f Genova with the use of permuted blocks of variable size. The GP regimen consisted of: gemcitabine 1250 mgm 2 on days 1 and 8 and cisplatin 80 mgm 2 on day 1 every 21 days. The GN regimen consisted of: gemcitabine 1250 mgm 2 on days 1 and 8 and vinorelbine 25 mgm 2 on days 1 and 8 every 21 days. The GIP regimen consisted of: gemcitabine 1000 mgm 2 on days 1 and 8, ifosfamide 2 gm 2 on day 1 and Canertinib CI-1033 cisplatin 80 mgm 2 on day 1 every 21 days. The GIN regimen consisted of: gemcitabine 1000 mgm 2 on days 1 and 8, ifosfamide 3 gm 2 on day 1 and vinorelbine 25 mgm 2 on days 1 and 8 every 21 days. Clinical examination was performed before every cycle and 21 days from the end of therapy. Complete blood cell count was performed on day 1 and between days 12 and 14.
Serum liver and renal functions were measured before each cycle of chemotherapy and at the end of the treatment. Dose reductions of single drugs and delay of each cycle were applied according to standard criteria defined by protocol schedules. Primary prophylaxis with G CSF was not allowed. The treatment was given for a maximum of six cycles unless there were disease progression, unacceptable toxicity or withdrawal of the consent. Statistical analyses The primary end point was OS. Secondary end points included characterisation of toxicities, objective RR and progression free survival. The study was designed to detect a 25% relative reduction of the mortality hazard, in both planned comparisons. We aimed to enrol enough patients to yield the occurrence of 385 deaths, which would give a statistical power of 80% to reject the null hypothesis of no significant difference in the OS time in the two planned comparisons, assuming a hazard ratio of 0.
75, a significance level of a two sided log rank test fixed at 5%, an accrual rate equal to 230 patients per year and a minimum follow up duration of 2 years. No adjustment for multiple comparisons was made. Overall survival was measured from the date of randomisation to the date of death from any cause. Progression free survival was measured from the date of randomisation to the first date of disease progression or of death from any cause. In both OS and PFS analyses, observation times were censored at the limit date of 30 September 2009 for patients in whom no event occurred. Objective response was evaluated according to RECIST criteria.
Response was assessed after three and six courses with a CT scan. The best overall response is the best response recorded from the start of treatment until disease progression. Patients who received at least one dose of chemotherapy were considered evaluable for response, any patient who died early, had early suspension of chemotherapy because of any cause or was not evaluated after randomisation was considered non responder. Toxicity grading, based on NCIC CTC toxicity criteria, was evaluated weekly. All efficacy analyses were based on the intention to treat principle. Safety was analysed on all subjects receiving at least one dose of study drugs, according to treatment actually received. Median period of follow up was calculated for the entire study cohort according to the reverse Kaplan Meier method. Non parametric estimates of the survivor functions and hazardDespite a high curability rate, approximately 25% of pati