Facial Dysmorphology – Diagnostic Approach

Assessment of the child and adult with dysmorphic facial features: terminology, measurement, computational tools and genetic testing strategy

Literature and regulatory information updated: 29 October 2026.

0. Initial Reference Table

Facial dysmorphism is not a disease but a clinical finding that may correspond to a familial variant of no significance, a deformation due to intrauterine constraint, a disruption, the effect of a teratogen or one of the many hundreds of chromosomal and monogenic syndromes with recognisable facial features. The table brings together the reference data for the approach; the specific entities are listed in section 6.

1. Introduction

1.1. Origin of dysmorphology and definition of dysmorphism

Dysmorphology is the recognition and study of congenital defects and of the syndromes that bring them together. The term was first used by David W. Smith, in the United States, in the 1960s, to designate the study of human congenital malformations and the patterns in which they cluster [2]. The discipline grew out of a small core of clinicians in the 1950s and later incorporated research in developmental biology, molecular genetics and metabolic medicine [8]. Smith's work, now in its 8th edition, retains the sections devoted to minor anomalies and to the measurement standards that underpin clinical observation [5, 21].

A child is described as dysmorphic when his or her physical features, especially facial ones, are not usual in a child of the same age and the same ethnic background [3]. In the strict sense, the term means an abnormality of structure, and it therefore also covers internal organs affected by the same mechanism that produced the visible features; knowledge of normal fetal development is a prerequisite for interpreting it [2]. Some features are abnormal in any circumstances, such as premature fusion of the cranial sutures; others, such as cutaneous syndactyly between the second and third toes, may be either a diagnostic clue or a familial trait of no significance [2, 3].

The expression “isolated facial dysmorphism” therefore describes a starting point and not a diagnosis: one or more unusual facial features, with no major malformation or other known manifestation at the time of examination. The clinical task is to decide whether the feature is a familial or population variant, a transient deformation, a sign of a syndrome or the effect of a prenatal exposure, and to choose an investigation proportionate to that probability.

1.2. Pathogenetic classification of morphological anomalies

The classical classification distinguishes the mechanisms by which a structural anomaly arises, because each mechanism has its own implications for prognosis and recurrence risk [2]. The definitions were later revised by the Elements of Morphology consortium in an article on general terms [22].

Source: Firth and Hurst [2]; general terms revised by the Elements of Morphology consortium [22].

Aase subdivided phenotypic anomalies into anomalies proper (malformations, deformations, disruptions and dysplasias) and minor variants; the latter are separated into minor anomalies, with a prevalence equal to or below 4% in the general population, and common variants, above that threshold [6]. Smith defines minor anomalies as unusual morphological features without serious medical or cosmetic consequence, whose value lies in indicating altered morphogenesis or in serving as a clue to a specific pattern of malformation [5].

1.3. The Elements of Morphology project

The descriptive language of dysmorphology evolved in an uncoordinated way, with synonymous, ambiguous and sometimes pejorative terms. In 2005 and 2006, an international group of 34 clinicians met in Bethesda and in Rome to standardise this nomenclature. The initial aim was to review the 683 terms of the London Dysmorphology Database; the group began with the cranium, the face and its components, and the hands and feet, in six teams, whose articles were published in 2009 in the American Journal of Medical Genetics [8]. The six articles cover the head and face, the periorbital region, the ear, the nose and philtrum, the lips, mouth and oral region, and the hands and feet [7, 23–27]. The series continued with general terms for congenital anomalies, in 2013, and with the trunk and limbs, in 2022 [22, 28].

The principles of the project remain useful in practice. Each feature has a preferred term, a definition and guidance on how to observe and measure it. Whenever possible, an objective definition (for example, a measurement more than 2 standard deviations from the mean) and a subjective one (the appearance of the feature) coexist, and the objective one is preferred; when there are several standards for the same measurement, each objective definition is equally valid [8]. The terms avoid implying a pathogenetic mechanism, list synonyms and terms to be abandoned, and propose alternatives for designations considered pejorative. Radiographs were excluded from the definitions, which are limited to surface examination [8].

Two principles deserve emphasis. The first is the absence of any judgement of normality: many features exist in isolation in the general population and their frequency varies between populations, so an objective finding is “abnormal” only in the statistical sense and the individual simply has the feature [8]. The second is the rejection of bundled terms, which gather several components under a single label. “Large nose”, for example, combines prominent nose, wide nasal bridge, prominent nasal tip and wide nasal base; these terms presume a pathogenesis, hide the components and have mostly been eliminated [8]. The authors did not include differential diagnoses by feature, considering that syndromes are recognised by combinations and not by isolated features, and they assigned that function to dysmorphology databases [8].

1.4. From description to computable data: HPO and next-generation phenotyping

The Human Phenotype Ontology (HPO), described by Robinson and colleagues in 2008 [2], organises and defines the phenotypic features of human diseases and enables computational inference, comparison by semantic similarity and machine learning [9]. The HPO definitions for facial features reproduce those of Elements of Morphology, including the objective thresholds [1, 7], and the 2024 release added 2239 terms and 49,235 annotations since the previous report, translations and a browser in ten languages other than English, the Medical Action Ontology (MAxO) for management actions, and integration with the GA4GH Phenopacket Schema in electronic health records [9]. Hennekam and Biesecker summarised this transition in 2012 with the formula “next-generation sequencing demands next-generation phenotyping” [2].

The expression “next-generation phenotyping” now refers mainly to the computational analysis of facial photographs. In 2019, DeepGestalt, trained on more than 17,000 images of more than 200 syndromes, placed the correct syndrome among the top ten suggestions in 91% of 502 images in a simulated clinical setting [14]. In 2022, GestaltMatcher began to represent each portrait as a point in a facial phenotype space defined by 17,560 patients with 1115 rare diseases, which makes it possible to bring together patients with the same molecular diagnosis even when the disease was not part of the training set [15]. These tools and their limitations are covered in section 5.

1.5. Scope

This article covers the diagnostic approach to facial dysmorphism: history, examination and measurement, terminology, interpretation of minor anomalies, support tools, choice of genetic tests, counselling without a diagnosis and follow-up. Specific syndromes are presented only as feature-oriented differentials.

2. Molecular Genetics and Pathophysiology

2.1. Embryology of the face and origin of minor anomalies

The face is formed from processes that fuse within narrow time windows, and many dysmorphic features are the visible record of a disturbance that occurred during those weeks. The auricle develops from six auricular hillocks, mesenchymal derivatives of the first and second branchial arches, around the first branchial groove; during embryonic and early fetal life, the ears ascend to the level of the eyes [29]. This is why low-set and posteriorly rotated ears, which usually coexist, reflect delayed morphogenesis, since this is the normal position of the auricle early in fetal life [5]. Preauricular tags, often with a cartilage core, correspond to accessory hillocks, and in craniofacial microsomia they lie along the line joining the tragus to the oral commissure, which marks the junction of the mandibular and maxillary processes [5, 29].

The slant of the palpebral fissures depends on the rate of early growth of the brain above the eye relative to that of the face below it: microcephaly with a narrow frontal region tends to produce an upward slant, and maxillary hypoplasia a downward slant [5]. The epicanthic fold corresponds to redundant skin over the inner canthus, almost always due to a low nasal root and, less often, to excess skin, as in cutis laxa; it is common in early childhood and disappears as the nasal bridge becomes prominent, whereas a unilateral epicanthic fold suggests torticollis [5].

The hair and eyebrow pattern provides particularly precise dating. Hair follicles grow down into the dermis between 10 and 16 weeks, and their slope is determined by the tension exerted on the skin by the growth of underlying structures, especially the brain; by 18 weeks the pattern is fixed [5]. Severe microcephaly may lead to absence of the parietal whorl (25%) or to an upsweep frontal whorl (70%), and gross alterations of the hair pattern often suggest severe intellectual disability, as they indicate a brain alteration preceding 16 weeks. Primary craniosynostosis, by contrast, does not alter the hair pattern, because the calvarium is not yet formed at that stage. The widow's peak corresponds to a lower than usual intersection of the periocular fields of hair growth suppression, which occurs in hypertelorism and frontonasal dysplasia; absence of the auricle is accompanied by absence of the sideburn, and a rudimentary auricle, as in Treacher Collins, may be associated with a tongue of hair over the cheek [5].

Palmar creases form at 11–12 weeks and dermatoglyphics between 13 and 19 weeks, so gross alterations of these patterns indicate an abnormality of hand shape or function during that period [5].

2.2. Mechanisms: deformation, disruption, teratogenesis and genetic alteration

Many minor anomalies, perhaps most, are deformations caused by mechanical forces on normal tissue; most minor anomalies of the auricle present at birth result from intrauterine constraint [5]. Deformation may, however, be secondary to a primary malformation, and this is the presumed mechanism of the association between minor anomalies and major malformations [5]. The skin also provides information: redundant skin indicates tangential traction from external constraint, whereas tight, thin skin over the joints indicates early immobility of neurological cause in a fetus that had room to move [5].

