1
Journal of Alzheimer’s Disease xx (20xx) x–xx DOI 10.3233/JAD-141413 IOS Press
3
roo f
2
Review
Risk Factors and Screening Methods for Detecting Dementia: A Narrative Review
4
Blossom C.M. Stephana,∗ and Carol Brayneb
6 5
a Institute
or P
1
of Healthy and Society, Newcastle University, UK of Public Health and Primary Care, Cambridge University, UK
b Department
uth
Accepted 16 September 2014
15
Keywords: Alzheimer’s disease, cognition, dementia, early diagnosis, screening
16
INTRODUCTION
13
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33
cte
12
rre
11
Dementia is a neurodegenerative disease with complex, multi-factorial pathogenesis that currently has no cure. As the population continues to age, the prevalence of dementia is predicted to rise [1]. Dementia, including clinically diagnosed Alzheimer’s disease (AD), is a slowly evolving disease, with a long preclinical phase possibly occurring 20 years before symptoms manifest. Clinical trials to meaningfully alter or halt disease progression post-diagnosis have been largely unsuccessful. As such, there have been calls to abandon trials at this clinical stage in favor of targeting individuals before symptom onset and possibly irreversible neuronal damage has occurred. In most other disorders for which such approaches have been promoted, there has been an existing effective treatment demonstrated at earlier stages that can be trialed alongside early detection. However, identifying risk factors
co
9 10
Un
8
dA
14
Abstract. Accurate detection of individuals with cognitive impairment and dementia in addition to identification of those at high risk of future disease is important to guide clinical care, and has research implications regarding clinical trial recruitment and development of dementia preventative strategies. In this narrative review, we describe new proposed criteria for early diagnosis of Alzheimer’s disease (AD). We also explore risk factors for dementia and evaluate methods for screening for increased risk of incident disease. We highlight variability in different diagnostic approaches. Additional work needs to be done to validate new methods across different settings (such as population-based, primary care, and memory clinics), age, and ethnic groups. Having an accurate method to assess for dementia and predict risk in routine clinical care will aid decision making and could ultimately lead to disease prevention.
7
∗ Correspondence
to: Dr. Blossom Stephan, Institute of Health and Society, Newcastle University, The Baddiley-Clark Building, Richardson Road, Newcastle upon Tyne NE2 4AX, UK. Tel.: +44 0 191 222 3811; E-mail:
[email protected].
for cognitive decline and dementia and using these to detect individuals in asymptomatic or preclinical stages is complex. To achieve this, numerous different strategies have been proposed. These strategies will be reviewed here. METHODS This is a narrative review providing an overview of the literature focused on a number of topics including: 1) current criteria for diagnosing dementia including its preclinical and clinical stages; 2) risk factors for dementia; and 3) models for predicting risk of incident dementia. Therefore, this article is not an exhaustive account of all the literature available on the different topics; which have already been the subject of systematic reviews. Our goal was to provide a broad overview of the literature focused on each topic. DETECTION: WHAT WE AIM TO ACHIEVE Dementia is strongly associated with age and most people if they are to develop dementia will do so
ISSN 1387-2877/14/$27.50 © 2014 – IOS Press and the authors. All rights reserved
34 35 36 37 38
39
40 41 42 43 44 45 46 47 48 49
50
51 52
60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104
roo f
59
or P
58
sions regarding future care and service needs of the older population. However, a review of the literature by the National Screen Committee in the UK (in 2003, with updates in 2006 and 2010) concluded that there is no evidence base to support the introduction of routine screening for dementia [9]. Further, clinical guidance produced by the National Institute of Clinical Excellence (NICE 2006) in the UK stated that ‘General population screening for dementia should not be undertaken’, of which one of the reasons for this was the lack of an accurate screening tool ([10], see also [11]). Committees outside the UK (e.g., the US Preventive Services Task Force [12, 13] and the Canadian Task Force on Preventive Health Care [14]) have also reached similar conclusions. However, given that perceived detection of dementia and cognitive impairment in community settings is low [15], with a large proportion of dementia cases being missed in primary care [12, 16], and with initiatives targeted at case finding being launched (e.g., in the USA the Annual Wellness Visit (AWV), legislated under The Patient Protection and Affordable Care Act), accurate and effective strategies to detect individuals with (and at-risk of) cognitive impairment and dementia in population-based and clinical samples are needed.
