Abstract
Background:
A high body mass index (BMI, kg/m2) is associated with decreased risk of breast cancer before menopause, but increased risk after menopause. Exactly when this reversal occurs in relation to menopause is unclear. Locating that change point could provide insight into the role of adiposity in breast cancer etiology.
Methods:
We examined the association between BMI and breast cancer risk in the Premenopausal Breast Cancer Collaborative Group, from age 45 up to breast cancer diagnosis, loss to follow-up, death, or age 55, whichever came first. Analyses included 609,880 women in 16 prospective studies, including 9,956 who developed breast cancer before age 55. We fitted three BMI hazard ratio (HR) models over age-time: constant, linear, or nonlinear (via splines), applying piecewise exponential additive mixed models, with age as the primary time scale. We divided person-time into four strata: premenopause; postmenopause due to natural menopause; postmenopause because of interventional loss of ovarian function (bilateral oophorectomy (BO) or chemotherapy); postmenopause due to hysterectomy without BO. Sensitivity analyses included stratifying by BMI in young adulthood, or excluding women using menopausal hormone therapy.
Results:
The constant BMI HR model provided the best fit for all four menopausal status groups. Under this model, the estimated association between a five-unit increment in BMI and breast cancer risk was HR=0.87 (95% CI: 0.85, 0.89) before menopause, HR=1.00 (95% CI: 0.96, 1.04) after natural menopause, HR=0.99 (95% CI: 0.93, 1.05) after interventional loss of ovarian function, and HR=0.88 (95% CI: 0.76, 1.02) after hysterectomy without BO.
Conclusion:
The BMI breast cancer HRs remained less than or near one during the 45-55 year age range indicating that the transition to a positive association between BMI and risk occurs after age 55.
Keywords: Middle Aged, Postmenopause, Premenopause, Body Mass Index, Breast Neoplasms
1. Introduction
Direction of associations between body mass index (BMI, kg/m2) and breast cancer risk depends on menopausal status1. A wide body of evidence supports an inverse association between BMI and breast cancer risk before menopause2-7. Higher young adult BMI is a protective factor for breast cancer incidence6,8 and may have the potential to modify patterns in estimated BMI HRs later in life. Following menopause, this inverse association changes to a positive association5,9-12. With this evidence prior to and after menopause, along this continuum on an age scale, it is clear that the association between BMI and breast cancer risk changes from a protective to a deleterious association. Although potential biological mechanisms have been proposed to explain the difference in associations between BMI and breast cancer by menopausal status1, the time-course for this reversal has not been well described, particularly in relation to type of menopause.
Including menopausal status as an effect measure modifier of the association beween BMI and breast cancer risk is one analytic approach to model associations between BMI and breast cancer risk dependent on menopause. Evaluation of these associations can entail inclusion of a product term between a continuous constant BMI variable and a binary menopause variable in a regression model13. These types of analyses assume a constant hazard ratio (HR) over time allowing for an abrupt shift at menopause, but age-time dependent changes in these associations remain unknown and may be more gradual or vary by type of menopause. Describing changes in associations between BMI and breast cancer risk around the typical age at menopausal transition may provide a better understanding of these temporal patterns.
Given the lack of evidence assessing changes in the associations between BMI and breast cancer risk around ages of menopausal transition, our primary aim was to describe patterns of change over age-time in the association between BMI and breast cancer risk from 45 to 55 years. We examined the change in the associations between BMI and breast cancer risk across the age-time scale by menopausal status using a large consortium-based pooled sample from 16 cohorts.
2. Methods
2.1. Study sample
The Premenopausal Breast Cancer Collaborative Group (PBCCG) includes over 1 million participants from 23 cohorts meeting eligibility requirements of at least 100 female breast cancer diagnoses during follow-up before age 55 years and data collection occurring at least at two time points14. To accommodate our primary aim of estimating breast cancer incidence in relation to an age-time-dependent menopause status variable, we included the 16 cohorts with participants reporting both age at menopause and a cause of menopause after the following exclusions. We excluded women: (i) without at least one BMI measurement (n=13,367); (ii) postmenopausal women with missing age at or cause of menopause (n=111,117); (iii) postmenopausal women with no BMI ascertained after menopause (n=512); (iv) no BMI ascertained between the ages of 40-55 years (n=138,005); (v) women from cohorts with fewer than 20 breast cancer diagnoses (n=155,813); (vi) women with no person-time between ages 45 and 55 years (n=17,279); and (vii) women with implausible BMI values (n=40). With these exclusions, 609,880 participants remained in the analysis (Supplemental Figure 1).
Approval from institutional review boards and individual consent for all cohorts in the PBCCG conformed to each study’s ethics review requirements.
2.2. Body Mass Index
We included BMI (in kg/m2) in the models as a continuous variable, with results reported per 5-unit increment. BMI values ascertained included BMI reported at study entry, which we updated for each age-time interval with any follow-up BMI to create a time-dependent covariate. If there was no BMI reported for a particular person-year, we used BMI from the preceding person-year for the individual. For example, if a person did not report any new BMI during the follow up period, their BMI at entry to their age-time period was used for each subsequent age-time interval.
