Mount Sinai estimate puts Phelan-McDermid syndrome at about 1 in 7,300 people

Start with who was counted. Everyone among the nearly 180,000 people in the pooled data was autistic and had already had a genetic test ^[4]. The records came from ten sources, among them the commercial labs GeneDx, Labcorp and Ambry Genetics, the SPARK study and the Autism Sequencing Consortium ^[5]. So the base population is tested autistic people. To get from there to a rate for everyone, the team corrected for undiagnosed cases, for the limits of genetic testing, and for people with the syndrome who do not meet autism criteria ^[6].

The design makes sense for this condition. The syndrome comes from a deletion or mutation involving SHANK3 on chromosome 22 ^[7], and most people who have it also meet the criteria for autism ^[8]. Autistic cohorts should therefore capture most cases, and the third correction adds back the rest. The result is 13.7 cases per 100,000 people ^[2]. Divide 100,000 by 13.7 and you get about 7,299, which is where the 1 in 7,300 comes from ^[1]. The US figure is that rate multiplied out: 45,000 people at 13.7 per 100,000 implies a population of about 328 million ^[2].

The release does not report how many SHANK3 cases turned up among the 180,000 before correction, how large each correction was, or the earlier estimate it calls a major change ^[15]. Those numbers would show how much of 1 in 7,300 was observed and how much was modelled.

Tess Levy, the first author, gave the authors' explanation for the gap. "The large gap between known and estimated cases is likely due in large part to the fact that many individuals with developmental disabilities and autism are never offered genetic testing. Families may also face insurance barriers or may receive tests that do not adequately evaluate the SHANK3 gene," she said ^[10]. That fits the estimate, but the study as described measures how many cases there should be. It does not directly measure why they were missed.

The funding matters here. Neuren Pharmaceuticals initiated the study with the Seaver center and CureSHANK, according to Neuren's Rachel Groth ^[12], and CureSHANK and Neuren supported it ^[13]. CureSHANK also put out the release ^[13]. "Patients cannot benefit from these advances if they never receive a diagnosis," Groth said ^[12]. In my view the ten-source pooling is the strongest part of the design. The corrections are the part an independent group would most need to reproduce, given who paid for the work.

The patient count matters now because several clinical trials are running, including precision-medicine approaches aimed at the biology behind the disorder ^[14]. Changes to SHANK3 are thought to account for as many as one percent of autism cases ^[9]. Joseph Buxbaum, the senior author, wants testing to be universal. "We recommend that every child with autism undergo genetic testing, because knowledge is power," he said ^[11]. He also offered a timeline: "I truly believe that within the next five years, we'll see successful examples of new treatments coming from these genetic discoveries" ^[16]. That five-year timeline is his own forecast.