**Table 3.** Results of neuropsychological assessment.

### 3.1.2. Speech

All but one individual demonstrated impaired articulation, characterized by imprecise production of consonants (7/8) and vowels (1/8). Only one child had acquired all speech sounds expected for his age. Most absent consonants were vibrant [r] (7/8), followed by voiced palatal lateral approximant [L] (6/8, fricatives [z], [s] (2/8) and affricates [dz], [ţ] (2/8), both voiced and voiceless. Phonological planning was strongly impaired in all children but one, who showed a mild impairment (mean z-score −15.24, SD 9.52; range −31.5 −1.29). All patients showed a high percentage of phonetically inaccurate production of words.

We identified simplification processes (backing 3/8; fronting 4/8; stopping 3/8; epenthesis 6/8; metathesis 2/8; cluster reduction 5/8; gliding 1/8; voicing2/8; de-voicing 3/8; affrication 2/8; affrication 3/8; assimilation 4/8; diphthong reduction 2/8) and atypical processes (stops deletion/reduction in clusters 3/8; conflicting processes 3/8; idiosyncratic processes 4/8). Speech was typified by the imprecise articulation of consonants and vowels, abnormal nasal resonance, low pitch, and prosodic errors (e.g., excessive stress on unstressed parts of speech, slow rate, short phrases). Evaluation of Diadochokinesis rate was significantly slower in 3/7 patients, mildly impaired in 1/7 (mean z-score 0.7, SD 3.48; range −7.86 0.8). Conversation speech intelligibility was adequate in seven out of eight patients. See Table 3.

### 3.1.3. Language

Lexical abilities were preserved in 6/8 patients, both in receptive (mean standard score 89.63, SD 13.42; range 75–113) and expressive vocabulary (mean z-score −0.91, SD 2.01; range −5.45 0.77). Receptive vocabulary resulted in the normal borderline range in 2/8 patients. Expressive vocabulary was severely impaired in one patient, moderately impaired in two patients, and normal in five patients. Receptive grammar was the most impaired domain (mean scaled score 4.75, SD 2.81; range 1–8): a severe impairment was seen in 4/8 patients, two patients showed a borderline normal score, and 2/8 patients obtained a normal score. Expressive grammar (mean scale score 7.37, SD 3.42; range 1–12) was found impaired just in 1/8 patients, 3/8 performed borderline normal and 4/10 normal. See Table 3.

### 3.1.4. Intelligence Quotient

Full-scale IQ (FSIQ) scores varied from moderately impaired to normal (range 41–96; median 73.38; SD 15.55). Verbal Comprehension Index (VCI) was more conserved compared to the remaining sub-IQ, with a median score of 85 (SD 11.11; range 66–98). See Table 3.

### *3.2. Correlations*

The results of the Spearman correlation indicated that there was a significant negative association between age at the babbling onset and phonemic fluency tasks' score and phonemic inventory. Age at the babbling onset was inversely related to phonemic fluency task and the number of consonants acquired: the greater the delay in babbling's onset, the smaller the number of words produced at the phonemic fluency task [ρ(5) = −0.89, *p* = 0.042] and the number of consonants acquired [ρ(7) = −0.90, *p* = 0.005]. Age at the first word onset significantly correlated with phonemic fluency tasks' score [ρ(5) = −0.91, *p* = 0.03] and phonemic inventory [ρ(8) = −0.71, *p* = 0.049]. Children who produced their first word earlier had a better performance in phonemic fluency tasks [ρ(5) = −0.91, *p* = 0.03] and they had a greater number of stable consonants [ρ(8) = −0.71, *p* = 0.049]. See Table 3.

Moreover, we attempted a correlation between the severity of language evaluations (lower scores in language and oromotor assessments) and genotype. Due to the small sample, we could not find a significance; in particular, as reported in literature about clinical phenotype severity [7], patients with deletions (*n* = 2) or truncating mutation (*n* = 1) do not seem to have a more impaired profile compared to patients with missense mutations (*n* = 5).

