Introduction

Chronic hepatitis C virus (HCV) infection affects more than 170 million people worldwide and is a major cause of cirrhosis and hepatocellular carcinoma (HCC) [1, 2]. Currently the standard of care for patients with chronic hepatitis C is peginterferon (PEG-IFN) plus ribavirin (RBV) therapy, which can induce a sustained virological response (SVR) in 40–50 % of treatment-naïve patients with genotype 1 and an SVR of approximately 80 % in treatment-naïve patients with genotypes 2 or 3 [3–8]. After an SVR is achieved, the risk of developing liver failure and HCC is greatly reduced [9]. However, this treatment is associated with various types of complications, some of which lead to fatal outcomes. Because death during treatment is a rare event, a large sample size is needed to accurately assess the mortality rate and risk factors. The aim of this systematic review was to assess the rates of serious adverse events (SAEs) and mortality during PEG-IFN/RBV therapy in a pooled large-sized sample and to assess the relationship between mortality and SAE rates and therapeutic characteristics.

Methods

Study search protocol

We searched MEDLINE, EMBASE, and the Cochrane Library to identify randomized controlled trials (RCTs) evaluating the efficacy and safety of PEG-IFN/RBV therapy published between December 1999 and October 2010. We used the following search terms: chronic hepatitis C, interferon, and pegylated or peg or peginterferon or pegasys or pegintron. The search was limited to the English language.

Inclusion criteria

Studies were included in the analysis if: (1) they were RCTs, (2) they included at least one PEG-IFN/RBV treatment group in patients with chronic hepatitis C, (3) they clearly specified adverse events, (4) patients were followed up until at least 24 weeks after the end of treatment, and (5) the studies had been published or accepted for publication as full-length articles. Studies were excluded if: (1) they dealt only with co-infection of HCV and HIV; (2) they dealt only with patients with a specific condition such as a comorbid disease (e.g., cryoglobulinemia), status after liver transplantation, or patients on dialysis; (3) they included patients under 18 years of age; or (4) they focused on specific adverse events or only on hemodynamic status. We restricted the included studies to RCTs on the premise that the quality of RCTs is superior to that of non-randomized or retrospective studies.

Data extraction

Two authors (T. M. and T. K.) independently screened titles and abstracts for potential eligibility and the full texts for final eligibility. Disagreements were resolved by consensus or by consulting a third author (R. T.). We extracted the data using a standardized data-collection form to record details of the study design, treatment doses and duration, number of patients in the arm, patient characteristics, and outcomes. A database using Microsoft Access 2010 (Microsoft, Redmond, WA, USA) was developed specifically for that purpose. Two authors independently entered data into the form, and the data were then compared. Any discrepancies were checked and resolved by consensus.

Statistical analysis

The primary and secondary outcome measures were mortality and SAE rates during PEG-IFN/RBV therapy, respectively. We recorded the number of SAEs and deaths observed in each arm of the included studies. We calculated crude mortality and SAE rates with 95 % confidence intervals (CI) by dividing the total number of deaths or SAEs observed by the total number of patients in the relevant group. Studies that did not discriminate serious adverse events from others were not included in the analyses of the corresponding outcome. We performed a subgroup analysis by comparing mortality and SAE rates between PEG-IFN alpha-2a and alpha-2b, high-dose and low-dose PEG-IFN, and shorter and longer treatment durations. We also performed a meta-regression analysis to investigate relationships between mortality and SAE rates and continuous variables (mean age; mean body weight; proportion of males; proportion of Caucasian, African, and Asian patients; and proportion of genotype 1 patients) using the random effects model. Heterogeneity was tested using the I 2 test to calculate the percentage of variation caused by heterogeneity rather than by chance alone [10]. The analyses were performed with S-plus Ver. 7.0 (Insightful, Seattle, WA, USA) and StatsDirect version 2.7.7 (StatsDirect, Chesire, UK). The threshold of the reported P value accepted as indicating significance was <0.05.

Results

Study characteristics

Figure 1 shows the results of the screening. Our initial database search retrieved 323 citations, of which 243 were excluded because they did not meet our inclusion criteria; therefore, a total of 80 RCTs with 153 arms that included 27569 patients were enrolled. Table 1 shows the characteristics of each enrolled study. All studies contained at least one treatment arm using PEG-IFN and RBV for chronic hepatitis C. There were 16797 male and 10254 females and the sex was not reported in the remaining 518 patients. The mean age was 45.9 years. A total of 14401 patients were treated with PEG-IFN alpha-2a/RBV, and 13168 with PEG-IFN alpha-2b/RBV. PEG-IFN alpha-2a was used at a fixed dose of 180–360 μg/body/week, and PEG-IFN alpha-2b at weight-based doses of 0.35–3.0 μg/kg/week. The RBV dose was fixed in 36 treatment arms and weight-based in 117 treatment arms. Treatment duration ranged from 12 to 72 weeks. The numbers of patients with HCV genotypes 1, 2, 3, and 4–6 were 18082, 3427, 3519, and 842, respectively. The genotype was not reported in the remaining 1699 patients. The treatment protocol is described in Supplementary Table 1.

