 ng/ml, respectively). No associations were observed between quartiles of serum folate and risk of cervical cancer after adjustment for other risk factors, and no interactions with established risk factors were observed. Folate levels were also unrelated to risk among women who might have compromised folate status because of recent or extended oral contraceptive usage or multiple pregnancies. Further, mean levels of folate were similar by stage of disease, arguing against an effect of disease progression on serum values. These results do not support a role for serum folate in the etiology of invasive cervical cancer.

Causal criteria in nutritional epidemiology.

Making nutrition recommendations involves complex judgments about the balance between benefits and risks associated with a nutrient or food. Causal criteria are central features of such judgments but are not sufficient. Other scientific considerations include study designs, statistical tests, bias, confounding, and measurement issues. At a minimum, the set of criteria includes consistency, strength of association, dose response, plausibility, and temporality. The current practice, methods, and theory of causal inference permit flexibility in the choice of criteria, their relative priority, and the rules of inference assigned to them. Our approach is as follows. Consistency across study designs is compelling when the studies are of high quality and are not subject to biases. A statistically significant risk estimate with a > 20% increase or decrease in risk is considered a positive finding. A statistically significant linear or otherwise regularly increasing trend reinforces the judgment in favor of a recommendation. A plausible hypothesis likewise reinforces a recommendation, although the rules of inference for biological evidence are highly variable and depend on the situation. Temporality is, for nutrition recommendations, more a consideration of the extent to which a dietary factor affects disease onset or progression. Evidence supporting these criteria provides a strong basis for making a nutrition recommendation, given due consideration of the balance between presumed benefits and presumed harms. Recommendations should make clear their breadth of application; a narrow recommendation involves a single disease or condition whereas a broad recommendation involves all relevant diseases or conditions.

Evaluation of 2 brief instruments and a food-frequency questionnaire to estimate daily number of servings of fruit and vegetables.

BACKGROUND: Measurement of fruit and vegetable intake is important in the surveillance of populations and in epidemiologic studies that examine the relations between diet and disease. Some situations require the use of brief dietary assessment tools. OBJECTIVE: Our objective was to evaluate the performance of 2 brief dietary assessment instruments, a 7-item standard screener and a new 16-item screener, and a complete food-frequency questionnaire (FFQ) in measuring total fruit and vegetable consumption. DESIGN: About 800 men and women from the National Institutes of Health-AARP Diet and Health Study completed an FFQ, 1 of the 2 screeners, and two 24-h dietary recalls. Fruit and vegetable intakes as measured by each screener and the FFQ were compared with estimated true usual intake by using a measurement-error model. RESULTS: Median daily servings of fruit and vegetables were underestimated by both screeners. The estimated agreement between true intake and the screener was higher for the new screener than for the standard screener and was higher for women than for men. The estimated agreement between true intake and the FFQ was higher than that for both screeners. Attenuation coefficients for the FFQ and screeners were comparable. CONCLUSIONS: For estimating median intakes of fruit and vegetables and the prevalence of recommended intakes being met, the use of screeners without appropriate adjustment is suboptimal. For estimating relative risks in the relations between fruit and vegetable intake and disease, screeners and this FFQ are similar in performance.

Breast cancer and dietary and plasma concentrations of carotenoids and vitamin A.

A case-control study of breast cancer was conducted in Buffalo. Participants completed a food frequency questionnaire and donated a fasting blood sample before definitive workup for breast masses. Dietary and plasma concentrations of carotenoids and retinol for 83 women found to have breast cancer were compared with those of 113 women found to be free of breast cancer (control subjects). There were no case-control differences in dietary estimates of vitamin A intake or in plasma alpha-carotene and lycopene. However, subjects with breast cancer had lower concentrations of plasma beta-carotene than did control subjects (P = 0.02). There was no overall association between plasma retinol and breast cancer but a positive relationship was observed between retinol and breast cancer in the subgroup with low beta-carotene values. These results suggest that low plasma beta-carotene is associated with increased risk of breast cancer. Other studies will need to determine whether low carotene concentrations are a subtle effect of the disease or might be causally related to breast cancer.

