The potential of whole genome NGS for infectious disease diagnosis

Abstract

Non-targeted identification of microbes is now possible directly in biological samples, based on whole-genome-NGS (WG-NGS) techniques that allow deep sequencing of nucleic acids, data mining and sorting out of sequences of pathogens without any a priori hypothesis. WG-NGS was first only used as a research tool due to its cost, complexity and lack of standardization. Recent improvements in sample preparation and bioinformatics pipelines and decrease in cost now allow actionable diagnostics in patients. The potency and limits of WG-NGS and possible future indications are discussed here. WG-NGS will likely soon become a standard procedure in microbiological diagnosis.

KEYWORDS:

• infectious diseases
• diagnostic
• NGS
• virus
• bacteria
• fungus

View correction statement:

Corrigendum

The laboratory diagnostic of bacterial, viral and fungal diseases has long been dominated by culture, a method that does not necessitate any diagnostic hypothesis, but is limited by the non-cultivability of many pathogen species. Whereas viral diagnostic has mostly shifted to targeted nucleic acid testing [PCR; Nucleic Acid Sequence-based Amplification (NASBA)] for years, bacterial diagnostic remains mainly based on culture and more rarely on the use of molecular tests, such as PCR and antigen detection. The positive predictive value of bacterial culture is very high, and, additionally, bacterial isolation allows for antibiotic resistance phenotyping. Nevertheless, bacterial culture is not a perfect diagnostic tool. For example, 90–95% of blood cultures remain negative in immunocompromised patients suspected of infection; even in the cases where bacterial or fungal sepsis is likely, negative results are recorded for 50% of the blood samples.[1, 2] Moreover, false positive results can be obtained as a consequence of contamination during the sampling procedure. Quantitative bacterial loads determination seems clinically valuable but is not routinely performed, as it is time and resource consuming. Finally although it has decreased with advanced blood culture techniques, time to results is inconstant, from 2 days to more than 1 week, and requires additional time for antibiotic resistance determination. New tools, like Matrix Assisted Laser Desorption Ionization Time-of-Flight (MALDI-TOF), speed up identification of bacterial species after the cultivation step [3] but do not profoundly modify the intrinsic limits of cultivation assays.

The advantages of targeted molecular diagnostic are numerous: speed, low cost, automation, sensitivity and specificity, but indeed targeted molecular diagnostic is only able to identify predefined targets. Moreover, for highly variable pathogens like RNA viruses or DNA viruses that comprise multiple genotypes, conserved loci targeted by these tests do not discriminate between types. Very recently new molecular tests for direct biological sample testing have been released. They use specific multiplex amplifications with a narrow (Biofire, Biomerieux or SeptiFast, Roche Diagnostics) or very broad range [PCR amplification coupled with electrospray ionization mass spectrometry (PCR-ESI-MS), PlexId or Iridica, Abbott Laboratories] to detect a list of agents, defined by syndromes or taxonomy (e.g., bacteria vs viruses). These tests enlarge the landscape of molecular testing and might represent a major advance, even if sequence-dependent amplification does not cover the complexity of etiologies of infections in some indications, like infections of immunocompromised people and require a frequent adaptation of the panel interrogated.

Non-targeted identification of microbes is now possible based on Next Generation Sequencing (NGS) techniques, which allows deep sequencing of biological samples, data mining and sorting out of sequences of pathogens without a priori. In this strategy also known as whole genome NGS (WG-NGS), amplification of nucleic acids is random. This avoids the very difficult and strenuous work of designing hundreds of specific primers able to function simultaneously without interference against multiple pathogens and to continuously adapt their sequences for each description of new variants and species. Other advantages of WG-NGS are its ability to provide more detailed taxonomic information than diagnostic PCRs, as viral subtypes or serotypes (see for example[4]) and its good relationship between the number of NGS reads and CT values.[5–7] It may also lead to the discovery of new pathogens.

Nevertheless, given the unavoidable co-amplification of human nucleic acids (NA) one needs to use high depth of sequencing to reach PCR sensitivity levels, and due to human NA concentration variability, analytical sensitivity of WG-NGS is more critically influenced by matrix properties than PCR (see discussion in [8]). A whole genome coverage is in principle necessary to predict with a good confidence phenotypes such as resistance to antimicrobials or virulence, but indeed partial coverage might target by chance the loci encoding the phenotypes. This good coverage would also benefit from increasing target/NA reads. Improvements in the treatment of the sample to increase this ratio is very important and is now well known to be key to decrease the sequencing and bioinformatics costs.

WG-NGS was first seen as a research tool due to its initial cost, complexity and lack of standardization. More recently, we and others have shown that improvements of the sample preparation and bioIT pipelines allow actionable diagnostics in a short and constant time.[9, 10] These studies have shown the medical benefit of WG-NGS as a “last intention” test in patients who tested negative by standard diagnostic, leading to adapted therapy. In all cases, positive results were not due to the discovery of new pathogens, but to the identification of known pathogens not considered by the physicians as possible etiology or for which corresponding diagnostic tests were not available in clinical laboratories. So, one of the first diagnostic applications of WG-NGS could be testing the samples that remain negative with routine diagnostics either for technical or practical reasons, or because of they were not suspected by the clinicians and so being used as an additional diagnostic tool with an added value.

