The Development and Validation of Dietary Inflammatory Index with hs-CRP and c-IMT Using Food Frequency Questionnaire in Adults Aged 40 to 79 based on the MRCohort

Abstract

Inflammation, an important risk factor for some chronic diseases, may result from a pro-inflammatory diet. Efforts have been made to create dietary inflammatory indices (DII) using mainly high sensitivity C-reactive protein (hs-CRP) have been developed to predict inflammatory response, which has several limitations to apply in the Korean population. In addition, common carotid arterial (CCA) intima medial wall thickness (c-IMT) is a well-known bio-indicator of chronic diseases, but there was no dietary atherogenic index (DAI). Therefore, we intended to develop and validate the DII and DAI using a validated food frequency questionnaire which was developed for the Korean Genome and Epidemiology Study (KoGES). Of the 3,564 adults aged ≥ 40 from Yangpyeong cohort Study, only those who were free of diagnosed cardiovascular disease and cancer, and participated in both the baseline (2005-2009) and 2nd visit. DII and DAI were developed in a total of 1,373 subjects, where excluded cases where the history of diagnosing and medication in type 2 diabetes and hypertension, ≥ 80 years, implausible dietary intake, and hs-CRP or c-IMT has missing or outlier values. A total of 106 food items from the food frequency questionnaire (FFQ) were re-classified into 38 food groups, with food groups as predictors and hs-CRP, IMT as response variables the DII and DAI were developed using stepwise linear regression (SLR) and reduced rank regression (RRR), respectively. We validated the association between DII/DAI and hs-CRP/c-IMT, respectively in total, men, and women by using the bootstrapping method, where excluded missing values in general characteristics (N = 1,346) for internal validation, and by using the Namwon, Goryeong cohort data set (N = 2,455) for external validation. In development, DII included seven by SLR, eleven by RRR food groups, and DAI included seventeen by SLR, eight by RRR food groups, respectively. Each index was calculated as a sum of weights from regression * corresponding intake as serving per day. In internal validation set, total 1000 sets of replicated DIIs and DAIs (total number of participants = 1,346 * 1000) were analyzed, the means of the hs-CRP and c-IMT in the original development data set were almost included in the 95% CI of bootstrapped values. Both appear to be included in bootstrapped CI, but the value of hs-CRP according to DII appears more clear than c-IMT according to DAI. In external validation, the consistency up to model 3 was significant only in DAI models using SLR (S-DAI), regardless of gender (p-trend in total = 0.003, p-trend in men = 0.0087, p-trend in women = 0.0165). In this study, we developed DII and DAI as a prediction index of the chronic inflammatory disease using SLR and RRR with FFQ data and hs-CRP, c-IMT, respectively. However, DIIs and R-DAI were not valid in external validation, S-DAI was the one that was valid in both validations. It is to say, we developed a novel, hypothesis-data-based named S-DAI, and was associated with c-IMT in the internal and external validation significantly. Therefore, S-DAI may be applied as a dietary index to evaluate the association with phenotypes of chronic diseases.