Detecting Cognitive Impairment in Amyloid-Positive Asymptomatic - - PowerPoint PPT Presentation
Detecting Cognitive Impairment in Amyloid-Positive Asymptomatic Populations. Jason Hassenstab, PhD Assistant Professor Neurology and Psychological & Brain Sciences Knight Alzheimers Disease Research Center Washington University in St.
Jason Hassenstab, PhD Assistant Professor Neurology and Psychological & Brain Sciences Knight Alzheimer’s Disease Research Center Washington University in St. Louis
Research Support ◉ NIH U19 AG032438 ◉ NIH U01 AG042791 ◉ NIH P01AG026276 ◉ NIH P50AG005681 ◉ NIH P01AG003991 ◉ NIH R01AG046179 ◉ NIH R01AG053267 ◉ Alzheimer’s Association ◉ GHR Foundation ◉ An Anonymous Foundation ◉ BrightFocus Foundation
Disclosures: Jason Hassenstab
Consultant/Advisory Boards Biogen Lundbeck Takeda Clinical Trials Cognition Core Director, Dominantly- Inherited Alzheimer Network-Trials Unit (DIAN-TU) I own no stocks or equities in any pharmaceutical or biotechnology company
Overview
◉ Basics of AD Prevention Trials ◉ Cognitive Decline in Preclinical to Symptomatic AD ◉ Endpoints in Active and Planned Secondary Prevention Trials ◉ Psychometric characteristics and gold standard measures ◉ Novel Approaches/Innovations
Prevention Trials in AD
Pillai & Cummings, 2013 Med Clin N Am
How does cognition change
Bateman et al., 2012 NEJM
Lim et al., 2014 Brain
Papp et al., 2015 Neuropsychologia Petersen et al., 2015 JAMA Neurol
Cog Domain ADCS-PACC APCC EPAD TOMMORROW* DIAN-TU Mental Status MMSE MMSE Orientation (5 items) MMSE Episodic Memory Logical Memory Delayed Recall RBANS List Recall RBANS Immediate Memory BVMT-R Delayed Logical Memory Delayed Recall FCSRT Total Recall RBANS Story Recall RBANS Delayed Memory CVLT-II Long Delay Recall ISLT Delayed Recall Executive Function Digit-Symbol Substitution RBANS Digit Coding Trails B Digit-Symbol Substitution Digit Span Backward Semantic Memory/Language RBANS Language Category Fluency Letter Fluency Attention RBANS Attention Digit Span Forward Trails A Other Ravens Matrices RBANS Line Orientation RBANS Visuospatial
Cognitive Endpoints in Secondary Prevention Trials
The ADCS-PACC
Measure
Ceiling? Floor? Practice Effects? MMSE Y N Anectodal (nowhere to go) FCSRT Free Recall N (rare) Y Y WMS-R Logical Memory N (rare) Y Large Digit-Symbol Substitution N (rare) N Minimal
Donohue et al., 2014 JAMA Neurology
Basics of Ceiling and Floor Effects Test is too easy Test is too difficult
MMSE: Ceiling Effect ◉ “Coin of the Realm” ◉ Has face validity for clinical meaningfulness ◉ Any decline on MMSE is typically very significant clinically Why EVER use a cognitive test with a ceiling or floor effect? Logical Memory: Floor Effect ◉ Narrative recall deficits commonly reported in AD ◉ Has face validity for clinical meaningfulness ◉ A score of zero (floor) provides meaningful (but limited) information about dementia severity.
Small baseline SD, large effect sizes, but high error ◉ Composite scores standardize to baseline (z-scores) ◉ Tests with ceiling effects will naturally produce large decline estimates (when they decline) Why EVER use a cognitive test with a ceiling effect?
