Secondary Logo

Journal Logo

Original Studies

Anal Human Papillomavirus Prevalence Among Vaccinated and Unvaccinated Gay, Bisexual, and Other Men Who Have Sex With Men in Canada

Chambers, Catharine MSc∗,†; Deeks, Shelley L. MD∗,‡; Sutradhar, Rinku PhD∗,§; Cox, Joseph MD; de Pokomandy, Alexandra MDCM; Grennan, Troy MD∗∗; Hart, Trevor A. PhD∗,††; Lambert, Gilles MD‡‡; Moore, David M. MD§§; Coutlée, François PhD¶¶,∥∥; Burchell, Ann N. PhD∗,†,∗∗∗;  the Engage-HPV Study Team

Author Information
doi: 10.1097/OLQ.0000000000001560

Human papillomavirus (HPV) vaccination is recommended for gay, bisexual, and other men who have sex with men (GBM) because of their high burden of HPV-associated disease, such as anal cancer.1 Starting in 2015, GBM 26 years or younger who disclose same-sex activity to a healthcare provider have been eligible for publicly funded HPV vaccine under most provincial and territorial programs in Canada. Despite these targeted programs, vaccine uptake remains low: less than half of self-identified GBM 26 years or younger in major Canadian cities have received ≥1 dose.2

Human papillomavirus is the most common sexually transmitted infection (STI). Before the introduction of male HPV vaccination, the prevalence of anal infection with at least one HPV type was 50% to 70% among HIV-negative GBM and even higher at 80% to 90% among GBM living with HIV.3,4 These estimates are 2 to 4 times higher than men who have sex with women.3 Infection with vaccine-preventable HPV types, most notably oncogenic types 16 and 18, follow similar demographic patterns.3,5,6 The majority of HPV prevalence studies published to date were conducted among unvaccinated men; few studies have measured HPV infection among young GBM who are eligible for publicly funded HPV vaccination.7

Human papillomavirus vaccine is highly efficacious in males, preventing up to 85% of incident anal infections with vaccine-preventable types in young GBM 26 years or younger who were HPV-naive at baseline and had ≤5 lifetime sexual partners.8 Current vaccination guidelines are predicated on receipt of HPV vaccine before sexual exposure to confer maximum benefit. However, less information is known about how well this vaccine works in real-world settings among sexually active GBM who may have had multiple exposures to HPV before vaccination.9–11 Our objective was to estimate the prevalence of anal HPV infection among young, sexually active GBM soon after implementation of targeted HPV vaccination programs in Canada and to compare the anal prevalence of vaccine-preventable types between vaccinated and unvaccinated men.

MATERIALS AND METHODS

Setting

Since 2012, national immunization guidelines have recommended HPV vaccine for all males aged 9 to 26 years and GBM 9 years or older without upper age restriction.1 Under Canada’s publicly funded healthcare system, self-identified GBM aged 9 to 26 years have been eligible for free HPV vaccine as of September 2015 in British Columbia (BC),12 January 2016 in Quebec,13 and September 2016 in Ontario.14 Gay, bisexual, and other men who have sex with men 27 years and older who are ineligible for the publicly funded programs can purchase the vaccine or may have coverage through private insurance. Nine-valent (9vHPV) replaced quadrivalent (4vHPV) vaccine for the targeted programs starting in May 2017 (BC and Quebec) and September 2017 (Ontario).15 Three doses are recommended for immunocompetent males who initiate their vaccine series at age 15 years or older,15 which would apply to all participants in the current study based on their age at vaccination; in Quebec, 3 doses are recommended for males 18 years and older.13 Although all 3 provinces expanded their school-based HPV vaccination programs to be gender neutral as of September 2016 (Quebec and Ontario) and August 2017 (BC), participants in the current study were outside the eligible birth cohorts for these programs.

Study Participants and Recruitment

The Engage sexual health study is a prospective cohort of GBM, including cisgender and transgender men, in Montreal, Toronto, and Vancouver—3 cities with the largest GBM populations in Canada.16 At enrollment, men 16 years or older were eligible if they self-identified as a man, reported having sex with another man within the past 6 months, read English or French, and provided written informed consent. Participants were recruited between February 2017 and August 2019 using respondent-driven sampling (RDS), a form of chain-referral sampling that aims to generate more representative samples for hard-to-reach populations.17,18 Briefly, initial participants (or seeds) were purposively selected to represent subgroups of the GBM community based on age group, gender, ethnocultural background, and HIV status. Seeds were provided with up to 6 coupons each, which were used to recruit members of their social networks. Recruitment continued through waves until the target sample size was reached in each city (Supplemental Table S1, https://links.lww.com/OLQ/A755). For the present study, participants received $60 CAD plus $15 CAD for each additional eligible participant recruited. Protocols were approved by research ethics boards at participating institutions.

Data Collection

Participants self-completed a questionnaire by computer-assisted self-interview including items on demographics and socioeconomic status; gender and sexual orientation; medical history, including history of STI and blood-borne infection testing; HIV status; lifetime and recent (past 6 months) sexual behaviors; and substance use. Questionnaire items were informed by the Sexual Health Framework and the Global AIDS Monitoring Indicators.19,20 Human papillomavirus vaccine-specific questionnaire items included the following: awareness of HPV vaccine and willingness to be vaccinated, HPV vaccination history, lifetime number of doses, health service location of most recent dose of HPV vaccine, and age at first dose.

At enrollment, young GBM aged 16 to 30 years were invited to provide an anal specimen for HPV genotyping. The HPV substudy was an optional add-on component to the larger Engage Cohort Study. Anal specimens were self-collected at study sites using moistened Dacron swabs inserted 3 to 5 cm into the anal canal using illustrated instructions.21 Samples were kept at +4°C and transported to the laboratory under wet ice conditions.

HPV DNA Genotyping

All specimens were screened for HPV DNA using an in-house generic probe assay.22 Human papillomavirus DNA–positive specimens were genotyped using the polymerase chain reaction–based Linear Array (Roche Molecular Systems) for the L1 HPV gene that detects 36 mucosal HPV genotypes, including all 9 vaccine-preventable types and 2 variants of HPV-82.23 This assay has been shown to have good agreement with conventional research-based assays (mean, 96.4% ± 2.4%; range, 86%–100% by HPV type) and greater sensitivity.23 Coamplification of a β-globin human DNA sequence was performed to assess specimen adequacy. Analyses were restricted to valid anal specimens, defined as those with detection of either β-globin or HPV DNA.

