Oral Presentation Clinical Oncology Society of Australia Annual Scientific Meeting 2017

Premenopausal and postmenopausal breast cancer burden attributable to health behaviours and hormonal factors (#94)

Maria E Arriaga 1 , Claire M Vajdic 1 , Peter Hull 1 , Karen Canfell 2 , Robert MacInnis 3 4 , Emily Banks 5 , Graham G Giles 3 4 , Paul Mitchell 6 , Barbara-Ann Adelstein 1 , Julie E Byles 7 , Dianna J Magliano 8 , Jonathan E Shaw 8 , Anne W Taylor 9 , Kay Price 10 , Vasant Hirani 6 , Louiza S Velentzis 2 , Maarit A Laaksonen 1
  1. University of New South Wales, Sydney, NSW, Australia
  2. Cancer Council New South Wales, Sydney, NSW, Australia
  3. Cancer Council Victoria, Melbourne, VIC, Australia
  4. University of Melbourne, Melbourne, VIC, Australia
  5. Australian National University, Canberra, ACT, Australia
  6. University of Sydney, Sydney, NSW, Australia
  7. University of Newcastle, Newcastle, NSW, Australia
  8. Baker IDI Heart and Diabetes Institute, Melbourne, VIC, Australia
  9. University of Adelaide, Adelaide, SA, Australia
  10. University of South Australia, Adelaide, SA, Australia

Aims:

To quantify the burden of premenopausal and postmenopausal breast cancer in Australian women attributable to behavioural and hormonal risk factors.

Methods:

Data on exposure to behavioural, hormonal and reproductive risk factors from six Australian cohort studies (N=214,536) were harmonised and pooled. The cohorts were linked to the Australian Cancer Database and National Death Index to identify cancers and deaths. The strength of the exposure-cancer association (hazard ratio, HR) was estimated using a proportional hazards model, adjusting for age, study and the other exposures. Exposure prevalences (P) were estimated from the Australian National Health Survey 2014-2015, LADY Survey 2013, and representative AusDiab cohort study 1999-2000. These estimates were then combined to calculate the Population Attributable Fractions (PAFs) and their 95% confidence intervals (CIs) using an advanced method accounting for competing risk of death.

Results:

During the first 10-years follow-up, 640 incident premenopausal breast cancers, 2,632 incident postmenopausal breast cancers and 8,761 deaths were ascertained. For premenopausal women, regular light, moderate or heavy alcohol consumption (HR=1.25,1.28,1.47, combined P=48%) explains 12% (CI=4-20%), current long-term use of the oral contraceptive pill (HR=1.33, P=22%) 7% (CI=0-13%), and the two factors combined 18% (CI=8-27%) of the breast cancer burden. For postmenopausal women, overweight (BMI 25-34.9 kg/m2) or obesity (≥35 kg/m2) (HR=1.18,1.42, combined P=58%) explains 12% (CI=7-16%), light or moderate/heavy alcohol consumption (HR=1.08,1.20, combined P=58%) 8% (CI=4-13%), current use of menopausal hormone therapy (HR=1.49, P=14%) 6% (CI=4-8%), and the three factors combined 25% (CI=19-31%) of the cancer burden. The adjusted PAF results for not breastfeeding and limited physical activity were not statistically significant.  

Conclusions:

These first Australian breast cancer PAF estimates based on cohort studies and accounting for competing risk of death allow us to reliably rank the main attributable causes of breast cancer, many of which are highly prevalent and modifiable.