Distinct Neisseria gonorrhoeae Transmission Networks Among Men Who Have Sex With Men in Amsterdam, the Netherlands

Abstract

Background. Molecular typing was used to elucidate Neisseria gonorrhoeae transmission networks among men who have sex with men (MSM) in Amsterdam, the Netherlands. We determined whether clusters of patients infected with specific N. gonorrhoeae genotypes were related to various epidemiological characteristics.

Methods. MSM (age ≥18 years) visiting the sexually transmitted infections (STI) clinic between July 2008 and August 2009 were eligible. After STI screening, participants completed a behavioral questionnaire concerning the previous 6 months. N. gonorrhoeae cultures were genotyped using multiple-locus variable-number tandem repeat analysis typing.

Results. We obtained 278 N. gonorrhoeae–positive isolates from 240 MSM. Five large clusters (≥10 isolates), a unique sixth cluster (n = 9), and 8 smaller clusters (5–9 isolates) were identified. Prevalence of human immunodeficiency virus differed between clusters I and VI (P = .003), ranging from 27.8% to 100%. Receptive unprotected anal intercourse was frequently reported by MSM (51.8%) but did not differ significantly among clusters. Significant differences were identified concerning the participant's history of syphilis (P = .030), having met partners at a popular sex venue in Amsterdam (P = .048), and meeting partners outside Amsterdam (P = .036).

Conclusions. Distinct N. gonorrhoeae transmission networks were present in a mixed high-risk MSM population; concordance between clusters and epidemiological characteristics was present but not marked.

The incidence of Neisseria gonorrhoeae infection among men who have sex with men (MSM) has increased substantially in many industrialized countries since the mid 1990s, posing a serious public health problem [1, 2]. The increase is often attributed to lower risk perception and increasing risk behavior due to the introduction of combination antiretroviral therapy for HIV and the change of HIV infection from a terminal illness to a chronic manageable disease [3, 4]. In addition, risk-reducing strategies other than condom use, such as serosorting, that is, having unprotected anal intercourse with only HIV seroconcordant partners, are considered important risk factors for the spread of sexually transmitted infections (STIs) such as gonorrhea [5]. Additional risk factors for anogenital gonococcal infection include oral–anogenital sexual contacts [6]. Pharyngeal gonorrhea has been associated with oral–anal sex, suggesting that the pharynx might act as reservoir for N. gonorrhoeae [7–10].

The structure of sexual networks is important for STI transmission, but elucidating these transmission networks is challenging. Cross-sectional surveys do not provide actual partner links, and respondent-driven sampling can be hampered by persons unwilling or unable to reveal their sexual partners. Incomplete partner data might obscure partner links, leading to an underestimation of the network size [11, 12]. Molecular typing combined with epidemiological data can provide better insight into N. gonorrhoeae transmission patterns, which can help to improve intervention strategies. To our knowledge, there have been only a few molecular epidemiological studies in which N. gonorrhoeae transmission networks within well-defined MSM populations were examined [11, 13–16]. In this study, a validated molecular typing method, multiple-locus variable-number tandem repeat analysis for N. gonorrhoeae (NG-MLVA), was used to elucidate these networks within the MSM population in Amsterdam, the Netherlands [17]. To elucidate the temporary nature of N. gonorrhoeae outbreaks, we determined whether clusters could be related to calendar time, whereby some strains might persist as endemic strains while others disappear shortly after introduction. In addition, we examined the association of N. gonorrhoeae clusters with coinfection by other STIs; human immunodeficiency virus (HIV) status; and other demographic, epidemiological, and behavioral characteristics of MSM.

MATERIALS AND METHODS

Study Design

The study population was recruited from the STI outpatient clinic of the Public Health Service of Amsterdam, the Netherlands. It is the only STI clinic in Amsterdam with a low-threshold character: it offers anonymous and free-of-charge testing and treatment to nearly 28 000 clients a year, regardless of having a residence permit, health insurance, or a letter of referral. Approximately 22% are MSM, defined as men reporting sexual contact with men in the preceding 6 months. MSM were eligible for participation if they were at least 18 years old, could understand written Dutch or English, and gave written informed consent. Individuals could participate more than once if they visited the clinic for a new consultation. Recruitment, which ran from July 2008 to August 2009, was twice interrupted by a biannual survey among clinic visitors. The study was approved by the Medical Ethics Committee of the Academic Medical Centre of Amsterdam, the Netherlands.

