1. Lifecourse progression of AD or modeling of AD biomarker trajectories: history and pitfalls
Clifford R Jack Jr MD
Professor of Radiology
The Alexander Family Professor
of Alzheimer's Disease Research
Mayo Clinic, Rochester, MN

2. Acknowledgements
Funded in part by Grant R13 AG from the National Institute on Aging
RO1 AG011378
RO1 AG041851
U01 AG06786
Alexander Family Professorship in Alzheimer's disease research
GHR Foundation
The views expressed in written conference materials or publications and by speakers and moderators do not necessarily reflect the official policies of the Department of Health and Human Services; nor does mention by trade names, commercial practices, or organizations imply endorsement by the U.S. Government.

3. outline AD biomarkers Motivation for 2010 Lancet Neurology model
2013 refinements of 2010 model
Evidence since publication of 2010 model
Non-AD pathology in elderly and SNAP
Summary

4. AD Biomarkers are proxies for AD pathophysiology: 6 Major – “2” categories
Measures of brain A deposition – amyloid plaques
Amyloid PET
CSF AB 42 – low
Measures of Neruofibrillary tangles (tau)
CSF tau (t-tau and p-tau) – high
Tau PET
Measures of Neurodegeneration (progressive loss of neurons or processes with corresponding impairment in neuronal function)
FDG PET – AD signature hypo metabolism
Structural MRI - AD signature atrophy

5. MRI AD-signature FDG AD-signature
Jack et al Nat Neurol Rev in press

6. specificity for AD pathology: ranking
amyloid PET > CSF Ab 42, CSF p-tau > tau PET> CSF total tau > FDG and MRI
FDG and MRI
sensitive markers of neurodegeneration,
correlate very well with cognition
but neurodegeneration not specific to AD
Atrophy and hypometabolism not specific to AD

7. Hippocampal Volume vs CA1 neuron counts
Bobinski, Neuroscience 95, 2000
Zarow, Ann Neurol, 2005

8. STAND algorithm for Individual Subject Diagnosis Vemuri, Neuroimage 2008; 39(3): and Neuroimage 2008; 42(2):559-67

9. Hippocampal W score by diagnosis in those with a single path Dx
HS = hippocampal sclerosis; DLBD = diffuse Lewy body disease;
FTD = frontotemporal degeneration; NFT = Neurofibrillary tangle-only dementia
Jack Neurology 2002

10. outline AD biomarkers Motivation for 2010 Lancet Neurology model
Order of biomarker events
Shape of curves
2013 refinements of 2010 model
Evidence since publication of 2010 model
Non-AD pathology in elderly and SNAP
Summary

11. Dominate theory of AD pathogenesis since 1990s
Sporadic AD: Failure of Ab42 clearance
The Amyloid Hypothesis of Alzheimer’s Disease: Progress and Problems on the Road to Therapeutics
Hardy and Selkoe, Science 2002
Dominate theory of AD pathogenesis since 1990s

12. Cross sectional dissociation – time and location
Problems with amyloid cascade hypothesis: state of biomarker studies ~ 2008/2009 was confusing
Cross sectional dissociation – time and location
direct relationship between neurodegenerative biomarker magnitude/topography and symptoms
indirect relationship between amyloid biomarkers and symptoms: 30% CN abnormal, topographic dissociation
Longitudinal dissociation
change in cognition closely coupled to rate of neurodegeneration
not to rate of amyloid deposition

13. AD vs. Cog Normal: Topography Jack et al, Brain 20085 4 3 2 1
p < 0.005
(unc)
MRI

14. Paradox
Atrophy, FDG hypo metabolism, CSF tau, symptoms group together in location and time, amyloid does not
Genetics all point to Ab as causative
early onset AD: Down syndrome and all known autosomal-dominant mutations - increase production of Ab42 or all Ab species
Late onset AD: APOE 4 facilitates Ab deposition
Protective genetics: APOE 2, Icelandic mutation
primary tauopathies lead to FTLD, CBD, PSP but never to pathological AD