Before considering a genetic cause, it is advisable to look for signs of abnormal physical forces in utero (deformation), of vascular interruption (disruption) or of teratogenic exposure, such as fetal alcohol syndrome, fetal anticonvulsant syndrome or diabetic embryopathy [2]. Germline genetic alterations disturb developmental programming and produce malformations and dysplasias by several mechanisms: failure of structural integrity (COL2A1 variants in Stickler syndrome), failure of regulation of cell number (ASPM in autosomal recessive primary microcephaly), failure of cell migration (MID1 in Opitz syndrome) and failure of a developmental “switch”, such as transcription factors and signalling pathways [2]. The number of genes involved ranges from many (aneuploidies), to several (microdeletions such as Williams), to a single gene, in which a single variant may be lethal, such as NM_000142.5:c.1948A>G p.(Lys650Glu) in FGFR3 in thanatophoric dysplasia type 2 [2].

The list of mechanisms to consider includes chromosomal anomalies, monogenic disorders (structural proteins, transcription factors, disturbances of expression such as imprinted genes), the effects of several genes, multifactorial disorders, mainly environmental causes, and chromosomal, monogenic or expression mosaicism [3].

2.3. Population and familial variation

Facial morphology varies between families and between populations, and this variation conditions the interpretation of any feature. Almost any minor anomaly may exist as a usual trait in a family; Smith illustrates a father and daughter with a chin dimple and a mother and daughter with a palatal torus [5]. Published measurement standards were obtained mainly in white populations and may not be accurate for other groups [21].

The expression of syndromes also depends on ancestry. In a study of 65 individuals with Down syndrome from 13 countries, brachycephaly, ear anomalies, clinodactyly, a gap between the first and second toes and excess nuchal skin were significantly less frequent in African individuals; on digital facial analysis, only the inner canthal and nasal alar angles were common to the Caucasian, African and Asian groups when compared with controls of the same background [33]. These data underpin the electronic atlas of malformation syndromes in diverse populations of the National Human Genome Research Institute [34].

2.4. Composite phenotypes

A child may have two distinct genetic disorders, and the resulting phenotype may not correspond to either of them. In the Deciphering Developmental Disorders (DDD) study, 17 of 1133 children with severe developmental disorders (1.5% of the cohort, about 5% of the children with a diagnosis) had pathogenic variants in two different genes, with a composite phenotype [35]. In a series of 7374 consecutive diagnostic exomes, 101 of the 2076 patients with a diagnosis (4.9%) had two or more molecular diagnoses, with distinct or overlapping phenotypes, at a proportion lower than that predicted by a model of independent occurrence, which suggests under-recognition [36]. The figure of “about 5%” cited in reference texts for the coexistence of two disorders therefore refers to children with a diagnosis and not to all the children assessed [3, 35].

3. Clinical Spectrum and Manifestations by System

3.1. Major and minor anomalies: the significance of the number

The relationship between minor anomalies and major malformations is the oldest and most useful empirical finding in dysmorphology. In the study by Marden, Smith and McDonald (1964), in newborns examined at the surface, 85% had no minor anomaly, 13.4% had one, 0.8% had two and 0.5% had three or more; the proportion of babies with at least one major malformation was 1.4%, 3%, 11% and 90%, respectively [5]. With one minor anomaly, the risk of a major malformation did not increase appreciably; with two, it was five times that of the general population; with three or more, most children had multiple major malformations. In two later studies, by Méhes and by Leppig, the proportion of newborns with three or more minor anomalies and a major malformation was 26% and 19.6%, lower than Marden's figures and attributed to differences in design [5]. The operational conclusion is that any infant with three or more minor anomalies should be assessed for major malformations, many of which are occult [5].

Source: Marden et al. (1964), Méhes et al. and Leppig et al., according to Jones, Jones and del Campo [5]. The three cohorts are of North American and European newborns from different periods and with different examination criteria; the values are not interchangeable.

Major congenital anomalies are present in 2–3% of singleton newborns and the proportion doubles with follow-up, as internal anomalies are identified; in monozygotic twins the frequency is two to three times higher, about 10% [2]. External minor anomalies are concentrated in regions of complex and variable morphology: the face, auricles, hands and feet [5].

3.2. Findings by facial region

The table summarises the standardised definitions of the most commonly used facial features, with the objective criterion where one exists, and the population frequency data available in the sources. Any given feature should always be interpreted in the light of age, ethnic background and the parents' features.

The definition of low-set ears differs between sources. The Oxford Desk Reference describes it as superior insertion below a line between the outer canthus and the occiput [29], whereas Smith and Elements of Morphology, followed by the HPO, use the horizontal line passing through the inner canthi, with the head in the Frankfurt plane [1, 5]. The standardised definition is the latter.

3.3. Other findings of significance in facial assessment

Some minor anomalies outside the face are routinely included in the assessment. A single palmar crease is present unilaterally in about 4% of normal infants and bilaterally in 1%, and is twice as common in males; fifth-finger clinodactyly up to 8° is normal; syndactyly between the second and third toes not exceeding one third of their length is a normal variant, and beyond that a minor anomaly [5]. In isolation, this syndactyly is a common autosomal dominant trait with variable expressivity, whereas the Y-shaped form occurs in 79.5% of children with Smith-Lemli-Opitz syndrome (OMIM 270400) [6, 39]. Blind sacral dimples are present in 2% of newborns and, without pigmentation, hair tuft or mass, behave as a normal variant; in a series of 75 babies with a dimple there were no anomalies on spinal ultrasound [6]. Inverted nipples are present in about 3% of women but, in a child with hypotonia and developmental delay, suggest a congenital disorder of glycosylation or propionic acidaemia [6].

Prominent lateral palatine ridges reflect a lack of tongue pressure against the palate and indicate a prolonged functional alteration, such as hypotonia; absence of the lingual frenulum associated with a single central incisor suggests holoprosencephaly [5].

3.4. Evolution of features with age

Facial features change throughout life, and this evolution is both an obstacle and a diagnostic tool. The hypoplastic nasal bridge of the infant with 22q11.2 deletion may become high in the child and adult [38]; the face in Williams syndrome becomes coarser and more angular in adulthood [30]; in Aarskog-Scott syndrome the facial features tend to normalise with age [37]; in Coffin-Lowry syndrome they become more marked over time and may be recognisable in some female carriers [30]; in mandibulofacial dysostosis with microcephaly, some patients catch up in head circumference in adulthood [29]. In some syndromes, diagnosis only becomes possible after a few years, and serial observation with photographs is part of the method [3]. Progressive facial coarsening has a particular significance, because it suggests a storage disorder and an autosomal recessive recurrence risk [30].

4. Clinical Recognition and Red Flags

4.1. When a facial feature merits genetic assessment

The decision to investigate is based on the combination of features and the context, not on an isolated feature. A single facial feature, common in the population (frequency above 4%) and present in one of the parents, is in most cases a familial variant [5, 6]. The level of suspicion increases when the feature is distinctive, that is, when it does not occur as a normal or familial variant and appears only in a small number of conditions; premature fusion of the cranial sutures is the classic example [2, 3]. Quantification also provides guidance: the further growth measurements deviate from the normal centiles, the greater the probability of a genetic diagnosis [3].

4.2. Emergency settings

Facial dysmorphism is rarely an emergency in itself, but some patterns require immediate action because of the associated malformations. Robin sequence, with micrognathia and cleft palate, may cause upper airway obstruction [2]. Bilateral choanal atresia, present in CHARGE syndrome and in carbimazole or thiamazole embryopathy, compromises breathing in the newborn [38]. A coarse face in an infant with hypotonia, metabolic acidosis, hyperammonaemia or cytopenias suggests a metabolic disorder prone to decompensation, such as propionic acidaemia [6]. Suspected craniosynostosis, including Crouzon and Apert syndromes, with signs of raised intracranial pressure or significant hypertelorism with midline malformations, requires assessment by a craniofacial team [32, 37].

4.3. Newborn screening and primary care

There is no population screening for facial dysmorphism. The newborn examination is the main opportunity for detection, and biochemical newborn screening detects some metabolic disorders that may have late facial expression; confirming that newborn screening was performed and was negative is part of the initial assessment in primary care [10, 40]. In primary care, the American Academy of Pediatrics recommends that, when there is an abnormal dysmorphological examination in a child with developmental delay, a case summary with a clinical photograph be sent to the genetics centre and the consultation awaited before ordering tests; if the examination is normal or equivocal, CMA, fragile X testing and metabolic screening may be started while awaiting referral [10]. In children with congenital heart disease, the presence of dysmorphism of the face (OR 2.57), forehead (OR 2.55), neck (OR 2.24), periorbital region (OR 1.57) or hands and feet (OR 1.90) was independently associated with a genetic diagnosis in a cohort of 1008 infants, of whom 24.1% had a diagnosis [42].

5. Diagnosis: Criteria, Tests and Algorithms

There are no diagnostic criteria for facial dysmorphism as a finding. Diagnosis is the process that leads from the finding to the cause, and it is organised in four steps: taking the history, systematic examination with measurement and photography, formulation of hypotheses with the support of databases and computational tools, and choice of tests.

5.1. Preparation and structure of the consultation

The consultation begins before the family arrives. It is helpful to know what question the referrer is asking, to review the clinical record and radiographs and, if useful, to contact the paediatrician or the family itself [2]. Geneticists are called upon to give a diagnostic opinion, interpret a test already performed, discuss the genetic aspects, propose investigations, provide information on prognosis and therapeutic options, calculate the recurrence risk and discuss prenatal diagnosis [2]. Increasingly, patients arrive with sequencing results ordered by other professionals, and it falls to the geneticist to assess the relevance of these findings to the clinical presentation [2].

