Introduction
The behavioral variant of frontotemporal dementia (bvFTD) is a subtype of frontotemporal lobe degeneration (FTLD), a group of clinical syndromes associated with focal atrophy of the frontal and anterior temporal lobes. This syndrome also includes semantic dementia and progressive nonfluent aphasia (McKhann et al., Reference McKhann, Albert, Grossman, Miller, Dickson and Trojanowski2001; Neary et al., Reference Neary, Snowden, Gustafson, Passant, Stuss, Black and Benson1998).
BvFTD is characterized by an early alteration in behavior and personality with a loss of insight, apathy, and disinhibition (Piguet, Hornberger, Mioshi, & Hodges, Reference Piguet, Hornberger, Mioshi and Hodges2011; Seelaar, Rohrer, Pijnenburg, Fox, & van Swieten, Reference Seelaar, Rohrer, Pijnenburg, Fox and van Swieten2011). Clinical symptoms in bvFTD have mostly been related to progressive hypometabolism and atrophy in the frontal and polar temporal lobes (Agosta, Canu, Sarro, Comi, & Filippi, Reference Agosta, Canu, Sarro, Comi and Filippi2012; Schroeter, Raczka, Neumann, & von Cramon, Reference Schroeter, Raczka, Neumann and von Cramon2008; Seeley et al., Reference Seeley, Crawford, Rascovsky, Kramer, Weiner, Miller and Gorno-Tempini2008), particularly in the medial prefrontal cortex (mPFC) (including the ventromedial prefrontal cortex), which is known to be affected in the early stages of the disease (Boccardi et al., Reference Boccardi, Sabattoli, Laakso, Testa, Rossi, Beltramello and Frisoni2005; Broe et al., Reference Broe, Hodges, Schofield, Shepherd, Kril and Halliday2003; Perry et al., Reference Perry, Graham, Williams, Rosen, Erzinclioglu, Weiner and Hodges2006; Schroeter, Raczka, Neumann, & von Cramon, 2007; Seeley et al., Reference Seeley, Crawford, Rascovsky, Kramer, Weiner, Miller and Gorno-Tempini2008). The mPFC is associated with social cognition, which is also known to be particularly impaired in the early stages of bvFTD (Adenzato, Cavallo, & Enrici, Reference Adenzato, Cavallo and Enrici2010; Gregory et al., Reference Gregory, Lough, Stone, Erzinclioglu, Martin, Baron-Cohen and Hodges2002). Additionally, the ventral part of mPFC has been linked to inhibition (Hornberger, Geng, & Hodges, Reference Hornberger, Geng and Hodges2011) and reversal-learning (Rahman, Sahakian, Hodges, Rogers, & Robbins, Reference Rahman, Sahakian, Hodges, Rogers and Robbins1999). However, the neural correlates of reversal-learning have never been investigated in bvFTD patients. Diagnostic criteria for bvFTD have been recently revised (Rascovsky et al., Reference Rascovsky, Hodges, Knopman, Mendez, Kramer, Neuhaus and Miller2011) and will allow for better identification of bvFTD patients, but neither dysfunctions of the medial and orbital PFC or social cognition impairments have been taken into account in this revision. Given their high and early prevalence in bvFTD, imaging criteria focused on mPFC atrophy and deficits in “frontomedial” functions (such as social cognition) could be used to diagnose or even predict bvFTD (Schroeter, Reference Schroeter2012). The neuropsychological evaluation of cognitive functions that drive social behavior, such as emotional processing, reward sensitivity, and affective shifting, would complement classical cognitive evaluation (Sarazin, Dubois, de Souza, & Bertoux, Reference Sarazin, Dubois, de Souza and Bertoux2012). These evaluations would also improve the assessment of behavioral symptoms that are currently based on subjective clinical scales and caregiver interviews.
Some of the neuropsychological studies focusing on the assessment of social cognition and emotion processing in bvFTD have demonstrated its effectiveness in distinguishing bvFTD patients from controls (Torralva, Roca, Gleichgerrcht, Bekinschtein, & Manes, Reference Torralva, Roca, Gleichgerrcht, Bekinschtein and Manes2009) and patients with Alzheimer's disease (Funkiewiez, Bertoux, de Souza, Levy, & Dubois, 2011) or depression (Bertoux et al., Reference Bertoux, Delavest, de Souza, Funkiewiez, Lepine, Fossati and Sarazin2012). The Executive and Social Cognition Battery (ESCB) (Torralva et al., Reference Torralva, Roca, Gleichgerrcht, Bekinschtein and Manes2009), the Social cognition and Emotional Assessment (SEA) (Funkiewiez et al., Reference Funkiewiez, Bertoux, de Souza, Levy and Dubois2012), and the reduced version of this last tool, the mini-SEA (Bertoux et al., Reference Bertoux, Delavest, de Souza, Funkiewiez, Lepine, Fossati and Sarazin2012), have been proposed as new diagnosis tools for bvFTD. The SEA is a battery composed of five different subtests: a theory of mind evaluation (Faux-Pas test), a facial emotion recognition test (these two subtests compose the mini-SEA), reversal-learning and behavioral-control tests, and an apathy scale. The functions evaluated by each of these tests have been shown to be impaired in bvFTD (Gregory et al., Reference Gregory, Lough, Stone, Erzinclioglu, Martin, Baron-Cohen and Hodges2002; Lavenu & Pasquier, Reference Lavenu and Pasquier2005; Lough et al., Reference Lough, Kipps, Treise, Watson, Blair and Hodges2006; Rahman et al., Reference Rahman, Sahakian, Hodges, Rogers and Robbins1999; Torralva et al., Reference Torralva, Kipps, Hodges, Clark, Bekinschtein, Roca and Manes2007; Zamboni, Huey, Krueger, Nichelli, & Grafman, Reference Zamboni, Huey, Krueger, Nichelli and Grafman2008).