DIAN-All CDR 0’s DIAN-TU (CDR 0-1) WUSTL ADRC CDR 0’s WUSTL ADRC PET-PIB+ CDR 0’s MMSE (mean, SD) 29.1 (1.2) 27.8 (3.3) 29.0 (1.2) 28.8 (1.5) Digit Symbol (mean, SD) 62.4 (12.0) 54.7 (18.2) 49.9 (11.1) 46.12 (11.3)
“Waterfall” Effects
Estimated Year to Symptom Onset
Z-score MMSE in Trial Eligible Subjects (n=146)
Estimated Year to Symptom Onset
Z-score Digit Symbol in Trial Eligible Subjects (n=146)
How do ceiling effects impact cognitive composites?
Lim et al., 2016 Alz Dem DADM
AIBL: Error in slope estimates is nearly doubled when including MMSE
How do ceiling effects impact cognitive composites?
Knight ADRC: Error in slope estimates is nearly doubled when including MMSE
Composite CDR 0 PIB-PET Negative (n=193) CDR 0 PIB-PET Positive (n=71) Mean Slope (SD) Mean Slope (SD) ADCS-PACC (FCSRT, DSST, LMDelayed, MMSE) 0.054 (0.24)
Knight ADRC PACC (FCSRT, DSST, Animal Naming, LMDelayed) 0.031 (0.13)
If Cognitive Tests Were Nearly Perfect…
What actually occurs…
Selective R Reminding: g: Fre ree R Reca call Log
ical M l Memory D y Dela layed R Recall ll Age ge
Data from Knight ADRC
Age ge
Using Composites Helps…A Bit
DIAN AN E Episodic dic Mem Memory C Composit ite EY EYO
Retest Reliabilities for Common Neuropsychological Tests
Test Pearson’s r WMS Logical Memory Delayed Recall 0.71 CVLT Trials 1-5 Total 0.72 CVLT Long Delay Free Recall 0.74 Trailmaking Test Part A 0.66 Trailmaking Test Part B 0.77 Letter Fluency (COWAT) 0.79 WAIS Letter-Number Sequencing 0.73 WAIS Digit Symbol 0.85
Calamia et al., 2013 TCN; Lo et al., 2012, J. Neuropsychology
Reliabilities in DIAN: 1-3 Years Between Assessments
Mutation Carriers CDR >0 (n = 92) Mutation Carriers CDR 0 (n = 167) Non-Carriers (n = 171) r (ICC) r (ICC) r (ICC) DIAN Word List Delayed Recall 0.54 0.57 0.40 WMS Logical Memory Immediate Recall 0.81 0.53 0.51 WMS Logical Memory Delayed Recall 0.85 0.58 0.53 Pair Binding (Assoc. Memory) 0.57 0.51 0.58 Trailmaking Test Part A 0.66 0.76 0.42 Trailmaking Test Part B 0.74 0.72 0.58 WAIS Digit Symbol 0.78 0.84 0.88 Letter Fluency (COWAT) 0.84 0.80 0.80 Category Fluency (Animals + Vegetables) 0.75 0.74 0.61
Reliab abiliti ties es i in Knight t ADRC: Annual al Asses essmen ents ts
Amyloid Positive CDR 0 (n = 95) Amyloid Negative CDR 0 (n = 155) Test r (ICC) r (ICC) Free and Cued SRT: Free Recall
0.60 0.57
WMS Logical Memory Immediate Recall
0.58 0.54
WMS Logical Memory Delayed Recall
0.53 0.59
Associate Memory
0.46 0.67
Trailmaking Test Part A
0.67 0.61
Trailmaking Test Part B
0.63 0.71
WAIS Digit Symbol
0.88 0.87
Animal Fluency
0.47 0.68
Vegetable Fluency
0.58 0.63
Courtesy Kathryn Papp, MGH. Presented at AAIC 2018
Courtesy Kathryn Papp, MGH. Presented at AAIC 2018
Within-Subject Repeated Measures Tasks
Shorter-term repeated measurement tasks: ◉Accelerated Long-Term Forgetting (ALF) ◉Associate Learning (Online Repeatable Cognitive Assessment; ORCA) ◉Measurement “Burst” Designs
Limitations of Traditional Cognitive Assessments
Figure courtesy of Martin Sliwinski, PhD
Weston et al. 2018 Lancet Neurology
Immediate Recall 30-minute Recall 7-day Recall Recognition
Uses a 7-day long term recall paradigm to look at % retention
Relative to n = 14 noncarriers, n = 21 CDR 0 (EYO -7.2y) carriers had 30% lower retention of word list, 20% lower story retention. **No differences on standard tests between groups
Accelerated Long Term Forgetting
Participants are played a sound and then shown a picture. They guess whether it is a match or not. Test is completed in about ~20mins/day for 6 consecutive days. The proportion of matches vs non- matches increases each day such that associations are learned implicitly. **This version uses Mandarin characters, but can be done with other pictures and words.