The structure suggested in the Oxford Desk Reference has eight steps: introductions and parental concerns; observation of the child during the consultation, including spontaneous language and interaction; history; physical examination with photographs; investigations; conclusions with risk assessment and counselling; correspondence; and follow-up [2]. Even when the diagnosis seems obvious at first sight, restraint is advised: building the relationship with the family and confirming that the history and examination support the impression before communicating it [2]. One should ask, and not assume, who the adults accompanying the child are [2].

5.2. History

The history carries the same diagnostic weight as the examination [3]. The essential elements are summarised in the table.

5.3. Dysmorphological examination

The examination rests on meticulous observation, measurement and documentation [2]. The first overall impression classifies the face as normal, with apparently familial dysmorphic features or with non-familial dysmorphic features; this is followed by a region-by-region description, using standardised terms and, when in doubt, everyday words or a drawn sketch [2, 41].

Observation should take place under standardised conditions. Elements of Morphology recommends that the head be in the Frankfurt horizontal plane, with the facial and neck muscles relaxed, the eyes open, the lips lightly in contact and a neutral expression, and that the observer's face be at the same height [7]. The head should be viewed from the front, from both sides and from above, especially to distinguish positional plagiocephaly from craniosynostosis [32].

The Oxford Desk Reference checklist covers: growth parameters (length or height, weight, OFC, with centile and standard deviation), proportions (arm span, upper and lower segments, sitting height), build, skin (birthmarks, neurocutaneous signs, pigmentation, texture), cranium and fontanelles, ears (size, position, structure, folds, pits, tags), eyes (deep-set or prominent, spacing, epicanthic folds, length, slant and eversion of the palpebral fissures, eyebrows), nose (length, bridge, tip, nostrils), philtrum, mouth (size, lips, tongue, palate, frenula, gums, teeth), chin, hands, feet and nails, chest and nipples, heart, abdomen and genitalia, spine, neurological examination, communication and behaviour [41]. The skin deserves particular attention, asking specifically about “birthmarks” and looking for pigmentary changes suggesting chromosomal mosaicism [3]. Hearing aids should be removed, because the tubing hides preauricular pits [29].

5.4. Measurement and standards

Measurement corrects the visual impression, which is frequently misleading: a low nasal root makes the eyes look widely spaced, and obesity makes the hands look small [21]. The record should state the number of standard deviations; saying that “the OFC is below the 0.4th centile, at −5.5 SD” conveys more information than saying only that it is below the 0.4th centile [3]. The OFC may also be compared with the height age to decide whether the head is small for age or proportionate to stature [41].

The norms collected in Smith were obtained mainly in white populations, separated by sex where there is a difference, with paired structures measured on the right, and many come from Feingold's Boston study; for other measurements the reader is referred to the Handbook of Physical Measurements [21]. Measurements made on photographs differ from those made in person: palpebral fissures measured on photographs are systematically shorter, and a norm obtained with photographic software fell 1.6 SD below the mean of the direct-examination curve [31]. The norms should therefore correspond to the measurement modality used.

5.5. Clinical photography and video

Photography is the best record of unusual features and allows their evolution with age to be followed [2, 3]. The minimum set includes the face from the front, the profile, the hands and any unusual feature; a photograph of the child with both parents helps to separate familial features from potentially significant ones [2, 3]. Video is useful when facial movement is abnormal, which occurs in several syndromes, especially microdeletion syndromes, and in cases of asymmetry of movement or characteristic behaviours [3, 32, 41]. Sharing photographs with colleagues, at departmental meetings or in expert networks, increases the probability of recognition and trains doctors in training [3]. Photographs used in computational tools require specific consent; in a qualitative study of 15 German geneticists, the consent process was one of the main barriers to integrating Face2Gene into routine practice [43]. Sharing portraits of patients with rare diseases for scientific purposes is compatible with the European Union General Data Protection Regulation (2016/679), which protects individuals and allows data to circulate for research [16].

5.6. Formulating hypotheses: gestalt, “handles” and databases

The Oxford Desk Reference proposes a six-step diagnostic framework: (1) answer basic questions (single or multiple malformation; likely multiple anomaly syndrome; deformations explicable by the pregnancy; contribution of family history); (2) think about the possible mechanisms; (3) ask whether something similar has been seen before, that is, recognise a gestalt; (4) consult the literature; (5) consult databases; (6) ask colleagues for help, sharing photographs [3]. Gestalt diagnosis, based on recognising a pattern seen before, depends on experience; the doctor in training should present the history and signs to a senior before drawing conclusions [2].

Databases are most useful when the search starts from distinctive features (“good handles”), such as a median cleft lip, and not from common or subjective features, such as “low-set ears” [3]. A weak handle is a feature that exists as a normal variant or in many syndromes [2]. The most widely used databases include POSSUM, the London Dysmorphology Database, OMIM, Orphanet, DECIPHER (which, since 2004, has displayed genomic variation of any size and phenotypic relationships annotated with the HPO), ClinVar, gnomAD and Gene2Phenotype, which annotates genes by allelic requirement, variant consequence and phenotypic category [13]. Coding the phenotype with HPO terms also enables computational gene prioritisation: in the case described in section 10, adding dysmorphological terms such as “short columella” moved the causal gene up to first place in the phenotypic similarity ranking [44].

5.7. Computational facial phenotyping

Facial analysis tools transform a frontal photograph into a numerical representation and compare it with patients with a known diagnosis. There are two main approaches. Classifiers, such as DeepGestalt (Face2Gene), assign a probability to each syndrome in the training set; GestaltMatcher calculates a facial descriptor that places the patient in a phenotype space, where the distance to other patients measures similarity, which allows it to bring together patients with ultra-rare diseases that were not part of the training set [14, 15]. PEDIA combines the facial score, HPO-based phenotypic similarity and the variant deleteriousness score to rank the candidate genes from an exome [16].

These studies measure classification accuracy in selected sets of patients with a confirmed diagnosis, mainly with syndromes for which sufficient images existed; they do not measure the probability of diagnosis in an unselected child with an isolated facial feature. Comparisons between studies are indirect, because galleries, populations and criteria for success differ. Performance improves with the number of syndromes and of patients per syndrome in training and plateaus from about 150 syndromes [15]; GestaltMatcher accuracy did not correlate with disease prevalence, which suggests that ultra-rare diseases have facial distinctiveness similar to that of more common ones [15]. Both tools performed better in rare than in ultra-rare syndromes in the Indian cohort [47].

Ancestry bias is the best-documented limitation. Human assessment depends on the assessor's background, and computational assessment on the patient's background; training with photographs of the population concerned corrected part of the difference in Down syndrome [45]. The PEDIA authors acknowledge that European ancestry is the most represented in the training data and the one that achieves the best performance [16]. In 2026, a paper was published on Eurocentric bias in the teaching of dysmorphology and medical genetics [49].

Interaction between the clinician and the tool carries its own risks. In a pilot study with 44 geneticists and 18 images, a correct prediction by the tool improved clinicians' accuracy, but an incorrect prediction reduced it; visual explanations by saliency maps were little valued and did not change decisions, which were influenced mainly by the probability presented by the tool [50]. Prudent use is therefore that of a second opinion that generates or ranks hypotheses, to be confirmed by examination, history and molecular testing.

Integration into variant interpretation is the most recent application. The PP4 criterion of the ACMG/AMP recommendations applies when the phenotype is highly specific for a disease with a single genetic cause, and it is traditionally assessed subjectively [44]. In 2026, Hsieh and colleagues derived, for Mowat-Wilson syndrome, GestaltMatcher score thresholds corresponding to the PP4 evidence levels of the ClinGen Bayesian framework: supporting 0.5111 (target likelihood ratio ≥2.08), moderate 0.5672 (≥4.3), strong 0.6609 (≥18.7) and very strong 0.7632 (≥350), based on 61 cases and 610 controls [44]. This is a proof of concept for one syndrome, in a small set and with simulated exomes, and the authors themselves stress that these tools complement the geneticist and do not replace them [44]. Studies on PIGQ-CDG and on germline histonopathies show, respectively, a shared gestalt that supports diagnosis and the absence of an established gestalt when data are scarce and the phenotype changes with age [51, 52].

5.8. Choice of tests

The choice depends on the probability of the clinical diagnosis. If the diagnosis is likely, targeted confirmatory testing is performed (karyotype or FISH for a recognisable aneuploidy, single-gene test or panel, enzyme study in a storage disorder) [10, 30]. If there is no likely diagnosis, testing is genomic. Guidance has changed in three stages. In 2010, the International Standard Cytogenomic Array (ISCA) consensus, based on 33 studies with 21,698 patients, recommended CMA instead of karyotyping as the first-tier cytogenetic test in patients with developmental delay or intellectual disability, autism spectrum disorder or multiple congenital anomalies, with a yield of 15–20% against about 3% for karyotyping (excluding Down syndrome and other recognisable chromosomal syndromes); in the text, the weighted mean yield of the 33 studies was 12.2% [11]. In 2014, the American Academy of Pediatrics adopted CMA as the first-tier test for every child with global delay or intellectual disability of unknown cause, with an estimated yield of 12%, together with fragile X testing and metabolic screening targeted at treatable causes [10]. In 2021, the American College of Medical Genetics and Genomics (ACMG) made a strong recommendation that exome or genome sequencing be considered a first- or second-tier test in patients with one or more congenital anomalies with onset before 1 year, or with developmental delay or intellectual disability with onset before 18 years [12].

