Recent dynamics in Neisseria gonorrhoeae genomic epidemiology in Brazil: antimicrobial resistance and genomic lineages in 2017–20 compared to 2015–16

Abstract

Objectives

Regular quality-assured WGS with antimicrobial resistance (AMR) and epidemiological data of patients is imperative to elucidate the shifting gonorrhoea epidemiology, nationally and internationally. We describe the dynamics of the gonococcal population in 11 cities in Brazil between 2017 and 2020 and elucidate emerging and disappearing gonococcal lineages associated with AMR, compare to Brazilian WGS and AMR data from 2015 to 2016, and explain recent changes in gonococcal AMR and gonorrhoea epidemiology.

Methods

WGS was performed using Illumina NextSeq 550 and genomes of 623 gonococcal isolates were used for downstream analysis. Molecular typing and AMR determinants were obtained and links between genomic lineages and AMR (determined by agar dilution/Etest) examined.

Results

Azithromycin resistance (15.6%, 97/623) had substantially increased and was mainly explained by clonal expansions of strains with 23S rRNA C2611T (mostly NG-STAR CC124) and mtr mosaics (mostly NG-STAR CC63, MLST ST9363). Resistance to ceftriaxone and cefixime remained at the same levels as in 2015–16, i.e. at 0% and 0.2% (1/623), respectively. Regarding novel gonorrhoea treatments, no known zoliflodacin-resistance gyrB mutations or gepotidacin-resistance gyrA mutations were found. Genomic lineages and sublineages showed a phylogenomic shift from sublineage A5 to sublineages A1–A4, while isolates within lineage B remained diverse in Brazil.

Conclusions

Azithromycin resistance, mainly caused by 23S rRNA C2611T and mtrD mosaics/semi-mosaics, had substantially increased in Brazil. This mostly low-level azithromycin resistance may threaten the recommended ceftriaxone-azithromycin therapy, but the lack of ceftriaxone resistance is encouraging. Enhanced gonococcal AMR surveillance, including WGS, is imperative in Brazil and other Latin American and Caribbean countries.

Introduction

Gonorrhoea and antimicrobial resistance (AMR) in Neisseria gonorrhoeae remain significant global public health concerns.^1–3  N. gonorrhoeae has a long history of adaptation and development of AMR to all first-line empiric therapies introduced over the years.^3,4 Currently, the first-line treatment for uncomplicated gonorrhoea is ceftriaxone, an extended-spectrum cephalosporin (ESC), either in a high-dose monotherapy or combined with azithromycin.^5–8 However, ceftriaxone, the last remaining treatment option, is also threatened due to resistance emergence internationally.^3–5,9 For example, the ceftriaxone-resistant strain FC428 (carrying mosaic penA-60)^10–17 has spread internationally since 2015 and some other sporadic mosaic penA-60-carrying strains with ceftriaxone resistance combined with high-level resistance to azithromycin have been confirmed in several countries.^18–20 Furthermore, ceftriaxone-resistant strains carrying mosaic penA-60 has also emerged in the more antimicrobial-susceptible genomic lineage B.²¹ This worrying spread of ceftriaxone resistance in N. gonorrhoeae urges expanded and improved national and international gonococcal antimicrobial surveillance programmes (GASP).⁹

In Brazil, syndromic STI management was introduced in 1993,²² and this created a lack of etiological diagnostics including culture of gonococci. Consequently, comprehensive AMR data from Brazil were very scarce before the initiation of the national Brazilian GASP in 2015.^23,24 The first Brazilian-GASP data from seven sentinel sites in 2015–16 showed a lack of resistance to ceftriaxone and cefixime, but high prevalence of resistance to ciprofloxacin, tetracycline and benzylpenicillin, and 5.1%–6.9% resistance to azithromycin.^23,24 The high levels of ciprofloxacin resistance across Brazil informed prompt revisions of the national treatment guidelines, i.e. in 2017 ciprofloxacin 500 mg was replaced with ceftriaxone (500 mg) in combination with azithromycin (1 g) as the recommended first-line empirical therapy for uncomplicated gonorrhoea.²² Furthermore, WGS of the isolates from 2015–16 was used to describe the national genomic baseline for N. gonorrhoeae in Brazil;²⁴ a high proportion belonged to genomic lineage A.^4,24–28 Lineage A includes most strains with resistance to the currently used antimicrobials such as ESCs and azithromycin and the corresponding AMR determinants: mosaic penA and penA A501 V/T (associated with ESC resistance) and 23S rRNA and mosaic mtr (associated with azithromycin resistance).

