The four types of Phelan McDermid syndrome

David has PMS Type 1
Originally created 12 September 2020
Updated 31 January 2022

I am going to describe four types of Phelan-McDermid syndrome (PMS). By type, I mean four different genetic arrangements that can result in PMS. These are technically called “genotypes”. The four genotypes do not address the mechanisms that create each genotype. For example, my son, David, has Type 1 (Genotype 1) from an unbalanced translocation, but others have PMS Type 1 from de novo deletions or ring chromosomes. The types can be used by parents as a very simple short-hand to share information about their child.

Phelan McDermid syndrome Type 1, terminal microdeletion

This is the original 22q13.3 microdeletion syndrome described by Katy Phelan and Heather McDermid. It is also the most common type among people identified with PMS.

A microdeletion occurs when multiple genes on a chromosome are removed. Most microdeletions are too small to see under a microscope, but Katy Phelan has described some microdeletions that are large enough to be visible. Deletion sizes range from about 0.2 Mb to 9.3 Mb, with an average of 4.5 Mb. Nearly all of the microdeletions observed are terminal deletions (continue to the end of the chromosome). Terminal deletions much smaller than 0.2 Mb do not include SHANK3 and do not produce a syndrome.

Phelan McDermid syndrome Type 2, interstitial microdeletion

An interstitial microdeletion is a deletion that removes multiple genes, but does not extend to the end of the chromosome. For PMS Type 2, the deletion does not reach SHANK3, which is near the end of the chromosome.

Interstitial deletions that produce PMS, but do not disrupt SHANK3 are relatively rare. Some scientist have argued against including them in the definition of PMS. However, a recent scientific consensus paper ( justifies the strong reasoning for including Type 2 in the definition of PMS.

Phelan McDermid syndrome Type 3, single SHANK3 rare variant

A rare variant is an atypical version of a single gene. PMS caused by a rare variant of SHANK3 is called Type 3. Sometimes this has been called a “mutation” of SHANK3. That terminology is not always technically correct, so “variant” is preferred.

Type 3 is currently the second-most common example of PMS. However, rare variants of SHANK3 have been found in large collections of DNA from people with autism spectrum disorder (ASD). Although only 1 to 2% of people with ASD might have a rare variant of SHANK3, this still a potentially large population of people who might have PMS Type 3. More research is needed.

Phelan McDermid syndrome Type 4, heterozygous deleterious variants

You child does not have PMS Type 4. PMS Type 4 is included here for completeness. Individuals can inherit two copies of unusual variants of SHANK3, one from each parent. This can produce PMS as a recessive disorder. There may be cases of Type 4, but they would be very rare, indeed.


The consensus paper (see the link, above) has provided a clear statement about how the PMS diagnosis should be applied. This is the classification that geneticists and scientist should use.

  1. PMS-SHANK3 related      (Types 1, 3, 4)
  2. PMS-SHANK3 unrelated   (Type 2)

Note that most cases of PMS fit one of the four PMS types. However, the reality of genetics is that some cases can be very messy. The four genotypes of PMS are not always clearly segregated in any given individual. An individual can have an interstitial deletion, yet also have a pathological variant of SHANK3. This could be described as both PMS Type 2 and PMS Type 3. The classification system from the consensus group, PMS-SHANK3 related versus PMS-SHANK3 unrelated,covers all possible cases without any overlap.

Sometimes two different genotypes are actually very similar

Most PMS Type 1 and PMS Type 3 cases are very different from each other. An individual with a large terminal deletion is likely to have problems not shared by an individual with a SHANK3 variant. With Type 1 there are many important genes involved, some known to produce profound effects. In Type 2, any effect would only result from disturbing SHANK3. Yet, someone with a small terminal deletion (PMS Type 1) might be very similar to someone else with PMS Type 3. Although in different categories, they could end up being very similar.

