Blood‐based biomarkers for predicting treatment response in the Apathy in Dementia Methylphenidate Trial 2 randomized clinical trial

Abstract

AbstractBackgroundThe Apathy in Dementia Methylphenidate Trial 2 (ADMET 2) found that methylphenidate (MPH) was modestly efficacious is treating apathy in Alzheimer’s disease (AD) with 66% response (> = 4 point decrease on Neuropsychiatric Inventory – apathy (NPI‐A). Here, we evaluated if blood‐based biomarkers potentially associated with treatment‐limiting neuronal damage predicted the probability of treatment response.MethodADMET 2 participants with baseline and endpoint (6 month) NPI‐A and baseline concentrations of (i) neuronal damage: neurofilament light (NFL) and S‐100B, (ii) inflammation: interleukin (IL)‐6, IL‐10, Tumor Necrosis Factor‐alpha (TNFα), and (iii) oxidative stress: lipid hydroperoxide (LPH), 4‐hydroxynonenal (4‐HNE), 8‐isoprostane (8‐ISO) were included. Biomarkers were normalized by log transformation and pareto scaling. Univariate models examined whether each biomarker interacted with the treatment condition to predict a change of at least 2 points on the NPI‐A. Next, biomarkers with significant associations were included in a multivariate logistical regression model that was used to predict treatment response to MPH and placebo. The index score (difference between response probabilities for MPH and placebo) of each participant indicated their favourability for MPH treatment. Participants were grouped into quartiles by their index score with confidence intervals estimated using non‐parametric bootstrapping.ResultForty‐nine participants (placebo = 26, MPH = 23, age = 75.4 years (standard deviation [SD] = 7.6), 57% males, Mini Mental State Examination = 19.7 [SD = 4.6]) were included. MPH was more efficacious than placebo in those with higher baseline NFL (‐5.8, t = ‐3.1, p = 0.003), and less efficacious in those with higher LPH (2.0, t = 1.48, p = 0.15). Apathy worsened in participants with high TNFα (2.71, t = 1.94, p = 0.06) in both MPH and placebo groups. Participants in the top quartile of the biomarker index score (66.7% on MPH) were more likely to respond to MPH than placebo (index score = 0.58, 95%CI: 0.48, 0.69).ConclusionIndividuals with greater neuronal injury were more likely to respond to MPH, suggesting that MPH supplementation provides more benefit when neuronal injury is more severe. Combining blood biomarkers yielded a prediction score where participants with high NFL and low LPH levels were more likely to respond to MPH. This approach utilizes biomarkers to address heterogeneity in treatment response to MPH.