Each day during the Baseline and Treatment Periods, patients were asked to call into an interactive voice response system (IVRS) to record their IBS-C symptoms. Although a number of symptoms were assessed, those used in these analyses included Abdominal Pain at its worst using an 11-point NRS, the number of BMs, and whether rescue medication was used. In addition, each BM was assessed for sensation of complete bowel emptying (yes/no).
Symptom-specific patient rating of change questions (PRCQs) for Abdominal Pain Relief and CSBM Frequency Improvement were assessed at all visits following randomization (i.e., weeks 2, 4, 8, and 12). For Abdominal Pain Relief, patients responded to the following PRCQ: ‘Compared to before you started this study, how would you rate your abdominal pain at its worst during the past 7 days?’ For CSBM Frequency Improvement, patients responded to the following PRCQ: ‘Compared to before you started this study, in the absence of laxative use, how would you rate your frequency of complete bowel movements (i.e., bowel movements where you felt like you completely emptied your bowels) during the past 7 days?’ Finally, the Degree of Relief of IBS symptoms PRCQ was assessed weekly via IVRS with the following question: ‘Compared to before you started this study, how would you rate your IBS symptoms during the past 7 days?’ For all three PRCQs, the response options were as follows: 1 = Completely relieved/improved, 2 = Considerably relieved/improved, 3 = Somewhat relieved/improved, 4 = Unchanged, 5 = Somewhat worse, 6 = Considerably worse, 7 = As bad as I can imagine.
Patients rated their current IBS and Constipation Symptom Severity weekly via IVRS using the following questions: ‘On average, how would you rate your IBS symptoms during the past 7 days?’ and ‘On average, how would you rate your constipation during the past 7 days’, respectively. For both questions, the response options were as follows: 1 = None, 2 = Mild, 3 = Moderate, 4 = Severe, 5 = Very severe. For these assessments, patients did not need to reflect back to before the study started, but rather rate their current symptoms during the past 7 days.
Endpoint Evaluation Methods
Anchor-based Estimates of Meaningful Change: An anchor-based methodology was used to estimate thresholds of meaningful change using the following assessments as anchors: symptom-specific (i.e., Abdominal Pain Relief and CSBM Frequency Improvement) PRCQs, global (i.e., Degree of Relief of IBS) PRCQs, and the IBS and Constipation Symptom Severity questions. Whereas the PRCQs require the patient to compare their current status with their status prior to starting the study, the IBS and Constipation Symptom Severity questions (these items will be referred to as ‘Current Severity Questions’) require patients to rate their current status over the past 7 days.
The anchor-based clinically meaningful threshold estimates were the change-from-baseline scores in Abdominal Pain or CSBMs averaged over those weeks the patient responded ‘Somewhat relieved/improved’ for the PRCQs.[12, 13] Similarly, the threshold estimates for the Current Severity Questions were the change-from-baseline scores in Abdominal Pain or CSBMs averaged over those weeks the patient had an improvement of one category. As a patient can have multiple weeks in which the anchor criterion was met (for either the PRCQs or the Current Severity Questions), a repeated measures mixed-model approach was performed with the use of the PROC MIXED procedure in sas software, version 9.1 (sas Institute) to estimate the mean of the change-from-baseline scores on those weeks when a patient met the anchor criterion. The mixed model had a fixed intercept, Gaussian distribution random intercepts for each patient, and the covariance pattern for the within-patient observations was unstructured. For CSBMs, as the distribution of the change-from-baseline data was non-Gaussian (i.e., the underlying distribution was skewed), clinically meaningful thresholds based on median scores were estimated in addition to the thresholds based on mean scores. For this CMC estimate, the median change-from-baseline score in CSBMs over those weeks the patient met the anchor criterion was used. If a patient had multiple weeks meeting the anchor criterion, the patient's median score was used to determine the overall change-from-baseline median score.
Sensitivity/Specificity: The diagnostic accuracy of the FDA Responder Endpoint was assessed using sensitivity/specificity-based methods. Sensitivity is defined as the ability of a measure to correctly identify patients who have a condition; it is the percentage of true positives (i.e., responders) classified as positive by the measure being evaluated. Specificity is the ability of a measure to correctly identify patients who do not have a condition; it is the percentage of true negatives (i.e., non-responders) classified as negative by the measure being evaluated. Accuracy is the percentage of true results, both true positives and true negatives.
The two symptom-specific PRCQs, Abdominal Pain Relief, and CSBM Frequency Improvement, averaged across the 12-week Treatment Period, were used as gold standards for categorizing each patient's status (improved or not improved) in the assessment of the diagnostic accuracy of the FDA Responder Endpoint. A patient's status was defined as ‘improved’ if his/her average response to both the Abdominal Pain Relief and CSBM Frequency Improvement PRCQs was ≤3.0 on the 7-point balanced scale (i.e., an average score of somewhat, considerably, or completely relieved/improved across the 12-week Treatment Period) and, conversely, ‘not improved’ if his/her average response to both the Abdominal Pain Relief and CSBM Frequency Improvement PRCQs was >3.0. In addition, the patient would be categorized as ‘not improved’ if any of the following occurred in the 12 weeks of treatment: the patient withdrew from the trial due to lack of efficacy, the patient withdrew from the trial due to a gastrointestinal adverse event, or the patient completed less than 6 weeks of treatment.
Therefore, the FDA Responder Endpoint is considered sensitive if it identifies as responders a large percentage of individuals who are categorized as improved based on PRCQ Abdominal Pain Relief and PRCQ CSBM Frequency Improvement (i.e., patients with an average score of at least ‘somewhat relieved’), and specific if it does not classify a large number of patients as responders who did not indicate relief of Abdominal Pain and improvement in CSBM Frequency. Accuracy is the percentage of patients who are categorized as both ‘improved’ and ‘responders’ or ‘not improved’ and ‘non-responders.’
To further assess the rigor of the FDA Responder Endpoint, we also evaluated its sensitivity, specificity, and accuracy while separately varying the number of weeks (from ≥6/12 weeks), the percentage change in Abdominal Pain (from ≥30%), and the increase in CSBMs from baseline (from ≥1) required to be identified as a responder. While varying one of these criteria, the other two criteria remained at the values defined for the FDA endpoint.
Percentage of Responders Based on Incremental Abdominal Pain and Weekly CSBM Criteria by Treatment Group: The percentages of patients meeting the FDA Responder Endpoint in the linaclotide and placebo treatment groups were compared using a Cochran-Mantel-Haenszel (CMH) test controlling for trial geographic region. In addition, the percentages of patients meeting incremental thresholds of reduction in Abdominal Pain and improvements in CSBM Frequency for at least 6 of the first 12 weeks of the Treatment Period in the linaclotide and placebo treatment groups were compared using Fisher's exact test.