Combining Cerebrospinal Fluid Biomarkers and Neuropsychological Assessment: A Simple and Cost-Effective Algorithm to Predict the Progression from Mild Cognitive Impairment to Alzheimer’s Disease Dementia

Abstract

Background: Correctly diagnosing Alzheimer’s disease (AD) in prodromal phases would allow the adoption of experimental therapeutic strategies that could selectively interrupt the pathogenetic process before neuronal damage becomes irreversible. Therefore, great efforts have been aimed at finding early reliable disease markers.

Objective: The aim of this study was to identify a simple, cost effective, and reliable diagnostic algorithm to predict conversion from mild cognitive impairment (MCI) to AD.

Methods: 96 consecutive MCI patients admitted to the Neurology department of San Raffaele Hospital in Milan between January 2009 and January 2015 were included. All patients underwent neuropsychological assessment and lumbar puncture with CSF analysis of amyloid-β 42 (Aβ₄₂), total tau (t-tau), and phosphorylated tau (p-tau) levels. Each patient underwent clinical and neuropsychological follow-up, in order to identify a possible progression from MCI to AD. The mean follow up time was 36.73 months.

Results: 37 out of 96 MCI converted to AD during follow up. CSF analysis and neuropsychological assessment reliably detected MCI patients who developed AD. In a subsample of 43 subjects, a Composite Cognitive Score (CCS) was calculated including episodic memory, executive function, and verbal fluency tests. Combining together CSF biomarkers and CCS increased the accuracy of the single predictors, correctly classifying 86% of patients with a specificity of 96% and a Positive Predictive Value of 93%.

Discussion: Even if preliminary, our data seem to suggest that CSF analysis and neuropsychological assessment could detect MCI patients who will convert to AD with high confidence. Their relative low cost and availability could make them worldwide essential tools in future clinical trials.

Keywords

composite cognitive score diagnostic algorithm mild cognitive impairment neuropsychological tests tau protein

INTRODUCTION

Alzheimer’s disease (AD) represents the number one cause of dementia, accounting for 50% –75% of the total, with a greater proportion in the higher age ranges [1], and affecting more than 35 million people worldwide, with 3 million new cases arising every year [2]. The incidence rate for dementia increases exponentially with age, and the number of affected individuals is predicted to grow with the increasing life expectancy, doubling between 2020 and 2040 [3]. AD greatly impacts on familial environment and social costs, thus requiring substantial efforts in improving diagnosis, treatment, and management. The therapeutic scenario in AD is currently discouraging: only symptomatic drugs are available and can be prescribed only in the full-blown dementia stages. Indeed, these drugs show a wide variability in individual response and a short period of effectiveness, which decreases in 12–24 months after the beginning of the treatment [4, 5]. Thence great attention has recently been addressed to identify possible disease-modifying drugs which could interfere with the AD pathological processes, i.e., anti-amyloid-β₄₂ (Aβ₄₂) antibodies or γ-secretase inhibitors. Unfortunately, up to now, all the candidates have failed in Phase II-III clinical trials [6]. These failures might have been due to the AD stages of patients at enrollment: indeed most clinical trials recruited AD patients in the mild to moderate stages, probably when the disease progression was too advanced to be stopped. Secondly, the absence of strict diagnostic criteria could have led clinicians to enroll misdiagnosed AD patients, thus underestimating the real effectiveness of the experimental drugs. Therefore, correctly diagnosing AD subjects at the very early disease stages will be of crucial importance to develop and test new disease-modifying therapies on patients who could actually benefit from an early intervention [7]. An early diagnosis would assure a better focused management of affected subjects, reducing familial burden and social costs [8]. For these reasons, growing attention has been focused on mild cognitive impairment (MCI) in the last years. MCI is an intermediate condition between the expected cognitive decline observed in normal aging and the pathological decline typical for dementia [9], representing in this case the early stage of the disease. Different conditions, either neurodegenerative or non-degenerative (cerebrovascular, infective, metabolic, or pharmacological), may underlie MCI. When an underlying AD pathology is supposed, MCI can be defined as “MCI due to Alzheimer” according to NIA-AA (National Institute on Aging–Alzheimer’s Association) criteria [10] or “Prodromal AD” according to IWG (International Working Group) [11] and IWG-2 [12] criteria. The new NIA-AA criteria on MCI emphasize the role of neuroimaging and cerebrospinal fluid (CSF) biomarkers in the diagnostic work up. In particular the CSF 42 amino acidic isoform of amyloid (Aβ₄₂), total tau (t-tau), and tau phosphorylated at position threonine 181 (p-Tau) proteins have been extensively investigated as potential in vivo biomarkers of AD pathology [13, 14], both in the early disease stages [15, 16] and in atypical variants [17, 18]. Several studies showed that low Aβ₄₂ associated with high t-tau and p-tau CSF concentrations not only can distinguish AD from other types of dementia [13, 16] but can also detect MCI patients who will progress to AD [19 –22] with high reliability.

On another hand the accuracy of cognitive tests to predict progression from MCI to AD has been largely investigated. Several studies showed that low scores in neuropsychological tests evaluating verbal and visuo-spatial episodic memory, abstract reasoning, learning, language, and executive functions [23 –27] could support the hypothesis that MCI would be due to AD. In particular, several word recall tests (Rey Auditory Verbal Learning Test, [28]; Story Recall, [29]) were found to have good accuracy in predicting progression from MCI to AD. Visual recall and recognition tests are less frequently reported as strong predictors of progression [29, 30]. Tests for non-memory domains did not show high predictive value, unless they were combined with episodic memory tests [31]. Up to now only few studies have investigated the diagnostic reliability and predictive values of CSF biomarkers and neuropsychological tests when combined together in MCI condition, finding an improvement in diagnostic accuracy when used in combination and not as single predictors [28 , 31–33]. In the present study, we firstly aimed to evaluate the diagnostic accuracy of CSF biomarkers and neuropsychological tests both as single and combined biomarkers in identifying MCI due to AD. Secondly, we aimed at defining a reliable diagnostic algorithm that could be easily and largely applied, helping clinicians to correctly diagnose those MCI patients who will develop AD, even in not hyper-specialized clinics.

