Developing a biomarker-stratified trial design in advanced colorectal cancer: the MRC FOCUS 3 feasibility study

It is likely that future trials for many solid tumours will be stratified by predictive/prognostic biomarkers in order to a) derive more homogeneous patient cohorts; b) "personalize" treatment interventions in relation to tumour (or host) phenotype; and c) improve the likelihood of discovering a beneficial effect, in at least a subset of patients with the disease. In some settings there are likely to be multiple relevant biomarkers and treatments to be tested. For logistical reasons it would be efficient to combine these into single trials. Moreover, the inclusion of overlapping or common treatment arms could serve to further validate the use of a given biomarker for selection and support retrospective testing of other biomarker selection algorithms that may arise during the course of a trial’s enrolment period. These requirements present significant challenges both for creating efficient designs and for making these comprehensible for prospective subjects. Methods: FOCUS 3 is a feasibility study designed to address these issues and test a novel design for examining the predictive role of biomarkers (KRAS/BRAF mutations and topo-1 expression) for 1st-line therapy of aCRC. It employs an unusual design in which the two biomarkers defined four possible cohorts. Specific treatment questions were relevant to each marker and, because each cohort was defined by 2 markers, there were 2 relevant experimental arms for each cohort, plus 1 common control arm for all cohorts (FOLFIRI). The study thus incorporates 5 treatment regimens, but only 3 possibilities for each patient, defined by marker phenotype. We will complete recruitment of 240 patients by April 2011. Paraffin blocks from 24 sites were obtained for central biomarker analysis and aim to feedback results within 10 working days. A 4 stage suite of patient information sheets (PIS) was designed to avoid patient overload and we are assessing patients’ comprehension in a qualitative study. Conclusion: Patient samples can be collected and analysed centrally within acceptable time constraints. Complex, stratified, multi arm designs can be made acceptable to patients through good PIS design, ensured by patient and carer input into their design