Prevalence of Phelan McDermid Syndrome Estimated To Be ~1:7300 Using a Multisource Model Authors: Tess Levy, David Lapidus, Kate Friedman, Paige Siper, Larry Glass, Liza Squires, Mary Hames, Joseph D. Buxbaum, Brooklynn Gasser, Brooklynn Gasser, Meghan Towne, Behrang Mahjani, Seulgi Jung, Stuart Schwartz, Kai Lee Yap, Mindy Li, Avinash V. Dharmadhikari, Gordana Raca, Dolores B. Estrine, Stefanie Biller, Amanda Singleton, Sarah Poll, Alexander Kolevzon Published: 2026 ABSTRACT ABSTRACTEstimating the prevalence of genetic disorders is complicated by many factors including sampling bias and differing methods of estimation. However, establishing the true prevalence of these disorders is critical for understanding disease burden, pharmacoeconomic modeling, and resource allocation for testing and care. Phelan‐McDermid syndrome (PMS) is a genetic neurodevelopmental disorder with an unknown true prevalence and previous estimates vary from 2.5–10 per million births. The study team reached out to a multitude of sources, including clinical genetic testing laboratories, research centers, and clinical centers. Sites provided the number of PMS diagnoses made out of the total number participants with autism tested at their site. Further extrapolations were made to adjust for the proportion of individuals with PMS who do not have autism, autism diagnosis age limitations, and type of genetic variant. Lastly, Centers for Disease Control estimates of autism rates were used to extrapolate to the general population. Ten sources participated and data from 179,837 autism cases were evaluated. The frequency of PMS diagnoses ranged from 1% to ~2.5%. However, the studies had different levels of sensitivity depending on the assay(s) used, and therefore raw results should not be directly compared. After applying extrapolations, including adjustments for assay sensitivity and other factors, the final weighted average was 13.7 per 100,000 (95% CI 10.02–18.60 per 100,000), indicating 1 in ~7300 individuals in the general population have PMS. We leveraged a diverse set of genetic data from multiple sources to generate a population‐level estimate of the prevalence of PMS. Our findings indicate that PMS affects approximately 13.7 per 100,000 individuals, substantially higher than previous estimates. Introduction Elucidating the prevalence of rare genetic disorders is complicated by various factors including barriers to diagnosis, lack of centralized data, and limited research. While these data are critical, there is currently no systematic way to estimate the true prevalence of genetic disorders, and most estimates are likely underestimated. Knowing the true prevalence is important for understanding the disease burden, gaining interest from industry and pharmaceutical companies, advising governments and insurance companies on resource allocation for testing and care, and for general awareness of the disorders. Phelan‐McDermid syndrome (PMS, OMIM: 606232) is a rare genetic neurodevelopmental disorder, with an unknown true prevalence. The disorder was originally identified as a microdeletion syndrome affecting 22q13 in the early 1990s (Phelan et al. 2001, 1992; Praphanphoj et al. 2000; Prasad et al. 2000; Romain et al. 1990). As the suspected region narrowed, SHANK3 became the candidate gene (Wilson et al. 2003; Bonaglia et al. 2001; Boeckers et al. 2002) and by 2007 the first case reports of SHANK3 pathogenic sequence variants were reported as causative of PMS (Durand et al. 2007). Today, the etiology of PMS is defined by haploinsufficiency of SHANK3, either by a 22q13 deletion or pathogenic sequence variant. The clinical features of PMS vary widely, with the most common findings including intellectual disability of varying degree, developmental delays, autism, gastrointestinal symptoms, hypotonia, and behavioral problems. Early reports of PMS tended to include more severely affected individuals, both because genetic testing was reserved for this population and because technology could detect only large deletions, typically associated with a more severe presentation. As utilization of genetic testing has increased and costs have decreased, individuals with smaller deletions or pathogenic sequence variants, often associated with a milder phenotype, have been identified (Levy et al. 2025). Currently, the prevalence of PMS is estimated at 2.5–10 per million births (0.25–1 per 100,000) (National Organization for Rare Disorders, Phelan–McDermid syndrome). The Phelan McDermid Syndrome Foundation (PMSF) collates known cases through family registration and serves as the best estimate of PMS worldwide, with 3600 cases known in 2025. However, there are