The interval between ages 45 and 55 was of most interest because the median age at menopause is around age 51 years 15-17. We included person-time based on BMI most proximal to this time range dependent on menopausal status. For premenopausal person-time, we included BMI reported at most five years prior to this interval. For analyses including postmenopausal person-time, we used BMI reported within three years prior to the age at menopause. For the 609,880 women in the study sample, there was a median (interquartile range) of 2 (1,3) BMI observations.
2.3. Statistical Methods
We used frequencies and percentages to describe categorical variables, and medians and interquartile ranges to describe the continuous variables for sample characteristics.
To estimate breast cancer HRs for BMI as continuous time-dependent variables by menopausal status, we used piecewise exponential additive mixed modeling (PAMM) 18 with age as the time scale. This method provides estimates similar to Cox proportional hazards models, and it is more versatile in specifying smooth age-time-varying coefficient estimates with cohort strata. In brief, the simpler piecewise exponential model (PEM) enabled us to divide person-time into year intervals on the age scale and estimate a constant hazard within each interval using a Poisson-likelihood method. The PAMM model extends the PEM with an additive component to allow for more complex smooth time-varying effects of covariates and the underlying baseline hazard. The PAMM approach allows estimation of 95% confidence intervals (CIs) for models including splines of time-varying covariates and coefficients in time-to-event analyses along with cohort strata, as we specified in our models.
Age was the primary time scale to estimate three time-related BMI breast HRs: constant, linear change over age-time, or nonlinear (via splines) change. We specified the best fitting model after evaluating the Akaike's Information Criterion (AIC) and residual deviance with chi-square tests comparing the three models mentioned above.
We evaluated heterogeneity of our estimates across cohorts before pooling the data 14. To do this, we estimated cohort-specific estimates and estimated the I2 term from a meta-analysis to evaluate heterogeneity of estimates across studies 19.
Sensitivity analyses included three different types of analyses. A first set of sensitivity analyses included estimation of BMI-breast cancer HRs after stratifying by overweight status (BMI≥25 kg/m2) at age 18-24 years. A second set of sensitivity analyses included assessing associations in a subgroup of women who reported never using menopausal hormone therapy (MHT) at baseline, because such use could cause inaccuracies in the reported age at menopause and consequently influence the estimated patterns of BMI HR over age-time. Also, the relationship between BMI and postmenopausal breast cancer may depend on MHT20. Sample size for this group only allowed estimates for the natural and premenopausal groups of women. Both overweight status at age 18-24 and MHT variables were not available across all 16 cohorts, and subsets were analyzed for each of these sensitivity analyses. Last, we changed the time scale for the postmenopausal person time to time since menopause instead of an age scale and adjusted for age (in years) at baseline.
2.4. Menopausal Status
All analyses were stratified into menopausal groups: 1) premenopausal, 2) natural menopause, 3) medically induced loss of ovarian function and 4) hysterectomy without bilateral oophorectomy (BO). The “loss of ovarian function” group (group 3) mainly included women reporting BO as well as chemotherapy. For the fourth group, the timing of hormonal menopause is not known, and we instead used person-time starting at the hysterectomy without BO. In the analyses, we divided person-time, as defined above, according to the four menopausal groups and fitted those person-times in separate models between 45 and 55 years of age. For example, a woman who experienced natural menopause after study entry had person-time that was partitioned into two separate models: 1) the premenopausal person-time, from her study entry until one year following the reported age at last menstrual period, and 2) the natural menopause person-time, which includes time from one year after cessation of menstruation until either an event or censoring, whichever occurred first. Menopausal status was lagged by one year for all participants to account for menopause occurrence due to breast cancer treatments in year of diagnosis6.
We used R software21, version 4.0.2, to manage the harmonized data from the PBCCG and conduct all statistical analyses.
3. Results
We used person-time in years from 609,880 women including incident breast cancer between ages 45-55 years and followed for a median (IQR) of 5.0 (4.0, 10.0) years from a total of 16 cohorts in the PBCCG (Table 1, Supplemental Table 1). The median age at entry (IQR) was 42.6 (37.0, 47.7) years. Most women self-identified as white or of European ancestry (n=391,561, 82.5%). Within the 45-55 year age interval, a total of 9,956 developed incident breast cancer, 387,623 women were postmenopausal at entry, and 183,744 reported menopause during follow-up.
