Semantic Memory as an Early Cognitive Marker of Alzheimer’s Disease: Role of Category and Phonological Verbal Fluency Tasks

Abstract

Background:

The assessment of semantic memory may be a useful marker to identify individuals with mild cognitive impairment (MCI) who will progress to Alzheimer’s disease (AD) in the early stages of the disease.

Objective:

The aim of this five-year follow-up longitudinal study is to assess whether semantic assessment could predict progression in MCI.

Methods:

A population of MCI (N = 251); mild (N = 178) and moderate AD (N = 114); and a sample of healthy participants (HP; N = 262) was investigated. The five-year follow-up of the MCI group was completed by 178 patients. Semantic and episodic memory measures were used, including a measure of the discrepancy between categorical and phonological verbal fluency, the semantic–phonological delta (SPD). The main outcome was the progression of MCI due to AD to dementia.

Results:

A general linear model showed a significant effect of diagnosis on SPD (Wilks’ Lambda = 0.591; p < 0.001). The estimated marginal means were –0.91 (SE = 0.185) in HP, –1.83 (SE = 0.187) in MCI, –1.16 (SE = 0.218) in mild AD, and –1.02 (SE = 0.275) in moderate AD. Post-hoc comparisons showed a significant difference between MCI and HP (p < 0.001). The follow-up was completed by 178 MCI individuals. SPD in MCI patients who progress to dementia was significantly lower than in MCI that will not progress (p = 0.003). Together with the Mini-Mental State Examination, the SPD was the only measure with a significant predicting effect at the five-years follow-up (p = 0.016).

Conclusion:

The SPD indicates the impairment of semantic memory in individuals with underlying AD at the MCI early stage, reflecting the early involvement of perirhinal and entorhinal cortices in the earliest stages of AD neuropathological process.

Keywords

Alzheimer’s disease category fluency task memory mild cognitive impairment

INTRODUCTION

According to the multisystem model proposed by Cohen and Squire in 1980 [1], declarative memory is divided into two subsystems: episodic memory (EM) and semantic memory (SM). EM permits the encoding, storage, and recollection of facts and events in a specific spatiotemporal context; SM is a system that processes conceptual knowledge, associated lexical meanings (either verbal or nonverbal), and the relations existing among these meanings. Furthermore, it refers to a complex architecture of abstract ideas, associations of images, and hierarchical representation of features specific to each category or shared among different categories. In the early stages of Alzheimer’s disease (AD), an EM deficit is considered the pathognomonic presentation of typical AD.

Several tests of EM have been proposed as a marker of AD [2] and several diagnostic markers of early AD have been developed to investigate qualitative and quantitative aspects of EM in individuals with mild cognitive impairment (MCI) [3 –5]. For example, the Free and Cued Selective Reminding Test (FCSRT) has been proposed as a typical test to detect EM disorders due to prodromal AD [6, 7]. Consequently, this task has often been considered specific to early diagnosis of AD, and it is widely used as a screening instrument to enter prodromal AD patients into pharmacological trials. However, clinical experience suggests that even patients in the mild dementia stage can score above cut-off on this test if they are highly educated.

In recent years, SM has been proposed as a marker of early AD based on both clinical and neuroimaging evidence [8]. The complexity of SM has a neurobiological counterpart, and several models have been proposed to explain how semantic information is organized in the brain. The complexity of this organization and the transcultural variability have probably prevented researchers from systematically studying this issue in AD.

Monsch et al. [9] were probably the first authors to suggest that from the early stages of dementia, SM, investigated through the category fluency task (CFT), could be useful in distinguishing AD patients from control participants. In addition, in the prodromal stage of AD, SM is persistently impaired since at the first stages of MCI condition [10].

Although SM has received less attention than EM as both a diagnostic marker of AD and a putative marker of progression of MCI patients, several reasons could encourage researchers to explore this field in AD.

Although it is well known that SM is widely distributed in different regions of the brain [11], and that the left temporal pole represents a hub region for conceptual knowledge, a critical role in the item-context analyses and knowledge is grounded in the medial perirhinal cortex (mPRC). This region has been reported to be involved at the earliest stage of the AD pathological process, even before lesions spread to the entorhinal cortex (ERC) and hippocampus [12], and this observation reasonably supports the idea that impairment of SM could precede that of episodic memory. Several studies [13 –15] have shown how in early AD patients, semantic processing of objects with different levels of specificity depends on the integrity of mPRC. Furthermore, the disambiguation of semantically confusable objects specifically relies on the integrity of mPRC (but not ERC) [16]. From a clinical point of view, an experimental assessment using a delayed matching-to-sample task has shown that performance on SM tasks has a high predictive value for progression from MCI to AD [17].

