Comparison of Different Drug Susceptibility Test Methods To Detect Rifampin Heteroresistance in Mycobacterium tuberculosis

Abstract

We compared the efficiencies of different drug susceptibility testing methods in detecting rifampin (RIF) heteroresistance in Mycobacterium tuberculosis. Our data revealed that the broth dilution method found more resistance than MGIT did (P = 0.046) for the low-resistance group. Similarly, the broth dilution method was more sensitive in detecting RIF heteroresistance in subpopulations with low growth rates than was MGIT (P = 0.033). In conclusion, our data demonstrated that the broth dilution method was more sensitive than MGIT in detecting RIF heteroresistance.

TEXT

Drug-resistant tuberculosis (TB) can be acquired through the transmission of already resistant strains, or resistance can happen de novo during treatment due to improper diagnosis and poor compliance of TB patients (1, 2). Under the latter condition, selection of drug-resistant organisms can occur during discontinuous exposure to subtherapeutic drug levels, producing initially heteroresistant and finally fully resistant bacterial populations (2–4). The early diagnosis of heteroresistance is essential to generate effective drug regimens and prevent the emergence of fully resistant populations (5, 6). Unfortunately, heteroresistance in clinical isolates is difficult to detect using conventional and genotypic drug susceptibility testing (DST) methods (7). In this study, our aim was to compare the efficiencies of different drug susceptibility testing methods in detecting rifampin (RIF) heteroresistance in Mycobacterium tuberculosis, including Bactec MGIT DST, the broth dilution method, and Sanger sequencing of the rpoB gene.

Seven RIF-resistant and one RIF-susceptible M. tuberculosis strain, classified in the Beijing family, were obtained from the National Tuberculosis Reference Laboratory of China (8, 9). The drug susceptibility of M. tuberculosis isolates was determined by the conventional proportion method recommended by the World Health Organization (10). Genomic DNA of M. tuberculosis was extracted from freshly cultured bacteria as previously reported (11). A 450-bp region of the rpoB gene containing the 81-bp RIF resistance-determining region (RRDR) was amplified by PCR (12). The primers were synthesized by Invitrogen Company (Shanghai, China). The amplicons were sent to Tsingke Company for DNA sequencing.

The Bactec MGIT 960 system (Becton Dickinson, USA) was used to evaluate the growth rate of RIF-resistant M. tuberculosis isolates. Tubes were incubated at 37°C in the Bactec MGIT 960, and time to detection (TTD) was calculated as the time between the date of culture inoculation and the earliest date that the instrument recorded positive growth. The RIF-resistant bacterial suspensions were serially diluted into a RIF-susceptible bacterial suspension to the following percentages: 0.5%, 1%, 2%, 5%, 10%, 20%, and 50%. Additionally, three separate experiments were performed for each diluted suspension to detect the reproducibility of different DST methods.

To determine MICs of M. tuberculosis strains with different proportions of RIF-resistant subpopulations, a microplate alamarBlue assay (MABA) was performed as described previously (12, 13). Final RIF concentrations were 0.125 to 256 μg/ml. The MIC breakpoint concentration for RIF was defined as 0.5 μg/ml (12, 14). In addition, the RIF susceptibility of the heteroresistant mixture was detected by the Bactec MGIT 960 automated system according to the manufacturer's instructions for DST for RIF (15).

We performed a chi-square test or Fisher exact test to evaluate the efficiency of drug susceptibility test methods in detecting RIF heteroresistance. Statistical analysis was performed in SPSS 11.5 (SPSS Inc., USA). Differences with a P value of 0.05 or less were considered statistically significant.

A total of five mutation types in the RRDR of the rpoB gene were observed among seven RIF-resistant isolates, including 2 isolates with Ser531Leu, 2 with His526Arg, 1 with Ser531Trp, 1 with Asp516Val, and 1 with Leu533Pro. The RIF MICs of strains with mutations at codons 531 and 526 were no lower than 256 μg/ml, while the strains harboring a 516Val or 533Pro mutation had MICs ranging from 16 to 32 μg/ml. The TTD was also analyzed with the MGIT system. As shown in Table 1, five RIF-resistant isolates showed growth rates similar to that of the RIF-susceptible control isolate, the TTDs of which ranged from 169 to 189 h. In addition, we also found that the TTDs of two RIF-resistant isolates were 248 and 264 h, respectively.

