Signals. Communication is one of the clinical domains that we talk about often in genetic neurodevelopmental disorders, but surprisingly rarely study across disorders in a systematic way. In a recent preprint, the authors used large-scale research datasets to map communicative abilities across many genetic conditions linked to neurodevelopmental disorders and autism, asking a simple but powerful question: can we see disorder-specific patterns when we look at communication at scale? Here is what the authors found.
Figure. Communication profiles across genetic neurodevelopmental disorders. Left: Heatmap reproduced from the recent preprint by Hsu et al. (2026), based on data from the Simons Searchlight and SPARK cohorts, showing multiple measures of communicative ability across genetic neurodevelopmental disorders. Measures include Vineland-3 expressive and receptive subdomains, speech milestone items, and single-item questionnaire responses. Across conditions, different measures of communication tend to agree, suggesting that both standardized scales and simple parent-reported questions capture related aspects of communicative function, although the strength of impairment differs by genetic etiology. Right: Scatterplot derived from the Vineland-3 values shown in the table on the left, plotting mean expressive versus receptive v-scores for each genetic condition. While communication abilities differ substantially across disorders, expressive and receptive skills generally move in parallel, with clear gene-specific profiles. STXBP1-related disorders show particularly low scores in both domains, consistent with the severe early developmental impairment observed clinically, whereas other conditions cluster at higher levels of communicative ability. Together, these data illustrate how large cohort datasets make it possible to map communication phenotypes across many genetic disorders simultaneously, revealing both shared patterns and disorder-specific signatures.
Preprint. Before diving into the publication, I want to point out that I am breaking one of my steadfast rules by commenting on a manuscript before it is published. The work by Hsu et al. is a preprint on medRxiv, but it has already attracted considerable attention. Therefore, I find it timely to comment on this study. Preprints are an interesting phenomenon that I will discuss more deeply in a future post. They came relatively late to biomedical research compared to physics and mathematics, where early sharing through arXiv has been common since the 1990s. For decades, biomedical results appeared only in journals. This changed with bioRxiv and later medRxiv, and the COVID-19 pandemic accelerated the use of these platforms. Today, the field lives in a hybrid system where findings often appear first as preprints and later in journals.
Searchlight and SPARK as phenotyping platforms. Back to the preprint by Hsu et al. Their study draws on two of the largest resources in rare disease and autism research, the Simons Searchlight and SPARK cohorts. These datasets combine genetic information with standardized phenotyping collected across thousands of individuals, including many different genetic etiologies. This makes it possible to move beyond single-gene case series and instead compare communication profiles across a wide range of neurodevelopmental disorders using the same measures.
Validating what clinicians already know at scale. One striking aspect of the study by Hsu et al. is that many of the results confirm clinical impressions that have accumulated over years. Disorders known to severely affect language and social communication show the expected impairments (e.g., STXBP1, SCN2A). However, the important point is that this is now demonstrated across large numbers of individuals and across many conditions simultaneously. Instead of learning about communication one gene at a time, we can see the broader landscape of communicative phenotypes across genetic neurodevelopmental disorders (NDD).
Communication in genetic NDD and autism. By combining genetic diagnoses with autism-focused cohorts, the preprint by Hsu et al. also highlights the overlap between rare genetic disorders and autism-related traits. Communication difficulties are not uniform across conditions, and some genetic etiologies are more severely affected than others. In particular, copy number variants (16p11.2 del/dup, 1q21 del/dup) show much milder communication impairments than many of the single-gene disorders. In addition, the authors include a large cohort of individuals with idiopathic autism and demonstrate how clearly the monogenic neurodevelopmental disorders separate from individuals with autism more broadly. The impairments are much more severe on average.
Established scales and simple questions. There is an interesting methodological aspect of the study by Hsu and colleagues that the authors may be able to emphasize even more in the future. In brief, the study combines standardized instruments such as the Vineland Adaptive Behavior Scales with single-item parent-reported questions that were taken from other scales. Overall, the measurements align, which emphasizes how various measures of communication converge and can be captured through proxy measures. While single-item questions such as the ability to use phrases or to have a conversation do not have the same validity as the Vineland-3 (the “workhorse of NDD research”), the results point toward the possibility that we might be able to infer standardized measures in the future based on categorical information in a patient’s medical record — essentially an imputation of phenotypic data, where standardized measures are typically absent outside of natural history studies or trials. The authors note that this needs to be explored further, especially when single-item measures are used to compare across disorders with very different developmental trajectories. However, their initial observation is highly interesting and may allow us to capture clinical information more comprehensively in the future.
STXBP1 as a distinct communication profile. One of the clearest gene-specific signals in the analysis is seen in STXBP1-related disorders, the condition that we know extremely well from our work in ENDD. In the study by Hsu et al., individuals with STXBP1-related disorders show particularly severe impairment in communication development. The authors suggest that there is evidence for stagnation over time rather than steady improvement, but it should be noted that much of the data is cross-sectional rather than following patients across their lifespan, with the potential for ascertainment bias. However, the core finding holds true. Most individuals with STXBP1 have profound communication impairments, and the preprint by Hsu et al. emphasizes that this impairment is more severe than in individuals with all other genetic etiologies included in their study. Of note, the Searchlight/SPARK analysis only includes a subset of genetic neurodevelopmental disorders, but it still provides a useful map of where different autism-related and monogenic disorders fall along the spectrum of communication ability.
Expressive versus receptive. I used the data from Figure 2 by Hsu et al. and compared expressive versus receptive v-scores. Vineland-3 v-scores are age-normed standardized scores with a mean of 15 and a standard deviation of 3, used for subdomains such as expressive and receptive communication. Because they are adjusted for age, v-scores reflect adaptive functioning relative to peers of the same age, allowing comparison across individuals of different ages. What is remarkable about the values is that expressive and receptive skills are not only correlated but show a near one-to-one relationship. This is important in two ways. First, receptive communication skills in individuals with more severe conditions are not higher than expressive skills (STXBP1, SCN2A, ASXL3). We know that initial natural history data already suggest this, but seeing this across multiple conditions is a new finding. Second, even conditions often linked to speech apraxia (such as SETBP1, MED13L) do not show higher expressive than receptive subdomain scores. While this may differ for individual patients, it does not appear at the cohort level. Virtually all conditions fall close to the identity line.
What you need to know. Using large-scale datasets from Simons Searchlight and SPARK, the authors mapped communication abilities across many genetic neurodevelopmental disorders and autism, showing that standardized scales and simple parent-reported measures capture consistent signals across conditions. The analysis confirms known clinical patterns, highlights gene-specific profiles such as the severe communication impairment in STXBP1 and illustrates how large cohort studies can define outcome domains that are relevant for both natural history studies and future clinical trials.
