Impact of direct acting antiviral therapy in patients with chronic hepatitis C and decompensated cirrhosis; Direct acting antivirals for decompensated cirrhosis. Efficacy and safety are now established - Editorial
Download the PDF here
Download the PDF here
original article below: "The efficacy of all oral antiviral regimens in the management of patients with compensated liver disease due to chronic HCV infection is now established [, , , ] and data on patients with decompensated cirrhosis are emerging [, , , ]. In this study we examined a fixed, 12 week, duration course of antiviral therapy in a large heterogeneous group of patients with decompensated cirrhosis or life-threatening complications of HCV infection, and we compared outcomes to an untreated cohort with equivalent disease and duration of follow-up.....In summary, this is a large, real-life study of all oral HCV therapy in patients with advanced HCV disease with a significant risk of death or irreversible damage within 12 months. Overall virological response was high and we found early improvement in liver parameters and in clinical outcomes after antiviral treatment, compared to untreated patients. The longer term benefits of therapy in patients with decompensated disease remain to be ascertained."
Editorial - Direct acting antivirals for decompensated cirrhosis. Efficacy and safety are now established
June 2016 Jnl of Hepatology
Michael P. Curry
Beth Israel Deaconess Medical Centre, Boston, MA, USA
Chronic liver injury of any etiology most commonly results in liver fibrosis and eventually cirrhosis and portal hypertension with the attendant risks of decompensated liver failure, hepatocellular carcinoma (HCC) and death. Several studies have shown that successful treatment or removal of the underlying liver injury can result in regression of liver fibrosis and cirrhosis in some patients . Reversal of liver cirrhosis was observed in 49% of patients with compensated hepatitis C virus (HCV) cirrhosis treated with interferon (IFN) based therapy. Sustained virological response (SVR) was the only independent predictor of reversal . Eradication of HCV infection with IFN based therapy also reduces the risk of liver failure, HCC, liver-related mortality, all cause mortality and the need for liver transplantation (LT) in those patients with compensated liver cirrhosis [, ].
The improved SVR rates and safety profiles of all oral direct acting antivirals (DAA) has led to the treatment of some patients who would not have received widespread treatment in the IFN era. One such group is the decompensated cirrhotic patients, who have a poor prognosis, have limited treatment options and make up a large proportion of those awaiting LT . Several open label clinical trials of DAA in decompensated HCV patients have recently demonstrated SVR rates above 80% and improvements have been observed in the Child-Pugh-Turcotte (CPT) and/or model for end-stage liver disease (MELD) scores in a significant proportion of patients after relatively short follow up [, , , ].
These improvements are largely attributable to changes in biochemical parameters of serum bilirubin and albumin. The results for SVR observed in these studies are indeed impressive given that the majority of these patients had several characteristics that would predict poor response. However, there are some limitations to the studies that make it hard to draw definitive conclusions about benefits other than viral eradication. The absence of comparator control groups makes it difficult to determine if viral eradication and improvement in MELD and CPT scores result in improved survival. The effects of viral eradication on the clinical aspects of hepatic decompensation are not well represented in these studies; and the high SVR rates achieved in these open label studies would likely now hamper enrolment in any future placebo control trials designed to answer these questions. We are therefore left wondering whether viral eradication results in a reduction in morbidity and mortality, improves transplant free survival and if DAA are indeed safe in this patient population. Given that LT is the only available treatment for decompensated cirrhosis and that there is a growing disparity between donor organ availability and the number of patients wait-listed for LT, we need to determine which population of patients should receive DAA and if treatment can improve liver function so that these patients have improved transplant free survival.
In a study published in this issue, Foster and colleagues have addressed some of the shortfalls of these open label studies using the early access program (EAP) for treatment in the HCV Research UK registry. Patients were treated with combination DAA for 12 weeks at the discretion of the investigator. The study cohort was prospectively enrolled from registry patients while the control cohort was retrospectively enrolled from the same registry prior to the availability of DAA EAP . Primary outcome was SVR at 12 weeks and secondary outcomes were changes in MELD score, and development of adverse outcomes such as decompensation, HCC, sepsis, LT, death and hospitalization observed over the 12 week treatment period and 12 weeks of follow-up.