Another reason that supports the usefulness of SM assessment in identifying MCI patients that will progress to AD is the general resilience of SM in normal aging. Specifically, SM shows a general resistance while EM declines naturally with age [18]. Moreover, in the normal elderly population, performance on the CFT is generally better than on the phonological verbal fluency (PVF) task, showing a discrepancy between pure lexical abilities and semantic attainment in older age [19]. A methodological consequence of these observations is that tests assessing SM would not be biased by normal age effect and could provide a more reliable method in distinguishing normal from pathological aging both in early- and late-onset AD.

Verbal fluency tasks have been already used to assess SM in the course of neurodegenerative diseases. Several papers showed that PVF and CFT are differently affected in frontotemporal dementia. Specifically, behavioral variant of frontotemporal dementia has been shown to be disadvantaged on PVF task while semantic dementia is disadvantaged on CFT [20].

CFT has been also shown to distinguish MCI from healthy individuals [21 –23]. CFT has been shown to be more reliable than PVF and naming tasks in distinguishing prodromal AD or preclinical MCI from normal controls, as well as in predicting MCI progression to AD [24]. At some extent, CFT performance depends upon an executive component, which makes the analyses of raw scores only a biased measure of SM. In order to disentangle the executive component from the semantic component in the CFT score, an adjusted score obtained by making a delta between the CFT and PVF scores [19] has been proposed. The use of the PVF/CFT ‘discrepancy score’ could be a reliable method to explore a pure semantic domain in the early stages of cognitive decline in AD.

The aim of the present study was to explore a large sample that included healthy controls, amnestic MCI (aMCI) patients due to AD and mild AD patients with the pattern of impairment in SM by studying CFT per se and the PVF/CFT discrepancy score. Furthermore, these semantic measures were compared to classic episodic measures in predicting the progression of MCI to dementia. We estimated a different effect of the PVF/CFT discrepancy score during the disease. According to the cognitive progression of the AD, while a decline in CFT performance is expected at the first stages of the disease, the decline in PVF is expected in the later stages. Based on these grounds, we expected that while CFT and PVF performances will be comparable in healthy participants (expected to obtain normal scores on both tests) and in AD (expected to obtain abnormal or borderline scores on both tests), a discrepancy between performances (CFT impaired and PVF still preserved) will be observable only in MCI patients. Our prediction was that this discrepancy would be even more evident in MCI patients that will progress to dementia compared to those that remain stable over time. Our hypothesis was that a prevalence of isolated SM disorders is expected to affect only MCI patients at risk of progression providing an early neuropsychological marker of AD in clinical practice.

MATERIALS AND METHODS

Participants

Participants were enrolled among those referred to the Neuropsychology Unit—Memory Clinic at the Policlinico A. Gemelli in Rome for cognitive disturbances.

The diagnosis of AD dementia was formulated based on current clinical criteria [25]; the sample was subdivided into mild (Mini-Mental State Examination [MMSE] score 18–23) and moderate (MMSE score 10–17) AD.

Individuals labelled as aMCI had to be impaired on EM tests in a single or in multiple domains (after correction for age and education) according to the Italian normative studies on the Rey’s Auditory Verbal Learning Test (RAVLT) [26] and delayed recall of Rey’s Osterrieth Complex Figure (ROCF) [27]. Exclusion criteria were as follows: co-occurrence of central nervous system diseases other than AD or MCI (e.g., other neurodegenerative disorders, cerebrovascular accidents, tumors, traumatic injury); major psychiatric disorders (e.g., bipolar disorder, major depressive disorder, etc.); substance abuse; medical conditions potentially able to interfere with cognitive functions (e.g., renal or hepatic failure; vitamin B12 deficiency, etc.). The assessment of efficiency in activities of daily living was based on narrative reports and the administration of the Activities of Daily Living [28] and Instrumental Activities of Daily Living [29] scales to the main informant (generally the spouse or children). All enrolled aMCI individuals obtained a Clinical Dementia Rating (CDR) score of 0.5. All aMCI patients underwent brain MRI and or FDG-PET examination confirming neurodegenerative pathology and excluding other pathologies. According to the clinical, neuropsychological, and imaging investigations, all MCI patients met the criteria for MCI due to AD at an intermediate level of certainty [30]. Healthy participants (HP) were recruited among relatives of individuals referred to our memory clinic. All of them had no report of any cognitive disturbances, scored normally on the MMSE [31], and satisfied all of the above-mentioned exclusion criteria.

All participants gave their informed consent to participate in the study following the requirements set by the Ethics Committee of the Catholic University.

