High incidence of Hepatitis C virus infection observed in the PROUD study of HIV pre‐exposure prophylaxis

HIV negative men who have sex with men (MSM) who access pre-exposure prophylaxis (PrEP) report sexual behaviours that could place them at high risk of hepatitis C virus infection (HCV). We report HCV prevalence and incidence from the PROUD trial of PrEP.PROUD was an open-label, wait-list design randomised trial of HIV PrEP for MSM.Participants were recruited between November 2012 and April 2014, and follow-up continued to October 2016. Initial HCV testing followed national guidelines, with screening "on indication", but was replaced by routine quarterly screening in the latter part of the study.We estimated HCV seroprevalence at enrolment and incidence overall and according to calendar year.544 participants were recruited to PROUD. 133 (24.4%) were screened for HCV at enrolment, and 490 (90.1%) were tested at least once during follow-up. Seroprevalence at enrolment was 2.1% (11/530; 95% CI: 1.0-3.7%). Median follow-up time was 2.6 (IQR: 2.1-3.0) years and total follow-up of 1188.8 person years (PY). Twenty-five participants had a new HCV infection during the trial, yielding an incidence rate of 2.1 per 100 PY (25/1188.8; 95% CI: 1.4-3.1), of which three were re-infections. There was some evidence that HCV incidence increased over calendar time (P-value for trend=0.09), reaching an estimated 4.0 per 100 PY (95% CI: 2.0-8.1)in 2016. In conclusion, participants in PROUD had a high, and possibly increasing, incidence of HCV infection. This high incidence of HCV supports the 2018 BHIVA/BASHH recommendation for quarterly HCV testing among HIV-negative MSM using PrEP in the UK.


| INTRODUC TI ON
In general, HIV-negative men who have sex with men (MSM) have been considered to be at low risk for hepatitis C virus (HCV) infection. 1,2 HIV-negative MSM who access pre-exposure prophylaxis (PrEP) have reported sexual behaviours that could place them at high risk of HCV, including high partner numbers, chemsex and injecting drug use. 3,4 Early diagnosis of HCV infection allows early linkage to treatment and care, and reduction in onward transmission of infection. 5 Data from European PrEP trials and cohort studies of PrEP users have reported high baseline HCV prevalence and incidence during follow-up. [6][7][8] In contrast, studies from North America have generally found low levels of HCV endemicity among PrEP users. [9][10][11] The PROUD study was an open-label trial of HIV PrEP among 544 HIV-negative MSM. 12,13 As there were no data on HCV incidence in HIV-negative MSM using PrEP in the UK, we implemented routine quarterly screening during the long-term follow-up phase when all participants had access to PrEP. We report HCV prevalence and incidence among participants in the PROUD study.

| MATERIAL S AND ME THODS
PROUD was an open-label, wait-list trial design that randomized MSM attending participating sexual health centres in England to receive HIV PrEP immediately or after a deferral period of 1 year (the deferred phase). Five hundred and forty-four participants were recruited between November 2012 and April 2014, and follow-up continued to October 2016. The protocol was modified in November 2014 following an interim analysis which showed PrEP to be highly effective, resulting in some participants in the deferred arm being offered PrEP earlier than one year (N = 163). 13 The initial PROUD protocol followed national guidelines on HCV If the first HCV antibody (anti-HCV) test was negative, the participant was assumed to be seronegative at enrolment. If the first test was positive, the Trial Management Group determined whether infection was most likely acquired before enrolment (and thus contributed to the seroprevalence analysis) or after enrolment (and thus contributed to the incidence analysis) based on alanine transaminase and HCV viral load measurements, in relation to time since randomization.

| Statistical analysis
The cumulative incidence of HCV infection (time to diagnosis) was estimated using Kaplan-Meier analysis and randomized groups compared with the log-rank test, censoring at the time of the last screen for HCV. Estimation of incidence by calendar year was complicated by the highly variable time between the last negative test and the first positive test which, in some cases, could span adjacent calendar years. To address this, the date of infection was imputed 1000 times assuming a uniform distribution, and calculating the incidence for each calendar year within each imputed dataset. 14 Estimates were obtained by averaging across the imputed datasets, and confidence intervals derived using Rubin's rule. 15 All analyses were done in STATA version 15.1.

| RE SULTS
Characteristics of the 544 study participants have been previously described. 13 Figure 1 illustrates the hepatitis C screening and infection among study participants. One hundred and thirty-three (24.4%) participants were screened for HCV at enrolment, and 499 (91.7%) were tested at least once during follow-up. Nine participants were only screened at baseline and therefore could not contribute to an incidence analysis. A HCV screen was conducted at 54.0% (3213/5946) of visits (higher during the phase of routine quarterly screening [80.6%] compared to the earlier phase of testing on indication [34.9%]), with a median of 6 (IQR: 3-8) screens per participant. Of the 45 participants who were never tested, 14 participants were also missing information on HCV history collected at enrolment.

