Weight status and breast cancer incidence in the UK Women's Cohort Study: a survival analysis

1002 Background: Physical activity is a modifiable breast cancer risk factor, perhaps mediating risk reduction through regulation of estrogen metabolism. Evidence regarding effect of physical activity is conflicting partly because breast cancer is a heterogenous constellation of different tumor subtypes with differing etiologies. No prospective study has examined the relationship between physical activity and breast cancer incidence based on ER/PR status or histological subtype. Objective: Examine effect of physical activity on breast cancer incidence based on ER/PR status and histological subtypes of breast cancer. Methods: The Iowa Women’s Health Study is a prospective cohort study of postmenopausal women (N=41,837). Physical activity was self-reported on baseline questionnaire, and three levels (high, medium and low) were defined. Breast cancer incidence, histologic subtype and ER/PR status, through 18 years of follow-up, were ascertained by linkage with the Iowa SEER Cancer Registry. Cox proportional hazards models were used to estimate multivariate relative risks (RRs) and 95% confidence intervals (CIs) of breast cancer, adjusting for other breast cancer risk factors. Results: During 554,819 person-years of follow-up, 2548 incident cases of breast cancer were observed. High physical activity was associated with decreased risk for breast cancer (RR 0.91, 95 % CI 0.81–1.01) compared to low activity. The protective effect was most marked in ER+/PR− (RR 0.66, CI 0.46–0.94), intermediate in ER−/PR− (RR 0.80, CI 0.56–1.15), weakest in ER+/PR+ (RR 0.94, CI 0.81–1.08), and elevated in ER-/PR+ (RR 1.42, CI 0.67–3.01) tumors. Higher physical activity was also associated with a decreased risk of invasive ductal/lobular carcinoma (RR 0.90, CI 0.80–1.02), but not with invasive breast cancer with a favorable histology (RR 1.19, CI 0.78–1.81). Conclusions: Higher physical activity was associated with a 10% decreased risk of breast cancer. Unexpectedly, risk reduction was most marked in PR- tumors, particularly ER+/PR-, and the more aggressive histologic forms. Further studies are needed to confirm these findings, and also evaluate other risk factors based on ER/PR status and histological subtypes. No significant financial relationships to disclose.

Download Full-text

Retrospective cohort study of breast cancer incidence, health service use and outcomes in Europe: a study of feasibility

European Journal of Public Health ◽

10.1093/eurpub/ckx127 ◽

2017 ◽

Vol 28 (2) ◽

pp. 327-332 ◽

Cited By ~ 1

Author(s):

Linda J Williams ◽

Eilidh Fletcher ◽

Anne Douglas ◽

Elaine D C Anderson ◽

Alison McCallum ◽

...

Keyword(s):

Breast Cancer ◽

Cohort Study ◽

Health Service ◽

Cancer Incidence ◽

Retrospective Cohort Study ◽

Retrospective Cohort ◽

Service Use ◽

Breast Cancer Incidence ◽

Health Service Use

Download Full-text

A comparison of time to event analysis methods, using weight status and breast cancer as a case study

Scientific Reports ◽

10.1038/s41598-021-92944-z ◽

2021 ◽

Vol 11 (1) ◽

Author(s):

Georgios Aivaliotis ◽

Jan Palczewski ◽

Rebecca Atkinson ◽

Janet E. Cade ◽

Michelle A. Morris

Keyword(s):

Breast Cancer ◽

Machine Learning ◽

Cohort Study ◽

Weight Status ◽

Cox Model ◽

Breast Cancer Incidence ◽

Time To Event ◽

Menopausal Women ◽

Post Menopausal ◽

Partial Dependence

AbstractSurvival analysis with cohort study data has been traditionally performed using Cox proportional hazards models. Random survival forests (RSFs), a machine learning method, now present an alternative method. Using the UK Women’s Cohort Study (n = 34,493) we evaluate two methods: a Cox model and an RSF, to investigate the association between Body Mass Index and time to breast cancer incidence. Robustness of the models were assessed by cross validation and bootstraping. Histograms of bootstrap coefficients are reported. C-Indices and Integrated Brier Scores are reported for all models. In post-menopausal women, the Cox model Hazard Ratios (HR) for Overweight (OW) and Obese (O) were 1.25 (1.04, 1.51) and 1.28 (0.98, 1.68) respectively and the RSF Odds Ratios (OR) with partial dependence on menopause for OW and O were 1.34 (1.31, 1.70) and 1.45 (1.42, 1.48). HR are non-significant results. Only the RSF appears confident about the effect of weight status on time to event. Bootstrapping demonstrated Cox model coefficients can vary significantly, weakening interpretation potential. An RSF was used to produce partial dependence plots (PDPs) showing OW and O weight status increase the probability of breast cancer incidence in post-menopausal women. All models have relatively low C-Index and high Integrated Brier Score. The RSF overfits the data. In our study, RSF can identify complex non-proportional hazard type patterns in the data, and allow more complicated relationships to be investigated using PDPs, but it overfits limiting extrapolation of results to new instances. Moreover, it is less easily interpreted than Cox models. The value of survival analysis remains paramount and therefore machine learning techniques like RSF should be considered as another method for analysis.

Download Full-text