MATERIALS AND METHODS

Participants and clinical assessment

We considered 137 consecutive patients with a mild cognitive impairment who referred to the Memory Center of IRCCS-San Raffaele Hospital, Milan, Italy, between April 2009 and November 2015, complaining of slight cognitive deficits. During the baseline visit, patients were evaluated by a team of neurologists and neuropsychologists with expertise in neurodegenerative disorders and underwent CSF collection. Clinical evaluation included a structured clinical interview, a full neurological examination, and a standard neuropsychological assessment. During the structured clinical interview with both the patient and the caregiver, clinicians assessed patients’ history evaluating both past and concurrent illnesses, patients’ functional status at work and in usual daily living activities, using ADL (Activities of Daily Living) [34] and IADL (Instrumental Activities of Daily Living) [35] scales, patients’ global cognitive status using general disease severity indices such as MMSE (Mini-Mental State Examination) [36, 37], CDR (Clinical Dementia Rating) and CDR-sb (Clinical Dementia Rating-sum of boxes) [38], patients’ behavioral psychiatric symptoms using NPI (Neuropsychiatric Inventory) [39]. ADL and IADL were expressed as the ratio of the preserved to the total evaluated functions per hundred parts. MMSE and CDR-sb were expressed as raw scores. Inclusion criteria were: 1) CDR equal to 0.5; 2) diagnosis of MCI according to the IWG diagnostic criteria [40]; 3) attainment of the clinical endpoint, i.e., conversion to AD according to the recently revised NINCDS-ADRDA criteria [41] during follow up, regardless of the follow up duration. Exclusion criteria were: 1) the presence of either past or concurrent illnesses which could account for the patients’ cognitive deficits; 2) the complete data loss of patients’ follow up; 3) a follow up visit at less than 12 months from the baseline visit for those patients who did not develop AD. From the initial sample, we excluded 41 patients who did not convert to AD with a follow up shorter than 12 months. Therefore we finally included96 patients. Among these, 6 converted to AD in 12 months from the baseline, while the remaining 90 patients had a follow up time ranging from 1 to 6 years. 72 patients underwent clinical and neuropsychological follow-up every 6 months. 18 patients did not refer to our Memory Centre for follow-up visits: they were interviewed over the phone with a caregiver to obtain relevant information on their cognitive status. Specifically, caregivers were asked whether patients were followed up by other clinicians specialized in cognitive disorders, received a diagnosis either of AD or other neurodegenerative disorder, were told to start therapy with acetylcholinesterase inhibitors or memantine, or showed a clear worsening of cognitive functions determining a reduction in ADL and IADL scale scores. A subset of 43 MCI patients was considered to build up a diagnostic algorithm combining CSF analysis and neuropsychological assessment. On the basis of progression to AD, MCI subjects were sub-grouped into converters (MCI-c) and non-converters (MCI-nc), in the latter including both stable MCI and MCI due to other dementias.

Neuropsychological assessment

We evaluated patients’ core cognitive domains using a well-standardized battery of 15 tests. To assess the memory domain we used the following tests: Rey auditory Verbal Learning test immediate recall (RAVLT-I), delayed recall (RVLT-D), and recognition (RVLT-R) [42], Story Recall (SR) [43]; for attention and executive functions we used: Attentive Matrices (AM) [44], Trail Making Test A, B and B-A (TMTa, TMTb, and TMT b-a) [45]; for language domain: Token Test (TT) [44], Phonemic and Semantic Verbal Fluency (PVF and SVF) [43]; for visuo-spatial abilities: Rey–Osterrieth Complex Figure copy and recall (RFC and RFR) [43] and Geometric Shapes Copy (GSC) [44]; for reasoning: Raven’s Coloured Progressive Matrices (CPM) [46]. Scores at each cognitive test were reported as z-scores.

CSF acquisition and analysis

During the baseline visit all patients underwent lumbar puncture in the L3-L4 or L4-L5 interspace after signing the written informed consent, and following detailed explanation of the procedure in accordance with the ethical standards of the Committee on Human Experimentation of our Institute. The procedure was always performed early in the morning. No serious adverse events were reported. CSF (8–10 ml) was collected in sterile polypropylene tubes. Routine chemical parameters were determined (i.e., leukocyte and erythrocyte cell count, glucose measurement, protein total content); the remaining CSF was centrifuged for 10 min at 4,000 g at 4°C. After centrifugation, CSF samples were stored at –80°C until the analysis. Then, measurements of Aβ₄₂, t-tau, and p-tau were performed in the local laboratory (LABORAF, San Raffaele Hospital, Milan, Italy) by technicians blinded to the clinical diagnosis, using a commercially available ELISA kits (Innogenetics®, Gent, Belgium), according to the manufacturer’s instructions. For each biomarker or biomarker ratio, according to the main literature on the topic, we defined the following normal values: Aβ₄₂ ≥500 ng/L [47]; t-tau ≤350 ng/L [48]; p-tau ≤60 ng/L [14]; t-tau/Aβ₄₂ ≤0.5 [49]; p-tau/Aβ₄₂ ≤0.12 [49].

Statistical analysis

Patient groups were characterized using means and standard deviations (SD). CSF biomarkers (Aβ₄₂, t-tau, and p-tau) were expressed in absolute values (in nanograms per liter). Scores at cognitive tests were reported as z-scores (z scores were calculated as the raw score of the patient, minus the mean score of Italian general population, divided by the standard deviation of Italian general population). We tested for the normality distribution of the data using the Kolmogorov-Smirnov test. Depending on the distribution of our data, we used non parametric Mann-Whitney U Tests for between groups’ comparisons and non-parametric Spearman’s ρ (rho) to evaluate correlations between groups’ numeric measures. We used chi-square test to compare categorical data. We used ROC curve analysis to evaluate diagnostic accuracy, i.e., sensitivity, specificity and AUC (Area Under the Curve) for each biomarker and neuropsychological test. Sensitivity and specificity were calculated considering the total amount of MCI who converted to AD (MCI-c) and who did not convert to AD (MCI-nc) over time until the last follow-up visit. Positive Predictive Value (PPV), Negative Predictive Value (NPV), and Odds ratio (OR) were also calculated for each parameter. A backward logistic regression was performed to find the best dichotomous predictors of conversion to AD. The dichotomizing cut-off values for the selected predictors were obtained from their individual ROC curves. All statistical analyses were performed with SPSS software v.13 (SPSS Inc., Chicago, USA). The significance level was set at p < 0.05.

RESULTS

Demographic and clinical features

At follow up, 37 out of 96 MCI subjects (38%) had converted to AD and were defined as MCI converters (MCI-c). The mean conversion time (±SD) was 21.58 (±9.95) months with an overall conversion rate from MCI to AD of 14.97% per year. 59 out of 96 MCI patients (62%) did not develop AD dementia and were defined as MCI non converters (MCI-nc). The mean follow up time in MCI-nc was 36.73 months (range: 13.23–74.70 months). During follow up, among MCI-nc subjects, 5 were diagnosed with vascular dementia, according to NINDS-AIREN criteria [50], 1 with semantic dementia, according to Gorno-Tempini criteria [51], 1 with Lewy body dementia according to McKeith criteria [52], 48 remained stable, and 4 were considered as MCI due to a depressive disorder (all of them had a completely normal brain FDG-PET, showed isolated memory deficit associated with deflection of mood or anxious symptoms evidenced at NPI scale, had no evidence of cognitive function worsening at follow up after introduction of adequate SSRI therapy) (Fig. 1). No significant differences were found between MCI-c and MCI-nc in sex, age, age at the disease onset, education, and disease duration. At baseline no differences were found in MMSE, CDR-sb scores, ADL, and IADL between MCI-c and MCI-nc (Table 1). For 18 patients (18.75 % of the total sample), we obtained follow up data by telephone interview. Among these, 14 did not convert to AD (23.72% of the total MCI-nc) while 4 converted to AD (10.8% of the total MCI-c). Finally, we combined neuropsychological tests and CSF analysis in a subsample of 43 MCI patients. Among these, we included 18 MCI-c and 25 MCI-nc. Only 3 MCI-nc had a follow up time shorter than 2 years (22.2, 22, and 17 months, respectively).

