Using Biomarkers to Predict Memantine Effects in Alzheimer's Disease: A Proposal and Proof-Of-Concept Demonstration

Abstract

Memantine's benefits in Alzheimer's disease (AD) are modest and heterogeneous. We tested the feasibility of using sensitivity to acute memantine challenge to predict an individual's clinical response. Eight participants completed a double-blind challenge study of memantine (placebo versus 20 mg) effects on autonomic, subjective, cognitive, and neurophysiological measures, followed by a 24-week unblinded active-dose therapeutic trial (10 mg bid). Study participation was well tolerated. Subgroups based on memantine sensitivity on specific laboratory measures differed in their clinical response to memantine, some by large effect sizes. It appears feasible to use biomarkers to predict clinical sensitivity to memantine.

Keywords: Alzheimer’s disease; event-related potentials; memantine; neurocognition; prepulse inhibition.

Fig. 1.

Study design and test schedule, described in the Methods. The study involved a screen day, two test days, and titration to 10 mg MEM bid for 24 weeks, with outcome measures after 8, 16, and 24 weeks. The detailed schedule of the two test days is shown at right: autonomic, subjective, neurocognitive, and neurophysiological measures were obtained on each test day after administration of either placebo or 20 mg MEM, in a double-blind, order-balanced design.

Fig. 2.

Five examples of changes from baseline clinical outcome measures (Y-Axis) after 8–24 weeks of MEM (10 mg bid), among subgroups defined by a low versus high (median split) response to acute MEM challenge (placebo versus 20 mg po). Larger Y-axis values reflect worsening of AD symptoms. A) Example of autonomic response to acute MEM challenge: Subjects with a mild bradycardic response to acute MEM challenge exhibited a 16-week delay in the progression of cognitive symptoms during MEM treatment (d = 1.38 and 0.91 at weeks 8 and 16, respectively). B) Example of subjective ratings after acute MEM challenge: Subjects who experienced the greatest “Happy” increase (shown here) and least “Anxiety” increase after acute MEM challenge were most likely to experience reductions in NPI-Q (shown here) and GDS scores, respectively. Less acute MEM-associated drowsiness predicted a positive cognitive (ADAS-cog) response to MEM. C) Example of pill “guess”: More favorable responses to MEM were detected in measures of ADAS-cog, GDS, and NPI-Q (shown here) among subjects who did (n = 5) versus did not (n = 3) correctly guess pill identity on the day that they received active MEM dose (d = 0.41, 1.13, and 0.88 for ADAS-cog, GDS, and NPI-Q, respectively, averaged across weeks). “Correct guess” versus “incorrect guess” groups did not differ in baseline ADAS-cog or MoCA scores (both Fs < 1). D) Example of cognitive response to acute MEM challenge: Progression of ADAS-cog scores for weeks 8–24 among subjects whose RBANS index score declined after acute MEM (mean decline = 3.8) versus those that didn’t decline (mean gain = 5.3) (d = 1.07 at week 16). This pattern was also seen for the GDS response to sustained MEM treatment (d = 1.17 at week 16). E) Example of neurophysiological response to acute MEM challenge: Subjects showing the greatest acute MEM-enhanced PPI exhibited greater gains in GDS scores (shown here; d = 1.20 by week 24) and a modestly delayed progression of cognitive deficits (d = 0.74 by week 16) and during MEM treatment. Two subjects had very low startle magnitude that would qualify them as “non-responders” in many studies of PPI (e.g., [12, 16]), but exclusion of these subjects did not impact the overall patterns of results. Subjects showing an acute MEM-induced increase in FC E/I also exhibited a 16-week delay in deterioration of ADAS-cog performance (not shown: d = 1.38 and 0.91 for weeks 8 and 16, respectively).