Using Bayesian methods to evaluate the diagnostic performance of blood-based diagnostic tests for American cutaneous leishmaniasis and Chagas disease in rural Panamanian dogs in the absence of a gold standard.

Abstract

Dogs are important sentinels and potential reservoirs of Leishmania (Viannia) panamensis and Trypanosoma cruzi, etiologic agents of American cutaneous leishmaniasis and Chagas disease, respectively. Accurate serological tools are essential for surveillance in endemic regions where both diseases co-occur, yet no single gold standard diagnostic test exists. We evaluated the performance of blood-based assays used to detect exposure to L. panamensis and T. cruzi in 311 owned dogs from 185 households across 12 rural communities in Coclé Province, Panama. We evaluated exposure to L. panamensis using enzyme-linked immunosorbent assay (ELISA) and indirect immunofluorescence antibody test (IFAT). We assessed T. cruzi exposure with IFAT, western blot (WB), multiplex microsphere immunoassay (MIA), and the Chagas Stat-Pak rapid immunochromatographic test. Agreement among assays was measured with Cohen's kappa coefficient, and diagnostic accuracy was estimated using Bayesian latent class models to account for conditional dependence in the absence of a gold standard. ELISA showed the highest accuracy for L. panamensis (sensitivity = 0.85; specificity = 0.87). The combination of MIA and the Chagas Stat-Pak test achieved the greatest diagnostic performance for T. cruzi (both sensitivity and specificity ≈ 0.90). However, because MIA is costly and not currently available in-country, the combination of WB and Chagas Stat-Pak rapid immunochromatographic test provides the most practical approach for T. cruzi diagnosis in this setting, with Chagas Stat-Pak rapid immunochromatographic test sensitivities ranging from 0.84 to 0.91 and specificities ranging from 0.74 to 0.75; and WB sensitivities ranging from 0.90 to 0.91 and specificity consistent at 0.97. Slight agreement between L. panamensis IFAT and T. cruzi assays suggests cross-reactivity between assays. These findings highlight the importance of region-specific assay validation and cost-effective diagnostic strategies to improve surveillance of canine vector-borne infections in Central America.