Background

Approximately 3 % of the world’s population is infected with hepatitis C virus (HCV), a blood borne infection that is almost entirely attributable to parenteral exposure via non-sterile injection equipment used in medical settings or to inject drugs [1]. After clinical or subclinical HCV infection up to 25 % of people will spontaneously clear the virus [2]. There is significant excess mortality attributable to liver-related injury in those with chronic HCV infection compared to the general population, and among those with chronic HCV infection, progression of the disease varies widely [3, 4].

Among people who inject drugs (PWID), high HCV incidence rates of 10–40 infections/100 person-years (PYs) contribute to a persistent and high population prevalence of 43–80 % [5–7]; as such, HCV is endemic among PWID [8].

Co-infection with HIV and HCV is common due to shared routes of disease transmission. HIV/HCV co-infection contributes to substantial, yet preventable, morbidity and mortality; specifically, liver disease progression is accelerated in HIV co-infected individuals [9, 10]. Co-infection with HIV and HCV is of considerable relevance to PWID as the majority of HIV-positive PWID also are infected with HCV (50–70 %) [11].

Within the HIV-positive population, sexual transmission of HCV also is a concern [12, 13]. In a recent meta-analysis, the incidence rate of HCV infection among HIV-positive men who have sex with men (HIV+ MSM) who are non-PWID was found to be 0.53/100 PYs [14]. While low, incidence of HCV in this population is expected to increase [14]. In a related meta-analysis, HCV prevalence was estimated to be 12 % among HIV+ MSM [15].

Characterizing spontaneous viral clearance of HCV infection among PWID and HIV+ MSM is important for assessing the burden of disease and the need for treatment in these populations. In this systematic review and meta-analysis, we synthesized the literature on the prevalence of spontaneous viral clearance within PWID and HIV+ MSM populations. This systematic review and meta-analysis and related simulations are conducted as part of the HCV Synthesis Project, which is funded to develop guidance and recommendations for HCV control strategies in the US [14–19].

Methods

Search strategy

Both electronic and manual searches for published literature were conducted. The databases of CINAHL, OVID, ProQuest, PubMed, and Web of Science were searched using the following terms: “HCV,” “hepatitis C,” “natural history,” “disease progression,” “clearance,” and “resolution.” For the HIV+ MSM group, the search string included variations of the terms “HIV,” “human immunodeficiency virus,” “AIDS,” “acquired immunodeficiency syndrome,” “men who have sex with men,” “homosexual,” and “gay.” Reports examining PWID were sought through the incorporation of the keywords “PWID,” “injection drug use,” and “intravenous drug use”. (See Additional file 1 for complete search strategies.)

Searches were refined using filters for publication date, peer-reviewed journal, and human studies. Additional literature was retrieved through manual searches of the reference lists of eligible reports, review articles, and methodological papers. The conduct and reporting of this project was guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [20]. The protocols for the HIV+ MSM and PWID research were registered with PROSPERO (CRD42013006462; CRD42014008805) and subsequently published [16, 18].

Inclusion and exclusion criteria

PWID

Reports that met the following criteria were eligible for inclusion in the review: (i) included participants with laboratory confirmed HCV infection and who reported current or previous injection drug use (hereafter referred to as PWID); (ii) presented original data on spontaneous viral clearance in a study sample comprised of at least 90 % PWID; (iii) published between January 1, 1990, and April 1, 2014; and, (iv) provided data on participants from upper-middle- or high-income countries. The condition in (iv) was based on the prevalence of the hepatitis B virus (HBV) carrier state, which is associated with lower rates of HCV clearance [21]. Across upper-middle- and high-income countries, the HBV carrier rate is less than 2 % whereas areas of high HBV endemicity are comprised of predominately lower-income countries such as Southeast Asia, Sub-Saharan Africa, and the Amazon Basin; HBV carrier rates in these regions are upwards of 8 % [22] and may be higher among those at risk of HIV or HCV. Reports were excluded if HIV or HBV co-infection was present in greater than 50 % of the PWID study sample, or if participants were receiving or previously had received HCV treatment. (Seven reports included HIV-positive PWID (range 1.8–57.1 %).)