STI Clinic Procedures

According to the STI clinic protocol, all participants were screened for urogenital Chlamydia trachomatis, pharyngeal and anogenital N. gonorrhoeae, syphilis, and hepatitis B (except when vaccination or natural immunity had been documented previously). HIV testing was offered to all participants on an opt-out basis [18] and was performed by rapid immunoassay (Abbott Determine HIV 1/2; Abbott Diagnostic Division, Hoofddorp, the Netherlands). Reactive samples were confirmed by HIV Ag/Ab Combo test (Axsym, Abbott Laboratories; Abbott Determine HIV 1/2, Abbott Diagnostics) and line immunoassay (Inno-Lia HIV I/II Score; Innogenetics, Ghent, Belgium). Hepatitis C serologic testing was offered to all HIV-positive MSM. If passive anal intercourse in the previous 6 months was reported, anogenital C. trachomatis was screened for; if anogenital C. trachomatis was detected, lymphogranuloma venereum was excluded by genovar-specific nucleic acid amplification testing. Chlamydia testing was performed with the APTIMA platform (APTIMA Combo 2 assay for C. trachomatis and N. gonorrhoeae; Gen-Probe, San Diego, CA). Urogenital and anogenital gonorrhea testing was performed by culture, and pharyngeal gonorrhea was determined by the APTIMA platform. All samples included in this study were derived from pure N. gonorrhoeae cultures.

From all participants with macroscopic signs of urethritis and/or proctitis, Gram-stained smears were prepared for microscopic examination. If >10 polymorphic nucleic leucocytes per high-power field were seen, a presumptuous diagnosis of urethritis and/or proctitis was made. If intracellular gram-negative diplococci were seen, a presumptuous diagnosis of gonococcal urethritis and/or proctitis was made. For all presumptuous diagnoses, standard antibiotic treatment was commenced immediately.

Bacterial Cultivation and Lysate Preparation

For cultivation of N. gonorrhoeae, urethral and rectal swab specimens were directly inoculated onto GC-Lect agar plates (Becton Dickinson, Franklin Lakes, NJ) and incubated in an aerobic, carbon dioxide–enriched environment at 37°C for 40–48 hours. N. gonorrhoeae colonies were identified by Gram staining and oxidase-, sugar utilization-, and aminopeptidase reactions. Positive colonies were confirmed by nucleic acid amplification testing (AccuProbe; Gen-Probe) and stored in 20% glycerine at −80°C. Culture plates were prepared from the stored glycerine-frozen cultures. For all isolates, antimicrobial susceptibility was evaluated by Etest (AB bioMérieux, Askim, Sweden). Subsequently, colonies were suspended in sterile saline (0.5 McFarland), lysed at 95°C for 10 minutes, and stored at −80°C prior to nucleic acid amplification.

Molecular Typing of N. gonorrhoeae

Amplification of the variable number of tandem repeat (VNTR) sequences was performed on a BioRad C1000 PCR system (BioRad, Hercules, CA). Five VNTR loci (VNTR04-03, VNTR04-10, VNTR07-02, VNTR15-02, and VNTR16-01) were amplified in 2 multiplex polymerase chain reactions (PCRs), as described previously [17]. The amplified samples were diluted 1:20 in water, and 2 μL of each diluted sample was mixed with 18 μL of a 1:450 in-water diluted GeneScan LIZ 500 size standard (Applied Biosystems, Carlsbad, CA). After heat denaturation for 5 minutes at 95°C, fragments were separated with an ABI 3130 automated sequencer (Applied Biosystems) using the fragment analysis module. Sizing and calculation of the number of repeats of each VNTR were performed with GeneMarker software, version 1.80 (SoftGenetics, State College, PA). A hierarchical cluster analysis of the NG-MLVA typing data was performed to compensate for the highly polymorphic nature of VNTR04-03, VNTR04-10, and VNTR07-02. This was done by assigning double weight to VNTR15-02 and VNTR-1601 because these VNTR loci were considered more stable (Bionumerics software, version 6.0; Applied Maths, Sint-Martens-Latern, Belgium). Single-locus variants (SLV) were considered to belong to the same N. gonorrhoeae strain. This genotyping method has been described previously [17].