HPV Vaccination Status

Vaccination status was defined as self-reported receipt of ≥1 lifetime HPV vaccine dose before study enrollment. Participants were asked if they had ever heard of the HPV vaccine and, if yes, to report if they had ever received 1 or more doses. Those who were unaware of the HPV vaccine were assumed to be unvaccinated, whereas those who reported an unknown HPV vaccination history were excluded. Alternative definitions of vaccination completion (≥2 doses or all 3 doses) were explored in sensitivity analyses. Time since vaccination was derived as the difference (in years) between self-reported age at first dose and age at enrollment; biological male participants reporting vaccination before the Canadian approval of the 4vHPV vaccine for males in February 2010 were excluded from analyses of vaccination timing.24 Data on the type of vaccine or timing of doses were not collected; we assumed that most vaccinated GBM received 4vHPV based on the earliest availability of 9vHPV in each province relative to self-reported age at vaccination.

Analyses

Type-specific HPV prevalence at enrollment was estimated for each genotype, along with composite outcomes for ≥1 4vHPV-preventable type (HPV-6/11/16/18) or 9vHPV-preventable type (HPV-6/11/16/18/31/33/45/52/58). We derived weighted HPV prevalence estimates and 95% confidence intervals (CIs) pooled across the 3 cities and accounting for strata by city. Because RDS relies on chain-referral sampling, individuals who have smaller social networks will be underrepresented in the RDS sample, whereas those who have larger social networks will be overrepresented. For this reason, prevalence estimates were weighted using RDS-II weights (Volz-Heckathorn estimator), which are inversely proportional to a participant's self-reported social network size within each city, to adjust for selection biases inherent to chain-referral sampling.25 Self-reported network size was truncated at a minimum of 1, based on study eligibility criteria, and a maximum of 150, based on the maximum number of possible current relationships.26 All other results are RDS unweighted, unless otherwise specified.

Prevalence ratios (PRs) and 95% CI comparing anal HPV prevalence between vaccinated and unvaccinated GBM were estimated using a modified Poisson regression with robust standard errors for binary outcomes including RDS seeds.27,28 For the focal comparison of HPV prevalence by vaccination status, we calculated PRs without RDS-II weights.29 However, because there is no agreed-upon method for multivariable regression using RDS data,30,31s we report RDS-weighted PRs in Supplemental Digital Content, https://links.lww.com/OLQ/A755.29 Because HPV genotyping results were only available for a subset of participants within each city, we did not account for potential clustering within RDS recruitment chains.31s

Potential confounders for multivariable regression models were identified based on prior literature and informed by directed acyclic graphs.32s,33s All potential confounders independently associated with both HPV infection and HPV vaccine uptake at a P value of <0.25 were considered. A backward elimination procedure was used to reduce the number of covariates included in the final multivariable models.34s,35s Age group and city were included in all models regardless of statistical significance. Final models were adjusted for age group (based on eligibility for publicly funded vaccine at enrollment), city, education (indicator for socioeconomic status), smoking history, self-reported STI diagnosis in lifetime (excluding HIV and anogenital warts), and number of condomless receptive anal sex encounters in the past 6 months (based on quintiles). Sensitivity analyses explored vaccination status based on self-reported number of doses, time since vaccination (restricted vaccinated participants to those who were vaccinated >1 or >2 years before enrollment), and age at first dose. We also conducted a sensitivity analysis restricted to participants who had ≥1 prevalent HPV infection (i.e., excluding participants who were HPV DNA negative) and compared the prevalence of vaccine-preventable types with nonvaccine types to control for HPV exposure risk. All analyses were conducted in SAS (Cary, NC). A P value of α < 0.05 was considered statistically significant.

RESULTS

Participant Characteristics

Of the 1003 eligible GBM aged 16 to 30 years at enrollment, 847 (84.4%) provided an anal specimen for HPV genotyping. Participants who provided anal specimens did not significantly differ from those who did not on any covariate, except for city (data not shown). Participants with anal specimens were more likely to be from Montreal (90.1%) or Toronto (86.3%) compared with Vancouver (75.2%) because of delays in introducing the optional collection of specimens for HPV testing at some sites. Of the 847 anal specimens available for genotyping, 645 (76.2%) were considered valid, including 402 (47.5%) that were β-globin positive. Human papillomavirus vaccine uptake did not significantly differ between participants who did and did not provide valid specimens (40.3% vs. 37.0%, P = 0.428).

Of the 645 participants included in the analysis, the median age was 26 years (interquartile range [IQR], 24–28 years; Table 1). Most participants were from Montreal (n = 270; 41.9%), with fewer from Toronto (n = 171; 26.5%) or Vancouver (n = 204; 31.6%). Most participants (80.5%) self-identified as gay; 10 (1.6%) identified as transgender. Few (n = 38; 5.9%) had a laboratory-confirmed HIV infection. Almost all participants (98.3%) reported having anal sex with a man in their lifetime. The median age at first anal sex with a man was 18 years (IQR, 16–20 years). Almost two-thirds (62.8%) reported engaging in condomless receptive anal sex in the past 6 months; the median number of times was 5 (IQR, 2–15 times) among participants who reported ≥1 episode. Almost one-fifth (18.5%) had a self-reported history of anogenital warts.

