Evaluation of whole genome sequencing for Mycobacterial species identification and drug susceptibility testing in a clinical setting: a large-scale prospective assessment of performance against line-probe assays and phenotyping.
Quan TP., Bawa Z., Foster D., Walker T., Del Ojo Elias C., Rathod P., MMM Informatics Group None., Iqbal Z., Bradley P., Mowbray J., Walker AS., Crook DW., Wyllie DH., Peto TEA., Smith EG.
Use of whole genome sequencing (WGS) for routine Mycobacterial species identification and drug susceptibility testing (DST) is becoming a reality. We compared performance of WGS and standard laboratory workflows prospectively, by parallel processing at a major Mycobacterial Reference Service over one year, for species identification, first-line Mycobacterium tuberculosis (TB) resistance prediction, and turnaround time. Of 2039 isolates with line-probe results for species identification, 74 (3.6%) failed sequencing or WGS species identification. Excluding these, clinically important species were identified in 1902 isolates, of which 1825 (96.0%) were identified by WGS as the same species. 2157 line-probe test results assaying resistance to the first-line drugs isoniazid and rifampicin were available from 728 TB complex isolates. Excluding 216 (10.0%) cases where there was insufficient sequencing data for WGS to make a prediction, overall concordance was 99.3% (95% CI 98.9-99.6), (sensitivity 97.6% (91.7-99.7), specificity 99.5% (99.0-99.7)). 2982 phenotypic DST results were available from 777 TB complex isolates. Of these, 356 (11.9%) had no WGS comparator due to insufficient sequencing data, and in 154 (5.2%) cases the WGS prediction was indeterminate due to discovery of novel, previously uncharacterized mutations. Excluding these, overall concordance was 99.2% (98.7-99.5), (sensitivity 94.2% (88.4-97.6), specificity 99.4% (99.0-99.7)). Median processing time for the routine laboratory versus WGS was similar overall, at 20 days (IQR 15,31) and 21 days (15, 29) respectively (p=0.41). In conclusion, WGS predicts species and drug susceptibility with great accuracy but work is needed to increase the proportion of predictions made.