Questionnaire

Participants completed a computer-assisted self-interview (CASI) about demographics and characteristics regarding themselves and their partners. This was done after the consultation or while waiting for preliminary results of the C. trachomatis, N. gonorrhoeae, and HIV tests, which are typically available within 45 minutes. During the first 2 months of the study, a paper version of the English CASI was used; it was later replaced by a CASI version. Later, in case of technical problems, an English and Dutch paper version of the CASI were available.

The questionnaire reflected the 6 months preceding the clinic visit and inquired as to partnership with a self-defined steady partner, if any, and the 3 most recent other partnerships: a steady partner, a traceable casual partner, or an anonymous casual partner, see Supplementary Information S.1. No a priori definition of partnership status was provided to allow participants to subjectively determine their partner's status. It was assumed that the perceived status would be an important determinant of behavior without necessarily being mediated by external criteria such as a minimum partnership duration. To determine whether a partnership was 1 of the 3 most recent partnerships, the date of last sexual contact was used, irrespective of the partnership start date.

Determinants

Age was divided into 3 categories based on the interquartile range (IQR). Self-defined ethnicity was categorized as Dutch, Western non-Dutch, and non-Western. In the questionnaire, lifestyle was determined by asking whether the participant characterized himself by code of dressing or as belonging to a certain social stream within the gay community, for example, a casual, leather, rubber, or military type.

The number of partners in the previous 6 months was categorized using IQRs. We distinguished between the number of self-defined steady partners, traceable casual partners, and untraceable anonymous casual partners.

Sexual risk behavior and substance use during last sexual contact in the previous 6 months were assessed per partnership. We made a distinction between insertive and receptive anal intercourse and fisting (ie, brachioproctic insertion). Unprotected anal intercourse was defined as never using or inconsistent use of condoms during anal intercourse.

With respect to use of sex-related substance, we made a distinction between amyl nitrate (poppers) and other substances, that is, noninjecting use of cocaine, ketamine, 3,4-methylenedioxymethamphetamine (XTC), gamma hydroxybutyrate (GHB), amphetamine, or methyl-amphetamine. The specific meeting location where the participant met his sexual partner (eg, the name of the Web site or the type of location, such as a club or gay sauna) was also assessed for each partnership.

Due to recruitment interruptions during the study period, calendar time was divided into 3 parts: period 1, July 2008 through October 2008; period 2, December 2008 through May 2009; and period 3, June 2009 through August 2009.

Statistical Analysis

To examine whether epidemiological characteristics differed according to being in a larger cluster or not, we performed χ² tests for independence and Mann-Whitney U tests for continuous variables. Then, we examined the associations between clusters and calendar time and epidemiological factors (ie, demographic characteristics, risk behavior, and meeting location), performing χ² tests for independence and Kruskal-Wallis tests for continuous variables. Fisher exact tests were performed when the expected value was less than 1. Analysis was done using Stata 11.1 (StataCorp., College Station, TX).

RESULTS

Identification of Distinct NG-MLVA Clusters

During the study period, 4513 MSM visited the STI clinic in 6532 visits, of whom 2492 MSM in 3050 visits participated in the study. MSM who tested negative for urethral and rectal N. gonorrhoeae were excluded, resulting in a study population of 240 MSM with 253 clinic visits. We identified 283 N. gonorrhoeae–positive isolates; 278 isolates were successfully genotyped, revealing 5 larger clusters of 10 or more isolates and 9 smaller clusters of 5 to 9 isolates (Figure 1 and Table 1).

Table 1.

Number of Isolates, Client Visits, and MSM in NG-MLVA Clusters I–VI, in a Small-Cluster Group and Not-Clustered Group, Included in the Study Among Clients of the Sexually Transmitted Infection Clinic in Amsterdam, 2008–2009

Cluster	Total No. of Isolates Identified (N = 278)	No. of Isolates Included (N = 259)a	No. of Client Visits (N = 253)b	No. of MSM (N = 240)c
I	37	34	34	34
II	19	18	18	18
III	28	25	25	25
IV	10	10	10	10
V	22	20	20	20
VI	9	9	9	8d
In a small cluster	50	49	49	49
Not in a cluster	103	94	91	90d

The numbers in the last 2 columns do not add up to the column totals because some men contributed more than 1 isolate per clinic visit (or contributed more than 1 visit), which were not always in the same cluster.