CSF analysis between groups

At baseline MCI-c patients showed significantly lower Aβ₄₂ and significantly higher t-tau and p-tau levels than MCI-nc. Moreover both t-tau/Aβ₄₂ and p-tau/Aβ₄₂ ratios were significantly higher in MCI-c than in MCI-nc. P-tau/t-tau ratio did not significantly differ between MCI-c and MCI-nc (Fig. 2 and Supplementary Table 1). We do not find any significant correlation between sex, age, MMSE, duration of illness, time of conversion to AD and Aβ₄₂, t-tau, and p-tau levels (data not shown).

Diagnostic accuracy of CSF biomarkers and international normative values

For each biomarker value and ratio, sensitivity and specificity in differentiating MCI-c from MCI-nc were calculated using ROC-curve analysis (Fig. 3). All the single CSF biomarkers offered an AUC >0.7, but only the calculated indices (i.e., t-tau/Aβ₄₂ and p-tau/Aβ₄₂ ratios) provided both sensitivity and specificity >70%, with an NPV higher than 85%, thus offering the best diagnostic reliability (Table 2). We also tested for the reliability of the CSF biomarker and ratio normative values as reported in the main literature in predicting conversion from MCI to AD in our sample. Cut-off values were overlapping with those found in the present study. Also for normative values, the best diagnostic accuracy was given by biomarker ratios (Table 2). Similarly to the biomarker ratios, a CSF profile typical for AD (AD profile) provided a very satisfying diagnostic accuracy with a sensitivity of 73% and a specificity of 85%. AD profile is characterized by low CSF Ab₄₂ plus increased CSF t-tau or p-tau levels according to the recent IWG-2 Criteria [12]. For AD profile only international normative values were considered. Sensitivity, specificity, PPV, NPV, and OR) of each parameter (using both our values and the international normative data) are reported in Table 2.

Neuropsychological assessment in MCI-c versus MCI-nc at baseline

At baseline, MCI-c performed significantly worse than MCI-nc on tests of memory domain (RAVLT-I, RAVLT-D, RAVLT-R), executive functions (TMTb, TMTa, TMTb-a), and language (SVF). No significant differences were found at baseline for all the other neuropsychological tests. (Fig. 4 and Supplementary Table 2).

Diagnostic accuracy of neuropsychological tests

Considering those neuropsychological tests that were significantly different at baseline between MCI-c and MCI-nc, we used ROC-curve analysis to find the most accurate neuropsychological tests in differentiating MCI-c from MCI-nc (Fig. 5). We considered for this analysis only 43 MCI subjects who performed all the selected tests (18 MCI-c, 25 MCI-nc). TMTb, TMT b-a, RAVLT-I, and RAVLT-D offered the best AUC (0.78, 0.74, 0.72, 0.72, respectively). TMTb offered the best sensitivity (89%) but a low specificity (64%). The best specificity was provided by RAVLT-I (80%), at the cost of a very low sensitivity (61%). Instead RAVLT-D had a more balanced diagnostic accuracy with an acceptable sensitivity (67%) and a good specificity (72%). RAVLT-R, TMT-A, and SVF showed lower AUC values (0.62, 0.70, and 0.67, respectively) with a sensitivity of 61%, 72% and 66%, respectively and a specificity of 72%, 64%, and 76%, respectively (Table 3). Then we selected the most accurate neuropsychological test for each major cognitive domain where MCI-c performed worse than MCI-nc at baseline (i.e., memory, executive functions, and language). Therefore we considered RAVLT-D for memory domain, TMTb for executive functions and SVF for language domain. Hence we calculated a Composite Cognitive test Score (CCS) summing up the three single z-scores each patient obtained at RAVLT-D, TMTb, and SVF tests. To calculate CCS, the additive inverse of TMTb z-score was entered into the equation. CCS showed a good diagnostic accuracy with a cut-off of –1.99 Z-score (AUC: 0.83, sensitivity: 77.8 specificity: 84.8;Table 3).

Combination of CSF biomarkers and neuropsychological tests

CSF biomarkers, biomarker ratios, AD profile, cognitive test scores, and CCS were entered into a backward logistic regression model with the outcome “conversion/non conversion to AD” as dependent variable. We considered for this analysis only 43 MCI subjects for whom all the predictors were available (18 MCI-c, 25 MCI-nc). All the predictors were dichotomized using their cut-off values, except for AD profile, which is already dichotomous. The best model including CSF AD profile, t-tau/Aβ42 and CCS was statistically significant (χ² = 19.491, p < 0.0005) and correctly classified 86.00% of cases with a sensitivity of 72.20%, a specificity of 96.00%, a PPV of 92.86%, an NPV of 82.76%, and an OR of 62.4 (Table 4).

Using the regression coefficients associated with the three predictors in the logistic model, we estimated the risk of AD conversion for each predictor combination. The following equation describes the regression model: $logit (p) = β_{0} + β_{1} x_{1} + β_{2} x_{2} + β_{3} x_{3} + . . . + β_{k} x_{k}$

$\log it (p) = ln (\frac{p}{1 - p})$ is the logit function where p represents the probability that the event (i.e., “conversion to AD”) might happen, and β is the corresponding regression coefficient associated to each predictor x. Entering the constant and the coefficients found in our logistic model, we obtained the following equation which enabled us to estimate the probability that the event “conversion” might happen. $\begin{matrix} ln (\frac{p}{1 - p}) = - 2.62 + 1.598 \times tau / A β + 0.729 \\ \times AD + 2.105 \times CCS \end{matrix}$

In the predictor algorithm, the probability of conversion was very low if no predictor was abnormal (6.79%), low to medium if only one predictor was abnormal (13% to 37%), medium to high if two predictors were abnormal (43% to 75%), very high if all the predictors were abnormal (86%) (Table 5).

DISCUSSION

We tried to define a simple and cost-effective diagnostic algorithm to reliably detect MCI patients who will convert to AD. To our knowledge this is one of the first studies to have combined easily available diagnostic tools, even in not-hyper specialized clinics (i.e., CSF analysis and neuropsychological assessment) in the early diagnosis of AD. Among 96 MCI subjects, 37 converted to AD (MCI-c) with an overall conversion rate of 14,97% per person-year, almost similar to the data reported in the literature [53, 54]. CSF biomarkers and neuropsychological tests showed good reliability at baseline in predicting progression from MCI to AD both when considered as single predictors and when combined together. The most striking finding about CSF analyses is probably the high diagnostic accuracy that the international normative values provided when applied in our setting (sensitivity higher than 80%; specificity about 80%, especially for biomarker ratios). Interestingly they were almost completely overlapping to our cut off values. Up to now, one of the most critical issues about the usefulness of CSF analysis and its worldwide application has been represented by high inter-laboratory variability [55 –57]. As a consequence, some authors [58, 59] have recently stated that each laboratory should use internally-validated CSF biomarker cut-off values. In disagreement with this statement, we believe that this issue could be easily overcome with the enhancement of the laboratory techniques and the adoption of well-standardized measurement protocols and kits, thus making the CSF analysis a worldwide pivotal diagnostic tool in the early diagnosis of AD. In line with other international studies [60 –62], the combination of the single biomarkers, in particular p-tau/Aβ₄₂ and t-tau/Aβ₄₂ ratios offered the best diagnostic accuracy in predicting conversion to AD.

