Impact of pegylated interferon and ribavirin on morbidity and mortality in patients with chronic hepatitis C and normal aminotransferases in France

Authors


  • Potential conflict of interest: S.D.-B. received an unrestricted grant from Roche to conduct this study. G.B. is a Roche employee. I.L.-D. is a former Roche employee and is now a Janssen-Cilag employee. P.D. has no conflicts of interest. V.C.-D. has no conflicts of interest. S.D. is a consultant for Roche. A.L. has no conflicts of interest. F.R.-T. is an investigator for Roche and Schering-Plough. P.M. was a paid speaker at symposiums held by Roche, Schering-Plough, Gilead Sciences, and Bristol-Myers Squibb. He is an investigator for Roche, Schering-Plough, Bristol-Myers Squibb, Gilead Sciences, Vertex Pharmaceuticals, and Bayer Healthcare. He is a member of the French Boards of Experts in Hepatology for Roche, Schering-Plough, Gilead Sciences, Bayer Healthcare, and Bristol-Myers Squibb. He is a consultant for Gilead Sciences and Vertex Pharmaceuticals.

Abstract

Clinicians continue to raise questions concerning the necessity of treating chronic hepatitis C virus (HCV)-infected patients with normal alanine aminotransferase (N-ALT), in light of their slower progression to cirrhosis than patients with elevated alanine aminotraferase (E-ALT). This study was undertaken to predict the impact of pegylated interferon (IFN) and ribavirin on HCV-related morbidity and mortality in patients with N-ALT. A previous Markov model was adapted to separately simulate patients with N-ALT (30%) and those with E-ALT (70%). The model estimates fibrosis progression rates according to age, sex, and whether ALT levels are normal or elevated, assuming that patients with E-ALT have a 2.6 times higher progression than those with N-ALT. It takes into account improvement in HCV screening and treatment and competitive mortality. We assumed that N-ALT patients were treated 80% less frequently between 2002 and 2004 and 70% less frequently from 2005 on, as obtained in real life from three multicentric cohorts (Hepatys, Adequation, Persee). Antiviral treatment of HCV-infected populations might reduce 2008-2025 HCV-related morbidity and mortality by 34,200 cases of cirrhosis (36%, 33,000-35,000), 22,400 complications (28%, 21,000-23,000) and 17,500 deaths (25%, 17,000-18,000), including 3000 cases of cirrhosis (22%, 2000-5000), 1200 complications (15%, 1000-1700), and 1000 deaths (14%, 900-1300) in the N-ALT population, despite a probability of receiving treatment that is three to five times less in this population. If N-ALT patients are treated at the same proportions as those with E-ALT, morbidity and mortality could be further reduced by 1400 cases of cirrhosis (13%, 1200-2200), 600 complications (9%, 600-1000), and 500 deaths (9%, 500-800). Conclusion: Treatment of N-ALT patients would decrease HCV morbidity and mortality. These patients should be considered candidates for treatment just as others are. (HEPATOLOGY 2009.)

Hepatitis C virus (HCV) infection is generally associated with a rise in transaminases, especially alanine aminotransferase (ALT). However, an average of 30% of screened HCV patients have ALT activity within the normal range.1–3 Longitudinal follow-up of ALT evolution showed that 20% to 30% of them had transient elevation of ALT that did not appear to affect disease progression. Data4–7 showing lower disease progression in patients with normal ALT (N-ALT) compared with those with elevated ALT (E-ALT) may cause confusion in terms of expected morbidity and mortality related to HCV in N-ALT patients. For example, the international consensus conference in 19998 recommended not treating such patients, mainly because of only minor liver injury. This implied that HCV-related morbidity and mortality were nonexistent in this subgroup of patients. However, the presence of hepatic lesions in some patients, with a negative impact on quality of life,9–12 and the fact that slower fibrosis progression does not prevent liver mortality,13 led to reconsidering indications for treating these patients. In addition, the improved efficacy of a pegylated interferon (IFN) combination14 with a similar sustained virological response (SVR) in patients with N-ALT to that in patients with E-ALT,15 led, in 2005, to extending pegylated IFN bitherapy to French patients with N-ALT. The impact of antiviral therapy on long-term HCV-related morbidity and mortality for these patients remains subject to debate, but it can be assessed at a population level by using mathematical modeling.16

The objectives of this study were to evaluate the impact of pegylated bitherapy on HCV morbidity (cirrhosis and its complications) and mortality in patients with chronic hepatitis C and N-ALT. In a second step, the model enabled estimating treatment impact in terms of lives saved according to an alternative scenario.

