Michael’s missed genetic diagnosis

Through interviews with Michael’s sister and his general practitioner as well as from screening previous medical records, we tried to retrospectively obtain relevant clinical information on Michael’s disorder and its course with particular focus on features potentially related to WBS. Michael’s sister, who also was his legal guardian, agrees with publication of the postmortem data on Michael.

Three publicly available portrait images of Michael were analyzed with the deep convolutional neural networks DeepGestalt, a technology powering Face2Gene (FDNA Inc., Boston, USA) [17], and its successor GestaltMatcher that was trained on the phenotypes of 830 different Mendelian disorders [4]. For each photo a coordinate is computed in a high-dimensional clinical face phenotype space (CFPS) that is populated by portraits of 8857 molecularly diagnosed individuals [5]. The cosine distance between each of these data points is the gestalt score and represents the syndromic similarity. An ordered list of gestalt scores results in the differential diagnoses that we measure and discuss in the following. For each disorder a positive likelihood ratio (PLR) can be defined, which is based on the entire cohort of individuals who are affected by the disease or not. The PLR is the probability of the specific test result in people who do have the disease to the probability in people who do not. The test result is positive if the gestalt score is below the computed threshold. For disorders with a highly characteristic phenotype and a PLR above 10, a gestalt score below the threshold can be considered as strong evidence supporting the diagnosis [6].

For longitudinal data, which are a time series of portrait photos of an individual at different ages, a trajectory through the CFPS can be computed. This pseudo-walk through time allows the gestalt scores for each of the differential diagnoses to be added up. This approach improves the signal-to-noise ratio and makes it possible to reduce artifacts from posture, facial expression, lighting conditions, and age [7]. Likewise, similarities to other disorders changes over time, and the analysis of longitudinal data can downgrade false positives.

According to the clinical records, Michael had fallen from the baby scales (from a height of presumedly 1 m) at the age of 10 weeks without signs of trauma or concussion. At 11 days after this event, Michael experienced a coma-like episode over 4 h with unconsciousness, pallor, hypoventilation, muscular hypotonia, and loss of pupillary reflex. Four weeks later, there was the first seizure-like episode with apnea, cyanosis, and clouded awareness lasting for approximately 1 min. This last event was considered a laryngospasm. Over the following 1.5 years, the aforementioned coma-like episodes recurred every few weeks to months, sometimes accompanied by extended positioning of the arms and foaming at the mouth. From today’s perspective, these episodes might equal non-convulsive status epilepticus. In addition, absence-like seizures of only a few seconds had also been observed. In 1945 at the age of 3 years, Michael started to experience occasional generalized tonic–clonic seizures. Over the years, anti-seizure medication included barbiturates and phenytoin among others. Seizures remained refractory to these medications.

In early childhood, there was apparently no marked delay in motor development. Michael had been able to walk adequately, and easily learned to feed as well as to dress himself. There was, however, a marked discrepancy between his verbal communication skills and fluid reasoning. His expressive verbal skills were described as very good with a distinguished vocabulary. He enjoyed play on words and liked reading. He had a remarkable memory for names and places. He had an affinity to music and even was able to recognize composers of classical music plays. By contrast, he had marked difficulties in decision-making and needed help and support in many aspects. He displayed friendly behavior, a gentle temper, and an outgoing personality.

In 1954 at the age of 12 years, generalized tonic–clonic seizures occurred every 6–8 weeks and were sometimes followed by hours of nausea and vomiting, which led to hospitalization at the children’s department of Mara Hospital, Bethel, Germany. Michael’s clinical records from back then revealed a height of 140 cm (−1.49 SDS; P7), weight of 31.6 kg (−1.46 SDS; P7), and microcephaly with a head circumference of 49 cm (−3.94 SDS; < P1). He had a low frontal hairline, pes planovalgus, strabismus convergens, weakness of the left n. facialis as well as of the right n. abducens, spastic paraparesis of the legs (right > left), spastic ataxia, and mild athetoid and choreiform hyperkinesia.

Repeated lumbar punctures, pneumo-encephalopathy, and ventriculography delivered normal results. The EEGs showed severe generalized dysrhythmia and hyperexcitability with continuous (perhaps status-like) spikes.

At the age of 15 years, seizures disappeared while being under the medical care of Prof. Dieter Janz.

Michael did not have signs of autistic behavior, sleep disorder, or anxiety. There were no dental or skeletal abnormalities, no hyperlaxity, and no problems of the joints or connective tissue (other than pes planovalgus). He had normal hearing and a normal voice. He had no intestinal or organ anomalies, particularly no apparent heart defect. Laboratory findings during routine check-ups did not reveal any abnormalities and in particular no hypercalcemia. He did not have diabetes. Overall, he was of relatively good health, also at an advanced age. In the past few years, he had mildly progressive coordination problems and became dependent on a wheelchair. Decades ago, he had lost one eye for reasons that could not be reconstructed. The other eye had myopia, most likely since childhood. Approximately 4 months prior to his death, Michael was diagnosed with a brain tumor that resembled a glioblastoma. Surgical intervention was not opted for; thus, a histological validation was not possible. Michael died in 2023 at the age of 80 years.

