Background. Treatment for hepatitis C virus (HCV) can lead to sustained virological response (SVR) in over 90% of people. Subsequent recurrence of HCV, either from late relapse or reinfection, reverses the beneficial effects of SVR.
Methods. A search identified studies analysing HCV recurrence post-SVR. The recurrence rate for each study was calculated using events/person years of follow-up (PYFU). Results were pooled using a random-effects model and used to calculate 5-year recurrence risk. Three patient groups were analysed: (1) Mono-HCV infected “low-risk” patients; (2) Mono-HCV infected “high-risk” patients (injecting drug users or prisoners); (3) human immunodeficiency virus (HIV)/HCV coinfected patients. Recurrence was defined as confirmed HCV RNA detectability post-SVR.
Results. In the 43 studies of HCV mono-infected “low-risk” patients (n = 7969) the pooled recurrence rate was 1.85/1000 PYFU (95% confidence interval [CI], .71–3.35; I2 = 73%) leading to a summary 5-year recurrence risk of 0.95% (95% CI, .35%–1.69%). For the 14 studies of HCV monoinfected “high-risk” patients (n = 771) the pooled recurrence rate was 22.32/1000 PYFU (95% CI, 13.07–33.46; I2 = 27%) leading to a summary 5-year risk of 10.67% (95% CI, 6.38%–15.66%). For the 4 studies of HIV/HCV coinfected patients the pooled recurrence rate was 32.02/1000 PYFU (95% CI, .00–123.49; I2 = 96%) leading to a summary 5-year risk of 15.02% (95% CI, .00%–48.26%). The higher pooled estimates of recurrence in the high-risk and coinfected cohorts were driven by an increase in reinfection rather than late relapse.
Conclusions. SVR appears durable in the majority of patients at 5 years post-treatment. The large difference in 5 year event rate by risk group is driven mainly by an increased reinfection risk.
Infection with the hepatitis C virus (HCV) is a significant public health concern associated with a high burden of morbidity and mortality [1, 2]. Recent estimates suggest that worldwide, of the 185 million individuals infected, over 700 000 people die annually as a result of infection [3, 4].
The attainment of a sustained virological response (SVR), defined as aviremia 12 or 24 weeks after the completion of antiviral therapy (SVR12 or SVR24), is associated with an improved prognosis compared with patients either untreated or failing therapy. These benefits include improved histology, reduced risk of hepatocellular carcinoma, and improved overall survival [5, 6].
Despite these benefits, treatment uptake for chronic HCV has been low due to complexities of treatment and poor success rates. The availability of new highly efficacious regimens provides the foundation for marked treatment scale-up; however, high costs are currently limiting access [7–10].
One challenge to treatment scale-up is the risk of HCV recurrence, either as late relapse post-SVR or reinfection following treatment. HCV recurrence is a particular concern in patients with ongoing high-risk behaviors, such as injecting drug users (IDUs), who are more susceptible to reinfection, and also patients coinfected with human immunodeficiency virus (HIV) who may be at increased risk of relapse due to their immunocompromised status [11–15].
A number of studies have been carried out to examine the durability of treatment-induced SVR in patients with chronic HCV in a variety of patient populations. Our aim was to systematically review the existing evidence and undertake meta-analysis to provide summary estimates of the recurrence rate by risk group. The secondary aim was to evaluate the contribution of late relapse and of reinfection to the recurrence rate. This work fits within the theme one of the PROGRESS framework for prognosis research (“fundamental prognosis research”) and will provide a clearer understanding of HCV recurrence to inform the provision of antiviral therapy [16].
Previous SectionNext Section
METHODS
Search Strategy and Inclusion Criteria
The MEDLINE database was searched from 1990 until 1 March 2015 for studies analyzing HCV recurrence post-SVR. A sensitive search string was developed using terms including hepatitis C, treatment, SVR, recurrence, relapse, and reinfection (Supplementary Appendix). The reference lists of articles were thoroughly searched to identify additional articles. Lastly, the proceedings of the following conferences were search for additional studies: International Liver Congress (EASL), The Liver Meeting (AASLD), Conference on Retroviruses and Opportunistic Infections, and the International AIDS Conference.
Studies included were to have enrolled adult patients (aged ≥18) who achieved SVR after antiviral treatment for acute or chronic HCV. SVR was defined as undetectable HCV RNA 12 or 24 weeks post-treatment. There was no stipulated method of HCV acquisition or specific antiviral treatment regimen. There were no restrictions on study design however all studies were to have a follow-up longer than 6 months post-SVR. Studies were excluded if they examined rate of recurrence after spontaneous clearance, or if they measured recurrences after the end of treatment, not allowing for the SVR time period to elapse.
Studies were categorized in to 3 groups: (1) Low-risk population, inclusive of studies of mono-HCV infected patients with no recognized risk factors for reinfection; (2) High-risk population, inclusive of studies of mono-HCV infected patients with at least 1 identified risk factor for reinfection; and (3) HIV/HCV coinfection populations, inclusive of all studies of HIV/HCV coinfected persons, regardless of the presence or absence of other risk factors. Risk factors for reinfection were defined as current or former IDU, imprisonment, and men who have sex with men (MSM). Studies of liver transplant recipients were excluded.
Quality Assessment
Articles meeting the inclusion criteria were assessed for methodological quality using the Newcastle–Ottawa Scale (NOS). The assessment was modified to allocate a maximum of 8 stars, for quality of selection, comparability, exposure, and outcome of study participants (Supplementary Appendix). Studies with a NOS rating ≥6 were considered high-quality.
