A pooled analysis of published, basket trials in cancer medicine

Background: There is widespread interest in cancer basket trials. However, to date, there has been no formal analysis of all published basket trials. Methods: We performed a systematic review to identify all published basket trials in cancer medicine. We included studies that (1) did not place restriction on tumour type and (2) used a biomarker for the basis of enrolment (3) administered at least one anti-cancer agent in prospective fashion and (4) reported responses by each tumour type included. We sought information regarding the tumour histology included and the response rate in basket trials. In addition, we used national cancer statistics to identify which tumour types have been overrepresented in basket studies (i.e. more representation in trials than their incidence would suggest) and those which are under-represented. Results: We identified eight articles with a combined enrolment of 1176 patients were included in our analysis, representing over 33 tumour types. Ovarian and fallopian tube cancers 221/1176 (19%), colorectal cancer 144/1176 (12%) and sarcoma 129 (11%) were the most common tumours represented, whereas renal cell cancer, seminoma, thymic carcinoma and neuroendocrine tumour and appendiceal carcinoma were the least represented with one case each. The overall response rate was 25%. Common cancers may be underrepresented compared with rarer tumour types (linear regression beta = 0.58, 95% confidence interval = −0.037–1.21) (slope < 1 implies under-representation, >1 over-representation). Conclusions: We found that, to date, over 1100 patients have been enrolled on published basket studies. Common cancers may be underrepresented compared with rarer tumours. The overall response rate was 25%. Patients enrolling on basket trials should be counselled appropriately that although these studies are highly promising, most patients did not respond. Future targets and drugs may improve on these results.