Genetic Testing in Neurodevelopmental Disorders

1With a combined prevalence of ~17% among 3- to 17-year-old children in the United States, neurodevelopmental disabilities are the most prevalent chronic medical conditions encountered in pediatric primary care. The vast majority of these individuals have diagnoses captured within the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM-5) neurodevelopmental disorders category that includes intellectual disability (ID), global developmental delay (GDD), communication disorders (language disorder, speech sound disorder, childhood onset fluency disorder, and social/pragmatic communication disorder), autism spectrum disorder (ASD), attention deficit/hyperactivity disorder (ADHD), specific learning disorder (involving reading, written expression, and/or mathematics), and motor disorders (developmental coordination disorder, stereotypic movement disorder, and tic disorders). Broader conceptualizations of neurodevelopmental disorders include conditions outside of the realm of the DSM-5, such as cerebral palsy (CP) and epilepsy, and sometimes neuropsychiatric disorders for which there is strong clinical and biological evidence of developmental origins, such as schizophrenia.

2Neurodevelopmental disorders are characterized by developmental deficits in cognition, language, behavior, and/or motor skills that cause impairment of personal, social, academic, and/or occupational functioning. These clinically and etiologically heterogeneous disorders represent manifestations of altered neural development and, as such, are typically diagnosed during infancy, childhood, or adolescence. Although subject to maturational changes, neurodevelopmental disorders are non-progressive and tend to follow a relatively steady trajectory rather than a pattern of remitting and relapsing. Tic disorders are an exception to the latter, since there is typically waxing and waning of core symptoms rather than a steady course. Because there is insufficient information to allow systematic classification of neurodevelopmental conditions based on etiology and pathophysiology, a descriptive (phenomenological) categorical nosology based on groups of signs and symptoms that define disorders or syndromes (e.g., ASD, ADHD, CP, etc.), has been adopted. Descriptive categorical diagnoses are useful heuristics for improving interrater reliability and enhancing information exchange, but their limited alignment with many clinical and biological findings is well-described. Pearn noted that “syndromic diagnosis is a concise shorthand for describing a constellation of clinical symptoms and signs - but is an acknowledgment of causal ignorance, which in turn demands differential reappraisal as new biochemical, genetic, or ultimately molecular causes of the syndrome are defined.” Comprehensive developmental care, in addition to identifying appropriate descriptive diagnoses and guiding families to evidence-based treatments and supports, includes a search for an underlying etiologic diagnosis.

3The scope of this review is limited to the genetic etiologic evaluation of children and adolescents with neurodevelopmental disorders. The focus is on individuals with diagnoses including GDD, ID, or ASD, for which there are existing guidelines for genetic testing, but we discuss the prospect of future genetic etiologic testing across broader neurodevelopmental disorders. Because of the focus on determining etiology using diagnostic genetic testing, the review does not cover pharmacogenomics, carrier testing, newborn screening, or pre-symptomatic/predictive testing. We provide an overview of the value of clinical genetic testing; the most commonly performed tests and the yield of these tests for determining a genetic diagnosis (focusing on established indications, ID/GDD and ASD); approaches to service delivery and implementation of genetic testing in clinical practice; and future directions, including emerging indications for genetic testing and forthcoming technologies.

4Identification of an underlying genetic etiology for a child's categorical neurodevelopmental diagnosis (or diagnoses) can provide both clinical and personal utility to patients and their families. Table 1 provides several examples of genetic causes of neurodevelopmental disorders and the potential utility of identifying such a diagnosis. Establishing a genetic basis for a child's neurodevelopmental phenotype can provide additional information about their prognosis and enable caretakers to understand potential areas of need and opportunities for increased support. For example, a genetic diagnosis, and the prognostic information it provides, may facilitate acquisition of educational, disability, and employment services. Although currently infrequent in the case of neurodevelopmental disorders, a genetic diagnosis might also provide access to etiology-specific treatments. Given the number of treatments that are currently being explored in animal models and clinical trials, such gene-specific therapies are likely to become more ubiquitous. In the interim, as new therapies are being developed, a genetic diagnosis can provide access to condition-specific research protocols enrolling human subjects. It is also important to note that elucidating etiologies and pathophysiologic processes is a necessary step toward developing animal models and human cell lines to use in pathogenesis studies, clinical trials, and, ultimately, mechanism-based targeted treatments for neurodevelopmental disorders. In addition to facilitating medical management and access to services, a genetic diagnosis can also enable families to avoid unnecessary diagnostic tests and, with an etiologic diagnosis in hand, families may be more empowered to avoid therapeutic interventions that are based on unfounded etiologic theories and are potentially harmful (e.g., chelation therapy). Additionally, genetic testing can afford patients and their caregivers the ability to identify, treat, and/or prevent medical comorbidities at the time of diagnosis, as well as conditions that may develop later in life.

5Furthermore, genetic testing allows healthcare providers to refine recurrence risk counseling for the family to inform reproductive decision making. When an etiology is determined, the risk of recurrence for an individual family varies depending on the specific genomic variant(s) identified. For parents of a child with a neurodevelopmental disorder, recurrence may be 50%, for example, in the case of an inherited, maternally-derived chromosome 15q11–q13 interstitial duplication, 25% in the case of an autosomal recessive disorder such as Smith-Lemli-Opitz syndrome, or approximately the prevalence rate of the particular disorder in the general population (e.g., ~1.5% for ASD) if the child has a de novo explanatory variant. This refined recurrence risk information offers patients, parents, and family members an understanding of their reproductive risks and informs reproductive decision making, restores reproductive confidence, and enables prenatal diagnosis. Furthermore, an understanding of recurrence risk in subsequent children and generations can enable families and their medical providers to identify neurodevelopmental disorders and initiate beneficial behavioral treatments and therapies at earlier ages. In the case of inherited etiologies, identifying a genetic diagnosis for the proband may also provide diagnoses to other family members with a history of a neurodevelopmental disorder.

