Proportion of invasive breast cancer attributable to risk factors modifiable after menopause
January, 2009
Commentary by Dr Anne Kricker
The article
Sprague BL, Trentham-Dietz A, Egan KM, et al. Proportion of Invasive Breast Cancer Attributable to Risk Factors Modifiable after Menopause. Am J Epidemiol 2008;168:404–11 published online 13 June 2008 doi:10.1093/aje/kwn143
Reviewer
Dr Anne Kricker is an epidemiologist and Research Director of the Cancer Genes, Environment and Behaviour Program at the University of Sydney.
Summary
Abbreviations
Hormone Replacement Therapy (HRT); Population Attributable Risk (PAR)
Study design
This article describes a large case-control study conducted in the USA between 1997 and 2001, investigating invasive breast cancer risk among postmenopausal women. The study includes data from 3499 breast cancer cases and 4213 controls.
Findings
The population attributable risks (PARs) of the main non-modifiable and modifiable risk factors are listed in Table 1 below. The PAR is the proportion of cases that would be eliminated from the population if the risk factor(s) was/were eliminated or, in the case of this study, shifted to a low-risk category.
Table 1. Population attributable risks of known risk factors for breast cancer
| Risk factor | PAR | Reference/low-risk category | |
| Non-modifiable risk factors | Summary | 57.3% | |
| Age at menarche | 18.8% | ≥ 15 years | |
| Age at menopause | 13.7% | < 45 years | |
| Parity | 13.3% | ≥ 4 children | |
| Modifiable risk factors | Summary | 40.7% | |
| Current postmenopausal hormone use | 4.6% | former users | |
| Recent alcohol consumption (past year) | 6.1% | < 1 drink/week | |
| Weight gain since age 18 years | 21.3% | lost ≤ 5-gained ≤ 5kg | |
| Recent recreational physical activity | 15.7% | > 5 hours/week | |
PAR – population attributable risk
The PAR estimates varied depending on the cut-points used to define reference categories, for example if the reference category for weight gain included gains up to 15kg, the PAR decreased to 8.8% and if the reference category for physical activity included 2 hours/week of exercise the PAR decreased to 3.9%. This study used extreme cut-points which provide an interesting public health perspective.
Conclusion
The authors conclude that a substantial proportion of breast cancer incidence in postmenopausal women could be prevented with the modification of known risk factors.
Commentary
What does this article add to existing clinical evidence in this area?
This article investigated four lifestyle factors (increased body weight, alcohol consumption, HRT use and low physical activity) that are known to increase the risk of breast cancer, a claim supported by their Harvard Cancer Risk Index listing as established or likely causes of breast cancer (see Colditz et al 20001). All factors are amenable to intervention in postmenopausal women to reduce their potential harms. This article takes up the challenge of estimating the likely population-wide benefits in sets of modifiable and non-modifiable risk factors from shifting postmenopausal women currently at unfavourable levels to the least harmful levels.
Most similar studies have focused on strong risk factors such as reproductive and menstrual history which are not modifiable, especially not in later life. The estimated contributions made by individual modifiable factors to breast cancer risk are reasonably similar to previous cited studies: substantial proportions of breast cancers were attributed to maximum weight gains after age 18 and to low average physical activity in the recent past (around 12-21% of breast cancers could be eliminated by shifting women from high to low risk levels), 6% of breast cancers to unfavourable alcohol consumption and 4.6% to current use of HRT at the time of diagnosis. Not unexpectedly, the influence of these lifestyle factors on breast cancer risk is consistent with advice from other highly reputable sources such as the Harvard Women’s Health Watch.2
Sprague et al (2008), however, make a special contribution by considering all four factors simultaneously: approximately 40% of postmenopausal breast cancers would be avoided in the population if women ceased or did not start postmenopausal hormone use, restricted alcohol consumption to 5kg after age 18 and exercised >5 hours a week.
The authors emphasise the need to shift the whole population to more favourable levels, a difficult task: they suggest that ‘dramatic’ lifestyle alterations are required for cancer prevention. Their estimates, however, are useful guidance for women looking to lower their risk profile. Rockhill (2001)3 makes a strong case for promoting individual decision making about behaviour as a major prevention option for populations, as does Colditz (2001).4 The data from this study supports promoting to individual women that the benefits of maintaining a healthy lifestyle after the menopause could include the potential to avoid up to 40% of breast cancers in postmenopausal women.
