Typing of Neisseria gonorrhoeae Isolates in Shenzhen, China, from 2014 to 2018 Reveals the Shift of Genotypes Significantly Associated with Antimicrobial Resistance

The growing antimicrobial resistance (AMR) in Neisseria gonorrhoeae is a serious global threat to gonococcal therapy. Molecular typing is an ideal tool to reveal the association between specific genotypes and resistance phenotypes that provide effective data for tracking the transmission of resistant clones of N. gonorrhoeae. In our study, we aimed to describe the molecular epidemiology of AMR and the distribution of resistance-associated genotypes in Shenzhen, China, during 2014 to 2018.

ABSTRACT

The growing antimicrobial resistance (AMR) in Neisseria gonorrhoeae is a serious global threat to gonococcal therapy. Molecular typing is an ideal tool to reveal the association between specific genotypes and resistance phenotypes that provide effective data for tracking the transmission of resistant clones of N. gonorrhoeae. In our study, we aimed to describe the molecular epidemiology of AMR and the distribution of resistance-associated genotypes in Shenzhen, China, during 2014 to 2018. In total, 909 isolates were collected from Shenzhen from 2014 to 2018. Two typing schemes, multilocus sequence typing (MLST) and N. gonorrhoeae sequence typing for antimicrobial resistance (NG-STAR), were performed for all isolates. The distribution of resistance-associated genotypes was described using goeBURST analysis combined with logistic regression data. Among 909 isolates, sequence type 8123 (ST₈₁₂₃), ST₇₃₆₃, ST₁₉₀₁, ST₇₃₆₅, and ST₇₃₆₀ were the most common MLST sequence types, and ST₃₄₈, ST₂₄₇₃, ST₄₉₇, and ST₁₉₉ were the most prevalent NG-STAR STs. Logistic regression analysis showed that NG-STAR_ST497, MLST_ST7365, and MLST_ST7360 were typically associated with decreased susceptibility to ceftriaxone. Furthermore, the internationally spreading extended-spectrum cephalosporin (ESC)-resistant clone MLST_ST1901 has been prevalent since at least 2014 in Shenzhen and showed a significant increase during 2014 to 2018. Additionally, MLST_ST7363 owns the potential to become the next internationally spreading ceftriaxone-resistant ST. In conclusion, we performed a comprehensive epidemiological study to explore the correlation between AMR and specific STs, which provided important data for future studies of the molecular epidemiology of AMR in N. gonorrhoeae. Besides, these findings provide insight for adjusting surveillance strategies and therapy management in Shenzhen.

INTRODUCTION

Antimicrobial resistance (AMR) in Neisseria gonorrhoeae has significantly increased in many countries, making it a global health problem (1–3). The World Health Organization (WHO) recommends a dual antimicrobial therapy, with cephalosporins (ceftriaxone 250 mg intramuscularly [i.m.] or cefixime 400 mg orally) and azithromycin, as the first-line treatment for uncomplicated gonorrhea, and the optimal treatment depends on up-to-date local resistance data (4). Recently, however, due to the increased number of azithromycin-resistant strains and the elevated MIC of ceftriaxone, the U.S. Centers for Disease Control and Prevention updated the treatment guidelines, which recommend ceftriaxone monotherapy but increase the ceftriaxone dosage to 500 mg for uncomplicated gonococcal infections, to ensure continued efficacy (5). Unfortunately, the occurrence of treatment failure by cephalosporins (mainly ceftriaxone), the last remaining option for empirical monotherapy in many countries, has recently been widely reported in countries that include Japan (6), Australia (7), China (8), Ireland (9), Canada (10), Denmark (11), and England (12). Hence, it is necessary to monitor the resistant strains and enhance the surveillance of the evolution, transmission, and emergence of AMR in N. gonorrhoeae. Comprehensive molecular epidemiological studies are currently one of the most important aspects of AMR surveillance in N. gonorrhoeae. Different typing schemes have been applied to epidemiological research, including N. gonorrhoeae multiantigen sequence typing (NG-MAST) (13), multilocus sequence typing (MLST) (14), and N. gonorrhoeae sequence typing for antimicrobial resistance (NG-STAR) methods (15), which serve as powerful tools for monitoring the emergence, variation, and further spread of drug resistance. Based on the sequencing of two highly polymorphic alleles (porB and tbpB), NG-MAST is often employed for short-term epidemiological studies and can identify clusters within a specific area in a timely manner for local public health purposes (13, 16).