The sensitivity and specificity of such tools may be mainly explained by the earliness and specificity of atrophy and hypoperfusion within OFC and mPFC in bvFTD compared to that of other neurodegenerative disorders such as Alzheimer's disease (AD) (Agosta et al., Reference Agosta, Canu, Sarro, Comi and Filippi2012; Apostolova & Thompson, Reference Apostolova and Thompson2008; Krueger et al., Reference Krueger, Dean, Rosen, Halabi, Weiner, Miller and Kramer2010; Rabinovici et al., Reference Rabinovici, Seeley, Kim, Gorno-Tempini, Rascovsky, Pagliaro and Rosen2007). Lesion and fMRI studies have shown that facial emotion recognition is linked to the mPFC and medial OFC as well as the amygdala (Adolphs, Reference Adolphs2002; Heberlin, Padon, Gillijan, Farah, & Fellows, 2008; Hornak et al., Reference Hornak, O'Doherty, Bramham, Rolls, Morris, Bullock and Polkey2004). Scores on theory of mind tests are strongly associated with the rostral mPFC, particularly with BA 10 (Carrington & Bailey, Reference Carrington and Bailey2009; Frith & Frith, Reference Frith and Frith2006; Gilbert et al., Reference Gilbert, Spengler, Simons, Steele, Lawrie, Frith and Burgess2006). Reversal-learning and inhibitory control are mostly related to the ventral mPFC or OFC (Fellows & Farah, Reference Fellows and Farah2003; Hornberger et al., Reference Hornberger, Geng and Hodges2011; Rolls, Hornak, Wade, & McGrath, Reference Rolls, Hornak, Wade and McGrath1994; Tsuchida, Doll, & Fellows, Reference Tsuchida, Doll and Fellows2010). The lateral PFC, mPFC and basal ganglia have been associated with motivation and behavioral initiation (Levy & Dubois, Reference Levy and Dubois2006). Therefore, the SEA may be sensitive and specific, because it includes tests that are related to mPFC/OFC functions. However, to our knowledge, no study has directly investigated the neural correlates of the tests that compose the SEA in bvFTD patients.
In an effort to demonstrate and understand the SEA's usefulness in assessing bvFTD, we investigated the relationship between performance on the SEA subtests (including the mini-SEA) and regional gray matter volume in patients with bvFTD using voxel-based morphometry (VBM), an automated method of structural brain analysis. We investigated how the regional gray matter volume varied with SEA subtests performance. Based on converging evidence from the literature and previous clinical results using this battery, we expected that reinforcement learning tests would be associated with medial and/or lateral OFC cortical volume, while theory of mind and emotion recognition would be related to mPFC gray matter volume.
Method
Patients
Twenty patients participated in the study: 10 women and 10 men, 68.5 ± 9.1 years old, range 54–83 years; years of education = 9.95 ± 4.75, range 3–17; disease duration = 2.7 years ± 1.2, range 0.5–6. All bvFTD patients were seen and evaluated at the National Reference Center for Rare Dementia in the Pitié-Salpêtrière Hospital, Paris, France. The final diagnosis was established by experts after the synthesis of multidisciplinary clinical assessments based on neuropsychological, neurological, biological, and neuroimaging evidence. The bvFTD patients were enrolled according to the Lund and Manchester criteria for bvFTD diagnosis (Neary et al., Reference Neary, Snowden, Gustafson, Passant, Stuss, Black and Benson1998) and fulfilled the recently revised diagnostic criteria (Rascovsky et al., Reference Rascovsky, Hodges, Knopman, Mendez, Kramer, Neuhaus and Miller2011). All patients presented with prominent changes in personality and social behavior that were established by their caregivers’ interviews. All patients underwent a neuropsychological examination that included the Mini Mental State Examination (Folstein, Folstein, & McHugh, Reference Folstein, Folstein and McHugh1975) and the Mattis Dementia Rating Scale (Mattis, Reference Mattis1976) for general efficiency, the Frontal Assessment Battery (Dubois, Slachevsky, Litvan, & Pillon, Reference Dubois, Slachevsky, Litvan and Pillon2000), verbal fluency (semantic fluency for animals and phonological fluency for the letter M, each tested in 1 min), modified Wisconsin Card Sorting Task (Nelson, Reference Nelson1976) for executive functions, free and cued selective reminding test (Van der Linden et al., Reference Van der Linden, Coyette, Thomas-Anterion, Sellal, Poitrenaud, Gély-Nargeot and Deweer2006) for verbal episodic memory, praxis evaluation, a word denomination task for language evaluation, and the five tests of the SEA (see details below). Clinical and neuropsychological data are shown in Table 1, and the SEA score and raw subscores are presented in Table 2.
Table 1 Demographic and neuropsychological data of bvFTD patients
Note. Pathological threshold were established according to each test normative data.
bvFTD = behavioral variant frontotemporal dementia; MMSE = Mini-Mental State Examination; FAB = Frontal Assessment Battery; MDRS = Mattis Dementia Rating Scale; WCST = Wisconsin Card Sorting Task; SD = Standard deviation; N.A. = Not applicable.
Table 2 Social cognition & Emotional Assessment data of bvFTD patients: SEA and mini-SEA composite score and raw scores (used for the VBM analysis) for the five subtests
Note. The number of patients that obtained a pathological SEA subscore and raw scores is indicated for each test. Pathological threshold were established according to normative data from Funkiewiez et al. (Reference Funkiewiez, Bertoux, de Souza, Levy and Dubois2012), reported in the last column.
VBM = voxel-based morphometry; bvFTD = behavioral variant frontotemporal dementia; SEA = Social cognition & Emotional Assessment; SD = standard deviation; N.A. = not available.
A structural MRI was performed for all patients and revealed frontal or fronto-temporal atrophy.
