CERAD neuropsychological battery and structural magnetic resonance imaging in the detection of mild cognitive impairment and prodromal Alzheimer's disease

(1)

DISSERTATIONS | TEEMU PAAJANEN | CERAD NEUROPSYCHOLOGICAL BATTERY AND STRUCTURAL... | No 421

uef.fi

PUBLICATIONS OF

THE UNIVERSITY OF EASTERN FINLAND Dissertations in Health Sciences

ISBN 978-952-61-2504-6 ISSN 1798-5706

Dissertations in Health Sciences

THE UNIVERSITY OF EASTERN FINLAND

TEEMU PAAJANEN

CERAD NEUROPSYCHOLOGICAL BATTERY AND STRUCTURAL MAGNETIC RESONANCE IMAGING IN THE DETECTION OF MILD COGNITIVE IMPAIRMENT AND PRODROMAL ALZHEIMER’S DISEASE

Alzheimer’s disease (AD) is the most common neurodegenerative disease leading to dementia

and thus a major public health challenge worldwide. Subjects with mild cognitive impairment (MCI) have an increased risk for

developing AD, however, there is no single diagnostic test for AD. Early detection of AD is important because current treatments alleviate symptoms and help to maintain daily functions.

In this thesis CERAD neuropsychological battery and brain volumes and cortical thickness were examined in the detection of MCI and prodromal

AD in a large prospective AddNeuroMed study.

TEEMU PAAJANEN

(2)

(3)

CERAD Neuropsychological Battery and Structural Magnetic Resonance Imaging in the Detection of Mild Cognitive Impairment

and Prodromal Alzheimer’s Disease

(4)

(5)

TEEMU PAAJANEN

CERAD Neuropsychological Battery and Structural Magnetic Resonance Imaging in the Detection of Mild Cognitive Impairment

and Prodromal Alzheimer’s Disease

To be presented by permission of the Philosophical Faculty, University of Eastern Finland for public examination in MD100, Mediteknia building, Kuopio, on Friday, May 26^th 2017 at 12 noon

Publications of the University of Eastern Finland Dissertations in Health Sciences

Number 421

Department of Neurology, Institute of Clinical Medicine School of Medicine, Faculty of Health Sciences

and

Department of Education and Psychology Philosophical Faculty

University of Eastern Finland Kuopio

2017

(6)

Grano Oy Jyväskylä, 2017

Series Editors:

Professor Tomi Laitinen, M.D., Ph.D.

Institute of Clinical Medicine, Clinical Radiology and Nuclear Medicine Faculty of Health Sciences

Professor Hannele Turunen, Ph.D.

Department of Nursing Science Faculty of Health Sciences

Professor Kai Kaarniranta, M.D., Ph.D.

Institute of Clinical Medicine, Ophthalmology Faculty of Health Sciences

Associate Professor (Tenure Track) Tarja Malm, Ph.D.

A.I. Virtanen Institute for Molecular Sciences Faculty of Health Sciences

Lecturer Veli-Pekka Ranta, Ph.D. (pharmacy) School of Pharmacy

Faculty of Health Sciences

Distributor:

University of Eastern Finland Kuopio Campus Library

P.O.Box 1627 FI-70211 Kuopio, Finland http://www.uef.fi/kirjasto

ISBN (print):978-952-61-2504-6 ISBN (pdf):978-952-61-2505-3

ISSN (print):1798-5706 ISSN (pdf):1798-5714

ISSN-L: 1798-5706

(7)

Author’s address: Department of Education and Psychology, and

Department of Neurology, Institute of Clinical Medicine University of Eastern Finland

KUOPIO FINLAND

Supervisors: Professor Hilkka Soininen, MD, Ph.D.

Department of Neurology

Institute of Clinical Medicine, School of Medicine University of Eastern Finland

KUOPIO FINLAND

Adjunct Professor Tuomo Hänninen, Ph.D.

Department of Neurology Kuopio University Hospital KUOPIO

FINLAND

Professor Hannu Räty, Ph.D.

Department of Education and Psychology University of Eastern Finland

JOENSUU FINLAND

Reviewers: Professor Emeritus Ove Almkvist, Ph.D.

Department of Psychology Stockholm University STOCKHOLM SWEDEN

Adjunct Professor Eero Vuoksimaa, Ph.D.

Institute for Molecular Medicine Finland University of Helsinki

HELSINKI FINLAND

Opponent: Adjunct Professor Mira Karrasch, Ph.D.

Department of Psychology Åbo Akademi University TURKU

FINLAND

(8)

(9)

Paajanen, Teemu

CERAD Neuropsychological Battery and Structural Magnetic Resonance Imaging in the Detection of Mild Cognitive Impairment and Prodromal Alzheimer’s Disease

University of Eastern Finland, Philosophical Faculty

Publications of the University of Eastern Finland. Dissertations in Health Sciences 421. 2017. 86 p.

ISBN (print): 978-952-61-2504-6 ISBN (pdf): 978-952-61-2505-3 ISSN (print): 1798-5706 ISSN (pdf): 1798-5714 ISSN-L: 1798-5706

ABSTRACT:

Neurodegenerative diseases leading to dementia are a major public health challenge all over the world. Alzheimer’s disease (AD) is the most common progressive neurodegenerative disease and its’ prevalence is known to increase nearly exponentially with old age. Typically, the clinical manifestation of AD starts with a slowly progressive episodic memory impairment followed by a decline in other cognitive functions, which later lead to dementia.

Even though the AD related neuropathological brain changes are known to start several years before the appearance of the first clinical symptoms, there is no single biological diagnostic test for AD nor is there any curative medication available. However, early detection of AD is highly important because current symptomatic treatments alleviate clinical symptoms and help the individual to maintain daily functions. Recent developments in biomarkers have been promising and biomarkers have been incorporated to the new diagnostic criteria for predementia phase of AD. Despite the progress in imaging and other biomarkers, cognitive deficits measured with neuropsychological tests still represent the core of the AD diagnosis. Neuropsychological tests have been shown to be sensitive and practical in detecting mild cognitive impairment (MCI) and AD. However, new more sensitive and better validated cognitive measures together with novel biomarkers are needed for identifying subjects already in the prodromal phase of AD.

The aim of this thesis was to explore cognitive performance and brain morphology among subjects with MCI and prodromal AD. One special emphasis was to examine the CERAD Neuropsychological Battery (NB) composite scores and automated structural magnetic resonance imaging (MRI) measures of cortical thickness and regional volumes for detecting subjects with prodromal AD. This thesis is based on the 1 year prospective data of the AddNeuroMed study investigating AD, MCI and healthy control participants recruited in six European countries. We found that CERAD-NB total scores were more accurate than the subtests and Mini-Mental State Examination in detecting MCI and prodromal AD in this multinational population. In the large imaging subgroup, cognitive measures were more accurate than structural MRI in differentiating MCI subjects from controls, but regional brain volumes were better at predicting MCI to AD conversion. The two newly designed episodic memory composite scores for CERAD-NB were the most accurate cognitive measures for detecting MCI to AD conversion in the larger clinical cohort. We also showed that cortical thickness signatures of CERAD total scores reflected accurately the atrophy pattern of prodromal AD. In conclusion, this doctoral thesis presents evidence that CERAD total scores reflect morphological brain changes of prodromal AD and can increase accuracy of CERAD- NB in detecting AD already during the MCI stage. The raw score based CERAD total score and the new Memory Total Score represent useful additions to the CERAD-NB, and are applicable in both scientific and clinical settings.