15. “solution”: modified amyloid cascade
b-amyloid facilitates spread of tau, effect of b-amyloid on cognition is indirect
AT  N  C
A is the upstream driver of TNC sequence although topology of A and TNC differ
Time shifts or ordering - biomarkers become abnormal in an ordered but temporally overlapping manner

16. amyloid precedes tauopathy/neurodegeneration, effect of amyloid on cognition is indirect
Inglesson & Hyman, Neurology 2004
Jack et al, Brain 2008 & 2009
Mormino & Jagust, Brain 2009
Perrin & Holztman, Nat Rev 2009

17. Lancet Neurol 2010
Ab Amyloid = CSF Ab42 or amyloid PET imaging; Tau Mediated Neuron Injury and Dysfunction = CSF tau or FDG PET; Brain Structure = structural MRI

18. Sigmoid shape

19. Rate of atrophy accelerates as approach dementia Chan and Fox, Lancet 2003

20. Annual change in PIB SUVR and ventricular volume by clinical diagnosis
Jack et al , Brain 2009

21. Lancet Neurol 2010
Ab Amyloid = CSF Ab42 or amyloid PET imaging; Tau Mediated Neuron Injury and Dysfunction = CSF tau or FDG PET; Brain Structure = structural MRI

22. outline AD biomarkers Motivation for 2010 Lancet Neurology model
2013 refinements of 2010 model
Evidence since publication of 2010 model
Non-AD pathology in elderly and SNAP
Summary

23. Lancet Neurology, Feb, 2013

24. Modulators of Biomarker Temporal Relationships (Fig 5) C- = cognition in the presence of co-morbid pathologies (e.g., Lewy bodies or vascular disease) or risk amplification genes, C+ = cognition in subjects with enhanced cognitive reserve or protective genes, Co = cognition in subjects without co-morbidity or enhanced cognitive reserve.
Jack et al Lancet Neurol 2010

25. medial temporal tauopathy often occurs without (“before”) Aβ deposition at autopsy (discussed but not incorporated into 2010 model)
Isolated medial temporal tauopathy - brain stem, entorhinal cortex, hippocampus, tauopathy is common in middle age and older subjects (as young as 6yo) with no amyloid plaques - Braak 1997, 2011; Price and Morris, Annals Neurol 1999; Haroutunian, Arch Neurol 1999

27. Late onset AD - MTL tauopathy precedes b-amyloid
Lancet Neurology, Feb, 2013
Late onset AD - MTL tauopathy precedes b-amyloid

28. outline AD biomarkers Motivation for 2010 Lancet Neurology model
2013 refinements of 2010 model
Evidence since publication of 2010 model
Non-AD pathology in elderly and SNAP
Summary

29. AD pathology in young vs old
Young – pathologically pure (except LB)
Old – plaques and tangles superimposed changes of
non AD pathologies
CVD
Non AD tauopathies – PART, grains, CTE, rarely PSP, CBD, and FTLD LB
Hipp sclerosis
TDP43
aging

30. DIAN, Bateman et al NEJM 2012
Cross sectional, years from parental age onset, difference between carriers and non-carriers

31. Evidence for temporal ordering of AD biomarkers model DIAN, Benzinger et al PNAS, 2012 (Mutation carries, PIB n = 121; FDG n= 116; MRI n= 137)
Subcortical MRI ROI analyses – hippocampus, amygdala, N accumbens all - 10 yrs in carriers

32. Fleisher et al JAMA Neurol 2015

33. Modeling studies in elderly
Support model
Caroli, 2010
Jack, 2011
Buchave, 2012
Villemagne, 2013
Young, 2014
Donohue, 2014
Do not support model
Jedynak, 2012

34. Cerebrospinal Fluid Levels of beta-Amyloid 1-42, but Not of Tau, Are Fully Changed Already 5 to 10 Years Before the Onset of Alzheimer Dementia Buchhave et al, Arch Gen Psych 2012