Patients were excluded if they presented any of the following: (1) language complaints (progressive non-fluent aphasia or semantic dementia); (2) systemic illnesses that could interfere with cognitive functioning; (3) vascular lesions seen on MRI or neurological history suggestive of vascular dementia; (4) motor-neuron disease; (5) major depression; or (6) use of anxiolytics or antipsychotics drugs. To improve diagnostic accuracy, all patients had at least an 18-month follow-up to validate the diagnosis according to their clinical evolution. Four patients were previously included in another study on the SEA (Funkiewiez et al., Reference Funkiewiez, Bertoux, de Souza, Levy and Dubois2012).
This study was conducted at the National Reference Centre for Rare Dementias and in the Neurology Department of Pitié-Salpêtrière Hospital. All clinical and neuroimaging data were generated during a routine clinical work-up and were extracted for the purpose of this study. According to French legislation, the patients were informed and consented that their data might be used in retrospective clinical research studies. All regulations concerning electronic filing were followed.
Social Cognition and Emotional Assessment
All patients underwent the SEA within six months of the MRI. The SEA consists of five subtests and provides six weighted composite scores. We briefly describe each SEA subtest below and how their raw scores were calculated. A detailed description of the tests, including instructions and scoring, is available elsewhere (Funkiewiez et al., Reference Funkiewiez, Bertoux, de Souza, Levy and Dubois2012). The mini-SEA includes the theory of mind and the emotional recognition tests alone (Bertoux et al., Reference Bertoux, Delavest, de Souza, Funkiewiez, Lepine, Fossati and Sarazin2012).
Theory of mind evaluation
We used a short (10 stories) version of the Faux-Pas test (Stone, Baron-Cohen, & Knight, Reference Stone, Baron-Cohen and Knight1998) to evaluate theory of mind. For this study, we calculated two scores: a “detection of faux-pas” score and an “explanation of faux-pas” score (What was the faux-pas? Who made it? Why? Was it intentional? How did the victim feel?).
Reversal-learning task
This test had been adapted from the original paradigm of Rolls et al. (Reference Rolls, Hornak, Wade and McGrath1994) and included three phases. In the learning phase, the patients were asked to choose an advantageously rewarded item. Subsequently (in the reversal phase), the contingencies were unexpectedly reversed: the previously non-rewarded item became the advantageous one and had to be chosen. Depending on the patients’ performance, there could be more than one reversal. Afterward, the patients had to refrain from choosing any items (extinction phase). Nine correct consecutive choices were needed to pass each phase. For this study, we scored the number of trials needed to pass the reversal phase (“reversal score”), the number of reversals (“global reversal score”), and the number of errors committed before reversing the first rule (“reversal errors”).
Behavioral-control test
In this test, the patients had to choose between two items on the basis of a virtual monetary reward or punishment. Three consecutive rules were to be found and followed. First, an alternation between the two items was required. Second, the patients had to choose the first item at each trial. Finally, the second item had to be chosen at each trial. Six correct consecutive correct choices were needed to validate each rule. For this study, we scored the number of errors committed before reaching the second rule (“reversal errors”).
Facial emotions recognition test
Thirty-five faces from Ekman pictures (Ekman & Friesen, Reference Ekman and Friesen1975) were presented. The patients were asked to identify which emotion was being expressed by choosing among a list of seven different emotions displayed at the top of the screen: fear, sadness, disgust, surprise, anger, happiness, and neutral. A percentage of accurate general recognition was calculated (“emotion recognition score”).
Apathy scale
Apathy was assessed by caregivers using a questionnaire adapted from Starkstein (the “Apathy Scale”) (Starkstein et al., Reference Starkstein, Mayberg, Preziosi, Andrezejewski, Leiguarda and Robinson1992).
Statistical Analysis
The STATISTICA 6 software (StaSoft, 2001) and SPSS 20 (IBM, 2011) were used for statistical evaluation where appropriate. Continuous variables were tested for normality using both the Shapiro-Wilk and Kolmogorov-Smirnov tests after a visual inspection of histograms. Pearson r coefficient correlation and Student's t test were then used.
Imaging acquisition
All patients underwent the same imaging protocol with whole-brain T1 images using a 1.5 Tesla GE-Medical System MRI scanner. The three-dimensional (3D) T1-weighted images were acquired as follows: axial orientation, matrix 256 × 192, 128 slices, 0.859375 × 0.859375 mm2 in-plane resolution, slice thickness 1.5 mm, TE/TR = 8/23 ms.
VBM preprocessing
The 3D T1-weighted sequences were analyzed with SPM8 (http://www.fil.ion.ucl.ac.uk/spm) and MATLAB 7.12 (Math-Works, Natick, MA). First, MR images were segmented into gray matter (GM), white matter, and cerebrospinal fluid using SPM8's new version of the default unified segmentation (Ashburner and Friston, Reference Ashburner and Friston2005). Second, GM population templates were generated from the entire image dataset using the “diffeomorphic anatomical registration using exponentiated lie algebra” (DARTEL) technique (Ashburner, Reference Ashburner2007). Third, after an initial affine registration of the GM DARTEL templates to the tissue probability maps in the Montreal Neurological Institute (MNI) space (http://www.mni.mcgill.ca/), non-linear warping of GM images to the DARTEL GM template in the MNI space was performed. Fourth, the images were then modulated to ensure that relative volumes of GM were preserved following the spatial normalization procedure. Lastly, the images were smoothed with an 8 mm full width at half maximum Gaussian kernel. After spatial pre-processing, the smoothed, modulated, normalized GM datasets were used for statistical analysis. Visual checks were performed at each stage of the SPM pre-processing to ensure accuracy. No manual editing was required.
Finally, based on our hypotheses, the analysis was performed inside a frontal mask (including anterior and middle cingulate cortices) imported from the WFU_PickAtlas software (http://www.fmri.wfubmc.edu/download.htm). The details of the areas composing the mask are presented in Supplementary Table S1 and Figure S3.
