March 9, 2024

Host Depletion in Clinical mNGS: A Feasibility Study

Host depletion is widely recognized as a critical step in metagenomic next-generation sequencing (mNGS) workflows, yet clinical feasibility data demonstrating its real-world impact remain limited. PathAI Diagnostics, a US-based clinical laboratory specializing in molecular infectious disease testing, conducted a systematic mNGS feasibility study evaluating host depletion filtration for urinary tract infection (UTI) pathogen detection. The study, presented as a poster at the 2023 Association for Molecular Pathology (AMP) conference in Salt Lake City, tested three objectives: spiked sample recovery, concordance with routine clinical methods, and freeze-thaw stability.

Key Takeaways

Host depletion filtration significantly reduced human reads and enriched microbial signal in both spiked and clinical urine samples
mNGS with host depletion achieved 100% concordance with culture and 91% concordance with a 17-pathogen PCR panel
mNGS detected additional clinically relevant organisms missed by both culture and PCR, including Klebsiella pneumoniae and Aspergillus species
Freeze-thaw cycles did not adversely affect filtration performance or pathogen detection sensitivity
The end-to-end workflow, from collection through analysis, completed within 24 hours

The Diagnostic Gap in Urinary Tract Infections

Urine culture, the traditional gold standard for UTI diagnosis, detects as few as 12% of clinically relevant species associated with urinary tract infection, according to peer-reviewed literature. Culture results take 24 to 48 hours for bacteria and up to 72 hours for fungal pathogens. Even molecular methods have coverage limitations: PathAI’s existing 17-pathogen multiplex PCR panel leaves 30-40% of symptomatic UTI cases without a definitive pathogen identification.

This diagnostic gap is not trivial. Unidentified pathogens in symptomatic patients lead to empiric antibiotic therapy, which drives antimicrobial resistance and increases treatment failures. mNGS offers an unbiased approach that can theoretically detect any organism present, but the overwhelming proportion of human DNA in urine samples has historically limited its clinical utility. Host cells are not the target of the assay, yet they consume the vast majority of sequencing capacity.

Study Design: Evaluating Host Depletion Across Three Objectives

The PathAI study was designed as a three-part evaluation, each addressing a distinct question about clinical feasibility.

Part A: Spiked Samples with Known Organisms

Negative urine samples (no pathogen detected) were spiked with four known organisms at concentrations ranging from 10^4 to 10^0 colony-forming units (CFU) per milliliter. Each concentration was processed in triplicate, with and without the host depletion filter. Sequencing data showed that filtered samples had dramatically reduced human reads compared to unfiltered controls. The reduction was consistent across all spike concentrations, confirming effective host cell removal even at very low pathogen loads. Microbial enrichment was correspondingly increased in filtered samples.

Part B: Routine Clinical UTI Specimens

Fresh, symptomatic urine specimens received for routine clinical testing were processed through three parallel methods: the existing 17-pathogen PCR panel, conventional culture and susceptibility testing, and mNGS with host depletion filtration. This design enabled direct comparison of mNGS against both existing clinical methods on the same patient specimens.

Results showed strong concordance across methods. Comparing culture to PCR yielded 14 out of 15 matches. mNGS achieved 15 out of 15 concordance with culture (100%) and 31 out of 34 concordance with PCR (91%). Importantly, where mNGS and PCR disagreed, the discrepancies occurred at high cycle threshold (CT) values (above 30), indicating very low pathogen loads near the limit of detection.

In one example, specimen USB-01 showed E. coli detection by both PCR (CT value 15.94) and mNGS, while culture reported only “>100,000 CFU/mL” without species-level identification. Both molecular methods provided species-specific results that culture could not.

Part C: Frozen Specimen Performance

Clinical specimens were frozen for at least 24 hours, thawed, and processed with and without the host depletion filter. Comparing fresh versus frozen filtered samples showed no significant difference in human read reduction or microbial read enrichment. Pathogen detection was not adversely affected by freeze-thaw, a critical finding for laboratories that need to batch-process or store specimens before mNGS analysis.

How Host Depletion Filtration Improved Pathogen Detection Sensitivity

Beyond concordance with existing methods, mNGS with host depletion detected additional organisms that both culture and PCR missed. These included clinically significant pathogens such as Klebsiella pneumoniae and Aspergillus species, as well as multiple Lactobacillus species.

For PathAI, this finding directly addresses their operational challenge: the 30-40% of symptomatic UTI cases where the 17-pathogen panel returns negative results. The study data suggest that mNGS with host depletion could cut that undiagnosed fraction roughly in half, recovering an additional 10-20% of cases with clinically actionable pathogen identifications. The laboratory’s planned implementation strategy is to reflex PCR-negative symptomatic specimens to mNGS rather than replacing PCR outright.

The clinical validation approach involves spiking known strains at defined concentrations and demonstrating detection by mNGS at levels below the sensitivity threshold of both PCR and culture. This establishes a limit of detection for the mNGS workflow that can be compared against existing methods.

Clinical Implications for mNGS Feasibility in UTI Diagnostics

The PathAI feasibility data support several conclusions relevant to clinical laboratories evaluating mNGS adoption:

Host depletion is not optional. Without filtration, human reads dominate the sequencing output to the point where pathogen detection sensitivity is insufficient for clinical use. The filter reduced host reads consistently across fresh, frozen, spiked, and real clinical specimens.

End-to-end workflow matters. The study used a complete, integrated workflow, from host depletion through extraction, library preparation, sequencing, and bioinformatic analysis. Concordance results reflect the performance of the entire system, not individual components in isolation.

mNGS complements rather than replaces existing tests. The highest clinical value may come from using mNGS as a reflex test for cases where targeted methods fail, rather than as a first-line replacement. This approach manages cost while capturing the diagnostic gap.

Frozen specimens are viable. Laboratories do not need to process samples immediately upon receipt. Freeze-thaw stability enables batch processing and broader specimen access, including samples collected at satellite clinics and shipped to a central laboratory.

End-to-End mNGS Workflow from Sample Collection to Pathogen Report

The study used Micronbrane Medical’s complete workflow: the Devin Host Depletion Filter for host cell removal, the Devin Microbial DNA Enrichment Kit for extraction and purification, and mNGS-grade library preparation reagents. All upstream components are currently compatible with Illumina sequencing platforms, with support for additional platforms planned. The total workflow from sample collection to pathogen identification completes within 24 hours.

Micronbrane Medical is extending this same workflow to additional sample types beyond urine, including whole blood for sepsis, joint fluid for prosthetic infections, and cerebrospinal fluid (CSF) for central nervous system infections.

About the Speaker

Anami Patel, PhD — Chief Science Officer and Lead Investigator, PathAI Diagnostics Mengchu Wu, PhD — Co-founder, CEO, and Chairwoman, Micronbrane Medical