In the full text, the recommendation is worded as first- or second-tier exome or genome sequencing, “guided by clinical judgement and often by shared decision-making after CMA or targeted testing”; the overall certainty of the evidence was rated very low, owing to the absence of randomised trials and reliance on observational studies, and that of comparative yield as moderate [12]. The economic model of the Ontario health technology assessment showed that exome sequencing used as a first or second test (for example, after CMA or targeted testing) yields more diagnoses at lower cost than when used only after extensive testing [12]. The same guideline keeps targeted testing first when the presentation strongly suggests a specific diagnosis, including suspected chromosomal disorder, known family history and situations in which sequencing may not be diagnostic, such as the methylation alterations of Prader-Willi and Angelman syndromes or fragile X syndrome [12]. The guidance diverges on the position of CMA. The 2010, 2014 and 2017 recommendations place it first; the 2019 consensus and the 2021 ACMG guideline allow exome or genome sequencing at the outset, with CMA remaining useful when the exome does not detect copy number variations with the same sensitivity. The systematic review that supported the 2021 guideline included 167 articles, mainly case reports and series with fewer than 20 patients, and the most frequently reported outcomes were changes in clinical management and in reproductive decisions; only two studies assessed mortality, morbidity or quality of life [53].

In the Clark meta-analysis, trio testing doubled the probability of diagnosis compared with the proband alone (OR 2.04; 95% CI 1.62–2.56), and clinical utility, measured as a change in management, was 27% with genome, 17% with exome and 6% with CMA [17]. In the DDD, trio recruitment was the factor with the greatest effect on the probability of diagnosis (OR 4.70; 95% CI 4.16–5.31); the probability was lower in children born between 22 and 27 weeks (OR 0.39), in those exposed in utero to antiepileptic drugs (OR 0.44), in children of diabetic mothers (OR 0.52) and in probands of African ancestry (OR 0.51) [19]. These data indicate that environmental and teratogenic causes compete with genetic ones and that the under-representation of some populations in reference databases reduces yield.

Fragile X testing remains a separate test in children with developmental delay, in the AAP and UK guidance; the syndrome affects about 1 in 5000 births and the facial features are not very apparent in the young child [10, 40]. Metabolic screening targeted at treatable causes is recommended even in the absence of specific signs by the AAP and the UK guidance, although its yield is low (1–5% in the AAP series; 0.2–4.6% in a review by the American Academy of Neurology) [10, 40]; the Oxford Desk Reference, citing the 1997 consensus, describes a very low yield of untargeted metabolic screening [3]. In a child with microcephaly without healthy siblings, maternal phenylketonuria should be considered [3]. Brain MRI is not routine with normal OFC and normal neurological examination; the yield was 28% with neurological signs and 7.5% with isolated delay [3, 10].

In the coarse face, the baseline investigation includes thyroid function, urinary metabolic screening (mucopolysaccharides, oligosaccharides, amino acids and organic acids), plasma amino acids, biochemistry with CK, glucose, alkaline phosphatase and phosphate, a search for vacuolated lymphocytes, skeletal radiographs, ophthalmological assessment, echocardiogram, abdominal ultrasound and, depending on the suspicion, leucocyte enzymes, very-long-chain fatty acids, insulin or skin biopsy [30]. Skin biopsy for fibroblast karyotyping is indicated in the presence of streaky pigmentation along Blaschko's lines, asymmetry or 3/4 syndactyly with 2/3 finger syndactyly and bulbous tips (diploid/triploid mosaicism), and is required for the diagnosis of Pallister-Killian syndrome, whose isochromosome 12p is present in fibroblasts but usually absent in lymphocytes [3, 30].

Most intoxication-type inborn errors of metabolism (amino acid and organic acid catabolism) do not interfere with embryofetal development or produce dysmorphism, despite prenatal exposure to toxic compounds; a coarse face at birth is rare in lysosomal disorders but, when present, has diagnostic value [54]. Facial dysmorphism with a characteristic metabolic marker occurs in Zellweger syndrome, Smith-Lemli-Opitz syndrome, PMM2-CDG and pyruvate dehydrogenase deficiency; maternal metabolic disturbances (untreated maternal phenylketonuria, alcohol, diabetes, drugs, riboflavin deficiency) also produce it, and maternal phenylketonuria may cause a fetopathy with intrauterine growth restriction that mimics fetal alcohol syndrome [54]. Defects in the synthesis of small molecules (serine, including Neu-Laxova syndrome, glutamine, asparagine; purines and pyrimidines, such as Miller syndrome due to DHODH; molybdenum cofactor or sulphite oxidase deficiency) and of complex molecules (cholesterol synthesis, glycosylation) complete the list [54]. In the newborn or infant with dysmorphism and/or seizures, the Oxford Desk Reference adds to the investigation plasma biotinidase, very-long-chain fatty acids, transferrin isoelectric focusing, cholesterol and 7-dehydrocholesterol, lysosomal enzymes and, for pyridoxine- or pyridoxal phosphate-dependent epilepsies, CSF pipecolate, urinary α-aminoadipic semialdehyde and CSF pyridoxal phosphate, followed by a therapeutic trial of pyridoxine; a wide anterior fontanelle suggests Zellweger syndrome [55].

5.9. Algorithm

1. History with a three-generation pedigree, systematic dysmorphological examination with measurements in SD, photographs and comparison with the parents [2, 41].
2. Classify: familial or common variant, deformation, disruption, teratogenic exposure, recognisable syndromic pattern or non-specific dysmorphism [2, 3].
3. If three or more minor anomalies, or a suspected malformation: look for occult malformations (echocardiogram, renal ultrasound, hearing and ophthalmological assessment according to the pattern) [5, 6].
4. If a clinical diagnosis is likely: targeted confirmatory testing [10].
5. If there is no likely diagnosis and there are congenital anomalies, developmental delay or intellectual disability: first-tier CMA or exome/genome, preferably as a trio, plus fragile X when there is developmental delay and metabolic screening targeted at treatable causes [10–12, 17].
6. If abnormal neurological examination, non-familial microcephaly or macrocephaly: brain MRI [10].
7. Facial analysis tool and databases as support for ranking hypotheses and interpreting variants [16, 46].
8. If no diagnosis: periodic reanalysis of genomic data, clinical and photographic reassessment and discussion in an expert network (ERN ITHACA) [3, 19, 20].

In truly isolated facial dysmorphism, in an infant with normal development, routine cytogenetic testing is not indicated, as the Oxford Desk Reference emphasises with regard to facial asymmetry [32]; in this scenario, observation over time and comparison with the parents are the first tool.

5.10. Limitations of the tests

A negative result does not exclude a genetic cause. CMA does not detect truly balanced rearrangements nor, in general, low-level mosaicism; apparently balanced rearrangements occur in about 1 in 500 patients with intellectual disability (de novo in 1 in 2000), but many have submicroscopic imbalances, and aneuploidy mosaicism was observed in 1% of 2019 patients [11]. Conventional 20-cell karyotyping detects mosaicism from about 14% [11]. The recommended CMA resolution is 400 kb or better across the genome, against about 5 Mb for karyotyping [11]. Mosaicism restricted to one tissue may be absent from blood, as in Pallister-Killian syndrome, in which diagnosis requires skin fibroblasts [30]. A negative exome or genome leaves a substantial fraction of children unexplained: in the DDD, about 59% of probands remained without a diagnosis even after iterative reanalysis [19]. Each genome contains 4 to 5 million variants, and in one study 122 of 406 (27%) variants published as causing severe disease were common polymorphisms or lacked direct evidence of pathogenicity [56]. The recommendations of MacArthur and colleagues for assigning causality insist on first assessing genes already associated with similar phenotypes, on requiring several unrelated patients to implicate a new gene, on not treating several in silico predictors as independent lines of evidence and on not assuming complete penetrance [56].

5.11. Report, variants of uncertain significance and segregation

The report should state the gene, the reference transcript with version, the HGVS nomenclature at cDNA and protein level, zygosity, inheritance (de novo or inherited, when the parents have been tested) and the ACMG/AMP class with the criteria applied. CNVs are classified as pathogenic, of uncertain significance and likely benign, and their interpretation should be carried out by a geneticist and genetic counsellor in collaboration with the laboratory [10]. Parental studies distinguish de novo from inherited variants and are the first step when faced with a variant of uncertain significance [11]. Detailed phenotyping contributes to classification: the case in section 10 shows a ZEB2 variant classified as of uncertain significance by the laboratory and reclassified as likely pathogenic after dysmorphological assessment and quantification of the PP4 criterion [44]. In a study cited by the same authors, clinical interpretation diverged from laboratory interpretation in 18 of 130 results (13.8%) [44]. In the DDD, 22% of probands had variants of uncertain significance in genes strongly associated with monogenic developmental disorders [19].