Similar to what was seen in the open label studies, SVR was achieved in over 80% of patients with severe HCV related liver disease and was more frequently observed in genotype 1 patients and in genotype 3 patients who received the pan-genotypic NS5A inhibitor, daclatasvir as part of their treatment regimen. Treated patients had an improvement in MELD score, were less likely to increase MELD score ⩾2 points, were less likely to have an increase MELD score ⩾2 points associated with an adverse outcome and were less likely to have a new decompensating event when compared with untreated cohort regardless of SVR outcome. SVR did not predict improvement in MELD score, but patients achieving SVR were less likely to increase MELD score ⩾2 points and were less likely to have an adverse outcome compared with those who did not achieve SVR. The authors identified baseline MELD score and albumin as predictors of adverse outcomes.
Additionally, the authors attempted to use baseline characteristics to create a model that would predict benefit. Patients older than 65 years with reduced hepatic synthetic function (serum albumin ≤35 g/L) were less likely to derive a benefit from therapy, however this model does not discriminate well enough (AUROC of 0.5484) to determine if patients should be treated or if treatment should be deferred in favor of LT.
The study showed no difference in incidence of HCC, sepsis, or death between treated and untreated patients over the 24 week study period. HCC occurred in 6.1% of the treated and 8.0% of the untreated cohort. SVR had no effect on the incidence of HCC, although this is not all that surprising given the 24 week time-frame. The relatively high rate of HCC over this short time period would suggest that some or all of these HCCs were present but not detected on enrolment into the study. It is also unlikely that eradication of HCV would alter the development of HCC within 12 weeks of achieving SVR given that half of the HCC that developed in a large Japanese cohort of patients successfully treated with IFN occurred ⩾37 months after SVR . Studies that have shown a difference in the incidence of HCC between SVR and non-SVR patients have had significantly longer follow up (>6 years) .
Importantly from the perspective of safety in treating decompensated patients with DAA, the authors compared the incidence of adverse outcomes between the treated and untreated cohorts and noted that the treatment was generally well tolerated. The study showed no difference in the incidence of hospitalization, sepsis and death between treated and untreated cohorts. This would suggest that these complications are related to the underlying liver disease and not related to the treatment regimen. The baseline median MELD score was 11 in both treated and untreated cohorts but some patients had MELD scores of up to 32. Almost 10% of the treated cohort were classified as CPT class C and over 40% had ascites indicating a significant degree of advanced liver disease that is unlikely to improve over the short term after eradication of HCV. It is likely that some of these patients with decompensated cirrhosis had reached a point where antiviral therapy was less effective in improving liver function. In the Ally-1 study of sofosbuvir/daclatasvir and ribavirin, Poordad and colleagues observed that CPT class C patients had lower SVR (56%) than CPT class A (92%) and B (94%) patients . In the Solar 2 study of sofosbuvir/ledipasvir and ribavirin, the most common reason for non-SVR in the CPT class C patients was death from progressive liver failure . This "point of no return" has also been illustrated in the past with the treatment of decompensated hepatitis B virus (HBV) cirrhosis. Nucleoside analogue treatment of decompensated HBV cirrhosis is associated with a reduction in all cause mortality when compared with control (RR = 0/5, 85% CI 0.3-0.8) . In a study of 70 patients with decompensated HBV cirrhosis treated with entecavir, the cumulative incidence of mortality or LT was 17% at 2 years for the whole group but 36% for CPT class C patients, despite viral suppression in 92% of patients and improvements in CPT and MELD scores .
We have now seen multiple reports that DAA are highly efficacious in a population of patients in most need of viral eradication. Viral eradication results in improvements in MELD and CPT scores denoting improvement in liver function in studies of short term follow up. It appears that these regimens are relatively safe, with most adverse events caused by the underlying disease process. However, there is a proportion of patients that do not seem to gain immediate benefit from DAA therapy. Whether these patients are slower to improve or will remain with decompensated liver disease and at risk of further complication requires longer follow up studies. Additionally, it will take time and further study to see if these patients have been disadvantaged in the prioritization system for organ allocation. Lastly we need the results of ongoing studies of portal pressure measurements (NCT01687257) and long-term registries (NCT02292706) to identify those patients less likely to benefit from DAA therapy.
