Sociodemographic and clinical characteristics of groups are reported in Table 1. Groups were comparable in terms of age at MRI, sex, handedness, and disease duration. Age of onset was similar among groups of patients, with an average of 60.7 years (SD 8.1). The average time elapsed between symptoms onset and MRI was 44 months (SD 32.3). Education was significantly higher in svPPA compared to bvFTD. As for disease severity, bvFTD patients scored higher compared to other groups at CDR-SB. Of note, 5 out of 15 patients (33%) classified as sbvFTD [7] formally met Rascovsky criteria [1] for bvFTD as well.

As outlined in Fig. 1, out of 14 sbvFTD patients for whom genetic analysis was available, 3 (21%) presented pathogenic variants, respectively, in the C9orf72 (N = 1), GRN (N = 1), and MAPT gene (N = 1). For what concerns bvFTD, out of 58 patients with genetic analysis available, a total of 19 patients (33%) presented pathogenic variants: C9orf72 (N = 4), GRN (N = 12), TREM2 (N = 1), FUS gene (N = 1), or in both C9orf72 and MAPT genes (double mutation, N = 1). Only 1 svPPA patient out of 17 with genetic analysis available (6%) carried a known pathogenic mutation in the MAPT gene. Supplementary Table 3 reports the details of each clinical variant identified in our cohort.

Supplementary Table 4 summarizes the first three symptoms developed by each sbvFTD patient as reported by caregivers and/or patients themselves. The most frequent earliest complaints experienced by patients included criteria-specific symptoms: word and object semantic loss (67%), person-specific semantic knowledge loss (67%), and complex compulsions and rigid though process (60%). Another 47% of patients reported experiencing apathy/inertia, 27% presented simple repetitive behaviors, hoarding or obsessions, as well as loss of empathy. Only 20% of patients described episodic memory loss and, eventually, 13% of patients reported hyperorality or dietary changes. Also, 27% of patients experienced early “extra-criteria” symptoms (i.e., non-criteria-specific). The main feature experienced by those patients was anxiety, with the tendency of suspiciousness and in general easy irritability. No patient had early lack of judgement and dysexecutive symptoms or disinhibition.

As the disease progressed, the number of patients experiencing general behavioral symptoms as well as episodic memory loss increased. Indeed, at the time of the visit, recollecting symptoms from disease onset up to referral to our center, up to 80% of patients had developed apathy/inertia, as well as complex compulsions and rigid though process, 73% words and objects semantic loss, 53% presented disinhibition and loss of empathy, 50% developed simple repetitive behaviors, hoarding and obsessions and episodic memory loss, 51% developed hyperorality or dietary changes, 27% developed lack of judgement or dysexecutive symptoms, while 47% developed other symptoms (all criteria-specific symptoms). The percentage of patients experiencing “extra-criteria” symptoms increased from 27 to 67%. No patient developed motor neuron disease signs. A visual representation of symptom unfolding is provided in Fig. 2.

Symptoms developed by semantic behavioral FTD (sbvFTD) patients. Spider chart depicting the number of sbvFTD patients affected by a given symptom. Blue line represents the first three symptoms reported by caregivers/patients; orange line represents all symptoms reported from disease onset to the time of the visit. “Extra-criteria” symptoms = anxiety, suspiciousness, agitation, irritability or depression

To better characterize the type of symptoms referred, we asked caregivers to provide real-life examples. Patients complaining complex compulsions presented with restricted food preference (e.g., only ate bread and pasta; healthy food; pre-cooked food; sophisticated food; cold dishes and fruit), had fixed daily routine (i.e., must set up the table 3 h before diner, had to spend exactly 1 h a day at the bakery shop), had the tendency to compulsively look for a job after having lost it, could not let the phone battery go below 80%, or experienced psychiatric symptoms such as potomania or opiate abuse. As for person-specific semantic knowledge loss, the main symptoms reported were the inability to recognize either famous or familiar faces (colleagues, relatives) or recalling proper names. Among other “extra-criteria” behavioral derangements were the inability to understand jokes, anxiety, preoccupation, histrionic-like behavior, and irritability.

Table 2 describes the main neuropsychological features of our cohort. sbvFTD patients were more preserved in terms of verbal working memory and selective attention (attentive matrices) compared to both bvFTD and svPPA, limb ideomotor praxis (Goldenberg’s test) and affect discrimination (CATS, affect discrimination) compared to bvFTD, global cognition (MMSE) and naming compared to svPPA.

bvFTD group had greater dietary changes and poorer performances at problem solving (MCST), left limb ideomotor praxis (Goldenberg’s test), identity discrimination (CATS, face discrimination), and complex visuospatial constructive abilities (Benson’s figure copy) compared to both sbvFTD and svPPA, and at affect discrimination compared to svPPA (CATS, affect discrimination-three faces). Patients with svPPA were more compromised at semantic fluency compared to both sbvFTD and bvFTD groups, also performing poorer at tests of ideational apraxia compared to bvFTD.