The sample was composed of 805 participants, subdivided into 262 HP, with a mean age of 68.1 years (standard deviation [SD] = 11.62) and mean education of 8.4 years (SD = 4.43); 251 participants with aMCI, with a mean age of 72.5 years (SD = 7.29) and mean education of 10.9 years (SD = 4.64); 292 AD patients, subdivided in mild AD (MMSE: 18–23; 178 patients; mean age 72.7 years; SD = 8.53; mean education 9.9 years; SD = 4.65) and moderate AD (MMSE: 10–17; 114 patients; mean age 73.9; SD = 6.73; mean education 7.3 years; SD = 4.69).

Assessment of phonological and semantic verbal fluency

Each participant completed the PVF and CFT tasks. The PVF [26] task required participants to generate as many words as possible starting with a given letter (F, A, S) in one minute (three minutes in total). There were no constraints on grammar classes (i.e., nouns, verbs, adverbs, and adjectives) or the use of compound words.

The CFT tasks required participants to generate nouns belonging to the categories “birds” and “pieces of furniture” [22]. Participants had one minute per category to produce as many words as possible.

Mean scores on PVF were 25.8 (SD = 12.08) in HP; 27.0 (SD = 9.78) in aMCI; 18.8 (SD = 10.07) in mild AD; and 10.0 (SD = 7.87) in moderate AD.

Mean scores on CFT were 16.1 (SD = 4.60) in HP; 13.3 (SD = 4.08) in aMCI; 9.9 (SD = 4.76); and 5.7 (SD = 4.45) in moderate AD.

To assess the specific contribution of a lexical–semantic deficit to the word retrieval difficulties that can be observed in MCI and AD, we computed the difference between CFT and PVF after adjusting (dividing) for the number of stimuli for each task. Such value was defined as ‘Semantic-Phonological Delta’ (SPD) and computed as follows: $SPD = \frac{CFT score}{2} - \frac{PVF score}{3}$

Negative value corresponds to worse semantic fluency output as compared to phonological fluency output.

Neuropsychological assessment and follow-up of participants with aMCI

Each aMCI participant was assessed with a comprehensive neuropsychological battery including, alongside the MMSE and the CFT and PVF, the following tests: RAVLT [26]; copy of figures with and without landmarks [26]; Raven’s Coloured Progressive Matrices [26]; Stroop test [32]; digit span forward and backward [33]; denomination of nouns and verbs [34]; and the Multiple Features Targets Cancellation (MFTC) [35]. During the five years, follow-up participants with MCI were assessed every six months, undergoing a complete neurological and medical examination and a neuropsychological assessment. At each follow-up visit, the progression to dementia was assessed by two neurologists who were blind to the results of the baseline neuropsychological examination. The diagnosis of dementia was formulated if the clinical criteria for dementia due to AD [25] were satisfied, and participants obtained a CDR score of 1. The diagnosis of AD was also confirmed in a following follow-up visit after six months.

Statistics

The effect of Group (HP, aMCI, mild AD, and moderate AD) on CFT, PVF, and SPD were assessed using a multivariate general linear model (GLM) controlled for age and education. Estimated marginal means were used to perform pairwise comparisons.

As for the assessment of the rate of progression to AD in MCI participants, Cox proportional hazard risk models were developed in which SPD entered as a possible predictor, alongside age, education, MMSE and neuropsychological scores associated with a p level lower than 0.1 in univariate analyses.

RESULTS

Comparison of semantic–phonological delta among groups

Multivariate tests of the GLM showed a significant effect of age, education, and group (Wilk’s lambda = 0.950, F_2,798 = 20.816, p < 0.001; Wilk’s lambda = 0.779, F_2,798 = 112.880, p < 0.001; Pillai’s trace = 0.591, F_6,1596 = 80.115, p < 0.001, respectively).

On between subjects tests, there was an effect of age on CFT (F = 41.606; p < 0.001) and PVF (F = 8.397; p = 004) but not on SPD (F = 2.598; p = 0.107). Education showed a significant effect on CFT (F = 15.504; p < 0.001), PVF (F = 220.232; p < 0.001) and SPD (F = 162.930; p < 0.001). Finally, a significant effect was found for group on CFT (F = 161.157; p = 0.001), PVF (F = 81.763; p < 0.001) and SPD (F = 4.398; p = 0.004).

