Analytical, experimental and clinical aspects of total 25-hydroxyvitamin D measurement by two ligand assays are differentially biased by disease-related factors.

Background: Pathological concentrations of plasma proteins may confound the results of binding assays. We compared two routinely used total 25-hydroxyvitamin D [t-25(OH)D] methods: a chemiluminescence-immuno-(CLIA) and an electro-chemiluminescence-protein-binding-(ECLPBA) assay.

Methods: Two sub-studies were performed: 1) In an "in vitro" study, exogenous albumin was added to pools of patients' sera with low albumin levels; and 2) In "ex vivo" studies of Cohort_1: sera of hospitalized patients with low albumin levels, and of healthy controls; and of Cohort_2: outpatients with chronic kidney disease in pre-dialysis stage, or on peritoneal dialysis and hemodialysis were investigated by the routine and LC_MS/MS methods.

Results: When increasing albumin concentrations were "in vitro" added, t-25(OH)D levels were overestimated by ECLPBA, and underestimated by CLIA. In patients' sera, positive correlations were detected between t-25(OH)D-vitamin D binding protein (DBP) values by both routine methods, and between t-25(OH)D-albumin values by all three methods. Much higher t-25(OH)D was measured by LC_MS/MS in all subgroups. When altering albumin levels with constant DBP concentration, the "in vitro" experiment revealed a higher sensitivity of ECLPBA. The "ex vivo" measurements demonstrated clinically relevant differences between the routine methods.

Conclusion: Both routine methods are dependent of the matrix effect in hospitalized patients, which is predicted by the DBP/Albumin ratios. In hemodialysis, ECLPBA is recommended because its outcomes differ less from those of LC_MS/MS. The results of LC_MS/MS are reliable, but not routinely available. A guidance would be valuable on how levels measured by the binding methods differ from those by LC-MS/MS in various clinical conditions.