Data Extraction
The following data were extracted for each study: location, design, recruitment, patient characteristics, average follow-up time, number of HCV recurrences, total PYFU, and frequency of HCV RNA assessment. HCV recurrence was defined as confirmed HCV RNA detectability post-SVR. Where possible, recurrence was characterized as either late relapse or as reinfection, with categorization carried out according to the original study definitions and techniques. In all studies using phylogenetic techniques late relapse was defined as detection of HCV RNA of the same virus lineage and reinfection as identification of a different virus. In the majority of studies, this classification was according to the protocol in the original article. In genotyping studies where no criteria for classification were given, the same definitions were applied by the authors of the current meta-analysis. In some studies, categorization was done by the study authors without confirmatory genotyping. In these studies, the decision to classify as late relapse or reinfection was usually made through consultation with patients to assess for the presence or absence of risk behaviors (eg, injecting drug use, unsafe procedures, etc.).
PYFU were accrued from the SVR time-point; in those studies where follow-up originated at the end-of-treatment, PYFU were appropriately adjusted. If total PYFU was not explicitly stated, it was estimated from the average follow-up time; studies in which PYFU was inestimable were excluded. In the case of study duplications, the article providing the most comprehensive account of the study population and longest follow-up period was used.
The literature search, data extraction, and quality assessment were carried out independently by 2 authors (B. S., J. S.), and any differences were resolved by consensus.
Data Synthesis
For each study, the incidence rate of HCV recurrence was calculated as the number of recurrences per 1000 PYFU and was reported with the corresponding 95% Wilson confidence interval (95% CI). Given the rarity of events, estimates were transformed using the Freeman-Tukey double arcsine transformation [17, 18]. A pooled estimate for recurrence was then calculated for each of the three groups separately using a random-effects model [19]. In addition, meta-analyses of the rate of late relapse and of the rate of reinfection were carried out including studies providing this data. The pooled estimates were used to calculate the 5-year event rate for recurrence, late relapse, and reinfection for each population. The summary 5-year risk was calculated using 1 – (1 – pooled incidence rate)5 and as such assumed that the pooled rate of recurrence was constant over the follow-up duration. For each calculation, the degree of heterogeneity between studies was quantitatively assessed using I2 and tau2, where an I2 ≥ 50% may indicate substantial heterogeneity and ≥75% is indicative of considerable heterogeneity. The existence of publication bias was evaluated by observational analysis of funnel plots. All analyses were conducted using STATA version 13 (StataCorp LP, Texas).
Previous SectionNext Section
RESULTS
As shown in Figure 1, a total of 1180 references were identified and screened for eligibility. Of these, results were available from 59 studies reporting on recurrence post-SVR in a total of 9049 patients. Two studies evaluated two distinct subgroups of monoinfected and HIV coinfected patients and as such were included in 2 analysis groups. Of the studies deemed possibly relevant and screened against inclusion criteria, the main reasons for exclusion were the assessment of recurrence rate after spontaneous clearance and the lack of an SVR time period after the end-of-treatment. The study and cohort characteristics are shown in Table 1. All identified studies evaluated SVR at 24 weeks post-treatment; no studies eligible for inclusion used SVR12 as the endpoint for analysis. Frequency of HCV RNA assessment varied from every 3 months to 1 single assessment during follow-up. For all 3 risk groups, funnel plots appeared symmetrical indicating no evidence of bias. Of all studies, 49/59 (83%) were considered high-quality (NOS score ≥6). The main biases observed were in determining PYFU and in accepting the authors' opinion regarding reinfection vs relapse.
View this table:
In this window
In a new window
Table 1.
Study Characteristics of Included Studies
Figure 1.
View larger version:
In this page
In a new window
Download as PowerPoint Slide
Figure 1.
Flow diagram of study selection for systematic review of hepatitis C virus (HCV) recurrence in patients achieving a sustained virologic response after treatment for HCV infection. Low-risk studies include those examining recurrence in general populations and high-risk studies include those studying patients with at least 1 reinfection risk factor (injecting drug use or prison populations). Human immunodeficiency virus (HIV)/HCV coinfected studies include all those of coinfected participants, regardless of risk factors. Total studies in the 3 groups does not equal the total number of studies identified as 2 studies examined 2 populations.
Low-risk Population
Forty-three articles were found evaluating the risk of recurrence in 7969 low-risk patients. Of these, 29 were prospective or retrospective cohorts, and 10 follow-up patients enrolled in randomized clinical trials (RCTs) or research protocols; study type was not recorded in 4 studies. All studies were carried out in patients with chronic HCV. In 39 studies, patients were treated with peg-IFN or IFN, either in combination with ribavirin or as monotherapy. In 3 studies, treatment consisted of peg-IFN, ribavirin, and a DAA (boceprevir n = 1, narlaprevir n = 1, unspecified n = 1); treatment regimen was not specified in the final study. The mean of the average follow-up post-SVR was 3.9 years (range, 1.0–8.7 years). Of the 28 studies with at least 1 recurrence, 11 used genotyping or sequencing to determine recurrence type, 5 relied on author judgment/terminology, and 12 did not classify the recurrence.
Overall, 108/7969 experienced HCV recurrence with individual study recurrence rates varying from 0.00/1000 PYFU to 70.18/1000 PYFU (Table 2). Following random effects meta-analysis, the pooled estimate for the recurrence rate was 1.85/1000 PYFU (95% CI, .71–3.35; Table 3); however, a high level of heterogeneity was observed (I2 = 73.0%). Based on this pooled estimate, the corresponding 5-year recurrence risk was 0.95% (95% CI, .35%–1.69%; Figure 2).