TABLE 1.

Mutations, MICs, and growth rates detected in RIF-resistant M. tuberculosis isolates

Strain identifier	Mutation	MIC (μg/ml)	TTDa (h)
TB1	Ser531Leu	256	169
TB2	Ser531Trp	>256	186
TB3	His526Arg	>256	248
TB4	His526Arg	>256	189
TB5	Asp516Val	16	182
TB6	Ser531Leu	256	264
TB7	Leu533Pro	32	179
TB8	NAb	0.25	173

^aTTD, time to detection.

^bNA, no mutation.

We first evaluated whether the mutant type or MIC level of RIF-resistant isolates affected the efficiency of DST methods in detecting RIF heteroresistance. The five RIF-resistant strains with similar growth rates were enrolled in this analysis. As shown in Table 2, the MGIT DST system showed various abilities to detect heteroresistance in isolates containing different mutations. For substitutions 531Leu, 531Trp, and 526Arg, conferring high RIF MICs, MGIT DST was able to find 2.0%, 1.0%, and 0.5% RIF resistance, respectively. For substitutions 516Val and 533Pro conferring low-level RIF resistance, we found that MGIT DST was able to detect RIF resistance when 10% and 5% of the bacteria were resistant, respectively. In comparison with MGIT DST, the broth dilution method had greater efficiency in detecting heteroresistance. For positions 531 and 526, the broth dilution method was able to detect RIF resistance if 0.5% resistant bacteria were present. For substitutions 516Val and 533Pro, the limits of detecting RIF heteroresistance had both been increased to 2% (Fig. 1A). We also analyzed the rpoB sequencing chromatograms of suspensions containing various proportions of RIF-resistant and RIF-susceptible bacteria. As shown in Fig. S1 in the supplemental material, the detection limit of dideoxynucleotide sequencing was 10% in the mutated populations in differential graphs, indicating that this method was less sensitive than MGIT DST and the broth dilution method. In contrast to phenotypic drug susceptibility testing methods, DNA sequencing showed similar abilities to detect heteroresistance in isolates harboring different mutation types (data not shown). When we compared the efficiencies of two phenotypic methods in the different MIC groups (1% heteroresistance was set as the cutoff value), statistical analysis revealed that the broth dilution method (12/14) found more resistance than did MGIT DST (6/14) (P = 0.046) for the low-MIC group (MIC, ≤32 μg/ml), while there was no significant difference observed among the high-MIC group members (MGIT, 20/21, versus broth dilution, 21/21; P = 1.000).

TABLE 2.

Phenotypic drug susceptibilities of mixtures with different proportions of RIF-resistant bacteria detected by Bactec MGIT

Proportion (%) of RIF-resistant bacteria	MGIT result by group and strain (mutation)a
	Fast groupb					Slow groupb
	TB1 (Ser531Leu)	TB2 (Ser531Trp)	TB4 (His526Arg)	TB5 (Asp516Val)	TB7 (Leu533Pro)	TB3 (His526Arg)	TB6 (Ser531Leu)
100	R	R	R	R	R	R	R
50	R	R	R	R	R	R	R
20	R	R	R	R	R	R	R
10	R	R	R	R	R	R	S
5.0	R	R	R	S	R	S	S
2.0	R	R	R	S	S	S	S
1.0	S	R	R	S	S	S	S
0.5	S	S	R	S	S	S	S

^aR, resistant; S, susceptible.

^bFast and slow groups were divided according to growth rate determined by TTD of MGIT.

FIG 1.

MICs of mixtures with different proportions of RIF-resistant bacteria detected by the broth dilution method. (A) MICs of RIF-resistant M. tuberculosis isolates with different types of mutations in the rpoB gene. (B) MICs of RIF-resistant M. tuberculosis isolates with different growth rates.

