It included both randomized controlled trials (RCTs) and one-arm studies on the sofosbuvir side, but only RCTs on the comparator side. It justified this by claiming that it wanted to reduce the number of hits of its literature search. However, the database for the comparison was different because of this and the comparison itself was therefore unsuitable. A first literature search by IQWiG showed that a number of studies were not considered in the dossier.
One direct comparative study on genotype 2
The situation was different for genotype 2: Here, the manufacturer cited an open-label RCT (FISSION), in which treatment-naive (i.e. non-pretreated) adults with genotype 2 and 3 hepatitis C infection were investigated. In the intervention arm, they received 12 weeks of sofosbuvir plus ribavirin, and in the control arm, they received 24 weeks of peginterferon alfa plus ribavirin.
Hence sofosbuvir was administered in compliance with the approval only for the therapeutic indication genotype 2; and, as a result, only these data were evaluable for the benefit assessment. For genotype 3, the Summary of Product Characteristics (SPC) specifies a treatment duration of 24 weeks. The manufacturer itself did not consider the FISSION study in its dossier for patients with genotype 3.
High risk of biased results
Overall, IQWiG assessed the risk of bias of the FISSION study as high. The main reason was that the manufacturer only included those participants in the analysis who had received at least one dose of the medication they were randomized to. However, particularly in the control arm, where not the new drug, but conventional drugs were administered, some patients refused to have their planned treatment.
This is a general problem of open-label studies, where it is kn
|Contact: Dr. Anna-Sabine Ernst|
Institute for Quality and Efficiency in Health Care