Table 1 displays the distribution of estimated marginal means among the groups. Pairwise comparisons showed that PVF scores were significantly different between moderate AD and mild AD (p < 0.001), MCI (p < 0.001), and HP (p < 0.001); between mild AD and both MCI (p < 0.001) and HP (p < 0.001); no difference was found between HP and MCI (p = 0.215). As for CFT, there was a statistically significant difference between moderate AD and mild AD (p < 0.001), MCI (p < 0.001), and HP (p < 0.001); between mild AD and both MCI (p < 0.001) and HP (p < 0.001); between HP and MCI (p < 0.001). The estimated marginal mean of SPD in MCI patients was significantly lower than in HP (p < 0.001), mild AD (p = 0.020) and moderate AD (p = 0.017); no differences were found in the pairwise comparisons between HP and mild AD (p = 0.370), HP and moderate AD (p = 0.732), and moderate AD and mild AD (p = 0.679). Finally, we repeated this last comparison after splitting the MCI group in stable or converters to AD dementia (Fig. 1). The estimated marginal mean of SPD in aMCI converters was significantly lower than in HP (–2.691±0.288 versus –0.827±0.198; p < 0.001, respectively), mild AD (1.106±0.234; p < 0.001) and moderate AD (–0.952±0.294; p < 0.001). The SPD in aMCI converters was also lower than the SPD in stable aMCI, but not reaching statistical significance (–1.612±0.326; p = 0.126).

Table 1

Estimated marginal means of Phonological Verbal Fluency (PVF), Category Fluency Task (CFT), and Semantic-Phonological Delta (SPD) after GLM

	HP (n:262)		aMCI (n:251)		mild AD (n:178)		moderate AD (n:114)
	Mean	SE	Mean	SE	Mean	SE	Mean	SE
PVF	26.51^a,b,c	0.582	25.45	0.589	18.37	0.685	12.44	0.866
CFT	15.86	0.267	13.32	0.270	9.91	0.314	6.25	0.397
SPD	–0.91	0.185	–1.83	0.187	–1.16	0.218	–1.02	0.275

^astatistically significant difference (p < 0.05) in comparison with aMCI. ^bstatistical significant difference (p < 0.05) in comparison with mild AD. ^cstatistical significant difference (p < 0.05) in comparison with moderate AD.

Fig. 1

Estimated marginal means on semantic phonologic delta in HP, aMCI stable and converters, mild and moderate AD.

Survival analysis

One-hundred seventy-eight of the initial aMCI sample completed the five-year follow-up; the main reasons for dropping-out were as follows: 25 participants refused or found it difficult to attend visits; 11 participants died during the follow-up period; 30 participants experienced major medical conditions that led to exclusion from the study following enrolment criteria; and 4 participants moved to another town. Three participants with aMCI progressed to other forms of dementia (2 Lewy body dementia and 1 frontotemporal dementia) and were not considered in the analyses. However, we did not find any differences in age, education, or baseline MMSE and neuropsychological scores between participants who were lost at follow-up and participants who completed the follow-up (Table 2).

Table 2

Comparison between the neuropsychological scores of participants who completed the five-year follow-up in comparison with participants who were lost at follow-up

	Lost at follow-up (n:73)		Follow-up (n:178)
	Mean	SD	Mean	SD	\|t\|	p
Age	72.74	7.775	73.06	6.769	0.33	0.744
Education	10.52	4.885	10.63	4.609	0.17	0.868
MMSE	25.92	2.284	25.56	1.977	1.26	0.210
RAVLT: immediate recall	24.23	6.286	24.04	6.214	0.22	0.823
RAVLT: delayed recall	1.78	1.635	1.67	1.525	0.52	0.604
RAVLT: recognition accuracy	0.72	0.116	0.73	0.118	0.43	0.669
Digit span forward	5.38	0.981	5.16	1.116	1.49	0.139
Digit span backward	3.53	0.867	3.63	0.958	0.73	0.468
Raven’s Progressive Colored Matrices	24.12	6.545	24.70	5.180	0.74	0.462
Copy of figures	9.26	1.703	9.44	1.772	0.71	0.480
Copy of figures with landmarks	65.40	5.120	65.45	6.189	0.06	0.956
Stroop’s test: time	58.91	41.740	52.04	29.376	1.43	0.154
Stroop’s test: errors	3.69	6.455	3.29	4.613	0.54	0.593
MFTC: false alarms	0.58	1.039	0.77	2.011	0.75	0.454
MFTC: accuracy	0.88	0.140	0.89	0.139	0.61	0.5407
MFTC: time	78.26	39.550	82.44	34.446	0.80	0.425
Phonological Verbal Fluency	27.73	10.669	26.37	9.367	0.75	0.456
Category Fluency Task	12.34	3.362	13.49	4.011	1.45	0.150

MMSE, Mini-Mental State Examination; RAVLT, Rey’s Auditory Verbal Learning Test; MFTC, Multiple Features Targets Cancellation.

