Model-Based Meta-Analysis for Time Courses of Percent Responders in Diabetic Peripheral Neuropathy, Postherpetic Neuralgia and Fibromyalgia Pain Trials Chih-Wei Lin1, Wei Liu1, Walid Awni1, Sandeep Dutta1 1Clinical Pharmacology and Pharmacometrics, AbbVie, North Chicago, IL, 60064, USA RESULTS (CONTINUED) BACKGROUND AND OBJECTIVES Figure 2. Observed and Model-predicted (Solid Lines) Responder Rates for PHN Treatments • Common primary objectives in neuropathic pain trials, including diabetic peripheral neuropathy (DPN), post-herpetic neuralgia (PHN) and fibromyalgia (FM), are to investigate treatment difference from placebo and dose-response relationship. Pregabalin 30% reduction Gabapentin 50% reduction 30% reduction 100 75 • Primary endpoint in these trials is usually mean pain score reduction from baseline. Previously a meta-analysis by Othman et. al. (ACoP 2011) summarized the dose-response relationships for neuropathic pain trials using mean pain score reduction as efficacy endpoint. 0 75 %Responder 50 150 300 25 %Responder Dose (mg/day) 600 0 Trial Number 100 • An important secondary endpoint that is also clinically relevant is responder rate. Responder rate is used for evaluating treatment difference from placebo, and can also be used to rank order drug response within or across classes. 25 2 4 6 8 10 12 14 0 2 4 6 8 100 %Responder 50 PHN FM 5-13 Gabapentin (alpha-2 delta ) 3 9 892 0/1800/2400 4-10 Pregabalin (alpha-2 delta) 6 19 1666 0/75/150/300/600 1-13 Duloxetine (SNRI) 2 6 1027 0/20/60/120 12-26 Milnacipran (SNRI) 4 11 3234 0/50/100/200 1-27 Pregabalin (alpha-2 delta) 3 12 2002 0/150/300/450/600 8-14 31 96 12881 - - Total P%Responder t , d drug logit -1 %Responder t , d drug , N t , d drug Trial Number 6 8 10 4 8 12 16 20 24 7 28 Dose (mg/day) 50% reduction %Responder 75 Dose (mg/day) 50 25 20 60 0 120 100 Trial Number 75 5 50 9 25 200 0 50 Trial Number 0 4 8 12 16 20 24 28 0 4 8 12 16 20 24 28 Time (weeks) 3 25 7 4 8 12 16 20 24 28 0 4 8 8 12 16 20 24 28 Trial Number Pregabalin 30% reduction 50% reduction 1 100 %Responder 2 75 4 50 Dose (mg/day) 0 25 150 300 0 4 8 12 0 4 8 450 12 600 Table 3a. Estimated Responder Rates for DPN Treatments Percent Responder at Maximum Suggested dose Drug ED50 Maximum Suggested (mg/day) Dose# (mg/day) ≥30% Reduction (95% CI) ≥ 50% Reduction (95% CI) LOCF BOCF LOCF BOCF Placebo - 0 43 (38-48) 34 (30-39) 29 (25-33) 22 (19-25) Duloxetine* 9.2 60 59 (49-68) 50 (40-59) 43 (34-52) 35 (26-43) Pregabalin* 270 300 53 (46-61) 44 (37-52) 38 (31-46) 29 (24-37) *FDA labeled indication for DPN • Evaluated drugs provide an additional 10%-16% increase in 30% responder rate and 7%-13% increase in 50% responder rate for DPN over placebo effect. Table 3b. Estimated Responder Rates for PHN Treatments Percent Responder at Maximum Suggested dose Drug ED50 Maximum Suggested (mg/day) Dose# (mg/day) ≥30% Reduction (95% CI) ≥ 50% Reduction (95% CI) LOCF BOCF LOCF BOCF Placebo - 0 30 (23-39) 22 (17-28) 17 (13-23) 12 (9-15) Gabapentin* 2200 1800 49 (35-66) 38 (26-54) 31 (21-47) 23 (15-36) Pregabalin* 180 300 56 (40-73) 45 (30-63) 38 (24-56) 28 (17-44) *FDA labeled indication for PHN • Evaluated drugs provide an additional 16%-23% increase in 30% responder rate and 11%-16% increase in 50% responder rate for PHN over placebo effect. Table 3c. Estimated Responder