5.12. Prenatal diagnosis

Without an aetiological diagnosis, prenatal diagnosis is generally limited to the second-trimester anomaly scan when the child has structural malformations attributable to the condition, such as a heart defect or polydactyly; it should involve fetal medicine and its limitations should be explained [3]. Normal ranges of interocular distance by gestational age exist, but only very marked hypertelorism, or very marked nasal hypoplasia, can be predicted with confidence by ultrasound [37, 38]. When the genetic cause is identified, prenatal diagnosis and preimplantation genetic testing become possible (section 8) [57].

6. Differential Diagnosis

6.1. Principle: differential by combination of features

The differential of an isolated feature is of limited use, because syndromes are recognised by combinations [8]. The useful approach organises the differential from the most distinctive feature and its association with growth, development, malformations and family history. The first distinction is between what is not a syndrome (familial variant, deformation, disruption) and what is; the second, within syndromic causes, between prenatal exposures, chromosomal anomalies and monogenic disorders.

6.2. Differential table by leading feature

6.3. Main differentials

Deformations and familial variants are the most frequent differentials of isolated facial dysmorphism. Deformations have a history of constraint (oligohydramnios, twins, abnormal presentation, bicornuate uterus, a sensation of the “head being stuck”), arise in intrinsically normal tissue and improve when the force ceases [32]. Large or asymmetrical ears in the newborn are often a consequence of constraint or torticollis, and the newborn ear, especially in preterm infants, may look crumpled and should be reassessed weeks later [5, 29]. Familial traits are recognised by examining the parents and comparing photographs; before counselling an apparently sporadic case of hypertelorism, the parents' ocular distances should be measured [37].

Among exposures, the fetal alcohol spectrum is the most important differential because of its frequency and because its facial features are minor anomalies with a quantitative definition. The criteria revised in 2016 require, for fetal alcohol syndrome, at least two of three facial features, growth deficiency, deficient brain growth or structural anomalies, and neurobehavioural impairment, and they expressly include genetic disorders and other teratogens in the differential [31].

Source: Hoyme et al. 2016 [31]. Neurobehavioural impairment is defined as ≥1.5 SD below the mean; isolated adaptive deficits are not sufficient. As OFC, growth and many cognitive characteristics have moderate to high heritability, data on the biological parents should be considered.

Storage disorders are distinguished from syndromic causes of a coarse face by progression, organomegaly and skeletal and ocular changes; metabolic investigation is guided by the suspicion, and enzymatic and molecular confirmation is straightforward when the diagnosis is likely [30]. Among syndromic causes, Noonan syndrome (OMIM 163950; part of the RASopathy group) [78], Williams, Smith-Magenis, Pallister-Killian, Costello and Coffin-Siris syndromes produce a coarse appearance without storage material [30]. Other syndromes with a recognisable facial gestalt that frequently enter the differential are Kabuki syndrome (OMIM 147920) and Cornelia de Lange syndrome (OMIM 122470) [79, 80].

Facial asymmetry requires identifying the affected side and knowing whether there is a malformation or only a difference in size; routine cytogenetics is not indicated in an infant with normal development and asymmetry alone [32]. Hemihyperplasia should prompt a search for signs of Beckwith-Wiedemann and Silver-Russell syndromes, and progressive hemiatrophy in an older child suggests progressive hemifacial atrophy (Parry-Romberg) [32].

7. Treatment and Risk Management Strategies

7.1. Principles

Facial dysmorphism has no pharmacological treatment of its own, and none of the sources consulted describes drugs directed at the finding; therapy depends on the aetiological diagnosis and is that of the entity identified. Management of the finding has three aims: to identify and treat the associated malformations and functional deficits, to establish the aetiological diagnosis whenever possible and to support the family when a diagnosis is not reached. A precise diagnosis allows accurate counselling, discussion of prognosis and natural history, targeted surveillance, prenatal diagnosis and access to support groups [2].

7.2. Management of associated malformations and deficits

Antiepileptic medication may modify facial features and cause gingival hypertrophy, which should be taken into account when interpreting a coarse face [30].

7.3. Management of the child without a diagnosis

When the diagnosis is not established, the plan includes the services the child and family need and a schedule for diagnostic reassessment agreed with the family and the attending physician [10]. Iterative reanalysis of genomic data is effective: in the DDD, the yield rose from 28% in the first analysis of 1133 children to about 41% in 13,449 probands with iterative analysis and clinical and computational classification [19, 35]. Discussing the case in expert networks is another route; the Clinical Patient Management System (CPMS) of the European Reference Networks allows cases to be shared with experts in other countries with data protection, and ERN ITHACA used it in its first year for 28 very rare cases from 13 centres in 7 countries [20]. Phenotype-matching tools, such as GestaltMatcher, can link unrelated families with the same novel disease [15].

7.4. Framework of recommendations across jurisdictions

7.5. Clinical utility studies and their limitations

Clinical utility was measured mainly as a change in management, and not as an improvement in health outcomes; comparisons between studies are indirect and yields depend on cohort selection, generally from genetics centres and with severe phenotypes [17, 53]. Studies of facial tools measure classification accuracy and not clinical benefit, and the 2026 study shows that an incorrect suggestion can reduce the clinician's accuracy [50]. The sources consulted do not include studies on the yield of genomic testing in truly isolated facial dysmorphism, without developmental delay or malformations.

7.6. Developments in the last 12 months

In 2026, preprints, not yet peer-reviewed, also appeared on federated learning with GestaltMatcher and on generating synthetic faces to protect privacy; they were not used as evidence.

7.7. Pregnancy, transition and other settings

In the pregnancy of a mother whose child has undiagnosed dysmorphism, counselling focuses on the recurrence risk and on the limitations of prenatal diagnosis (section 8). Identifying a teratogenic exposure, such as alcohol, has preventive consequences for future pregnancies [31]. In adolescents and adults, childhood photographs are particularly useful, because many features become less marked or change with age [2]. Patients of non-European ancestry are at greater risk of misinterpretation, both by human assessment and by computational tools, which justifies comparison with relatives and the use of appropriate norms and atlases [33, 34, 45].

8. Reproductive Options and Family Counselling

8.1. With an aetiological diagnosis

When the diagnosis is established, the recurrence risk, carrier testing and prenatal diagnosis follow the specific condition, and counselling is generally straightforward [3]. Testing the parents is always part of the process: in several syndromes with variable expressivity, a parent may have mild features that are only recognised with targeted examination or molecular testing, as in Crouzon syndrome [38], and in conditions such as craniofacial microsomia there are families with dominant transmission even though most cases are sporadic [32]. When a variant is de novo, the recurrence risk is low but not zero, because parental germline mosaicism cannot be excluded; the laboratory in the case described in section 10 flagged this possibility in the report [44]. In frontonasal dysplasia and in syndromes in which a de novo variant has been demonstrated, recurrence is low [37].

8.2. Without an aetiological diagnosis

The absence of a diagnosis does not preclude counselling, but it requires explaining why it was not possible to reach one and that counselling is based on the probability that the cause is genetic [3]. The Oxford Desk Reference identifies situations in which the recurrence risk may reach 25%: a previous affected child; stillbirths or late miscarriages due to fetal anomalies; a family history of a similarly affected individual (X-linked, chromosomal or dominant with variable expression); consanguinity; neurodevelopmental regression; progressive coarsening of facial features; symmetrical structural defects already described in autosomal recessive disorders; and maternal influences, such as myasthenia with transplacental antibodies, with a very high risk in subsequent pregnancies [3]. In the absence of these elements, an empirical risk of about 5% may be given, explained as a composite risk: very low for most families and high for some, and liable to change if a diagnosis is made [3]. In an undiagnosed coarse face, clear progression suggests an autosomal recessive risk, whereas a non-progressive picture with malformations suggests a syndrome with a lower recurrence risk [30].

8.3. Carriers, prenatal and preimplantation diagnosis

Relatives at risk may be identified from the pedigree, for example in X-linked inheritance [3]. Without a molecular diagnosis, it is generally not possible to offer prenatal diagnosis, except a second-trimester anomaly scan when the child has structural malformations attributable to the condition, with an explanation of its limitations [3]. With an identified cause, the range of reproductive options includes accepting the risk, having no more children, adoption, donor gametes, prenatal diagnosis and preimplantation genetic testing [57]. Prenatal diagnosis by chorionic villus sampling or amniocentesis is possible in cytogenetic disorders, in monogenic disorders with a known pathogenic variant and in the great majority of biochemical disorders, and ultrasound is useful in conditions with major structural malformations; the type of sample should be agreed with the laboratory [30, 57]. Preimplantation genetic testing involves ovarian stimulation, in vitro fertilisation and embryo biopsy; the Oxford Desk Reference reports, for monogenic disorders, pregnancy rates of 21% per oocyte retrieval and 25% per embryo transfer, with limited safety data [57]. Follow-up of the family also has reproductive value: the Oxford Desk Reference recommends maintaining review appointments, among other reasons, to discuss the recurrence risk in the light of new tests [3].

8.4. Psychosocial aspects

Making a diagnosis is satisfying, but the time spent confirming that it is correct, building the relationship with the parents and assessing their readiness to receive it is valuable when the time comes to communicate it [2]. The expression “dysmorphic features” and some descriptive terms may be felt to be pejorative; Elements of Morphology proposed alternatives and noted, for example, that “coarse” may be considered pejorative, and it is advisable to explain the meaning of the terms to parents [7, 8, 30]. Families of children without a diagnosis can be put in touch with others in the same situation through organisations such as SWAN UK (Syndromes Without A Name) and Contact a Family [3, 13].

