Abstract / Summary
The atherogenic index of plasma (AIP) is a recognized predictor of cardiovascular risk. While vitamin D status is linked to metabolic outcomes, its specific relationship with AIP in diabetic populations remains unclear. This study evaluated the potential non-linear relationship between serum 25-hydroxyvitamin D [25(OH)D] levels and AIP. We analyzed 285 patients with type 2 diabetes (median age 64 years) in this cross-sectional study. A generalized additive model (GAM) with smoothing curve fitting was applied to identify non-linear associations, followed by a threshold-effect model to calculate a statistical threshold. Spearman’s correlation and subgroup analyses further assessed the stability of the relationship across clinical strata. Smoothing curve fitting revealed an “L-shaped” non-linear association, with a statistical threshold identified at 12.3 ng/mL (95% CI: 11.3–13.7; Likelihood Ratio Test P = 0.019). Below this exploratory threshold, 25(OH)D was significantly and inversely associated with AIP (β = −0.021, P = 0.0049); however, this association plateaued at higher 25(OH)D levels (β = 0.000, P = 0.9635). Compared to the higher group, patients with 25(OH)D < 12.3 ng/mL presented with significantly elevated HbA1c ( P = 0.025) and higher AIP (0.20 vs. 0.13, P = 0.020). Subgroup analysis indicated this inverse correlation was consistent across age and BMI categories ( P for interaction > 0.05). Serum 25(OH)D levels were non-linearly associated with AIP in this cohort of T2DM patients, with an exploratory statistical breakpoint identified at 12.3 ng/mL. The inverse association between vitamin D and AIP was most pronounced in patients with severe deficiency but appeared to plateau once levels exceeded this threshold. These observational, hypothesis-generating findings suggest that the potential link between vitamin D and atherogenic markers may be concentrated in severely deficient populations; however, the identified 12.3 ng/mL threshold is a data-derived exploratory value that requires external validation in independent populations before its potential clinical relevance can be established. Not applicable.