Impact of direct acting antiviral therapy in patients with chronic hepatitis C and decompensated cirrhosis
Jnl of Hepatology June 2016 Graham R. Foster1,⇑,y, William L. Irving2,⇑,y, Michelle C.M. Cheung3, Alex J. Walker4, Benjamin E. Hudson5, Suman Verma6, John McLauchlan7, David J. Mutimer8, Ashley Brown9, William T.H. Gelson10, Douglas C. MacDonald11, Kosh Agarwal6, on behalf of HCV Research, UK
1Queen Mary University of London, London, United Kingdom; 2NIHR Nottingham Digestive Diseases Biomedical Research Unit, United Kingdom; 3Liver Unit, Blizard Institute, Queen Mary University of London, United Kingdom; 4Faculty of Medicine & Health Sciences, University of Nottingham, United Kingdom; 5University Hospitals Bristol NHS Trust, United Kingdom; 6Institute of Liver Studies, King's College London, United Kingdom; 7MRC-University of Glasgow Centre for Virus Research, United Kingdom; 8Centre for Liver Research and NIHR Biomedical Research Unit, Queen Elizabeth Hospital, Birmingham, United Kingdom; 9Department of Hepatology, St Mary's Hospital, Imperial College London, United Kingdom; 10Department of Hepatology, Cambridge University Hospitals NHS Foundation Trust, United Kingdom; 11UCL Institute for Liver and Digestive Health, University College London, United Kingdom
Background & Aims
All oral direct acting antivirals (DAAs) effectively treat chronic hepatitis C virus (HCV) infection, but the benefits in advanced liver disease are unclear. We compared outcomes in treated and untreated patients with decompensated cirrhosis.
Patients with HCV and decompensated cirrhosis or at risk of irreversible disease were treated in an expanded access programme (EAP) in 2014. Treatment, by clinician choice, was with sofosbuvir, ledipasvir or daclatasvir, with or without ribavirin. For functional outcome comparison, untreated patients with HCV and decompensated cirrhosis who were registered on a database 6 months before treatment was available were retrospectively studied. Primary endpoint was sustained virological response 12 weeks post antiviral treatment (treated cohort) and the secondary endpoint (both cohorts) was adverse outcomes (worsening in MELD score or serious adverse event) within 6 months.
467 patients received treatment (409 decompensated cirrhosis). Viral clearance was achieved in 381 patients (81.6%) - 209 from 231 (90.5%) with genotype 1 and 132 from 192 (68.8%) with genotype 3. MELD scores improved in treated patients (mean change -0.85) but worsened in untreated patients (mean + 0.75) (p <0.0001). Patients with initial serum albumin <35 g/L, aged >65 or with low (<135 mmol/L) baseline serum sodium concentrations were least likely to benefit from therapy.
All oral DAAs effectively cured HCV in patients with advanced liver disease. Viral clearance was associated with improvement in liver function within 6 months compared to untreated patients. The longer term impact of HCV treatment in patients with decompensated cirrhosis remains to be determined.
All oral direct acting antiviral (DAA) drugs for patients with chronic hepatitis C virus (HCV) infection have transformed treatment options. High response rates are observed in patients with compensated cirrhosis in clinical trials [, , , ] and in real-life observational studies . Even in patients with decompensated cirrhosis, for whom pegylated interferon (PegIFN) is contraindicated and who otherwise have a poor prognosis, HCV can now be cured safely and effectively . Experience of different DAA combinations in the real world is growing within a number of compassionate use programs [, ]. However, virological response with all oral DAAs in decompensated cirrhosis, particularly in HCV genotype 3, is lower than in patients with less advanced liver disease [, ]. Remaining uncertainties in treating such patients include the optimal duration of therapy, whether or not the regimen should include ribavirin, and what, if any, clinical benefits accrue to patients with such advanced liver disease following clearance of virus.
Viral eradication in patients with chronic HCV infection has been shown to result in improvements in medium and long term outcomes - assessed both by patient reported quality of life measurements and mortality/morbidity [, , ]. However, in patients with advanced, decompensated cirrhosis it is uncertain whether treatment and viral clearance is beneficial, and if meaningful functional hepatic recovery is possible. Clearly if treatment were able to reverse liver dysfunction and perhaps avoid the need for transplantation then therapy should be recommended. Therapy, although generally safe, might be associated with adverse events, particularly in unstable patients with advanced cirrhosis  and may delay access to transplantation. It is conceivable that patients with decompensated cirrhosis may eliminate virus, stabilise their disease and not progress to a stage where transplantation is indicated. However, they may not recover to any meaningful extent and post therapy, may be left without access to transplantation but with a poor quality of life (so called 'MELD purgatory'). In the SOLAR study of patients with advanced liver disease infected with HCV genotypes 1 and 4, viral eradication was rapidly associated with an improvement in severity of liver disease scores (MELD scores)  but no comparator group was included, and it is unclear whether benefits were related to viral clearance or to improved management of the patients in a clinical trial setting in specialist centres [, , ].