9. Special Considerations: Population Diversity and Epidemiology

9.1. Frequency of congenital anomalies

In the United States, the Centers for Disease Control and Prevention estimate that about 1 in 33 babies is born with a birth defect [4]. The Oxford Desk Reference gives 2–3% of obvious major anomalies in singleton newborns, with this proportion doubling during follow-up [2]. Data from the EUROCAT registry of the northern Netherlands (births from 1981 to 2024, June 2026 update), which include live births, stillbirths, fetal deaths and terminations of pregnancy for fetal anomaly, give an idea of the order of magnitude of anomalies relevant to the face, per 10,000 births: anomalies of the ear, face and neck 3.3 (3.0 in live births), of which anotia and auditory canal atresia 1.5; orofacial clefts 21.8 (19.8), of which cleft lip with or without cleft palate 14.4 and isolated cleft palate 7.4; genetic disorders as a whole (syndromes, dysplasias and chromosomal anomalies) 56.2 (35.8); Down syndrome 17.8 (10.0); craniosynostosis 2.3; Robin sequence 1.2; holoprosencephaly 1.2; amniotic bands 0.8 [82]. These values describe a regional registry and not European prevalence; they count only major anomalies according to the EUROCAT definitions, and a child with several anomalies is counted more than once [82].

Craniofacial microsomia has been estimated at about 1 in 5600 births and CHARGE syndrome at about 1 in 12,000 [29, 32]. Genetic syndromes as a whole were estimated at 8% of the population in the DeepGestalt article and rare diseases at more than 6.2% of the world population in the GestaltMatcher article; both are citations by the authors and not their own measurements [14, 15].

9.2. Populations and ancestry

The measurement norms used in practice derive mainly from white North American and European children [21, 31], and the lip/philtrum guide used in the fetal alcohol spectrum was built with photographs of more than 800 white children in the USA; the criteria recommend population-standardised guides where they exist [31]. The frequency of minor anomalies differs between populations: preauricular pits are more frequent in black than in white people and twice as frequent in females [5]; in Down syndrome, several classic features are less frequent in African individuals [33]. Agreement between African and European clinicians in classifying African children as dysmorphic was only fair (kappa 0.29) [45]. In the DDD, African ancestry was associated with a lower probability of a molecular diagnosis (OR 0.51; 95% CI 0.31–0.78) [19]. The 2015 DDD cohort was about 90% of north-western European ancestry [35], which limits the generalisability of the yields.

The sources consulted do not present specific data from Portugal or Brazil on the assessment of facial dysmorphism, on the performance of computational tools in these populations or on the yield of genomic testing; atlases of diverse populations and training tools with local photographs are the responses proposed in the literature for this kind of gap [34, 45].

9.3. Special populations

Monozygotic twins have a two- to three-fold higher frequency of congenital anomalies, and frontonasal dysplasia is more frequent in twins [2, 37]. In preterm infants, measurements should be referred to corrected age and the ear may appear malformed owing to cartilage immaturity [21, 29]. Children born between 22 and 27 weeks, those exposed to antiepileptic drugs and children of diabetic mothers had a lower probability of a molecular diagnosis in the DDD, which points to non-genetic or multifactorial causes in these groups [19].

9.4. European networks and patient organisations

The relevant European Reference Network is ERN ITHACA (Intellectual disability, TeleHealth, Autism and Congenital Anomalies), dedicated to rare malformation syndromes, intellectual disability and neurodevelopmental disorders, with virtual case discussion through the CPMS (https://ern-ithaca.eu) [20]. For specific craniofacial syndromes, the relevant network is ERN CRANIO. Resources for families and professionals include Unique (rarechromo.org, information on chromosomal anomalies), SWAN UK (families without a diagnosis), Genetic Alliance UK and the National Organization for Rare Disorders in the USA [13], and, for professionals, DECIPHER, POSSUM, Orphanet, OMIM and the public GestaltMatcher database (GMDB, with 4306 images of 3693 individuals and 257 syndromes at the time of the 2022 publication) [13, 15].

10. Illustrative Clinical Case

The case published by Hsieh and colleagues in 2026 shows how dysmorphological examination, HPO coding and computational facial analysis combine in the interpretation of a variant [44].

A girl was followed from the age of 1 year because of parental concern about her development, although the ASQ-3 questionnaire at 12 months was within the reference values. The three-generation pedigree revealed no consanguinity, genetic disorders, recurrent miscarriages or birth defects; paternal ancestry was Hispanic and maternal ancestry unknown, and a brother had hypospadias. At 18 months she had marked gross motor delay, with rudimentary crawling, and limited expressive language, with delay in all ASQ-3 domains; assessment for autism spectrum disorder was inconclusive. At 21 months, an interdisciplinary assessment confirmed global developmental delay. At 23 months she underwent surgery for bilateral non-accommodative esotropia. At 3 years, neuromuscular assessment showed no weakness; there was fluctuating sialorrhoea and CK was normal [44].

Brain MRI under sedation, at 19 months, suggested subtle thinning of the body of the corpus callosum, and CMA was normal. A clinical exome identified no pathogenic or likely pathogenic variants but reported a de novo heterozygous variant of uncertain significance in ZEB2, NM_014795.4:c.3237del, p.(Tyr1080ThrfsTer28), in the last exon (exon 10), with no expected nonsense-mediated mRNA decay and absent from gnomAD. The laboratory applied the criteria PVS1 moderate, PS2 moderate and PM2 supporting and maintained the classification of uncertain significance, noting that parental germline mosaicism could not be excluded [44]. The nomenclature was confirmed in VariantValidator (GRCh38 NC_000002.12:g.144389860del; NP_055610.1:p.(Tyr1080ThrfsTer28)) [83].

At the genetics clinic, at 4 years, the child had a cheerful disposition, broad forehead, thick eyebrows, rounded nasal tip with a high bridge and short columella, low-set and posteriorly rotated ears with uplifted lobules, full lower lip, pointed chin, hypotonic facial impression, left esotropia and clinodactyly of the third to fifth toes (figure 2 of the article: https://pmc.ncbi.nlm.nih.gov/articles/PMC13326781/figure/F2/). The team concluded that the phenotype was compatible with Mowat-Wilson syndrome and reclassified the variant as likely pathogenic, adding the PP4 moderate criterion on the basis of phenotyping [44].

GestaltMatcher ranked Mowat-Wilson syndrome first in all four available photographs, with scores between 0.547 and 0.647; three images reached the PP4 moderate threshold and one the supporting threshold, according to thresholds derived from 61 patients with Mowat-Wilson syndrome and 610 controls (figure 1: https://pmc.ncbi.nlm.nih.gov/articles/PMC13326781/figure/F1/). In PEDIA prioritisation, with a simulated exome, ZEB2 ranked first, whereas the molecular score alone (CADD) placed it fifth; the progressive addition of dysmorphological HPO terms, starting with “short columella”, improved the gene's ranking (figure 3: https://pmc.ncbi.nlm.nih.gov/articles/PMC13326781/figure/F3/). The team recommended an echocardiogram and renal ultrasound for specific surveillance and, in the absence of seizures, clinical surveillance with an electroencephalogram if suspicious episodes arose [44]. The authors also applied the tool to the photograph of an infant with confirmed Mowat-Wilson syndrome (NM_014795.4:c.1489C>T, p.Gln497Ter), and the syndrome was likewise ranked first [44].

The case illustrates three points. The first is that dysmorphological assessment can change the classification of a variant, through the PP4 criterion. The second is that the terms that contributed most to prioritisation were fine facial features, such as the short columella, which depend on specialist examination. The third is caution: the authors themselves acknowledge that an experienced geneticist would have reached the diagnosis without the tool, that the thresholds were derived in a small set and with simulated exomes, and that this is a proof of concept in a single patient; the use of facial score thresholds for PP4 does not, at present, constitute practice recommended by guidelines [44].

11. Prognosis and Long-Term Follow-Up

11.1. Prognosis

The prognosis of a child with facial dysmorphism is that of its cause. An isolated familial or common feature, with no other anomalies, generally has a normal prognosis; a single minor anomaly did not appreciably increase the risk of a major malformation in the Marden cohort [5]. Deformations improve with time once the deforming force has ceased [32]. The prognosis of syndromic forms depends on the diagnosis, and the main uncertainty concerns children without a diagnosis: even in cohorts with exome and array and iterative reanalysis, about 59% of DDD probands remained without a molecular diagnosis [19]. In these children, observation of the natural history is an essential part of management and may allow diagnosis later, as tests improve and syndromes are delineated [3].

11.2. Follow-up of the child without a diagnosis

The Oxford Desk Reference proposes a second appointment to confirm that investigations are complete and that the information has been shared, a review one year later and, from then on, reviews agreed with the family, for example according to reproductive plans [3]. Reviews serve to offer new tests, confirm that no medical problems have arisen and that development is progressing, increase the probability of recognising syndromes that only become evident after a few years and discuss the recurrence risk [3]. Repeated clinical and dysmorphological assessments are more informative than a single assessment [40].

The sources consulted do not set an interval for genomic reanalysis; the AAP recommends agreeing the frequency and timing of diagnostic reassessment with the family and the attending physician [10].