| Hepatitis C seroprevalence at enrolment
The seroprevalence at enrolment was 2.1% (11/530; 95% CI: 1.0%-3.7%). The 11 cases were identified as follows: eight participants were reported by the clinician at enrolment to have had a previous diagnosis of HCV; two were diagnosed with HCV a few days before enrolment; two participants had HCV viraemia detected at their first post-enrolment test (they were not tested at enrolment), of which one was judged to have acquired infection before enrolment.

| Hepatitis C incidence
HCV incidence is based on the 490 participants who were considered HCV seronegative at enrolment, or who had previously cleared HCV infection prior to enrolment, and had at least one postenrolment HCV test. Table 1  Three of these were re-infections (the previous infections had cleared spontaneously or with treatment before enrolment), and one participant experienced two infections during follow-up (only the first infection was included in the incidence calculation). Two HCV infections were acquired after diagnosis of HIV whilst the participant was no longer on PrEP but was being actively followed up.
Excluding these cases from the HCV incidence calculation reduced the estimate only slightly (1.9 per 100 PY [95% CI: 1.2-2.9]). Use of nonprescribed injected drugs was reported by 11 participants at the suspected time of HCV infection, was denied by 12 and was unknown for two. Figure 2 shows the cumulative incidence of time to a new HCV diagnosis, stratified by randomized arm (P-value log-rank test = 0.87).
As only three infections (one immediate, two deferred) were observed during the deferred phase of the study (ie the period before the DEF arm had access to PrEP), the trial essentially provides no randomized information of whether access to PrEP affects the risk of acquiring HCV infection. Accounting for uncertainty in the time of acquisition of HCV infection, HCV incidence appeared to increase over calendar time (Table 2)   The disparity in HCV incidence seen between European and North

| D ISCUSS I ON
American PrEP studies could be due to a different risk of exposure to the virus (ie variation in the prevalence of active infection among MSM) or differences in the extent of high-risk sexual behaviours.
PROUD participants were at the far end of this spectrum, reporting an average of 10 sexual partners in the three months before enrolment and 37% having used post-exposure prophylaxis in the previous 12 months (Table 1). Although the increase in HCV incidence over calendar timein PROUD was not conventionally statistically significant, the low seroprevalence at enrolment (2.1%, coincidently identical to the overall incidence of 2.1% per year) suggests that the increasing incidence is genuine. This increase occurred despite an overall decline in HCV prevalence in England in recent years. 18

TA B L E 2 HCV incidence stratified by calendar year
and HIV-positive MSM in a French clinic 19,20 suggest that there was substantial sexual mixing between HIV-negative and HIV-positive populations, who have high HCV prevalence 21 and incidence, 2,22 and it is likely that PrEP facilitates this mixing. Although the frequency of HCV testing increased part way through the study and probably resulted in more rapid diagnosis of infections, we used an imputation method that should correct for any bias arising from this.
The 2.1% seroprevalence of HCV in PROUD participants at enrolment was considerably lower than that reported in the AmPrEP demonstration project (4.8%). 19 The IPERGAY study reported only one HCV infection at enrolment. 7 However, HCV seroprevalence changes rapidly in the context of a high incidence and comparisons between studies will be affected by the calendar time over which seroprevalence was calculated.
Our study has two major limitations. Regular care for PrEP provides an opportunity to screen for and provide early intervention for HCV. The increasing availability of HCV antigen/antibody testing makes screening in this population more feasible. 24 The high incidence of HCV that we observed in PROUD supports the 2018 BHIVA/BASHH recommendation for quarterly HCV testing among HIV-negative MSM using PrEP in the UK, in line with other STIs. 25 Also, a recent modelling study has indicated that screening and treating PrEP users for HCV at least every 12 months can reduce HCV incidence by 67.3% (uncertainty range 52.7%-79.2%). 26 However, HCV incidence can vary markedly by location and time, and guidance in the UK and other countries 27,28 should be regularly reviewed in the light of local evidence.

CO N FLI C T S O F I NTE R E S T S
The PROUD study was provided drug free of charge by Gilead Sciences plc. that also distributed it to participating clinics and provided funds for additional diagnostic tests for HCV and drug levels.
EW university fees and stipend funded by Gilead Science plc. SM reports grants from the European Union H2020 scheme, EDCTP 2, the