Here we report on the outcomes of the NHS England expanded access programme (EAP), which treated patients with severe liver disease of all viral genotypes, who were "at significant risk of death or irreversible damage within 12 months due to hepatic or extrahepatic manifestations". Sofosbuvir combined with the NS5A inhibitors ledipasvir or daclatasvir, with or without ribavirin, was used for a fixed duration of 12 weeks and patients were enrolled in the UK hepatitis C registry - HCV Research UK. To address the question of whether antiviral therapy is beneficial in unselected patients with decompensated cirrhosis infected with all HCV genotypes, we examined the functional outcomes of patients with equivalent disease stage, who enrolled into the same registry for at least 6 months prior to the start of EAP, and hence were not able to receive HCV treatment for at least 6 months of follow-up. We show that HCV treatment improves outcomes and we present a model to predict those patients who are likely to derive most benefit from therapy.
Patients and methods
Study design and patients
For treated patients this was a prospective, observational cohort study. Patients enrolled in the HCV Research UK registry who received antiviral therapy as part of the EAP between 1 April 2014 and 11 November 2014 were studied. Eligible patients were those at significant risk of death or irreversible damage from HCV infection within 12 months, irrespective of genotype. Criteria for inclusion were decompensated cirrhosis - ascites, variceal bleed or encephalopathy (past or current), Child Pugh score ⩾7, or non-hepatic manifestation of HCV likely to lead to irreversible damage in 12 months and intolerant to, or failed, PegIFN and ribavirin therapy, or exceptional circumstances (determined by a review panel). Treatment was chosen by the prescribing clinician and involved either ledipasvir/sofosbuvir or sofosbuvir/daclatasvir, both with or without ribavirin for a total of 12 weeks. Prescribing was restricted to 20 English centres selected by NHS England in a competitive tender process, and comprised experienced HCV treatment centres. A criterion for participation in the bid was participation in HCV Research UK ensuring that data could be collected from all treatment sites. All patients receiving therapy were asked to consent to contribute anonymised data to the national registry, and those who agreed were included.
For the comparator population (untreated) we used a retrospective, observational study design. Patients were selected from the HCV Research UK database. Inclusion criteria were decompensated cirrhosis (defined by the criteria above) who were enrolled before 1 October 2013 (i.e. enrolled in the database 6 months before the EAP was initiated), and patients subsequently included in the EAP who had been enrolled in the database 6 months prior to the initiation of therapy.
The study conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the institution's human research committee. Ethics approval for HCV Research UK was given by NRES Committee East Midlands - Derby 1 (Research Ethics Committee reference 11/EM/0314) and informed consent was obtained from each patient included in the study. Patients who declined to enrol in HCV Research UK were given antiviral therapy but data was not collected.
All patients received a maximum of 12 weeks therapy. Sofosbuvir (400 mg per day) was purchased from Gilead and clinicians chose to combine it with either ledipasvir (90 mg per day), provided by Gilead as a single tablet co-formulation with sofosbuvir, or with daclatasvir (60 mg per day, with dose adjustment to 30 mg or 90 mg per day as recommended in patients with relevant potential drug-drug interactions) which was provided by Bristol-Myers Squibb. Inclusion and dosage of ribavirin was discretionary according to the treating clinician.
Patients receiving therapy were reviewed at treatment weeks 2, 4, 8 and 12, and post treatment weeks 4 and 12. Clinical events were recorded on a standardised form and local accredited laboratories measured serum creatinine, bilirubin, albumin, alanine aminotransferase (ALT), sodium, HCV RNA level, full blood count and clotting profile, which were recorded for each study visit. Missing or late values during therapy were ignored, and missing values from the initiation visit or week 12 post treatment visit were derived from the nearest adjacent test value taken before or after therapy respectively. MELD scores were calculated centrally using site-derived laboratory parameters.
Serious adverse events, specifically any hospital admissions, decompensation events, liver transplantation, other complications of end-stage liver disease (including hepatocellular carcinoma) and death were recorded.
Baseline demographic data (age, gender, ethnicity, body mass index (BMI), alcohol use, previous treatment history, decompensation events and HIV serostatus) and HCV genotype were taken from the original registration form recorded on the HCV Research UK database.