National Library of Medicine Classification:WT 155;WL 141.5.N46; WL 141.5.M2

Medical Subject Headings: Alzheimer’s disease, mild cognitive impairment, neuropsychology, CERAD, cognition, memory, magnetic resonance imaging, brain volumes, cortical thickness

(10)

(11)

Paajanen, Teemu

CERAD-kognitiivinen tehtäväsarja ja magneettikuvantaminen lievän kognitiivisen heikentymän ja varhaisen Alzheimerin taudin tunnistamisessa

Itä-Suomen yliopisto, Filosofinen tiedekunta

Publications of the University of Eastern Finland. Dissertations in Health Sciences 421. 2017. 86 s.

ISBN (print): 978-952-61-2504-6 ISBN (pdf): 978-952-61-2505-3 ISSN (print): 1798-5706 ISSN (pdf): 1798-5714 ISSN-L: 1798-5706

TIIVISTELMÄ:

Dementiaan johtavat muistisairaudet ovat suuri kansanterveydellinen haaste kaikkialla maailmassa. Alzheimerin tauti (AT) on yleisin etenevä muistisairaus ja sen esiintyvyys kasvaa voimakkaasti väestön ikääntyessä. Tyypillisesti AT:n oireet alkavat hitaasti etenevänä tapahtumamuistin heikkoutena, jota seuraa muiden tiedonkäsittelyn alueiden heikentyminen, ja lopulta dementoituminen. AT:n neuropatologiset aivomuutokset alkavat useita vuosia ennen ensimmäisiä kliinisiä oireita, eikä sairauteen ole tällä hetkellä olemassa yksittäistä biologista diagnostista testiä, eikä parantavaa hoitoa. AT:n varhainen tunnistaminen on kuitenkin tärkeää, koska nykyisillä lääkkeillä voidaan lievittää oireita ja ylläpitää toimintakykyä. Viimeaikainen kehitys biomarkkereissa on ollut lupaavaa ja uudet näitä hyödyntävät varhaisen (=prodromaali) AT:n diagnostiset kriteerit on julkaistu.

Aivokuvantamisessa ja muissa biomarkkereissa tapahtuneesta edistyksestä huolimatta neuropsykologisilla testeillä todennetut vaikeudet muodostavat edelleen AT:n diagnostiikan ytimen. Neuropsykologisten testien on osoitettu olevan herkkiä ja käytännöllisiä lievän kognitiivisen heikentymän (MCI) ja AT:n tunnistamisessa. Uusia herkempiä ja validoituja kognitiivisia mittareita yhdessä biomarkkereiden kanssa kuitenkin tarvitaan, jotta sairastuneet voitaisiin tunnistaa jo AT:n prodromaalissa vaiheessa.

Tämän väitöskirjan tavoitteena oli tutkia kognitiivista suoriutumista ja aivojen rakennetta MCI:ssä ja prodromaalissa AT:ssa. Erityisenä kiinnostuksena olivat CERAD-tehtäväsarjan yhdistelmäpistemäärät ja automaattisesta rakenteellisesta magneettikuvauksesta (MRI) saadut aivoalueiden tilaavuudet ja aivokuoren paksuudet prodromaalin AT:n tunnistamisessa. Tutkimus perustuu ison prospektiivisen AddNeuroMed-tutkimuksen 1 vuoden seuranta-aineistoon, jossa oli AT-potilaita, MCI-henkilöitä, sekä terveitä verrokkeja kuudesta Euroopan maasta. Tutkimuksessa osoitimme CERAD-kokonaispistemäärien olevan yksittäisiä osatehtäviä ja Mini-Mental State Examination-testiä tarkempia MCI:n ja prodromaalin AT:n tunnistamisessa. Isossa MRI-otoksessa havaitsimme kognitiivisten testien olevan MRI-muuttujia parempia erottelemaan MCI-henkilöt verrokeista, mutta aivoalueiden tilavuudet ennustivat tarkemmin MCI henkilöiden sairastumista AT:iin. Kaksi uutta CERAD:sta laskettavaa episodisen muistin yhteispistemäärää osoittautuivat suuremmassa kliinisessä aineistossa tarkimmiksi menetelmiksi tunnistamaan ne MCI- henkilöt, jotka myöhemmin sairastuivat AT:iin. Lisäksi osoitimme CERAD- kokonaispistemäärien heijastavan aivokuoren paksuutta varsin tarkasti samoilla aivoalueilla, jotka atrofioituvat varhaisessa AT:ssa. Kokonaisuutena tulokset osoittavat, että CERAD-kokonaispistemäärät heijastavat varhaiseen AT:iin liittyviä aivojen rakenteellisia muutoksia ja parantavat testistön tarkkuutta AT:n tunnistamisessa jo MCI-vaiheessa.

Raakapisteisiin perustuvat kokonaiskognition ja muistin yhteispistemäärät ovat hyödyllinen lisä CERAD-testistöön ja sovellettavissa sekä tieteellisessä että kliinisessä käytössä.

Luokitus: WT 155;WL 141.5.N46; WL 141.5.M2

Yleinen suomalainen asiasanasto: Alzheimerin tauti, lievä kognitiivinen heikentymä, neuropsykologia, CERAD, kognitio, muisti, magneettikuvantaminen, aivoalueiden tilavuudet, aivokuoren paksuus

(12)

(13)

Acknowledgements

This doctoral thesis was carried out between 2009-2017 in the Department of Neurology and Department of Education and Psychology of the University of Eastern Finland, and Department of Neurology and Clinical Radiology, Kuopio University Hospital.

I want to express my sincere thanks to all of the people who have participated in this work, including the large number of researchers and colleagues from the AddNeuroMed consortium who contributed to the collection and management of the clinical and MRI data in seven European countries.

In particular, I wish to express my deepest gratitude to my supervisors.

My principal supervisor, Professor Hilkka Soininen, for her always precise, up-to-date and invaluable guidance during this project. Thank you for giving me the opportunity to join this eminent multinational workgroup and for the trust you placed in me when I started as a young researcher. It has been truly inspiring to learn neurology and neuroimaging under the guidance of such an expert. Your support has been crucial for the success of this work.

Adjunct Professor Tuomo Hänninen, for all the expert guidance regarding the neuropsychology of neurodegenerative disorders. I have been more than grateful to have you as my supervisor.

Professor Hannu Räty for all the support and practical guidance, especially during the final parts of this project.

I owe my deepest gratitude to Adjunct Professor Yawu Liu for the scientific collaboration and guidance in the field of MR imaging. In addition, I want to thank Dr. Andy Simmons and Dr. Andy Aitken from the King’s College London for all the invaluable help with cortical thickness analyses.