The same correlation was attempted between language evaluation and CSF/serum glucose ratio without significative results due to the small numbers of patients (*n* = 6) and the small range of values (mean 0.35, range 0.27 −0.043).

It has not been feasible to search for different functioning trajectories according to age of KDTs initiation. Nevertheless, in our sample, patients who started KDTs after six years of age achieved lower scores in language assessment (*n* = 4, mean 100.25 months, range 88–121 months). Regarding oromotor skills, we did not observe the same trend.

### **4. Discussion**

Speech and language impairment have already been recognized in patients with GLUT1DS, but have not been fully characterized compared to other disease symptoms. In available studies, language functioning in GLUT1DS is depicted as extremely variable, ranging from no apparent deficit to the absence of expressive speech, with most affected individuals having reduced language skills [22,23]. In the present study, we deeply investigated speech and language profile in eight Italian-speaking children with GLUT1DS.

Based on parental reports, we documented a delay of early vocal behavior and early language milestones with a late onset of first word and combinatory speech in the majority of patients. We also found a significant negative association between babbling onset and the number of words produced in the phonemic fluency task and phonetic inventory. The delay in the mean age of babbling onset represents a crucial finding, since several studies support the predictive value of babbling onset timing and characteristics to determine subsequent speech and language abilities and communication disorders [24,25]. Babbling represents a linguistic and articulatory exercise and the experience of frequent selfproducing consonants and vowels syllables makes infants more aware of similar patterns in their environmental language, acting as potential building blocks for word representations [25]. Moreover, in our sample the age at the first word onset significantly correlated with phonemic fluency tasks' score and phonemic inventory, meaning that children who produced their first word earlier had a better performance in phonemic fluency tasks and a greater number of stable consonants. Importantly, it is often hypothesized that the first speech-like articulation and the babbling phase, which occur at approximately ten months of age, allow infants to develop a link between articulatory settings and the resulting auditory consequences, thus contributing to the development of the phonetic inventory and adaptation to the ambient language [26]. In this connection, the early signs of speech and language deviance and slow acquisition of expressive words in the second and third years of life may set off a cascade, negatively affecting a variety of following additional linguistic capabilities [24]. This scenario, which is frequently reported in cognitive and language disorders, has never been described as associated with GLUT1DS previously.

Oral-motor skills were impaired in most subjects in our sample. Development of orofacial praxis is impaired in a series of developmental disorders such as Developmental Coordination Disorder, Developmental Apraxia of Speech and Speech disorder [27]. These conditions have in common the combined presence of motor and language deficits, as observed in patients with GLUT1DS.

Speech was often characterized by phonetically inaccurate production of words, imprecise articulation of consonants and vowels, abnormal nasal resonance, low pitch, and prosodic errors. The most represented impairment was found in the phonological planning. This task resulted as severely deficient in seven out of eight patients, confirming the presence of a speech and language disorder, still active in some patients, and partially compensated in others. Receptive and expressive language abilities revealed different degrees of impairment in our patients; some of them showed severe receptive and expressive linguistic deficits, others had a mild impairment and only one had a normal profile. In all patients, a more conserved expressive and receptive lexical competence was observed, while linguistic grammar ability was impaired with a greater compromise of the receptive abilities. We may assume that a severe impairment at the morpho-syntactic level of lan-

guage organization could be interpreted as the less likely domain to recover in patients with a previous speech and language disorder, as observed in GLUT1DS patients.

Several reports describe a mild-to-severe intellectual disability of GLUT1DS patients, in most cases proportional to the disease's severity [6–8]. In our sample, FSIQ scores varied from moderately impaired to normal, one child showed a normal intelligence, five patients had a borderline intellectual functioning, two patients received a diagnosis of intellectual disability on mild and moderate ranges. VCI showed up as more conserved compared to the remaining sub-IQ: these data confirm the results of our previous work, where PRI was more affected than VCI [4]. A less impaired verbal quotient could lead at first to a misidentification of language deficits but, as shown by our results, an impairment of several linguistic domains can be documented with focused tests.