Abbreviations: MSM, men who have sex with men; NG-MLVA, multiple-locus variable-number tandem repeat analysis for Neisseria gonorrhoeae.

^a Over the study period, 278 N. gonorrhoeae isolates were obtained; 19 samples were excluded from MSM who had 2 positive isolates at 1 clinic visit with the same NG-MLVA type.

^b Over 6 clinic visits, individuals had multiple isolates with divergent NG-MLVA types because, over 3 visits, samples were “not in a cluster,” 94 isolates are listed in that group for 91 visits.

^c Of the MSM, 240 visited the STI clinic; 228 contributed 1 N. gonorrhoeae–positive visit, 11 contributed 2 visits, and 1 contributed 3 visits.

^d The isolates of 2 men at 2 separate clinic visits were classified to the same group at both visits (cluster VI and “not in a cluster”).

Figure 1.

Minimum spanning tree of 259 isolates. Each circle represents 1 multiple-locus variable-number tandem repeat analysis (MLVA) type. The size of the circle is proportional to the number of isolates with that identical MLVA type. Bold lines connect types that differ in a single variable-number of tandem repeat (VNTR) locus; dotted lines connect types that differ in 2 loci (double locus variants); colored halos indicate distinct clusters, defined by chains of MLVA types that did not differ by more than 1 VNTR locus. Roman numerals I–V represent 5 larger clusters (≥10 isolates), and VI represents 1 smaller cluster (9 isolates). Red indicates human immunodeficiency virus (HIV)-positive status; white indicates HIV-negative status.

In 25 individuals, infections at 2 anatomical locations (both urethral and rectal) were diagnosed at a single visit. Nineteen of the 25 MSM (76.0%) had isolates with identical NG-MLVA types at both locations. We excluded their 19 rectal isolates from further analysis. For the remaining 6 MSM who had isolates with divergent NG-MLVA types (multilocus variants), both samples were included.

Five couples were identified (ie, MSM who each reported being the other's sexual partner). In 4 of the 5 couples (80%), the partners had isolates containing NG-MLVA types that were considered identical (ie, identical NG-MLVA types or with SLVs). The fifth couple was discordant, having rectal infections with identical NG-MLVA types but urethral isolates that differed.

One MSM visited the clinic 3 times and was diagnosed with N. gonorrhoeae at 2 anatomical locations at the first visit. Eleven MSM were diagnosed with N. gonorrhoeae at 2 separate visits (1 to 10 months apart). One of them had identical NG-MLVA types at the first and second visit (separated by 3 weeks), suggesting reinfection with the same strain or possible treatment failure. In 9 others, 9 isolates obtained at the second visit belonged to different NG-MLVA clusters than the isolates from the first visit. The strains from 1 recidivist were different from each other and not classified to a cluster on either of his visits.

Epidemiological Analysis of N. gonorrhoeae MLVA Clusters

The study population was predominantly Dutch (70.2%), and the median age was 37 (IQR 30–42) years. A median number of 10 (IQR 4–20) sexual partners in the previous 6 months was reported (Table 2). In the first analysis we determined whether MSM in 1 of the larger clusters (cluster I, II, III, IV, or V) differed from MSM who were not in a larger cluster (Table 3). The HIV prevalence was higher among MSM who were not in a larger cluster (P = .048), but no other epidemiological, statistically significant differences were found.

Table 2.

Epidemiological Characteristics Recorded at 253 Visits of MSM Diagnosed With Rectal or Urethral Gonorrhea at the Sexually Transmitted Infection Clinic in Amsterdam, 2008–2009

Epidemiological Characteristic	Client visits (N = 253)a
Demographic characteristics
Median age, in years (IQR)	37 (30–42)
Ethnicity
Dutch	177 (70.2%)
Western, non-Dutch	24 (9.5%)
Non-Western	51 (20.2%)
Partner types in the previous 6 mo
At least 1 steady partner	154 (60.9%)
At least 1 traceable casual partner	160 (63.2%)
At least 1 anonymous partner	119 (47.0%)
Median no. of partners of the index patient in the previous 6 mo (IQR)
Total no. of partners	10 (4–20)
No. of steady partners (IQR)	1 (0–1)
No. of traceable casual partners (IQR)	3 (1–5)
No. of anonymous partners (IQR)	4 (0–13)
Coinfections and substance use
Chlamydia trachomatis coinfection	70 (27.7%)
HIV status
Negative	123 (48.8%)
Known positive	110 (43.7%)
Newly tested positive	10 (4.0%)
Not tested	9 (3.6%)
Sex-related drug use with at least 1 partner in the previous 6 mob	104 (41.1%)

Numbers might not add up to the column total due to missing data.

