Cognitive and Behavioral Profiles of Left and Right Semantic Dementia: Differential Diagnosis with Behavioral Variant Frontotemporal Dementia and Alzheimer’s Disease

INTRODUCTION

Neuropsychology has played a critical role in characterizing the cognitive changes associated with dementing disorders. This has improved our ability to accurately diagnose dementia subtypes, to identify subtle cognitive changes that occur in the prodromal phase of the disease, and to track clinical progression [1]. In patients with suspected dementia, a comprehensive neuropsychological and neuropsychiatric evaluation is an indispensable diagnostic element.

Semantic dementia (SD) is a subtype of frontotemporal dementia (FTD) commonly associated to TDP-43 protein deposits, and to a lesser extent to tau protein [2, 3]. Clinically, SD is a primary progressive aphasia syndrome which is characterized by specific semantic memory loss and preserved abilities of other cognitive functions [4, 5]. The initial cognitive profile shows a loss of semantic knowledge [6, 7], with relatively preserved episodic memory, in particular, visual memory, visuospatial capacity, and executive functions. Tests of visual confrontation naming, single-word comprehension, category fluency, and word-picture matching are useful to confirm the loss of semantic knowledge in this group of patients [8, 9]. It has been reported that SD patients also perform poorly on tests of verbal episodic memory, such as list learning and story recall, and recognition memory for visual materials [10]. Patients with early Alzheimer’s disease (AD) generally show preserved semantic knowledge for ordinary objects/animals. On tests of autobiographical memory, SD patients recall recent life events significantly better than their memory for early life periods, whereas AD patients exhibit an inverse pattern [11].

Although SD typically presents with a left predominant anterior temporal atrophy, up to 30% of cases exhibit right predominant involvement at onset [12]. In each case, the contralateral anterior temporal lobe becomes affected as the disease progresses [13]. Thus, SD can be divided into two clinical presentations: Left-sided SD (L-SD) and right-sided SD (R-SD). L-SD patients show more naming and comprehension difficulties [14], whereas R-SD patients typically exhibit behavioral changes and prosopagnosia [6, 16]. Behavioral changes most commonly referred include disinhibition, apathy, aggression, lack of interpersonal engagement, loss of empathy, compulsive and rigid behaviors, restricted food preference in favor of sweet food and hyper-religiosity [17, 18].

The prominence of behavioral symptoms described in R-SD patients requires differential diagnosis with the behavioral variant frontotemporal dementia (bvFTD). bvFTD is characterized by progressive and insidious behavioral alterations associated with a pattern of cognitive performance defined by the relative preservation of episodic memory and visuospatial functions with executive dysfunction [7]. Thus, differential diagnosis may be particularly challenging, especially in the early disease phase of bvFTD [19] when the neuropsychological performance can be normal. However, there are conflicting reports concerning the extent to which episodic memory is preserved [20]. A growing number of studies have shown that bvFTD may have a moderate episodic memory disorder comparable to the one seen in AD [20, 21] which cannot be entirely explained by executive dysfunction but reflects some degree of disruption to memory storage and its consolidation processes [22, 23]. Two discrete behavioral syndromes have been described among patients with bvFTD: 1) the apathetic type, distinguished by declined volition and motivation, loss of socio-emotional awareness, isolating behaviors and increased latency to pain response; and 2) the disinhibited type which is characterized by hyper-orality, preference for sweet foods, perseverative behaviors, and motor stereotypies [24]. Increased disinhibition and impulsivity can lead to inappropriate comments (e.g., sexually explicit), inadequate social behavior, overspending, pathological gambling, and more rarely, hyper-religiosity [25–27]. R-SD share symptoms with bvFTD, such as lack of empathy, disinhibition, changes in diet preferences, and a decline in personal care [28]. However, while patients with R-SD typically show prosopagnosia, obsessions, rigid behavior, personality changes, and disorientation, bvFTD patients show higher degrees of apathy and executive dysfunction [16, 28].

