The Relationship Between Insertive Oral and Anal Sex and Select Measures of the Composition of the Urethral Microbiota Among Men Who Have Sex With Men

Abstract

Background:

Sexual behavior may influence the composition of the male urethral microbiota, but this hypothesis has not been tested in longitudinal studies of men who have sex with men (MSM).

Methods:

From 12/2014–7/2018, we enrolled MSM with non-gonococcal urethritis (NGU) attending a sexual health clinic. Men attended five in-clinic visits at 3-week intervals, collected weekly urine specimens at home, and reported daily antibiotics and sexual activity on weekly diaries. We applied broad-range 16S rRNA gene sequencing to urine. We used generalized estimating equations to estimate the association between urethral sexual exposures in the prior 7 days (insertive oral sex [IOS] only, condomless insertive anal intercourse [CIAI] only, IOS with CIAI [IOS+CIAI], or none) and Shannon index, number of species (observed, oral indicator, and rectal indicator), and specific taxa, adjusting for recent antibiotics, age, race/ethnicity, HIV, and pre-exposure prophylaxis.

Results:

Ninety-six of 108 MSM with NGU attended ≥1 follow-up visit. They contributed 1,140 person-weeks of behavioral data and 1,006 urine specimens. Compared to those with no urethral sexual exposures, those with IOS only had higher Shannon index (P=0.03) but similar number of species and presence of specific taxa considered, adjusting for confounders; the exception was an association with Haemophilus parainfluenzae. CIAI only was not associated with measured aspects of the urethral microbiota. IOS+CIAI was only associated with presence of H. parainfluenzae and Haemophilus.

Conclusions:

Among MSM after NGU, IOS and CIAI did not appear to have a substantial influence on measured aspects of the composition of the urethral microbiota.

SUMMARY

Among men who have sex with men after non-gonococcal urethritis, insertive oral and anal sex did not appear to have a substantial influence on the composition of the urethral microbiota.

INTRODUCTION

Although many non-gonococcal urethritis (NGU) cases are caused by Chlamydia trachomatis (CT) and Mycoplasma genitalium (MG), 30–50% of cases are of unknown etiology.^1–3 Clinicians frequently treat NGU syndromically,⁴ with treatment failure rates of up to 20%.⁵ While antibiotic resistant MG^5,6 and reinfection account for some persistent and recurrent NGU cases, difficult to cultivate bacteria and polymicrobial communities that are not susceptible to the antibiotics used to treat NGU may also play a role.

Studies suggest that specific sexual exposures can influence the composition of the male urethral microbiota, in addition to transmitting pathogens. Among adolescent men, certain bacterial genera were only detected in urine from sexually active men, suggesting colonization via urethral sexual exposures.⁷ Some cross-sectional studies have detected associations between vaginal sex and specific bacteria/taxa or community type;^8,9 others have not detected associations with sexual exposures.^10–12 One cross-sectional study suggested vaginal sex impacts the urethral microbiota but identified little to no evidence of an association between insertive oral sex (IOS) or insertive anal intercourse (IAI) and the urethral microbiota.⁹ One cohort study found that heterosexual men reporting recent condom use and a longer time since last sexual episode had less diverse penile microbiota.¹³

However, most prior studies evaluating the relationship between sex and the male urethral microbiota considered broad measures of sexual activity (e.g., exposures in the last few months), potentially obscuring short-term influences of specific exposures, and were relatively small, cross-sectional, or predominantly among heterosexual men.^7,9,10,12,13 Many men who have sex with men (MSM) are exposed to different anatomic sites than heterosexual men, resulting in different influences on their urethral microbiota. We conducted a cohort study of MSM with NGU and estimated the association between recent IOS and condomless IAI (CIAI) and select measures of the composition of the urethral microbiota.

MATERIALS AND METHODS

Study Design and Procedures

We recruited Public Health–Seattle and King County (PHSKC) Sexual Health Clinic (SHC) patients age ≥16 years with NGU who were assigned male sex at birth and only had sex with people assigned male sex at birth in the past year. Patients reporting no sexual activity in the past 60 days, antibiotic therapy in the past 30 days, or urethral contact to Neisseria gonorrhoeae (GC) were ineligible. Patients were required to have a freezer at home to store urine specimens.