6In addition to the clinical utility conferred through a genetic diagnosis, identifying an etiology can also provide psychosocial benefit to families. A genetic diagnosis can provide families with an explanation for their child's developmental history and, therefore, bring an end to the child's “diagnostic odyssey” that may have included years of uncertainty, anxiety, and evaluations, and can also guide patients and families to condition-specific resources and supports. Receiving a diagnosis has also been shown to increase knowledge, provide a sense of empowerment, result in peace of mind, increase parental quality of life, decrease parental guilt, and foster increased acceptance.

7In addition to wanting to know their child's clinical diagnosis and prognosis, families typically want to know the cause of the child's developmental disability. In the last two decades, rapid advances in the development of genetic testing technologies and application of these technologies to well-characterized patient cohorts have revolutionized our ability to make specific genetic diagnoses in patients presenting with neurodevelopmental disorders. New genes are being implicated in neurodevelopment at a rapid pace. Online resources such as the Geisinger Developmental Brain Disorder Genes Database (https://dbd.geisingeradmi.org/), the Clinical Genome Resource's Gene Validity Curations and Dosage Sensitivity Map (https://search.clinicalgenome.org/kb/gene-validity/; https://dosage.clinicalgenome.org/), and DECIPHER's Development Disorder Genotype - Phenotype Database (https://decipher.sanger.ac.uk/ddd#ddgenes) can provide up to date information about genes' and genomic variants' relationship to neurodevelopmental disorders.

8Genetic testing is now the standard of care for several neurodevelopmental disorders and the indications for testing will almost certainly broaden in the future. In practice, a genetic etiologic diagnosis may be suspected clinically and confirmed by gene-specific genetic testing or, more commonly, it may be revealed by chromosomal microarray (CMA) analysis, exome sequencing (ES), or FMR1 CGG repeat analysis for fragile X syndrome completed as a routine part of the evaluation of a patient with ID/GDD or ASD in the absence of a clinically recognizable syndrome. It is important for pediatric clinicians to understand these common tests and their role in the care of children with neurodevelopmental disorders.

9Genome-wide CMA has been endorsed as a first-tier test for several indications including in patients with ASD, ID, GDD, and/or multiple congenital anomalies. CMA technologies (Comparative Genomic Hybridization (CGH) array CGH and Single Nucleotide Polymorphism (SNP)-based testing) detect copy number variants (CNVs) – gains or losses of chromosomal material. In some cases, such gains or losses affect gene function and impact health and development. The resolution, or size of gains and losses that can be detected by CMA, varies and is determined by the specific technology used and the genomic distance between DNA probes.

10CNVs detected through CMA should be categorized into the following categories in accordance with the American College of Medical Genetics and Genomics (ACMG) and Clinical Genome Resource (ClinGen) guidelines.

11Pathogenic – CNVs that are thought to be associated with disease. Pathogenic CNVs may include those that explain the patient's phenotype, those that are associated with carrier status for a recessive condition, and those that indicate disease risk for an unrelated phenotype. Pathogenic CNVs that explain a patient's neurodevelopmental history can include recurrent deletions or duplications such as 22q11.2 deletion syndrome or the 17p11.2 recurrent microdeletion that causes Smith-Magenis syndrome, or novel gains or losses that impact dosage sensitive gene(s).

12Likely pathogenic – CNVs that have considerable evidence to suggest that they are associated with disease but where additional evidence could further clarify the variant's pathogenicity.

13Uncertain Significance – CNVs that do not have enough information to determine if they are pathogenic or benign. When a CNV with uncertain significance is identified, parental studies might be used to provide additional information to clarify pathogenicity.

14Likely Benign – CNVs that have considerable evidence to suggest that they are not associated with disease but where additional evidence could further clarify this.

15Benign – CNVs that are not thought to be associated with disease. These are often present in phenotypically normal individuals or in the general population.

16Interpretations may change over time as new evidence emerges informing a variants' pathogenicity. Because CMA employs genome-wide testing, incidental findings, or those that are unrelated to the primary indication, may be identified. CNVs often include multiple genes and, in rare instances, a CNV might explain a child's neurodevelopmental history, but given the gene content, may also confer risk for an unrelated condition. In other cases, a CNV that is unrelated to a patient's phenotype but that has implications for care may be identified and reported through testing.

17In addition to CNVs, laboratories that employ SNP-based CMA can also detect regions of homozygosity (chromosomal segments that are identical to one another). Identification of regions of homozygosity enables the potential detection of conditions that can be caused by uniparental isodisomy (UPD) such as Silver-Russell (maternal UPD chromosome 7), Angelman (paternal UPD chromosome 15), and Prader-Willi syndromes (maternal UPD chromosome 15). In other instances, regions of homozygosity may be indicative of ancestral homozygosity or parental consanguinity. Regions of homozygosity can also be indicative of an increased risk for an autosomal recessive condition due to the potential for homozygous variation in single gene.

18Although CMA is able to detect CNVs, this technology has limited ability to detect balanced chromosomal rearrangements (translocations, inversions), trinucleotide repeat expansions, imbalances in the mitochondrial genome, epigenetic abnormalities (e.g., methylation abnormalities), sequence level variants, or low-level mosaicism for CNVs (Table 2). As stated above, the size of deletions and/or duplications that can be detected by CMA varies. Current clinical CMA platforms can detect CNVs ~400 kb in size with many laboratories detecting those >250 kilobases. Certain regions may be more specifically targeted enabling even smaller CNV detection (Table 2).

19Additional testing discussed in this review can aid in detection of some variants that CMA is unable to detect, including FMR1 CGG repeat analysis for the trinucleotide repeat expansion that causes fragile X syndrome and ES for detection of exonic sequence variants. Cytogenetic testing, including G-banded karyotype and/or florescence in situ hybridization (FISH) as well as methylation testing, can detect additional variants or clarify results from CMA. As reviewed elsewhere, such testing should be considered on a case by case basis (e.g., for detection of balanced rearrangements and other complex rearrangements or mosaicism for partial or whole chromosome aneuploidy, or for clarification of the location of a duplication).