Assay reproducibility of hormone measurements in postmenopausal women.

As part of a breast cancer case-control study of serum hormones conducted in Columbia, MO, we included several replicate quality control samples to monitor the consistency of laboratory assays. Sera were obtained from three postmenopausal women; from each woman, three samples were placed randomly in each of nine batches with the laboratory unaware of which sample corresponded to whom. Laboratory assays for estrone (E1), estradiol (E2), testosterone, androstenedione (Adione), E1SO4, dehydroepiandrosterone sulfate (DHEAS), follicle-stimulating hormone (FSH), sex hormone binding globulin (SHBG), and percentages of free and albumin-bound E2 were done at a single academic facility. ANOVA results showed that hormone values varied considerably from one batch to the next. The overall coefficients of variation (CVs) estimated for E2, percentage of unbound E2, and percentage of albumin-bound E2 were higher than 15%, but of these, only percentage of unbound E2 had both inter- and intra-assay CVs greater than 10%. Intraclass correlations (ICC) for FSH, SHBG, and DHEAS were high, suggesting that these assays are suitable for population-based studies attempting to link hormone levels to disease risk. The ICC estimated for E1SO4 was quite low due to aberrant values reported in a single batch. For the remaining hormones, the ICCs were fair (ranging from 47% for albumin-bound E2 to 67% for Adione), and studies using these assays would require a substantial increase in the sample size to detect small case-control differences.

Effects of breast cancer treatments on plasma nutrient levels: implications for epidemiological studies.

The interpretation of case-control studies in which blood nutrient levels are examined as etiological factors in cancer is complicated by the possibility that either the disease or its treatment may alter these levels. Circulating levels of selected nutrients were examined prior to diagnostic biopsy and compared with levels 3 to 4 months after diagnosis among 71 women with breast cancer and 95 women with benign breast disease. Among women with benign breast disease or women with breast cancer who were not given postsurgical adjuvant drug therapy, levels of alpha-carotene, lycopene, alpha-tocopherol, cholesterol, and triglycerides did not change over time. In contrast, women who received chemotherapy had increased levels of cholesterol, retinol, and alpha- and gamma-tocopherol, and women on antiestrogen therapy showed increased levels of triglycerides and alpha-tocopherol. Overall, the concentrations of carotenoids (lycopene, alpha-carotene, and beta-carotene) did not change in breast cancer cases, although subgroup analyses showed increased levels of beta-carotene among cases not receiving drug treatment and decreased levels among those receiving antiestrogens. In summary, blood levels of some nutrients did not appear to be affected by breast cancer or its treatments, but changes were noted for levels of plasma lipids, tocopherols, retinol, and beta-carotene. Those investigating the etiological relationship between breast cancer and circulating nutrients need to consider these effects in designing and interpreting epidemiological studies.

Endometrial cancer risk in relation to serum lipids and lipoprotein levels.

Blood lipids are useful biochemical indicators for assessing the risk of a number of chronic diseases, particularly those associated with obesity. In a multicenter case-control study that included 256 cases and 185 controls less than 75 years old, we studied the risk of endometrial cancer in relation to serum cholesterol, low density lipoprotein cholesterol, high density lipoprotein cholesterol, and triglycerides. Contrary to expectation, blood lipids were, in general, lower among cases compared with controls. The effects of low blood lipids, specifically cholesterol and low density lipoprotein cholesterol, were limited to older women (> or = 55 years). Risk of the disease in this subgroup of 177 cases and 110 controls was increased 3-4-fold among those with the lowest cholesterol or low density lipoprotein cholesterol values. For example, after adjustment for age, education, smoking status, obesity, and body fat distribution, the relative risks of endometrial cancer across decreasing quartiles of serum cholesterol were 1.0, 2.5, 2.4, and 4.2 (P for trend < 0.01). We examined blood lipid levels by disease stage. The low lipid values of older cases did not appear to be a consequence of the disease. While we cannot rule out the possibility that hypocholesterolemia is a predisposing factor for endometrial cancer, there is no obvious biological explanation for the inverse association.

