fax +39 0263632714
Liver Failure/Cirrhosis/Portal Hypertension
Sustained virologic response prevents the development of esophageal varices in compensated, Child-Pugh class A hepatitis C virus–induced cirrhosis. A 12-year prospective follow-up study†
Article first published online: 27 JAN 2010
Copyright © 2010 American Association for the Study of Liver Diseases
Volume 51, Issue 6, pages 2069–2076, June 2010
How to Cite
Bruno, S., Crosignani, A., Facciotto, C., Rossi, S., Roffi, L., Redaelli, A., de Franchis, R., Almasio, P. L. and Maisonneuve, P. (2010), Sustained virologic response prevents the development of esophageal varices in compensated, Child-Pugh class A hepatitis C virus–induced cirrhosis. A 12-year prospective follow-up study. Hepatology, 51: 2069–2076. doi: 10.1002/hep.23528
Potential conflict of interest: Nothing to report.
- Issue published online: 23 MAY 2010
- Article first published online: 27 JAN 2010
- Accepted manuscript online: 27 JAN 2010 12:00AM EST
- Manuscript Accepted: 8 JAN 2010
- Manuscript Received: 20 NOV 2009
The incidence of de novo development of esophageal varices (EV) in patients with compensated liver cirrhosis has been determined by few studies in the short term and never in the long term. The aims of the present study were to determine the incidence and the risk factors associated with the development of EV and to assess whether antiviral treatment and achievement of sustained virologic response (SVR) may prevent de novo EV development in patients with HCV-induced cirrhosis. We studied 218 patients with compensated EV-free, HCV-induced cirrhosis consecutively enrolled between 1989 and 1992 at three referral centers in Milan, Italy. Endoscopic surveillance was performed at 3-year intervals according to international guidelines. SVR was defined as undetectable serum HCV-RNA 24 weeks after treatment discontinuation. During a median follow-up of 11.4 years, 149/218 (68%) patients received antiviral treatment and 34 (22.8%) achieved SVR. None of the SVR patients developed EV compared with 22 (31.8%) of the 69 untreated subjects (P < 0.0001) and 45 (39.1%) of the 115 non-SVR patients (P < 0.0001). On multivariate analysis, HCV genotype 1b (hazard ratio [HR] 2.40; 95% confidence interval [CI] 1.17-4.90) and baseline model for end-stage liver disease (MELD) score (HR 1.20; 95% CI 1.07-1.35 for 1 point increase) were independent predictors of EV. Conclusion: In the long term, the achievement of SVR prevents the development of EV in patients with compensated HCV-induced cirrhosis. Therefore, in these patients, endoscopic surveillance can be safely delayed or avoided. Genotype 1b infection and MELD score identify the subset of patients at higher risk of EV development who need tailored endoscopic surveillance. Hepatology 2010
The natural history of patients with compensated liver cirrhosis, induced by hepatitis C virus (HCV) is now well characterized.1-5 However, even if it is well established that the onset of esophageal varices (EV) marks a crucial turning point in the outcome of the disease,6 only one study has assessed the incidence of de novo development of EV prospectively over a short period. However, this study combined patients with alcoholic, HCV-related, hepatitis B virus–related and cryptogenic cirrhosis.7 In addition, the effect of antiviral therapy and the achievement of sustained virologic response (SVR) on EV development has never been evaluated. Finally, apart from the level of the hepatic vein pressure gradient (HVPG), which has recently been shown to be associated with the risk of EV development,7 no other reliable predictor of EV has been described.
The availability of a large prospective cohort study5 that included consecutive patients with compensated HCV-induced cirrhosis followed for up to 18 years allowed us to evaluate the incidence of, the risk factors associated with, and the impact of antiviral treatment, including the effect of SVR, on de novo development of EV in the long term.
Patients and Methods
The present study is a subgroup analysis based on data obtained from a large-scale prospective study aimed at evaluating the long-term outcome of patients with compensated HCV-induced cirrhosis, which has been described elsewhere.5 Briefly, from January 1989 to December 1992, all consecutive patients with compensated Child-Pugh class A cirrhosis who presented at three referral centers in the Milan area (Northern Italy) and tested positive for serum anti-HCV were enrolled. The present study followed the guidelines for subgroup analyses8 and included only subjects who had agreed to undergo upper endoscopy at the time of enrollment (±6 months) and were found to be EV-free.