20Authors reporting CMA yield use variable nomenclature to describe variants' pathogenicity and diverse interpretation practices, thus limiting between-study comparisons and pooling of data across studies of CMA diagnostic yield. The diagnostic yield of CMA is the proportion of tests performed that identify a variant that is considered causative for a patient's phenotype. Taken together, a causative result can be identified by CMA in 15–20% of individuals with ID/DD, ASD, and/or multiple congenital anomalies. For the purpose of trying to capture yield specific to neurodevelopmental disorders, one can examine 19 studies that report on the yield of CMA in more than 150 individuals with ID/GDD where the diagnostic yield ranged from 4.5 to 28.0% (median 13.7%). Additionally, one can look to 11 studies limited to patients with ASD (each with a sample size ≥50), in which CMA identified a causative variant in 1.5 to 20.5% of subjects (median 8.1%). Because karyotype was the standard etiologic approach before the clinical integration of CMA, many publications summarizing diagnostic yields of CMA excluded patients with an abnormal karyotype. Since the majority of pathogenic results identified by karyotype would be detected by CMA, a number of publications may be reporting yields ~3.7% percent lower, on average, than if CMA had been applied as a first-tier evaluation. Overall, the diagnostic yield of CMA in patients with ID, GDD, and/or ASD suggests that CNVs comprise a substantial proportion of underlying genetic etiologies.

21Although the diagnostic yield of CMA has been well-examined in the literature, there are few studies exploring the frequency that incidental findings are identified with this testing technology. In two studies, 0.15–0.48% of patients had a CNV that included a gene associated with hereditary cancer pre-disposition, and in a third study, 1.2% had a CNV that included a gene associated with an adult-onset condition The pathogenicity of these CNVs were not reported by the study authors, so the true rate of pathogenic incidental findings in these study populations is somewhat unclear. Taken together, however, these studies suggest that clinically significant incidental findings likely occur in <1% of patients undergoing CMA.

22In the last decade, the arrival of high-throughput sequencing technologies, collectively referred to as next generation sequencing (NGS) or massively parallel sequencing, has reduced the cost and increased the speed of sequencing, enabling laboratories to sequence large amounts of DNA. Rather than testing a single gene or several genes, laboratories can now offer sequencing of extensive gene panels, the exome, or the genome.

23In ES, protein coding regions, or exons, of the genome are sequenced. The exome makes up ~1.5–2% of the genome, but the vast majority of alleles underlying Mendelian disorders impact coding sequences. ES allows detection of sequence-level, single gene variants across almost all of the exome and a small number of intronic nucleotides at the boundaries of each exon. In addition to sequence-level variant detection, several clinical laboratories are now incorporating CNV calling into ES and are able to detect multi-exon deletions and duplications and more laboratories will likely add this to their exome analyses moving forward. Shorter CNVs including those that involve one to two exons are less reliabily detected using ES. Given the breadth of information that is generated using ES, parental samples are recommended for analysis (ideally trio analysis or duo analysis if only one parent is available) since this can reduce the number of candidate variants that require review and facilitate variant interpretation.

24As with CMA, variants detected by ES are categorized as pathogenic, likely pathogenic, uncertain significance, likely benign, or benign in accordance with ACMG/ Association for Molecular Pathology (AMP) sequence variant interpretation guidelines. Due to the rapid increase in our understanding of genes and genomic variants, improved variant annotation and filtering, and evolving patient phenotypes, an iterative reanalysis of ES data may enable additional diagnoses. Over time, a variant's pathogenicity may be updated as new information emerges. ES also includes analysis of genes not yet associated with a specific phenotype. As a result, iterative reanalysis might also identify a novel variant.

25ES, like CMA, may identify results that are clinically relevant but unrelated to the patient's neurodevelopmental disorder. ES includes almost all of the coding portions of known genes, providing an opportunity to examine sequence data for pathogenic/likely pathogenic variants in medically actionable genes that are unrelated to the indication for testing (secondary findings). The ACMG has recommended that clinicians notify their patient if a variant known or expected to increase disease risk was identified in a list of, currently 59, clinically actionable genes. ACMG has specified that patients and families have the ability to opt out of such findings and will continue to update this list of clinically actionable genes as additional evidence emerges. In addition to secondary findings, the use of parental samples in ES can result in incidental identification of possible misattributed parentage; reporting and disclosure of this information varies between laboratories and clinicians.

26Although ES captures the majority of the exome, the exome is not covered in its entirety and coverage may differ across platforms and laboratories. As a result, ES may not detect all coding, sequence variants. ES also has lower sensitivity for detection of mosaicism and exon-level deletions and/or duplications compared to gene panels that include deletion/duplication analysis. An increasing number of laboratories are calling CNVs from ES data; laboratories that are detecting and reporting CNVs have limited ability to detect deletions or duplications involving only one to two exons, however (Table 2). Furthermore, ES is unable to detect repetitive DNA sequences including trinucleotide repeats (e.g., the CGG repeat expansion that causes fragile X syndrome), intronic/non-coding variants, mitochondrial variants, epigenetic variants (e.g., methylation abnormalities), or balanced chromosomal rearrangements.

27The diagnostic yield is the proportion of exome analyses with variants that are determined to be pathogenic or likely pathogenic and explain the patient's phenotype. A recent meta-analysis of 21 ES studies that focused on isolated neurodevelopmental disorders (GDD, ID, and/or ASD; n = 3,173) identified a diagnostic yield of 31% (95% CI 25–38%). When nine additional studies of individuals with these neurodevelopmental disorders plus associated neurological or syndromic conditions or clinical characteristics were included, increasing the total number of participants to 3,350, the yield was 36% (95% CI 30–43%). In a large, laboratory-based study not included in the meta-analysis because the cohort included a potentially broader group of neurodevelopmental phenotypes, the diagnostic yield among individuals undergoing ES due to neurodevelopmental disorders (ID, ASD, developmental delay, or speech delay) with or without involvement of other organ systems was 25.4% (425 of 1,673).