many points of potential exclusion which limit the validity of this estimate (Figure 1). Receiving proper referrals for a genetic consultation and being offered genetic testing is the first major barrier to diagnosis. Though the standard of care has included genetic testing for individuals with neurodevelopmental phenotypes for nearly two decades (Miller et al. 2010), a minority of individuals are offered testing (Peabody et al. 2015; Brown et al. 2025; Abreu et al. 2024; Vande Wydeven et al. 2012; Moreno‐De‐Luca et al. 2020). For individuals who are offered testing, coverage of testing is the next hurdle. Private insurances and Medicaid each establish criteria for medical necessity for genetic testing, which are often not aligned with current care guidelines, limiting the availability of genetic testing to the population. When testing is approved, many individuals will only receive a chromosomal microarray (CMA), which will identify copy number variation (CNV) but not sequence variation (SV), therefore missing approximately half of the PMS cases. Even when exome sequencing or panel sequencing is offered, there are multiple labs where SHANK3 is not analyzed due to its complex nature, again reducing the number of individuals who receive a diagnosis. For individuals who receive a rigorous evaluation and are diagnosed with PMS, not all will be aware of PMSF and its efforts to collate all known cases or of ongoing research studies. For families who are aware of PMSF, not all will be interested in joining the Foundation or ongoing research, again reducing the number of known cases. Though there are many more points of potential exclusion, these key factors illustrate how the true population is likely to be substantially larger than the known population. Factors such as socioeconomic status, limiting access to testing if not covered by insurance, geographic factors, and limiting access to tertiary medical centers with genetics departments, play a role in the underestimation of PMS. Points of Potential Exclusion in Count of Phelan‐McDermid Syndrome Patients. Illustration showing points of potential exclusion in the count of Phelan‐McDermid syndrome patients in the Foundation's registry count. CNV, copy number variant; N, no; PMS, Phelan‐McDermid syndrome; SV, sequence variant; Y, yes. To estimate a ‘closer‐to‐true’ prevalence number for PMS, we sought data from multiple sources including clinical genetic testing laboratories, hospital‐based clinical centers, and research studies. By combining data across sources, we aimed to reduce the impact of the biases associated with each source. Herein, we present data from 10 data sources leading to a new estimated prevalence of PMS. Methods Sites We contacted research centers, clinical genetic testing laboratories, and clinical centers within academic medical centers and requested prevalence rates of PMS. The number of individuals diagnosed at the site and the total number of tests completed were the minimum necessary data to be shared to be part of the collaboration. Sites were instructed to include PMS cases as either a 22q13 deletion that included the SHANK3 gene or a likely pathogenic or pathogenic sequence variant in SHANK3. Additional information, including age of diagnosis, specific genetic variant, insurance type, and sex, was requested as supplemental data. Each center had individual limitations to the prevalence estimates based on their sample and how they could parse their data; the limitations are described below. Data from peer‐reviewed publications were also included. In order to create comparable estimates, we requested the prevalence of autism cases from each site. Because many of the larger sources only included those with autism, we requested updated numbers from the other sources to match. Therefore, calculations in the final analyses included the number of individuals with autism who were diagnosed with PMS divided by the number of individuals tested at the site with a diagnosis of autism. To reduce bias and increase the robustness of our estimate, we integrated data from diverse sources: research studies, clinical centers, genetic testing laboratories, and published manuscripts. Each source brings inherent biases, but they are different in nature and can partially offset one another. For example, research cohorts such as the Autism Sequencing Consortium (ASC) and Simons Foundation Powering Autism Research for Knowledge (SPARK) offer no‐cost genetic testing to eligible participants, which helps eliminate financial barriers. However, participation is limited by recruitment criteria and scope and individual willingness to engage in research. On the other