PMS is a syndrome

PMS is defined by its genetics. However, central to all syndromes is that the individual must have a phenotype reflecting the syndrome. There are some individuals with interstitial deletions and other individuals with with SHANK3 variants who simply do not have significant features of Phelan-McDermid syndrome. These are very interesting cases, but they cannot be called PMS.

This simple system of PMS Types 1 to 4 can help parents quickly share the basic genetics of their children. PMS Types 1 and 2 have a deletion size associated with them. PMS Type 3 does not. Some manifestations of PMS, like lymphedema are not seen with Type 3. Sharing your child’s PMS Type can be an ice breaker during introductions and help a parent share key information without having to be a genetics expert.


Earlier posts related to this one:

Understanding deletion size

Gene deletion versus mutation: sometimes missing a gene is better.

PMS, IQ and why interstitial deletions matter

How can the same deletion have such different consequences?

We need to study interstitial deletions to cure PMS

Defining Phelan McDermid syndrome


15 thoughts on “The four types of Phelan McDermid syndrome

  1. So where does Ring 22 with complete loss of Shank 3 fall under? Which type?
    Thoroughly enjoyed this blog. Deserves a double read. Speak in lay terms as much as possible.


    • Glad you enjoyed the post. Your geneticist is the best source for interpreting your test results. Many Ring 22 chromosomes cause terminal deletions. In those cases, the result is PMS Type 1.


  2. Andy, et al., as we in the medical, clinical, & scientific community begin to increase engagement in biological therapy clinical trails and work to engage the pharmaceutical industry to invest in PMS, it would seem prudent to lump your arbitrary Type 1, 3, and 4 into a single category. They all involve the SHANK3 gene. As a field, I think the jury is still out as to whether the SHANK3-related biology is one thing or many different things, but if we separate the pie into smaller and smaller pieces before we have clear evidence to support clearly different mechanisms in humans, we may be shooting ourselves in the foot. As you know, likely gene deleterious de novo mutations (falling under your Type 3 category) are currently included in the same clinical trials as your Type 1 category, especially if the biological target is the synapse or a pathway related to SHANK3. If we do not know that the biological mechanisms in humans are different, what is your rationale to “split” them (Type 1 and Type 3)?


    • Craig. Your input is always welcome. I think you make an important point about engaging the pharmaceutical industry and promoting clinical trials. You and I have had many conversations on the topic of PMS classifications and I think you get credit for noting that I am a splitter and you a lumper! We are likely to never agree. But, I am pleased to have this opportunity to defend my approach.

      I am not worried that pharmaceutical companies may have to advertise for “PMS Types 1, 3 and 4”. Pharmaceutical companies seem rather resilient to these sorts of complexities. PMS Types should be a separate entry field in the PMSF International Registry, but pharmaceutical companies can still leverage the more detailed genotype data when selecting subjects for a study. The classification adds no new burden.

      Phenotype studies like De Rubeis et al (2018) on point mutations are specific for PMS Type 3 and represent an important category in both the scientific literature and something that the pharmaceutical industry should, perhaps, consider with great care. It is also important to note that nearly every animal study of “Phelan McDermid syndrome” is a study of, essentially, either PMS Type 3 or PMS Type 4. To my knowledge, PMS Type 1, the most common form in humans, has never been studied in animals. (cf. iPSC studies).

      Science is not the only important venue for establishing a meaningful classification scheme. When I listen to parents try to discuss the phenotype traits of their children, especially when looking for help with a problem their child has, I experience the dire need for parents to have a common language. Parents often deal with medical problems like lymphedema or cardiac arrhythmia. Some of these issues do not exist in PMS Types 3 or 4. Families benefit from knowing where to turn for advise and whether their own experiences are relevant to another family’s issue.