Due to the small number of patients included, it has not been feasible to obtain a phenotype-genotype and/or a phenotype/glycorrhachia correlation, as well as to search for different functioning trajectories according to age of KDTs initiation or total IQ level.

Nevertheless, in our sample, patients who started KDTs later in life (mean 8.5 years) achieved lower scores in language assessment and the patient with lowest IQ achieved one of the worst performances. Definitely larger samples are needed to assess whether KDTs initiation timing and mutation type might influence chances of recovery of speech and language. Unfortunately, in our sample KDTs introduction was late for all included patients.

Children with GLUT1DS are at a disadvantage in the development of cognitive functions since the disease itself causes a lower supply of energy for the correct functioning of the brain, resulting in a multilevel dysfunction affecting cognitive, speech and language abilities, as evidenced by the neuropsychological and language assessment carried in our sample [4]. Our data confirm the presence of a potentially heterogeneous cognitive and linguistic profile with different degrees of impairment in multiple speech and language areas. The variability of the linguistic profiles observed could be explained based on the general theoretical framework of neuroconstructivism [12].

This model is suitable to understand the interaction between biological and socioenvironmental factors determining the linguistic development of patients with GLUT1DS.

The neuroconstructivism approach highlights how tiny variations in the initial state could give rise to domain-specific differences in end states [12]. If brain energy requirements are not satisfied in the first years of life, an impairment of input processing and starting points such as language circuitry will occur. Variability of genetic mutations, adaptive strategies, successful behavior as well as intact domains leads to inter-individual outcome differences, that could explain the relative heterogeneity of language profile in our small sample. We did not find factors determining language outcome; nevertheless, we believe that focus must be placed in at-risk populations in early infancy, even before onset of language, and that this time window should represents the optimal timing to start therapy, namely KDTs.

Limitations of this study are represented by the small number of subjects included also due to the low prevalence of the disease—the age heterogeneity and the absence of a language and speech assessment before KDTs introduction.

### **5. Conclusions**

In conclusion, GLUT1DS can be considered a multilevel condition affecting cognitive, motor, speech, and language competences. Our results confirm the importance of a complete speech and language evaluation to obtain a detailed profile, that is crucial to plan early and specific rehabilitative intervention.

GLUT1DS patients are often diagnosed with aspecific language disorder or delay in the first years of life, before other symptoms manifest. In this scenario, recognizing typical and atypical language fragilities and searching for a common linguistic phenotype in these patients could help to guide early diagnosis. An early diagnosis of GLUT1DS would allow a prompt start of target dietary treatment and of rehabilitative intervention inclusive of speech and language training. Further studies are needed to evaluate the effects of KDTs on language function.

**Author Contributions:** Conceptualization, M.P.Z. and L.P. (Ludovica Pasca) and B.V.V. and A.F.; methodology, M.P.Z., L.P. (Ludovica Pasca), B.V.V., V.D.G.; formal analysis, S.G. and L.P. (Livio Provenzi); writing—original draft preparation, M.P.Z. and L.P. (Ludovica Pasca); writing—review and editing, L.P. (Ludovica Pasca) C.V., V.D.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Italian Ministry of Health RC 2020–2021.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by our Ethical Committee (P-20190033749), IRCCS Mondino Foundation, Pavia.

**Informed Consent Statement:** Informed consent was obtained from the parent of each subject and all children agreed to participate in the study.

**Acknowledgments:** Our thanks to the Italian association GLUT1DS ONLUS and the families. This work was supported by the Italian Ministry of Health RC 2017–2019.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Children* Editorial Office E-mail: children@mdpi.com www.mdpi.com/journal/children

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34

www.mdpi.com ISBN 978-3-0365-6840-9