This prospective cohort study included in-clinic visits every three weeks for three months (five visits total). Participants collected weekly urine specimens at home and completed a weekly symptom and sex diary.

Enrollment Visit.

One of two study clinicians offered enrollment to eligible patients. The clinician conducted a standard examination, collecting a urethral swab for Gram staining and 30–45 mL of first-catch urine. The clinician examined the Gram-stained slide of urethral exudates to quantitate polymorphonuclear leukocytes (PMNs) and check for evidence of GC infection.⁴ We defined NGU as urethral symptoms or visible urethral discharge, with ≥5 PMNs per high power field (HPF). Participants received presumptive NGU treatment according to current clinic standard of care (azithromycin [1g, single-dose] or doxycycline [100mg, twice daily, 7 days]).⁴ Participants provided sociobehavioral data on a questionnaire developed with Research Electronic Data Capture (REDCap).¹⁴

Follow-Up Visits.

Follow-up visits included the same procedures as the enrollment visit, irrespective of signs and symptoms. Participants with urethral symptoms between scheduled visits returned for an interim visit with the same procedures and treatment as indicated.

Home Collection of Urine.

Throughout follow-up, participants collected 30–45 mL of first-void, first-catch urine at home weekly. They were instructed to collect the specimen every Monday morning, store them in their freezer, and bring them to their next study visit. We provided participants with specimen tubes, biohazard bags, labels, an opaque container for private storage, and an insulated bag with ice packs to use during transport to the clinic.

Web-Based, Mobile Phone-Enabled Diaries.

Throughout follow-up, participants completed weekly diaries, reporting daily instances of sexual activity, antibiotic therapy (for any reason), and urethral symptoms. Participants reported the specific types of sex at each sexual episode and condom use for IAI and receptive anal intercourse (RAI). Diary questions are available in the Supplement. We programmed and automated the diaries using REDCap and timed-out incomplete diaries after 7 days. Participants could opt to complete paper diaries, but none did.

Laboratory Testing

We tested clinic-collected urine specimens from symptomatic and asymptomatic participants for GC, CT, and MG using Aptima assays (Hologic, Inc., San Diego, CA). Clinicians treated participants with positive tests according to clinic standard of care.

We applied broad-range 16S rRNA gene PCR and Illumina MiSeq sequencing to all clinic- and home-collected urine specimens (methods previously described).^3,15 Sequences were deposited to the NCBI Short Read Archive (accession number PRJNA972692).

Statistical Analysis

We summarized baseline characteristics and antibiotic exposures with percentages. We compared continuous measures of the composition of the urethral microbiota using medians and interquartile ranges (IQRs). Specimens with insufficient bacterial DNA for sequencing or <1,000 sequence reads were excluded from analyses. Additionally, analyses excluded specimens collected within the first 7 days after enrollment because participants had not yet provided sufficient diary data to classify their recent exposures.

We estimated the association between urethral sexual exposures in the 7 days before specimen collection and the composition of the urethral microbiota using a permutational multivariate analysis of variance (PERMANOVA) test based on weighted UniFrac distance. Although this method does not account for repeated measurements on participants, it incorporates the relative abundance of bacterial taxa, permitting the identification of specific bacterial communities rather than simpler summary measures (e.g., alpha diversity). We considered urethral sexual exposures using four mutually exclusive categories: (1) IOS (presumed condomless) but not CIAI, (2) CIAI but not IOS, (3) both IOS and CIAI, (4) none (i.e., neither IOS nor CIAI). Our focus was on urethral exposures, so we did not include RAI. Participants who reported vaginal sex during follow-up were excluded from all analyses. When the PERMANOVA test was significant, we conducted principal coordinates analyses to visualize potential differences in urethral community composition between urethral sexual exposure groups using weighted UniFrac distance.

