2022/01/27 by David C Gaston, David C. Gaston, Patricia J Simner +1 · 1 citation
Medicine · #Cytomegalovirus and herpesvirus research #Respiratory viral infections research #Mycobacterium research and diagnosis
paper · pdf · doi:10.1093/clinchem/hvac007
Pathogen detection and identification is a fundamental service provided by clinical microbiology laboratories. Metagenomic next-generation sequencing (mNGS) as applied to infectious disease diagnostics is a developing field (1, 2). mNGS for this purpose attempts to identify nucleic acids derived from pathogens present in a sample without a priori suspicion of a specific etiologic agent in an untargeted fashion. The 2020 Nature Medicine article from Gu and colleagues provides insight into important questions facing this field as methodologies advance toward broader clinical use in clinical microbiology laboratories (3). There is no single standardized approach to mNGS for infectious disease diagnostics, and methods utilizing different approaches have varied benefits depending on the microbial group (e.g., bacterial, viral, parasites, fungi) and sequencing goal (4). The use of mNGS leverages powerful sequencing capabilities for the objective of providing broad, accurate pathogen identification in the shortest amount of time. In the Gu study, the authors develop an innovative method of dual-barcoding to compare workflows using both Illumina and Nanopore technologies, as well as adapt automated bioinformatic analysis pipelines for the 2 sequencing approaches. The analytes measured are sequenced fragments of cell-free DNA (cfDNA), mostly <100 nucleotides but up to 500 to 600 nucleotides, which are released into an infected space during the course of disease. Each sequenced fragment is referred to as a read, and each read is either human or microbial derived. Read counts for an individual pathogen are made comparable between samples by calculating the number of organism reads per every million total sample reads (reads per million; RPM). A conversion factor generates the normalized RPM (nRPM) metric, making this assay semiquantitative and grossly aligned with abundance of cultured pathogens. This provides a method of interpretation that can be conceptualized similarly to standard microbiology methods, potentially aiding pathogen identification. Illumina and Nanopore sequencing workflows differ in the chemistry used to produce data, the availability of data during a sequencing run, the sequencing depth obtained, and the number of errors present in the data. Illumina sequencing is based upon “sequencing-by-synthesis” technology, in which data are generated by the addition of fluorescent nucleotides to DNA affixed to an Illumina flow cell. Illumina-based sequencing can achieve profound sequencing depths with exceedingly low error rates, though sequencing data are not available for analysis until the sequencing run is complete. Nanopore sequencing utilizes synthetic protein pores present in the Nanopore flow cell that generate nucleotide-specific electric signals as DNA passes through the pores. Sequencing data are available for analysis in real-time, though a higher error rate is present when compared to Illumina sequencing. The same bioinformatic analysis pipeline and reference database is utilized in this study to analyze data from both workflows, with the analysis pipeline adapted for the Nanopore workflow to analyze data as they are generated. Importantly, the authors found only minor differences between the ability of Illumina and Nanopore workflows to detect and identify pathogens in validation sample sets. Sensitivity calculated by comparison to standard microbiology methods differed by 4.2% for bacterial pathogens (Illumina 79.2%, Nanopore 75.0%) and by only 0.3% for fungal pathogens (Illumina 90.6%, Nanopore 90.0%). Specificity diverged more, with a difference of 9.2% for bacterial pathogens (Illumina 90.6%, Nanopore 81.4%) and 11.0% for fungal pathogens (Illumina 89.0%, Nanopore 100.0%). Clinical performance for bacterial pathogens calculated by comparison to a composite standard demonstrated a difference in positive percent agreement of 1.0% (Illumina 80.0%, Nanopore 81.1%) and in negative percent agreement of 2.3% (Illumina 95.3%, Nanopore 93.0%). Smaller numbers of samples were tested for Nanopore (n = 43) than for Illumina (n = 127), impacting these performance measures. Regardless, these analyses demonstrate both workflows performed overall at a level appropriate for use in patient care with comparable performance characteristics. They also demonstrate the higher error rate with Nanopore can be, for the most part, overcome by bioinformatic approaches and does not significantly impact performance for this application. A pressing question in the field of mNGS for infectious disease diagnostics is the optimal sample source to test. Conventional approaches hold that sampling an infected site is preferred to sampling distant or communicating sites. However, sampling infected sites may sequence an abundance of host reads that limits the ability to detect pathogens. Other sample sources containing cfDNA, notably plasma, could offer alternative samples with fewer host reads and that avoid invasive sampling. This study utilized samples from multiple sources, including cerebrospinal fluid, synovial, pleural, peritoneal, bronchoalveolar lavage fluid, urine, abscess contents, and plasma. Pathogen reads were detected from all sources, and nRPM counts were not significantly different between the sequencing workflows for all sample types combined. The authors compared nRPM counts from the infected sample source to those obtained from plasma in a subset of samples. Although pathogen cfDNA was detected in plasma, reads were more consistently identified and above the detection threshold from infected sites. Accordingly, plasma cfDNA testing may be inferior to testing samples from infected sites, and testing infected samples may be best pursued, if available, in lieu of plasma cfDNA testing. An anticipated component of the power of mNGS