The Ethics Committees of all participating centers approved the design of the study.
The study includes Child-Pugh class A patients with HCV-related cirrhosis <70 years of age. Patients with bridging fibrosis were excluded. The diagnosis of cirrhosis was based on liver biopsy or on clinical criteria.9 Patients with concurrent hepatitis B or human immunodeficiency virus infection, and patients who were not willing to attend regular follow-up endoscopy were excluded. Patients with previous episodes of decompensation or with hepatocellular carcinoma (HCC) within 6 months from enrollment were also excluded. Anti-HCV was assessed by first- and second-generation enzyme-linked immunosorbent assay. HCV genotype was determined on frozen sera by nested reverse-transcription polymerase chain reaction of HCV core sequences using type-specific primers10 and confirmed on fresh sera by INNO-LiPA (HCVII, Innogenetics, Ghent, Belgium). Information on alcohol intake (>60 g/day for women, >80 g/day for men, for at least 5 years) was obtained during baseline interview and confirmed by relatives.
All patients underwent a 3-month run-in period, and a database including demographic, clinical, laboratory, and endoscopic examinations obtained at entry was set up. Endoscopic procedures for EV assessment were performed in each center by three skilled endoscopists. The size of EV was determined at medium insufflation and classified according to North Italian Endoscopic Club score.11 Child-Pugh score was assessed according to current criteria,12 whereas the MELD score13 was retrospectively calculated using the information collected at baseline. international normalized ratio was calculated by conversion of prothrombin time or activity at baseline.
All patients had a regular follow-up that included clinical, laboratory, and abdominal ultrasound surveillance every 6 months. Examination of the upper gastrointestinal tract was planned at 3-year intervals both for treated and untreated patients, as recommended by international guidelines.14 Esophago-gastro-duodenoscopy (EGD) was also performed at the initiation of antiviral treatment or when clinically required. Additional EGD surveillance after the diagnosis of HCC was not scheduled, unless patients were eligible for surgical resection. At the time of EV detection, beta-blockers were prescribed to patients with medium or large (F2 or F3) varices, while no prophylaxis was scheduled for those with small (F1) varices.
In order to assess the effect of SVR on portal hypertension progression, the measurement of HVPG was also offered to patients who achieved SVR and agreed to perform this procedure. The first HVPG measurement (Time 0) was made 12 months following discontinuation of antiviral treatment and repeated every 3 years (at the same time as EGD). HVPG was measured according to established recommendations.15
Treatment of HCV Infection.
Recombinant interferon (IFN)-α monotherapy or combination with both IFN and ribavirin were offered over time. IFN monotherapy, at the dose of 3 MIU three times a week, was administered, regardless of HCV genotype, for at least 6 months and for an additional 6-month period in patients who achieved a complete biochemical response. Combination therapy with IFN or pegylated IFN and ribavirin was administered in agreement with international guidelines. SVR was defined as undetectable serum HCV-RNA (<50 IU/mL) 6 months after stopping therapy.
Continuous variables were compared using the Mann-Whitney test. Fisher's exact test and the Mantel-Haenszel chi-square test for trend were used to assess differences in the baseline characteristics between patients who did and did not receive antiviral treatment and between treated patients who did or did not achieve an SVR. Time-to-event was calculated from the date of enrollment to the date of first detection of EV or to the date of last EGD, or death. The cumulative incidence of EV during follow-up was plotted using the Kaplan-Meier method, and the log-rank test was used to assess difference between strata. When specified, observation time was censored at the time of SVR. All patients were included and the time of observation of SVR subjects was censored when successful antiviral treatment was initiated. This choice was justified by the fact that SVR patients were still HCV-RNA–positive in the same way as untreated and non-SVR treated patients until they started antiviral treatment. Univariate and multivariate Cox proportional hazards regression models were used to identify factors associated with EV development. Intercurrent events such as initiation of antiviral therapy, SVR achievement, or HCC occurrence were set as time-dependent covariables. Analyses were performed with SAS version 8.2 (Cary, NC). All statistical tests were two-sided, and P < 0.05 was considered significant.