Fig. 1
figure 1

The literature-search and study-selection process. SAE serious adverse events

Full size image
Table 1 Characteristics of the enrolled studies
Full size table

Primary outcome

A total of 50 deaths from all causes were observed in the enrolled studies. Of these, sixteen were considered by the authors to be treatment-related. The crude overall and treatment-related mortality rates were 0.18 % (95 % CI 0.13–0.24 %) and 0.058 % (0.033–0.094 %), respectively. There was no evidence of heterogeneity among studies for mortality and treatment-related mortality (I 2 = 0 in both). The causes of mortality were suicide (N = 6), drug intoxication (N = 6), myocardial infarction (N = 3), sepsis (N = 3), aortic dissection (N = 2), traffic accident (N = 2), HCC (N = 2), rupture of a cerebral aneurysm (N = 1), bronchitis (N = 1), syncope (N = 1), pulmonary tuberculosis (N = 1), and unknown causes (N = 22). The six cases of suicide, the three of sepsis, two of drug intoxication, two of myocardial infarction, one of HCC, one of syncope, and one of pulmonary tuberculosis were considered by the investigators to be treatment-related mortality.

Subgroup analysis did not reveal any difference in mortality according to type and dose of PEG-IFN or in relation to duration of treatment (Fig. 2).

Fig. 2
figure 2

Forest plot of mortality, comparing treatment regimens. Sizes of the boxes reflect sample sizes, with the bars showing the 95 % confidence interval (CI)

Full size image

Secondary outcome

Seventy-two studies with 135 treatment arms including 23996 patients reported SAEs. They reported SAEs such as anemia requiring transfusion, neutropenia below 500/mm3, hypothyroidism, psychosis, pneumonia, and cellulitis. SAEs were not discriminated from other adverse events in the remaining eighteen studies. The crude SAE rate was 7.08 % (95 % CI 6.75–7.41 %). Significant heterogeneity among studies was found for this outcome (I 2 = 94.1 %). SAE rates were higher in PEG-IFN alpha-2a than in PEG-IFN alpha-2b (7.45 vs. 6.74 %). In a subgroup analysis of the type and dose of PEG-IFN and duration of treatment, higher SAE rates were observed for intensive (270–360 μg) than for standard (180 μg) doses and for extended (>48 weeks) than for standard (48 weeks) treatment duration in patients treated with PEG-IFN alpha-2a, and for standard (1.5 μg/kg) than for lower (≤1.0 μg/kg) doses in patients treated with PEG-IFN alpha-2b (Fig. 3). However, heterogeneity remained evident in all subgroups.

Fig. 3
figure 3

Forest plot of SAEs, comparing treatment regimens. Sizes of the boxes reflect sample sizes, with the bars showing the 95 % confidence interval

Full size image

Meta-regression analysis showed that greater body weight, an increased proportion of male patients, an increased proportion of HCV genotype 1, and an increased proportion of Caucasian patients and decreased proportion of Asian patients were significantly associated with increased SAE rates (Table 2). There was no significant association between increased SAE rates and the mean patient age or the proportion of African patients.

Table 2 Meta-regression analysis for continuous variables
Full size table

Discussion

According to a report by the World Health Organization (WHO), age-specific annual death rates from all causes in individuals aged 45–49 years in the United States, Italy, and Japan (locations of the majority of our enrolled studies) were 409, 216, and 248 per 100000, respectively [11]. The crude mortality of 0.18 % (180 per 100000) found in the present study is low by comparison, even allowing for the biased population tolerable to PEG-IFN/RBV. Furthermore, the annual mortality rate could have been lower than the crude mortality rate, considering that the study period was longer than 1 year (including the follow-up period) in most enrolled studies. The annual treatment-related mortality rate could have been lower than our finding of a treatment-related mortality of 0.06 %. However, the treatment-related mortality rate may be an underestimate, as assessment of the causal relationship between treatment and mortality can be subjective and/or biased. Nonetheless, these PEG-IFN/RBV-related mortality rates would be acceptable considering the high SVR rates and considering that SVR drastically reduces adverse events related to chronic hepatitis C infection. In the present study, the most common cause of mortality was suicide, and all of the suicides were considered as treatment-related. This finding should alert treating physicians when they are treating patients with a history of psychiatric illness.

Two types of PEG-IFNs (i.e., PEG-IFN alpha-2a and 2b) are approved for the treatment of chronic hepatitis C. PEG-IFN alpha-2a has a molecular mass of 40 kDa and PEG-IFN alpha-2b a mass of 12 kDa. In comparison with PEG-IFN alpha-2b, PEG-IFN alpha-2a is less effectively cleared by the kidneys and therefore has a longer half-life. In fact, pharmacokinetic analysis in 22 patients showed that PEG-IFN alpha-2a was still detectable in 10 patients 168 h after the administration of 180 μg/week, whereas the administration of 1.0 μg/kg/week of PEG-IFN alpha-2b was undetectable in 11 of 12 patients at the same time point [12]. PEG-IFN alpha-2a is thought to be more effective than PEG-IFN alpha-2b because of its longer half-life. A recent meta-analysis showed a higher SVR rate after treatment with PEG-IFN alpha-2a than after treatment with PEG-IFN alpha-2b [13]. On the other hand, the half-life of each PEG-IFN may be related to its safety profile. However, among studies that have directly compared the safety of the two PEG-IFNs, only one reported a significant difference between SAE rates for PEG-IFN alpha-2a and PEG-IFN alpha-2b (11.7 vs. 8.6 %, P = 0.02) [14–18]. The inability of the other studies to detect such a difference may have been due to small sample sizes. In fact, a difference in SAE rates between the two PEG-IFNs was observed in pooled samples in our study.