hand, clinical genetic testing laboratories provide large, geographically diverse datasets but are constrained by disparities in insurance approval and access to testing. By combining data from these distinct sources, we sought to minimize the influence of any single source's limitation. Clinical Genetic Testing Laboratories Twelve clinical laboratories were contacted and five agreed to participate. This included two academic hospital‐based laboratories, Lurie Children's Hospital of Chicago and the Children's Hospital Los Angeles (CHLA), along with three private laboratories, Ambry Genetics, Labcorp, and GeneDx. Lurie Children's Hospital has a molecular genetics laboratory that performs CMAs in‐house. The medical genetics team, in collaboration with the molecular laboratory, shared the number of positive PMS diagnoses made by CMAs from 2011 to 2024, with an indication of autism. The total number of CMAs completed for an indication of autism was also provided. CHLA has a molecular genetics laboratory which performs CMAs and other genetic testing in‐house. CHLA provided the number of PMS diagnoses made by CMA related to the number of total CMAs completed for probands with an indication of autism. Ambry Genetics is a private genetic testing laboratory based in California. Ambry provided detailed information about their PMS cases including indications for testing, variant type, and insurance type. They shared results from both exome sequencing and chromosomal microarrays. The total number of tests reflects the total number of exome and microarrays done for an indication of autism at their laboratory. Labcorp Genetics and Genomics Lab is a private genetic testing laboratory and has completed over 12 million genetic tests since its inception. Labcorp shared CNV prevalence estimates for PMS based on the total number of postnatal chromosomal microarrays completed for an indication of autism. GeneDx is a genetic laboratory specializing in exome and genome sequencing. GeneDx shared the total number of exomes and genomes completed with an indication of autism, along with the total number of Phelan‐McDermid syndrome diagnoses, both SHANK3 pathogenic sequence variants and deletions. Twelve clinical centers were contacted and one agreed to participate. Geisinger Medical Center has an autism clinic where patients are offered genetic testing. This group provided information about PMS diagnoses made at their clinic. Patients within the clinic may be offered different types of testing depending on the year they were seen, their presentation, and their insurance coverage. The numbers provided represent the number of PMS diagnoses made compared to the number of individuals seen in their clinic who received any type of genetic testing. Nineteen research centers were contacted and two agreed to participate. ASC and Genomics of Autism in Latinx Ancestry (GALA). The ASC and GALA are NIH‐funded large‐scale autism gene discovery efforts led by the Seaver Autism Center at the Icahn School of Medicine at Mount Sinai. Both studies enroll individuals with autism and their family members and complete exome sequencing. Both CNVs and SVs are called from the exome sequencing. SPARK is a foundation‐funded initiative which enrolls individuals with autism and their family members to complete exome sequencing. Raw data is available upon request from the Simons Foundation. Researchers at the Seaver Autism Center analyzed the raw data and provided CNV and SV calls based on the most recent SPARK data. Lastly, data from peer‐reviewed manuscripts were included. Researchers from The Hospital for Sick Children in Toronto reviewed results from a whole genome sequencing (WGS) study which enrolled individuals with autism. Diagnostic results were shared in the manuscript, including diagnoses of PMS (Selvanayagam et al. 2025). Researchers from France enrolled autistic individuals at clinics throughout France and Sweden; an analysis was completed to assess the prevalence of SHANK1‐3 CNV and SVs in the cohort (Leblond et al. 2014). Calculations The PMS frequencies reported in the autism cohorts can be extrapolated to the US national population by applying additional assumptions that bridge from the cohorts to the general population. The extrapolation uses the following steps: Prevalence of Autism in the Pediatric General Population The CDC reports biannual estimates of autism prevalence at age 8 years old. We assume that these cases are detectable as early as age 3 years, though in practice many are not diagnosed until older. For each autism cohort, the prevalence value(s) corresponding to its year(s) of recruitment is applied. For cohorts which were recruited