In the United States, one in eight women develops breast cancer after the age of 70 years [
Each 2 kg/m2 increase in body mass index (BMI) is associated with 5% increased risk of postmenopausal breast cancer (Relative Risk/RR = 1.05, 95% CI: 1.03–1.07), 13% increased risk among overweight women (BMI = 25–29 kg/m2), and 25% among obese women (BMI ≥ 30 kg/m2) [
Currently, postmenopausal breast cancer (PBC) is the most common cancer among women in the United States [
Data from NHANES I Epidemiologic Follow-Up Study (NHEFS) includes subjects of 25–74 years of age who completed a medical examination at NHANES I in 1971–1975 and were followed up in 1982, 1986, 1987, and 1992. The database does not include any private information about participants and is available and open for public use. Methodology of the NHEFS has been described elsewhere [
The analytic cohort was formed from women enrolled in NHANES I study. They self-reported weight at 25, 40, and 65 years of age, whereas other weight assessments were a combination of physical exams and questionnaires from the first exam in 1971–1974 through 1992 [
Women who did not have BMI assessment at age 25 and women with less than three BMI assessments before the postmenopausal breast cancer diagnosis/end of follow-up (for control) were excluded from the study. Lastly, because the major interest in the study is to investigate the association between weight fluctuation and PBC among women who gained weight in adult years, women with lost weight were also excluded.
Definition of weight fluctuation was based on the results from a simple linear regression model for each woman with her BMI regressed on age. The extent of BMI fluctuation for each participant was estimated with the root-mean-square-error (RMSE) statistic. The subset of women with a positive beta coefficient for age (
The other postmenopausal breast cancer risk factors were included in the conditional logistic regression as the following covariates: parity or live births (0, 1, or more), age at the first child greater than 30 (yes/no), history of breast cancer in mother (yes/no), and use of hormone replacement therapy (yes/no).
In 1971–1975, weight and height were measured. In the following years, data were collected on live births, age when first child was born, history of breast cancer in first-degree female relatives (mother), and use of hormone replacement therapy.
A conditional logistic regression model (PROC LOGISTIC in SAS®) was developed to assess the increase in risk of PBC for every unit increase in weight fluctuation measured by root-mean-square-error (RMSE) derived from a simple linear regression for BMI on age (from 25 years). The other risk factors for postmenopausal breast cancer, such as parity, age at the first child greater than 30, history of breast cancer in first-degree female relatives (mother), and use of hormone replacement therapy, were included in the logistic regression model as the covariates.
SAS version 9.2 was used for all analyses where significance tests were two-sided with the level of significance equaling 5%.
A total of 79 new cases of breast cancer for women of 50 years of age and older were included in this study and matched to 79 controls by the years of follow-up and BMI status at 25 years of age. The study population was weight gainers and had at least three BMI assessments before the diagnosis/end of follow-up (for controls). Distribution by 50–60, 61–70, 71–80, and >80 years of age was 51.9%, 26.58%, 17.72%, and 3.8%, respectively. The mean age at cancer diagnosis was about 63 years (standard deviation = 9.26). Most women (82.28%) had normal BMI status at age of 25 years. The proportion of women with live births “1 or more” was a little higher among controls (87.34%) versus cases (84.81%). The proportion of women with the history of breast cancer in female relatives (mother) was slightly higher among cases (83.54%) compared to controls (79.75%). More women with PBC reported that they “ever used the hormonal therapy” among cases (92.41%) versus controls (56.96%). The summary of demographic and baseline characteristics is presented in Table
Demographics and baseline characteristics.
