The efficacy and safety of direct-acting antiviral regimens for end-stage renal disease patients with HCV infection: a systematic review and network meta-analysis

Background Hepatitis C virus (HCV) infection is an independent risk factor associated with adverse outcomes in patients with end-stage renal disease (ESRD). Due to the wide variety of direct-acting antiviral regimens (DAAs) and the factor of renal insufficiency, careless selection of anti-hepatitis C treatment can lead to treatment failure and safety problems. The integrated evidence for optimized therapies for these patients is lacking. This study would conduct comparisons of different DAAs and facilitate clinical decision-making. Methods We conducted a systematic literature search in multiple databases (PubMed, Ovid, Embase, Cochrane Library, and Web of Science) up to 7 August 2023. Study data that contained patient characteristics, study design, treatment regimens, intention-to-treat sustained virologic response (SVR), and adverse event (AE) data per regimen were extracted into a structured electronic database and analyzed. The network meta-analysis of the estimation was performed by the Bayesian Markov Chain Monte Carlo methods. Results Our search identified 5,278 articles; removing the studies with duplicates and ineligible criteria, a total of 62 studies (comprising 4,554 patients) were included. Overall, the analyses contained more than 2,489 male individuals, at least 202 patients with cirrhosis, and no less than 2,377 patients under hemodialysis. Network meta-analyses of the DAAs found that receiving ombitasvir (OBV)/paritaprevir (PTV)/ritonavir (R) plus dasabuvir (DSV), glecaprevir (G)/pibrentasvir (P), and sofosbuvir (SOF)/ledipasvir (LDV) ranked as the top three efficacy factors for the HCV-infected ESRD patients. Stratified by genotype, the G/P would prioritize genotype 1 and 2 patients with 98.9%−100% SVR, the SOF/DCV regimen had the greatest SVR rates (98.7%; 95% CI, 93.0%−100.0%) in genotype 3, and the OBV/PTV/R regimen was the best choice for genotype 4, with the highest SVR of 98.1% (95% CI, 94.4%−99.9%). In the pan-genotypic DAAs comparison, the G/P regimen showed the best pooled SVR of 99.4% (95% CI, 98.6%−100%). DAA regimens without Ribavirin or SOF showed the lowest rates of AEs (49.9%; 95% CI, 38.4%−61.5%) in HCV-infected ESRD patients. Conclusion The G/P could be recommended as the best option for the treatment of pan-genotypic HCV-infected ESRD patients. The OBV/PTV/R plus DSV, SOF/Velpatasvir (VEL), SOF/Ledipasvir (LDV), and SOF/DCV would be reliable alternatives for HCV treatment with comparable efficacy and safety profiles. Systematic review registration https://www.crd.york.ac.uk/prospero/#searchadvanced, PROSPERO: CRD42021242359.


Introduction
Chronic kidney disease (CKD) patients, especially those under maintenance hemodialysis, are at increased risk of hepatitis C virus (HCV) infection (1). The prevalence of CKD in HCV patients is significantly higher than that in the general population (2)(3)(4). HCV infection is an independent risk factor for accelerated CKD progression and is associated with adverse outcomes in patients with end-stage renal disease (ESRD), including hepaticrelated hospitalizations, mortality, and poor health-related life quality (2,3).
The advent of direct-acting antiviral regimens (DAAs) has transformed the treatment of HCV in patients with CKD. There is poor evidence comparing and assessing the efficacy and safety of DAAs in ESRD. The guidelines recommend that HCVinfected CKD patients should be assessed for DAA therapy, with the specific regimen determined by HCV genotype, viral load, treatment history, estimated glomerular filtration rate (eGFR), hepatic fibrosis stage, kidney and liver transplant candidacy, and after consideration of drug-drug interactions (4,5). Although the sustained virologic response (SVR) rate could even reach 90%−100% with few adverse events (AEs) (4), the choice of DAAs for patients with ESRD has not been elucidated. Renal clearance is the major elimination pathway of Sofosbuvir (SOF), so SOF-based regimens have been questioned for use in ESRD patients. Nevertheless, based on several studies on the safety and efficacy of SOF-based regimens in patients with severe CKD, the Drug Administration of most countries has removed the restricted label for renal impairment. A variety of DAAs are permitted for the treatment of HCV infection in patients with impaired renal function. The comparisons of different DAA regimens applied in HCV-infected ESRD patients are not completely understood, which is necessary to steer guideline development and clinical practice. Moreover, pan-genotypic DAAs simplified the treatment algorithm and supported the campaign to eliminate HCV infection all over the world. Whether these drugs still have excellent performance in ESRD remains to be further confirmed.
To guide best practices for DAAs in patients with CKD and chronic hepatitis C, we performed this systematic review and network meta-analysis to quantify the efficacy and safety of different DAAs for the treatment of HCV-infected ESRD patients. To understand which specific interventions work best, their effects should be explored separately and compared against those of other regimens using the two alternative Bayesian models that can accommodate disconnected networks. The study will facilitate informed clinical decision-making and drafting of HCV treatment guidelines.