VBM analysis
We ran multiple regressions analyses between gray matter volume and each of the SEA subscores in a brain analysis focused on the frontal lobes. Age, Mini-Mental State Examination (MMSE), and Frontal Assessment Battery (FAB) scores were co-varied out in the linear regression model to control for aging, global cognitive efficiency, or executive dysfunctions and to improve the specificity of our results.
Data were also corrected for individual total GM volume using a proportional normalization in the model.
To obtain a larger range of values, we used raw SEA subscores. Six patients did not undergo the faux-pas test due to extreme fatigability during the assessment; thus, the size of the group for this analysis was 14 patients. Six patients failed to learn the first rule in the reversal-learning test after more than 40 trials; thus, the size of the group for this analysis was 14. For correlation with the apathy scale, the size of the group was 17 because three caregivers did not rate it. For all other analyses, the size of the group was 20 patients.
For each statistical test, we investigated significant results at p < .05 corrected for family-wise error at the cluster level and at an uncorrected threshold of p < .005 with a minimal cluster size of 100 voxels.
The signal in each significant maximum at this uncorrected threshold was extracted using the MarsBar region of interest toolbox for SPM (http://marsbar.sourceforge.net). The ROIs were defined as 1.5 × 1.5 × 1.5 mm (voxel-size) boxes, centered on each significant maximum. As these ROIs were not independent from the global analysis, our aim was not to run further analyses on these ROIs, but to better describe the effects observed in the global analysis.
For each ROI and each given subscore, the individual MRI GM values were plotted against this subscore to verify that individual data had a regular dispersion (see Table 4; Figures 1, 2, and 3; and Supplementary Figures S1 and S2) and against the other subscores of the SEA. Pearson R correlations were also calculated.
Fig. 1 Correlation between gray matter volume and explanation of social faux-pas in the faux-pas test superimposed on the canonical template provided in SPM. Normalized grey matter intensity in the left Brodmann area (BA) 10 (A) at −9 48 13 and left BA 13 (B) at 46 2 12 are plotted against the “faux-pas” score.
Fig. 2 Correlation between gray matter volume and reversal errors in the reversal-learning test superimposed on the canonical template provided in SPM. Normalized gray matter intensity in the left Brodmann area (BA) 11 (C) at −18 38 −20 and right BA 24 (D) at 6 15 27 are plotted against the “reversal errors” score.
Fig. 3 Correlation between gray matter volume and performance at the emotion recognition test superimposed on the canonical template provided in SPM. Normalized gray matter intensity in the right Brodmann area (BA) 9 (E) at 4 50 24 and right BA 8 (F) at 3 32 49 are plotted against the “emotion recognition” score.
Results
Social Cognition and Emotional Assessment
Each patient had a composite SEA score below the pathological threshold established in a previous study (Funkiewiez et al., Reference Funkiewiez, Bertoux, de Souza, Levy and Dubois2012). The number of patients presenting an abnormal score at each of the SEA tests is provided in Table 2.
Voxel Based Morphometry
At an FWE-corrected threshold at the cluster level, two significant results were found (see Table 3). First, the “explanation of faux-pas” score exploring TOM positively correlated to the medial rostral PFC (medial BA 10 extending to BA 32; region A). Second, the number of errors in the reversal-learning test (“reversal errors” score) negatively correlated to one cluster in the anterior cingulate cortex (BA 24; region D) and to one cluster in the medial OFC (BA 11; region C) extending laterally into BA 47.
Table 3 MNI coordinates of cluster maxima for regions in which gray matter volume correlated with SEA performances in bvFTD patients
BA = Brodmann area; k = cluster size; G = gyrus; Lat. = lateral; Med. = medial; Sup. = superior; Inf. = inferior; Mid. = middle; z = Z scores; p uncor = uncorrected p value; * indicates a significant p value (p<.05) FWE-corrected at the cluster level.
At an uncorrected threshold (p < .005; 100 voxels extend), the “reversal errors” and the “explanation of faux-pas” scores were associated with several additional regions detailed in Table 3 and Figures 1 and 2. The correlation between GM volume and the “detection of faux-pas” score was non-significant. Performance on the other subtests was associated with cortical volume, particularly the prefrontal regions at this uncorrected threshold. Significant positive correlations were found between the “emotion recognition score” and GM in the bilateral medial superior frontal gyrus [bilateral BA 8 (region F) and right BA 9 (region E)] (Figure 3). Performance in the behavioral-control task (“reversal errors”) was associated (negative correlation) with small clusters in the premotor (BA 6; region G) and lateral prefrontal (BA 8/9; region H) regions (Supplementary Figure S1). Finally, a significant negative correlation was found between the “apathy scale” and GM volume in the right middle/inferior frontal gyrus (BA 44/45; region I) (Supplementary Figure S2).
Table 4 details correlations between the most significant maxima observed in the global analysis for the different subscores and each of the other SEA subscores. These correlations show that subregions associated with a given SEA subscore were not significantly associated with the other SEA subscores. This finding suggests that each SEA subtest may have specific cerebral correlates in FTD patients.
Table 4 Coefficient of correlations between SEA raw subscores and individual MRI gray matter values from ROIs centered on the most significant maxima observed in the global analysis
Note. Results in bold indicate the subtest that allowed identifying each of the ROIs in the global analysis. * Indicates bilateral significance (p<.05).
SEA = Social cognition & Emotional Assessment; MRI = magnetic resonance imaging; ROI = region of interest; BA = Brodmann area; mPFC = medial prefrontal cortex; OFC = Orbito-frontal cortex; G = gyrus; Lat. = lateral; Med. = medial; Sup. = superior; Inf. = inferior; Mid. = middle.