We also analyzed the effect of growth rate on the ability of phenotypic DST methods to detect heteroresistance. Our data revealed that the heteroresistant bacterial suspension containing RIF-resistant populations with high growth rates showed a higher MIC than did the suspension with a low growth rate. For substitution 531Leu, MGIT DST was able to detect only 20% for the low-growth-rate group, and the broth dilution method was able to find resistance if 2.0% resistant bacteria were present. Similarly, compared with the high-growth-rate group harboring the His526Arg substitution, the detection limit of MGIT DST had been decreased from 1.0% to 20% for the low-growth-rate group, while the broth dilution method was able to detect the recommended 1.0% RIF resistance (Table 2 and Fig. 1B). We also found that the broth dilution method (13/14) was more sensitive in detecting RIF heteroresistance with low-growth-rate RIF-resistant subpopulations than was MGIT DST (7/14) (P = 0.033).

Heteroresistance with the presence of susceptible and resistant subpopulations in M. tuberculosis has been considered a major reason for discordant phenotypic DST results (16). First, RIF heteroresistance with specific ropB mutations conferring low-level RIF resistance is detected by MGIT DST with difficulty. Compared with M. tuberculosis populations harboring isolates with high MICs, the low-level-resistant isolates may reveal lower fitness in the presence of RIF. Consequently, the growth of RIF-resistant populations will be severely inhibited due to the loss of fitness, failing to reach a growth unit (GU) value of 100 by the end of detection. Second, our data reveal that the slow growth is another important factor responsible for the ability of phenotypic DST methods to detect RIF heteroresistance. The low growth rate is associated with loss of fitness, which might explain why MGIT failed to detect RIF heteroresistance in our study. Surprisingly, another growth-based method, the broth dilution method, presents higher efficiency in detecting RIF heteroresistance in M. tuberculosis. One possible explanation is the larger inoculum size for the broth dilution method. Similar to our findings, a recent study by Van Deun found that low-level RIF resistance related to specific mutations was easily missed by the rapid MGIT DST system (17). The different detection limits of RIF heteroresistance with various rpoB mutations and low growth rate highlight the requirement for revision of the MGIT DST methods (16). In addition, the RIF critical concentration used in Bactec methods may be too high, and lowering the RIF breakpoint as suggested by Suo et al. may be another, more accurate variant of the MGIT DST system (16, 18). Considering the complexity of modifying the MGIT procedures, our findings indicate that the microdilution MIC method may be a satisfactory alternative for identifying RIF heteroresistance. Recently, several articles in the literature revealed that the broth dilution method might be more specific for identifying isoniazid (INH) and ethambutol (EMB) resistance than the conventional proportion method (19, 20). Taken together, we recommend that the broth dilution method be regarded as an excellent method able to provide rapid and reliable results of RIF susceptibility testing.

Several works in the literature have demonstrated that the existing molecular methods have difficulty in detecting RIF-heteroresistant M. tuberculosis, especially direct DNA sequencing (3, 11). In contrast to former findings, we found that DNA sequencing could detect only 10% RIF heteroresistance, even by reviewing the original sequencing chromatograms. Although phenotypic DST methods show better efficiency in detecting RIF heteroresistance, an obvious advantage of DNA sequencing, that various rpoB mutations and low growth rate have no influence on detection of heteroresistant populations, should not be ignored. In order to overcome the shortcoming of Sanger sequencing for amplifying low-abundance mutations, numerous novel techniques have been developed to enable preferential amplification of minority alleles from a mixture of wild-type and mutant sequences (11). By combining coamplification at lower denaturation temperature (COLD)-PCR with high-resolution melt technology, the mutation detection limit of RIF heteroresistance in M. tuberculosis has been increased from 20% to 2.5%, which is higher than the 5% rate of the MTBDRplus test found by Folkvardsen et al. (3, 11).

In conclusion, our data demonstrated that the broth dilution method was more sensitive than MGIT DST in detecting RIF heteroresistance in M. tuberculosis. The ability to identify RIF heteroresistance may be determined by both various rpoB mutations and the growth rate of RIF-resistant populations in mixtures. Our findings indicate that the microbroth dilution MIC method may serve as an alternative technique in detecting RIF heteroresistance in M. tuberculosis isolates.