View this table:
In this window
In a new window
Table 2.
Hepatitis C Virus Recurrences and Rate of Recurrence in Included Studies
View this table:
In this window
In a new window
Table 3.
Meta-analysis of Recurrence
Figure 2.
View larger version:
In this page
In a new window
Download as PowerPoint Slide
Figure 2.
Summary 5-year risk (95% confidence interval) of recurrence post-sustained virological response (SVR), by risk group. Presented are the pooled estimates for the 5-year risk of recurrence after achieving an SVR. Also shown are the number of studies that were included to derive each estimate. Abbreviations: HCV, hepatitis C virus; HIV, human immunodeficiency virus.
The pooled estimate was 0.82/1000 PYFU (95% CI, .08–2.05) for late relapse, and 0.00/1000 PYFU (95% CI, .00–.00) for reinfection (Table 2). These estimates led to 5-year late relapse and reinfection rates of 0.40% (95% CI, .35%–1.05%) and 0.00% (95% CI, .00%–.00%), respectively (Figure 2).
High-risk Population
In total, 14 articles were found that assessed HCV recurrence in high-risk patients. Of these studies, 12 evaluated the risk in IDUs (n = 617) and 2 in prisoners (n = 154). In sum, 10 of 12 IDU studies were cohorts, and 2 were the long-term follow-up from RCTs. Both studies of prisoners were retrospective cohorts of patients receiving treatment while under detention. Twelve of the studies were conducted in patients with chronic HCV exclusively, and 2 studies enrolled patients with acute and chronic HCV. Patients received peg-IFN or IFN with or without ribavirin in 9 studies and either peg-IFN plus ribavirin or a DAA regimen in 1 study; 4 studies did not specify the antiviral regimen. The average of the mean follow-up post-SVR was 2.8 years (range 1.4–4.9 years). Overall, 9/13 studies with at least 1 recurrence used genotyping to classify the recurrence type.
In total, 42 recurrences were observed in a total of 771 patients. The recurrence rate varied from 0.00/1000 PYFU to 63.09/1000 PYFU in each study (Table 2); the pooled estimate for recurrence was 22.32/1000 PYFU (95% CI, 13.07–33.46) and a low level of heterogeneity was observed (I2 = 27.3%; Table 3). As shown in Figure 2, this estimate led to a 5-year recurrence rate of 10.67% (95% CI, 6.38%–15.66%) and was driven mainly by reinfection (19.06/1000 PYFU, 95% CI, 11.42–28.16) rather than late relapse.
HIV/HCV Coinfected Population
Of the 4 studies identified assessing recurrence in the HIV/HCV coinfected patients, 1 was carried out exclusively in MSM, 1 enrolled incarcerated patients only, and the remaining 2 recruited a mixed population. Two studies were cohort studies (n = 132) and two (n = 177) were long-term follow-up of RCTs. Three of the studies enrolled patients with chronic HCV, and the remaining study enrolled patients with both acute and chronic disease. Patients received peg-IFN or IFN with or without ribavirin in 3 studies; 1 study did not specify the regimen. In sum, 3 of the 4 studies reported the proportion of patients receiving antiretroviral therapy for HIV infection. In total, 78% of patients were receiving treatment ranging from 53% to 100% in the 3 studies. Of the 4 studies, 2 excluded patients with active IDU, and 2 enrolled patients with either a history of IDU or drug use during or after treatment. The average of the mean follow-up post-SVR was 3.3 years (1.6–4.3 years). One of the 3 studies reporting at least 1 recurrence used genotyping techniques to classify the recurrence.
Overall, 31/309 patients experienced a recurrence for a pooled recurrence rate of 32.02/1000 PYFU (95% CI, .00–123.49; Table 3); however, a substantial level of heterogeneity was observed and individual study recurrence rates varied from 0.00 to 133.93/1000 PYFU. The pooled rate led to a 5-year recurrence rate of 15.02% (95% CI, .00%–48.26%; Figure 2).
By recurrence type, the pooled estimate for late relapse was 0.00/1000 PYFU (95% CI, .00–.03) and for reinfection it was 32.02/1000 PYFU (95% CI, .00–123.49), leading to a 5-year risk of 0.0% (95% CI, .0%–.01%) and 15.02% (95% CI, .00%–48.26%), respectively. The uncertainties of the reinfection estimate are reflected by the wide 95% CI and the high level of heterogeneity observed.
To attempt to understand the heterogeneity, an analysis of RCTs compared with unselected patient cohorts was conducted. The pooled estimate of recurrence was significantly lower for patients followed-up after RCTs, leading to a significantly lower 5-year recurrence rate compared to the unselected cohorts (0.46% [95% CI, .00–2.65] vs 45.86% [95% CI, 32.86–58.27]). These data however should be interpreted with caution given the small number of studies available for evaluation (2 studies in each group) and the substantial between study heterogeneity observed (I2 = 98.7%).
Previous SectionNext Section
DISCUSSION
Achieving SVR substantially reduces the risk of hepatocellular carcinoma, cirrhosis, and mortality, however these benefits are lost following recurrent infection [80]. In this meta-analysis, the risk of HCV recurrence after treatment-induced SVR was found to be 1.85/1000 PYFU in the low-risk group and rose to 22.32 and 32.02/1000 PYFU in the high-risk and HIV/HCV coinfection populations, respectively. These incidence rates led to estimated 5-year recurrence rates of 0.95%, 10.67%, and 15.02% in the low-risk, high-risk, and coinfection groups, respectively. Thus, despite higher recurrence rates in those with identified ongoing risk behaviors and/or HIV infection, SVR is durable, and the great majority of patients have SVR at 5 years post-treatment.