over multiple years, the average prevalence value for the recruitment period is applied. Where needed, prevalence is interpolated for the years between the CDC's biannual reports. Percentage of Autism Cases That Proceed to Sequencing For some autism cases, genetic sequencing via autism panels is not required because a clinically evident syndrome is present. Such cases would not appear in the autism cohorts used in this study, so they must be excluded from the extrapolation. For this study, it is estimated that 95% of autism cases lack a clinically obvious diagnosis (Tammimies et al. 2015). Therefore, the vast majority of cases are potential recruits to the autism cohorts. Percentage of PMS Cases With Autism Autism is a common feature of PMS, but it is absent from 38%–40% of PMS cases (Schön et al. 2023; Levy et al. 2022). This value is used to adjust for the prevalence of non‐autism PMS cases that would be absent from the autism cohorts. In a review of 315 participants examined by Schön et al. (2023) and a cohort of 17s0 participants examined by Levy et al. (2022), consensus diagnoses were made based on the Autism Diagnostic Observation Schedule, Second Edition (Lord et al. 2012), the Autism Diagnostic Interview—Revised (Lord et al. 1994), and psychiatric evaluation based on DSM‐5 criteria for autism spectrum disorder (Schön et al. 2023; Levy et al. 2022). This comprehensive workup provides the best estimate of autism in this population. However, it should be noted that autism is a common indication for genetic testing and those with autism are more likely to receive coverage for testing than those without autism, inflating the estimate of autism in PMS. This would result in an underestimate of total PMS prevalence since the true number of PMS cases missing from the autism cohorts would be higher than implied by these cohorts. Assays' Sensitivity to PMS Genetic Defects Most of the autism cohorts were tested via either CMA, whole exome sequencing (WES), or WGS. CMA can detect CNVs. Older WES/WGS assays were effective at detecting SVs, but not CNVs. Newer WES/WGS can detect both CNVs and SVs. In an analysis of 154 participants with PMS, 74% of individuals had a deletion while 26% had a sequence variant (Levy et al. 2025). A chronological analysis of cases revealed equal cases of deletions and sequence variants diagnosed each year after 2010 (Levy et al. 2025). This evidence suggests a ~50:50 ratio of deletions and pathogenic sequence variants in PMS10. Prior estimates showing a larger proportion of deletions than pathogenic sequence variants are likely due to lack of sequencing assays and availability. For autism cohorts that were tested only via CMA or older WES/WGS, half of PMS cases would remain undiagnosed. Therefore, this study adjusts the PMS frequency in each cohort depending on the type of assay(s) used. The ~50:50 ratio of deletions and sequence variants is based on the smallest sample size of all elements in the national extrapolation (25 cases out of 51 (Levy et al. 2025)). Because of the small sample size, this assumption generates the broadest 95% CI among all the extrapolation inputs. To account for the statistical uncertainty generated by this assumption, its 95% CI range is propagated through to the final prevalence output. PMS Diagnosed Before Onset of Autism Symptoms Autism is rarely diagnosed before age 3, though PMS may be diagnosed earlier due to its other symptoms. As a conservative assumption, it is assumed that PMS cases diagnosed during age 0–24 months are absent from the autism cohorts. Data from two cohorts (Labcorp and CHLA) show that 25% of PMS diagnoses occur during age 0–24 months. Among these, 38% would not be expected to develop autism (Schön et al. 2023; Levy et al. 2022), so the net share of PMS cases who are diagnosed before age 24 months and would not develop autism is 15.4%. This adjustment is applied to the PMS prevalence estimate to capture early diagnosed cases who would not be captured in the autism cohorts. Calculation of Average Across All Autism Cohorts and National Prevalence For each autism cohort, the share of PMS cases is determined (SHANK3 positive/total autism cases) and then adjusted as described above to extrapolate a national PMS prevalence. The prevalence values from all autism cohorts are weighted according to their size (total autism cases) to calculate a weighted average that represents the estimated population prevalence of PMS among persons aged 0–17 years. Results PMS Frequency in Autism Cohorts Details of each autism cohort, including the frequency of cases with SHANK3 variants, are shown in Table 1. A total of 179,837 autism cases were evaluated