The glycemic index and insulin index assess postprandial glycemic and insulin response to foods, respectively, which may not reflect long‐term effects of diet on glycemia or insulinemia. We developed and validated three empirical dietary indices (EDI) to assess the insulinemic potential of usual diet:hyperinsulinemia (EDIH), insulin resistance (EDIR) and both hyperinsulinemia and insulin resistance (CDHR), using dietary and biomarker data from the Nurses’ Health Study (NHS, n=5,986 for EDIH, n=3,988 for EDIR). We entered 39food frequency questionnaire‐derived food groups in stepwise linear regression models and defined the indices as dietary patterns most predictive of C‐peptide(EDIH) and triglycerides/high density lipoprotein‐cholesterol (TG/HDL) ratio (EDIR). The CDHR was derived by combining components of the EDIH and EDIR. Indices were validated in two independent samples from NHS‐II and Health Professionals Follow‐up Study (HPFS) using C‐peptide and TG/HDL as construct validators in multivariable‐adjusted linear regression analyses. The EDIH comprised of 19 food groups;11 were positively and eight were negatively associated with C‐peptide concentrations. The EDIR comprised of 18 food groups; nine positively and nine negatively associated with TG/HDL ratio, while twenty‐six unique food groups comprised the CDHR. In NHS, the Pearson correlation coefficient was 0.22 for EDIH and C‐peptide and 0.31 for EDIR and TG/HDL ratio. Similar correlations were found in HPFS and NHS‐II. In HPFS and NHS‐II, all three indices significantly predicted biomarker concentrations. E.g., comparing extreme index quintiles (Q5/Q1)(95%CI) biomarker concentrations in HPFS were: EDIH, 1.25 (1.18, 1.33), ptrend<0.0001; EDIR, 1.42(1.32, 1.52), ptrend<0.0001;CDHR, 1.26 (1.19, 1.34), ptrend<0.0001for C‐peptide and 1.31 (1.22, 1.41), ptrend<0.0001 for TG/HDL. The strong validity of these novel hypothesis‐driven indices in independent populations suggests their usefulness in assessing the ability of whole diets to stimulate and/or sustain insulin secretion.Support or Funding InformationDrs. Jorge E. Chavarro and Frank B. Hu were supported by National Institutes of Health (NIH) grants P30DK046200 and U54 CA155426. The HPFS, NHS and NHS‐II cohorts are supported by the following NIH grants: UM1 CA 167552, UM1 CA 186107 and UM1 CA 176726 respectively. Mulltivariable‐adjusted11 relative concentrations2 of biomarkers in quintiles of insulinemic dietary patterns in the NHS, HPFS and NHS‐II3 cohorts Quintile1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 4Ptrend Empirical dietary index for hyperinsulinemia (EDIH) C‐peptide (NHS, n=5,986) 1.00 1.11 (1.07, 1.16) 1.20 (1.15, 1.25) 1.25 (1.20, 1.31) 1.42 (1.35, 1.48) <0.0001 C‐peptide (HPFS, n=3,984) 1.00 1.02 (0.97, 1.08) 1.18 (1.11, 1.25) 1.17 (1.11, 1.24) 1.25 (1.18, 1.33) <0.0001 TG/HDL ratio (HPFS, n=3,559) 1.00 1.10 (1.01, 1.16) 1.12 (1.04, 1.20) 1.24 (1.15, 1.33) 1.26 (1.18, 1.36) <0.0001 C‐peptide (NHS‐II, n=1,736) 1.00 1.08 (0.99, 1.18) 1.09 (1.00, 1.19) 1.21 (1.10, 1.32) 1.30 (1.19, 1.43) <0.0001 Empirical dietary index for insulin resistance (EDIR) TG/HDL ratio (NHS, n=3,988) 1.00 1.17 (1.10, 1.25) 1.31 (1.23, 1.40) 1.42 (1.32, 1.52) 1.67 (1.56, 1.79) <0.0001 TG/HDL ratio (HPFS, n=3,559) 1.00 1.17 (1.09, 1.25) 1.21 (1.12, 1.29) 1.38 (1.29, 1.48) 1.42 (1.32, 1.52) <0.0001 C‐peptide (HPFS, n=3,984) 1.00 0.99 (0.93, 1.05) 1.07 (1.01, 1.13) 1.09 (1.03, 1.16) 1.14 (1.07, 1.21) <0.0001 Composite dietary index for hyperinsulinemia and insulin resistance (CDHR) C‐peptide (HPFS, n=3,984) 1.00 1.06 (1.00, 1.12) 1.08 (1.02, 1.15) 1.14 (1.08, 1.21) 1.26 (1.19, 1.34) <0.0001 TG/HDL ratio (HPFS, n=3,539) 1.00 1.10 (1.03, 1.18) 1.14 (1.06, 1.22) 1.23 (1.15, 1.33) 1.31 (1.22, 1.41) <0.0001 Multivariable models were adjusted for regular aspirin/NSAIDs use, age, smoking status, chronic diseases/conditions, case‐control status; in women, models were additionally adjusted for menopausal status and postmenopausal hormone use; relative concentrations are the concentrations of biomarkers in higher dietary density index quintiles divided by the reference quintile (Q1); NHS=Nurses’ Health Study and HPFS=Health Professional Follow‐up Study; the p‐value for trend was the p‐value of the continuous dietary index Variable for the potential confounders listed in footnote #1.
Read full abstract