Increasing Azithromycin Resistance in Neisseria gonorrhoeae Due to NG-MAST 12302 Clonal Spread in Canada, 2015 to 2018

ABSTRACT Azithromycin-resistant (AZIR) gonorrhea has been steadily increasing in Canada over the past decade, which is cause for alarm, as azithromycin (AZI) has been part of the combination therapy recommended by the Canadian Guidelines on Sexually Transmitted Infections (CGSTI) since 2012. Neisseria gonorrhoeae with AZI MICs ≥1 mg/L collected between 2015 and 2018 as part of the Gonococcal Antimicrobial Surveillance Program-Canada underwent antimicrobial susceptibility testing, molecular typing, and whole-genome sequencing. Regional, demographic, and clinical isolation site comparisons were made to aid in our understanding of AZI susceptibility trending. We identified 3,447 N. gonorrhoeae with AZI MICs of ≥1 mg/L in Canada, which increased from 6.3% in 2015 to 26.5% of isolates in 2018. Central Canada had the highest proportion, rising from 9.2% in 2015 to 31.2% in 2018. We identified 273 different N. gonorrhoeae multiantigen sequence types (NG-MAST) among these isolates, with ST-12302 the most prevalent (50.9%). Whole-genome sequencing identified the Neisseria lactamica-like mosaic mtr locus as the mechanism of AZIR in isolates of ST-12302 and isolates genetically similar (differing by ≤5 bp), designated the ST-12302 genogroup, accounting for 65.2% of study isolates which were originally identified in central Canada but spread to other regions by 2018. Genomic analysis indicated that AZIR in Canadian N. gonorrhoeae expanded rapidly due to clonal spread of the ST-12302 genogroup. The rapid expansion of this AZIR clonal group in all regions of Canada is of concern. CGSTI are currently under review to address the increase in AZIR in Canada.

the 2012 update of Canadian Guidelines on Sexually Transmitted Infections to recommend combination therapy with 2 antibiotics, ceftriaxone or cefixime and azithromycin (AZI) as first-line treatment for high-risk cases (https://www.canada.ca/en/public-health/services/ infectious-diseases/sexual-health-sexually-transmitted-infections/canadian-guidelines.html).
The Canadian Gonococcal Antimicrobial Surveillance Program (GASP-Canada), initiated in the mid-1980s, is a culture-based laboratory surveillance system monitoring AMR trends and N. gonorrhoeae multiantigen sequence types (NG-MAST). Over the last few years and since the introduction of combination therapy for the treatment of gonorrhea in Canada, laboratory surveillance has reported that decreased susceptibility to the ESCs has been declining in Canada, but AZI resistance (AZIR; MIC $ 2 mg/L) has been increasing, with 0.9% (26/3036) in 2012 to 7.6% (427/5,607) in 2018 (1). Increases in AZIR have been documented in other countries as well, including the United States (6), United Kingdom (7)(8)(9), Australia (10), and Japan (11). Treatment failures have also been documented in Canada (12), the United Kingdom (13,14), and Australia (14). AZIR rates have now exceeded WHO thresholds of 5% in some regions of the world, warranting changes to the prescribed gonorrhea treatments (3).
There is some variability in the AZIR breakpoints being used globally. In 2019, the Clinical and Laboratory Standards Institute (CLSI) defined AZI susceptibility for N. gonorrhoeae as #1 mg/L but did not establish an MIC indicating resistance (15). The European Committee on Antimicrobial Susceptibility Testing (EUCAST [https://www .eucast.org/clinical_breakpoints]) reported the epidemiological cutoff value for azithromycin to be 1 mg/L (16). Although GASP-Canada uses an MIC of $2 mg/L as the AZIR breakpoint, as agar dilution MICs are accepted within 61-log dilution, in this study, we included all isolates with MICs of $1 mg/L, and for convenience, these isolates are referred to as AZIR.
Through GASP-Canada, we collected AZIR N. gonorrhoeae isolated in Canada between 2015 and 2018. Phenotypic, demographic information, and whole-genome sequencing (WGS) data were analyzed to aid in our understanding of AZIR trends and the molecular epidemiology of these isolates.