The current analysis suggests that the greater recurrence risk in the high-risk and HIV coinfected populations is driven by an increased likelihood of reinfection, highlighting the need for prevention campaigns targeted at individuals who continue to place themselves at high-risk of HCV re-exposure. According to the inclusion criteria, the meta-analysis evaluated the risk of recurrence post-treatment. Consequently, studies evaluating spontaneous cleared were excluded [81–86]. The data from these studies support the notion that the risk of recurrence is driven by reinfection in those with high-risk behaviors [87, 88].
Included studies reported contradictory results about the risk of HCV recurrence among patients with HIV. There remains a question as to whether higher recurrence rates in HIV patients are a consequence of HIV and related immune suppression or to the presence of risk behaviors associated with HCV acquisition. Given that RCTs tend to have more restricted inclusion criteria than open cohorts, we compared recurrence between the 2 types of study. Although the number of studies was low, evidence from RCTs suggested a significantly lower recurrence rate than data from open cohorts, supporting the notion that reinfection in these patient groups, rather than an increased propensity to relapse, is the main driver to recurrence [45, 76, 78, 79].
It is important to highlight that the majority of studies included analyzed recurrence after treatment with interferon-based therapies. The use of such regimens is decreasing in favor of interferon-free regimens, and although there is no evidence to support the notion that recurrence rates may differ with new treatments, it is possible that this will be the case, particularly if the consequences of reinfection are perceived to be low. Thus, collecting prospective data on recurrence rates after treatment with newer therapies is important.
There are a number of limitations to the present study. First, it is likely that a number of spontaneously clearing recurrent infections were missed, leading to an underestimate of recurrence. Evidence indicates that the probability of spontaneously clearing recurrent infection is high, and the duration of spontaneously clearing infection is about one month [89]. Thus, HCV RNA assessment at intervals of 6–12 months, as was the case in the majority of studies, is unlikely to capture all recurrences. Second, the analysis was limited by the detection and sequencing methods utilized in the original studies. Evidence from more sensitive detection methods indicates that long-term persistence of low levels of HCV RNA is possible [90, 91]. While the clinical significance is unclear, it suggests that some patients thought to have achieved SVR may still harbor the HCV.
The use of insensitive sequencing methods has particular implications for the late relapse/reinfection subanalysis. Recent evidence with more sensitive deep sequencing techniques suggests that a number of reinfections may be wrongly classified and are actually the emergence of preexisting resistant minority variants rather than reinfection [92]. Despite this, previous evidence corroborates this analysis showing that late relapse following SVR is rare, occurring in <1% of mono- and coinfected individuals [45]. Furthermore, many recurrent cases have good outcomes, in terms of high spontaneous and treatment-induced clearance rates, supporting the mechanism of reinfection with novel susceptible virus, rather than the emergence of resistant low-level variants [93]. The distinction between late relapse and reinfection is particularly important when the epidemiological differences between risk groups are considered. In some populations, epidemics are concentrated, limiting genetic diversity such that reinfection will likely be with a highly similar strain, and thus will require better techniques to distinguish late relapse from reinfection [94].
In those studies not utilizing genotyping methods, bias may have been introduced by the tendency of study authors to classify recurrent infection as late relapse vs reinfection. Indeed, the late relapse rate was highest in the low risk group, suggesting recurrences were more likely attributed to late relapse over reinfection, possibly overestimating the relapse rate in this population. Similarly, in high-risk groups, relapse may have been underestimated by the tendency to classify recurrence as reinfection when uncertain. Finally, the estimates of late relapse and reinfection may have been biased by the availability of studies for inclusion in these analyses. Studies not classifying recurrence were excluded meaning that zero event studies were overrepresented in calculations, possibly leading to an underestimate of the true relapse and reinfection rates.
Despite the limitations, the results of the analysis will be helpful to inform treatment scale-up and modeling of strategies, which prioritize different groups for therapy with the ultimate goal of disease eradication. Although the probability of late relapse is low, reinfection in high-risk groups such as IDUs, prisoners, and HIV-positive MSM present both a challenge and an opportunity for epidemic control. As such, strategies to minimize the risk of reinfection in high-risk groups need to be intensified in parallel to introduction of interferon-free regimens in order to curtail onward transmission. The current analysis highlights the notion that estimates from RCTs may underestimate recurrence and emphasizes the need for real-life analyses and an updated analysis once the results of long-term interferon-free studies are available.
Previous SectionNext Section
Supplementary Data
Supplementary materials are available at http://cid.oxfordjournals.org. Consisting of data provided by the author to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the author, so questions or comments should be addressed to the author.
Previous SectionNext Section
Notes
Acknowledgments. We would like to thank Emma Thomson for her advice and helpful comments.
Financial support. This work was funded in part by UNITAID. G. S. C. is funded in part by the Biomedical Research Centre of Imperial College National Health Service trust and supported by the Medical Research Council (MRC) STOP hepatitis C virus (HCV) consortium.