Both ratios provided a good sensitivity (about or higher than 80%) and an optimal NPV (equal or higher than 85%), but a suboptimal specificity (less than 80%) and an unsatisfactory PPV (about or less than 70%). The suboptimal specificity and PPV shown both by our data and in other studies [63, 64] could be due to a follow up time not long enough to detect the majority of MCI patients who will convert to AD, since at least five years are required before the most MCI patients with AD pathology convert to a full blown AD dementia [53]. Therefore, a shorter follow-up period could determine an underestimation of the real amount of MCI patients who will develop AD, thus lowering both specificity and positive predictive values. When longer observation times are established, biomarker diagnostic accuracy significantly improves [14, 65]. Moreover, neuropsychological tests provided a good diagnostic accuracy in the early diagnosis of AD. Several prospective studies suggested cognitive deficits are detectable up to 12 years before the clinical diagnosis of AD dementia [66 –73]. In line with the international literature on the topic, in our sample, MCI-c subjects performed worse than MCI-nc subjects on tests about memory (RAVLT) [74 –76], executive functions (TMTa, TMTb, TMTb-a) [76] and language domain (SVF) [74, 76], revealing already at baseline a wider, albeit slight, multi-domain cognitive involvement.

As single predictors, the best accuracy in predicting conversion to AD is provided by long delayed cued recall tests [77 –79], delayed memory tests [80], and executive function tests [81, 82], even if evidences are still controversial, since other studies evidenced that neuropsychological tests did not adequately perform in the prediction of AD conversion [81, 83]. In line with these findings, we found that the single most accurate neuropsychological tests are a long delayed recall test (RAVLT-D) and an executive function test (TMTb). The neuropsychological tests’ cut-off values we found in the present study are different from normative values generally adopted in the clinical practice (i.e., 2 standard deviations under the general population mean). These normative values in fact are suitable for the diagnosis of a full-blown dementia but not for detecting those MCI patients who will convert to AD, since they are not yet demented subjects, by definition. In accordance with these observations, Grundman [74] stated that MCI subjects obtain scores on neuropsychological tests ranging from –2 and –1 SD below the general population mean, without identifying clear-cut normative values. This lack could have determined the very high heterogeneity several studies adopting different cut-off values showed in estimating the real neuropsychological test reliability in predicting conversion from MCI to AD [84 –87]. For this reason, in agreement with other authors [88, 89], we think that MCI specific neuropsychological cut-off values should be established to distinguish MCI converters from MCI non converters, rather than healthy from demented subjects. To ameliorate the single tests’ accuracy, we calculated a composite cognitive score (CCS) summing up the z-scores of the three most accurate neuropsychological tests, one per each cognitive domain significantly involved in MCI-c, i.e., memory (RAVLT-D), executive functions (TMT-b), and language (SVF). CCS strongly improved the single tests’ accuracy in differentiating MCI-c from MCI-nc.Similarly, Rajan and colleagues [90] built up a composite cognitive test score combining four brief cognitive tests including two tests of episodic memory (Immediate and Delayed Recall of the East Boston Story), a test of executive function (Symbol Digits Modalities Test), and a test of general orientation and global cognition (MMSE). In a prospective study conducted on a sample of 2,125 participants without AD dementia, followed up over intervals ranging from 1 to 18 years, they found that lower scores at this composite cognitive test at baseline were associated with an improved risk to develop AD dementia over time. Another two studies [91, 92] used a similar composite cognitive score to track preclinical AD decline better than the single most sensitive tests. Even if a similar approach may represent a statistical artifact, thus raising some criticism and needing further evidence to be validated, nevertheless we think that it could be reasonable from a clinical prospective. Our and other groups’ data, in fact, seem to confirm that a simple and time-sparing composite cognitive score could provide a more tailored and complete neuropsychological characterization than other generic tools (i.e., MMSE) in addition to a higher diagnostic accuracy. Hence we think that future clinical trials could adopt such a composite score either as screening and/or as outcome measurement. Finally, we succeeded in creating a simple and widely usable diagnostic algorithm combining both CSF biomarkers and neuropsychological assessment. Entering all the CSF biomarkers and neuropsychological tests into a regression model, only three predictors, i.e., CCS, AD profile, and t-tau/Aβ₄₂ ratio, are sufficient to provide the best diagnostic accuracy, correctly classifying 86% of cases and improving the single predictors’ accuracy, in particular specificity (96%) and PPV (92%). Moreover this simple multi-marker based model could enable clinicians to stratify the risk of AD conversion. Based on the number of abnormal predictors (from 0 to 3), we identified four different risk classes: in the first (no abnormal predictor), the probability of AD conversion was very low (6.79%); in the second (only one abnormal predictor), the risk was low to medium (13% to 37%); in the third (two abnormal predictors), it was medium to high (43% to 75%); finally, in the fourth (all the predictors abnormal), it was very high (86%) (Table 5). Interestingly, in the first class, the risk was low but not null, probably because MCI subjects are not healthy subjects, since they have an objective, albeit slight, cognitive deficit, anyhow being at very low risk of developing other neurodegenerative disorders in the future that can be misdiagnosed as AD. In the present algorithm, the single predictor with the strongest impact on the AD conversion risk is CCS. Even if this could represent a sort of “selection bias” since the MCI diagnosis is mainly based upon neuropsychological criteria; nevertheless we think that this finding supports such a composite cognitive score as a possible useful and inexpensive screening tool for future clinicaltrials.

