Fig. 2: ImP is associated with subclinical atherosclerosis in humans.

a,c, Plasma ImP in healthy individuals (Ctrl) and individuals with subclinical atherosclerosis (AT) from the PESA (a; Ctrl, n = 105; AT, n = 295) and IGT (c; Ctrl, n = 529; AT, n = 1,315) cohorts. b,d, Dose–response curves of plasma ImP concentration and endpoints in the PESA (b) and IGT (d) cohorts. e,f, Spearman correlation matrices between ImP and atherosclerosis traits, diet and microbiota in the PESA (e) and IGT (f) cohorts. Benjamini–Hochberg adjusted P values. ALT, alanine aminotransferase; AST, aspartate aminotransferase; BM, bone marrow; ECO2D, 2D vascular ultrasound; ECO3D, 3D vascular ultrasound; F, familial; GFR, glomerular filtration rate; GGT, gamma-glutamyl transferase; HTN, hypertension. g, Adjusted regression models for the association of ImP with atherosclerosis (left) and extent of atherosclerosis (right) in the PESA (top, n = 400) and IGT (bottom, n = 1,844) cohorts. Effect estimates were controlled for age, sex, smoking, creatinine, family history of CVD, haemoglobin, hypertension and LDL-C (PESA cohort) and age, smoking, family history of CVD, hypertension, LDL-C, Hb1Ac, hs-CRP and ALT (IGT cohort). Error bars show 95% confidence intervals. OR, odds ratio. h,j, Plasma ImP in individuals with inactive atherosclerosis (FDG⁻, n = 74) and active atherosclerosis (FDG⁺, n = 221) (h). FDG⁺ individuals were further stratified according to vascular inflammation (arterial ¹⁸F-FDG uptake, FDG⁺_A, n = 57), bone marrow activation (¹⁸F-FDG uptake in bone marrow, FDG⁺_BM, n = 40) and systemic inflammation (concurrent arterial and bone marrow ¹⁸F-FDG uptake, FDG⁺_SYS, n = 124) (j). FDG^– refers to the group of inactive atherosclerosis. a,c,h,j, Horizontal line represents median and error bars show interquartile range (Supplementary Tables 1a, 2a and 4). Two-tailed Mann–Whitney U-test. i, Dose–response curve for plasma ImP effect on active atherosclerosis for the PESA cohort. b,d,i, Dashed lines represent the 95% confidence interval; vertical lines delineate tertiles. k, Multinomial logistic regression for subclinical atherosclerosis in the atherosclerosis subgroups (FDG⁺_SYS, FDG⁺_BM, FDG⁺_A and FDG⁻) versus controls according to plasma ImP. Odds ratio was adjusted for age, sex, smoking, glucose, hs-CRP and haemoglobin concentration. g,k, Whiskers indicate 95% confidence intervals (values in Supplementary Tables 3 and 5). ***P < 0.001.

Source Data