To the best of our knowledge, this is the first family to be described and carrying the MAPT p.K298E mutation. Notably, we describe here the variable clinical presentation and progression with different degrees of severity and neuroimaging appearance, expanding the genotype–phenotype correlations of this mutation. We hypothesize that the MAPT p.K298E is associated with early onset (the average onset age was around 57 years old considering all described cases) and a behavioral phenotype (bvFTD), consistent with the predominant phenotype and demographic features of other MAPT mutations [8, 12]. Apathy was a common initial behavioral manifestation, predating cognitive impairment. This is in line with a recent study in genetic FTD, showing that apathy is present even in presymptomatic MAPT mutation carriers, and is followed by cognitive impairment (but not vice versa) [13].

Interestingly, an initially less prominent speech impairment was present in almost all our patients (anomia, semantic, and phonemic paraphasia), while the case described by Iovino was diagnosed with progressive non-fluent aphasia [3]. Semantic dementia, albeit with behavioral features, is also a possible phenotype of MAPT mutations [8]. It should be noticed that an isolated presentation of semantic dementia in MAPT mutations is quite rare, and limited to sporadic case reports [14], and most commonly semantic deficits co-occur with a behavioral phenotype due to a common tropism of MAPT mutations. Analogously, non-fluent aphasia phenotypes have been described with exon 10 MAPT mutations [15]. In a GENFI study, 80% of MAPT-related FTD exhibited language impairment, with various patterns [16], and naming performances were the worst among genetic FTD forms. Correlates of anomia in MAPT mutation carriers were found in the anterior temporal lobes and anterior insula [17]. The described phenotype is probably not unique to MAPT p.K298E mutation, but rather typical of MAPT mutations resulting in a preferential increase of tau 4R isoforms. Indeed, it seems that MAPT mutations resulting in predominant tau 4R isoforms present with linguistic features more often than mutations leading to predominant 3R isoforms, as shown by a recent systematic review [18]. This is also supported by neuropathological data showing a higher burden of tau lesions in case of predominant 4R isoforms compared to predominant 3R isoforms, with distinct tropism for different cerebral regions [19].

Clinical follow-up revealed motor symptoms development in all patients, usually contralateral to the most affected side at neuroimaging. Patients presented with bradykinesia, limb weakness, gait instability, and, in one case, distal dyskinesia; muscle ultrasound performed in our center did not show significant fasciculations. Elements of parkinsonism and basal ganglia degeneration were noted also in the case of MAPT p.K298E mutation described by Iovino in 2014 (notably, with a normal DaT-Scan). While extensive phosphorylated tau was reported, also affecting the caudate nucleus and the substantia nigra (which seems typical for mutations in exon 10 of MAPT [8, 20]), no α-synuclein pathology was present in her brain [3]. Moreover, extrapyramidal symptoms have been described in the context of MAPT mutations affecting exon 10 splicing [8], and an abnormal striatal tau isoform content with an excess of 4R isoforms seems responsible for clinical parkinsonism in murine models of tauopathy resulting from altered exon 10 splicing [21]. Therefore, the parkinsonian features also present in our cases might arise from a tau-related degeneration of key extrapyramidal regions.

Motor symptoms in our family were not limited to extrapyramidal features, but upper and lower motor signs were not sufficient to pose a diagnosis of amyotrophic lateral sclerosis (ALS) in any case [22, 23]. Indeed, a constellation of nonspecific motor signs like bradykinesia and ALS-like features such as spasticity, limb weakness, and fasciculations are present in around 40% of MAPT mutation carriers, although less common than the Parkinson’s-like motor phenotype [24].

In two cases, mild memory impairment was noticed. This is also consistent with previous literature showing impaired performance on memory tasks even in presymptomatic MAPT mutation carriers [25].

Brain MRI showed asymmetrical frontotemporal atrophy and the metabolic PET confirmed asymmetrical frontotemporal involvement, with greater right involvement in two patients and left-side predominance in the other two patients. Indeed, MAPT mutations affecting exon 10 splicing seem to be predominantly associated with structural and metabolic medial temporal lobe involvement [8]. In the aforementioned GENFI cohort, the temporal pole was atrophic in up to 70% of MAPT mutation carriers, significantly differing from other mutation groups [16]. Temporal lobe volumes show the fastest decline over time even in presymptomatic MAPT mutation carriers [26], and it was even more accelerated in those who converted to dementia, who also showed frontal and parietal longitudinal atrophy [27]. Mesial temporal gray matter atrophy may be present in presymptomatic carriers as early as their thirties [28]. However, the pattern of atrophy seems to be dependent on the specific MAPT mutation, with distinct clinic-radiological peculiarities [29].

It would be interesting to speculate on the role of SGSH heterozygosity as a potential risk factor for neurodegeneration, as is the case for mutations in other genes involved in lysosome storage disorder, such as those in glucosaminidase (GBA) and Parkinson’s disease. SGSH mutations are responsible for mucopolysaccharidosis type IIIA, the most common form of Sanfilippo syndrome, an autosomal recessive disorder of lysosome storage with onset in childhood [30]. It results in developmental delay arising between 2 and 6 years of age, with subsequent cognitive decline, sleep disorders, hyperactivity, and aggressive behavior, and finally, loss of mobility, swallowing troubles, and epilepsy, leading to premature death [31, 32]. The p.L146P is a missense mutation previously described only in one Italian patient, drastically changing the protein structure, and leading to a severe phenotype [33]. Neurodegeneration in mucopolysaccharidosis III has been demonstrated independently of the specific genetic mutation [34,35,36], being the result of axonal dystrophy with an accumulation of ubiquitin-positive lesions containing phosphorylated-tau, wildtype α-synuclein, APP and β-amyloid, as shown by clinical and preclinical evidence [37,38,39]. However, mice carrying heterozygous p.D31N mutation in SGSH did not show an increased presence of brain pathology or neurodegeneration compared to controls, nor a significant accumulation of heparan sulfate despite a 50% reduction in SGSH levels. Nevertheless, they exhibited significantly worse performances on the negative geotaxis test and significant increase in the length of the outer-most dendritic processes with aging [40]. While this would support the idea that p.D31N SGSH heterozygosity does not overtly hasten age-related neurological decline, it is possible that such a decline, if indeed present, might be mutation-specific. Regrettably, we do not have enough data to evaluate whether heterozygous carriers of p.L146P mutation exhibit worse neurological outcomes. Hence, despite the speculations, the most plausible explanation for the simultaneous presence of the SGSH p.L146P heterozygous mutation and MAPT p.K298E mutations in the proband is likely mere coincidence.

Our report has some limitations. Pathological confirmation was not available, although almost all the cases satisfy the criteria for bvFTD [11], with the exception of case II-8, who was most likely affected by rtvFTD. Analogously, we could not explore brain tau accumulation in vivo with tau-PET, which shows preclinical alterations in other MAPT mutations [41], as this was not available at our institution. Moreover, given either the advanced stage of dementia at presentation or the refusal of subjects involved, it was not possible to explore CSF or blood biomarkers in this family.

In conclusion, the p.K298E mutation belongs to a small group of exon 10 mutations and it is responsible for early onset bvFTD with a variegated phenotype, including language and motor symptoms, with predominant fronto-temporal hypometabolism. At present, it is necessary to implement our knowledge about MAPT p.K298E to better clarify its clinical and epidemiological role. Continuous follow-up of the family described in this paper, and the identification of other families, will help increase our understanding of MAPT exon 10 mutations.