HIV+ MSM

The following eligibility criteria were applied: (i) included male participants co-infected with HIV infection and with laboratory confirmed acute HCV infection and who reported having sex with other men (hereafter referred to as MSM); (ii) presented original data on spontaneous viral clearance in a study sample comprised of at least 90 % HIV+ MSM; (iii) published between January 1, 1996, and April 1, 2014; and, (iv) provided data on participants from upper-middle- or high-income countries. We established the condition in (iii) to account for the introduction in 1996 of highly active anti-retroviral therapy (HAART), which represented a significant change in the standard of care, and thus the natural history, of HIV infection. Accordingly, HAART may affect the disease progression of individuals co-infected with HIV and acute HCV. The restriction that participants had acute HCV allowed for the observation of the effect of HIV infection on the course of newly acquired HCV infection. The condition in (iv) follows from the assumption that health outcomes are influenced by the disparity between income-based country groups in the accessibility and completeness of coverage of HAART to HIV-positive individuals; a wide gap exists between treatment and need in low-income countries [23].

HCV infection measure

The primary exposure of interest was acute or chronic HCV infection. The preferred criteria for defining acute HCV infection employed in our review was that endorsed by the European AIDS Treatment Network (NEAT) for which the criteria were seroconversion or a positive HCV RNA test following a documented negative HCV RNA or negative HCV antibody test in the previous 12 months [24]. Our alternative criterion for defining acute HCV infection required a statement in the report that all patients were acutely infected. Chronic infection was defined by HCV RNA positivity.

The importance of defining whether or not HCV was acute was less critical to those without HIV infection; however, in those with HIV the demonstration of acute infection (either by the preferred or alternative criteria) was relevant in asking whether HIV co-infection impacts spontaneous viral clearance. Among the reports presenting data on spontaneous viral clearance among HIV+ MSM, all participants acquired HIV infection prior to HCV infection; among PWID the sequence of disease acquisition was less clear.

Outcome measures

The outcome of interest was the prevalence of spontaneous viral clearance. Spontaneous viral clearance was measured in cross-sectional studies as at least one negative or undetectable RNA test result and in longitudinal studies as consecutive negative or undetectable RNA test results. In some reports the definition of spontaneous viral clearance was not presented, but data were available on the outcome of interest.

Screening and data collection

Two research assistants (RAs) screened abstracts and extracted data. The project director and the principal investigator reviewed all eligible reports and made final decisions on inclusion in the review and meta-analysis. We collected from the included reports data on the following domains: citation information; study cohort, period, and location; study design and methods; incidence and prevalence of spontaneous viral clearance; disease duration; and participant characteristics, particularly factors understood to be associated with clearance (e.g., age, sex, and HIV co-infection). For any report with missing or inconsistent data, we contacted the corresponding author for additional information or clarification. Among the set of reports examining PWID, six authors were contacted, and four fulfilled our data requests (67 %). Three of six authors provided additional data on HIV+ MSM (50 %).

Report quality

Quality appraisal of each report included in this systematic review was based on an adapted version of the Quality In Prognosis Studies (QUIPS) tool, which was developed to assess potential biases in studies of prognostic factors [25, 26]. (The complete adapted instrument is available by request.) Each report was assigned an overall rating of high, moderate, or low, which indicated the extent to which the study design and analysis controlled for the influence of selection bias, misclassification, and confounding.

Data analysis

Report-level prevalence of spontaneous viral clearance was estimated using the binomial distribution. Pooled estimates of spontaneous viral clearance were generated using both fixed-effect and random-effects models. The Cochran’s Q and I2 [27] measures provided assessments of heterogeneity, and random-effects meta-regression was performed to examine variability among the report-level estimates of spontaneous viral clearance.

We examined the possibility of assessing the role of previous infection with HIV on spontaneous viral clearance among PWID; however, because most studies did not specify which of HIV or HCV was acquired first, we were not able to directly examine this. Those PWID who were HIV positive and had unequivocal evidence of spontaneous viral clearance most likely represented an individual in whom the sequence of events was HCV infection, followed by spontaneous viral clearance, and subsequently HIV infection. All statistical analysis was conducted using Stata 13.1 [28].

Results

Meta-analysis

PWID

A total of 7,488 reports were retrieved from the literature searches, and 28 were included in the final review (see Fig. 1). The reports are described in Table 1. The majority of reports described studies located in Europe (12; 43 %); the remaining samples were from the United States (7; 25 %), Australia (5; 18 %), China (2; 7 %), Canada (1; 4 %), and Iran (1; 4 %). Data on participant attributes were presented in greater than 50 % of all reports. Sixteen reports (57 %) presented age at enrollment, eighteen (64 %) provided the sex distribution, and sixteen (57 %) stated the proportion of HIV-positive individuals.