Compared with NG-MAST, MLST shows a high ability to differentiate between isolates by comparing the genetic variation of seven conserved housekeeping gene fragments (abcZ, adk, fumC, aroE, pdhC, gdh, and pgm), and it is a suitable tool for indexing the relatedness and transmission pattern among global isolates (14). Previous epidemiological studies have shown that MLST sequence type 1901 (MLST_ST1901) has a high prevalence among clones with decreased susceptibility to cephalosporin in many countries, such as South Korea (3), Argentina (17), and Japan (18). In addition, due to the typical ceftriaxone resistance phenotype, MLST_ST1903 has also been reported worldwide (8–10, 19), especially the internationally spreading ceftriaxone-resistant N. gonorrhoeae FC428 clone, which was first identified in Japan (6). The NG-STAR schemes are an ideal method for analyzing the correlation between seven AMR-associated alleles (mtrR, penA, ponA, porB, gyrA, 23S rRNA, and parC) and the resistance phenotype. Of most concern, the resistance mosaic penA-60.001 has shown a trend of international spread since it was initially identified in Japan, and it is responsible for ceftriaxone resistance in many isolates (6, 10, 20, 21). Furthermore, NG-STAR_ST996 was found to have a high level of resistance (MIC ≥ 256 mg/liter) to azithromycin in both England (12) and Australia (7), which poses a threat to recommended therapy.

In a previous study, we conducted a retrospective analysis of antimicrobial susceptibility data (2010 to 2017) and NG-MAST genotyping data (2012 to 2017) in the Shenzhen region of China. The high level of diversity in the distribution of STs also suggested that further typing schemes would be more effective for understanding the evolution and dissemination of antimicrobial-resistant clones in the long term. In addition, the specific AMR distribution in Shenzhen is discussed further, and an obvious increment is observed among isolates with decreased susceptibility to ceftriaxone collected in 2017, which suggests that we should further monitor this trend and track the spread of drug-resistant N. gonorrhoeae clones.

In this work, we aim to describe the distribution, variation, and spread of AMR by performing MLST and NG-STAR on isolates collected between 2014 and 2018, in order to provide comprehensive data to improve surveillance and optimize treatment strategies in the Shenzhen region. As one of the most important ports and with the largest proportion of floating population of China, Shenzhen provides unique geographical conditions for the international spread of AMR, so close monitoring of AMR in the Shenzhen region is a very important part of our national surveillance strategy. To our knowledge, this is the largest-scale epidemiological study in China that uses both MLST and NG-STAR methods, which is of great significance to adjusting the drug resistance monitoring strategy and improving monitoring levels in real time.

RESULTS

The demographic data show that most infected individuals were male (91.97%), and the mean age was 30.60 ± 8.61 years. Additionally, most patients (97.80%) were heterosexually oriented, and 98.68% of patients showed symptoms of abnormal discharge (see Table S3 in the supplemental material). Among 909 isolates, the prevalence of isolates with azithromycin resistance (Azi-R), spectinomycin resistance (Spec-R), and ciprofloxacin resistance (Cipr-R) remained stable in 2018, and all isolates were sensitive to spectinomycin.

In contrast, the proportion of isolates with decreased ceftriaxone susceptibility (Cro-DS) showed a fluctuation across 2014 to 2016 (from 2.58% to 6.04% to 2.22%) and subsequently showed an increase from 2.22% in 2016 to 14.80% in 2018, which indicated that ceftriaxone is less effective in Shenzhen since 2016. It is worth mentioning that the proportion of Cro-DS isolates between 2014 and 2017 was as stable as previously reported (22), but it clearly increased in 2018, which indicates that the distribution of AMR may change in a short time, reminding us of the importance of real-time AMR epidemiology surveillance.

Molecular typing with NG-STAR.

For NG-STAR genotyping, 909 isolates were assigned to 373 STs, of which 278 (n = 427 isolates) were novel STs to the NG-STAR database. Briefly, the predominant type was ST₃₄₈ (n = 62), followed by ST₂₄₇₃ (n = 42), ST₄₉₇ (n = 32), ST₁₉₉ (n = 27), ST₁₄₆₃ (n = 23), ST₃₈ (n = 22), and 248 STs represented by one isolate each. The frequencies of all ST are listed in Table S4 in the supplemental material. From 2014 to 2018, the most prevalent STs were ST₂₄₇₃ (n = 13), ST₃₄₈ (n = 6), ST₃₄₈ (n = 13), ST₃₄₈ (n = 20), and ST₄₉₇ (n = 17), respectively. The distribution of different ST types among different years is shown in Table S5 in the supplemental material. Compared with other predominant STs, ST₂₄₇₃ was a new dominant ST, but it has become the dominant ST since 2014, which means that we may miss some important AMR information due to a lack of monitoring capability. Compared to ST₂₄₇₃, which showed a decrease from 2014 to 2018, both ST₃₄₈ and ST₄₉₇ showed an increasing trend (Fig. 1a).