Abbreviations: IQR, interquartile range; MSM, men who have sex with men.

^a Of the MSM, 228 contributed 1 visit, 11 contributed 2 visits, and 1 contributed 3 visits.

^b Drug use is defined as noninjecting recreational use of cocaine, 3,4-methylenedioxymethamphetamine, gamma hydroxybutyrate, ketamines, amphetamines, or methylamphetamines.

Table 3.

Epidemiological Characteristics of MSM by NG-MLVA Cluster, Classified by NG-MLVA Typing, Amsterdam, 2008–2009

	Larger Cluster I (N = 34)	Larger Cluster II (N = 18)	Larger Cluster III (N = 25)	Larger Cluster IV (N = 10)	Larger Cluster V (N = 20)	Smaller Cluster VI (N = 9)	Not in Clusters I–V (N = 152)a	Pb	Pc	Pd
Demographics of the Index Patient
Median age, in years (IQR)	37 (31–45)	34.5 (27–39)	38 (31–43)	33 (31–42)	37.5 (29–41)	40 (31–42)	36 (29–42)	.704	.768	.709
Ethnicity
Dutch	25 (73.5%)	15 (83.3%)	19 (76.0%)	7 (70.0%)	12 (60.0%)	8 (88.9%)	105 (69.5%)	.605	.728e	.715e
Western, non-Dutch	3 (8.8%)	2 (11.1%)	2 (8.0%)	0	4 (20.0%)	0	12 (8.6%)
Non-Western	6 (17.7%)	1 (5.6%)	4 (16.0%)	3 (30.0%)	4 (20.0%)	1 (11.1%)	33 (21.9%)
Leather lifestyle	3 (9.4%)	1 (5.6%)	4 (16.7%)	2 (20.0%)	1 (5.3%)	0	23 (15.9%)	.566	.535	.587
Military lifestyle	1 (3.1%)	2 (11.1%)	2 (8.3%)	1 (10.0%)	0	1 (12.5%)	12 (8.3%)	.432e	.422e	.647e
Rubber/Lycra lifestyle	4 (12.5%)	2 (11.1%)	3 (12.5%)	2 (20.0%)	0	0	9 (6.2%)	.496	.472e	.309e
Median no. of partners of the index patient in the previous 6 mo (IQR)
Total partners	10 (4–20)	7.5 (3–15)	24.5 (6–35)	10 (4–20)	7 (3–26)	22.5 (10–40)	10 (5–20)	.278	.135	.302
Steady partners	1 (0–1)	1 (1–1)	1 (1–3)	1 (0–2)	1 (0–1)	0 (0–1)	1 (0–1)	.013	.011	.024
Traceable casual partners	2 (1–5)	3 (1–5)	4.5 (1.5–14)	2.5 (0–7)	3 (0–5)	4.5 (3–11.5)	3 (1–6)	.602	.472	.692
Anonymous partners	5 (1–15.5)	2.5 (0–10)	7.5 (1–24.5)	5 (1–9)	4 (0–20)	14 (8–30)	4 (0–11)	.561	.194	.623
Coinfections
Chlamydia trachomatis	10 (29.4%)	6 (33.3%)	7 (28.0%)	0	6 (30.0)	1 (11.1)	43 (28.3%)	.372	.358	.516
Human immunodeficiency virusf	20 (58.8%)	5 (27.8%)	11 (44.0%)	3 (30.0%)	6 (31.6%)	9 (100.0%)	80 (55.9%)	.142	.003	.048
Serological evidence of past/present syphilis	9 (26.5%)	5 (27.8%)	13 (52.0%)	4 (40.0%)	3 (15.0%)	6 (66.7%)	53 (35.1%)	.085	.030	.142
Sexual risk behavior
Insertive anal intercourse
None	7 (21.9%)	7 (38.9%)	2 (8.7%)	1 (10.0%)	2 (10.5%)	1 (12.5%)	36 (24.8%)	.072	.052	.126
Safe only	6 (18.8%)	3 (16.7%)	4 (17.4%)	3 (30.0%)	9 (47.4%)	0	38 (26.2%)
Unprotected with at least 1 partner	19 (59.4%)	8 (44.4%)	17 (73.9%)	6 (60.0%)	8 (42.1%)	7 (87.5%)	71 (49.0%)
Receptive anal intercourse
None	6 (18.8%)	4 (22.2%)	3 (13.0%)	0	8 (42.1%)	1 (12.5%)	34 (23.5%)	.150	.117	.198
Safe only	5 (15.6%)	4 (22.2%)	7 (30.4%)	2 (20.0%)	5 (26.3%)	0	41 (28.3%)
Unprotected with at least 1 partner	21 (65.6%)	10 (55.6%)	13 (56.5%)	8 (80.0%)	6 (31.6%)	7 (87.5%)	70 (48.3%)
Insertive fisting with at least 1 partner	3 (8.8%)	2 (11.1%)	5 (20.0%)	3 (30.0%)	0	0	17 (11.2%)	.109	.096	.159
Receptive fisting with at least 1 partner	2 (5.9%)	0	3 (12.0%)	2 (20.0%)	1 (5.0%)	0	15 (9.9%)	.283	.347	.479
Sex-related substance use with at least 1 partner
Poppers use	19 (59.4%)	10 (55.6%)	16 (69.6%)	8 (80.0%)	9 (47.4%)	6 (75.0%)	72 (49.7%)	.411	.462	.237
Drug useg	17 (50.0%)	11 (61.1%)	9 (36.0%)	5 (50.0%)	6 (30.0%)	8 (88.9%)	61 (40.1%)	.290	.045	.345
Meeting location of at least 1 partner
Internet	23 (67.7%)	8 (44.4%)	17 (68.0%)	6 (60.0%)	13 (65.0%)	8 (88.9%)	92 (60.5%)	.527e	.318	.643
Amsterdam	11 (32.4%)	8 (44.4%)	10 (40.0%)	7 (70.0%)	9 (45.0%)	3 (33.3%)	60 (39.5%)	.322	.421	.428
Gay sauna in Amsterdam	1 (2.9%)	5 (27.8%)	1 (4.0%)	1 (10.0%)	2 (10.0%)	2 (22.2%)	18 (11.8%)	.049e	.048e	.111
In the Netherlands, not in Amsterdam	1 (2.9%)	1 (5.6%)	6 (24.0%)	1 (10.0%)	0	0	9 (5.9%)	.026e	.036e	.035e