RESULTS
From 2015 to 2018, the number of N. gonorrhoeae isolates identified with azithromycin MICs of $1 mg/L (AZIR) increased each year from 6.3% in 2015 (262/4,190) Table 2). Whole-genome sequencing analysis. The Illumina MiSeq platform (Illumina, San Diego, CA) generated paired-end, 300-bp indexed reads with an average genome coverage of 86 times and an average of 1,104,350 reads per genome. The average contig length generated was 23,043 bp, and the average N 50 contig length was 62,832 bp.
Phylogenetic analysis. Phylogenetic analysis (Fig. 3) 19.0% (4/21) rectal isolates. The mtrR promoter 235A D or 23S rRNA C2611T mutations caused AZIR in this clade. One cluster of concern within clade D consists of NG-MAST 5308 isolates (n = 6) that were identified in 2018 in central and western Canada. They were AZIR and had decreased susceptibility to ESCs. Their AZIR was due to three 23S rRNA alleles with the C2611T mutation. Five out of the 6 isolates (83.3%) were from females, and four (66.7%) were from pharyngeal infections (Fig. 5).

DISCUSSION
In Canada, the increase of AZIR N. gonorrhoeae between 2015 and 2018 is primarily due to the multidrug-resistant NG-MAST 12302 clone and STs that are in its genogroup, totaling 2,250 or 65.2% (2,250/3,447) of all study isolates. The mosaic structure of the MtrCDE efflux pump is the mechanism of the low-level AZIR in these isolates (24,25). This is in contrast to previous years when the 23S rRNA C2611T mutation was the primary source of AZIR (20).
Over 80% of ST-12302 isolates were from males, with approximately 20% identified from rectal infections suggesting a gay, bisexual, and other population of men who have sex with men. These proportions are similar for all of our study isolates with the mosaic mtr locus. Isolates without the mosaic locus had slightly lower proportions (approximately 65% males with 16% rectal infections). The development of AZIR in rectal infections is possibly due to the high bioavailability of AZI in rectal tissues (29,31).
In   The pervasiveness of this clone may be due to increased biological fitness caused by the same mechanism that causes AZIR. The overexpression of the mutated MtrCDE efflux pump causes enhanced survival of N. gonorrhoeae in the presence of human neutrophils (32). In contrast, the 23S rRNA mutations appear sporadically and may have a fitness cost (26).
Of interest, ST-12302 is a result of 2 existing alleles (porB-908 and tbpB-267) combining as opposed to a newly mutated porB or tbpB allele. ST-12302 has the same porB as the internationally identified clone ST-1407 known for high-level resistance to ESCs (33), which was prevalent in Canada from 2010 to 2012 (34). ST-1407 isolates were primarily susceptible to AZI and did not have the mosaic mtr locus. ST-14994 differs from ST-12302g in that it has the A39T mtrR mutation instead of the mosaic D79N/S183N/M197I, has a partial-mosaic mtrE allele, and has a PorB1a protein structure ( Table 3). Isolates of this group tend to have a lower average AZI MIC, most likely due to the absence of the mosaic mtrR mutations (24) and possibly due to the lack of porB mutations that have an additive effect on AZIR (21). The PorB1a protein structure increases the isolate's resistance to complement-dependent killing in normal human serum (35) and is often identified in disseminated gonococcal infections (DGIs). Five DGIs were associated with ST-14994 study isolates in 2018 and are cause for concern, as DGIs are on the rise in Canada. In 2015, 0.3% (4/4,190) of isolates submitted to the NML were DGIs; in 2019, it increased to 1.4% (70/4,859) (National Microbiology Laboratory; I. Martin, P. Sawatzky, and G. Liu, unpublished data). The United States has also identified DGIs in 2019 and 2020 with the same PorB1a structure and MLST as Canadian ST-14994, though NG-MAST was not cited (36).
ST-16288, identified in 2017 and 2018, also has a slightly lower average AZI MIC (Table 3) than ST12302g isolates, which may be due to the absence of a porb1b G120 mutation that is present in ST-12302g isolates. ST-16288 isolates were primarily identified in western Canada (92.5%, 86/93), with only 7.5% (7/93) found in central Canada. There is no published data that NG-MAST 16288 has been identified outside Canada.