Discussion
The aim of this study was to explore the cerebral correlates of deficits from SEA subtests that assess social and emotional functions in bvFTD patients. The main findings are summarized in Figure 4. Using voxel-based morphometry, we found correlations between several subscores of the SEA and gray matter within the mPFC and OFC. The number of errors committed in the reversal-learning task predicted the degree of atrophy (GM volume reduction) in the medial OFC/mPFC (BA 11), lateral OFC (BA 47), and in anterior cingulated cortex (BA 24) at a corrected threshold. The “explanation of faux-pas” was associated with GM volume in the medial BA 10 at a corrected threshold and in the frontoinsular cortex in bilateral BA 13 at a more permissive threshold. Uncorrected statistical maps also show a relationship between the “emotion recognition score” and gray matter volume in medial BA 9 and BA 8, between the “apathy scale” and the lateral PFC (BA 44/45), and between the behavioral-control test and small clusters in BA 6 and 8/9.
Fig. 4 Correlations between gray matter volume and the faux-as test (red), the reversal-learning test (green) and the emotional recognition test (purple) within the left medial prefrontal cortex (mPFC; left; X = −6) and the right mPFC (right; X = 5). Statistical maps are superimposed on a canonical brain using MRIcro.
Neuroimaging Correlations with Theory of Mind Evaluation
The ability to explain social faux-pas in the faux-pas test significantly predicted gray matter volume in the left mPFC (medial BA 10; region A). The link between this particular area within the prefrontal cortex and theory of mind have already been discussed by Adenzato et al. (Reference Adenzato, Cavallo and Enrici2010), but to the best of our knowledge, this is the first time that this link has been shown in bvFTD using the faux-pas test. The critical involvement of the mPFC—especially of the medial BA 10—in theory of mind tasks was, however, previously shown in two lesion studies (Roca et al., Reference Roca, Torralva, Gleichgerrcht, Woolgar, Thompson, Duncan and Manes2011; Shamay-Tsoory, Tomer, Berger, Goldsher, & Aharon-Peretz, Reference Shamay-Tsoory, Tomer, Berger, Goldsher and Aharon-Peretz2005).
The mPFC has been reported to be involved in 93% of the functional neuroimaging studies exploring theory of mind (Carrington & Bailey, Reference Carrington and Bailey2009), including either the recognition of other's mental state task (Baron-Cohen et al., Reference Baron-Cohen, Ring, Moriarty, Schmitz, Costa and Ell1994), verbal stories (Vogeley et al., Reference Vogeley, Bussfeld, Newen, Herrmann, Happe, Falkai and Zilles2001), or cartoon stories (Gallagher et al., Reference Gallagher, Happe, Brunswick, Fletcher, Frith and Frith2000). A meta-analysis focusing on BA 10 function, conducted by Gilbert et al. (Reference Gilbert, Spengler, Simons, Steele, Lawrie, Frith and Burgess2006), showed that the medial part of BA 10 seems to be a key cerebral area for mentalizing (i.e., reflecting on the mental states of another agent). The activity of BA 10 has been proposed to reflect self-referential mental processes and a more general mechanism for integrating social information that allows for a representation of norms, mental state, and intentionality (Amodio & Frith, Reference Amodio and Frith2006; Gilbert et al., Reference Gilbert, Spengler, Simons, Steele, Lawrie, Frith and Burgess2006; van Overwalle, Reference Van Overwalle2009). In another recent meta-analysis, Gilbert, Gonen-Yaacovi, Benoit, Volle, and Burgess (Reference Gilbert, Gonen-Yaacovi, Benoit, Volle and Burgess2010) explored distant co-activation with BA 10 in functional imaging studies. These studies showed that regions that were co-activated with medial BA 10 depended, in some cases, on the type of task being performed, especially during mentalizing tasks. For example, dorsolateral PFC activation was associated with medial BA 10 activation in mentalizing tasks, but with the lateral BA 10 in other tasks. Thus, the lateral PFC may be involved in mentalizing tasks, together with the medial BA 10, which would be consistent with the association of the faux-pas task and BA 45 observed in the current study.
At a lower (uncorrected) threshold, we found two additional regions associated with the “explanation of faux-pas” score: the premotor (BA 6) and frontoinsular cortices (BA 13; region B). BA6, which is strongly connected to the mPFC (Amodio & Frith, Reference Amodio and Frith2006), is considered by several authors to be important for the emotional and empathic-substrates model called the “motor theory of empathy” (Leslie, Johnson-Frey, & Grafton, Reference Leslie, Johnson-Frey and Grafton2004; Zaki, Weber, Bolger, & Ochsner, Reference Zaki, Weber, Bolger and Ochsner2009). This model emphasizes that the premotor cortex is involved in the process of recognizing an emotional expression by simulating the observed emotion in one's own emotional circuitry. The frontoinsular (FI) region also seems to play a role in mentalizing in bvFTD patients (Lamm, Batson, & Decety, Reference Lamm, Batson and Decety2007). According to a recent theory, FI degeneration in bvFTD patients may impair their ability to represent the emotional impact of their actions (Seeley, Reference Seeley2010). The insula has also been reported to be involved in representation of others and self-visceral states, and the anterior insula has been linked to a conscious introspection of this state (Keysers & Gazzola, Reference Keysers and Gazzola2007; Lutz, Greischar, Perlman, & Davidson, Reference Lutz, Greischar, Perlman and Davidson2009). In this context, the FI region would be involved in detecting and explaining social faux-pas by identifying the visceral state caused by shame, anger, regret, … in “victims” of faux-pas. This is concordant with our current findings that suggest that FI atrophy may lead to the difficulties that bvFTD patients show in accessing these representations. More generally, previous studies have shown a correlation between FI atrophy and behavioral disturbances in bvFTD (Rosen et al., Reference Rosen, Allison, Schauer, Gorno-Tempini, Weiner and Miller2005; Viskontas, Possin, & Miller, Reference Viskontas, Possin and Miller2007), and it has been reported that neurons such as Von Economo neurons within the FI cortex (as well as in the anterior cingulate cortex) may be particularly vulnerable to degeneration in bvFTD and may be related to the early impairments in social functioning observed in these patients (Allman et al., Reference Allman, Tetreault, Hakeem, Manaye, Semendeferi, Erwin and Hof2011; Seeley et al., Reference Seeley, Crawford, Rascovsky, Kramer, Weiner, Miller and Gorno-Tempini2008; Seeley, Zhou, & Kim, Reference Seeley, Zhou and Kim2011).