Numbers might not add up to the column totals due to missing data.

Abbreviations: IQR, interquartile range; MSM, men who have sex with men; NG-MLVA, multiple-locus variable-number tandem repeat analysis for Neisseria gonorrhoeae.

^a Isolates of MSM who were in 1 of the smaller clusters or not in a cluster at all.

^bP value for χ² tests and Kruskal-Wallis tests of cluster I–V.

^cP value for χ² tests and Kruskal-Wallis tests of cluster I–VI.

^dP value for χ² tests and Kruskal-Wallis tests comparing “being in a larger cluster” with “being not in a larger cluster.”

^eP value for Fisher exact test.

^f There were 10 newly diagnosed HIV-positive MSM: 1 in clusters I, II, and V and 7 in the group “not in a larger cluster.” There were 9 MSM who were not tested for HIV: 1 in cluster V and 8 in the group “not in a larger cluster.”

^g Drug use is defined as not injecting use of cocaine, 3,4-methylenedioxymethamphetamine, gamma hydroxybutyrate, ketamines, amphetamines, or methylamphetamines.

The second analysis, in which we determined whether clusters differed according to epidemiological characteristics, was performed on larger clusters I–V. Because the isolates (N = 9) in smaller cluster VI contained identical NG-MLVA types that were very different from all other types, a third analysis was performed on clusters I–VI. In both analyses, demographic characteristics did not differ significantly among clusters, nor did belonging to a specific MSM behavioral subculture (eg, being a leather, military, or rubber type), indicating that N. gonorrhoeae clusters were not occurring in distinct subgroups of the MSM population.

Significant differences by cluster were found for the number of steady partners (P = .013 for clusters I–V; P = .011 for clusters I–VI), while the number of traceable casual and anonymous partners did not differ by cluster (Table 3).