Table 1:
Descriptive statistics for analytic sample
| Overall | |
|---|---|
| Number of women | 609,880 |
| Age at menopause (at baseline) | 47.0 [42.0, 50.0] |
| Participant age at entry | 42.6 [37.0, 47.7] |
| Age at menarche | 13.0 [12.0, 13.0] |
| Follow-up time (years) between age 45-55 years | 5.0 [4.0, 10.0] |
| BMI | 23.3 [21.2, 26.6] |
| BMI, 18-24 years | 20.8 [19.3, 22.7] |
| BMI>=25 kg/m2, 18-24 years | 46,579 (10.8) |
| Cause of menopause, baseline | |
| Hysterectomy without bilateral oophorectomy | 35,525 (9.2) |
| Medically induced loss of ovarian function | 70,432 (18.2) |
| Natural | 281,666 (72.7) |
| Age at first birth | |
| Nulliparous | 121,033 (19.8) |
| ≤20 | 65,983 (10.8) |
| 21-24 | 172,932 (28.4) |
| 25-29 | 174,825 (28.7) |
| 30-55 | 75,107 (12.3) |
| Ethnicity | |
| African ancestry | 58,132 (12.2) |
| Asian | 17,430 (3.7) |
| European ancestry | 391,561 (82.5) |
| Other | 7,714 (1.6) |
| Menopausal hormone therapy (at baseline) | 53,174 (21.7) |
| Breast Cancers | 9,956 (1.6) |
Note:
All age variables are in year units, continuous variables are characterized by median
[interquartile range], and categorical variables are characterized by number of observations (percent)
In assessing change in the association between BMI and breast cancer risk over age-time across the cohorts we did not find strong heterogeneity across samples for any of the menopausal groups, which supported the validity of pooling data from all the cohorts (Supplemental Figure 2). Similarly, the between-cohort heterogeneity in constant BMI HR estimates was low enough to allow pooling of the data across cohorts (Supplemental Figure 3).
After evaluating the three models to assess the BMI HR change over age-time between 45 and 55 years of age for the four menopause groups, we found no evidence to support either a linear or non-linear change in the BMI HR over age-time compared with a constant BMI HR model for any of the menopause categories (Table 2, Supplemental Table 2, and Supplemental Figure 4). The estimated ratio of BMI-breast cancer HRs for a one-year change in age-time was close to one for all four menopausal groups suggesting little change in BMI breast cancer HRs over this age interval. Similarly, the estimated BMI breast cancer HRs at ages 45, 50, and 54 years in the nonlinear over age-time spline model indicated BMI breast cancer HRs close to the HR estimated from the constant HR model (Table 2). The constant BMI (per 5 kg/m2 increment) HR (95% CI) was 0.87 (0.85, 0.89) for the premenopausal group, 0.88 (0.76, 1.02) for the hysterectomy without bilateral oophorectomy group, 1.00 (0.96, 1.04) for the natural menopause group, 0.99 (0.93, 1.05) for the loss of ovarian function group (Table 2). Comparing models with and without the constant BMI term, we found no evidence for a BMI HR that differed from 1 for the postmenopausal groups, suggesting a null association between BMI and breast cancer risk for postmenopausal women under age 55 years.
Table 2:
Breast cancer hazard ratiosa (HRs) per 5-unit change in BMI by model type
| Model 1: Constant BMI |
Model 2: Linear BMI | Model 3: HR estimates based on spline model |
||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Menopause status |
n, cases |
person- years |
HR (95% CI) |
p- valueb |
linear HR intercept termc (95% CI) |
linear HR slope termd (95% CI) |
p-valuee | HR at age 45 |
HR at age 50 |
HR at age 54 |
| Premeno pausal | 6,886 | 2,541,743 | 0.87 (0.85,0.89) | 0 | 0.86 (0.83,0.90) | 1.00 (0.99,1.01) | 0.079 | 0.86 (0.82,0.90) | 0.87 (0.85,0.90) | 0.88 (0.83,0.94) |
| Hysterectomy without | 184 | 57,467 | 0.88 (0.76,1.02) | 0.087 | 0.93 (0.64,1.35) | 0.99 (0.94,1.05) | 0.773 | 0.93 (0.63,1.36) | 0.89 (0.76,1.05) | 0.86 (0.69,1.07) |
| Bilateral oophorectomy | 2,058 | 818,476 | 1.00 (0.96,1.04) | 1 | 0.93 (0.79,1.10) | 1.01 (0.99,1.03) | 0.242 | 0.93 (0.79,1.10) | 0.98 (0.92,1.04) | 1.02 (0.96,1.08) |
| Natural loss of ovarian function | 828 | 367,675 | 0.99 (0.93,1.05) | 1 | 0.98 (0.84,1.13) | 1.00 (0.98,1.03) | 1 | 0.98 (0.84,1.14) | 0.98 (0.92,1.05) | 0.99 (0.90,1.09) |
All analyses are stratified by cohort.
p-value for Chi-square test for difference from null model (no BMI term).
HR at age 45 years
Ratio of HR for a one-year change in age-time.
p-value for Chi-square test for difference from constant BMI model.
We conducted sensitivity analyses to determine any changes in the interpretation of our findings. The first and second set of analyses either stratified by young adult overweight status or restricting the sample to women who did not report ever taking menopausal hormone therapy at study entry. Both the first and second set of sensitivity analyses included a subset of cohorts from the primary sample that had information on either early life BMI or menopausal hormone therapy, and the HRs for the subsets were slightly different from those from the primary analyses. Last, we evaluated models with time since menopause as the time scale.
