Introduction

Mycobacterium abscessus is implicated in many clinically important infections including respiratory infections and skin infections. It is the third most common non-tuberculous mycobacteria (NTM) implicated in respiratory infections, only outnumbered by M. kansasii and M. avium.infections [1]. In 2009, Leao et al proposed that M. abscessus isolates identified via 16s rRNA gene sequencing should be differentiated into 3 subspecies, M. abscessus subsp. abscessus, subsp. massiliense and subsp. bolletii using differences in their hsp65 and rpoB gene sequences [2]. After a re-evaluation of that taxonomy, the M. abscessus subsp. massiliense and M.abscessus subsp. boletii subsp. were combined into a single group based on the application of the Bacteriological Code [2]. More recently the combination of these two subspecies is being re-evaluated, with one of the major items of contention being inducible macrolide resistance and therefore potential treatment differences between the three previously proposed subspecies [3–6].

Macrolides are considered a standard tool in the treatment of NTM infections[1]. The erythromycin ribosomal methlylase (erm) gene, which is associated with inducible resistance to macrolide antibiotics, has been identified in several clinically relevant rapidly growing mycobacteria (RGM), including the erm(41) gene in the M. abscessus group [7–9]. In order to detect inducible macrolide resistance phenotypically, the Clinical Laboratory Standards Institute (CLSI) M24 A-2 standard suggests a 14 day incubation time for broth microdilution-based macrolide sensitivity testing, but also states that further studies including large numbers of isolates are required to determine the optimal incubation time for interpretation of minimum inhibitory concentration (MIC) for macrolides [10]. Detection of inducible macrolide resistance can be expedited by sequencing the erm(41) gene that is present in organisms. Truncated erm(41) genes, or T28C mutations in the erm(41) gene render the enzyme non-functional and do not result in inducible macrolide resistance [7,11]. The erm(41) gene sequence can further be separated into sequevars based on 13 single nucleotide polymorphisms (SNPs) throughout the gene. So far, the T28C SNP is the only SNP in a full length erm(41) gene that has been associated with sensitivity. This study aimed to determine the ideal incubation time for the detection of inducible macrolide resistance and establish whether there is a correlation between the length of time it takes to detect inducible macrolide resistance in M.abscessus group organisms and their erm(41) sequevars.

Methods

Results

Twenty one M. chelonae strains, which lack the mechanism for inducible macrolide resistance (Nash, 2009), were inoculated into panels and incubated for 21 days to evaluate the stability of the clarithromycin in the panel at that extended time point. Clarithromycin MIC values were relatively stable over the 21 day incubation period for 20/21 strains (S1 Table). One strain did achieve an MIC of 8 μg/mL at day 10, but due to a rapid jump in MIC over a 3 day period, this is suspected to be due to contamination caused by exposing the panel to the air. Based on these findings, we concluded that the clarithromycin present in the panel was sufficiently active over the entire 21 day incubation period.

Identifications, macrolide sensitivity interpretations and erm(41) sequevars are presented in Table 1. All 20 M. abscessus subsp. massiliense isolates and 2 isolates closely related to M. abscessus (1 bp difference by 16s rRNA gene sequencing) had a truncated erm(41) gene, and were sensitive to clarithromycin. Sixteen (3 M. abscessus subsp. bolletii, 1 close to M. abscessus and 12 M. abscessus subsp. abscessus) isolates had T28C mutations and belonged to Sequevar02, with 87.5% of those showing sensitivity to clarithromycin after the 14 day incubation period. Additionally, within Sequevar02, 2 M. abscessus subsp. abscessus isolates with T28C mutations intermediate interpretations at day 14 and 1 strain closely related to M. abscessus (1 bp away by 16s rRNA gene sequencing) had an intermediate interpretation at day 21. Interestingly, the two strains with intermediate interpretations at day 14, along with two other M. abscessus subsp. abscessus isolates with T28C mutations initially showed inducible macrolide resistance. Repeat testing resulted in two isolates having MICs in the sensitive range on day 14, with the other two having intermediate MICs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Subspeciation, days to detection of inducible macrolide resistance, erm(41) sequence and Sequevar assignment for study strains¹.

https://doi.org/10.1371/journal.pone.0158723.t001

Two of the remaining M. abscessus subsp. bolletii isolates and the 1 remaining strain closely related to M. abscessus (1 bp away by 16 rRNA gene sequencing) showed inducible resistance at day 7. The last remaining M. abscessus supsp. bolletii strain was macrolide resistant upon initial interpretation at day 3. The erm(41) gene sequences for these 4 isolates did not cluster closely with the remaining M. abscessus subsp. abscessus isolates and were assigned provisional sequevar numbers for the purpose of this study only.