Neuroimaging Correlations with Reversal-Learning and Behavioral-Control Tests
We found a relationship between the number of errors committed before reversing a learned rule in the reversal-learning test and GM volume in the medial OFC/ventral mPFC (BA 11; region C), extending to the lateral OFC (BA 47), and in the anterior cingulate cortex (BA 24; region D). Additional regions were also found at a lower threshold in the posterior portion of the middle cingulate cortex (BA 23/31) and in the lateral premotor cortex (BA 6).
The (negative) correlation between GM volume in the ventral mPFC and OFC and the number of reversal errors is highly concordant with previous results in lesion and functional imaging studies. This is the first time that this link has been specifically examined and observed in bvFTD patients. It is, however, well established that patients with ventral mPFC damage show strong deficits in tasks requiring a behavioral change after modification of environmental reinforcement contingencies (Hornak et al., Reference Hornak, O'Doherty, Bramham, Rolls, Morris, Bullock and Polkey2004; Fellows, Reference Fellows2007; Rolls et al., Reference Rolls, Hornak, Wade and McGrath1994). The ventral mPFC appears to play a critical role in binding an environmental stimulus with a given outcome and in rapidly changing this association when the stimulus is no longer rewarding (Fellows, Reference Fellows2007; Kringelbach & Rolls, Reference Kringelbach and Rolls2004; Wheeler & Fellows, Reference Wheeler and Fellows2008). Patients with ventral mPFC damage, including bvFTD patients, tend to perseverate on a previously rewarded response and fail to switch to another response after an unexpected negative outcome. Such impairment has been specifically linked to lesions in the ventral mPFC area (BA 11) in patient studies (Fellows & Farrah, 2003; Tsushida et al., 2010), while functional neuroimaging data also support the role of the lateral OFC in the reversal process (Zald & Andreotti, Reference Zald and Andreotti2010).
Evidence from functional neuroimaging suggests that the medial and lateral portions of the ventral PFC region contribute differently to reversal learning and behavioral adaptation by encoding predicted reward values of a stimulus (medial BA 11), maintaining a representation of the predicted reward value (medial BA 11 and lateral BA 47), orienting future behavioral choices (BA 11/47), and suppressing previously rewarded responses (lateral BA 47) (Elliott, Dolan, & Frith, Reference Elliott, Dolan and Frith2000; O'Doherty & Dolan, Reference O'Doherty and Dolan2006). In reinforcement-learning studies, a functional segregation within the OFC according to the valence of reward has been proposed. For example, the medial OFC (BA 11) may be involved in positive rewarding outcomes, while the lateral OFC (BA 47) may be involved in punishing/negative outcomes (Cools, Clark, Owen, & Robbins, Reference Cools, Clark, Owen and Robbins2002; Mitchell, Rhodes, Pine, & Blair, Reference Mitchell, Rhodes, Pine and Blair2008; O'Doherty, Kringelbach, Rolls, Hornak, & Andrews, Reference O'Doherty, Kringelbach, Rolls, Hornak and Andrews2001). In another branch of research, the right BA 47 has been associated with the suppression of a prepotent or overlearned rule or action. Using go/no-go tasks, the Hayling suppression task, the Stroop task, or neuropsychiatric scales, both patient studies (Aron, Fletcher, Bullmore, Sahakian, & Robbins, Reference Aron, Fletcher, Bullmore, Sahakian and Robbins2003; Hornberger et al., Reference Hornberger, Geng and Hodges2011; Nakano et al., Reference Nakano, Asada, Yamashita, Kitamura, Matsuda, Hirai and Yamada2006; Picton et al., Reference Picton, Stuss, Alexander, Shallice, Binns and Gillingham2007) and functional imaging studies (Goya-Maldonado et al., Reference Goya-Maldonado, Walther, Simon, Stippich, Weisbrod and Kaiser2010; Walther, Goya-Maldonado, Stippich, Weisbrod, & Kaiser, Reference Walther, Goya-Maldonado, Stippich, Weisbrod and Kaiser2010) have identified the right BA 47 as a key region for response suppression.
In summary, our results suggest that the ventral mPFC (BA 11) is a critical cerebral area for behavioral adaptation to new contingencies. When capable of learning a rule, bvFTD patients presenting with atrophy in the ventral mPFC have difficulties suppressing a previously learned and rewarded behavior and/or performing an alternative behavior. Although the ventral prefrontal regions have been previously associated with reversal learning and decision making in many functional imaging studies and in some lesion studies, to the best of our knowledge, the anatomical correlates of the reversal deficit in bvFTD patients have not been explored previously.
