Comparison of three different methods for 25(oh)-vitamin d determination and vitamin d status in general population – Serbian experience / Poređenje tri različite metode za određivanje 25(oh)-vitamina D i statusa vitamina D u opštoj populaciji – srpsko iskustvo

Abstract

Summary Determination of 25-hydroxyvitamin D [25(OH)D] represents a unique challenge, considering its lipophilic na - ture. Considering the widespread prevalence of vitamin D deficiency, which leads to increasing number of requests for 25(OH)D determination, immunoassay measurements adjusted to automated analyzers are being developed. Because of the variability among assays, it is often difficult to monitor vitamin D status and supplementation. The aim of this study was to compare the results of two immunoassays with high performance liquid chromatography with ultraviolet detection (HPLC-UV). Also, the aim was to estimate vitamin D status, since up to date the prevalence of vitamin D deficiency in Serbia was not examined. We have evaluated analytical characteristics of two automated immunoassays for 25(OH)D determination, from Roche (CobasR e601) and Abbott (Architect). For comparison studies we used HPLC analysis of 25-(OH)-Vitamin D3/D2 from Chromsystems (Waters isocratic system). In order to estimate vitamin D status in general population, we have searched the database of the laboratory information system and analyzed the data from 533 patients whose 25(OH)D was determined together with intact parathyroid hormone (iPTH). For imprecision assessment, four serum pools were prepared with following 25(OH)D concentrations: 35 nmol/L, ∼50 nmol/L, ∼75 nmol/L and ∼125 nmol/L. Obtai ned CVs for Roche method were 1.5-2.8% for within-run and 4.0-6.7% for between-run imprecision. For Abbott method, CVs were 0.7-4.4% for within- run and 3.8-7.2% for between-run imprecision. Inaccuracy was analyzed with commercial control sera. Obtained deviations from target value were 2.1% for Roche assay and 1.3-1.5% for Abbott method, and were not statistically significant (P>0.05). Comparison of Roche and HPLC-UV methods using Passing-Bablok regression analysis gave the following equation for the regression line y=0.937x+9.518 (r=0.739; n=97) and the regression line equation from the comparison of Abbott and HPLC-UV methods was y=0.745x+10.343 (r=0.793; n=97). Mean difference and SD for Bland-Altman plot were -4.5 nmol/L and 21.75 nmo/L, respectively for Roche method and 6.4 nmol/L and 18.8 nmol/L, respectively for Abbott. Statistical analysis (Chi-square test) of frequency distribution among different vitamin D status categories (<25 nmol/L severe deficiency, 25-50 nmol/L deficiency, 50-75 nmol/L insufficiency and >75 nmol/L sufficiency) showed that the frequency distribution obtained with Abbott method was significantly different from the distribution of the HPLC results, in contrast to Roche results frequency distribution which did not differ significantly. Also, statistical analysis of the agreement between the three methods for each vitamin D status category showed that results of both Roche and Abbott methods were significantly higher than HPLC in the two deficiency categories (P=0.005 for Roche, P=0.0407 for Abbott), and in the sufficiency category Abbott method significantly underestimated concentration of 25(OH)D compared to HPLC results (P<0.0001). Median population values of 25(OH)D and iPTH were 41.8 nmol/L and 76.6 ng/L, respectively. ANOVA analyses showed significant (P<0.05) decrease in iPTH and Ca2+ concentrations across the 25(OH)D concentration categories. Stepwise multiple linear regression analysis indicated independent correlation of iPTH with 25(OH)D concentration (b=-0.290, P=0.0008). Also, one-way ANOVA with Student-Newman-Keuls test demonstrated that 25(OH)D concentrations measured in summer and autumn were significantly (P<0.001) higher compared to those determined in winter and spring. Despite acceptable imprecision and inaccuracy of both examined methods, results obtained with them did not correlate well with HPLC-UV (r<0.9), which was used as a reference. However, methods showed satisfactory ability to classify patients into vitamin D status categories, which is important for diagnosis of vitamin D deficiency and therapy follow-up. About two thirds (68.5%) of the examined po pulation had vitamin D deficiency (25(OH)D<50 nmol/L) and only 8% had sufficient 25(OH)D concentration (>75 nmol/L).