Phase II trials in patients with carcinoma of unknown primary: a pooled data analysis

Abstract

Background: The management of carcinoma of unknown primary site remains debatable. The literature data consists of about 29 phase-II studies investigating 38 regimens, providing a broad range of response rates (RR).

Methods: We performed a pooled published data analysis to identify the factors influencing RR in these 29 studies.

Results: In front-line treatment, the overall RR was 31% (430/1,380). At least nine factors significantly influenced the reported RR, excluding drugs under investigation, such as the impact factor of the journal, single-centre study, some eligibility criteria, central radiological review, sample size calculation based on statistical hypothesis and stratification. Two drugs seem to improve RR: cisplatin and doxorubicin. Two drugs seem to be associated with a worse RR: irinotecan and carboplatin.

Conclusions: This pooled data analysis illustrates that in a phase II trial setting, it is impossible to attribute the variation in RR solely to the modification of drugs under investigation.