Many patients with chronic hepatitis C have not yet been able to obtain SVR with anti-HCV therapies, and thus become a candidate likely to develop progressive liver disease in the long term, such as cirrhosis or HCC with the possibility of needing liver transplantation [18–20]. Thus, infection in patients with chronic hepatitis C relapsers or non-responders to previous therapy has been an important public health problem and the possibility of an alternative to retreatment has been the focus of clinical investigations.
So, this cohort study was able to assess the effectiveness of retreatment with peginterferon alfa in patients with chronic hepatitis C genotype 2 or 3 in a healthcare environment of the Brazilian public health system. Baseline characteristics such as age, gender, initial body weight, body mass index and baseline ALT and AST, showed no differences from the baseline features presented in several published studies, indicating that the populations for these characteristics were similar.
In our study group of patients non-responders to previous therapy with interferon alfa combined with ribavirin showed an SVR rate of 34.4% when retreated with PEG-IFN alpha-2a or 2b combined with ribavirin for 48 weeks. This result is similar to that described by Sherman et al. who found an SVR rate of 37% in non-responders retreated with PEG-IFN alpha-2a and RBV . In the study of HALT-C Shiffman et al. to assess a sample of 604 patients and all non-responders with advanced fibrosis (METAVIR F3-F4) obtained SVR rate was 18%  as well as the study of Carnicer et al. to assess a sample of 35 patients non-responders, 17% with fibrosis F3-F4 and 45.8% with HCV RNA > 850.000 UI/mL showed an SVR rate of only 27.3% . However, some multicenter studies, sponsored by pharmaceutical companies as Parise et al. with PEG-IFN alpha-2a plus RBV  and Krawitt EL et al. with PEG-IFN alpha-2b plus RBV , has demonstrated higher values in the SVR rates in patients non-responders to prior therapy of 46% and 57%, respectively.
For the group of relapsed patients, in our study we found an SVR rate of 50%. This result is confirmed favorably to the data presented in studies by Mathew et al. who found an SVR rate of 50% of patients relapsed when retreated with PEG-IFN alpha-2b plus RBV  as well as the study of Sherman et al. who found an SVR rate of 51% for relapsed patients, genotype 2 and 3 when retreated with PEG-IFN alpha-2a plus RBV . Meanwhile, other authors have presented their studies in SVR rates for patients relapsed to prior therapy, superior to that found in our study, as presented by multicenter studies by Parise et al. PEG-IFN alpha-2a plus RBV  and Krawitt E.L et al. PEG-IFN alpha-2b plus RBV , SVR rates of 70% and 59%, respectively. Other studies also showed a higher SVR rate than our results, as well as the one presented by Basso et al. to portray patients with recurrent PEG-IFN alpha-2b plus RBV where the SVR rate achieved was 78.6%  and Gonçalves Jr et al. who found an SVR rate of 62%, to portray with PEG-IFN alpha-2b plus RBV a sample where only 26% of patients had liver diagnostics F3-F4 and 30% had baseline viral load HCV RNA > 800.000 UI/mL .
According to Mitchell L. Shiffman clinical and virological factors may be useful to predict the likelihood of response to retreatment such as: type of response to previous treatment (non-responders or relapsers), race, type of viral genotype, liver disease severity, current alcohol consumption in which higher response rates can be obtained when patients are carefully selected. Therefore, we concluded that retreatment with PEG-IFN and RBV should be reserved for patients relapsed to prior therapy, which had been done with IFN-α monotherapy and in patients with viral genotype 2 and 3 .
Based on these previous findings, current studies of retreatment have evaluated the influence of factors predictive of SVR rates. In our study, among the evaluated parameter settings (type of response to previous treatment, age, gender, BMI, ALT, liver fibrosis and serum HCV RNA), only the type of response to previous treatment, age, the initial level of ALT and the degree of hepatic fibrosis, showed statistical significance in achieving SVR.
In the evaluation of factors independently associated with SVR rate was applied Poisson regression analysis and were identified as predictors of SVR: age, be relapsed to previous treatment and present degree of liver fibrosis F0-F2, indicating that these patients are more likely to have SVR. For the analysis of average initial pretreatment ALT variable was not statistically significant, indicating that this parameter in this study did not contribute to increase the rate of SVR.
