Enrolment in Clinical Trials: Do Patients Benefit?

A commonly held belief among clinicians is that enrolling patients with cancer in clinical trials improves patient outcomes and quality of care. The possibility of individual benefit is frequently, and justifiably, used as part of the discussion with patients surrounding involvement in trials. However, is this actively the case?

It is a very difficult question to answer with data; clinical trials are heterogenous enough without accounting for differences in disease biology, existing available treatments, comparative arms, blinding, bias and demographic considerations. However, that is exactly what Dr Renata Iskander and colleagues have attempted to do. In a meta-analysis published in the Journal of the American Medical Association (JAMA) in May 2024, the authors examined 39 studies (with a total of 85 comparisons) to delineate between what they termed the treatment effect, the participation effect, and the trial effect.

Given the lack of standardised nomenclature in this space, the authors have had to engineer a the following novel definitions:

• Treatment effect: outcomes that are mediated by the assignment to the experimental intervention within a trial

• Participation effect: effects of trial that are not mediated to arm assignment

• Trial effect: combination of the treatment and trial effects

The components of the trial, treatment and participation effects are summarised below:

Courtesy JAMA (2024)

The authors’ definition of the participation effect attempted to control for prognostic confounders. These include:

• Demographic factors: include age, sex, ethnicity

• Pre-existing illnesses

• Cancer-specific variables: include performance status, histology, stage

• Pre-trial treatment factors: previous surgical and (neo-)adjuvant systemic therapy

To be included in the analysis, studies had to conform to the following inclusion criteria:

• Compare overall survival in a cohort of patients receiving experimental treatment to routine care. Results must be given as hazard ratios (HR)

• Treatment includes a drug/biological agent being tested in patients with cancer

In all, 39 studies were included in the analysis. Information extracted included patient demographics, treatment characteristics, and factors required for adjustment for participation effect. All OS HRs were also extracted regardless of assigned arm; when more than one HR was extracted, data that reflected the most adjustments for confounding factors was used.

Another challenge for the authors is the lack of standard tools for assessing the impact of trial and participation effects as defined in this analysis. To overcome this, the authors first created an acrylic graph to identify factors that could cause a participation effect (below). Sixteen factors associated with confounding or bias were identified. From this, the authors created a 16-point scoring scale that examined how well a trial compared and accounted for these factors, with higher scores indicating a greater degree of comparison.

Courtesy JAMA (2024)

Of the 39 studies included in the final analysis, 32 comparisons assessed the trial effect. The median sample sizes for trial and routine care patient groups were 209 and 409 patients, respectively. The majority of studies were published between 2018-2022, with haematological cancers the most common cancer subtype. Other common subtypes included breast, lung and “other”, which combined bladder, cervical, colorectal, oesophagogastric, head and neck, kidney, ovarian and mixed solid tumours.

The original pooled HR for all included studies was 0.76. However, this was in a highly heterogenous cohort (I-squared score: 83%, indicating substantial heterogeneity). When studies were grouped according to the author’s quality score, studies scoring <8 demonstrated a survival benefit for trial participants (HR 0.70), while those scoring ≥8 demonstrated no significant survival benefit (HR 0.91). Funnel plot and Egger test analyses (p=0.007) suggested possible publication bias against studies failing to show a participation effect.

The Bottom Line:

This is an incredibly ambitious attempt to not only answer whether patients benefit from participation in clinical trials, but to quantify how. The study’s composition and analysis take a very interesting approach to differentiating trials’ superior care processes — associated with closer follow-up, more investigations and treatment at a tertiary or quaternary cancer centre — from the agents provided by the trial.

The lack of significant benefit amongst studies deemed to be “high quality” may seem slightly disheartening at first glance. However, the degree of heterogeneity among studies indicates that finding a pattern in the collected data is difficult. Studies investigating all cancer types were included; outcomes for patients with haematological malignancies are stereotypically superior to advanced solid organ malignancies, with higher cure and remission rates. Outcomes also vary between types of solid organ tumours; the expected prognosis for patients with oesophageal, gastric and lung cancers is usually much shorter than those with renal cell or colorectal cancer, for example.

For so-called “orphan cancers”, the degree of benefit when compared against patients for whom “routine care” is limited to best supportive care and symptom management. No data were provided on the comparison arms for each trial of what “routine care” constituted, which may increase the heterogeneity of results.

The authors also acknowledge the possibility of publication bias, where negative studies are not published, which may further impact the study’s results. Their attempts to statistically correct for potential bias “produced an HR that was inconsistent.”

Ultimately, the study leaves many questions unanswered, and as a result, the best practice remains to review potential trial benefits on a case-by-case basis. 

Source:

Iskander R, Moyer H, Vigneault K, Mahmud SM, Kimmelman J. Survival Benefit Associated With Participation in Clinical Trials of Anticancer Drugs: A Systematic Review and Meta-analysis. JAMA. 2024 May 20:e246281. doi: 10.1001/jama.2024.6281.