FIG 1.

(a) Trends of change in the main NG-STAR STs. (b) Trends of change in the main MLST STs.

The result of our phylogenetic analysis shows that 373 STs are divided into two clusters, L1 (n = 140, 47 STs) and L2 (n = 769, 326 STs). The STs (all 373 STs) represented by more than two Azi-R isolates were concentrated on the L2 cluster, and the proportion of Cro-DS isolates of L1 (21/140, 15%) was significantly higher than that in L2 (55/769, 7.15%) (χ² test, P < 0.05). In addition, two of the top three STs (ST₄₉₇ and ST₃₄₈) showed a high rate of Cro-DS, as indicated in the pie chart (Fig. 2), which is consistent with the logistic regression analysis results.

FIG 2.

Circular phylogenetic tree visualized using MEGA 6 of 373 NG-STAR STs. The external color strips range from 1 (inner) to 4 (outer). Strip 1, 373 ST divided into two clusters, as follows: L1, pink; L2, blue. Strip 2, novel STs to NG-STAR database (deep red). Strip 3, number of Cro-DS isolates among each ST, as follows: n = 0, gray; n = 1, red; n = 2, mint green; n ≥ 3, green. Strip 4, number of Azi-R isolates among each ST, as follows: n = 0, green; n = 1, orange; n = 2, purple; n ≥ 3, mint green. The proportion of Cro-DS isolates among the top 3 STs is represented by a pie chart: isolates with decreased susceptibility to ceftriaxone, deep purple; ceftriaxone-susceptible isolates, light purple.

Molecular typing with MLST.

For MLST genotyping, 96 different STs were identified among 909 isolates, of which 20 (n = 29) were new STs to the database, and no new STs showed resistance to ceftriaxone, while three new STs (ST₁₅₂₃₈, ST₁₅₂₅₀, and ST₁₅₂₅₃) showed resistance to azithromycin. The most common ST was ST₈₁₂₃ (n = 89), followed by ST₇₃₆₃ (n = 85), ST₁₉₀₁ (n = 76), ST₇₃₆₅ (n = 56), ST₇₃₆₀ (n = 55), ST₁₆₀₀ (n = 54), ST₇₈₂₂ (n = 54), and ST₇₃₆₇ (n = 52); each of these types was found in more than 50 isolates. The predominant ST types among different years are shown in Table S5.

The proportion of the internationally spreading extended-spectrum cephalosporin (ESC)-resistant clone ST₁₉₀₁ kept increasing from 2014 (4/155; 2.58%) to 2017 (21/166; 12.65%) and decreased slightly in 2018 (23/304; 7.57%), but remained at a high level. Like ST₁₉₀₁, ST₇₃₆₀ also showed an increasing trend during 2014 to 2018. In contrast, ST₇₈₂₇, the high-prevalence clone in China (23), showed a typical decrease (19/155 [12.26%] maximum to 2/166 [1.20%] minimum) from 2014 to 2018 (Fig. 1b). Additionally, the other local clones, including ST₇₃₆₅ (24/304 [7.89%] maximum to 6/135 [4.44%] minimum) and ST₁₆₀₀ (15/155 [9.68%] maximum to 12/304 [3.95%] minimum), remained stable (Table S5).

