Premium
P1‐315: Prediction of Progression Rates in AMCI Subjects Using Subject‐Specific Statistical Modelling
Author(s) -
Guerrero Ricardo,
Wolz Robin,
Rueckert Daniel,
Hill Derek L.
Publication year - 2016
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2016.06.1066
Subject(s) - post hoc analysis , cohort , post hoc , clinical trial , population , medicine , alzheimer's disease neuroimaging initiative , oncology , dementia , disease , environmental health
Background: Recent clinical trial results suggest improved enrichment and stratification of subjects could increase chance of trial success. For example, a post-hoc analysis in the Gantenerumab trial show that around 30% of the (amyloid +ve) subjects did not progress on the clinical endpoints (Lasser CTAD 15), and that a positive treatment effect was shown in those subjects that were predicted to progress more rapidly by a population-based mixed effects model incorporating clinical scales and hippocampal volume at baseline (Delor 2013, Lasser AAIC 15). We extend our recently presented subject-specific mixed effects model (Guerrero AAIC 15 and applied this to the ADNI 1 cohort prospectively stratify into those predicted to progress fast, intermediate and slow, and compared against published results from the Ganteneurmab post-hoc analysis. Methods:We used the 114 aMCI amyloid positive (CSFAbeta) subjects from the ADNI I cohort that have a clinical assessment at baseline and month 24 with MMSE, CDR-SB, ADAS-Cog13 and FAQ. The four clinical scales and hippocampal volume were used to model subject-specific disease progression. We predicted 2-year change in CDR-SB, MMSE and ADAS-Cog13 using (Guerrero AAIC 15) and validated actual progression rates in the 30% that are predicted to progress most slowly and the 35% of subjects that are predicted to progress most rapidly to allow comparison with the published Gantenerumab post-hoc analysis. Results:Table 1 shows 2-year mean change in the whole ADNI population and in the subgroups predicted to progress most rapidly and most slowly. The mean CDR-SB 2-year change in the SCarlet RoAD placebo group were presented as w1.262 and the 35% of subjects that were predicted to be fast progressors by (Delor 2013) showed a median ADAS-Cog 13 of 6 (Lasser AAIC 2015). For comparison, selecting the 35% of subjects predicted to progress most rapidly with our model, improves ADAS-Cog 13 in ADNI from 4 to 6.5. Conclusions: The subject-specific disease model accurately identifies a non-progressing sub-group in an amyloid positive aMCI population (<50% average clinical change compared to the whole population). The model efficiently identifies rapidly progressing subjects, increasing median ADAS-Cog 13 2-year change by 63%.