Understanding the N. gonorrhoeae genomic epidemiology and basis of AMR, including the acquisition and dissemination of AMR determinants, is crucial for implementing targeted and evidence-based approaches to combat the escalating threat of AMR in N. gonorrhoeae.^4,24–35 The population dynamics and transmission patterns of N. gonorrhoeae are complex, driven by various factors such as sexual networks, high-risk populations and social determinants of health. Genomic epidemiology offers not only an unparalleled opportunity to decipher the emergence and spread of AMR, but also the interconnectedness of bacterial strains, providing valuable insights into the routes of transmission, clustering of infections and potential outbreaks.^29,31,33,34

In the present study, we analysed WGS data in conjunction with AMR data for 623 N. gonorrhoeae isolates obtained at 11 sentinel cities across all the five macroregions of Brazil in 2017–20, compared to WGS and AMR data from 2015 to 2016 (seven sentinel sites),²⁴ to elucidate the circulating gonococcal genomic lineages/sublineages and their AMR determinants, and enhance our understanding of the fluctuations in N. gonorrhoeae epidemiology and AMR in Brazil.

Materials and methods

Neisseria gonorrhoeae isolates

N. gonorrhoeae isolates (633; one per patient/gonorrhoea episode) were cultured from consecutive men with urethral discharge attending 11 sentinel cities across Brazil in 2017–20.³⁶ The men were 18–70 years old, with a median (mean) age of 26 (28) years. The isolates represented all five macroregions in Brazil, i.e. (i) the Northern region: Amazonas (Manaus) n = 97; (ii) the Northeastern region: Bahia (Salvador) n = 10 and Pernambuco (Recife) n = 97; (iii) the Central-Western region: Federal District (Brasília) n = 66; (iv) the Southeastern region: Minas Gerais (Belo Horizonte) n = 98 and São Paulo [São Paulo (n = 15), Ribeirão Preto (n = 95), São José dos Campos (n = 13)] and (v) the Southern region: Santa Catarina (Florianópolis) n = 75, Rio Grande do Sul (Porto Alegre) n = 54 and Paraná (Curitiba) n = 13.

In the present study, isolates were after shipment to the WHO Collaborating Centre for Gonorrhoea and Other STIs, Sweden cultured from frozen stocks (−70°C) on GCAGP agar media for 20–24 h in a humid 5% CO₂-enriched atmosphere at 36 ± 1°C, as previously described.³⁷ If any dubious colony morphology was observed, isolates were verified or falsified as N. gonorrhoeae using MALDI-TOF MS (Bruker Daltonics, Bremen, Germany). Ten (1.6%) isolates, i.e. from Manaus (n = 3), Recife (n = 1), Brasilia (n = 1), Belo Horizonte (n = 1) and Porto Alegre (n = 4), were not viable or verified as N. gonorrhoeae. Accordingly, 623 (98.4%) isolates from 2017 (n = 4), 2018 (n = 85), 2019 (n = 382) and 2020 (n = 152) were available for WGS. One subculturing using identical incubation parameters was performed to ensure the purity of the isolates before DNA extraction.

Antimicrobial susceptibility testing

MICs of eight antimicrobials were determined using agar dilution, in accordance with the CLSI.³⁸ High MICs of ceftriaxone, cefixime and azithromycin were confirmed using Etest, according to the manufacturer’s instructions (bioMérieux, Marcy-l´Etoile, France). Furthermore, isolates with discrepancies between antimicrobial MICs and WGS-derived AMR determinants were also retested using Etest. Where available, EUCAST clinical breakpoints³⁹ were used to determine susceptibility, susceptibility with increased exposure and resistance. For azithromycin, the epidemiological cut-off (ECOFF) of MIC > 1.0 mg/L³⁹ was used to indicate isolates with azithromycin-resistance determinants (referred to as azithromycin resistant hereafter). Resistance to ceftriaxone and cefixime was, in accordance with EUCAST clinical breakpoints,³⁹ defined as MIC > 0.125 mg/L. However, decreased susceptibility (DS) to ceftriaxone and cefixime was also defined, i.e. as MIC = 0.125 mg/L, because isolates with this MIC have previously caused treatment failures.^40–42 The 2016 WHO reference strains⁴³ were used for quality control of the antimicrobial susceptibility testing.

