Impact of direct-acting antiviral agents on liver function in patients with chronic hepatitis C virus infection

Whilst the benefit of direct-acting antiviral agents (DAAs) in achieving sustained virological response (SVR) is now well-accepted, their impact on liver function, particularly in relation to achievement of SVR, has not been well documented. We studied 2394 patients with chronic HCV infection, 1276 receiving DAAs and 1118 interferon-based therapy. Liver function was assessed by the albumin-bilirubin (ALBI) score or grade. Overall survival according to SVR status and baseline ALBI grade was exam-ined. We also studied time to first decompensation according to ALBI grade, as well as longitudinal changes in ALBI score over time according to SVR. Among the patients receiving DAAs, 89% achieved SVR (Japan = 99%, UK = 78%). Amongst the decompensated patients in the UK cohort, three distinct risk groups according to ALBI grade at baseline were observed. The UK patients receiving DAAs, who had


| INTRODUC TI ON
The clinical benefits of achieving a sustained virological response (SVR) to antiviral therapy in patients with chronic hepatitis C virus (HCV) infection have been extensively documented. These include a decrease in liver-related complications (such as hepatocellular carcinoma [HCC] and hepatic decompensation), non-liver-related complications (such as development of insulin resistance and lymphoma), together with improvements in quality of life for patients and economic benefit for healthcare providers. [1][2][3][4][5] Improvement in fibrosis and even reversal thereof has also been observed 6 and, it has been assumed, there is also improvement in liver function. Nonetheless, since DAAs have only recently been introduced, there is very short follow-up, leading a recent Cochrane review to conclude that 'there are no long-term trials that have assessed whether or not DAA treatment improves morbidity or mortality' 7 although this opinion has been robustly challenged. 1 Furthermore, it has even been suggested that the 'uncertain clinical benefit', along with the risks and costs of DAA therapy, should be discussed with patients to allow a shared decision on treatment to be made. 7 We have recently shown that several of the factors involved in the Childs-Pugh score (CPS), the conventional measure of liver function, are 'redundant'. Thus, the CPS can be refined in a simple algorithm to involve only serum albumin and bilirubin (the albumin-bilirubin score [ALBI]). 8 The model was originally developed for patients with HCC and has been extensively validated. 9 More recently, the model has been validated in patients with chronic liver disease without HCC. 10 The ALBI score has the advantage that, as well as avoiding the necessity of using highly subjective variables (ascites and encephalopathy) as required for the Child-Pugh score, it has sufficient granularity to permit sensitive quantification of serial changes in liver function over time.
Our primary aim here is to report changes in liver function in patients with decompensated disease consequent upon treatment with direct-acting antiviral agents (DAAs), but for comparison, we also include a large cohort of patients with compensated disease also treated with DAAs. A cohort of patients receiving interferon-based therapies is also included to give an indication of how liver function can change over the long term in response to effective antiviral treatment. Patients were classified as having achieved SVR, relapse after initial response (REL) or no-response (NR) according to internationally agreed criteria. 11 Because there were only very minor differences between those with REL and NR with regard to any of the recorded parameters, REL and NR were combined and analysed as a 'non-SVR' cohort. A small group of Japanese patients who underwent a second course of DAA treatment (n = 17) were similarly classified; those who had response unclassified at the time of analysis were excluded.

| The UK DAA cohort
This observational cohort comprised 602 patients prospectively collected and treated for liver disease within the NHS England Expanded Access Programme (EAP), data for which are held by predominantly decompensated disease, showed clear evidence of improvement of liver function detectable within 2 years of the start of treatment, especially in those achieving SVR. These early changes in liver function were very similar to those observed in the first 2-3 years after interferon-based therapy. DAAs improve liver function especially in those with decompensated disease who achieve SVR. Experience with interferon-based therapy suggests that failure to achieve SVR is associated with long-term decline in liver function and, in contrast, patients who do achieve SVR can expect long-term disease improvement and subsequent stabilization of liver function.
Our initial analysis suggests that those receiving DAAs are likely, in the long term, to follow a similar course.

K E Y W O R D S
ALBI score, direct-acting antivirals, hepatitis C virus, interferon therapy, liver function, sustained virological response the UK Biobank, HCV Research UK. Patients eligible for treatment within the EAP were those at significant risk of death or irreversible damage from HCV infection within 12 months, regardless of genotype. Patients were treated with 12 weeks of Sofosbuvir along with either Daclatasvir or Ledipasvir (at clinician discretion) and ribavirin (at clinician discretion). All patients included within this study consented to contribute their data to the database.

| The Japanese interferonbased therapy cohort
We

| Kaplan-Meier survival analysis
Overall survival according to (a) each of the three cohorts, (b   The fractional polynomial method requires a minimum of 4 observations in order to construct a curve for a patient. Due to long follow-up, this was possible to undertake in the Japanese IFN patients. However, in the DAA cohorts, follow-up was shorter (about 24 months) and the average number of observations per patient was less. In this case, linear regression was used to construct the aggregate curves.