Relation of endometrial cancer risk to past and contemporary body size and body fat distribution.

In a multicenter case-control study that included 403 cases and 297 controls, we examined the relation of past and contemporary body size, including body fat distribution, to the risk of endometrial cancer. The relative contributions of past and contemporary body size were assessed by examining weight and height histories provided by the subjects. Anthropometric indicators thought to reflect early environmental influences (e.g., height and sitting height), current weight, and fat distribution patterns were measured directly. Height was not a risk factor for endometrial cancer, but inexplicably, sitting height was inversely associated with risk. Weight during early adulthood appeared to be directly related to disease risk, but the association was explained by contemporary weight and thus weight gain during adulthood. While contemporary weight was associated with risk of endometrial cancer, the effect was restricted to those in the top quartile. Women whose measured weight at interview exceeded 78 kg had 2.3 times the risk of those weighing less than 58 kg (95% confidence interval, 1.4 to 3.7). Upper-body obesity (waist-to-thigh circumference ratio) was a risk factor independent of body weight. After adjustment for weight, the relative risks of endometrial cancer across increasing quartiles of upper-body obesity were 1.0, 1.5, 1.8, and 2.6 (P for trend < 0.001). These data indicate that both obesity and the distribution of adipose tissue accumulated during adult life increase endometrial cancer risk substantially.

Diet and cancer: future etiologic research.

In light of several credible diet and cancer hypotheses, we suggest strategies for advancing our understanding in this area. Two conceptual approaches can be taken in defining dietary exposure: the decompositional approach focuses on specific nutrients and other chemical constituents of food, whereas the integrative approach emphasizes the action of whole foods or food patterns (cuisines). Diet-cancer hypotheses can be organized according to this conceptual framework. We review four types of scientific investigation available to us for advancing the diet and cancer field: metabolic (clinical nutrition) studies; animal studies; observational epidemiologic investigations; and clinical trials. Each of these designs has its strengths and limitations. Observational epidemiologic studies and trials have the particular advantage of examining explicit cancer end points in humans. Results from metabolic and animal research, however, can complement the findings from epidemiologic studies and trials. Finally, we briefly review strategies for evaluating promising hypotheses linking diet to cancers of the large bowel, lung, breast, and prostate.

Workshop to identify critical windows of exposure for children's health: cancer work group summary.

We considered whether there are discrete windows of vulnerability in the development of cancer and which time periods may be of the greatest importance. Cancer was considered broadly, including cancers in childhood as well as adult cancers that may have an in utero or childhood origin. We concluded that there was evidence from animal and epidemiologic studies for causal relationships for preconceptional, in utero, and childhood exposures and cancer occurrence in children and adults. However, the evidence is incomplete and all relevant critical windows may not have been identified. The comprehensive evaluation of the relative importance of specific time windows of exposure is limited. Improvements in the design of epidemiologic studies and additional animal studies of mechanisms are warranted.

Breast cancer risk among women under 55 years of age by joint effects of usage of oral contraceptives and hormone replacement therapy.