Whole-genome sequencing and analysis

DNA was extracted using the automated Qiasymphony DSP Virus/Pathogen kit (Qiagen, Hilden, Germany). Illumina DNA Prep (Illumina, Inc. San Diego, CA, USA) libraries were quality controlled using Qubit (Thermo Fischer Scientific, Waltham, MA, USA) and Tapestation (Agilent Technologies, Santa Clara, CA, USA), according to the manufacturer’s instructions before library normalization. Libraries were sequenced on an Illumina NextSeq 550 using Mid Output v.2.5 kits (Illumina) to acquire 149 paired-end reads with an average coverage of >100×.

All reads were quality controlled, trimmed (Phred quality score Q30) and tested for contamination using kmer spectra against the 2016 WHO reference strains⁴³ with at least 95% mapped reads using customized CLC Genomics Workbench (v.22.0.1) that includes an assembly pipeline as well as in silico detection of AMR determinants and typing schemes (MLST (https://pubmlst.org/) and N. gonorrhoeae Sequence Typing for Antimicrobial Resistance (NG-STAR: https://ngstar.canada.ca/)). PyngoST (https://github.com/leosanbu/pyngoST) was used to acquire new alleles for submission to PubMLST and NG-STAR databases and to acquire all NG-STAR clonal complexes (CCs)⁴⁴ based on the NG-STAR database (downloaded 5 May 2023). NG-STAR alleles and sequence types (STs) were analysed using goeBURST⁴⁵ to explore the correlations between different NG-STAR CCs and AMR determinants using PHYLOViZ v.2.0⁴⁶ to generate a minimum spanning tree of isolates from Brazil, 2015–20.

A pipeline to generate a multiple sequence alignment (MSA), phylogenetic tree and additional quality control reports was performed using nullarbor (v.2.0.20191013) (https://github.com/tseemann/nullarbor) with WHO F (accession NZ_LT591897.1)⁴³ as reference with the following parameters; –ref NZ_LT591897.1.fasta –assembler spades –assembler-opt ‘–careful’ –taxoner kraken2. The MSA was used for phylogenetic inference using IQ-TREE (v.1.6.12)⁴⁷ based on 17 907 parsimony-informative sites with subsequent removal of recombinant regions using Gubbins.⁴⁸ Simultaneous, phylogenomic analysis as before was performed based on 64 552 parsimony-informative sites using data from Brazil 2015–16 (PRJEB36607),²⁴ Argentina (PRJEB36608, PRJEB41007)^49,50 as well as the 2016 WHO reference strains (PRJEB14020),⁴³ WHO R (DRR124693)¹⁰ and WHO Q (ERR2560139)¹⁸ for comparison. The phylogenetic trees were visualized using Microreact.⁵¹

Isolates from Brazil (n = 623) were additionally compared to 33 306 publicly available gonococcal genomes (downloaded 31 January 2023) in a global phylogenomic tree, as previously described.³² This phylogenomic tree was divided into genomic lineages and sublineages using TreeCluster⁵² with the average clade mode. The method generated 84 distinct clades in the two main lineages A and B, and clades with >1000 isolates were designated sublineages B1–B7 and A1–A5 (Figure S1, available as Supplementary data at JAC Online). Isolates that did not belong to any of these sublineages were designated lineage A or B. Lineage and sublineages for all isolates from Brazil, Argentina and the WHO reference strains (n = 1441) were redesignated to fit this more stable model of the global gonococcal population for genomic molecular epidemiology. The MSA was used in MEGA11⁵³ to divide the isolates into the identified lineages and compute the interpopulation diversity and mean distance between the lineages.