      Finally, from a scientific and medical perspective (sharply focused by my own experience as a parent), I do not approve of lumping all SHANK3 disruptions into the same bin when other high pLI genes clearly contribute to the core features of PMS in PMS Type 1. I have laid out my reasoning both in this blog space and in Mitz et al (2018). Although I am convinced by my own arguments, the best demonstration would be recruitment and careful phenotyping of participants with PMS Type 2. The results of such a study are of crucial importance to testing pharmaceuticals specifically targeting the SHANK3 gene. Without a clear picture of PMS Type 2, we might miss the value of a drug targeted for SHANK3. What does PMS look like after SHANK3 is fixed in PMS Type 1 participants? PMS Type 2 studies can tell us.

      The PMS classification I have proposed is of huge potential value to clinical studies, science and a families supporting each other. Too much lumping and these benefits go away.


      • Andy, I think I see your point about symptoms for parents. This may change as we learn more. I would suggest that these “types” be changed to “genotypes”. That would provide a means of “splitting” for the families while still allowing a simplified system to “lump” for certain types of advocacy and to maintain as much unity as is reasonable among families, advocates, et al. This is not political, but rather strategic and pragmatic. It harms no one and has the upside of helping as many as humanly possible. There is strength in numbers. Science would dictate, for example, that we “split” these genotypes even further than you have proposed in the name of accuracy and to determine the actual genotype/phenotype correlations. As you suggested, many of our famies’ genotypes are rare enough that comparison with others may be inaccurate. The goal here, in my opinion, is and should be to maintain a common sense of purpose, to advocate as strongly as possible for the group, and to be nimble enough to evolve as our scientific understanding progresses.

        Liked by 1 person

      • I am glad you recognize the social benefits of establishing the four PMS Types. However, the categorization strategy is science based.
        Lumping PMS Type 1 and PMS Type 3 (and 4)
        into a single SHANK3 category is based largely on a common, but poorly supported assumption about the relationship between PMS Type 1 and PMS Type 3: that pathology resulting from indels and other variant disruptions of the SHANK3 gene primarily reflect haploinsufficiency. It also assumes, with little evidence, that patients with large deletions would see substantial benefits from a SHANK3 treatment. (I share that hope.) PMS Type 1 is driven, in part, by true haploinsufficiency, along with the haploinsufficiency of additional genes. PMS Type 3 is far less clear. Numerous lines of evidence suggest gain-of-function and other isoform effects.
        PMS is a developmental disorder. iPSC studies suggest that variant RNAs are active during development, a busy time for SHANK3. SHANK3 is a large gene that produces likely 100s of isoforms. As you know small differences in the genotype of mouse models create very different phenotypes. In humans, sometimes the same mutation leads to substantial phenotype differences. While these could be generic background, the other salient explanation is that the developmental trajectory diverged in stochastically different directions. Thus, PMS Type 3 is a mix between haploinsufficiency, a developmental disorder and gain-of-function pathology. We have some tools, but insufficient tools, to predict and study these pathology mechanics separately. For sure, some PMS Type 3 are very likely pure haploinsufficiency.
        Placing PMS Type 3 into the same category as PMS Type 1 promotes turning a blind eye to the enormous complexity of SHANK3 variants. PMS Type 1 includes true SHANK3 haploinsufficiency. In any individual, the phenotype of PMS Type 3 may or may not be driven by haploinsufficiency. This is the observed reality of SHANK3. Smaller genes with fewer domains and less pliotropy do not have this issue.
        Evidence from animal studies support the benefit of treating haploinsufficiency in mutation models. I don’t argue with the pursuit of treatments aimed at bolstering SHANK3 levels in adults. Perhaps many PMS Type 3 patients will benefit in some ways. Likewise, despite the pathology of large PMS Type 1 deletions likely driven by very high pLI genes lost throughout 22q13, I see merit in testing SHANK3-centric treatments. But, these are practical choices driven by a limited test population. The evidence that the core human pathology of PMS Type 1 and PMS Type 3 are a common mechanism is weak, except among limited cases, e.g. small terminal deletions and large interstitial deletions within the SHANK3 gene.
        I heartily agree that we should continue to explore the overlap between PMS Type 1 and PMS Type 3. We can hope for the best with drug studies. But, given the very different mechanisms of pathology in humans, polygenic loss and complete SHANK3 deletion in PMS Type 1 versus a mixture of gain-of-function, haploinsufficiency and developmental errors in PMS Type 3, the science and medicine of PMS is best served by keeping these PMS Types separate. At the very least, it forces scientists to remember that animal studies are ultimately low fidelity models of the human disorder.