I want to express my warm thanks to the official reviewers, Professor Emeritus Ove Almkvist from Stockholm University and Adjunct Professor Eero Vuoksimaa from the University of Helsinki for the valuable comments that improved the final thesis.

I would like to thank all the co-authors and researchers with whom I had opportunity to collaborate. Especially I would like to thank the principal investigators of the AddNeuroMed centers; Professor Simon Lovestone from King’s College London, Professor Tomasz Sobow from the Medical University of Lodz, Professor Patrizia Mecocci from the University of Perugia, Professor Magda Tsolaki from the Aristotle University of Thessaloniki and Professor Bruno Vellas from the Paul Sabatier University. In addition, I am grateful to Professor Lars- Olof Wahlund and Associate Professor Eric Westman from Karolinska Institutet who were responsible for the MR imaging data coordination center in Stockholm.

I want to express my warmest gratitude to the research director, neurologist, Dr. Merja Hallikainen who worked as the group leader at the clinical practice level in the Kuopio AddNeuroMed team. Thank you for the supportive work environment and interesting discussions of science and all other issues in life. I want to thank my psychologist colleagues Ilona Hallikainen PhD and Susanna Tervo MPsych who had initially started the collection of AddNeuroMed data at the Kuopio site and introduced me to the cognitive part of this project.

I also want to thank Tarja Lappalainen, Helena Mäkelä, Sirkka Tanskanen, Seija Hynynen,

(14)

Marja-Leena Soininen and all the people who participated in the AddNeuroMed project data collection. In addition, I want to thank Docent Seppo Helisalmi, Docent Sanna-Kaisa Herukka, MD Lasse Nieminen, Noora Suhonen, MPsych, Valtteri Julkunen, PhD, Mr.

Markku Kalinen and all the great people with whom I have had privilege to work in the Brain Research Unit during the years in Kuopio.

I want to thank Marja-Leena Hannila and Vesa Kiviniemi for the guidance in the statistical analyses and Ewen MacDonald for the language revision of this thesis. I also want to thank my dear friend Janne Ulvinen for the peer support and guidance in the use of image layout program. I want to express my gratitude also to Mari Tikkanen, Esa Koivisto, Tuija Parsons and all the UEF neurology folk for creating such convivial surroundings in which to do the scientific work.

I want to thank my father Jussi and mother Pirjo for the support and their always encouraging attitude, supporting me in all my decisions throughout my life. I also want to thank Anneli, my mother-in-law, for the help with my family during these years.

I am grateful to my lovely wife Noora and my two splendid kids, Lenny and Lumi, for love and support during this long project. You kept my feet firmly on the ground and reminded me of all those issues in life that ultimately matter.

Finally, I express my thanks to all patients and caregivers who participated in the AddNeuroMed project and made this study possible.

Helsinki, April 2017

Teemu Paajanen

AddNeuroMed project was funded by the European Union, AddNeuroMed/Innovative Medicines LSHB-CT-2005-518170 (2005-2008), Health Research Council of The Academy of Finland, grant 121038, and EVO grants 5772709 and 5772720 from Kuopio University Hospital. This thesis was funded by EVO grant 53/2010 from Kuopio University Hospital;

Finnish Brain Research and Rehabilitation Centre – Neuron; Finnish Brain Foundation;

Instrumentarium Science Foundation; Finnish Cultural Foundation, North Savo Regional fund and University of Eastern Finland.

(15)

List of the original publications

This dissertation is based on the following original publications:

I Paajanen T, Hänninen T, Tunnard C, Mecocci P, Sobow T, Tsolaki M, Vellas B, Lovestone S, Soininen H and AddNeuroMed Consortium. CERAD

neuropsychological battery total score in multinational mild cognitive impairment and control populations: the AddNeuroMed study. Journal of Alzheimer’s Disease 22:1089-1097, 2010.

II Liu Y, Paajanen T, Zhang Y, Westman E, Wahlund LO, Simmons A, Tunnard C, Sobow T, Mecocci P, Tsolaki M, Vellas B, Muehlboeck S, Evans A, Spenger C, Lovestone S, Soininen H and AddNeuroMed Consortium. Analysis of regional MRI volumes and thicknesses as predictors of conversion from mild cognitive impairment to Alzheimer's disease. Neurobiology of Aging 31:1375-1385, 2010.

III Paajanen T, Hänninen T, Aitken A, Hallikainen M, Westman E, Wahlund LO, Sobow T, Mecocci P, Tsolaki M, Vellas B, Muehlboeck S, Spenger C, Lovestone S, Simmons A, Soininen H and AddNeuroMed Consortium. CERAD

Neuropsychological Total Scores Reflect Cortical Thinning in Prodromal Alzheimer's Disease. Dementia and Geriatric Cognitive Disorders Extra 3:446-458, 2013.

IV Paajanen T, Hänninen T, Tunnard C, Hallikainen M, Mecocci P, Sobow T, Tsolaki M, Vellas B, Lovestone S, Soininen H. CERAD neuropsychological compound scores are accurate in detecting prodromal Alzheimer's disease: a prospective AddNeuroMed study. Journal of Alzheimer’s Disease 39:679-690, 2014.

The publications were adapted with the permission of the copyright owners.

(16)

(17)

Abbreviations

AACD Aging-Associated Cognitive Decline

AAMI Age-Associated Memory Impairment

AD Alzheimer’s disease ADAS-Cog Alzheimer’s Disease

Assessment Scale – Cognitive subscale

ADAS1 First subtest on the ADAS- Cog

ADCS-ADL Alzheimer’s Disease Cooperative Study – Activities of Daily Living Inventory

ADNI Alzheimer's Disease Neuroimaging Initiative

ANCOVA Analysis of covariance

ANM AddNeuroMed

ANOVA Analysis of variance APOE ɛ4 Apolipoprotein E ɛ4 allele AUC Area under the curve Aβ Amyloid beta

BOLD Blood oxygen level dependent

CA1 Cornu ammnonis CDR-SOB Clinical Dementia Rating

scale-Sum Of Boxes

CERAD-NB Consortium to Establish a Registry for Alzheimer’s Disease-Neuropsychological Battery

CP Constructional praxis CR Constructional praxis recall CSC Cortical signature of

cognition

CSF Cerebrospinal fluid CT Computed tomography CTH Cortical thickness CVD Cerebrovascular disease DBM Deformation-based

morphometry

Depr Depression

DSM-IV-TR Diagnostic and Statistical Manual of Mental Disorders fourth edition

DTI Diffusion tensor imaging EOAD Early onset Alzheimer’s

disease

ERC Entorhinal cortex FCSRT Free and Cued Selective

Reminding Test

FDG Fluorodeoxyglucose fMRI Functional magnetic

resonance imaging

FTD Frontotemporal lobar degeneration

(20)

GDS Geriatric Depression rating Scale

GMV Grey matter volume HAROLD Hemispheric asymmetry in

the old

HC Healthy control

Hc Hippocampus

LBD Lewy body disease LOAD Late onset Alzheimer’s

disease

MCI Mild Cognitive Impairment

MI Memory Index

MMSE Mini-Mental State Examination

MoCA Montreal Cognitive Assessment

MRI Magnetic Resonance Imaging MTL Medial temporal lobe

MTS Memory Total Score NFT Neurofibrillary tangles NIA-AA National Institute on Aging-

