Background

Hepatitis C virus (HCV) infection and chronic kidney disease (CKD) are epidemically correlated and clinically challenging.

It’s estimated that 71 million people were chronically infected with HCV globally, and around 10% of them live with CKD [1, 2]. For patients with CKD, particularly patients receiving hemodialysis, the incidence of HCV is much higher than general population, ranging from 3 to 50% [3]. HCV infection significantly elevates renal disease progression, and clearing HCV has proved to reduce liver related mortality/complications as well as risk of HCV transmission, therefore, HCV cure is of great importance to the patients with dual burden [4,5,6].

Sofosbuvir, a nonstructural NS5B polymerase inhibitor, was approved in 2013 and has revolutionized HCV treatment, enhancing the cure bar to above 90% [7,8,9]. Sofosbuvir is mainly eliminated through renal pathway, and its use in patients with stage 4 and 5 CKD, defined according to KDIGO guidelines (GFR < 30 mL/min/1.73 m2), is not indicated in label [10, 11]. EASL Recommendations on Treatment of Hepatitis C 2018 suggested that sofosbuvir should be used with caution in patients with an eGFR< 30 ml/min/1. 73m2 or with end-stage renal disease, only if alternative treatment is not available [12]. AASLD Guidance: Recommendations for testing, managing and treating hepatitis C mentioned the safe and effective dose of sofosbuvir in persons with an eGFR< 30 ml/min have not been established. However, there is accumulating evidence on use of sofosbuvir-based regimen in those with an eGFR< 30 ml/min [13]. Therefore, we performed a systematic review and meta-analysis to evaluate the safety and effectiveness of SOF-containing therapy for this group of patients.

Methods

Literature search strategy

We followed PRISMA (Preferred reporting items for systematic review and meta-analyses) statement guidelines to conduct this study [14]. Systematic literature search in PubMed, EMBASE, Web of science, and Google Scholar was performed by two reviewers independently, without publishing date or language limitation. The searching strategy used was: (sofosbuvir OR Sovaldi OR Harvoni OR Epclusa OR Vosevi) AND (severe kidney impairment OR severe renal impairment OR end-stage renal disease OR dialysis OR renal failure OR ESRD OR renal insufficiency OR hepatorenal syndrome OR HRS). References listed in these literatures were also reviewed. Literature search was lastly updated on August 2018.

Inclusion criteria

Studies were included when following criteria was met:

  1. 1.

    Subject: HCV patients with stage 4 or 5 chronic kidney disease.

  2. 2.

    Intervention: SOF-based regimen.

  3. 3.

    Publication: articles, abstracts or letters.

Exclusion criteria

  1. 1.

    Patients with normal kidney function or early stage (1–3) CKD.

  2. 2.

    Number of enrolled patients or number of patients 12 weeks after treatment completion less than 10.

  3. 3.

    Insufficient data on SVR12/24, which is defined as undetectable HCV RNA 12 weeks (SVR12) or 24 weeks (SVR24) after treatment completion [12].

  4. 4.

    Insufficient HCV treatment combination information.

  5. 5.

    Case report.

Data extraction

Two authors independently extracted data of study design, patients demographics and characteristics, treatment strategy, SVR12/24, renal function, SAE, discontinuation due to AE. Disagreement was resolved by consensus.

Quality assessment

Newcastle-Ottawa scale (NOS) was applied to evaluate the quality and risk of bias of each study by two authors [15]. Studies were judged on three aspects, namely selection of study groups, the comparability of the groups and the exposure or outcome of participants. A score system was used for quality assessment in which a cumulative 7–9 score indicates high quality, 4–6 as fair quality.

Data analysis

SVR12/24 rate, SAE rate were combined and assessed by fixed effect model/random effect model via R software. Heterogeneity among studies was evaluated by I2 index, with value > 50 implying substantial heterogeneity [16]. Fixed effect model was applied in the absence or minor heterogeneity, and random effect model was adopted for significant heterogeneity [17]. Comparison was made in subgroup analysis between studies that adopted full dose and decreased dose of sofosbuvir, RBV-containing and RBV-free regimen, studies that enrolled dialysis-dependent and dialysis-independent patients, as well as studies of different geographic origin. The relative risk (RR) with 95% CI was used to examine the impact of cirrhosis status to sustained virological response. We conducted sensitivity analysis to examine the robustness of primary findings. Publication bias was assessed by the Egger test for funnel plots asymmetry.

Results

A total of 496 literatures were identified after preliminary search. Four hundred forty-two were excluded for duplication or irrelevancy. After further judgement by inclusion/exclusion criteria, final 21 studies were included for our review and meta-analysis, and IRB approval information were reported in 12 studies. Figure 1 shows the process of literature review and selection.

Fig. 1
figure 1

Flow diagram of literature search and selection

Full size image

Studies and patients’ characteristics

Twenty-one studies included manuscripts (n = 13), abstracts (n = 6) and letters (n = 2). All of them were prospective or retrospective cohort studies, with NOS score ranging 4–5 (two literatures scored 3). The absence of high quality is mainly due to absence of control arm. All of the studies met the inclusion criteria while one study enrolled patients with eGFR< 35 mL/min/1.73 m2 [18].