Among the 352 patients included in the original study,5 26 did not agree to undergo EGD at the time of enrollment, and 54 had EV at baseline. In the remaining 272 patients free of varices at baseline, 54 (19.9%) did not adhere to the endoscopic surveillance program and were thus excluded from the present study (Fig. 1). The baseline characteristics of the 218 patients included in the study, stratified according to antiviral treatment and response to therapy, are shown in Table 1.
|Variable||All Patients||Antiviral Treatment||SVR|
|No||Yes||P Value||No||Yes||P Value|
|<50||45 (20.6%)||6 (8.7%)||39 (26.2%)||27 (23.5%)||12 (35.3%)|
|50-54||31 (14.2%)||7 (10.1%)||24 (16.1%)||18 (15.7%)||6 (17.7%)|
|55-59||52 (23.9%)||12 (17.4%)||40 (26.9%)||31 (27.0%)||9 (26.5%)|
|60-64||67 (30.7%)||30 (43.5%)||37 (24.8%)||33 (28.7%)||4 (11.8%)|
|65+||23 (10.6%)||14 (20.3%)||9 (6.0%)||<0.0001||6 (5.2%)||3 (8.8%)||0.14|
|Male||111 (50.9%)||30 (43.5%)||81 (54.4%)||61 (53.0%)||20 (58.8%)|
|Female||107 (49.1%)||39 (56.5%)||68 (45.6%)||0.15||54 (47.0%)||14 (41.2%)||0.55|
|1b||134 (61.5%)||38 (55.1%)||96 (64.4%)||85 (73.9%)||11 (32.4%)|
|2a||62 (28.4%)||17 (24.6%)||45 (30.2%)||0.89||26 (22.6%)||19 (55.9%)||<0.0001|
|Other/unknown||22 (10.1%)||14 (20.3%)||8 (5.4%)||4 (3.5%)||4 (11.8%)|
|No||167 (76.6%)||49 (71.0%)||118 (79.2%)||94 (81.7%)||24 (70.6%)|
|Yes||51 (23.4%)||20 (29.0%)||31 (20.8%)||0.18||21 (18.3%)||10 (29.4%)||0.16|
|<8||85 (39.0%)||17 (24.6%)||68 (45.6%)||52 (45.2%)||16 (47.1%)|
|8-9||69 (31.7%)||23 (33.3%)||46 (30.9%)||37 (32.2%)||9 (26.5%)|
|≥10||56 (25.7%)||26 (37.7%)||30 (20.1%)||0.0009||23 (20.0%)||7 (20.6%)||0.88|
|Unknown||8 (3.7%)||3 (4.4%)||5 (3.4%)||3 (2.6%)||2 (5.9%)|
|Median (25%-75% quartiles)||8 (7-10)||9 (7-10)||8 (7-9)||0.003||8 (7-9)||7 (6-9)||0.68|
|<5 ng/mL||90 (41.3%)||31 (44.9%)||59 (39.6%)||35 (30.4%)||24 (70.6%)|
|5-10 ng/mL||73 (33.5%)||16 (23.2%)||57 (38.3%)||50 (43.5%)||7 (20.6%)|
|10-20 ng/mL||27 (12.4%)||11 (15.9%)||16 (10.7%)||15 (13.0%)||1 (2.9%)|
|>20 ng/mL||26 (11.9%)||11 (15.9%)||15 (10.1%)||0.27||14 (12.2%)||1 (2.9%)||0.0003|
|Unknown||2 (0.9%)||0 (0.0%)||2 (1.3%)||1 (0.9%)||1 (2.9%)|
|Median (25%-75% quartiles)||6 (3-10)||6 (3-12)||6 (3-9)||0.64||7 (4-10)||3 (3-5)||<0.0001|
One hundred sixty-three patients (74.8%) had a diagnosis of cirrhosis obtained by way of liver biopsy at the time of inclusion in the study. Half of them were males (50.9%), the majority infected with HCV genotype 1b (61.5%). Approximately one quarter of patients admitted an excess of alcohol consumption during their life. Treated patients were significantly younger, were less likely to be past heavy alcohol drinkers, and had a lower MELD score compared with untreated ones. Among treated patients, only HCV genotype and baseline alpha-fetoprotein serum levels were associated with SVR (Table 1).