28Delineation of the diagnostic yield of ES for cohorts ascertained for ID/GDD vs. ASD is limited by the phenotype data reported in the literature. Srivastava et al. reported in their meta-analysis that among studies including individuals with primarily ID (n = 10), the diagnostic yield was 39% (95% CI 29–50%), whereas studies including individuals with primarily ASD (n = 5) had a yield of 16% (95% CI 11–24%) and those with a more heterogenous mix of ID and/or ASD (n = 6) identified a diagnosis in 37% of participants (95% CI 29–46%). Among seven ES studies each involving more than 50 individuals with ID/GDD, a diagnosis was established in 34% of patients (range 28–43%). Studies restricted to clinically ascertained samples of patients with ASD and analyses of ASD subgroups within the large, laboratory-based samples have reported lower yields of 8–26% (median 15%). There is substantial variability among these studies regarding the information provided about cognitive status and the stringency of ASD diagnosis. In most cases the patients included had previously had genetic testing including fragile X analysis and CMA that did not result in a molecular diagnosis. ES reveals two or, rarely, three molecular diagnoses in ~1% of individuals undergoing clinical testing. When this occurs, the patient typically has a “blended phenotype,” with features that are accounted for by each of the pathogenic variants and not by a single molecular diagnosis.

29Several studies suggest that the diagnostic yield of ES is higher when trios (probands and parents) are tested than when only the DNA of probands is sequenced. Because of the rapid advances in gene and variant curation and factors such as evolution of the phenotype in an individual over time, periodic reanalysis of ES data may enhance the diagnostic yield considerably. There is wide variability in the rate of secondary findings among individuals who undergo ES; several recent large studies have reported rates of 2–6%. These reportable secondary findings, which are most commonly related to cardiomyopathies, cardiac conduction disorders, hereditary cancer pre-disposition, or familial hypercholesterolemia, often result in additional testing and/or surveillance of the proband and relatives.

30In addition to ES, single gene testing and targeted next-generation sequencing panels have been used in evaluation of patients with neurodevelopmental disorders including ID, GDD, and ASD historically, and a number of clinical laboratories currently offer such testing. Compared to ES, more targeted sequencing allows for increased read depth and sequence coverage thus increasing detection of mosaicism. Moreover, single gene testing and gene panels are better able to detect indels and those that include deletion/duplication analysis are better able to detect exon-level deletions and/or duplications. While ES identifies 100–200 potentially deleterious sequence variants on average, more targeted analysis identifies a smaller number of variants requiring less analysis and reducing the potential for variants of uncertain significance.

31Although there are some advantages to more targeted testing and clinical scenarios that warrant such testing, the genetic heterogeneity of ID, GDD, and ASD and frequency of causative de novo variants, ES is an appropriate approach for many patients with these diagnoses. Studies have found that ES can detect more than 98% of pathogenic variants identified on gene panels. Additionally, the gene content of targeted gene panels vary significantly between laboratories meaning that diagnostic yields also vary. In Hoang et al., comparison of ASD gene panels across 21 laboratories found that the number of genes included on ASD-related panels ranged from 11–2,562 and only a single gene (MECP2) was included on all panels. In a simulation study comparing ES to gene panels, providers were asked to choose a commercially available gene panel as an alternative when ordering ES for a patient; of the patients receiving a diagnosis through ES, 23% would not have had their variant identified through the provider-chosen gene panel. Additionally, as an increasing number of genes are being implicated in Mendelian conditions including neurodevelopmental disorders, panels quickly can become out of date and updating panels can be a time-consuming process. After undergoing gene panel testing, 11% of panel-negative cases in one cohort received a diagnosis via ES; many of these diagnoses were attributed to a gene-disease relationship being identified after the panel assay was established. In addition to being able to more readily report recently described genes implicated in neurodevelopmental disorders, ES also enables increasing diagnoses over time due to reanalysis. Unlike gene panels, ES includes analysis of genes not yet associated with a specific phenotype. As a result, iterative reanalysis enables reporting of novel variants recently implicated in neurodevelopment and can allow for diagnoses into the future while gene panels typically only enable reanalysis of the sequenced genes. Finally, more targeted testing has been viewed as a less expensive testing option; however, more recent publications have suggested that ES is cost-effective.

32Although ES and CMA have revolutionized our ability to detect genomic variants, there are limitations in our ability to interpret such variants and determine their impact on health and development. Copy number and sequence variants identified through CMA and ES might be interpreted as variants of uncertain significance (VUS) based on available evidence. VUS include variants in known disease genes that have insufficient evidence to be classified as benign or pathogenic as well as sequence and copy number variants that involve genes that are not yet associated with a disease or phenotype (also referred to as genes of uncertain significance/ candidate genes). VUS in patients undergoing CMA due to a history of developmental delay, ID, ASD, and/or multiple congenital anomalies have been reported to carry at least one VUS 7.9–19% of the time. The frequency of VUS in patients undergoing ES depends on a number of factors including the reporting laboratory's reporting practices, phenotypic features provided by the ordering clinician, and inclusion of parental samples. Several studies examined VUS rates among patients undergoing ES for a variety of indications and found VUS rates of 25.3–86%.

33Uncertain results, although not unique to genetics, can pose challenges for patients and providers alike. Compared to other potential results, providers are least comfortable explaining VUS results to families and report that additional preparation is required for such results. Non-genetics providers express a need for additional education and access to genetics professionals' expertise to facilitate understanding and disclosure of such results. Despite expressing less comfort with VUS results and calls for more support and education, non-genetics, pediatric providers that regularly order testing appear to be able to understand and interpret VUS results accurately.