OBJECTIVE: To assess effects on breast cancer risk of exposure to both oral contraceptives and menopausal hormones, an increasingly common exposure. DESIGN: A case-control study of breast cancer among women under the age of 55 years in Atlanta, GA involving 1,031 cases and 919 population controls was conducted. RESULTS: Ever use of oral contraceptives was associated with a relative risk of 1.1 (95% 0.9-1.4), whereas the relative risk for hormone replacement therapy was 0.9 (95% CI 0.7-1.2). Seventeen percent of the cases versus 19% of the population controls reported exposure to both agents, resulting in a relative risk of 1.0 (95% CI 0.7-1.4) relative to those unexposed to either preparation. Although there was little variation in risk associated with joint effects by either age or race, there were statistically nonsignificant elevations in risk for this exposure among women who had experienced a natural menopause (relative risk = 2.0, 95% CI 0.7-5.6), were relatively thin (relative risk = 1.5, 0.8-3.0), or who had a first degree relative with breast cancer (relative risk = 2.0, 0.6-7.0). When joint effects of longer term use of both agents were considered, subjects who reported use of oral contraceptives for 10 or more years and hormone replacement for 3 or more years had a relative risk of 3.2 (95% CI 1.4-7.4) compared with nonusers of either preparation. CONCLUSIONS: Although our results must be cautiously interpreted given small numbers within subgroups, they raise concern and emphasize the need for further evaluation on breast cancer risk of the increasingly common exposure to both oral contraceptives and hormone replacement therapy.

Breast cancer risk in young women and history of selected medical conditions.

BACKGROUND: Several common medical conditions are associated with altered hormone levels, and may thus plausibly influence breast cancer risk. Few studies have examined such relationships, and we utilized a population-based case-control study of young women in the US to examine breast cancer risk following a history of various medical conditions. Relationships between breast cancer and each medical condition examined are biologically plausible, and relevant in terms of public health. METHODS: The study included 2173 breast cancer cases and 1990 population-based controls from three areas of the US, under 55 years, who were administered a questionnaire including details of physician-diagnosed medical conditions. RESULTS: No significantly increased or decreased breast cancer risk was associated with a history of thyroid disease, gallbladder disease, colorectal polyps, diabetes, high blood pressure, high cholesterol or surgery for endometriosis. There was some evidence of an increased breast cancer risk associated with ovarian cysts among women who did not receive an oophorectomy (relative risk [RR] = 1.94, 95% CI: 1.0-3.9). Non-significant increases in breast cancer risk were observed following diagnoses of several other cancers, including thyroid cancer, basal cell carcinoma, Hodgkin's disease and malignant melanoma. CONCLUSIONS: To conclude, our generally null results from this large, population-based study support results from previous studies in providing reassurance that women with a history of several common medical conditions do not appear to be at an increased risk of breast cancer at a young age.

Characteristics of respondents and non-respondents from a case-control study of breast cancer in younger women.

BACKGROUND: This study assessed the nature of potential biases by comparing respondents with non-respondents from a case-control study of breast cancer in younger women. METHODS: The case-control study was conducted in three regions in the US: Atlanta GA, Seattle/Puget Sound WA, and central New Jersey. An abbreviated interview or mailed questionnaire was completed by willing non-respondents, most of whom had refused participation in the main study. RESULTS: Respondents and non-respondents appeared similar with respect to age, race, relative weight, smoking, family history of breast cancer, number of births, age at first birth, and several dietary items. Compared to non-respondents, case and control respondents were of shorter stature, and reported less frequent consumption of doughnuts/pastries. Respondent cases, compared with non-respondent cases, were more highly educated and more likely to have consumed alcohol regularly; similar but not statistically significant tendencies were observed for controls. Respondent cases experienced menarche earlier than non-respondents. Respondent controls were more likely to have used oral contraceptives than non-respondents; a similar but not statistically significant tendency was observed in cases. Comparisons of crude and simulated relative risks using available non-respondents' data generally showed a low impact of non-response on relative risks in this study. CONCLUSIONS: Our results suggest that non-response would not greatly affect relative risk estimates in this study, except possibly regarding height. However, we were limited by the numbers of informative non-respondents and the amount of data collected. Collecting similar information in future studies would be useful, especially since varying methods used to encourage participation may lead to differences in respondents' characteristics.

Improving food frequency questionnaires: a qualitative approach using cognitive interviewing.

