Original Title: MSAC application no. 1646

Authors' objectives: To assess the safety, effectiveness and cost-effectiveness of whole genome sequencing (WGS) of the Mycobacterium tuberculosis pathogen, to assess for antimicrobial drug susceptibility (compared to phenotypic drug susceptibility testing (pDST) alone).

Authors' results and conclusions: No direct from test to health outcomes evidence was identified comparing the effectiveness of WGS plus pDST versus pDST alone. A linked evidence assessment was therefore used, linking the accuracy of WGS versus pDST, data on the turnaround time of WGS and pDST, clinical guidelines to suggest how management of patients would likely change based on the accuracy and turnaround time, and data on how appropriate versus inappropriate treatment changes health outcomes. The incremental benefit of WGS can be inferred from these studies. Jajou et al. (2019) performed a retrospective cohort study on patients with MTBC from the Netherlands (with a very similar rate of drug resistance to Australia). WGS detected more cases of resistance to rifampicin, isoniazid and ethambutol than pDST, with a slight loss of specificity. It is therefore logical that WGS ± pDST would be more accurate than pDST alone, although conflicting results between WGS and pDST can make it difficult to interpret the patients’ resistome profile appropriately. Per 1000 patients tested, WGS would result in 77 additional cases of resistance to isoniazid, ethambutol or pyrazinamide being identified. Up to 10 per cent of patients who had their mycobacterial infection tested for resistance to second- or later-line drugs had isolates with resistance-conferring variants to at least one antibiotic (103/1000 for ethionamide), which was not detected by pDST. However, from South Australian data, none of the 5/75 (6.7%) patients whose infection had ethionamide resistance-conferring variants would be considered for this treatment (as they were all likely to respond to at least three first-line drugs). The clinical impact of this discordance is therefore negligible. Despite the majority of patients’ infections having the same antimicrobial resistance results based on WGS and pDST, the use of WGS is proposed by the applicants to result in a change in management, due to the faster average turnaround of WGS results compared to pDST. NSW data on 32 isolates from Martinez et al. (2016) reported that WGS results had a median turnaround time of 11 days, compared to pDST, which took a median of 16 days. Personal communication with the authors stated that their WGS results are now being turned around within 8-10 days . Although not all laboratories in Australia are currently equipped to turnaround WGS results as fast as NSW, the expectation is that in the near future, the other reference laboratories will have similar in-house capacity . Slightly longer turnaround times have been reported in the international literature, with mean turnaround times for WGS ranging from 3 to a mean 15 days, whereas turnaround time for pDST ranged from 13 to 24 days for first-line drugs, and 21 to 47 days for second-line drugs. No data were found on the impact of WGS on reducing transmission of TB. The incremental cost of WGS per correct diagnosis after initial pDST was calculated at AU$4,551. The ICER is most sensitive to changes in the difference in test turnaround times and the proportion of cases with MDR-TB in hospital isolation at the time of testing.