Potential conflicts of interest. A. H. has received consultancy payments from Janssen, not connected with this project. G. S. C. has received consultancy payments and funding for HCV clinical trials from pharmaceutical companies not connected with this project. R. D. R. reports no conflicts but has received funding from the MRC for HCV treatment and outcomes and prognosis research work. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
Methods. A search identified studies analysing HCV recurrence post-SVR. The recurrence rate for each study was calculated using events/person years of follow-up (PYFU). Results were pooled using a random-effects model and used to calculate 5-year recurrence risk. Three patient groups were analysed: (1) Mono-HCV infected “low-risk” patients; (2) Mono-HCV infected “high-risk” patients (injecting drug users or prisoners); (3) human immunodeficiency virus (HIV)/HCV coinfected patients. Recurrence was defined as confirmed HCV RNA detectability post-SVR.
Results. In the 43 studies of HCV mono-infected “low-risk” patients (n = 7969) the pooled recurrence rate was 1.85/1000 PYFU (95% confidence interval [CI], .71–3.35; I2 = 73%) leading to a summary 5-year recurrence risk of 0.95% (95% CI, .35%–1.69%). For the 14 studies of HCV monoinfected “high-risk” patients (n = 771) the pooled recurrence rate was 22.32/1000 PYFU (95% CI, 13.07–33.46; I2 = 27%) leading to a summary 5-year risk of 10.67% (95% CI, 6.38%–15.66%). For the 4 studies of HIV/HCV coinfected patients the pooled recurrence rate was 32.02/1000 PYFU (95% CI, .00–123.49; I2 = 96%) leading to a summary 5-year risk of 15.02% (95% CI, .00%–48.26%). The higher pooled estimates of recurrence in the high-risk and coinfected cohorts were driven by an increase in reinfection rather than late relapse.
Conclusions. SVR appears durable in the majority of patients at 5 years post-treatment. The large difference in 5 year event rate by risk group is driven mainly by an increased reinfection risk.
Infection with the hepatitis C virus (HCV) is a significant public health concern associated with a high burden of morbidity and mortality [1, 2]. Recent estimates suggest that worldwide, of the 185 million individuals infected, over 700 000 people die annually as a result of infection [3, 4].
The attainment of a sustained virological response (SVR), defined as aviremia 12 or 24 weeks after the completion of antiviral therapy (SVR12 or SVR24), is associated with an improved prognosis compared with patients either untreated or failing therapy. These benefits include improved histology, reduced risk of hepatocellular carcinoma, and improved overall survival [5, 6].
Despite these benefits, treatment uptake for chronic HCV has been low due to complexities of treatment and poor success rates. The availability of new highly efficacious regimens provides the foundation for marked treatment scale-up; however, high costs are currently limiting access [7–10].
One challenge to treatment scale-up is the risk of HCV recurrence, either as late relapse post-SVR or reinfection following treatment. HCV recurrence is a particular concern in patients with ongoing high-risk behaviors, such as injecting drug users (IDUs), who are more susceptible to reinfection, and also patients coinfected with human immunodeficiency virus (HIV) who may be at increased risk of relapse due to their immunocompromised status [11–15].
A number of studies have been carried out to examine the durability of treatment-induced SVR in patients with chronic HCV in a variety of patient populations. Our aim was to systematically review the existing evidence and undertake meta-analysis to provide summary estimates of the recurrence rate by risk group. The secondary aim was to evaluate the contribution of late relapse and of reinfection to the recurrence rate. This work fits within the theme one of the PROGRESS framework for prognosis research (“fundamental prognosis research”) and will provide a clearer understanding of HCV recurrence to inform the provision of antiviral therapy [16].
Previous SectionNext Section
METHODS
Search Strategy and Inclusion Criteria
The MEDLINE database was searched from 1990 until 1 March 2015 for studies analyzing HCV recurrence post-SVR. A sensitive search string was developed using terms including hepatitis C, treatment, SVR, recurrence, relapse, and reinfection (Supplementary Appendix). The reference lists of articles were thoroughly searched to identify additional articles. Lastly, the proceedings of the following conferences were search for additional studies: International Liver Congress (EASL), The Liver Meeting (AASLD), Conference on Retroviruses and Opportunistic Infections, and the International AIDS Conference.
Studies included were to have enrolled adult patients (aged ≥18) who achieved SVR after antiviral treatment for acute or chronic HCV. SVR was defined as undetectable HCV RNA 12 or 24 weeks post-treatment. There was no stipulated method of HCV acquisition or specific antiviral treatment regimen. There were no restrictions on study design however all studies were to have a follow-up longer than 6 months post-SVR. Studies were excluded if they examined rate of recurrence after spontaneous clearance, or if they measured recurrences after the end of treatment, not allowing for the SVR time period to elapse.
Studies were categorized in to 3 groups: (1) Low-risk population, inclusive of studies of mono-HCV infected patients with no recognized risk factors for reinfection; (2) High-risk population, inclusive of studies of mono-HCV infected patients with at least 1 identified risk factor for reinfection; and (3) HIV/HCV coinfection populations, inclusive of all studies of HIV/HCV coinfected persons, regardless of the presence or absence of other risk factors. Risk factors for reinfection were defined as current or former IDU, imprisonment, and men who have sex with men (MSM). Studies of liver transplant recipients were excluded.
Quality Assessment
Articles meeting the inclusion criteria were assessed for methodological quality using the Newcastle–Ottawa Scale (NOS). The assessment was modified to allocate a maximum of 8 stars, for quality of selection, comparability, exposure, and outcome of study participants (Supplementary Appendix). Studies with a NOS rating ≥6 were considered high-quality.
