Detection of Salmonella Typhi nucleic acid by RT-PCR and anti-HlyE, -CdtB, -PilL, and -Vi IgM by ELISA at sites in Ghana, Madagascar and Ethiopia

We performed species-specific real-time polymerase reaction (RT-PCR) to identify bacterial nucleic acid, and enzyme-linked immunosorbent assay (ELISA) for detecting HlyE/STY1498-, CdtB/STY1886-, pilL/STY4539- and Vi-antigens in blood and biopsy specimens of febrile and non-febrile subjects. We generated antigen-specific ELISA proxy cut-offs by change-point analyses, and utilized cumulative sum as detection method coupled with 1000 repetitive bootstrap analyses. We computed prevalence rates in addition to odds ratios to assess correlations between ELISA outcomes and participant characteristics.

Definitive positive RT-PCR results were obtained from samples of febrile subjects originating from Adama Zuria/Ethiopia (1.9%, 2/104), Wolayita Sodo/Ethiopia (1.0%, 1/100), Diego/Madagascar (1.0%, 1/100), and Kintampo/Ghana (1.0%, 1/100), and from samples of non-febrile subjects from Wolayita Sodo/Ethiopia (1%, 2/201). While IgM antibodies against all antigens were identified across all sites, prevalence rates were highest at all Ethiopian sites, albeit in non-febrile populations. Significant correlations in febrile subjects aged < 15 years versus ≥ 15 years were detected for Vi (Odds Ratio (OR): 8.00, p = 0.034) in Adama Zuria/Ethiopia, STY1498 (OR: 3.21, p = 0.008), STY1886 (OR: 2.31, p = 0.054) and STY4539 (OR: 2.82, p = 0.022) in Diego/Madagascar, and STY1498 (OR: 2.45, p = 0.034) in Kintampo/Ghana. We found statistical significance in non-febrile male versus female subjects for STY1498 (OR: 1.96, p = 0.020) in Adama Zuria/Ethiopia, Vi (OR: 2.84, p = 0.048) in Diego/Madagascar, and STY4539 (OR: 0.46, p = 0.009) in Kintampo/Ghana.

Findings indicate non-discriminatory stages of acute infections, though with site-specific differences. Immune responses among non-febrile, presumably healthy participants may mask recall and/or reporting bias leading to misclassification, or asymptomatic, subclinical infection signs induced by suppression of inflammatory responses. As most Ethiopian participants were ≥ 15 years of age and not at high-risk, the true S. Typhi burden was likely missed. Change-point analyses for generating ELISA proxy cut-offs appeared robust, though misclassification is possible. Our findings provided important information that may be useful to assess sites prior to implementing surveillance for febrile illness including Salmonella disease.