According to our goeBURST analysis results (see Fig. 3a), all 96 MLST STs were divided into four large groups, i.e., group 0 (no. of STs = 93, no. of isolates = 904), group 1 (STs = 1, isolates = 3), group 2 (STs = 1, isolates = 1), and group 3 (STs = 1, isolates = 1). Most isolates (904/909) were found in group 0. ST₇₈₂₇ had the most single-locus variants (SLVs), 14, among all clone complexes within the same group, and the founding genotype was assigned to ST₇₈₂₇ through the goeBURST algorithm. Results suggest that ST₈₁₂₃, ST₇₃₆₃, ST₇₃₆₅, ST₇₈₂₂, and ST₁₉₀₁ were the founders of subgroups with a high frequency. The common ST₇₈₂₂ showed a close genetic association with the internationally spreading ESC-resistant clone ST₁₉₀₁ and the Cro-DS-associated clone ST₇₃₆₅. Similarly, both ST₈₁₂₃ and ST₇₈₂₇ were SLVs of ST₇₃₆₃, and ST₇₃₆₃ was closely related to increased MIC of ceftriaxone according to our logistic regression data, which means we may face the risk that predominant STs with low resistance rates could act as a reservoir for resistance-associated STs (Fig. 3a). A dominant MLST ST can be found in isolates with each of the six most common NG-STAR STs (n ≥ 20); such low diversity may mean that there is a potential correlation between the two typing schemes (Fig. 3b). This can also explain the results that MLST_ST7363/NG-STAR_ST348, and MLST_ST7360/NG-STAR_ST497 showed consistency in logistic regression analysis.

FIG 3.

(a) Snapshot of group 0. The goeBURST analysis was based on the seven loci used for MLST typing. The occurrence frequency is represented by the font size of MLST STs. The predicted pattern of evolution was represented by three colors of ST node, as follows: group founder, dark green; subgroup founder, light green; common node, light purple. Link colors: black, link drawn without recourse to tiebreak rules; blue, link drawn using tie break rule 1 (number of single-locus variants [SLVs]); green, link drawn using tie break rule 2 (number of double-locus variants [DLVs]). (b) The distribution of MLST STs among the six most prevalent NS-STAR STs (n ≥ 20).

Relationships between molecular types and AMR phenotypes.

The method of logistic regression analysis has been used to evaluate the correlation between MICs and NG-MAST STs in previous research, which provided many important clues for future study (24). The results of our logistic regression analysis indicated that NG-STAR_ST348 and NG-STAR_ST497 were typically associated with increased MICs of ceftriaxone (odds ratio [OR] > 1, P < 0.05). In agreement with a report from the Shanghai region (China), all isolates (n = 4) with NG-STAR_ST202 were Azi-R isolates (38). For MLST, logistic regression analysis was performed to describe the associations between the top 8 STs and MICs (mainly azithromycin and ceftriaxone) (Table 1). The results indicated that ST₇₃₆₅ (56/909; 6.16%) and ST₇₃₆₀ (55/909; 6.05%) were typically associated with decreased susceptibility to ceftriaxone (OR > 1, P < 0.05), and both STs contained a high proportion of Cro-DS isolates (ST₇₃₆₅, 12/56 [21.43%] Cro-DS; ST₇₃₆₀, 10/55 [18.18%] Cro-DS). In addition, the increased MIC of azithromycin showed a close correlation with ST₇₈₂₂ (15/54 [27.78%] Azi-R).

TABLE 1.

Results of logistic regression analysis^a

ST	AZIb				CRO
	MIC		R		MIC		DS
	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P
NG-STAR 348	0.39 (0.31–0.5)	<0.001	0.08 (0.01–0.61)	0.014	1.6 (1.28–1.98)c	<0.001c	1.19 (0.49–2.89)	0.702
NG-STAR 2473	0.98 (0.81–1.19)	0.851	0.69 (0.27–1.79)	0.447	0.8 (0.61–1.04)	0.096	0.27 (0.04–2.03)	0.205
NG-STAR 497	0.57 (0.42–0.77)	<0.001	0.17 (0.02–1.23)	0.079	1.57 (1.19–2.08)c	0.001c	2.57 (1.02–6.5)c	0.046c
MLST 8123	1.02 (0.89–1.19)	0.747	0.99 (0.54–1.84)	0.98	1.04 (0.86–1.25)	0.679	0.63 (0.22–1.85)	0.399
MLST 7363	0.46 (0.38–0.57)	<0.001	0.18 (0.05–0.58)	0.004	1.4 (1.15–1.7)c	0.001c	1.41 (0.61–3.24)	0.423
MLST 1901	0.86 (0.72–1.02)	0.089	0.74 (0.36–1.51)	0.407	0.91 (0.74–1.12)	0.373	1.36 (0.57–3.25)	0.487
MLST 7365	0.86 (0.7–1.06)	0.153	0.18 (0.04–0.75)	0.019	1.38 (1.1–1.72)c	0.005c	3.69 (1.74–7.87)c	0.001c
MLST 7360	0.68 (0.54–0.86)	0.001	0.38 (0.13–1.09)	0.071	1.48 (1.18–1.85)c	0.001c	3 (1.36–6.63)c	0.007c
MLST 1600	1.14 (0.97–1.34)	0.12	0.98 (0.46–2.11)	0.963	0.76 (0.59–0.97)	0.027	0.52 (0.12–2.23)	0.375
MLST 7822	1.22 (1.05–1.42)c	0.011c	1.88 (0.98–3.62)	0.057	0.62 (0.47–0.8)	<0.001	0.78 (0.23–2.67)	0.691

^aTwo resistance phenotypes/MICs, age, and sex were used as predictive variables.

