Sorafenib Plus Capecitabine Efficacy Assessment in Patients With Advanced Pre-treated Colorectal Cancer (SoMore)

Treatment doses :

Sorafenib 200mg in the morning,400mg in the evening; escalation to 400mg twice daily after 1 cycle Capecitabine 850mg/m2 twice daily Oral Days 1-14, weeks 1-2

Fluoro-Deoxy Glucose Positron Emitting Tomography (FDG-PET)imaging at baseline and after 17-21 days while on therapy.

Sample size justification/statistical analysis

Sample size has been estimated in order to be able to test the null hypothesis that the overall survival rate at 6 months is less than 30%. This hypothesis will be tested using a binomial distribution. The study should be able to reject the null hypothesis, using a 1-sided test with α = 0.025, with a power of 90% in case of a true overall survival ≥ 50% (rate at 6 months). The sample size required is 66 eligible patients (to be followed for 6 months minimum). Analysis will be done on all registered patients using an ITT approach on all eligible patients.

A co-primary endpoint is to compare the overall survival of patients assessed as early PET responders and of patients assessed as early PET non responders (the clinicians will remain blinded for PET response assessment). For this primary analysis, patients who will undergo the second PET assessment will be eligible and time zero for measuring survival will be the date of this second PET examination. It is anticipated that 95% of the patients will be eligible for the analysis with a 50% expected rate of early PET non-responders (result obtained from an unpublished study conducted at Jules-Bordet Institute). With 66 patients registered, we anticipate then that 63 patients will be available for the co-primary endpoint. With 63 patients and our assumption that the HR for the comparison between the survival distributions will be around 0.385 (based on the previously mentioned unpublished study), we will need using a two-sided logrank test at the 2.5% level (2.5% chosen because of the existence of 2 co-primary endpoints), 54 events (power of 90%). With 63 patients and a follow-up after accrual of 1 year, we should reach this number of 54 events. However, to account for another possible 5% drop-out (patient's refusal for undergoing the second PET examination for instance), sample size should be increased to 70 eligible patients.

However, our estimation of 50% expected rate of early PET responders is coming from a prospective unicentric cohort of 38 patients undergoing chemotherapy for 1st line or 2nd line treatment of advanced colorectal cancer with a few of them having received biological agents together with chemotherapy. Our estimation may then not be reliable due to small sample size and different patients population. If this rate of early PET responders proves to be higher, we should be prepared to increase our sample size for targeting the same power of 90%. For instance, if the rate is 67% instead of 50%, the required number of events would be 62 in-stead of 54. If the rate is 75%, the number of events should be increased to 73. The number of patients would have then to be adapted according to the rate of evaluable patients for this PET objective and the rate of patients lost to follow-up. For reassessing the required numbers of events, we did not change our hypothesis of detecting, if true, a hazard ratio of 0.385 in favor of early PET responders.

We plan, during the course of accrual, to assess the rate of patients evaluable for the PET objective, of early PET responders and, if possible the rate of patients lost to follow-up in order to check whether we need to review our planned sample size. However, no interim analysis will be done on the primary endpoints.

Secondary endpoints are to estimate progression-free survival and objective response rate, and to describe the adverse reactions associated with the study regimen in the study population. Also, to determine the correlation of early metabolic response, as assessed by FDG-PET/CT immediately before the first and the second cycles of treatment with the study regimen, with overall survival, progression-free survival, and response rate, and to determine the correlation of growth modulation index (GMI), defined as the time to progression under the study regimen over the time to progression under the latest regimen administered to the patient, with overall survival and progression-free survival.

The study is designed as a single-arm phase II study, with all patients accrued in one stage. No early stopping rules will be used.