An in-house mtrD database was used to define mosaic and semi-mosaic mtrD variants. The encoded MtrRCDE proteins have regions with high disorder outside the predicted membrane segments suggesting motility. Mosaicism was defined as having recombinations from closely related species in all such regions. We defined semi-mosaicism as mosaic structures in only one of these regions. MtrD semi-mosaics additionally harboured previously described MtrD amino acid alterations associated with increased MICs of azithromycin, i.e. in amino acids R714 and/or K823. All mtrD variants and their nucleotide sequences are available through Microreact (https://microreact.org/project/golparian-et-al-mtrd-mosaic).

All raw sequence reads from the present study are available through the European Nucleotide Archive (project accession number PRJEB62806).

Statistical analyses

Statistical analysis of AMR associations with genomic sublineages was performed using logistic regression analysis and Fisher’s exact test using statsmodels (v.0.14.0).⁵⁴ Statistical significance was defined by P < 0.05.

Results

Antimicrobial susceptibility in Brazil, 2017–20

The level of resistance in 2017–20 to ciprofloxacin, tetracycline, benzylpenicillin, azithromycin and cefixime was 66.3% (11.6% increase since 2015–16²⁴), 40.4% [22.0% decrease; 23.8% had a MIC > 8 mg/L (high-level resistance)], 24.6% (14.5% decrease), 15.6% (10.5% increase) and 0.2% (no changes), respectively (Table S1). The DS to ceftriaxone and cefixime decreased from 0.4% to 0.3% and 6.4% to 4.5%, respectively. No resistance to ceftriaxone or spectinomycin was found, and the highest gentamicin MIC was 16 mg/L.

Molecular epidemiology and antimicrobial resistance determinants

Phylogenomic tree and all metadata are available through Microreact (https://microreact.org/project/6rU79pwTnmd4z9DH3mNtyG-brazil2020). Molecular epidemiology was conducted using genomic epidemiological tools as well as MLST STs and NG-STAR STs and CCs in Brazil.

In total, 93 MLST STs were found among the 623 isolates from Brazil in 2017–20; most common were ST1901 (n = 78), ST1588 (n = 67), ST8143 (n = 48), ST7363 (n = 41) and ST9363 (n = 32). Forty-three MLST STs were represented by a single isolate and 15 were new. Furthermore, the isolates were assigned to 232 NG-STAR STs; most prevalent were ST426 (n = 31), ST1025 (n = 28), ST442 (n = 25), ST90 (n = 19) and ST2148 (n = 17), and 137 STs were represented by a single isolate and 85 were new. The goeBURST analysis of the NG-STAR STs revealed 65 CCs and six ungroupable isolates. The largest NG-STAR CCs were CC309 (n = 69), CC426 (n = 57), CC124 (n = 49), CC63 (n = 40) and CC90 (n = 39). The shifts of MLST and NG-STAR STs and CCs since 2015–16²⁴ in the Brazil sites are summarized in Table 1. Although MLST ST1901 and ST1588 remained the most common MLST STs, their prevalence decreased from 2015–16²⁴ to 2017–20 by 4.7% and 6.0%, respectively. By contrast, some STs, e.g. ST8143, ST7363 and ST7827 increased by 6.4%, 4.6% and 4.3%, respectively (Figure S2). Shifts in NG-STAR CCs were also observed, for example, CC309 and CC90 that were the most prevalent in 2015–16²⁴ had decreased in prevalence by 5.7% and 8.9%, respectively. While CC426, CC124 and CC442 instead had increased by 7.9%, 7.1% and 4.6%, respectively (Figure S3). Furthermore, NG-STAR CC38 and CC1387 had emerged in 2017–20 (Figure 1).