      • The science some of us are reading shows very little genotype/phenotype correlation among various deletion sizes of your Genotype 1 category, save for a few non-brain organ issues.

        I repeat my suggestion that you call these genotypes rather than types.

        I need to disengage due to other commitments.


      • All of your points are well taken. I understand that you are simply asking for a nomenclature adjustment. I infer this is to make room for another, somewhat parallel scheme, that would benefit from using the “Type” nomenclature. Although I would like to avoid the mouthful of “PMS Genotype 1” for the sake of the parents, I am also very eager to support efforts at formalizing Phelan McDermid syndrome across the scientific and medical communities. Let’s talk on the phone so I can better understand your goals and so any change in the nomenclature I use does not (again) accidentally interfere with your goals.


  3. As you know, Andy, this topic has been a priority for our scientific advisory committee with the intent to acquire an ICD-10 code, often requested by families. You’ve been advised of the project and kept informed along the way. I always respond to your queries. The work of the SAC is not mentioned here and the implication is that there has been no movement. The pandemic certainly slowed the process but there is already a draft response signed by every member. We recently were made aware of efforts not unlike ours among other European researchers and felt that we should be speaking with them as well so we don’t have further division. You may not like the timetable but I do wish you’d acknowledged that this is being addressed and prepared for a peer reviewed journal through committee consensus.


    • Because I am not a member of the PMS Foundation SAC, the board of directors, or any related committee, I have not been privy to any details of the Foundation’s efforts. Contrary to what your note seems to suggest, I have only heard the most general of comments about SAC efforts. From my perspective, this has been a closed-door process. I should not be expected to know more than the Foundation’s interest in the matter and know that some colleagues agree with my strong belief in including PMS Type 2 cases. In truth, I have heard very little from colleagues. They have protected the privacy of the process. I respect it, too. I also respect the scientists who make up the current PMSF SAC. Of the members I know, the Foundation has done well to find a group of scientist of high caliber and genuine interest in families.

      That said, from my perspective there has been a 10 year drought in developing a classification scheme for PMS. Note, the PMS Foundation has published self-contradictory descriptions of Phelan McDermid syndrome over the years. My initial attempts to encourage the establishment of a PMS classification scheme was to propose a process for others to use. I developed a scheme to facilitate a conversation (described in one of my blogs). That proposal, years ago, was rejected and I was instructed by the Foundation to not participate as a parent or as a scientist.

      In May of 2015, in response to the lack of progress, especially progress in educating families and other scientists, I started writing blogs on PMS. Many of these blogs explored various aspects of classification. How does a deletion compare to a mutation (variant)? How can we use clustering to identify genotype-phenotype correlation? Why should interstitial deletions be included in PMS? I have moved very cautiously, establishing a scientific basis for my classifications. My presentation of the four types of PMS is a culmination (albeit also a work in progress) of slow, deliberate and I hope thoughtful approach to a very important issue. My previous comment shows how I envision the importance.

      My blog faults the scientific community for infighting, which I see both at meetings and in the published literature. It is not aimed at the Foundation or its SAC. The decision to publish this blog was simply because it was time. The argument that a different formalized classification scheme is “just around the corner” looses luster after many years. I hope the Foundation will recognize the value of the classification scheme I have proposed and embrace its positive characteristics. I encourage American, Asian and European researchers to consider my rationale, and to recognize that this classification scheme benefits science, medicine, pharmaceutical companies, and — importantly — families.