34Studies of parents have suggested that they are interested in receiving uncertain results and those that receive such results post-natally see the them as important, often demonstrate an understanding that the result is uncertain, and acknowledge that future advancements can increase understanding. Although parents seem to accurately recall the concept of uncertainty, parents report difficulty understanding how the result impacts them and their child, and some over interpret the variant as being causative. Emotional reactions to VUS vary between families and over time.

35Given the potential for misunderstandings and variable emotional reactions, providers need to be equipped to discuss uncertain results or refer on to providers with genetics expertise. The option for in-person, timely discussions with empathic and honest providers and access to supplemental information including documentation of the result have been suggested to improve understanding of VUS. In contrast, internet searches have been shown to increase uncertainty; therefore, anticipatory guidance about misleading or irrelevant online information should be provided to families receiving VUS results. In addition to providing families with appropriate information, support, and resources to facilitate understanding of uncertain results, providers returning VUS should consider if additional evaluations that could inform the pathogenicity of the variant are indicated such as parental or familial testing, imaging, or specialist referrals. Furthermore, knowledge of genomic variants and their relationship to health and development will continue to improve. As such, VUS will be updated to benign or pathogenic over time. For example, in one study of 2,250 patients undergoing ES reanalysis, 23 had variants initially reported as VUS upgraded to diagnostic. Patients receiving VUS results need to be informed of the potential for interpretation updates and ordering providers should discuss the process for reassessing variants.

36Fragile X syndrome, caused by loss of function of the FMR1 gene, is thought to be one of the most common inherited genetic causes of ID and ASD. Ninety-nine percent of cases of fragile X syndrome are caused by an expansion of the unstable CGG repeat sequence in the 5′ untranslated region (UTR) of the FMR1 gene. Most individuals in the general population have 44 or fewer CGG repeats, while more than 200 CGG repeats in the FMR1 gene results in hypermethylation and, consequently, transcriptional silencing of the gene. The FMR1 gene is located on the X chromosome (Xq27.3) and, consequently, mutations can cause variable phenotypes in males and females. FMR1 CGG repeat analysis is usually completed using polymerase chain reaction (PCR) analysis and Southern Blot Analysis. FMR1 CGG repeat analysis can yield four main categories of results:

37Normal: ≤44 repeats.

38Intermediate (Inconclusive, Borderline, Gray Zone): 45–54 repeats. Alleles with 45–54 repeats have not been observed to expand to a full mutation in one generation. Because minor increases or decreases in repeat size can occur, alleles of this size could be associated with fragile X syndrome in future generations.

39Pre-mutation: 55–200 repeats. Expansions in FMR1 of this size are unstably passed from parent to child and, when passed from the mother, expansion from the pre-mutation to a full mutation may occur. The risk for expansion is greatest in those with larger repeat sizes. Furthermore, the presence of AGG interruptions within the CGG repeat tract is associated with decreased risk for expansion.

40Full Mutation: >200 repeats (typically several hundred to several thousand repeats).

41In addition to the risk of expansion to a full mutation in offspring, female pre-mutation carriers are at risk for hypergonadotropic hypogonadism (fragile X pre-mature ovarian insufficiency) before age 40 years. Although more prevalent in males, both male and female pre-mutation carriers are also at an increased risk for fragile X-Associated Tremor/Ataxia Syndrome (FXTAS), a late-onset neurodegenerative condition characterized by cerebellar ataxia, intention tremor, cognitive impairment with increased penetrance observed in males. When offering FMR1 CGG repeat testing, pre-mutation carrier status may be identified in the proband and/or relatives and incidentally identify risks for these adult-onset conditions.

42FMR1 CGG repeat testing cannot detect other variants causing loss of function of the FMR1 gene (e.g., deletions, sequence variants that result in protein truncation), which are rare. Testing does not detect other genomic causes of neurodevelopmental disorders (e.g., sequence variants in other genes, copy number variants, epigenetic abnormalities) (Table 2).

43The yield of fragile X testing among individuals with ID/GDD and/or ASD varies widely based on study design and characteristics of the population being tested, such as severity of cognitive impairment, whether both males and females are included, and whether clinical judgement or phenotypic feature checklists were used to exclude some potential participants. For example, the diagnostic yield varied from 0.5 to 6% among 14 studies reviewed by Peprah that included at least 200 individuals with ID who were ascertained either through clinical referral or special needs service utilization. Hunter et al. identified 15 studies that estimated the frequency of the full mutation in populations with ID before extrapolating to the total population, with the goal of including them in a meta-analysis, but it was not possible to combine the data and calculate a valid mean prevalence due to variability in study methods and lack of reported measures of uncertainty.

44Several studies that attempted to capture populations of males with neurodevelopmental disorders for assessment of the rate of fragile X syndrome reported full mutations in 8/611 (1.3%) with unexplained ID, 20/3,738 (0.5%) with special education needs related to cognitive deficiencies, and 7/2,471 (0.28%) receiving special education services (excluding isolated speech therapy or gifted services). An epidemiological study of a representative sample of 3,313 people with ID (56% male) in the Netherlands included 1,143 individuals (55% male) who were not eligible for testing because they had other etiologic diagnoses, including fragile X syndrome (30 males and 2 females; 4.75 and 0.39%, respectively), or had previous negative clinical testing, and 2,170 (57% male) who were eligible for testing as part of the study. Among those who were eligible, 1,520 individuals (57% male) consented and were tested. Full mutations were identified in 9/866 males (1.0%) and 2/654 females (0.31%). Allowing the unknown diagnostic yield for those who were eligible but did not consent to testing to vary from 0.5 to 2.0 times the sex-specific diagnostic yield of those who were tested and including the ineligible individuals, the overall prevalence estimates are 2.2–2.5% for males with ID and 1.3–1.6% for females with ID. The rates of full FMR1 mutations identified among clinical ASD cohorts are generally lower than those identified in clinical ID cohorts. Although early studies suggested higher rates, the larger studies published in the last decade (n = 142 to n = 861) have identified fragile X full mutations in only 0.23 to 1.2% of individuals ascertained for ASD diagnosis. The combined yield of these studies was nine full mutations in 1,984 individuals tested (0.45%), including seven males and two females. Fragile X syndrome has been identified in females with ASD with and without ID.

