PWE-027 Mediation Analysis Supports a Direct Effect of Linaclotide (LIN) on Abdominal Pain (AP) Relief Independent of Constipation Improvement

Abstract

Introduction: Linaclotide (LIN) demonstrated statistically significant improvements over placebo (PBO) in abdominal pain and CSBM frequency in two Phase 3 trials of IBS-C patients. IBS-C patients suffer from recurrent abdominal pain that may be arising from increased visceral hypersensitivity, exacerbated by constipation. The relative direct and mediated (by increasing CSBMs) treatment effects of LIN on abdominal pain improvement have not been examined. Aims: To estimate the direct effect of LIN on improving abdominal pain by controlling for a concurrent increase in CSBMs. Methods: Patients meeting Rome II criteria for IBS-C were randomized to oral LIN or PBO for 26 weeks. Patients reported abdominal and bowel symptoms, and rescue medication use daily. A multilevel mediation analysis was performed on percent improvement from baseline in abdominal pain scores to estimate the proportion of the LIN treatment effect attributable to increased CSBM rate. CSBMs occurring on the day of the reported abdominal pain score and also on the previous 6 days were considered as mediation variables. This analysis was performed on Weeks 13- 26 of a Phase 3 trial because the effect of LIN and PBO on abdominal pain was generally constant over this time period. An additional analysis summarizing patient improvement in abdominal pain on a particular day was also performed using a 2-way CSBM stratification: the number of days since a patient's last CSBM (0, 1, 2, and ≥3 days) and the number of CSBMs a patient had through the previous 3 days (0, 1, 2, and ≥3 CSBMs). Results: The results of the mediation analysis indicated that the 20% effect of LIN on abdominal pain above the PBO effect was a combination of direct effect (18%) on abdominal pain and an indirect effect (2%) mediated by increasing a patient's CSBM rate (Figure 1). Spontaneous bowel movements (SBMs) that were not associated with a sense of completeness and BMs associated with rescue medication use did not qualitatively increase the amount of mediated LIN pain effect. The results of the 2-way CSBM stratification analysis (Table 1) indicated that improvement in abdominal pain was influenced by the time since last CSBM and the number of recent CSBMs. However, consistent with the predominant direct effect of LIN on abdominal pain shown in the mediation analysis, in each cell of Table 1, the LIN-treated patients had greater abdominal pain relief than PBO-treated patients when controlling for these CSBM factors. Conclusions: The results of these analyses are consistent with the hypothesis that the LIN effect on abdominal pain (over PBO) was predominantly a direct effect and, to a lesser extent, a mediated effect of increasing CSBM frequency. (Table Presented).