Diagnostic and Prognostic Value of the Combination of Two Measures of Verbal Memory in Mild Cognitive Impairment due to Alzheimer’s Disease

Abstract

Background: Episodic memory impairment is the core feature of typical Alzheimer’s disease.

Objective: To evaluate the performance of two commonly used verbal memory tests to detect mild cognitive impairment due to Alzheimer’s disease (MCI-AD) and to predict progression to Alzheimer’s disease dementia (AD-d).

Methods: Prospective study of MCI patients in a tertiary memory disorder unit. Patients underwent an extensive neuropsychological battery including two tests of declarative verbal memory: The Free and Cued Selective Reminding Test (FCSRT) and the word list learning task from the Consortium to Establish a Registry for Alzheimer’s disease (CERAD-WL). Cerebrospinal fluid (CSF) was obtained from all patients and MCI-AD was defined by means of the t-Tau/Aβ_1–42 ratio. Logistic regression analyses tested whether the combination of FCSRT and CERAD-WL measures significantly improved the prediction of MCI-AD. Progression to AD-d was analyzed in a Cox regression model.

Results: A total of 202 MCI patients with a mean follow-up of 34.2±24.2 months were included and 98 (48.5%) met the criteria for MCI-AD. The combination of FCSRT and CERAD-WL measures improved MCI-AD classification accuracy based on CSF biomarkers. Both tests yielded similar global predictive values (59.9–65.3% and 59.4–62.8% for FCSRT and CERAD-WL, respectively). MCI-AD patients with deficits in both FCSRT and CERAD-WL had a faster progression to AD-d than patients with deficits in only one test.

Conclusions: The combination of FCSRT and CERAD-WL improves the classification of MCI-AD and defines different prognostic profiles. These findings have important implications for clinical practice and the design of clinical trials.

Keywords

Alzheimer’s disease biomarkers cerebrospinal fluid episodic memory mild cognitive impairment mild neurocognitive disorder neuropsychological test

INTRODUCTION

Mild cognitive impairment (MCI) is a broad, heterogeneous clinical syndrome that can include multiple neurodegenerative and non-neurodegenerative conditions [1]. The increasing use of biomarkers in recent years has enabled the identification of a subgroup of MCI patients with underlying Alzheimer’s disease (AD). Imaging and cerebrospinal fluid (CSF) biomarkers, in particular, have shown to be excellent tools for the diagnosis of MCI due to AD (MCI-AD) and both have been incorporated into current diagnostic criteria [2, 3].

The core feature of typical AD is episodic memory impairment. This deficit can be assessed by episodic memory tests, [4] but studies comparing their diagnostic performance in biomarker-supported or autopsy-proven patient samples are scarce [5, 6].

One of the most commonly used tests world-wide is the Free and Cued Selective Reminding Test (FCSRT). The FCSRT is considered a typical episodic memory test that ensures encoding by semantic cueing and has been particularly recommended for the diagnosis of prodromal AD [7]. Other tests that use verbal word lists are the Consortium to Establish a Registry for Alzheimer’s Disease (abbreviated herein as CERAD-WL) or the Rey Auditory Verbal Learning Test (RAVLT). The CERAD-WL test includes a delayed recognition task which has been tightly related to the integrity of the entorhinal and perirhinal cortex, regions known to be selectively vulnerable to AD [8]. Although the current evidence indicate that these tests are reliable tools for the early diagnosis of AD, there is insufficient data about their direct comparison [9]. Some studies claim the use of the FCSRT based on longitudinal studies in subjects with MCI [10 –13]. However, in these studies the progression to AD dementia was assessed during a short follow-up [10 –13]. This approach has two major limitations: first, a limited follow-up may underestimate the full clinical spectrum of AD as some cases may not have the time to progress to the dementia stage. In addition, MCI-AD is heterogeneous and multiple patterns of neurodegeneration and longitudinal trajectories have been recognized [14]. Second, the sensitivity for the diagnosis of AD dementia based only on clinical criteria is limited, as up to a third of the patients may have non-AD pathologies or insufficient AD-related neuropathological changes at autopsy [15, 16]. Thus, we believe that it is necessary to explore the diagnostic accuracy of verbal memory tests in biomarker-supported MCI studies. In this study, we compared the diagnostic accuracy of two commonly used verbal memory tests, the FCSRT and CERAD-WL to predict MCI-AD defined by CSF biomarkers. Importantly, we further provide information about follow-up and the impact of verbal memory profiles on the progression to AD dementia. We hypothesized that the combination of FCSRT and CERAD-WL measures could improve the prediction of MCI-AD and define different prognostic profiles as it would unveil a more accurate picture of verbal memory disintegration along the heterogeneous MCI-AD continuum [14]. This information is crucial to select candidates for biomarker testing in clinical practice and clinical trials [17].

MATERIALS AND METHODS

Study population

Patients were recruited between June 2009 and April 2016 at the Memory Unit at Hospital de la Santa Creu i Sant Pau (Barcelona, Spain) as part of the biomarker program. Briefly, patients underwent a uniform set of clinical, neuropsychological, neuroimaging, and laboratory assessments, including CSF sampling. Subjects were referred by general physicians or neurologists because of cognitive or behavioral complaints. A total of 202 patients with a diagnosis of MCI and CSF biomarker data available were included. We excluded subjects meeting criteria for the diagnosis of dementia [18] at baseline, a history of stroke, other cerebral lesions, substance abuse, psychiatric comorbidities, or other comorbid conditions with excess mortality.