11.3. Transition

In adolescence and adulthood, follow-up without a diagnosis continues to have diagnostic value, because some features become more marked and others less so, and childhood photographs allow the natural history to be reconstructed [2, 30]. The DDD inclusion criteria also accepted adults after an initial limit of 16 years, and the median age of probands was 7 years (range 0–63) [19].

12. Summary and Concluding Remarks

Facial dysmorphism is a finding and not a diagnosis. Its interpretation depends on distinguishing between familial and population variants, deformations, disruptions, teratogenic exposures and syndromes, and this distinction is made with a careful history, a three-generation pedigree, a systematic examination with measurements expressed in standard deviations, photographs and comparison with the parents. The number of minor anomalies has demonstrated prognostic value: three or more are associated with major malformations in a high proportion of newborns, whereas an isolated anomaly has little significance.

The standardised terminology of Elements of Morphology and the HPO allow features to be described with objective criteria whenever possible and transformed into computable data, useful for databases, gene prioritisation and case sharing. Facial analysis tools achieve high accuracy in sets of patients with a confirmed diagnosis, mainly of European ancestry, and are beginning to be used in variant interpretation; their performance is lower in under-represented populations and in ultra-rare diseases, and an incorrect suggestion may mislead the clinician.

When there are congenital anomalies, developmental delay or intellectual disability without a likely clinical diagnosis, genomic testing is the next step: CMA or, according to the 2021 ACMG guideline, exome or genome sequencing as a first- or second-tier test, preferably as a trio, complemented by fragile X testing and investigation of treatable causes. A negative test does not exclude a genetic cause, and reanalysis, serial clinical reassessment and discussion in networks such as ERN ITHACA keep the possibility of diagnosis open. In the absence of a diagnosis, counselling is based on the signs that point to a high recurrence risk and, in their absence, on a composite empirical risk of about 5%.