The primary outcome was sustained virological response 12 weeks post treatment (SVR12) defined as undetectable HCV RNA measured at an accredited local laboratory with a lower limit of quantification (LLQ) of <30 IU/ml. Secondary outcomes included change in MELD score, and adverse clinical events during the study period. Overall adverse clinical outcome was defined as the composite of a MELD score increase by 2 points or more, and/or occurrence of any serious adverse event. For patients who were enrolled in the comparator cohort, the responsible clinician reported the same events in a pre-defined form.
All analyses were by intent to treat - patients who received a liver transplant during therapy who did not complete 12 weeks' therapy were regarded as having achieved SVR12 if they had undetectable viral load at 24 weeks from start of treatment; subjects who were lost to follow-up were regarded as treatment failures.
For the primary analysis (SVR12) comparisons were made using Chi-square. Predictors of relapse were also assessed using a logistic regression analysis in which only patients with a known virological outcome were included. Initially we used a univariable model, then all factors were included in a multivariable model, except for those that were likely to be highly collinear (i.e. Child Pugh score which was removed in favour of MELD score).
For all functional outcomes analysis, patients were excluded if they were transplanted before treatment or if treatment was for an extrahepatic indication. The proportion of patients with adverse clinical outcomes was determined and subsequently stratified by various baseline characteristics. The significance of any difference between the treated and comparison group were determined using Chi-square tests. For calculations of MELD changes, we excluded patients who received a liver transplant during the study period, in order to evaluate MELD progression in patients with decompensated cirrhosis over 6 months with and without treatment. Similarly patients who died before study end, who did not have a MELD score at 6 months, were excluded from this analysis. The EAP and comparison cohort were then plotted in separate waterfall plots. We performed a sensitivity analysis excluding current alcohol users (of any number of units, within one month of study start date) to assess the potential impact of differences in alcohol use between the treated and untreated cohorts. Predictors of clinical benefit at baseline (defined as a lack of serious adverse event and no increase in MELD score) was analysed using logistic regression. Predictors were assessed if they were determined a priori to plausibly influence outcome, including age, gender, body mass index (BMI), alcohol use (never, past or current), baseline MELD score, creatinine, albumin, platelets and sodium. The interactions between albumin and age, and baseline sodium, were also explored a priori. Internal validation was carried out using bootstrapping (50 iterations), with a 70% derivation sample. This generated a receiver-operating characteristic (ROC) curve, from which the area under the curve (AUC) was calculated. Model simplicity was prioritised, with additional factors only being accepted into the model if the AUC increased by more than 0.02.
Data handling and statistical analysis were performed using STATA 13.1 with additional analysis performed using Orange 2.7.8.
A total of 480 patients received therapy as part of the EAP between the start of the programme (1 April 2014) to 11 November 2014; 467 patients consented to provide data to the HCV Research UK database (Fig. 1). At treatment initiation 409 (88%) patients had decompensated cirrhosis and/or Child Pugh score ⩾7; 44 (9%) had undergone liver transplantation with aggressive HCV recurrence without decompensation. The remaining 14 (3%) patients were treated for extrahepatic indications. Table 1 and Supplementary Table 1 show the baseline characteristics of the treated cohort. A higher proportion of genotype 3 patients received daclatasvir compared to genotype 1 (125/192 (65%) vs. 46/231(20%)), and clinicians chose to add ribavirin to the regimen for most (427/467 (91%)) patients (Table 2).
From the HCV Research UK database we identified 261 patients with decompensated HCV cirrhosis who fulfilled the criteria for inclusion in the study (Fig. 1). Study start dates for these patients were between 5 October 2012 to 11 September 2014 and 177 (68%) patients subsequently received antiviral therapy on the EAP after its initiation on 1 April 2014. The treated and untreated cohorts had similar baseline characteristics and liver disease severity scores (Table 1). Amongst the comparator patients who did not receive EAP therapy, there was a higher proportion of active alcohol users (23%) compared to those who subsequently received treatment (12.6%).