MCI definition

Patients were diagnosed as MCI using a broad definition [1] and met the following criteria: (1) Mini-Mental State Examination (MMSE) score between 24 and 30; (2) cognitive global status of 3 in the Global Deterioration Scale (GDS); and (3) impaired cognitive performance (defined as <1.5 SD on age- and education-adjusted normative scores) in at least one cognitive domain.

Neuropsychological evaluation

Patients underwent a formal neuropsychological evaluation close to CSF sampling (mean interval 0.9±3 months). A detailed description of the neuropsychological testing protocol and verbal memory tests is provided in the Supplementary Material. Declarative verbal memory was evaluated by FCSRT and CERAD-WL. For FCSRT, four scores were considered: total free recall (FCSRT total free recall), total recall (FCSRT total recall), delayed free recall (FCSRT delayed free recall) and delayed total recall (FCSRT delayed total recall). For CERAD-WL, three measures were considered: recall on trial 3 (CERAD-WL trial 3), delayed recall (CERAD-WL recall) and delayed recognition (CERAD-WL recognition). In the recognition task, true positive (items previously presented) and true negative items (novel items) were added to give a maximum score of 20.

Cut-off values for verbal memory tests

We defined specific cut-offs for this study because normalization procedures used for FCSRT and CERAD-WL tests differed [19, 20] and this could interfere with the comparison between both measures. We z-transformed raw values using means and standard deviations (SD) of 48 age-matched, cognitively normal control subjects (33% male, mean age: 67±6.54, education: 13±4.48, MMSE: 29.2±1.05 all with a CDR sum of boxes of 0) [21]. Controls had normal CSF values and thus were classified as stage 0, according to the NIA-AA staging of preclinical AD [22]. The impairment threshold was set at –1.5 SD for each test as in previous studies in patients with MCI [23].

CSF biochemical analysis

CSF was obtained by lumbar puncture and collected following previously reported, international consensus recommendations [24]. All samples were stored in polypropylene tubes at –80°C. We used commercially available ELISA kits to determine levels of Aβ_1 - 42 (Innotest β-amyloid_1 - 42, Fujirebio-Europe, Gent, Belgium), t-Tau (Innotest hTAU Ag, Fujirebio-Europe), p-Tau (Innotest Phospho-Tau _181P, Fujirebio-Europe) following the manufacturers’ recommendations. Our laboratory has experience in CSF biomarker determination and participates in the Alzheimer’s Association external quality control program for CSF biomarkers [25].

CSF biomarker cut-offs

We applied the t-Tau/Aβ_1 - 42 ratio for the diagnosis of MCI-AD because this parameter has shown an excellent diagnostic accuracy in autopsy-proven case-series [26]. We applied our previously validated cut-off of 0.52, which has shown a good diagnostic performance to diagnose AD in our cohort [21] and in other cohorts [26, 27].

Diagnosis of MCI-AD

In accordance with the NIA-AA diagnostic criteria, MCI-AD was applied to any MCI patient with CSF biomarker evidence of AD [3] based on thet-Tau/Aß_1–42 ratio. The remaining MCI patients were labeled as MCI-nonAD.

Genetic analysis

APOE was genotyped according to previously described methods [28].

Longitudinal follow-up

Subjects underwent a clinical and neuropsychological evaluation at least once a year unless marked clinical deterioration occurred. Additional visits were scheduled if judged necessary. As in previous longitudinal studies, [29] cognitive progression to AD dementia was operationalized as a loss of more than 3 points between first and last MMSE administration, or a MMSE score <24, or a clinical diagnosis of dementia at follow-up. Dementia with Lewy bodies (DLB), vascular cognitive impairment and frontotemporal dementia (FTD) were diagnosed according to current clinical criteria [18 , 30–32]. Patients who converted to non-AD dementias (n = 12, 11 DLB and 1 FTD) and MCI patients with a follow-up of less than 6 months (n = 27) were excluded from the longitudinal analysis, as in similar studies.

Statistical analysis

Continuous variables are described as mean and standard deviation (SD) and categorical variables are described as percentages. Differences in baseline characteristics between MCI groups were assessed using the t-test for continuous variables and the Chi-square for dichotomous or categorical data. Nonparametric tests were applied when variables did not follow a normal distribution. The performance of MCI groups was compared in a series of analyses of covariance using age, sex and years of education as covariates. Effect sizes (Cohen d) were derived from the means and SD of adjusted scores. Logistic regression analyses tested whether the combination of FCSRT and CERAD-WL measures significantly improved the prediction of MCI-AD. We calculated the interclass correlation coefficient (raw values) and Cohen’s Kappa index (dichotomous classification) to test agreement between CERAD-WL and FCSRT measures. We tested the ability of the various verbal memory tests to discriminate the MCI-AD group by means of receiver operating characteristics curve analysis. We used the Cox proportional-hazards model to estimate the risk of progression from MCI-AD to AD dementia for verbal memory profiles based on CERAD-WL and FCSRT measures. These statistical models were adjusted for age, sex and education. Statistical significance for all tests was set at 5% (α= 0.05) and all statistical tests were two-sided. All analyses were performed using SPSS 20.0 (Armonk, NY: IBM Corp.).

Standard protocol approvals, registrations, and patient consent

The study was approved by the local ethics committee and was conducted in accordance with the Declaration of Helsinki. All participants gave their written informed consent to participate in the study.