Data Extraction
The following data were extracted for each study: location, design, recruitment, patient characteristics, average follow-up time, number of HCV recurrences, total PYFU, and frequency of HCV RNA assessment. HCV recurrence was defined as confirmed HCV RNA detectability post-SVR. Where possible, recurrence was characterized as either late relapse or as reinfection, with categorization carried out according to the original study definitions and techniques. In all studies using phylogenetic techniques late relapse was defined as detection of HCV RNA of the same virus lineage and reinfection as identification of a different virus. In the majority of studies, this classification was according to the protocol in the original article. In genotyping studies where no criteria for classification were given, the same definitions were applied by the authors of the current meta-analysis. In some studies, categorization was done by the study authors without confirmatory genotyping. In these studies, the decision to classify as late relapse or reinfection was usually made through consultation with patients to assess for the presence or absence of risk behaviors (eg, injecting drug use, unsafe procedures, etc.).
PYFU were accrued from the SVR time-point; in those studies where follow-up originated at the end-of-treatment, PYFU were appropriately adjusted. If total PYFU was not explicitly stated, it was estimated from the average follow-up time; studies in which PYFU was inestimable were excluded. In the case of study duplications, the article providing the most comprehensive account of the study population and longest follow-up period was used.
The literature search, data extraction, and quality assessment were carried out independently by 2 authors (B. S., J. S.), and any differences were resolved by consensus.
Data Synthesis
For each study, the incidence rate of HCV recurrence was calculated as the number of recurrences per 1000 PYFU and was reported with the corresponding 95% Wilson confidence interval (95% CI). Given the rarity of events, estimates were transformed using the Freeman-Tukey double arcsine transformation [17, 18]. A pooled estimate for recurrence was then calculated for each of the three groups separately using a random-effects model [19]. In addition, meta-analyses of the rate of late relapse and of the rate of reinfection were carried out including studies providing this data. The pooled estimates were used to calculate the 5-year event rate for recurrence, late relapse, and reinfection for each population. The summary 5-year risk was calculated using 1 – (1 – pooled incidence rate)5 and as such assumed that the pooled rate of recurrence was constant over the follow-up duration. For each calculation, the degree of heterogeneity between studies was quantitatively assessed using I2 and tau2, where an I2 ≥ 50% may indicate substantial heterogeneity and ≥75% is indicative of considerable heterogeneity. The existence of publication bias was evaluated by observational analysis of funnel plots. All analyses were conducted using STATA version 13 (StataCorp LP, Texas).
Previous SectionNext Section
RESULTS
As shown in Figure 1, a total of 1180 references were identified and screened for eligibility. Of these, results were available from 59 studies reporting on recurrence post-SVR in a total of 9049 patients. Two studies evaluated two distinct subgroups of monoinfected and HIV coinfected patients and as such were included in 2 analysis groups. Of the studies deemed possibly relevant and screened against inclusion criteria, the main reasons for exclusion were the assessment of recurrence rate after spontaneous clearance and the lack of an SVR time period after the end-of-treatment. The study and cohort characteristics are shown in Table 1. All identified studies evaluated SVR at 24 weeks post-treatment; no studies eligible for inclusion used SVR12 as the endpoint for analysis. Frequency of HCV RNA assessment varied from every 3 months to 1 single assessment during follow-up. For all 3 risk groups, funnel plots appeared symmetrical indicating no evidence of bias. Of all studies, 49/59 (83%) were considered high-quality (NOS score ≥6). The main biases observed were in determining PYFU and in accepting the authors' opinion regarding reinfection vs relapse.
View this table:
In this window
In a new window
Table 1.
Study Characteristics of Included Studies
Figure 1.
View larger version:
In this page
In a new window
Download as PowerPoint Slide
Figure 1.
Flow diagram of study selection for systematic review of hepatitis C virus (HCV) recurrence in patients achieving a sustained virologic response after treatment for HCV infection. Low-risk studies include those examining recurrence in general populations and high-risk studies include those studying patients with at least 1 reinfection risk factor (injecting drug use or prison populations). Human immunodeficiency virus (HIV)/HCV coinfected studies include all those of coinfected participants, regardless of risk factors. Total studies in the 3 groups does not equal the total number of studies identified as 2 studies examined 2 populations.
Low-risk Population
Forty-three articles were found evaluating the risk of recurrence in 7969 low-risk patients. Of these, 29 were prospective or retrospective cohorts, and 10 follow-up patients enrolled in randomized clinical trials (RCTs) or research protocols; study type was not recorded in 4 studies. All studies were carried out in patients with chronic HCV. In 39 studies, patients were treated with peg-IFN or IFN, either in combination with ribavirin or as monotherapy. In 3 studies, treatment consisted of peg-IFN, ribavirin, and a DAA (boceprevir n = 1, narlaprevir n = 1, unspecified n = 1); treatment regimen was not specified in the final study. The mean of the average follow-up post-SVR was 3.9 years (range, 1.0–8.7 years). Of the 28 studies with at least 1 recurrence, 11 used genotyping or sequencing to determine recurrence type, 5 relied on author judgment/terminology, and 12 did not classify the recurrence.
Overall, 108/7969 experienced HCV recurrence with individual study recurrence rates varying from 0.00/1000 PYFU to 70.18/1000 PYFU (Table 2). Following random effects meta-analysis, the pooled estimate for the recurrence rate was 1.85/1000 PYFU (95% CI, .71–3.35; Table 3); however, a high level of heterogeneity was observed (I2 = 73.0%). Based on this pooled estimate, the corresponding 5-year recurrence risk was 0.95% (95% CI, .35%–1.69%; Figure 2).
