Evaluation of a mNGS Workflow for Infection Diagnosis Using Oxford Nanopore Sequencing.

An External Evaluation of a Metagenomic Next Generation Sequencing Workflow for Infection Diagnosis Using Oxford Nanopore Sequencing.

This is a laboratory evaluation of a new testing methodology for microbiological diagnosis, whereby participant samples received as part of routine care will be divided between the standard diagnostic pathway and this new pathway: metagenomic next generation sequencing (mNGS). Results obtained from the mNGS pathway will be compared against the standard diagnostic pathway in terms of sensitivity, specificity, accuracy and clinical impact. The samples will be identified at Wellington Southern Community Laboratories (WSCL), which provides laboratory services for Capital and Coast District Health Board, and forwarded to the Institute of Environmental Science and Research (ESR) to undergo mNGS testing.

Diagnostic microbiology has traditionally involved culture of organisms to diagnose infection, which is time consuming, insensitive for organisms that are difficult to grow, and compromised by prior antimicrobial therapy. Molecular diagnostics, predominantly in the form of nucleic acid amplification tests (NAAT), e.g. PCR, overcome some of these limitations and are now in widespread and increasing use. NAAT-based tests are however limited by only being able to detect a small number of pre-specified organisms and can offer limited to no antimicrobial susceptibility information.

Metagenomic next generation sequencing (mNGS) works by directly sequencing all of the nucleic acid in a microbiological sample, thus allowing identification of all microorganisms that are present in sufficient quantity, along with the potential to infer antimicrobial susceptibility patterns based on the presence or absence of relevant genes. Unlike NAAT, no pre-specification of target pathogen(s) is required, so mNGS has the potential to identify important pathogens that may have not been tested for otherwise. Host (human) sequences will also be present in the sample, so are removed from the analysis either by preventing them from being sequenced, or deleting them during the initial analysis steps.

mNGS therefore has the ability to overcome the limitations of both culture-based and NAAT-based infection diagnosis, with the potential to offer rapid diagnostics with greater levels of antimicrobial susceptibility detail, which is less affected by whether the organism is viable/culturable. Rapid infection diagnostics has the ability to significantly improve patient care, whereby appropriately targeted antimicrobial therapy can be instituted promptly (or ceased if e.g. a viral pathogen is identified). This is of particular importance given ongoing global increases in antimicrobial resistance. Rapid diagnostics with mNGS may also reduce the need for multiple other lines of investigation. There are likely to be certain groups of patients where this technology can be particularly targeted for maximal benefit either due to the rapidity of the results or the ability to diagnose infections that may not have been clinically suspected or detected with standard processes. In the investigators' department, several cases have been seen recently where patients have had very poor outcomes due to delays in diagnosis, where mNGS would have had the potential to markedly improve their outcomes. There are also potential benefits on a population level, such as reducing exposure of the population to overly broad-spectrum antibiotics, rapid identification and surveillance of communicable diseases that may require a public health response, and expediting appropriate management and flow of patients through an already congested hospital system. mNGS also has the ability to detect novel pathogens. As an example, the rapid identification and dissemination of information relating to SARS-CoV-2 was due to the availability of rapid 'agnostic' sequencing technologies.

Next generation sequencing has typically been too expensive to be used as a front-line diagnostic test, with its use confined to larger research-affiliated institutions. However, nanopore sequencing (Oxford Nanopore Technologies [ONT]), now offers a relatively inexpensive option, with a small physical footprint and an ability to generate a large amount of sequence data rapidly, making it a potentially viable option for front-line diagnostic microbiology laboratories. As such, there is considerable interest in the use of nanopore sequencing for mNGS. A number of publications have reported on its use in clinical diagnostics, and it is already in use in a number of healthcare settings overseas

. Continuous Quality Improvement (QI) via the evaluation of new diagnostic assays is a critically important component of clinical laboratory medicine. In line with this, the investigators are interested in evaluating the use of mNGS in their laboratory as a QI initiative to enhance the diagnostic service, increase the sensitivity of infection diagnostic testing, and compare existing standard diagnostic procedures against mNGS. The investigators plan to undertake this in the form of an external evaluation, whereby samples from Wellington Southern Community Laboratories (WSCL) would be forwarded to the Institute of Environmental Sciences and Research (ESR) for mNGS testing. ESR has existing expertise in sequencing and bioinformatics and have already developed mNGS capability, however have not comprehensively evaluated it on real patient samples. The initial evaluation would occur at ESR, with the aim of producing a workflow that could be usable at WSCL.