45The diagnostic yield of fragile X testing has also been described by several clinical laboratories, typically with little information available about phenotype. For example, a large-volume commercial diagnostic laboratory reported FMR1 full mutation alleles in 1.4% of 59,707 males and 0.61% of 59,525 females tested post-natally over a 14-year period (1992–2006). More recently, two university hospital diagnostic laboratories reported yields of 0% (0/654) and 0.9% (11/1,177) in males under age 22 and 19 years, respectively. A similar yield of 0.8% (43/5,401) was reported by another hospital-based genetics laboratory that did not describe the diagnostic rate in males and females separately. Borch et al. found a diagnostic yield of 1.2% (30/2,486) among pediatric patients who had fragile X testing, including 1.3% of males (25/1,919) and 0.9% of females (5/567). The vast majority, 96% (29/30), had clinical features and/or family history that were suggestive of fragile X syndrome.

46Although studies vary in the level of phenotypic detail provided, several authors reporting on yields of CMA, ES, and FMR1 CGG repeat analysis in patients with neurodevelopmental disorders suggest that specific characteristics and additional diagnoses are associated with increased diagnostic yields. Lower IQ, dysmorphic features, and congenital anomalies have been found to be associated with higher diagnostic yield of CMA and ES in some studies examining yield in patients with ASD. Similarly, in Fan et al., co-occurring congenital heart defects, facial dysmorphisms, microcephaly, and hypotonia were associated with increased diagnostic yield of CMA in patients with developmental delay or ID. A higher diagnostic yield has also been reported for moderate to severe ID than for mild ID. Among individuals with severe ID, it is estimated that more than 60% harbor causative CNVs or exonic sequence variants. Finally, in a meta-analysis of published studies using clinical checklists among patients with ID undergoing testing for fragile X, soft and velvety skin on the palms with redundancy on the dorsum of the hand, large prominent ears, pale blue eyes, family history of ID, autistic behavior, flat feet, and plantar crease all are correlated with an increased likelihood of identifying a FMR1 full mutation. Although specific features and additional diagnoses have been correlated with increased diagnostic yield, the yield in patients with isolated ID/GDD and/or ASD is high enough to warrant an etiologic genetic investigation following the application of these categorical diagnoses.

47Although heterogeneous study designs and methodologies limit conclusions that can be drawn regarding diagnosis-specific diagnostic yields, current evidence suggests that among patients with ID or GDD, the diagnostic yields are at least 15% for CMA, 35% for ES, and 1% (in females) to 2% (in males) for FMR1 CGG repeat analysis. For those with primarily ASD, the diagnostic yields are at least 10% for CMA and 15% for ES, and 0.5% or less for FMR1 CGG repeat analysis. Across all of these tests and clinical diagnoses, the diagnostic yield rates are higher in the presence of lower IQ, neurological comorbidities, dysmorphic features, and congenital anomalies. The cumulative diagnostic yield for all three tests, therefore, is over 50% for individuals with ID/GDD and over 25% for individuals with primarily ASD.

48Due to the yields of genetic testing and corresponding potential for clinical and personal utility, there is consensus that genetic testing should be offered to patients with GDD, ID, and/or ASD diagnoses. The specific testing algorithm varies between guidelines and consensus statements, and many recommendations predate the broad integration of exome sequencing into clinical care (Table 3).

49Despite widespread recommendations for genetic testing for neurodevelopmental disorders in pediatric patients, surveys, interviews and retrospective studies of clinical cohorts suggest that that a minority of children with ASD (16.5%-45%) and ID (43%) in the United States have undergone any genetic testing. Furthermore, a study examining pediatric providers' use of genetic testing in the care of simulated patients with ID, GDD, and/or ASD suggest that genetic testing is underutilized. In Europe, rates of testing vary from country to country with 61.7% French parents reporting that their child with ASD underwent some diagnostic genetic test compared to 13% of children with ASD in Spain. Although there is consensus agreement that a genetic evaluation should be offered to individuals with GDD, ID, and/or ASD, this does not appear to be occurring in practice. Pediatric providers caring for patients with neurodevelopmental disorders should work to address this gap in care.

50Pediatric providers, from general practitioners to neurodevelopmental pediatricians, are in a position to empower themselves to consent for and order genetic testing. Given the increasing availability of genetic testing and limited number of genetics professionals, genetics providers cannot and need not be involved in all instances of consent for genetic testing and disclosure of results. Genetic testing for patients with neurodevelopmental disorders has shifted to be within the purview of non-genetics providers and is something that pediatric providers can be equipped to undertake. Below we describe a structured but flexible approach to facilitate integration of genetic testing into clinical practice across pediatric specialties (Table 4).

51In most cases of patients with ID, GDD, and/or ASD, developmental, medical, and family histories do not point to a specific genetic etiology. In these cases, the approach to genetic testing should include broad analysis for copy number and exonic sequence variants (Table 4). This can be achieved by ordering a single genetic test - ES with CNV analysis. Diagnostic yields of ES and emerging evidence that ES is cost-efficient in the etiologic evaluation of children with neurodevelopmental disorders support this assertion that ES is the most appropriate first test to pursue. Payers are even altering their policies to support increasing efficiency through such testing algorithms.

52It is important to note that, at present, some genetic testing laboratories may not include CNV analysis as part of their ES and some payers may require completion of CMA prior to coverage of ES. As a result, ES (without CNV analysis) and CMA may need to be pursued in a stepwise manner. In cases where both CMA and ES are pursued independently, the order of testing will be dictated by individual factors such as insurance requirements (e.g., payer may only cover ES after a negative CMA) and additional phenotypic features (e.g., non-specific epilepsy phenotype in addition to ASD increasing the likelihood of a sequence variant).

