Abstract / Summary
Objectives. To establish how targeted (tNGS) and metagenomic (mNGS) next-generation sequencing differ in analytical and diagnostic performance in lower respiratory tract infection (LRTI), how much of any difference is a property of the individual panel rather than of the platform class, and what now limits clinical value. Methods. We searched PubMed on 15 August 2026 for studies of targeted or metagenomic sequencing in respiratory infection, combining sequencing terms ("targeted next-generation sequencing", "metagenomic next-generation sequencing", tNGS, mNGS, nanopore targeted sequencing) with terms for the lung and lower respiratory tract and with population terms (immunocompromised, transplant, haematological malignancy, paediatric, intensive care). The search was repeated across seven clinical dimensions — technical comparison, immunocompromised hosts, paediatric populations, critical care, resistance determinants, clinical impact, and diagnostic accuracy. This review analyses 121 articles available in full text, together with 38 further records available only as PubMed abstracts and one record for which no abstract could be retrieved. Every quantitative statement was traced to a specified line of the source document. Results. Within the spectrum a panel covers, tNGS is diagnostically comparable to mNGS: a network meta-analysis of studies in suspected pneumonia found no significant difference in overall sensitivity (OR 1.54, 95% CI 0.47–5.06) or specificity (OR 1.29, 95% CI 0.39–4.28). The losses are complementary and design-determined rather than ranked. In one prospective cohort of 251 patients in which causality was adjudicated, 18 causal pathogens were detected by mNGS alone, while 23 were detected by targeted methods but missed by mNGS, of which nine were fungi — seven of them Pneumocystis jirovecii — and seven were RNA viruses. Conclusions. Detection is no longer rate-limiting; interpretation is. The adjudicated positive predictive value of a tNGS-positive finding was 80.3% overall and 18.2% for fungal targets, and within a single multicentre cohort of 631 patients, moving from qualitative reporting to a model incorporating normalised read counts and pathogen copy number raised sensitivity from 78.5% to 82.4% and specificity from 76.6% to 85.0% without any change to the assay. Standardised panel reporting, validated interpretive thresholds, and trials that test sequencing as a triage strategy rather than as an addition to conventional testing are the priorities.