These findings resemble those reported by Shiffman et al. that showed that in their multivariate regression analysis the following factors: previous treatment with IFN-α monotherapy, virus C genotype 2 and 3, a serum HCV RNA less than 1.5 million IU/mL, an AST: ALT ratio less than 1.0 and absence cirrhosis on liver biopsy as associated with an increased probability of achieving SVR .
Our results also are similar to those presented Sherman et al. Direct comparisons are difficult to achieve, since, Sherman et al. stratified the results among the group of relapsed patients and non-responders to initial treatment to assess predictors of SVR: in the group of relapsed patients, the predictors of SVR were HCV genotype 2 and 3, the Caucasian race and a low initial HCV viral load, and in the patients non-responders, were gender, body weight, BMI, degree of hepatic fibrosis and a low initial HCV viral load as predictors of SVR .
No association was found between parameters BMI (< or ≥ 30 Kg/m2) and baseline viral load of HCV RNA (< or ≥ 800.000 IU/mL) at the rate of SVR. This lack of association probably could not be established in our study because the sample was lost for the two parameters of 33.8% and 28.2%, respectively. Loss of data collection for these parameters represented a failure in data collection of the study, not allowing the establishment of relationship with SVR. The study of Sherman et al. identified as a predictor of SVR low initial load HCV RNA in both groups of patients as in non-responders relapsed; whereas for this parameter only BMI was identified as a predictor of response in patients non-responders . As the study of Sherman et al. and HALT-C study of Shiffman et al. also found that an initial viral load less than 1.500.000 IU/mL is associated with an increased probability of SVR .
Our study showed that patients with less advanced age are more likely to respond to retreatment regimen, since the patients with SVR had a lower mean age (51.3 ± 9.4 years) compared with patients who had failures in SVR with significantly higher mean age (55.2 ± 8.2 years). This result was also observed by Krawitt et al. to identify non-responders that patients aged 40 years or younger have a higher SVR rate, but for the group of relapsed patients, this significance was not found .
It is important to note the poor liver profile of these patients’ with high percentage (76% presented F3/F4) with results in liver biopsy can justify the number of interruptions of treatment and deaths from decompensated cirrhosis (ascites and encephalopathy) found in our study. Some effectiveness studies in the literature revealed different percentages described in cirrhotic patients, and international studies with a smaller percentage as observed by Carnicer et al.  and Sherman et al.  where 20% and 26%, respectively of the samples were composed of patients with cirrhosis. However, in studies in Brazil are found, generally higher percentages of patients with cirrhosis as shown by the studies of Gonçalves et al.  and Parise et al. 40% and 33% respectively of their samples composed of patients with cirrhosis .
The percentage of discontinuation due to adverse events (18.5%) in our study was slightly different from controlled clinical studies (5-14%)  but similar to studies of effectiveness [21, 25, 27]. This difference in profile is justified by the presence of comorbidities in this population when are conducted effectiveness studies that portray the reality of drug use, without prior selection of the sample. Almost a third of patients eligible to participate in clinical trials is excluded from the criteria established for inclusion and exclusion. Thus, patients with decompensated cirrhosis, hepatitis B, HIV, kidney disease, neuropsychiatric, coronary, cerebrovascular, or hematologic diseases are usually excluded from controlled clinical studies, favoring the non-appearance of adverse events associated with these comorbidities .
Eight patients died during the retreatment, representing 20% of interruptions due to adverse events. The causal relationship between death and retreatment has not been established. However, it is remarkable that in controlled clinical studies of effectiveness investigated there are no reports of death during the retreatment [21–32]. The average age of patients who died was 57.3 years and all had degree of liver fibrosis by F4 rating Metavir, characterizing the presence of cirrhosis in the whole population. These criteria may have compromised the continuity of treatment, as well as contributed to the evolutions for the deaths, but since these criteria are included in the general criteria for inclusion of patients candidates for retreatment with alfapeguinterferona + ribavirin in accordance with Ordinance No. 34 of 28 September 2007, the Ministry of Health of Brazil, we could not exclude these patients the possibility for retreatment. For this reason, studies like this, which represent the description of a population treated in real life, without prior sample selection, and was able to demonstrate in their analyzes the importance of establishing predictors of response, should be considered and made public in order to prevent individuals who are at risk of a treatment which will not benefit them, producing individual damage and to the public health system.