According to the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition, Text Revision (DSM-5-TR), Autism Spectrum Disorder (ASD) is defined as a neurodevelopmental condition. It is characterized by deficits in social communication and interaction, as well as repetitive and restricted patterns of behavior and interests (e.g., repetitive body movements such as hand flapping, sensory sensitivities, and circumscribed interests) (American Psychiatric Association, 2022). The male sex is widely recognized as one of the most established etiological factors for ASD, with an estimated prevalence of 3.8 times higher among boys than girls (Maenner et al., 2023), leading to the concept of a “female protective effect,” where females may require a greater etiological burden to exhibit the same level of impact as males. The current global prevalence of ASD is estimated to be approximately 1% (Zeidan et al., 2022) and has significantly increased over the last 20 years (Baio et al., 2018). Research conducted over the past 4 decades has revealed not only the substantial variability of the underlying causal factors of autism but also that autism is one of the most heritable conditions (Rutter, 2011). Recent molecular genetic studies have led to the identification of an increasing number of genes associated with autism and have stimulated a growing body of literature on potential biological insights (Vorstman et al., 2017; Vorstman and Scherer, 2023; Vicari et al., 2019; Quesnel-Vallières et al., 2019; Savatt and Myers, 2021). There exists a category of monogenic disorders, such as Rett syndrome, tuberous sclerosis complex (TSC) and fragile X syndrome, which exhibit highly distinctive physical characteristics and may be associated with autism. These monogenic disorders are infrequent, and none of these genes can account for more than 1%–2% of ASD cases; however, collectively, they are estimated to represent about 5% of ASD cases (Sztainberg and Zoghbi, 2016). Some researchers classify these disorders as “syndromic autism”, distinguishing them from “nonsyndromic” autism based solely on clinical criteria. The term “syndromic” denotes conditions in which ASD occurs alongside additional phenotypes and/or dysmorphic traits. In contrast, “nonsyndromic” generally refers to “classic autism,” as characterized by Kanner, in which no additional symptoms are observed (Vorstman and Scherer, 2023). For the majority of nonsyndromic ASD cases, the underlying cause remains unknown, and the term “idiopathic autism” has been alternatively used, although a genetic component has also been acknowledged in these instances (Sandin et al., 2014). These discoveries have propelled research on genomic loci linked to genetic susceptibility for ASD (Weiss et al., 2009; Weiss et al., 2008) and the pursuit of rare and common variants that contribute to the risk of developing ASD (Anney et al., 2012). In the early 2000s, array Comparative Genomic Hybridization (CGH) technology enabled researchers to identify an enrichment of copy number variations, such as duplications and deletions in regions 15q11-13, 16p11.2, 22q11.2, and 7q11.23, in both syndromic and nonsyndromic autistic patients (Sanders et al., 2011). It is evident from molecular genetics studies that autism is not only clinically heterogeneous but also highly diverse in terms of genetic causation at the molecular level. The rare variants associated with autism, like common variants, are pleiotropic. Consequently, there is likely no definitive distinction between rare monogenic and common multifactorial autism (Vorstman and Scherer, 2023). Nevertheless, ASD cannot be entirely elucidated by genetics: the influence of environmental factors has been extensively investigated (Mandy and Lai, 2016), emphasizing that environmental risks and genetic predispositions are interconnected rather than independent factors, linked through gene-environment correlation (Rutter et al., 2006), which can be either passive or active (Thapar and Rutter, 2021).

Beyond global insights into ASD genetics, a growing body of research has investigated the genetic bases of speech and language impairment in autism. Several candidate genes implicated in language development have also been associated with ASD, most notably FOXP2 and related transcriptional networks influencing cortico-striatal and cerebellar circuits (Vargha-Khadem and Gadian, 2005; Fisher and Scharff, 2009). Mutations in genes such as CNTNAP2 and SHANK3 have been linked to both autism and severe speech/language phenotypes, including minimally verbal presentations (Alarcón et al., 2008; Durand et al., 2007). Importantly, animal model studies have provided converging evidence: mouse knockouts of Shank3, Cntnap2, and Foxp2 display atypical ultrasonic vocalizations, supporting their role in communication-related pathways (Ey et al., 2012; Peñagarikano et al., 2011; Castellucci et al., 2016; Abdavinejad et al., 2024). These findings suggest that speech and language impairment in autism may arise from partially distinct genetic mechanisms, some of which overlap with broader neurodevelopmental risk loci. However, despite these advances, systematic studies focusing specifically on the genetic underpinnings of minimally verbal autism remain limited.

