Abstract
Background: Between Nov 3, 2014, and Feb 24, 2017, 70 cases of high-level azithromycin-resistant (HL-AziR; minimum inhibitory concentration [MIC] ≥256 mg/L) Neisseria gonorrhoeae were reported from across England. Whole-genome sequencing was done to investigate this outbreak to determine whether the ongoing outbreak represented clonal spread of an HL-AziR N gonorrhoeae strain identified in Leeds. We also wanted to elucidate the molecular mechanisms of azithromycin resistance in N gonorrhoeae in the UK. Methods: In this observational study, whole-genome sequencing was done on the HL-AziR N gonorrhoeae isolates from England. As comparators, 110 isolates from the UK and Ireland with a range of azithromycin MICs were also sequenced, including eight isolates from Scotland with azithromycin MICs ranging from 0·12 mg/L to 1·00 mg/L that were N gonorrhoeae multi-antigen sequence type 9768 (ST9768), which was the sequence type initially responsible for the outbreak. The presence of mutations or genes associated with azithromycin resistance was also investigated. Findings: 37 of the 60 HL-AziR isolates from England belonged to ST9768, and were genetically similar (mean 4·3 single-nucleotide polymorphisms). A 2059A→G mutation was detected in three or all four alleles of the 23S rRNA gene. Five susceptible ST9768 isolates had one mutated 23S rRNA allele and one low-level resistant ST9768 isolate had two mutated alleles. Interpretation: Sustained transmission of a successful HL-AziR clone was seen across England. Mutation 2059A→G was found in isolates with lower azithromycin MICs. Azithromycin exposure might have provided the selection pressure for one or two mutated copies of the 23S rRNA gene to recombine with wild-type copies, leading to three or four mutated copies and the HL-AziR phenotype. HL-AziR could emerge in isolates with low azithromycin MICs and eliminate the effectiveness of azithromycin as part of dual therapy for the treatment of gonorrhoea. Funding: Public Health England.
| Original language | English |
|---|---|
| Pages (from-to) | 573-581 |
| Number of pages | 9 |
| Journal | The Lancet Infectious Diseases |
| Volume | 18 |
| Issue number | 5 |
| DOIs | |
| Publication status | Published - 1 May 2018 |
Bibliographical note
Funding Information:This is the first report of sustained transmission of a clonal outbreak of HL-AziR N gonorrhoeae over several years. Previously the HL-AziR phenotype has been observed sporadically and in small clusters in the UK and elsewhere. 9,10,16,20–23 Our study included the largest number of HL-AziR isolates sequenced to date; we found that they clustered into three phylogenetic clades and were distinct from the majority of the samples with low-level azithromycin resistance. We have also shown that high-level resistance can emerge from susceptible strains or strains with low-level resistance in a short amount of time. The molecular clock of N gonorrhoeae needs to be understood to predict the likelihood of transmission. However, the SNP phylogeny was difficult to interpret due to the little context available with regard to other English samples because N gonorrhoeae is not routinely sequenced as part of Public Health England's surveillance programme. Data from De Silva and colleagues 24 suggest that within a 12 month period the mean number of expected substitutions per genome is four, with an upper 95% confidence limit of 14. Within clade 1, the earliest sample was from April, 2014, and the latest sample was February, 2017—a period of 2 years 10 months. The isolates of ST9768 differed by fewer than 14 SNPs, which was close to the mean of four SNPs per genome per year. Together with our BEAST analysis, this finding supports the conclusion that these isolates shared a common ancestor indicative of very recent transmission. By contrast, the isolates from clade 2 differed from all isolates in clade 1 by at least 14 SNPs. There are isolates within clade 2 with similar isolation dates to those in clade 1 from the early part of the outbreak period. This finding, together with the topology of the phylogeny and BEAST analysis, suggests that clades 1 and clades 2 are unlikely to be part of the same recent transmission chain but shared a common ancestor 6·5 years ago. These lineages might have evolved from a UK-derived common ancestor and have been circulating in the UK population, or might be due to separate introductions of an internationally successful clone that diversified to produce the extant variation observed in clades 1 and 2. The minimum distance between any two isolates in clades 1 and 3 was seven SNPs. Considering that the earliest sample in clade 3 was from April, 2016, and the most ancestral part of clade 1 was from late 2014, we believe that the common ancestor probably diverged into two clonal expansions, one of which manifested as the successful ST9768 clade and the other as clade 3, but that we have not detected cases in the intervening period. The SNP phylogeny gave a level of discrimination beyond that of NG-MAST in terms of determining which isolates were likely to be linked by recent transmission—for example, NG-MAST ST649 was found throughout the phylogeny. A ST649 HL-AziR N gonorrhoeae cluster was identified in the UK in 2007, 9,10 and this sequence type has been associated with HL-AziR N gonorrhoeae internationally. 