53Less often, a patient presenting with ID, GDD, and/or ASD may have a constellation of features and/or history that are suggestive of a particular genetic etiology. In these instances, if a specific disorder is suspected, specific genetic evaluation(s) should be pursued (Table 4). For example, in a male patient with developmental delay, head circumference that is more than 2.5 standard deviations above the mean for age, and penile freckling, PTEN gene analysis may be pursued as an initial evaluation given that these features are highly suggestive of PTEN hamartoma tumor syndrome. Similarly, in the case of a female with severe developmental delay, acquired microcephaly, seizures, and stereotypic hand movements, MECP2 gene analysis may be pursued given that these features are suggestive of Rett syndrome.

54Recommendations for when to pursue FMR1 repeat analysis are diverse with some publications recommending that all individuals with ASD and/or ID/GDD undergo testing while others make a distinction based on sex and recommend testing for all males. Finally, some authors suggest ordering FMR1 repeat analysis as a second-tier test or restricting testing to those with the highest likelihood of testing positive for an FMR1 full mutation based on specific clinical findings could reduce the number of individuals tested while maintaining high sensitivity. With the relatively low cost of FMR1 repeat analysis and reproductive risks for family members should a child have fragile X, there should be a low threshold for offering this testing if testing is not offered to all patients with ID, GDD. and/or ASD.

55Once a testing approach is determined, the ordering provider should engage families in a conversation prior to testing so they understand the nature and scope of the test(s) (e.g., purpose and potential results including uncertain results), benefits, limitations (e.g., does not detect all genetic causes of neurodevelopmental disorders), risks (e.g., uncertain results and familial implications), and costs. Following receipt of results, providers will need to review any variants identified and put them in the context of their patient's clinical features and what is known about the gene, communicate results to the family, establish a follow-up plan, and collaborate with and refer to subspecialists (e.g., medical geneticist) as needed.

56Some non-genetics providers express a lack of knowledge and/or discomfort in ordering and interpreting genetic testing, but genomics education, relationships with genetic testing laboratory staff, and the ability to refer to genetics once results are received can afford providers with knowledge and support to integrate genomic medicine into their practice. In addition to consenting for and ordering testing independently, pediatric providers can integrate genetic counselors into their clinics for additional support and expertise and a multidisciplinary approach to care. Genetic counselors provide care in at least 28 countries and are typically graduate level providers that have training in both genetics and counseling enabling them to interpret genetic test results and support patients and families in their care. Increasingly, genetic counselors are working as experts in genetic testing alongside non-genetics providers in various clinical settings, including pediatric neurology and neurodevelopmental clinics. With the ability to foster and maintain interdisciplinary relationships included in their code of ethics and practice-based competencies, genetic counselors can educate other providers about genetic testing and the genetics of neurodevelopmental disorders. Furthermore, genetic counselors' expertise makes them well-suited to work alongside pediatric providers to obtain consent for genetic testing, aid in ordering of appropriate testing, interpret genetic test results, and facilitate a family's understanding of and adjustment to genetic information. Genetic counselors' skill sets can enable them to be important members of a multidisciplinary neurodevelopmental care team along with other specialists such as neurodevelopmental pediatricians.

57Pediatric healthcare providers may also elect to refer a subset of patients with neurodevelopmental disorders on to an external genetics provider. Medical geneticists have expertise in genetic diagnostics, management of genetic conditions, and counseling services. Pediatric healthcare providers might choose to refer to a genetics provider in cases where a specific genetic condition is suspected, additional etiological work-up is indicated after non-diagnostic testing, the provider is not equipped to facilitate discussion and ordering of specific testing, or where insurance requires a genetics evaluation before coverage of genetic testing. Following receipt of results, pediatric healthcare providers could also refer to medical geneticists given that medical geneticists are well-poised to facilitate clinical correlation of genomic variants, evaluation of VUS, and management of patients with genetic diagnoses requiring ongoing surveillance and care.

58Given the diagnostic yields and implications of genetic test results, genetic testing should be considered for any patient with ID, GDD, or ASD across the globe (Table 4). With that said, approaches to testing might be influenced by country and/or regional-specific factors. The healthcare system (universal health coverage vs. primarily private, for profit insurers), recognition of genetic professionals, access to testing due to availability and cost, and genetics education for non-genetics professionals all affect integration of genetics into the care of patients with neurodevelopmental disorders.

59Although the majority of the literature concerning genetic testing for patients with neurodevelopmental disorders focuses on genetic testing for three neurodevelopmental conditions – ASD, ID, and GDD; the rapid evolution of genomic medicine and expansion of the literature will continue to change the landscape of genetic testing shifting who is offered genetic testing, the test offerings available, and our understanding of genomic variants.

60In addition to ASD, ID, and GDD, increasing evidence supports consideration of genetic testing for other neurodevelopmental disorders. In epilepsy, between 4 and 78% of selected patients have genomic variants that are probable or definitive explanations for their epilepsy phenotypes. In a recent meta-analysis, of studies reporting on the yield of CMA, epilepsy gene panels, and ES, the diagnostic yield was 8, 23, and 45%, respectively. Specific epilepsy phenotypes, such as epileptic encephalopathies and comorbid ID have been associated with increased yields of testing. Importantly, an increasing number of genetic conditions associated with seizure phenotypes influence therapeutic decision making; therefore, identification of a genetic etiology increasingly can enable personalized care. Despite genetic testing yields similar to those if ID, GDD, and ASD, no formal practice guidelines regarding genetic testing for individuals with epilepsy have been published by major medical societies; although various approaches to etiologic genetic testing in this patient population have been suggested by independent authors.

