Determining Colonisation from Infection Using PCR-Based Diagnostics in HIV-Negative Individuals.

Background: pneumonia is increasingly diagnosed with highly sensitive PCR diagnostics in immunocompromised, HIV-negative individuals. We assessed the performance of our in-house quantitative PCR with the aim to optimise interpretation.

Methods: Retrospective audit of all positive qPCRs on induced sputum or BAL fluid at a single centre from 2012 to 2023. Medical and laboratory records were analysed and people with HIV were excluded. Cases were categorised as colonisation, high-probability PCP or uncertain PCP infection against a clinical gold standard incorporating clinico-radiological data. Quantitative PCR assay targeting the 5s gene was utilised throughout the time period.

Results: Of the 82 positive qPCRs, 28 were categorised as high-probability PCP infection, 30 as uncertain PCP and 24 as colonisation. There was a significant difference in qPCR values stratified by clinical category but not respiratory sample type. Current assay performance with a cutoff of 2.5 × 10 copies/mL had a sensitivity of 50% (95% CI, 30.65-69.35%) and specificity of 83.33% (95% CI, 62.62-95.26%). Youden Index calculated at 6.5 × 10 copies/mL had a sensitivity of 75% (56.64-87.32%, 95% CI) and specificity of 66.67% (46.71-82.03%, 95% CI). High and low cutoffs were explored. Significant variables associated with infection were age > 70 years old, the presence of fever, hypoxia or ground glass changes.

Conclusions: A single qPCR cutoff cannot reliably determine infection from colonisation. Low and high cutoffs are useful, however, a large "possible infection" cohort will remain where interpretation of clinic-radiological factors remains essential. Standardisation of assays with prospective validation in specific immunocompromised groups will allow greater generalisability and allow large-scale prospective assay validation to be performed.