Abbreviations

ALT, alanine aminotransferase; E-ALT, elevated alanine aminotransferase; HCV, hepatitis C virus; IFN, interferon; N-ALT, normal alanine aminotransferase; SVR, sustained virological response.

Patients and Methods

Model Structure.

We used a previously published Markov model to simulate progression of newly HCV-infected cohorts,16 from onset of HCV infection to death (Fig. 1), except that HCV patients were divided into N-ALT and E-ALT. Briefly, 75% of acutely infected individuals entered the first stage of chronic hepatitis (F0) during the first year after infection.17 The annual probabilities of fibrosis progression were estimated according to sex, age, and ALT level (normal or elevated), assuming that patients with E-ALT have a 2.6 times higher progression than those with N-ALT.7 Once individuals had cirrhosis (F4), they were at risk of liver failure depending on the ALT level18 and at risk of HCC according to age and sex.19 Finally, HCV-related mortality was defined as death of liver failure, taking into account complications of cirrhosis such as ascites, variceal bleeding, and encephalopathy,20 and as death from HCC assuming period-dependent and age-dependent probabilities.21 Competitive mortality was derived from French life tables.22

Figure 1.

Natural history model of HCV disease.16 PCHC is the probability of chronic hepatitis progression 1 year after infection and is set at 75%.17Ps,a,i is the annual probability of fibrosis progression according to sex, age, and ALT level, to be estimated assuming that patients with elevated ALT (E-ALT) have a 2.6 times higher progression than those with normal ALT (N-ALT)7; PLF is the annual probability of progression from cirrhosis to liver failure depending on the ALT level18: 4% for N-ALT and 10% for E-ALT; PHCC is the annual probability of progression from cirrhosis to hepatocellular carcinoma depending on age and sex: 1.7% for females and 3.6% for males at 57 years of age, with an age effect of 1.05 by year of age19; PDLF is the annual probability of death of liver failure, 39% during the first year and 11% thereafter20; and PDHCC is the annual probability of death from HCC depending on age and period: at 20 to 39 years of age, 70.2% during the first year and 23.1% thereafter; at 40 to 59 years of age, between 75.2% and 85.7% during the first year and between 26.9% and 38.1% thereafter; at 60 years of age or more, between 79.1% and 89.2% during the first year and between 31.9% and 40.0% thereafter.21 Death rates from causes other than HCV are assumed to be equal to those of the general population (competitive mortality22).

Data.

We used past incidence of HCV infections as estimated by our previous model.16 Estimates were based on two assumptions: that past HCV infections occurred according to a logistic function until 1989, and that they decreased in the same proportions as those observed in the United States thereafter (93% decrease between 1989 and 2005).23

Age is a determining factor in the natural history of HCV; it is thus essential to take into account the age distribution of subjects at the time of their contamination with HCV. This distribution was obtained from French cohorts according to sex, transmission mode, and genotype,24, 25 assuming that 30% of subjects were N-ALT,26 60% of these were women, and 70% were E-ALT, 40% of whom were women,4, 6, 7 independently of age or genotype. These data permitted characterizing cohorts of infected patients by sex, age, genotype, and ALT level (Table 1).