To complement PERMANOVA analyses and account for repeated measurements, we estimated the association between specific urethral sexual exposures in the prior 7 days and summary measures of bacterial diversity using generalized estimating equations (GEE). We assessed alpha diversity (Shannon index¹⁶) and number of species (observed, oral indicator, and rectal indicator). We log-10 transformed each of the numbers of species variables. Bacterial species were classified as potential oral or rectal species based on literature reports of their presence in human gut or oral samples (Supplemental Table S1).^17–20 We fit a single model for each outcome, specifying GEE with a Gaussian error distribution, identity link, exchangeable working correlation, and robust standard errors. Models were adjusted for the following hypothesized confounders: antibiotic use in the prior 7 days (yes, no), age (continuous), race (Black, White, other/multiple/unknown), known Hispanic or Latino ethnicity (yes, no), and HIV and pre-exposure prophylaxis (PrEP) status (previously diagnosed with HIV, never diagnosed with HIV and not on PrEP, never diagnosed with HIV and on PrEP). We excluded specimens with missing data on sexual or antibiotic exposures in the prior 7 days from multivariable models.

We conducted three sensitivity analyses. First, we used GEE as described above but with a Poisson error distribution and log link to consider the presence (yes, no) of oral and rectal indicator species. Second, we restricted analyses to specimens without antibiotic exposure in the prior 7 days to remove the effect of antibiotics on the urethral microbiota. Third, we restricted analyses to home-collected specimens to explore the impact of the differing collection protocols. We also estimated the association between specimen collection method and presence of insufficient bacterial DNA.

In addition to analyses of community composition, we used GEE as described above to consider the presence of 14 taxa significantly associated with IOS and/or IAI in the prior 60 days in a prior cross-sectional study,⁹ including Actinomyces turicensis, Corynebacterium pseudogenitalium/tuberculostearicum, Corynebacterium, Gardnerella vaginalis, Haemophilus parainfluenzae, Haemophilus, Peptoniphilus asaccharolyticus/grossensis/harei, Prevotella bivia, Prevotella buccalis, Prevotella timonensis, Streptococcus anginosus group, Propionimicrobium lymphophilum, Actinomces radingae and Streptococcus pseudopneumoniae. P-values were adjusted for multiple comparisons using the Benjamini-Hochberg method. Finally, we used GEE as described above for an exploratory analysis of whether select taxa previously identified only in men who reported vaginal sex⁹ were more common in specimens from MSM with versus without a lifetime history of vaginal sex, including Lactobacillus iners, G. vaginalis, Atopobium vaginae, Mageeibacillus indolicus, Prevotella amnii, Sneathia amnii, Streptococcus anginosus, and Veilonella montpellierensis.

This study was approved by the University of Washington Human Subjects Division. Participants provided written, informed consent. We compensated participants $25–60 per visit and $5–20 per diary, for a total of $360. We conducted analyses using the phyloseq²¹ and vegan²² packages in R version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria) and Stata 17 (StataCorp, College Station, Texas), using two-sided tests and significance-level alpha=0.05.

RESULTS

Between December 16, 2014, and July 27, 2018, we approached 502 patients attending the PHSKC SHC, of whom 204 (41%) were ineligible, 187 (37%) declined, and 111 (22%) enrolled. The main reasons for declining were inability to stay to enroll (21%) or attend follow-up visits (42%). We excluded one participant (1%) who self-identified as a transgender woman due to insufficient sample size to stratify analyses by gender and two participants (2%) who reported sex with a female partner during follow-up. The remaining 108 cisgender MSM comprised the analysis sample.

Baseline Characteristics

Among these MSM with NGU, median age was 30 years (range=20–58), and most (62%) were White. Thirteen men (12%) were known to be living with HIV. Of men not previously diagnosed with HIV, 19 (20%) were taking PrEP. Over half of men reported a previous GC (56%) and/or CT (59%) infection. Most men had urethral symptoms (94%), visible urethral discharge (91%), and ≥10 PMNs/HPF (81%). All men were treated presumptively for NGU, including 101 (94%) with azithromycin, 6 (6%) with doxycycline, and 1 (1%) who reported contact to MG with moxifloxacin (Supplemental Table S2). One man (1%) was treated with ceftriaxone and one (1%) with gentamicin for proctitis. Based on urethral Aptima testing, 37 men (34%) had CT infection, 25 (23%) had MG infection, and three (3%) had CT/MG co-infection.

Sexual behaviors reported at baseline are summarized in Supplemental Table S3. Although all men reported exclusively male partners in the prior year, 45 (42%) reported vaginal sex in their lifetime.

