Can Circulating Cell-Free Microbial DNA Carry Us into the Future of Culture Independent Microbiology?

Organisms take time to grow, and advances in medical microbiology have sought to minimize the need for growth from primary specimens. Despite the speed of some nucleic acid amplification tests, they have several limitations: testing only for specific organisms, limited sample types, and few purely molecular markers for antimicrobial resistance. In the current state, culture-based methods remain fast enough given their comprehensiveness, clinical utility, and recovery of isolates for antimicrobial susceptibility testing.

Metagenomics refers to the en masse sequencing of DNA in a target unbiased manner, and metagenomic next-generation sequencing (mNGS)² has been used to discover previously uncultured organisms in epidemiology and human microbiome studies (1). Unlike culture-based diagnostics, mNGS is independent of target organism isolation and propagation and can detect a broad range of organisms. Outside of microbiology, a particular application of mNGS, plasma cell-free DNA (cfDNA) sequencing, has been explored clinically for prenatal screening and “liquid biopsies” for cancer diagnostics.

In studies of plasma cfDNA mNGS, small numbers of microbial DNA sequences were consistently identified in participants without infections (2). Experiments excluded contamination as the source of these sequences; they were thought to represent DNA from components of the human microbiome. Further analysis identified a substantial and previously uncharacterized virome and over 1000 novel bacterial taxa. This work established that plasma cfDNA mNGS could identify microbial DNA shed into the bloodstream from tissue-resident organisms, and the recently described Karius test in Nature Microbiology extends this methodology to clinical microbiology in the hope of assisting the diagnosis of infectious disease.

In this report, Blauwkamp and colleagues describe the Karius test, an mNGS assay that detects microbial cfDNA shed into the plasma from sites of infection (3). When looking at a new assay like this, 2 questions need to be answered: (a) is the method analytically valid and (b) is the method clinically useful? Blauwkamp and colleagues offer useful first steps toward the rigorous validation of these complex assays; however, the clinical implementation, interpretation, and stewardship thereof raise a series of important and open questions.

The Karius test work flow consists of plasma separation, cfDNA extraction, sample and library preparation, single-end 75 base-pair sequencing, data cleanup, comparison to a microbial sequence database, and report generation. The assay can potentially identify 1250 pathogens across 16 phyla and 364 genera. Detected organisms are reported quantitatively as molecules per microliter (MPM). Importantly, the developed work flow can provide results in a clinically relevant time frame to affect patient care. The described validation is thorough and largely adheres to a recently outlined approach for validation of mNGS testing for infectious diseases (4). A preliminary look at the clinical performance of the assay is provided by the Karius-sponsored Sep-Seq trial (NCT02730468) wherein patients presenting with sepsis were tested by both the Karius test and standard microbiological methods at presentation.

Much of the validation was performed with a set of 13 diverse organisms added into samples, but several important classes of organisms were not represented in the validation set (e.g., yeast, non-Staphylococcal Firmicutes including Streptococci and Clostridia, Actinobacteria, Bacteroidetes, human herpes viruses). It would be important to know if the in silico analysis (i.e., data supplementing studies) included these organisms and if it accurately predicted assay performance by doing targeted limit-of-detection and specificity testing. The combination of quantitative reporting with a limited validation data set creates challenges for test interpretation. This limitation is reinforced by the observed higher limit of detection for the important pathogen Pseudomonas aeruginosa. This is likely related to the presence of nucleic acid from this organism in the environment, but immunosuppressed patients are particularly susceptible to infection from environmental organisms. Thus, it will be important to ascertain the performance of the Karius test in different patient populations and through on-going validation studies.

The Karius test uses control sequences added directly to specimens to assess analytic processing. These sequences are used to normalize the number of detected microbial sequences generating the MPM value. In general, quantitative measurement of microbial cfDNA should help identify the most relevant pathogen to providers when multiple organisms are detected. However, there are no clear cutoffs differentiating pathogens from nonpathogens, and culture-confirmed pathogens may not have the highest MPM in a sample, as indicated by the supplemental data. Additional clinical data may help resolve this complexity, and it will be important to provide such information in future reports of this and similar tests to guide clinical decision-making.

A final concern with the method described is the lack of detection of RNA viruses such as enteroviruses, some hepatitis viruses, most respiratory viruses including measles, arboviruses, and hemorrhagic fever viruses. RNA viruses are prominently featured in the microbial differential diagnosis of patients with fulminant hepatitis, acute or subacute meningoencephalitis, infants and children with culture-negative sepsis, and travelers with unexplained febrile illnesses, among others. Providers often don't consider the phylogeny and underlying molecular biology of the pathogens in their differential diagnosis, and clear communication with clinicians about the limitations of these methods is required. Alternative diagnostic approaches should be considered if these organisms are in the differential diagnosis.

