Ting Zhang

Ph.D.

National Cancer Institute

Postdoctoral Fellow

PLCO (Learn more about this study)

PLCO-1686

Sep 30, 2024

Metabolic biomarkers of habitual alcohol use and risk of five cancers

Background: Alcohol is an established risk factor for cancers of the oral cavity, pharynx, and larynx; esophagus; liver; and female breast. The associations between alcohol intake, particularly low-to-moderate intake, and other cancers, such as pancreatic cancer, endometrial cancer, glioma and prostate cancer, have been inconsistent, with an excess risk for a higher intake (>45 grams of ethanol or 3 drinks a day) with pancreatic cancer. Alcohol-cancer association might be biased due to measurement error inherent to self-reported alcohol intake and residual confounding by smoking in observational studies. A previous study using a biomarker (i.e., 2-hydroxy-3-methylbutyric acid) of habitual alcohol intake, which may be less prone to measurement error, has suggested a positive association of the biomarker but not alcohol intake with pancreatic cancer in the European Prospective Investigation into Cancer and Nutrition study. However, whether the findings could replicate in other populations and whether other candidate alcohol biomarkers previously identified also exert associations with pancreatic cancer are unknown. It is also understudied whether alcohol biomarkers are associated with endometrial cancer, glioma, or prostate cancer risk. We aim to identify serum metabolites correlated with habitual alcohol intake, and to examine associations of the alcohol biomarkers with five cancers risk, including esophageal cancer as a positive control outcome.
Study populations: Participants from nine nested cancer case-control studies with pre-diagnostic serum metabolites measured, including two PDAC studies in PLCO (360 cases/360 controls) and ATBC (373 cases/373 controls),4 two esophageal cancer studies in PLCO (131 cases/131 controls) and ATBC (74 cases/62 controls), two glioma studies in PLCO (164 cases/165 controls) and ATBC (61 cases/64 controls), two prostate cancer in mostly whites (380 cases/380 controls)5 and Blacks (224 cases/224 controls) in PLCO, and a endometrial cancer (193 cases/193 controls) in PLCO.
Exposures: Alcohol and its subtypes (beer, liquor, wine) intake, assessed by using FFQ at baseline (DQX for PLCO).
Outcomes: Metabolomics measurements; pancreatic cancer, endometrial cancer, glioma, prostate cancer, and esophageal cancer.
Statistical analysis:
We will use partial Spearman correlation between alcohol and its subtypes intake (g/day) and metabolites (median-scaled and log transformed) in each subset and combined using fixed-effect meta-analysis (with Fisher’s r to z transformation). Covariates include age at blood draw (years), sex (PLCO), race (PLCO), smoking (ATBC: cigarettes per day and years smoked; PLCO: never smoking, former smoking quit ≥15 years, former smoking quit <15 years, current smoking), diabetes, BMI (continuous), HEI-2020 dietary scores (continuous), daily energy intake (kcal), and coffee intake (continuous). For alcohol subtypes, we will adjust for the sum of other alcohol intake. FDR Q<0.05 in both cohorts will be considered statistically significant.
We will use unconditional or conditional logistic regressions for associations of alcohol intake (per 1 SD of log values or quintiles) and alcohol-correlated metabolites with each cancer in each cohort and combined using fixed-effect meta-analysis if necessary. Model will be adjusted for age at blood draw (years) and additionally adjusted for smoking, diabetes, BMI, and daily energy intake (continuous).