During the follow-up period, 117 participants progressed to AD, with a median survival time of 42 months. Table 3 shows the results of the univariate Cox proportional hazard models in which each neuropsychological test was set individually as a predictor. As shown, the SPD had a predictive effect on progression to dementia (HR = 0.92; 95%CI =0.881–0.968; p = 0.001); the same effect was observed for the CFT (HR = 0.91; 95%CI = 0.859–0.956; p < 0.001), for scores obtained on RAVLT immediate and delayed recall (HR = 0.97, 95%CI =0.936–0.997, p = 0.029 and HR = 0.81; 95%CI =0.707–0.931; p = 0.003, respectively) and for the accuracy on MFTC (HR = 0.13; 95%CI = 0.040–0.419; p = 0.001); the number of errors on the Stroop test was associated with a slight reduction of time to dementia (HR = 1.04; 95%CI = 1.000–1.076; p =0.049).

Table 3

Results of the Cox regression analyses. In bold, the variables significantly associated with a modification of the hazard; in italic, the variables with a significance level of p < 0.1 that entered the multiple variable model

	Univariate analyses				Multivariate analysis
	HR	95%CI		p	HR	95%CI		p
MMSE	0.72	0.652	0.794	< 0.001	0.70	0.620	0.792	0.001
RAVLT: immediate recall	0.97	0.936	0.997	0.029	1.00	0.960	1.038	0.926
RAVLT: delayed recall	0.81	0.707	0.931	0.003	0.87	0.721	1.051	0.148
RAVLT: recognition accuracy	0.29	0.072	1.188	0.085	0.64	0.121	3.369	0.597
Digit span forward	0.98	0.842	1.139	0.782
Digit span backward	0.93	0.773	1.112	0.415
Raven’s Progressive Colored Matrices	0.99	0.952	1.021	0.421
Copy of figures	0.91	0.826	1.002	0.056	1.02	0.898	1.156	0.776
Copy of figures with landmarks	0.98	0.961	1.008	0.190
Stroop’s test: time	1.00	0.998	1.011	0.171
Stroop’s test: errors	1.04	1.000	1.076	0.049	1.03	0.986	1.072	0.197
MFTC: false alarms	0.99	0.898	1.088	0.814
MFTC: accuracy	0.13	0.040	0.419	0.001	0.51	0.075	3.434	0.488
MFTC: time	1.00	0.997	1.009	0.300
Phonological Verbal Fluency	1.00	0.985	1.023	0.720
Category Fluency Task	0.91	0.859	0.956	< 0.000	0.94	0.885	1.004	0.067
SPD	0.92	0.881	0.968	< 0.001	0.92	0.853	0.984	0.016

HR, hazard ratio; MMSE, Mini-Mental State Examination; RAVLT, Rey’s Auditory Verbal Learning Test; MFTC, Multiple Features Targets Cancellation; SPD, Semantic-Phonological Delta.

These variables, alongside with the scores obtained on the RAVLT—accuracy and on the copy of figure (both showing a significance level of p < 0.1), were set as predictors in a multivariate Cox model controlled for age and education. As shown in Table 3 (right columns), together with the MMSE score, the SPD was the only variable able to predict progression to dementia.

To detect the timing of progression to dementia, a survival analysis was carried out with the sample stratified according to the SPD median (–2.33). As shown by the Kaplan–Mayer curve (Fig. 2), participants with SPD below the median value had a shorter time of conversion to dementia than participants with SPD above the mean (median time to conversion: 30 months versus 48 months, respectively).

Fig. 2

Kaplan-Mayer curves of the sample stratified according to SPD median value reflecting the progression to dementia, with the sample stratified according to the SPD median (–2.33).

DISCUSSION

Storage capacity, rate of forgetting and processing of memory are the most extensively researched episodic long-term memory neuropsychological indices that could be used to identify memory disorders due to temporal lobe dysfunction and consequently become early markers of AD [6, 7]. In the last few years, there has been increasing focus on SM as a reliable and early neuropsychological marker of AD (see Venneri et al., 2018 [36] for a review). SM disorders have shown correlations with brain amyloid load and lower cortical thickness, supporting the hypothesis that SM could be a reliable cognitive marker of initial neurodegeneration and may serve as an early and distinguishing domain for preclinical AD [36 –38].