RESULTS

Patients’ characteristics

Ninety-eight of 202 MCI patients (48.5%) had a CSF profile consistent with AD (MCI-AD). Table 1 shows their demographic, clinical, and CSF biomarker characteristics. The MCI-AD group was older (70.7 versus 66.9 years; p = 0.002) and had a higher frequency of women (61.2% versus 47.1%; p = 0.044) and APOE ɛ4 carriers than the MCI-nonAD group (57.1% versus 21.2%; p < 0.001). Global cognitive impairment as evaluated using the Clinical Dementia Rating Scale Sum of Boxes (CDR-SOB) was higher in the MCI-AD group (2.12 versus 1.33; p < 0.001) but functional impairment (assessed by the Interview for Deterioration in Daily living activities in Dementia (IDDD) was similar in both groups (p = 0.450).

Table 1

Comparison of patient characteristics between MCI-AD and MCI-nonAD groups

	MCI-AD	MCI-nonAD		All MCI
	n = 98	n= 104		n = 202
	(48.5%)	(51.5%)	P	(100%)
Age at diagnosis, years	70.7±7.1	66.9±8.7	0.002	68.7±8.2
Women	60 (61.2)	49 (47.1)	0.044	109 (54)
Education, years	10.3±4.6	11.2±5.1	0.203	10.8±4.9
Time to MCI diagnosis, months	33.3±29.6	32.3±25.2	0.998	32.8±27.4
APOEɛ4	56 (57.1)	22 (21.2)	< 0.001	78 (38.6)
IDDD at diagnosis	37.95±5.1	38±6.8	0.450	38±5.9
MMSE score at diagnosis	27±2.7	27.6±2	0.130	27.3±2.4
GDS	10.8±6.6	11.2±6.8	0.704	11±6.7
CDR-SOB	2.12±1.38	1.33±0.98	< 0.001	1.70±1.24
Follow-up, months	33.4±24	35.1±24.7	0.580	34.2±24.2
CSF Aβ_1–42, pg/mL	467.7±142.9	856.9±249.6	< 0.001	668.1±282.5
CSF t-Tau, pg/mL	575.3±306.4	240.7±87.8	< 0.001	403.1±278.1
CSF p-Tau, pg/mL	88.5±40.2	48.1±14.4	< 0.001	67.7±36

Results are mean±SD (standard deviation) for continuous variables or frequency (%). In bold, p < 0.05. MCI-AD, mild cognitive impairment due to Alzheimer’s disease; MCI-nonAD, mild cognitive impairment not due to Alzheimer’s disease; MCI, mild cognitive impairment; IDDD, Interview for Deterioration of Daily Life in Dementia; MMSE, Mini-Mental State Examination; GDS, Geriatric Depression Scale; CDR-SOB, Clinical Dementia Rating Sum of Boxes; CSF, cerebrospinal Fluid.

Diagnostic accuracy of FCSRT and CERAD-WL for the diagnosis of MCI-AD

Table 2 shows the diagnostic accuracy for each verbal memory measured alone and in combination. FCSRT total free recall had the highest sensitivity (90.8%) while CERAD-WL recognition showed the highest specificity (79.6%). These two tests yielded similar global predictive values (59.9–65.3% and 59.4–62.8% for FCSRT and CERAD-WL, respectively). The positive predictive and negative predictive values were highest for FCSRT total free recall and CERAD-WL recognition (66.1% and 82.3%, respectively). Globally, FCSRT measures were more sensitive than CERAD-WL, but CERAD-WL measures were more specific. Combined measures increased sensitivity at the expense of decreasing specificity. For example, sensitivity was found to be highest for the combination (at least one abnormal test) of FCSRT total free recall and FCSRT total recall measures (93.9%). However, this was at the expense of a low specificity (26.9%). The combination of abnormal FCSRT total free recall and FCSRT total recall measures had the highest global predictive value (65.3%) and yielded a sensitivity of 84.7% and a specificity of 47.1%. FCSRT AUC values were slightly superior for raw values, but when adjusting for age, sex and education, the two tests yielded a similar diagnostic performance.

Table 2

Diagnostic performance of verbal memory tests and combined memory profiles in MCI-AD subjects

	Cut-off	S	E	PPV	NPV	GPV
		(%)	(%)	(%)	(%)	(%)	AUC	AUC^*
Verbal memory tests
CERAD-WL trial 3	≤6	76.6	43.7	55.4	67.2	59.4	0.697	0.597
CERAD-WL recall	≤2	79.6	44.7	56.5	70.8	61.2	0.702	0.601
CERAD-WL recognition	≤15	44.1	79.6	66.1	61.2	62.8	0.685	0.623
FCSRT total free recall	≤17	90.8	40.4	58.9	82.3	64.8	0.746	0.601
FCSRT total recall	≤38	87.7	33.6	55.5	74.5	59.9	0.711	0.600
FCSRT delayed free recall	≤6	78.6	50	59.7	71.2	65.3	0.716	0.593
FCSRT delayed total recall	≤13	84.7	37.9	56.5	72.2	60.7	0.734	0.597
Combined verbal memory profiles
CERAD-WL recall and CERAD-WL recognition abnormal	–	39.8	84.5	69.8	60.8	63.3	–	–
FCSRT delayed free recall and FCSRT delayed total recall abnormal	–	74.5	56.3	61.9	69.9	65.2	–	–
FCSRT total free recall or CERAD-WL recall abnormal		93.9	24	53.8	80.6	57.9	–	–
FCSRT total free recall and CERAD-WL recognition abnormal		41.8	84.6	71.9	60.7	63.8	–	–

^*Adjusted for age, sex and years of education. S, sensitivity; E, specificity; PPV, positive predictive value; NPV, negative predictive value; GPV, global predictive value; AUC, area under the ROC curve; CERAD-WL, Word List from the Consortium to Establish a Registry for Alzheimer’s Disease; FCSRT, Free and Cued Selective Reminding Test; MCI, mild cognitive impairment.