Increasing the dose intensity of PEG-IFN and prolonging treatment duration have been attempted to achieve higher IFN levels in blood for longer periods, eventually resulting in a higher SVR rate. Treatment dose and duration are also expected to be related to the safety profile. The higher SAE rates in regimens with more intensive dosing observed for PEG-IFN alpha-2a and 2b and longer treatment duration observed for PEG-IFN alpha-2a support this hypothesis. The higher SAE rates in regimens with longer treatment duration were not observed for PEG-IFN alpha-2b, probably due to small sample sizes in regimens with longer treatment duration of PEG-IFN alpha-2b.

As mortality and SAE during PEG-IFN/RBV treatment are rare, most studies reported no such events. Therefore, the proportion calculated using the DerSimonian and Laird weight for the random-effect model showed considerable discrepancies between crude and pooled rates. In fact, pooled and treatment-related mortalities calculated using the random-effects (DerSimonian and Laird) model were 0.30 % (0.24–0.37 %) and 0.17 % (0.12–0.22 %), respectively, which were considerably different from the crude rates of each outcome (data not shown). Thus, we adopted crude instead of pooled rates for mortality and SAE.

Our meta-regression analysis showed a significant association between increased SAEs and HCV genotype 1. It is plausible that patients with genotype 1, which is difficult to treat, received a higher dose and longer duration of treatment. This is consistent with the results of the subgroup analysis.

A significant positive association between the SAE rate and the proportion of Caucasian patients, and an inverse relationship between SAEs and the proportion of Asian patients were also observed. This result may suggest a role of genetic diversity in the mechanisms underlying the adverse effects of PEG-IFN/RBV. Indeed, inosine triphosphate pyrophosphatase (ITPA) gene variants are associated with RBV-induced hemolytic anemia, and genetic polymorphisms near the interleukin-28B (IL-28B) gene were reported to be associated with response to HCV treatment with PEG-IFN and RBV, and the frequency of the variants differed between ethnic groups [19, 20].

We found that greater body weight was associated with a higher SAE rate. Of note, in the PEG-IFN alpha-2a-based regimen, the starting dose was fixed regardless of body weight; thus, with the PEG-IFN alpha-2a regimen, there might have been an overdose for patients of lower weight, leading to SAEs. However, whether such overdosing occurred was not clear in this study because there was a positive correlation between body weight and the SAE rate in patients receiving the PEG-IFN alpha-2a regimen. The reason for this positive relationship remains unclear; however, it may be because obesity is itself associated with various medical comorbidities.

We also found that an increased proportion of male patients in a study was associated with a higher SAE rate. It has been reported that female gender was an independent factor contributing to severe anemia [21], so the reason for the present finding of the increased proportion of male patients remains unclear; it may be correlated with increased body weight which caused a higher SAE rate. However, whether the proportions of individuals with obesity differed between male and female patients is not clear, because data on body mass index was often lacking.

In the present study increased mean age was not associated with a higher SAE rate, whereas discontinuation, dose reduction, and grade 3 adverse events were more frequent in older patients in previous studies [22, 23]. The lack of an association between mean age and the SAE rate in the present study could be due in part to the patients’ mean age of 45.9 years, and the proportion of patients over 60 being small. Low-risk patients tend to be included in RCTs. This is one of the limitations of this study.

Recently, the use of HCV nonstructural 3/4A serine protease inhibitors combined with PEG-IFN and RBV were reported to achieve higher SVR rates in genotype 1 patients compared with conventional PEG-IFN/RBV. These triple therapies are considered to be the next standard of care for chronic hepatitis C [24, 25]. Adverse events during triple therapies could include those related to PEG-IFN/RBV, as these regimens include PEG-IFN/RBV.

We extracted only RCTs for our analysis in order to obtain highly reliable data and minimize the influence of recall bias because RCTs are prospectively designed, and SAEs should be defined a priori. However, several limitations are still worth noting. The latent limitation of this study is inter-study variability in the definition of SAE. The precise meaning of ‘serious’ has not been determined, and some discrepancies between studies exist. These discrepancies may diminish the accuracy of the pooled SAE rate in this study. Second, even by choosing only RCTs, we could not completely exclude the influence of publication bias.

Overall, PEG-IFN/RBV treatment is relatively safe, with low mortality, considering the fact that chronic hepatitis C patients carry a high risk of cirrhosis and HCC. Nevertheless, the SAE rate with this treatment is not negligible and the development of safer regimens should be, and is, encouraged.