Accuracy of polymerase chain reaction assays for diagnosis of feline immunodeficiency virus infection in cats.

To determine the sensitivity, specificity, and overall diagnostic accuracy of polymerase chain reaction (PCR) assays offered by commercial diagnostic laboratories for diagnosis of FIV infection in cats. Prospective clinical trial. 124 cats. Blood was collected from cats that were neither infected with nor vaccinated against FIV, uninfected cats that were vaccinated with a licensed FIV vaccine, and cats experimentally and naturally infected with FIV representing subtypes A, B, and C. Coded blood samples were submitted to 3 laboratories in the United States and Canada offering PCR assays for diagnosis of FIV infection to veterinary practitioners. All laboratories tested fresh blood samples, and 1 laboratory also tested samples submitted as dried blood smears. The FIV infection status in all cats was confirmed by virus isolation. Sensitivity, specificity, and correct results were calculated for each PCR assay. Sensitivity ranged from 41% to 93%. Specificity ranged from 81% to 100% in unvaccinated cats and 44% to 95% in cats vaccinated against FIV. Correct results were obtained in 58% to 90% of 124 cats tested. All tests misidentified both uninfected and infected cats. False-positive results by all laboratories were higher in cats vaccinated against FIV than in unvaccinated cats, suggesting that vaccination interferes with the performance or interpretation of PCR assays used for diagnosis of FIV infection. PCR assays used for diagnosis of FIV infection presently marketed to veterinary practitioners in North America vary significantly in diagnostic accuracy and did not resolve the diagnostic dilemma resulting from vaccination of cats against FIV.