20–22 The ST9768 clone might be a descendant of one of the lineages of the globally successful HL-AziR ST649. Interestingly, the Scottish ST9768 isolates, which were susceptible or demonstrated low-level resistance to azithromycin, were found within the HL-AziR ST9768 clade. The phylogeny showed that the HL-AziR ST9768 isolates were descendants of the low-level azithromycin-resistant isolates, which were in turn, descendants of the susceptible isolates. However, the maximum likelihood and BEAST phylogenies both suggest that the susceptible ST9768 Scottish isolates originated from an azithromycin-resistant parent but became susceptible through reversion of mutant alleles of the 23S rRNA gene to wild type, either through back mutation or recombination. This change might occur if HL-AziR is associated with a fitness cost. We hypothesise that azithromycin exposure might then have provided selection pressure for the one or two mutated alleles to recombine with wild-type copies without the mutation. Alternatively, azithromycin exposure might have induced additional 2059A→G mutations in the wild-type alleles. Either scenario would lead to three or four mutated copies, which would confer the HL-AziR phenotype, and has been shown to occur in the laboratory. 25 However, without more comprehensive sampling, particularly of isolates from England, we cannot know this for certain. Azithromycin is used in the treatment of other sexually transmitted infections, particularly chlamydia; as such, concurrent undiagnosed N gonorrhoeae could be exposed to subtherapeutic levels of azithromycin during the treatment of these other infections and select for resistance. Additionally, inappropriate azithromycin monotherapy for N gonorrhoeae could lead to treatment failure by providing selection pressure for the development of resistance. We found that most HL-AziR N gonorrhoeae isolates had the 2059A→G mutation in all four copies of the 23S rRNA genes. Low-level azithromycin resistance (MICs 1·0–32·0 mg/L) is commonly associated with a 2611C→T 23S rRNA gene mutation. Whole-genome sequencing of 75 azithromycin-resistant isolates from Europe 19 identified the 2611C→T mutation in two to four alleles of the 23S rRNA gene in isolates with MICs ranging from 4 mg/L to 8 mg/L, and in all four alleles of isolates with MICs of 16–32 mg/L. The 2059A→G mutation was detected in all four alleles of isolates with MICs of 256 mg/L or higher (n=4), but not in isolates with lower MICs. Additionally, mutations in mtrR and its promoter, leading to overexpression of the MtrCDE efflux pump, occurred in isolates across the whole MIC range. A Canadian study, 17 which included five isolates with HL-AziR, had similar findings. In both studies, the resistant isolates clustered clonally into distinct lineages, but the HL-AziR isolates were found across the phylogenetic tree. By contrast, a US study 18 found that isolates with azithromycin resistance were more diverse and showed less clonal expansion than the European and Canadian studies. Only two isolates had HL-AziR, both with four alleles with the 2059A→G mutation. The UK HL-AziR isolates from 2014 onwards were distinct from those seen previously in the UK, from the international HL-AziR N gonorrhoeae sequences available from the studies that we have previously described, 17–19 and also from a cluster of seven cases of HL-AziR N gonorrhoeae reported in Hawaii in 2016. 16 A retrospective prevalence study 26 from Hangzhou, China, in 2012, found HL-AziR in 21 of 118 isolates tested, suggesting that HL-AziR might be widespread in this region. These isolates belonged to seven different NG-MAST sequence types, although some of the sequence types were more than 99% similar; 26 these were different sequence types to those found in our study. HL-AziR might lead to a fitness cost, which could explain why this phenotype has previously been seen only sporadically or in non-sustained clusters. We do not know why sustained transmission of ST9768 has occurred in England; perhaps compensatory mutations to preserve this HL-AziR phenotype are present. Further research to investigate this hypothesis is needed. In our study, cases from both heterosexual and MSM