The 62 remaining M. abscessus subsp. abscessus isolates were separated into 3 sequevars with multiple strains, and 8 sequevars containing only a single strain. Five isolates were resistant upon initial read, 4 were intermediate upon initial read and resistant at day 7, 46 showed inducible resistance at day 7, 5 were inducible at day 10 and, 2 showed inducible resistance at day 14 (Table 1).

Sequevars 06, 08 and 15 contained the largest number of samples that did not contain the T28C mutation. These three sequevars were analyzed to determine if there was any correlation between the sequevar and the time to detection of inducible resistance using an Fischer’s Exact Test analysis in R. There was no statistically significant association between sequevar and time to detection of inducible resistance (p = 0.06). Further, the low number of samples in sequevar08 appear to drive the p value down giving the false impression of possible significance. There were not enough subsp. bolletii isolates available for this study to analyze if subspecies was correlated with a difference in time to detection of inducible resistance.

Overall, the majority of inducible resistance (87.2%) was detected by day 7, with an additional 9.1% of inducible resistance being detected at day 10. There were two isolates (3.6%) with MIC interpretations that did not reach “resistant” until day 14. Both of these isolates had an MIC of 8 μg/mL at day 14. These isolates belonged to Sequevar06 (predicted to have inducible resistance). Thirty-five isolates had interpretations of sensitive at day 14 of incubation. All of these isolates either belonged to Sequevar02 (containing a T28C mutation) or had truncated erm(41) genes. Of the 14 isolates incubated until day 21, only one had a 21 day read of intermediate, the rest remained sensitive. This isolate had a T28C mutation and was therefore expected to be sensitive.

Discussion

The identification of a functional erm(41) gene in Mycobacterium abscessus isolates has called into question the clinical usefulness of macrolides in the treatment of these infections [7,15,16], although the CLSI suggests inclusion of macrolides in the case of inducible resistance due to a lack of sufficient alternate options, and emperical evidence showing some usefulness. Clinical management of cases is further confused by controversy in the subspeciation of the M. abscessus group of organisms, which were previously split into three subspecies, with two of those subspecies, bolletii and massiliense eventually being combined into subspecies bolletii, and the fact that many labs do not even differentiate the M. abscessus/ chelonae complex.

The CLSI has partially addressed concerns about the detection of inducible resistance by recommending a 14 day incubation period for non-pigmented rapidly growing NTM, but this recommendation was provisional pending the accumulation of further data [10]. Based on our findings, we would suggest that the addition of a day 7 read to the day 14 read of the panels would improve turn around times for the vast majority of M. abscessus isolates exhibiting inducible macrolide resistance, but the addition of a 21 day read provided no added value to the test. There were two strains that converted to resistance at day 14 from Sequevar06, and one strain that had an interpretation of intermediate at day 21(Sequevar02). Based on the trends for Sequevars02 and 06, the isolates from Sequevar02 would be expected to be sensitive and the isolates from Sequevar06 would be expected to exhibit inducible resistance. These results call into question the accuracy of interpretations at (or beyond) the day 14 mark as well as the clinical relevance of this level of “resistance”. Like Brown-Elliot et al, we suggest that isolates with a 14 day MIC of 4 or 8 μg/mL be interpreted with caution and be repeated, or preferably, have gene sequencing completed to aid in interpretation [14].