Delving further, to our knowledge, there is a gap in the current literature regarding the relationship between individuals with ASD who have minimally verbal abilities and genetic factors.

For example, the KAT6A (Arboleda-Tham) syndrome, a Mendelian disorder caused by pathogenic variants in the lysine acetyltransferase 6A (KAT6A) gene (Ng et al., 2024), is characterized by intellectual disability and autism-related features (particularly restricted interests and repetitive behaviors) with an important speech/language impairment (e.g., minimally verbal).

A recent study (Barone et al., 2023) analyzed the autistic phenotype profile in two siblings with PARK2 microdeletion, which appears to be involved in heterogeneous autistic variability, in some cases associated with significant impairment of the language profile.

It also emerges the need to identify objective outcome measures of expressive language for use in children with ASD and minimally verbal abilities as recently investigated in a sample with Phelan-McDermid Syndrome (Rankine et al., 2017). Indeed, the development of verbal language in autistic children differs from neurotypical development and shows wide variability in expression (Tek et al., 2014) with language profiles in autism being variable (Kissine et al., 2023). Around the age of three, 50%–60% of autistic children are still non- or minimally speaking, but by the age of six or seven, around 60%–80% of autistic children develop expressive and receptive structural language, often at levels close or even superior to typical. During this 3- to 7-year of age interval there is a variety of language trajectories, from steady and relatively modest growth curves to surprisingly steep language gains, but after this point, language scores usually progress in relatively predictable fashion. Some authors speculated that many autistic children display an atypical and clinically recognizable trajectory in the acquisition of structural language (phonology, vocabulary and morphosyntax) arguing that this trajectory, called bayonet-shaped development, includes early language onset, regression, a plateau, and a certain level of language recovery (Mottron and Gagnon, 2023). The authors assumed that this developmental trajectory is associated with a prototypical autism phenotype considered as a discernible, rapidly identifiable, unique pattern during the plateau period (period without major language gains, usually in the 2–5-years period). Eventually about 25%–30% of autistic children do not develop or fail to develop functional spoken language and remain minimally verbal (MV) past age 5 (Anderson et al., 2007; Norrelgen et al., 2015), with implications for social and adaptive functioning in adulthood (Tager-Flusberg and Kasari, 2013). The usually refers to children who exhibit a limited vocabulary consisting of a small number of spoken words and fixed phrases (Tager-Flusberg and Kasari, 2013). There are studies that define MV on the basis of expressive vocabulary size (e.g., fewer than 20 or 30 words), others by the inability to produce multi-word utterances, or by the persistence of single-word speech beyond a certain age threshold (Kasari et al., 2006; Thurm et al., 2007). Additional approaches adopt parent-report measures or standardized language assessments to determine functional speech abilities (Pickles et al., 2014; Almirall et al., 2016; La Valle et al., 2024). A widely accepted definition of minimally verbal (MV) autistic children is derived from the Autism Diagnostic Observation Schedule, Second Edition (ADOS-2) (Lord et al., 2012). Within this framework, the term “minimally verbal” designates individuals evaluated with Module 1 of the ADOS-2, which is designed for children over 30 months of age who either produce no speech or use only a very limited repertoire of words in simple combinations (Bal et al., 2016; Fok and Bal, 2019; Lord et al., 2012; Thurm et al., 2015; Williams et al., 2018). It is important to note that, the use of standardized assessments commonly applied in language therapy to evaluate the linguistic abilities of MV autistic children has been critically examined, as such measures often present challenges due to the complexity of task instructions and the requirement for active child participation (Schaeffer et al., 2023).

Language in autistic children remains imperfectly understood, and identifying the factors that may predict, favor or shape linguistic trajectories within the crucial 3-to-7 period remains one of the major scientific challenges in autism research (Kissine et al., 2023), including possible associated genetic profiles. Studies have been conducted in the literature to search for predictive factors on language development; for example, Kissine M. and colleagues in recent research (Kissine et al., 2023) speculated that joint attention plays a pivotal role for the emergence of language, but some autistic children may acquire language independently of joint attention skills.