35. Brain Beta Amyloid Load Approaches a Plateau Jack et al, Neurology 2013

36. Brain Beta Amyloid Load Approaches a Plateau Jack et al, Neurology 2013
Relating the inverted U-shaped amyloid rates as a function of baseline SUVR to sigmoid
shaped trajectory of amyloid accumulation with time – red = 195; blue = 158

37. Sigmoid shape - amyloid
Villemange amyloid PET
Landau and Jagust amyloid PET
Shaw CSF

38. outline AD biomarkers Motivation for 2010 Lancet Neurology model
2013 refinements of 2010 model
Evidence since publication of 2010 model
Non-AD pathology in elderly and SNAP
Summary

40. NIA-AA Preclinical AD staging in relation to 2010 Lancet Neurol biomarker model Jack, Annals, 2012

41. Operationalize the NIA-AA criteria
Objectives
Operationalize the NIA-AA criteria
How do cognitively normal subjects (n=450) in MCSA distribute in the NIA-AA scheme?
Annals Neurol 2012

42. Distribution of 450 CN in MCSA by NIA-AA Preclinical Stage
Annals, 2012
0 – 43%; %; 2 – 12%; 3 – 3%’; SNAP – 23%; Unclassif – 4%

43. CSF Ab42 and tau (either ptau or ttatu)
Characteristics of SNAP subjects in different study cohorts Cognitively normal subjects
Source
Population characteristics
Biomarkers used for classification
Number (%) of SNAP/ number in cohort
Age in SNAP group
Number (%) of men in SNAP group
Number (%) of APOE4 in SNAP group
Clinical outcome
Follow up time across cohort
reference
Mayo Clinic Study of Aging
population based, cognitively normal
Amyloid PET, FDG PET or hipp volume
103 (23%) / 450
79 yrs
(IQR 76,84)
62 (60%)
12 (13%)
Jack et al 2012
69 (23%) / 296
81 yrs
43 (62%)
8 (12%)
CN to MCI or dementia:
st 0 – 5%,
st 1 – 11%,
st 2 – 21%,
st 3 – 43%,
SNAP – 10%
1.3 yrs
(range 1.1 – 5.1)
Knopman et al 2012
Washington University
community dwelling, cognitively normal
CSF Ab42 and tau (either ptau or ttatu)
72 (23%) / 311
73.6 yrs
(SD 5.8)
30 (40%)
22 (31%)
CDR 0 to >= .5 AD dementia:
st 0 – 2%,
st 2 – 26%,
st 3 – 56%,
SNAP – 5%
3.9 yrs
(range 1 – 15)
Vos et al 2013
CDR 0 to >= 0.5: survival
HR A-N- ref
A+N- 2.58
A+N+ 8.41
SNAP 1.12
3.70 yrs
(SD 1.46)
Roe et al 2013
Berkeley Aging Cohort
Amyloid PET, FDG PET, hipp volume and MRI cortical ROI
19 (26%) / 72
Wirth et al 2013
Harvard Aging Brain Study
38 (23%) / 166
(IQR 75,82)
248 (63%)
7 (19%)
SNAP greater decline than stage 0, less decline than A+N+
2.09 yrs
(IQR 1.9 – 2.3)
Mormino et al 2014
ADNI
clinical trial sites, cognitively normal
CSF Ab42, and either CSF tau or hipp volume
54 (23%) / 238
survival HR stage 0 ref
st 1 – 2.6,
st 2 – 1.8,
st 3 – 11.3,
SNAP – 2.4
6 yrs
(IQR 3.0 – 7.0)
Toledo et al 2014
Amsterdam Dementia Cohort
memory clinic,
subjective memory complaint
CSF Ab42 and tau
31 (23%) / 132
st 0 – 3%,
st 1 – 18%,
st 2 – 60%,
SNAP – 10%
1.8 yrs
(SD 1.3)
Van Harten et al 2013