TABLE 1 - Baseline Characteristics of Gay, Bisexual, and Other Men Who Have Sex With Men Aged 16 to 30 Years by Self-Reported HPV Vaccination Status, Engage Cohort Study, 2017 to 2019
Characteristic Overall* (n = 645) Unvaccinated (n = 363) Vaccinated (n = 245) P
Age group at enrollment, n (%) <0.001
 16–26 y 354 (54.9) 175 (48.2) 156 (63.7)
 27–30 y 291 (45.1) 188 (51.8) 89 (36.3)
City, n (%) 0.048
 Montreal 270 (41.9) 163 (44.9) 87 (35.5)
 Toronto 171 (26.5) 95 (26.2) 68 (27.8)
 Vancouver 204 (31.6) 105 (28.9) 90 (36.7)
Education, n (%) 0.144
 High school or less§ 122 (18.9) 74 (20.4) 40 (16.3)
 Postsecondary 412 (63.9) 219 (60.3) 167 (68.2)
 Graduate or professional degree 111 (17.2) 70 (19.3) 38 (15.5)
Ethnicity, n (%) 0.170
 English or French Canadian 295 (45.7) 150 (41.3) 123 (50.2)
 Other European 122 (18.9) 69 (19.0) 50 (20.4)
 Asian 73 (11.3) 46 (12.7) 23 (9.4)
 Black, African, Caribbean 18 (2.8) 11 (3.0) 7 (2.9)
 Indigenous 9 (1.4) 7 (1.9) 2 (0.8)
 Other or mixed 128 (19.8) 80 (22.0) 40 (16.3)
Sexual orientation, n (%) 0.024
 Gay 519 (80.5) 293 (80.7) 198 (80.8)
 Queer 69 (10.7) 35 (9.6) 32 (13.1)
 Bisexual 30 (4.7) 21 (5.8) 8 (3.3)
 Other 27 (4.2) 14 (3.9) 7 (2.9)
Has a regular current partner, n (%) 301 (46.7) 171 (47.1) 112 (45.7) 0.736
Laboratory-confirmed HIV infection, n (%) 38 (5.9) 18 (5.0) 15 (6.2) 0.529
Self-reported STI diagnosis, lifetime, n (%) 353 (55.9) 166 (47.2) 170 (70.0) <0.001
Smoking history, lifetime, n (%) 0.030
 Never smoker 199 (31.1) 100 (27.8) 87 (36.0)
 Current smoker 299 (46.8) 184 (51.1) 98 (40.5)
 Former smoker 141 (22.1) 76 (21.1) 57 (23.6)
Alcohol risk, past 6 mo, n (%)** 0.082
 Lower risk 392 (63.3) 225 (64.7) 146 (62.1)
 Moderate risk 190 (30.7) 97 (27.9) 80 (34.0)
 High risk 37 (6.0) 26 (7.5) 9 (3.8)
Any illicit drug use, lifetime, n (%) 512 (80.5) 276 (77.3) 205 (84.7) 0.026
Poppers use, lifetime, n (%) 372 (58.4) 189 (52.8) 162 (66.7) 0.001
Male anal sex partners, past 6 mo, n (%) <0.001
 0–1 partners 147 (22.8) 97 (26.7) 40 (16.3)
 2–5 partners 248 (38.4) 147 (40.5) 85 (34.7)
 6–10 partners 121 (18.8) 65 (17.9) 54 (22.0)
 >10 partners 129 (20.0) 54 (14.9) 66 (26.9)
Condomless receptive anal sex, past 6 mo, n (%) 0.003
 0 times 240 (37.2) 153 (42.1) 73 (29.8)
 1–2 times 123 (19.1) 77 (21.2) 42 (17.1)
 3–5 times 100 (15.5) 48 (13.2) 45 (18.4)
 6–15 times 88 (13.6) 41 (11.3) 42 (17.1)
 >15 times 94 (14.6) 44 (12.1) 43 (17.6)
Rimming (received), past 6 mo, n (%) 519 (80.5) 287 (79.1) 205 (83.7) 0.156
Fisting (received), past 6 mo, n (%) 28 (4.3) 14 (3.9) 12 (4.9) 0.534
RDS network size, median (IQR)†† 30 (15–55) 25 (10–50) 35 (20–78) <0.001
RDS-II weights, median (IQR)‡‡ 0.53 (0.28–1.06) 0.63 (0.31–1.39) 0.45 (0.18–0.79) <0.001
*Overall column includes 37 participants with missing data for HPV vaccination status.
Self-reported receipt of ≥1 dose of HPV vaccine.
P value comparing participant characteristics between unvaccinated and vaccinated participants from χ2 test.
§Includes participants with trade, vocational, or technical institute training.
Other sexual orientations include straight, pansexual, or other.
Excludes HIV and anogenital warts.
**Alcohol risk classified according to the World Health Organization's Alcohol, Smoking and Substance Involvement Screening Test as lower (scores 0–10), moderate (scores 11–26), or high (scores ≥27) risk.
††Based on response to “How many men who have sex with men aged 16 years or older, including trans men, do you know who live or work in the [city] metropolitan area (whether they identify as gay or otherwise)? This includes gay/bi guys you see or speak to regularly, e.g. close friends, boyfriends, spouses, regular sex partners, roommates, relatives, people you regularly hang out with, etc.” Values truncated at a minimum of 1 and a maximum of 150.
‡‡RDS-II weights are inversely proportional to self-reported RDS network size within each city; RDS-II weights sum to total number of participants in each city with a mean of 1.00.
HPV indicates human papillomavirus; IQR, interquartile range; RDS, respondent-driven sampling; STI, sexually transmitted infection.

HPV Vaccine Uptake

Of the 608 of 645 participants (94.3%) with known vaccination history, 245 (40.3%) self-reported receiving ≥1 dose of HPV vaccine; the corresponding RDS-weighted proportion was 30.6%. Among vaccinated participants, 34 (15.5%) received 1 dose, 50 (22.7%) received 2 doses, and 136 (61.8%) received all 3 recommended doses; 25 (10.2%) vaccinated participants had an unknown number of doses. The median age at first dose was 23 years (IQR, 21–25 years); 12 (4.9%) vaccinated participants had missing or invalid data for age at first dose. The majority (88.8%) of vaccinated participants, including 69.4% of participants who were aged 27 to 30 years at enrollment, were vaccinated before age 27 years as part of publicly funded programs. The median time from HPV vaccination to study enrollment was 2 years (IQR, 1–3 years). Only 9 (3.9%) participants reported receiving their first dose of HPV vaccine before their first anal sexual episode; the median time from first anal sex to HPV vaccination was 5 years (IQR, 2–8 years).

Anal HPV Prevalence

Overall RDS-weighted anal HPV prevalence was 66.7% (95% CI, 59.7%–73.6%) for ≥1 tested HPV type, 35.1% (95% CI, 28.7%–41.5%) for ≥1 9vHPV-preventable type, and 25.4% (95% CI, 19.5%–31.3%) for ≥1 4vHPV-preventable type (Table 2). Among unvaccinated participants (n = 363), the corresponding prevalence estimates were 63.2% (95% CI, 53.7%–72.8%), 35.9% (95% CI, 27.6%–44.1%), and 26.4% (95% CI, 18.9%–34.0%), respectively. Vaccine-preventable types HPV-16 (RDS-weighted, 12.0%) and HPV-6 (9.1%) were among the most commonly detected, along with HPV-51 (10.9%) and HPV-39 (8.0%) in unvaccinated participants; lower prevalence was seen for 4vHPV types HPV-11 (6.3%) and HPV-18 (2.9%; Fig. 1). Vaccine-preventable types were less common overall among vaccinated participants (n = 245) but remained high at 30.4% (95% CI, 19.9%–41.0%) for ≥1 9vHPV type and 19.8% (95% CI, 10.6%–29.0%) for ≥1 4vHPV type. Type-specific estimates were 8.7% for HPV-11, 8.2% for HPV-16, 7.9% for HPV-18, and 5.9% for HPV-6 in vaccinated participants.