Other studies investigating intellectual functioning in MV population have yielded highly heterogeneous results (Bal et al., 2016). It has been reported that, in MV autistic preschoolers, nonverbal intelligence quotient is one of the major predictors of subsequent language gains for those children who acquire some language before the age of 5 (Ellis Weismer and Kover, 2015; Thurm et al., 2007; Thurm et al., 2015; Wodka et al., 2013). Additionally, general intellectual disability has been reported in MV autistic individuals (Hewitt et al., 2012; Luyster et al., 2008; Slušná et al., 2021). However, Bal and colleagues (Bal et al., 2016) reported in 2016 that 16% of children in their sample had non-verbal cognitive abilities within average limits, contrasting with the hypothesis that minimal verbalization is synonymous with cognitive impairment. These results suggest that the mechanisms underlying the lack of language development are not solely attributable to cognitive level, but rather to a heterogeneous set of predictive factors, some of which are precursors to language that vary across the autism spectrum (Pecukonis et al., 2019; Thurm et al., 2007).

As we described, there are several studies that analyze the autistic phenotype in different syndromes (e.g., case reports and syndrome-specific cohorts), but none focus specifically on patients with syndromes that present both autism and are found to be minimally verbal.

Given the notable gap in current literature regarding the clinical and genetic profiles of autistic individuals with minimal verbal abilities, the goal of this study was to assess the prevalence of genetic conditions within MV autistic population and identify genetic features linked to specific syndromes. Furthermore, by analyzing the neuropsychological profiles of a subgroup within this cohort, the study sought to identify clinical phenotypes associated with these genetic factors that is as primary neuropsycological outcome the severity of ASD characteristics and as secondary outcome if children with genetic alterations would exhibit differences in cognitive, adaptive, or behavioural outcomes compared to those without.

We finally hypothesised whether, in the MV autistic population, certain characteristics of the MV profile could be fully explained by the specific genetic condition identified or not.

The novelty of this study is that we tried to apply a systematic approach including both a detailed neuropsychological profile in MV autistic individuals and a in-depth genetic evaluation. Moreover, considering the relatively low prevalence of autism in syndromes, the size of the cohort inclued was significant.

The first aim of this study was to conduct a genetic analysis of a large sample of autistic children and adolescents with MV characteristics. The finding was a 22.6% prevalence of genetic disorders, characterized by a diverse distribution of duplication and deletion and variations patterns affecting various gene loci, each associated with distinct clinical phenotypes, as detailed in Table 1.

To our knowledge, the data available in the literature indicate that approximately 10% of all autistic individuals typically exhibit dysmorphic traits and/or malformations and are classified as having syndromic autism (Zafeiriou et al., 2013; Styles et al., 2020). Therefore, in our sample, the prevalence of genetic syndromes was found to be twice as high as in the general autistic population.

“Syndromic” autism includes deletions and duplications syndromes (e.g., 1q24.2, 15q11-q13, 16p11.2, 22q11), chromosomal aneuploidies (e.g., Down syndrome) and monogenic syndromes (e.g., Fragile X syndrome, Angelman syndrome, Rett syndrome, Tuberous Sclerosis Complex - TSC, Neurofibromatosis and PTEN hamartoma tumor syndrome) (Ziats et al., 2021; Genovese and Butler, 2023).

Some genetic syndromes have been reported to have higher autism prevalence rates, more similar to those observed in our MV cohort: for Kleefstra Syndrome the prevalence of behavioral disorders, such as autism spectrum disorder, is about 38% (Frazier et al., 2025); in Myotonic Dystrophy type 1 the prevalence of ASD was 14%, with congenital onset being 79% more common than juvenile onset (Pascual-Morena et al., 2024); the reported prevalence of ASD in Down syndrome is 16% (Santoro et al., 2021). Adviento and colleagues found in four major RASopathies (neurofibromatosis type 1 - NF1, Noonan syndrome - NS, Costello syndrome - CS, and cardio-facio-cutaneous syndrome - CFC) a high prevalence of ASD traits using Social Communication Questionnaire (SCQ) (85% scored at the threshold of 15 or higher) (Adviento et al., 2014).

