Study Supports More Inclusive Clinical Trials With Broader Eligibility Criteria

—No difference in adverse events, clinical benefit, OS for patients with or without waivers

by Charles Bankhead

Patients who received protocol waivers in a clinical trial of targeted anticancer therapy had outcomes similar to those who met all enrollment criteria, a large retrospective review showed.

Patients who received waivers had a clinical benefit rate (CBR; response plus stable disease) of 40% as compared with 33% for patients who did not. Median overall survival (OS) numerically favored the waiver group (11 vs 8 months), but the difference did not reach statistical significance. Serious adverse events (SAEs) occurred in a similar proportion of patients in the two groups.

The findings make a case for broader enrollment criteria to allow for a more diverse patient population that more closely resembles patients treated in clinical practice, wrote Hans Gelderblom, MD, of Leiden University Medical Center in the Netherlands, and co-authors in Clinical Cancer Research.

“It is well known that results in an ‘ideal’ population do not always translate to the real-world population,” Gelderblom said in a statement. “Eligibility criteria are often too strict, and educated exemptions by experienced investigators can help individual patients, especially in a last-resort trial.”

Exemptions often involve minor deviations from eligibility criteria or lab testing, such as a slightly out-of-range lab result, imaging completed outside the recommended time window, or a tumor that cannot be biopsied for safety reasons. Such exemptions allow patients to participate in clinical trials for which they otherwise would be ineligible.

“These findings advocate for a broader and more inclusive design when establishing novel trials, paving the way for a more effective and tailored application of cancer therapies in patients with advanced or refractory disease,” said Gelderblom.

The study supports pragmatic trial design for which a growing number of clinical researchers have been advocating, said Roy Herbst, MD, PhD, of Yale Cancer Center in New Haven, Connecticut.

“We have been promoting pragmatic design that will lead to reasonable but limited eligibility and enrollment criteria,” he told MedPage Today. “[The trials] can be done simply so that they can be more inclusive and more patients can get in. This study seems to show, in what I think is sort of an interesting way, that allowing these deviations didn’t have any harmful effects.”

Clinical trials show whether an intervention or treatment is safe and effective, but many questions remain unanswered.

“Take a [finished] trial of immunotherapy, for example,” said Herbst. “We still don’t know how long we should treat. Do we have the right dose? Is the schedule right? Those questions can be addressed in more pragmatic designs after drugs are approved, or in parallel with drug approval, using the designs like we’re talking about, which have more inclusive enrollment criteria and much less data collection.”

Gelderblom and co-authors reported findings from the Drug Rediscovery Protocol, a retrospective analysis of a pan-cancer basket/umbrella trial that matched patients with treatment-refractory cancers to off-label use of targeted therapies chosen on the basis of individual tumor genomics. The analysis included 1,019 patients, 82 of whom had received protocol waivers.

They grouped the waivers into four categories: eligibility criteria exceptions, out-of-window testing, treatment exceptions, and testing exceptions. The most common waivers involved exceptions to eligibility criteria, often because of out-of-range lab tests. Testing exceptions were the second most common type of waiver and often involved exemptions from a biopsy.

The primary outcomes of interest were the incidence of SAEs and the CBR at 16 weeks. The data showed that 49 SAEs, irrespective of cause, occurred in 32 of 82 (39%) patients who received waivers. The investigators determined that waivers “possibly contributed” to seven (14%) of the SAEs compared with 385 of 937 (41%) SAEs among patients who did not have waivers. Grade 5 SAE rates were 4% in the waiver group and 5% in the non-waiver group. A protocol exception might have contributed to one of the two fatal SAEs in the waiver group, they noted.

At the 16-week cutoff for treatment effect, 33 of 82 patients in the waiver group had a partial response (n=14) or stable disease (n=19), resulting in a CBR of 40%. That difference from the non-waiver group (CBR 33%) did not achieve statistical significance (P=0.43). The 3-month absolute difference in OS (11 vs 8 months) translated into a 13% reduction in the hazard ratio (95% CI 0.66-1.15, P=0.33).

Gelderblom and colleagues noted that 17 waivers were granted for 12 patients who were not evaluable. The analysis suggested that a relationship between waivers and non-evaluability was “certain” for five patients, “possible” for six, and “unlikely” for six. They concluded that “the granted waivers in our trial did not significantly contribute to the number of non-evaluable patients and therefore did not compromise the internal validity of the study.”

The authors acknowledged several limitations to their study: inclusion of patients with a broad variety of tumor types, different molecular profiles and different assigned treatments, variability in the reasons for waivers, and a limited sample size.