The number of errors committed before reversing the rule in the reversal-learning test was also negatively correlated with GM volume in the anterior cingulate cortex. The anterior cingulate cortex has been shown to be involved in the evaluation of action outcomes, particularly when they are negative (Bush et al., 2002; Gehring & Willoughby, 2002), and in integrating the reward-related information about past actions to avoid negative events in future responses (Rushworth, Walton, Kennerley, & Bannerman, 2004). In reinforcement-learning tasks, the anterior cingulate cortex has been found to be especially involved during the reversal phase (Ghahremani, Monterosso, Jentsch, Bilder, & Poldrack, 2010; O'Doherty et al., Reference O'Doherty, Kringelbach, Rolls, Hornak and Andrews2001) and may accelerate and stabilize relearning via the inhibition of prior incorrect responses (Ghahremani et al., 2010). A similar region in the anterior cingulate cortex is also thought to have an important role in conflict monitoring and error detection (Beckmann et al., 2009; Botvinick, 2007; Rushworth, Behrens, Rudebeck, & Walton, 2007), particularly when a situation involves errors or requires overriding a prepotent response (Botvinick, Braver, Barch, Carter, & Cohen, 2001; Braver, Barch, Grey, Molfese, & Snyder, 2001). The exact role of this region in these processes is still debated, and the link between the anterior cingulate cortex functions in conflict monitoring and outcome evaluation is not fully understood (Botvinick, 2007). It is also not definitively established how the ventral mPFC/medial OFC and anterior cingulate cortex interact during reward-guided decision-making (Rushworth & Behrens, 2008; see also Kennerley, Behrens, Wallis, 2011).
We did not find the same correlations within the ventral prefrontal regions using the behavioral-control test, which is also based on reversal-learning processes. This finding may be due to an extreme interindividual variability in this test regardless of the disease (see Table 1), as suggested by the high variability of the performance observed in healthy controls for this particular test in Funkiewiez et al. (Reference Funkiewiez, Bertoux, de Souza, Levy and Dubois2012). At an uncorrected threshold, the reversal score in the behavioral-control test was associated with BA 6 (region G) within the lateral premotor cortex as well as with lateral BA 8/9 (region H) in a more dorsal region of the PFC. The negative correlation with BA 6 may reflect some difficulties in the selection of movement. As the premotor cortex is a key region for motor output, the meaning of this correlation is more difficult to interpret. During the behavioral-control task, a computerized test that requires touching the stimuli on a tactile screen to answer, some bvFTD patients presented repetitive or stereotyped patterns of answers that impaired performance. This type of abnormal motor behavior has been shown to be correlated with premotor cortex atrophy in dementia (Rosen et al., Reference Rosen, Allison, Schauer, Gorno-Tempini, Weiner and Miller2005) or to be related to a motivation decrease (Roesch & Olson, Reference Roesch and Olson2003), a deficit in movement programming (Mitz, Godschalk, & Wise Reference Mitz, Godschalk and Wise1991), or a motor switching impairment (Dove, Pollmann, Schubert, Wiggins, & von Cramon, Reference Dove, Pollmann, Schubert, Wiggins and von Cramon2000). Alternatively, the association between the behavioral-control test and the premotor BA 6 in addition to the caudal lateral PFC (BA 8/9) may reflect the need for cognitive control to perform this task. Cognitive control allows for the selection of appropriate stimulus-response associations in goal-directed behaviors. Recent models describe three hierarchical levels of cognitive control that depend on the amount and type of information required for action selection (Badre, Reference Badre2008; Koechlin, Ody, & Kouneiher, Reference Koechlin, Ody and Kouneiher2003; Koechlin & Summerfield, Reference Koechlin and Summerfield2007). The need for cognitive control in the behavioral-control task may explain the regions we observed (premotor and caudal lateral PFC), which are close to the regions described by Koechlin et al. (Reference Koechlin, Ody and Kouneiher2003).
Neuroimaging Correlations with Emotional Faces Identification
Performance on identification in the facial emotions recognition test was significantly related to GM volume in the right mPFC (BA 8 and 9; regions F and E). In brain-damaged patients, Hornak et al. (2003) have previously shown that surgical lesions of BA 9 were followed by alterations of the emotional face processing. Similarly, Heberlein et al. (Reference Heberlein, Padon, Gillihan, Farah and Fellows2008) and Shamay-Tsoory, Aharon-Peretz, and Perry (Reference Shamay-Tsoory, Aharon-Peretz and Perry2009) found that lesions of the medial or ventromedial portions of the PFC were associated with emotional recognition impairment.
Functional imaging studies have highlighted several different relationships between emotion processing and BA 9. This area has been shown to be activated when normal subjects are asked to identify a facial emotion or to attribute an emotion to a person (Lane et al., Reference Lane, Reiman, Bradley, Lang, Ahern, Davidson and Schwartz1997; Ochsner et al., Reference Ochsner, Knierim, Ludlow, Hanelin, Ramachandran, Glover and Mackey2004). It has also been found to be activated during emotional regulation (i.e., the reinterpretation of a stimulus that no longer elicits a negative or a positive response or the inhibition of the emotional reaction to this stimulus) (Blair et al., Reference Blair, Smith, Mitchell, Morton, Vythilingam, Pessoa and Blair2007; Peelen, Atkinson, & Vuilleumier, Reference Peelen, Atkinson and Vuilleumier2010). These results have led Peelen and colleagues (2010) to suggest that the mPFC, and BA 9 in particular, may be a region involved with supra-modal representations of emotion.
Neuroimaging Correlations With the Apathy Scale
We found a negative correlation between GM volume in the lateral PFC (BA 44/45; region I) and the severity of apathetic symptoms as rated by caregivers in the apathy scale. The mechanisms and neural correlates of apathy are poorly known. The few lesion studies that have been performed have associated apathy with the dorsal anterior cingulate cortex (Kumral, Bayulkem, Evyapan, & Yunten, Reference Kumral, Bayulkem, Evyapan and Yunten2002) or the caudate nucleus (Kumral, Evyapan, & Balkir, Reference Kumral, Evyapan and Balkir1999) and more generally with the lateral and/or ventral prefrontal–basal ganglia circuits (Levy & Dubois, Reference Levy and Dubois2006).
