P2‐011: PHARMACOKINETIC‐PHARMACODYNAMIC MODELING OF THE THERAPEUTIC EFFECT OF ORAL P38 MAPK ALPHA ANTAGONISM ON EPISODIC MEMORY IN PATIENTS WITH EARLY ALZHEIMER'S DISEASE (AD)

Background:Neflamapimod is an oral antagonist of p38a, a kinase implicated in inflammation-induced hippocampal synaptic dysfunction. Phase 2a clinical trials (one of 6and one of 12-weeks duration) with neflamapimod in early AD demonstrated improvement in measures of episodic memory (CTAD, 2016; AAIC, 2017). In the absence of a placebo control in these trials, a specific neflamapimod-driven treatment effect was indicated by pharmacokinetic-pharmacokinetic (PK-PD) modeling of the longer duration study that demonstrated a relationship between plasma drug concentration exposure and improvement in episodic memory (AAIC, 2017). We report herein additional PK-PD modeling analyses than reported previously and assessment of the robustness of the best-fit model. Methods: Fifteen patients with MCI due to AD or mild AD (MMSE 20-28) received neflamapimod at 40 mg or 125 mg twice daily for 12 weeks. Episodic memory function assessed with Wechsler Memory Scale (WMS)–IV, analyzed as Immediate-Recall composite and Delayed-Recall composites. Individual plasma drug exposure (Area-under-curve over dosing interval of 12 hours) was derived from a population pharmacokinetics analysis of the two phase 2a studies and used in the PK-PD analysis. Results: A series of models were assessed for episodic memory scores with a set of covariates drug exposure, patient characteristics and some baseline AD metrics. Drug exposure was a statistically significant predictor of the change from baseline of combined (immediate + delayed) recall score for the best-fit model (p-value < 0.0001, and r1⁄40.696). Separately, there was also an important relationship between the drug exposure and change from baseline of either immediate or delay recall score (both with p-value < 0.001). Finally, a sensitivity analysis of the bestfit model for change from baseline of combined immediate and delayed-recall was performed using the remove of one subject-at-time jackknife resampling approach to determine model robustness. The r values for the jackknife models ranged from 0.63 to 0.77, indicating that no one individual subject disproportionately influenced the result. Conclusions: PK-PD modeling indicates that there is a robust relationship between plasma drug concentration of neflamapimod and improvement in episodic memory, as assessed by immediate and delayed-recall on the WMS.