AZIR N. gonorrhoeae associated with mosaic mtr alleles has been documented in the United States (26,27), Australia (28), and Germany (29). Germany also identified NG-MAST 12302g as the predominant clone with the mosaic mtr locus from 2016 to 2018 (29). Year, region, sex, age, isolation site, AMR, AZI MICs, STs, and mutations related to AZIR are indicated. Clade B is represented by NG-MAST 14994, Clade C by NG-MAST 10451g, and Clade D is fairly diverse. Clade D-1 is represented by ST-5308 and is AZIR and has decreased susceptibility to extended-spectrum cephalosporins (ESCDS). mosaic mtr alleles, suggesting these mutations were acquired after this clone emerged in Canada. The first Canadian ST-14994 N. gonorrhoeae, identified in May 2017, had AZI MICs of 0.125 to 0.25 mg/L. In September 2017, ST-14994 isolates with AZI MICs of 1 mg/L and mosaic mtrR promoter and mtrD alleles emerged. While the more resistant version of ST-14994 is more prevalent, the other strain is still identified in Canada.
High-level AZIR is also a concern both internationally (38-45) and here in Canada (1). The 5 high-level AZIR isolates identified in Canada during our study period were isolated from 2016 to 2018 in western and central regions. Four were isolated from the pharynx, where AZI is less effective and commensal Neisseria transfer of resistance determinants and reduced tissue penetration of AZI can lead to increased resistance (46). High-level AZIR is primarily due to the 23S rRNA A2059G mutation; however, a new mutation, A2058G, which confers similar resistance, was identified in the United States between 2016 and 2019 (22). This mutation was not identified in our study isolates.
In the last 20 years, we have seen a decline in the proportion of cultures received at the NML due to the shift to nucleic acid amplification testing (NAAT) for the diagnosis of gonorrhea in lieu of culturing. This is especially an issue for the smaller provinces and the remote regions of Canada, as culturing and shipping N. gonorrhoeae is difficult. Provinces/territories that culture N. gonorrhoeae follow regional protocols for doing so that are often dictated by risk factors of the patient and site of infection. Most of the larger centers that do their own AST primarily send cultures that are nonsusceptible to the NML for testing. As cultures of N. gonorrhoeae are required for AST and WGS, all regions of Canada, as well as populations that are not high risk for sexually transmitted infections, may not be represented equally. This problem can also cause the proportions of AZIR to be overrepresented in some regions (for example, the eastern region) or not identified in some regions (such as the northern region).
Another limitation of this study is our lack of resources to perform WGS on all cultures received at the NML, and while we endeavored to choose a representative number of isolates from the different regions and with various NG-MASTs, we focused on resistant isolates and, therefore, could not compare them to a representative number of susceptible isolates. Also, we looked for specific mutations that have been identified to affect AZI susceptibility and may have missed mutations that have not been previously documented.
The clonal dissemination of ST-12302 occurred across Canada and was also identified internationally. It is possible that this clone emerged due to sublethal azithromycin concentrations triggering an N. gonorrhoeae isolate which already has the widespread porB-908 allele to mutate to an ST-12302 clone in the presence of the commensal Neisseria lactamica. While ST-12302 has the lactamica-like mosaic mtr locus, a meningitidis-like mosaic mtr locus has also been identified as a contributor to AZIR (25), although not in Canada. The increased biological fitness of this clone and others with the mosaic mtr locus enhances their ability to survive and spread through sexual networks. Monitoring of antimicrobial susceptibilities and sequence types of N. gonorrhoeae and using WGS to provide molecular antibiotic resistance, virulence, and fitness determinants can help us understand the higher transmission rates of certain lineages and track their spread. Combining phenotypic and molecular testing will help us monitor emerging mutations and their effects on AMR and transmission. Identification of specific mutations associated with antimicrobial resistance is necessary to support the development of NAAT assays for AMR prediction and point-of-care tests. Data generated from this surveillance are used to update the CGSTI with the most effective treatment given current antimicrobial resistance and, consequently, reduce the spread of drug-resistant gonorrhea. These guidelines are currently under review to address the increase in AZIR in Canada.

