Evaluation of diagnostic cut-off values of immunoglobulinG antibodies against pertussis toxin

Abstract number: P1755

de Greeff S.C., Notermans D.W., Elvers B., de Melker H.E., Teunis P., Schellekens J.F.

Objectives: Measuring IgG antibodies against pertussis toxin (IgG-PT) with ELISA can be used to diagnose pertussis infection; however, cut-off points have not unanimously been determined. We applied binary mixture models to interprete IgG-PT antibodies in diagnostics samples obtained from patients with suspected pertussis using single serum samples and the increase in paired sera, to determine optimal cut-offs.

Methods: IgG-PT levels were used from clinical diagnostic samples between 1st October, 2003 and 31st December, 2007 and assayed with an in-house IgG-PT ELISA that was calibrated with the international FDA lot 3 IgG-PT reference serum. The validity of cut-off levels in a single serum sample was assessed by selecting patients sampled within 100 days after onset of symptoms (n = 11655). Increase in antibody titre was studied in patients with two serum samples where the second sample was taken within 10–28 days after the first sample (n = 2205).

Children eligible for the acellular booster vaccination at four years of age or vaccination with an acellular in infancy were excluded, to avoid potential interference of a vaccine induced IgG-PT rise.

Binary distribution mixtures were fitted to the age stratified data, using the two fitted components to calculate ROC curves and determine optimum cut-off levels for positives in single samples and increase (seroconversion) in paired samples.

Results: Given the distributions of antibody levels, an IgG-PT titre of 79 U/ml yielded the highest sensitivity (96.5%) and specificity (94.4%) as diagnostic cut-off level. Mixtures fitted to separate age categories produced similar results: for the age-groups 1–9, 10–19, 20–39, 40–59 and >60 years a cut-off for IgG-PT titre of 66 U/ml, 92 U/ml, 81 U/ml, 78 U/ml and 107 U/ml respectively yielded the highest sensitivity (97.2, 95.3, 97.2, 96.8, 97.3%) and specificity (95.8, 91.5, 96.1, 96.0, 95.9%). In addition, a 3.4 fold increase in IgG-PT titre in paired serum samples yielded a high diagnostic sensitivity (99.4%) and specificity (99.7%) for all ages.

Conclusion: We show that diagnostic samples for pertussis can be classified into two subpopulations: a highly reactive population (persons with infection) and a baseline population (persons without infection); from these bimodal distributions, cut-offs for absolute values and dynamics of IgG-PT can be derived that are associated with very high diagnostic sensitivity and specificity.