Evaluation of four molecular detection assays for the diagnosis of Chlamydia pecorum infection in Victorian koalas (Phascolarctos cinereus) using Bayesian latent class analysis

Chlamydiosis poses a significant threat to koala health and population viability. Accurate diagnosis of chlamydial infection in koalas at veterinary hospitals informs patient prognosis, biosecurity and determines treatment. However, there is no standardised approach to the diagnosis of chlamydiosis in wildlife hospitals. The performance of available diagnostic tests has not previously been rigorously assessed, and there is a lack of sufficient well-characterised reference samples. This study aimed to estimate the diagnostic accuracy of four molecular detection assays for Chlamydia pecorum infection in koalas using Bayesian latent class analysis (BLCA). Swabs were collected from koalas presented to a wildlife hospital (n = 45) and historical population surveys (n = 48). The four assays that were compared were a qPCR with melt curve that targets 16S rRNA (16SG), a probe-based qPCR assay that targets ompB, and two loop-mediated isothermal amplification (LAMP) assays, targeting MreC and a hypothetical protein (CpecG_0573). The 16SG qPCR and CpecG_0573 LAMP showed the highest overall diagnostic accuracy, with Youden's indices of 83.8 % (95 % posterior credible interval [95 % PCI]: 66.8 %, 93.4 %), and 81.5 % (95 % PCI: 62.5 %, 93.7 %) respectively. The underlying true prevalence of infection in hospitalised koalas (i.e., the latent class) was estimated to be 35.2 % (95 % PCI 21.4, 51.2 %). Combining the 16SG qPCR with the CpecG_0573 LAMP assay at this estimated infection prevalence yielded an excellent positive predictive value (PPV) > 96 %. Therefore, a testing protocol applying these tests in series is recommended for diagnosing chlamydiosis in this population of hospitalised koalas.