Literature search
Databases including PubMed, Ovid (BIOSIS Previews Embase), Cochrane Library, and Web of Science were systematically searched under the direction of a medical librarian. The final search was completed on 7 August 2023. The bibliographies of relevant studies and reviews were scrutinized for any additional eligible studies not covered by the literature search. The literature search combined the terms and descriptors related to DAA, HCV, and ESRD concerning literature published in English (details of the searching strategy are available in Supplementary File 1). Conference abstracts and comments were not considered.

Outcome measures
The primary outcome of the study was the mean estimated probability of SVR in various studied DAA regimens for HCVinfected ESRD patients. The SVR was defined as a sustained virologic response at 12 weeks after the end of therapy (SVR12) for patients in the treatment group. The relative rank of efficacy would be calculated by network meta-analysis. For secondary outcomes, the AEs reported in the studies, particularly the serious adverse events (SAEs), discontinuation of treatment, or death, were extracted. AEs evaluation included physical examinations, clinical laboratory tests, and symptoms. SAEs were defined as any event leading to hospital admission or resulting in death, or any event considered serious in the opinion of the treating physician.

Data extraction
Study characteristics (first author, publication year, location, study design, study period), SVR, and AE data per regimen were extracted into a structured electronic database, while two researchers (Ruochan Chen and Yinghui Xiong) completed a crosscheck procedure. The Methodological Index for Non-Randomized Studies (MINORS) and the Newcastle-Ottawa Score (NOS) were used to assess the quality of trials and cohort studies, respectively (literature evaluations are available in Supplementary File 2) (7, 8). Disagreements were resolved by consensus and arbitration by a panel of other investigators within the review team (Yanyang Zeng and Yixiang Zheng).

Statistical analysis
The network meta-analysis in this study would be regarded as a "disconnected network, " while many single-arm studies of DAA were included. On account of the promising efficacy and safety results, the FDA updated their 2017 guidance to industry on the design and analysis of clinical trials of DAAs to recommend the use of single-arm/historical controls as well as a placebo-deferred trial design (9).
The disconnected network analysis was conducted according to the National Institute for Health and Clinical Excellence Guideline (10). With neither direct comparisons nor a common comparator through which to derive indirect comparisons of comparator treatments, the evidence base will be structured as a disconnected network. The Bayesian Markov Chain Monte Carlo (MCMC) method was used to estimate the pooled SVR of each DAA regimen. The random-effects model with binomial likelihood was implemented to predict the distribution of baseline and treatment effects in the analysis. Bayesian models could accommodate disconnected networks; assuming that the variance of the baseline response is fixed at zero, we applied the absolute response as a means to evaluate the efficacy and safety of the regimens. All the studied DAA regimens were respectively combined to estimate the probability of SVR with a 95% equal tail credible interval (95% CI). Relative ranks of the efficacy of different DAA regimens were established in the analysis. Primary calculations were done according to modified intention to treat (mITT) analysis, where only patients who received at least one dose of DAAs and had an assessment of HCV RNA at 12 weeks after completion of treatment were included for SVR analysis. Additional sensitivity analysis was done using intention to treat (ITT). For ITT, all patients who received at least one dose of DAA regimens were analyzed. Subgroup analyses were pre-specified to separate the distinct kinds of HCV genotype, CKD stage, cirrhosis, and hemodialysis. We checked whether the MCMC procedure had reached convergence by visually inspecting the history trace plots and the autocorrelation plots for irregularities. Since the included articles were almost single-arm studies, a single-rate meta-analysis with a randomeffects model was used for safety evaluation and subgroup analyses. All the statistical analyses were performed using WinBUGs (version 1.4.3) and R version 4.1.0.