Before us, Palmqvist [59] found that CSF and cognitive tests (MMSE and the clock drawing test) when combined together in MCI patients provided an overall accuracy quite identical to the ours, correctly classifying 85% of the cases who developed AD. The overall accuracy is further increased when other diagnostic tools (i.e., brain MRI or FDG-PET) are considered [28]. However Ewers [31] found out that a combination of TMTb, right hippocampal volume, CSF p-tau/Aβ₄₂ ratio, and age increased the accuracy of each single predictor, without joining the statistical significance. Hence the author argued that some sparser and economic single-predictor models may be as good as any more complex model for the prediction of progression from MCI to AD. Partially disagreeing with Ewers, we think that a multi-marker based model is preferable to a single-predictor model, because, in addition to a higher accuracy, it may offer a stratification of the risk of conversion to AD, thus helping the clinicians to better interpret possible contrasting results deriving from the single diagnostic tests. In this regard, Devanand et al. [32], in a 3-year follow-up study on 148 MCI patients, developed a 5 predictor-based model, including Functional Activities Questionnaire, University of Pennsylvania Smell Identification Test, Selective Reminding Test, MRI hippocampal volume, and MRI entorhinal cortex volume, but not CSF biomarker analysis. This model provided a risk stratification ranging from 0% if no predictor was pathological to 100% if all the predictors were abnormal, with a specificity of 90% and a sensitivity of 85.2%, confirming that a multi-marker based model could be useful both as a diagnostic and a prognostic tool. The present study has some limitations: first of all the lack of other fundamental diagnostic techniques, such as brain MRI or FDG-PET, has prevented us from elaborating a complete and really powerful diagnostic algorithm, but the aim of the present study was to find a reliable, easy, and largely applicable diagnostic tool which could help clinicians to correctly diagnose those MCI patients who will develop AD, even in not hyper-specialized clinics. In particular, we think that CCS, a very cheap and time-sparing tool could be proposed as a possible screening test for future clinical trials, being confirmed at a second level analysis by CSF biomarkers. In particular, the high specificity (96%) and PPV (93%) of our model would allow clinicians to detect an underlying AD pathology with very high confidence. This could be of great importance for future clinical trials, since correctly diagnosing AD in the early phases would avoid enrolling “false AD” patients, one of the principal reasons accounting for the failure of all the previous experimental clinical trials on AD. On the other hand, our model could be useful in excluding AD pathology, too. In fact, if all the three predictors of our model are negative, the risk of progression to AD is about 6%. However it should be noticed that excluding progression to AD, does not exclude progression to other types of dementia. In this context a more complex diagnostic approach, including brain MRI, FDG-PET, and a more detailed neuropsychological assessment, is mandatory to fully exclude a neurodegenerative process accounting for the cognitive impairment patients refer. Other limitations are represented by the small sample size and the sub-optimal follow up time length that could have determined a loss of statistical power. Moreover, taking into account the small sample size, multiple analyses were performed and the standard p-value of 0.05 might have led to some false positive findings. Another limit is represented by the inclusion of subjects for whom only follow up information by telephone interview were available. Finally, our findings and cut-off values need validation in independent samples of MCI subjects. For this reason, further studies with larger sample size, longer follow-up time, and other diagnostic techniques, such as functional and morphological neuroimaging, are necessary to validate the present findings and eventually elaborate a very strong diagnostic and prognostic algorithm in the early diagnosis of AD dementia.

Conclusions

Despite some limitations we already described, we think that our study provided an interesting, simple, and reliable diagnostic tool, which obviously needs further validation, but which could be very useful both in clinical practice and in future clinical trials, allowing clinicians to detect those MCI patients who will develop AD dementia with high confidence. The relative low cost and wide availability of CSF biomarkers and neuropsychological assessment could make them essential tools to face the great and growing challenge represented by AD dementia.

DISCLOSURE STATEMENT

Authors’ disclosures available online (http://j-alz.com/manuscript-disclosures/160360).

Footnotes

Appendix

References

Braak

, Del

Tredici K

(2012) Where, when, and in what form does sporadic Alzheimer’s disease begin? Curr Opin Neurol 25, 708–714.

Wimo

, Jönsson

, Bond

, Prince

, Winblad

, Alzheimer

Disease International

(2013) The worldwide economic impact of dementia 2010. Alzheimers Dement 9, 1–11.e3.

Ferri

, Prince

, Brayne

, Brodaty

, Fratiglioni

, Ganguli

, Hall

, Hasegawa

, Hendrie

, Huang

, Jorm

, Mathers

, Menezes

, Rimmer

, Scazufca

, Alzheimer’s Disease, International (2005) Global prevalence of dementia: A Delphi consensus study. Lancet 366, 2112–2117.

Birks

(2006) Cholinesterase inhibitors for Alzheimer’s disease. Cochrane Database Syst. Rev CD005593.

McShane

, Areosa Sastre

, Minakaran

(2006) Memantine for dementia. Cochrane Database Syst. Rev CD003154.

Rafii

, Aisen

(2015) Advances in Alzheimer’s disease drug development. BMC Med 13, 62.

Dubois

, Feldman

, Jacova

, Dekosky

, Barberger-Gateau

, Cummings

, Delacourte

, Galasko

, Gauthier

, Jicha

, Meguro

, O’brien

, Pasquier

, Robert

, Rossor

, Salloway

, Stern

, Visser

, Scheltens

(2007) Research criteria for the diagnosis of Alzheimer’s disease: Revising the NINCDS-ADRDA criteria. Lancet Neurol 6, 734–746.

Ashford

, Borson

, O’Hara

, Dash

, Frank

, Robert

, Shankle

, Tierney

, Brodaty

, Schmitt

, Kraemer

, Buschke

, Fillit

(2007) Should older adults be screened for dementia? It is important to screen for evidence of dementia!. Alzheimers Dement 3, 75–80.

Petersen

(2004) Mild cognitive impairment as a diagnostic entity. J Intern Med 256, 183–194.

10.

Albert

, DeKosky

, Dickson

, Dubois

, Feldman

, Fox

, Gamst

, Holtzman

, Jagust

, Petersen

, Snyder

, Carrillo

, Thies

, Phelps

(2011) The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 270–279.

11.

Dubois

, Feldman

, Jacova

, Cummings

, Dekosky

, Barberger-Gateau

, Delacourte

, Frisoni

, Fox

, Galasko

, Gauthier

, Hampel

, Jicha

, Meguro

, O’Brien

, Pasquier

, Robert

, Rossor

, Salloway

, Sarazin

, de Souza

, Stern

, Visser

, Scheltens

(2010) Revising the definition of Alzheimer’s disease: A new lexicon. Lancet Neurol 9, 1118–1127.

12.

Dubois

, Feldman

, Jacova

, Hampel

, Molinuevo

, Blennow

, DeKosky

, Gauthier

, Selkoe

, Bateman

, Cappa

, Crutch

, Engelborghs

, Frisoni

, Fox

, Galasko

, Habert

M-O

, Jicha

, Nordberg

, Pasquier

, Rabinovici

, Robert

, Rowe

, Salloway

, Sarazin

, Epelbaum

, de Souza

, Vellas

, Visser

, Schneider

, Stern

, Scheltens

, Cummings

(2014) Advancing research diagnostic criteria for Alzheimer’s disease: The IWG-2 criteria. Lancet Neurol 13, 614–629.

13.

Mattsson

, Zetterberg

, Hansson

, Andreasen

, Parnetti

, Jonsson

, Herukka

S-K

, van der Flier

, Blankenstein

, Ewers

, Rich

, Kaiser

, Verbeek

, Tsolaki

, Mulugeta

, Rosén

, Aarsland

, Visser

, Schröder

, Marcusson

, de Leon

, Hampel

, Scheltens

, Pirttilä

, Wallin

, Jönhagen

, Minthon

, Winblad

, Blennow

(2009) CSF biomarkers and incipient Alzheimer disease in patients with mild cognitive impairment. JAMA 302, 385–393.

14.

Hansson

, Zetterberg

, Buchhave

, Londos

, Blennow

, Minthon

(2006) Association between CSF biomarkers and incipient Alzheimer’s disease in patients with mild cognitive impairment: A follow-up study. Lancet Neurol 5, 228–234.

15.

Blennow

(2004) Cerebrospinal fluid protein biomarkers for Alzheimer’s disease. NeuroRx 1, 213–225.

16.