In analyses stratifying by young adult overweight status (BMI>=25 kg/m2) at 18-24 years of age (Table 3), the best fitting model remained the constant BMI breast cancer HR. For women reporting overweight at ages 18-24 years, the BMI breast cancer HR (95% CI) estimate was 0.75 (0.45, 1.22) for the hysterectomy without BO group, 1.07 (0.93, 1.23) for the natural menopause group, and 1.10 (0.93, 1.31) for the loss of ovarian function menopause group. Among women not reporting overweight at ages 18-24 years, the estimates were similar. Except for the natural menopause group who were not overweight, we found no evidence of the constant BMI breast cancer HRs differing from the ones for the postmenopausal groups in either strata. The BMI HR for the premenopausal group of women not reporting overweight in young adulthood suggested a protective relationship that was similar to that for the group of women reporting overweight in young adulthood.
Table 3:
Breast cancer hazard ratiosa (HRs) per 5-unit change in BMI by overweight (BMI>=25 kg/m2) strata at age 18-24 years
| Overweight at ages 18-24 years | Not overweight at ages 18-24 years | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Model 1: Constant BMI |
Model 2: Linear BMI |
Model 1: Constant BMI |
Model 2: Linear BMI |
|||||||||||
| Menopause status |
n, cases |
person- years |
HR (95% CI) |
p- valueb |
linear HR interceptc (95% CI) |
linear HR termd (95 % CI) |
p- valuee |
n, cases |
person- | HR (95% CI) |
p- valueb |
linear HR interceptc (95 % CI) |
linear HR termd (95 % CI) |
p- valuee |
| Premeno pausal | 351 | 196,081 | 0.94 (0.87,1.02) | 0.116 | 0.88 (0.77,1.01) | 1.02 (0.99,1.05) | 0.229 | 4,381 | 1,689,521 | 0.92 (0.88,0.95) | <0.001 | 0.91 (0.86,0.97) | 1.00 (0.99,1.02) | 0.487 |
| Hysterectomy without bilateral oophorectomy | 14 | 6,392 | 0.75 (0.45,1.22) | 0.228 | 1.49 (0.36,6.22) | 0.90 (0.73,1.11) | 0.351 | 165 | 49,943 | 0.94 (0.79,1.13) | 0.503 | 0.97 (0.63,1.51) | 0.99 (0.93,1.06) | 0.872 |
| Natural | 93 | 57,008 | 1.07 (0.93,1.23) | 0.336 | 1.29 (0.76,2.20) | 0.97 (0.91,1.05) | 0.477 | 1,092 | 468,214 | 1.06 (1.00,1.13) | 0.364 | 1.00 (0.81,1.23) | 1.01 (0.98,1.04) | 0.525 |
| Loss of ovarian function | 63 | 33,265 | 1.10 (0.93,1.31) | 0.273 | 0.91 (0.58,1.45) | 1.03 (0.96,1.10) | 0.391 | 521 | 249,676 | 0.97 (0.89,1.06) | 0.997 | 0.91 (0.73,1.14) | 1.01 (0.98,1.04) | 0.525 |
All analyses are stratified by cohort.
p-value for Chi-square test for difference from null model (no BMI term).
HR at age 45 years
Ratio of HR for a one-year change in age-time.
p-value for Chi-square test for difference from constant BMI model.
Note: The cohorts in this analytic sample were: Breakthrough Generations Study, California Teachers Study, Canadian Study of Diet, lifestyle and Health, Clue II, EPIC, Melbourne Collaborative Cohort Study, Nurses’ Health Study I, Nurses’ Health Study II, Southern Cmmunity Cohort, US Radiologic Technologists Cohort, Women’s Lifestyle and Health Study (Sweden) , Black Women’s Health Study
When we restricted analyses to women who, at the start of the 45-55 year interval, reported never using menopausal hormone therapy, the models with a constant BMI HR over age-time remained the best fitting (Table 4). For the natural menopause group, estimates for the never use group were similar to ones using the full sample. As in the primary analyses, the premenopausal group not reporting menopausal hormone therapy use had an inverse HR (0.89 (95% CI: 0.85, 0.93)). In the last set of sensitivity analyses, the best fitting model was the constant BMI breast cancer HR over time since menopause (Supplemental Table 3). These results did not substantively change from the primary analyses using age as the time scale.