^bR, resistance; DS, decreased susceptibility; OR, odds ratio; CI, confidence interval.

^cOR > 1 and P < 0.05 in both continuous variable and categorical variable models.

Figure 4 shows the MICs of ceftriaxone and azithromycin of isolates with main STs (more than 20 isolates). For ceftriaxone, NG-STAR_ST497, MLST_ST7365, and MLST_ST7360 isolates showed a high proportion of Cro-DS isolates, which was consistent with the results of logistic regression analysis. In addition, isolates with a typical resistance phenotype (MIC = 0.5 mg/liter) were found in the NG-STAR_ST348, MLST_ST7363, and MLST_ST1901 clones, respectively. For azithromycin, isolates with ST₇₈₂₂ showed a higher proportion of Azi-R, while isolates with other STs represented relatively lower MIC levels overall.

FIG 4.

Violin plots of observed MICs for NG-STAR and MLSTs STs (n ≥ 20). MICs are on the y axis, and different STs are on the x axis; Red dashed horizontal lines indicate quartiles; black dashed horizontal lines indicate medians; blue dashed horizontal lines indicate breakpoints of ceftriaxone and azithromycin; widths of violin plots represent the proportion of enrichment of isolates.

Characterization of AMR determinants.

In 909 isolates, 140/909 (15.4%) were observed to harbor a mosaic-type penA, and the most frequent type was penA-10.001 (n = 87), followed by penA-10.003 (n = 25), penA-34.007 (n = 8), and penA-34.001 (n = 7). Overall, the proportion of mosaic penA significantly increased during 2014 to 2018 (from 5.4% to 24.7%), during which the mosaic penA alleles played an important role in the emergence of resistance to ESCs (25–27) (Table S5). The proportion of penA-10.001 to total mosaic penA varied from 27/51 (52.94%) to 7/8 (87.50%) during 2014 to 2018 (Table S5). The penA-34.001 type, which is associated globally with resistance and decreased susceptibility to ESCs, (17, 28–30) has increased in prevalence and produced three different variants (penA-34.007, penA-34.006, and penA-34.018) since 2017 (Table S5). Besides, two new mosaic penA alleles were submitted to the NG-STAR database, including penA-145.001 (n = 1) and penA-124.001 (n = 1), the latter containing the A311V mutation, which shows a high correlation with ceftriaxone resistance (26, 31, 32).

Among 76 Cro-DS (MIC ≥ 0.125 mg/liter) isolates, penA-10.001 and penA-13.001 were the most common types of mosaic penA and nonmosaic penA, respectively. For molecular typing, ST₇₃₆₅ (12/76; 15.79%), which was the most prevalent MLST ST, and NG-STAR_ST348 (6/76; 7.9%) were observed in most isolates. AMR determinant analysis suggests that the most prevalent mutations in penA, mtrR, and porB, all alleles showing high correlation with ceftriaxone resistance, were PenA P551L/S (32/76; 42.11%), a single-nucleotide deletion (A) in the mtrR promoter (51/76; 67.11%), and PorB G120K/A121D (49/76; 64.47%). Furthermore, 10/76 isolates showed additional resistance to azithromycin, and only two isolates harbored the 23S rRNA C2611T mutation, but both two showed typical resistance to azithromycin (MIC = 8 mg/liter) (Table 2). In addition, we found that 17/909 isolates harbored the 23S rRNA A2059G mutation while 22/909 isolates harbored the 23S rRNA C2611T mutation, and 37/39 (94.87%) showed significant resistance to azithromycin (MIC ≥ 4 mg/liter), which means that the mutation located in 23S rRNA shows a high correlation with Azi-R.

TABLE 2.

