Burden of disease from breast cancer attributable to smoking and second‐hand smoke exposure in Europe

Smoking and second‐hand smoke (SHS) exposure have been recently linked to a higher risk of breast cancer in women. The aim of this work is to estimate the number of deaths and disability‐adjusted life years (DALYs) from breast cancer attributable to these two risk factors in the European Union (EU‐28) in 2017. The comparative risk assessment method was used. Data on prevalence of smoking and SHS exposure were extracted from the Eurobarometer surveys, relative risks from a recent meta‐analysis, and data on mortality and DALYs from breast cancer were estimated from the Global Burden of Disease, Injuries and Risk Factors Study. In 2017, 82 239 DALYs and 3354 deaths from breast cancer in the EU‐28 could have been avoided by removing exposure to these two risk factors (smoking and SHS exposure). The proportion of DALYs from breast cancer lost respectively from smoking and SHS exposure was 2.6% and 1.0%, although geographically distributed with significant heterogeneity. These results represent the first estimates of breast cancer burden in women attributable to smoking and SHS exposure for the EU‐28. It is important to increase awareness among women, health professionals and wider society of the association between smoking, SHS exposure and breast cancer, a relationship that is not widely recognised or discussed.

exposure was 2.6% and 1.0%, although geographically distributed with significant heterogeneity. These results represent the first estimates of breast cancer burden in women attributable to smoking and SHS exposure for the EU-28. It is important to increase awareness among women, health professionals and wider society of the association between smoking, SHS exposure and breast cancer, a relationship that is not widely recognised or discussed. General, did not report any association between smoking and breast cancer. The 1986 IARC monograph even indicated that smoking reduced breast cancer risk. 2 The 2005 CalEPA Report for the first time concluded that there was a positive association between smoking and breast cancer risk, 3 followed a few years later by a report by a group of Canadian experts, 4 and then by the 2012 monograph of the IARC, which for the first time, mentioned a causal relationship. 5 Also, the 2014 US Surgeon General reported an increased risk of breast cancer in smokers, particularly in postmenopausal women. 6 Regarding second-hand smoke (SHS) exposure, both the 2005 CalEPA Report and the Canadian study reported a relationship consistent with causality between SHS exposure and occurrence of breast cancer, particularly in women before menopause, whereas the 2014 US Surgeon General reported a possible association in nonsmoking premenopausal women. 3,4,6 The 2015 meta-analysis of all published studies on this research topic reported significant 9% and 20% increases in breast cancer risk respectively in smoking women and among nonsmoking women exposed to SHS, surprisingly. However, high heterogeneity was observed among studies, and respectively a 10% and 7% risk increase was estimated if considering only prospective studies accounting for heterogeneity. 7 Several theories have been put forward to explain why SHS exposure could have a similar or stronger effect on breast cancer than active smoking, and the debate is ongoing. According to one theory, smoking has an association with breast cancer weaker than expected due to the role of being both an anti-estrogenic protective factor and a risk factor for breast cancer. 2 According to others, SHS exposure has a predominant effect in premenopausal cancers, a hypothesis strongly supported by a Japanese cohort study that found three times higher risk of developing breast cancer among SHS exposed women in premenopausal age, but not in postmenopausal age. 8 Overall, smoking prevalence in European Union (EU-28) has shown a slight decrease in recent years, yet with heterogeneous trends among countries, and, since the widespread implementation of smoking bans, important reductions in SHS exposure in EU-28 have been observed. 9 It is important to highlight the impact that both smoking and SHS exposure have on women's health and to quantify the number of deaths and disability-adjusted life years (DALYs) lost that could be avoided. The aim of this work, conducted within the TackSHS project, 10 is to estimate the number of deaths and DALYs from breast cancer attributable to smoking and SHS exposure in the EU-28 in 2017.

| MATERIALS AND METHODS
The burden from breast cancer attributable to smoking and SHS exposure was obtained using the comparative risk assessment