Will WG-NGS be used in the future as a first intention test? The response depends on the cost/medical benefit ratio. Infections in immunocompromised patients are a critical issue in clinical practice. Their number is increasing as a consequence of temporary or in many cases even long-term alterations of immune functions due to the development of new immunosuppressive therapies, for example organ transplantation or cancer treatment, and more generally due to the aging population and the growing prevalence of chronic diseases. We recently undertook a study to evaluate the usefulness of WG-NGS as a “first-intention” test by comparison with standard pipelines of laboratory in such a population of immunocompromised patients with very encouraging results. WG-NGS identifies bacterial genomes despite a lack of growth of cognate bacteria by standard culture. This result is not specific to WG-NGS as already evidenced by other molecular techniques based on specific amplification. For example, among 906 specimens from patients with suspected bloodstream infection, 33 were PCR-ESI-MS-positive but culture-negative: medical documentation analysis suggested true bacteremia in 31 cases.[11] Using a similar test with improved sensitivity in 331 specimens collected from patients with suspected bloodstream infections demonstrated 35 positive specimens compared with 18 positive by culture.[12] The most probable cause of this discrepancy is likely that in most infections, the load of cultivable bacteria is several order of magnitude lower than the genome copy number,[12] and this is amplified in patients under antimicrobial therapy.[13]

What could be the indication of such a wide-range test in immunocompromised patients suspected of infection? Time to results for WG-NGS is 2–3 days and the assay cannot be done as frequently of culture for economic reasons. Nevertheless, in most cases, multiple sampling for culture does not impact significantly the detection rate of pathogens but increase the risks of false positive, whereas other studies support the necessity of 2–4 consecutive blood cultures to detect >99% of all identifiable bacteriemic patients.[14] In practice, febrile immunocompromised patients are almost always under antibiotics at the time or even before the culture is started and the main interest of WG-NGS could be to adapt the antibiotic therapy, in particular in patients that do not respond to the first line. “Salvage microbiology” is a term used for the application of molecular diagnostic techniques in the detection of bacterial DNA directly from clinical specimens submitted for culture from patients who were treated with antimicrobial therapy.[15] Clinicians would probably do not take the risk of not using antibiotics even in the case of negative results delivered by such a broad-range molecular test, but, on the other hand, testing by WG-NGS can be done at inclusion, in parallel with culture, without the limitation of culture if an antibiotic therapy has already be initiated. Identification of major antimicrobial resistance genes would be also important for critical therapy decision. Results would be available most often before those of culture and would help adapt the treatment for patients not returning to apyrexia. Indeed testing can also be done as a second intention test, in patients who do not respond to antibiotics, but at the price of an additional delay. Such strategies should be evaluated in interventional studies, which are also needed to estimate the medical benefit and associated added- or saved costs.

Finally, it is likely that the main challenge for WG-NGS is not technical, but resides in the interpretation of the results by the physicians, a statement that is also valid for all broad range, without a priori tests, in the field of infectious diseases as in other medical disciplines (genomic, cancerology). Like bacterial culture, but with a much wider range of detection, WG-NGS identifies all agents present in a sample, independently of their responsibility in the disease. Interpretation requires appropriate training and is certainly easier in normally sterile samples, like blood, than in polymicrobial samples (respiratory samples, feces, etc.). Nevertheless, even in blood, the situation is sometimes complicated as translocation of compounds from gut bacteria is possible and might be associated with fever, even if live bacteria do not multiply and so are not detected by standard culture.[16] Another example is given by respiratory diseases that we have analyzed sputum for cystic fibrosis patients during clinical stability and at exacerbation and like others [17] have found that WG-NGS provides bacterial and viral composition in a single analysis that is more comprehensive and better covers the taxonomic diversity than cultures and PCRs (in particular by revealing the anaerobic flora, fastidious bacteria and viruses not screened in routine diagnostic), and could therefore be helpful in patients with recurrent or chronic bacterial bronchial colonization. Nevertheless, in our experience, when physicians are confronted to the results, they often do not know how to use them for the management of the patients. This added value of WG-NGS in clinical managements of patients will have to be evaluated, and additional work to associate typologies of microbiota with patient status and ability to respond to antibiotic therapy be needed.

WG-NGS has become very rapidly a standard tool in pathogen discovery but is currently in its infancy with regard to its use in medical diagnostic of infectious diseases. Powerful and automated sample preparation pipelines, sequencers in central laboratories and validated pipelines for read, sorting and taxonomic assignation are expanding rapidly, in parallel with decreasing costs. The progressive awareness of physicians to the patients’ benefit will certainly help WG-NGS to become standard in the practice of conducting microbiological diagnosis in some indications.

Financial & competing interests disclosure

M Eloit is the founder and CSO of Pathoquest (Paris, France), a spin out of Institut Pasteur dedicated to diagnostic of infectious diseases by NGS. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Additional information

Notes on contributors

Marc Lecuit

Marc Eloit