Table 1. Age Distribution (%) of Patients at Time of Infection According to Sex, Genotype, and ALT Level
AgeMenWomen
G1/4G2/3G1/4G2/3
ALT NALT EALT NALT EALT NALT EALT NALT E
HCV Infections Before 1990
0–90.20.60.10.20.10.10.00.0
10–191.55.20.72.61.62.50.81.2
20–293.311.51.65.64.46.91.93.0
30–391.13.90.51.72.94.51.01.6
40–491.03.60.41.42.23.40.91.3
50–590.82.70.31.01.42.10.50.8
60–690.31.20.10.40.30.40.10.1
70–790.10.30.00.10.00.00.00.0
Total8.329.03.713.012.919.95.28.0
HCV Infections Since 1990
0–90.10.30.00.20.00.00.00.0
10–191.96.51.13.82.64.11.52.4
20–293.913.52.37.95.28.03.04.7
30–390.83.00.51.70.91.40.50.8
40–490.51.80.31.11.72.71.01.6
50–590.30.90.20.50.81.30.50.7
60–690.10.30.10.20.10.10.00.1
70–790.00.20.00.10.00.00.00.0
Total7.626.54.515.511.317.66.510.3

As previously done, antiviral treatment effects were incorporated by estimating the likelihood of being screened for HCV, of being treated, and of reaching SVR after treatment according to the HCV genotype for naive and previously treated patients. We extrapolated annual likelihood of being screened for HCV from the two French prevalence studies, which estimated that 24% and 56% of the HCV-infected population were aware of their infection in 1994 and 2004.26, 27 We fit the annual likelihood of treatment among patients aware of their infection to 2002-2005 data obtained from the Groupement pour l'Elaboration et la Réalisation de Statistiques, a nonprofit economic group in charge of monthly collection of all units of drugs (including peg-IFNs) sold in France. The main assumptions for treatment were: (1) patients eligible for treatment were those screened, aged 18 to 70 years, and in F0 to F4; (2) the likelihood of treatment was identical for genotypes 1 and 4 and for genotypes 2 and 3 with IFN (IFN) monotherapy (period 1991-1998) because, during that period, genotyping was not reimbursed in France and therefore had not been performed before treatment except for research purposes. Conversely, since 1999, genotyping has been performed in routine practice for assessment of patients with HCV infection. Based on lower SVR rates observed in genotypes 1 and 4 with the recommended 48-week treatment duration, we therefore assumed that the likelihood of treatment was lower for genotypes 1 and 4 than for genotypes 2 and 3 with standard combination therapy (1999-2001) or with pegylated bitherapy (from 2002 on); (3) for patients with N-ALT, the likelihood of treatment was null before 2002, 80% lower in 2002-2004, and 70% lower from 2005 on than for patients with E-ALT, and the likelihood of retreatment was null. Indeed, in 1999, the European Association for the Study of the Liver International Consensus Conference on Hepatitis C8 did not recommend treatment for patients with N-ALT. In 2002, the French consensus conference recommended that patients with chronic HCV infection associated with repeatedly normal aminotransferase levels not be excluded from treatment.28 After this consensus conference, French clinicians proposed treatment for some N-ALT patients. After a randomized controlled trial of pegylated IFN with ribavirin in HCV patients with N-ALT,4 marketing authorization was extended to all patients with N-ALT in 2005; (4) the likelihood of treatment and retreatment for patients with fibrosis stage lower than F2 was 80% lower than for patients with fibrosis stage F2 or higher.29 Although the likelihood of treatment was not directly estimated based on age and sex, those two variables indirectly influence the probability of treatment. Indeed, the probability of treatment depends on the stage of fibrosis, which is affected by age and sex; (5) patients achieving SVR were withdrawn from the number of patients at the different stages, except for patients with cirrhosis (F4), who remained at risk of developing complications at the same rate as if they were not treated. The latter assumption was considered to be the most conservative possible so as not to overestimate treatment impact; (6) E-ALT relapsers after IFN monotherapy or standard bitherapy might be re-treated once with pegylated bitherapy (available since 2002); (7) E-ALT relapsers after pegylated bitherapy might be retreated once 3 years later. The likelihood of reaching SVR after treatment was obtained from the literature.30–33

The model was fitted to the reported 400,000 HCV prevalence in 200427 and to age-specific annual HCC deaths22 related to HCV.34 As a model validation, we compared the distribution of the modeled population with that obtained in real-life data from Hepatys, a French multicentric observatory of 2101 HCV patients,29 Adequation, a French multicentric observatory of 784 HCV patients,35 and Persee, a French multicentric observatory of 2306 patients.36

Procedure.