In an attempt to improve data quality and ease of administration of standard self-administered food frequency questionnaires, various alternative approaches were tried for inquiring about frequency of consumption, portion size, seasonal intake, and food preparation. Evaluation consisted of a cognitive interviewing method in which respondents verbalize their thought process while completing several variations of a questionnaire. Interviewers observed and asked follow-up probe questions to evaluate problems or inconsistencies verbalized by respondents. Consensus and judgment by interviewers and observers suggested several problematic features of food frequency questionnaires: formatting of questions about frequency and portion size; computing average frequencies for aggregated food items or for foods eaten seasonally; comprehension of many items; and ordering of foods. These findings led to cognitive refinement and innovations, which included detailed questions regarding preparation or use of low-fat varieties or other alternatives to help better describe specifics of intake for some foods; questions on seasonal intake for several foods; inclusion of portion size ranges; and additional response categories for frequency of intake. Cognitive interviewing is an important step in pinpointing cognitive problems in dietary questionnaires.

Perinatal characteristics and breast cancer risk in daughters: a Scandinavian population-based study.

The in utero origins of breast cancer are an increasing focus of research. However, the long time period between exposure and disease diagnosis, and the lack of standardized perinatal data collection makes this research challenging. We assessed perinatal factors, as proxies for in utero exposures, and breast cancer risk using pooled, population-based birth and cancer registry data. Birth registries provided information on perinatal exposures. Cases were females born in Norway, Sweden or Denmark who were subsequently diagnosed with primary, invasive breast cancer (n = 1419). Ten controls for each case were selected from the birth registries matched on country and birth year (n = 14,190). Relative risks (RRs) and 95% confidence intervals (CIs) were estimated using unconditional regression models. Breast cancer risk rose 7% (95% CI 2-13%) with every 500 g (roughly 1 s.d.) increase in birth weight and 7% for every 1 s.d. increase in birth length (95% CI 1-14%). The association with birth length was attenuated after adjustment for birth weight, while the increase in risk with birth weight remained with adjustment for birth length. Ponderal index and small- and large-for-gestational-age status were not better predictors of risk than either weight or length alone. Risk was not associated with maternal education or age, gestational duration, delivery type or birth order, or with several pregnancy complications, including preeclampsia. These data confirm the positive association between birth weight and breast cancer risk. Other pregnancy characteristics, including complications such as preeclampsia, do not appear to be involved in later breast carcinogenesis in young women.

Exposures in early life: associations with DNA promoter methylation in breast tumors.

There is evidence that epigenetic changes occur early in breast carcinogenesis. We hypothesized that early-life exposures associated with breast cancer would be associated with epigenetic alterations in breast tumors. In particular, we examined DNA methylation patterns in breast tumors in association with several early-life exposures in a population-based case-control study. Promoter methylation of E-cadherin, p16 and RAR-ß2 genes was assessed in archived tumor blocks from 803 cases with real-time methylation-specific PCR. Unconditional logistic regression was used for case-case comparisons of those with and without promoter methylation. We found no differences in the prevalence of DNA methylation of the individual genes by age at menarche, age at first live birth and weight at age 20. In case-case comparisons of premenopausal breast cancer, lower birth weight was associated with increased likelihood of E-cadherin promoter methylation (OR = 2.79, 95% CI, 1.15-6.82, for ⩽2.5 v. 2.6-2.9 kg); higher adult height with RAR-ß2 methylation (OR = 3.34, 95% CI, 1.19-9.39, for ⩾1.65 v. <1.60 m); and not having been breastfed with p16 methylation (OR = 2.75, 95% CI, 1.14-6.62). Among postmenopausal breast cancers, birth order was associated with increased likelihood of p16 promoter methylation. Being other than first in the birth order was inversely associated with likelihood of ⩾1 of the three genes being methylated for premenopausal breast cancers, but positively associated with methylation in postmenopausal women. These results suggest that there may be alterations in methylation associated with early-life exposures that persist into adulthood and affect breast cancer risk.