TABLE 2 - Anal HPV Prevalence by Self-Reported Vaccination Status Among Gay, Bisexual, and Other Men Who Have Sex With Men Aged 16 to 30 Years, Engage Cohort Study, 2017 to 2019
HPV Type Overall* (n = 645) Unvaccinated (n = 363) Vaccinated (n = 245)
n Sample % Weighted% (95% CI) n Sample % Weighted % (95% CI) n Sample % Weighted % (95% CI)
≥1 HPV type 471 73.0 66.7 (59.7–73.6) 259 71.3 63.2 (53.7–72.8) 184 75.1 74.0 (65.6–82.5)
≥1 4vHPV type 167 25.9 25.4 (19.5–31.3) 102 28.1 26.4 (18.9–34.0) 54 22.0 19.8 (10.6–29.0)
 HPV-6 65 10.1 8.2 (5.0–11.5) 38 10.5 9.1 (4.3–13.8) 21 8.6 5.9 (2.6–9.2)
 HPV-11 37 5.7 8.4 (4.3–12.4) 27 7.4 6.3 (2.9–9.7) 7 2.9 8.7 (0.0–17.6)
 HPV-16 70 10.9 11.0 (6.4–15.6) 42 11.6 12.0 (6.2–17.8) 25 10.2 8.2 (0.1–16.2)
 HPV-18 29 4.5 4.5 (1.7–7.4) 14 3.9 2.9 (0.9–5.0) 13 5.3 7.9 (0.0–16.2)
≥1 additional 9vHPV type 104 16.1 14.0 (9.7–18.3) 58 16.0 13.4 (8.6–18.3) 38 15.5 12.7 (5.0–20.4)
 HPV-31 21 3.3 3.3 (1.2–5.5) 13 3.6 4.1 (0.9–7.4) 6 2.4 1.5 (0.0–3.2)
 HPV-33 13 2.0 2.3 (0–4.6) 6 1.7 1.2 (0.0–2.4) 5 2.0 1.3 (0.0–3.0)
 HPV-45 29 4.5 4.3 (1.4–7.2) 16 4.4 2.5 (0.8–4.3) 9 3.7 4.1 (0.0–9.5)
 HPV-52 33 5.1 4.4 (1.7–7.2) 20 5.5 3.9 (1.4–6.4) 12 4.9 2.9 (0.6–5.2)
 HPV-58 28 4.3 4.7 (1.7–7.8) 12 3.3 3.7 (1.0–6.3) 14 5.7 4.7 (0.0–10.1)
≥1 9vHPV type 236 36.6 35.1 (28.7–41.5) 139 38.3 35.9 (27.6–44.1) 79 32.2 30.4 (19.9–41.0)
≥1 non-9vHPV type 394 61.1 52.3 (45.4–59.2) 212 58.4 49.0 (40.0–58.0) 160 65.3 62.0 (51.9–72.2)
4vHPV types include HPV-6/11/16/18; 9vHPV types include HPV-6/11/16/18/31/33/45/52/58.
*Overall column includes 37 participants with missing data for HPV vaccination status.
Self-reported receipt of ≥1 dose of HPV vaccine.
Prevalence estimates weighted using RDS-II weights to account for the RDS recruitment approach.
CI indicates confidence interval; HPV, human papillomavirus.

Figure 1
Figure 1:
Type-specific anal HPV prevalence among unvaccinated (A) and vaccinated (B) gay, bisexual, and other men who have sex with men aged 16 to 30 years, Engage Cohort Study, 2017 to 2019. HPV indicates human papillomavirus.

In general, prevalence estimates did not significantly differ by city (Table 3 and Supplemental Table S2, https://links.lww.com/OLQ/A755). In unweighted analyses, anal prevalence of vaccine-preventable types was significantly higher among current or former smokers and participants who reported a higher number of anal sex partners in the past 6 months, engaged in condomless receptive anal sex in the past 6 months, and self-reported an STI diagnosis in their lifetime (Table 3). In RDS-weighted analyses, only number of anal sex partners in the past 6 months remained statistically significant. Associations were similar among unvaccinated and vaccinated participants, although some failed to reach statistical significance because of the smaller sample size in stratified analyses (data not shown).

