Analytical and clinical validation of a microbial cell-free DNA sequencing test for infectious disease.
Blauwkamp TA, Thair S, Rosen MJ, Blair L, Lindner MS, Vilfan ID, Kawli T, Christians FC, Venkatasubrahmanyam S, Wall GD, Cheung A, Rogers ZN, Meshulam-Simon G, Huijse L, Balakrishnan S, Quinn JV, Hollemon D, Hong DK, Vaughn ML, Kertesz M, Bercovici S, Wilber JC, Yang S
Nat Microbiol. Feb 2019. doi: 10.1038/s41564-018-0349-6
COMMENT: This work shows the potential use of next-generation sequencing of cell-free DNA (cfDNA) from pathogens for clinical infectious disease testing. The Karius test isolates cfDNA from blood plasma, cfDNA is sequenced on Illumina NextSeq500 using a 75-cycle single end, and a custom-built analysis pipeline uses a clinical-grade database to identify the microbial DNA fragments found in plasma.
Here we describe the analytical and clinical validation of a quantitative microbial cfDNA sequencing test that identifies 1,250 human pathogens based on the fragments of genomic DNA these pathogens leave in blood (the Karius Test). Performance characterization was based on Clinical and Laboratory Standards Institute guidelines and informed by the US Food and Drug Administration draft guidance for characterizing NGS-based infectious disease tests.
The authors attempt to address all challenges to supplying of high-quality diagnostic testing for clinical metagenomics applications through a combination of traditional and metagenomic-specific validation strategies. A reference panel of 13 microorganisms was designed to characterize test performance across several potential sources of bias.
In silico simulations have tested the integrity of the bioinformatics pipeline in the face of clinical-isolate divergence, 580 clinical samples that assessed performance in patient samples and 2,000 samples that were run through CLIA-certified laboratory and reported in real-time
Elevated levels of human cfDNA background in the sample had only minor effects on sensitivity and precision
The level of environmental contamination did influence test sensitivity, but only for the 5–8 microorganisms with the highest environmental backgrounds among 1,250 microorganisms probed.
The sequencing depth also influenced the limit of detection (LoD), but the processing methods for this test are designed to provide similar sequencing depth for all samples, such that 95% of the samples tested fell into a range of sequencing depths where sensitivity was consistent
The positive agreement between this test and blood culture (93.7%) of patients with a sepsis alert is equal to or better than other direct molecular diagnostic methods, including real-time PCR panels and PCR combined with electrospray ionization
The sensitivity and breadth of microorganisms detected, combined with the diversity of the microbiome compositions across patients, makes it challenging to achieve high diagnostic specificity.
Analytical validation experiments demonstrated very low levels of falsely reporting of microbial cfDNA that was not in the original plasma sample, consistent with high reproducibility of cfDNA detection across independent runs
Although it would be convenient if the cfDNA concentration alone were indicative of true infection, both the microbe identity and the location of infection are likely to influence the concentration of microbial cfDNA observed in plasma.
The cfDNA concentrations associated with definite calls spanned four orders of magnitude (20–450,000 MPM). Therefore, high concentrations of microbial cfDNA were typically associated with true infections, whereas low concentrations were associated with both true infections and commensal/colonizer/contaminant microorganisms of unknown clinical significance.
The performance characteristics reported here point towards clinical applications in which the benefits of sensitivity, non-invasive sampling and broad testing outweigh the limitations of sequencing cost, turn-around time and identification of microorganisms...
Regardless of the cfDNA concentrations, the entire clinical picture must be considered when determining the clinical significance of a microbe detected by cfDNA sequencing.
Use cases that could potentially benefit from this approach include rule-out testing for sepsis when the standard of care was not sensitive enough or too invasive, testing immunocompromised patients presenting with non-specific symptoms of infection or testing before invasive procedures with high cost or morbidity.