Clinical and Genetic Characteristics and Prenatal Diagnosis of Rare Monogenic Global Developmental Delay and Intellectual Disability.

Background: The estimated worldwide prevalence of global developmental delay (GDD) and intellectual disability (ID) is 1-3%. Rare monogenic GDD/ID is poorly characterized because its low prevalence limits research. In this study, we aimed to describe the diagnostic courses and clinical and genetic characteristics of a cohort with rare monogenic GDD/ID. Method: We retrospectively analyzed the diagnostic courses, clinical characteristics, and genetic spectra of rare monogenic GDD/ID patients. We also conducted a follow-up study on prenatal diagnosis in these families. Mutation pathogenicity was interpreted by molecular geneticists and clinicians according to the guidelines of the American College of Medical Genetics and Genomics. Results: Among genetic usually presented moderate to severe GDD/ID. The most common coexisting conditions were epilepsy (68%), facial dysmorphism (14%) and microcephaly (13%). In total, 149 different pathogenic variants were found in 81 different genes among the 108 pedigrees, and 71 variants were novel. The most common inheritance patterns in this outbred Chinese population were autosomal recessive (AR; 46.3%), autosomal dominant (AD; 37%), and X-linked (XL; 16.7%). GLB1, PLA2G6, SCN2A, SHANK3 and STXBP1 were important causal genes. Hot-spot mutations were rarely found. By the follow-up, 43 families, including 24 ARID, 13 ADID and 6 XLID families, had undergone prenatal diagnosis. The offspring of 6 ARID, 2 ADID and 2 XLID families had the same pathogenic variants as the probands. Conclusion:


Introduction
Global developmental delay (GDD) is characterized by a delay in achieving developmental milestones in at least two of the following domains: motor skills, speech and language, cognitive skills, and social and emotional skills [1]. Many patients with GDD demonstrate intellectual disability (ID), which is characterized by an intelligence quotient below 70 and limitations on adaptability [1,2]. The prevalence of GDD/ID in the world population is estimated to be 1-3% [3], and the average lifetime costs (direct and indirect) to support an individual with ID have reached $1 million [4,5].
Various environmental and genetic factors can result in GDD/ID [6]. Genetic reasons, including aneuploidy, copy number variants and single gene variants, account for 30-50% of cases [7], and Down syndrome, MECP2-related Rett syndrome and fragile X syndrome are the most common forms of genetic GDD/ID [6]. The modes of inheritance of monogenic GDD/ID include autosome recessive (AR), autosome dominant (AD), X-linked (XL) and maternal inheritance [8]. On this basis, some scholars divide monogenic GDD/ID into ARID, ADID and XLID [9][10][11]. With the improvement of next-generation sequencing (NGS) [12], monogenic causes are being found in previously unexplained or idiopathic cases of GDD/ID. Some scholars estimated that the diagnostic yield of exome sequencing in neurodevelopmental delay was 36%, higher than the power of chromosomal microarray (CMA) testing (15-20%) [13]. To date, nearly 1334 causative genes and 1159 candidate genes have been identi ed as related to GDD/ID [14], and the number continues to grow.
Although numerous new candidate genes and novel variants have been identi ed, most forms of monogenic GDD/ID have low prevalence, which has led to many challenging problems [15]. One such problem is the di culty of genetic diagnosis; many GDD/ID patients do not obtain accurate diagnoses. The characteristics of rare monogenic GDD/ID have not been well studied, and a limited number of articles have assessed the disparity among ARID, ADID and XLID. Hence, in this retrospective study, our rst aim was to describe the diagnostic courses and clinical and genetic features of a cohort of rare monogenic GDD/ID patients and to explore potential disparities among ARID, ADID and XLID.
In addition, effective and speci c treatments for most forms of monogenic GDD/ID are still in development, and prenatal molecular diagnosis is an important method to prevent recurrence. Since studies concerning the prenatal diagnosis of these diseases are rare, the second aim of this study is to report the results of prenatal tests for GDD/ID. Material And Methods

