Hepatitis C Eradication With Direct-acting Anti-virals Reduces the Risk of Variceal Bleeding

Andrew M. Moon; Pamela K. Green; Don C. Rockey; Kristin Berry; George N. Ioannou


Aliment Pharmacol Ther. 2020;51(3):364-373. 

In This Article


Data Source

The VAHS is the largest integrated healthcare provider of HCV anti-viral treatment in the United States.[8] Nationwide, the VA uses a single comprehensive electronic healthcare information network which integrates all care applications into a single, common database. We obtained data on all patients who initiated anti-viral therapy for chronic HCV in the VA system using the VA Corporate Data Warehouse (CDW), a national, continually updated repository of healthcare data including all patient pharmacy prescriptions, demographics, inpatient and outpatient visits, problem lists, procedures, vital signs, diagnostic tests and laboratory tests.[10] The study was approved by the Institutional Review Board of the Veterans Affairs Puget Sound Healthcare System.

Study Population

Using VA pharmacy prescription data, we identified all DAA-only HCV anti-viral regimens initiated in the VA from 2013 to 2015 (Figure 1). We defined sustained virologic response (SVR) as a serum HCV RNA viral load test below the lower limit of detection performed at least 12 weeks after the end of anti-viral treatment.[11] We excluded patients with missing SVR data or prior liver transplant. Patients were excluded if variceal bleed, death or last follow-up visit occurred within the first 90 days after the start of DAA treatment. For patients who received multiple DAA regimens, we analysed only the results of the first one and censored them at the point of a subsequent regimen that resulted in SVR, if one existed. The most common DAA regimen was sofosbuvir/ledipasvir (58.1%) followed by Paritaprevir/Ritonavir/Ombitasvir/Dasabuvir (19.1%), Sofosbuvir (±daclatasvir) (13.1%) and Sofosbuvir + Simeprevir (9.6%).

Figure 1.

Flow diagram of HCV patients included in our analysis cohort. Our cohort included all DAA-treated patients within the national VA from 2013 to 2015, excluding those with missing data, prior liver transplant, or variceal bleed, death or liver transplant within 90 days of DAA treatment. Of our resulting cohort of 33 582 patients, 29 998 (89.3%) had SVR and 3584 (10.6%) had no SVR. Patients were followed for a mean of 3.1 years to assess for the development of variceal bleeding

Gastro-oesophageal Varices

Gastro-oesophageal varices without bleeding were defined by the presence of appropriate diagnostic codes (ICD-9 codes 456.1, 456.21 or ICD-10 codes I85.00, I86.4 or I85.10) recorded at least twice. These diagnostic codes have been validated within the VA and have a positive predictive value of approximately 90% for identifying oesophageal varices compared to chart review.[12,13] Gastro-oesophageal varices with bleeding were defined by the presence of appropriate diagnostic codes (ICD-9 codes 456.0, 456.20 or ICD-10 codes I85.10, I85.01, I86.41 or I85.11) recorded at least once. A single recording was required for bleeding varices (rather than two or more) because many patients may only have a single bleeding episode.

Nonselective beta blockers (NSBBs), frequently used to prevent variceal bleeding, were not included in the diagnostic criteria for gastro-oesophageal varices because they have many other indications (hypertension, angina prophylaxis, essential tremor, migraine prophylaxis, post-traumatic stress disorder) and therefore were not considered specific enough. However, we did evaluate NSBB as a potential confounder of the association between SVR and variceal bleeding.

Baseline Patient Characteristics and Potential Confounders

We collected baseline data including age, sex, race/ethnicity, diabetes, body mass index (BMI), HCV genotype, HCV viral load and receipt of prior anti-viral treatment. We extracted all clinical factors and laboratory tests [including components of the model for end-stage liver disease (MELD) score[14]] prior to treatment and recorded the value of each test closest to the treatment starting date within the preceding 6 months. We defined hepatitis B virus (HBV) co-infection by positive HBV surface antigen or viral load. We also determined the presence of cirrhosis, or complications of cirrhosis (ascites, spontaneous bacterial peritonitis, encephalopathy, gastro-oesophageal varices and hepatorenal syndrome), type 2 diabetes mellitus, alcohol use disorders, substance use disorders, based on appropriate ICD-9 or ICD-10 codes recorded at least twice prior to treatment initiation in any inpatient or outpatient encounter. These ICD-based definitions of cirrhosis and other comorbidities have been widely used and validated in studies using VA medical records.[15–20] We assessed treatment with NSBBs (ie nadolol, propranolol or carvedilol) at the time of DAA initiation, since NSBBs may reduce the risk of variceal bleeding or may be associated with the presence of large gastro-oesophageal varices or prior variceal bleeding requiring prophylaxis.[21–23]