AMR determinants decreasing ESC susceptibility such as mosaic penA, penA encoding A501 substitutions, mtrR and porB1b were found in 9.0%, 10.9%, 35.6% and 30.0%, respectively, of isolates (Table 2). Forty-one penA alleles (seven new) were found, of which six were mosaic. The most common penA alleles were penA-2.001/2.002/2.006/2.008/2.059 (n = 224), penA-5.002/5.016 (n = 88) and penA-19.001 (n = 83). Mosaic penA or penA encoding A501 alteration [penA-34.001 (n = 42), 13.001 (n = 30), 44.001 (n = 30), 34.007 (n = 10), 43.002 (n = 3), 44.004 (n = 3) 103.002 (n = 2), 34.006 (n = 1), 34.028 (n = 1), 219.001 (n = 1) and 223.001 (n = 1)] in combination with mtrR (−35 A-deletion, G45D and/or mtr₁₂₀ (n = 203)) and penB (n = 187) were present in 63 (10.1%) of isolates, of which 16 (25.4%) and 2 (3.2%) had resistance or DS to cefixime and ceftriaxone, respectively. Of these 63 isolates, 22 (34.9%) belonged to MLST ST1901 and 25 (39.7%) to NG-STAR CC90, which is in contrast to 2015–16²⁴ when nearly all isolates with decreased ESC susceptibility belonged to NG-STAR CC90 (94.9%). This is also reflected by the decreased prevalence of mosaic penA alleles from 2015–16²⁴ to 2017–20 (Table 2). The single cefixime-resistant isolate carried mosaic penA-34.001 and belonged to NG-STAR CC90.

No isolates with mutations in 23S rRNA A2058 or A2059 were found. However, 75 (12.0%) isolates had 23S rRNA C2611T mutations (1–4 mutated alleles) and nine (12.0%) of these 75 isolates also had an mtrD mosaic. All these isolates were azithromycin resistant (MIC = 2–64 mg/L), while no azithromycin-susceptible isolates had any 23S rRNA C2611T mutations. The increasing azithromycin resistance due to 23S rRNA C2611T, which was only found in 10 isolates (1.8%) in 2015–16,²⁴ was primarily caused by the expansion of NG-STAR CC124 (Figure 1, Figure S3) across Brazil. The remaining azithromycin-resistant isolates (n = 22) had an mtrD mosaic (n = 20) or semi-mosaic (n = 2). Accordingly, the increase in azithromycin resistance from 5.1% in 2015–16²⁴ to 15.6% in 2017–20 was caused by the increase in prevalence of 23S rRNA C2611T and mtrD mosaics (6.3% and 2.4% increase, respectively) (Table 2). However, 11 and one of the azithromycin-susceptible isolates in 2017–20 also had mtrD mosaic and semi-mosaic, respectively. Notably, 6.6% (n = 41), 11.2% (n = 70) and 10.0% (n = 62) of isolates contained mosaic/semi-mosaic mtrC (26 alleles), mtrD (41 alleles) and/or mtrE (16 alleles), respectively. All 34 isolates with mosaic/semi-mosaic mtrD alleles (https://microreact.org/project/golparian-et-al-mtrd-mosaic) (13 of 41 mtrD alleles) also had mosaic sequence in mtrR, mtrC and/or mtrE with azithromycin MICs ranging from 1–8 mg/L, excluding isolates with 23S rRNA C2611T mutations (n = 9) or mtrC GC hexarepeat deletion (n = 27), which can increase the susceptibility to azithromycin.⁵⁵ Finally, 14 isolates had rplD G70D,⁵⁶ with no alterations in 23S rRNA or mtrD and an azithromycin MIC-range of <0.032–0.5 mg/L.

All ciprofloxacin-resistant isolates carried gyrA S91F (n = 413) and no ciprofloxacin-susceptible isolate had this mutation. The isolates with gyrA S91F (n = 413) also had gyrA D95A (n = 248), D95G (n = 156) and D95N (n = 9). Notably, the prevalence of parC D86N, which predisposes to emergence of resistance to the new antimicrobial gepotidacin,⁵⁷ had dramatically increased from 2.0% in 2015–16²⁴ to 14.0% in 2017–20 (Table 2).

Plasmid-mediated high-level tetracycline resistance (tetM) and chromosomal tetracycline resistance determinant (rpsJ V57M) were found in 179 (28.7%) and 571 (91.7%), respectively, of isolates. All tetracycline non-susceptible isolates (n = 326) had tetM (n = 172) and/or rpsJ V57M (n = 320), beside one isolate with MIC = 1 mg/L. Plasmid-mediated (β-lactamase) high-level penicillin resistance was detected in 142 (22.8%) isolates.