Supplementary Table 5 describes the groups of patients in terms of frequency of pathological performances (based on Italian norms) at each neuropsychological test. Of note, no sbvFTD patient had pathological scores at selective attention (attentive matrices) and showed lower frequency of pathological scores at visual naming and semantic fluency compared to svPPA. Compared to sbvFTD and svPPA patients, bvFTD cases showed higher percentages of pathological scores in tests assessing complex visuospatial constructive abilities (Benson’s figure copy), problem solving and perseverations (MCST), and face discrimination.

Qualitatively, 100% of sbvFTD patients failed in test assessing famous face recognition and affect selecting, and more than 80% failed in tests assessing word comprehension, written metaphor, and humor comprehension subtests of the BLED. On the other hand, no one (0%) failed in tests assessing set-shifting (TMB B, B-A), ideomotor apraxia with the imitation of non-meaning gestures (Goldenberg’ test), and affect discrimination.

As shown by Fig. 3 and Supplementary Table 6, different patterns of atrophy emerged at VBM when comparing group of patients to controls. BvFTD patients (Fig. 3A) showed a pattern of widespread bilateral frontotemporal atrophy, which extended also to occipito-parietal areas and basal ganglia. svPPA patients (Fig. 3B) presented lateralized volume loss in temporal regions, mainly on the left hemisphere, with a relative sparing of frontal cortices but a tendency to spread toward ipsilateral middle frontal gyrus and medial temporal areas. sbvFTD patients (Fig. 3C) presented a mixed phenotype of bilateral ATL atrophy, mainly on the right side with some extension to the ipsilateral cingulum, fusiform, and supramarginal gyrus. Evaluating the patterns of atrophy observed in svPPA and sbvFTD patients, as compared with healthy controls (Fig. 3D), an almost specular pattern of atrophy was found, with rightward regions more affected in sbvFTD and leftward areas more atrophic in svPPA. Still, overlapping areas emerge in ATL, bilaterally. When comparing bvFTD and svPPA patients to sbvFTD, the latter resulted more atrophic in an extensive cluster involving the right ATL (Fig. 4, Supplementary Table 6). Compared to sbvFTD, bvFTD were more atrophic in the left precentral and caudate areas, whereas svPPA showed greater atrophy in the left middle temporal gyrus (Fig. 4, Supplementary Table 6).

Patterns of gray matter atrophy in patients with behavioral variant frontotemporal dementia (bvFTD), semantic variant primary progressive aphasia (svPPA), and semantic behavioral frontotemporal dementia (sbvFTD). Results of voxel-based morphometry analysis showing regions of significant GM atrophy in bvFTD patients compared to healthy controls (HC) (A), svPPA compared to HC (B), and sbvFTD compared to HC (C). Composite image showing svPPA and sbvFTD patterns of atrophy, combined with green areas showing regions of overlapping volume loss (D). Significant clusters are overlaid on sections of the Montreal Neurologic Institute standard brain. Analyses were corrected for age, sex, and total intracranial volume. Statistical threshold for significance was p < 0.05, family-wise error corrected for multiple comparisons. bvFTD behavioral variant frontotemporal dementia, sbvFTD semantic behavioral variant frontotemporal dementia, svPPA semantic variant primary progressive aphasia

Patterns of gray matter atrophy in patients with semantic behavioral variant frontotemporal dementia (sbvFTD) compared to behavioral variant frontotemporal dementia (bvFTD) and semantic variant primary progressive aphasia (svPPA). Results of voxel-based morphometry showing regions of significant GM atrophy in bvFTD compared to sbvFTD (A), sbvFTD compared to bvFTD (B), svPPA compared to sbvFTD (C), sbvFTD compared to svPPA (D). Significant clusters are overlaid on sections of the Montreal Neurologic Institute standard brain. Analyses were corrected for age, sex, and total intracranial volume. Statistical threshold for significance was p < 0.05, family-wise error corrected for multiple comparisons. For uncorrected results, threshold of significance was p < 0.001. bvFTD behavioral variant frontotemporal dementia, sbvFTD semantic behavioral variant frontotemporal dementia, svPPA semantic variant primary progressive aphasia