TABLE 3 - Anal HPV Prevalence by Covariates Among Gay, Bisexual, and Other Men Who Have Sex With Men Aged 16 to 30 Years, Engage Cohort Study, 2017 to 2019
Characteristic N ≥1 4vHPV Type ≥1 9vHPV Type
n Sample % Weighted %* (95% CI) n Sample % Weighted %* (95% CI)
Overall 645 167 25.9 25.4 (19.5–31.3) 236 36.6 35.1 (28.7–41.5)
Age group at enrollment
 16–26 y 354 90 25.4 25.9 (18.1–33.8) 122 34.5 34.6 (26.2–43.1)
 27–30 y 291 77 26.5 24.7 (15.8–33.5) 114 39.2 35.7 (26.1–45.3)
City
 Montreal 270 71 26.3 27.3 (18.3–36.3) 99 36.7 33.8 (24.5–43.0)
 Toronto 171 49 28.7 27.3 (14.9–39.8) 68 39.8 43.5 (29.3–57.8)
 Vancouver 204 47 23.0 21.2 (11.6–30.7) 69 33.8 29.8 (19.2–40.4)
Education
 High school or less 122 27 22.1 19.0 (6.1–32.0) 37 30.3 30.7 (16.1–45.3)
 Postsecondary 412 112 27.2 25.8 (18.6–33.0) 160 38.8 35.4 (27.4–43.3)
 Graduate or professional degree 111 28 25.2 32.6 (17.6–47.5) 39 35.1 40.2 (25.3–55.1)
Ethnicity
 English or French Canadian 295 74 25.1 20.6 (12.9–28.3) 107 36.3 28.7 (19.9–37.5)
 Other European 122 33 27.0 22.0 (11.7–32.4) 47 38.5 31.9 (19.3–44.6)
 Asian 73 16 21.9 28.0 (8.7–47.3) 24 32.9 41.6 (22.7–60.4)
 Black, African, Caribbean 18 5 27.8 44.5 (0.0–89.5) 5 27.8 44.5 (0.0–89.5)
 Indigenous 9 3 33.3 64.6 (19.9–100.0) 3 33.3 64.6 (19.9–100.0)
 Other or mixed 128 36 28.1 29.3 (16.6–42.1) 50 39.1 42.4 (28.8–55.9)
Self-identifies as gay
 No 126 26 20.6 22.4 (8.4–36.3) 39 31.0 33.5 (19.0–48.1)
 Yes 519 141 27.2 26.2 (19.7–32.7) 197 38.0 35.5 (28.4–42.6)
Has a regular current partner
 No 344 95 27.6 27.7 (18.6–36.7) 129 37.5 35.3 (25.9–44.7)
 Yes 301 72 23.9 22.8 (15.6–29.9) 107 35.5 34.9 (26.5–43.3)
Laboratory-confirmed HIV infection
 Negative 603 154 25.5 24.2 (18.3–30.0) 216 35.8 34.0 (27.5–40.4)
 Positive 38 12 31.6 40.4 (10.8–70.0) 19 50.0 49.9 (22.3–77.6)
Self-reported STI diagnosis, lifetime§
 No 279 66 23.7 23.7 (16.0–31.3) 86 30.8 30.1 (21.9–38.4)
 Yes 353 97 27.5 28.0 (18.6–37.3) 144 40.8 40.1 (30.2–50.0)
Smoking history, lifetime
 Never smoker 199 41 20.6 20.0 (9.9–30.2) 62 31.2 31.3 (20.0–42.6)
 Current smoker 299 83 27.8 29.0 (19.7–38.2) 106 35.5 36.2 (26.6–45.9)
 Former smoker 141 42 29.8 28.7 (17.2–40.2) 65 46.1 40.7 (27.8–53.7)
Alcohol risk, past 6 mo
 Lower risk 392 97 24.7 22.8 (16.0–29.7) 137 34.9 34.7 (26.9–42.5)
 Moderate risk 190 53 27.9 27.2 (16.2–38.3) 77 40.5 33.9 (22.3–45.5)
 High risk 37 9 24.3 60.1 (31.8–88.5) 11 29.7 65.7 (40.5–90.8)
Any illicit drug use, lifetime
 No 124 27 21.8 19.6 (8.2–30.9) 39 31.5 30.2 (17.2–43.3)
 Yes 512 138 27.0 28.3 (21.5–35.0) 195 38.1 37.8 (30.7–44.9)
Poppers use, lifetime
 No 265 62 23.4 21.8 (13.8–29.8) 88 33.2 31.8 (22.8–40.9)
 Yes 372 103 27.7 30.7 (22.1–39.4) 146 39.2 40.6 (31.8–49.4)
Male anal sex partners, past 6 mo
 0–1 partners 147 26 17.7 16.7 (7.5–25.9) 37 25.2 25.8 (15.0–36.6)
 2–5 partners 248 67 27.0 29.2 (19.3–39.0) 95 38.3 38.9 (28.6–49.3)
 6–10 partners 121 30 24.8 19.8 (9.1–30.5) 41 33.9 28.8 (15.4–42.2)
 >10 partners 129 44 34.1 41.3 (24.4–58.2) 63 48.8 53.5 (37.8–69.3)
Condomless receptive anal sex, past 6 mo
 0 times 240 59 24.6 24.4 (15.2–33.5) 74 30.8 30.1 (20.5–39.7)
 1–2 times 123 28 22.8 18.8 (9.3–28.4) 41 33.3 30.4 (17.7–43.1)
 3–5 times 100 31 31.0 37.8 (21.2–54.3) 36 36.0 47.0 (30.9–63.0)
 6–15 times 88 23 26.1 28.8 (10.6–47.1) 41 46.6 46.8 (28.6–65.0)
 >15 times 94 26 27.7 23.0 (8.4–37.6) 44 46.8 40.8 (24.4–57.2)
Rimming (received), past 6 mo
 No 126 28 22.2 26.6 (12.7–40.6) 38 30.2 32.6 (18.1–47.1)
 Yes 519 139 26.8 24.9 (18.8–31.1) 198 38.2 36.0 (29.2–42.9)
Fisting (received), past 6 mo
 No 617 158 25.6 25.3 (19.3–31.3) 222 36.0 35.2 (28.7–41.7)
 Yes 28 9 32.1 28.1 (0.0–56.8) 14 50.0 32.7 (2.7–62.8)
4vHPV types include HPV-6/11/16/18; 9vHPV types include HPV-6/11/16/18/31/33/45/52/58.
*Prevalence estimates weighted using RDS-II weights to account for the RDS recruitment approach.
Includes participants with trade, vocational, or technical institute training.
Versus other sexual orientations including bisexual, queer, straight, pansexual, and other.
§Excludes HIV and anogenital warts.
Alcohol risk classified according to the World Health Organization's Alcohol, Smoking and Substance Involvement Screening Test as lower (scores 0–10), moderate (scores 11–26), or high (scores ≥27) risk.
CI indicates confidence interval; HPV, human papillomavirus.

PRs by HPV Vaccination Status

In unadjusted models, anal prevalence of vaccine-preventable types did not significantly differ between vaccinated and unvaccinated participants (Table 4). After adjustment for potential confounders, RDS-unweighted PRs for receipt of ≥1 dose were 0.73 (95% CI, 0.54–1.00) for 4vHPV types and 0.72 (95% CI, 0.57–0.91) for 9vHPV types. Findings were similar when vaccination status was defined as either ≥2 or 3 doses and for RDS-weighted analyses (Supplemental Table S3, https://links.lww.com/OLQ/A755).

