Beyond sitosterol: A multivariable sterol model improves diagnosis of pediatric sitosterolemia in the gray zone

Patients with sitosterolemia (ST) are often misdiagnosed as familial hypercholesterolemia (FH) because of overlapping lipid phenotypes. Sitosterol is considered a disease-specific biomarker; however, its diagnostic utility in highly heterogeneous populations-particularly among children-remains unclear.

To evaluate the diagnostic value of phytosterol biomarkers and develop a multivariate model to improve the diagnostic accuracy of ST.

We conducted a cross-sectional study of 379 children with suspected lipid disorders: ST (n = 38), ABCG5/8 heterozygous carriers (n = 12), genetically confirmed FH (n = 54), individuals with FH-like (n = 50), and healthy controls (n = 225). Clinical characteristics, lipid profiles, phytosterols, liver enzymes, and genetic data were collected. The diagnostic performance of single biomarkers and a multivariate model was evaluated. A clinical gray zone based on sitosterol levels was defined to assess the models' ability.

Sitosterol demonstrated near-perfect discrimination for ST in the overall population area under the curve ([AUC] 0.994; 95% CI, 0.988-1.000), outperforming conventional lipid markers. However, substantial overlap in phytosterol distributions was observed among ST, heterozygotes, and FH-related phenotypes. Within the gray zone (sitosterol: 17.7-50 μg/mL), the diagnostic performance of sitosterol declined markedly (AUC 0.653). The multivariable model demonstrated comparable overall performance (AUC 0.983) but significantly improved discrimination in the gray zone (AUC 0.806), with good calibration and greater net clinical benefit.

Although sitosterol is highly effective for identifying ST, its diagnostic performance declines significantly in clinically significant gray zone cases. A combined sterol-based model improves diagnostic discrimination in these challenging scenarios and supports a stepwise diagnostic strategy for children with suspected ST.