After estimation of annual probabilities of fibrosis progression, we simulated HCV progression up until 2025 for all HCV infections occurring up until 2005, assuming that current treatment practices would be continued up to 2025. We then assessed an alternative scenario assuming that patients with N-ALT were treated at the same proportions as patients with E-ALT from 2005 on.

Sensitivity Analysis.

We evaluated the implications of alternative assumptions by performing six sensitivity analyses of a drop in new infections, competitive mortality, percentages of spontaneous HCV clearance, evolution of the percentage of patients aware of having HCV, proportion of patients with N-ALT, and progression to complications of cirrhosis for patients with SVR.

Results

Baseline Analysis.

The optimization process led to good adjustment between observed deaths by HCC related to HCV and those estimated by the model. For example, the number of observed HCV-related HCC deaths in 2001 was 899,34 whereas the model prediction was 922. The median relative difference between estimated and observed HCV-related HCC deaths was 7% (interquartile range, 3%-16%).

The model estimated the probability of fibrosis progression according to current age, sex, and ALT level (Table 2). These increased according to age and were estimated, among N-ALT, between 0.028 and 0.062 for women and between 0.030 and 0.285 for men, and among E-ALT, between 0.07 and 0.153 for women and between 0.077 and 0.582 for men.

Table 2. Estimates of Annual Probability of Fibrosis Progression (95% CI) According to Current Age, Sex, and ALT Level
Age group*WomenMen
ALT NALT EALT NALT E
  • *

    Current age.

≤40 years0.0276 (0.0255–0.0298)0.0703 (0.0650–0.756)0.0304 (0.0299–0.0309)0.0772 (0.0760–0.0784)
41–50 years0.0231 (0.0198–0.0265)0.0591 (0.0508–0.0673)0.0372 (0.0362–0.0381)0.0937 (0.0916–0.0959)
51–60 years0.0356 (0.0334–0.0377)0.0898 (0.0846–0.0950)0.0885 (0.0873–0.0896)0.2140 (0.2114–0.2166)
61–70 years0.0446 (0.0432–0.0460)0.1118 (0.1086–0.1150)0.1328 (0.1307–0.1350)0.3096 (0.3052–0.3139)
>70 years0.0621 (0.0608–0.0633)0.1534 (0.1507–0.1562)0.2848 (0.2789–0.2908)0.5817 (0.5728–0.5906)

At each time point, the model estimated the number of patients at each stage of the disease. We confronted model estimates of the distribution of treated patients according to fibrosis stage and genotype in 2004 with data obtained in real life from a multicentric observatory, Hepatys.29 The model accurately reproduces the distribution of treated patients in 2004: in genotypes 1 and 4, 29% in F0 to F1 (versus 31% in HEPATYS), 35% in F2 (versus 33% in HEPATYS), and 36% in F3 to F4 (versus 36% in HEPATYS); in genotypes 2 and 3, 39% in F0 to F1 (versus 38% in HEPATYS), 32% in F2 (versus 28% in HEPATYS), and 28% in F3 to F4 (versus 34% in HEPATYS). We also compared the distribution of treated patients in 2006 in our model with that of two other multicentric observatories, Adequation35 and Persee.36 We found 13.4% of patients with N-ALT compared with 12.7% in the Adequation observatory and 13.7% in the Persee observatory; 36% of patients with genotypes 2 and 3 (versus 38% in the Adequation observatory) and 67% of patients with fibrosis of at least F2 (versus 71% in the Adequation observatory).