Alzheimer's Association

NINCDS-ADRDA National Institute of Neurological and

Communicative Disorders and Stroke and the

Alzheimer’s Disease Related Disorders Association NPI Neuropsychiatric inventory PCA Posterior cortical atrophy PD Parkinson’s disease

PET Positron emission tomography

PiB Pittsburgh compound B PMCI Progressive Mild Cognitive

Impairment

PPA Primary progressive aphasia p-tau Hyperphosphorylated tau ROC Receiver operating

characteristic ROI Region of interest SD Semantic dementia SMCI Stable Mild Cognitive

Impairment

SPECT Single photon emission computed tomography

SPSS Statistical Package for Social Sciences

STAC Scaffolding theory of age and cognition

TBM Tensor-based morphometry UTSW University of Texas

Southwestern

VaD Vascular dementia

VBM Voxel-based morphometry VF Verbal fluency

WLM Word list memory WLR Word list recall WLRc Word list recognition WMI White matter integrity

(21)

(22)

(23)

1 Introduction

Alzheimer’s disease (AD) is the most common progressive neurodegenerative disease leading to dementia in old age (Lobo et al., 2000). Although AD related neuropathological changes in the brain, namely development of beta-amyloid plaques and neurofibrillary tangles (Nelson et al., 2012), are known to start years before the first clinical symptoms, there is currently no single biological test capable of diagnosing the disease (Jack et al., 2013). Typically, the first clinical manifestation of AD is a slowly progressive impairment in episodic memory and other cognitive functions, eventually leading to dementia and subsequently to hospitalization and death.

Even although there is currently no cure for AD, early diagnosis is important, because if symptomatic medication is administered during the disease’s earliest stages, this can slow down the cognitive deterioration and alleviate neuropsychiatric symptoms. Subjects with mild cognitive impairment (MCI) (Petersen et al., 2001) are known to have an increased risk for developing AD and thus have been investigated extensively during the last decades. Recent developments in biomarkers have been promising and new diagnostic criteria have been proposed for identifying the predementia phase of AD (Albert et al., 2011; Dubois et al., 2007). Despite the progress in imaging and biomarkers, cognitive deficits, as measured with neuropsychological tests, still form the core of the AD diagnosis. Neuropsychological tests have been shown to be highly sensitive and practical in screening for MCI and AD (Amieva et al., 2014; Sarazin et al., 2007; Tabert et al., 2006). However, more sensitive and specific neuropsychological measures are needed together with novel biomarkers for identifying subjects already in the prodromal phase of AD.

The Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) neuropsychological battery (Morris et al., 1989) is a widely used cognitive battery and it has been shown to be accurate in detecting MCI and AD. However, cognitive total scores for this test battery have not been widely studied in prodromal AD, and the test battery also lacks a validated memory compound score. This study provides new knowledge on the use of the CERAD battery alone and together with automated structural magnetic resonance imaging (MRI) in detecting MCI and AD.

(24)

2 Review of the literature

2.1 DEMENTIA AND ALZHEIMER’S DISEASE

2.1.1 Socioeconomic significance of AD and dementia

Neurodegenerative diseases leading to dementia are becoming a major public health problem all over the world. The prevalence of memory diseases is known to increase nearly exponentially with age after an individual’s 65^th birthday(Hofman et al., 1991;

Lobo et al., 2000) and this, together with longevity of the population, will dramatically increase the prevalence of dementia. Over 35 million people suffered from dementia in 2010 and it has been estimated that these numbers will double every 20 years leading to over 115 million people living with dementia in year 2050 (Prince et al., 2013).

Alzheimer’s disease (AD) is the most common type of neurodegenerative disease in old age, accounting for more than half of all dementia cases (Lobo et al., 2000). Other less common causes of dementia are cerebrovascular disease (CVD) (small and large vessel diseases), Lewy body disease (LBD), Parkinson’s disease (PD) and frontotemporal lobar degeneration (FTD) including behavioral variant of FTD, semantic dementia (SD) and primary progressive aphasia (PPA) (Levy & Chelune, 2007). Neuropathological studies have revealed that AD related neuropathological changes can be found in over 80% subjects previously diagnosed with dementia (Jellinger & Attems, 2010). However, it is important to note that also mixed pathologies are very common especially among the oldest-old. A recent study indicated that up to 84% of those subjects with AD pathology also had morphological signs of CVD (Attems & Jellinger, 2014).

In addition to the huge burden to individuals and their caregivers, AD and other dementias result in enormous socio-economic costs. It has been estimated to be nearly 5 million people are living with dementia in the United States and in this population, the direct costs of AD and other dementias were estimated to be 183 billion dollars.

Additionally, another over 200 billion dollars has been estimated to be consumed in unpaid caregiving by family members, relatives and friends (Stefanacci, 2011). Even though a disease modifying medication for AD has yet to be discovered, it is known that several modifiable factors that will be discussed later are associated with the risk of Alzheimer’s disease and dementia. Early detection of Alzheimer’s disease is important because initiation of currently available symptomatic medications (Birks, 2006) and lifestyle changes (Ngandu et al., 2015) affect cognition, may slow down the disease progression, decrease neuropsychiatric symptoms, and maintain daily functions and thus also reduce socioeconomic costs (Pouryamout, Dams, Wasem, Dodel, & Neumann, 2012). In addition, early diagnosis can alleviate anxiety, allow sufficient time for making the mentally competent life arrangements. Thus, early diagnosis would benefit patients, their caregivers and society as whole.

2.1.2 Neuropathology and multifactorial aetiology of AD

Three major neuropathological changes have been identified in AD: 1) Accumulation of extraneuronal beta-amyloid (Aβ) protein plaques, 2) formation of neurofibrillary tangles (NFT) of tau-protein inside the neuron, and 3) loss of neurons and their

(25)

synaptic connections (Perl, 2010). Today, it is known that these AD-related neuropathological processes begin several years before the onset of first clinical symptoms and different neurobiological changes typically proceed in a certain order (Jack et al., 2013). In this proposed model of AD, initially there is an asymptomatic phase of several years when the initial neuropathological changes can be detected in cerebrospinal fluid (CSF) markers, first in lowered beta-amyloid levels and second in elevated concentrations of tau and hyperphosphorylated tau protein (p-tau). After the progression of neuropathological processes, subtle cognitive changes, typically first in the memory domain, become evident. Mild cognitive impairment (MCI) describes a clinical condition where cognitive deficits have been verified, but activities of daily living are intact and criteria of dementia are not fulfilled (Winblad et al., 2004). An individual with MCI has been shown to carry an increased risk for developing AD and other neurodegenerative diseases and thus this stage is referred to as a transition stage between normal ageing and dementia.