View this table:
In this window
In a new window
Table 2.
Hepatitis C Virus Recurrences and Rate of Recurrence in Included Studies
View this table:
In this window
In a new window
Table 3.
Meta-analysis of Recurrence
Figure 2.
View larger version:
In this page
In a new window
Download as PowerPoint Slide
Figure 2.
Summary 5-year risk (95% confidence interval) of recurrence post-sustained virological response (SVR), by risk group. Presented are the pooled estimates for the 5-year risk of recurrence after achieving an SVR. Also shown are the number of studies that were included to derive each estimate. Abbreviations: HCV, hepatitis C virus; HIV, human immunodeficiency virus.
The pooled estimate was 0.82/1000 PYFU (95% CI, .08–2.05) for late relapse, and 0.00/1000 PYFU (95% CI, .00–.00) for reinfection (Table 2). These estimates led to 5-year late relapse and reinfection rates of 0.40% (95% CI, .35%–1.05%) and 0.00% (95% CI, .00%–.00%), respectively (Figure 2).
High-risk Population
In total, 14 articles were found that assessed HCV recurrence in high-risk patients. Of these studies, 12 evaluated the risk in IDUs (n = 617) and 2 in prisoners (n = 154). In sum, 10 of 12 IDU studies were cohorts, and 2 were the long-term follow-up from RCTs. Both studies of prisoners were retrospective cohorts of patients receiving treatment while under detention. Twelve of the studies were conducted in patients with chronic HCV exclusively, and 2 studies enrolled patients with acute and chronic HCV. Patients received peg-IFN or IFN with or without ribavirin in 9 studies and either peg-IFN plus ribavirin or a DAA regimen in 1 study; 4 studies did not specify the antiviral regimen. The average of the mean follow-up post-SVR was 2.8 years (range 1.4–4.9 years). Overall, 9/13 studies with at least 1 recurrence used genotyping to classify the recurrence type.
In total, 42 recurrences were observed in a total of 771 patients. The recurrence rate varied from 0.00/1000 PYFU to 63.09/1000 PYFU in each study (Table 2); the pooled estimate for recurrence was 22.32/1000 PYFU (95% CI, 13.07–33.46) and a low level of heterogeneity was observed (I2 = 27.3%; Table 3). As shown in Figure 2, this estimate led to a 5-year recurrence rate of 10.67% (95% CI, 6.38%–15.66%) and was driven mainly by reinfection (19.06/1000 PYFU, 95% CI, 11.42–28.16) rather than late relapse.
HIV/HCV Coinfected Population
Of the 4 studies identified assessing recurrence in the HIV/HCV coinfected patients, 1 was carried out exclusively in MSM, 1 enrolled incarcerated patients only, and the remaining 2 recruited a mixed population. Two studies were cohort studies (n = 132) and two (n = 177) were long-term follow-up of RCTs. Three of the studies enrolled patients with chronic HCV, and the remaining study enrolled patients with both acute and chronic disease. Patients received peg-IFN or IFN with or without ribavirin in 3 studies; 1 study did not specify the regimen. In sum, 3 of the 4 studies reported the proportion of patients receiving antiretroviral therapy for HIV infection. In total, 78% of patients were receiving treatment ranging from 53% to 100% in the 3 studies. Of the 4 studies, 2 excluded patients with active IDU, and 2 enrolled patients with either a history of IDU or drug use during or after treatment. The average of the mean follow-up post-SVR was 3.3 years (1.6–4.3 years). One of the 3 studies reporting at least 1 recurrence used genotyping techniques to classify the recurrence.
Overall, 31/309 patients experienced a recurrence for a pooled recurrence rate of 32.02/1000 PYFU (95% CI, .00–123.49; Table 3); however, a substantial level of heterogeneity was observed and individual study recurrence rates varied from 0.00 to 133.93/1000 PYFU. The pooled rate led to a 5-year recurrence rate of 15.02% (95% CI, .00%–48.26%; Figure 2).
By recurrence type, the pooled estimate for late relapse was 0.00/1000 PYFU (95% CI, .00–.03) and for reinfection it was 32.02/1000 PYFU (95% CI, .00–123.49), leading to a 5-year risk of 0.0% (95% CI, .0%–.01%) and 15.02% (95% CI, .00%–48.26%), respectively. The uncertainties of the reinfection estimate are reflected by the wide 95% CI and the high level of heterogeneity observed.
To attempt to understand the heterogeneity, an analysis of RCTs compared with unselected patient cohorts was conducted. The pooled estimate of recurrence was significantly lower for patients followed-up after RCTs, leading to a significantly lower 5-year recurrence rate compared to the unselected cohorts (0.46% [95% CI, .00–2.65] vs 45.86% [95% CI, 32.86–58.27]). These data however should be interpreted with caution given the small number of studies available for evaluation (2 studies in each group) and the substantial between study heterogeneity observed (I2 = 98.7%).
Previous SectionNext Section
DISCUSSION
Achieving SVR substantially reduces the risk of hepatocellular carcinoma, cirrhosis, and mortality, however these benefits are lost following recurrent infection [80]. In this meta-analysis, the risk of HCV recurrence after treatment-induced SVR was found to be 1.85/1000 PYFU in the low-risk group and rose to 22.32 and 32.02/1000 PYFU in the high-risk and HIV/HCV coinfection populations, respectively. These incidence rates led to estimated 5-year recurrence rates of 0.95%, 10.67%, and 15.02% in the low-risk, high-risk, and coinfection groups, respectively. Thus, despite higher recurrence rates in those with identified ongoing risk behaviors and/or HIV infection, SVR is durable, and the great majority of patients have SVR at 5 years post-treatment.
