April 17, 2026

mNGS Validation for Sterile Body Fluids: Methods and Results

Diagnosing infections in sterile body fluids — cerebrospinal fluid (CSF), synovial fluid, vitreous humor — is one of the most consequential challenges in clinical microbiology. Standard culture and PCR miss a substantial proportion of cases: up to half of meningitis patients receive no confirmed microbiological diagnosis. Clinical validation of metagenomic next-generation sequencing (mNGS) for sterile body fluids is now producing data that quantifies this diagnostic gap and demonstrates a path to close it. In this symposium presentation, Dr. Amos Adler shared validation results from a 1,500-bed medical center in Tel Aviv, reporting 92.1% overall sensitivity and 100% specificity across 40 positive samples and 11 true-negative controls.

Key Takeaways

mNGS achieved 92.1% overall sensitivity and 100% specificity for pathogen detection in sterile body fluids, validated against culture and PCR reference standards.
The three false-negative cases were all Staphylococcus aureus from samples with suspected DNA degradation due to intense inflammation — not a platform limitation.
The RPM ratio (RPM in sample divided by RPM in no-template control) with a threshold of 10 provided an effective method for distinguishing true pathogens from background contamination.
Ophthalmologic infections — endophthalmitis and uveitis — represent a high-value clinical niche for mNGS due to extremely small sample volumes, broad differential diagnosis, and the inability to collect repeat specimens.
Operational challenges include batching economics (cost per flow cell), cloud-based bioinformatics concerns around patient data privacy, and the need for multidisciplinary diagnostic boards to interpret results.

Why Sterile Body Fluids Need Unbiased Diagnostics

The diagnostic landscape for sterile site infections presents a fundamental tension between clinical urgency and methodological limitations. Culture, while theoretically broad in identification range, suffers from sensitivity problems — particularly after prior antimicrobial treatment and with fastidious organisms. Specific PCR offers better sensitivity and analytic specificity, but only detects what it targets. Broad-range PCR (16S/18S) is less sensitive than targeted PCR, more prone to contamination, and produces ambiguous results at non-trivial cost.

Dr. Adler positioned mNGS as the method that addresses all three limitations simultaneously: high sensitivity, broad detection range, and the ability to identify organisms that clinicians are not even suspecting. The tradeoff, as he noted candidly, is that hypothesis-free testing generates results that clinicians may not expect and may struggle to interpret — and reagent costs remain high.

Clinical Validation Methodology and Specimen Types

Positive and Negative Sample Sets

The validation study used almost exclusively sterile body fluids, including CSF (the largest sample type), synovial fluid, vitreous fluid, and a small number of brain abscess samples. The 40 positive samples were defined by prior positive culture or PCR results, encompassing mainly bacteria with some fungal and viral pathogens.

Critically, the 11 negative control samples were not simply culture-negative specimens — they were collected from patients with no suspicion of infectious disease. These included patients undergoing lumbar puncture for normal pressure hydrocephalus evaluation and patients receiving intrathecal chemotherapy for lymphocytic leukemia. This rigorous negative control design strengthens the specificity calculation considerably.

The minimum sample volume was 200 microliters — an important practical constraint for ophthalmologic and CSF specimens where available volume is extremely limited.

Bioinformatic Analysis and the RPM Ratio

The pipeline used Kraken with a comprehensive taxonomic database for classification. A key challenge in metagenomic analysis is distinguishing true pathogens from the inevitable background microbial DNA present even in sterile fluids. Dr. Adler’s team adopted the RPM ratio (RPMR) — the reads per million for a given taxon in the clinical sample divided by the RPM for that same taxon in the no-template control (NTC).

Using an RPMR threshold of 10, the team demonstrated effective separation of true pathogens from background noise. For example, Cutibacterium acnes (a common skin commensal and frequent contaminant) showed an RPMR above 30 in a confirmed shunt infection case but well below threshold in negative samples. While the threshold is acknowledged as somewhat arbitrary, it provides a systematic and reproducible framework for clinical reporting.

Dr. Adler noted that RPMR alone is insufficient — it does not account for genome coverage breadth. A high read count mapping to a narrow genomic region may represent misclassification rather than true detection. Combined metrics incorporating both abundance and coverage (such as the EDCC score) may provide more robust classification in future implementations.

Sensitivity, Specificity, and Ophthalmologic Applications

At a minimum detection threshold of 100 genome equivalents per sample, overall sensitivity was 92.1% with 100% specificity. The three false-negative cases were all Staphylococcus aureus from joint infections. The team attributed these failures to DNA degradation from intense inflammatory environments — by the time validation testing occurred, the DNA had likely deteriorated beyond recovery, while culture (which requires viable organisms at the time of plating) had succeeded earlier.

Endophthalmitis and Uveitis — A Clinical Niche for mNGS

Dr. Adler identified ophthalmologic infections as a particularly compelling use case for mNGS. Endophthalmitis is a sight-threatening emergency where samples are collected surgically (vitreous tap or anterior chamber aspirate), volumes are extremely small (often under 200 microliters), and the specimen collection is typically a one-time opportunity — there is no second sample.

The differential diagnosis for intraocular inflammation spans infectious and non-infectious etiologies: bacterial, mycobacterial, fungal, viral (HSV, VZV, CMV), and parasitic (Toxoplasma) causes must all be considered alongside autoimmune uveitis. Traditional testing requires splitting a tiny sample across multiple assays, often exhausting it before all possibilities are covered.

mNGS addresses this by testing for all pathogen classes from a single specimen. In their clinical experience, the team identified CMV, varicella-zoster virus, Toxoplasma, and Bartonella in uveitis cases, and Listeria in endophthalmitis — diagnoses confirmed retrospectively by targeted methods. Some cases produced unexpected taxa (such as Nocardia) that required the clinical board’s judgment to determine significance.

Operational Challenges in Clinical Metagenomics

Dr. Adler outlined several operational realities for laboratories implementing clinical mNGS. Batching economics remain a significant barrier: the flow cell represents the largest single cost, so laboratories need sufficient sample volume to run efficiently. His current schedule of 3–5 samples every two weeks reflects this constraint.

Samples with high human DNA abundance (biopsies, tissue) consume disproportionate sequencing capacity — tens of millions of reads — making them impractical to multiplex with other sample types without effective host depletion using filters such as the Devin filter. Cloud-based bioinformatics is preferred for database currency but raises patient data privacy concerns — what Dr. Adler termed “cloud phobia” — since metagenomic data inherently contains the patient’s own genome alongside microbial sequences.

Looking ahead, Dr. Adler identified three priority development areas: processing high-human-DNA samples (biopsies), adding RNA virus detection (requiring dual libraries at nearly double cost), and — most importantly — combined microbial detection with antimicrobial resistance gene characterization, which he considers the area where mNGS can provide the greatest added value over syndromic PCR testing.

About the Speaker

Dr. Amos Adler is the director of clinical microbiology and a professor at a major medical center and university in Tel Aviv, Israel. His research focuses on antimicrobial resistance mechanisms, hospital-acquired infection transmission, and the clinical application of microbial genomics.