Details of 76 Cro-DS isolates from Shenzhen in 2014 to 2018

Parameter	Description
No. of isolates	76
No. of MDRa isolates/total	10/76
No. of MP/NMPb	21 MP, 55 NMP
Predominant MP/NMP allele types (n)	MP:10.001 (10), NMP:13.001 (17)
Mutations
penA (n)	P551S/L (32), G545S (21), G542S (13), A311V (1)
MtrR (n)	ΔAd (51), G45D (16), H105Y (38)
PorB (n)	G120K/A121D (49), G120K/A121N (7)
23S (n)	C2611T (2)
Predominant NG-STAR STs (proportion of Cro-DSc)	348 (6/62, 9.68%), 497 (6/32, 18.75%), 1,696 (4/16, 25%), 501 (3/10, 30%), 1,707 (3/7, 42.86%)
Predominant MLST STs (proportion of Cro-DSc)	7,365 (12/56, 21.43%), 7,360 (10/55, 18.18%), 7,363 (8/85, 9.41%), 1,901 (7/76, 9.21%), 8,123 (4/89, 4.49%)

^aMDR, multiple drug resistance.

^bMP, mosaic penA; NMP, nonmosaic penA.

^cProportion of Cro-DS isolates among all isolates that share the same ST.

^dΔA, a single-nucleotide (A) deletion in the mtrR promoter.

DISCUSSION

Over the past decades, multiple platforms were involved in epidemiological research to support the monitoring of AMR (33), including real-time PCR with high-resolution melting analysis (34), whole-genome sequencing (17, 23, 35), and multi-PCR coupled with mass spectrometry (36). All such platforms have been used to investigate the distribution and transmission of gonococcal resistance (17, 30, 37), but the high cost, technical requirements, and lack of uniform criteria limit their wide application in remote areas and large-scale research. In contrast, the NG-STAR and MLST methods provide a high resolution and a uniform standard for revealing the transmission pattern and analyzing the correlation between genotype and phenotype in different regions and at different scales (3, 38). In this work, we conducted a comprehensive epidemiological study using MLST and NG-STAR methods to describe the distribution of AMR and discuss the association between genotype and phenotype.

Of the samples used in this study, 91.97% were collected from male patients. The uneven gender ratio of cases can be attributed to the following reasons. First, in a previous study, we found that among female patients, the rates of untested, false negative, and unreported were higher than those of male patients (39). Another possible reason is that asymptomatic gonorrheal infection is more common in women, and therefore female patients are less likely to seek treatment and have their samples collected (39). Likewise, a gender imbalance caused by similar reasons also existed in a nationwide study, with 91.91% of samples collected from males, which was close to that of our study (91.97%) (1).

Of isolates collected in 2018, 14.8% show Cro-DS, representing a huge threat to the treatment regimen recommended. In fact, we found that the percentage of isolates with MIC = 0.06 mg/liter increased since 2014, which pose a reservoir for the growth of Cro-DS (22). Many factors contribute to the emergence of AMR in N. gonorrhoeae. The abuse of antibiotics, especially the widely inappropriate use of ceftriaxone in the treatment of uncomplicated gonorrhea could play a major role in China (40). In the emergence and spread of ESC strains in Shenzhen, whether domestic selective pressure or importation from other regions plays a prominent role needs further investigation. The prevalence of Azi-R strains has remained at a high level during the past few years. Fortunately, isolates still show sensitivity to spectinomycin. Spectinomycin is also available and is recommended as monotherapy for gonorrhea in China (41). Nevertheless, Spec-R isolates have been reported sporadically worldwide (42–44). Therefore, it is vital to enhance monitoring of the emergence of Spec-R strains and their transmission patterns.

In previous studies in China, the internationally spreading ESC-resistant clone MLST_ST1901 was identified in Chongqing in 2015 (23, 45), Zhejiang in 2016 (45), and Wenzhou in 2017 (46). In our study, we found that the spread of the MLST_ST1901 clone may be affected by geographical factors, and MLST_ST1901 has been prevalent at least since 2014 and has quickly become a dominant clone in the Shenzhen region. To the best of our knowledge, the rare MLST_ST8123, which has not been reported internationally and only sporadically in Wenzhou, China, has a surprisingly high proportion in Shenzhen (46), which may mean that the Shenzhen region has a special distribution of STs because of the special geographical location. According to our MLST data, we found MLST_ST7365 (21.43% Cro-DS) and MLST_ST7360 (18.18% Cro-DS) were significantly associated with decreased susceptibility to ceftriaxone (OR > 1, P < 0.05), especially MLST_ST7365, which has been reported in China to be resistant typically to both ceftriaxone and azithromycin (8). Besides, MLST_ST7363 was observed in many isolates (85/909, 9.35%), and has shown a significant correlation with increased MICs of ceftriaxone (OR > 1, P < 0.05). In addition to China, MLST_ST7363 is also considered a common ST in Ukraine (29), South Korea (3), Japan (47), Vietnam (30), and some European countries (24). Unfortunately, MLST_ST7363 also showed a trend of global transmission and has been reported as a Cro-R-associated ST in Australia (48, 49), Japan (25, 27), and South Korea (3), indicating that it may become the next international ceftriaxone-resistant clone behind MLST_ST1903. Therefore, in addition to those ST types with high proportions (MLST_ST1901 and MLST_ST8123), these special resistance-associated ST types (like MLST_ST7365, and MLST_ST7360) should also be the focus of monitoring.