Table 4:
Breast cancer HRsa per 5-unit change in BMI by menopausal hormone therapy status
| Cohorts with information on HRT use | Restricted to no menopausal hormone therapy | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Model 1: Constant BMI |
Model 2: Linear BMI |
Model 1: Constant BMI |
Model 2: Linear BMI |
|||||||||||
| Menopause status |
n, cases |
person- years |
HR (9 5 % CI) |
p- valueb |
linear HR intercept termc (95 % CI) |
linear HR termd (95 % CI) |
p- valuee |
n, cases |
person- years |
HR (9 5 % CI) |
p- valueb |
linear HR intercept termc (95 %CI) |
linear HR termd (95 % CI) |
p- valuee |
| Premeno pausal | 3,188 | 1,264,645 | 0.87 (0.84,0.91) | 0 | 0.88 (0.83,0.94) | 1.00 (0.98,1.01) | 0.005 | 1,662 | 573,545 | 0.89 (0.85,0.93) | <0.001 | 0.92 (0.85,0.99) | 0.99 (0.97,1.01) | 0.644 |
| Hysterec tomy without bilateral oophorectomy | 153 | 52,691 | 0.87 (0.74,1.03) | 0.671 | 0.97 (0.57,1.65) | 0.98 (0.91,1.06) | 0.756 | |||||||
| Natural | 902 | 417,231 | 1.05 (0.99,1.12) | 0.992 | 1.05 (0.83,1.33) | 1.00 (0.97,1.03) | 0.986 | 420 | 207,149 | 1.09 (1.01,1.18) | 0.103 | 1.00 (0.73,1.38) | 1.01 (0.97,1.06) | 0.019 |
| Loss of ovarian function | 473 | 211,834 | 1.05 (0.97,1.14) | 0.194 | 1.07 (0.89,1.29) | 1.00 (0.97,1.03) | 0.924 | |||||||
All analyses are stratified by cohort.
p-value for Chi-square test for difference from null model (no BMI term).
HR at age 45 years
Ratio of HR for a one-year change in age-time.
p-value for Chi-square test for difference from constant BMI model.
Insufficient sample size for estimates restricting to no menopausal hormone therapy
Note: The cohorts in this analytic sample were: Breakthrough Generations Study, Canadian Study of Diet, Lifestyle and Health, Clue II, EPIC, Melbourne Collaborative Cohort Study, Nurses’ Health Study I, Singapore Chinese Health Study, Sister Study, Southern Community Cohort, US Radiologic Technologists Cohort, Black Women’s Health Study
4. Discussion
Considering each of the four menopausal groups separately, we found little evidence of changes in the BMI breast cancer HRs over age-time during the age range covering the most common period of transition to menopause, ages 45-55 years. BMI was inversely associated with breast cancer risk for premenopausal women. The BMI breast cancer HR was close to one for women who had experienced natural or loss of ovarian function and were between 45 and 55 years; we found little evidence of any change in this association between BMI and breast cancer risk with increasing age in our primary analysis. These findings did not differ by overweight in early adulthood or for women not on MHT at baseline.
Our results confirm previous findings in this consortium of inverse associations between BMI and premenopausal breast cancer6. A meta-analysis of risk estimates from 34 data sources yielded a BMI relative risk (95% CI) of 0.92 (per 5-unit increment in BMI (kg/m2)) (0.88, 0.97) for premenopausal breast cancer5. Analysis of four studies found a pooled relative risk (95% CI) of 0.89 (0.81, 0.97)7. These estimates were similar to those reported previously in the PBCCG6.
Our findings for postmenopausal women are not consistent with those from previous publications, but most have included postmenopausal women across much broader age ranges. For example, a meta-analysis of 31 studies examining the association between postmenopausal breast cancer incidence and BMI indicated a relative risk (95% CI) of 1.12 (1.08, 1.16)5 for a 5 unit increment. In the United States Women’s Health Initiative9, the overall invasive breast cancer HR was 1.17 (95% CI: 1.06, 1.29) for women with overweight compared to women who were not with overweight between 50 and 79 years with a median 13 years follow-up time. The BMI HR (95% CI) increased with age — the HR (95% CI) comparing overweight to non-overweight groups was 1.02 (0.87, 1.20) for women 50-59 years at baseline and 1.29 (1.03, 1.62) for women 70-79 years old at baseline. Another study showed a breast cancer relative risk (95% CI) of 1.29 (per 5 unit difference) (1.22, 1.36) for women with obesity compared to the non-obese BMI group in postmenopausal women who at baseline reported never using MHT10. The narrower and earlier age range covered by our analysis thus may explain the lower HR found in our analysis of postmenopausal women. Given our null postmenopausal BMI HR estimates compared with previous analyses, our data are most consistent with the possibility that the directional change in the BMI HR for breast cancer occurs after age 55 years.
We did not find evidence to support changes over time for the association between BMI and breast cancer risk between ages 45-55 years. To fit within our findings, any proposed mechanisms must take many years to influence risk or initiate after menopause. For example, before menopause, the influence of BMI on hormonal factors such as estrogen and progesterone is one hypothesized mechanism explaining the inverse association between BMI and breast cancer risk22,23. Estrogen production shifts from the ovaries to adipose tissue after menopause. This mechanism could support our findings if the change of estrogen production to adipose tissue and its adverse associations with breast cancer risk 24-28 occur gradually after menopause supporting a gradual increase over time without any sudden shift in the association between BMI and breast cancer risk immediately following menopause. A study supporting this mechanism demonstrated a positive association between breast cancer and BMI for a cohort diagnosed after age 60 years, not before12. The BMI associated risks for pre- and postmenopausal groups following hormonal changes may begin to diverge in this interval, with the differences persisting, and widening over time, resulting in stronger associations between BMI and risk in older ages.