For other genetic conditions identified in this study, the prevalence of them in the autistic population is known: about 10%–20% of autistic individuals and macrocephaly have germline PTEN mutations (Yehia et al., 2020).

It is important to note that the overall prevalence of ASD within the syndrome has been found to be significantly higher than the prevalence of the syndrome among ASD cases (Styles et al., 2020). Indeed, for the best characterized syndromes, we are aware of both the prevalence of the syndrome in autistic individuals and the prevalence of ASD in the syndrome: for TSC different studies have shown that 3%–4% of autistic subjects may have TSC; conversely, studies of TSC patients have identified much higher rates of autism, ranging from 17% to 61% (Curatolo et al., 2004). For Phelan-McDermid syndrome (PMS) approximately 1% of individuals with autism are estimated to have SHANK3 mutations while up to 84% of individuals with PMS met criteria for ASD (Frank, 2021).

For 1q21.1 duplication syndrome (Xavier et al., 2018), Greig cephalopolysyndactyly syndrome (Siracusano et al., 2019) and Intellectual developmental disorder with autism and speech delay associated to a heterozygous pathogenic variation in TBR1 (Cao et al., 2021) we do not have prevalence data regarding ASD but merely case reports.

For intellectual developmental disorder with hypotonia, impaired speech, and dysmorphic facies due to a heterozygous variant in TNPO2 (Goodman et al., 2021) and CAKUTHED due to heterozygous variant in PBX1 (Alankarage et al., 2020) there are only functional studies.

To our knowledge, none of the cited studies analyze the prevalence of MV autistic individuals in the genetic syndromes studied, athough there is evidence of a gap in the literature concerning the exploration of factors that may affect the language development in autistic individuals (Vogindroukas et al., 2022) and the necessity of examining large cohorts of autistic individuals with MV traits (Koegel et al., 2019). It is broadly acknowledged that an underlying etiological diagnosis, mainly through genetic testing, particularly for autistic children and adolescents with severe symptoms and notable language impairments, is a critical prerequisite for comprehensive management (Hyman et al., 2020). This underscores the significance, in clinical practice, of performing in-depth genetic screening in MV autistic individuals to detect a syndrome that could influence prognostic outcomes for patient care (for instance, consider the cancer surveillance guidelines for individuals with PTEN hamartoma tumor syndrome) (Tischkowitz et al., 2020).

Going now to analyze the neuropsychological profile in a subgroup of patients, our study identified an interesting second finding: the total raw ADOS-2 scores of MV subjects with genetic causative alterations were less severe than those of MV individuals without genetic causative alterations, in a total population that nevertheless by characteristics has high ADOS scores (mean of ADOS-2 total raw scores: 15.38 in MV autistic group with genetic causative alterations versus 19.66 in MV autistic group without genetic causative alterations). It has already been described in the literature that the clinical syndromic presentations of autism with genetic causative alterations are extremely heterogeneous, which could be attributed to differences in genetic background, eventual epigenetic influences, and to the distinct characteristics of the population (Styles et al., 2020). With respect to the characteristics of the population we studied: there are no studies in the literature specifically describing the autistic phenotype in syndromic MV individuals. The information on the severity of ASD (e.g., using ADOS-2 Module 1) and language impairment (which we can trace back to a MV autistic population) comes indirectly from case reports analyzing autistic functioning in some of the more studied syndromes. In a recent study (Barone et al., 2023), the autistic phenotype profile was analyzed in two siblings with PARK2 microdeletion as copy number variations (CNVs) involving PARK2 have been associated with ASD. This study pointed out that the younger sibling had minimally verbal autism and severe intellectual disability, whereas his older brother presented high-functioning autism and preserved speech. This is consistent with the hypothesis indicating that polygenic effects acting in summation may explain the heritability and clinical heterogeneity of ASD.

