Diagnostic performance of serological assays for anti-HBs testing: Results from a quality assessment program

- EQA programs are indispensable to achieve standardisation among laboratories.
- Anti-HBs assays produce different results around clinically relevant cutoff values.
- Lack of agreement between assays is mostly due to false-negative results of two assays.

BackgroundPost-vaccination testing after hepatitis B vaccination is indispensable to evaluate long-term immunological protection. Using a threshold level of antibodies against hepatitis B surface antigen (anti-HBs) to define serological protection, implies reproducible and valid measurements of different diagnostic assays.ObjectivesIn this study we assess the performance of currently used anti-HBs assays.Study designIn 2013, 45 laboratories participated in an external quality assessment program using pooled anti-HBs serum samples around the cutoff values 10 IU/l and 100 IU/l. Laboratories used either Axsym (Abbott Laboratories), Architect (Abbott Laboratories), Access (Beckman-Coulter), ADVIA Centaur anti-HBs2 (Siemens Healthcare Diagnostics), Elecsys, Modular or Cobas (Roche Diagnostics) or Vidas Total Quick (Biomerieux) for anti-HBs titre quantification. We analysed covariance using mixed-model repeated measures. To assess sensitivity/specificity and agreement, a true positive or true negative result was defined as an anti-HBs titre respectively above or below the cutoff value by ≥4 of 6 assays.ResultsDifferent anti-HBs assays were associated with statistically significant (P < 0.05) differences in anti-HBs titres in all dilutions. Sensitivity and specificity ranged respectively from 64%-100% and 95%-100%. Agreement between assays around an anti-HBs titre cutoff value of 10 IU/l ranged from 93%-100% and was 44% for a cutoff value of 100 IU/l.ConclusionsAround a cutoff value of 10 IU/l use of the Access assay may result in false-negative results. Concerning the cutoff value of 100 IU/l, a sample being classified below or above this cutoff relied heavily on the specific assay used, with both the Architect and the Access resulting in false-negative results.