populations were seen across the phylogenetic tree, providing evidence of transmission between MSM and heterosexual networks on several occasions during the course of this relatively short time period. This suggests that there might be greater fluidity between sexual networks than previously supposed, which, in view of the large numbers of partners reported in both populations in this outbreak, has implications for infection control. It also suggests that when resistance emerges in one population it can soon spread to the other. Of particular concern is that 80% of contacts were not traceable, which could mean a substantial burden of unidentified cases given that 77% of contacts who were traced were found to be positive. Fortunately, there were no confirmed treatment failures in any of these cases in this study, probably because the isolates were all susceptible to ceftriaxone. The cluster of seven cases of HL-AziR N gonorrhoeae reported in Hawaii in 2016 16 is concerning because these isolates also showed decreased susceptibility to ceftriaxone (MIC 0·125 mg/L). It is reassuring that all of the patients in Hawaii were successfully treated with 250 mg ceftriaxone plus 1 g azithromycin, and there have been no further cases. Given that the prevalence of low-level azithromycin resistance (MICs >0·5 mg/L) among culture-positive detected cases in England is now 5%, 7 there is concern that HL-AziR N gonorrhoeae might arise more frequently if a proportion of the low-level resistant isolates already harbour an 2059A→G mutation in a single allele. Even low-level azithromycin resistance is likely to render the azithromycin component of dual therapy ineffective. In a Japanese study, 27 treatment failures in men with gonococcal urethritis treated with an extended-release 2 g azithromycin single dose were associated with azithromycin MICs of higher than 0·5 mg/L. Additionally, the in-vitro MIC of azithromycin does not necessarily correlate with clinical treatment outcome, and treatment failures occur in patients with isolates identified as azithromycin susceptible in the laboratory. 28 In conclusion, sustained transmission of a successful HL-AziR N gonorrhoeae clone has been observed across England. Whole-genome sequencing provided characterisation of an outbreak substantially beyond that achieved by NG-MAST. Whole-genome sequencing also progressed our understanding of the emergence and mechanisms of resistance, particularly the finding that high-level resistance can emerge from low-level resistance. Dual therapy for gonorrhoea using azithromycin with ceftriaxone is clearly under threat and we might not be able to rely on azithromycin to protect ceftriaxone. Surveillance of resistance, regular review of treatment guidelines, and detection of treatment failures are crucial towards keeping gonorrhoea as a treatable infection in the future. Contributors HF prepared the manuscript with input from MC, GH, SP, NW, and AU. AU and US prepared the figures. All other authors contributed to the review of the final manuscript. MC managed the laboratory work and assisted with the analysis. NM prepared the cultures for whole-genome sequencing. US, RM, and AU developed and performed the bioinformatic analysis. HF and AU analysed the data. HF and GH coordinated the national outbreak control team. SP, AW, and CS performed data collection, analysis, and interpretation for epidemiological information for the outbreak cases. KT and JS managed the Scottish Reference service and identified ST9768 isolates in their laboratory. Declaration of interests Public Health England's Antimicrobial Resistance and Healthcare Associated Infections Reference Unit has received financial support for conference attendance, lectures, research projects, or contracted evaluations from numerous sources, including Accelerate, Achaogen, Allecra, Amplex, AstraZeneca, Basilea, Becton Dickinson Diagnostics, bioMérieux, Bio-Rad Laboratories, the BSAC, Cepheid, Check-Points, Cubist Pharmaceuticals, Department of Health, European Centre for Disease Prevention and Control, Enigma Diagnostics, Food Standards Agency, GlaxoSmithKline, Henry Stewart Talks, IHMA, Kalidex, Melinta, Merck Sharpe & Dohme, Meiji, Mobidiag, Momentum Biosciences, Nordic, Norgine, Rempex, Roche, Rokitan, Smith & Nephew, Trius, VenatoRx, and Wockhardt. HF is a member of the Scientific Advisory Board for Discuva. All other authors declare no competing interests. Acknowledgments We would like to thank the laboratory staff of Public Health England who helped perform this work; the Leeds Genitourinary Medicine Clinic and Health Protection team for their work in the early stages of the outbreak; and all microbiology laboratories for submitting isolates to the Public Health England national reference service.