Study design and participants
From June 2015 to June 2019, 108 consecutive subjects under 18 years old with rare monogenic GDD/ID were recruited. The clinical diagnosis of GDD/ID was made according to the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-V) [2]. GDD was de ned by delays in the achievement of motor or mental milestones in the following domains: gross and ne motor skills, speech and language, adaptability and social skills. A developmental scale for children aged 0-6 years[16] was used to assess the Developmental Quotient (DQ) for children who were under 6 years old or failed to nish the intelligence test. Patients with a DQ of less than 70 in at least two of ve developmental domains were diagnosed with GDD. For patients over 6 years old, we used the Wechsler Intelligence Scale for Children (WISC) to quantify IQ. Those who had IQ scores lower than 70 and adaptability di culties were diagnosed with ID. The tests were performed by specialists in child development.
For etiological diagnosis, all patients were examined systematically to exclude nongenetic causes and underwent necessary genetic tests, such as G-band karyotyping, FMR1 CGG repeat testing [17], and CMA testing [18], to exclude other genetic reasons. Sanger sequencing or Trio-NGS [19,20] (targeted exome sequencing or whole exome sequencing) was performed depending on clinical judgment. The details of the detection methods are reported elsewhere.
The nal clinical and genetic diagnoses were determined by a group of pediatric neurologists, clinical geneticists and molecular geneticists. The Ethics Committee of Peking University First Hospital approved the study (2020 − 333). Informed consent was obtained from all participants.

Data collection
Demographic data, medical history, laboratory and genetic ndings were collected. The severity of GDD/ID was classi ed into four groups: mild, moderate, severe and profound, de ned by DQ or IQ scores of 50-69, [35][36][37][38][39][40][41][42][43][44][45][46][47][48][49][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34], and below 20, respectively. The age at disease onset was calculated as the interval from the date of birth to the date when the rst symptom was noticed. The age at diagnosis was calculated as the interval from the date of birth to the date when genetic diagnosis was con rmed. The interval between symptom onset and diagnosis was obtained as the age at diagnosis minus the age at disease onset. The date of genetic counseling was the time when the patient was referred to outpatient genetic counseling. The duration from genetic counseling to diagnosis was obtained by subtracting the date of diagnosis from the date of genetic counseling.
The normal standardized reference ranges of height, weight and head circumference for children at different ages were obtained from two national growth surveys of children in China [21,22]. Microcephaly, macrocephaly, short stature and facial dysmorphism were de ned in accordance with the Human Phenotype Ontology (HPO). Positive family history was de ned as having family members who presented similar traits to the probands, with or without genetic con rmation. Abnormal birth history was de ned as irregular events occurring during delivery or the neonatal period, such as amniotic uid pollution or neonatal pathological jaundice. Abnormal prenatal ultrasound ndings, such as delayed brain development, biparietal diameter anomaly and intrauterine growth retardation, were also recorded. The last follow-up for clinical outcome information (i.e., improvement, deterioration, and mortality) was in November 2019.

Criteria For Variant Interpretation
Standard gene variant nomenclature informed by the Human Genome Variation Society (HGVS)[23] was adopted to unify the description of variants. According to the 2015 American College of Medical Genetics and Genomics (ACMG) guidelines [24], variants were classi ed as "pathogenic", "likely pathogenic", "uncertain signi cance (VUS)", "likely benign", or "benign". In order to avoid biases, patients with pathogenic or likely pathogenic variants in known genes were recruited, while patients with VUS variants in known genes or variants in candidate genes were excluded.

Prenatal Diagnostic Testing
DNA from chorionic villi or amniotic uid was extracted using the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany). PCR sequencing was performed using an ABI3730 xl (Applied Biosystems, USA) to detect the causative variants harbored by probands in the family. Linkage analyses with two to ve short tandem repeat (STR) markers were performed to exclude contamination with maternal DNA and con rm the originality of a variant.

Statistical Analysis
In this study, all continuous variables were found to be nonnormally distributed; accordingly, they were described as the median (lower quartile, upper quartile) values and compared using the Kruskal-Wallis H test among three groups or the Mann-Whitney U test between two groups. Categorical variables were expressed as frequency rates (percentages). The chi-squared test was used to compare categorical data from at least two groups, and Fisher's exact test was used when the samples were limited. If differences among three groups reached statistical signi cance, multiple comparisons were performed using the Bonferroni method. All data analyses were performed using SPSS 23.0 software (SPSS Inc., Chicago, IL, USA). A two-sided α of less than 0.05 was used to de ne statistical signi cance. For multiple testing, the Bonferroni correction was used to reduce the probability of a type I error, and a two-sided α less than 0.017 (0.05/3) was considered statistically signi cant.