A critical point concerns the interpretation of ADOS-2 severity scores. The finding that MV children without genetic causative alteration exhibited greater autism symptom severity may in part reflect differences in expressive language abilities rather than true differences in core autistic features. Several studies have noted that ADOS calibrated severity scores, particularly in Modules 1 and 2, are influenced by verbal ability, with lower levels of spoken language associated with higher severity ratings (Hus et al., 2014; Gotham et al., 2009). Thus, it is possible that the apparently greater severity observed in the group without genetic causative alteration partly reflects their language profile rather than a genuine difference in autism symptomatology.

One of the most studied syndromes is the Phelan-McDermid Syndrome, which as we have already reported, is associated with ASD but also with Intellectual Disability (ID) and speech and language impairment (absent to severely delayed speech) which are present in all cases (Frank, 2021). Considering Down Syndrome (DS) ad ASD, Godfrey and colleagues in 2019 compared children with co-occurring ASD + DS versus children with DS versus cognitively matched ASD children (Godfrey et al., 2019). The authors found fewer global ASD symptoms on the ADI-R for autistic children with DS compared to matched autistic children. However, autistic children with DS had more evenly distributed social communication and restricted and repetitive behavior scores while autistic children only had more severe SC than RRB scores. Another study examining domain scores on ADOS Module 1 by Oxelgren et al. (2019). Similarly found lower communication scores for the ASD + DS group compared to an idiopathic ASD only group: autistic children with DS showed significantly lower scores on stereotyped use of phrases (Oxelgren et al., 2019). To explain these findings, Godfrey M. and colleagues thought that it might be related to the ASD + DS group’s low verbal abilities (Godfrey et al., 2019). Indeed, since most observational measures of ASD symptom severity are strongly influenced by language level (Anderson et al., 2007; Schaeffer et al., 2023), it is still necessary to understand whether there is a possible association between the severity of core symptoms and language abilities. Our hypothesis is that, in the MV autistic population without genetic causative alteration, there are specific and unknown characteristics of the MV profile (absence of language development beyond 5 years of age) which have a greater impact than the individual genetic condition reported but this is still speculative and requires longitudinal validation.

We hypothesize that, in MV autistic individuals without a genetic causative alteration, specific and yet unknown features of the MV phenotype may have a greater influence on clinical presentation than the presence of a known genetic condition. For instance, it is possible that, in the absence of a single high-impact genetic variant, the cumulative effect of multiple genetic and neurodevelopmental factors, together with early language deficits, contributes to the more severe presentation observed in this subgroup. However, this interpretation remains speculative and requires validation through longitudinal and multi-omic studies aimed at disentangling the respective contributions of language, cognition, and genetics to the MV phenotype.

Our finding that minimally verbal children without a known genetic condition showed greater autism severity than those with genetic causative alterations requires careful contextualization. Previous studies comparing idiopathic and syndromic ASD have reported mixed results, with some evidence suggesting that syndromic cases present with broader developmental impairments, while idiopathic ASD may show relatively greater severity in core social-communication domains (Oxelgren et al., 2019; Thurm et al., 2019). One possible explanation, as we already told, is that ADOS-2 severity scores may be disproportionately influenced by expressive language abilities (Gotham et al., 2009), which are often more profoundly impaired in MV children without genetic causative alteration. However, alternative mechanisms should also be considered. It is plausible that children without genetic causative alteration may carry a higher polygenic risk load for ASD, in line with recent genomic studies highlighting cumulative contributions of common variants to autism severity (Weiner et al., 2017). Epigenetic mechanisms, such as environmentally mediated changes in gene expression, may also modulate phenotypic severity independently of rare, highly penetrant variants (Loke et al., 2015). Finally, the possibility of undetected genetic variants—including non-coding mutations, structural rearrangements, or low-level mosaicism not captured by standard diagnostic pipelines—cannot be excluded.

Taken together, these considerations suggest that the apparently greater severity in MV children without genetic causative alteration may reflect not only assessment bias linked to language profile, but also a combination of polygenic, epigenetic, and unidentified genetic contributions. This highlights the need for comprehensive genomic testing (e.g., whole-genome sequencing, epigenomic profiling) in future studies of MV populations to disentangle the relative roles of rare syndromic mutations versus broader genetic and environmental liabilities in shaping the clinical phenotype.