TABLE 4 - RDS-Unweighted Prevalence Ratios for Anal HPV Infection With Vaccine-Preventable Types Comparing Vaccinated With Unvaccinated Gay, Bisexual, and Other Men Who Have Sex With Men Aged 16 to 30 Years, Engage Cohort Study, 2017 to 2019
Model* Anal HPV Prevalence Unadjusted PR (95% CI) Adjusted PR (95% CI)
n/N %
4vHPV types
 Unvaccinated 98/349 28.1 Ref Ref
 Vaccinated (≥1 dose) 53/241 22.0 0.78 (0.59–1.05) 0.73 (0.54–1.00)
 No. doses
  ≥2 doses§ 43/184 23.4 0.83 (0.61–1.14) 0.77 (0.55–1.07)
  3 doses 31/136 22.8 0.81 (0.57–1.15) 0.75 (0.52–1.10)
 Time since vaccination
  >1 y before enrollment 28/121 23.1 0.82 (0.57–1.19) 0.77 (0.52–1.13)
  >2 y before enrollment 9/61 14.8 0.53 (0.28–0.98) 0.47 (0.25–0.86)
 Age at first dose
  ≤23 y 23/118 19.5 0.69 (0.46–1.04) 0.64 (0.42–0.99)
  >23 y 28/112 25.0 0.89 (0.62–1.28) 0.82 (0.55–1.20)
9vHPV types
 Unvaccinated 134/349 38.4 Ref Ref
 Vaccinated (≥1 dose) 76/241 31.5 0.82 (0.65–1.03) 0.72 (0.57–0.91)
 No. doses
  ≥2 doses§ 63/184 34.2 0.89 (0.70–1.13) 0.76 (0.59–0.98)
  3 doses 44/136 32.4 0.84 (0.64–1.11) 0.70 (0.52–0.94)
 Time since vaccination
  >1 y before enrollment 40/121 33.1 0.86 (0.65–1.15) 0.73 (0.54–0.99)
  >2 y before enrollment 16/61 26.2 0.68 (0.44–1.06) 0.55 (0.36–0.85)
 Age at first dose
  ≤23 y 34/118 28.8 0.75 (0.55–1.03) 0.69 (0.49–0.96)
  >23 y 40/112 35.7 0.93 (0.70–1.23) 0.76 (0.56–1.02)
4vHPV types include HPV-6/11/16/18; 9vHPV types include HPV-6/11/16/18/31/33/45/52/58.
*Analyses restricted to 590 (97.0%) participants who had complete data for all covariates.
Analyses adjusted for age group, city, education, lifetime smoking history, lifetime history of STIs (excluding HIV and anogenital warts), and number of condomless receptive anal sex encounters in the past 6 months.
HPV vaccination status defined as self-reported receipt of ≥1 dose or unknown number of doses (vs. 0 doses).
§HPV vaccination status defined as self-reported receipt of ≥2 doses (vs. 0 doses); excludes participants who self-reported 1 dose or had an unknown number of doses.
HPV vaccination status defined as self-reported receipt of 3 doses (vs. 0 doses); excludes participants who self-reported 1 or 2 doses or had an unknown number of doses.
CI indicates confidence interval; HPV, human papillomavirus; PR, prevalence ratio; RDS, respondent-driven sampling.
Boldface text indicates statistically significant results (P < 0.05).

Prevalence ratios were comparable when vaccinated participants were restricted to those vaccinated >1 year before enrollment (4vHPV: adjusted PR [aPR], 0.77 [95% CI, 0.52–1.13]; 9vHPV: aPR, 0.73 [95% CI, 0.54–0.99]) but departed further from the null when restricted to those vaccinated >2 years ago (4vHPV: aPR, 0.47 [95% CI, 0.25–0.86]; 9vHPV: aPR, 0.55 [95% CI, 0.36–0.85]). We observed lower PRs among men who received their first vaccine dose at age ≤23 years (median age at first dose in sample) compared with those vaccinated at age >23 years, although CIs overlapped: 0.64 (95% CI, 0.42–0.99) versus 0.82 (95% CI, 0.55–1.20) for 4vHPV types and 0.69 (95% CI, 0.49–0.96) versus 0.76 (95% CI, 0.56–1.02) for 9vHPV types. Restricting the analysis to participants who had ≥1 prevalent HPV infection, the PRs were 0.77 (95% CI, 0.57–1.03) for 4vHPV types and 0.76 (95% CI, 0.61–0.94) for 9vHPV types.

DISCUSSION

In this sexually active cohort of young GBM, anal HPV infection with vaccine-preventable types was highly prevalent. After accounting for RDS recruitment, we estimated that more than one-third of unvaccinated GBM aged 16 to 30 years living in Canada’s 3 largest cities were infected with 9vHPV-preventable types and more than one-quarter were infected with 4vHPV-preventable types. These estimates are comparable with prior studies conducted in the prevaccine era that measured anal HPV prevalence against vaccine-preventable types among young, HIV-negative GBM recruited in community settings,36s–39s but are lower than prevalence estimates from clinic-based samples.40s–44s Anal HPV prevalence remained high in our sample despite targeted HPV vaccination programs for GBM that were implemented 2 to 4 years before study enrollment with 40% vaccine uptake.

We did not find a statistically significant difference in HPV prevalence between vaccinated and unvaccinated GBM. Potential explanations for this nonsignificant finding include lack of vaccine effectiveness against prevalence outcomes in previously infected men, confounding between exposure groups, or differential misclassification of HPV vaccination history. Vaccinated GBM were more likely to engage in condomless receptive anal sex and self-report a prior STI diagnosis, suggesting that targeted vaccination efforts are likely reaching those most at risk for HPV exposure.2 After adjusting for this confounding between exposure groups, we found that the prevalence of anal infection with vaccine-preventable types was about 30% lower in participants who received ≥1 dose. As expected, this vaccine effectiveness estimate is lower than the vaccine efficacy of 84% (95% CI, 69%–93%) against incident 4vHPV detection measured in clinical trials among GBM who were HPV-naive and had received a complete 3-dose series.8 Our results are more comparable with the observed efficacy of 49% (95% CI, 32%–61%) in the intent-to-treat sample of GBM who may have been previously infected with HPV and received ≥1 dose.8 In that analysis, more than one-quarter of participants had evidence of infection with 4vHPV types before vaccination.8 Differences in study populations should be taken into account, including younger age, limited number of sexual partners, and absence of anogenital warts/lesions and HIV infection among clinical trial participants.

