Diagnosis value of targeted and metagenomic sequencing in respiratory tract infection

Targeted next-generation sequencing (tNGS) has become a trending tool in the field of infection diagnosis, but concerns are also raising about its performance compared with metagenomic next-generation sequencing (mNGS). This study aims to explore the clinical feasibility of a tNGS panel for respiratory tract infection diagnosis and compare it with mNGS in the same cohort of inpatients.

180 bronchoalveolar lavage fluid samples were collected and sent to two centers for mNGS and tNGS blinded tests, respectively. The concordance between pathogen reports of both methods and the clinical significance among samples with/without known etiology was further evaluated.

Overall, both methods displayed high agreement on pathogen reports, as the average percent agreement reached 95.29%. But tNGS presented a slightly higher detection rate per species than mNGS (P_Wilcoxon=1.212e-05; standard mean difference = 0.2887091), as detection rates for 32 out of 48 species were higher than those of mNGS. Due to limitations of panel coverage, tNGS identified 28 fewer species than mNGS, among which only 3 were considered clinically relevant. In reference to composite reference standard, accuracy, sensitivity, and specificity combining both tNGS and mNGS reached 95.61%, 96.71%, and 95.68%, respectively, while positive prediction value (PPV) was low at 48.13%, which was caused by low agreement regarding opportunistic pathogens. tNGS and mNGS improved the etiology identification in 30.6% (55/180) and 33.9% (61/180) cases, respectively.

Collectively, tNGS presented a similar overall performance in pathogen identification compared to mNGS, but outperformed in some pathogens. This study also demonstrated that deployment of tNGS significantly improves etiology identification in routine practice and provides hints for clinical decisions. The low agreement between clinical diagnosis and NGS reports towards opportunistic pathogens implies that adjudication is essential for report interpretation. Finally, We proposed tNGS as a diagnosis option in clinical practice due to its cost-efficiency.