Longitudinal Analyses

Ninety-six of the 108 men (89%) were included in the longitudinal analyses (12 men were lost to follow-up or withdrew after enrollment). Men included and excluded from the longitudinal analyses had similar characteristics, except excluded men were younger (P=0.05), more likely to be non-Hispanic Black (P=0.03), and had lower education (P=0.001). The 96 included men provided 1,140 person-weeks of behavioral data (median=86 days/man, IQR=82–92), completing 1,212 (93%) of 1,308 diaries.

Based on clinical interviews, 50 men (52%) had at least one additional antibiotic during follow-up; most were provided at the study clinic, although 10 men (10%) received an antibiotic from another clinic for non-STI infections. Azithromycin (28%), moxifloxacin (21%), and ceftriaxone (11%) were the antibiotics most commonly taken during follow-up.

These 96 men collected 1,370 urine specimens during the analysis period (n=377 [28%] clinic-collected, n=993 [72%] home-collected). Of these, 280 (20%) had insufficient bacterial DNA for sequencing, and three (<1%) had <1,000 sequence reads. Insufficient bacterial DNA was more common in clinic-collected than home-collected urine (P=0.004).

Among the 1,087 specimens with sufficient bacterial DNA and sequence reads (median=12 specimens/man, IQR=10–14), median Shannon index was 1.31 (IQR=0.73–1.99), and the median number of observed species was 14 (IQR=9–24). Overall, 397 specimens (37%) had oral indicator species (median=1 species, IQR=1–2), and 312 (29%) had rectal indicator species (median=1 species, IQR=1–2). Additionally, 13% of specimens were collected within 7 days of antibiotic therapy.

The most common urethral sexual exposures in the 7 days before specimen collection were IOS and CIAI together (29%), followed by IOS only (17%) and CIAI only (4%) (Table 1). Participants who reported no urethral sexual exposures in the prior 7 days were somewhat more likely to have had antibiotics in the prior 7 days than those reporting urethral sexual exposures (19% vs. 7–10%). Seventy-six men (79%) reported at least one new sex partner during follow-up.

Table 1.

Characteristics of participating MSM among longitudinal clinic- and home-collected specimens by specific urethral sexual exposures in the prior 7 days (2014–2018)

Characteristic	None N=483 n (%)	IOS only N=183 n (%)	CIAI only N=40 n (%)	Both IOS and CIAI N=317 n (%)	Unknown N=64 n (%)
Age group (years)
21–29	206 (43)	79 (43)	17 (43)	97 (31)	37 (58)
30–39	157 (33)	69 (38)	14 (35)	124 (39)	14 (22)
≥40	120 (25)	35 (19)	9 (23)	96 (30)	13 (20)
Race
White	312 (65)	105 (57)	19 (48)	213 (67)	38 (59)
Black	49 (10)	14 (8)	5 (13)	32 (10)	12 (19)
Other or unknown	122 (25)	64 (35)	16 (40)	72 (23)	14 (22)
Ethnicity
Not known Hispanic or Latino	385 (80)	147 (80)	30 (75)	252 (80)	51 (80)
Known Hispanic or Latino	98 (20)	36 (20)	10 (25)	65 (21)	13 (20)
HIV and PrEP status
HIV-negative, not on PrEP	324 (67)	131 (72)	23 (58)	218 (69)	31 (48)
HIV-negative, on PrEP	98 (20)	33 (18)	10 (25)	59 (19)	13 (20)
HIV-positive	61 (13)	19 (10)	7 (18)	40 (13)	20 (31)
Antibiotics in prior 7 days
No	393 (81)	165 (90)	37 (93)	278 (88)	12 (19)
Yes	90 (19)	18 (10)	3 (8)	22 (7)	9 (14)
Unknown	0 (0)	0 (0)	0 (0)	17 (5)	43 (67)

Abbreviation: CIAI, condomless insertive anal intercourse; IOS, insertive oral sex; MSM, men who have sex with men; PrEP, HIV pre-exposure prophylaxis.

Urethral Sexual Exposures and Community Composition

PERMANOVA tests for differences in the urethral microbiota by urethral sexual exposures (P=0.001) and antibiotic exposure (P=0.001) in the prior 7 days suggested there may be differences in the bacterial communities. However, in principal coordinates analyses, the two axes that explained the greatest percentage of variation in community composition (axes 1 and 2, Figure 1) did not demonstrate obvious separation based on recent sexual or antibiotic exposures. Example timelines for five men summarizing relative abundances of the 20 most abundant taxa, summary measures of urethral microbiota, and sexual and antibiotic exposures are available in Supplemental Figures S1–S5.