for infectious disease diagnosis is the possibility of detecting pathogens otherwise missed by standard approaches. In this study the authors report a case series of 12 patients with infectious syndromes that had either incomplete or negative microbiologic diagnoses by standard testing. mNGS testing provided likely pathogen identifications in 9 of these cases, 5 of which were not achieved by targeted molecular testing. Importantly, 4 of these cases represented invasive fungal infections for which standard testing is currently lacking. The performance characteristics for fungi by both Illumina and Nanopore workflows viewed in the context of detecting otherwise missed fungal pathogens adds to anticipation that mNGS can be a useful tool in the diagnosis of invasive fungal infections. However, not all cases identified by standard testing were identified by mNGS approaches, and the performance characteristics exhibit a need for additional optimization. An important implication is that mNGS testing can serve to supplement but not replace standard microbiologic testing. Although the sequencing workflows demonstrated overall similar performance characteristics across multiple sample sources, differences between the workflows relate to practical considerations for implementation in clinical microbiology laboratories. Cost varied between the workflows, with the cost per sample ranging from 27.20 to 61.40 for the Illumina workflow (based upon the amount of batch testing utilized) vs 269.70 per sample for the Nanopore workflow (also using a batch testing strategy). It is notable, however, that sequencing costs are individualized to laboratories and highly dependent upon batch testing. Batching is a cost-efficient method for potential clinical application, but can increase the turnaround time required to provide a result considerably. Using the methods in this study, the purported time from sample processing to results was approximately 6 h for the Nanopore workflow and 24 h for Illumina, a time differential attributable to the real-time sequencing capabilities of Nanopore workflows. However, this approximates only the laboratory time in an ideal scenario, and does not incorporate clinical factors that would lengthen turnaround time. At present, mNGS for infectious disease diagnostics is not an on-demand testing strategy as are other molecular diagnostics. Decreasing the reporting turnaround time for pathogen identification can directly impact administration of the most appropriate antibiotics and care pathways, with potential impacts on such important outcome measures as appropriate antimicrobial use, total antimicrobial days, total admission days, and mortality. Furthermore, despite the simplified workflows and automation via liquid handlers proposed by the authors, the methodologies still require considerable oversight in the laboratory. mNGS for infectious disease diagnostics is currently in limited use for patient care but likely will be more broadly utilized as testing and experience become increasingly available. This study from Gu and colleagues presents helpful perspectives into performance characteristics for 2 sequencing workflows and the utility of these workflows across multiple infected sample sources. Many questions remain when considering how to best utilize mNGS testing for infectious disease diagnostics, including if testing should be limited to specific patient populations, when in the course of disease should testing be submitted, and how mNGS testing can provide insight into antimicrobial resistance, virulence, and organism viability. Ongoing research is needed to investigate these and other questions in the field. Ultimately, we foresee mNGS testing developing into a precision medicine approach addressing occult pathogens, microbiome composition, host response, and even non-infectious etiologies such as cryptogenic malignancies (5), among other uses. The insights provided in this study are beneficial to clinicians considering how to utilize available mNGS testing, as well as for clinical microbiologists considering how mNGS can be incorporated into testing offered by a laboratory. It is important to note that sequencing workflows other than those included in this study are available with varied benefits and limitations, such that laboratories must consider differences in start-up and testing costs, turnaround time, capacity for bioinformatic analysis, and data management strategies when seeking to offer clinically validated mNGS testing. At present, this testing requires substantial infrastructure and expertise that is not widely available. Advances in the field, such as those presented in this article, help bring this testing strategy forward and toward broader use for patient care. All authors confirmed they have contributed to the intellectual content of this paper and have met the following 4 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved. Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest: D.C. Gaston, leadership in the American Society of Transplantation (Consensus Conference for Advanced Diagnostics for Transplant Infectious Diseases) and the Infectious Disease Society of America (Fellows Subcommittee); P.J. Simner, Subcommittee on Antimicrobial Susceptibility Testing for Clinical and Laboratory Standards Institute (CLSI), College of American Pathologists (CAP) Microbiology Committee, and Antibacterial Resistant Leadership Group Diagnostic Committee. P.J. Simner, OpGen, Inc., BD Diagnostics, Shionogi, and GeneCapture. P.J. Simner, GeneCapture. P.J. Simner, GenMark, Dx. D.C. Gaston, grants or contracts from the Johns Hopkins Fisher Center Discovery Program and the Johns Hopkins Accelerated Translational Incubator Pilot Program, receipt of research materials from IDbyDNA and Illumina; P.J. Simner, grants or contracts from Affinity Biosensors, OpGen, Inc., BD Diagnostics, and Qiagen, receipt of research materials from Illumina, IDbyDNA, and ArcBio. None declared. None declared. metagenomic next generation sequencing cell free deoxyribonucleic acid