Low agreement between FCSRT and CERAD-WL

A low agreement in interclass correlation coefficient (ICC) and Cohen’s Kappa index (κ) was noted between FCSRT and CERAD-WL verbal memory measures in this population (Supplementary Tables 2 and 3). Importantly, similar agreement rates were noted when using both raw data (ICC) and dichotomized groups (κ), indicating that low agreement rates were not related to the cut-offs used.

FCSRT and CERAD-WL for the prediction of MCI-AD

All verbal memory measures were lower in the MCI-AD group than in the MCI-nonAD group (Supplementary Table 1). Measures with the highest effect size for each test were the FCSRT delayed total recall (d = 0.976) and CERAD-WL trial 3 (d = 0.956). The model fit using forward stepwise logistic regression was best when the CERAD-WL trial 3 and FCSRT delayed total recall were included (Nagelkerke R² = 0.319). Models based on FCSRT measures improved significantly when CERAD-WL measures were added (see Table 3).

Table 3

Logistic regression analyses to predict MCI-AD in MCI subjects

Variables introduced in the logistic regression analysis*		Difference		Nagelkerke
	–2LL^†	χ ²	p	R ²
FCSRT delayed total recall	226.7	–	–	0.297
FCSRT delayed total recall + CERAD-WL trial 3	217.8	8.9	0.004	0.319
FCSRT delayed total recall + CERAD-WL recognition	218.9	7.8	0.005	0.307
FCSRT delayed total recall + CERAD-WL recall	216.8	9.9	< 0.001	0.318
FCSRT total free recall	233.5	–	–	0.274
FCSRT total free recall + CERAD-WL trial 3	223.0	10.5	< 0.001	0.297
FCSRT total free recall + CERAD-WL recognition	222.5	11	< 0.001	0.294
FCSRT total free recall + CERAD-WL recall	222.7	10.8	< 0.001	0.292
FCSRT delayed free recall	238.5	–	–	0.241
FCSRT delayed free recall + CERAD-WL trial 3	227.0	11.5	< 0.001	0.270
FCSRT delayed free recall + CERAD-WL recognition	228.1	10.4	< 0.001	0.258
FCSRT delayed free recall + CERAD-WL recall	228.2	10.3	< 0.001	0.257

^*Age, sex, and years of education were introduced first as covariates in all analyses; †, Lower scores indicate better model fit; In bold, p < 0.05. MCI-AD, mild cognitive impairment due to Alzheimer’s disease; CERAD-WL, Word list of the Consortium to Establish a Registry for Alzheimer’s Disease; FCSRT, Free and Cued Selective Reminding Test.

Use of FCSRT and CERAD-WL to predict conversion to AD dementia in MCI-AD

Of the 163 patients selected for longitudinal analysis, 78 (47.8%) had an AD CSF profile (MCI-AD). After a mean follow-up of 34.2±24.2 months, almost half of these patients (35/78, 44.9%) had progressed to AD dementia. Four patients labeled as MCI nonAD (2.4%) at baseline converted to AD dementia during follow-up (two had borderline Aβ_1 - 42 levels and two had a CSF profile compatible with suspected non-Alzheimer disease pathophysiology (SNAP) (for a detailed description of patients’ characteristics see Supplementary Table 4). As shown in Fig. 1, the combination of FCSRT and CERAD-WL subtests defined different prognostic profiles in MCI-AD patients. After adjusting for age, sex, and education, MCI-AD patients who failed both FCSRT delayed free recall and CERAD-WL recall had a faster progression to AD dementia than patients who failed only one test (HR = 4.7; p = 0.018). Similarly, patients who failed both FCSRT delayed total recall and CERAD-WL recognition tests had a faster progression to AD dementia than patients who failed only one test (HR = 2.39; p = 0.019, Supplementary Table 5). As shown in Fig. 1, amnestic profiles did not differ in the non-amnestic global z-scores, suggesting that progression rates to AD dementia were not driven by global cognitive impairment (data not shown).

Fig.1

Survival plots for the progression to AD dementia in MCI-AD. A) Survival plots of FCSRT delayed free recall and CERAD-WL delayed recall amnestic profiles; B) Survival plots of FCSRT delayed total recall and CERAD-WL recognition amnestic profiles; ^*, p < 0.05 (Mantel-Cox); ns, no statistically significant difference; +, abnormal result according to the calculated cut-off; -, normal result according to the calculated cut-off; MCI-AD, mild cognitive impairment due to Alzheimer’s disease; FCSRT, Free and Cued Selective Reminding Test; CERAD-WL, Word list of the Consortium to Establish a Registry for Alzheimer’s Disease.

DISCUSSION

In this study, we found that the combination of FCSRT and CERAD-WL measures improves the identification of subjects with MCI-AD defined by CSF biomarkers. In addition, the combination of a semantically-cued test and a free list-learning test with a recognition task defined distinct prognostic profiles in MCI-AD.