DAA therapy subgroups analyses by grade of ESRD
Overall, 35 studies were identified to evaluate the efficacy of DAAs in HCV-infected patients with CKD5 or hemodialysis. Results show the overall SVR to be 97.5% (95% CI, 96.7%−98.4%). We also analyzed the efficacy of DAA in CHC patients with CKD4, with the overall SVR being 99.4% (95% CI, 97.4%−100.0%). Based on the most heavily weighted studies and patients, GZR/EBR and G/P had the highest SVR rates, almost 100%. Further comparison based on 10 studies involving GZR/EBR, G/P, OBV/PTV/R plus DSV or SOF-based regimens showed that the OR of achieving SVR in CKD4 vs. CKD5 was 0.75 (95% CI, 0.31-1.84) without significant difference (Supplementary Figure 5).

Subgroup analyses by cirrhosis
In this analysis, the included regimens of G/P, SOF/DCV, and OBV/PTV/R ± DSV ± RBV were included in the regimens for the majority of cirrhotic patients, with a high SVR of almost 100%. The OR of achieving SVR in cirrhotic compared to non-cirrhotic patients was 0.31 (95% CI, 0.14-0.69). No heterogeneity (I 2 = 0) was found among these studies (Supplementary Figure 6).

Discussion
This systematic review and network meta-analysis aimed to establish a hierarchy of available treatment regimens for HCV infection among patients with ESRD. To the best of our knowledge, Frontiers in Public Health frontiersin.org . /fpubh. . this is the most comprehensive overview of the available efficacy and safety data for oral DAA regimens, and the main findings can be summarized below.
The key finding was that the OBV/PTV/R ± DSV regimen achieved the highest efficacy in HCV-infected ESRD patients, and similar estimated SVR rates could be achieved using the GP regimen. In addition, SOF/LDV, GZR/EBR, and SOF/VEL had only 1%−2% lower estimated SVR rates and remained alternative options for treatment.
We established the unprofitable value of Ribavirin, regardless of the difference in DAAs. The RBV did not improve the SVR of OBV/PTV/R ± DSV regimens in HCV-infected ESRD patients. The SOF-RBV and SOF-ad-RBV have the lowest SVR rates, poor performance, and should be considered obsolete.
Identifying certain genotypes before initiating therapy remains useful and may be required when drugs permit limitations or optimize treatment regimens. This study also gave priority to the selection of DAAs based on different genotypes. For HCV genotype 1a patients, in addition to OBV/PTV/R ± DSV and G/P as the optimal selection, two other combinations with SVR rates over 95% (SOF/LDV and GZR/EBR) would be recommended. Genotype 1b patients achieved excellent efficacy both in OBV/PTV/R ± DSV, GZR/EBR, and DCV/ASV, with SVR rates of approximately 99%. The G/P regimen would be the optimal solution for HCV genotype 2 ESRD patients based on the most evidence. As a relatively easyto-treat type, HCV genotype 4 ESRD patients could achieve a higher SVR rate through OBV/PTV/R, and there were reasonable reasons to believe in the efficacy of other regimens. However, HCV genotype 3 is considered the most hard-to-treat type due to the increased incidence of cirrhosis that may reduce the SVR rate. In this meta-analysis, the SVR rate of SOF/DCV was close to 98%, and even a dose reduction of SOF combined with DCV also achieved an SVR rate of more than 95%. Thus, SOF/DCV would be a priority for genotype 3 ESRD patients, which was consistent with a previous network meta-analysis of optimal DAAs for HCV genotype 3 infection (73). As a pan-genotypic HCV drug regimen, the G/P regimen can be used to treat individuals without identifying their HCV genotype and subtype (74). In ESRD patients, G/P also showed a therapeutic superiority in all genotype subgroup analyses. The 2020 EASL (74) and 2019 AASLD (75) treatment guidelines now suggest two main regimens for G/P and SOF/VEL with pangenotypic antiviral activity to simplify the treatment algorithm. The tolerability and effectiveness of pan-genotypic DAAs in ESRD are still unclear. By comparing those DAAs, we found the SVR of G/P was close to perfect and slightly better than SOL/VEL. Subgroup analyses of cirrhosis suggested that ESRD patients with cirrhosis were 69% less likely to achieve SVR than those without cirrhosis. However, the influence of cirrhosis on efficacy was limited to ESRD patients using G/P and SOF/DCV regimens. Regardless of glomerular filtration rate, GZR/EBR or G/P used in ESRD patients both showed significant and comparable efficacy between CKD4 and CKD5 patients. When HCV-infected ESRD patients undergo hemodialysis, the OBV/PTV/R plus DSV and GP regimens are preferentially recommended due to the excellent efficacy of SVR, which exceeds 95%, which is much higher than other regimens. In addition, SOF/LDV, SOF/VEL, and GZR/EBR would also be good substitutes for hemodialytic patients as their SVR rates were around 94%.