Cumulative Incidence of EV.
During a median follow-up of 11.4 years, de novo EV were detected in 67 patients, at the same frequency in untreated 22/69 (31.8%) and treated 45/115 (31.8%) patients. Overall, eight patients developed EV during the first 3 years of follow-up. At the time of detection, the size of EV was small (F1) in 51 (76.1%) patients, medium (F2) in eight (11.9%) patients, and large (F3) in eight (11.9%) patients. Only one of the eight patients with F2 varices but five of the eight patients with F3 varices bled at the time of recognition. The median time between enrollment and detection of F3 varices was 8 years (range, 3-17).
By contrast, no single EV was detected among the 34 patients who achieved SVR (P < 0.0001 versus both untreated and non-SVR patients). Of interest, the median duration of follow-up from enrollment to last EGD was 7.5 years for untreated patients, 10.7 years for non-SVR patients, and 15.9 years for SVR patients (P < 0.0001). Among treated patients, the median duration between initiation of antiviral therapy and last EGD was 8.8 years for non-SVR patients and 12.0 years for SVR patients (P = 0.006). The cumulative incidence of EV among treated patients according to response to therapy is shown in Fig. 2.
During follow-up, 16 of the patients who had developed EV subsequently bled at the same rate in untreated and non-SVR treated subjects (Fig. 1).
Figure 3 provides information on the number of patients, person-years of observations events and the 10-year cumulative incidence of EV in the whole series of patients and stratified according to HCV genotype and MELD score.
Incidence of EV and HCC Development.
During follow-up, 66 patients developed HCC, seven of whom had achieved an SVR (Fig. 1). HCC was diagnosed as a single nodule (<2 cm) in 60% of the cases, and 76% of the patients fulfilled the Milan criteria.19 In 57 cases (86%), patients were free of varices at the last EGD performed before tumor diagnosis. Following HCC diagnosis, EV were detected in 16 patients (in 12 cases within 3 years of tumor occurrence).
Predictors of EV Development.
On univariate analysis, HCV genotype 1b (hazard ratio [HR] 2.38; 95% confidence interval [CI] 1.23-4.59), history of heavy alcohol consumption (HR 2.00; 95% CI 1.16-3.43), elevated serum creatinine (HR 2.41; 95% CI 1.03-5.66), higher MELD score (HR for 1 point increase 1.21; 95% CI 1.08-1.35), alpha-fetoprotein serum leveln ≥10 hg/mL (HR 1.99; 95% CI 1.20-3.29) and occurrence of HCC (HR 2.65; 95% CI 1.45-4.86) were associated with development of de novo EV (Table 2). We found no association with platelets, albumin, international normalized ratio, or bilirubin levels at baseline.
|Baseline Characteristics||Comparison Groups||Univariate||Multivariate 1||Multivariate 2|
|Age||≥55 years versus <55 years||0.90 (0.55-1.47)||0.83 (0.48-1.42)||0.67 (0.38-1.17)|
|Sex||Male versus female||1.17 (0.72-1.90)||0.84 (0.48-1.47)||0.71 (0.40-1.26)|
|Genotype||1b versus 2a/2c||2.38 (1.23-4.59)||2.40 (1.17-4.90)||2.23 (1.09-4.57)|
|Alcohol intake||Yes versus no||2.00 (1.16-3.43)||1.34 (0.74-2.43)||1.20 (0.65-2.21)|
|Platelets||<80,000/mL versus ≥80,000/mL||1.72 (0.54-5.53)|
|Albumin||≤3.5 g/dL versus >3.5 g/dL||1.25 (0.17-9.03)|
|International normalized ratio*||>1.2 versus ≤1.2||1.61 (0.98-2.63)|
|Bilirubin*||≥1.2 mg/dL versus <1.2 mg/dL||1.67 (0.91-3.08)|
|Creatinine*||≥1.2 mg/dL versus <1.2 mg/dL||2.41 (1.03-5.66)|
|MELD score||1-point increase||1.21 (1.08-1.35)||1.20 (1.07-1.35)||1.25 (1.10-1.41)|
|Alpha-fetoprotein||≥10 ng/mL versus <10 ng/mL||1.99 (1.20-3.29)||1.61 (0.95-2.74)||1.45 (0.84-2.49)|
|Antibody to hepatitis B core antigen||+ve versus −ve||1.35 (0.82-2.20)|
|Diabetes||Yes versus no||1.50 (0.76-2.98)|
|Antiviral treatment**||Yes versus no||1.02 (0.60-1.72)|
|SVR†||Yes versus no||No events|
|HCC‡||Yes versus no||2.65 (1.45-4.86)||2.87 (1.46-5.64)|
On multivariate analysis, HCV genotype 1b and baseline MELD score remained independent predictors of EV (Table 2). When intercurrent events were included in the multivariate model, antiviral treatment in absence of SVR did not show any protective effect on EV development (HR 1.02; 95% CI 0.60-1.72). By contrast, HCC occurrence was associated with an HR of 2.87 (95% CI 1.46-5.64) on the development of EV (Table 2). This latter result was not confirmed when the analysis was restricted to the subset of treated patients (Table 3).