dA
57
cte
56
rre
55
by their ninth and tenth decades of life [2]. However, approximately half of those individuals who will develop dementia will be younger than these ages with most being in their eighth decade of life. Work on early detection and characterization of dementia must therefore take age into account. Indeed, at younger ages pathology tends to be ‘purer’ in nature (e.g., just reflecting AD or vascular dementia rather than a mixed phenotype) and length of survival is relatively long (e.g., there is nearly a seven year difference in survival between young-old and oldest-old people with dementia [2]). In contrast, at older ages, individuals tend to show more mixed etiology, with shorter survival which often becomes combined with terminal decline [3]. Prevention can be conducted in many ways. One approach is to understand whole population risk, with an example being salt intake and hypertension. Here a whole population’s profile of blood pressure could possibly be shifted through food policies and societal changes focused on salt consumption (e.g., [4–7]). An alternative approach is the high-risk approach where individuals are identified as being at high risk through screening of some kind. This is a more expensive and less effective method when compared with populationbased approaches [8]. A high proportion of research investment is focused on the high-risk approach. Detection of individuals at high risk is assumed to be useful in the following ways: 1) to inform clinical decision-making and initiate interventions appropriate to an individual’s risk profile (e.g., “personalized medicine”); 2) to tailor frequency of monitoring (e.g., high risk cases can be referred to specialist dementia services while low risk cases can be excluded from immediate follow-up); and 3) to provide an accurate method for helping to plan recruitment for clinical trials, design prevention strategies, and undertake further research into whether interventions targeted at different risk factors (alone or combination) are effective in lowering dementia risk. At the individual patient level when risk is identified as sufficiently high, knowledge of dementia risk may be important for future planning (e.g., care and financial planning when the individual still has capacity to be involved in decision making), and where risk is low could lead to reduced anxiety. In terms of public health, availability of a strategy for detecting high risk and preclinical cases could be used to inform choices about appropriate decisions on policy focused on population screening. In addition, knowing the profile of high risk cases and the proportion of the population likely to fall into this risk category could inform budgeting deci-
co
54
Un
53
B.C.M. Stephan and C. Brayne / Dementia Detection
uth
2
MILD COGNITIVE IMPAIRMENT (MCI) Much work on the identification of individuals at increased risk of dementia, compared to the background population, has focused on the concept of MCI. MCI captures intermediate cognition between high functioning and clearly impaired states. Use of such concepts in clinical settings has shown higher risk of dementia for those individuals attending and being referred to such settings [17]. However, numerous problems have been identified. First, several definitions of intermediate cognitive states exist and there is no established agreement on single sets of criteria or standardization of diagnostic methodology, with constantly changing criteria as new evidence and measurement methods emerge [18, 19]. This makes comparisons over time and setting difficult. Second, risk of dementia in MCI cases varies across populations (clinical versus population-based) and the definition of MCI used and the impairment captured in MCI is not always progressive, with some cases reverting to normal or remaining stable at follow-up [17, 20, 21]. Third, neuropathology in people with a diagnosis of MCI and dementia looks very similar with
105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130
131
132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153
B.C.M. Stephan and C. Brayne / Dementia Detection
162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205
roo f
161
or P
160
uth
159
dA
158
in combination with positive ratings on pathophysiological and topographical markers of AD including: 1) brain amyloid- (A) accumulation (including: low cerebrospinal fluid (CSF) A42 level and positive positron emission tomography (PET) amyloid imaging); and 2) biomarkers of neurodegeneration (including: MRI/PET evidence of volumetric changes including hippocampal or whole brain atrophy, functional impairment (18F-fluorodexoxyglucose PET or functional MRI) and CSF tau levels) [31, 32]. Using this framework, different disease stages have been defined including: Stage 0 (no biomarker evidence of AD pathology), suspected non-AD (evidence of neurodegeneration and negative results on biomarkers of amyloid accumulation); Stage 1 (asymptomatic cerebral amyloidosis); Stage 2 (amyloidosis and