Neuronal loss and dysfunction in AD have been shown to be associated especially with cholinergic system in the basal forebrain and depletion of acetylcholine is known to be related with both Aβ deposition and cognitive functions (Mufson, Counts, Perez,

& Ginsberg, 2008; Perry et al., 1978). One of the leading theoretical models to explain the pathological process underlying AD is the so-called “amyloid-cascade hypothesis”

which suggests that the neurotoxic effects of overproduced Aβ leads to a synaptic dysfunction (Ondrejcak et al., 2010) and intraneuronal NFT formulation (Giaccone et al., 1996), eventually resulting in synaptic and neuronal loss. Although Aβ-plaques are considered to be an essential neuropathological marker of the disease and commonly accepted to play a key role in its pathogenesis, it has been also shown that amyloid accumulation and neuronal injury are not necessarily linked (Chetelat et al., 2013). Conversely, NFT accumulation seems to reflect better the clinical and cognitive changes of the AD (Nelson et al., 2012). The initial neurofibrillary changes are found in transentorhinal and entorhinal cortex, from which the pathology proceeds to hippocampus, limbic lobe and eventually to all of the neocortical regions (Braak &

Braak, 1995). MTL structures, i.e. entorhinal cortex (ERC) and hippocampus (Hc) are known to be one of the most crucial brain areas for memory encoding and consolidation (Insel & Takehara-Nishiuchi, 2013; Preston & Eichenbaum, 2013). In a typical case of sporadic AD, the first deficits are thus commonly seen on episodic memory and followed by gradual decline in other cognitive functions (Pena- Casanova, Sanchez-Benavides, de Sola, Manero-Borras, & Casals-Coll, 2012). Even though there are several known risk factors (discussed in detail later) and increased knowledge of the neuropathological process in AD, the ultimate reason for this debilitating progressive disease remains a mystery.

Epidemiological studies have identified several factors associated with the increased risk for AD and dementia. In addition to higher age, the most commonly verified risk factors are related with cardiovascular health, lifestyle, and psychological and genetic factors (Imtiaz, Tolppanen, Kivipelto, & Soininen, 2014). In a large meta- analysis, especially diabetes, the ɛ4 allele of apolipoprotein E (APOE) gene, smoking and depression showed the most consistent association increased risk for AD and cognitive decline across multiple studies (Williams, Plassman, Burke, & Benjamin, 2010). Until today, the most remarkable landmarks in AD genetics have been the discovery of mutations in Amyloid Precursor Protein, Presenilin 1 and Presenilin 2 genes causing autosomal dominant AD, while the APOE ɛ4 allele is the most important genetic risk factor for both late-onset and early-onset AD (Van

(26)

Cauwenberghe, Van Broeckhoven, & Sleegers, 2016). Despite the continuous expansion of knowledge on AD genetics and the identification of over 20 additional genetic risk factors, none of these have achieved the same significance as APOE ɛ4.

Even although the explanation power of genetic risk factors is currently not sufficiently powerful to be used in clinical diagnosis, in the future, it has been claimed that genetic information will hold the key to the personalized treatment of AD (Van Cauwenberghe et al., 2016).

Interests in prevention of AD and dementia have increased dramatically since several possible modifiable preventive factors have been identified. In epidemiological studies, potential factors for decreasing the risk for AD include physical and cognitive activity, Mediterranean-type diet, folic acid-intake and low or moderate alcohol intake (Daviglus et al., 2011). Recently, it has also been demonstrated in a multidomain-intervention that including intensive cardiovascular disease management, dietary guidance, and physical and cognitive exercising can exert positive effects on cognition in those elderly subjects with a slightly increased risk of dementia (Ngandu et al., 2015). In addition, the most recent large epidemiological study in United States indicated that the prevalence of dementia among the people over 65 years had decreased from 11.6% to 8.8% between the years 2000-2012. Interestingly, the observed decrease in dementia prevalence was evident even although the cardiovascular risk profile figures had worsened. Increased educational attainment was found to explain part of the change, however, the complete picture of the social, behavioral and medical factors is still unclear (Langa et al., 2017).

2.1.3 Development of clinical AD diagnosis

The criteria for clinical diagnosis of Alzheimer’s disease were established by the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association (NINCDS/ADRDA) workgroup over 30 years ago (McKhann et al., 1984). In the criteria, AD diagnosis is based solely on clinical symptoms and exclusion of other reasons possibly causing dementia. The diagnosis of AD requires an insidious onset combined with progressive impairment of memory and other cognitive domains. Neuropsychological tests were suggested to be used as a confirmatory method of objective cognitive deficits and laboratory tests and brain imaging were applied for excluding other possible causes of dementia. These criteria proposed a probabilistic diagnostic classification including probable, possible and definite Alzheimer’s disease (McKhann et al., 1984).Until the recent years have NINCDS/ADRDA criteria combined with Diagnostic and Statistical Manual of Mental Disorders fourth edition (DSM-IV) (American Psychiatric Association, 1994) served as diagnostic standards for AD. The remaining challenge in the clinical diagnosis of AD is the lack of an unambiguous biological disease marker and the fact that definite diagnosis can be established only by post-mortem verification of the disease due to the appropriate neuropathological changes in the brain. In spite of their probabilistic nature, NINCDS/ADRDA criteria have been demonstrated to be rather good when compared against the post-mortem diagnosis.

A review of validation studies indicated that the diagnostic accuracy for “probable AD” according to NINCDS/ADRDA and “Dementia of Alzheimer’s type” according DSM-IIIR criteria achieved 81% sensitivity and 70% specificity. The diagnostic sensitivity for “possible AD” was found to be even higher (average 93%), however, with significantly lower specificity (average 47%) (Knopman et al., 2001). The low

(27)

specificity against other neurodegenerative diseases and the requirement that cognitive deterioration has to be classified as dementia before diagnosis have been the main limitations of the criteria.

Increased knowledge on AD biomarkers, especially on brain imaging and cerebrospinal fluid (CSF) markers, has advanced our understanding of the nature of disease and its progression. Today, it is widely accepted that the first neuropathological changes related to AD develop many years or even decades before the first clinical symptoms and there also are several dynamic biomarkers which reflect the disease course (Jack et al., 2013). The first move towards revising the long established diagnostic standards (NINCDS-ADRDA and the DSM-IV-TR) occurred in 2007 when new research criteria for AD were published (Dubois et al., 2007). The Dubois research criteria for AD aimed to provide a diagnostic framework that can be applied to increase our understanding of the distinctive and reliable biomarkers to capture also the earlier stages of AD before full-blown dementia. The criteria suggest that AD diagnosis should be made based on the core criteria of amnestic type of memory deficit accompanied with at least one supportive biomarker finding. The putative biomarkers were MTL atrophy in structural brain imaging; disease markers on molecular neuroimaging with positron emission tomography (PET), abnormal CSF-marker (either lowered Aβ or elevated p-tau protein), abnormal SPECT (Single photon emission computed tomography) imaging result or familial AD gene. The novel aspect in these research criteria was that it was intended to achieve an AD diagnosis before the stage of dementia (Dubois, Picard, & Sarazin, 2009) (Table 1).

