This preprint investigated whether a pro-inflammatory dietary pattern, measured by the Dietary Inflammatory Index (DII), is associated with metabolic syndrome (MetS) in a sex-specific manner and whether genetic evidence supports an inflammatory pathway. The authors also assessed whether the DII demonstrates consistent construct validity across different national dietary contexts.
The analysis harmonised individual participant data (IPD) from three nationally representative surveys: US NHANES, Korean KNHANES, and Mexican ENSANUT, yielding 56,475 adult participants. Primary observational analyses used survey-weighted, sex-stratified logistic regression models to estimate associations between DII and MetS. The DII scoring relied on the 20-item construct derived from the global reference database of Shivappa et al. (2014).
To appraise causality and the inflammatory pathway, the authors implemented three complementary genetic approaches: a CRP-based Mendelian randomization (MR) analysis (primary instrument set: 84 female-matched SNPs; sensitivity subset: 56-SNP sex-overlap), exploratory direct diet MR, and a proxy gap decomposition to appraise instrument-to-exposure gaps. Country-specific population attributable fractions (PAFs) quantified potential population-level diet-attributable MetS burden by nation. The authors reported both multiplicative interaction (sex × DII) and additive interaction (RERI).
CRP-based MR was the principal genetic strategy to evaluate whether systemic inflammation mediates a diet–MetS relationship. The primary CRP instrument set comprised 84 female-matched SNPs; a sensitivity analysis used a 56-SNP sex-overlap subset. MR results included inverse-variance weighted (IVW) estimates and pleiotropy-robust MR-RAPS estimates. An exploratory direct diet MR was performed to assess consistency of directionality when instruments more directly capture diet-related exposures. The authors used external GWAS summary statistics for CRP and MetS and LD clumping with the 1000 Genomes Phase 3 EUR panel.
Observational IPD analyses identified a robust female-specific association between a pro-inflammatory diet and MetS. The overall female OR for the DII–MetS association was 1.15 (P < 0.0001). Comparing extremes of DII quintiles, Q5 versus Q1 in women yielded OR = 1.56. In men, associations were null. The multiplicative sex interaction was statistically significant (P < 0.0001). However, the additive interaction measured by RERI was not significant (RERI P > 0.05).
Country-specific PAFs highlighted cross-cultural differences in how the 20-item DII relates to MetS. For women, the US PAF was reported as 20.2%, indicating a substantial estimated population-level diet-attributable component, whereas Korea had a negative PAF of -8.0% and Mexico a negative PAF of -33.1%. These negative PAFs imply the DII construct may not align with inflammation-related dietary patterns in those non-Western contexts and point to a construct validity failure of the standard 20-item DII outside the population contexts in which it was developed.
Triangulation combined the observational findings and the genetic analyses. The authors report a formal Fisher combined P = 4.4e-8, while noting partial evidence-line dependency across sources. CRP-based MR provided conservative genetic support: the primary 84-SNP IVW estimate was attenuated (IVW OR = 1.03, P = 0.31), whereas the MR-RAPS method, designed to be robust to pleiotropy, indicated a modest signal (OR = 1.03–1.05, P ≤ 0.01). The exploratory direct diet MR was directionally consistent for females but did not reach conventional statistical significance (female OR = 0.91, P = 0.10).
The combined evidence supports a consistent observational, female-specific association of a pro-inflammatory diet with MetS across three nations. Genetic analyses give conservative support for the inflammatory pathway hypothesis via CRP-based MR, though IVW estimates were attenuated and MR-RAPS suggested only modest effect sizes (OR ~1.03–1.05). A salient study finding is the poor cross-national construct validity of the standard 20-item DII in Korea and Mexico, which undermines straightforward interpretation of PAFs outside the US context.
Because the report is a preprint and observational associations plus modest MR signals do not establish causality, the authors conclude that randomized trials are needed to test whether reducing dietary inflammatory potential lowers MetS risk. The significant multiplicative sex interaction alongside a non-significant additive interaction suggests that population-level dietary guidance might still capture benefits, but sex-specific effects warrant consideration in research design and public health messaging.
Supplementary results, analysis scripts (R version 4.6.1 and Python version 3.13), and aggregated data are provided as Supplementary Material accompanying the manuscript. GWAS summary statistics used for MR are cited with public repositories; LD clumping used the 1000 Genomes Project Phase 3 EUR panel. The DII calculation referenced Shivappa et al. (2014).
The manuscript is posted as a medRxiv preprint and has not been peer reviewed; the authors state no competing interests and confirm ethical approvals and participant consent. The report notes limitations inherent to cross-sectional observational IPD, cross-national harmonisation, and partial dependency across evidence lines used for triangulation. Details beyond those reported in the preprint (for example, full instrument lists, all sensitivity analyses, or trial data) were not reported in the source.