Cancer patients who participate in clinical trials hoping for better outcomes fare no better than those who do not, when setting aside the new treatment’s effect, according to the results of a study published in the Journal of the American Medical Association. The analysis found that while overall, trials had a positive benefit, this effect diminished after accounting for various factors common to trial participants such as being younger. Evidence of publication bias was also uncovered.
Participation in a clinical trial may confer a survival benefit to cancer patients is known as a trial effect, and results from access to effective new therapies (the treatment effect), but it is also thought that a trial’s closer monitoring provides a distinct benefit as well (the participation effect). The treatment effect only applies if the treatment proves to be effective, while the participation effect should apply regardless of treatment effect. But the evidence for the participation effect has been conflicting. A pair of reviews, one conducted in 2001 and the other in 2004, found no evidence of a participation effect.
The researchers therefore sought to account for biases and confounding in differences between routine care patients and trial patients. A search was performed for studies comparing survival outcomes for the two groups between January 1 2000 and August 31 2022, which turned up 12 791 records. After screening for eligibility and duplicates, this yielded 39 studies (85 comparisons) for analysis. These comparisons involved haematologic (21%), breast (16%), lung (14%), central nervous system (7%), prostate (7%), and pancreatic cancers (5%), as well as melanoma (6%). The remaining 24% consisted of bladder, cervical, colorectal, oesophageal, gastric, head and neck, kidney, ovarian, and solid mix tumours. One-third of the comparisons involved advanced or metastatic cancer.
Initially, the meta-analysis revealed a statistically significant overall survival benefit for trial participants (HR [hazard ratio], 0.76) when all studies were pooled without regard to their design or quality. But in study subsets matching trial participants and routine care patients for eligibility criteria, the survival benefits diminished (HR, 0.85). Finally, the survival benefit disappeared when only high-quality studies were pooled (HR, 0.91). They also disappeared when estimates were adjusted for potential publication bias (HR, 0.94).
Further analysis (using funnel plots and Egger’s regression test) indicated there was a publication bias against studies which lacked a participation effect.
In an accompanying editorial, Wilson et al. note that the participation effect explains that, “Patients in trials are generally younger, fitter, have fewer comorbidities, and come from higher socioeconomic groups; this enrollment bias largely explains the participation effect. The implications of this finding are important for understanding how trials are often viewed in clinical practice. The participation effect is often used to promote the view that “a clinical trial is the best treatment option, ‘but this may be a false narrative.”
Corresponding author Jonathan Kimmelman, PhD concluded: “Our findings provide reassurance that inability to enroll in a cancer trial doesn’t disadvantage a patient, at least in terms of survival. Our findings can help patients (and physicians) focus their consent discussions on the most relevant and evidence-based benefits of trial participation: the prospects of advancing the care of future patients.”