We found a relatively low agreement between FCSRT and CERAD-WL, indicating that each of these tests measures different but overlapping aspects of declarative memory [4]. Although both tests evaluate free retrieval, the FCSRT could more closely reflect the hippocampal-mediated consolidationprocess because registration is ensured by semantic cueing. The CERAD-WL test, in contrast, includes a recognition task that implies both recollection and familiarity, functions that depend more on extra-hippocampal areas [8]. Previous studies have examined the neural substrate of the different components of episodic verbal memory using the RAVLT, a non-cued list-learning test with a delayed recognition task [8]. A study using MRI found a close relationship between RAVLT recognition and atrophy in perirhinal and entorhinal regions [33], while a study based on ¹⁸F-fluoro-D-glucose positron emission tomography found that the delayed recall performance was associated with hypometabolism in orbitofrontal, cingulate-precuneus, and parietal areas [34]. Conversely, other imaging studies have shown that cued-recall measures of the FCSRT are related to medial temporal lobe atrophy, although other temporal and parietal regions have been also involved [6 , 36]. Moreover, a recent study has highlighted heterogeneity in amnestic FCSRT profiles among MCI-AD patients as defined by CSF core AD biomarkers [37]. Taken together, these data indicate that different verbal memory tests capture different aspects of episodic memory that reflect interrelated but distinct neural circuits.

Comparative data are scarce with regard to the diagnostic performance of verbal memory tests in biomarker-supported or autopsy-proven patient samples and most studies have relied on clinical diagnosis [10 –13]. In the only comparable study so far, the authors found that the FCSRT test performed better than the CERAD-WL test to predict MCI-AD [5]. Their sample size (185 patients) was similar to ours (202 patients) and they used a similar broad definition of MCI, but we were unable to replicate their findings. However, we identified some methodological differences that might explain the discrepant results. First, in our study we added both true positive and true negative responses to the CERAD-WL recognition final score, while Wagner et al. considered only true positive responses. Second, we included the CERAD-WL trial 3 score in the analysis but Wagner et al. did not. It is of note that both these subtests appeared to be important predictors of MCI-AD in our study.

We observed that the FCSRT showed a slightly higher sensitivity than CERAD-WL in the diagnosis of MCI-AD (specially, FCSRT total free recall and FCSRT total recall) but both tests had low specificities, reinforcing the notion that MCI is a heterogeneous syndrome [38, 39]. It is known that MCI can be caused by pathophysiological processes other than AD that target the medial temporal lobe structures. Hippocampal sclerosis, TDP-43 pathology, primary age-related tauopathy, and other non-neurodegenerative conditions may explain some of these MCI-nonAD cases [40 –42]. Our results could have implications for clinical practice and clinical trials. In clinical practice, when selecting MCI patients for biomarker testing, FCSRT could be used first and then CERAD-WL could be applied if FCSRT is normal. This would maximize the sensitivity to detect MCI-AD in clinical practice. In a clinical trial, measures with high specificity, such as CERAD-WL recognition, could be used in order to minimize the number of screening failures due to biomarker negative results. Further studies are needed to test if the addition of a delayed recognition task to the FCSRT may help increasing specificity for the diagnosis of MCI-AD.

The progression rates to AD dementia during the longitudinal follow up of the patients were similar to those in previous large collaborative studies [23, 43]. Of note, 4 patients with MCI-nonAD converted to AD dementia during follow-up, possibly due to an atypical CSF biomarker AD profile (SNAP or borderline Aβ_1–42 levels) or phenocopies due to non-AD conditions [23, 41].

It is also of note in our study that MCI-AD patients with deficits in both FCSRT and CERAD-WL progressed to dementia faster than those with deficits in only one memory test. These data is in agreement with other previous studies that have described different patterns of neurodegeneration in MCI-AD [14, 39]. We speculate that because FCSRT and CERAD-WL reflect slightly different neural substrates, MCI patients with deficits in both tests may have a higher pathology burden in AD-related structures and their disorder could therefore be at a more advanced stage in the disease continuum. Further imaging studies are needed to confirm this hypothesis.

This study has several limitations. First, we did not analyze measures of visual memory and we acknowledge that the combination of verbal and visual memory measures could further increase the ability to detect MCI-AD [17]. Second, the inclusion of a second verbal memory test could have caused interference when administering the two tests. To minimize this effect, however, we allowed a minimum time between both memory tests and included a non-verbal task. Nonetheless, interference is a potential limitation of all studies that introduce multiple declarative memory tests. Third, we defined MCI-AD using the t-Tau/Aβ_1–42 ratio instead of decreased Aβ_1–42 levels together with increased t-Tau or p-Tau. Our decision to do so was based on large neuropathological series where t-Tau/Aβ_1–42 ratio has proven to be the most sensitive CSF biomarker for AD [26, 44]. Finally, the fact that we used broad criteria to define MCI could be seen as a limitation in a heterogeneous syndrome such as MCI. However, the frequency of MCI-AD being consistent with previous CSF-based studies on MCI population [23, 43] suggests that the MCI sample in our study is comparable to many previous studies.

In summary, we provide evidence that the classification of MCI-AD can be improved by combining FCSRT and CERAD-WL. The diagnostic performance of each test should be considered when MCI-AD is suspected and when selecting patients for biomarker testing in clinical trials. Importantly, the combination had prognostic implications. Our findings highlight the heterogeneity of the amnestic syndrome in the earliest symptomatic phase of AD and reinforce the need to use pathophysiological markers to diagnose AD at this stage.