Hampel

, Bürger

, Teipel

, Bokde

ALW

, Zetterberg

, Blennow

(2008) Core candidate neurochemical and imaging biomarkers of Alzheimer’s disease. Alzheimers Dement 4, 38–48.

17.

Coppi

, Ferrari

, Santangelo

, Caso

, Pinto

, Passerini

, Comi

, Magnani

(2014) Further evidence about the crucial role of CSF biomarkers in diagnosis of posterior cortical atrophy. Neurol Sci 35, 785–787.

18.

Santangelo

, Coppi

, Ferrari

, Bernasconi

, Pinto

, Passerini

, Comi

, Magnani

(2015) Cerebrospinal fluid biomarkers can play a pivotal role in the diagnostic work up of primary progressive aphasia. J Alzheimers Dis 43, 1429–1440.

19.

Diniz

, Pinto Júnior

, Forlenza

(2007) Do CSF total tau, phosphorylated tau, and beta-amyloid 42 help to predict progression of mild cognitive impairment to Alzheimer’s disease? A systematic review and meta-analysis of the literature. World J Biol Psychiatry 9, 172–182.

20.

, Toledo

, Zee

, Wolk

, Xie

, Ou

, Shacklett

, Parmpi

, Shaw

, Trojanowski

, Davatzikos

(2014) Integration and relative value of biomarkers for prediction of MCI to AD progression: Spatial patterns of brain atrophy, cognitive scores, APOE genotype and CSF biomarkers. Neuroimage Clin 4, 164–173.

21.

Andreasen

, Vanmechelen

, Vanderstichele

, Davidsson

, Blennow

(2003) Cerebrospinal fluid levels of total-tau, phospho-tau and A beta 42 predicts development of Alzheimer’s disease in patients with mild cognitive impairment. Acta Neurol Scand 107, 47–51.

22.

Schmand

, Huizenga

, van Gool

(2010) Meta-analysis of CSF and MRI biomarkers for detecting preclinical Alzheimer’s disease. Psychol Med 40, 135–145.

23.

Arnáiz

, Almkvist

(2003) Neuropsychological features of mild cognitive impairment and preclinical Alzheimer’s disease. Acta Neurol Scand Suppl 179, 34–41.

24.

Newman

, Warrington

, Kennedy

, Rossor

(1994) The earliest cognitive change in a person with familial Alzheimer’s disease: Presymptomatic neuropsychological features in a pedigree with familial Alzheimer’s disease confirmed at necropsy. J Neurol Neurosurg Psychiatry 57, 967–972.

25.

Jacobs

, Sano

, Dooneief

, Marder

, Bell

, Stern

(1995) Neuropsychological detection and characterization of preclinical Alzheimer’s disease. Neurology 45, 957–962.

26.

Masur

, Sliwinski

, Lipton

, Blau

, Crystal

(1994) Neuropsychological prediction of dementia and the absence of dementia in healthy elderly persons. Neurology 44, 1427–1432.

27.

Small

, Herlitz

, Fratiglioni

, Almkvist

, Bäckman

(1997) Cognitive predictors of incident Alzheimer’s disease: A prospective longitudinal study. Neuropsychology 11, 413–420.

28.

Eckerström

, Olsson

, Bjerke

, Malmgren

, Edman

, Wallin

, Nordlund

(2013) A combination of neuropsychological, neuroimaging, and cerebrospinal fluid markers predicts conversion from mild cognitive impairment to dementia. J Alzheimers Dis 36, 421–431.

29.

Didic

, Felician

, Barbeau

, Mancini

, Latger-Florence

, Tramoni

, Ceccaldi

(2013) Impaired visual recognition memory predicts Alzheimer’s disease in amnestic mild cognitive impairment. Dement Geriatr Cogn Disord 35, 291–299.

30.

Griffith

, Netson

, Harrell

, Zamrini

, Brockington

, Marson

(2006) Amnestic mild cognitive impairment: Diagnostic outcomes and clinical prediction over a two-year time period. J Int Neuropsychol Soc 12, 166–175.

31.

Ewers

, Walsh

, Trojanowski

, Shaw

, Petersen

, Jack

, Feldman

, Bokde

ALW

, Alexander

, Scheltens

, Vellas

, Dubois

, Weiner

, Hampel

, North American Alzheimer’s Disease Neuroimaging Initiative (ADNI) (2012) Prediction of conversion from mild cognitive impairment to Alzheimer’s disease dementia based upon biomarkers and neuropsychological test performance. Neurobiol Aging 33, 1203–1214.

32.

Davangere

, Devanand

(2008) Combining early markers strongly predicts conversion from mild cognitive impairment to Alzheimer’s disease. Biol Psychiatry 64, 871–879.

33.

Cui

, Liu

, Luo

, Zhen

, Fan

, Liu

, Zhu

, Park

, Jiang

, Jin

, Alzheimer’s Disease Neuroimaging, Initiative (2011) Identification of conversion from mild cognitive impairment to Alzheimer’s disease using multivariate predictors. PloS One 6, e21896.

34.

Katz

, Ford

, Moskowitz

, Jackson

, Jaffe

(1963) Studies of illness in the aged. The index of ADL: A standardized measure of biological and psychosocial function. JAMA 185, 914–919.

35.

Lawton

, Brody

(1969) Assessment of older people: Self-maintaining and instrumental activities of daily living. Gerontologist 9, 179–186.

36.

Folstein

, Folstein

, McHugh

(1975) “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

37.

Magni

, Binetti

, Bianchetti

, Rozzini

, Trabucchi

(1996) Mini-Mental State Examination: A normative study in Italian elderly population. Eur J Neurol 3, 198–202.

38.

Hughes

, Berg

, Danziger

, Coben

, Martin

(1982) A new clinical scale for the staging of dementia. Br. J Psychiatry J Ment Sci 140, 566–572.

39.

Cummings

, Mega

, Gray

, Rosenberg-Thompson

, Carusi

, Gornbein

(1994) The Neuropsychiatric Inventory: Comprehensive assessment of psychopathology in dementia. Neurology 44, 2308–2314.

40.

Winblad

, Palmer

, Kivipelto

, Jelic

, Fratiglioni

, Wahlund

L-O

, Nordberg

, Bäckman

, Albert

, Almkvist

, Arai

, Basun

, Blennow

, De Leon

, DeCarli

, Erkinjuntti

, Giacobini

, Graff

, Hardy

, Jack

, Jorm

, Ritchie

, Van Duijn

, Visser

, Petersen

R c

(2004) Mild cognitive impairment –beyond controversies, towards a consensus: Report of the International Working Group on Mild Cognitive Impairment. J Intern Med 256, 240–246.

41.

McKhann

, Knopman

, Chertkow

, Hyman

, Jack

, Kawas

, Klunk

, Koroshetz

, Manly

, Mayeux

, Mohs

, Morris

, Rossor

, Scheltens

, Carrillo

, Thies

, Weintraub

, Phelps

(2011) The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 263–269.

42.

Mauri

, Carlesimo

, Graceffa

AMS

, Loasses

, Lorusso

, Sinforiani

, Bono

, Caltagirone

(1997) Standardizzazione di due nuovi test di memoria: Apprendimento di liste di parole correlate e non correlate semanticamente. Arch Psicol Neurol Psichiatr 58, 621–645.