|Baseline Characteristics||Comparison Groups||Univariate||Multivariate 1||Multivariate 2|
|Age||≥55 years versus <55 years||0.79 (0.44-1.43)||0.64 (0.32-1.27)||0.57 (0.29-1.15)|
|Sex||Male versus female||1.07 (0.59-1.92)||0.64 (0.31-1.32)||0.60 (0.29-1.23)|
|Genotype||1b versus 2a/2c||2.46 (1.03-5.87)||3.21 (1.28-8.03)||3.00 (1.19-7.52)|
|Alcohol intake||Yes versus no||2.10 (1.05-4.21)||1.56 (0.74-3.27)||1.61 (0.76-3.41)|
|Platelets||<80,000/mL versus ≥80,000/mL||0.74 (0.10-5.39)|
|Albumin||≤3.5 g/dL versus >3.5 g/dL||4.72 (0.61-36.4)|
|International normalized ratio*||>1.2 versus ≤1.2||1.67 (0.89-3.14)|
|Bilirubin*||≥1.2 mg/dL versus <1.2 mg/dL||1.63 (0.76-3.53)|
|Creatinine*||≥1.2 mg/dL versus <1.2 mg/dL||1.66 (0.59-4.66)|
|MELD score||1-point increase||1.19 (1.03-1.39)||1.21 (1.02-1.44)||1.25 (1.05-1.48)|
|Alpha-fetoprotein||≥10 ng/mL versus <10 ng/mL||1.45 (0.77-2.72)||1.48 (0.76-2.91)||1.40 (0.71-2.76)|
|Antibody to hepatitis B core antigen||+ve versus −ve||1.21 (0.68-2.19)|
|Diabetes||Yes versus no||1.80 (0.82-3.95)|
|SVR†||Yes versus no||No events|
|HCC‡||Yes versus no||1.60 (0.73-3.52)||2.07 (0.86-5.00)|
When we limited the analysis to F2 or F3 varices (n = 17), a history of heavy alcohol consumption (HR 5.79; 95% CI 1.99-16.8), higher MELD score (HR for 1 point increase 1.27; 95% CI 1.01-1.58), alpha-fetoprotein serum level ≥10 hg/mL (HR 2.92; 95% CI 1.09-7.87), and occurrence of HCC (HR 4.32; 95% CI 1.41-13.2) were associated with development of medium or large varices in univariate analysis, whereas MELD score and development of HCC remained independent predictors of EV on multivariate analysis (data not shown).
In recent years, several studies have described disease outcomes in patients with compensated HCV-induced cirrhosis.1-5 However, although these investigations provide relevant information, none of them report data about the de novo development of EV. In addition, even if there is growing evidence that antiviral treatment may reduce the rate of complications related to portal hypertension when leading to SVR, no study has evaluated whether viral eradication could prevent EV development in the long term.