Sustained virological response (SVR12) was 381/467 (81.6%) overall for this population of patients with severe HCV disease at risk of death or irreversible harm within 12 months (Table 2 and Fig. 2). 51 patients (10.9%) responded but relapsed, 2 (0.43%) were non-responders, 17 (3.6%) died and 16 (3.4%) were lost to follow-up (Supplementary Table 2). Response was markedly influenced by genotype - SVR12 was achieved in 209 of 231 (90.5%) patients infected with genotype 1 compared to 132 of 192 (68.8%) patients with genotype 3 (p <0.0001). There was no statistically significant difference in response between ledipasvir- or daclatasvir-containing regimens in either genotype groups. SVR12 with ledipasvir/sofosbuvir and sofosbuvir plus daclatasvir, with or without ribavirin, was 91.9% (170/185) and 84.8% (39/46) respectively (p = 0.14) for genotype 1 infected patients, and 61.2% (41/67) and 72.8% (91/125) (p = 0.098) for genotype 3 infected patients. The addition of ribavirin numerically increased response (352/427 (82.4%) vs. 29/40 (72.5%), p = 0.11).
Of note, 19 patients had virus detectable at levels below the LLQ (30 IU/ml) after 12 weeks of therapy, and a further 4 patients "blipped" detectable virus below LLQ at 16 weeks (i.e. 4 weeks after end of therapy). 22 of these 23 patients subsequently achieved SVR12, including all 4 of the latter group.
We examined baseline factors that might predict virological failure (Supplementary Table 3). In this analysis we included only patients with documented virological failure and by multivariable analysis, genotype 3 (OR = 10.3, 95% CI 4.4-24.6), BMI >30 kg/m2 (OR = 2.9, 95% CI 1.1-8.2) and detectable virus at treatment week 2 (OR = 2.6, 95% CI 1.1-6.3) were significantly associated with virological failure. The impact of detectable week 2 viral load was greater in patients with genotype 3 than genotype 1. For patients with genotype 3 HCV, virological failure in those with and without detectable virus at treatment week 2 was 32/105 (30.5%) and 7/70 (10.0%) respectively (p = 0.00143). For those with genotype 1 HCV, virological failure in patients with detectable virus at week 2 was 8/141 (5.7%) compared to 4/79 (5.1%) in patients with undetectable virus (p = 0.84). Therapy without ribavirin (OR = 9.0, 95% CI 2.5-31.0) and therapy with ledipasvir compared to daclatasvir (OR = 2.7 95% CI 1.2-6.3) were also significantly associated with virological failure, although this latter finding was likely driven by the differential response in patients with genotype 3 infection, which showed numerical advantage with daclatasvir over ledipasvir.
Treatment with sofosbuvir and an NS5A inhibitor in this patient group with advanced liver disease and/or complex extrahepatic complications was well tolerated. Serious adverse events comprised mainly complications of end-stage liver disease (Supplementary Table 4). Twenty six patients (5.6%) discontinued treatment prematurely, including 7 who died on treatment. Most patients (306/427) treated with ribavirin received the recommended dose of 1.2 g per day for patients weighing >75 kg and 1 g per day for others, 18.7% of patients receiving ribavirin required dose reductions and 9.6% discontinued ribavirin. The median ribavirin dose was 1 g per day. There were only 8 patients with baseline haemoglobin (Hb) below 80 g/L, all but one had ribavirin omitted from their regimen. Grade 3/4 anaemia (Hb ≤80 g/L) occurred in 23 (5.4%) patients receiving ribavirin. Development of acute kidney injury on treatment (defined as creatinine increase of 1.5-fold or higher at treatment week 12 compared to baseline) was infrequent, occurring in 13/467 (2.8%) patients.
We examined only patients with decompensated cirrhosis and/or Child Pugh score B or worse in this analysis (n = 409). Liver function over a 6 month period (3 months on therapy and 3 months post therapy) was compared to the untreated cohort. Fig. 3 and Supplementary Table 5 show MELD score changes and development of adverse clinical outcomes within the two cohorts. Treated patients had a mean negative change in MELD (-0.85, SD 2.54) representing improvement in liver function, whereas untreated patients had a mean positive change (0.75, SD 3.54) representing worsening in liver function (p <0.0001). Significantly fewer treated patients developed a profound worsening in MELD (increase of 2 points or more) over 6 months compared to untreated patients (23.0% vs. 37.9%, p = 0.05). The proportion of patients with at least one decompensating event during the study period was reduced in the treated compared to untreated cohort, apart from the subgroup with baseline MELD score >14. Rates of new decompensation (i.e. development of decompensating event(s) in a patient with recompensated disease at baseline) were significantly lower in the treated cohort (3.7% vs. 10.0%, p = 0.0009) compared to the untreated cohort. There were no significant differences between treated and untreated patients in the incidence of hepatocellular carcinoma, sepsis or death. Overall adverse clinical outcomes (composite of MELD worsening by 2 or more and/or any serious adverse event) were 52.3% in the treated group and 63.6% in the untreated group (p = 0.004).