Due to the identical phenotype (typical resistance to ceftriaxone), NG-STAR_ST233 has been reported previously from different cities in China from 2016 (19, 36, 50), suggesting that this clone has spread to many parts of China. The internationally spreading ceftriaxone-resistant clone NG-STAR_ST233 was represented by only one isolate with a typical phenotype (MIC = 0.5 mg/liter) in our study. Fortunately, this clone has appeared sporadically, and there is no apparent trend indicating that NG-STAR_ST233 will become a predominant clone. In contrast, we found that NG-STAR_ST348 and NG-STAR_ST497 have a high proportion, and both two types showed a high correlation with increased MICs of ceftriaxone (OR > 1, P < 0.05). Additionally, NG-STAR_ST348 was also closely related to reduced susceptibility to cefixime in a previous study in Shanghai, China (38), suggesting that NG-STAR_ST348 may have the potential to become an ESC-resistant clone. Unfortunately, NG-STAR_ST348 has previously been reported as the predominant ST in Shanghai (38), and it has also been reported in the Americas (28), suggesting we need to strengthen the surveillance of this ST to prevent international transmission. Additionally, we must enhance monitoring for the clones with a low frequency but a high proportion of Cro-DS (such as NG-STAR_ST1696, NG-STAR_ST501, and NG-STAR_ST1707), to track emerging resistant clones.

Phylogenetic analysis showed that cluster L1 had a high rate of Cro-DS, and two Cro-DS-associated NG-STAR STs (ST₂₃₃ and ST₃₄₈) belong to L1, which means that different resistance-associated STs may contain similar combinations of seven AMR loci. For the MLST STs, the goeBURST results indicated that the azithromycin-resistant clone MLST_ST7822 (27.78% Azi-R) was the single-locus variant of two common ceftriaxone-resistant clones (MLST_ST1901 and MLST_ST7365) (Fig. 3a), which indicated all three STs had a close genetic association. Two predominant STs, MLST_ST8123 and MLST_ST7827, both showed high similarity with the Cro-DS MLST_ST7363, except for one different base in one of the MLST loci (ST_8123aroE373G → ST_7363aroE373T; ST_7827fumC138G → ST_7363fumC138C), which means we face the risk that locally prevalent clones may acquire new phenotypes through producing variations during expansion. Therefore, in order to prevent those dominant STs from acquiring new resistance phenotypes and becoming reservoirs of resistance, it is necessary to enhance monitoring of changes in AMR in dominant STs. Likewise, all STs with an increasing trend during 2014 to 2018 (Fig. 1a) showed a close relation with increased MICs of ceftriaxone in our logistic regression analysis.

There are some limitations in our study. Compared with asymptomatic patients, patients with symptomatic infection had a higher rate of outpatient visit, which may lead to a potential sampling bias. Another limitation is the uneven gender ratio of cases, which may influence the generalizability of our findings. Therefore, the sampling bias should be improved in our future monitoring strategy, as more weight should be given to female populations and asymptomatic patients.

A rise in the proportion of Cro-DS isolates has driven doctors to prescribe an increased dose of ceftriaxone to maintain the cure rate. Consequently, it is urgent to develop new antimicrobials or find alternatives like spectinomycin, to deal with the dilemma of dwindling antimicrobial options for gonorrhea treatment in Shenzhen. For gonococcal AMR surveillance, MLST and NG-STAR analysis should be applied in company with the routine MIC monitoring. This study demonstrated that molecular typing methods can serve as powerful tools for tracking the transmission of resistance-associated clones (like MLST_ST7363 and NG-STAR_ST348) and monitoring in real time the changes of resistance phenotype in dominant STs. In conclusion, the results from our research not only improve our understanding of the distribution and transmission of AMR in N. gonorrhoeae but also provide effective AMR data for improving surveillance strategies in China.

