Fred Tabung

PhD, MSPH

The Ohio State University College of Medicine

Assistant Professor

PLCO (Learn more about this study)

PLCO-553

Nov 25, 2019

Associations of Dietary Inflammatory and Insulinemic Potential with Risk of Cancer Development

Systemic inflammation and hyperinsulinemia play an important role in cancer development including prostate cancer. Previously, we developed the empirical dietary inflammatory pattern (EDIP) and empirical dietary index for hyperinsulinemia (EDIH) scores to assess the potential of diet to contribute to systemic inflammation and to hyperinsulinemia. Each score consist of a weighted sum of 18 food groups identified from the Nurses’ Health Study (NHS). Our objectives are to; (a) conduct a construct validation of the two dietary scores using biomarker data in PLCO (c-peptide and other insulin/IGF signalling markers for EDIH and inflammatory markers for EDIP), stratifying by race/ethnicity and (b) calculate each dietary score at baseline and examine associations with total PCa and PCa subtypes (total PCa, advanced, lethal, fatal and by Gleason’s grade) stratifying by race/ethnicity. With antecedent dietary information and more than 2.6 million biological samples collected, the PLCO Trial is a unique and valuable resource for etiologic and early marker studies. We identified 23 circulating biomarkers of systemic inflammation and insulin response/IGF signaling that have been assessed in prior studies within the PLCO cohort that we plan to include in our proposed analysis. We will assess EDIH and EDIP scores from the semi-quantitative Food Frequency Questionnaire (FFQ) based Diet Questionnaire (DQX) and diet History Questionnaite (DHQ). We will use dietary data from the questionnaires closest to the blood draw. To assess the association of EDIP and EDIH scores with concentrations of biomarkers, we will conduct age-adjusted and multivariable-adjusted linear regression analyses, fitting separate models for each biomarker. For analyses of linear trend, we will use the EDIP and EDIH quintile medians assigned to each participant, in multivariable-adjusted models and interpret the P value of the ordinal variable as the P value for linear trend. All tests will be two-sided. With 23 outcome markers, we will adjust for P value for statistical significance using the Bonferroni approach (0.05/23=0.002). Cox proportional hazards (PH) regression models will be used to calculate HRs, 95% CIs and linear trends for risk of PCa and Pca subtypes, in EDIP and EDIH quintiles, and with adjustment for covariates. Tests of linear trend between total PCa and PCa subtypes incidence and increments of EDIP and EDIH scores adjusted for covariates will be computed by assigning the median value of each quintile to each participant in the quintile, and this variable will be entered into models as ordinal values. All analyses will be done using SAS version 9.4. All tests will be two-sided. Statistical significance will be defined by P < 0.05. We expect the inflammatory and insulinemic potential of diet to significantly predict circulating concentrations of inflammatory and insulin/IGF signaling biomarkers and also the EDIP and EDIH scores to be predictive of prostate cancer risk with differences by race/ethnicity.