No known mutations causing resistance to zoliflodacin (gyrB D429, K450, S467),³² gepotidacin (gyrA A92P),⁵⁷ gentamicin (fusA)⁵⁸ or spectinomycin (16S rRNA, rpsE)³ was found.

Phylogenomic epidemiology in Brazil

The phylogenomic analysis of isolates from Brazil 2017–20 (n = 623) showed that both genomic lineage A (43.5%) and B (56.5%) of the global gonococcal population (Figure S1) were highly prevalent in Brazil (Figure 2). Most isolates (71.6%) could be designated to one of 12 sublineages (Table S2), i.e. sublineages A1–A5 and B1–B7. Accordingly, 3.7% (10/271) and 44.4% (167/352) of the isolates in lineage A and B, respectively, did not belong to any of the sublineages. The large difference in the proportions of isolates not belonging to any of the sublineages emphasizes the substantially higher diversity within lineage B. This was further confirmed by the mean diversity within the lineages; lineage B [5595 single nucleotide polymorphisms (SNPs)] and lineage A (4144 SNPs).

The most common sublineages within lineage A were A5 (n = 130), A2 (n = 56) and A1 (n = 31), and within linage B the most prevalent ones were B4 (n = 60), B3 (n = 58) and B1 (n = 34). Using logistic regression analysis, azithromycin resistance was associated with sublineages A2, A3 and A5, and Fisher’s exact test confirmed these associations (P = 0.004, P = 0.004 and P = 0.006, respectively). Resistance plus DS to cefixime was associated with sublineage A5 (P < 0.001). Resistance to several of the outdated antimicrobials were associated with many sublineages, i.e. sublineages A3 (P = 0.015), A4 (P = 0.002), A5 (P < 0.001), B1 (P < 0.001), B3 (P < 0.001) and B5 (P=<0.001) with ciprofloxacin resistance; sublineages A3 (P < 0.001), A4 (P < 0.001), A5 (P = 0.007), B1 (P = 0.001), B4 (P < 0.001) and B5 (P = 0.015) with tetracycline resistance and sublineages A5 (P < 0.001), B3 (P < 0.001) and B4 (P < 0.001) with benzylpenicillin resistance. Consequently, sublineage A5 was associated with resistance to all these antimicrobials.

Regarding AMR determinants, 23S rRNA C2611T SNP was associated with sublineages A3 (P = 0.004), A5 (P = 0.006) and B5 (P = 0.043) while mtrD mosaics were associated with sublineage A2 (P < 0.001). Mosaic penA was not associated with sublineage A5 (P = 0.083), despite that the majority (69.0%) of isolates with cefixime DS was within this sublineage. However, penA A501V and A501T mutations were associated with sublineages A4 (P < 0.001) and B1 (P < 0.001), respectively. Sublineages B4 and B5 were associated (P < 0.001) with plasmid-mediated resistance to both tetracycline and benzylpenicillin, sublineage A3 with only tetM (P < 0.001), and sublineage B3 with solely β-lactamase (P < 0.001). Notably, sublineages A4 (n = 29) and B3 (n = 58), which were not present in Brazil in 2015–16,²⁴ with 100% and 77.6% prevalence of parC D86N, respectively, were the main reasons for the dramatically increased prevalence of this mutation (Table 2).