A few years after Dubois et al. had presented research criteria for AD, also the National Institute on Aging-Alzheimer's Association (NIA-AA) instigated a large workgroup to revise the original McKhann criteria for AD; this created three different work groups each focusing on the different phases of the disease, namely; a) dementia phase, b) symptomatic predementia phase, and c) asymptomatic preclinical phase of AD (Jack et al., 2011). The NIA-AA criteria also suggest that new biomarker information can be applied in the AD diagnosis, however, it is more conservative and adheres to the probabilistic nature of the diagnosis and does not consider biomarker information mandatory. For example, according to the NIA-AA criteria, “Dementia due to AD” (McKhann et al., 2011) can be set solely on the basis of clinical information but certainty of the diagnosis can be elevated with biomarkers. The NIA-AA workgroup concentrating on symptomatic predementia phase of AD suggested new diagnostic guidelines for “Mild cognitive impairment due to AD” (Albert et al., 2011).

These criteria also include the concept that a “predementia AD diagnosis” can be made on the basis of clinical information and diagnostic certainty can be classified into four different levels depending on the availability and quality of biomarker information. The third NIA-AA workgroup which focused on the preclinical phase of the AD, did not suggest diagnostic criteria for clinical practice but formulated a recommendation which aims to guide future research in the AD diagnostics and treatment (Sperling et al., 2011).

(28)

Table 1. New diagnostic criteria for Alzheimer’s disease (AD) proposed by Dubois et al.

Diagnosis of AD requires the cognitive core criterion (major) and at least one supportive biomarker criterion (minor). Modified from (Dubois et al., 2009).

Major criteria (cognition) Minor criteria (biomarkers) Presence of early and significant episodic

memory impairment including the following (1-3) features below:

At least one of the following (A-C) supportive biomarker founding below:

1. Gradual/progressive deterioration in memory (over 6 months) reported by patient or informant

A) Presence of medial temporal lobe atrophy:

 volume loss of hippocampus, entorhinal cortex, amygdala (evidenced on MRI) 2. Significantly impaired episodic memory on

objective testing. Deficit in delayed recall that does not improve notably/normalize with cuing or recognition testing, even effective encoding has previously ensured.

3. The episodic memory impairment can be isolated or associated with other cognitive changes at the onset of AD or as AD progress.

B) Abnormal CSF biomarker:

 Low amyloid β 1-42 concentrations, increased total tau and/or increased phospho-tau concentrations

 Other well-validated CSF markers to be discovered in the future

C) Specific pattern on functional PET imaging:

 Reduced glucose metabolism in bilateral temporal parietal regions

 Other well-validated ligands. Including those that foreseeably will emerge such as PiB or FDDNP.

D) Proven AD autosomal dominant mutation within the immediate family

MRI = magnetic resonance imaging; CSF = cerebrospinal fluid; PET = positron emission tomography; PiB = Pittsburgh compound B; FDDNP = amyloid and neurofibrillary tangle ligand

From the current clinical practice viewpoint, it is important to note that despite the increased significance of biomarker information in all of the putative clinical criteria (Albert et al., 2011; Dubois et al., 2007; McKhann et al., 2011) nonetheless, an objective cognitive impairment, as measured with the neuropsychological tests, represents the core in the AD diagnosis. Figure 1 describes the typical course of AD according to the presence of changes in (a) some cognitive functions and (b) in levels of biomarkers.

Frisoni and coworkers proposed that, according to the known natural history, different cognitive and biological markers are sensitive to early pathological changes and others to later developments in disease progression (Frisoni, Fox, Jack, Scheltens,

& Thompson, 2010).

According to the development of AD related neuropathological changes (Braak &

Braak, 1995) it can be postulated that brain changes especially in the medial temporal lobe structures and hippocampal type of episodic memory deficit would be the most potential markers for early detection of AD.

(29)

Figure 1. Natural progression of cognition and biological markers of Alzheimer’s disease:

a theoretical model by Frisoni and coworkers (Frisoni et al., 2010). a) Memory tests are typically useful in the MCI phase diagnostic (1) but, soon reach a maximal level of impairment (2), not very sensitive to later changes of disease progression. Whereas, verbal comprehension tests lack sensitivity in the early diagnostic (4), but are more sensitive to later cognitive changes in AD (5). Similarly biological markers have been suggested to have their own typical trajectories in their relationship with the disease course (b). Reprinted by permission from Macmillan Ltd: Nature Reviews Neurology, copyright (2010).

(30)

2.1.4 Neuropsychological symptoms of AD

The symptoms of the neurodegenerative diseases can be categorized into three main groups: the neurological, cognitive, and neuropsychiatric symptoms of dementia (Finkel, 2001). Neuropsychological symptoms can sometimes be considered as a large umbrella term which, in addition to cognition, includes also some behavioral and higher neurological functions. Most often in research into the neurodegenerative disorders, neuropsychological symptoms are referred simply to neurocognitive or cognitive deficits. Typically AD starts with the hippocampal type of episodic memory deficits, which include a deterioration both in the memory encoding and retrieval (Pena-Casanova et al., 2012), which are later followed by an impairment in language, spatial cognition and executive functions, eventually leading to dementia (Salmon &

Bondi, 2009). Nowadays, it is also known that neuropsychiatric symptoms, such depression, apathy and anxiety, are not only present in late AD but certain manifestations are rather common already in the very early phase of the disease (Lyketsos et al., 2011).

Research on different episodic memory processes and their relationship with broad memory network has highlighted the importance of comprehensive testing of memory functions in AD. Despite the crucial role of the MTL structures in the memory encoding and storage processes (Squire, Stark, & Clark, 2004), it is known that prefrontal and parietal regions play a very important part in working and long term memory processes (Blumenfeld & Ranganath, 2007; Champod & Petrides, 2007). For example, in mild AD, an immediate recall trial in a classic word list task was not associated with the cortical morphometry of the MTL structures, but was related to the fronto-temporal and inferior parietal regions. After rehearsal, the immediate recall of verbal material was associated to a large memory network including MTL and isocortical structures (especially left temporal gyri and precuneus). In addition, delayed recall performance correlated only with the volume of Hc and the accuracy of delayed recognition was associated most strongly with parahippocampal gyrus and ERC (Wolk, Dickerson, & Alzheimer's Disease Neuroimaging Initiative, 2011). On the other hand, morphological fragmentation of episodic memory functions is not necessarily so distinct, as involvement of different temporal, parietal and frontal areas may vary depending on the clinical condition (Walhovd et al., 2010). In studies applying functional magnetic resonance imaging (fMRI), changes in broad memory networks have been verified in prodromal AD and even among healthy elderly individuals with high amyloid deposition (Sperling et al., 2010). Episodic memory impairment is known to be the most relevant factor in predicting the future development of AD (Jungwirth, Zehetmayer, Hinterberger, Tragl, & Fischer, 2012;

Lehrner et al., 2005; Ravaglia et al., 2006). However, several studies have shown that the subjects who have both memory and executive problems are especially at risk for developing this disease (Chapman et al., 2011; Rozzini et al., 2007; Tabert et al., 2006).

In addition to episodic memory and executive function dysfunction, also working memory deficits have been found to be related to AD (Kirova, Bays, & Lagalwar, 2015). Thus, it has been suggested that in addition to memory, also other cognitive domains should be assessed when suspecting early AD (Pena-Casanova et al., 2012).