For treated patients, SVR12 status did not predict MELD score change. However, a smaller proportion of patients who achieve SVR12 had an increase in MELD score ⩾2 compared to those who did not achieve SVR12 (11.9% vs. 68.8%). Adverse outcomes for patients with and without SVR12 were 45.0% and 82.5% (Supplementary Table 5).
To determine whether or not alcohol consumption (which differed in the treated and untreated populations) impacted outcomes, we analysed functional response after excluding patients who were active alcohol users (of any amount) at baseline (Supplementary Table 6). Amongst patients who were past or never users of alcohol, greater extents of improvement were observe in those who received treatment than untreated patients (mean change in MELD score -0.86 vs. 0.68; composite adverse outcome in 52.0% vs. 61.7%).
Overall adverse outcomes were more frequent in patients with high baseline MELD and low albumin (≤35 g/L) for both treated and untreated groups (Supplementary Table 5). For patients with decompensated cirrhosis, we assessed whether baseline characteristics might be useful in predicting functional benefit gained from therapy. We defined functional benefit as no MELD score worsening and no development of serious adverse events following treatment. Patients older than 65 years with reduced synthetic function (serum albumin ≤35 g/L) had the lowest odds of deriving functional benefit with antiviral treatment (Table 3). However the model was not robust with a ROC AUC of 0.5484 (Supplementary Fig. 1). Alternatively, using baseline sodium level with a cut-off at 135 mmol/L increases the AUC to 0.5797 (Supplementary Fig. 2). We did not study the interaction of all three factors due to insufficient number of patients within the subgroups.
The efficacy of all oral antiviral regimens in the management of patients with compensated liver disease due to chronic HCV infection is now established [, , , ] and data on patients with decompensated cirrhosis are emerging [, , , ]. In this study we examined a fixed, 12 week, duration course of antiviral therapy in a large heterogeneous group of patients with decompensated cirrhosis or life-threatening complications of HCV infection, and we compared outcomes to an untreated cohort with equivalent disease and duration of follow-up.
The EAP study was robustly conducted with prospective, standardised monitoring and reporting through a central database - HCV Research UK. The majority of treated patients were included in this report (only 13 patients declined to participate) thus reflecting a true 'real-life' cohort, and a large number of patients infected with genotype 3 were included for the first time. The inclusion criteria for EAP treatment was patients with significant risk of death or irreversible harm within 12 months, mostly with decompensated cirrhosis. Markers of liver disease severity such as MELD score and platelet count were similar to other studies of HCV treatment in decompensated patients (SOLAR-1, ASTRAL-4) [, ]. Within the decompensated subgroup, 17% patients were Child Pugh class A at baseline, but had past decompensation events, and therefore remained at significant risk from severe liver disease. Importantly for the comparator group we selected patients who enrolled in the cohort before therapy was available, thereby reducing the inherent bias in selecting treated patients with prior follow-up.
We noted excellent virological response rates in patients infected with genotype 1 HCV regardless of choice of NS5A inhibitor, but response was statistically significantly lower in patients with genotype 3 HCV. In this non-randomised study SVR was numerically higher in patients with genotype 3 infection who received daclatasvir compared to ledipasvir, although the clinical relevance of this observation is not addressed in this study.
Aside from treatment regimen, we found the presence of detectable HCV RNA at treatment week 2 was an independent predictor of virological failure, and was a stronger predictor than previous treatment history or markers of severe liver disease such as high MELD score or thrombocytopenia. This finding is novel to other analyses of treatment response evaluating only genotype 1 patients with less advanced cirrhosis  and was most marked in patients with genotype 3 infection. On treatment viral response may help to guide the need for extension to 24 weeks of therapy in patients with a lower likelihood of a clinical response, specifically patients with genotype 3 infection.
The beneficial role of ribavirin in treating patients with decompensated cirrhosis is suggested by improved virological response rates in ribavirin-containing regimens (whether combined with ledipasvir or daclatasvir), particularly for patients infected with genotype 3. However the total number of patients treated without ribavirin was small and although no clear adverse events were associated with ribavirin, robust conclusions cannot be drawn. While the clinical trial of ledipasvir/sofosbuvir in patients with decompensated cirrhosis did not include a ribavirin-free arm (SOLAR-1), in ASTRAL-4 which treated 267 patients infected with genotypes 1 to 6 and decompensated cirrhosis with 12 weeks of sofosbuvir and velpatasvir, with or without ribavirin, and 24 weeks of sofosbuvir and velpatasvir alone, ribavirin showed a clear advantage. The study was not powered to detect significance between the three treatment groups, but the addition of ribavirin increased SVR, and to a greater extent than the 24 week duration, particularly in genotype 3 infected patients .
