Apparent Deceleration of Microstructural Neurodegeneration in a Real-World Study of AD Patients Treated with Lecanemab

Preliminary analysis of a diffusion MRI measure of cortical microstructure in real-world longitudinal data from Lecanemab-treated AD patients shows a tendency for apparent within-subject deceleration, consistent with a treatment benefit slowing neurodegeneration

Introduction

The amyloid-beta immunotherapy Lecanemab was associated with a 27% slowing of CDR-SB progression (van Dyck et al., 2023), but exhibited accelerated brain volume loss (Alves et al., 2023). It has been noted (Belder et al., 2024) that the hippocampi were relatively spared despite being strongly associated with atrophy in untreated AD, and it might be hypothesised that ‘pseudo-atrophy’ would be spatially distributed, with treatment-related benefit more localised in AD-specific regions. We explore this hypothesis using a diffusion-MRI-derived measure of cortical microstructural integrity or disarray.

Methods

This real-world study (jRCT1031240123) recruited patients seeking disease-modifying therapy at NHO Niigata National Hospital between January 2024 and November 2025. 15 patients (13 with baseline CDR=0.5, 2 with CDR=1) had more than three time-points at the time of analysis (one with four, nine with five, and 5 with six: baseline, 2m, 3m, 6m, 12m and 18m). All participants underwent MRI including 32-direction diffusion tensor imaging and 3D T1-weighted imaging. A cortical microstructural measure, PerpPD+, was calculated (Torso et al., 2022), across Desikan-Killiany regions, an AD Signature meta-region (Torso et al., 2025), and over the whole-brain. Simple quadratic regression models were fitted within-subject, and the quadratic term (positive for acceleration and negative for deceleration) was extracted for cross-subject analysis.

Results

Figure 1 shows the raw longitudinal trajectories of PerpPD+ for whole-brain and AD Signature. Figure 2 shows the distribution over the 15 subjects of the quadratic terms for each, centred close to zero for whole-brain, but with some evidence of deceleration for the AD Signature. Figure 3 shows a regional visualisation with the count of negative within-subject terms over the 15 patients; the regions with the greatest number of subjects showing apparent deceleration include left pars triangularis (12/15), left inferior-temporal (11/15) and right supra-marginal (11/15). The prominence of the temporal lobe is consistent with the hypothesis of slowing of neurodegeneration in the regions most affected in untreated AD.

Conclusions

This preliminary analysis of a microstructural measure of cortical neurodegeneration shows a tendency for apparent deceleration in regions commonly affected by neurodegeneration, potentially consistent with a hypothesised spatial separation of pseudo-atrophy and treatment-related slowing of neurodegeneration.