| RE SULTS
There was a significant difference in survival between the UK group with decompensated disease and the Japanese with compensated disease ( Figure 1). Those who achieved SVR survived longer in the UK cohort (P < .0001) but the number of events in the Japanese data set was too small for meaningful statistical analysis. (Figure 2

| Overall survival and clinically significant deterioration according to ALBI grade
There was some survival discrimination according to baseline ALBI grade in the UK decompensated cohort (Figure 3). In the same cohort, baseline ALBI grade also showed three distinct groups with different time to first decompensation ( Figure 4).
Equivalent analysis in the Japanese DAA patients was not possible as the number of patients with the relevant outcomes was very low.

| Changes in ALBI score over time-impact of treatment on the ALBI score
For a subset of patients, longitudinal data were available for 571 Japanese patients (including 9 achieving SVR) and 469 UK patients (including 77 achieving SVR). These were used to examine changes in ALBI score over time.

| UK DAA cohort
The mean ALBI score at baseline was −1.76 (CI −1.82, −1.72), while the mean ALBI score for the last ALBI was a significantly better (P < .01) at −2.04 (CI −2.09, −1.99). The median interval between these first and last measurements was 8.9 months (IQR 5.7, 15.2), meaning that for a large majority of patients, the last ALBI measurement was taken well after treatment was completed. Patients who achieved an SVR within 12 weeks of the end of treatment improved significantly more (change of −0.31) than those who did not (change of −0.14), P = .0031 ( Figure 5A,B).

| Japanese Interferon cohort of non-SVR patients
In the interferon-based group, survival at 20 years was 89.00% (95% CI 84.21, 92.41) in the SVR group compared to 66.13% (95% CI 59.25, 72.13) in the non-SVR group. Among the 134 patients in the Japanese interferon-based cohort who died, the major cause of death was HCC (45.5%) and, of these, the great majority (90.2%) occurred among the group who did not achieve SVR. Overall only 5.22% of those who died (or 0.63% overall) were recorded as dying from liver failure, the great majority of these (85.7%) also being in the non-SVR group.
Among the SVR cohort, the ALBI score improved slightly over the first 4 years and then remained largely stable for up to 24 years.
By contrast, in the non-SVR group, after a small initial improvement there was a steady deterioration in liver function ( Figure 5).
It is of particular interest to note that the early changes in ALBI, over the 2 years following initiation of DAA treatment, are very closely reflected in the first few years of the interferon-based treatment cohort in that in both SVR and non-SVR, there is an improvement in liver function. However, with insight into the long-term outcome it is apparent that among the SVR patients liver function stabilized whereas there was a steady deterioration in the non-SVR group.

| DISCUSS ION
Our study design permits us to assess the impact of a broad range of the UK, they were initially confined to patients with decompensated disease on the grounds that these were at the most imminent risk of death from liver failure. However, in both instances the follow-up is, of necessity, short, and therefore, our long-term data on the impact of interferon-based therapy on survival and liver function allow us to reasonably extrapolate DAA-related changes in liver function over the longer term.
Here, we document that liver function improves in patients receiving DAAs, particularly in those who achieve SVR, and most strikingly in those who have decompensated disease. The changes in the DAA cohort in a 3-year period are very similar in character to those seen in the interferon-treated groups over the first 3 years suggesting that the DAA cohort is likely to experience similar beneficial changes over the longer term as was seen in the interferon-treated group.
Our method of irregular time series analysis, together with the extensive serial data and use of the ALBI score (rather than the Child-Pugh score), allows us to visualize and quantify these changes over time. It is apparent, in all datasets, that even in those who do not achieve SVR, there is modest improvement in liver function over the first 3 years after initiation of therapy. We assume that even in the non-SVR group, there is transient improvement in liver function associated with a fall in viral load on treatment (indeed we know this to be the case because they include some patients who relapsed after initial response) that persists for a short time following treatment, but recurs with the post-treatment rise in viral load and subsequent recurrence of liver injury.
It is also striking that, associated with the improvement in liver function, there is, in the UK cohort with decompensated disease, clear evidence of survival benefit in those achieving SVR (compared to those not achieving SVR) even over the short period of follow-up.
Again, extrapolating from the Interferon-treated group there is clear evidence of long-term survival benefit with the achievement of SVR although it is important to note that this benefit is consequent upon the decreased incidence of HCC rather than any improvement in mortality from liver failure, the incidence of which was, in fact, very low. survival and the likelihood of further adverse liver-related events.
The changes in liver function and survival with DAAs mirror, in the short term over which they have been observed, analogous results in a population receiving interferon-based therapy. In the latter cohort, such changes were, on long follow-up, associated with improved liver function and survival. It is likely that the DAA-mediated improvement of liver function in patients who achieve SVR will translate into survival benefit. F I G U R E 5 A and B, Changes in albumin-bilirubin (ALBI) score over time in the (A) UK DAA and the (B) Japanese interferon cohorts. Note the similarity in the changes between the two cohorts over the first 3 y