61In CP, studies have suggested a yield of 9.6–31.0% from CMA and 10.6–65.0% using ES. In addition to emerging evidence that CP has a sizable genetic underpinning, genetic testing in CP has already resulted in changes to care. Although a practice parameter published more than 15 years ago, the 2004 joint practice parameter from the American Academy of Neurology and Child Neurology Society, indicates that genetic testing should not be pursued in children with CP, no guidelines have been published since the emergence of data that demonstrates a significant diagnostic yield of genetic testing in CP patients.

62Additional neurodevelopmental disorders, including speech disorders, Tourette syndrome ADHD, and schizophrenia have emerging evidence that genetic testing can provide an etiology to a portion of patients. Further studies are needed to understand the diagnostic yields and inform testing algorithms for these conditions.

63The overlapping symptoms, shared genetic underpinnings, and comorbidity of neurodevelopmental and neuropsychiatric diagnoses, have prompted many to advocate that neurodevelopmental and neuropsychiatric conditions collectively be considered as a continuum of developmental brain dysfunction. Familial aggregation and genetic studies have further supported the heterogenous nature of neurodevelopment and neuropsychiatric disorders. Given this and the emerging evidence of utility of genetic testing in a number of neurodevelopmental disorders, in the future, evidence may support broad implementation of genetic testing across neurodevelopmental and neuropsychiatric conditions.

64In addition to considering genetic testing for additional neurodevelopmental diagnoses, current testing technology will continue to improve, and new testing offerings will emerge. For example, ES analysis did not routinely detect copy number variants and uniparental disomy, but now some laboratories are identifying and reporting these potentially causative variants. With the rapid improvement of testing technologies and bioinformatics pipelines, ES may one day be able to detect additional genetic causes of neurodevelopmental disorders, including trinucleotide repeat expansions.

65Not only will existing testing continue to improve, the clinical integration of other testing modalities such as genome sequencing, that increases coverage of exons, improves detection of structural variants, and interrogates non-coding regions; RNA-sequencing that analyzes gene expression and mRNA splicing; and genome-wide methylation studies that assesses epigenetic modification of the genome that can impact gene expression will increase diagnostic yield. It is estimated that up to 3% of all patients with negative exomes could be explained by variants in the non-coding, regulatory regions that would be detectable by genome sequencing and could be further assessed through technologies such as RNA-sequencing and genome-wide methylation studies. In addition to informing pathogenicity of non-coding variants, RNA-sequencing and genome-wide methylation studies could also detect epigenetic changes that impact gene expression but do not alter the DNA sequence and are thus undetectable through sequencing.

66Furthermore, genome sequencing and single-cell sequencing will also enable increased detection of somatic mosaicism that, using current technologies, appear to constitute 5–10% of de novo variants. Because somatic variants are not always detectable in the blood, future testing that includes additional cell types will also increase the identification of genetic etiologies for patients and families. Availability of relevant tissues, such as single neurons, may be limited, however.

67In addition to advancements in testing technology, the understanding of genes and variants and their relationships to disease will continue to improve. Currently, only a fraction of the 20,000 human coding genes have an established disease relationship, and variants in known disease genes may be uncertain given the current knowledge base. As reference databases are bolstered, data sharing of variants and candidate genes increases, and additional technologies such RNA-sequencing and genome-wide methylation studies are integrated into testing algorithms, our understanding of genes' relationship to neurodevelopment disorders and variants' pathogenicity will improve, thus increasing diagnostic yield of testing.

68As new tests emerge and previous test results are reanalyzed or updated, recontacting of patients should be a shared responsibility among health-care providers, the clinical testing laboratories, and the patients. As such, pediatric providers ordering genetic testing and those with an ongoing relationship with the patient and family, should remain informed of advancements that might provide additional information about a reported variant or indicate that additional testing or evaluation is warranted. Providers should bear in mind that, even if an individual has had genetic testing, there may be additional testing (e.g., exome reanalysis, CMA if ES does not include adequate CNV detection, genome sequencing, and novel testing such as genome sequencing) to consider over time and it is important that clinicians continue to consider the utility of genetic testing in individuals with neurodevelopmental diagnoses.

69Currently, diagnostic genetic testing in neurodevelopmental disorders is focused on identifying rare variants of large effect size. As our ability to detect and interpret genetic causes of NDD increases, including those of more modest effect size, more families will be afforded the clinical and personal utility of genetic diagnoses. Many genetic underpinnings of NDD are characterized by variable expressivity, however, often leaving families with questions about recurrence and prognosis. In addition to the important impact of environmental and stochastic variation, recent advances suggest that the observed variability in phenotypic expression of large-effect rare variants may be explained, in part, by the additive effects of additional high-impact variants and background polygenic risk conferred by a large number of common variants of small individual effect size. Larger sample sizes and improved study methodologies are expected to lead to identification of rare variants of smaller effect size and elucidate the potential clinical role for polygenic scores in the near future. Increased understanding of additional genetic and other factors that influence phenotypes will enable tailored counseling and care for families receiving a genetic etiology for their child's NDD.

70Neurodevelopmental disorders are common, and a significant proportion are caused by rare copy number and exonic sequence variants of large effect size that can be identified by genetic testing. Identifying a genetic etiologic diagnosis can allow clinicians to provide more accurate prognostication and recurrence risk counseling, identify and treat or prevent medical comorbidities, guide patients and families to condition-specific resources and supports and, in some cases, refine treatment options. Genetic testing is now the standard of care for individuals with ID/GDD and/or ASD, and the indications for testing will almost certainly broaden to include other neurodevelopmental disorders in the future. Most genetic etiologic diagnoses in patients with ID/GDD and/or ASD can be made with broad examination of copy number and exonic sequence variants that is most efficiently achieved using ES with CNV calling. It is important for clinicians who provide healthcare to children and adolescents with neurodevelopmental disorders to gain an understanding of these common tests and their role in providing the best medical care for patients and to work to facilitate the genetic etiologic evaluation of these patients by ordering testing or partnering with genetic providers.

71JS and SM contributed to the literature review and synthesis of this review article.

72The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.