Discussion

The present study provides important insights into the levels of AMR and genomic diversity of N. gonorrhoeae in Brazil, 2015–20. Increasing resistance to azithromycin and persistently high levels of resistance to ciprofloxacin were observed across Brazil, 2015–20 (Table S1). However, resistance to cefixime remained rare (0.2%), ceftriaxone resistance remained lacking and DS to both ESCs had decreased since 2015–16.^23,24 This decreased ESC DS was due to a reduced prevalence of isolates with mosaic penA. However, it is worrying that resistance to azithromycin had increased from 5.1%²⁴ to 15.6% (P < 0.001), which correlates especially with the increased prevalence of isolates with 23S rRNA C2611T, i.e. from 1.8%²⁴ to 12.0% (P < 0.001) (Table 2). This increase was mainly caused by the rapid expansion of one clade in sublineage A5 belonging to NG-STAR CC124 and MLST1901 (https://microreact.org/project/6rU79pwTnmd4z9DH3mNtyG-brazil2020#1j7l-23srrnac2611t). The spread of this clade was primarily in Southeastern Brazil, and this potentially represents the first successful N. gonorrhoeae clonal expansion of a 23S rRNA C2611T strain. These findings are worrying for the first-line empirical treatment in Brazil (ceftriaxone plus azithromycin)²² and NG-STAR CC124 should be carefully monitored. Furthermore, mosaic/semi-mosaic mtrD alleles causing resistance to azithromycin have also increased in prevalence, from 2.7%²⁴ to 5.6%, and included two new mtrD mosaics (variant 77 and variant 79) and three semi-mosaics (variant 67, 69 and 71) (https://microreact.org/project/ghXEPmnhVr7Svhf8VMPfn1-mtrd-mosaicism). As in Argentina,⁵⁰ Europe^33,34 and the USA,⁵⁹ the isolates with mosaic mtrD primarily belonged to NG-STAR CC63 (MLST ST9363). The increasing azithromycin resistance in gonococci in Brazil (and many other countries) is most probably a consequence of the widespread use of azithromycin therapy to treat C. trachomatis and M. genitalium infections, non-gonococcal urethritis, respiratory tract infections and general use of macrolides for other infections.^5,60,61 However, it cannot be excluded that the vast azithromycin overuse during the COVID-19 pandemic in Brazil⁶² has also contributed to the increasing azithromycin resistance in gonococci.

All ciprofloxacin-resistant isolates had gyrA S91F mutations that increased by 11.6% since 2015–16.²⁴ But the largest increase in ciprofloxacin-resistance determinants was for parC D86N, in the two sublineages A4 (NG-STAR CC38) and B3 (NG-STAR CC426) (https://microreact.org/project/6rU79pwTnmd4z9DH3mNtyG-brazil2020#wkos-d86n38426427). Notably, more than half of the isolates were from men who have sex with men in sublineage A4 (51.7%) and 34.9% in sublineage B3. Both NG-STAR CC38 and CC426 had increased (P < 0.001) from 2015–16 (one and seven isolates, respectively)²⁴ to 2017–20 (29 and 57 isolates, respectively) (Figure S3). In addition to being involved in ciprofloxacin resistance, the parC D86N mutation is predisposing for the emergence of resistance to the novel antimicrobial gepotidacin (GlaxoSmithKline, Brentford, UK); together with zoliflodacin, these are the only two antimicrobials in the later stage of clinical development for gonorrhoea treatment.^57,63–65

In the present study, gonococcal genomic lineages and sublineages were for the first time defined in a standardized procedure using >33 000 publicly available gonococcal genomes to acquire a global gonococcal phylogenomic structure, which is more stable and accurate. This was used to redesignate lineages and sublineages from Brazil²⁴ and Argentina.^49,50 Compared to these previous WGS data from Brazil²⁴ and Argentina^49,50 (Figure 3), the main lineages A (45.9%) and B (54.1%) were found in similar proportions (Figure S4). The gonococcal populations in some sites remained unchanged, while sites in Brasilia and Porto Alegre observed an increase in lineage A isolates. Within Brazil, the phylogenomic shifts that occurred from 2015–16²⁴ to 2017–20 within lineage A, with more AMR, might be evolutionarily driven by antibiotic use, i.e. from sublineage A5 to sublineages A1–A4 to a higher degree (Table S2). Thus, sublineage A5 was associated with mosaic penA-carrying isolates and the strains with ESC DS circulating in Brazil²⁴ and Argentina during 2013–16,⁴⁹ which are now being replaced by azithromycin-resistant strains in sublineages, such as A2, associated (P < 0.001) with mosaic mtrD. Lineage B has remained genomically diverse and less affected by antimicrobial use.^4,25 The fluctuations in the gonococcal population in Brazil are also observed using MLST and NG-STAR CC (Figure S2, Figure S3). MLST ST1901 (NG-STAR CC90) frequently with ESC DS has declined, while ESC-susceptible STs such as ST7827 (NG-STAR CC38), ST8143 (NG-STAR CC426), ST8156 (NG-STAR CC442) have expanded, and ST1588 (NG-STAR CC309) in sublineage B4 has remained stable.