In our study, we have examined in a sample of HP, aMCI, and mild to moderate AD patients the potential role of SM as an early predictor of progression to AD dementia. As expected, our results have shown that in mild and moderate AD, both CFT and PVF score were significantly worse than in HP. At the variance, MCI performed like HP on the PVF task but were significantly worse on the CFT task. This discrepancy is even more highlighted by the discrepancy score (SPD). The SPD was conceived as an indicator of semantic impairment, in which the weight of executive functions connected with word generation could be smoothed. The score was lower in the aMCI patients than in HP and patients with mild to moderate AD, and it was even lower in the MCI group that progressed to AD, showing that this index has a good predictive value for progression to AD.

This result supports the hypothesis of an early decline in SM that contrasts with the long-lasting preservation of verbal initiation as explored by a letter fluency task. The lack of any significance between letter fluency and progression of the disease indicates that it is the semantic processing, and not any generic lexical processing, that characterizes the early stages of neurodegeneration in AD, and this effect is highlighted using the SPD. The effect of the SPD has already been reported [19 –23], but how much this marker is useful for diagnostic purposes and how it can have a different value during the progression of the disease has not yet been explored.

Our results show that this index is mainly detectable in an early stage of the MCI condition since it tends to disappear when the neurodegeneration progresses to an overt AD dementia condition, losing its sensitivity when the MCI severity approaches progression to dementia.

In our sample, we studied patients with a long follow-up and examined patients’ conversion from 1 to 5 years. In addition to long-term memory, in the univariate Cox regression analyses, several variables showed predictive values for progression, mainly involving executive functions (MFTC accuracy and Stroop test Interference score) and semantic attainment (SPD and CFT). When we considered a multivariate hazard ratio, only MMSE and SPD maintained a significant value, predicting the risk of progression to AD dementia.

In our view, the SPD, as demonstrated by the comparisons between HP and AD patients, is a rather early marker in the course of the disease and could have more relevance in identifying MCI individuals who will progress to dementia after the first two to three years of follow-ups. This observation is consistent with several studies that examined the vocabulary of the participants to the ‘Nun study’ or the number of words used by some famous writers affected by AD up to 10 years before the disease became apparent [39]. All these studies found an early reduction of semantic competence and in the words utilized in procedural speaking among individuals that will develop AD.

In the preclinical phase, corresponding to Braak stage I [40], AD neurofibrillary pathology appears in the lateral transentorhinal region of the perirhinal cortex (including the mPRC), and only at a later stage AD pathology spreads to the hippocampus. Sub-hippocampal structures, in particular perirhinal cortex (PRC), are involved in semantic memory. PRC and entorhinal cortices (ERC) are affected very early by tau pathology in the transentorhinal stage of AD [40], even before the involvement of hippocampus (limbic stage). According to this interpretation, our group has recently reported that patients with MCI who eventually converted to dementia displayed a reduced level of semantic proximity between items produced during the CFT task [41], which may be a proxy for impaired ability to retrieve shared features among items belonging to a semantic category. These deficits are rarely reported by patients and caregivers since the reduction of vocabulary and difficulties in word-finding in procedural speaking produce far less dramatic consequences in everyday life than episodic memory failures.

The lack of significance of EM in predicting progression to dementia in our sample deserves some further consideration. One main explanation could be due to the rather long follow-up period of our study. This condition could have highlighted the predictive effect of the performance on SM tasks that is already impaired when EM tasks are only slightly involved. Thus, the results of the survival analyses are no more surprising, confirm these hypotheses, and are clearer after these considerations. Furthermore, our data are consistent with those of Papp et al. [38], who showed that the main effect of the CFT tasks in predicting progression from MCI to overt AD dementia is observed four years before the time of conversion.

The study has several limitations. Data were gathered in a clinical setting by different raters that could have partially interfered with the omogeneity of data recording also if CFT and PVF tasks have a high interrater reliability. Most important, MCI patients were diagnosed as MCI due to AD at an intermediate level of certainty and are lacking biomarker data. However, the rather long follow-up period and the repeated assessments reduce the possibility of inaccurate identification of MCI individuals both at the baseline and at the follow-ups. Moreover, considering the independent control made on the outcome of the MCI patients in our sample, we could not have included incorrectly identified patients both at the baseline and at the final follow-up.

In conclusion, our data support the hypothesis that the investigation of SM using score extracted by fluency tests may represent a proxy for early AD changes independent of episodic memory. Specifically, the SPD score might also prove to be useful in the preclinical stage of AD when EM impairment is still not detectable by neuropsychological tests. An extensive and prospective study in a sample of individuals with subjective cognitive complaints could help to support the usefulness of SPD in the early stages of the disease.

