Factors associated with phylogenetic clustering of hepatitis C among people who inject drugs in Baltimore

Hepatitis C virus (HCV) infection is a major public health challenge that disproportionately affects people who inject drugs (PWID) with a prevalence of 50–80% [1]. Shared use of drug preparation equipment (including cotton, cookers, and water) is the major route of HCV transmission among PWID [2]. This results in persistently high HCV transmission rates among PWID, with annual incidence rates ranging from 3 to 66 infections/100 person-years [3‐6]. Recent increases in injection opioid use have further escalated HCV incidence and contributed to increased mortality due to HCV-related conditions [7, 8].

The availability of effective, oral direct acting antivirals (DAAs) for HCV treatment has fueled optimism for HCV elimination, including a call by the World Health Organization for the elimination of HCV as a public health challenge by 2030 [9, 10]. Identifying characteristics of individuals in HCV transmission networks would provide critical information for the development and implementation of targeted HCV treatment interventions.

Phylogenetic analyses have been used to document transmission of both HIV and HCV infections between partners and within social networks [11‐14]. Recent phylogenetic analyses suggested dense networks of HIV transmission in a rural community in Indiana, United States linked to injection use of oxymorphone [12]. This analysis also demonstrated that HIV spreads rapidly through existing HCV transmission networks [15]; highlighting the importance of understanding HCV transmission networks for prevention of HCV and HIV transmission among PWID. Less is known about HCV transmission networks in urban areas of the United States.

In this study, we sought to examine phylogenetic clustering, cluster composition and factors associated with clustering among participants enrolled in the AIDS Linked to the Intravenous Experience (ALIVE) cohort, a longstanding prospective cohort of current and former PWID followed in Baltimore, Maryland, United States.

HIV infection was independently associated with being in a cluster in this study. This association of HIV infection with HCV clustering has previously been described [24, 25] and suggests that individuals with HIV/HCV co-infection represent a group at greater risk of behaviors leading to onward HCV transmission, for which targeted interventions, including HCV treatment for prevention of transmission, could provide disproportionate benefit [24‐27]. Of note, in this group of HIV infected PWID, a significant proportion had detectable HIV viremia and only a minority reported accessing syringe service programs. Although HCV is approximately 10 times more transmissible than HIV [28], continued high-risk injection practices could also result in HIV transmission, as noted in the recent HIV transmission outbreak in Indiana [15]. This is especially concerning given the high proportion of HIV uninfected individuals who had potential HCV transmission partners who were HIV infected. Our findings reinforce the need for interventions to increase uptake of harm reduction interventions, including syringe service programs, medication assisted treatment, frequent testing for both HIV and HCV, and immediate treatment to prevent and break transmission cycles for both HIV and HCV among PWID [29‐32].

We also found female sex to be associated with HCV clustering. Increased HCV clustering in females may be linked to differences in the way women experience substance use compared to men. Women are more likely to have a sexual partner facilitate their initiation of injection drug use by guiding or administering their first injection [33‐36]. Further, individuals who are injected by someone else at initiation are more likely to report receptive syringe and other equipment sharing at initiation and within the past 6 months [37, 38]. Sexual relationships have also been associated with increased sharing of drug use paraphernalia [39]. Male sexual partners may particularly influence injection practices of women [33, 34, 40]. It is thus not surprising that a higher risk of incident HCV infection has been documented in women compared to men in the setting of heterosexual relationships in another study [41]. While our analysis cannot confirm direct transmission, existing sexual relationships, or information on behaviors at injection initiation, most women who were part of a cluster in our analysis clustered with men, which is supportive of our hypothesis of transmission of HCV through injection drug use with individuals that may also be sexual partners. Interventions taking into account the experiences of women who use drugs are needed to reduce rates of HCV infection and reinfection in female PWID, especially given recent increases in incident HCV infection among women [42].

Overall rates of HCV clustering found in our study were lower than those of 22–37% reported in other studies, including one from this cohort [23‐25, 43‐45]. This study differs from the previous study evaluating clustering of HCV in the ALIVE cohort in that samples were sequenced from earlier in the HCV epidemic (1988) at a time when majority of the participants were younger (median age < 35 years compared to the median age of 53 years of participants in this current study) suggestive of shorter duration of HCV infection. Lower rates of HCV infection clustering in our study is likely due to the long duration of HCV infection in most individuals sampled in this analysis. The hepatitis C virus uses an error prone polymerase to replicate to high viral loads resulting in the generation in each individual host infected with HCV, genetically distinct viral variants. This, in addition to the impact of host immune response on viral evolution within individuals, will make it challenging to uncover infections from a common source occurring many years earlier [46, 47]. Our sensitivity analyses revealed intra but not inter person clustering of HCV sequences over time among sequences from women in the anti-D cohort, who were known to have been infected from a common source exposure to HCV, suggesting that clustering as defined in this analysis represents more recent transmission events. Low rates of clustering found in our study may also reflect differences in recruitment across studies, sampling density, or methods used for phylogenetic and cluster analyses. In the ALIVE cohort, the study population is recruited in discrete waves with targeted efforts to recruit a geographically diverse cohort that is representative of the estimated 19,000 people who use drugs in Baltimore [48]. It is thus possible that a low sampling density of HCV infected PWID may have reduced the sensitivity of finding evidence of clustering [49]. Additionally, most analyses have used a longer fragment of the HCV genome; for example, the Core to E2 region (1142 nucleotides), which is much longer than the Core to E1 region (342 nucleotides) used in our analyses, thus the latter may have lower sensitivity of cluster detection. However, the sensitivity of sequence recovery of the Core E1 amplicon (over 90% of HCV RNA positive samples are successfully sequenced) over that of longer amplicons, may balance the short sequence length by generating less missing sequence data for participants and has identified clustering from samples collected closer to infection [23].

Our study is limited by being restricted to a phylogenetic analysis of a sample of PWID enrolled in the ALIVE cohort, which may not be representative of all PWID in Baltimore or in other urban areas of the United States. Additionally, these analyses cannot determine all linked transmissions, as there are likely individuals who were not participants in this study, but may have acquired HCV and then transmitted to others as components of transmission networks. We also do not have information on potential social relationships between individuals identified as being in a cluster. Our analyses do, however, identify factors associated with being in a cluster suggestive of linked transmission.

In the era of availability of effective tools, such as oral DAA HCV therapies, syringe service programs, and medication-assisted treatment, with the potential to reduce HCV transmission, molecular epidemiology data such as these analyses, provide critical information to guide effective public health interventions to control HCV. Our study highlights the potential role of combination interventions targeting HIV infected PWID and addressing HCV infection prevention needs of women to reduce HCV infection transmission.