Rates for FM Treatments 1 exp k t Percent Responder at Maximum Suggested dose Drug ED50 Maximum Suggested (mg/day) Dose# (mg/day) ≥30% Reduction (95% CI) ≥ 50% Reduction (95% CI) LOCF BOCF LOCF BOCF Placebo - 0 31 (27-35) 26 (23-29) 19 (16-22) 15 (13-18) Duloxetine* 3.9 60 42 (33-48) 36 (28-42) 27 (20-33) 23 (17-27) Milnacipran* 68 200 40 (35-46) 34 (30-39) 26 (22-31) 21 (18-25) Pregabalin* 120 450 40 (34-47) 35 (29-40) 26 (21-31) 22 (17-26) *FDA labeled indication for FM • Evaluated drugs provide an additional 8%-10% increase in 30% responder rate and 6%-8% increase in 50% responder rate for FM over placebo effect. Table 4. Estimated Durations to achieve 50% and 97% effect for DNP, PHN and FM • The correlation between observations within an arm was evaluated by a compound symmetry correlation structure. • Analyses were performed in S-PLUS 8.1 and graphs were created in R 2.14.1. Parameter DPN PHN FM Time to achieve 50% effect (wk) 1.0 1.7 1.7 Time to achieve 97% effect (wk) 5.0 8.5 8.5 • The times to achieve drug effect plateau for DPN, PHN and FM were estimated to be 5.0, 8.5, and 8.5 weeks. RESULTS SUMMARY Table 2. Parameter Estimate • The placebo response rate for 30% pain reduction (BOCF) was estimated to be 22%-34%, and for 50% pain reduction was estimated to be 12%-22% for DPN, PHN and FM. Model Parameter DPN PHN FM E0_30%reduction -0.65 -1.3 -1.1 E0_50%reduction -1.3 -2.0 -1.7 Emax 0.79 1.8 0.55 LNED50 Duloxetine 2.2 - 1.4 LNED50 Pregabalin 5.6 5.2 4.8 LNED50 Gabapentin - 7.7 - LNED50 Milnacipran - - 4.2 LOCF effect 0.37 0.44 0.25 • It takes 5 to 8 weeks to maximize drug effect for DPN, PHN or FM. k 0.69 0.41 0.41 • Response rate is 4%-11% higher using LOCF imputation than BOCF. ωBT (SD) 0.25 0.27 0.21 ρ (Within arm correlation) 0.60 0.47 0.02 • Evaluated drugs provide an additional 8%-23% increase in 30% responder rate and 6%-16% increase in 50% responder rate for DPN, PHN, and FM over placebo effect. • Response rate over placebo are higher (3%-13%) in PHN than DNP and FM. • Among approved drugs, at their maximum approved doses • Duloxetine showed the highest BOCF responder rate in DPN with estimated rates of 50% and 35% to achieve 30% and 50% pain reduction. Figure 1. Observed and Model-predicted (Solid Lines) Responder Rates for DPN Treatments Duloxetine 30% reduction Pregabalin 50% reduction 30% reduction 100 50% reduction 100 75 50 Dose (mg/day) 75 0 50 75 150 300 60 120 0 100 Trial Number 75 9 25 Trial Number 0 1 100 2 75 3 10 50 11 13 25 • Duloxetine gave the highest BOCF responder rate in FM with estimated rates of 36% and 23% to achieve 30% and 50% pain reduction. 600 %Responder 25 Dose (mg/day) • Pregabalin had the highest BOCF responder rate in PHN with estimated rates of 45% and 28% to achieve 30% and 50% pain reduction. 0 20 4 50 5 6 25 CONCLUSIONS These results suggest the trial durations for DPN, PHN and FM of at least 5 to 8 weeks would be required to maximize the separations of pain response between drugs included in this analysis from placebo. Duloxetine was most efficacious in DPN and FM, while pregabalin was most efficacious in PHN. 7 0 8 0 0 2 4 6 8 10 12 0 2 Time (weeks) 4 6 8 10 12 0 2 4 6 8 10 12 0 2 Time (weeks) 4 6 8 10 12 2400 50% reduction Time (weeks) • Where: Nresponder(t,ddrug): number of responders at time t receiving daily drug dose of