Are Clinical Trial Exclusion Criteria Too Strict?

New research suggests that eligibility criteria for randomized controlled trials (RCTs) in hematologic malignancies might be too strict, limiting their applicability to real-world clinical practice. The research, conducted by Abby Statler, MPH, MA, from Taussig Cancer Institute at Cleveland Clinic in Cleveland, Ohio, and colleagues looked at the exclusion criteria among published RCTs articles and whether or not these criteria reflected expected or observed adverse event (AE) rates among enrolled patients.

“We have this push and pull between internal and external validity – between the trial’s ability to demonstrate a cause-effect relationship and the actual generalizability of those results,” Ms. Statler said in an interview with ASH Clinical News. “It’s a challenge to balance these two things, so we asked, ‘Are we compromising external validity by focusing on internal validity?’”

To maximize the potential for registration, Ms. Statler and colleagues wrote, “eligibility criteria for RCTs may be overly restrictive so as to avoid toxicities that may be attributed to the study drug.” To test their hypothesis that RCTs exclude patients regardless of the AEs that would be expected (or ultimately observed) based on drug class, the authors identified 98 phase II and III RCTs in hematologic malignancies that were published in high-impact medical journals (an impact factor ≥5) between January 2010 and January 2015.

Of these 98 trials, 32 (33%) were leukemia trials, 27 (28%) were lymphoma, 34 (35%) were multiple myeloma, and 5 (5%) were myelodysplastic syndromes or myelofibrosis. The majority of studies were phase III (n=77; 79%), multi-center (n=92; 94%), and/or multi-national (n=63; 64%). Twelve trials contributed pivotal data leading to a label change or new FDA approval for eight drugs.

The researchers collected AE data from package inserts or published manuscripts for the following drug classes: alkylators, antimetabolites, anthracyclines, topoisomerase inhibitors, microtubule inhibitors, proteasome inhibitors, and monoclonal antibodies. Toxicities of these drug classes were then compared with corresponding trial exclusion criteria using the exact binomial test.

Using the threshold applied in medication labels (reported AEs that occurred in ≥10% of trial participants), Ms. Statler calculated the binomial probability for each AE, particularly those relevant to the most commonly used organ function exclusion criteria (hepatic, kidney, cardiac, and neurologic).

The most commonly applied exclusion criteria were:

  • medical comorbidities (e.g., active or prior cancer, previous cardiac conditions, HIV infection, hepatitis, or psychiatric disease): 97% of studies
  • inadequate organ function: 89%
  • poor performance status: 67%

“Exclusion criteria relevant to baseline organ function did not reflect established safety profiles,” the authors noted, with the proportion of studies excluding patients with specific organ dysfunction far outweighing the proportion of drug classes with known toxicities in these realms:

  • Hepatic toxicities: 87% (proportion of studies with organ-specific exclusion criteria) vs. 75% (proportion of drugs classes with known toxicities), p=.007
  • Cardiac toxicities: 74.5% vs. 62.5%, p=0.02
  • Renal toxicities: 73.5% vs. 50%, p<0.0001

Interestingly, the proportion of studies excluding patients with neurologic deficits was actually lower than the proportion of drug classes with known neurologic toxicities: 22 percent versus 50 percent (p<0.0001), respectively.

Given the toxicity profile of the investigational products in these RCTs, the widespread use of restrictive eligibility criteria seems inappropriate, Ms. Statler and co-authors concluded.

“The proportion of RCTs in hematologic malignancies published in high-impact medical journals excluding patients with comorbidities and/or organ function abnormalities does not reflect the expected or observed AEs in these patients. [Landmark RCTs], with an eye to registration, may be overly conservative in using restrictive exclusion criteria.”

However, the authors added, the number of studies that reported grade ≥1 toxicities in ≥10 percent or more participants was considerably lower than expected, assuming study treatments lead to an AE in 10 percent of patients (FIGURE). The difference was lowest for clinical trials that excluded patients with peripheral neuropathy (14 vs. 11; p=0.95).

“We were quite surprised by this. Our hypothesis was that we would see fewer studies report toxicities than we expected, but we did not anticipate that the difference would be so dramatic,” Ms. Statler said. “What may be driving this is the inclusion and exclusion criteria – we are enrolling patients who are very clinically fit and, therefore, do not have many toxicities.”

Although the strict eligibility criteria result in “very good safety data,” she added, they may not accurately reflect what occurs in daily practice. “We may be able to expand exclusion criteria a little bit to include more patients and we would not significantly impact the safety data of the investigational product.”


Statler A, Radivoyevitch T, Siebenaller C, et al. Eligibility criteria are not associated with expected or observed adverse events in randomized controlled trials (RCTs) of hematologic malignancies. Abstract #635. Presented at the 2015 ASH Annual Meeting, December 7, 2015; Orlando, Florida.