Authors' recommendations: No direct from test to health outcomes evidence was identified comparing the effectiveness of WGS plus pDST versus pDST alone. A linked evidence assessment was therefore used, linking the accuracy of WGS versus pDST, data on the turnaround time of WGS and pDST, clinical guidelines to suggest how management of patients would likely change based on the accuracy and turnaround time, and data on how appropriate versus inappropriate treatment changes health outcomes. The incremental benefit of WGS can be inferred from these studies. Jajou et al. (2019) performed a retrospective cohort study on patients with MTBC from the Netherlands (with a very similar rate of drug resistance to Australia). WGS detected more cases of resistance to rifampicin, isoniazid and ethambutol than pDST, with a slight loss of specificity. It is therefore logical that WGS ± pDST would be more accurate than pDST alone, although conflicting results between WGS and pDST can make it difficult to interpret the patients’ resistome profile appropriately. Per 1000 patients tested, WGS would result in 77 additional cases of resistance to isoniazid, ethambutol or pyrazinamide being identified. Up to 10 per cent of patients who had their mycobacterial infection tested for resistance to second- or later-line drugs had isolates with resistance-conferring variants to at least one antibiotic (103/1000 for ethionamide), which was not detected by pDST. However, from South Australian data, none of the 5/75 (6.7%) patients whose infection had ethionamide resistance-conferring variants would be considered for this treatment (as they were all likely to respond to at least three first-line drugs). The clinical impact of this discordance is therefore negligible. Despite the majority of patients’ infections having the same antimicrobial resistance results based on WGS and pDST, the use of WGS is proposed by the applicants to result in a change in management, due to the faster average turnaround of WGS results compared to pDST. NSW data on 32 isolates from Martinez et al. (2016) reported that WGS results had a median turnaround time of 11 days, compared to pDST, which took a median of 16 days. Personal communication with the authors stated that their WGS results are now being turned around within 8-10 days . Although not all laboratories in Australia are currently equipped to turnaround WGS results as fast as NSW, the expectation is that in the near future, the other reference laboratories will have similar in-house capacity . Slightly longer turnaround times have been reported in the international literature, with mean turnaround times for WGS ranging from 3 to a mean 15 days, whereas turnaround time for pDST ranged from 13 to 24 days for first-line drugs, and 21 to 47 days for second-line drugs. No data were found on the impact of WGS on reducing transmission of TB. The incremental cost of WGS per correct diagnosis after initial pDST was calculated at AU$4,551. The ICER is most sensitive to changes in the difference in test turnaround times and the proportion of cases with MDR-TB in hospital isolation at the time of testing.

Authors' methods: Insufficient direct evidence was identified during scoping searches of the topic, so a linked evidence approach has been used to assess the clinical utility of WGS for identification and characterisation of antimicrobial resistance in mycobacteria. Although the proposed MBS item is for any type of mycobacterial infection, the exemplar is M. tuberculosis and the assessment of other types of mycobacteria has been facilitated. A systematic review of evidence on the use of WGS for drug resistance testing in mycobacteria was performed. Population/Intervention/Comparator/Outcomes (PICO) criteria were established for a linked evidence approach. A protocol was registered a priori with the International Prospective Register of Systematic Reviews (PROSPERO 2021 CRD42021271969). Search terms (text terms and MeSH terms or equivalent) related to WGS, drug susceptibility testing, and tuberculosis were used. These terms allowed identification of studies on the accuracy of WGS, and the impact WGS has on the management of patients. PubMed (including pre-Medline), Embase, Cochrane, ClinicalTrials.gov, International Clinical Trials Registry Platform, Australian Clinical trials Registry, INAHTA HTA database and PROSPERO were searched. Studies were restricted to those which included both WGS and pDST (i.e. did not include non-comparative studies). Studies in a language other than English were considered if the title/abstract was in English, but would only have been translated if it appeared to provide a higher level of evidence than available in English. Two reviewers looked at titles and abstracts of the identified citations, with an overlap of approximately 70 per cent, and a high concordance rate. Full text articles were assessed by one reviewer. Due to the high volume of articles on the concordance of WGS compared to the comparator pDST, a third cull was performed to limit these studies to those which provided 2x2 data.

Project Status: Completed

URL for project: https://www.msac.gov.au/applications/1646

URL for protocol: https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=271969

Year Published: 2021

URL for published report: https://www.msac.gov.au/applications/1646

Requestor: Department of Health and Aged Care / Medical Services Advisory Committee (MSAC)

English language abstract: An English language summary is available

Publication Type: Full HTA

Country: Australia

Organisation Name: Adelaide Health Technology Assessment

Contact Address: School of Public Health, Mail Drop 545, University of Adelaide, Adelaide SA 5005, AUSTRALIA, Tel: +61 8 8313 4617

Contact Name: ahta@adelaide.edu.au

Contact Email: ahta@adelaide.edu.au

Copyright: <p>Adelaide Health Technology Assessment (AHTA) on behalf of NICS</p>