evidence of neurodegeneration or neuronal injury); and Stage 3 (positive on biomarkers of A accumulation and neurodegeneration and evidence of subtle cognitive decline) [31, 33]. While there is preliminary support for this model [32] much more reliability and validation work and evidence synthesis of findings in support (and opposed) to these new diagnostic methods across different settings is needed. Lastly, the latest revision of the Diagnostic and Statistical Manual of Mental Disorders (5th edition: DSM-5) proposes a reclassification of dementia and related illnesses such as MCI captured in the new term Mild (versus Major) Neurocognitive Disorder (NCD) due to AD [34]. This new definition is broad and encompasses not only Petersen et al. [35] defined amnestic MCI, but also states such as Cognitive Impairment no Dementia (CIND) [12]. The criteria for capturing MCI using the most recent definitions are outlined in Table 1. New criteria attempt to improve certainty of diagnosis and harmonize methods for detecting clinical and preclinical cases in research settings and clinical practice. However, there are a number of limitations. First, while there is some overlap (e.g., focus on cognition), there are differences (e.g., in the type of cognitive impairment and the necessity for biomarkers in diagnosis). Second, individuals with no dementia or MCI can be pathologyand biomarker-positive. This raises questions regarding the validity of such measures in diagnostic criteria. Third, no established cut-offs exist for the different biomarkers and cut-offs may vary by factors such as age and the nature of the population being investigated (e.g., unselected versus highly selected clinical sample). Fourth, there are also the issues of cost of attaining each biomarker and their availability, which influences utility. Fifth, having so many different criteria raises
cte
157
rre
156
both Alzheimer-type pathology and vascular pathology [22]. Fourth, results from clinical trials have been largely negative raising questions regarding the suitability of MCI as a treatment target including the interpretation that the MCI stage might be too late for intervention to change natural history. Attempts to keep the disease concepts alive and the ability to measure potential neurobiological phenomena in vivo has led to the awareness of the lack of robust evidence linking these neurobiological phenomena to clinical manifestations. However, instead of looking at the biological measures to see what their meaning is, studies have largely been designed to prove the utility or value of the measure in selected samples. Such studies have been the driver of how AD and MCI is now conceptualized. Revision to the ever changing “lexicon of AD”, by the International Working Group (IWG) in 2007, 2010, and again in 2014, has resulted in a shift of the boundary between clinical and non-clinical states [23–25]. With regard to MCI, rather than being considered as an at-risk state, MCI is instead seen as the earliest manifestation of clinical AD and encompassed in the new term “prodromal AD”. Going back to earlier concepts, this now fits better with previous efforts to characterize intermediate states such as the Cambridge Examination for Mental Disorders of the Elderly (CAMDEX) definition of minimal dementia [26]. Based on the IWG criteria, the diagnosis of prodromal AD includes an amnestic deficit and “abnormalities” on one or more biomarkers thought to reflect biological, structural, or molecular evidence of AD. In addition to revision by the IWG, there have also been attempts to stage MCI into early versus late MCI, where early MCI cases have milder episodic memory impairment compared to the late MCI group [27]. Moreover, attempts have been made to develop new criteria to capture early preclinical states including for example, pre-MCI that captures individuals with impaired executive function and language, higher apathy scores, and lower left hippocampal volumes compared to normal controls [28]. The National Institute on Aging and the Alzheimer’s Association (NIA-AA) have also proposed new clinical and research criteria for detecting AD [29], “MCI due to AD” [30], and preclinical AD [31]. Key differences to the old MCI criteria include extension beyond neuropsychological impairment and inclusion of biomarkers to aid with improving certainty in diagnosis as with the IWG criteria. With regard to detecting preclinical AD, a new framework, designed specifically for research, has been proposed based on neuropsychological impairment
co
155
Un
154
3
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257
4
B.C.M. Stephan and C. Brayne / Dementia Detection Table 1 Updates to diagnostic criteria for mild cognitive impairment (MCI) Terminology for MCI
Criteria
Petersen et al. [35]
Amnestic MCI (aMCI)