One of the limitations of the present study was that no isolates from women or extra-genital sites were examined. The strengths of the study included the relatively high number of isolates examined by both AMR testing and WGS, and representation of all the five macroregions of Brazil, by surveillance in 11 sentinel cities.

In conclusion, the N. gonorrhoeae population in Brazil substantially changed in regard to AMR and genomic structure from 2015–16²⁴ to 2017–20. Azithromycin resistance, mainly caused by 23S rRNA C2611T and mosaic/semi-mosaic mtrD, had substantially increased, however, 56.7% showed low-level azithromycin resistance (MIC = 2–4 mg/L). This mostly low-level azithromycin resistance may threaten the recommended ceftriaxone-azithromycin therapy, but the lack of ceftriaxone resistance is encouraging. Expanded surveillance of gonococcal AMR, including WGS, is imperative in Brazil. Furthermore, enhanced surveillance in other countries in South America, Central America and the Caribbean also remains essential, to gain a more comprehensive understanding of the circulating gonococcal populations, their AMR, factors contributing to this AMR and the dissemination of AMR.

Members of the Brazilian-GASP network

Belo Horizonte: Simone Veloso Faria de Carvalho, Maria Rita Rabelo Costa, Luciane Guimarães Dias, Joana D'arc Pinheiro Feitosa, Mariana Isabella Maciel, Sibele Corrêa Neto; Brasília: Elly Rodrigo Porto, Lidiane da Fonseca Andrade, Glaura Regina de Castro e Caldo Lima, Viviane Furlan Lozano; Florianópolis: Maria Luiza Bazzo, Felipe de Rocco, Fernando Hartmann Barazzetti, Guilherme Kerber, Hanalydia de Melo Machado, Jéssica Motta Martins, Ketlyn Buss, Mara Cristina Scheffer, Marcos André Schörner, Ronaldo Zonta; Porto Alegre: Mauro Cunha Ramos, Maria Rita Castilhos Nicola, Maria Cristina Cecconi; Manaus: Barbara Suely Souza de Noronha; Cleiby Andrade dos Santos; Francinete Motta Lopes; Jairo de Souza Gomes; Jamile Izan Lopes Palhesta Júnior; Paulo Tadeu Cavalcante Saif; Willian Antunes Ferreira; Salvador: Miralba Freire, André Maurício Costa Ramos, Felipe Nogueira M. Carvalho, Aida Politano; São Paulo: Roberto José Carvalho da Silva; Sandra de Araújo; Claudio Campos do Porto; Roberta Alessandra Lima Bocalon; Ursula de Oliveira Machado de Souza; Curitiba: Rafael Mialski, Keite da Silva Nogueira; Natal: Mônica Baumgardt Bay, Manoella do Monte Alves, Erianna Yadja Lucina de Macedo; São José dos Campos: Juliana Cintra Campos, Luíz Fernando Aires Junior, Larissa de Oliveira Camargo; Ribeirão Preto: Lis Aparecida de Souza Neves, Ana Paula Luchetta Paes, Felipe Barufaldi, Henrique Dib Oliveira Reis, Luiz Sérgio D’Oliveira Rocha, Marta Inês Cazentini Ribeiro, Paulo da Silva, Fabiana Rezende Amaral; Recife: François José de Figueiroa, Anesia Maria Siqueira Barbosa, Ana Albertina Araujo, Maria Goretti Varejão, Fernanda Garnier de França Mendes, Valdelucia Oliveira Cavalcanti, Paulo Gabriel Lima Ribeiro, Bruno Ishigami, Lucas Caheté; Laboratório Santa Luzia: Cássia Maria Zoccoli.

Funding

This study was supported by the Örebro County Council Research Committee (2021) and the Foundation for Medical Research at Örebro University Hospital (2020), Örebro, Sweden and the Brazilian Ministry of Health, through its Secretariat for Health Surveillance and its Department of Chronic Conditions and Sexually Transmitted Infection (2017-2020).

Transparency declarations

None to declare.

Supplementary data

Figures S1 to S4 and Tables S1 and S2 are available as Supplementary data at JAC Online.

References

Author notes