Key points are not available for this paper at this time.
complex (MAC), are increasingly recognized as an important clinical entity, yet distinguishing active pulmonary disease from asymptomatic colonization remains challenging because current diagnosis relies on composite criteria. In this study, we aimed to identify blood transcriptomic signatures that discriminate MAC pulmonary disease (MAC-PD) from pulmonary colonization (MAC-PC) and to evaluate their potential as candidate biomarkers. MAC-positive patients from two medical centers in Taiwan were enrolled as training and external validation cohorts, and peripheral blood transcriptomes were profiled. Candidate genes were identified using least absolute shrinkage and selection operator regression and recursive feature elimination across multiple random seeds, followed by filtering based on statistical robustness and biological relevance. Machine-learning models were then trained and externally validated. Among 120 patients (training cohort: 46 MAC-PD and 28 MAC-PC; validation cohort: 25 MAC-PD and 21 MAC-PC), seven enriched gene ontology terms were prioritized. Three models demonstrated robust performance in the validation cohort, with areas under the receiver operating characteristic curve of 0.78, 0.78, and 0.75, and accuracies of 0.76, 0.74, and 0.74, respectively. These models shared a five-gene core signature consisting of IGKV1D-39, IGKV6-21, OVCH1, PLAU, and DMD, highlighting convergent biological signals related to immune responses and tissue remodeling. Overall, blood transcriptomic profiling shows promise in differentiating MAC-PD from MAC-PC in our cohorts, and the identified five-gene core signature represents a biologically coherent, minimally invasive candidate biomarker panel warranting further prospective validation.
Lin et al. (Sat,) studied this question.