Fig. 4

Longitudinal atrophy rates of NGF AD patients plotted against subtype-specific z-scores from the ADNI data. First, using ADNI data, we calculated subtype-specific cutoffs for longitudinal atrophy rates based on the upper 10th percentile (+ 1.3 standard deviations) [39]. Second, NGF AD patients were classified into one of the four subtypes and their longitudinal atrophy rates were compared to the subtype-specific cutoffs derived from the ADNI data. Patients above the cutoff reflect slower atrophy rate than expected, hence suggesting a possible treatment effect. a, b The longitudinal atrophy rates at 6 months (calculated as volume at 6-month follow-up minus volume at baseline, and transformed to z-scores using subtype-specific ADNI reference data) and at 12 months (volume at 12-month follow-up minus volume at baseline, and transformed to z-scores using subtype-specific ADNI reference data) for the basal forebrain, hippocampus, precuneus, and primary somatosensory cortex (PSC). Analyses were done separately over 6 and 12 months because MRI scans were available only at 6-month follow-up for four of the NGF patients, while they were available only at 12-month follow-up for the other six NGF patients. “Expected atrophy rate” reflects the average atrophy rate of each subtype from the ADNI cohort (Z-score = 0) with a two-tailed confidence interval of 1.3 as determined by the upper 10th percentile cutoff [39], represented by the orange dashed line. “Slower rate atrophy” represents the area from this cutoff and above, where some NGF AD patients had presumable less atrophy over time. Symbols correspond to NGF individuals’ atrophy rate. Color correspondence represents the limbic-predominant subtype (in green), typical AD subtype (in blue), and hippocampal-sparing (in purple)