Genomic Antibiotic Resistance Prediction in Tuberculosis: A Comprehensive Evaluation of Whole Genome Sequencing for Routine Diagnostics in Germany

Phenotypic drug susceptibility testing (DST) remains gold standard for evaluating drug resistance in Germany, despite limitations. Since 2020, the National Reference Center (NRC) for Mycobacteria, has integrated whole genome sequencing of tuberculosis pathogens as part of the Robert Koch Institute (RKI) coordinated IMS (integrated molecular surveillance)-TB project. Accordingly, a comprehensive data set is available to test the potential of genomic resistance prediction.

This study aims to evaluate the efficacy of genome-based antibiotic resistance prediction using the expanded WHO catalog, within a low-burden setting for routine diagnostics.

Between 2020 and 2022, over 3,000 isolates underwent phenotypic DST at the NRC using the proportion method at critical concentration in the MGIT system. As part of the IMS-TB project, each of these isolates was analyzed using next generation sequencing (NGS). Genomic identification of resistance-causing variants from the extended WHO catalogue (2021) predicted resistance against first-line drugs and group A second-line drugs (Levofloxacin, Moxifloxacin, Bedaquiline and Linezolid). The results were compared with phenotypic outcome.

For 3,148 isolates, a complete phenotypic and genotypic first-line resistance profile was produced. Among these 2,279 (72.4%) were fully susceptible against first-line drugs. NGS prediction demonstrated sensitivities above 95 % for each drug except pyrazinamide (90.4%), specificities exceeding 97 % for all four first-line drugs. The negative predictive value exceeded 99% for each of all first-line antibiotics. Group A drug analysis included 341 full datasets including 66 Fluorquinolone, 11 Bedaquiline and 5 Linezolid resistant isolates. Revealing high sensitivity and specificity for the detection of Fluorquinolone resistance (95.5% and 99.6%), while Bedaquiline and Linezolid resistant isolates must be evaluated individually due to limited numbers.

NGS-based genomic DST reliably predicts first-line susceptibility. Standard sequencing of all culture-positive isolates may replace phenotypic first-line testing, given that the technical and quality management requirements can be met.