MATERIALS AND METHODS

Ethics statement.

The study protocol was approved by the Medical Ethics Committee at the Institute of Dermatology, the Chinese Academy of Medical Sciences and Peking Union Medical College and the National Center for Sexually Transmitted Disease Control, Nanjing, China (approval number 2014-LS-026).

N. gonorrhoeae isolates and antimicrobial susceptibility testing.

A total of 909 N. gonorrhoeae isolates collected between 2014 and 2018 across Shenzhen were included. Isolates were stored at −80°C. The isolates from 2014 to 2017 were the same as those in our previous publication, except that information for a few isolates was not available due to the failure of recovery (22). The corresponding MIC profiles and demographic characteristics (including age, sex, symptoms, and history of antibiotic usage) were provided by sentinel surveillance within the China Gonococcal Resistance Surveillance Program. MICs of five antimicrobials (azithromycin, ceftriaxone, penicillin, spectinomycin, and ciprofloxacin) were determined as previously described (16). The criteria of resistance were interpreted by the China Gonococcal Antimicrobial Susceptibility Surveillance Program (GASP) (51). The following threshold values of the five antibiotics were considered: penicillin resistance (Pen-R), MIC ≥ 2 mg/liter; azithromycin resistance (Azi-R), MIC ≥ 1 mg/liter; ciprofloxacin resistance (Cipr-R), MIC ≥ 1 mg/liter; and spectinomycin resistance (Spec-R), MIC ≥ 128 mg/liter. Decreased ceftriaxone susceptibility (Cro-DS) was defined as follows: MIC ≥ 0.125 mg/liter (23, 36).

DNA extraction.

DNA was extracted from bacterial suspensions using the MagNA Pure LC nucleic acid isolation kit (Roche Diagnostics, USA) on a MagNA Pure LC 2.0 instrument (Roche Diagnostics, USA), according to the user’s manual.

Molecular typing.

To genotype the isolates, the fragments of seven MLST housekeeping alleles (abcZ, adk, fumC, aroE, pdhC, gdh, and pgm), seven NG-STAR AMR alleles (mtrR, penA, ponA, porB, gyrA, 23S rRNA, and parC), and two NG-MAST alleles (porB and tbpB) were amplified using Premix Taq (Ex Taq version 2.0 plus dye; TaKaRa, Japan) (primers and PCR conditions can be found in previous reports [13–15], and primers designed in this study are listed in Table S1 in the supplemental material). Both the MLST database (https://pubmlst.org/neisseria/) and the NG-STAR database (https://ngstar.canada.ca) were used to conduct the data analysis. All allele fragments were assembled with SeqMan software (DNASTAR, Inc., USA).

Four AMR alleles (penA, porB, mtrR, and 23S rRNA) of Cro-DS isolates were subjected to a BLAST search with the corresponding wild-type gene sequence using BioEdit software (Ibis Biosciences, Inc., USA) to acquire AMR profiles. Details of wild-type genes are shown in Table S2 in the supplemental material.

Logistic regression analysis.

The associations of the top three NG-STAR STs and the top eight MLST STs with MICs (azithromycin and ceftriaxone) and resistance phenotypes (Azi-R and Cro-DS) were tested using multivariate logistic regression in SAS (version 9.4; SAS Institute, Inc.). Antimicrobial phenotypes/MIC values, age, and sex were used as predictive variables; the NG-STAR STs and the MLST STs were used as the outcome and tested for odds ratio (OR) in logistic regression, respectively. Here, only the NG-STAR STs with more than 30 isolates and the MLST STs with more than 50 isolates were tested for OR. For analyses of MIC values, data were log₂ transformed prior to analysis. A 2-sided P value of <0.05 was considered significant.

Phylogenetic analysis.

Phylogenetic analysis was based on the seven AMR determinants of NG-STAR and a neighbor-joining phylogeny tree with 100 bootstrapped replicates using MEGA 6. (https://www.megasoftware.net/). Modification of the phylogenetic tree was performed on the iTOL website (http://itol.embl.de/).

GoeBURST analysis (http://www.phyloviz.net/goeburst/) was used to infer a hypothetical phylogenetic tree based on seven MLST allelic profiles (52). The formation of clonal complexes is based on the definition that all STs in the same clonal complex should share at least six MLST loci with at least one member from the identical clonal complex.