Genotyping of Mycobacterium tuberculosis Rifampin Resistance-Associated Mutations by Use of Data from Xpert MTB/RIF Ultra Enables Large-Scale Tuberculosis Molecular Epidemiology Studies

Molecular epidemiology studies of tuberculosis have been empowered in recent years by the availability of whole-genome sequencing, which has allowed a new focus on the adaptive significance of drug resistance mutations. Genome sequencing technology remains expensive, however, limiting the potential for larger studies.

ABSTRACT

Molecular epidemiology studies of tuberculosis have been empowered in recent years by the availability of whole-genome sequencing, which has allowed a new focus on the adaptive significance of drug resistance mutations. Genome sequencing technology remains expensive, however, limiting the potential for larger studies. Conversely, during this same time the GeneXpert molecular diagnostic method has been deployed globally and now serves as a cornerstone of tuberculosis diagnosis and drug sensitivity testing. In this issue, Y. Cao, H. Parmar, A. M. Simmons, D. Kale, et al. (J Clin Microbiol 57:e00907-19, 2019, https://doi.org/10.1128/JCM.00907-19) report the development of an algorithm that can use high-resolution melting temperature data generated in the course of analysis using the next-generation Xpert MTB/RIF Ultra assay to accurately genotype rifampin resistance-associated mutations. When paired with a system to aggregate data from diagnostic laboratories, this technique has the potential to enable studies on the global scale of the epidemiology of tuberculosis drug resistance.

TEXT

Since the earliest days of the antibiotic era, when effective chemotherapeutics were first introduced, it has been clear that drug resistance would be an obstacle to successful treatment. Few infections present the challenge posed by tuberculosis (TB), in which monotherapy nearly ensures the emergence of resistance, necessitating combination treatment to achieve a cure (1). The frequency of multidrug-resistant TB (MDR-TB), which is resistant to the first-line drugs rifampin and isoniazid, estimated to be 5% globally and increasing, requires drug sensitivity testing (DST) of the infecting Mycobacterium tuberculosis strain to determine an effective combination to treat each patient (2). Though DST has clear utility in tailoring treatment to individual infections and in estimating the frequency of drug-resistant infections in a particular population over time, there are additional aspects of the epidemiology of drug resistance that would benefit from a more detailed view of the responsible mutations. How does drug resistance arise in circulating strains of M. tuberculosis? What are the relative potentials for specific genotypes to be transmitted? What is the overall fitness of resistant strains? The answers to these questions could substantially impact efforts to control and reverse the emergence of MDR-TB.

Genotypic data on rifampin resistance (RR) could contribute to improved public health interventions against MDR-TB, but such data are expensive to collect. In the wealthier world, public health authorities are moving toward the whole-genome sequencing of every tuberculosis isolate (3). In coming years, these growing data sets will be invaluable to epidemiological researchers, but the relative dearth of such efforts in low- and middle-income countries, some of which have the greatest burdens of active transmission and drug resistance, represents a missed opportunity. It is in this context that the study of Cao et al. presents an innovative method of collecting genotypic data on rifampin resistance, deriving these data from those collected in the course of molecular DST with the widely used Xpert MTB/RIF Ultra assay (4).

Accurate, rapid, and affordable Mycobacterium tuberculosis identification and drug sensitivity testing are critical components of efforts to limit the suffering from tuberculosis. In recent years, tremendous progress has been made toward this goal with the introduction of the GeneXpert MTB/RIF nucleic acid-based diagnostic method. This test amplifies and detects DNA sequences specific for M. tuberculosis and the rifampin resistance-determining region (RRDR) of its genome (5). While the test has had a successful global rollout, it has some limitations with regard to limits of detection, identification of heteroresistance, and false-positive resistance calls for paucibacillary samples (6). Hence, Xpert MTB/RIF Ultra, a next-generation test, has been developed to overcome some of these limitations. By switching to longer sloppy molecular beacons for detection, the new test can distinguish genotypes based on melting temperature (T_m) analysis, which allows positive identification of mutants as an alternative to determining absence of detection of an RRDR amplicon at a particular cycle threshold (C_T) in the amplification reaction (7). Beyond the determination of resistance, Cao et al. have hypothesized that by characterizing a library of diverse RRDR genotypes, an algorithm could be created that would determine the RRDR genotype of clinical samples from the T_m data alone.

Cao and colleagues assembled a panel of samples with 41 different RRDR genotypes as a training set and determined the mean T_m within each of 9 T_m windows for 4 probes. After generating this T_m signature library, the same training set was used to validate an algorithm that defined a T_m difference (TD) value based on how much a given signature differs from a specific RRDR genotype T_m signature across each of the 9 T_m windows. The TD values of the two closest matches were used to identify the putative genotype of each sample. In most cases this worked well, except in very similar variants within the same codon, which were not distinguishable. A new set of 33 clinical isolates was then tested, and the algorithm identified all but 3 which had genotypes absent from the reference set. The accuracy of identification was above 90%, with bad calls identified to the codon level but not to the level of the correct variant within it.

The utility of the added dimension provided by T_m data relative to C_T data with respect to information on genotype was recognized early in the development of molecular TB DST based on high-resolution melting curve analysis (8). More recently, a proof-of-concept study validated the approach of using T_m data from the Xpert MTB/RIF Ultra assay to specifically identify 33 distinct rifampin resistance-conferring mutations (9). The work of Cao et al. has now systematized this approach by developing an algorithm that can determine the genotype of a strain within the current breadth of clinically observed rifampin resistance-conferring mutations from T_m data alone with a high degree of accuracy.

Given the potential of this new tool to dramatically lower the cost of studies of the epidemiology of tuberculosis drug resistance, it is imperative that mechanisms are put in place to ease the collection of RRDR genotype data from diagnostic laboratories in regions of intense TB transmission and high prevalence of drug resistance. Fortunately, the value of data acquired by molecular TB diagnostics has been recognized and advances in this field are accruing (10, 11). The development of tools to simplify the collection of RRDR genotype data will require continued support, whether deployed as automated and networked data aggregation systems or, in many cases, as easy-to-use tools to export data to portable storage at sites that lack Internet access. These data aggregation systems will also need the ability to include relevant clinical details, while being sufficiently anonymized to protect patient privacy.

Whole-genome sequencing has revolutionized TB molecular epidemiology studies and has allowed investigators to address new issues concerning fitness costs and transmission potential associated with particular drug resistance mutations (12, 13). Though costs have dropped significantly, whole-genome sequencing remains relatively expensive and is not universally available. By making use of data that are collected at no extra cost in the course of operation of a globally deployed diagnostic, the approach described by Cao et al. will enable a range of investigations that can substantially improve our understanding of the epidemiology of M. tuberculosis drug resistance. Valuable data should no longer be left on the table but instead should be put to use to help reduce the suffering from drug-resistant tuberculosis.

The views expressed in this article do not necessarily reflect the views of the journal or of ASM.