Larger clusters I–V did not differ according to sexual risk behavior. When cluster VI was included, borderline significant differences were found for insertive anal intercourse (P = .072 for cluster I–V; P = .052 for I–VI). In this cluster 87.5% of the MSM reported unprotected insertive anal intercourse. Overall, insertive and receptive unprotected anal intercourse were common, as well as among HIV negative MSM of whom 45.3% reported insertive unprotected anal intercourse and 38.5% reported receptive unprotected anal intercourse.

HIV prevalence ranged from 27.8% to 58.8% in clusters I–V and did not differ significantly by cluster. Nevertheless, a significant difference among clusters was seen when cluster VI was included in the analysis (P = .003). The HIV prevalence in this cluster was 100%. Significant differences between clusters were also found with respect to serological evidence of past or present syphilis infection (P = .030 for clusters I–VI), which is a marker of lifetime risk behavior. This serological evidence was found in only 15% of the MSM in cluster V but in 52.0% of cluster III and 66.7% of cluster VI (Table 3).

Sex-related substance use occurred in all clusters. Amyl nitrate was used by 54.3% of the population and sex-related noninjecting drug use was reported by 42.1%. In cluster VI, 88.9% of the MSM reported sex-related drug use; in the analyses including cluster VI, significant cluster differences were found (P = .045; Table 3).

Overall, 40.8% of the MSM met at least 1 partner at a location in Amsterdam. Meeting in a gay sauna in Amsterdam differed significantly by cluster (P = .049 for clusters I–V; P = .048 for clusters I–VI); such meetings were more often reported by MSM in cluster II. MSM in cluster III were more likely to have met a partner outside of Amsterdam; 50% of them were linked to the region of Rotterdam/The Hague (the P values for cluster differences were P = .026 for clusters I–V and P = .036 for cluster I–VI). The Internet was used as a meeting location by 61.2% of the total population; cluster differences were not significant. In cluster VI, 8 of the 9 MSM (88.9%) reported online partners; 6 of the 8 (75%) met partners at the same Web site (Table 3).

Next, we examined whether the occurrence of NG-MLVA clustered in time. The study period was divided into 3 parts. During the first period, 39.7% of the total number of gonorrhea cases was identified. The second period accounted for 42.6% of the cases, and the last period for 17.7% of the cases. The presence of different N. gonorrhoeae strains varied significantly among time periods (P = .001 for clusters I–V; P < .001 for clusters I–VI). Strains belonging to clusters II, IV, and V were seen only in periods 1 and 2, and the strain in cluster VI was seen only in period 2, whereas the strains of clusters I and III were identified throughout the study period (Figure 2).

Figure 2.

Identified Neisseriagonorrhoeae clusters among 278 isolates from 240 men who have sex with men with N. gonorrhoeae infection who visited the sexually transmitted infection clinic in Amsterdam, by time since start of study. Six clusters (I–VI), identified by multiple-locus variable-number tandem repeat analysis for N. gonorrhoeae, are shown according to the day of diagnosis since the start of the study. Individuals were recruited from July 2008 through August 2009. Recruitment was halted twice (gray areas) because of a biannual survey among clinic visitors, dividing the study into 3 parts: period 1, July 2008 through October 2008; period 2, December 2008 through May 2009; and period 3, June through August 2009. The number of isolates per time period, shown in the supporting table, differed significantly (P = .001 for clusters I–V; P < .001 for clusters I–VI).

DISCUSSION

NG-MLVA was used to identify N. gonorrhoeae transmission networks among MSM in Amsterdam, the Netherlands. Hierarchical cluster analysis of NG-MLVA profiles revealed 6 clusters (large clusters I–V and small cluster VI). These clusters differed to some degree in various epidemiological characteristics. The limited overall concordance between N. gonorrhoeae clusters and epidemiological characteristics suggests that transmission of distinct N. gonorrhoeae strains is not related to distinct subgroups of the high-risk MSM population in Amsterdam.

The HIV prevalence differed significantly among clusters I–VI. Interestingly, all members in cluster VI were HIV-infected and were often active on a single large dating Web site, which might facilitate HIV disclosure. They visited the STI clinic within a 2-month period. Although we did not measure serosorting explicitly, these findings suggest that serosorting was practiced. The fact that all 9 men in this cluster were HIV-positive was unusual and suggests that they disclosed their HIV status prior to having unprotected sex or in order to have unprotected sex. MSM in cluster VI more often than MSM in other clusters used sex-related drugs, had anonymous partners, and had serological evidence of past or present syphilis infection, indicating more risky sexual behavior. Those in other clusters were not exclusively HIV-negative or exclusively HIV-positive. Given the high HIV prevalence, the mix of HIV-negative and HIV-positive MSM, and high-risk behaviors in all clusters, it is likely that many HIV-negative individuals within the networks were at risk of acquiring HIV.

