Background/Purpose: Fibromyalgia (FM) is characterized by symptoms that include widespread pain and sleep disruption. Clinical studies that rely on patient self-reported outcome measures such as pain scales may be analyzed by responder analysis (comparisons of proportions of treated patients achieving a predefined clinically meaningful improvement threshold) and by group mean changes.  In a Phase 2b trial of TNX-102 SL¹, a proprietary eutectic sublingual (SL) tablet formulation of low-dose cyclobenzaprine HCl (2.8 mg) in FM patients (BESTFIT), we compared these approaches to the evaluation of changes in pain and FM symptoms. 

Methods: 205 FM patients who satisfied the ACR 2010 FM criteria from 17 US investigational sites were randomized in a double-blinded fashion to placebo (n=102) or TNX-102 SL (n=103) for 12 weeks. Topline results from BESTFIT are presented elsewhere.  For this analysis, we compared changes in pain score, patient global impression of change (PGIC), and Fibromyalgia Impact Questionnaire-Revised (FIQ-R), analyzed by both responder analyses and group mean change from baseline techniques.

Results: For the daily pain diary, the responder analysis (=>30% improvement from baseline) resulted in 34.0% on TNX-102 SL compared to 20.6% on placebo, p=0.033.  The same daily pain data when analyzed as mean change from baseline, TNX-102 SL had improvement of -1.50 units vs. -1.0 units on placebo, p=0.086.  Using the FIQ-R pain item as the pain measure resulted in a response rate of 38.8% on TNX-102 SL vs. 23.5% on placebo, p=0.025, while the mean change was -1.8 on TNX-102 SL vs -0.7 on placebo, p=0.004.  PGIC is a 7 point Likert scale and defining response “very much improved” or “much improved”, the response rate was 30.1% on TNX-102 SL vs. 16.7% on placebo, p=0.025, and the mean change was 3.1 on TNX-102 SL vs 3.6 on placebo, p=0.025. The FIQ-R was evaluated as 21 discrete items, three defined domains (symptoms, overall impact and function) and a total score. The FIQ-R total score and symptom domain were statistically different between treatment groups by both responder and mean analyses. The function domain separated on the responder analysis but was not significant (p=0.059) on mean change.  Systemic adverse events (AEs) were infrequent.  Local administration site reactions (transient tongue or sublingual numbness) occurred in 42% of treated patients.

Conclusion: Results from this Phase 2b trial support the finding that responder analyses for analgesia²and other patient-reported outcomes may reveal significant and meaningful effects that are missed by group mean changes.  In the particular case of the BESTFIT study of TNX-102 SL, the pain questionnaire was not specific to central or regional  pain and TNX-102 SL targets non-restorative sleep with secondary improvement in pain. An ongoing confirmatory study is utilizing pain and other responder analyses as the primary and secondary endpoints which are more appropriate and sensitive measures for the evaluation of TNX-102 SL for FM.

¹TNX-102 SL is an Investigational New Drug and has not been approved for any indication.

²Witter J, Simon LS, Dianne R. Are means meaningless? The application of individual responder analysis to analgesic drug development. APS Bulletin. 2003;13:4–7.

« Back to 2015 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/responder-compared-to-mean-change-analyses-in-a-fibromyalgia-phase-2b-clinical-study-of-bedtime-rapidly-absorbed-sublingual-cyclobenzaprine-tnx-102-sl/