In addition to determining the history of alcohol use disorders by ICD-9/10 codes, we used the Alcohol Use Disorders Identification Test-Consumption (AUDIT-C) score to estimate the severity of alcohol use at baseline. AUDIT-C is a validated screening tool for assessing alcohol misuse and scores range from 0 to 12, with higher scores reflecting higher amounts of alcohol consumption.[24,25] Since 2004, AUDIT-C has been used to screen all veterans for unhealthy alcohol use annually in the outpatient setting.[26] Baseline alcohol use was defined by the AUDIT-C score reported within 1 year before the initiation of anti-viral treatment and categorised into abstinent (score of 0), low-level drinking (score of 1–3 in men, 1–2 in women), and unhealthy drinking (score of 4–12 in men, 3–12 in women).[27,28]

Statistical Analysis

We used Cox proportional hazards regression to compare patients who achieved SVR to those who did not achieve SVR with respect to the risk of developing gastro-oesophageal variceal bleeding after anti-viral treatment (or, more accurately 90 days after the initiation of anti-viral treatment). Our comparison group was DAA-treated patients who did not achieve SVR, rather than patients who were never treated with DAAs, to avoid the risk of health initiator bias (selective provision of treatments to healthy and health-conscious patients and avoidance of treatment of frail individuals[29,30]) and immortal time bias (treated patients experience "immortal time" prior to treatment during which outcomes cannot occur). Any episodes of variceal bleeding that occurred within 90 days of initiation of treatment were excluded because they occurred during the DAA treatment and might have caused the failure of treatment, thus resulting in a spurious association between failure of treatment and variceal haemorrhage.

Analyses were adjusted for the following potential confounders that may be associated with both SVR and the risk of progressive liver disease and variceal bleeding: cirrhosis, prior history of varices, variceal bleeding, ascites, bacterial peritonitis or encephalopathy, age, sex, race/ethnicity, body mass index, HBV co-infection, type 2 diabetes mellitus, hepatocellular carcinoma, alcohol use disorders, substance use disorder, baseline alcohol use, NSBB use, platelet count, serum bilirubin, serum creatinine, serum albumin, INR and haemoglobin levels. Continuous variables were categorised and modelled as dummy categorical variables. Cutoffs were chosen to correspond to clinically useful categories, when applicable. BMI cutoffs for overweight/obese/morbidly obese, platelet cutoffs to correspond to Baveno criteria for variceal screening, and MELD categories were chosen based on values demonstrated to predict mortality among patients with variceal bleeding.[1,31] Haemoglobin cutoffs were chosen based on quartiles to equalise sample sizes between groups since we had no a priori clinically important cutoffs. Follow-up for the development of variceal bleeding extended until 1 January 2019 so that even the patients treated in late 2015 (ie the most recent in our cohort) had adequate follow-up. Patients without incident variceal bleeding were censored at the time of death or last follow-up in the VA. We presented subgroup analyses according to prior history of varices or variceal bleeding, cirrhosis, MELD score, and alcohol use disorders (suggesting comorbid alcohol-related liver disease). A global test of the proportional hazards assumption was performed for the Cox proportional hazards models (Table 3 and Table 4). To prevent multiple comparisons bias, the global test was only performed for the entire cohort and not for other subgroups.

Survival analyses were stratified by the Veterans Affairs facility at which the anti-viral treatment was administered. We analysed only the first anti-viral regimen administered between 2013 and 2015. Patients who did not achieve SVR with this regimen and were subsequently treated again at any point up to 1 January 2019 and achieved SVR, were censored at the time of initiation of the regimen that led to SVR.