Figure 1.

Principal coordinates analysis plots based on the two axes that explain the greatest percentage of variation in the urethral community composition. Each point represents the bacterial community at a single timepoint in a single participating MSM (2014–2018). No separation in bacterial communities was noted based on (A) urethral sexual exposures in the prior 7 days and (B) antibiotic exposure in the prior 7 days.

Abbreviations: IAI, insertive anal intercourse; IOS, insertive oral sex; MSM, men who have sex with men.

Using GEE, compared to specimens with no urethral sexual exposures in the prior 7 days, those with IOS only in the prior 7 days had significantly higher Shannon index (coefficient=0.11, 95% confidence interval=0.011–0.213; P=0.03), adjusting for recent antibiotics and baseline age, race, ethnicity, and HIV/PrEP status (Figure 2, Supplemental Table S4). However, IOS only in the prior 7 days was not associated with log-10 number of species (observed, oral indicator, or rectal indicator). Neither CIAI only nor both IOS and CIAI was associated with any summary measures of community composition, nor was antibiotic exposure in the prior 7 days.

Figure 2.

Forest plots of the unadjusted and adjusted* association between urethral sexual exposures^† and antibiotic exposure^‡ in the prior 7 days and (A) Shannon index, (B) log-10 number of observed species, (C) log-10 number of oral indicator species, and (D) log-10 number of rectal indicator species in the urethral microbiota among MSM in the three months after NGU diagnosis (2014–2018)..

Abbreviations: CIAI, condomless insertive anal intercourse; IOS, insertive oral sex; MSM, men who have sex with men.

*Adjusted for the other time-varying exposure variable (i.e., urethral sexual exposures or antibiotic exposure in the prior 7 days), age, race, ethnicity, and baseline HIV and PrEP status.

†Referent group: no urethral sexual exposures.

‡Referent group: no antibiotics.

In sensitivity analyses considering the presence of oral and rectal indicator species, restricting to specimens without antibiotic exposure in the prior 7 days, and restricting to home-collected specimens, the results were generally similar (Supplemental Tables S4–S5, Supplemental Figures S6–S7). Among specimens without recent antibiotic exposure, IOS only in the prior 7 days was not associated with Shannon index, and CIAI only in the prior 7 days was inversely associated with log-10 number of observed species (P=0.05). Among home-collected specimens, IOS with CIAI was associated with log-10 number of oral indicator species (P=0.02).

Urethral Sexual Exposures and Specific Taxa

In analyses of 14 taxa previously associated with IOS and/or IAI,⁹ IOS only in the prior 7 days was positively associated with presence of H. parainfluenzae after adjusting for multiple comparisons (Table 2). Additionally, both IOS and CIAI in the prior 7 days was positively associated with H. parainfluenzae and Haemophilus.

Table 2.

Multivariable associations between recent urethral sexual exposures and the presence of specific tax in the urethral microbiota among MSM in the three months after NGU diagnosis (2014–2018)