44. Characteristics of SNAP subjects in different study cohorts Impaired subjects
source
Population characteristics
Biomarkers used for classification
Number (%) of SNAP/ number in cohort
Age in SNAP group
Number (%) of men in SNAP group
Number (%) of APOE4 in SNAP group
Clinical outcome
Follow up time across cohort
reference
3- site European consortium
memory clinics,
MCI
CSF Ab42, hipp volume, FDG PET
15 (20%) / 73
MCI to dementia:
A-N- 5%
A+N- 27%
A+N+ 100%
SNAP 47%
Progressor MCI 23.3 (2-76) mo.
stable MCI 31.8 (12-84) mo
Prestia et al 2013
Mayo Clinic Study of Aging
Population based, MCI
Amyloid PET, FDG PET or hipp volume
36 (29%) / 126
82 yrs
(IQR 78,85)
28 (78%)
4 (11%)
A-N- 8%
A+N- 0
A+N+ 16%
SNAP 21%
15 mo
Petersen et al 2013
ADNI
clinical trial sites, MCI
10 (17%) / 58
77 yrs
(IQR 73,83)
7 (70%)
4 (40%)
A-N- 11%
A+N+ 42%
SNAP 25%
12 mo
clinical trial, 7 European sites
memory clinics, MCI
Amyloid PET, medial temporal atrophy
7 (35%) / 20
Duara et al 2013
clinical trial sites, AD dementia
Amyloid PET, CSF Ab42, FDG PET or hipp volume
6 (7%) / 92
Lowe et al 2013
ADNI plus 4- site European consortium
clinical trial sites and memory clinics, MCI
34(17%) / 201
70.6 yrs
(SD 9.2)
23 (68%)
10 (31%)
Percent progressors*:
A+N- 34%
A+N+ 71%
SNAP 56%
26.4 months (SD 16.8)
Among SNAP
Caroli et al 2015
ADNI plus multi- site European consortium
CSF Ab42 and tau, medial temp atrophy, or FDG PET
220 (29%) / 776
69.4 yrs
(SD 8.3)
116 (53%)
62 (32%)
progression at last follow up: to AD dementia 21%,
to non AD dementia 10%
2.5 (SD1.3) yrs
Vos et al 2015

45. SNAP
SNAP is a biomarker based construct denoting amyloid negative neurodegeneration positive individuals
Common in cognitively normal elderly (roughly 23%) and in mild cognitive impairment (roughly 25%)
APOE4 is markedly underrepresented compared to amyloid positive individuals (A+N- and A+N+) individuals
Suspected to be pathologically heterogeneous, composed of a variety of non-AD etiologies common in aging

46. Non AD pathologies in elderly
CVD
Non AD tauopathies – PART, grains, CTE, rarely PSP, CBD, and FTLD LB
Hipp sclerosis
TDP43
“aging”

47. outline AD biomarkers Motivation for 2010 Lancet Neurology model
2013 refinements of 2010 model
Evidence since publication of 2010 model
Non-AD pathology in elderly and SNAP
Summary

48. Difficulties with empiric modeling of AD with biomarkers in elderly
Can not measure full extent of disease (creates x axis and y axis problem)
Non-AD pathology (SNAP)
Proportional etiological substrates of neurodegeneration and cognitive impairment unknown
Mixed pathology
No specific biomarkers for important pathologies
Neurodegeneration and its biomarkers not specific for AD
Account for aging effects on cognition/neurodegeneration

49. Lancet Neurol 2010
Ab Amyloid = CSF Ab42 or amyloid PET imaging; Tau Mediated Neuron Injury and Dysfunction = CSF tau or FDG PET; Brain Structure = structural MRI

50. Mixed AD: proportional etiological substrates of neurodegeneration are unknown and vary from person to person
Amyloid-first sequence
Neurodegeneration-first sequence
Jack and Holtzman, Neuron 2013