How adequate was the methodology used in addressing the aim of the study?
The study aimed to extend what is known about risk factors that cause breast cancer to estimate how many breast cancers could be eliminated if risk factor prevalence in the population was shifted from high to low risk levels. Such studies often draw criticism on the grounds that they consider exposures individually and do not engage with the complexity of multiple exposures in women’s’ lives5 or that they consider only those well established risk factors that are not modifiable.6 A strength of this study comes from addressing both these issues. The quantity estimated for each factor, the population attributable risk (PAR), can be calculated for individual risk factors alone but these PARs cannot be totalled for a summary PAR across all factors. This study, however, used multivariable regression to do so.
The authors acknowledge potential limitations such as possible correlations of earlier behaviours with recent physical activity, body weight and alcohol consumption and that exposure cannot be altered completely later in life – HRT users cannot become nonusers. They explore the issue of an expected stronger effect of HRT and obesity on estrogen receptor-positive breast cancers which are not separately identified in their study. They also discuss the characterisation of ‘exposed’ when heavy bodyweight or minimal physical activity clearly are relative terms and assigning women to the ‘exposed’ category is clearly influenced by decisions on how to assign cut-points.
What are the implications of this study for clinical practice in Australia?
Importantly for Australian women, this study offers guidance on ways to help reduce breast cancer risk later in life. There is no doubt that evidence is accumulating on lowering risk through certain healthy lifestyle choices: stopping or not starting postmenopausal hormone use unless the value of relieving symptoms outweighs the hazards, restricting alcohol consumption, avoiding excessive weight gain after age 18 or losing weight in later life,7 and exercising at recommended weekly levels. While factors affecting risk for breast cancer are very complex and an individual woman may or may not develop breast cancer regardless of her risk factor profile, the findings of this study support adopting lifestyle modifications in later life to prevent breast cancer in substantial numbers of women.
References
1. Colditz GA, Atwood KA, Emmons K et al. Harvard report on cancer prevention Volume 4: Harvard Cancer Risk Index. Cancer Causes Control 2000;11:477-88.
2. Harvard Women’s Health Watch. Seven things you should know about breast cancer risk. www.health.harvard.edu/newsweek/seven-things-you-should-know-about-breast-cancer-risk.htm
3. Rockhill B. The Privatization of Risk. Am J Public Health 2001;91:365-8.
4. Colditz GA. Cancer Culture: epidemics, human behaviour, and the dubious search for new risk factors. Am J Public Health 2001;91:357-9.
5. Eide GE, Heuch I, Albrektsen G. Re: population attributable risk for breast cancer: diet, nutrition and physical exercise. J Natl Cancer Inst 2000;92:843-4
6. Rockhill B, Weinberg CR, Newman B. Population attributable fraction estimation for established risk factors: considering the issues of high prevalence and unmodifiablilty. Am J Epidemiol 1998;147:826-33.
7. Eliassen AH, Colditz GA, Rosner B, Willett WC, Hankinson SE. Adult weight change and risk of postmenopausal breast cancer. JAMA 2006;296:193-201.
Editor: Alison Pearce, Program Manager, National Breast and Ovarian Cancer Centre.
Editorial Committee: Mr John Collins - Surgeon, Ms Jo Keyser - Specialist Breast Nurse, Dr Warwick Lee - Radiologist, A/Prof Liz Lobb – Senior Research Fellow, Dr Sue-Anne McLachlan - Medical Oncologist, Dr Sally Meade - Breast Surgeon, Dr Sue Pendlebury - Radiation Oncologist, A/Prof Martin Stockler - Medical Oncologist.
Disclaimer
Clinical Update - Breast Cancer is produced by National Breast and Ovarian Cancer Centre (NBOCC) and is intended to provide health professionals with timely expert commentary on new research in breast cancer. Commentaries included in Clinical Update - Breast Cancer do not replace recommendations included in NBOCC clinical practice guidelines.
Information contained in Clinical Update - Breast Cancer is not intended to be used as substitute for an independent health professional's advice. NBOCC does not accept any liability for any injury, loss or damage incurred by use of or reliance on the information contained in Clinical Update - Breast Cancer. NBOCC develops material based on the best available evidence however cannot guarantee and assumes no legal liability or responsibility for the currency or completeness of the information.