The link between NG-MLVA clusters and some epidemiological determinants mentioned earlier supports the presence of N. gonorrhoeae transmission networks. Yet no clusters could be identified by epidemiological data alone. That is, identified molecular clusters were not exclusively characterized by specific epidemiological determinants, as is reported for heterosexual networks [11, 17, 19]. Our findings, particularly the fact that 11 MSM were diagnosed at 1 visit with different N. gonorrhoeae strains at 2 anatomical sites, suggests that N. gonorrhoeae transmission networks overlap, with many MSM active in more than 1 network. Sensitivity analysis was performed to assess the impact of multiple isolates per individual; omission of multiple isolates did not affect the results.

Clusters I and III probably represent ongoing transmission of endemic N. gonorrhoeae strains in large MSM transmission networks because their strains were identified throughout the study period. The molecular epidemiology of cluster VI, on the other hand, was markedly divergent from that of the other clusters, and the sudden emergence and relatively short circulation time of this MLVA type might represent a local outbreak. The disappearance of clusters II, IV, V, and VI may indicate their replacement by other N. gonorrhoeae strains or by successful STI prevention and intervention strategies.

In accordance with previous findings [17], the validity of cluster classification was substantiated by the relatedness between isolates of patients diagnosed at a single visit with infections at 2 sites (76%) and between isolates shared by members of couples (80%). In addition, previous studies have shown that the NG-MLVA performs well and that acceptable concordance was found between the results of NG-MLVA and 2 frequently used N. gonorrhoeae genotyping techniques (ie, NG-MAST and porB sequencing) [17, 20].

Our results should be interpreted with caution because the study population included only STI clinic visitors and is therefore not a random sample of the general MSM population. In addition, pharyngeal samples were not evaluated in this study because NG-MLVA, as it currently performs, requires cultured samples prior to amplification and fragment analysis. Our pharyngeal samples were not cultured (unlike urethral and rectal samples) but were amplified directly from sampled material. Because commensal Neisseria strains are known to colonize the pharynx, these samples were considered unsuitable. Various reports described oro–penile and oro–anal male-to-male contact as risk factors for acquiring pharyngeal gonorrhea [6, 9, 16]. Because the pharynx may act as a reservoir for gonococcal recombination, we may have missed an important group or at least anatomical locations that significantly contribute to the introduction and spread of new or recombined N. gonorrhoeae strains. The actual size of the transmission network might thus be underestimated due to these missing transmission links [11].

An understanding of the transmission networks of N. gonorrhoeae and also the sexual network structure can contribute to optimization of intervention strategies, for example, by targeting high-risk populations or populations that contribute most to the spread of the epidemic. Even though we did not sample the entire MSM population of Amsterdam and therefore not the entire transmission network, we believe that a representative group was assessed to provide valid additional insight and understanding of the underlying N. gonorrhoeae transmission network structure. We have shown that due to mixing of HIV-negative and HIV-infected MSM in nearly all clusters and the high frequency of unprotected anal intercourse, many HIV-negative MSM are at risk of acquiring HIV infections. The concordance between identified clusters and epidemiological characteristics was present, though limited. We therefore conclude that distinct N. gonorrhoeae transmission networks in Amsterdam exist, but the network structure does not reflect distinct subgroups of the high-risk MSM population.

Notes

Acknowledgments. The authors thank all other members of the study team including Han Fennema, Mirjam Kretzschmar, Maria Xiridou, Maria Prins, Reinier Bom, and Jan Prins, for their contributions to the design and implementation of the study. Also, we thank M. van Rooijen for the data management, L. Philips for editing the manuscript, and the anonymous referees for their suggestions.

Financial support. This work was funded by the Netherlands Organisation for Health Research (grant number 125010008).

Potential conflicts of interest. All authors: No reported conflicts.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Author notes

Data were presented previously at the 19th ISSTDR Conference, July 2011, Québec, Canada.

a

R. H. and A. A. M. contributed equally to this paper.