The other principal differential diagnosis of SD is AD. Clinically, typical AD is characterized by progressive cognitive decline, particularly involving episodic memory. This episodic memory impairment is usually the earliest and most salient aspect of the AD dementia syndrome. It has been shown that the memory deficit reflects an inability to encode and store new information effectively. Additionally, some studies are reporting that semantic memory is early impaired in AD patients, affecting verbal fluency and naming. The semantic memory loss in AD may occur several years before diagnosis [29]. Additionally, some AD patients, typically exhibiting younger onset ages, present a disproportionately greater executive dysfunction than memory impairment [30, 31]. Furthermore, distinct studies have also evidenced that typical AD patients show mild to severe executive impairment across a variety of neuropsychological tests such as attention, verbal fluency, and working memory, concluding that AD subjects might perform similarly to bvFTD patients [32–34]. However, AD manifestations are not limited to cognitive symptoms; but they can also include a wide range of neuropsychiatric symptoms like depression, apathy, sleep disorder, agitation, and psychosis [35].

Thus, AD, especially young-onset AD, bvFTD, and SD present an overlap of cognitive and neuropsychiatric symptoms, which make it necessary to carry out a comprehensive differential diagnosis. Previous studies have already tried to address this issue; however, they had some limitations. First of all, SD is a rare disease, and most publications include relatively small samples from single centers. (Table 1 shows a review of the main published series which included at least more than 15 patients with SD). Secondly, the right temporal atrophy onset is commonly underdiagnosed being usually underrepresented in SD series, since, apart from structural magnetic resonance imaging (MRI), most published studies did not use disease biomarkers to precisely categorize patients.

One of the breakthroughs in the field has been the incorporation of molecular imaging to dementia diagnosis. Positron emission tomography (PET) techniques have provided significant advances, promoting novel approaches to support an earlier and more accurate differential dementia diagnosis. These diagnostic tools have robustly shown its utility in the differential diagnosis of AD versus FTD syndromes [36–38].

We propose a study systematically assessing cognitive performances and neuropsychiatric symptoms in one of the largest cohorts of SD to date, with substantial representation of R-SD, compared to a population of young AD and bvFTD defined by the support of molecular imaging (amyloid PET (Pittsburgh Compound-B (PiB)) and 2-[¹⁸F] fluoro-2-deoxy-D-glucose (FDG) PET). We aimed to describe SD profile and to evaluate the accuracy of various neuropsychological parameters to distinguish these neurodegenerative syndromes from young AD and bvFTD. In addition, we aimed to compare the neuropsychiatric and neuropsychological assessment between L-SD, R-SD, and bvFTD.

We hypothesize that the AD group will obtain worse scores in episodic verbal memory and visuospatial capacity; the bvFTD group will score significantly worse in tasks that measure executive functions and emotion recognition and will show marked behavioral disorders; finally, SD patients will obtain worse performance in tasks that measure semantic memory, a test of visual confrontation naming, categorical fluency, as well as in tasks that evaluate prosopagnosia. We hypothesized that of all neuropsychological assessment, a detailed evaluation of language, episodic memory, and prosopagnosia will help us with the differential diagnosis.

METHODS

RESULTS

Toward a more parsimonious assessment of neuropsychological and neuropsychiatric symptoms

A forward stepwise selection method was used to evaluate how the neuropsychological assessment differentiated between the groups, and to identify which of those variables had the greatest discriminant value. Figure 1 shows in a schematic way the results obtained in the logistic regression.

OPEN IN VIEWER

Fig.1

Cognitive and neuropsychiatric variables that best predict diagnoses AD versus bvFTD, SD versus bvFTD, and AD versus SD. Note: Neuropsychiatric symptoms are in red font and neuropsychological symptoms are in black font.

DISCUSSION

This study aimed to describe the main neuropsychological and neuropsychiatric features of a relatively large SD population diagnosed with the support of FDG and PiB-PET; trying to distinguish those items that allow to discriminate them from young-onset AD and bvFTD patients.

We decided to recruit patients with young AD (over 50 years old) with a positive PiB-PET because we believe that this population, especially at early disease stages, comprise the most challenging differential diagnosis with bvFTD and SD.

Our results revealed an overlap of many of the cognitive and neuropsychiatric symptoms. However, a prosopagnosia task, an episodic memory test that included semantic cues, a visuospatial, and a naming test were the four most helpful cognitive tests for the differential diagnosis. Besides, we described several behavioral abnormalities with discriminative power. Among them, aggression, personal care, and eating habits, specifically over-eating and altered food preference, were the most powerful items for group categorization.

In our study, we found differences across the distinct groups in all the neuropsychological domains explored with the only exception of attention and executives function tests which appeared as the tasks with less discriminative power between the groups. Even though a dysexecutive pattern is a part of the bvFTD diagnostic criteria [7] when compared with SD and early AD cases, we found that differences were not statistically significant.