43.

Novelli

, Papagno

, Capitani

, Laiacona

, Cappa

, Vallar

(1986) Tre test clinici di memoria verbale a lungo termine: Taratura su soggetti normali. [Three clinical tests for the assessment of verbal long-term memory function: Norms from 320 normal subjects.]. Arch Psicol Neurol Psichiatr 47, 278–296.

44.

Spinnler

, Tognoni

(1987) Standardizzazione e taratura italiana di test neuropsicologici: Gruppo italiano per lo studio neuropsicologico dell’invecchiamento,Masson Italia Periodici, Milano.

45.

Giovagnoli

, Del Pesce

, Mascheroni

, Simoncelli

, Laiacona

, Capitani

(1996) Trail making test: Normative values from 287 normal adult controls. Ital J Neurol Sci 17, 305–309.

46.

Basso

, Capitani

, Laiacona

(1987) Raven’s coloured progressive matrices: Normative values on 305 adult normal controls. Funct Neurol 2, 189–194.

47.

Sjögren

, Vanderstichele

, Agren

, Zachrisson

, Edsbagge

, Wikkelsø

, Skoog

, Wallin

, Wahlund

, Marcusson

, Nägga

, Andreasen

, Davidsson

, Vanmechelen

, Blennow

(2001) Tau and Abeta42 in cerebrospinal fluid from healthy adults 21-93 years of age: Establishment of reference values. Clin Chem 47, 1776–1781.

48.

Tapiola

, Alafuzoff

, Herukka

S-K

, Parkkinen

, Hartikainen

, Soininen

, Pirttilä

(2009) Cerebrospinal fluid beta-amyloid 42 and tau proteins as biomarkers of Alzheimer-type pathologic changes in the brain. Arch Neurol 66, 382–389.

49.

de Jong

, Jansen

RWMM

, Kremer

BPH

, Verbeek

(2006) Cerebrospinal fluid amyloid beta42/phosphorylated tau ratio discriminates between Alzheimer’s disease and vascular dementia. J Gerontol A Biol Sci Med Sci 61, 755–758.

50.

Román

, Tatemichi

, Erkinjuntti

, Cummings

, Masdeu

, Garcia

, Amaducci

, Orgogozo

, Brun

, Hofman

(1993) Vascular dementia: Diagnostic criteria for research studies. Report of the NINDS-AIREN International Workshop. Neurology 43, 250–260.

51.

Gorno-Tempini

, Hillis

, Weintraub

, Kertesz

, Mendez

, Cappa

, Ogar

, Rohrer

, Black

, Boeve

, Manes

, Dronkers

, Vandenberghe

, Rascovsky

, Patterson

, Miller

, Knopman

, Hodges

, Mesulam

, Grossman

(2011) Classification of primary progressive aphasia and its variants. Neurology 76, 1006–1014.

52.

McKeith

(2004) Dementia with Lewy bodies. Dialogues Clin Neurosci 6, 333–341.

53.

Petersen

, Smith

, Waring

, Ivnik

, Tangalos

, Kokmen

(1999) Mild cognitive impairment: Clinical characterization and outcome. Arch Neurol 56, 303–308.

54.

Ward

, Tardiff

, Dye

, Arrighi

(2013) Rate of conversion from prodromal Alzheimer’s disease to Alzheimer’s dementia: A systematic review of the literature. Dement Geriatr Cogn Disord EXTRA 3, 320–332.

55.

Hort

, Bartos

, Pirttilä

, Scheltens

(2010) Use of cerebrospinal fluid biomarkers in diagnosis of dementia across Europe. Eur J Neurol 17, 90–96.

56.

Verwey

, van der Flier

, Blennow

, Clark

, Sokolow

, De Deyn

, Galasko

, Hampel

, Hartmann

, Kapaki

, Lannfelt

, Mehta

, Parnetti

, Petzold

, Pirttila

, Saleh

, Skinningsrud

, Swieten

, Verbeek

, Wiltfang

, Younkin

, Scheltens

, Blankenstein

(2009) A worldwide multicentre comparison of assays for cerebrospinal fluid biomarkers in Alzheimer’s disease. Ann Clin Biochem 46, 235–240.

57.

Fourier

, Portelius

, Zetterberg

, Blennow

, Quadrio

, Perret-Liaudet

(2015) Pre-analytical and analytical factors influencing Alzheimer’s disease cerebrospinal fluid biomarker variability. Clin Chim Acta 449, 9–15.

58.

Ferreira

, Perestelo-Pérez

, Westman

, Wahlund

L-O

, Sarría

, Serrano-Aguilar

(2014) Meta-review of CSF core biomarkers in Alzheimer’s disease: The state-of-the-art after the new revised diagnostic criteria. Front Aging Neurosci 6, 47.

59.

Palmqvist

, Hertze

, Minthon

, Wattmo

, Zetterberg

, Blennow

, Londos

, Hansson

(2012) Comparison of brief cognitive tests and CSF biomarkers in predicting Alzheimer’s disease in mild cognitive impairment: six-year follow-up study. PLoS One 7, e38639.

60.

Mattsson

, Blennow

, Zetterberg

(2009) CSF biomarkers: Pinpointing Alzheimer pathogenesis. Ann N Y Acad Sci 1180, 28–35.

61.

Buchhave

, Minthon

, Zetterberg

, Wallin

, Blennow

, Hansson

(2012) Cerebrospinal fluid levels of β-amyloid 1-42, but not of tau, are fully changed already 5 to 10 years before the onset of Alzheimer dementia. Arch Gen Psychiatry 69, 98–106.

62.

van Rossum

, Vos

, Handels

, Visser

(2010) Biomarkers as predictors for conversion from mild cognitive impairment to Alzheimer-type dementia: Implications for trial design. J Alzheimers Dis 20, 881–891.

63.

Parnetti

, Chiasserini

, Eusebi

, Giannandrea

, Bellomo

, De Carlo

, Padiglioni

, Mastrocola

, Lisetti

, Calabresi

(2012) Performance of Aβ1-40, Aβ1-42, total tau, and phosphorylated tau as predictors of dementia in a cohort of patients with mild cognitive impairment. J Alzheimers Dis 29, 229–238.

64.

Gaser

, Franke

, Klöppel

, Koutsouleris

, Sauer

, Alzheimer’s Disease Neuroimaging, Initiative (2013) BrainAGE in mild cognitive impaired patients: Predicting the conversion to Alzheimer’s disease. PloS One 8, e67346.

65.

Hertze

, Minthon

, Zetterberg

, Vanmechelen

, Blennow

, Hansson

(2010) Evaluation of CSF biomarkers as predictors of Alzheimer’s disease: A clinical follow-up study of 4.7 years. J Alzheimers Dis 21, 1119–1128.

66.

Amieva

, Jacqmin-Gadda

, Orgogozo

J-M

, Le Carret

, Helmer

, Letenneur

, Barberger-Gateau

, Fabrigoule

, Dartigues

J-F

(2005) The 9 year cognitive decline before dementia of the Alzheimer type: A prospective population-based study. Brain 128, 1093–1101.