Therefore, future studies should analyze the genetic profile of large cohorts of MV autistic children and adolescents to evaluate the possible association with the neuropsychological profile and to better characterize a specific phenotype. Furthermore, when comparing non-verbal intellectual and adaptive functioning, as well as emotional and behavioral problems, no significant differences emerged between the group of MV autistic children and adolescents with associated genetic causes and those without them. According to the literature, most of the syndromic conditions identified in our population are more frequently linked to lower intellectual and adaptive functioning (Samanta, 2020; Xavier et al., 2018; Bradbury et al., 2022; Frank, 2021; see Table 1) compared to the general population. This effect is further amplified in ASD populations when a genetic syndrome (e.g., Down syndrome) is present (Thurm et al., 2019). However, it is important to note that our entire cohort of children exhibits an average NVIQ below 70. As previously reported, general intellectual disability has been described in MV autistic individuals (Hewitt et al., 2012; Luyster et al., 2008; Slušná et al., 2021), yet considerable heterogeneity has been observed, highlighting that language impairment cannot be fully explained by cognitive difficulties (Table 1). It is likely that in our sample, differences between the two groups are not apparent due to the relatively homogeneous cognitive and adaptive functioning of the population studied. Once again, these findings suggest that MV characteristics may play a significant role in our cohort of patients. Additionally, although few studies have examined how emotional and behavioral problems change in children with different language abilities (Fok and Bal, 2019), some psychiatric comorbidities (e.g., anxiety, depression) have been shown to manifest differently in individuals with varied intellectual and language abilities (Einfeld and Aman, 1995). Therefore, the lack of differences in emotional and behavioral profiles, regardless of the presence of associated genetic causes, may be attributed to our sample’s inclusion of individuals who either did not develop language or produced only a few words beyond the age of six.

This study has limitations that should be considered: the limited neuropsychological assessments conducted on a reduced subset of patients rather than the entire sample (given the lack of availability of complete neuropsychological assessments for the entire sample), the retrospective nature of the analyses, and the classification of MV. Consistent with previously selected samples (Guerrera et al., 2025), our sample was classified as minimally verbal (MV) according to one of the definitions reported in the literature, based on both linguistic functioning and age. However, given the absence of a universally established operational definition of MV in autistic individuals, the findings of the present study cannot be considered fully comparable to those of previous research. A similar caveat applies to the assessment tools employed to define the sample. Although we used Module 1 of the ADOS, which is recognized in the literature as valid under certain definitions, variability in the instruments adopted across studies further limits direct comparability. This variability reflects both differences in research aims and the lack of consensus on a universally accepted threshold. In future studies a more precise and consistent operational definition of MV would therefore enhance comparability, facilitate meta-analytic synthesis, and improve translation of research into clinical practice.

Furthermore, although the CBCL is among the most widely used tools for behavioral and psychological screening in autistic children, it includes numerous items that are directly influenced by verbal language skills, which may not be entirely appropriate for autistic and MV individuals (DiStefano et al., 2016). Another limitation of our study lies in its retrospective design, which resulted in heterogeneity in the genetic testing performed across participants.

While all participants underwent first-tier genetic testing (karyotype, FMR1, and CMA), only two underwent exome sequencing, based on clinical evaluation and the presence of peculiar somatic features or syndromic features. Consequently, a large group of individuals negative classified as “without genetic causative alteration” may in fact carry a yet undetected pathogenic variant. This methodological limitation should be addressed in future prospective studies through a more systematic use of uniform, high-resolution genomic testing (e.g., WES or WGS) for all participants in order to enhance the diagnosis rate and to allow a more accurate genotype–phenotype characterization.

Moreover, future research should focus on the need for uniform testing protocols in order to allow more robust genotype–phenotype correlations.

Nevertheless, to our knowledge, this is the first study to investigate genetic causes and neuropsychological characteristics in a relatively large sample of MV autistic children and adolescents and to define the neuropsychological profile within this population. It examines the prevalence of genetic syndromes associated with MV autistic traits and compares the clinical profile of a subgroup of individuals with associated genetic features to a subgroup that did not show genetic alterations in the analyses performed. Hence there is the need to carry out specific studies on the MV autistic population, which is still poorly known, on the characteristics of this population in terms of autistic symptoms, intellectual and linguistic functioning and severity and on the association with genetic profiles.