Fig. 1
figure 1

Flow diagram of the literature search and eligibility assessment for PWID. 1Reasons for exclusion are not mutually exclusive. A report may have been ineligible due to multiple reasons. Here, only one reason for exclusion is provided

Full size image
Table 1 Studies investigating spontaneous viral clearance of HCV among PWID
Full size table

All 28 reports were included in the meta-analysis. Among 5,569 PWID, whose mean age at enrollment was 29.6 years (median 27.4; 15 reports), spontaneous viral clearance was observed in 1,236 participants. The random-effects meta-analysis estimate of the prevalence of spontaneous viral clearance was 24.3 % (95 % CI 19.5, 29.1; Q = 638.51, p < 0.001; I2 = 95.8 %). The estimate from each study is presented in the forest plot in Fig. 2.

Fig. 2
figure 2

Forest plot of the estimated prevalence and 95 % CI of HCV clearance among PWID for each study

Full size image

Pooled subgroup estimates also were generated to account for HIV status and quality rating. Within the sets of reports for which there were data on the HIV status of PWID, spontaneous viral clearance occurred in 25.7 % (95 % CI 16.4, 35.0; 13 reports) of HIV-negative participants and in 16.1 % (95 % CI 12.5, 19.6; 3 reports) of HIV-positive participants. In stratifying the reports by quality rating, as depicted in Table 3, the prevalence of spontaneous viral clearance was 27.6, 24.1, and 15.4 % among high-, moderate-, and low-quality reports, respectively.

HIV+ MSM

Following from Fig. 3, the literature searches yielded 2,417 reports, of which 10 were included in the present analysis; these reports are detailed in Table 2. Seven of the reports (70 %) were on studies with cohorts drawn only from Europe. One report (10 %) examined a sample from the United States, and another report (10 %) assessed samples from North America, Australia, and Europe. The study location was not described in one report. Few participant characteristics were summarized consistently across the reports. Age at enrollment was presented in five reports (50 %), proportion on HAART in three reports (30 %), and duration of HIV infection in one report (10 %).

Fig. 3
figure 3

Flow diagram of the literature search and eligibility assessment for HIV+ MSM. 1Reasons for exclusion are not mutually exclusive. A report may have been ineligible due to multiple reasons. Here, only one reason for exclusion is provided

Full size image
Table 2 Studies investigating spontaneous viral clearance of HCV among HIV+ MSM
Full size table

Only 9 of the 10 reports were included in the meta-analysis. One report provided a count of zero spontaneous clearance events in the sample. We chose the conservative approach to handling the zero-count issue, which was to exclude the report from the pooled analysis.

In aggregate, there were 588 HIV+ MSM with a mean age of 40.0 years (median 40.7; 4 reports); 96 participants experienced spontaneous viral clearance. The random-effects meta-analysis estimate of the prevalence of spontaneous viral clearance was 15.4 % (95 % CI 11.5, 19.3; Q = 13.29, p = 0.102; I2 = 39.8 %). The forest plot in Fig. 4 provides the estimates from the contributing reports. Subgroup estimates based on quality rating also were obtained. As shown in Table 3, among high-, moderate-, and low-quality reports, prevalence of spontaneous viral clearance was 19.2, 15.8, and 11.3 %, respectively.

Fig. 4
figure 4

Forest plot of the estimated prevalence and 95 % CI of HCV clearance among HIV+ MSM for each study

Full size image
Table 3 Prevalence of spontaneous viral clearance among PWID and HIV+ MSM stratified by quality rating
Full size table

Meta-regression

To examine the effect of report and participant characteristics on the prevalence of spontaneous viral clearance, we conducted random-effects meta-regression. Because covariates hypothesized to contribute to spontaneous viral clearance were not consistently collected or reported across reports, our meta-regression was limited to a small set of factors. Both univariate and multivariate meta-regression were performed.

PWID

Only two variables were significant in univariate analysis: male sex and age (p < 0.05). Spontaneous viral clearance was lower in both males and older individuals. We examined the relationship among pairs of variables (specifically, male sex, HIV-positivity, and age) through contingency tables. Fisher’s exact tests revealed no significant associations between factor dyads (results not shown; see Additional file 2). Of particular interest to our study was the impact of HIV. In our preferred multivariate model, presented in Table 4, male sex (p < 0.021) and HIV positivity (p < 0.036) were significant predictors.