		Presence (vs. absence) of specific taxa*
		Coefficient (95%CI)‡	Coefficient (95%CI)‡	Coefficient (95%CI)‡	Coefficient (95%CI)‡
Urethral sexual exposures in prior 7 days	Obs. N=1,006 † n (%)	Actinomyces turicensis	Corynebacterium pseudogenitalium/tuberculostearicum	Gardnerella vaginalis	Haemophilus parainfluenzae
None	483 (48)	Ref.	Ref.	Ref.	Ref.
IOS only	183 (18)	1.19 (0.91, 1.56)	1.06 (0.99, 1.13)	0.99 (0.70, 1.41)	1.40 (1.16, 1.69)
CIAI only	40 (4)	1.16 (0.56, 2.41)	0.91 (0.76, 1.07)	0.98 (0.39, 2.42)	0.74 (0.49, 1.14)
Both IOS and CIAI	300 (30)	0.88 (0.56, 1.38)	0.93 (0.86, 1.00)	1.29 (0.86, 1.91)	1.44 (1.17, 1.79)
		Peptoniphilus asaccharolyticus/grossensis/harei	Prevotella bivia	Prevotella buccalis	Prevotella timonensis
None	483 (48)	Ref.	Ref.	Ref.	Ref.
IOS only	183 (18)	0.79 (0.53, 1.18)	1.01 (0.77, 1.34)	1.02 (0.65, 1.60)	0.90 (0.64, 1.25)
CIAI only	40 (4)	0.75 (0.21, 2.63)	1.20 (0.67, 2.15)	0.58 (0.18, 1.82)	0.86 (0.49, 1.51)
Both IOS and CIAI	300 (30)	0.76 (0.54, 1.05)	1.22 (0.92, 1.63)	0.81 (0.54, 1.20)	0.99 (0.70, 1.41)
		Streptococcus anginosus group	Propionimicrobium lymphophilum	Corynebacterium	Haemophilus
None	483 (48)	Ref.	Ref.	Ref.	Ref.
IOS only	183 (18)	1.40 (1.03, 1.90)	1.08 (0.82, 1.42)	1.15 (0.84, 1.57)	1.70 (0.90, 3.18)
CIAI only	40 (4)	1.34 (0.83, 2.16)	1.05 (0.62, 1.78)	1.54 (0.92, 2.57)	0.62 (0.07, 5.90)
Both IOS and CIAI	300 (30)	1.16 (0.74, 1.83)	0.76 (0.49, 1.17)	1.28 (0.94, 1.74)	2.25 (1.24, 4.09)

Abbreviation: CI, confidence interval; CIAI, condomless insertive anal intercourse; IOS, insertive oral sex; MSM, men who have sex with men; NGU, non-gonococcal urethritis; PrEP, HIV pre-exposure prophylaxis.

^*Corynebacterium pseudogenitalium/tuberculostearicum has been referred to as Corynebacterium pseudogenitalium in other publications. Corynebacterium has been referred to as Corynebacterium sp NML108 in other publications. Haemophilus has been referred to as Haemophilus sp. HMSC71H0 in other publications. Actinomyces radingae was not detected and Streptococcus pseudopneumoniae could not be differentiated at the species level.

^†Multivariable models include 1,006 time-points with urethral microbiota assessment and data on sexual and antibiotic exposures in the prior 7 days from 96 men.

^‡Adjusted for antibiotic use in the prior 7 days, age, race, ethnicity, and HIV and PrEP status. Bold indicates statistical significance, based on P<0.05 using the false discovery rate adjustment method of Benjamini-Hochberg.

History of Vaginal Sex

In exploratory analyses of taxa previously detected only among men with a history of vaginal sex,⁹ L. iners was only detected in one specimen from a man with a history of vaginal sex (Supplemental Table S6). Detection of other taxa considered was similar in men with and without a history of vaginal sex, except for G. vaginalis which was more common among men who had never had vaginal sex (P=0.03).

DISCUSSION

Among a cohort of MSM attending a SHC in Seattle with NGU at baseline, PERMANOVA tests suggested there may be differences in the urethral microbiota based on recent (i.e., prior 7 days) urethral sexual exposures and antibiotic exposure; however, there was not obvious separation between groups based on the first two axes of principal coordinates analyses. In multivariable analyses adjusted for confounders, recent IOS only was associated with higher Shannon index but not with log-10 number of species (observed, oral indicator, or rectal indicator). Recent CIAI only, IOS with CIAI, and antibiotic exposure were not associated with any summary measure of the urethral microbiota. Of 14 specific taxa previously associated with IOS and/or IAI, recent IOS only was positively associated with H. parainfluenzae, while recent IOS and CIAI was positively associated with H. parainfluenzae and Haemophilus.

