Reply : comprehensive validation of the FAIM3:PLAC8 ratio in time-matched public gene expression data

Abstract

We are thoroughly grateful to Dr. Sweeney and Dr. Khatri for their positive comments regarding our study on the FAIM3:PLAC8 gene expression biomarker for the rapid diagnosis of community-acquired pneumonia (CAP) at intensive care unit (ICU) admission (1). Moreover, the additional analyses across different cohorts certainly add value to our proposed biomarker. We definitely concur with the authors in making use of publicly available datasets for further validation of proposed biomarkers. However, we would also like to point out that contrary to our study, which was centered on a specific population of ICU patients with suspected CAP, the publicly available data sets differ with regard to both cases (sepsis rather than CAP) and systemic inflammatory response syndrome (SIRS) controls (rather than no-CAP) (2–4). Only one study specifically investigated CAP cases, albeit with general SIRS patients as control subjects (5). Notwithstanding the potential contribution of almost inevitable nonexperimental hybridization chip effects, which we assessed and corrected for in our own study, differences in patient selection may indeed contribute to a large extent to the poor areas under receiver operating characteristic curves observed in some cohorts. Furthermore, as Sweeney and Khatri highlighted, the derivation and implementation of simple models such as our two-gene ratio will certainly aid in future validation testing not only across other cohorts but also technologies other than hybridization microarrays. [Extract from the article]