Table 4 Random-effects meta-regression results for spontaneous viral clearance among PWID
Full size table

HIV+ MSM

Neither univariate nor multivariate analysis provided any evidence of an effect on spontaneous viral clearance of any of the variables (i.e., age, proportion on highly active antiretroviral therapy, recruitment site, and quality rating) evaluated.

Discussion

In this systematic review and meta-analysis, we estimated that the prevalence of spontaneous viral clearance is 24.4 % in PWID and 15.4 % in HIV+ MSM. Although the estimates for PWID and HIV+ MSM were not directly compared here, the difference may be related to the impact of HIV co-infection on the natural history of HCV. The rates of spontaneous viral clearance were similar in HIV+ MSM (15.4 %) and PWID with HIV infection at the time that HCV clearance was evaluated (16.1 %). Our estimate of 24.4 % among PWID was very closely similar to the estimate from a pooled analysis of clearance among 632 participants in multiple studies (25 %) who were observed following acute infection [29].

Among HIV-positive PWID, lower clearance cannot reliably be attributed to the effect of HIV infection on viral kinetics because the temporal relation between HIV and HCV infections in these individuals is unclear. Moreover, estimates of spontaneous viral clearance from most studies do not necessarily represent true rates but rather the prevalence of cleared infection. Evidence of clearance at any given time in an individual is the cumulative result of behaviors that led to infection events, including multiple re-infections, and the host and viral characteristics that govern response to acute HCV infection. Although re-infection post-SVR is higher among HIV + MSM than among PWID, it cannot be concluded that the rates of HCV re-infection in treated and untreated HIV + MSM are higher than among treated and untreated PWID because studies of re-infection in treated PWID have in many cases excluded active injectors [14, 30]. There is no research comparing the frequency of HCV transmission behavior between these groups, and thus, there is insufficient evidence to attribute differences in clearance to differences in behavioral risk.

Our finding that female gender was associated with higher proportions of spontaneous viral clearance events is consistent with published literature demonstrating that females are more likely to clear HCV than their male counterparts in a variety of settings and other patient groups [31, 32].

Evaluation of the report-level spontaneous viral clearance data for each of the populations indicates that the meta-analysis estimates were affected by notable degrees of heterogeneity. Among PWID, both male sex and age were associated with lower proportions of spontaneous viral clearance events. In examining spontaneous viral clearance estimates in HIV+ MSM, the results did not suggest the effect of any of the factors considered. Given the null findings in univariate meta-regression on quality rating (p = 0.163) and study design (p = 0.182), and noting the low volume of reports for which we extracted data on participant characteristics, heterogeneity was most likely due to unmeasured clinical, patient-level characteristics (e.g., ethnicity, genotype, other viral infections) rather than methodological characteristics. However, our analysis of sources of heterogeneity was limited by the dearth of information about time to event, viral factors, and other participant attributes.

Limitations

There are limitations to the meta-analysis that should be considered. One critical point is that the results of this review suggest, but do not explain, the mechanisms that lead to spontaneous viral clearance. Another important consideration is that the study methods and data provided by some of the contributing reports to this review presented challenges to the characterization of clearance in PWID and HIV+ MSM. We discuss here the main issues related to the contributing reports.

First, the samples of PWID examined in the included reports were not composed solely of individuals with a single HCV infection event, and, therefore, the estimates likely represented clearance in relation to cumulative HCV exposures via ongoing injection risk behavior [33]. Indeed, among PWID, rates of re-infection following spontaneous viral clearance are as high as 47 cases per 100 person-years [2]. Additionally, low rates of clearance observed among HIV+ PWID may represent frequent risk behavior that led to HIV infection and HCV re-infection. The high heterogeneity in the clearance estimates for both PWID and HIV+ MSM may in fact represent variability in the number of HCV infection events, in addition to genetic and other factors.

Second, the cross-sectional study design of most of the reports also limits interpreting the estimates vis-à-vis the underlying process.

Third, and related to the previous point, the definition of spontaneous viral clearance also was not uniform across studies. In particular, the criteria for spontaneous viral clearance events were notably different between cross-sectional and longitudinal reports.

Conclusion

This systematic review and meta-analysis suggests that the prevalence of spontaneous viral clearance is higher among PWID compared to HIV+ MSM. Our findings also showed that, among PWID, male sex and HIV co-infection are negatively correlated with clearance. The data we presented are useful for modeling future morbidity, mortality, and costs related to HCV infection. Improved research methodology and examination of individual characteristics in future studies would help to determine the natural course of HCV among the high-risk populations of PWID and HIV+ MSM, and appropriate allocation of resources for HCV treatment.