It is important to conduct a comprehensive neuropsychological evaluation also for differentiation between AD and other dementias (Karantzoulis & Galvin, 2011) as well as in detecting AD which presents with non-typical cognitive deficits. Atypical AD is commonly divided into a) logopenic progressive aphasia, b) posterior cortical atrophy (PCA) and c) frontal variant of AD; it can be clinically very challenging to differentiate

(31)

from the other rare forms of dementia (Wolk, 2013). In the late onset AD (LOAD), the atypical form is relatively rare, accounting for approximately 6-7 % of patients, however, in early onset AD (EOAD), the atypical non-amnestic form is significantly more common, and can be encountered in 30-60% of all patients (Mendez, Lee, Joshi,

& Shapira, 2012; Smits et al., 2012). It has also been shown that in EOAD, major cognitive deficits are loaded according to the disease related neuropathological changes. The most common major cognitive deficit can be present in either visuospatial or language functions and correspondingly the most evident brain changes are detected in the parietal and left frontal cortex, respectively (Mendez et al., 2012).

2.1.5 Brain imaging and other biomarkers of AD

Knowledge on the AD biomarkers has increased massively during recent years, and extensive research has produced several potential imaging and other biomarkers for AD. A large number of promising candidate biomarkers for AD have been suggested, such as Hc and ERC volumes, cortical thickness (CTH), basal forebrain nuclei, brain changes measured by deformation-based and voxel-based morphometry, functional changes in fMRI, brain connectivity in diffusion tensor imaging (DTI) and CSF markers such as Abeta42, total tau, and p-tau (Hampel et al., 2008). Jack and coworkers suggested that five biological and imaging markers of AD have been well enough validated to be used both in clinical and research settings (Jack et al., 2010). The currently recognized biomarkers of AD can be divided into two main categories: 1) biomarkers reflecting brain β-amyloid deposition, and 2) biomarkers of neuronal degeneration or injury. The NIA-AA criteria for “Dementia due to Alzheimer’s disease” (G. M. McKhann et al., 2011) suggest two biomarkers of Aβ a) CSF Aβ1–42 levels and b) Aβ imaging in PET, and three biomarkers of neuronal injury c) elevated CSF total tau or p-tau, d) decreased fluorodeoxyglucose (FDG) activity in PET and e) disproportionate atrophy in specified areas of temporal lobe and/or medial parietal cortex measured with structural brain imaging. Research on biomarkers has revealed that different biological processes are expressed in different phases of AD and a model describing temporal ordering of the biomarkers in relationship with cognitive and other clinical symptoms has been postulated (Jack et al., 2010; Jack et al., 2013). Due to topic of this thesis, a more detailed focus will be placed on structural magnetic resonance imaging (MRI).

2.1.5.1 Structural magnetic resonance imaging

The role of brain imaging in AD has changed remarkably during recent decades.

Initially, structural brain imaging, first computed tomography (CT) and later magnetic resonance imaging (MRI), was applied in the diagnostics of AD to exclude other pathologies such vascular dementia and head traumas (McKhann et al., 1984).

In the late 1980s, hippocampal atrophy in early AD was verified in the CT studies (de Leon, George, Stylopoulos, Smith, & Miller, 1989) and later researchers suggested an MRI-based visual rating scale for MTL atrophy could be used in clinical diagnostics (Scheltens et al., 1992). Although CT imaging is suitable for detecting acute surgically treatable reasons and new CT devices are able to measure total brain and MTL atrophy and also white matter changes (Wattjes et al., 2009), MRI is superior as it provides better soft tissue contrast resolution and does not require exposing the patient to ionizing radiation (Symms, Jager, Schmierer, & Yousry, 2004).

(32)

The MRI technique is based on the phenomenon that protons have an angular momentum which can become polarized in a strong magnetic field. Radiofrequency pulses are used for changing the energy state of the protons. After the pulses are no longer applied, the protons return to their own energy state and at the same time, they produce radiofrequencies which can be recorded. By combining different pulses and gradients, MR imaging is able to discriminate very accurately different brain tissue characteristics. MRI methods are typically classified as “structural” when they provide static anatomic information (for example, volumetrics of the brain tissue) and

“functional” when the information gathered is physiologically dynamic such BOLD (blood oxygen level dependent). Structural MR imaging produces basically two kinds of information of the brain: 1) volumetry, reflecting the cell loss (atrophy) and 2) changes in the tissue characteristics, such as white matter integrity (WMI) (Symms et al., 2004).

The evaluation of the extent of brain atrophy from the MR images can be done using several different procedures. The first rating scale for AD related brain changes developed by Scheltens et al. was based on the visual inspection of hippocampal atrophy from the coronal T1-weighted MR images parallel to the brain stem axis. In this system, the severity of the Hc atrophy is classified on a 0-4 scale, where 0 represents no atrophy and 4 indicates severe atrophy (Scheltens et al., 1992). Scheltens’

rating system has been shown to yield up to 92% sensitivity and 93% specificity in discriminating AD patients from the controls (Desmond et al., 1994), however, the downsides of visual assessments are their subjective nature and the high inter-rater variation in the evaluations (Scheltens, Launer, Barkhof, Weinstein, & van Gool, 1995).

Recently, a visual rating scale has been developed also for posterior cortical atrophy, which can be helpful in the evaluation of brain atrophy of atypical forms of AD and other dementias (Koedam et al., 2011). Visual assessments are applicable in clinical practice when a rapid assessment of the presence of brain atrophy is needed, however, more exact quantitative information of the Hc volumetry can be obtained with manual segmentation procedures (Konrad et al., 2009).

Manual labeling of the Hc is currently considered to be the golden standard for measuring Hc volumes and it has proved to be accurate in differentiating AD patients from MCI subjects and controls (Bottino et al., 2002; Jack et al., 1998). However, the challenge in the manual Hc tracing is the extensive variation in the different segmentation protocols (Konrad et al., 2009). In addition, the manual segmentation of Hc and other brain areas is time consuming and needs to be performed by an experienced professional.

2.1.5.1.1 Automated MRI methods

Previously, many MRI studies have used manual segmentation procedures and thus only a few brain regions have been evaluated. Visual rating of the MTL atrophy is convenient in the clinical AD diagnostics, however, as noted earlier, its evaluation is vulnerable to subjectivity and inter-rater variability. Much research has been conducted on the processing of MR images to eliminate human error and variation in the analytics. It is also recognized that AD-related structural and functional brain changes are not limited to MTL but cover broad structural and functional brain networks (Bakkour, Morris, & Dickerson, 2009; Sperling et al., 2010). Thus, automated MR imaging methods have been developed to assess whole brain structures in an explorative fashion and additionally enabling specific regions of interest (ROI) to be defined beforehand.