MATERIALS AND METHODS
Antimicrobial susceptibility testing and molecular testing. Between 2015 and 2018, 19,625 N. gonorrhoeae isolates were tested nationally, either in provincial laboratories or at the National Microbiology Laboratory (NML). The NML has antimicrobial susceptibility data for 14,816 of these isolates, NG-MAST sequence types for 11,993 isolates, and WGS data for 2,763 isolates (Table 1).
MICs were determined using agar dilution following CLSI methodology (47) or Etest (48). Isolates with an AZI MIC of $1 mg/L were included in this study (n = 3,447). Molecular genotyping using the NG-MAST method was successfully performed on 98.0% (3,377/3,447) of them as previously described (49). ST genogroups were defined as STs that differed by less than 5 bp. WGS was performed as previously described (20) on 1,544 of these isolates (2015, n = 56; 2016, n = 105; 2017, n = 682; 2018, n = 701), selected to include a convenience sample of geographical and temporal distribution of isolates as well as a range of NG-MAST and MLST sequence types (STs) ( Table 1). A subset of these isolates can be found in BioProject under accession no. PRJNA785548. ST-12302 isolates that were identical according to the WGS data analyzed (mutations and molecular typing) were represented by one isolate in this subset to reduce the number of isolates uploaded to 715.
Study isolates were grouped into regions based on the province they were identified in: the western region included isolates from British Columbia, Alberta, Saskatchewan, and Manitoba; central region isolates were from Ontario and Quebec; and eastern region isolates were from Nova Scotia and New Brunswick.
Whole-genome sequencing and assembly. The WGS analyses were conducted on 1,544 AZIR N. gonorrhoeae isolates collected from provinces in Canada at the NML as previously described. DNA samples were prepared using Epicentre MasterPure Complete DNA and RNA extraction kit (Mandel Scientific, Guelph, Ontario, Canada), and libraries were created with Nextera sample preparation kits (Illumina, San Diego, CA) with 300-bp paired-end indexed reads generated on the Illumina NextSeq platform (Illumina, San Diego, CA). The quality of the reads was assessed using FastQC version 0.11.4 (https://www.bioinformatics .babraham.ac.uk/projects/fastqc/) and assembled using Shovill (Galaxy Version 1.0.41galaxy) (50). Core single nucleotide variant (SNV) phylogenetic analysis was conducted using a custom Galaxy SNVphyl (50) phylogenomics workflow (Galaxy version SNVPhyl v1.0.1b Paired-end) using NCCP11945 (GenBank accession no. NC_011035) as a mapping reference with thresholds of minimum coverage of 7, minimum mean mapping quality of 30, and alternative allele ratio of 0.75 and removing highly recombinant regions containing .5 SNVs per 500 bp. Phylogenetic trees were visualized using FigTree v1.4.3 (http://tree.bio.ed.ac.uk/software/figtree/). Phylogenetic clades were determined visually and correlated by cluster analysis using ClusterPicker (51) with the following settings: initial and main support thresholds, 0.9; genetic distance threshold, 0.045; and the large cluster threshold, 10.
Pathogenwatch (pathogen.watch) (30) was used to produce a core distance-based neighbor-joining tree of 911 Canadian NG-MAST 12302 isolates with WGS data. Pathogenwatch was also used to confirm AMR determinant genes and mutations identified (https://cgps.gitbook.io/pathogenwatch). Clades were determined visually and confirmed with the ClusterPicker program (51) using a main support threshold of 0.9, a genetic distance threshold of 4.5, and large cluster threshold of 10.
Statistical analysis. Absolute and relative frequencies were calculated for categorical variables. Fisher's exact test was applied to proportion comparisons of AZIR between years with a 95% confidence interval using EpiCalc 2000 (version 1.02; Brixton Health).
Data availability. A subset of the isolates used in this study was deposited in BioProject under accession no. PRJNA785548.