Xpert Ultra Can Unambiguously Identify Specific Rifampin Resistance-Conferring Mutations

LETTER

The deluge of data produced by the Xpert MTB/RIF test (Cepheid) can help improve global rifampin-resistant tuberculosis (RR-TB) control strategies through molecular epidemiological surveillance (1, 2). Recently, a new version of the test, Xpert Ultra (hereinafter called Ultra), was released (3). Determining the relationship between RR-conferring rpoB mutations, Ultra probes, and melting temperature shifts (ΔT_m), i.e., the difference between mutant and wild-type melting temperatures, allows Ultra results to be utilized for rapid detection of RR-TB strains and related underlying rpoB mutations.

To determine the reliability of Ultra results for predicting specific mutations, we tested 13 rifampin-susceptible (RS)-TB strains and 104 RR-TB strains harboring 33 unique RR-conferring mutations from the Belgian Coordinated Collections of Microorganisms in the Institute of Tropical Medicine Antwerp according to a protocol previously described (2) (see the supplemental material). Of note, the Glu250Gly (n = 2) and Arg299Cys (n = 1) mutations were among the RS-TB strains. We then compared Ultra raw results with available rpoB sequences of the strains.

Overall, 29/30 (97%) mutations inside the rifampin resistance-determining region (RRDR) were correctly identified by Ultra. Of concern, mutation His445Arg gave a “RIF Resistance Indeterminate” result among 3/4 strains tested, while it was reported as RR in the initial validation study (3). The silent mutation Thr444Thr was not reported as RR (Fig. 1). The RR-conferring mutations on codons 170 and 491 situated outside the RRDR were not detected.

• Open in new tab
• Download powerpoint

FIG 1

Overview of Xpert Ultra test results. The observed probe reactions for each RRDR mutation were laid over the claimed probe coverage (light gray). Shown in black are probe reactions concordant with manufacturer claims, in blue are probe reactions missed by one probe but captured by another probe, and in red is a probe reaction representing a “RIF Resistance Indeterminate” result from 3 out of 4 strains tested. Results in the hatched pattern were superimposed for greater visibility.

The probe reactions observed were largely in agreement with previous results (3), although we noted that mutations Met434Val, Met434Thr, and those in codon 435 were captured only by probe rpoB2, Ser450Leu and Ser450Trp were captured by both probe rpoB3 and probe rpoB4a, His445Arg was captured only by probe rpoB3, and Lys446Gln was captured only by probe rpoB4.

All mutations except those in codon 450 were associated with a negative ΔT_m (Fig. 2). The combination of ΔT_m values with the capturing probes enabled us to differentiate mutations in codons 430, 431, 434, 435, 441, 446, 450, and 452, including disputed mutations (4) (Table 1). Mutation Asp435Tyr was unambiguously distinguished from Asp435Val with the ∣ΔT_m∣ of probe rpoB2, while mutations Ser441Gln and Ser441Leu were discriminated from the rest by the ∣ΔT_m∣ values of probes rpoB2 and rpoB3. Mutations His445Asp and His445Tyr were distinguished from disputed mutations His445Leu and His445Asn through the ∣ΔT_m∣ of probe rpoB3. Ser450Leu was distinguished from Ser450Trp by the ∣ΔT_m∣ of probe rpoB4A. The indeterminate result associated with His445Arg may be caused by its ∣ΔT_m∣ being equal to 1.8°C, unlike the ∣ΔT_m∣ values for other mutations, which typically exceed 2°C. Our recent experience with Ultra on diagnostic sputum samples pertained only to the Ser450Leu and His445Asp mutations, for which the ΔT_m corresponded exactly with the ΔT_m that we observed for bacterial thermolysates. This should be validated more extensively, which is beyond the scope of our present study.

• Open in new tab
• Download powerpoint

FIG 2

Melting temperature shifts (ΔT_ms) observed upon detection of a rifampin resistance (RR)-conferring rpoB mutation in the RR-determining region (RRDR) by Xpert Ultra. The y axis reflects the melting temperature difference (ΔT_m) between mutant and wild-type probe-amplicon hybrids, while the x axis shows the mutations that we tested. The data points on the graph are ΔT_m values grouped by their associated Ultra probes (differentiated by color), which correspond to a specific rpoB mutation. x axis labels in brown are disputed mutations.

Our findings confirm the ability of Ultra to unambiguously identify a wide range of RRDR mutations. With the unprecedented rollout of Xpert MTB/RIF and associated connectivity solutions, such as DataToCare (Savics, Belgium) and GXAlert (SystemOne, USA) (2), Ultra results may allow us to rule out transmission between RR-TB patients in a specific setting (Fig. S1), distinguish relapse from reinfection (5) (Fig. S2), and resolve discordance between an RR Ultra result and a low-level RS phenotypic result due to a disputed mutation. For such applications, it is key that ΔT_m values are included in the exported results.