Footnotes

ACKNOWLEDGMENTS

This work was supported by research grants from the Carlos III Institute of Health, Spain (grants PI11/02425 and PI14/01126 to Juan Fortea, grants PI10/1878 and PI13/01532 to Rafael Blesa, grants PI11/03035 and PI14/1561 to Alberto Lleó) and the CIBERNED program (Program 1, Alzheimer Disease to Alberto Lleó and SIGNAL study, ), partly funded by Fondo Europeo de Desarrollo Regional (FEDER), Unión Europea, “Una manera de hacer Europa”. This work has also been supported by a “Marató TV3” grant (20141210 to Juan Fortea) and by by Generalitat de Catalunya (2014SGR-0235). I. Illán-Gala is supported by the i-PFIS grant (IF15/00060) from the FIS, Instituto de Salud Carlos III.

The authors thank the patients and their relatives for their support for this study. We also thank Laia Muñoz and Carolyn Newey for technical and editorial assistance.

Authors’ disclosures available online ().

Appendix

References

Petersen

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

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: TheIWG-2 criteria. Lancet Neurol 13, 614–629.

Albert

, Dekosky

, Dickson

, Dubois

, Feldman

, Fox

, Gamst

, Holtzman

, Jagust

, Petersen

, Snyder

, Carrillo

, Thies

, Phelps

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

Lezak

, Howieson

, Bigler

, Tranel

(2012) Neuropsychological Assessment, Oxford University Press.

Wagner

, Wolf

, Reischies

, Daerr

, Wolfsgruber

, Jessen

, Popp

, Maier

, Hüll

, Frolich

, 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.

Quenon

, Dricot

, Woodard

, Hanseeuw

, Gilis

, Lhommel

, Ivanoiu

(2016) Prediction of free and cued selective reminding test performance using volumetric and amyloid-based biomarkers of Alzheimer’s disease. J Int Neuropsychol Soc 22, 991–1004.

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.

Wolk

, Dunfee

, Dickerson

, Aizenstein

, Dekosky

(2011) A medial temporal lobe division of labor: Insights from memory in aging and early Alzheimer disease. Hippocampus 21, 461–466.

Fillenbaum

, van Belle

, Morris

, Mohs

, Mirra

, Davis

, Tariot

, Silverman

, Clark

, Welsh-Bohmer

, Heyman

(2008) Consortium to Establish a Registry for Alzheimer’s Disease (CERAD): The first twenty years. Alzheimers Dement 4, 96–109.

10.

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.

11.

Derby

, Burns

, Wang

, Katz

, Zimmerman

, L’italien

, Guo

, Berman

, Lipton

(2013) Screening for predementia AD: Time-dependent operating characteristics of episodic memory tests. Neurology 80, 1307–1314.

12.

Mura

, Proust-Lima

, Jacqmin-Gadda

, Akbaraly

, Touchon

, Dubois

, Berr

(2014) Measuring cognitivechange in subjects with prodromal Alzheimer’s disease. J Neurol Neurosurg Psychiatry 85, 363–370.

13.

Lemos

, Afonso

, Martins

, Waters

, Blanco

, Simões

, Santana

(2015) Selective reminding and free and cued selectivereminding in mild cognitive impairment and Alzheimer disease. Appl Neuropsychol Adult 23, 85–93.

14.

Dong

, Toledo

, Honnorat

, Doshi

, Varol

, Sotiras

, Wolk

, Trojanowski

, Davatzikos

, Alzheimer’s Disease Neuroimaging Initiative (2017) Heterogeneity of neuroanatomical patterns in prodromal Alzheimer’s disease: Links to cognition, progression and biomarkers. Brain 140, 735–747.

15.

Blennow

, Dubois

, Fagan

, Lewczuk

, de Leon

, Hampel

(2015) Clinical utility of cerebrospinal fluid biomarkers in the diagnosis of early Alzheimer’s disease. Alzheimers Dement 11, 58–69.

16.

Serrano-Pozo

, Qian

, Monsell

, Blacker

, Gomez-Isla

, Betensky

, Growdon

, Johnson

, Frosch

, Sperling

, Hyman

(2014) Mild to moderate Alzheimer dementia with insufficient neuropathological changes. Ann Neurol 75, 597–601.

17.

Gainotti

, Quaranta

, Vita

, Marra

(2014) Neuropsychologicalpredictors of conversion from mild cognitive impairment toAlzheimer’s disease. J Alzheimers Dis 38, 481–495.

18.

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’sAssociation workgroups on diagnostic guidelines for Alzheimer’sdisease. Alzheimers Dement 7, 263–269.

19.

Peña-Casanova

, Blesa

, Aguilar

, Gramunt-Fombuena

, Gómez-Ansón

, Oliva

, Molinuevo

, Robles

, Barquero

, Antúnez

, Martínez-Parra

, Frank-García

, Fernández

, Alfonso

, Sol

, Neuronorma Study Team (2009) Spanish Multicenter Normative Studies (NEURONORMA Project): Methods and sample characteristics. Arch Clin Neuropsychol 24, 307–319.

20.

Van Den Eeden

, Nelson

, Tanner

, McGuire

(2004) RegionalNorth American Annual Meeting of the World Federation of Neurology–Research Group on Neuroepidemiology. Neuroepidemiology 23, 151–158.

21.