In total 717 patients were enrolled, including 419 (58.4%) hemodialysis or peritoneal dialysis recipients. Patients’ characteristics were shown in Table 1. Mean/median age ranged from 35 to 62 among studies. Eighteen studies provided genotype information, overall GT 1 was predominant genotype (67%), followed by GT 3 (20%) and GT2 (8%). Cirrhotic patient was eligible for most studies, and 4 studies included patients with decompensated cirrhosis. SOF-based regimen included: SOF + SMV ± RBV, SOF + PR, SOF + RBV, SOF + DCV ± RBV and SOF/LDV ± RBV, with varied administration of sofosbuvir: 400 mg daily (QD), 200 mg QD, 400 mg/48 h, 400 mg three times a week (TIW). Dose of sofosbuvir during treatment was rarely adjusted, except 2 patients increased from 200 mg QD to 400 mg QD after 4–6 weeks, and 3 reduced dosing due to sepsis, digestive discomfort or headache.

Table 1 Characteristics of studies and patients
Full size table

Sustained virological response

Per protocol (PP) analysis set was applied for sustained virological response analysis. The pooled SVR12/24 rate was 97.1% with random effect model (95% CI 93.9–99.3%, I2 = 61%) (Fig. 2). By aggregating dialysis-dependent patients (n = 306) and the others who were not on dialysis (n = 88) based on data available, we found significant difference of SVR12/24 between these two groups (95.1% vs 100%, p = 0.019). (Additional file 1: Figure S1).

Fig. 2
figure 2

Forest plots showing the results of pooled SVR12/24

Full size image

Studies with full sofosbuvir dose (400 mg QD) or decreased dose were compared for subgroup analysis. The difference between full dose (97.1, 95% CI: 92.1–99.9%) and decreased dose (96.2, 95%CI: 88.3–100%) was not significant (p = 0.72) (Additional file 2: Figure S2). Studies that applied single therapy of decreased-dose of sofosbuvir (i.e. 200 mg QD or 400 mg/48 h or 400 mg TIW) were further selected and compared, concluding that 200 mg QD and 400 mg/48 h demonstrated similar treatment effect on total population (100% vs 97.7%, p = 0.30) (Additional file 3: Figure S3) [19,20,21,22,23,24,25]. For patients on dialysis, SVR12/24 was also comparable among varied doses of sofosbuvir (p = 0.25) (Additional file 4: Figure S4) [19,20,21,22, 24,25,26,27,28,29,30]. Ten studies adopted RBV-free regimen (SOF/LDV, SOF + DCV, SOF + SMV) with pooled SVR12/24 99.1% (95% CI: 96.2–100%), higher than that of studies using RBV-containing therapy (94.0%,95 CI: 87.5–98.6%) (p = 0.035). Seven studies provided sufficient data for comparison between patients with or without cirrhosis [21,22,23, 29, 31,32,33]. Non-cirrhotic patients trended to have higher sustained virological response, while there was no significant difference between these two subgroups (RR 0.93, 95% CI 0.85–1.02, I2 = 33%) (Fig. 3).

Fig. 3
figure 3

Forest plots showing the results of meta-analysis comparing SVR12/24 in patients with cirrhosis versus patients without cirrhosis

Full size image

Most of the studies in this analysis originated from Asia and Norther America, and the pooled SVR12/24 was comparable among different regions (p = 0.15) (Additional file 5: Figure S5).

Serious adverse event

Information of SAE rate was provided in 16 studies, of which the pooled incidence was 4.8% (95% CI 2.1–10.3%, I2 = 60.0%) (Fig. 4). SAE occurred in 5 studies, and 4 provided case description (Table 2). Subgroup analysis was conducted for studies with full dose and decreased-dose sofosbuvir, resulting no significant difference (8.8% vs 2.9%, P = 0.13).

Fig. 4
figure 4

Forest plots showing the results of pooled SAE rate

Full size image
Table 2 Serious adverse events
Full size table

Change in renal function

Four studies reported the dynamic of kidney function [18, 22, 34, 35], with detailed data before and after treatment (Table 3). Generally, eGFR was stable during treatment, and 2 cases reported discontinuation from hemodialysis due to eGFR improvement [35].

Table 3 Kidney function before and after treatment
Full size table

Sensitivity analysis

SVR12/24 of 13 manuscripts (PP analysis set) was aggregated, with pooled result of 97.0% (95% CI: 92.4–99.7%, I2 = 66%). 90.3% SVR12/24 (95% CI: 83.4–95.6%, I2 = 84%) was resulted in intention to treat (ITT) analysis set for 21 studies. Not reaching SVR12 time point and lost follow-up were two major reasons for relative lower result in ITT analysis set.

Publication bias

P value of Egger test for funnel plots asymmetry was not significant for ITT analysis set (0.537), and bordered significance level for PP analysis set (0.0498).

