A potential new enriching trial design for selecting non-small-cell lung cancer patients with no predictive biomarker for trials based on both histology and early tumor response: further analysis of a thalidomide trial.

Author(s): Lee SM(1), Hackshaw A.

Affiliation(s): Author information: (1)Department of Oncology, University College London (UCL) Hospitals and UCL Cancer Institute, 250 Euston Road, London, NW1 2PG, United Kingdom. sm.lee@uclh.nhs.uk

Publication date & source: 2013, Cancer Med. , 2(3):360-6

There are few predictive biomarkers for antiangiogenic trials in lung cancer. We examine a potential treatment strategy in which a patient group is enriched using both histology and an early assessment of response during standard chemotherapy, and where a new agent is given for the remainder of chemotherapy and as maintenance. We performed a retrospective analysis of 722 stage IIIB/IV non-small-cell lung cancer patients from a double-blind placebo-controlled trial of thalidomide or placebo 100-200 mg/day, combined with gemcitabine/carboplatin (for up to four cycles), then given as single agent maintenance therapy. There was a significant statistical interaction between treatment and histology, with a possible benefit among squamous cell cancer (SCC) patients. We examined 150 SCC patients who were "nonprogressors" (stable disease or complete/partial response) after completing the second chemotherapy cycle. Endpoints were progression-free survival (PFS) and overall survival (OS). Among the 150 patients nonprogressors after cycle 2 (thalidomide, n = 72; placebo, n = 78; baseline characteristics were similar), the hazard ratios (HRs) were: OS = 0.76 (95% CI: 0.54-1.07) and PFS = 0.69 (95% CI: 0.50-0.97). In 57 patients who had a complete/partial response, the HRs were: OS = 0.63 (95% CI: 0.34-1.15) and PFS = 0.50 (95% CI: 0.28-0.88). SCC patients who were nonprogressors after 2 cycles of standard chemotherapy showed evidence of a benefit from thalidomide when taken for the remainder of chemotherapy and as maintenance. This strategy based on histology and, importantly, early assessment of tumor response, as a means of patient enrichment, could be examined in other lung cancer studies. Such an approach might be suitable for trials where there are no predictive biomarkers.