Differential Impairment of Cognitive and Affective Mentalizing Abilities in Neurodegenerative Dementias: Evidence from Behavioral Variant of Frontotemporal Dementia,Alzheimer’s Disease,and Mild Cognitive Impairment

INTRODUCTION

Theory of mind (ToM) has been classically described as the process by which “an individual imputes mental states to himself and others” [1]. It is widely recognized as a multidimensional process [2, 3] requiring the integration of several components. Among them, the ability to attribute emotion (EA) and intention (IA) to others plays a key role in the mentalizing construct [4].

The distinction between affective (i.e., EA) and cognitive (i.e., IA) facets of ToM has been assessed using different tests, with respect to both their cognitive (e.g., reasoning about belief) or affective (e.g., reasoning about feelings) demands. First- and second-order false-beliefs, generally used to assess cognitive ToM ability [5, 6], differ in difficulty, as second-order false-belief tasks require high-level ToM skills [5]. On the contrary, affective ToM is classically investigated with tasks such as the Reading the Mind in the Eyes or the Yoni, which require subjects to mentalize based on eye gaze or facial expression [7–9].

Functional MRI studies on healthy subjects have identified brain correlates of affective and cognitive ToM [10–12], highlighting the engagement of both common and differential brain networks in the attribution of intentions and emotions. In particular, posterior temporo-parietal regions (e.g., temporo-parietal junction, posterior superior temporal sulcus, and precuneus) are key components of mentalizing networks [13, 14], while fronto-limbic regions (e.g., ventromedial prefrontal cortex [11], amygdala [12], inferior frontal gyrus [15, 16], and anterior cingulate cortex [12, 15]) are additionally engaged in tasks requiring inference on other’s mental state based on affective cues.

Focal neurological disorders [9, 17] and neurodegenerative conditions (see [18] for a review) may affect mentalizing abilities with specific and differential patterns of deficits, according to the topographical distribution of brain damage. In particular, affective mentalizing deficits have been reported in patients with frontal brain lesions due to the selective damage of the inferior frontal gyrus [16] and the ventromedial prefrontal cortex [9, 17]. Specific involvement of affective ToM has also been reported in amyotrophic lateral sclerosis [19] in association to gray-matter reduction in the inferior frontal gyrus and anterior cingulate cortex.

More widespread ToM deficits may be present in other neurodegenerative diseases such as Alzheimer’s disease (AD), primary progressive aphasias or progressive supranuclear palsy [18], Parkinson’s disease [20], as well as in some psychiatric conditions as major depression [21] and schizophrenia [22].

Due to a highly selective pathology-driven disruption of the structures engaged in the ToM-related brain networks [23], mentalizing impairments are among the main features of the behavioral variant of the frontotemporal dementia (bvFTD), the second most common early-onset dementia [24, 25]. The clinical picture of bvFTD is usually characterized by social cognition disorders, particularly loss of empathy and emotion recognition deficits, associated with progressive and insidious behavioral alteration and executive disorders [26, 27].

Many studies explored both intention and emotion attribution deficits in bvFTD patients. Overall, these studies highlighted a widespread deficit in both affective and cognitive ToM [18, 28–30], but EA and IA have been often assessed using different tasks [18, 29], making the results open to alternative interpretation (e.g., task difficulty).

Since impairments in basic cognitive abilities (e.g., executive functioning deficits of bvFTD patients [31]) may affect ToM performances, some paradigms included a control condition that matches the ToM conditions in the general cognitive demands, but can be solved without any mentalistic inference. The control condition performance may thus help in elucidating whether task impairments reflect pure ToM deficits or mirror impairments on other cognitive abilities (e.g., executive functioning, working memory, visuo-spatial abilities). Though evidence of single bvFTD studies are controversial, ranging from selective mentalizing impairments to broadened deficits of both ToM and basic cognitive functioning, results of a recent review supported that the ToM deficits seen in bvFTD do not simply reflect a general cognitive impairment [29].

In contrast to bvFTD, global cognitive functioning is considered to influence performances on mentalizing tasks in AD patients [32]. In particular, false-belief tasks with highly demanding cognitive load (i.e., second-order) are more impaired compared to first-order conditions [18, 33]. This evidence strongly suggests a prominent role of global cognitive functioning in the resulting performance on ToM task in AD. Since temporo-parietal regions are selectively damaged in AD dementia [34], posterior components of the mentalizing networks may be affected in this neurodegenerative condition. Moreover, with the progression of the disease and the extension of the pathological process to more anterior brain regions [35, 36], it is likely that also AD patients may present affective ToM deficits. While evidence on cognitive ToM impairments in AD are consistent [18, 33], reports of affective ToM deficits are sparse and discordant, even with the use of the same ToM paradigm (i.e., the Reading the Mind in the Eye; [37–39]).

