Clinical Heterogeneity in Alzheimer’s Disease: A Possible New Amnesic Phenotype

Is the CM-phenotype really a syndrome?

In two previous articles, we described a small group of patients diagnosed with biomarker-supported AD who had a common set of cognitive and behavioral features that we termed the CM-phenotype [1, 2]. These data are not sufficient to determine whether the CM-phenotype should be considered a variant of AD, rather than simply a description of a set of patients with AD who differ slightly from the typical phenotype. There are at least three reasons for this null result.

First, to support that the CM-phenotype can be described as a syndrome it is necessary that symptoms occur in association, as we described in previous works [1, 2]. Coherently, some studies found that confabulations in AD are often associated with delusions [8, 9] or delusions and aggression [10], consistent with the symptoms’ cluster of the CM phenotype. Moreover, some studies on patients with AD who show psychotic symptoms, by applying factor and cluster analyses to behavioral data, identified a misidentification subtype that is well distinguished from a paranoid subtype and a nonpsychotic subtype [11, 12]. The misidentification subtype is characterized by misidentifications, misidentification delusions, and hallucinations exactly as in the case of CM-phenotype [11, 12]. Finally, the association of the CM-phenotype symptoms was well known in the early 1900s as presbyophrenia [3–5].

Second, to support the existence of the CM-phenotype it is necessary that it is clearly distinguished from the classical phenotype and does not show only small variations from it. In this regard, the most salient symptoms of the CM-phenotype, i.e., confabulation and misidentification, are also considered relatively common in typical AD. This overlap may undermine the existence of the CM-phenotype as a distinct syndrome. However, especially considering confabulation there are important differences in the manifestation of this symptom between CM- and CA-phenotype, concerning the time of onset, the type of confabulation, and their content. Indeed, confabulations in classical AD become more frequent with the progression of dementia [13], from infrequent at the early stages [13–15] to relatively more frequent at the advanced stages [13, 17]. Furthermore, at the early stages of classical AD confabulations are almost exclusively provoked [18], whereas spontaneous confabulations [18] occur predominantly or exclusively at the advanced stages of the disease [19]. Finally, confabulations at the early stages of classical AD are almost simple intrusions [20] into memory test content, e.g., production of unstudied words in word-list recall, or habits [21], i.e., personal semantic contents recalled in place of forgotten episodic information. By contrast, confabulations in the CM-phenotype are very frequent at the early stages of dementia [2] and in some cases are the first symptom of the whole clinical picture [1]. More importantly, at the early stages of CM-phenotype mainly spontaneous confabulations appear in addition to those provoked on testing [1, 2]. Finally, considering the content of confabulations, it consists of not only habits, but also and especially misplacements, memory fabrications and confusions [1].

Even when considering misidentifications, we found differences between classical AD and CM-AD [2]. First, our data show that misidentifications regardless of the stage of onset, are more frequent in CM- (59%) [2] than in CA-AD (a median prevalence among different studies of 25.6%) [12, 22–26]. Furthermore, in classical AD, similar to confabulations, misidentifications become increasingly frequent when progressing from early to more advanced stages of dementia [24, 27]. Conversely, misidentification in the CM-phenotype is a frequent symptom already at mild dementia stages (41.2% in our sample) [2] and is also present at the MCI stage (23.5% in our sample) [2]. Finally, in one patient misidentification was the first symptom in the whole clinical picture [2].

In addition, in a previous study in which we compared patients with CM- and CA-phenotypes, we found multiple differences in further cognitive and behavioral aspects (see Table 2), that seemed to be maintained even at follow-up points [2].

Third, to sustain the hypothesis of a new dementia syndrome, it is mandatory to prove that it has an underlying specific neuroanatomical basis. In this regard, presbyophrenia was found to be very similar to Korsakoff syndrome (KS) [3–5]. KS is considered the prototype of diencephalic amnesia and has a precise neuroanatomical correlate that is clearly distinguishable from the hippocampal involvement of classic AD [28, 29]. Hence, the similarity between presbyophrenia and KS suggests that the CM-phenotype has specific anatomical features and is a peculiar presentation of AD.

