Tau-Connectome Subtypes and Solanezumab Response in Preclinical Alzheimer’s Disease
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Importance
This study identifies clinically meaningful subtypes within preclinical Alzheimer’s disease (AD) using the functional connectome and introduces tau-predictive functional connectivity as a new tool to stratify risk and treatment response.
Objective
To determine whether fMRI–derived functional connectivity can identify subgroups with differential cognitive outcomes and treatment responses among amyloid-positive, cognitively unimpaired adults.
Design, Setting and Participants
A post-hoc analysis of the Anti-Amyloid in Asymptomatic Alzheimer’s disease (A4) study, a randomized, double-blinded, placebo-controlled clinical trial. The A4 trial was a multi-center clinical trial in the United States, Canada, Australia, and Japan. This trial enrolled cognitively unimpaired adults aged 65-85, of whom 1,490 had successful functional magnetic resonance imaging data preprocessing: A β -(n = 445) and A β + (n = 1045). Longitudinal data were drawn from both the primary trial and the open-label time period (up to 455 weeks). Hierarchical clustering was performed on baseline functional connectivity patterns, followed by spline-based generalized linear modeling to evaluate differences in longitudinal clinical and biomarker outcomes.
Intervention(s) or Exposure(s)
Tau-predictive functional connectivity, derived from previously generated connectome-based tau predictive models, was used to cluster individuals at baseline. Subsequent analyses compared longitudinal outcomes within the placebo-only group and within the treatment group.
Main Outcome(s) and Measure(s)
For the original A4 trial, the primary endpoint was the Preclinical Alzheimer’s Cognitive Composite (PACC) score, with secondary clinical outcomes including the Clinical Dementia Rating Sum of Boxes (CDR-SB), Cognitive Function Index (CFI), and Activities of Daily Living (ADL). Secondary biomarker endpoints included plasma phosphorylated tau-217 (pTau-217) and cortical amyloid SUVR (Standardized Uptake Value Ratio). We assessed significance using either permutation testing or bootstrapped models (n=5000). All confidence intervals were defined by bootstrapped models.
Results
Hierarchical clustering of connectivity data revealed two distinct subgroups among amyloid-positive individuals in the A4 trial. Individuals had distinct tau PET spatial distributions: one subgroup (Cluster 1, n=618) with a typical limbic-predominant distribution, and the other (Cluster 2, n=427) showing an atypical cortical-predominant distribution. In the placebo arm, the cortical-predominant group exhibited significantly more cognitive decline on both PACC (marginal mean difference: -3.5, 95% CI: -6.4 to -0.6, p = 0.009) and secondary cognitive and clinical outcomes (CDR-Sum of Boxes: +1.8 (+0.6 to +3.1), p =0.005; Activities of Daily Living: - 8.7 (-16.5 to -1.2), p = 0.010; Cognitive Function Index: +4.4 (+1.1 to +7.9) p=0.010; Plasma pTau-217 not significant: p = 0.67; FDR-adjusted for 4 tests) at the end of the study (Week 455). Remarkably, in the cortical-predominant cluster, those treated with solanezumab had significantly improved PACC scores compared to placebo counterparts (marginal mean difference at 455 weeks: +3.3, 95% CI: +0.4 to +6.4, p = 0.03). This corresponds to 48% less cognitive decline, equivalent to a 103-week difference in modeled cognitive trajectories over the 9-year study period, despite the trial’s overall null results. Secondary clinical outcomes were not significant in either cluster. In typical tau group, solanezumab showed no effect in secondary outcomes (CDR-SB: -0.1, 95% CI: –0.8 to +0.7; p=0.80; ADL: +0.9, 95% CI: –3.3 to +5.1 p=0.80; CFI: –0.2, 95% CI: –2.1 to +1.5, p=0.80; FDR-adjusted for 5 tests). In the cortical-predominant group, effects trended in favor of solanezumab on CDR-SB: -1.1, 95% CI: –0.4 to +2.8; p=0.20; ADL: +2.5, 95% CI: –6.4 to +12.1; p=0.60; CFI: –2.5, 95% CI: –6.4 to +0.8; p=0.20; FDR-adjusted for 5 tests) but none reached significance. Amyloid clearance was significantly reduced in both groups at equal levels (Typical tau Solanezumab-Placebo Δ SUVR: -0.06, 95% CI: -0.08 to -0.04, p =<0.001; Cortical-predominant tau Solanezumab-Placebo Δ SUVR: -0.05, 95% CI: -0.08 to -0.02, p = 0.008; FDR-adjusted for 5 tests). There was a significant reduction in plasma pTau-217 for cortical-predominant solanezumab patients (-0.29; 95% CI: -0.49 to -0.11, p=0.008; FDR-adjusted for 5 tests), while the typical group showed no associated differences (-0.08; 95% CI: -0.22 to +0.05, p = 0.63; FDR-adjusted for 5 tests).
Conclusions and Relevance
We show that baseline clustering using tau-predictive functional connectivity identified an AD subtype with worse overall outcomes and stronger responsiveness to solanezumab. These findings highlight the potential of the functional connectome to personalize interventions in AD and in future clinical trials.
Key Points
Question
Can tau-predictive functional connectome patterns derived from baseline functional magnetic resonance imaging (fMRI) identify clinically meaningful subgroups in preclinical Alzheimer’s disease?
Findings
We performed a post-hoc analysis of the Anti-Amyloid in Asymptomatic Alzheimer’s (A4) study, a clinical trial of solanezumab in preclinical Alzheimer’s patients. Hierarchical clustering of baseline tau-predictive functional connectome signatures identified two distinct subgroups. Groups showed no baseline differences in demographic or biomarker variables, except for a modest difference in racial distribution. Spatial tau distributions differed among the groups, with one group showing a classical limbic-predominant pattern and the other having a cortical-predominant tau pattern. Among participants in the placebo arm, the cortical-predominant subgroup demonstrated poorer longitudinal scores on the A4 trial’s primary endpoint, the Preclinical Alzheimer’s Cognitive Composite (-3.3 points; lower score indicates greater cognitive impairment), relative to the other group. Importantly, within the treatment arm, individuals belonging to the cortical predominant subgroup exhibited positive treatment effects relative to the placebo group, despite the trial’s overall null results.
Meaning
Our clustering analyses used the baseline fMRI-derived functional connectome to identify a vulnerable subgroup in preclinical AD individuals that benefited from the drug in the A4 null trial of solanezumab. These findings support the use of fMRI-based biomarkers for patient stratification and the development of more personalized treatments in future clinical trials of Alzheimer’s disease.