To our knowledge, only one other observational study has measured real-world HPV vaccine effectiveness in this population. In a convenience sample of GBM aged 18 to 26 years recruited at community centers or clinics in 3 US cities during the period 2016–2018, Meites et al.9 found that the prevalence of ≥1 4vHPV type in anal and/or oral specimens was about 30% lower among GBM who self-reported receiving ≥1 HPV vaccine dose. Higher vaccine effectiveness (~60%) was observed in GBM who received their first dose at age ≤18 years.9 In the HYPER2 study in Australia, Chow et al.7 found a significantly lower anal prevalence of 4vHPV types in GBM who were eligible for universal school-based vaccination (vaccine uptake of ≥1 dose, 75%) compared with a vaccine-ineligible cohort (7% vs. 28%; aPR, 0.24; 95% CI, 0.14–0.42). In females, early evidence of HPV vaccine impact in real-world settings has been observed, with vaccine effectiveness estimates for ≥1 dose against prevalent infection with 4vHPV types ranging from 36% to 90% (compared with >90% observed in clinical trials), with greater impact in younger cohorts vaccinated before HPV exposure and with high vaccine uptake.45s–48s

In our cross-sectional analysis, the timing of vaccination relative to HPV acquisition was unknown. Human papillomavirus outcomes may be capturing prevalent infections present at the time of HPV vaccination rather than incident infection acquired afterward. Because most individuals will acquire an incident HPV infection shortly after sexual debut,49s,50s participants were likely infected with at least one vaccine-preventable HPV type before vaccination, which would make the vaccine less effective. We found a stronger association between HPV vaccination and anal prevalence among those who initiated vaccination at younger ages and likely had fewer exposures to HPV before vaccination. However, because nearly all participants reported being sexually active for multiple years before HPV vaccination, differences in PRs between participants who were vaccinated at ≤23 years old compared with >23 years old were not statistically significant. Current HPV vaccines are not approved for therapeutic indications.1 Analyses including participants who were recently vaccinated may not fully account for the necessary time to complete the 3-dose vaccine series (6 months)15 or time to clear prevalent anal infection at vaccination (typically 6–12 months depending on type).51s In sensitivity analyses restricted to participants who were vaccinated >2 years before enrollment, which more likely captures the effect of vaccination against incident infections acquired after HPV vaccination, vaccine effectiveness estimates exceeded 50%.

Differences in anal HPV prevalence between vaccinated and unvaccinated participants were driven by 4vHPV types, especially HPV-16, which was the most prevalent vaccine-preventable type in our study. Human papillomavirus type 16 is consistently associated with higher incidence and longer time to clearance, underscoring its higher oncogenic potential.52s Conversely, anal prevalence of HPV-18 was higher in vaccinated compared with unvaccinated participants, although differences were not statistically significant. Immunogenicity studies have found lower immune response and greater antibody waning over time for HPV-18.53s,54s However, there was no evidence of reduced vaccine efficacy for the HPV-18 component in the clinical trial of GBM8 or in longer-term follow-up of vaccinated cohorts.55s

Strengths of this analysis include the observational study design to measure real-world vaccine effectiveness and the RDS recruitment to estimate population-based HPV prevalence, which is likely more representative than clinic-based samples.40s–44s It is one of the largest community-recruited studies of HPV prevalence among GBM to date, including both unvaccinated and vaccinated men. Limitations include the nonnegligible number of invalid specimens and self-reported HPV vaccination status. Although we failed to detect human β-globin in some anal specimens, our proportion of valid specimens was similar to other studies among GBM using self-collection methods.37s Vaccine uptake was similar between those with and without valid specimens, suggesting that our analysis restricted to valid specimens is unbiased. Although self-reported HPV vaccination status has high sensitivity (>90%) and moderate specificity (>75%) in adults,56s self-report may be associated with nondifferential misclassification, which would bias our estimates toward the null. We have attempted to address potential confounding associated with nonrandom allocation of HPV vaccine through multivariable regression adjustment. Although some residual confounding may remain, findings were similar in sensitivity analyses restricted to participants who had ≥1 prevalent HPV infection that controlled for differences in HPV exposure risk. The few people living with HIV in our sample of young GBM precluded us from looking at differences in vaccine effectiveness by HIV status; HIV was not a significant confounder in our analysis because of its lack of association with HPV vaccine uptake. Men were recruited using RDS, which relies on several assumptions inherent to chain-referral sampling, including accurate reporting of network size.18,30 Respondent-driven sampling-II weights were applied to minimize biases in HPV prevalence estimates, although this may increase variability in weighted regression estimates.31s Although RDS is better able to recruit members of the GBM community who would not traditionally be captured in research studies, men who are less engaged with their health, and thus have higher rates of HPV infection and/or lower vaccination, may have been missed.

Using observational data, we show that HPV vaccination was associated with a lower anal prevalence of vaccine-preventable types among young, sexually active GBM soon after implementation of publicly funded HPV vaccination. This protective association was observed despite the high incidence of HPV infection in this population. Many participants likely had sexual exposure before HPV vaccination and may not have had the opportunity to benefit from these recently launched programs. Lower point estimates suggestive of better vaccine protection were observed in those vaccinated >2 years before enrollment and in those who initiated vaccination at younger ages. Overall, our findings provide further support for current universal HPV vaccination policies targeting school-aged youth before sexual debut but also suggest some vaccine benefit in high-risk programs for young GBM. These findings will help inform shared decision making around HPV vaccination for GBM and their healthcare providers.57s,58s Future analyses will explore vaccine effectiveness against clinically relevant outcomes, including longitudinal end points such as HPV incidence and persistence.