13. References

1. Human Phenotype Ontology. Definitions of terms HP:0000316, HP:0000506, HP:0012745, HP:0000369, HP:0000358, HP:0000286, HP:0000582, HP:0000494, HP:0000343, HP:0000319, HP:0000219, HP:0000347, HP:0000252, HP:0000256, HP:0000280, HP:0001999, HP:0000431 [Internet]. Bar Harbor: The Jackson Laboratory; [accessed 2026-10-29]. Available from: https://hpo.jax.org
2. Firth HV, Hurst JA. Approach to the consultation with a child with dysmorphism, congenital malformation, or developmental delay. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 4–7.
3. Firth HV, Hurst JA. Dysmorphic child. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 134–136.
4. Centers for Disease Control and Prevention. Data and statistics on birth defects [Internet]. Atlanta: CDC; updated 2026 Apr 21 [accessed 2026-10-29]. Available from: https://www.cdc.gov/birth-defects/data-research/facts-stats/index.html
5. Jones KL, Jones MC, del Campo M. Minor anomalies: clues to more serious problems and to the recognition of malformation syndromes. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 955–972.
6. Firth HV, Hurst JA. Minor congenital anomalies. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 240–241.
7. Allanson JE, Cunniff C, Hoyme HE, McGaughran J, Muenke M, Neri G. Elements of morphology: standard terminology for the head and face. Am J Med Genet A. 2009;149A(1):6-28. doi:10.1002/ajmg.a.32612
8. Allanson JE, Biesecker LG, Carey JC, Hennekam RC. Elements of morphology: introduction. Am J Med Genet A. 2009;149A(1):2-5. doi:10.1002/ajmg.a.32601
9. Gargano MA, Matentzoglu N, Coleman B, Addo-Lartey EB, Anagnostopoulos AV, Anderton J, et al. The Human Phenotype Ontology in 2024: phenotypes around the world. Nucleic Acids Res. 2024;52(D1):D1333-46. doi:10.1093/nar/gkad1005
10. Moeschler JB, Shevell M; Committee on Genetics. Comprehensive evaluation of the child with intellectual disability or global developmental delays. Pediatrics. 2014;134(3):e903-18. doi:10.1542/peds.2014-1839
11. Miller DT, Adam MP, Aradhya S, Biesecker LG, Brothman AR, Carter NP, et al. Consensus statement: chromosomal microarray is a first-tier clinical diagnostic test for individuals with developmental disabilities or congenital anomalies. Am J Hum Genet. 2010;86(5):749-64. doi:10.1016/j.ajhg.2010.04.006
12. Manickam K, McClain MR, Demmer LA, Biswas S, Kearney HM, Malinowski J, et al. Exome and genome sequencing for pediatric patients with congenital anomalies or intellectual disability: an evidence-based clinical guideline of the American College of Medical Genetics and Genomics (ACMG). Genet Med. 2021;23(11):2029-37. doi:10.1038/s41436-021-01242-6
13. Firth HV, Hurst JA. Useful resources. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 44–45.
14. Gurovich Y, Hanani Y, Bar O, Nadav G, Fleischer N, Gelbman D, et al. Identifying facial phenotypes of genetic disorders using deep learning. Nat Med. 2019;25(1):60-4. doi:10.1038/s41591-018-0279-0
15. Hsieh TC, Bar-Haim A, Moosa S, Ehmke N, Gripp KW, Pantel JT, et al. GestaltMatcher facilitates rare disease matching using facial phenotype descriptors. Nat Genet. 2022;54(3):349-57. doi:10.1038/s41588-021-01010-x
16. Hsieh TC, Mensah MA, Pantel JT, Aguilar D, Bar O, Bayat A, et al. PEDIA: prioritization of exome data by image analysis. Genet Med. 2019;21(12):2807-14. doi:10.1038/s41436-019-0566-2
17. Clark MM, Stark Z, Farnaes L, Tan TY, White SM, Dimmock D, et al. Meta-analysis of the diagnostic and clinical utility of genome and exome sequencing and chromosomal microarray in children with suspected genetic diseases. NPJ Genom Med. 2018;3:16. doi:10.1038/s41525-018-0053-8
18. Srivastava S, Love-Nichols JA, Dies KA, Ledbetter DH, Martin CL, Chung WK, et al. Meta-analysis and multidisciplinary consensus statement: exome sequencing is a first-tier clinical diagnostic test for individuals with neurodevelopmental disorders. Genet Med. 2019;21(11):2413-21. doi:10.1038/s41436-019-0554-6
19. Wright CF, Campbell P, Eberhardt RY, Aitken S, Perrett D, Brent S, et al. Genomic diagnosis of rare pediatric disease in the United Kingdom and Ireland. N Engl J Med. 2023;388(17):1559-71. doi:10.1056/NEJMoa2209046
20. Smith M, Alexander E, Marcinkute R, Dan D, Rawson M, Banka S, et al. Telemedicine strategy of the European Reference Network ITHACA for the diagnosis and management of patients with rare developmental disorders. Orphanet J Rare Dis. 2020;15(1):103. doi:10.1186/s13023-020-1349-1
21. Jones KL, Jones MC, del Campo M. Normal standards. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 973–998.
22. Hennekam RC, Biesecker LG, Allanson JE, Hall JG, Opitz JM, Temple IK, et al. Elements of morphology: general terms for congenital anomalies. Am J Med Genet A. 2013;161A(11):2726-33. doi:10.1002/ajmg.a.36249
23. Hall BD, Graham JM Jr, Cassidy SB, Opitz JM. Elements of morphology: standard terminology for the periorbital region. Am J Med Genet A. 2009;149A(1):29-39. doi:10.1002/ajmg.a.32597
24. Hunter A, Frias JL, Gillessen-Kaesbach G, Hughes H, Jones KL, Wilson L. Elements of morphology: standard terminology for the ear. Am J Med Genet A. 2009;149A(1):40-60. doi:10.1002/ajmg.a.32599
25. Hennekam RC, Cormier-Daire V, Hall JG, Méhes K, Patton M, Stevenson RE. Elements of morphology: standard terminology for the nose and philtrum. Am J Med Genet A. 2009;149A(1):61-76. doi:10.1002/ajmg.a.32600
26. Carey JC, Cohen MM Jr, Curry CJ, Devriendt K, Holmes LB, Verloes A. Elements of morphology: standard terminology for the lips, mouth, and oral region. Am J Med Genet A. 2009;149A(1):77-92. doi:10.1002/ajmg.a.32602
27. Biesecker LG, Aase JM, Clericuzio C, Gurrieri F, Temple IK, Toriello H. Elements of morphology: standard terminology for the hands and feet. Am J Med Genet A. 2009;149A(1):93-127. doi:10.1002/ajmg.a.32596
28. Biesecker LG, Adam MP, Chung BH, Kosaki K, Menke LA, White S, et al. Elements of morphology: standard terminology for the trunk and limbs. Am J Med Genet A. 2022;188(11):3191-3228. doi:10.1002/ajmg.a.62965
29. Firth HV, Hurst JA. Ear anomalies. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 142–144.
30. Firth HV, Hurst JA. Coarse facial features. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 106–108.
31. Hoyme HE, Kalberg WO, Elliott AJ, Blankenship J, Buckley D, Marais AS, et al. Updated clinical guidelines for diagnosing fetal alcohol spectrum disorders. Pediatrics. 2016;138(2):e20154256. doi:10.1542/peds.2015-4256
32. Firth HV, Hurst JA. Facial asymmetry. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 146–148.
33. Kruszka P, Porras AR, Sobering AK, Ikolo FA, La Qua S, Shotelersuk V, et al. Down syndrome in diverse populations. Am J Med Genet A. 2017;173(1):42-53. doi:10.1002/ajmg.a.38043
34. Muenke M, Adeyemo A, Kruszka P. An electronic atlas of human malformation syndromes in diverse populations. Genet Med. 2016;18(11):1085-7. doi:10.1038/gim.2016.3
35. Deciphering Developmental Disorders Study. Large-scale discovery of novel genetic causes of developmental disorders. Nature. 2015;519(7542):223-8. doi:10.1038/nature14135
36. Posey JE, Harel T, Liu P, Rosenfeld JA, James RA, Coban Akdemir ZH, et al. Resolution of disease phenotypes resulting from multilocus genomic variation. N Engl J Med. 2017;376(1):21-31. doi:10.1056/NEJMoa1516767
37. Firth HV, Hurst JA. Ocular hypertelorism. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 258–259.
38. Firth HV, Hurst JA. Nasal anomalies. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 242–244.
39. Online Mendelian Inheritance in Man (OMIM). MIM 270400: Smith-Lemli-Opitz syndrome; SLOS [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/270400
40. Mithyantha R, Kneen R, McCann E, Gladstone M. Current evidence-based recommendations on investigating children with global developmental delay. Arch Dis Child. 2017;102(11):1071-6. doi:10.1136/archdischild-2016-311271
41. Firth HV, Hurst JA. Dysmorphology examination checklist. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 830.
42. Helm BM, Wetherill L, Landis BJ, Ware SM. Body region dysmorphology is predictive of genetic diagnoses in infants with congenital heart disease. Mol Genet Genomic Med. 2026;14(7):e70260. doi:10.1002/mgg3.70260
43. Wenderott K, Krups J, Zaruchas F, Krawitz P, Weigl M, Lesmann H. Workflow analysis and evaluation of a next-generation phenotyping tool: a qualitative study of Face2Gene. Eur J Hum Genet. 2025;33(10):1333-41. doi:10.1038/s41431-025-01875-0
44. Hsieh TC, Todd D, Warner T, Blankenship K, Saade D, Weiland H, et al. Leveraging next-generation phenotyping in dysmorphology to support variant interpretation in Mowat-Wilson syndrome. Neurol Genet. 2026;12(4):e200384. doi:10.1212/NXG.0000000000200384
45. Lumaka A, Cosemans N, Lulebo Mampasi A, Mubungu G, Mvuama N, Lubala T, et al. Facial dysmorphism is influenced by ethnic background of the patient and of the evaluator. Clin Genet. 2017;92(2):166-71. doi:10.1111/cge.12948
46. Schmidt A, Danyel M, Grundmann K, Brunet T, Klinkhammer H, Hsieh TC, et al. Next-generation phenotyping integrated in a national framework for patients with ultrarare disorders improves genetic diagnostics and yields new molecular findings. Nat Genet. 2024;56(8):1644-53. doi:10.1038/s41588-024-01836-1
47. Gupta S, Bawa P, Kumari A, Panigrahi I, Srivastava P, Kaur A. Real-world performance of Face2Gene and GestaltMatcher for facial image analysis in a large Indian ethnic cohort. Eur J Med Genet. 2025;78:105063. doi:10.1016/j.ejmg.2025.105063
48. Hennocq Q, Lienhard O, Rao D, Amiel J, Benichou L, Bongibault T, et al. Next generation phenotyping and synthetic faces in Coffin Siris syndrome. Clin Genet. 2025;107(5):495-504. doi:10.1111/cge.14682
49. Prendergast AB, Sullivan E, Mackley MP, Faghfoury H. Eurocentric bias in dysmorphology and medical genetics education. Am J Med Genet A. 2026;200(6):e70081. doi:10.1002/ajmg.a.70081
50. Sümer Ö, Huber T, Cheng J, Duong D, Ledgister Hanchard SE, Conati C, et al. Application of deep learning and explainable AI-supported medical decision-making for facial phenotyping in genetic syndromes. Clin Dysmorphol. 2026. doi:10.1097/MCD.0000000000000581
51. Kušíková K, Hsieh TC, Pfeifer M, Fauth C, Murakami Y, Laccone F, et al. Two novel cases with PIGQ-CDG: expansion of the genotype-phenotype spectrum and evaluation of GestaltMatcher as a diagnostic tool. Front Genet. 2025;16:1598602. doi:10.3389/fgene.2025.1598602
52. Lubin EE, Gonzalez EM, Sangree AK, Durham EL, Klinkhammer H, Li JM, et al. Coupling deep phenotypic quantification with next-generation phenotyping for 192 individuals with germline histonopathies. HGG Adv. 2025;6(3):100440. doi:10.1016/j.xhgg.2025.100440
53. Malinowski J, Miller DT, Demmer L, Gannon J, Pereira EM, Schroeder MC, et al. Systematic evidence-based review: outcomes from exome and genome sequencing for pediatric patients with congenital anomalies or intellectual disability. Genet Med. 2020;22(6):986-1004. doi:10.1038/s41436-020-0771-z
54. Saudubray JM, García-Cazorla Á. Clinical Approach to Inborn Errors of Metabolism in Paediatrics. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 1, section 1.1.1.2.
55. Firth HV, Hurst JA. Neonatal encephalopathy and intractable seizures. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 246–251.
56. Firth HV, Hurst JA. Genomic sequencing and interpretation of data from WES or WGS analyses. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 32–33.
57. Firth HV, Hurst JA. Reproductive options. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 36–37.
58. Online Mendelian Inheritance in Man (OMIM). MIM 164210: Craniofacial microsomia 1; CFM1 [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/164210
59. Online Mendelian Inheritance in Man (OMIM). MIM 154500: Treacher Collins syndrome 1; TCS1 [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/154500
60. Online Mendelian Inheritance in Man (OMIM). MIM 154400: Acrofacial dysostosis 1, Nager type; AFD1 [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/154400
61. Online Mendelian Inheritance in Man (OMIM). MIM 263750: Miller syndrome [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/263750
62. Online Mendelian Inheritance in Man (OMIM). MIM 610536: Mandibulofacial dysostosis, Guion-Almeida type; MFDGA [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/610536
63. Online Mendelian Inheritance in Man (OMIM). MIM 113650: Branchiootorenal syndrome 1; BOR1 [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/113650
64. Online Mendelian Inheritance in Man (OMIM). MIM 107480: Townes-Brocks syndrome 1; TBS1 [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/107480
65. Online Mendelian Inheritance in Man (OMIM). MIM 214800: CHARGE syndrome [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/214800
66. Online Mendelian Inheritance in Man (OMIM). MIM 157900: Moebius syndrome; MBS [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/157900
67. Online Mendelian Inheritance in Man (OMIM). MIM 304110: Craniofrontonasal syndrome; CFNS [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/304110
68. Online Mendelian Inheritance in Man (OMIM). MIM 300000: Opitz GBBB syndrome; GBBB [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/300000
69. Online Mendelian Inheritance in Man (OMIM). MIM 175700: Greig cephalopolysyndactyly syndrome; GCPS [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/175700
70. Online Mendelian Inheritance in Man (OMIM). MIM 305400: Aarskog-Scott syndrome; AAS [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/305400
71. Online Mendelian Inheritance in Man (OMIM). MIM 180849: Rubinstein-Taybi syndrome 1; RSTS1 [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/180849
72. Online Mendelian Inheritance in Man (OMIM). MIM 194190: Wolf-Hirschhorn syndrome; WHS [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/194190
73. Online Mendelian Inheritance in Man (OMIM). MIM 188400: DiGeorge syndrome; DGS [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/188400
74. Online Mendelian Inheritance in Man (OMIM). MIM 194050: Williams-Beuren syndrome; WBS [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/194050
75. Online Mendelian Inheritance in Man (OMIM). MIM 136140: Floating-Harbor syndrome; FLHS [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/136140
76. Online Mendelian Inheritance in Man (OMIM). MIM 601803: Pallister-Killian syndrome; PKS [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/601803
77. Online Mendelian Inheritance in Man (OMIM). MIM 115470: Cat eye syndrome; CES [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/115470
78. Online Mendelian Inheritance in Man (OMIM). MIM 163950: Noonan syndrome 1; NS1 [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/163950
79. Online Mendelian Inheritance in Man (OMIM). MIM 147920: Kabuki syndrome 1; KABUK1 [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/147920
80. Online Mendelian Inheritance in Man (OMIM). MIM 122470: Cornelia de Lange syndrome 1; CDLS1 [Internet]. Baltimore: Johns Hopkins University; [accessed 2026-10-29]. Available from: https://omim.org/entry/122470
81. GeneGuidance. Developmental delay and intellectual disability — library (diagnosis). 28 July 2022 [accessed 2026-09-26].
82. EUROCAT Northern Netherlands. Prevalence of congenital anomalies, 1981-2024 (update June 2026) [Internet]. Groningen: University Medical Center Groningen; 2026 [accessed 2026-10-29]. Available from: https://umcgresearch.org/documents/d/research/eurocat-prevalence-of-congenital-anomalies
83. VariantValidator. NM_014795.4:c.3237del (GRCh38) [Internet]. [accessed 2026-10-29]. Available from: https://variantvalidator.org