for the presence of a SHANK3 variant. The frequency of SHANK3 variants ranged from 0.089% (CHLA) to 2.475% (SickKids Toronto). These raw results should not be directly compared because the assays had different levels of sensitivity, but the extrapolation made adjustments for these and other factors (Figure S1). TABLE 1Autism cohorts and SHANK3 frequency.Cohort nameAutism prevalence periodAutism prevalence per 1000Variants: CNV, SV, or bothTotal autism cohort SHANK3 P/LP cases SHANK3 percentAutism Sequencing Consortium2010–202220.9Both15,036390.259%Geisinger2020–202229.9CNV3000120.400%SPARK2010–202220.9SV11,052150.136%Lurie Children's2009–202220.3CNV6143140.228%Ambry Genetics2018–202227.6Both7365130.177%PARIS (Paris Autism Research International Sibpair)2010–201014.7CNV5657100.177%SickKids Toronto2012–202020.0Both20252.475%CHLA2012–202222.0CNV112310.089%Labcorp2017–202226.5CNV44,000410.093%GeneDx2014–202522.7Both86,2591610.186% Note: Raw data indicating the years of data, the CDC autism prevalence for these years, if the data source provided CNV data, SV data, or both, the total autism cohort seen per site, and the number of PMS cases within those sites. Lastly the percentage of PMS cases per site.Abbreviations: CNV, copy number variant; SV, sequence variant. Autism cohorts and SHANK3 frequency. Note: Raw data indicating the years of data, the CDC autism prevalence for these years, if the data source provided CNV data, SV data, or both, the total autism cohort seen per site, and the number of PMS cases within those sites. Lastly the percentage of PMS cases per site. Abbreviations: CNV, copy number variant; SV, sequence variant. Extrapolation to National Prevalence After applying the adjustments to extrapolate the national PMS prevalence implied by each study, the weighted average across all studies was 11.60 cases per 100,000. The final adjustment incorporated cases who were diagnosed before the onset of autism symptoms, bringing the prevalence estimate to 13.7 cases per 100,000 among the population age 0–17 years (95% CI 10.02–18.60 per 100,000). Assuming minimal disease‐specific mortality for PMS during this age range, the US prevalent population is approximately 17,700 persons with PMS age 0–17 years. Discussion In this study, we used a multisource analytic approach to estimate the prevalence of PMS. Our estimate, 13.7 per 100,000, is substantially higher than previously reported figures. Prior estimates vary widely by publication; a commonly cited range is 2.5–10 per million births (0.25–1 per 100,000) (Cammarata‐Scalisi et al. 2022), though the methodology behind this figure is unclear. Other reports offer estimates closer to this study, such as 1 in 30,000 births (Burdeus‐Olavarrieta et al. 2023), though again without transparent methods. To our knowledge, the current estimate represents the highest prevalence published to date. Our findings suggest that more than 45,000 individuals in the United States are currently living with PMS, assuming an average lifespan. In contrast, fewer than 1800 are enrolled with the PMS Foundation (PMSF), underscoring the significant underdiagnosis of this condition (PMS Foundation 2025). This discrepancy likely reflects multiple factors. Some families may not be aware of the Foundation or may choose not to register, but the most significant factor is that many children with PMS remain undiagnosed. Barriers to diagnosis are substantial and multifaceted. These include access to genetic testing, insufficient insurance coverage, lack of referrals to genetic specialists, and gaps in genetic counseling. Even when testing is ordered, many clinical genetic testing laboratories inadequately sequence SHANK3, which is a challenging gene to sequence, resulting in more missed diagnoses (Hames et al. 2025). Despite long‐standing recommendations, the rates of genetic testing in individuals with autism and developmental delay remain low in clinical practice (Peabody et al. 2015; Brown et al. 2025; Abreu et al. 2024; Vande Wydeven et al. 2012; Moreno‐De‐Luca et al. 2020). This is especially concerning as genetic diagnoses are becoming increasingly actionable. Beyond providing families with a better understanding of their child's condition and future trajectory, a genetic diagnosis may qualify individuals for clinical trials or targeted therapies that could significantly alter outcomes. This is true in PMS, where several clinical trials, including at least one gene therapy, are ongoing. It is therefore critical that all individuals with neurodevelopmental disorders are offered genetic testing as early as possible. This recommendation aligns with the care guidelines, recommending either