Our analyses suggest that, with the possible exception of recent IOS in the absence of CIAI, urethral sexual exposures do not have a substantial influence on the urethral microbiota among MSM in the timeframe that we studied. Although we had hypothesized that MSM would acquire highly diverse oral bacteria after IOS,²³ we did not observe an association between recent IOS only and number of oral indicator species, and the association between recent IOS only and higher diversity may have been due to chance. Findings from this longitudinal study among MSM are consistent with a recent cross-sectional study among predominantly heterosexual men, which similarly found little to no association between IOS and IAI and the urethral microbiota.⁹ Prior cross-sectional studies also did not detect associations between these exposures and specific bacteria/taxa^8,10,11 or community composition.¹² In contrast, vaginal sex has previously been associated with some bacteria/taxa^8,9 and community type.⁹ Additionally, studies have detected bacterial vaginosis (BV)-associated bacteria in the male urethra and coronal sulcus,^7,9,11,24 suggesting these bacteria can be acquired from or shared with female partners. Interestingly, though Toh et al. detected BV-associated bacteria only among men who had ever had vaginal sex,⁹ this was not the case in our study (except for one man with L. iners), perhaps due to differences in study populations, to prevalent vs. incident colonization, or to the existence of other routes of transmission for these bacteria. Toh et al. proposed that vaginal sex may be more likely to influence the urethral microbiota than IOS or IAI because the vagina is less often exposed to other sources of bacteria than the oral mucosa or gut.⁹ Alternatively, bacteria that inhabit the oral cavity or gut may not be adapted to survive in the urethra, whereas the vaginal and urethral environments are more similar.

Our observation that recent antibiotic exposure did not influence the urethral microbiota was unexpected. Studies of the microbiota at multiple anatomic sites suggest that changes in bacterial diversity following antibiotic treatment are common, though many effects differed across studies, anatomic sites, and/or antibiotics.²⁵ Although we excluded specimens collected in the first 7 days after enrollment, 142 specimens (13%) were still collected within 7 days of antibiotic exposure throughout follow-up. Prior work suggests that some shifts in microbial diversity may persist for 8 weeks to 4 months,²⁵ and the presumptive therapy for NGU provided at enrollment (typically azithromycin, with a half-life of 2–4 days) and interim antibiotics may have had lasting effects on the urethral microbiota. Longitudinal studies of the influence of antibiotic and sexual exposures on the urethral microbiota among men who have not received antibiotics within 6 months would be useful, as would daily sampling to assess shorter-term changes in the urethral microbiota.

Our study was strengthened by its longitudinal design, which documented the temporal sequence of exposures and outcomes allowing for better causal inference. Moreover, the weekly web-based and mobile phone-enabled diaries used to collect sexual behavior data minimized recall and social desirability bias^26,27 and had high response rates (93%). Our bioinformatics pipeline classified bacteria to the species-level in most cases,²⁸ allowing more accurate measures of urethral community composition and reducing bias in associations when grouped at higher taxonomic levels.

This study also had important limitations. First, 26% of urine specimens had insufficient bacterial DNA for sequencing analyses. Exclusion of these specimens from our statistical analyses may have biased our results if having insufficient bacterial DNA was not random. Home-collected urine specimens less often had insufficient bacterial DNA, perhaps because early morning first-void, first-catch urine has greater concentrations of bacteria than urine collected later in the day or because inadequate or inconsistent home storage and transport conditions contributed to bacterial growth. However, adjusting for clinic- vs. home-collection and restricting analyses to home-collected specimens did not meaningfully change our results. Second, we only evaluated select measures of the composition of the urethral microbiota. Sexual exposures may have influenced the microbial composition in other unmeasured ways. Future research considering the relative abundance of oral and/or rectal indicator species, as well as the presence, number, and/or relative abundance of other individual species, would be useful. Third, although we adjusted for recent antibiotic exposure and baseline characteristics previously associated with sexual behavior and the urethral microbiota, our estimates may be subject to residual confounding. In particular, although we adjusted for antibiotic therapy in the 7 days before specimen collection, we did not have day-level data on antibiotic type, and different antibiotics have different half-lives and may influence different aspects of the microbiota for different periods of time. Nevertheless, our findings were generally similar in analyses restricted to specimens without recent antibiotic exposure. Finally, we performed a large number of statistical tests, and the associations detected may be due to chance.

In conclusion, although recent IOS in the absence of CIAI was associated with higher urethral bacterial diversity among MSM with a recent history of NGU, this finding was not consistent across other measures of microbial community composition. We did not detect any evidence of an association between recent CIAI (with or without IOS) nor recent antibiotic exposure on the urethral microbiota of MSM. Similar studies of the impact of these exposures on the urethral microbiota among MSM without antibiotic exposure in the prior 6 months would be useful to determine whether antibiotic treatment for NGU had a lasting effect in these men. IOS and IAI may perturb the urethral microenvironment among MSM, but these data suggest that any effect on the microbial community as a whole is subtle.