RESUMO
BACKGROUND: American visceral leishmaniasis is caused by the protozoan Leishmania infantum. Dogs are the main reservoirs in the domestic transmission cycle. The limited accuracy of diagnostic tests for canine leishmaniasis may contribute to the lack of impact of control measures recommended by the Brazilian Ministry of Health. The objective of this study was to estimate the accuracy of two enzyme-linked immunosorbent assays employing L. major or L. infantum antigens and their reliability between three laboratories of different levels of complexity. METHODS: A validation study of ELISA techniques using L. major or L. infantum antigens was conducted. Direct visualization of the parasite in hematoxylin/eosin-stained histopathological sections, immunohistochemistry, and isolation of the parasite in culture.were used as gold standard. An animal that was positive in at least one of the tests was defined as infected with L. infantum. Serum samples collected from 1,425 dogs were analyzed. Samples were separated in three aliquots and tested in three different laboratories. Sensitivity, specificity and the area under de ROC curve were calculated and the reliability was evaluated between the participant laboratories. RESULTS: The sensitivity was 91.8% and 89.8% for the L. major and L. infantum assays, respectively. The specificity was 83.75% and 82.7% for the L. major and L. infantum assays, respectively. The area under de ROC curve was 0.920 and 0.898 for L. major and L. infantum, respectively. The mean intraclass correlation coefficients between laboratories ranged from 0.890 to 0.948 when L. major was used as antigen, and from 0.818 to 0.879 when L. infantum was used. INTERPRETATION: ELISA tests using L. major or L. infantum antigens have similar accuracy and reliability. Our results do not support the substitution of the L. major antigen of the ELISA test currently used for the diagnosis of canine visceral leishmaniasis in Brazil.