Association of Traditional and Novel Lipid Indicators With the Hemorrhagic Phenotype in Adult Moyamoya Disease: Implications for Phenotype Stratification.

To investigate the associations of traditional and novel lipid parameters with the hemorrhagic phenotype in adults with primary moyamoya disease (MMD) and to evaluate their value for phenotype discrimination and stratification. This retrospective dual-center cross-sectional study included 1,176 adults with primary MMD treated at Beijing Hospital and Beijing Tiantan Hospital between January 2022 and January 2026, including 857 patients with a non-hemorrhagic phenotype and 319 with a hemorrhagic phenotype. Traditional and derived lipid parameters were analyzed. Missing body mass index values were handled using multiple imputation. Multivariable logistic regression, restricted cubic spline analysis, receiver operating characteristic analysis, incremental discrimination analysis, sensitivity analyses, subgroup analyses, and exploratory mediation analysis were performed. Compared with patients with the non-hemorrhagic phenotype, those with the hemorrhagic phenotype had lower body mass index and higher levels of several cholesterol-related lipid parameters. In fully adjusted models, total cholesterol, low-density lipoprotein cholesterol, and non-high-density lipoprotein cholesterol were significantly associated with the hemorrhagic phenotype, with odds ratios for the highest versus lowest quartile of 3.435, 3.197, and 3.150, respectively. These indicators showed modest individual discriminative ability and provided limited incremental improvement when added to the basic clinical model. Sensitivity and subgroup analyses were generally consistent. Exploratory analyses identified BMI-related negative indirect effects. Cholesterol-related lipid parameters, particularly total cholesterol, low-density lipoprotein cholesterol, and non-high-density lipoprotein cholesterol, were associated with the hemorrhagic phenotype in adult primary MMD and may serve as supplementary markers for phenotype stratification.