Skip to Main Content

An official website of the United States government

Examining the association between the dietary inflammatory index (DII) and pancreatic cancer risk and potential mediation by type-2 Diabetes Mellitus (DM)

Principal Investigator
Name
Susan Steck
Degrees
PhD, MPH, RD, FAND
Institution
University of South Carolina
Position Title
Associate Professor
Email
About this CDAS Project
Study
PLCO (Learn more about this study)
Project ID
PLCO-171
Initial CDAS Request Approval
Oct 23, 2015
Title
Examining the association between the dietary inflammatory index (DII) and pancreatic cancer risk and potential mediation by type-2 Diabetes Mellitus (DM)
Summary
Background: Substantial biologic evidence supports that inflammation is causally linked with the pathogenesis of PC.1-3 Dietary factors have consistently been shown to affect inflammation through several biologic mechanisms.4 So far two case-control studies examined the inflammation potential of diet [using dietary inflammatory index (DII) developed at University of South Carolina] and PC risk and both reported significantly higher PC risk among subjects in the highest quintile of DII compared to those in the lowest.5,6 Our proposed study plans to utilize the data from PLCO to investigate the DII and PC association prospectively, and investigate the effect modification by sex, body mass index (BMI), smoking and nonsteroidal anti-inflammatory drugs (NSAIDs) use as these factors are also associated with inflammation.7-9 In addition, to further elucidate the pathway from DII to PC and given type-2 DM is a persistent inflammatory state,10 and associated with PC risk,11 we propose to investigate if the DII-PC association is mediated through type-2 DM. We will pool data from PLCO and the NIH-AARP study (NIH- American Association of Retired Persons Diet and Health study) which is another large prospective cohort to obtain a large enough sample size of PC cases to perform the mediation analyses, given these two studies have very similar study populations and dietary questionnaires.
Method: The development and construct validation of the DII have been described previously.12-15 The main exposure of DII score will be calculated based on inflammatory effect scores for each food parameter included in the DII and diet data from the PLCO diet history questionnaire (DHQ). Cox proportional hazard modeling will be used to estimate the hazard ratios and 95% confidence intervals for incident exocrine PC risk by quintiles of DII, and additional Cox models including effect modification by sex, BMI, smoking and NSAIDs use will also be established separately. For the pooled analysis, we will use causal mediation analysis approach with interaction between mediator incident type-2 DM and DII to assess the mediation effect on the DII and PC association.16 Type-2 DM will be first regressed on DII using logistic regression with adjustment for confounding variables. PC will be subsequently regressed on type-2 DM and DII with inclusion of DII and type-2 DM interaction and confounding variables using a multiple logistic regression model.17 Parameters derived from integration of the two models give way to three essential components (controlled direct effect, natural direct effect and natural indirect effect) used to calculate the mediation proportion by type-2 DM by using a SAS macro developed by Valeri and VanderWeele.18

Significance: Evidence form our study may suggest a potential prevention approach for PC from the dietary inflammatory perspective and also provide clues for the mechanism of this association.
Aims

1. Investigate the association between DII and PC risk using Cox proportional hazard modeling. DII will be treated as both continuous variable and categorized as quintiles to examine the dose-response relationship (first manuscript)
2. Investigate if the association between DII and PC risk is modified by sex, BMI, smoking and NSAIDs use. As BMI, smoking information, NSAIDs was updated in PLCO supplemental questionnaire, we plan to use three approaches (baseline, updated and average of baseline and updated data) in the separate analysis to compare results. For the variable smoking, we will use duration of smoking, number of cigarette smoked per day, pack-years of smoking, smoking status, duration of exposure to second-hand smoking, and other tobacco use information, to examine the interaction with DII on PC risk separately. For NSAIDs use at baseline, we will combine aspirin and ibuprofen use to create a new variable representing amount of pills taken per day from any of the two anti-inflammatory drugs. (first manuscript)
3. Pool data from PLCO and NIH-AARP to investigate the causal mediation by incident type-2 DM on the association between DII and PC risk. (second manuscript)

Collaborators

PI: Susan Steck, University of South Carolina
Co-investigators:
Jiali Zheng, University of South Carolina
James Hebert, University of South Carolina
Anwar Merchant, University of South Carolina
Nitin Shivappa, University of South Carolina
Azza Shoaibi, University of South Carolina
Rachael Stolzenberg-Solomon, National Cancer Institute
Michael Wirth, University of South Carolina
Jiajia Zhang, University of South Carolina

Related Publications