• Subjective memory complaint (self or informant) • Normal general cognitive function • Objective decline in memory • No impairment in Activities of Daily Living • No dementia • Amnestic syndrome of the hippocampal type (assessed using the Free and Cued Selective Reminding Test). Impairment can be isolated or associated with other cognitive/behavioral changes • Positive on one or more biomarkers reflecting structural (atrophy of medial temporal lobe on MRI), biological (changes in cerebrospinal fluid markers), metabolic (regional hypometabolism on PET) or molecular (amyloid ligand retention on PET) evidence of disease • Concern regarding cognitive change (individual, informant, or clinician) • Cognitive impairment (one or more domains) and longitudinal evidence of decline (where available) • Preservation of independence in functional abilities • Positive or negative on biomarkers reflecting molecular neuropathology (to determine that the etiology of MCI and cognitive symptoms are consistent with AD)∗ • Reported history consistent with AD genetic factors (where relevant) • Exclusions: vascular and other medical causes of cognitive decline (where possible) • Cognitive decline from a previous level of performance in one or more cognitive domains (complex attention, executive function, learning and memory, language, perceptual-motor, or social cognition) based on: 1) concern of the individual, a knowledgeable informant, or the clinician that there has been a mild decline in cognitive function; and 2) a modest impairment in cognitive performance, preferably documented by standardized neuropsychological testing or, in its absence, another quantified clinical assessment • Cognitive impairment does not interfere with independence (i.e., complex instrumental activities of daily living such as paying bills or managing medications are preserved, but greater effort, compensatory strategies, or accommodation may be required) • Cognitive symptoms are NOT due to delirium • Cognitive symptoms are NOT due to another mental disorder (e.g., major depressive disorder, schizophrenia)
Prodromal dementia
MCI due to AD
dA
uth
NIA-AA clinical (and research) criteria [30]
or P
IWG [23]
roo f
Source
Mild Neurocognitive Disorder
∗ Recommended
259 260 261 262 263 264 265 266 267 268 269 270 271
for research only.
the question of which to apply and where the different terms sit relative to each other [36, 37]. Lastly, the focus has been mainly on AD and therefore the new criteria (with the exception of the DSM-5 where Mild and Major Neurocognitive Disorder due to vascular disease, traumatic brain injury, Lewy Body Disease (LBD), and several others are also specified), generally fail to address dementia associated with other wellknown etiologies, such as vascular disease, LBD, and dementia of mixed etiology (e.g., co-occurring AD and vascular pathology). Detection of preclinical at-risk states associated with, for example, vascular dementia and LBD has been suggested using criteria for Vascular Cognitive Impairment no Dementia [38] and LBD-
Un
258
co
rre
cte
DSM-5 [34]
MCI [39], respectively. However, relative to MCI due to AD these criteria have received much less attention and at present lack consensus and evidence of clinical utility. Even less so is their combined presence in so many people with ‘usual’ dementia.
PREDICTIVE VALUE OF AD BIOMARKERS IN MCI In order to better distinguish progressive from non-progressive MCI, the added value of clinical observation (e.g., neuropsychological screening or clinical opinion), health related variables (e.g., ane-
272 273 274 275 276
277 278
279 280 281 282
B.C.M. Stephan and C. Brayne / Dementia Detection
5
Table 2 Non-modifiable and potentially modifiable risk factors for cognitive decline and dementia Potentially modifiable risk factors
• Age (increasing) • Gender (female)
• Educational and occupational attainment (low) • Socio-economic status (poor) • Social characteristics (e.g., loneliness, involvement in cognitive stimulating activities) • Health status (cardio-metabolic and cerebrovascular conditions such as diabetes, hypertension, high lipids, stroke, obesity) • Psychiatric conditions (e.g., depression)
roo f
Non-modifiable risk factors
• Family History • Genetics (e.g., APOE, non-APOE, presenilin1 and presenilin2) • Down’s Syndrome
288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304
305
306 307 308 309 310 311 312 313 314 315 316
uth
287
sive work to identify risk and protective factors, their utility for ascertaining risk of cognitive decline and dementia through prediction models remains unclear. Further, associations between risk factors and cognition can vary due to age (e.g., timing of risk factor assessment), ethnicity, gender (e.g., possibly due to differences in longevity and hormones between men and women), educational attainment (e.g., the cognitive reserve hypothesis), and the population being studied (clinical versus population-based). Therefore any method for assessing risk factors for cognitive impairment and dementia will need to take into consideration effect modifiers (e.g., age, gender, and ethnicity) and the possible interaction between variables themselves and across the life course.
cte
286
rre
285
mia, depression), CFS (e.g., Total Tau, A42 ), MRI (e.g., medial temporal lobe and hippocampal atrophy), 18 F-FDG PET and genetic (e.g., APOE, CLU, and PICALM), and blood-based biomarkers (e.g., plasma metabolite changes) have been tested. Generally, supplementing MCI criteria has been found to improve prediction of conversion from MCI to AD [40–45]. However, most studies have been undertaken in selected samples (e.g., participants in the Alzheimer’s Disease Neuroimaging Initiative: ADNI) and whether results replicate across settings (e.g., clinical versus community or population-based) needs to be tested. In addition, further research is needed to determine what combination of variables offer the highest predictive power (e.g., looking at simple versus complex variables or a combination of both types of variables). Indeed, MRI, CSF, or blood-based biomarkers are costly, invasive, time-consuming, and require specialist expertise for their attainment (and analysis), and therefore their use in risk prediction would be expected statistically as well as clinically to enhance risk prediction.
co
284
Un
283
dA
• Head trauma
or P
• Poor neuropsychological test performance (e.g., memory, non-memory and global cognitive function) • Lifestyle factors (physical inactivity, poor diet/nutrition, smoking and excessive alcohol use)
RISK AND PROTECTIVE FACTORS
Various non-modifiable and potentially modifiable factors have been linked to an increased risk of cognitive impairment and dementia including, for example, demographic (e.g., age, gender), socio-economic (e.g., educational and occupational attainment), health (e.g., diabetes, hypertension, obesity, stroke), lifestyle/exposure (e.g., smoking, alcohol, pesticides, Mediterranean diet and fish intake), and genetic (Apolipoprotein (APOE) 4 status) variables, as highlighted in Table 2. For a recent review of risk factors for dementia, see [46]. However, despite exten-
MODELS FOR PREDICTING RISK OF COGNITIVE DECLINE AND DEMENTIA Numerous models have been constructed for predicting future risk of dementia, and more specifically AD (for a systematic review of current dementia risk prediction models, see [47]). Examples of variables incorporated into the different risk models are shown in Fig. 1. Based on their component variables risk models can be broadly dividing into: 1) cognitive based models; 2) health indices; 3) genetic risk scores; 4) blood-based molecular signatures; and 5) multivariable models typically incorporating demographic, health, and genetic factors. In total, over 30 different models have been identified [47]. However, a systematic review of risk models concluded that at present none could be recommended for dementia risk screening because of a number of weaknesses [47]. First, an appraisal of indices of model performance such as discriminative accuracy found
317 318 319 320 321 322 323 324 325 326 327 328 329 330 331
332 333
334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350
B.C.M. Stephan and C. Brayne / Dementia Detection
uth
or P
roo f
6
355 356 357 358 359 360 361 362 363 364 365 366 367 368 369
370
371 372 373 374 375 376
cte
354
rre
353
that most models lack sufficient diagnostic power. Predictive accuracy, measured using the area under the curve (AUC) or the c-statistic, ranged from low (AUC = 0.49) [48] to high (AUC = 0.91) [49]. Second, the majority of models have been built in Caucasian volunteer samples that tend to have biases such as high education levels and better health than the general population. As such, current risk prediction models may not be generalizable to other populations, particularly those with different ethnic and socio-demographic characteristics. Third, very few models (
