Evolution of the cancer clinical trial over three decades: problems of design, analysis, reporting and interpretation


Year:

Session type:

Ian Tannock
Princess Margaret Hospital, Toronto, Canada

Abstract

<p>The randomised controlled trial (RCT) is the gold standard for establishing new therapies in clinical oncology.  We have reviewed RCTs evaluating systemic therapy for breast, colorectal (CRC), and non-small-cell lung cancer (NSCLC) published since 1975 in major journals.</p><p>We conclude:</p><p>1. RCTs in oncology have become larger and are more likely to be sponsored by industry.  </p><p>2. Increasing size of RCTs allows detection of smaller benefits from new agents; similar benefits might be obtained with older drugs that were not evaluated in large trials</p><p>3. There was increasing use of time-to-event measures as the primary endpoint </p><p>4. Effect size remained stable or decreased with time</p><p>5. Publication bias is evident, even for relatively large RCTs</p><p>6. Approval of drugs has been based often on endpoints such as relapse- or progression-free   survival, often without evidence that they are valid surrogates for overall survival </p><p>7. Authors have become more likely to strongly endorse the experimental arm, especially in trials sponsored by industry. Bias is evident in reporting of many trials, including conclusions based on secondary endpoints  </p><p>8.Comparison of time-to-event curves is usually made at a single point (e.g. median) with the often false assumption of proportional hazards, and ignoring much of the data.  Area between curves is proposed as a superior measure of absolute benefit.</p><p>9. A minority of RCTs evaluate quality-of-life (QL), even for patients with incurable cancer, and the quality of QL assessment has been poor.</p><p>10. There is under-reporting of harm in RCTs; important side-effects are often identified after drug approval</p><p>11. Cost-effectiveness is almost never addressed, and the cost of approved drugs has increased >100 fold </p><br>