Evaluation Of A Metagenomic Next-Generation Sequencing Assay With A Novel Host Depletion Method For Pathogen Identification In Septic Patients

BackgroundThe traditional diagnosis of sepsis has always been based on microbial blood culture (BC). However, BC suffers from (1) long culture cycle, leading to delay in results, and (2) low diagnostic yields. Metagenomic next-generation sequencing (mNGS) has been proposed as an efficient and agnostic option that potentially overcomes these issues. In this study, a mNGS workflow utilizing a novel filter to specifically capture white blood cells and deplete host DNA background, was evaluated against BC results, as well as mNGS without host depletion, for pathogen identification. Materials and MethodsPatients admitted to Taipei Veterans General Hospital (TVGH) with suspected sepsis were recruited to the study approved by the IRB. Blood sample was taken for BC (designated as BC1) before any antibiotic exposure. Upon patient enrolment, blood was taken again and divided in 3 portions with one used for the 2nd BC (BC2). The other two were used for mNGS with one processed with the filter and the other without filtering, to assess the effectiveness of host-depletion by the filter. ResultsA total of 50 patients were recruited among which 45 had results for all 4 tests. mNGS with filter had the highest positive rate of 74.4%, followed by mNGS without filter and BC1 (51.1% and 50.0% respectively), while the 2nd BC had the lowest positive rate of 22.0%. Further, mNGS was less sensitive to antibiotics exposure as compared to BC. The overall correlation between samples with vs without filtration (R2=0.96) confirmed that filtration does not affect microbial composition in a sample. For the BC positive samples, the effect of host depletion by filtration increased microbial target reads/million QC reads from 46 reads to 243 reads on average. Microbial reads enrichment by the filter appeared to be more effective for the samples with lower microbial concentration, thus increasing the test sensitivity over mNGS without filter. Using the 2nd BC results as reference, mNGS with filter and mNGS without filter exhibited sensitivities of 81.8% and 63.6%. ConclusionThe mNGS with filter was able to recover most of the pathogens identified by clinical BC and achieved the highest diagnostic yield. With the clinical implementation to complete the workflow within 24 hours, it has the potential to overcome slow turnaround and low diagnostic yield issues of traditional BC.