Example of the training procedure.
Associate Learning: Online Repeated Cognitive Assessment (ORCA)
From AIBL: n = 80 (n = 20 AB+, n = 30 AB-) CDR 0s completed task plus n = 30 Controls (ages 18-40). Significant interaction of group by
AB+ had reduced learning. Importantly: No difference between AB+/AB- on standard memory measures.
Associate Learning: Online Repeated Cognitive Assessment (ORCA)
Different rates of memory acquisition in young controls, Aβ- older adults, and Aβ+
Review
Cohen’s d = 1.9
M T W Th F S Su
7-day “Burst” of Cognitive Assessments
Time 7a 8a 9a 10a 11a 12p 1p 2p 3p 4p 5p 6p 7p 8p 9p 10p Test Test Test Test Test Test Test Test Test Test Test Test Test Test Test Test Test Test Test Test Test Test Test Test Test Test Test Test
Measurement “Burst” Design: Ambulatory Research in Cognition (ARC)
Reliab ability ty of ARC M Meas asures res
All Participants Symbols (n = 208) Grids (n = 208) Prices (n = 113) Reliability of 1 Test (ICC)
0.80 0.38 0.34
Reliability 1 day (4 tests)
0.91 0.67 0.73
Reliability 2 day (~7 tests)
0.96 0.78 0.81
Reliability 5 day (~16 tests)
0.99 0.91 0.90
Reliability 7 day (~22 tests)
0.99 0.93 0.92
ARC Biomarker Correlations: CSF Amyloid. 100% Cognitively Normal (CDR 0)
CSF Amyloid Beta1-42 and ARC smartphone assessments (n = 39). r = 0.24, p = 0.13 CSF Amyloid Beta1-42 and standard In-clinic assessments (n = 37). r = 0.06, p = 0.72
ARC Biomarker Correlations: CSF Tau. 100% Cognitively Normal (CDR 0)
CSF Total Tau and ARC smartphone assessments (n = 39). r = -0.34, p = 0.03 CSF Total Tau and standard In-clinic assessments (n = 37). r = -0.06, p = 0.69
Conclusions
Detecting amyloid-positivity in normal adults is difficult! ◉Most “one-shot” composites include multiple measures of memory and mental status exams. ◉In cognitive normal populations, avoid measures with restricted range, ceiling effects in particular (i.e. MMSE). ◉Reliability of standard measures is poor, hence large sample sizes typically needed in cognitively normal samples.
Conclusions
◉Within-subject repeated measures provide LARGE improvements in reliability. More likely to be successful in differentiating amyloid-positive from amyloid-negative. ◉But, they take more time, resources, and are subject to poor adherence.
Acknowledgements
Andrew Aschenbrenner, PhD Yen Ying Lim, PhD Paul Maruff, PhD Kathryn Papp, PhD Eric McDade, DO Guoqiao “Peter” Wang, PhD John C. Morris, MD Randall Bateman, MD
DIAN Participants and Families Knight ADRC Participants and Families