Zhaoling Meng (1), James Rogers (3), Jonathan Sidi (3), Qi Tang (1), Dimple Patel (1), Nadia Gaudel-Dedieu (2) and David Delvart (2)
(1) Translational Informatics, Sanofi, Bridgewater, NJ, USA, (2) Translational Informatics, Sanofi, Chilly-Mazarin, FRANCE (3) Metrum Research Group, USA
Objectives: Cardiovascular (CV) safety outcome study is routinely required in diabetes drug approval. Recent empagliflozin approval for CV indication provides an additional risk reduction option for type II diabetes (T2DM) patients with high CV risks [1] and, at the same time, presents a potential confounding in the CV effect assessment for future studies. Some patients might already take empaglifozin at the study start and some patients might initiate it during the study. Concomitant administration of empaglifozin with the study drug increases the CV effect assessment uncertainty, especially when there is imbalanced empaglifozin addition between treatments during the study. A CV outcome study for glucagon-like peptide-1 receptor agonists (GLP-1ra) class drug presents such a case. Although historical GLP-1ra CV outcome studies (LEADER [2] and SUSTAIN 6 [3]) can provide good assumptions for GLP-1ra CV effect compared to standard of care (SOC), currently, there is no clinical study available to assess the CV effect of concomitant administration of SGLT-2i and GLP-1ra. We analyzed real world evidence data to estimate this effect. With estimated effects, necessary models of the study drug and planned study design, clinical trial simulations (CTS) were used to assess the impact of this confounding and the study probability of success (POS) for a GLP-1ra drug CV outcome study.
Methods: First, PopPK and PK/HbA1c exposure-response models were used to simulate patients’ HbA1c over time for GLP-1ra and SOC arms based on the planned CV outcome study design. Secondly, literature and internal CV outcome studies were used to understand and model antidiabetic medication addition during the study, especially imbalanced addition between treatments due to differential HbA1c control. Then, a real world claim database, Truven, was used to estimate the CV effects of SGLT-2i addition, either concomitantly to GLP1-ra or alone. SGLT-2i class was assessed in the analysis considering AstraZeneca’s CVD-REAL study [4] showing SGLT-2i, as a class, significantly reduced CV risks versus other T2DM medicines. Using the Truven database, T2DM patients started on 1st GLP-1ra were included in the analysis. Baseline characteristics matched T2DM patients never used GLP-1ra but started other new antidiabetic medication during the same time period were included as the SOC arm. Patients’ age, gender and SGLT-2i baseline usage etc. were used in the matching. The estimated GLP-1ra vs. SOC and GLP-1ra + SGLT-2i vs. SGLT-2i effects were estimated and integrated in the CTS along with PK, PK/PD, concomitant medication addition models, and various design factors such as sample size, enrollment rate, event rates and study dropout rates. Based on the marketing prediction of empaglifozin patient utilization during the expected study period, different percentages of patients on empaglifozin at the study start and initiation during the study were simulated. The influential factor(s) for the study outcome were explored and identified.
Results: Imbalanced antidiabetic medication additions were consistently observed in historical GLP-1ra CV outcome studies, where ~20% and ~30% concomitant antidiabetic medication were observed for GLP-1ra and SOC arms, respectively. The imbalanced additions were hypothesized as due to lack of HbA1c control in SOC arm compared to GLP-1ra arm. An empirical concomitant antidiabetic medication addition model during the blinded study phase under differential HbA1c control of SOC and GLP-1ra arms was established using an internal historical CV outcome study. The real world evidence data estimated a GLP-1ra vs. SOC CV benefit ~10% reduction and smaller GLP-1ra+SGLT-2i vs. SGLT-2i CV benefit. 10% to 50% SGLT2-i patient usage prevalence were assumed and tested in the CTS. The simulation indicated a small impact of differential SGLT-2i addition during the blinded study phase unless there was a fairly large % SGLT-2i patient usage. Therefore, the percentage of patients on SGLT-2i can be monitored during the study to mitigate the risk.
Conclusion: In this exercise, explicit and informative assumptions are essential in addition to appropriately established modeling framework to mimic a future CV study. Real world data was used to estimate the concomitant CV effects with/without empagliflozin and inform the CTS.
References:
[1] Empagliflozin, Cardiovascular Outcomes, and Mortality in Type 2 Diabetes. Zinman B, et. Al. N Engl J Med 2015; 373:2117-2128.
[2] Marso S, et al. Liraglutide and cardiovascular outcomes in type 2 diabetes. N Engl J Med 2016;375(4):311–22.
[3] Ipp E et. al. N Engl J Med. 2017 Mar 2;376(9):890-1. doi: 10.1056/NEJMc1615712.Semaglutide and Cardiovascular Outcomes in Patients with Type 2 Diabetes.
[4] Kosiborod M, et. AL. Circulation. 2017;136:249-259. Lower Risk of Heart Failure and Death in Patients Initiated on Sodium-Glucose Cotransporter-2 Inhibitors Versus Other Glucose-Lowering DrugsClinical Perspective. The CVD-REAL Study (Comparative Effectiveness of Cardiovascular Outcomes in New Users of Sodium-Glucose Cotransporter-2 Inhibitors)
Reference: PAGE 27 (2018) Abstr 8576 [www.page-meeting.org/?abstract=8576]
Poster: Methodology - Other topics