Impact of Alcohol Use on Cancer Incidence and Mortality

Subproject

Pooling Project of Prospective Studies of Diet and Cancer (DCCP)

Paul Brennan

Financial support for the project will be sought with grant applications to relevant funding bodies: US NIAAA (NIH), UK WCRF, French INCA. Support will be needed to centralize and harmonize data from participating cohorts, and to carry out statistical analyses. For the first task, PIs of other consortial projects will be contacted to take profit of already ongoing harmonization efforts, including the BMI Pooling Project, the Diet and Cancer Pooling Project and the NCI/EGRP Cancer Epidemiology Consortia.

Recent epidemiological evidence suggests a causal link between alcohol consumption and the incidence of cancer of the oral cavity, pharynx, larynx, oesophagus, liver, colorectum, and female breast (Bagnardi et al., Br J Cancer, 2001; Boffetta & Hashibe, Lancet Oncol, 2006; Corrao et al., Prev Med, 2004a; WCRF/AICR, 2007). A causal association is also suspected for other cancer sites, including pancreatic cancer (Pelucchi et al., Nutr Cancer, 2011), while recent evidence is suggestive of a protective role of alcohol use for kidney cancer (Ljungberg et al., Eur Urol, 2011). The International Agency for Research on Cancer (IARC) has recently systematically re-evaluated the association between alcohol and different cancers in the Monograph program (Cogliano et al., J Natl Cancer Inst, 2011), and classified ethanol in alcoholic beverages as a definite carcinogen to humans (Group 1), notably with respect to cancers of the upper aero-digestive tract (UADT), colorectum, liver and female breast. Also, evidence was suggestive of an association between alcohol use and cancer of the pancreas, while for cancer of the kidney and non-Hodgkin lymphoma, lack of carcinogenicity was observed (Cogliano et al., J Natl Cancer Inst, 2011).

The burden on cancer incidence attributable to alcohol intake was recently evaluated using evidence from the EPIC study (Schutze et al., BMJ, 2011), where former and current alcohol consumption was found to be responsible for 10% of the incidence of total cancer in men and 3% in women. In parallel, alcohol intake has been associated with increased risk of death from some cancers, notably cancers of the upper-aero digestive tract, colorectum, liver and breast (Bergmann et al., Int J Epidemiol, 2013), not to mention digestive tract conditions, from liver cirrhosis, chronic pancreatitis, hypertension, injuries and violence (Corrao et al., Prev Med, 2004b; Rehm et al., Addiction, 2010; Thun et al., N Engl J Med, 1997). A recent IARC study based on a cohort of 150,000 adults from Russia found that men who drank three or more bottles of vodka a week had much higher risks of death than men who drank less than one bottle a week (Zaridze et al., Lancet, 2014). In the EPIC study lifetime alcohol intake was significantly associated to mortality overall and to deaths attributed to alcohol related cancers, whereas no association was observed with CVD mortality among drinkers (Ferrari et al., BMJ OPEN, submitted).

The present work is aimed at estimating the burden of alcohol consumption on site-specific cancer incidence and total mortality using data from the Cohort consortium. Dose-response relationships will be investigated using baseline and lifetime alcohol uses.

- Investigate the association between alcohol use and risk of cancer of the upper-aerordigestive tract (UADT), female breast, liver, colorectum, kidney and all other cancers, as well as overall mortality;

- Evaluate whether the association is differential with respect to smoking status, i.e. in never and current smokers, particularly for smoking related-cancers (UADT, colorectum);

- Lifetime and baseline alcohol use will be investigated, and specific aspects of alcohol use, i.e. binge drinking, alcohol subtypes, age at start drinking;

- Absolute risks related to incidence and mortality will be estimated for combinations of categories of alcohol use, age, Region (North America, Europe, Asia, etc.) and specific cancer sites.

- The burden of alcohol with respect to cancer incidence and overall mortality will be estimated

Data from the Cohort Consortium will be used to systematically evaluate the associations between alcohol use and site-specific cancer incidence using Cox models. Similar models will be employed for overall mortality. A strategy involving augmented data within a competing risks framework will be carried out to investigate whether associations with alcohol use are heterogeneous across cancer sites.

Whereas evidence on alcohol use and cancer based on retrospective studies flawed, the anticipated large size of the Cohort Consortium will provide the ideal setting to deeply investigate specific features of alcohol use in relation to cancer risk and mortality in a prospective setting. A consortium approach allows the extent of alcohol on disease risk to be evaluated in in non-smokers, by socio-economic status, by sex and region.

Information from any prospective study will be used

Incidence data on cancer-specific sites and overall mortality.

Baseline and lifetime (when available) alcohol use.

Chosen based on a priori knowledge: centre/country/study, age at baseline, age at first cancer or censoring, age of death, sex, smoking status and intensity, anthropometric factors (weight, height, BMI), dietary factors (total energy intake, main food groups), baseline prevalent conditions (cancer, diabetes, CVD, etc), physical activity, education level, smoking status. The list pf confounders will be adapted to specific analysis, for example, reproductive factors will be considered when focusing on breast cancer. Overall, a conservative approach will be used in the list of factors to be included in aetiological models during statistical analyses.

No