What's new?
Smoking and exposure to secondhand smoke are not considered a major risk factor for breast cancer risk in Europe.
Here, the authors publish the first analysis estimating how many deaths and years of disability can be attributed to smoking. Using the comparable risk assessment method, they determined that eliminating smoking could have saved 60 733 years of life with disability and 2719 deaths from breast cancer. Given the burden of disease from breast cancer attributable to smoking, it is important to increase awareness among health professionals and public. method. 11 Briefly, the smoking and SHS attributable fraction (AF) for each country and age-class was first estimated using Levin's where RR is the relative risk for exposed to smoking or SHS compared to nonexposed.
In the estimation of the AF from smoking p is the smoking impact ratio (SIR) where C LC and N LC are the age-specific lung cancer mortality rate respectively for the overall country under study and for never smokers only, and S * LC and N * LC are lung cancer mortality rates for smokers and never-smokers, respectively, in a reference population. The SIR represents the accumulated risk from smoking, using lung cancer mortality excess as a biological marker for accumulated smoking hazards. Since the effect of smoking on breast cancer depends on the smoking exposure history, such as age of starting smoking and number of cigarettes smoked per day, the SIR is usually preferred to the smoking prevalence alone that is an insufficient indicator of accumulated risk from smoking. 12 In the estimation of the AF from SHS, p is the 10-years lagged age and country-specific prevalence of SHS exposure. A 10-year lag between SHS exposure and breast cancer death/occurrence was assumed in computing the AF, due to an expected long latency, as for active smoking. 11 The number of breast cancer deaths/DALYs attributable to smoking was then obtained by multiplying the age-and countryspecific number of breast cancer deaths/DALYs by the corresponding AF, and the burden attributable to SHS exposure was estimated among nonsmoking women, because the impact of smoking could mask the effect due to SHS. 11 The RR for smoking women compared to nonsmokers was 1.10 (95% confidence interval [95% CI]: 1.09-1.12), and the RR for women exposed to SHS compared to nonexposed was 1.07 (95% CI: 1.02-1.13). 7 The sources and the data used in the analysis are reported in Tables 1 and 2 7 Moreover, an analysis estimating the burden form smoking by using the 10 years lagged prevalence of smoking instead of the SIR in the AF estimation was also performed.

| RESULTS
In the EU-28 in 2017, the number of DALYs from breast cancer attrib- In terms of both DALYs and deaths, the highest burden due to both risk factors (smoking and SHS exposure) was estimated in Denmark, Malta, Croatia, Hungary and in the United Kingdom, with a proportion on the total breast cancer DALYs and deaths higher than 4% and 5%, respectively. The lowest burden was estimated in Cyprus, Lithuania, Latvia, Italy and Estonia with a proportion less than 1.5%.
The proportion of DALYs and deaths from breast cancer attributable to smoking was higher than that due to SHS exposure for all EU-28 countries (Figure 1).
In the sensitivity analysis using meta-analytical RR of breast cancer with a high heterogeneity, the rank of the countries by the burden Sources of data used for the analyses

Variable Source
Country and age-specific prevalence of SHS exposure at home in nonsmoking women in 2006 for the estimation of attributable fractions (defined as being ever exposed in the home daily) Eurobarometer survey 9 Country and age-specific prevalence of smoking women in 2017 for the estimation of nonsmoking women Eurobarometer survey 9 Relative risk of breast cancer for exposed to SHS compared to not exposed and for smokers compared to nonsmokers Greece, Bulgaria and Hungary. By considering a higher risk from SHS exposure, even if estimated with a large heterogeneity, the same countries showed proportions of breast cancer DALYs attributable to SHS exposure over 1.6%. In the South-Eastern countries, the burden from breast cancer due to smoking was lower than that due to SHS exposure as a consequence of the lower smoking prevalence than that recorded among women of North-Western countries.
Differently from other studies, the analyses on SHS exposure were carried out by considering household exposure only to explore the burden unrelated to the current legislation.
The SIR approach in the estimation of the burden attributable to smoking is usually the preferred method because it considers the accumulated risk from smoking, but also because it uses lung cancer mortality data, which are easily available for all countries. 16 Using a lagged prevalence of smoking to take into account for the time period between exposure and cancer occurrence, gives smaller attributable fractions than SIR-based estimates among females, 17 and this is confirmed in our sensitivity analysis.
The GBD framework, which provides a comprehensive assessment of risk factor exposure and attributable burden of disease, estimated for 2017 in EU-28 a PAF to smoking of 6.8% and 5.7% for DALYs and deaths, respectively. 13 Another study on the burden from smoking on cancers that analysed five European cohorts reported a proportion of attributable DALYs of 4.7%, 18 whereas a study carried out in Norway estimated a PAF to smoking of 11.9%. 19  In conclusion, to the best of our knowledge, this is the first estimate on the burden from breast cancer attributable to smoking and SHS exposure in the EU-28. Although smoking (including SHS exposure) is not considered a major risk factor for breast cancer risk, the burden of disease from breast cancer attributable to smoking and SHS exposure is large in EU-28, and, given that awareness of both smoking and SHS exposure as risk factors for breast cancer remains low among the population, it is important to spread this link among the public and all stakeholders. Smoking and exposure to SHS are modifiable risk factors, and therefore important objects of primary prevention policies, in addition to the current population-based breast cancer early detection programmes across Europe.