In order to assess general or condition-related ToM deficits, we explored affective and cognitive facets of mentalizing abilities in bvFTD and AD patients using a single task (i.e., Story-based Empathy task, SET) [28] in its standardized version [40]. Moreover, as the SET also includes a control condition (i.e., physical causality), we evaluated the weight of basic cognitive functions on the resulting ToM skills. Performances of AD dementia patients are compared with those of a group of amnestic mild cognitive impairment patients (aMCI) fulfilling IWG criteria [41, 42] for AD in predementia phase (i.e., cerebrospinal fluid (CSF) positive for AD), in order to evaluate whether the ToM deficits reported in AD are associated to “AD dementia” or to “AD pathology” condition.

MATERIALS AND METHODS

RESULTS

One-way ANOVA on SET performances highlighted significant differences between groups in all the SET conditions (Table 2). In particular, post-hoc analyses proved significantly lower performances in SET-CI (F(3,108) = 5.65, p <  0.001, η² = 0.136), IA (F(3,108) = 17.84, p <  0.001, η² = 0.331), and EA (F(3,108) = 16.88, p <  0.001, η² = 0.319) conditions in both bvFTD and AD compared to HC (Table 2). No significant difference was found between the two dementia groups. Noteworthy, aMCI patients showed no significant difference in any SET condition compared to HC, but their performances significantly differed from those of AD patients (Table 2).

In the vectorial analysis both AD and bvFTD groups revealed a significant lower performance measured by the d index (d_SET - IA (F(3,108) = 15.46, p <  0.001, η² = 0.3), and d_SET - EA (F(3,108) = 13.01, p <  0.001, η² = 0.265) (Table 2 and Fig. 2). A specific imbalance between the affective ToM condition (EA) and the basic abilities (CI) measured by α_SET - EA, which was significantly different between bvFTD and HC (F(3,108) = 5.012, p <  0.01, η² = 0.122), was found only in the bvFTD group. No imbalance between cognitive ToM condition (IA) and the basic abilities (CI) was found in any group. Consistently with the results of the main statistical analysis (see above), the vectorial analysis showed no differences in d or α index in the aMCI group compared to HC (see Fig. 2). In sum, the vectorial analyses revealed: (1) The aMCI group did not differ from the HC group on overall performance or on the pattern (balance) across IA and CI and across EA and CI— indicating this patient group is “normal” on ToM and on basic cognitive abilities. (2) The AD group, compared to the HC group, performed more poorly on the IA and EA tasks but this impairment in performance on ToM can be explained entirely by a corresponding impairment in basic cognitive ability (as indexed by CI). (3) The bvFTD group, compared to HC group, performed more poorly on the IA and EA tasks, with the diminished performance by the bvFTD group on IA explicable in terms of a corresponding impairment in basic cognitive ability whereas the diminished performance by this group on EA explicable by a specific change in a specific form of social cognition (in contrast to a basic change in cognitive ability).

In order to provide a further confirmation of the imbalance between EA and CI abilities in bvFTD, we performed additional statistical analysis on SET-EA using SET-CI score as covariate. Consistent with the findings of vectorial analysis, we found a significant statistical effect of the group (F(3,107) = 11.21, p <  0.001). Post-hoc analyses revealed significant differences between bvFTD and both HC (p <  0.001) and aMCI (p <  0.05). We then compared the performances at SET conditions within groups. BvFTD was the only group in which we detected a significant effect (F(2,38) = 4.06, p <  0.05), due to the poorer performance in EA sub-task compared to the control condition (p <  0.05).

Correlation analyses showed a positive correlation between EA condition of SET task and both the IRI global score (Pearson r = 0.451, p <  0.05) and emotional empathy subscales considered together (Pearson r = 0.378, p <  0.05). No further significant correlation emerged.

DISCUSSION

In the present study, we investigated the ability to attribute mental states using a single task (i.e., Story-based Empathy Task, SET) based on affective and cognitive cues in a sample of neurodegenerative dementia (i.e., bvFTD and AD) and predementia (i.e., aMCI) patients. The use of a single ToM paradigm allowed us to better compare patients’ performances in the different facets of mentalizing and to evaluate the weight of basic cognitive functions on the resulting ToM performance through the introduction of a control condition, which equates the ToM task in the general cognitive requirements, but which can be solved without any mentalistic reading. A vectorial analysis was applied to evaluate the selectivity of social functioning deficit by means of the balance between basic functions (SET-CI) and ToM abilities (SET-IA and SET-EA) (see Fig. 2).

