Evaluation and validation of a commercial ELISA versus the in vitro toxin neutralization assay for determination of diphtheria anti-toxin in human serum

Abstract

Introduction. Diphtheria is a potentially life-threatening infection and remains endemic in many low- and middle-income countries (LMICs). A reliable, low-cost method for serosurveys in LMICs is warranted to estimate the accurate population immunity to control diphtheria. Hypothesis/Gap Statement. The correlation between the ELISA results against diphtheria toxoid and the gold standard diphtheria toxin neutralization test (TNT) values is poor when ELISA values are <0.1 IU ml−1, which results in inaccurate estimates of susceptibility in populations when ELISA is used for measuring antibody levels. Aim. To explore methods to accurately predict population immunity and TNT-derived anti-toxin titres from ELISA anti-toxoid results. Methodology. A total of 96 paired serum and dried blood spot (DBS) samples collected in Vietnam were used for comparison of TNT and ELISA. The diagnostic accuracy of ELISA measurement with reference to TNT was assessed by area under the receiver operating characteristic (ROC) curve (AUC) and other parameters. Optimal ELISA cut-off values corresponding to TNT cut-off values of 0.01 and 0.1 IU ml−1 were identified by ROC analysis. A method based on the multiple imputation approach was also applied to estimate TNT measurements in a dataset that only included ELISA results. These two approaches were then applied to ELISA results previously generated from 510 subjects in a serosurvey in Vietnam. Results. The ELISA results on DBS samples showed a good diagnostic performance compared to TNT. The cut-off values for ELISA measurement corresponding to the TNT cut-off values of 0.01 IU ml−1 were 0.060 IU ml−1 in serum samples, and 0.044 IU ml−1 in DBS samples. When a cut-off value of 0.06 IU ml−1 was applied to the 510 subject serosurvey data, 54 % of the population were considered susceptible (<0.01 IU ml−1). The multiple imputation-based approach estimated that 35 % of the population were susceptible. These proportions were much larger than the susceptible proportion estimated by the original ELISA measurements. Conclusion. Testing a subset of sera by TNT combined with ROC analysis or a multiple imputation approach helps to adjust ELISA thresholds or values to assess population susceptibility more accurately. DBS is an effective low-cost alternative to serum for future serological studies for diphtheria.