Characteristics | Cases |
Controls |
---|---|---|
Age of diagnosis/years of follow-up |
||
50–60 | 41 (51.90%) | |
61–70 | 21 (26.58%) | |
71–80 | 14 (17.72%) | |
80 and above | 3 (3.80%) | |
Mean (standard deviation) | 62.82 (9.26) | |
Minimum-maximum | 50–87 | |
|
||
BMI at age 25 | ||
Mean (standard deviation) | 21.11 (2.65) | |
Minimum-maximum | 16.83–33.77 | |
|
||
BMI status at age 25 | ||
Underweight | 9 (11.39%) | |
Normal | 65 (82.28%) | |
Overweight | 4 (5.06%) | |
Obese | 1 (1.27%) | |
|
||
Live births | ||
0 | 12 (15.19%) | 10 (12.66%) |
1 or more | 67 (84.81%) | 69 (87.34%) |
|
||
Age > 30 for birth of first child | ||
Yes ( |
4 (5.06%) | 4 (5.06%) |
No ( |
64 (81.01%) | 68 (86.08%) |
N/A | 11 (13.92%) | 7 (8.86%) |
|
||
History of BC female relatives (mother) | ||
Yes | 13 (16.46%) | 16 (20.25%) |
No | 66 (83.54%) | 63 (79.75%) |
|
||
Have ever used hormone therapy (HRT) | ||
Yes | 6 (7.59%) | 34 (43.04%) |
No | 73 (92.41%) | 45 (56.96%) |
All 158 women self-reported weight assessments at age 25 and had three or more weight assessments before postmenopausal breast cancer diagnosis/end of follow-up (for controls). For cases, three weight assessments before cancer were collected for eight women (10.13%) while the majority of cases (88.61%) had five or more weight assessments. All controls (100%) reported five or more weight assessments for the period of follow-up. The weight fluctuation was measured as root-mean-square-error statistic from simple linear regression of dependent BMI from independent age. It was higher in cases with mean (standard deviation) that is equal to 1.51 (0.89) compared to controls 1.37 (0.86).
The assumptions underlying the regression models were checked with diagnostics statistics and plots.
It is known that weight gain is a risk factor for postmenopausal breast cancer [
The odds ratio estimate for weight fluctuation from conditional logistic regression adjusted for weight gain (
The results from conditional logistic regression adjusted for parity, first child after 30 years of age, use of hormonal therapy, history of mother’s breast cancer, and weight gain (
The association between weight fluctuation and postmenopausal breast cancer.
Model | Odds ratio | 95% Wald confidence limits |
|
|
---|---|---|---|---|
# 1: weight fluctuation (RMSE) | 1.674 | 1.056 | 2.655 | 0.0285 |
# 2: weight fluctuation (RMSE) | 2.917 | 1.385 | 6.142 | 0.0048 |
Model #1 included weight gain.
Model #2 included weight gain, parity, first child at >30 years of age, history of cancer (mother), and have ever used hormonal therapy.
The major finding of this study is that weight fluctuation during gaining weight in adult years is an independent risk factor for postmenopausal breast cancer.
National Task Force on the Prevention and Treatment of Obesity summarized the forty-three (43) English-language articles that evaluated the effects of weight fluctuation/cycling on humans or animals done from 1966 through 1994 [
Studies with weight cycling as risk factor for postmenopausal breast cancer.
Author(s)/year/[ref] | Study design |
Cancer |
Exposure: weight cycling/analysis and covariates | Results/Conclusion | Comments |
---|---|---|---|---|---|
French et al., |
|
|
|
|
First study |
|
|||||
Trentham-Dietz et al., 2000 |
|
|
|
≥1 cycle versus |
~18.6% cases and ~17% controls |
|
|||||
Eng et al., 2005 |
|
|
|
|
Twofold increase in risk versus mainlining the “low” status |
French et al. investigated the association of weight variability and incidence of diseases including breast cancer [
In 2000, Trentham-Dietz et al. conducted a large case-control study of postmenopausal women to investigate the risk of breast cancer associated with different patterns of weight changes [
In 2005, Eng et al. conducted a population-based case-control study using participants in the Long Island Breast Cancer Study Project (LIBCP: 1996-1997) to investigate the effects of patterns of body size change throughout the lifetime on postmenopausal breast cancer risk [
The weight fluctuation/cycling were defined differently in three studies of postmenopausal women. The authors acknowledged that estimation of the risk of weight cycling is the hard task because of gaps in weight assessments in existing databases with long-term follow-up and lack of unified definition of weight cycling.
This study defines weight fluctuation similar to French et al. [
This is a novel study to investigate the effect of weight fluctuation during adulthood weight gain on PBC. The major strength of this study is use of national database. The major limitation is relatively small number of PBC cases. In addition, the majority of weight assessments were self-reported that introduced a recall bias (especially for 25, 40, and 65 years of age). Further studies are needed to confirm the results.
Understanding the relationship between weight fluctuation and postmenopausal breast cancer is very important. Many women go through weight loss programs to reduce weight gain. Yet, unsuccessful weight losses lead to weight regains and additional weight gain. This study demonstrated that weight fluctuation during weight gain is an independent risk factor for postmenopausal breast cancer. In summary, this project is one step forward to the goal of reducing the risk and the incidence of the disease. It will help to set up clear directions in weight control/management programs where successful interventions may improve health of many women.
The author declares that there is no conflict of interests regarding the publication of this paper.
The author thanks Drs. Michael Szarek, Michael Walsh, and Elizabeth Helzner for assistance with development of this proposal.