As expected, dementia patients showed decreased ToM performances (d_SET - EA and d_SET - IA). In particular, both AD and bvFTD patients showed reduced scores in all the SET conditions. The evidence of a reduced performance on the control condition suggests the presence of basic cognitive dysfunctions in such patients that may also account for reduced ToM scores [51]. Since neurodegenerative dementia patients usually present simultaneous impairments of different cognitive abilities, mentalizing deficits may be coexistent with dysfunctions on other cognitive domains. These latter deficits may crucially influence performances on cognitive highly demanding task such as ToM paradigms [51, 52].

The analyses on the overall performance (i.e., d index) in AD patients suggest that defective performance in affective and cognitive metalizing may be at least partially explained by basic cognitive deficits. In particular, according to Castelli and colleagues [37], AD ToM deficits may be secondary to other cognitive impairments, with high-level ToM abilities (both affective and cognitive) being the first to be affected, followed then by skills that are more basic in the advanced stages of the disease. Cortical atrophy in AD involves temporal posterior regions as the posterior cingulate cortex, the precuneus and the superior temporal sulcus [36], which underpin cognitive functions related to social abilities, such as mental imagery [53], representation of complex goals [14], and perspective-taking [54]. Damage to these regions may thus elicit, in AD, a deficit in the basic processes underlying the performance of ToM tasks.

On the contrary, socio-emotional processing disorders are core features of bvFTD clinical picture and usually represent key symptoms for the diagnosis [24, 56], suggesting a selective damage of mentalizing and other social cognition networks in this neurodegenerative condition [57]. In particular, bvFTD patients appear to be impaired in other-oriented emotional reactions, which, conversely to intentionality comprehension, are independent from executive functioning or to the general cognitive status [58].

In agreement with this, our data showed a reduced α_SET - EA index compared to HC only in the bvFTD group, proving an imbalance between emotion attribution and causal inference abilities. The introduction of a control condition is highly recommended in ToM tasks to improve the interpretation of defective performances [29]. In particular, the vectorial analysis provides the first evidence of the fact that, in contrast to AD, bvFTD patients present a mentalizing impairment in addition to global cognitive deficits. This evidence is in line with the specific degeneration of fronto-limbic networks in bvFTD [59] that disrupts critical hubs within the so-called “social brain” [60] and results in a severe breakdown of the affective facets of mentalizing ability.

Since the ability to attribute affective states to others (i.e., affective ToM) requires the integration of both cognitive and affective aspects of empathy, with the involvement to some extent of emotional empathy (e.g., emotional contagion, empathic concern, personal distress) [4], which is well known to be impaired in bvFTD patients [27, 61], we tested the relationship between EA performances and empathic attitude. In line with this hypothesis, we found a positive correlation between α_SET - EA index and the IRI global and emotional empathy scores in demented patients. Although this finding suggests low EA performance as good index of impaired affective empathy reflecting the social skills of subjects in daily life [62], the lack of correlation between EA and IRI sub-scales in our bvFTD sample are not in line with this finding. Further studies are thus needed to better determine the relationship between impaired performances in social tasks and altered social behaviors in daily life.

Unlike dementia patients, the aMCI group did not show any impairment either in the SET or in the vectorial analysis, compared with controls. Significant lower performances on SET emerged in AD compared to prodromal AD/aMCI patients. This result suggests that ToM deficits in AD highly depend on the degree of global cognitive impairments rather than being a signature of the AD pathology. Indeed, although previous reports provided evidence of ToM deficits in aMCI patients [63–65], this result may be due to the use of cognitively demanding tasks [64]. Different from AD, in which ToM deficits seem to be related to the dementia stage, bvFTD patients show markedly diminished ToM performances, particularly in the affective component, even in the mild disease stages [66], when daily functioning is not impaired and no other cognitive deficits are present. This finding supports the concept that mentalizing dysfunction based on affective cues is a core signature of social cognition disorders in bvFTD patients.

Noteworthy, we did not find an imbalance between cognitive ToM condition and basic control abilities (i.e., α_SET - IA) either in AD or in bvFTD patients. This result may be due to the limitations of the study (e.g., small sample size) or to the specific design of the SET-IA condition. Thus, large studies are needed to clarify whether EA is the only component selectively impaired in bvFTD patients and to better analyze the weight of specific cognitive functions in emotion and intention attribution tasks.

In conclusion, the results of our study provide the first direct evidence of a disproportion in affective and cognitive ToM deficits between bvFTD and AD. Even though EA and IA deficits are related to basic cognitive dysfunctions in both dementia conditions, the results of the vectorial analysis suggest that these groups experienced ToM difficulties for different reasons. In particular, AD-related ToM deficits are secondary to more general cognitive difficulties typical of AD dementia. Affective ToM difficulties instead are a core disturbance of bvFTD that may not only be attributed to general cognitive demands. Finally, our data underline the importance of introducing validated tasks exploring affective ToM component in the neuropsychological assessment of patients suspected for bvFTD, in order to provide an early and more accurate differential diagnosis.