Previous evaluations of erm(41) gene sequences have suggested that M. abscessus subsp. massiliense isolates invariably have truncated erm(41) genes and this finding is replicated in this study [7,17]. There were also two strains that were mismatched from M. abscessus by 16s rRNA gene sequencing that had the truncated erm(41) gene evaluated in this study. Considering the fact that in mycobacteriology, even single nucleotide polymorphisms in the 16s rRNA gene sequences have been used to differentiate at the species level, it can be hard at times to assign organisms such as these an appropriate identification. Sequencing of the erm(41) gene can therefore be valuable in assisting with proper identification and clinical management of organisms such as these.

The number of erm(41) sequevars found in this study exceeded the numbers expected based on previously reported [14] that found only 7 sequevars in a sample of 85 M. abscessus subsp.abscessus isolates. This study identified 15 sequevars of subsp. abscessus isolates and 3 sequvars comprised of a single subsp. bolletii isolate each. Our study strains fell into 6 of the 10 previously published sequevars and 9 new sequevars. This points to further diversity than previously published within the erm(41) gene. The data presented here shows that, within each of the sequevars, diversity is present both in the presence of immediately detectable macrolide resistance (likely caused by mutations in the rrl gene [8,11]) and inducible macrolide resistance. Further, while the majority of all inducible resistance was detected by day 7 of incubation, there was also variability in this characteristic within individual sequevars. Therefore, clustering via sequevar gave us no additional information as to why some resistance is detected at day 7 and some not until day 14.

As in previous studies, truncated erm(41) genes are associated with M. abscessus subsp. massiliense and are always associated with macrolide sensitivity [7,14,17]. M. abscessus subsp. bolletii isolates can benefit from erm(41) gene sequencing for the prediction of inducible macrolide resistance. Two M. abscessus subsp. abscessus isolates with a T28C mutation showed intermediate MICs at day 14, along with one isolate that had an intermediate clarithromycin MIC at day 21. This could indicate that there is an alternate resistance mechanism at paly, but based on the trends within each of the sequevars represented, along with two strains that showed sensitivity upon repeat testing, it is more likely that bacterial contamination possibly affected the MIC or the efficacy of the clarithromycin in the MIC plates may wane after extended incubation, although this effect was not observed during the validation performed for this study. In any case, extending the incubation past 14 days appears to be of little value and erm(41) gene sequencing has a better potential for accurate detection of inducible macrolide resistance. Incubation times for the identification of macrolide resistance did not correlate with sequevar, therefore time to detection of inducible macrolide resistance is not associated with, and cannot be predicted by, erm(41) gene sequence.

Conclusions

Ultimately, the results of this study lead us to recommend that the addition of a 7 day read of the broth microdiultion panels to the CLSI guidelines would improve turn around times for the majority (>80%) of M. abscessus isolates with inducible macrolide resistance [10]. Further, there is no additional value to adding a 21 day read to the Sensititre RAPMYCO panels. Finally, the routine differentiation of the members of the M. abscessus/M.chelonae complex down to the subspecies level, with the addition of of erm(41) gene sequencing for the detection of T28C mutations, would be an advantageous addition to phenotypic testing for the clinical management of M.abscessus cases. It is important to keep in mind, these results are related only to strains that are sensitive to macrolides after the standard 3-5day incubation period.

Based on the data presented here, presence of a truncated erm(41) gene in a macrolide-sensitive strain, accurately predicted the absence of inducible resistance in all cases, as did the presence of a T28C mutation. Any further characterization of the erm(41) did not appear to be predictive of resistance characteristics in this study. Our results suggest that molecular results would be sufficient to rule-in the use of a macrolide, although based on empirical evidence; macrolides may still be included in therapy in cases demonstrating inducible resistance. Further studies linking phenotypic and genotypic data from isolates demonstrating inducible resistance to patient outcomes will be required to determine if the molecular or phenotypic data correlates more closely with patient outcomes and if clinical recommendations should be changed. Assessment of whole genome sequence data from such cases may lead to insights into alternate causes for the continual “creeping-up” of the clarithromycin MIC in these isolates. Until such a time as that data is available, the use of both methodologies is recommended.

Supporting Information

Acknowledgments

We would like to thank Alisa McGurran, Nancy Smart and Debra Janella from the National Reference Centre for Mycobacteriology for their technical assistance with this project.