As a major behavioral symptom of the disease, neural correlates of apathy have been investigated in several studies in bvFTD through clinicometabolism or gray matter correlations. In these studies, the mPFC and lateral PFC were both found to be associated with the severity of apathy symptoms (Franceschi et al., Reference Franceschi, Anchisi, Pelati, Zuffi, Matarrese, Moresco and Perani2005; Massimo et al., Reference Massimo, Powers, Moore, Vesely, Avants, Gee and Grossman2009; Zamboni et al., Reference Zamboni, Huey, Krueger, Nichelli and Grafman2008). The correlation with the lateral PFC could reveal the “cognitive inertia”, described by Levy & Dubois (Reference Levy and Dubois2006), which results from difficulty in planning, generating, and executing goal-directed behaviors. Our patients had low scores in fluency and free recall, which could be related to cognitive inertia. However, the apathy scale is a heterogeneous and subjective measure, and this study does not allow for determining which mechanisms are responsible for apathy in our patients.
SEA and Mini-SEA Composite Score
No significant regional correlations between GM volume and the SEA or mini-SEA composite scores were found. This result may be explained by the fact that these two scores combine several subscores, assessing different cognitive functions likely associated with different cerebral areas. The ROI analyses showed that GM volume in each identified region was correlated with poor performance in only one given subtest. Although the specificity of each region was not statistically tested, our findings suggest a specialization within the mPFC for distinct social cognition, emotional processing and reversal-learning functions. However, given the small size of the group, we cannot exclude the possibility that the study lacked the power to show smaller effects, and conversely, we cannot rule out that some of the correlations reported at the uncorrected threshold may include false positives. In addition, we focused our analyses on a sample of bvFTD patients only because the aim of our study was not to describe the general atrophy profile of bvFTD compared to healthy subjects, but to look for correlations between SEA subscores and GM volume. Although our results are consistent with current knowledge in the field, further studies in larger groups and/or including other patient and control groups would be useful to confirm these results. Finally, in the absence of pathological or a genetic confirmation of the clinical diagnosis of bvFTD, we cannot exclude that some patients did not have a pathological diagnosis of FTD. Our study also included three patients older than 75 years, which is not within the typical age range for this disease. However, (1) pathological series have reported patients with bvFTD older than 80 years (Seilhean et al., Reference Seilhean, Le Ber, Sarazin, Lacomblez, Millecamps, Salachas and Duyckaerts2011), (2) some authors have reported that 5% of bvFTD patients were older than 75 years at their first hospital admission (Ibach et al., 2003), and (3) one study showed that the prevalence of bvFTD in 85-year-olds was 3% (Gislason, Sjogren, Larsson, & Skoog, Reference Gislason, Sjogren, Larsson and Skoog2003).
Conclusion
The aim of this study was to investigate the neural correlates of SEA subtests in bvFTD. These subtests have been shown to be sensitive to social and emotional cognitive impairments in bvFTD and could be used as an objective measure of this disease's behavioral dysfunctions. The specificity of the SEA and its reduced version (the mini-SEA) for the diagnosis of bvFTD was demonstrated in previous studies, even at an early stage of the disease (Bertoux et al., Reference Bertoux, Delavest, de Souza, Funkiewiez, Lepine, Fossati and Sarazin2012; Funkiewiez et al., Reference Funkiewiez, Bertoux, de Souza, Levy and Dubois2012). In the current study, we showed a strong relationship between the faux-pas test and the rostral mPFC, while reversal-learning was mainly related to more inferior regions within the medial and lateral ventral prefrontal areas. We also found a relationship between emotion recognition and the dorsal and rostral mPFC. It appears that the mPFC is a heterogeneous region in which separate subregions are critical for distinct aspects of social and/or affective processing.
It is of clinical interest to consider that the two most sensitive and specific tests of the SEA—the emotional recognition and the faux-pas tests—which were chosen to compose the mini-SEA, covaried with GM volume in the rostral mPFC. These findings provide anatomical support for the use of the mini-SEA to evaluate mPFC integrity in bvFTD, the core region affected by bvFTD (Agosta et al., Reference Agosta, Canu, Sarro, Comi and Filippi2012; Salmon et al., Reference Salmon, Garraux, Delbeuck, Collette, Kalbe, Zuendorf and Herholz2003; Schroeter et al., Reference Schroeter, Raczka, Neumann and von Cramon2007, Reference Schroeter, Raczka, Neumann and Yves von Cramon2008). The findings from these tests also highlight the need for a clinical consensus on imaging investigation of the mPFC or on clinical tests tapping into this region (Rascovsky et al., Reference Rascovsky, Hodges, Knopman, Mendez, Kramer, Grossman and Miller2012; Schroeter, Reference Schroeter2012).
Finally, the results of this study support the fact that social, emotional, and reversal-learning assessments provide good insights into mPFC and OFC functioning in bvFTD and suggest that the SEA may be helpful for the evaluation of these functions in other pathologies affecting the PFC, including vascular or tumoral damage.
Acknowledgments
E.V. was supported by the “Agence Nationale de la Recherche” [grant number ANR-09-RPDOC-004-01]. M.B. was supported by the “Centre National de la Recherche Scientifique” and the French ministry of defense (CNRS – DGA). The authors would like to thank Goulven Josse and Isabelle Le Ber for their helpful comments and suggestions.
Disclosures
Maxime L Bertoux reports no conflicts of interests, no financial interest and no disclosures. Emmanunelle Volle reports no conflicts of interests, no financial interest and no disclosures. Aurélie Funkiewiez reports no conflicts of interests, no financial interest and no disclosures. Leonardo Cruz de Souza reports no conflicts of interests and no financial interest. He received speaker honoraria from Lundbeck. Delphine Leclercq reports no conflicts of interests, no financial interest and no disclosures. Bruno Dubois reports no conflicts of interests and no financial interest. He has consulted or served on advisory board for Bristol-Myers Squibb, Roche, Elan, Eli Lilly, Eisai, Janssen. His institution has received grants from Novartis and Sanofi-Aventis.
