Introduction:
Clinical trials are integral to evidence generation, yet enrollment rates remain low. One possible contributor is overly restrictive eligibility criteria that unnecessarily limit the eligible participating populations. For example, some trials may exclude participants based on kidney function despite testing interventions that are neither renally excreted nor impact kidney function. However, how often this occurs and the effects on enrollment are unknown. For these reasons, we used natural language processing (NLP) to extract data on phase 3 urologic oncology trials from ClinicalTrials.gov and compare trials’ kidney function eligibility criteria to the kidney function effects of the trials’ interventions.
Methods:
We accessed all phase 3 urologic oncology clinical trials registered on ClinicalTrials.gov. We used NLP to extract kidney function requirements (i.e., creatinine, creatinine clearance/GFR) from each trial’s free-text record. We trained our NLP algorithm on 245 prostate cancer trials, then validated the algorithm with 50 kidney and bladder cancer trials. Accuracy was 90%, with only 1 kidney function miss by the algorithm and 4 minor disagreements in the level of kidney restriction (e.g., GFR>45 vs. GFR>60). We extracted intervention names for each trial from trial records using Python. We manually searched for possible effects on the kidney on FDA.gov, PubMed.gov, or company websites for each intervention and coded as “no kidney effect”, “excreted by the kidney”, and/or “harmful to the kidney”. We considered “strict exclusion” as trials excluding participants with GFR<45 or GFR<60, as a GFR<30 reflects competing mortality risk in addition to drug toxicity and clearance-related dosing risks.
Results:
Of 850 trials involving 1,116 interventions, 297 (34%) listed kidney function eligibility restrictions, and 421 (50%) tested an intervention with possible renal effects (i.e., potential renal toxicity or significant renal clearance). Of 285 trials with kidney function exclusions, 55% tested interventions with no kidney effects. Most trials with kidney function exclusions restricted only for GFR<30 (n=165), with fewer trials excluding for GFR<45 (n=89) or GFR<60 (n=31). Of the 120 trials with the strictest exclusions (GFR<45 or GFR<60), 43% tested interventions with no significant renal clearance or renal toxicity. Conversely, of 421 trials testing interventions with potential renal toxicity or significant renal clearance, only 169 (37%) had kidney function exclusions in the eligibility criteria.
Conclusion:
A third of phase 3 urologic oncology trials have kidney function eligibility restrictions, and 14% excluded patients strictly based on kidney function. However, only about half of these trials test interventions with potential renal clearance or renal toxicity. Further, many trials of interventions with potential kidney effects did not have kidney related eligibility criteria. While many factors contribute to clinical trial enrollment, eligibility criteria must be strongly considered when designing trials to optimize both trial enrollment and generalizability of trial findings. Additional work is needed to further describe existing trial eligibility criteria and the potential effects on both trial enrollment and the generalizability of expanding eligibility criteria. Our NLP approach allows for scalable, fast, and reliable characterization of clinical trial kidney function eligibility criteria, and is a proof of concept for a platform to analyze additional eligibility criteria of clinical trials on a broader scale.
Funding: NCI T32 CA180984
Image(s) (click to enlarge):
ARE PHASE 3 UROLOGIC ONCOLOGY CLINICAL TRIAL KIDNEY FUNCTION ELIGIBILITY REQUIREMENTS APPROPRIATE? APPLYING NATURAL LANGUAGE PROCESSING TO CLINICAL TRIAL ELIGIBILITY CRITERIA
Category
Health Services
Description
Poster #25
Wednesday, Dec 1
3:00 p.m. - 4:00 p.m.
Health Services/Penile Cancer
Presented By: Kristian Stensland
Authors:
Kristian Stensland
Merrick Bank
Todd Morgan
Zachery Reichert
William Jackson
Samuel Kaffenberger
Jeffrey Montgomery
Randy Vince
Ted Skolarus