Next steps to substantiate diencephalic variant of AD

It should be emphasized that we were able to identify the CM-phenotype mainly by adopting a thorough clinical examination developed by our group, by taking a detailed patient’s clinical history, and by observing signs and symptoms of cognitive dysfunction during the interview and visit [30, 31]. Conversely, the standard battery of neuropsychological tests was found to be unable to bring out the phenotype. Thus, the observational examination of cognitive and behavioral features seems crucial for early diagnosis of the CM-phenotype. In this regard, the detection of spontaneous confabulations can be difficult, as they are often entirely plausible and not bizarre accounts [1]. Therefore, it is essential that the interview with the relative includes specific questions about confabulations and/or that specific tests are used to induce confabulations in patients [32, 33]. Moreover, the interview with the patient should preliminarily investigate retrograde memory (e.g., questions about medical history, job, family, etc.) to stimulate the appearance of confabulation. In addition, retrograde memory deficits are one of the main features of amnesia in the CM-phenotype. Based on the evidence that in some patients with CM phenotype, spontaneous confabulation preceded the onset of recent memory disorder [1], it is particularly important to investigate the occurrence of confabulation in addition to memory disorder to early detect CM-AD. Interestingly, the finding that confabulations precede amnesia is consistent with the idea that memory must be at least partially unimpaired, and hippocampus preserved for spontaneous confabulations to occur [34, 35]. Moreover, considering the relationship between psychotic symptoms and cholinergic deficits in AD [36–39], the early detection of CM-AD might lead CMs to benefit more significantly than other ADs from early therapy with cholinesteraseinhibitors.

The need of deep clinical examination seems crucial not only for early diagnosis of the CM-phenotype in clinic, but also for its study in research. Indeed, both confabulation and misidentification are not frequently assessed in a standardized neuropsychological test battery, which makes it hard to retrospectively identify the CM-phenotype in existing datasets. This, in turn, severely hamper the scientific community to contribute to a conceptualization of this new phenotype.

The call for extensive clinical studies to recognize CM-phenotype is even more urgent considering the fact that in our case series CM-phenotype appears to be much more frequent than the other known variants of AD. Specifically, we retrieved 50 patients with early-stage AD from the registry of AD outpatients with positive biomarkers in the CSF enrolled consecutively over a 10-year interval in our Geriatrics unit [2]. Among them, 30 (60%) had a classical amnesic phenotype, 17 (34%) a CM-phenotype, 1 (2%) a logopenic variant, 1 (2%) a corticobasal syndrome-AD, and 1 (2%) a PCA-syndrome [2]. However, this result should be considered with caution, as the CM-phenotype estimate is probably amplified in our sample. In fact, in our Geriatrics unit we perform few spinal taps due to the advanced age of patients, and cases with particular symptoms are more likely than classic AD to perform the examination.

Taken together, all these considerations inform us that one of the main goals of future research is to confirm the occurrence of the CM-phenotype in a larger sample of patients. Actually, we found in the scientific literature a recent study that presents patients with AD who show a phenotype very similar to the CM-phenotype, but it is only two cases [40]. Furthermore, since we (re)discovered the CM-phenotype by studying elderly patients in a geriatric setting, the cases we observed were exclusively late-onset AD (LOAD) due to selection bias. Thus, it remains to be evaluated whether the same phenotype also occurs in early-onset AD (EOAD). In addition, our description of the CM phenotype is mainly based on clinical observations collected at the first visit and lacks data from comprehensive longitudinal studies to be confirmed [2].

Regarding the neuroanatomical substrate, we assessed atrophy only by visual rating scales on MRI images [2, 42], while the FDG-PET imaging was performed only on a subset of participants. Thus, a clearer characterization of the neuroanatomical network underlying the syndrome is still lacking. Our preliminary data of possible greater right hemispheric involvement in some anterior brain regions (i.e., frontoinsular), more symmetrical involvement at the temporal level, and less posterior involvement (i.e., dorso-parietal and precuneous) than in CA-AD could open the way for further studies. Moreover, based on previous evidence [7, 43], we speculated that a limbic-diencephalic network (involving thalamic nuclei, mammillary bodies, medial and posterior orbitofrontal cortex, hippocampus, parahippocampal gyrus, transentorhinal cortex, and amygdala) might be the neuroanatomical substrate of the CM phenotype [2]. Interestingly, this hypothesized network is consistent with a well-known model of the limbic system [44].

