4.1.1 In the first line setting, five published randomised controlled trials (RCT) were identified. One trial compares gemcitabine with a bolus infusion of 5-FU and another with intra-arterial 5-FU (abstract only). In addition, three trials compare gemcitabine to metallomatrix proteinase inhibitors (two trials involve marimastat and one BAY12 9566).
4.1.2 There is thus only one fully reported single-blind RCT, reported by Burris et al, which compares gemcitabine to treatment with 5-FU as a first line treatment used on patients with a Karnofsky score of 50 or more (n=126). In this trial, patients randomised to gemcitabine had better one-year survival (18% vs 2%, p=0.0002), better median survival (5.6 vs 4.4 months, p=0.0025) and improved median progression free survival (2.3 vs 0.9 months, p=0.0002). However, in this trial the comparator, 5-FU, was given by bolus injection, not the usual current means of administration; and the results of this trial may be prone to bias due to lack of blinding of the investigators. Furthermore, the 12-month survival rate of 2% in 5-FU group is unusually low when compared with other published 5-FU studies.
4.1.3 The trial also evaluated the impact of gemcitabine on quality of life in terms of Clinical Benefit Response (CBR) and demonstrated that the administration of gemcitabine led to more clinical benefit responders compared to 5-FU (24% vs 5%, p=0.0022). However, although it has professional support CBR is not a validated tool and its ability to measure the effectiveness of palliative chemotherapy is still investigational.
4.1.4 No relevant RCTs were identified which examine the effect of gemcitabine as a second line treatment in patients with relapsed disease. One non-randomised Phase II trial, involving exclusively individuals with relapsed disease studied the effect of gemcitabine in 74 patients with metastatic pancreatic cancer that had progressed despite the administration of 5-FU. This study demonstrated a median survival of 4 months, one-year survival of 4% and CBR rate of 27%.
4.1.5 In the Burris trial, both gemcitabine and 5-FU were generally well tolerated. However laboratory toxicity was worse for gemcitabine; 26% of patients treated with gemcitabine had Grade 3 or 4 neutropenia compared with 5% in 5-FU group (p<0.01).
4.1.6 The role of gemcitabine in the treatment of pancreatic cancer in comparison with other agents including combination regimens is being investigated in 11 ongoing RCTs and 24 Phase II trials.
4.2.1 Two published economic evaluations of gemcitabine as first line therapy for pancreatic cancer were identified, one of which was only available in abstract form. An economic evaluation of gemcitabine, in both first and second line treatment, was included in the Eli Lilly submission. ScHARR, on behalf of NICE, has undertaken its own analysis.
4.2.2 All of the economic analyses submitted drew on the effectiveness data from the single RCT by Burris et al. For first line treatment the estimates for cost per life year gained ranged from approximately £7,200 to £18,700 dependent on the 5-FU regimen used as comparator. These figures are very sensitive to reduced estimates of survival benefit over comparators.
4.2.3 Cost effectiveness evidence therefore suggests that gemcitabine provides a reasonably cost-effective alternative in the first-line treatment of pancreatic cancer but this conclusion is based on the results from one clinical trial.