Our study has some strengths and limitations. First, the consortium, pooling data from many study participants, allowed us to estimate HRs for several postmenopausal groups with more precision than in smaller samples. The enriched concentration of person-time around the age of a typical transition to menopause combined with the large sample size via the consortium, allowed us to evaluate BMI HRs of women who were pre- and peri-menopausal and those who were transitioning beyond menopause.
Limitations are similar to those discussed for the estimation of BMI breast cancer HRs before menopause6, including the fact that BMI was self-reported for most studies and therefore subject to error. The Sister Study and and the Melbourne Collaborative Cohort Study are exceptions because these cohorts provided examiner-based BMI measures at baseline. A previous analysis comparing the self-reported to examiner-based BMI suggested self-reported BMI are reliable for women in the middle BMI range whereas there is a tendency for women with overweight to under-report their weight and women who are underweight to over-report their weight29. As discussed regarding this sample and this type of misclassification6, there could be the potential for bias in estimating the associations between BMI and breast cancer risk and consequently the premenopausal protective association between BMI and breast cancer risk could be closer to 1.0 than what we estimated. However, if this bias is consistent across the age range we examined, this would not impact our findings that assess changes in these associations over age-time. Also, the potential for unbiased estimates to be closer to the null would not change our current findings. Another limitation related to BMI is our approach to carry forward BMI measures if we do not have updated values, which may lead to misclassification. However, in the U.S. BMI for women tends to be quite stable over the interval 45-55 years. Last, BMI may not reflect body fat distribution, which could impact the role of estrogen and adipokines1.
In summary, among postmenopausal women aged 45-55, the HR for the BMI-breast cancer association was near null, with no evidence of meaningful changes in the HR over age-time within that interval. Among groups of women who were premenopausal or reported hysterectomy without bilateral oophorectomy, the association of BMI with breast cancer risk remained protective with no evidence of changing HRs with age. Evidence supports a near-constant BMI breast cancer HR over age-time during this period among women with natural menopause, suggesting that the transition to BMI breast cancer HRs greater than one reported for postmenopausal women occurs after age 55 years. Factors that may explain this qualitative change are not understood. Future studies should extend to ages beyond 55 years, an age at which almost all women are postmenopausal, to further characterize the dynamics of this change.
Supplementary Material
Highlights.
Between 45-55 years of age, the BMI breast cancer HR varied little across age-time.
After natural menopause, the BMI breast cancer time-dependent HR remained near null.
The rise to a positive BMI breast cancer HR may be gradual and occur after age 55.
Acknowledgements:
We would like to thank Dr. Shanshan Zhao and Dr. Christine R. Langton for their critical review of the manuscript.
The authors would like to acknowledge the contribution to this study from central cancer registries supported through the Centers for Disease Control and Prevention’s National Program of Cancer Registries (NPCR) and/or the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Program. Central registries may also be supported by state agencies, universities, and cancer centers. Participating central cancer registries include the following: AL, AR, AZ, CA, CO, CT, DE, DC, FL, GA, HI, IA, IL, IN, KY, LA, MD, MA, MI, MO, MS, NE, NJ, NM, NY, NC, OH, OK, OR, PA, SC, TN, TX, VA, WA, WI. The content is solely the responsibility of the authors and does not necessarily represent the official views of the U.S. Department of Health and Human Services, the National Institutes of Health, the National Cancer Institute, or the state cancer registries. We thank participants and staff of the BWHS for their contributions.
This study was supported by the Division of Intramural Research at NIEHS, in the National Institutes of Health, under projects Z01-ES044005 for D.P. Sandler and Z01-ES102245 for C.R. Weinberg. This work was supported by the Intramural Research Program of the National Cancer Institute, National Institutes of Health. The Black Women’s Health Study was supported by NIH grants (No. R01-CA058420 and U01-CA164974). The Sister Study was supported by the Intramural Research Program of the National Institutes of Health, National Institute of Environmental Health Sciences (project Z01-ES044005 to DPS) which also partially supported the PMBCCG. The Singapore Chinese Health Study was supported by NIH grants (No. R01-CA144034 and UM1-CA182876). The coordination of EPIC is financially supported by International Agency for Research on Cancer (IARC) and also by the Department of Epidemiology and Biostatistics, School of Public Health, Imperial College London which has additional infrastructure support provided by the NIHR Imperial Biomedical Research Centre (BRC). The national cohorts are supported by: Danish Cancer Society (Denmark); Ligue Contre le Cancer, Institut Gustave Roussy, Mutuelle Générale de l’Education Nationale, Institut National de la Santé et de la Recherche Médicale (INSERM) (France); German Cancer Aid, German Cancer Research Center (DKFZ), German Institute of Human Nutrition Potsdam-Rehbruecke (DIfE), Federal Ministry of Education and Research (BMBF) (Germany); Associazione Italiana per la Ricerca sul Cancro-AIRC-Italy, Compagnia di SanPaolo and National Research Council (Italy); Dutch Ministry of Public Health, Welfare and Sports (VWS), Netherlands Cancer Registry (NKR), LK Research Funds, Dutch Prevention Funds, Dutch ZON (Zorg Onderzoek Nederland), World Cancer Research Fund (WCRF), Statistics Netherlands (The Netherlands); Health Research Fund (FIS) - Instituto de Salud Carlos III (ISCIII), Regional Governments of Andalucía, Asturias, Basque Country, Murcia and Navarra, and the Catalan Institute of Oncology - ICO (Spain); Swedish Cancer Society, Swedish Research Council and County Councils of Skåne and Västerbotten (Sweden); Cancer Research UK (14136 to EPIC-Norfolk; C8221/A29017 to EPIC-Oxford), Medical Research Council (1000143 to EPIC-Norfolk; MR/M012190/1 to EPIC-Oxford). (United Kingdom); The Generations Study is supported by Breast Cancer Now and The Institute of Cancer Research, United Kingdom. Thomas E. Rohan is supported in part by the Breast Cancer Research Foundation (BCRF-22-140).