Our study of biomarkers in CSF strongly supported an etiologic diagnosis of AD in patients presenting with the CM-phenotype [2]. However, considering that the CM-phenotype resembles DLB syndrome, it might be crucial to study the possible association of Lewy bodies pathology with AD in the CM phenotype. Furthermore, given the known affinity of TPD-43 pathology for limbic structures (amygdala, hippocampus, middle frontal gyrus) in the elderly (limbic-predominant age-related TDP-43 encephalopathy, LATE) [45] and the probable limbic involvement in the CM-phenotype, it seems noteworthy to study the possible contribution of TDP-43 pathology to CM-AD. In addition, given the known propensity of APOE ɛ4 for medial temporal lobe damage [46], it seems important to replicate our preliminary data of a higher frequency of the ɛ4 genotype in CM- than in CA-ADs. This finding is in line with the lesser posterior atrophy and the more frequent and symmetrical temporo-mesial hypometabolism found in CM-ADs [2]. Moreover, it is consistent with previous studies that found a correlation between the number of APOE ɛ4 alleles and the occurrence of delusions in AD [47–51]. Establishing the higher frequency of the ɛ4 genotype in CM-ADs might be crucial also because the use of new anti-Aβ antibody drugs for AD might result in a greater danger for cerebrovascular adverse events (ARIAs) in CM-AD [52]. In this case, this would be another important reason to make an early diagnosis of CM-AD. Finally, the probable involvement of the limbic system in the CM-phenotype suggests the importance of studying the possible association especially with the neuropathological substrate of AD known as limbic predominant [53–55].

Given the close overlap between misidentification subtype [11, 57] and features of CM-phenotype, a further important question for future research is whether CM-phenotype and misidentification subtype are actually the same syndrome. Some data seem to support this interpretation. Regarding symptoms, consistent with the CM-phenotype, a study on misidentification subtype pointed out that patients showed a memory disorder from the onset [56]. In another study, the authors reported that misidentifications were associated with content-related confabulations and delusional memories and that this association entailed an underlying failure of episodic retrograde memory (i.e., memory for past personal events) [58].

Regarding anatomical aspects, our studies on CM-phenotype [1, 2] did not allow a valid comparison with studies on misidentification subtype [12] as we only did a preliminary assessment of atrophy using visual rating scales in a small subgroup of participants [2]. However, the marked mesial and anterior temporal involvement in the CM-phenotype revealed by imaging data [2] seems to coincide with the involvement of limbic and paralimbic regions (especially the hippocampus and its projections such as the parahippocampal gyrus and the transentorhinal cortex) found in the misidentification subtype [12, 58–65]. Furthermore, our studies [2] found preliminary hints of possible greater right hemispheric involvement in some regions (i.e., right fronto-insular). These findings are in line with data about preferential right hemispheric involvement in the misidentification subtype [12, 66–69]. Finally, despite our imaging studies [2] did not show marked anterior involvement as found in studies on misidentification subtype [12, 69], multiple signs from clinical examination of the CM-phenotype (i.e., confabulations, logorrhea, euphoria/fatuity, mild social dishnibition, reduction of insight, executive deficits, attention deficits,) are consistent with frontal and fronto-parietal involvement [1, 2].

Despite the commonalities, misidentification subtype and CM-phenotype differ according to some aspects. In particular, patients with misidentification subtype, but not patients with the CM-phenotype (see Table 2) [2], show more severe global cognitive impairment and also faster rate of cognitive decline than the other subtypes [12, 70]. Furthermore, studies on the misidentification subtype show deficits in visuoperceptive and visuospatial tasks [56, 65] and signs of involvement of right posterior temporo-occipital regions corresponding to the ventral visual pathway [62, 69]. By contrast, in our CM-phenotype studies [2], compared to data on CA-ADs, there is no evidence neither of preminent visuoperceptive and visuospatial disorders, nor signs of involvement of posterior brain regions (see Table 2). However, results on differences between CM-phenotype and misidentification subtype should be considered with caution. Indeed, despite few exceptions [58, 70], most studies on misidentification subtype include patients at more advanced stages of dementia compared to CM-phenotype patients we included in our studies [1, 2].

An intriguing possibility is that the CM-phenotype and the misidentification subtype are the same syndrome but studied at different times in the progression of dementia. Thus, differences in some aspects between the two phenotypes could be explained by the changes that a single dementia syndrome undergoes over time during its evolution. Following this hypothesis, we can speculate that an initial limbic-paralimbic and only partially frontal, or thalamus-frontal involvement, associated with the appearance of amnesia and spontaneous confabulations, is followed by a more pronounced frontal and right-sided involvement, associated with the appearance of executive deficits and delusions including misidentifications. Then, at later stages of dementia progression, consistent with connections between frontal and parietal regions, and between hippocampus/parahippocampus and the ventral visual pathway [62], there could be the involvement of more posterior, fronto-parietal or fronto-parieto-occipital, and temporal-occipital regions, associated with the appearance of visuospatial and visuoperceptive disturbances.