Alcolea

, Martínez-Lage

, Sánchez-Juan

, Olazarán

, Antúnez

, Izagirre

, Ecay-Torres

, Estanga

, Clerigué

, Guisasola

, Sánchez Ruiz

, MarínMuñoz

, Calero

, Blesa

, Clarimon

, Carmona-Iragui

, Morenas-Rodríguez

, Rodriguez-Rodriguez

, Vázquez-Higuera

, Fortea

, Lleo

(2015) Amyloid precursorprotein metabolism and inflammation markers in preclinicalAlzheimer disease. Neurology 85, 626–633.

22.

Sperling

, Aisen

, Beckett

, Bennett

, Craft

, Fagan

, Iwatsubo

, Jack

, Kaye

, Montine

, Park

, Reiman

, Rowe

, Siemers

, Stern

, Yaffe

, Carrillo

, Thies

, Morrison-Bogorad

, Wagster

, Phelps

(2011) Toward definingthe preclinical stages of Alzheimer’s disease: Recommendationsfrom the National Institute on Aging-Alzheimer’s Associationworkgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 280–292.

23.

Vos

SJB

, Verhey

, Frölich

, Kornhuber

, Wiltfang

, Maier

, Peters

, Rüther

, Nobili

, Morbelli

, Frisoni

, Drzezga

, Didic

, van Berckel

BNM

, Simmons

, Soininen

, Kłoszewska

, Mecocci

, Tsolaki

, Vellas

, Lovestone

, Muscio

, Herukka

, Salmon

, Bastin

, Wallin

, Nordlund

, deMendonça

, Silva

, Santana

, Lemos

, Engelborghs

, Vander Mussele

, Alzheimer’s Disease Neuroimaging Initiative, Freund-Levi

, Wallin

ÅK

, Hampel

, van der Flier

, Scheltens

, Visser

(2015) Prevalence and prognosis of Alzheimer’sdisease at the mild cognitive impairment stage. Brain 138, 1327–1338.

24.

Alcolea

, Carmona-Iragui

, Suárez-Calvet

, Sánchez-Saudinós

, Sala

, Antón-Aguirre

, Blesa

, Clarimon

, Fortea

, Lleo

(2014) Relationship between β-Secretase, inflammation and core cerebrospinal fluid biomarkers for Alzheimer’s disease. J Alzheimers Dis 42, 157–167.

25.

Mattsson

, Andreasson

, Persson

, Carrillo

, Collins

, Chalbot

, Cutler

, Dufour-Rainfray

, Fagan

, Heegaard

NHH

, Robin Hsiung

G-Y

, Hyman

, Iqbal

, Kaeser

, Käser

, Lachno

, Lleo

, Lewczuk

, Molinuevo

, Parchi

, Regeniter

, Rissman

, Rosenmann

, Sancesario

, Schröder

, Shaw

, Teunissen

, Trojanowski

, Vanderstichele

, Vandijck

, Verbeek

, Zetterberg

, Blennow

, Alzheimer’s Association QC Program Work Group (2013) CSF biomarker variability in the Alzheimer’s Association quality control program. Alzheimers Dement 9, 251–261.

26.

Shaw

, Vanderstichele

, Knapik-Czajka

, Clark

, Aisen

, Petersen

, Blennow

, Soares

, Simon

, Lewczuk

, Dean

, Siemers

, Potter

, Lee

VM-Y

, Trojanowski

, Alzheimer’s Disease Neuroimaging Initiative (2009) Cerebrospinal fluid biomarker signature in Alzheimer’s disease neuroimaging initiative subjects. Ann Neurol 65, 403–413.

27.

Duits

, Teunissen

, Bouwman

, Visser

, Mattsson

, Zetterberg

, Blennow

, Hansson

, Minthon

, Andreasen

, Marcusson

, Wallin

, Rikkert

, Tsolaki

, Parnetti

, Herukka

, Hampel

, de Leon

, Schröder

, Aarsland

, Blankenstein

, Scheltens

, van der Flier

(2014) The cerebrospinal fluid “Alzheimer profile”: Easily said, but what does it mean? Alzheimers Dement 10, 713–723.e2.

28.

Calero

, Hortigüela

, Bullido

, Calero

(2009) Apolipoprotein E genotyping method by real time PCR, a fast and cost-effective alternative to the TaqMan and FRET assays. J Neurosci Methods 183, 238–240.

29.

Caroli

, Prestia

, Galluzzi

, Ferrari

, van der Flier

, Ossenkoppele

, Van Berckel

, Barkhof

, Teunissen

, Wall

, Carter

, Schöll

, Choo

, Grimmer

, Redolfi

, Nordberg

, Scheltens

, Drzezga

, Frisoni

, Alzheimer’s Disease Neuroimaging Initiative (2015) Mild cognitive impairment with suspected nonamyloid pathology (SNAP): Prediction of progression. Neurology 84, 508–515.

30.

McKeith

, Dickson

, Lowe

, Emre

, O’Brien

, Feldman

, Cummings

, Duda

, Lippa

, Perry

, Aarsland

, Arai

, Ballard

, Boeve

, Burn

, Costa

, del Ser

, Dubois

, Galasko

, Gauthier

, Goetz

, Gómez-Tortosa

, Halliday

, Hansen

, Hardy

, Iwatsubo

, Kalaria

, Kaufer

, Kenny

, Korczyn

, Kosaka

, Lee

VMY

, Lees

, Litvan

, Londos

, Lopez

, Minoshima

, Mizuno

, Molina

, Mukaetova-Ladinska

, Pasquier

, Perry

, Schulz

, Trojanowski

, Yamada

, Consortium on DLB (2005) Diagnosis and management of dementia with Lewy bodies: Third report of the DLB Consortium. Neurology 65, 1863–1872.

