1. Using Whole Genome Sequencing for the diagnosis, antimicrobial resistance profiling and cluster detection of TB direct from MGIT culture: An international pilot study
L. J. Pankhurst1 on behalf of the COMPASS-TB* and MMM Consortiumǂ
1NDM, University of Oxford, John Radcliffe Hospital, Oxford, UK.
Introduction and Purpose
Whole Genome Sequencing (WGS) has the potential to replace the current diagnostic system for M. tuberculosis (MTB). Current systems rely on culture, taking up to 2 months to yield full species and sensitivity profiles due to the slow growth rate of Mycobacterium spp. (Figure 1)1. WGS could yield species, susceptibility profiles and genotyping within 2 days if performed from newly positive Mycobacterial Growth Indicator Tubes (MGITs). We investigated the feasibility of WGS for Mycobacterium diagnosis in an ongoing international multicentre pilot study.
Methods
Eight international sites in the United Kingdom (Public Health England-Leeds, PHE-Birmingham, PHE-Oxford and PHE-Brighton), Republic of Ireland-Trinity College Dublin/St James's Hospital, Germany-Borstel, France-Lille and Canada-Vancouver, processed just-positive (1–5 days) MGIT tubes from hospital routine laboratories. DNA was extracted from 1mL boilates using mechanical disruption followed by ethanol precipitation. Sequencing libraries were prepared for the Illumina MiSeq platform using a modified version of the Nextera XT (Illumina) protocol. Data were uploaded via Illumina-BaseSpace to an in-house pipeline.
Gene presence-absence analysis was used to identify species from raw read data. Reads were subsequently mapped to a TB reference genome and the presence of any known resistance conferring mutations identified. Finally, assembled MTB complex genomes were compared to a catalogue of previously WGS isolates to find potential transmission events (genetically nearest neighbours)2. Results were manually reported within as little as 2 days of completing WGS.
Results
249 of 292 samples sequenced to date had routine laboratory species identification. 235 (94.4%) WGS were concordant for species complex or failure to identify species (Figure 2). There were 3 discordant results: 1 M. malmoense (routine laboratory)/M. haemophilum (WGS); 1 M. absessus complex (routine laboratory)/mixture of M. abscessus complex and M. avium-complex (WGS); and 1 M. abscessus complex (routine laboratory)/M. avium complex (WGS). Failure to identify species occurred in both assays (9) in routine laboratory only (3) and by sequencing only (8). Sequencing failure occurred owing to low sequencing read-depth and contaminating human or nasopharyngeal-flora DNA.
Routine laboratory and WGS antibiotic sensitivity results were compared in 103 MTB complex samples (Table 1). Drug sensitivities were correctly predicted by WGS in 92% of the 86 isolates for which full data was available. In addition, by WGS 3 isolates had predicted resistance to second-line drugs not tested in the reference laboratory (Amikacin: 1; Quinolones: 2). There was one discordant result, with a BCG isolate phenotypically resistant to Ethambutol and sensitive by WGS. Poor sequencing read-depth and/or contamination with human or nasopharyngeal DNA prevented full WGS resistance prediction in 8 isolates (Amikacin:6; Rifampicin:1; Pyrazinamide:1); and any prediction in 6.
Nearest neighbour comparison with a catalogue of >2000 previously sequenced samples revealed seven UK MTB cases belonged to well-characterised UK outbreaks (Figure 3). This included cases in Leeds and Brighton linked to a well characterised outbreak in Birmingham2 and 4 Oxfordshire cases linked to previous clusters in this low transmission setting3. Isolates were considered to be related if they were 12 or fewer single nucleotide polymorphisms (SNPs) apart2.
Figure 3: Significant (≤12 SNP) relationships between samples and known UK outbreaks; each colour represents a different outbreak. N = number of new members attributed to each outbreak.
N = 1
Pilot study sample(s) linked to an outbreak
Known outbreak
N = 2
Reference laboratory
Resistant
Sensitive
Not Available
WGS R S F
Isoniazid 5 80 2 10 1
Rifampicin 83 6
Ethambutol
Pyrazinamide 4 81
Streptomycin 84 9
Quinolones 93
Amikacin/Capreomycin/Kanamycin 87 12
Figure 1: Current sample flow for MTB diagnosis and potential flow with WGS
*ZN positive = BACTEC MGIT positive 7 – 10 days. ZN negative = BACTEC MGIT positive up to 35 days
Table 1: Resistance found by reference laboratories and WGS in 103 MTB complex isolates. R = resistant; S = sensitive; F = failed. Not available = isolate not tested or results not yet available for this analysis.
Conclusions
This ongoing pilot study has provided the proof of principle that WGS can be used diagnostically on an international scale. All the diagnostic results for clinical and public health action could be simultaneously available in as little as 2 days from completing sequencing. This includes information in one-step that could not be obtained by conventional methods in the same time-scale; such as, species identification, resistance prediction to first- and second-line antibiotics and recognition of outbreaks from genomic matches to previously sequenced isolates.
A limitation to WGS analysis was low read-depth associated with contamination by non-mycobacterial DNA. This limitation could be addressed by increasing the volume of MGIT sample taken for DNA extraction and ensuring effective decontamination of sample prior to MGIT inoculation
Figure 2: Species found. Mismatch = failure of either method or discordant results
References
1. Köser, C. U., et al. (2013) Whole-genome sequencing for rapid susceptibility testing of M. tuberculosis. NEJM 369(3): 290 – 292
2. Walker, T. M., et al. (2013) Whole-genome sequencing to delineate Mycobacterium tuberculosis outbreaks: a retrospective observational study. Lancet Infect Dis 13: 137 – 146
3. Walker, T. M., et al. (2014) Assessment of Mycobacterium tuberculosis transmission in Oxfordshire, UK, , with whole pathogen genome sequences: an observational study. Lancet Respir Med 2:
Thank-you to all members of ǂModernising Medical Microbiology (MMM) Consortium ( and the routine laboratory staff at each participating centre. *COMplete PAthogen Sequencing Solution (COMPASS)-TB participants as follows: PHE-Oxford: A Votintseva , T Walker, C del Ojo Elias, L Pankhurst, D Crook, TEA Peto, AS Walker, D Wyllie, H Barker, M Achayra, C Crichton, J Davies, L Madrid Marquez, M Morgan, S Wordsworth, J Fermont. PHE-Leeds: D Gascoyne-Binzi, R O’Hara, M Wilcox. PHE-Birmingham: T Collins, L Xu, G Smith, J Evans. PHE-Brighton: K Cole, J Paul. Trinity College Dublin/St James’s Hospital: E Roycroft ,T Rogers, M Fitzgibbon. Genoscreen Lille: C Gaudin, M Ramaroson, P Supply, M Mairey, C Allix-Béguec. Lille University Hospital: N Lemaitre. FZ-Borstel: T Kohl, S Niemann. BCCDC Vancouver: J Gardy, P Tang, C Kong, M Rodrigues. PHE: P Monk, J Magee, F Drobniewski, M Zambon, S Gharbia, I Abubakar, L Thomas, L Anderson, J Green.