Sample selection and referral from WSCL to ESR

1. Residual samples that have been collected as part of routine patient care and sent to WSCL microbiology laboratory for the purposes of diagnosing infection will be used. It is standard accepted practise in the laboratory for the microbiologist to arrange additional (unrequested) testing on certain clinical samples to optimise the diagnostic process, including referral of samples to an external laboratory such as ESR. This evaluation would follow a similar procedure.
2. Samples will be identified at WSCL by the clinical microbiologist(s) involved in the project. A variety of sample types will be evaluated. The specific sample types that will be assessed in the evaluation will vary during the course of the evaluation, based on how well the mNGS workflow functions on different sample types. Initially samples from normally sterile sites and those with potentially the greatest positive impact on patient care will be chosen (e.g. cerebrospinal fluid, joint fluid, pleural fluid, blood) followed by samples from non-sterile sites (e.g. sputum, urine, wound fluids) if the initial results are encouraging.
3. WSCL processes microbiology samples for the entire Capital & Coast and Hutt Valley DHB regions, so samples would be sourced from patients in these regions. The majority of samples will be from hospitalised patients, rather than community-based. Samples from a variety of different clinical specialties will be evaluated, including augmented care units e.g. both adult and neonatal intensive care units, and general ward patients.

Sample size 1. A specific sample size has not been calculated for this evaluation, as the total number of samples tested will be contingent upon how much refinement of the mNGS testing process is required, and the evaluation is likely to need to be an ongoing process. The investigators have set a maximum sample size at 400.

mNGS methodology

1. Clinical samples will be processed to enrich for bacterial cells using solid phase reversible immobilisation (SPRI) magnetic beads functionalised with poly-lysine. Host (human) DNA will be depleted using differential lysis and nuclease treatment.
2. Total nucleic acid will be extracted from the processed samples using commercially available DNA extraction kits.
3. ONT rapid sequencing kit will be used to sequence the resulting DNA.
4. To identify potential pathogen species in the sample, all non-human sequences generated will be will taxonomically classified to the species level using a fast minimiser-based approximate mapping algorithm against customised pathogen databases.

Avoidance of host (human) genome sequencing Robust processes involving several different published strategies will be put in place to avoid the possibility of inadvertent human genome sequencing.

1. Reducing host DNA in the sample:

   a. Bacterial cell enrichment and chemical depletion of host DNA during the initial sample processing stage of the protocol will be the first line in avoiding sequencing human DNA by reducing the amount of host DNA in the sample.

2. Ejecting host DNA from the sequencer:

   a. The second filter to reduce host sequencing will the use of the ONT 'Read Until' API. This process automatically prevents prespecified DNA sequences from entering the detector by reversing the molecule's direction of travel through the detector within less than one second, preventing any full-length host molecules from being sequenced. Given the raw error rate of the sequence data, any short host reads that pass this filter carry insufficient information to be analysed.

3. Deleting host sequences:

   a. The final step to avoid exposing host sequence to analysis is automatically and permanently deleting any residual human sequence data as it is produced, prior to the data stream entering the analysis step.

4. Mapping sequences only against microbial databases:

   1. In the very unlikely event that host sequence passes the above steps, a further safety step is that any sequences will only be mapped against microbial databases. This means that any human sequences would not create a match, so would not form part of the analysis.

Reporting of results

1. Results will be reported by ESR directly to WSCL via the same secure pathways used for other reference testing that ESR performs for WSCL.
2. Results will be reviewed by the clinical microbiologist(s) involved in the project so that an assessment of their clinical relevance can be made prior to results being made available to clinical teams.
3. Results will be entered onto the WSCL laboratory information system so that they can be reported to the clinical team. A comment will be attached to the text report explaining that the results have been generated using mNGS, which is still being evaluated, and the microbiologist will discuss directly with the clinical team regarding the results if there is any potential for uncertainty surrounding the result.

Evaluation of results

1. The results of the parallel mNGS testing will be compared over time to the results produced by routine laboratory testing on the same sample, to assess the sensitivity, specificity, and level of agreement between methods. Given that mNGS is potentially more sensitive than routine methods, mNGS results will also be assessed by the clinical microbiologist(s) involved in the project against other routine diagnostic (orthogonal) testing, which could include other laboratory tests, radiology, and the overall clinical assessment of the patient.
2. The clinical impact of the mNGS testing will also be assessed and recorded: for each mNGS result the clinical microbiologist(s) will make an assessment as to the clinical impact the mNGS result had on the overall assessment and management of the patient.

Inclusion Criteria:

• All samples received by the WSCL microbiology laboratory for testing for the purposes of diagnosing infection will be eligible.

Exclusion Criteria:

• Use of residual sample for mNGS testing may leave too little remaining sample and compromise standard diagnostic testing.
• Patients who have requested that their residual samples be returned to them.