As expected, AD group showed verbal and visual memory impairment, and they did not benefit from the controlled learning afforded by category cues. FCSRT enables the identification of memory storage failure defining the amnesic syndrome of the hippocampal type [59], primarily characterized by insensitivity to cueing and by a low total recall. Similar to other authors, we found that patients with AD were more impaired in episodic memory tests than bvFTD patients, with total recall the most helpful test to distinguish between these two diseases. Conversely, AD and SD performed similarly in episodic memory tests. Both the AD and SD groups were significantly impaired on verbal memory when compared to the bvFTD, while only patients with AD were impaired on visual memory. These results are consistent with previous studies that have demonstrated that SD scored below the normative values in episodic memory tasks [60]. We believe that the involvement of verbal language in the episodic memory task can explain these results. The bvFTD group was characterized by a memory profile that improved with category cues, unlike the memory profile found in AD and SD. Another expected finding was the AD patients’ lower performance in the visuospatial subtests from the VOSP. This test was especially helpful in the discrimination of AD versus SD.

Another result was that globally, patients with bvFTD showed worse performance on the copy condition of the RCF relative to the other clinical groups. However, differences were only significant when compared to SD. This result goes against the trend found in previous studies that presented patients with bvFTD with relative preservation of visuoconstruction skills [41, 61]. Interestingly, in simpler tasks, such as the CERAD Figures Copy, our bvFTD patients did not perform worse than AD or SD, which is strongly suggestive of a primary deficit in the planning and organization skills necessary for the correct execution of the RCF test rather than visuospatial or visuoconstruction disturbances per se.

SD group showed a pronounced impairment in the BNT and scored lower in animal fluency. In the description of objects task, that evaluates semantic memory, patients with SD performed worse than AD and bvFTD, with a higher number of “don’t know” and “I recognize it but ... ” “never heard it” responses and over-inclusively categories like “animal” to describe a camel or “vegetable” to describe an eggplant, ignoring the distinctive characteristics of each of them. The observed pattern of neuropsychological findings was generally consistent with previous reports [4, 62].

In the literature, about 30% of SD patients present with a predominant right temporal lobe atrophy. Frequently, it is difficult to differentiate R-SD from a bvFTD, the nosology of R-SD remaining uncertain [16, 63]. We compared neuropsychological performance between L-SD and R-SD and each of them with bvFTD. Surprisingly, we found significant differences in verbal repetition; being it relatively poorer in L-SD patients. L-SD group repeats worse than the R-SD, however, both groups continue to score within normal limits. More studies exploring the quality of the errors elicited on verbal repetition of low probability sentences, rather than a total achievement score, may shed light on this finding.

A remarkable result was the absence of statistically significant differences in the BNT when comparing R-SD with bvFTD. This finding is broadly in line with studies that argue that a predominant right anterior temporal lobe atrophy is associated with behavioral changes and prosopagnosia while naming and comprehension are disproportionately impaired in patients with more severe left atrophy [14]. The prosopagnosia task allowed differentiating patients with L-SD versus bvFTD and R-SD versus bvFTD, both SD group (especially R-SD) presenting a manifested difficulty for the famous face recognition when compared with bvFTD.

Neuropsychiatric symptoms were present across the three diagnostic groups. Patients with bvFTD showed more neuropsychiatric symptoms, followed by SD and finally AD, the differences being of statistical significance between bvFTD and AD. Our results are in line with Gainotti’s 2015 review which concludes that right temporal atrophy subsumes non-verbal representations and the left temporal atrophy lexical-semantic representations [64].

In our study, apathy was prevalent in patients with AD, bvFTD, and SD. Consistently with Chow et al. [65], it was particularly frequent in bvFTD, although no significant differences between groups were noted. We found that aggression, loss of empathy, the decline in personal care and eating habits, like over-eating and altered food preference, were the main neuropsychiatric features that permitted differentiation between AD and the FTD diseases.

Patients with bvFTD and SD share many behavioral symptoms, such as lack of empathy, disinhibition, changes in diet, and decline in personal care [28]. We confirmed these results, adding apathy as one of the most frequent non-cognitive symptoms in both groups. Aggression, most frequent among bvFTD patients, was the most discriminative neuropsychiatric symptom between bvFTD and SD, specifically with R-SD.