genome sequencing or exome sequencing with a microarray for all individuals with a concern for global developmental delay or intellectual disability (Rodan et al. 2025). Other techniques to predict genetic disease prevalence exist, including statistical methods to predict de novo sequence variants on a gene level. Lopez‐Rivera et al. (2020) predicted de novo rate for 101 genes associated with neurodevelopment. The incidence of SHANK3 protein truncating single nucleotide variants was estimated at 0.72 (0.65–0.73) per 100,000. After adjustment for the rate of single nucleotide variants to insertion/deletions (~1:2.6) and the rate of SV to CNV (~1:1), the estimate from this method would be ~5.3 per 100,000. These methods depend on ever‐improving estimates of mutation rates and are based, in part, on high‐quality sequence coverage. Mutation rate estimates for SHANK3 are still improving, and, as indicated above, whole‐exome and whole‐genome analyses of SHANK3 still result in incomplete coverage because of very challenging sequences in the gene. Given the evolving estimates of mutation rates and problems with gene coverage, we suggest that our estimates are closer to ground truth but recognize that there are significant limitations with both estimates. To better integrate the laboratory and research datasets, we filtered laboratory data to include only cases with autism as an indication for testing, in line with the eligibility criteria for the research cohorts and to facilitate extrapolation using CDC autism prevalence estimates. However, test indication fields are often incomplete, which may have led to exclusion of PMS cases with undocumented autism diagnoses. In fact, over 40 PMS diagnoses identified at Labcorp were excluded due to missing phenotypic information in the test indication. Quest Diagnostics, a laboratory that agreed to participate, did not have enough completed indication fields to meet their reporting threshold, and therefore was not able to contribute their prevalence estimate to the final manuscript. Using more broad capture terms, such as developmental delay or intellectual disability, was considered as alternative methods; however, the estimated population prevalence of these features is less well characterized. Another limitation is the challenge of reproducibility for other genetic disorders. Data collection was logistically complex and time consuming. Many potential collaborators did not respond after three attempts, declined to participate, or ultimately were unable to share data due to resource constraints. Clinical laboratories often lacked the capacity to support data requests for research purposes or had internal guidelines not allowing them to share total testing numbers. Clinical centers often lacked the capacity to provide a total number of patients with completed testing, as that was not a metric recorded. While our methodology produced what we believe is a high confidence estimate for PMS, it may not be feasible to apply the same strategy to other rare disorders. Given the low rate of participation across site types, it is also plausible to assume potential bias in those who participated, for example, labs who remember diagnosing PMS or centers who remember following PMS patients. We leveraged a diverse set of genetic data to generate a population‐level estimate of PMS prevalence in the US. By focusing on individuals with autism who underwent genetic analysis, and applying adjustments for age, subtype, and time period, we extrapolated to the broader US population using CDC and published PMS data. Our findings indicate that PMS affects approximately 13.7 per 100,000 individuals, substantially higher than previously thought, highlighting the urgent need to improve access to genetic testing and diagnosis for neurodevelopmental disorders. Funding This project was funded by CureSHANK and the Seaver Autism Center. The work of D.L. was funded by Neuren Pharmaceuticals. Ethics Statement This project was reviewed by the Program for Protection of Human Subjects at the Icahn School of Medicine at Mount Sinai and was deemed Human Subject Exempt. Conflicts of Interest T.L. consults for the Phelan McDermid Syndrome Foundation. D.L. consults for Neuren Pharmaceuticals. M.H. consults for BioLogic Pharma Solutions and Neuren Pharmaceuticals A.K. consults to Neuren Pharmaceuticals, PYC Therapeutics, and receives research support from Jaguar Gene Therapy. P.S. consults for Neuren Pharmaceuticals and Jaguar Gene Therapy. L.G. and L.S. executive at Neuren Pharmaceuticals and may hold stock or stock options. Supporting information Figure S1: Illustration of the methodology for prevalence calculation.