67.

Johnson

, Storandt

, Morris

, Galvin

(2009) Longitudinal study of the transition from healthy aging to Alzheimer disease. Arch Neurol 66, 1254–1259.

68.

Small

, Bäckman

(2007) Longitudinal trajectories of cognitive change in preclinical Alzheimer’s disease: A growth mixture modeling analysis. Cortex 43, 826–834.

69.

Grober

, Hall

, Lipton

, Zonderman

, Resnick

, Kawas

(2008) Memory impairment, executive dysfunction, and intellectual decline in preclinical Alzheimer’s disease. J Int Neuropsychol Soc 14, 266–278.

70.

Howieson

, Carlson

, Moore

, Wasserman

, Abendroth

, Payne-Murphy

, Kaye

(2008) Trajectory of mild cognitive impairment onset. J Int Neuropsychol Soc 14, 192–198.

71.

Wilson

, Leurgans

, Boyle

, Bennett

(2011) Cognitive decline in prodromal Alzheimer disease and mild cognitive impairment. Arch Neurol 68, 351–356.

72.

Amieva

, Le

Goff M

, Millet

, Orgogozo

, Pérès

, Barberger-Gateau

, Jacqmin-Gadda

, Dartigues

(2008) Prodromal Alzheimer’s disease: Successive emergence of the clinical symptoms. Ann Neurol 64, 492–498.

73.

Elias

, Beiser

, Wolf

, Au

, White

, D’Agostino

(2000) The preclinical phase of alzheimer disease: A 22-year prospective study of the Framingham Cohort. Arch Neurol 57, 808–813.

74.

Grundman

, Petersen

, Ferris

, Thomas

, Aisen

, Bennett

, Foster

, Jack

, Galasko

, Doody

, Kaye

, Sano

, Mohs

, Gauthier

, Kim

, Jin

, Schultz

, Schafer

, Mulnard

, van Dyck

, Mintzer

, Zamrini

, Cahn-Weiner

, Thal

, Alzheimer’s Disease Cooperative, Study (2004) Mild cognitive impairment can be distinguished from Alzheimer disease and normal aging for clinical trials. Arch Neurol 61, 59–66.

75.

Mitchell

, Arnold

, Dawson

, Nestor

, Hodges

(2009) Outcome in subgroups of mild cognitive impairment (MCI) is highly predictable using a simple algorithm. J Neurol 256, 1500–1509.

76.

Kandel

, Avants

, Gee

, Arnold

, Wolk

(2015) Neuropsychological testing predicts cerebrospinal fluid amyloid-β in mild cognitive impairment. J Alzheimers Dis 46, 901–912.

77.

Dierckx

, Engelborghs

, De Raedt

, Van Buggenhout

, De Deyn

, Verté

, Ponjaert-Kristoffersen

(2009) Verbal cued recall as a predictor of conversion to Alzheimer’s disease in mild cognitive impairment. Int J Geriatr Psychiatry 24, 1094–1100.

78.

Sarazin

, Berr

, De Rotrou

, Fabrigoule

, Pasquier

, Legrain

, Michel

, Puel

, Volteau

, Touchon

, Verny

, Dubois

(2007) Amnestic syndrome of the medial temporal type identifies prodromal AD: A longitudinal study. Neurology 69, 1859–1867.

79.

Wagner

, Wolf

, Reischies

, Daerr

, Wolfsgruber

, Jessen

, Popp

, Maier

, Hüll

, Frölich

, Hampel

, Perneczky

, Peters

, Jahn

, Luckhaus

, Gertz

H-J

, Schröder

, Pantel

, Lewczuk

, Kornhuber

, Wiltfang

(2012) Biomarker validation of a cued recall memory deficit in prodromal Alzheimer disease. Neurology 78, 379–386.

80.

Klages

, Fisk

, Rockwood

(2005) APOE genotype, vascular risk factors, memory test performance and the five-year risk of vascular cognitive impairment or Alzheimer’s disease. Dement Geriatr Cogn Disord 20, 292–297.

81.

Aretouli

, Tsilidis

, Brandt

(2013) Four-year outcome of mild cognitive impairment: The contribution of executive dysfunction. Neuropsychology 27, 95–106.

82.

Brandt

, Aretouli

, Neijstrom

, Samek

, Manning

, Albert

, Bandeen-Roche

(2009) Selectivity of executive function deficits in mild cognitive impairment. Neuropsychology 23, 607–618.

83.

Belleville

, Gauthier

, Lepage

, Kergoat

M-J

, Gilbert

(2014) Predicting decline in mild cognitive impairment: A prospective cognitive study. Neuropsychology 28, 643–652.

84.

Bickel

, Mosch

, Seigerschmidt

, Siemen

, Forstl

(2006) Prevalence and persistence of mild cognitive impairment among elderly patients in general hospitals. Dement Geriatr Cogn Disord 21, 242–250.

85.

Bondi

, Jak

, Delano-Wood

, Jacobson

, Delis

, Salmon

(2008) Neuropsychological contributions to the early identification of Alzheimer’s disease. Neuropsychol Rev 18, 73–90.

86.

Busse

, Hensel

, Gühne

, Angermeyer

, Riedel-Heller

(2006) Mild cognitive impairment: Long-term course of four clinical subtypes. Neurology 67, 2176–2185.

87.

Zanetti

, Ballabio

, Abbate

, Cutaia

, Vergani

, Bergamaschini

(2006) Mild cognitive impairment subtypes and vascular dementia in community-dwelling elderly people: A 3-year follow-up study. J Am Geriatr Soc 54, 580–586.

88.

Christa Maree Stephan

, Minett

, Pagett

, Siervo

, Brayne

, McKeith

(2013) Diagnosing mild cognitive impairment (MCI) in clinical trials: A systematic review. BMJ Open 3, pii:e001909.

89.

Jak

, Bondi

, Delano-Wood

, Wierenga

, Corey-Bloom

, Salmon

, Delis

(2009) Quantification of five neuropsychological approaches to defining mild cognitive impairment. Am J Geriatr Psychiatry 17, 368–375.

90.

Rajan

, Wilson

, Weuve

, Barnes

, Evans

(2015) Cognitive impairment 18 years before clinical diagnosis of Alzheimer disease dementia. Neurology 85, 898–904.

91.

Ayutyanont

, Langbaum

, Hendrix

, Chen

, Fleisher

, Friesenhahn

, Ward

, Aguirre

, Acosta-Baena

, Madrigal

, Muñoz

, Tirado

, Moreno

, Tariot

, Lopera

, Reiman

(2014) The Alzheimer’s Prevention Initiative composite cognitive test score: Sample size estimates for the evaluation of preclinical Alzheimer’s disease treatments in presenilin 1 E280A mutation carriers. J Clin Psychiatry 75, 652–660.

92.

Langbaum

, Hendrix

, Ayutyanont

, Chen

, Fleisher

, Shah

, Barnes

, Bennett

, Tariot

, Reiman

(2014) An empirically derived composite cognitive test score with improved power to track and evaluate treatments for preclinical Alzheimer’s disease. Alzheimers Dement 10, 666–674.