Discussion

Patients living with HCV infection and end-stage renal disease (ESRD) are special population for HCV treatment. Although current guidelines recommend the first-line therapies as elbasvir/grazoprevir, glecaprevir/pibrentasvir, paritaprevir/ ritonavir/ombitasvir/dasabuvir (PrOD) [12, 13], unmet medical needs still exist at some cases (e.g. comorbidity of advanced liver disease, non-GT1 genotype) and when other therapies are not available. In these circumstances, sofosbuvir might be applied after weighing risk and benefit. In vivo, sofosbuvir undergoes intra-hepatic metabolism to form the pharmacologically active uridine analog triphosphate (GS-461203), which eventually results in ultimate metabolite GS-331007 via dephosphorylation [36]. Sofosbuvir and GS331007 are mainly eliminated through kidney. Compared with subjects of normal renal function, area under the curve (AUC0-inf) of sofosbuvir and GS-331007 is 171 and 451% higher for patients with eGFR< 30 ml/min (not receiving hemodialysis) [11]. Desnoyer examined plasma concentrations of sofosbuvir (full dose) and GS331007 on hemodialysis patients and concluded they did not accumulate throughout the treatment course and between hemodialysis sessions [32]. Whether sofosbuvir and its metabolite accumulate in ESRD patients who are not on dialysis needs to be answered by further study.

Our meta-analysis included 21 studies, with a total of 717 patients. Pooled SVR12/24 was satisfying (97.1%), similar or higher than that of non-SOF-based therapies [37,38,39]. Patients who were on dialysis also achieved a SVR12/24 as high as 95.1%. Although it was lower than that of patients without dialysis, we assume it might not necessarily be the case given limited number of dialysis-free patients in our sub-analysis. Further well-designed RCT are needed to conclude whether effect of sofosbuvir is influenced by dialysis. Aggregated SAE rate was 4.8%, slightly higher than that in HCV infected patients with normal renal function [8, 9], which is reasonable since the patients involved in our meta-analysis had quite a few safety risk factors: old age, severe renal dysfunction, liver/renal transplant recipient, and advanced liver fibrosis. In another meta-analysis, SAE rate was 12.1% for direct-acting antivirals-based antiviral therapies in HCV/Stage 4–5 CKD patients [40]. Common SAEs included renal failure, cirrhosis complications, indicating special attention is needed on renal and liver function during treatment (Table 2).

Treatment strategy of SOF-containing therapy has been under broad discussion. Many physicians explored unconventional dose of sofosbuvir for safety concern, although there is no established pharmacokinetics profile for administration at 200 mg QD, 400 mg/48 h or 400 mg three times a week. Our subgroup analysis suggests that regimen with full and decreased dose of sofosbuvir might be both considerable at acceptable SAE rate (8.8% vs 2.9%) and high SVR12/24 (97.1% vs 96.2%). We assume that sofosbuvir could be alternatively applied at a lower dose as half as normal or at a frequency extended to once every 2 days without compromising its efficacy significantly. While administering three times a week might not be optimal, given that median terminal half-lives of sofosbuvir and its metabolite GS331007 were 0.4 and 27 h (healthy subjects), and possible low SVR12/24 (60% reported in very small sample size) [32]. Ribavirin should be used with caution, considering higher risk of anemia for RBV-containing regimen than RBV-free treatment. Anemia was the most frequently reported AE (6 to 43.7%) in studies of our review, and RBV was included in almost all of these studies (8/9). Manoj reported that 65.4% (17/26) patients in SOF + RBV group developed anemia, and 30% had to discontinue ribavirin [28]. What is worth mentioning is that we found RBV-containing regimen reached lower SVR12/24. One of the reasons is likely to be tolerability issue of ribavirin, another reason might be potent combination drug with sofosbuvir in most RBV-free regimen, 9 studies applied SOF/LDV or SOF + DCV, and 7 achieved 100% SVR12/24.

Kidney function deterioration is a concern for sofosbuvir usage. On one hand, there are few case reports proposed the correlation of nephrotoxicity and SOF-based therapy [41, 42], on the other hand, several large retrospective cohorts conclude SOF-based regimen does not introduce higher acute kidney injury for HCV patients compared to SOF-free treatment [43, 44]. Studies that reported renal function in our review had generally stable eGFR and serum creatinine during treatment. Two SAE cases of AKI and acute renal failure were considered not related to study drug.

There are some limitations to our study. First, we could not perform subgroup analysis according to HCV genotype, CKD stage for lack of enough information. Second, geographical origin of studies enrolled is mainly USA and India, and that limits the representativeness of our analysis. Third, all of the studies were observational studies without control group, and most studies were of medium quality. Heterogeneity was substantial which might attribute to varied sample size and treatment therapy. Furthermore, some newly approved SOF-based therapy (e.g. SOF/VEL) was not included in this analysis for lack of evidence and analysis time frame. The factors above compromise the quality of this review. Prospective and well-controlled studies are expected in near future to provide more robust evidence.

Conclusions

This systematic review and meta-analysis evaluated SOF-based therapy for HCV infected patients with comorbidity of stage 4–5 CKD. Data from this study suggests satisfying sustained virological response and tolerance. For treatment strategy, both full and decreased dose of sofosbuvir could be appropriate. Caution is still needed at clinical practice.