A safety evaluation would identify incidences of AEs, discontinuation of treatment, and death in ESRD patients with different DAA regimens. AEs with the DAA regimens were common in ESRD patients, with an incidence of up to 59.9%. In contrast, the pooled occurrence rates of SAEs, discontinuation of treatment, and death were relatively much lower, at 8.4%, .
/fpubh. . 0.8%, and 0.4%, respectively, among ESRD patients receiving DAA regimens. DAA regimens without SOF or RBV had a lower risk of common AEs. Anemia was the most common of the reported AEs, with a pooled prevalence of 44.4%. Since anemia is a common complication of CKD, we further explored the side effects of RBV on anemia in ESRD patients. The results suggested that more than one-third of ESRD patients treated with DAAs would experience anemia, and the RBV-containing regimens increased the incidence of anemia by 30% compared with the RBV-free regimens. Moreover, anemia exacerbation was more common in patients on RBV-containing regimens. The probabilities of SAEs and deaths were much higher with OBV/PTV/R plus DSV + RBV and SOF + RBV regimens, which should be taken seriously. In addition, the SOF + RBV regimens had the highest discontinuation rate. Therefore, the use of RBV in the antiviral protocol for HCV-infected ESRD patients should be avoided. The SOF-based regimens like SOF/LDV and SOF/DCV both showed satisfactory safety profiles, which further confirmed their applicability in HCV-infected ESRD patients. The GP also had a solid safety profile with a 10% reduction in SAE rate and a 3% reduction in discontinuation rate, which had an extremely low risk of death as low as 0.1% in ESRD patients.
However, several limitations should be expounded and warrant further discussion. First, differences in the details of the study design resulted in significant heterogeneity among the included studies, which may compromise comparability. In response to this, we controlled for and explored sources of heterogeneity by choosing a random-effects model rather than a fixed-effects model for the analyses and completing subgroup analyses by genotype, cirrhosis, and ESRD class. Second, most publications reported SVR as the major outcome, with a brief accompanying safety assessment. Due to the small number of studies, it was not possible to compare SOF-based regimens with SOF-free regimens. Future studies will require detailed comparisons of safety among different DAAs. Third, the efficacy class of DAA regimens for specific HCV genotypes could not be determined, and the estimated pooled data for some DAA regimens related to specific genotypes could not be extracted separately due to the small number of patients. Fourth, we were unable to formally assess publication bias because the studies per regimen ranged from 1 to 11. Fifth, the efficacy in patients with ESRD and decompensated cirrhosis could not be explored due to the paucity of data. Sixth, the limitation of our network meta-analysis was the risk of conceptual heterogeneity, reflecting differences between trials that may impair comparability. We used several strategies to target heterogeneity: (1) we used a random-effects model (by including a study effect in our model); (2) we split the analyses for patients according to genotype, cirrhosis, and CKD grade; and (3) we performed analyses between mITT study and ITT study to increase homogeneity, which showed similar results. Moreover, SVR is an objective outcome that decreases the risk of heterogeneity. Nonetheless, we do not expect publication bias as the HCV field is rapidly evolving.

Conclusion
The G/P would be recommended as the best option for the treatment of pan-genotypic HCV-infected ESRD patients due to its highest efficacy and safety; the SOF/VEL would be a suboptimal option. SOF/DCV had an advantage in the treatment of genotype 3 HCV patients. SOF-based DAA regimens had satisfactory safety profiles in HCV-infected ESRD patients; meanwhile, RBV should be counted out from HCV antiviral regimens in ESRD patients.

Data availability statement
The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

Author contributions
YYZ and YHX carried out the literature screening. YHX and RCC made the assessment and data extraction. YXZ and RCC wrote the article and prepared the figures and tables. YXZ designed the research and made the data analysis. YZX and XLW reviewed all the data, code, analysis, and results. All authors contributed to the article and approved the submitted version.