REFERENCES

1. Public Health Agency of Canada. Canadian Immunization Guide: Part 4—Active Vaccines. Ottawa, Canada: Government of Canada, 2018. Available at: https://www.canada.ca/en/public-health/services/publications/healthy-living/canadian-immunization-guide-part-4-active-vaccines.html. Accessed October 16, 2020.
2. Grewal R, Deeks SL, Hart TA, et al. Human papillomavirus (HPV) vaccine uptake among a community-recruited sample of gay, bisexual, and other men who have sex with men in the three largest cities in Canada from 2017 to 2019. Vaccine 2021; 39:3756–3766.
3. Machalek DA, Poynten M, Jin F, et al. Anal human papillomavirus infection and associated neoplastic lesions in men who have sex with men: A systematic review and meta-analysis. Lancet Oncol 2012; 13:487–500.
4. Marra E, Lin C, Clifford GM. Type-specific anal human papillomavirus prevalence among men, according to sexual preference and HIV status: A systematic literature review and meta-analysis. J Infect Dis 2019; 219:590–598.
5. Nyitray AG, Carvalho da Silva RJ, Baggio ML, et al. Six-month incidence, persistence, and factors associated with persistence of anal human papillomavirus in men: The HPV in men study. J Infect Dis 2011; 204:1711–1722.
6. Mooij SH, van Santen DK, Geskus RB, et al. The effect of HIV infection on anal and penile human papillomavirus incidence and clearance: A cohort study among MSM. AIDS 2016; 30:121–132.
7. Chow EPF, Tabrizi SN, Fairley CK, et al. Prevalence of human papillomavirus in young men who have sex with men after the implementation of gender-neutral HPV vaccination: A repeated cross-sectional study. Lancet Infect Dis 2021; 21:1448–1457.
8. Palefsky JM, Giuliano AR, Goldstone S, et al. HPV vaccine against anal HPV infection and anal intraepithelial neoplasia. N Engl J Med 2011; 365:1576–1585.
9. Meites E, Winer RL, Newcomb ME, et al. Vaccine effectiveness against prevalent anal and oral human papillomavirus infection among men who have sex with men-United States, 2016–2018. J Infect Dis 2020; 222:2052–2060.
10. Garland SM, Molesworth EG, Machalek DA, et al. How to best measure the effectiveness of male human papillomavirus vaccine programmes? Clin Microbiol Infect 2015; 21:834–841.
11. Drolet M, Bénard É, Pérez N, et al. Population-level impact and herd effects following the introduction of human papillomavirus vaccination programmes: Updated systematic review and meta-analysis. Lancet 2019; 394:497–509.
12. BC Centre for Disease Control. Chapter 2: Immunization. Communicable Disease Control Manual. Vancouver, Canada: British Columbia Centre for Disease Control, 2020. Available at: http://www.bccdc.ca/health-professionals/clinical-resources/communicable-disease-control-manual/immunization. Accessed February 13, 2019.
13. Gouvernement du Québec. Vaccination contre les virus du papillome humain (VPH): information à l'intention des vaccinateurs. Québec, Canada: Gouvernement du Québec, 2020. Available at: https://publications.msss.gouv.qc.ca/msss/fichiers/2020/20-291-03W.pdf. Accessed November 13, 2020.
14. Ontario Ministry of Health and Long-term Care. Publicly funded immunization schedules for Ontario—December 2016. Toronto, Canada: Queen’s Printer for Ontario, 2016. Available at: http://www.health.gov.on.ca/en/pro/programs/immunization/schedule.aspx. Accessed February 13, 2019.
15. National Advisory Committee on Immunization. Updated Recommendations on Human Papillomavirus (HPV) Vaccines: 9-Valent HPV Vaccine 2-Dose Immunization Schedule and the Use of HPV Vaccines in Immunocompromised Populations. Ottawa: Government of Canada, 2017. Available at: https://www.canada.ca/en/public-health/services/publications/healthy-living/updated-recommendations-human-papillomavirus-immunization-schedule-immunocompromised-populations.html. Accessed October 16, 2020.
16. Hart TA, Moore DM, Noor SW, et al. Prevalence of HIV and sexually transmitted and blood-borne infections, and related preventive and risk behaviours, among gay, bisexual and other men who have sex with men in Montreal, Toronto and Vancouver: Results from the Engage Study. Can J Public Health 2021; 112:1020–1029.
17. Heckathorn DD. Respondent-driven sampling: A new approach to the study of hidden populations. Soc Probl 1997; 44:174–199.
18. World Health Organization. Introduction to HIV/AIDS and Sexually Transmitted Infection Surveillance. Module 4: Introduction to Respondent-Driven Sampling. Geneva, Switzerland: World Health Organization, 2013. Available at: https://apps.who.int/iris/handle/10665/116864. Accessed October 16, 2020.
19. Ivankovich MB, Fenton KA, Douglas JM Jr. Considerations for national public health leadership in advancing sexual health. Public Health Rep 2013; 128 Suppl 1(Suppl 1):102–110.
20. Joint United Nations Programme on HIV/AIDS. Global AIDS Monitoring 2018: Indicators for Monitoring the 2016 United Nations Political Declaration on Ending AIDS. Geneva, Switzerland: UNAIDS, 2016. Available at: https://indicatorregistry.unaids.org/sites/default/files/2017-global-aids-monitoring_en.pdf. Accessed October 26, 2020.
21. Lampinen TM, Miller ML, Chan K, et al. Randomized clinical evaluation of self-screening for anal cancer precursors in men who have sex with men. Cytojournal 2006; 3:4.
22. Coutlée F, Trottier AM, Ghattas G, et al. Risk factors for oral human papillomavirus in adults infected and not infected with human immunodeficiency virus. Sex Transm Dis 1997; 24:23–31.
23. Coutlée F, Rouleau D, Petignat P, et al. Enhanced detection and typing of human papillomavirus (HPV) DNA in anogenital samples with PGMY primers and the Linear array HPV genotyping test. J Clin Microbiol 2006; 44:1998–2006.
24. National Advisory Committee on Immunization. Update on human papillomavirus (HPV) vaccines. Can Commun Dis Rep 2012; 38(ACS-1):62.
25. Volz E, Heckathorn D. Probability-based estimation theory for respondent driven sampling. J Off Stat 2008; 24:79–97.
26. Dunbar R. How Many Friends Does One Person Need? Dunbar’s Number and Other Evolutionary Quirks. Cambridge, United Kingdom: Harvard University Press, 2010.
27. Spiegelman D, Hertzmark E. Easy SAS calculations for risk or prevalence ratios and differences. Am J Epidemiol 2005; 162:199–200.
28. Zou G. A modified Poisson regression approach to prospective studies with binary data. Am J Epidemiol 2004; 159:702–706.
29. White RG, Hakim AJ, Salganik MJ, et al. Strengthening the Reporting of Observational Studies in Epidemiology for respondent-driven sampling studies: “STROBE-RDS” statement. J Clin Epidemiol 2015; 68:1463–1471.
30. Johnston LG, Luthra R. Analyzing data in RDS. In: Tyldum G, Johnston LG, eds. Applying Respondent Driven Sampling to Migrant Populations: Lessons From the Field. London, United Kingdom: Palgrave Pivot, 2014:84–100.

For further references, please see “Supplemental References,” https://links.lww.com/OLQ/A755.

Supplemental Digital Content

Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Sexually Transmitted Diseases Association.