In 2008, the model estimated that 386,000 patients were HCV positive, among whom 152,000 were HCV-RNA negative, whereas 234,000 were HCV-RNA positive. Among HCV-RNA–negative patients, 68% resolved spontaneously, and 32% cleared after antiviral therapy. Taking the whole population of HCV-positive patients, the proportion of cleared patients after antiviral therapy was 13%. Among HCV-RNA–positive patients, 41% were unaware of their infection, 40% were aware of their HCV status but had never been treated, and 19% had been treated and were nonresponders (9% once and 10% twice). Thus, the proportion of treated HCV infections was 24% with respect to the entire population of HCV-positive patients, but 33% for HCV patients who were candidates for antiviral therapy (after exclusion of patients with spontaneous resolution).

Among chronic hepatitis C patients, more then 50% had minimal fibrosis (F0-F1), and 17% were F4 (11% without and 6% with complications). The distribution in F4 according to ALT level led to 4% without complications and 1% with complications among N-ALT, and 16% and 9%, respectively, among E-ALT (P < 0.001).

Figure 2 illustrates the proportion of treated patients from the start of antiviral therapy (1991 for patients with E-ALT and 2002 for patients with N-ALT) to the end of simulations (2025), that is, the ratio of cumulative annual number of treated patients by the number of candidates for antiviral therapy after exclusion of those with spontaneous clearance. This estimated proportion varies from 0% in 2002 to 38% in 2025 in patients with N-ALT (Fig. 2A, baseline analysis) and from 0% in 1991 to 71% in 2025 in patients with E-ALT (Fig. 2B).

Figure 2.

Evolution of the proportions of treated patients over time: (A) in the N-ALT population according to baseline and alternative scenario; and (B) in the E-ALT population. Each year, the proportion of treated patients is the ratio of cumulative annual number of treated patients since the beginning (2002 for N-ALT and 1991 for E-ALT) to the number of candidates to treatment after exclusion of those with spontaneous clearance.

Figure 3 illustrates the specific impact of treatment on mortality over time. For 2008 to 2025, the model predicts that 1000 (900-1300) deaths would be avoided by antiviral therapy in the N-ALT population, a 14% reduction over the whole period. In comparison, 16,700 (16,000-17,000) deaths would be avoided by antiviral therapy in the E-ALT population, a decrease of 26% over the whole period.

Figure 3.

Incidence of HCV-related deaths in elevated ALT (upper curves) and normal ALT populations (lower curves) over time. Specific impact of treatment on 2008 to 2025 HCV-related mortality given in relative reduction of the expected number of deaths in the scenario without treatment (in each subpopulation) over the period 2008 to 2025.

Figure 4 illustrates the specific impact of treatment on morbidity in the N-ALT population over time. For 2008-2025, the model predicts that 3000 (2000-5000) cases of cirrhosis and 1200 (1000-2000) complications from cirrhosis would be avoided by antiviral therapy in the N-ALT population, a 22% reduction and a 16% reduction, respectively, over the whole period. In comparison, 31,300 (29,000-32,000) cases of cirrhosis and 21,300 (20,000-22,000) complications from cirrhosis would be avoided in the E-ALT population, a 39% reduction and a 30% reduction, respectively, over the whole period (results not shown).

Figure 4.

Incidence of HCV-related cirrhosis (upper curves) and complications of cirrhosis (lower curves) in the normal ALT population over time. Specific impact of treatment on 2008 to 2025 HCV-related morbidity given in relative reduction of the expected number of events in the scenario without treatment (in the N-ALT subpopulation) over the period 2008 to 2025.

Alternative Scenario.

Figure 2A illustrates the change in the proportion of treated patients with N-ALT over time if we assume that patients with N-ALT were treated at the same proportions as patients with E-ALT starting from 2005. In this scenario, HCV-related morbidity would be reduced compared with baseline analysis in light of current treatment practices, by 1400 (1200-2200) cases of cirrhosis (−13%) and 600 (600-1000) complications from cirrhosis (−9%) in the N-ALT population. HCV-related mortality would be reduced by a further 500 (500-800) deaths (−9%) in the N-ALT population (results not shown).

Sensitivity Analysis.

