Validation of real-time polymerase chain reaction tests for diagnosing feline immunodeficiency virus infection in domestic cats using Bayesian latent class models

John M. Morton, Richard J. McCoy, Rebecca K.C. Kann, Ian A. Gardner, Joanne Meers

The objectives of the current study were to estimate the sensitivity and specificity of three real-time polymerase chain reaction (PCR) tests for diagnosis of feline immunodeficiency virus (FIV) infection in domestic cats, both individually and when interpreted in series with one of two serological tests, separately in populations of cats at low and high risk of being infected with FIV. One PCR test targeted the pol gene and two targeted the gag gene of FIV. For comparison, sensitivities and specificities of the individual serological tests (IDEXX SNAP(®) test and AGEN Simplify(®) test) were also estimated. The study populations consisted of domestic cats thought to be not vaccinated against FIV. Low-risk (males aged 4 years or less and females; n=128) and high-risk (males over 4 years; n=128) cats were selected from those where blood samples were submitted to a commercial clinical pathology service. Bayesian latent class models were used to obtain posterior probability distributions for sensitivity and specificity for each test, based on prior distributions obtained from three experts. Medians of the posterior sensitivity distributions for the PCR tests based on the pol gene and two regions of the gag gene tests ranged from 0.85 to 0.89, compared to 0.89-0.97 for the two serological tests. The medians of posterior specificity distributions for these PCR tests were 0.94-0.96, and 0.95-0.97 for the serological tests. In contrast, the PCR based on one region of the gag gene had lower median sensitivity. Sensitivities of combinations of these serological and PCR tests interpreted in series were low; medians of posterior sensitivity distributions ranged from 0.75 to 0.83. Relative to the low-risk population, median sensitivities in the high-risk population were lower for all tests other than the AGEN Simplify(®) test; specificities were similar in both populations. We conclude that the sensitivities of the two PCR tests based on the pol gene and two regions of the gag gene, respectively, in non-vaccinated cats are probably lower than the sensitivities of the two serological tests we assessed. We do not recommend screening cats whose FIV vaccination status is uncertain with one of these serological tests and then testing positives with one of these PCR tests because in non-vaccinates, the sensitivities of combinations of these serological and PCR tests interpreted in series are low. Assessment of the validity of these PCR assays in FIV-vaccinated cats is required.