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Declaration of Competing Interest
none
Disclaimer: Where authors are identified as personnel of the International Agency for Research on Cancer / World Health Organization, the authors alone are responsible for the views expressed in this article and they do not necessarily represent the decisions, policy or views of the International Agency for Research on Cancer / World Health Organization.
Declarations of interest: none
References
- 1.García-Estévez L, Cortés J, Pérez S, Calvo I, Gallegos I, Moreno-Bueno G. Obesity and Breast Cancer: A Paradoxical and Controversial Relationship Influenced by Menopausal Status. Frontiers in Oncology. 2021;11:3114. doi: 10.3389/fonc.2021.705911 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Bergström A, Pisani P, Tenet V, Wolk A, Adami HO. Overweight as an avoidable cause of cancer in Europe. International Journal of Cancer. 2001;91(3):421–430. doi: [DOI] [PubMed] [Google Scholar]
- 3.Berstad P, Coates RJ, Bernstein L, et al. A case-control study of body mass index and breast cancer risk in white and African-American women. Cancer Epidemiol Biomarkers Prev. 2010;19(6):1532–1544. doi: 10.1158/1055-9965.EPI-10-0025 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Harris HR, Willett WC, Terry KL, Michels KB. Body Fat Distribution and Risk of Premenopausal Breast Cancer in the Nurses’ Health Study II. JNCI: Journal of the National Cancer Institute. 2011;103(3):273–278. doi: 10.1093/jnci/djq500 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Renehan AG, Tyson M, Egger M, Heller RF, Zwahlen M. Body-mass index and incidence of cancer: a systematic review and meta-analysis of prospective observational studies. The Lancet. 2008;371(9612):569–578. doi: 10.1016/S0140-6736(08)60269-X [DOI] [PubMed] [Google Scholar]
- 6.The Premenopausal Breast Cancer Collaborative Group, Schoemaker MJ, Nichols HB, et al. Association of Body Mass Index and Age With Subsequent Breast Cancer Risk in Premenopausal Women. JAMA Oncology. Published online June 21, 2018:e181771. doi: 10.1001/jamaoncol.2018.1771 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.van den Brandt PA, Spiegelman D, Yaun SS, et al. Pooled Analysis of Prospective Cohort Studies on Height, Weight, and Breast Cancer Risk. American Journal of Epidemiology. 2000; 152(6):514–527. doi: 10.1093/aje/152.6.514 [DOI] [PubMed] [Google Scholar]
- 8.World Cancer Research Fund/American Institute for Cancer Research: Diet, nutrition, physical activity and breast cancer. Published online 2017:124. [Google Scholar]
- 9.Neuhouser ML, Aragaki AK, Prentice RL, et al. Overweight, Obesity, and Postmenopausal Invasive Breast Cancer Risk: A Secondary Analysis of the Women’s Health Initiative Randomized Clinical Trials. JAMA Oncology. 2015;1(5):611–621. doi: 10.1001/jamaoncol.2015.1546 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Reeves GK, Pirie K, Beral V, Green J, Spencer E, Bull D. Cancer incidence and mortality in relation to body mass index in the Million Women Study: cohort study. BMJ. 2007;335(7630):1134. doi: 10.1136/bmj.39367.495995.AE [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Rosenberg LU, Einarsdóttir K, Friman EI, et al. Risk Factors for Hormone Receptor-Defined Breast Cancer in Postmenopausal Women. Cancer Epidemiol Biomarkers Prev. 2006;15(12):2482–2488. doi: 10.1158/1055-9965.EPI-06-0489 [DOI] [PubMed] [Google Scholar]
- 12.Ritte R, Lukanova A, Berrino F, et al. Adiposity, hormone replacement therapy use and breast cancer risk by age and hormone receptor status: a large prospective cohort study. Breast Cancer Res. 2012;14(3):R76. doi: 10.1186/bcr3186 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.VanderWeele TJ. On the Distinction Between Interaction and Effect Modification: Epidemiology. 2009;20(6):863–871. doi: 10.1097/EDE.0b013e3181ba333c [DOI] [PubMed] [Google Scholar]