Not unexpectedly from this population with advanced disease, serious adverse events were common, comprising mainly events related to end-stage liver disease. However, treatment was well tolerated with few premature discontinuations.
To assess the impact of antiviral treatment in decompensated cirrhosis, we selected a comparator cohort from patients registered in the national database HCV Research UK who had decompensation, and retrospectively monitored the change in MELD scores and development of serious adverse events over a 6 month period. Patients from both treatment and comparator cohorts were selected from the same network of experienced HCV treatment centres, thus reducing the impact of 'improved care' in patients receiving antiviral therapy. Amongst treated patients a greater proportion showed an improvement in MELD scores, and for patients who had worsened MELD the degree of worsening was less compared to untreated patients, suggesting benefits of therapy within 6 months. This was observed even for patients who failed treatment (mean MELD change -0.63), who nevertheless experienced several months of non-viraemia prior to relapse. We studied a composite endpoint of MELD worsening and development of any serious adverse events, rather than limiting to liver-related events, which we felt was the most clinically useful endpoint. Treated patients had significantly fewer adverse outcomes, with MELD score change and number of decompensation events being the main contribution to the outcome difference, while liver cancers, transplants and deaths were not significantly different over this 6 month period. Treated patients who achieved SVR12 had considerably better functional outcomes than those who were treated but failed to achieve SVR12 (adverse outcomes 45.0% and 82.5% respectively).
This study was not a randomised controlled trial of treatment vs. no treatment, which would be unethical. Wherever possible we attempted to reduce biases in the untreated comparison population. Patients were selected from the HCV Research UK database prior to treatment becoming available. While this population may contain patients who were not treated subsequently because they died or deteriorated to a stage where treatment could not be considered, it also contains patients who survived and were fit enough to receive treatment through the EAP. By using the entire available 'pre-EAP' population, we reduced the possibility of selecting only patients who were fit enough to receive antiviral therapy and we minimised bias. It is possible that differences in severity of liver disease remained between the treated and untreated groups, but these were not evident at baseline measurements, with the exception of alcohol consumption which was greater in the comparator cohort. As this may have contributed to the increased adverse outcomes in the untreated population, we performed a post-hoc analysis excluding active alcohol users in both groups, and found a similar improvement in MELD in treated patients, worsening in MELD and increased adverse outcomes for untreated patients. Monitoring of the untreated group was retrospective but was prospective for the treated group, which might bias the study towards a relative under-reporting of events within the comparator cohort.
Finally we attempted to develop a prediction model for functional outcomes after antiviral treatment to identify patients most likely to benefit from therapy. Regression analysis of baseline characteristics and association with MELD change yielded baseline MELD score as the only significant independent factor. We combined age and serum albumin data and found older patients with poor synthetic function to have a substantially lower chance of benefit compared to younger patients with preserved synthetic function (39.3% vs. 60.5%, p = 0.050). Given the importance of sodium as a variable in predicting mortality risk in addition to MELD score, we analysed the odds ratio of benefit at a cut-off of 135 mmol/L and showed that patients with normal range sodium levels to have higher chance of treatment benefit than patients with hyponatraemia (51.1% vs. 39.1%, p = 0.023). This model did not show sufficient discrimination to allow it to be used to recommend that treatment should be withheld nor is it robust enough to recommend that a patient proceed with liver transplantation, if appropriate. However, given that IFN-free therapy has opened up treatment options for vulnerable patients, the risk estimates provided within the modelling may help clinicians when they discuss the risks and benefits of all oral treatment with patients. Reports of decompensation on DAA therapy for patients with advanced cirrhosis  reiterates the importance of such discussions and the need for a careful evaluation of the different options on a case by case basis.
In summary, this is a large, real-life study of all oral HCV therapy in patients with advanced HCV disease with a significant risk of death or irreversible damage within 12 months. Overall virological response was high and we found early improvement in liver parameters and in clinical outcomes after antiviral treatment, compared to untreated patients. The longer term benefits of therapy in patients with decompensated disease remain to be ascertained.