Comprehensive Comparison of Three Automated 25-Hydroxyvitamin D Immunoassays with a VDSCP Traceable LC-MS/MS Method Across Clinically Relevant Decision Thresholds

Background/Objective: Accurate measurement of serum 25-hydroxyvitamin D (25OHD) is essential for the diagnosis and management of vitamin D deficiency. Although automated immunoassays are widely used in routine clinical practice, analytical variability between platforms may lead to inconsistent patient classification around clinically relevant decision thresholds. This study comprehensively compared three automated immunoassays (DiaSorin Liaison XL, Beckman Coulter DxI 9000, and Roche cobas e602) with a Vitamin D Standardization Certification Program (VDSCP)-traceable liquid chromatography–tandem mass spectrometry (LC-MS/MS) reference method. Methods: A total of 400 anonymized adult outpatient serum samples were prospectively collected and equally distributed across four predefined vitamin D concentration categories (<25, 25–49, 50–74, and ≥75 nmol/L). Samples were stored at −70 °C before simultaneous analysis on the three automated immunoassays and the reference LC-MS/MS platform. Method comparison included Deming regression, Bland–Altman analysis, and evaluation of diagnostic performance at clinically relevant thresholds (<25, <50, and <75 nmol/L). The effects of sample storage, vitamin D metabolites (C3-epi-25OHD3 and 24,25-dihydroxyvitamin D3), and external quality assessment (DEQAS) performance were also investigated. Results: Significant analytical differences were observed between all immunoassays and LC-MS/MS. Roche cobas e602 consistently overestimated 25OHD concentrations, resulting in reduced sensitivity but excellent specificity and positive predictive value across all clinical thresholds. Beckman Coulter demonstrated the highest sensitivity for detecting vitamin D deficiency but generated more false-positive classifications, particularly at the <75 nmol/L threshold. DiaSorin Liaison XL achieved the best overall balance between sensitivity, specificity, accuracy, and agreement with LC-MS/MS, with Cohen’s κ values ranging from 0.818 to 0.895 across decision thresholds. Higher concentrations of C3-epi-25OHD3 and 24,25-dihydroxyvitamin D3 were associated with greater positive bias in Roche measurements, while Beckman Coulter showed no significant metabolite-related interference, although the present correlation analyses could not establish independent analytical interference. DEQAS performance did not consistently reflect agreement with LC-MS/MS in patient samples. Conclusions: Despite ongoing international standardization efforts, clinically important differences remain among automated 25OHD immunoassays. Assay-specific analytical characteristics significantly influence patient classification around established decision thresholds. DiaSorin Liaison XL demonstrated the closest overall agreement with the reference LC-MS/MS method, whereas Beckman Coulter prioritized sensitivity and Roche specificity. These findings highlight the importance of method-aware interpretation of vitamin D results and support continued assay harmonization to improve comparability in routine clinical practice.