- 14.Nichols HB, Schoemaker MJ, Wright LB, et al. The Premenopausal Breast Cancer Collaboration: A Pooling Project of Studies Participating in the National Cancer Institute Cohort Consortium. Cancer Epidemiology Biomarkers & Prevention. 2017;26(9):1360–1369. doi: 10.1158/1055-9965.EPI-17-0246 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.InterLACE Study Team. Variations in reproductive events across life: a pooled analysis of data from 505 147 women across 10 countries. Human Reproduction. 2019;34(5):881–893. doi: 10.1093/humrep/dez015 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Gold EB. Factors Associated with Age at Natural Menopause in a Multiethnic Sample of Midlife Women. American Journal of Epidemiology. 2001;153(9):865–874. doi: 10.1093/aje/153.9.865 [DOI] [PubMed] [Google Scholar]
- 17.Lin WT, Beattie M, may Chen L, et al. Comparison of age at natural menopause in BRCA1/2 mutation carriers with a non-clinic-based sample of women in northern California: BRCA1 / 2 and Early Natural Menopause. Cancer. 2013;119(9):1652–1659. doi: 10.1002/cncr.27952 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bender A, Groll A, Scheipl F. A generalized additive model approach to time-to-event analysis. Statistical Modelling. 2018;18(3-4):299–321. doi: 10.1177/1471082X17748083 [DOI] [Google Scholar]
- 19.West SL, Gartlehner G, Mansfield AJ, et al. Comparative Effectiveness Review Methods: Clinical Heterogeneity. Agency for Healthcare Research and Quality (US); 2010. Accessed September 28, 2021. http://www.ncbi.nlm.nih.gov/books/NBK53310/ [PubMed] [Google Scholar]
- 20.Type and timing of menopausal hormone therapy and breast cancer risk: individual participant meta-analysis of the worldwide epidemiological evidence. The Lancet. 2019;394(10204):1159–1168. doi: 10.1016/S0140-6736(19)31709-X [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.R Core Team. R: A Language and Environment for Statistical Computing.; 2020. https://www.R-project.org/
- 22.Freeman EW, Sammel MD, Lin H, Gracia CR. Obesity and Reproductive Hormone Levels in the Transition to Menopause. Menopause. 2010;17(4):718–726. doi: 10.1097/gme.0b013e3181cec85d [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Dowsett M, Folkerd E. Reduced progesterone levels explain the reduced risk of breast cancer in obese premenopausal women: a new hypothesis. Breast Cancer Res Treat. 2015;149(1):1–4. doi: 10.1007/s10549-014-3211-4 [DOI] [PubMed] [Google Scholar]
- 24.Thomas HV, Reeves GK, Key TJA. Endogenous estrogen and postmenopausal breast cancer: a quantitative review. Cancer Causes Control. 1997;8(6):922–928. doi: 10.1023/A:1018476631561 [DOI] [PubMed] [Google Scholar]
- 25.Hankinson SE, Willett WC, Manson JE, et al. Plasma Sex Steroid Hormone Levels and Risk of Breast Cancer in Postmenopausal Women. JNCI: Journal of the National Cancer Institute. 1998;90(17):1292–1299. doi: 10.1093/jnci/90.17.1292 [DOI] [PubMed] [Google Scholar]
- 26.Lukanova A, Lundin E, Zeleniuch-Jacquotte A, et al. Body mass index, circulating levels of sex-steroid hormones, IGF-I and IGF-binding protein-3: a cross-sectional study in healthy women. Eur J Endocrinol. 2004;150(2):161–171. doi: 10.1530/eje.0.1500161 [DOI] [PubMed] [Google Scholar]
- 27.Missmer SA, Eliassen AH, Barbieri RL, Hankinson SE. Endogenous Estrogen, Androgen, and Progesterone Concentrations and Breast Cancer Risk Among Postmenopausal Women. JNCI Journal of the National Cancer Institute. 2004;96(24):1856–1865. doi: 10.1093/jnci/djh336 [DOI] [PubMed] [Google Scholar]
- 28.The Endogenous Hormones and Breast Cancer Collaborative Group. Endogenous Sex Hormones and Breast Cancer in Postmenopausal Women: Reanalysis of Nine Prospective Studies. JNCI: Journal of the National Cancer Institute. 2002;94(8):606–616. doi: 10.1093/jnci/94.8.606 [DOI] [PubMed] [Google Scholar]
- 29.Lin CJ, DeRoo LA, Jacobs SR, Sandler DP. Accuracy and reliability of self-reported weight and height in the Sister Study. Public Health Nutr. 2012;15(6):989–999. doi: 10.1017/S1368980011003193 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.