The present investigation, which fulfills the quality criteria requirements for both observational20 and subgroup analysis studies,8 provides several important data. First, it provides an accurate estimate of the 10-year cumulative incidence of EV in this population of patients. Our figures are lower than those recently reported by Groszmann et al.7 in a 5-year follow-up study. This discrepancy could be easily explained by the different characteristics of patients included in the two studies. In the latter study, more than one third of subjects had nonviral cirrhosis (predominantly alcohol-related), 12% had more advanced disease (Child-Pugh score B), all patients with hepatitis B virus–induced disease did not assume antiviral therapy, and the timing of endoscopic surveillance between studies was different (every 3 years, as suggested by the international guidelines in clinical practice, instead of annually). Moreover, a high proportion of the patients included in the study by Groszmann et al. had a high HVPG (63% had at least 10 mm Hg, with a median of 11 mm Hg). By contrast, the majority of our patients had the diagnosis of cirrhosis made by liver biopsy at the time of enrollment (i.e. at an early, less severe stage of the disease). On the other hand, it is well-established that the stage of compensated cirrhosis is extremely heterogeneous because it includes patients with different but still unrevealed clinical conditions. Moreover, approximately 20% of our patients did not agree to fulfill the planned schedule of endoscopic surveillance and this fact might have led to underestimation of the rate of EV detection.
A major finding of our study—and one of great importance in clinical practice—is that the achievement of SVR abolishes the development of EV in the long term. The reliability of our result is guaranteed by the ample length of observation among this group of patients (median follow-up after initiation of antiviral therapy being 12 years).
The biological plausibility of our result was supported by the finding that HVPG decreased over the years in the four SVR patients who agreed to undergo this procedure and, in all of them, the HVPG values reached the threshold level (<10 mm Hg), which has been reported to be protective for EV development (data not shown). These data are also supported by other studies in which the achievement of virological response following antiviral therapy was associated with a significant decrease of HVPG values,21, 22 fibrosis regression,23, 24 and reduction of portal hypertension–related complications in patients with similar clinical characteristics.25, 26
This study failed to identify a reliable predictor associated with high risk for early EV development. This was due to the low incidence of events in the first 3 years of follow-up. However, we found that a higher MELD score at entry and HCV genotype 1b infection are independent predictors of EV development in the long-term. Whereas the higher risk of EV in patients infected with genotype 1b could be attributable to the low rate of SVR in these patients, the statistically significant association in multivariate analysis suggests that genotype 1b may induce a faster progression of portal hypertension as suggested by a recent meta-analysis.27
Together with an HVPG value ≥10 reported by Groszmann et al.,7 HCV genotype and MELD score may be easily used to identify subjects in whom more frequent, tailored EGD surveillance is warranted.
Another finding emerging from this study is that the risk of EV development increases once HCC has occurred. This result reveals a limitation in our study, however, because frequency of EGD surveillance was performed at 3-year intervals, whereas ultrasound screening for HCC was performed every 6 months. Therefore, we cannot exclude that some patients had developed EV before the emergence of HCC and that the observed rates of EV after the diagnosis of HCC have been overestimated.
In conclusion, the 10-year cumulative incidence of de novo EV in EV-free patients with compensated Child-Pugh class A HCV-induced cirrhosis is fairly low. Despite the presence of established cirrhosis, achievement of SVR after antiviral therapy prevents the development of EV in the long-term. In routine clinical practice, serial surveillance by EGD can be safely delayed or avoided in SVR patients, sparing significant amount of useless invasive and costly procedures. It is conceivable that, as a result of new promising therapies,28 the rate of SVR in patients with chronic hepatitis C will further increase and the burden of portal hypertension-induced complications will decrease over time. Finally, the evidence that genotype 1b infection and higher MELD score increases the risk of EV indicates that tailored, more frequent endoscopic surveillance in this subset of patients may be warranted.
We thank Guido Rainoldi and Felice Cosentino for endoscopic assessment.
- 9Decision-making model for a non-invasive diagnosis of compensated liver cirrhosis. Ital J Gastroenterol 1983; 25: 1-8., , , , , , et al.
- 12Surgery and portal hypertension. In: ChildCG, ed. The Liver and Portal Hypertension. Philadelphia: Saunders; 1964: 50-64., .
- 21Antiviral therapy decreases hepatic venous pressure gradient in patients with chronic hepatitis C and advanced fibrosis. Am J Gastroenterol 2006; 101: 2269-2274., , , , , , et al.Direct Link: