Age-related mild cognitive deficit: a ready-to-use concept?

Abstract

For better management of mild cognitive impairment in elderly patients, clinicians should be provided with instruments to detect early changes and predict their progression. To define this cognitive status between optimal and pathological aging, many concepts have been proposed, which actually describe various conditions and provide more or less precise criteria, leaving room for variable implementation. As a consequence, application of these criteria gave highly variable prevalence rates, Neuropathological studies indicate that the different criteria have variable power in detecting incipient Alzheimer's disease (AD) and suggest that the transition between mild cognitive impairment and ÀD is not merely quantitative. Follow-up studies have produced, according to the criteria used, a 2.5% to 16,6% annual rate for progression toward dementia, and have also shown that the criteria differ in their stability and predictive power. Baseline cognitive performances have some predictive value, but are difficult to apply in first-line medicine. Investigational techniques (structural and functional imaging, magnetic resonance spectroscopy, magnetization transfer imaging, cerebrospinal fluid neuro-chemistry, and apolipoprotein E genotype) are promising tools in the early diagnosis of AD, which remains the most frequent type of dementia in elderly people and probably the most frequent type developed by patients with mild cognitive deficit. The final goal is to offer early treatment to those patients who will evolve towards dementia, once they can be identified, in the case of AD, recent findings question the adequacy of cholinergic replacement therapies. In its current state, the criteria for mild cognitive deficit are hardly transferable to first-line medicine. However, disseminating the concept could help increase the sensitivity of general practitioners to the importance of cognitive complaints and signs in their elderly patients.

Para un mejor manejo del deterioro cognitivo leve en los pacientes ancianos, los clínicos deben contar con instrumentas para detectar cambios precoces y predecir su progresión. Para definir este estado cognitivo entre el envejecimiento óptimo y el patológico se han propuesto muchos conceptos, los cuales incluyen diversas condiciones y proporcionan criterios más o menos precisos, dejando libertad para una implementación variable. Como consecuencia de esto, la aplicación de estos criterios determinó frecuencias de prevalencia altamente variables. Los estudios neuropatológicos indican que los diferentes criterios tienen un poder variable para detectar la Enfermedad de Alzheimer (EA) incipiente y sugieren que la transición entre el deterioro cognitivo leve y la EA no es meramente cuantitativa. Los estudios de seguimienio han determinado - de acuerdo con los criterios utilizados - una frecuencia anual de progresión hacia la demencia entre un 2,5% y un 16,6%, y también han mostrado que los criterios difieren en su estabilidad y poder predictor. Los resultados cognitivos basales tienen algún valor predictor, pero son difíciles de aplicar en la atenciôn primaria. Las têcnicas paradinicas (las imágenes estructurales y funcionales, la resonancia magnética por espectroscopía, las imágenes por transferencia de magnetización, la neuroquímica del líquido céfalo-raquídeo y el genotipo de la apolipoproteína E) constituyen herramientas promisorias en el diagnóstico precoz de la EA, la cual se mantiene como el tipo más frecuente de demencia en la población anciana y probablemente también el más frecuente que se desarrolla en pacientes con déficit cognitivo leve. El objetivo final es ofrecer un tratamiento precoz a aquellos pacientes que evolucionarán hacia la demencia, una vez que ellos se hayan podido identificar. En el caso de la EA los hallazgos recientes cuestionan la conveniencia de terapias colinérgicas de reemplazo. Actualmente los criterios para el déficit cognitivo leve son difíciles de aplicar a la atención primaria. Sin embargo, la divulgación del concepto podría ayudar a aumentar la sensibilización de los medicos generales respecto a la imporiancia de signos y quejas cognitivas en sus pacientes ancianos.

La prise en charge efficace des troubles cognitifs légers chez les patients âgés implique que les médecins puissent disposer d'instruments capables de détecter des modifications précoces et de prévoir leur évolution. Divers concepts, décrivant en fait des situations variées et utilisant des critères plus ou moins précis, ont été proposés pour définir cet état cognitif intermédiaire entre vieillissement optimal et pathologique. Par conséquent, l'application de ces critères a fourni des taux de prévalence eux-mêmes très variables. Les études neuropathologiques montrent que les différents critères ont un pouvoir variable pour détecter une maladie d'Alzheimer débutante (MA) et suggèrent que la transition entre trouble cognitif léger et MA avérée n'est pas seulement d'ordre quantitatif. Les études de suivi ont montré que les critères utilisés diffèrent quant à leur pouvoir prédictif et leur stabilité, aboutissant à des chiffres annuels de progression vers la démence variant entre 2,5 % et 16,6 %. Les performances cognitives basales ont une certaine valeur prédictive mais sont difficiles à appliquer en médecine de premier recours. Les examens paradiniques (imagerie fonctionnelle et structurelle, spectroscopie par résonance magnétique, imagerie par transfert de magnétisation, neurochimie du liquide céphalorachidien et génotype des apolipoprotéines E) sont des outils prometteurs dans le diagnostic précoce de la MA qui reste le type de démence le plus fréquent chez les sujets âgés et probablement la pathologie qui se développe le plus souvent chez les patients atteints de troubles cognitifs légers. Le but final est d'offrir un traitement précoce à ceux qui évolueront vers la démence dès qu'ils auront pu être identifiés. Pour ce qui est de la MA, des observations récentes remettent en cause les traitements substitutifs cholinergiques. Dans leur état actuel, les critères du déficit cognitif léger sont difficilement applicables en médecine de premier recours. Néanmoins, la diffusion de ce concept pourrait aider à sensibiliser les médecins généralistes à l'importance des manifestations tant subjectives qu'objectives indiquant une atteinte cognitive chez leurs patients âgés.

Keywords:

• aging
• cognition
• dementia
• Alzheimer's disease
• treatment

While promising therapeutic strategics are being explored, our capacity to diagnose dementias early in their evolution remains poor. Degenerative dementias are insidious and progressive in nature. It is therefore conceivable that a dementia picture is preceded by a “preclinical state” (ie, pathognomonic cerebral lesions coexisting with normal cognition) as described in Alzheimer's disease (AD),^{1, 2} followed by mild deficits first experienced by patients themselves, then suspected by their family members, and eventually demonstrated through neuropsychological examination.

It is generally assumed that, normal aging involves cognitive changes, displaying large inter- and intraindividual variability.³ Some studies challenged this common view, showing that the use of strict, criteria for the inclusion of cognitively normal subjects in longitudinal studies demonstrated long-term stability.^4-7 It. was therefore argued that the elderly populations who were the basis of the “normal cognitive aging” concept were contaminated by individuals with very mild dementia.^{1, 8, 9}

As a result, there is currently no consensus on the definition or on the meaning of mild cognitive deficit, in an older individual, or on the attitude it should trigger in physicians. Periodic reassessment until the criteria for a dementia syndrome are fulfilled, as is currently practiced, avoids the risks of overdiagnosis, but conveys those of delaying the initiation of an effective treatment. This daily clinical dilemma would be resolved if physicians were provided with simple instruments allowing a clear differentiation between normal and prodromal cognitive status in a given elderly patient. The goal of this review is to assess to what, extent this need is currently met.

Main concepts and criteria

Since Kral's benign senescent forgetfulness,¹⁰ several concepts have been proposed to understand this shadowy zone between optimal and pathological cognitive aging (Table I).^10-23

Table I Age-related mild cognitive deficit: definitions and criteria.^10-23 AACD, aging-associated cognitive decline; AAMI, age-associated memory impairment; ACMI, age-consistent memory impairment; BVRT, Benton Visual Retention Test; CAMDEX, Cambridge Examination for Mental Disorders in the Elderly; CDR, Clinical Dementia Rating; DSM-III-R, Diagnostic and Statistical Manual of Mental Health Disorders. 3rd ed, revised; GDS, Global Deterioration Scale; HIS, Hachinski Ischemia score; HRSD, Hamilton Rating Scale for Depression; ICD-10, International Statistical Classification of Diseases and Health-related Problems. 10th revision; LLF, Late-life forgetfulness; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; WAIS, Wechsler Adult Intelligence Scale; WMS, Wechsler Memory Scale.

Cognitive impairment-no dementia (CIND) identifies cognitive impairment associated with various conditions, ranging from age-associated memory impairment (AAMI)¹⁵ to cerebrovascular or general vascular diseases, to depression.^{22, 24} Mild cognitive disorder in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) ²⁰ and mild cognitive disorder in the International Statistical Classification of Diseases, 10th Revision (ICD-10) ¹⁸ refer to the cognitive consequences of somatic disorders. Limited dementia¹² and minimal dementia¹⁴ clearly refer to a dementia state. AAMI¹⁵ and age-consistent memory impairment. (ACMI)¹⁶ address normal cognitive aging. Zaudig's mild cognitive impairment,¹⁷ mild cognitive decline,¹³ and questionable dementia¹¹ rely on global scores on cognitive -behavioral rating scales. The major drawback of this approach is that the same score can reflect different clinical profiles, making clinicopatho logical correlation and betwecn-study comparison difficult. Late life forgetfulness (LLF)¹⁶ assesses cognitive deficits relative to what, is considered as normal for age, and aging-associated cognitive decline ( AACD)¹⁹ compares in addition with an education- and gender-matched relatively healthy sample. Both provide explicit, inclusion and exclusion criteria and - in the case of LLF - examples of tests. LLF focuses on memory impairment, whereas AACD considers additional cognitive domains (attention and concentration, problem-solving and abstraction, language, and visuospatial function) and enforces a 6-month duration of decline. The Mayo Clinic criteria for mild cognitive impairment (MCI)²¹ are less precise and their formulation has changed with time (Table II, page 66) .^{21, 25-33} As a consequence, the heading ”MCI“ covers highly variable diagnostic methodologies, hampering comparisons of studies from different research teams.

Table II. Definition and criteria for mild cognitive impairment (MCI).^{21, 25-26} ADL, activities of daily living; CDR, Clinical Dementia Rating; DSM-III-R, Diagnostic and Statistical Manual of Mental Health Disorders. 3rd ed, revised; MMSE, Mini-Mental State Examination. *Although CDR score^{32, 33} was not a criterion, MCI subjects had CDR=0.5.

These different concepts and criteria have seldom been compared in the same population. In a recent, study,³⁴ 111 subjects with informant, evidence of cognitive decline were classified as AAMI (n=37, 33.3%) after clinical assessment. When AACD criteria were also applied, they were fulfilled by 39 subjects (35.1 %), including 20 (54%)” of the AAMIs. Moreover, as illustrated in Figure 1 (seepage 66), the cognitive profiles of subjects with AACD or AAMI were different, with 35.9% of AACDs vs 27% of AAMIs impaired in the memory and learning domain according to AACD criteria (ie, at least 1 SD below age-appropriate norms), and 35.9% AACDs vs 18.9% AAMls impaired in more than one cognitive domain.³⁴

Figure 1. Cognitive profile in age-associated memory impairment (AAMI) and aging-associated cognitive decline (AACD) subjects, according to the data in reference 34. Memory: according to AACD criteria (at least 1 SD below age-appropriate norms). IMI, isolated memory impairment; >1 function, impairment in more than one function.

As expected according to their individual definitions and goals, the AAMI and AACD concepts only modestly overlap one another; the latter captures a more severe impairment.

In the Canadian Study of Health and Aging, specific criteria were applied in subjects classified as CIND.²² Sixty-five percent did not meet any of them; none met the AAMI criteria of Bradford and LaRue.¹⁶ When inclusion criteria were applied alone, 8.1% fitted the criteria for AAMI. 5.9% for ACMI, 7.4 % for LLF, and 34% AACD; after applying exclusion criteria, these figures dropped to 1.2 % (AAMI), 0.9 % (ACMI), 0 % (LLF), and 13 % (AACD). These data highlight the importance of exclusion criteria resulting from comprehensive clinical evaluation. Only 24% of those meeting one set of criteria also met one other or more (19.2 % met two, 3.8 % three, and 0.8 % four), suggesting that the different sets of criteria are mutually exclusive. In a sample of 60 - to 64 - year-old healthy people, ³⁵ 13.5% met criteria for AAMI, 6.5 % for ACMI, 1.5 % for LLF, and 23.5 % for AACD. Among subjects with AAMI, 22 % met the criteria for ACMI, 11 % for LLF, and 63 % for AACD. All the LLF subjects also fulfilled criteria for both AAMI and AACD.

Together these results are not very surprising. First, the different, sets of criteria refer to different concepts (AAMI and ACMI versus LLF and AACD, sec above). Second, they basically consist, of a priori constructs, differing in of clinical populations. Third, because of criteria's methodological vagueness (eg, no firm reference tests; no indication on whether one function should be assessed using one or several tests), they offer room for different, implementation across teams. The impact, of introducing changes in criteria is illustrated by the Eugeria Project.³⁶ Of 833 subjects recruited, 308 fulfilled the first two criteria for MCI (subjective memory complaint and normal general intellectual functioning, as assessed by performance on a vocabulary test); of these, 103 had a decrement of more than 1 SD on a memory task, relative to normal values for age and educational level (criterion 3); exclusion of subjects with difficulties in any other cognitive domain left only 27 subjects fulfilling the criteria; application of criterion 4 (normal activities of daily living) had no influence. Thus, modification of the third criterion reduced the prevalence of MCI from 12.4% to 3.2%. The AACD criteria applied to the same population identified 174 participants (20.8%), which included all the M.C1 subjects.

Neuropathological correlates

To the best of our knowledge, the only concept that has been compared with neuropathological examination is MCI as defined by the Mayo Clinic team.²¹ In a followup study,³⁷ 6 out. of 6 subjects with a Clinical Dementia Rating (CDR)³² score 0.5 resulting from memory impairment alone were found to meet modified³⁸ Kachaturian³⁹ neuropathological criteria for AD. This confirmed previous data⁴⁰ showing that 10 out of 10 subjects with CDR=0.5 had histopathological AD, versus none of 4 with a score of 0. In another study,⁴¹ subjects with a CDR>0.5 had large senile plaque densities in the neocortex and the degree of dementia seemed related to an increase in the ratio of neuritic to diffuse plaques. While cognitively healthy controls - and even individuals with preclinical AD - had no significant, decrease in neuronal count in the entorhinal cortex (ERC) as a whole, in ERC layer II or in the CA1 hippocampal field, the brains of subjects with CDR=0.5 were characterized by a significant neuronal loss in these areas.⁴² These studies suggest. that “questionable dementia” or isolated memory impairment sufficient to yield CDR=0.5 actually represent very mild AD. It can be questioned whether CDR=0.5 equates to MCI. A series of studies^43-45 compared MCI subjects (defined as being impaired in one domain on neuropsychological testing, but. not being found to have dementia by the examining neurologist according to NINCDS/ ADRDA [National Institute of Neurological and Communicative Diseases and Stroke/ Alzheimer's Disease and Related Disorders Association] criteria⁴⁶) with normal controls (NCs) and AD patients, all from a group of catholic clergy participating in the Religious Order Study (Table III). Both AD patients and MCI subjects had a lower ERC layer II neuronal count, remaining neuronal volume, and global volume. However, MCIs and AD patients significantly differed in layer II global volume only, despite a group effect in analysis of variance, while all the AD and NC values were significantly different. Global and neuronal atrophy were correlated with impairment of delayed and immediate recall.⁴⁴ Quantified ERC β-amyloid (βA) load⁴³ in MCIs was intermediate, but not significantly different from that found in NCs and AD patients respectively (again, NC vs AD values were significantly different), although analysis of variance revealed a significant group effect with a trend to linear increase from NCs to MCIs to AD patients. It is noteworthy that, some NC or MCI subjects had PA load equal to or higher than that seen in many AD patients, and two MCIs had no detectable βA. The inverse correlation between Mini-Mental State Examination (MMSFs)²⁷ scores (mean scores: 27.3 in NCs and MCI patients, and 24 in AD patients) and βA load was not significant. In this study, 6 of 12 MCI subjects had a neuropathological diagnosis of possible AD according to the criteria of the Consortium to Establish a Registry for Alzheimer's Disease (CERAD).⁴⁷ In a third study,⁴⁵ neurofibrillary tangles (NFTs) and neuropil threads (NTs) were present in perirhinal cortex and ERC in NCs, MCI subjects, and AD patients; the average number of NFTs increased with the diagnosis from NC to M'CI to AD. Retween-group differences analysis again found that MCI subjects were intermediate, but not significantly different from either NCs or AD patients. NFT density was inversely correlated with episodic memory score, but not with other, nonmemory, cognitive abilities across the three groups.

Table III. Neuropathological characteristics of mild cognitive impairment in the Religious Order Study.^43-45 CERAD, Consortium to Establish a Registry for Alzheimer's disease; MCI, mild cognitive impairment; NS, not significant; S, significant. Changes in MCI and AD group are given relative to normal controls.

These studies first, show that no more than 50% of MCIs were incipient. AD. This is less than in the studies by Morris^{37, 40} and Price,⁴¹ and suggests that the populations described were not equivalent, although the use of different neuropathological diagnostic criteria makes the comparison difficult. Approximately the same proportion of NCs (45%) were also diagnosed as possible AD; this finding suggests that the clinical diagnostic tools were neither sensitive nor specific in the detection of incipient AD. This can be explained by the fact, that both the NC and MCI groups had high and similar MMSE scores, but the concept of MCI precisely intends to detect cases missed by more global testing. Nevertheless, MCI subjects globally were, for most. F,RC lesions, intermediate between NC and AD cases. This suggests that ERC lesions could be a better neuropathological marker of MCI than the presence of those required for a diagnosis of AD. However, the fact, that group-to-group comparisons failed to distinguish MCIs from ADs makes currently impossible to determine practically useful cutoff values. If this failure is due to sample size, larger studies should solve it. However, it. could also be due to heterogeneity; clinicopathological studies seeking pathological markers of both non- AD dementias and AD should confirm or rule out this possibility Awaiting further studies, the lack of significant difference between MCI and AD, which was also found for high (trkA)⁴⁸ and low (p75^NTR) ⁴⁹ expression of nerve growth factor receptors, suggests that the transition from MCI to AD is not merely quantitative.

Predictive value

Another way of understanding these concepts and criteria is through their ability to predict, the progression of patients. Follow-up studies^{21, 25, 36, 37, 50-59} differ in their durations, making comparisons difficult; dividing the frequency of progression toward dementia by duration of follow-up gives an estimate of the annual rates of “conversion” (Table IV).

Table

Thus, a significant proportion of subjects did not become demented. It could be argued that a longer follow-up would increase the “conversion” rate. However, data from some studies reporting multiple evaluations^{21, 58, 60-62} suggest that the incidence of dementia could decrease over time. In a recent study with assessments at 3 and 6 years in subjects with CIND, aged 80 years or older,⁶¹ it was found that, according to the severity of impairment at baseline, 84% to 89% of those who were demented at 6 years had already received the diagnosis at 3 years. In another study in oldest old (84 to 90 years old at baseline) over 6 years,⁶⁰ a decrease in the progression from MCI to dementia with time was also reported. 'Ms attenuation of the rate of progression with time could be an artifact, since in these two studies - and also in one in slightly younger subjects⁵⁷ - MCI increased the risk of death by 1.7⁵⁸ to 7⁶⁰ during a 4-year period. In this case, there should be a correlation between the severity of cognitive impairment at baseline and the risk of death. Such a trend was found in one study,⁵⁸ but. not. in another,⁶¹ and in a third⁶⁰ baseline performances in the deceased group were lower than those of survivors, but. higher than for those who progressed to dementia. Thus, the issue of the slope of the rate of progression deserves further attention, particularly in relation to age at. onset, of cognitive impairment.

Because the main criteria were set. to capture degenerative cognitive impairment (ie, without identifiable medical cause), an intriguing finding is that a substantial proportion of subjects were found to improve over time (4.8% after 3 years in subjects with CDR=0.5⁶³; 19.5% after 2.7 years in MCI as defined by Zaudig⁵⁴; 25% after 3 years and 12% to 17% after 6 years in CIND⁶¹). In clinical practice, such an outcome would be ascribed to a diagnostic error (ie, impairment, was due to a unidentified medical condition). An alternative explanation is that the underlying process is different from AD. Indeed, a fluctuating course is classically described in vascular dementia (VaD)⁶¹ and dementia with Lewy bodies (DLB).⁶⁵ In line with this hypothesis is the finding that, in a sample of MCI subjects, 20.5% developed VaD within 3.9 years; nothing in their baseline cognitive profile or their progression (based on MMSE) differentiated them from those who progressed to AD (47.9%).^62-66 A third explanation is that the criteria do not. describe a stable state. The Eugeria Project compared MCI (with impairment, in memory, but not in any other domain) and AACD over 3 years,³⁶ and showed that. 7.5 % of MCI subjects retained the diagnosis from the first, to second assessment and 17.4 % from the second to third; the corresponding figures for AACD subjects were 56.3% and 59.4 %. Apart from those who became demented, subjects met criteria for the alternative diagnosis (from MCI to AACD and vice versa) or were found to be normal. In this study, the AACD diagnosis had a sensitivity of 94.7 % and a specificity of 54.1 %, whereas the MCI diagnosis had a sensitivity of 5.3 % and specificity of 91.3 % in the prediction of progression toward dementia after 2 years. In another community-based French study,⁵⁹ the MCT diagnosis was also found to be unstable.

According to the cited studies, there is no doubt that mild cognitive deficit in elderly subjects, whatever its definition and criteria, increases the risk of developing dementia. The available data provide a rather broad range of annual incidence of dementia and are not all in favor of a linear prevalence-time relationship in mildly impaired patients. The proposed sets of criteria have different, stability and predictive values. Also, they do not allow identification of individuals who will develop dementia or - more importantly - the type of dementia toward which they could evolve.

Beyond the criteria themselves, several studies found predictors of progression to dementia or even to AD in measures derived from the MMSE,⁶² the CDR,⁶³ or impairment in memory, verbal fluency, and attention on more conventional neuropsychological tests.^{52, 67-69} As pointed out by Tuokko and Frcrichs,⁷⁰ a major shortcoming of these data is that, they are retrospective. No combination of cognitive tests has yet been assessed prospectively for its ability to predict, outcome in mildly impaired patients. If it were done using neuropsychological batteries that, were sufficiently refined for early identification of the characteristic signs of the major dementing diseases and determination of reliable cutoff scores, then this type of investigation would be reserved for specialized teams; however, the first person who people with cognitive complaints see is their general practitioner. It is expected that this dilemma will be partly solved in the near future by recourse to investigational techniques.

Contribution of investigational techniques

For decades, the scientific community has been seeking biomarkers of AD, using genetics, neurochemistry, and imaging techniques. It was rational to apply these techniques to mild cognitive deficit, in order to characterize these states and identify predictors of progression to AD. Neuropathological studies have shown the hippocampus to be one of the earliest, affected structures in AD,⁷¹ and so it is a region of choice for neuroimaging studies. Although hippocampal atrophy, as measured by volumetric techniques is not entirely specific, it is now considered to be a hallmark of AD,⁷² and its absence in addition to minor or unilateral atrophy is believed to be strong evidence against the diagnosis. In mild cognitive deficit, several studies have shown lesser^73-76 or similar⁷⁷ hippocampal atrophy to that found in AD. Age transformation of combined hippocampal and amygdala volume increases the accuracy of classifying AD, MCI, and normal elderly subjects.⁷⁸ MCI subjects had hippocampal volume correlated with cognitive and performance measures⁷⁹ and those who declined over time had also a greater annualized rate of hippocampal atrophy than nondeclincrs, close to that of AD patients.⁸⁰ Atrophy of various regions at baseline, including hippocampus^{79, 81, 82} ERC,^83-84 fusiform gyrus,⁸⁵ caudal cingulate cortex,⁸³ and medial temporal lobe,⁷⁶ was found predictive of progression to AD. White matter lesions have been found to be associated with subjective cognitive decline,⁸⁶ lowered attention and speed of mental processing,⁸⁷ and progression to dementia.⁸⁸

There is an agreement on the fact that established AD is characterized by altered cerebral blood flow (CBF) and metabolism in posterior parietal and temporal lobes as well as by, according to stage and neuropsychological profile, frontal cortex deficits, and hemispheric asymmetry.⁸⁹ That functional imaging is able to detect preclinical AD is suggested by positron emission tomography (PET) studies, which found regional cerebral glucose metabolism (rCMRGlu) alterations in nondemented subjects at risk of AD (ie, those carrying the apolipoprotein E type 4 allele [ ApoE ε4] and with familial history of AD)^{90, 91}90-91; those in the inferior parietal and posterior cingulate cortices correlated with later memory decline.⁹²

Studies comparing CBF and rCMRGlu in normal and mildly impaired subjects found deficits in the latter, in various regions including bilateral parietal cortex,⁵⁶ hippocampus,⁷⁷ and posterior cingulate gyrus.^93-95 Prediction of outcome was found for defects in parietal or temporoparietal cortex,^{56, 96} posterior cingulate gyrus,^{94, 95} and for temporoparietal asymmetry⁹⁷ and lowered postcroanterior ratio⁸⁹; others were predictive when combined with performance on specific cognitive tasks^{98, 99} and/or demographic characteristics.⁹⁹

Progress in functional imaging can come from activation studies. Comparisons of young to elderly healthy subjects have shown that poorer performances in tasks such as conflict resolution and episodic memory in the elderly corresponded to underactivation, whereas a performance similar to that of young controls in working memoir}' tasks was accompanied with recruitment, of additional brain regions.¹⁰⁰ Studies comparing activation during cognitive tasks in AD patients and controls^101-105 showed that, together with lower performances, AD patients had activation patterns characterized by absence of activation in some brain areas, activation with shifted peak foci, expansion of normally activated zones, and recruitment, of remote areas.¹⁰³ These differences were generally interpreted as due to compensation efforts; complementary interpretations are disconnection between regions normally involved in the task and predominant processing of accessory aspects of the stimuli (eg, emotional appearance in face recognition).¹⁰⁵ Passive pattern-flash stimulation elicited less activation in AD patients; this failure requires a less demanding stimulation to be disclosed in the modcrate-to-scvere group than in the mild group.¹⁰⁶

Cognitively normal subjects at risk for AD (defined as the presence of at least one ApoE ε4 allele, alone¹⁰⁷ or combined with a history of AD in at. least one firstdegree relative¹⁰⁸) were compared with low-risk controls for activation induced by cognitive tasks they performed with the same accuracy level. In the high-risk group, some regions were activated to a greater extent or magnitude (eg, nearly twice as much as in controls in hippocampal regions¹⁰⁷); others displayed lower activation.¹⁰⁸ After a 2-year follow-up,¹⁰⁷ decline in verbal recall correlated with the number of regions activated in the left hemisphere at baseline.

Using a functional magnetic resonance imaging (fMRI) protocol specifically developed for hippocampal region analysis, one study¹⁰⁹ compared cognitively NCs, subjects with isolated memory impairment (IMI), and AD patients during a simple task (gender discrimination of presented faces); all subjects performed the task with 100% accuracy. AD patients had lesser activation of the three regions studied, ie, ERC, subiculum, and the hippocampus proper. Among the IMI subjects, one third had an activation pattern similar to that of AD patients and the others displayed lesser activation in the subiculum only. Follow-up data would be necessary to determine whether the differences described in this study are predictive, but together these activation studies indicate that properly chosen activation paradigms could help identify AD in subjects with mild cognitive deficits.

Nuclear magnetic resonance affords additional approaches. Magnetic resonance spectroscopy (MRS) can assess the biochemical composition of living brain regions. To date, the most, consistent findings in AD¹¹⁰ have been obtained with proton MRS showing a decrease in N-acetylaspartate (NAA) and an increase in myoinositol (MI). NAA and MI changes arc specific to neither AD nor brain disease, but the NAA/MI ratio can discriminate possible AD cases from NCs. In addition, NAA/MI, NAA, and the MI/creatine (Cr) ratio were shown to be correlated with MMSE score in controls and patients with probable AD¹¹¹; in a 12-month follow-up study¹¹² NAA/Cr and NAA/MI at baseline were correlated with the progression of MMSE. scores. Few studies have used MRS in mild cognitive deficit. Ml/Cr was found to be higher in MCI subjects^{113, 114} and NAA lower in A AMI subjects¹¹⁵ and AD patients than in controls, whereas MI values were intermediate between AD patients and controls.¹¹⁵ Follow-up studies are necessary to confirm the predictive value of such findings.

The magnetization transfer imaging (MTI)¹¹⁶ signal arises from the magnetization exchange between waterand macromolecule -bound protons; this technique is useful in the study of membranes and membrane-linked diseases such as multiple sclerosis, in which decreased magnetic transfer ratio (MTR) is a marker of demyelination¹¹⁷ and axonal density loss.¹¹⁸ MTI studies involving AD patients^119-125 agree on decreased values compared with NCs, expressing structural changes in the temporal lobe, and also the frontal lobe and the whole brain.^119-120 Hippocampal MTR had a discrimination rate relative to controls of 85% in mild AD (CDR=0.5), 89% in mild AD (CDR=1), and 100% in moderate AD (CDR=2); the values for visually rated atrophy were of 73%, 80%, and 91% respectively.¹²⁴ MTR was also able to differentiate AD from non-AD dementia with a success rate of 77%.¹²³ Studies comparing MCI subjects with AD patients and healthy controls^119-122 identified structural changes in MCI in the absence of significant, atrophy; they were located in gray matter, whereas those found in AD patients involved white and gray matters.¹²² These changes were found to be correlated with cognitive impairment.^119-120 MTI thus seems able to identify structural changes before atrophy is manifest. Follow-up studies should confirm its predictive value and comparison with functional imaging should assess which technique detects the earlier changes.

The ApoE ε4 allele is acknowledged to be a risk factor for AD.^{126, 127} Few studies have specifically addressed its influence on the evolution of MCI subjects. The ApoE ε4 carrier status was found the best predictor of conversion to AD (risk ratio=4.36),²¹ a nonsignificant predictor (relative risk of 1 .49)⁷⁹ or to have no predictive value.^{54, 128} In subjects with memory impairment and Global Deterioration Scale (GDS) score of 2 to 3,¹²⁹ ApoE ε4 alone predicted progression to dementia with a 73.8% accuracy; combining genotype and memory scores increased the accuracy to 92.5%. In subjects with MMSE, scores of 21 to 26, the ApoE, genotype was found to be associated with an odds ratio for progression to dementia of 3.31¹³⁰ and with memory decline.¹³¹ Among the various substances that have been assayed in blood and CSF,¹³² increased CSF-tau and decreased βA_1-42 proteins are the best markers for AD to date.^132-134 Their combination yielded a 94% and 88% sensitivity for probable and possible AD, respectively, and specificity of 100% versus psychiatric disorder, 89% versus no dementia, 67% versus DLB, and 48% versus VaD in clinical practice conditions.¹³⁵ CSF-tau levels were found to be higher in MCI than in healthy controls,^136-140 lower than¹⁴¹ or similar to^{139, 140} those found in AD patients. In follow-up studies, it identified MCI subjects who evolved to AD with a sensitivity of 65% to 68% and a specificity of 100% versus patients with memory complaints¹⁴² and 93% versus healthy controls¹³⁹; baseline values in converters were higher than in nonconvcrtcrs.¹³⁷ In one study,¹³⁹ combining CSF-tau and βA, values did not. improve the predictability obtained with CSF-tau alone. In others, combined CSF-tau/βA_1-42 values differentiated converters from healthy controls with 88% sensitivity and 80% specificity¹³⁸ and from nonconverters with a 90% sensitivity and specificity.¹⁴⁰

Medical diagnoses are rarely reached using a single marker; most often it results from the combination of different approaches, including thorough clinical evaluation. Once a consensus is obtained on cutoff values for the different techniques mentioned, it is likely that a combination of different markers for AD will allow early diagnosis with high sensitivity and specificity in individual cases.

Therapeutic aspects

The final goal of constructing criteria for age-associated mild cognitive deficit, is to treat, it, and many therapeutic approaches are available.^143-144

Some benefits have been reported, in terms of global stability and improved memory with the acetylcholinesterase inhibitors (AChEI) donepezil¹⁴⁵ and rivastigmine¹⁴⁶ and the dopamine receptor agonist/α₂ antagonist, piribedil¹⁴⁷; trials are underway using donepezil and vitamin E, rivastigmine, the cyclooxygenase-2 (COX-2) inhibitors celccoxib and rofecoxib with the goal of delaying patients' progression to dementia¹⁴⁴ and the Ampakine® CX516 with the aim of short-term symptomatic improvement.¹⁴⁸ Treating MCI using approaches initially intended for AD premises either that MCI equates to early AD in all cases, which is unlikely, or that the underlying mechanism is the same in both cases, the difference being merely a matter of intensity, which is not. confirmed by neuropathological data. Even if improved criteria or techniques were able to predict, the progression to AD in a given patient, this strategy deserves discussion. As regards the cholinergic and glutamatergic systems in AD, it has been proposed that, the final deficiency state is preceded, in the early stages, by a hyperactive state^{149, 150} originating in βA-induced N-methyl-D-aspartate (NMDA) receptor hypersensitivity. It was recently found that the number of choline acetyltransferase (ChAT)-positive neurons in the nucleus basalis of Meynert was no lower in MCI than in NCs.¹⁵¹ Further, ChAT activity was shown to be increased in hippocampus and in the superior frontal cortex compared with NCs, and similar in anterior cingulate, superior temporal, and inferior parietal cortices.¹⁵² Although such data are lacking for the glutamatergic system, these findings suggest that, among the currently available treatments, those aiming at downmodulating the NMDA receptor responsiveness could be more appropriate than replacement therapies in degenerative, AD-type MCI. Mild cognitive deficit will be the condition of choice for administration of future treatments addressing basic mechanisms of degenerative dementias, provided they can be reliably identified in these patients.

Conclusion

Introducing criteria for mild cognitive deficit should:

• Help its detection, mostly in first-line medicine.

• Improve the accuracy of early dementia diagnosis.

• Through harmonization of practice in research settings, permit progress in pathophysiological and therapeutic research.

All the sets of criteria, whatever their formulation, require the input, of a neuropsychologist and a thorough, comprehensive examination. This approach is available in specialized centers only, and not transferable to first-line medicine. The criteria by themselves do not predict which individual will progress to dementia and still less the nature of this potential disease. On the other hand, early, reliable diagnosis of AD through a proper combination of investigational procedures can be expected in the near future. Epidemiological data show that AD remains the most, frequent dementia type in elderly people.¹⁵³ Followup studies suggest that it. is also the most, frequent dementia type developed by subjects with mild cognitive deficit. Therefore, early identification or rejection of AD would solve the majority of cases. It can thus be thought that the priority is to optimize our battery of investigational tools by defining appropriate cutoff values, comparing them in the same patient samples, and defining their individual powers.

From a practical and clinical point of view, refining the sets of criteria to improve their specificity, which implies skilled professional intervention, is probably useless. Efforts should rather be made to define simple, sensitive tools, usable by general practitioners.

From a research point of view, it seems mandatory to reach a consensus on several points. Should the reference population be matched for age only, or for gender and education as well? Using age-related references implies admitting that cognitive decline occurs in healthy aging; using education-related ones implies that, low education is an independent risk factor for cognitive decline^{154, 155}; using gender requires taking into account the hormonal status of women. Should impairment, in a cognitive domain be established on the basis of a single test, or of several ones addressing the same function? Should studies include subjects with memory impairment, only, or with decline in other domains as well? Stratifying the participants according to their cognitive profile would allow assessment, of the predictivity of this item, relative to the underlying disease, as proposed by Petersen et al.¹⁵⁶ It would also permit comparison of the effect of treatments in these subsets.

These issues cannot, be solved by any single research team. Collaborative or, at least, comparable studies require the strict definition of common basic inclusion (eg, the tests to be used with standard cutoff scores) and exclusion criteria. Before being applicable in daily practice, the available sets of criteria need to be further defined and standardized. The current lack of treatment is a hurdle to its acceptance. However, disseminating the concept could help increase the sensitivity of general practitioners to the importance of cognitive complaints and signs in their elderly patients.

Selected abbreviations and acronyms

AACD	=	aging-associated cognitive decline
AAMI	=	age-associated memory impairment
ACMI	=	age-consistent memory impairment
AD	=	Alzheimer's disease
CDR	=	Clinical Dementia Rating
CERAD	=	Consortium to Establish a Registry for Alzheimer's Disease
CIND	=	cognitive impairment-no dementia
DLB	=	dementia with Lewy bodies
ERC	=	entorhinal cortex
IMI	=	isolated memory impairment
LLF	=	late-life forgetfulness
MCI	=	mild cognitive impairment
MMSE	=	Mini-Mental State Examination
MTI	=	magnetization transfer imaging
NC	=	normal control
NFT	=	neurofibrillary tangle
NT	=	neuropil thread
VaD	=	vascular dementia
βA	=	β-amyloid