The current analysis suggests that the greater recurrence risk in the high-risk and HIV coinfected populations is driven by an increased likelihood of reinfection, highlighting the need for prevention campaigns targeted at individuals who continue to place themselves at high-risk of HCV re-exposure. According to the inclusion criteria, the meta-analysis evaluated the risk of recurrence post-treatment. Consequently, studies evaluating spontaneous cleared were excluded [81–86]. The data from these studies support the notion that the risk of recurrence is driven by reinfection in those with high-risk behaviors [87, 88].
Included studies reported contradictory results about the risk of HCV recurrence among patients with HIV. There remains a question as to whether higher recurrence rates in HIV patients are a consequence of HIV and related immune suppression or to the presence of risk behaviors associated with HCV acquisition. Given that RCTs tend to have more restricted inclusion criteria than open cohorts, we compared recurrence between the 2 types of study. Although the number of studies was low, evidence from RCTs suggested a significantly lower recurrence rate than data from open cohorts, supporting the notion that reinfection in these patient groups, rather than an increased propensity to relapse, is the main driver to recurrence [45, 76, 78, 79].
It is important to highlight that the majority of studies included analyzed recurrence after treatment with interferon-based therapies. The use of such regimens is decreasing in favor of interferon-free regimens, and although there is no evidence to support the notion that recurrence rates may differ with new treatments, it is possible that this will be the case, particularly if the consequences of reinfection are perceived to be low. Thus, collecting prospective data on recurrence rates after treatment with newer therapies is important.
There are a number of limitations to the present study. First, it is likely that a number of spontaneously clearing recurrent infections were missed, leading to an underestimate of recurrence. Evidence indicates that the probability of spontaneously clearing recurrent infection is high, and the duration of spontaneously clearing infection is about one month [89]. Thus, HCV RNA assessment at intervals of 6–12 months, as was the case in the majority of studies, is unlikely to capture all recurrences. Second, the analysis was limited by the detection and sequencing methods utilized in the original studies. Evidence from more sensitive detection methods indicates that long-term persistence of low levels of HCV RNA is possible [90, 91]. While the clinical significance is unclear, it suggests that some patients thought to have achieved SVR may still harbor the HCV.
The use of insensitive sequencing methods has particular implications for the late relapse/reinfection subanalysis. Recent evidence with more sensitive deep sequencing techniques suggests that a number of reinfections may be wrongly classified and are actually the emergence of preexisting resistant minority variants rather than reinfection [92]. Despite this, previous evidence corroborates this analysis showing that late relapse following SVR is rare, occurring in <1% of mono- and coinfected individuals [45]. Furthermore, many recurrent cases have good outcomes, in terms of high spontaneous and treatment-induced clearance rates, supporting the mechanism of reinfection with novel susceptible virus, rather than the emergence of resistant low-level variants [93]. The distinction between late relapse and reinfection is particularly important when the epidemiological differences between risk groups are considered. In some populations, epidemics are concentrated, limiting genetic diversity such that reinfection will likely be with a highly similar strain, and thus will require better techniques to distinguish late relapse from reinfection [94].
In those studies not utilizing genotyping methods, bias may have been introduced by the tendency of study authors to classify recurrent infection as late relapse vs reinfection. Indeed, the late relapse rate was highest in the low risk group, suggesting recurrences were more likely attributed to late relapse over reinfection, possibly overestimating the relapse rate in this population. Similarly, in high-risk groups, relapse may have been underestimated by the tendency to classify recurrence as reinfection when uncertain. Finally, the estimates of late relapse and reinfection may have been biased by the availability of studies for inclusion in these analyses. Studies not classifying recurrence were excluded meaning that zero event studies were overrepresented in calculations, possibly leading to an underestimate of the true relapse and reinfection rates.
Despite the limitations, the results of the analysis will be helpful to inform treatment scale-up and modeling of strategies, which prioritize different groups for therapy with the ultimate goal of disease eradication. Although the probability of late relapse is low, reinfection in high-risk groups such as IDUs, prisoners, and HIV-positive MSM present both a challenge and an opportunity for epidemic control. As such, strategies to minimize the risk of reinfection in high-risk groups need to be intensified in parallel to introduction of interferon-free regimens in order to curtail onward transmission. The current analysis highlights the notion that estimates from RCTs may underestimate recurrence and emphasizes the need for real-life analyses and an updated analysis once the results of long-term interferon-free studies are available.
Previous SectionNext Section
Supplementary Data
Supplementary materials are available at http://cid.oxfordjournals.org. Consisting of data provided by the author to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the author, so questions or comments should be addressed to the author.
Previous SectionNext Section
Notes
Acknowledgments. We would like to thank Emma Thomson for her advice and helpful comments.
Financial support. This work was funded in part by UNITAID. G. S. C. is funded in part by the Biomedical Research Centre of Imperial College National Health Service trust and supported by the Medical Research Council (MRC) STOP hepatitis C virus (HCV) consortium.
Potential conflicts of interest. A. H. has received consultancy payments from Janssen, not connected with this project. G. S. C. has received consultancy payments and funding for HCV clinical trials from pharmaceutical companies not connected with this project. R. D. R. reports no conflicts but has received funding from the MRC for HCV treatment and outcomes and prognosis research work. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
No comments:
Post a Comment