(33)

Voxel-based morphometry (VBM) was one of the first developed automated MRI analyzing methods initially described in detail by Ashburner & Friston (Ashburner &

Friston, 2000). The VBM technique utilizes a standard stereotactic space in which high quality MR images are registered. Subsequently, grey matter is segmented from the other brain tissues and parametric voxel-level statistical analyses are tabulated between the studied subject groups. VBM provides a method for measuring whole brain structures without a priori defined areas and has proved to be an applicable method in group level analyses for discriminating between AD, MCI and control subjects (Baron et al., 2001; Chetelat et al., 2005; Pennanen et al., 2004). The results of the VBM studies correspond to the tau-pathology model of AD (Braak & Braak, 1991).

The brain atrophy pattern of AD including MTL, posterior cingulate, precuneus and areas of temporo-parietal junction has been verified (Baron et al., 2001) and rather similar regions have been found to suffer the highest grey matter volume loss during the follow-up among those MCI subjects who developed AD (Chetelat et al., 2005). It has been also suggested that ERC atrophy may precede Hc volume loss and thus be an especially important early biomarker of AD (Pennanen et al., 2004). Due to known relevance of MTL structures in AD, automated procedures specially designed for segmentation of the Hc formation have been developed. Automated Hc assessments are typically based on an atlas-based labeling procedure, of these, template-library methods using multiple atlases have proved to better than single template approaches (Barnes et al., 2008). The negative side of the multi-template segmentation procedures has been that until now, they have required very long computation times. The recently developed very fast multi-atlas labeling procedure has, however, been claimed to yield discrimination accuracies around 80% in a control vs. AD comparison and around 65% in the stable MCI vs. progressive MCI comparison depending on the study population (Lotjonen et al., 2011). Some other promising automated MRI analyzing methods have been proposed for the early detection of AD e.g. tensor-based morphometry (TBM) (Koikkalainen et al., 2011), deformation-based morphometry (DBM) (Teipel et al., 2013) and manifold-based learning (MBL) (Wolz et al., 2011).

Measuring the thickness of cortical mantle is a relatively novel approach for analyzing brain structures. The human cortex is highly folded, which make the manual cortical thickness (CTH) assessments of MR images very difficult, arduous and susceptible to measurement errors. Automated computer-based methods developed for measuring CTH have been claimed to improve the assessment and diagnosis of a variety of neurodegenerative and psychiatric disorders (Fischl & Dale, 2000; Lerch & Evans, 2005).

Typically, CTH analyses start with a Talairach transformation of T1-weighted MR images. Next, atlas location, intensity and local neighbors are applied in order to identify the main body of white matter. The skull is removed from the image, leaving only the brain in the image. The remaining voxels are classified then as white matter and non-white matter. An initial surface for each hemisphere is created on the edge of white matter and grey matter. Next, this surface is pushed outward until the intensity gradient change and the pial surface are reached (Fischl & Dale, 2000). Subsequently, the sulcal and gyral patterns are allocated to the artificial average brain surface and then resampled into a common reference space. This procedure enables accurate CTH analyses across participants’ cortex either vertex-by-vertex or regionally. The development of MR image analysis algorithms and software has led to new analytical tools that automatically split human brain according to distinct anatomic regions and make it possible to quantify volumes and thickness of the brain tissue in these regions

(34)

(Desikan et al., 2009; Fischl et al., 2002). The benefits of this kind of automated methods are decrease in the rater-dependent bias and an increase in regions that can be studied in parallel as potential biomarkers. As compared to more traditional MRI methods such VBM, whole brain parcellation MRI-software tools provide information that can be used at the individual level.

Several studies have shown that the CTH measurement can be very accurate in separating AD patients and MCI subjects from controls and according to some studies, CTH measures may be more sensitive than cognitive tests also in predicting conversion from MCI to AD (Desikan et al., 2009; Julkunen et al., 2010; Peters, Villeneuve, & Belleville, 2014; Querbes et al., 2009). A particular cortical thinning pattern, the so-called “cortical atrophy signature of AD” including temporal lobe structures (such parahippocampal, entorhinal and medial temporal cortex), and inner parietal lobe structures (such posterior cingulate and precuneus) has been found to be related to AD conversion and thus suggested as a novel biomarker (Bakkour et al., 2009; Dickerson et al., 2009; Verfaillie et al., 2016). In addition to regional cortical volumes and thicknesses, also changes in sulcal shape have recently been proposed as a potential biomarker for AD. A greater sulcal widening has been claimed to be a very early structural change occurring in MCI (Im et al., 2008).

2.2 HEALTHY AGEING, MILD COGNITIVE IMPAIRMENT AND PRODROMAL AD

2.2.1 Brain and cognition in healthy ageing

The early detection of the cognitive impairment related to AD or other neurodegenerative disorders is challenging because several non-organic factors (for example mood and sleep problems, attention and motivational factors) can affect cognitive performance and also healthy ageing evokes both structural and functional changes in the brain. Age-related changes in several cognitive functions have been documented in both cross-sectional and longitudinal samples (Park et al., 2002;

Salthouse, 2010).

The ageing brain declines in volume especially in the frontal cortex, however, volume decrease rates have been found also in other parts of the brain (Bartzokis et al., 2001; Giorgio et al., 2010; Meguro et al., 2001). In normal ageing cortical volume decline has been suggested to be explained mainly by cortical thinning and to a lesser extend by decrease on cortical surface area (Storsve et al., 2014). DTI results have revealed an age-associated WMI loss in frontal areas and distributed throughout the brain, explaining the diffuse cognitive consequences (Barrick, Charlton, Clark, &

Markus, 2010; Maniega et al., 2015). The age-related integrity loss in white matter affects widely distributed neural connections and has been shown to be associated with cognitive performance, especially those requiring perceptual speed, executive functions and memory (Madden et al., 2012), however, an association with general cognitive status has also been found (Shenkin et al., 2005). In addition to structural and functional changes in frontal cortex, ageing is also linked with a decline in Hc volumes. Nonetheless, the age-associated Hc volume loss is smaller than occurring in AD (Jack et al., 1998) and it is also known that Hc is not a homogeneous formation but composed of several subregions which may be differently affected in normal ageing and AD (de Flores, La Joie, & Chetelat, 2015; Pini et al., 2016). In addition to a decrease in brain volumes and WMI, also CTH is shown to undergo age-associated brain changes (Thambisetty et al., 2010).

CERAD neuropsychological battery and structural magnetic resonance imaging in the detection of mild cognitive impairment and prodromal Alzheimer's disease

uef.fi

Dissertations in Health Sciences

TEEMU PAAJANEN

CERAD NEUROPSYCHOLOGICAL BATTERY AND STRUCTURAL MAGNETIC RESONANCE IMAGING IN THE DETECTION OF MILD COGNITIVE IMPAIRMENT AND PRODROMAL ALZHEIMER’S DISEASE

TEEMU PAAJANEN

CERAD Neuropsychological Battery and Structural Magnetic Resonance Imaging in the Detection of Mild Cognitive Impairment

and Prodromal Alzheimer’s Disease

CERAD Neuropsychological Battery and Structural Magnetic Resonance Imaging in the Detection of Mild Cognitive Impairment

and Prodromal Alzheimer’s Disease

Acknowledgements

List of the original publications

Contents

Abbreviations

1 Introduction

2 Review of the literature