Role of whole genome- and targeted deep- sequencing in antimicrobial resistance detection: An integrative analysis of clinical MDR-TB isolates
Sembulingam Tamilzhalagan, Ashok Selvaraj, Siva Kumar Shanmugam, Chittibabu Suganthi, Sudha Solaiyappan, Suresh Babu Ramalingam, Alangudi Natarajan Palaniappan, Padmapriyadarshini Chandrasekaran, Soumya Swaminathan, Mohan Natarajan, Srikanth Prasad Tripathy, and Uma Devi Ranganathan*
ICMR-National Institute for Research in Tuberculosis, Chennai 600031, India.
*Corresponding author: Uma Devi Ranganathan, Scientist E & Head, Department of Immunology, ICMR-National Institute for Research in Tuberculosis, Chennai 600031, India. Email: firstname.lastname@example.org; Phone: +91 44 28369620.
Background: Drug resistance (DR) detection in tuberculosis has been improved significantly with the advent of next-generation sequencing (NGS) technologies. The accuracy and turn-around time for DR detection has also seen an improvement. However, the nature of specimen and culture generation of clinical isolates play key role in sequencing based DR detection. Sequencing depth is also an important factor to be considered in hetero-resistance detection.
Objective: To identify the DR detection efficacy of NGS from primary and secondary culture derived TB isolates in comparison with phenotypic drug-susceptibility testing (pDST).
Methods: Direct sputum culture and pDST for rapid DR detection was attempted in 7 clinical M. tuberculosis isolates in comparison with secondary LJ culture by genotype and phenotype. The clinical sputum samples derived from four MDR-TB subjects were further processed and cultured either directly on broth (primary culture) or LJ medium for two passages (secondary culture). Genomic DNA was isolated from all isolates and whole genome sequencing (WGS) and targeted deep next-generation sequencing (tNGS) were performed. The pDST was performed for a panel of 14 first- and second-line drugs in MGIT.
Results: Among the 7 isolates, WGS data from 3 isolates were found to be comparable between the passages. The remaining four that had variation in WGS data between the passages were further analysed by tNGS. The tNGS depth of coverage for primary and secondary MGIT culture ranges from 41X to 4869X and 313X to 5063X respectively. Up-to 59 folds reduction in the depth of coverage of primary culture WGS reads was observed after reference sequence alignment, due to the presence of other bacterial contaminants. Thus lead to chances of missed detection of DR mutations due to less mycobacterial genome coverage. Besides, the evolutionarily conserved rrs genetic loci from Streptococcal contaminants was often misdetected as G888A, T1264G and C1483T mutations of mycobacteria in primary culture WGS. The contamination issues were reduced in secondary LJ culture due to mycobacterial selection. However, the rpoB S450L mutation representing less than 10% of the population as detected by tNGS was lost when the cultures were passaged for two generations as LJ culture. WGS pipeline was able to detect inhA I21T and embA C-16T mutations which were not covered by tNGS. Despite pncA Q10P mutation being detected by both the sequencing methods, phenotypic resistance to pyrazinamide was not identified due to contaminants in primary culture, but was detected in secondary LJ culture MGIT DST demonstrates the importance of culture purity in pDST.
Conclusions: Despite the broader application of WGS, extreme caution is required when interpreting genotypic results for DR especially by taking into the account other bacterial species contaminants in the isolated DNA in difficult to treat cases. We also observed that tNGS as a secondary test could provide accurate results, especially when the primary MGIT culture has high percentage of other species contaminants. Though in this study we observed that WGS based DR data obtained from secondary LJ culture was found to be more accurate, we still have to be cautious about rare mutants that could be lost during the passage of strains.