Validation of commercially available ELISAs for the detection of circulating sclerostin in hemodialysis patients

Abstract

Background: Sclerostin is an endocrine regulator in chronic kidney disease - mineral and bone disorder (CKD-MBD). Validation of assay comparability and pre-analytical handling is mandatory for establishment of sclerostin as a biomarker.

Methods: Blood samples (serum, EDTA, heparin and citrate plasma) were obtained from 12 hemodialysis (HD) patients after the long dialysis interval. Passing-Bablok regression analysis and Bland-Altman difference plots were used to evaluate the agreement between sclerostin levels measured with two commercially available ELISAs from TECOmedical and Biomedica.

Results: Independent of the sample type, the agreement of the two assays was poor with a strong proportional but no systematic bias. Compared to the TECOmedical assay, the Biomedica test yielded almost 2-fold higher sclerostin values throughout all sample types. Spike recovery and linear dilution studies revealed a higher accuracy of the TECOmedical assay (97% and 96%) compared to the Biomedica assay (118% and 78%). Sclerostin levels were stable within 4 hours after sample collection, in particular when analyzed in plasma. In contrast to the Biomedica assay, the TECOmedical showed a systematic but no proportional bias between serum and plasma samples with higher values for plasma samples. Among the 3 different plasma samples no systematic error could be documented.

Conclusion: Careful consideration of the pre-analytical handling and comparative assay validation are necessary to facilitate a more differentiated interpretation of studies reporting circulating sclerostin levels. The presence of a proportional bias demonstrates that in HD patients the two ELISAs for measuring sclerostin should not be used interchangeably. Furthermore, caution is necessary when comparing sclerostin results obtained from different blood sample types.

Keywords: CKD-MBD; biomarker; haemodialysis; sclerostin; vascular calcification.

Figure 1

Sclerostin levels as determined by the Biomedica (black bars) and TECOmedical (grey bars) ELISAs in various sample types as indicated. Data represent mean ±SD

Figure 2

Bland-Altman difference plots visualizing the difference of the measurements obtained with the Biomedica and TECOmedical ELISAs. The differences or the bias (Biomedica – TECOmedical; Y-axis in ng/mL) is plotted versus the average of the two readings (Biomedica + TECOmedical / 2; X-axis in ng/mL). As an increase in variability of the differences is observed as the magnitude of the measurement increases, the lower right Bland-Altman plot displays the relationship between the average and the percentage of difference. Thin continuous horizontal lines represent the mean, thin dotted horizontal lines indicate the 95% limits of agreement (average difference ± 1.96 SD of the difference).

Figure 3

Plots displaying the Passing-Bablok regression analysis of the agreement between the Biomedica and TECOmedical ELISAs for various sample types as indicated. The solid line indicates the regression line, the black dotted lines show the line of identity and the gray dashed lines represent the 95% confidence interval for the regression line.

Figure 4

Figure 4A. Plots displaying the Passing-Bablok regression analysis of the agreement between the various samples types as indicated. Analysis was performed for results obtained by the Biomedica. The solid line indicates the regression line, the black dotted lines show the line of identity and the gray dashed lines represent the 95% confidence interval for the regression line. Figure 4B. Plots displaying the Passing-Bablok regression analysis of the agreement between the various samples types as indicated. Analysis was performed for results obtained by the TECOmedical ELISA. The solid line indicates the regression line, the black dotted lines show the line of identity and the gray dashed lines represent the 95% confidence interval for the regression line.

Figure 4

Figure 4A. Plots displaying the Passing-Bablok regression analysis of the agreement between the various samples types as indicated. Analysis was performed for results obtained by the Biomedica. The solid line indicates the regression line, the black dotted lines show the line of identity and the gray dashed lines represent the 95% confidence interval for the regression line. Figure 4B. Plots displaying the Passing-Bablok regression analysis of the agreement between the various samples types as indicated. Analysis was performed for results obtained by the TECOmedical ELISA. The solid line indicates the regression line, the black dotted lines show the line of identity and the gray dashed lines represent the 95% confidence interval for the regression line.