Review of the Sep-Seq data presented in the report (3) and supplemental data demonstrates both the challenges and promise of this complex data. Patient selection for this trial may not match its real-world application, but the Karius test was able to detect a likely pathogen in over half the cases. However, many cases had multiple microbes identified, and the additional detections were largely components of the human microbiome. This result should not be surprising given the use of plasma cfDNA to characterize the microbiome (2). However, it is important to consider that tissue-based infections (e.g., liver abscess) can be polymicrobial, and all organisms may not be recovered in culture and/or individually identified (i.e., cultures reported as “mixed”). The multiple detections may reflect the polymicrobial nature of these infections, but the lack of localizing information complicates the clinical interpretation of polymicrobial reports especially because testing may be performed on patients at risk of mucosal barrier disruption and mucositis.

The provision of empiric antibiotics to patients before obtaining samples for culture complicates their evaluation and management. Dead bugs don't grow, but they may continue to shed detectable cfDNA. Although the Karius test does not provide antimicrobial susceptibly data, it may remain positive after cultures become sterile after antibiotic treatment. This phenomenon likely explains the detection of pathogens causing community-acquired pneumonia that were not confirmed by culture. Empiric therapy regimens for community-acquired pneumonia consider these pathogens, but the lack of an identified pathogen in tissue-based infections can lead to prolonged courses of broad-spectrum antibiotics. Identification of a pathogen can inform decision-making about the required spectrum and duration of therapy, promoting antimicrobial stewardship. However, detection of organisms outside of standard empiric treatment regimens may prompt providers to unnecessarily broaden or prolong antibiotic therapy. It will be interesting to investigate the effects of this testing on antimicrobial stewardship.

Among the patients in the Sep-Seq trial, there were several “definite” or “probable” pathogens detected whose identification would have likely been delayed or missed with conventional methods. Depending on patient population and local testing practices, Pneumocystis jirovecii, Mycobacterium tuberculosis, Mycoplasma pneumoniae, human herpes viruses, Legionella longbeachae, and Borrelia hermsii may not be included in first- or second-line testing. In patients with subacute or chronic infections who may have been exposed to antibiotics, we expect this list to grow and broaden to encompass more atypical bacterial pathogens, Mycobacteria, and fungi.

A large, unmet clinical need that plasma cfDNA mNGS may address is the diagnosis of invasive infections, especially fungal and mycobacterial infections, in the immunocompromised and those refractory to invasive testing. These patients often receive empiric antibacterial agents, but the decision to start antifungal or antimycobacterial therapy is guided by clinical decision-making, imaging studies, and laboratory tests with suboptimal performance characteristics. Thorough culture evaluation requires invasive sampling and extended incubations and are often negative. Additionally, antifungal and antimycobacterial therapies can be expensive and/or toxic, and identification of a bona fide pathogen would facilitate clinical decision-making. Future studies should be performed to identify the negative predictive value for plasma cfDNA mNGS to help make treatment decisions.

Among nonculture based assays available, Karius is the first (and currently only) vendor offering plasma cfDNA mNGS for infectious disease diagnosis; however, mNGS analyses of cerebrospinal fluid and respiratory specimens and tissue-based targeted PCR and sequencing are also available. Faced with these choices for patient testing, laboratories and clinicians must decide locally what their priorities are and how to best achieve those priorities. We have laid out some questions (Table 1) that we think are important to consider when deciding how to implement a comprehensive testing program potentially incorporating on-site nucleic acid amplification tests, site-specific mNGS, cfDNA mNGS, culture, and site-specific targeted sequencing. A larger question outside the scope of this Perspective is how the human sequence data generated by these methods should be incorporated into existing frameworks for the management of genomic data.

Finally, we cannot ignore the cost of these assays. All currently available clinically validated mNGS assays performed on primary specimens are expensive, with the Karius test retailing for over $2000. Because patients undergoing this testing will most likely be hospitalized, institutions will bear the majority of this cost, and a business case for testing will depend on savings realized through improved outcomes and avoided diagnostic costs. The current cost of the test largely precludes longitudinal testing of patients, but such data could be helpful for facilitating the clinical interpretation, especially in light of the detection of components of the microbiome.

Direct marketing of highly complex commercial tests to providers who lack adequate laboratory expertise to evaluate the merits of the claims is increasing, and it is important for studies such as this to be published. Karius has opened up their test for evaluation and potential criticism by the laboratory and infectious disease community, allowing an informed dialog among stakeholders in deciding how to implement this test. We now have a number of choices for the non–culture-based evaluation of patients, and we need more data about how to best implement the tools we now have at our disposal including the Karius test.

Footnotes

Author Contributions: 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.

Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:

Employment or Leadership: None declared.

Consultant or Advisory Role: None declared.

Stock Ownership: D.R. Peaper, Tangen Biosciences.

Honoraria: None declared.

Research Funding: None declared.

Expert Testimony: None declared.

Patents: None declared.

References