ddrug; P%responder(t, ddrug): probability of responders at time t receiving daily drug dose ddrug; N(t, ddrug): number of subjects at time t receiving daily drug dose ddrug; E0,endpt: placebo response for 30% or 50% responder rate in the logit domain; ηtrial: random effect to account for between trial variability; ILOCF: effect of LOCF imputation over BOCF; Emax: maximum pain reduction effect; LNED50,drug: dose to achieve 50% maximum response in log scale; k: rate constant for the accumulation of pain reduction effect. %Responder 6 0 0 ( E max d drug ) E trial I LOCF 0 , endpt d drug exp LNED 50 , drug 4 75 %Responder 25 0 • The responder rates for different times and doses were described by a model that combined an inverse logit Emax mixed function to characterize the dose-response relationships, and a cumulative exponential function to describe the time course of the responder rate for each indication. P 2 6 • In the final model the number of responders for each indication was described by a binomial distribution. N responder t , d drug ~ binomial 0 0 100 1-12 0/75/150/300/600 10 100 200 100 Time Course (weeks) 2735 8 100 Time (weeks) 27 6 30% reduction 50 75 30% reduction Table 1. Summary of Analyzed Neuropathic Pain Trials 9 4 0 0 Pregabalin (alpha-2 delta) 2 Time (weeks) Dose (mg/day) 30% reduction • A MBMA was conducted to describe the time courses of 30% or 50% pain reduction responder rate over the treatment period and the dose-response relationship for the drugs used in the trials. The effect of using baseline observation carried forward (BOCF) or last observation carried forward (LOCF) imputation on responder rates was also evaluated. DPN 0 Gabapentin Milnacipran • The initial database included summary-level efficacy and safety data for 81 trials. Based on efficacy endpoints, imputation methods and medications of interest, the final analyzed data contained summary level data from 31 trials in ~13000 subjects for 4 medications approved by FDA for treating neuropathic pain: duloxetine, pregabalin, gabapentin, and milnacipran. 0/20/60/120 1800 0 Figure 3. Observed and Model-predicted (Solid Lines) Responder Rates for FM Treatments • A database was developed by a systemic search for all publicly available efficacy and safety information on drugs evaluated for treatment of neuropathic pain including DPN, PHN or FM. 1325 0 10 12 14 Time (weeks) 12 25 Time (weeks) 0 4 Dose (mg/day) 7 0 METHODS Duloxetine (SNRI) 50 6 0 Dose (mg/day) 9 5 • The objective of this work was to characterize the time courses of percent responders and relative efficacy of medications evaluated in clinical trials for neuropathic pain using a model-based meta-analysis (MBMA). Trials (n) Arms (n) Subjects (n) 8 75 4 50 • Meta-analyses of responder rate typically use only the response rate at study endpoint for comparing different regimens. Dose vs. response-rate relationships and the time course of responder rate across trials have not been fully investigated. Drug 3 2 0 Indication 100 1 75 Trial Number 50% reduction 12 DISCLOSURE The study design, study conduct, analysis, and financial support of the study were provided by AbbVie and AbbVie personnel were involved in the writing, reviewing and approving of this publication. Chih-Wei Lin, Wei Liu, Sandeep Dutta and Walid Awni are employees of AbbVie.