On the other hand, considering the biological definition of AD, recently conceptualized as a continuum [42], an operational definition of the disease using CFT and SPD scores could also be of some interest in preclinical stages of AD in identifying in advance individuals who will progress to prodromal AD in order to select individuals for biomarkers investigation, and to plan earlier prevention strategies and pharmacological intervention.

DISCLOSURE STATEMENT

Authors’ disclosures available online (https://www.j-alz.com/manuscript-disclosures/20-1452r1).

References

Cohen

, Squire

(1980) Preserved learning and retention of pattern-analyzing skill in amnesia: Dissociation of knowing how and knowing that. Science 210, 207–210.

Hort

, O’Brien

, Gainotti

, Pirttila

, Popescu

, Rektorova

, Sorbi

, Scheltens

, EFNS Scientist Panel on Dementia (2010) EFNS guidelines for the diagnosis and management of Alzheimer’s disease. Eur J Neurol 17, 1236–1248.

Chapman

, Mapstone

, McCrary

, Gardner

, Porsteinsson

, Sandoval

, Guillily

, Degrush

, Reilly

(2011) Predicting conversion from mild cognitive impairment to Alzheimer’s disease using neuropsychological tests and multivariate methods. J Clin Exp Neuropsychol 33, 187–199.

Rabin

, Paré

, Saykin

, Brown

, Wishart

, Flashman

, Santulli

(2009) Differential memory test sensitivity for diagnosing amnestic mild cognitive impairment and predicting conversion to Alzheimer’s disease. Neuropsychol Dev Cogn B Aging Neuropsychol Cogn 16, 357–376.

Marra

, Gainotti

, Fadda

, Perri

, Lacidogna

, Scaricamazza

, Piccininni

, Quaranta

(2015) Usefulness of an integrated analysis of different memory tasks to predict the progression from mild cognitive impairment to Alzheimer’s disease: The episodic memory score (EMS). J Alzheimers Dis 50, 61–70.

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.

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.

Venneri

, Mitolo

, Beltrachini

, Varma

, Della Pietà

, Jahn-Carta

, Frangi

, De Marco

(2019) Beyond episodic memory: Semantic processing as independent predictor of hippocampal/perirhinal volume in aging and mild cognitive impairment due to Alzheimer’s disease. Neuropsychology 33, 523–533.

Monsch

, Bondi

, Butters

, Salmon

, Katzman

, Thal

(1992) Comparisons of verbal fluency tasks in the detection of dementia of the Alzheimer type. Arch Neurol 49, 1253–1258.

10.

Adlam

A-LR

, Bozeat

, Arnold

, Watson

, Hodges

(2006) Semantic knowledge in mild cognitive impairment and mild Alzheimer’s disease. Cortex 42, 675–684.

11.

Huth

, de Heer

, Griffiths

, Theunissen

, Gallant

(2016) Natural speech reveals the semantic maps that tile human cerebral cortex. Nature 532, 453–458.

12.

Taylor

, Probst

(2008) Anatomic localization of the transentorhinal region of the perirhinal cortex. Neurobiol Aging 29, 1591–1596.

13.

Tyler

, Stamatakis

, Bright

, Acres

, Abdallah

, Rodd

, Moss

(2004) Processing objects at different levels of specificity. J Cogn Neurosci 16, 351–362.

14.

Taylor

, Moss

, Stamatakis

, Tyler

(2006) Binding crossmodal object features in perirhinal cortex. Proc Natl Acad Sci U S A 103, 8239–8244.

15.

Davies

, Graham

, Xuereb

, Williams

, Hodges

(2004) The human perirhinal cortex and semantic memory. Eur J Neurosci 20, 2441–2446.

16.

Kivisaari

, Tyler

, Monsch

, Taylor

(2012) Medial perirhinal cortex disambiguates confusable objects. Brain J Neurol 135, 3757–3769.

17.

Barbeau

, Didic

, Tramoni

, Felician

, Joubert

, Sontheimer

, Ceccaldi

, Poncet

(2004) Evaluation of visual recognition memory in MCI patients. Neurology 62, 1317–1322.

18.

Verhaeghen

(2003) Aging and vocabulary scores: A meta-analysis. Psychol Aging 18, 332–339.

19.

Vaughan

, Coen

, Kenny

, Lawlor

(2018) Semantic and phonemic verbal fluency discrepancy in mild cognitive impairment: Potential predictor of progression to Alzheimer’s disease. J Am Geriatr Soc 66, 755–759.

20.

Hodges

, Patterson

, Ward

, Garrard

, Bak

, Perry

, Gregory

(1999) The differentiation of semantic dementia and frontal lobe dementia (temporal and frontal variants of frontotemporal dementia) from early Alzheimer’s disease: A comparative neuropsychological study. Neuropsychology 13, 31–40.