Sensitivity analyses were performed on six model parameters and assumptions (Tables 3 and 4). These tables present model results obtained via baseline analysis and during sensitivity analyses (in lines) considering two scenarios, one without treatment and one with treatment (in columns). Treatment impact was calculated between these two scenarios in absolute and relative (%) numbers (last two columns). Table 3 presents sensitivity analyses for mortality in global and N-ALT populations. Table 4 presents sensitivity analyses for cirrhosis in an N-ALT population only.

Table 3. Sensitivity Analysis of Baseline Scenario for HCV-Related Mortality
 Scenario without TreatmentScenario with TreatmentTreatment Impact
(A) General Population    
Baseline analysis71,30053,80017,500−25%
Sensitivity analyses    
 HCV incidence plateau at 4000 new infections/year72,00054,40017,600−24%
 Higher competitive mortality69,80052,80017,000−24%
Progression to chronic hepatitis C    
 65%66,60049,90016,700−25%
 85%73,20055,50017,700−24%
Screening evolution    
 Progression twice as slow71,20054,80016,400−23%
 Plateau at 56%71,30056,30015,000−21%
% N-ALT patients    
 20%74,10056,00018,100−24%
 40%68,10051,40016,700−25%
Lower progression to complications of cirrhosis in F4 patients with SVR71,50050,40021,100−30%
(B) N-ALT population    
Baseline analysis700060001000−14%
Sensitivity analyses    
 HCV incidence plateau at 4000 new infections/year710061001000−14%
 Higher competitive mortality760065001100−14%
Progression to chronic hepatitis C
 65%770065001200−15%
 85%65005700800−13%
Screening evolution    
 Progression twice as slow70006100900−13%
 Plateau at 56%70006100900−13%
% N-ALT patients    
 20%39003400500−13%
 40%10,70091001600−15%
Lower progression to complications of cirrhosis in F4 patients with SVR710058001300−18%
Table 4. Sensitivity Analysis of Baseline Scenario for HCV-Related Cirrhosis Expected in 2008-2025: N-ALT Population
 Scenario without TreatmentScenario with TreatmentTreatment Impact
Baseline analysis13,60010,6003000−22%
Sensitivity analyses    
 HCV incidence plateau at 4000 new infections/year13,70010,8002900−21%
 Higher competitive mortality14,60011,3003300−23%
Progression to chronic hepatitis C    
 65%14,20010,7003500−25%
 85%12,90010,3002600−20%
Screening evolution    
 Progression twice as slow13,60010,80028000−21%
 Plateau at 56%13,60011,000260−19%
% N-ALT patients    
 20%780063001500−19%
 40%20,20015,6004600−23%
Lower progression to complications of cirrhosis in F4 patients with SVR13,60010,6003000−22%

Treatment impact on mortality in the overall HCV population varied between 15,000 and 21,100 deaths avoided (21%-30% decrease compared with the scenario without treatment), in which 500 to 1600 deaths would be avoided in an N-ALT population (13%-18% decrease) (Table 3). Treatment impact on morbidity in the total HCV population varied between 27,300 and 35,500 cirrhoses avoided (29%-38% decrease compared with the scenario without treatment) and between 18,800 and 25,800 complications from cirrhosis avoided (24%-33% decrease) (results not shown). Treatment impact on morbidity in the N-ALT population varied between 1500 and 4600 cases of cirrhosis avoided (19%-25% decrease compared with scenario without treatment) (Table 4) and between 600 and 1900 complications from cirrhosis avoided (14%-18% decrease) (results not shown).

Discussion

The challenge for clinicians when deciding whether to treat HCV patients lies in the fact that therapy is intended to prevent risks of morbidity and mortality that may occur several decades later. This challenge is even more clear-cut for N-ALT patients with known slow disease progression, and clinicians still question the benefits of antiviral therapy in this subgroup of patients. Consequently, the probability of being treated is five to three times lower for N-ALT than for E-ALT HCV patients. However, it is now accepted that advanced hepatic lesions occur in N-ALT patients.37 To assess the benefit of treating N-ALT patients and to help in clinician decision-making, we used a modeling approach evaluating treatment impact on HCV-related morbidity and mortality. We show for the first time that treating N-ALT populations will have an impact on HCV-related morbidity and mortality despite a lower probability of treatment: 3000 cases of cirrhosis, 1200 complications from cirrhosis, and 1000 deaths would be avoided from 2008-2025. These results add to the rationale for treating N-ALT patients, although the long-term benefit is lower than that predicted for E-ALT patients.

