Pharmacodynamic and clinical endpoints for functional colonic disorders: Statistical considerations

Background: Intra- and inter-subject coefficients of variation (COV) of scintigraphic colonic transit (SCT) are well characterized. SCT response to therapy predicts clinical efficacy of experimental medications in lower functional gastrointestinal disorders (FGID). Aim: To compare COVs for bowel function with pharmacodynamic (PD) colonic transit geometric center (GC) as endpoints in lower FGID studies. Methods: We evaluated data from placebo arm of 9 phase IIA, parallel-group, clinical trials of PD effects of linaclotide, dexloxiglumide, renzapride, elobixibat, ROSE 010, and chenodeoxycholate in lower FGID with constipation, and pexacerafont, VSL#3, and colesevelam in lower FGID with diarrhea. Patients completed daily diaries for at least 7 days of stool frequency, consistency (7-point Bristol Stool Form Scale), and ease of passage (7-point scale from manual disimpaction to incontinence). Seventeen patients received placebo in 2 separate studies allowing assessment of intra-patient COVs. We calculated sample sizes required to demonstrate a 30 % effect size for colonic transit, stool frequency, consistency and ease of passage for patients with lower FGID with constipation and, separately, diarrhea. Results: COV _inter from 87 patients and COV_intra from 17 patients are reported. Generally, COV_intra is somewhat greater than COV _inter. The COVs for PD endpoints are lower than for clinical endpoints; however, clinically relevant effects can be identified with modest (~50 %) increases in the sample size using parallel-group design studies. Conclusion: Phase IIA studies that incorporate clinical and PD endpoints are feasible in lower FGID associated with constipation or diarrhea. Crossover design would require lower sample size for most endpoints compared to parallel-group studies.