According to several photographs, Michael had a few dysmorphic features, such as bitemporal narrowing, broad nasal tip, long philtrum, thick vermilion of the upper and lower lips, wide mouth, full cheeks, and long face.

For all three portrait images, the distances from the high-dimensional CFPS were projected into 2D with MDS and visualized for the best-known photograph of Michael from the website of the Michael Foundation (https://​www.​michael-foundation.​de/​ueberstiftung/​EN/​ueberstiftung.​php, Fig. 1). This dimensional reduction can distort the actual distances from within the CFPS.

Michael’s unique and specific cognitive profile with strengths in verbal memory and language skills combined with more severely impaired cognitive function in the range of mild intellectual disability is indicative of WBS. This suspicion is further supported by his outgoing personality with a gentle and kind temper. Similar to Michael, height below average as well as microcephaly each occur in 30–50% of individuals with WBS [8]. Lastly, his facial gestalt immediately appeared suggestive of WBS.

However, some clinical anomalies frequently observed in individuals with WBS were not seen in Michael, such as cardiovascular disease, connective tissue abnormalities, and growth as well as endocrine abnormalities. Also, his bitemporal narrowing with broad nasal tip could have been—in part—the physical–morphological expression of administration of phenytoin during the Michael’s growth phase [18].

Epilepsy has frequently been diagnosed in individuals with WBS [9]; however, its exact prevalence in this disorder remains unknown. The spectrum of epileptic seizures observed in WBS comprises focal dyscognitive, focal motor, myoclonic, atonic, tonic, generalized tonic–clonic, seizures, and epileptic spasms [9‐11]. In some individuals, epilepsy was explicitly described as refractory [11, 12]. However, in at least one case, this was assigned to a second underlying genetic disease [11]. Thus, Michael’s refractory epilepsy appears compatible with WBS, although it may not necessarily have had the same etiology as his remaining phenotype.

Williams–Beuren syndrome belongs to the class of syndromes with a highly characteristic facial gestalt, and the positive likelihood ratio for correctly recognizing this disorder in an actual case is over 90%. Since age can be a confounder in the analysis, the best results are obtained if comparisons can be made with multiple images showing the individual at different ages. For Michael, three portraits were analyzed and gestalt scores consistently showed WBS as the likeliest diagnosis.

Michael’s symptoms, particularly his specific cognitive profile as well as his facial dysmorphic features, analyzed both by the human eye and artificial intelligence therefore establish WBS as the most likely clinical diagnosis by far.

Would an earlier diagnosis of WBS have made any difference to Michael or his family, relatives, or caregivers? The answer to this question is complex.

Yes, it would have made a difference in taking care of Michael. Recommended evaluations following the diagnosis of WBS comprise (a) screening for behavioral concerns, including attention, anxiety, and sleep disorders; (b) cardiovascular imaging; (c) assessment of reduced vision, hearing loss, and dental anomalies; (d) checking for hypercalcemia; and (e) thyroid function tests [8]. As Michael did not have relevant clinical problems in these areas (or it was already taken care of, e.g., myopia), he apparently did not require specific medical care. However, caregivers would generally have been better prepared if a diagnosis of WBS had been known earlier, even if Michael’s individual treatment would likely have stayed unchanged. No particular recommendations exist to treat epilepsy in WBS.

Yes, it would have made a difference to the family, particularly to his parents. Obviously, even after 80 years and long after the death of both parents, an exogenic cause (fall from baby scales) was considered a possible etiologic explanation for Michael’s disorder [2]. Knowledge of an inborn genetic disorder may have relieved the parents from a potential feeling of guilt. Furthermore, the anticipated potential proof of a de novo origin of a common WBS-associated microdeletion on 7q11.23 would have allowed for counseling of the family on inheritance and recurrence risks.

Yes, it would have made a difference in the perception of his disorder. Michael would perhaps have been recognized as an individual with WBS with refractory epilepsy rather than an individual with primarily refractory epilepsy whose unique but specific additional symptoms remained disregarded or perhaps attributed to being a consequence of his epilepsy and not being a complex genetic disorder. It, furthermore, would have helped to shed light on the usually barely understood late and geriatric course of an otherwise well-known genetic disorder.

Thus, genetic testing of older individuals is helpful not solely for the affected individuals themselves, but also because a clear diagnosis may help in the individualization of care, in establishing precision medicine approaches, and last but not least in guiding families and providing knowledge on the course and the inheritance pattern of the respective disorder [3, 4, 13‐16]. Michael’s case should remind us that we should never overlook the search for an underlying diagnosis.