Several years ago, Davis et al.38 examined the impact of treating HCV-infected patients with N-ALT in the United States. Their results may appear to differ from ours, but in fact were reported in a different manner. In our study, the reduction in morbidity and mortality in the N-ALT population is given in relative reduction of the expected number of events in an N-ALT population compared with the number of events in an N-ALT population without treatment. In contrast, in the Davis et al. study, the reduction in morbidity and mortality in the N-ALT population is given in relative reduction of the expected number of events in an N-ALT population compared with the number of events in the entire population without treatment. As an example, in 2020, Davis predicted a 3.5% treatment-related reduction in cirrhosis complications. In 2020, our model predicts that 4500 complications from cirrhosis would occur in the overall population (among which 500 would occur in the N-ALT population) without treatment, and that 100 complications of cirrhosis would be prevented in the N-ALT population by treatment. Using our presentation, treatment impact is 20%. Using the same presentation as that of Davis, the treatment impact is 2.2%, which is close to the 3.5% of Davis. The difference between 3.5% and 2.2% is at least partly explained by the fact that Davis' results are based on a scenario treating the entire HCV population in 2000. Such a scenario is subject to debate, because it does not adhere to real-life medical practice. Indeed, Davis' presentation of results combined impact of treatment in an ALT population with part of this subpopulation in overall expected events. Therefore, we believe that that presentation is not adapted to estimating treatment impact in an N-ALT population according to its specific natural history.

It appears that France exhibits the highest treatment rate across Europe,39 which is, for example, 1.5 times higher than in Germany, 2.7 times higher than in Spain, 4.6 times higher than in Italy, and 5.3 times higher than in the United Kingdom, and probably drastically higher than in the United States. This is the result of multiple screening campaigns and national plans to fight against hepatitis B and C, followed by effective case detection (detection has more than doubled in 10 years). Results from the current study would appear to be quite accurate, because they were validated by real-life data from large cohorts ranging from 784 to 2101 HCV patients treated with pegylated IFN alpha 2a or alpha 2b and ribavirin. In addition, we varied baseline parameter values and assumptions so as to evaluate the impact of data uncertainties on our results. As shown during sensitivity analyses, although substantial differences in absolute estimates of treatment impact were observed in the N-ALT population, relative estimates did not greatly vary. The annual probability of being treated in France, as estimated in our study, was based on clinicians' decisions that include contraindications and side effect profiles. Unfortunately, we have no data on the contribution of these two barriers to treatment. Therefore, we cannot perform sensitivity analysis of these two parameters.

The question persists as to whether the decision to treat should remain based at least in part on abnormality in ALT. To address this issue, the scenario we used from 2005 on assumed that patients with N-ALT were treated at the same proportions as E-ALT patients. It revealed a positive impact of treatment on N-ALT patients, resulting in an additional 12% reduction in morbidity and a 9% reduction in mortality from 2008 to 2025. Moreover, in current practice, HCV-related morbidity and mortality in N-ALT patients would continue to rise, whereas they would stabilize or even decrease if this population were to be treated at the same proportions as the E-ALT population.

One limitation of the current study lies in the definition of the N-ALT population. Prospective follow-ups showed that transient ALT elevation occurs in approximately 20% to 30% of patients independently of the definition of the N-ALT population. Our model did not differentiate between patients with persistent N-ALT and those with transient ALT elevation because the latter is not associated with a significant modification in disease progression37 and therefore should not impact on model predictions.

In summary, the current study shows that antiviral therapy will decrease morbidity and mortality in N-ALT HCV patients. It may help experts in the elaboration of therapeutic recommendations, taking into account the impact of viral eradication on HCV-related morbidity and mortality. HCV patients with normal ALT should be considered as candidates for treatment in the same way as others are.

Ancillary