In our study, we aimed to explore whether quantitative MRI markers derived using AI-based analysis, we use the ICO brain of ICO metrics, could help predict treatment outcomes in patients receiving lecanemab in our real-world clinical practice. And we include patients with biomarker-confirmed early Alzheimer’s disease who started treatment at our center and analyzed their baseline MRI scans using an automated software that provides quantitative measurements of whole brain and regional brain volumes and microhemorrhage burden...
In our study, we aimed to explore whether quantitative MRI markers derived using AI-based analysis, we use the ICO brain of ICO metrics, could help predict treatment outcomes in patients receiving lecanemab in our real-world clinical practice. And we include patients with biomarker-confirmed early Alzheimer’s disease who started treatment at our center and analyzed their baseline MRI scans using an automated software that provides quantitative measurements of whole brain and regional brain volumes and microhemorrhage burden. And this is a different approach from standard radiology assessment because instead of relying only on visual interpretation, the software generates objective volumetric measures and also compares them to normative population data. And this allows us to understand how advanced and structural neurodegeneration is in each patient at treatment initiation. And I think that one of the most interesting findings was that baseline gray matter volume was the strongest predictor of cognitive outcome. Patients who had higher global gray matter volume at the start of treatment experienced less cognitive decline over the following 12 months. And this suggests that the degree of neurodegeneration at treatment initiation at baseline actually matters, and that patients with relatively preserved brain structure may have a greater chance of benefiting from anti-amyloid therapy, even though the whole population was early Alzheimer’s disease in clinical parameters. And another important finding was related to safety. We saw that baseline microhemorrhage burden was the strongest predictor of ARIA. And in fact, each additional microhemorrhage at baseline increased the risk of ARIA by approximately threefold, even more than threefold. And this is clinically relevant because even though patients with a large number of microhemorrhages were excluded from treatment, our findings suggest that even within the eligible range, microvascular fragility still plays an important role in higher risk. And we also examined how well the AI system performed in detecting microhemorrhages, and the agreement between the automated analysis and the expert neuroradiology readings was excellent, with an intraclass correlation coefficient of 0.89. And this is important because it suggests that AI-supported analysis can provide reliable but also scalable measurements in clinical practice. So overall, the results have some key ideas. First, that baseline brain structure reflects disease stage and that this may influence treatment response. Second, that microvascular pathology plays an important role in higher risk. But more broadly, we show that quantitative MRI tools may help us move beyond simple eligibility assessment toward more biologically informed treatment decisions.
This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.
