Materials and methods

The study examined clinical trials on anticancer drugs published between January 1997 and October 2007 that included an interim analysis, retrieved via Medline. The search criteria focused on publications containing the terms "interim" and "analys*" limited to human studies, clinical trials, cancer, and in English. A total of 231 reports were identified.

To ensure comprehensive results, the researchers conducted an additional search for potentially missed articles published in three major peer-reviewed journals—the Lancet, The New England Journal of Medicine, and The Journal of Clinical Oncology—between October 2006 and October 2007. This search yielded two additional reports. These journals were selected based on their frequency of publication in the initial Medline search.

For improve specificity, the initial screening was based on abstracts. Of the 233 reports, 140 were excluded based on predefined inclusion and exclusion criteria (see Figure 1). Trials excluded included phase I studies, growth factor trials, and those focused solely on surgery or radiotherapy. Trials comparing drug dose regimens, those testing palliative therapies (such as antiemetics), and study protocols were also excluded.

Ultimately, only trials of anticancer drugs that included an interim analysis were considered for analysis, resulting in 93 eligible papers describing 93 trials. Of these, 65 were excluded for various reasons: 4 trials were halted due to toxicity, 28 for futility (lack of efficacy), and 33 because they continued after the interim analysis, making them ongoing trials.

The final sample included 28 papers that met the inclusion criteria—clinical trials testing anticancer drugs that were truncated for benefit after an interim analysis. However, one paper was not retrievable, and two studies were reported in multiple journals; in both cases, the earlier publication was retained. After adjustments, 25 papers describing 25 trials were included in the final analysis.

The analysis assessed various parameters, including disease type, study duration, publication date, presence of a Data and Safety Monitoring Committee (DSMC), endpoints, sample size, rationale for interim analysis, the type of analysis conducted, consequences on the trial and patients, and characteristics of the control group. Both the primary endpoint planned for final efficacy analysis and the endpoint used for the interim analysis were recorded, as was the sample size. In some cases, certain study characteristics were not specified in the articles and, thus, not included in the analysis. The data collection form was pre-defined by the investigators, and the papers were independently evaluated by FT and GT, with the results cross-checked for consistency. In case of discrepancies, a consensus decision was reached through discussion.

Summary

• The study reviewed clinical trials on anticancer drugs from January 1997 to October 2007, focusing on those with interim analyses, identified through Medline.
• The search resulted in 231 reports with a focus on English language, human studies, and cancer clinical trials using the keywords "interim" and "analys*".
• An additional search in three major journals (The Lancet, The New England Journal of Medicine, The Journal of Clinical Oncology) yielded 2 more relevant articles.
• Initial screening of abstracts led to the exclusion of 140 reports based on specific inclusion and exclusion criteria, such as phase I studies and trials focused solely on surgical or radiotherapy approaches.
• After detailed considerations, 93 trials were initially deemed eligible; however, 65 were excluded (4 for toxicity, 28 for futility, and 33 ongoing).
• A final sample of 28 eligible papers was determined; one was not retrievable, and adjustments led to a final inclusion of 25 trials.
• The analysis encompassed various parameters such as disease type, study duration, publication date, Data and Safety Monitoring Committee presence, endpoints, sample size, rationale for interim analysis, and trial implications.
• Data evaluation was conducted independently by researchers, with consensus reached for discrepancies.