Metagenomic next-generation sequencing, or mNGS, had a higher reported diagnostic yield than culture-based tests in a structured review of infective-endocarditis studies. Here, diagnostic yield means the reported rate at which a test produced a microbial diagnosis. The pooled estimate for mNGS microbial detection was 0.87, compared with 0.43 for blood culture.
The paper focused on mNGS in blood culture-negative endocarditis, or BCNE, and searched PubMed and Google Scholar for infective-endocarditis literature that included BCNE cases and sequencing methods. It compared reported yields for mNGS, blood culture and valve-tissue culture, including a separate pooled estimate for mNGS on valve tissue.
For mNGS applied to valve tissue, the reported pooled diagnostic yield was 0.92, with an interval of 0.80 to 0.97. The reported valve-tissue culture yield was 0.23, with an interval of 0.12 to 0.39. In other words, the review's reported estimate for mNGS on valve tissue was higher than its estimate for valve-tissue culture.
The blood-culture estimate was 0.43, with a reported interval of 0.28 to 0.58. Relative to mNGS, the review reported blood culture as having a two-fold lower diagnostic yield and valve-tissue culture as having a four-fold lower yield. These are comparisons of reported pooled proportions, not measurements of patient benefit.
The comparison has clear limits
The review's numbers are about diagnostic yield, not about what happened to patients after testing. It does not report patient-outcome comparisons. The authors interpret mNGS as increasingly implemented in routine infective-endocarditis diagnostics and as having consistently high diagnostic yield across heterogeneous studies, especially in valve tissue compared with blood and valve-tissue culture.
That interpretation needs to be read alongside the review's stated caution. The included studies were heterogeneous, and the pooled estimates were explicitly described as illustrative. The figures therefore summarize the reported literature rather than a single direct trial result.
The search identified 12 clinical studies involving 794 patients. Ten were prospective and two were retrospective. No randomised controlled trials were identified, so the review offers no randomised comparative evidence. Those study-design details matter because the review is comparing reported yields across clinical studies, not reporting a randomised head-to-head trial.
To generate the combined figures, the reviewers used random-effects meta-analyses of proportions, a way of combining reported proportions across studies, and displayed the pooled estimates in forest plots. For the relative comparison, they used ratios of pooled proportions. The review describes the resulting pooled estimates as illustrative.
What the authors say is still needed
For BCNE, the authors identify consensus on diagnostic algorithms, standardised mNGS testing and randomised controlled trials as needed to further define the role of mNGS. The review's conclusion supports the reported pooled estimates from the included studies while leaving those broader questions open.
The work was supported by the Murermester Lauritz Peter Christensen og hustru Kirsten Sigrid Christensens Fond. The authors declared no competing interests.
Paper data and sources
Original title: Metagenomic next-generation sequencing in blood culture-negative endocarditis: a structured review with illustrative pooled estimates.
Authors: Pontus Westerström
Journal/Repository: Infection
Status: Peer-reviewed
First online: 2026-08-21
DOI: 10.1007/s15010-026-02901-z
Original paper · Full text