21.

Dudas

, Clague

, Thompson

, Graham

, Hodges

(2005) Episodic and semantic memory in mild cognitive impairment. Neuropsychologia 43, 1266–1276.

22.

Quaranta

, Caprara

, Piccininni

, Vita

, Gainotti

, Marra

(2016) Standardization, clinical validation, and typicality norms of a new test assessing semantic verbal fluency. Arch Clin Neuropsychol 31, 434–445.

23.

Chasles

M-J

, Tremblay

, Escudier

, Lajeunesse

, Benoit

, Langlois

, Joubert

, Rouleau

(2019) An examination of semantic impairment in amnestic MCI and AD: What can we learn from verbal fluency? Arch Clin Neuropsychol 35, 22–30.

24.

Taler

, Phillips

(2008) Language performance in Alzheimer’s disease and mild cognitive impairment: A comparative review. J Clin Exp Neuropsychol 30, 501–556.

25.

McKhann

, Knopman

, Chertkow

, Hyman

, Jack

, Kawas

, Klunk

, Koroshetz

, Manly

, Mayeux

, others (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.

26.

Carlesimo

, Caltagirone

, Gainotti

, Fadda

, Gallassi

, Lorusso

, Marfia

, Marra

, Nocentini

, Parnetti

(1996) The mental deterioration battery: Normative data, diagnostic reliability and qualitative analyses of cognitive impairment. Eur Neurol 36, 378–384.

27.

Caffarra

, Vezzadini

, Dieci

, Zonato

, Venneri

(2002) Rey-Osterrieth complex figure: Normative values in an Italian population sample. Neurol Sci 22, 443–447.

28.

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.

29.

Lawton

, Brody

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

30.

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.

31.

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.

32.

Caffarra

, Vezzadini

, Dieci

, Zonato

, Venneri

(2002) Una versione abbreviata del test di Stroop: Dati normativi nella popolazione italiana. Nuova Riv Neurol 12, 111–115.

33.

Monaco

, Costa

, Caltagirone

, Carlesimo

(2013) Forward and backward span for verbal and visuo-spatial data: Standardization and normative data from an Italian adult population. Neurol Sci 34, 749–754.

34.

Miceli

, Laudanna

, Burani

, Capasso

(1994) Batteria per l’Analisi del Deficit Afasico. BADA [BADA A battery for the assessment of aphasic disorders].

35.

Marra

, Gainotti

, Scaricamazza

, Piccininni

, Ferraccioli

, Quaranta

(2013) The Multiple Features Target Cancellation (MFTC): An attentional visual conjunction search test. Normative values for the Italian population. Neurol Sci 34, 173–180.

36.

Venneri

, Jahn-Carta

, de Marco

, Quaranta

, Marra

(2018) Diagnostic and prognostic role of semantic processing in preclinical Alzheimer’s disease. Biomark Med 12, 637–651.

37.

Mirandez

, Aprahamian

, Talib

, Forlenza

, Radanovic

(2017) Multiple category verbal fluency in mild cognitive impairment and correlation with CSF biomarkers for Alzheimer’s disease. Int Psychogeriatr 29, 949–958.

38.

Papp

, Mormino

, Amariglio

, Munro

, Dagley

, Schultz

, Johnson

, Sperling

, Rentz

(2016) Biomarker validation of a decline in semantic processing in preclinical Alzheimer’s disease. Neuropsychology 30, 624–630.

39.

Garrard

, Maloney

, Hodges

, Patterson

(2005) The effects of very early Alzheimer’s disease on the characteristics of writing by a renowned author. Brain 128, 250–260.

40.

Braak

, Alafuzoff

, Arzberger

, Kretzschmar

, Del Tredici

(2006) Staging of Alzheimer disease-associated neurofibrillary pathology using paraffin sections and immunocytochemistry. Acta Neuropathol (Berl) 112, 389–404.

41.

Quaranta

, Piccininni

, Caprara

, Malandrino

, Gainotti

, Marra

(2019) Semantic relations in a categorical verbal fluency test: An exploratory investigation in mild cognitive impairment. Front Psychol 10, 2797.

42.

Jack

, Bennett

, Blennow

, Carrillo

, Dunn

, Haeberlein

, Holtzman

, Jagust

, Jessen

, Karlawish

, Liu

, Molinuevo

, Montine

, Phelps

, Rankin

, Rowe

, Scheltens

, Siemers

, Snyder

, Sperling

, Contributors (2018) NIA-AA Research Framework: Toward a biological definition of Alzheimer’s disease. Alzheimers Dement 14, 535–562.