31.

Rascovsky

, Hodges

, Knopman

, Mendez

, Kramer

, Neuhaus

, van Swieten

, Seelaar

, Dopper

EGP

, Onyike

, Hillis

, Josephs

, Boeve

, Kertesz

, Seeley

, Rankin

, Johnson

, Gorno-Tempini

M-L

, Rosen

, Prioleau-Latham

, Lee

, Kipps

, Lillo

, Piguet

, Rohrer

, Rossor

, Warren

, Fox

, Galasko

, Salmon

, Black

, Mesulam

, Weintraub

, Dickerson

, Diehl-Schmid

, Pasquier

, Deramecourt

, Lebert

, Pijnenburg

, Chow

, Manes

, Grafman

, Cappa

, Freedman

, Grossman

, Miller

(2011) Sensitivity of revised diagnosticcriteria for the behavioural variant of frontotemporal dementia. Neurology 134, 2456–2477.

32.

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.

33.

Braak

, Braak

(1991) Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol 82, 239–259.

34.

Brugnolo

, Morbelli

, Arnaldi

, De Carli

, Accardo

, Bossert

, Dessi

, Famà

, Ferrara

, Girtler

, Picco

, Rodriguez

, Sambuceti

, Nobili

(2014) Metabolic correlates of Reyauditory verbal learning test in elderly subjects with memorycomplaints. J Alzheimers Dis 39, 103–113.

35.

Sarazin

, Chauviré

, Gerardin

, Colliot

, Kinkingnéhun

, de Souza

, Hugonot-Diener

, Garnero

, Lehericy

, Chupin

, Dubois

(2010) The amnestic syndrome of hippocampal type in Alzheimer’s disease: An MRI study. J Alzheimers Dis 22, 285–294.

36.

Philippi

, Noblet

, Duron

, Cretin

, Boully

, Wisniewski

, Seux

, Martin-Hunyadi

, Chaussade

, Demuynck

, Kremer

, Lehéricy

, Gounot

, Armspach

, Hanon

, Blanc

(2016) Exploring anterograde memory: A volumetric MRI study in patients with mild cognitive impairment. Alzheimers Res Ther 8, 1–14.

37.

Xie

, Gabelle

, Dorey

, Garnier-Crussard

, Perret-Liaudet

, Delphin-Combe

, Bathsavanis

, Dauphinot

, Lehmann

, Mercier

, Desestret

, Roullet-Solignac

, Vighetto

, Krolak-Salmon

(2014) Initial memory deficit profiles in patients with a cerebrospinal fluid Alzheimer’s disease signature. J Alzheimers Dis 41, 1109–1116.

38.

Edmonds

, Delano-Wood

, Clark

, Jak

, Nation

, McDonald

, Libon

, Au

, Galasko

, Salmon

, Bondi

, Alzheimer’s Disease Neuroimaging Initiative (2015) Susceptibility of the conventional criteria for mild cognitive impairment to false-positive diagnostic errors. Alzheimers Dement 11, 415–424.

39.

Edmonds

, Eppig

, Bondi

, Leyden

, Goodwin

, Delano-Wood

, McDonald

, Alzheimer’s Disease Neuroimaging Initiative (2016) Heterogeneous cortical atrophy patterns in MCI not capturedby conventional diagnostic criteria. Neurology 87, 2108–2116.

40.

Crary

, Trojanowski

, Schneider

, Abisambra

, Abner

, Alafuzoff

, Arnold

, Attems

, Beach

, Bigio

, Cairns

, Dickson

, Gearing

, Grinberg

, Hof

, Hyman

, Jellinger

, Jicha

, Kovacs

, Knopman

, Kofler

, Kukull

, Mackenzie

, Masliah

, McKee

, Montine

, Murray

, Neltner

, Santa-Maria

, Seeley

, Serrano-Pozo

, Shelanski

, Stein

, Takao

, Thal

, Toledo

, Troncoso

, Vonsattel

, White

, Wisniewski

, Woltjer

, Yamada

, Nelson

(2014) Primary age-related tauopathy (PART): A common pathology associated with human aging. Acta Neuropathol 128, 755–766.

41.

Jack

, Knopman

, Chételat

, Dickson

, Fagan

, Frisoni

, Jagust

, Mormino

, Petersen

, Sperling

, van der Flier

, Villemagne

, Visser

, Vos

SJB

(2016) Suspected non-Alzheimer disease pathophysiology –concept and controversy. Nat Rev Neurol 12, 117–124.

42.

Bertoux

, de Souza

, Corlier

, Lamari

, Bottlaender

, Dubois

, Sarazin

(2014) Two distinct amnesic profiles in behavioral variant frontotemporal dementia. Biol Psychiatry 75, 582–588.

43.

Petersen

, Aisen

, Boeve

, Geda

, Ivnik

, Knopman

, Mielke

, Pankratz

, Roberts

, Rocca

, Weigand

, Weiner

, Wiste

, Jack

(2013) Mild cognitive impairment due to Alzheimer disease in the community. Ann Neurol 74, 199–208.

44.

Toledo

, Brettschneider

, Grossman

, Arnold

, Hu

, Xie

, Lee

VMY

, Shaw

, Trojanowski

(2012) CSF biomarkers cutoffs:The importance of coincidentneuropathological diseases. Acta Neuropathol 124, 23–35.