  4. Hi Andy, thank you for this thoughtful blog. I agree that some sort of grouping of the different types of mutations associated with PMS is necessary not just for scientific research, and more targeted interventions, but most importantly for family members to help them understand what is going on. As an alternate viewpoint, within the field of autism spectrum disorder, scientists have argued for years about how to possibly stratify the disorder according to phenotype and genotype and there continues to be ongoing debate. While the genetics of ASD are obviously more complex than the genetics of PMS, other scientists may not completely agree with the way you have organized them. I look forward to an ongoing discussion about this issue, as I believe it can inform ASD research. However, other scientific voices should weigh in on these categories you have described.


    • I welcome other scientific voices, although historically some of these voices have been clearly more political than scientific. Some decisions about classification are arbitrary (or at least a matter of preference), like how much to split and how much to lump.

      Autism spectrum disorder received a new classification scheme in DSM 5, partly because of the revolution in genetics. I don’t know if PMS can instruct ASD, but I think PMS is a fertile ground to understand the interaction between common variants (polygenic risk) and deleterious rare variants. In PMS you do not have a large research population. You essentially cannot do a genome-wide association study. However, with PMS Type 1 you know with exquisite detail which genes are the rare variants. There are also far too few studies on the cumulative impact of heterozygous losses. In PMS Type 1 we expect a cumulative impact on intellectual function with the inclusion of more high pLI genes with larger deletions. Also, 22q13.3 has some rather special genes that impact mitochondria. New information on SULT4A1 regulation of mitochondrial activity and the suspected role of RABL2B on mitochondrial transport to synapses might provide important insights into brain energetics.

      Thus, I see PMS as a sort of research tool in addition to a disorder that deserves treatment.


  5. Quote – “Terminal deletions smaller than about 133 kbp do not include SHANK3 and do not produce a syndrome.”
    My child has a partial Shank3 terminal deletion of 112kb (breakpoint somewhere b/w exons 8-9). ID, Autistic Behavior, and receptive language issues. Does this fall into Type 1?


    • I have see somewhat different ways to describe deletion size. Sometimes the size reported is based on the markers (probes) on a microarray. So, the distance between the deleted probes might be 112 kb. However, when a geneticist evaluates the results, he/she applies their professional judgement, and might conclude that it is a terminal deletion. The distance from Exon 8 to the end of the chromosome is just under 183 kb (genome build hg19). I am not a geneticist and I don’t know why the distance between probes is reported rather than the distance to the end of the chromosome. That type of reporting seems common and might be the prescribed way to describe a deletion. Regardless, your report describes the breakpoint quite clearly and makes the assessment that your child has a terminal deletion. From a genotype point of view, this is PMS Type 1.

      But genetics can get complicated.

      It turns out that the DNA near the breakpoint of a 22q13.3 deletion goes through some changes in order to keep the end of the chromosome covered by certain DNA patterns. If it is not covered, a ring chromosome 22 can result. It would take high resolution mapping to see exactly what has happened to the first 8 exons of the SHANK3 gene. Let’s just suppose two cases: 1) the 8 exons and their promotor region are intact, and 2) the 8 exons are mixed in with nonsense DNA making their transcription unlikely. In the first case, you have a SHANK3 variant (mutation). Although the PMS Type is PMS Type 1, it has produced a variant. One could argue that it is also PMS Type 3. In the second case it is simply PMS Type 1.

      Just to muddy the waters even more, your child’s terminal deletion falls into the category I describe in the updated blog, where two different genotypes can be very similar. There is good evidence that the main impact of certain SHANK3 mutations (PMS Type 3) is to stop the gene’s production of the main (most important) isoforms of SHANK3. Your child’s deletion would undoubtedly have that effect, with very little other impact. Thus, a very small terminal deletion (PMS Type 1) can be quite similar to some (perhaps many) de novo SHANK3 mutations (PMS Type 3).

      I am certain this is much more than you asked! However, your child’s genotype is an opportunity to point to my statement that an individual’s genotype can be complicated. This is a silver lining, however!

      In the end, you child have a rather specific and probably very complete loss of SHANK3, with little else affected. Said that way, it is actually quite easy to describe.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.