In line with the existing literature [6, 29], we found that R-SD presented more neuropsychiatric features than L-SD; however, those differences were not statistically significant in our population. Kumfor et al. in 2016 [66] obtained similar results to us, concluding that L-SD and R-SD evolve into a similar clinical profile, with social cognition affected in both phenotypes. Differences between studies might be related to several factors like the proportion of R-SD included, the assessment time point during the disease course, sample size or patient’s selection methods (Table 1). We think that the systematic use of molecular imaging and the relatively large percentage of R-SD ascertained are two strong points in our study.

bvFTD patients typically show deficits in facial emotion recognition [67–69]. A recently published meta-analysis described severe deficits in total emotion recognition compared to AD [70]. However, in our study, these differences did not reach statistical significance. At odds with previous studies, we found that the SD group was significantly associated with more severe deficits than AD and bvFTD in face emotion recognition. These results differ with several studies that indicate that patients with bvFTD show impairments on tests of face emotion processing [69, 71] and other studies that demonstrate that AD patients show similar deficits on tests of face emotion processing to those seen in bvFTD [72, 73]. A putative explanation for this phenomenon is the fact that prosopagnosia might contribute, to a certain degree, to face emotion recognition impairment in SD patients. The fact that after adjustment for prosopagnosia, the facial emotion recognition was no longer significantly discriminative supports this hypothesis. Additionally, language deficits present in SD patients can also contribute to explain this finding. Perry et al., 2001 [74] reported that emotion recognition might be particularly compromised in R-SD, supporting a differential contribution of left and right anterior temporal lobe regions to face processing [75]. Nevertheless, we found that both L-SD and R-SD showed equally impaired ability to recognize emotions and that there were no significant differences between these two groups; predominant language deficits in L-SD and prosopagnosia in R-SD might again complicate interpretation of these findings.

Although we do not replicate fully the results obtained by other studies regarding prosopagnosia and recognition of emotions, our work has found similar trends. A likely explanation is that the neuropsychological tools to evaluate these areas have not been sensitive enough to detect these differences, this being an important limitation to consider. Another limitation of the present study was the loss of statistical power in our multivariate analysis because of missing values, mainly in neuropsychological variables. The missing values for neuropsychological variables was 39.1%, with the majority corresponding to the group of SD patients (62.8% available). Multivariate analysis could, however, be performed with 78.1% AD and 75% bvFTD. For behavioral symptoms and NPI, there was no significant missing values and the multivariate analysis could be performed with 97.7% of the total sample (96.8% AD, 100% bvFTD, and 97.1% SD).

Few studies have compared cognitive and neuropsychiatric performance in young AD, bvFTD, and SD. Previous reports have studied the neuropsychological patterns in FTD and AD [41, 76] but did not include the assessment of the main neuropsychiatric symptoms, as well as tasks to evaluate prosopagnosia and emotions recognition, which, as shown by our data, are key elements for the differential diagnosis. Our study evaluated all these aspects in a relatively large and well-characterized cohort of patients. Specifically, the SD population included in this article is one of the largest and more comprehensively-studied samples of this disease described so far (see Table 1), with a very high representation of R-SD cases. Another important advantage of the current study is that most clinical diagnosis (all patients with young AD and bvFTD, and in 24 out of 35 patients with SD) were supported by typical patterns of hypometabolism on FDG-PET and PiB-PET results. Thus, minimizing patient’s misclassification, a capital factor increasing the validity of our observations. Very few SD series have studied patients using FDG and PiB-PET and of the existing ones, the largest is composed of only 16 patients studied with FDG-PET [77].

SD is a clinical diagnosis with a high predictive value of TDP-43 pathology. Thus, an accurate diagnostic categorization is of particular interest, as this phenotype could potentially be used in patient’s selection for clinical trials targeting TDP43 pathology. In the present study, we were able to demonstrate the utility of an adequate neuropsychological evaluation for the differential diagnosis between SD, bvFTD, and young AD. In addition to neuropsychological tests, the information offered by families about non-cognitive symptoms was particularly important, altogether offering clinical clues to differentiate between the three clinical syndromes. Despite the latest advances in biomarkers and neuroimaging techniques, this paper demonstrates the significant contribution of a detailed and theoretically motivated neuropsychological evaluation in the differential diagnosis between SD, bvFTD and young AD.