Association of systemic inflammatory indices NLR, SIRI, and MHR with carotid plaque in middle-aged and older adults: a retrospective cross-sectional study

Abstract

To examine the associations of the neutrophil-to-lymphocyte ratio (NLR), systemic inflammation response index (SIRI), and monocyte-to-high-density lipoprotein cholesterol ratio (MHR) with carotid plaque in middle-aged and older adults. This retrospective cross-sectional study included 433 patients aged ≥ 40 years who underwent carotid ultrasonography at Kunshan Second People’s Hospital between January 1, 2024, and February 28, 2026. Clinical and laboratory data were collected, and NLR, SIRI, and MHR were calculated. The primary multivariable analysis used complete cases ( n = 374) because high-sensitivity C-reactive protein (hsCRP), a prespecified covariate, was missing for 59 participants (13.6%). Sensitivity analyses used the full cohort ( n = 433) in models excluding hsCRP. Associations were examined per standard deviation increase, and restricted cubic spline models evaluated dose-response patterns. NLR, SIRI, and MHR were higher in the plaque group ( n = 244) than in the non-plaque group ( n = 189) (all P < 0.001). In fully adjusted complete-case models, NLR (OR = 1.35, 95% CI: 1.02–1.79; P = 0.039) and SIRI (OR = 1.47, 95% CI: 1.05–2.07; P = 0.026) were associated with carotid plaque, whereas MHR was not (OR = 1.19, 95% CI: 0.84–1.70; P = 0.326). In full-cohort sensitivity analyses, estimates remained directionally similar but were attenuated for NLR (OR = 1.27, 95% CI: 0.99–1.62; P = 0.059) and SIRI (OR = 1.26, 95% CI: 0.96–1.64; P = 0.095). SIRI showed a graded association across quartiles (P for trend = 0.031), and no statistically significant interactions were detected. Higher NLR and SIRI were associated with carotid plaque in complete-case analyses, although the associations were attenuated in full-cohort sensitivity analyses. The modest discrimination (AUC 0.610–0.642) indicates that these indices should be considered exploratory adjunctive markers rather than standalone screening tools. Prospective validation is required.