Results
As a tertiary genetic counseling and prenatal diagnosis center, our center served 290 families with individuals suspected of rare monogenic diseases during the 4-year study, and 142 (nearly 50%) of those patients had GDD/ID. After excluding 13 patients for missing information, 18 patients for uncertain diagnosis and 3 patients who had pathogenic variants along with atypical manifestations that could not be explained by the variants, we considered a total of 108 subjects (Fig. 1).
The majority of affected individuals were simplex cases (a single occurrence in a family), and only 9 (8.6%) patients had a positive family history. Notably, 13 (12.3%) individuals had abnormal antenatal ndings; among them, 8 patients had abnormal prenatal ultrasound results. In addition, 15 (14.2%) patients had an abnormal birth history. These parameters did not show signi cant differences among the three groups (Table 2). (The details are listed in Supplement 2.)

Variant spectra in 108 monogenic GDD/ID patients
In total, 149 different pathogenic variants were found in 81 different genes among the 108 pedigrees. Of these genes, 42 genes were transmitted in the AR pattern, 26 in the AD pattern and 13 in the XL pattern. In order to analyze the disparity in genetic spectra between different inherited models, repeated variants were included in the calculation. The results are presented in Table 3   Loss of function variants include nonsense, frameshift, start lost, single or multiple exons deletion and canonical ± 1 or 2 splice sites.
Gene ontology accumulation analyses indicated that those genes took part in multiple biological processes, including nervous system development, nervous impulse transmission, positive regulation of GTPase activity and energy metabolism. Genes associated with ion channel transport and nervous system development were mainly inherited in the AD model, while genes related to metabolism were mainly transmitted in AR or XL patterns ( Supplementary Fig. 1 [37][38][39] have reported the pathogenicity of these variants, suggesting that the 1510th base pair in the coding sequence of HEXA (NM_000520) was a common variant position.
Notably, 76 (46.9%) variants were identi ed as novel variants, and 86 (53.1%) variants have been included in disease databases (ClinVar or HGMD) or reported in PubMed articles. The rate was similar to that in previous studies [40][41][42][43][44]. The proportions of novel variants in ARID, ADID and XLID were 40%, 62% and 50%, respectively. This suggests that variant spectra in known ID genes have not been fully explored in all inheritance patterns. The higher rate of novel variants in ADID might be explained by the fact that most variants arose de novo in the AD pattern.
The major difference among ARID, ADID and XLID lies in the origin of variants. Of the 50 patients with ARID, 44 (88%) patients carried compound heterozygous variants, and 6 (12%) patients harbored homozygous variants. We con rmed that in all patients, the two abnormal alleles were separately inherited from healthy outbred parents who carried the heterozygous variants. Among 40 patients with ADID, 39 (97.5%) variants arose de novo. Of the 18 patients with XLID, 11 (61.1%) patients (2 male, 9 female) had de novo variants, 5 male patients harbored hemizygous variants inherited from their asymptomatic heterozygous mother, and 1 female (patient 70) inherited the heterozygous variant c.445C > T in PCDH19 from her non-symptomatic father. This unique characteristic was supported by previous reports [45].
In addition, parental somatic mosaicism was found in 2 cases. Patient 33, who presented with facial dysmorphism and GDD, had a c.941del in GATAD2B. The variant was also detected at a low frequency in his paternal peripheral blood genomic DNA but absent in samples of his healthy mother and sister. Therefore, it is likely that the father carries somatic and germline mosaicism for this variant. In addition, patient 93 harbored a hemizygous c.1153C > T in SLC9A6, and his mother was suspected to have the variant in mosaic state with a low peak in her peripheral blood Sanger sequencing.

Prenatal diagnosis results
In total, 43 families underwent prenatal tests to determine whether the next child would harbor the same pathogenic variants as the proband in the fetal period. As demonstrated in Table 4 and Supplement 4, among them, 24 cases were ARID, 13 cases were ADID and 6 were XLID. Thirty-six (83.7%) patients chose amniocentesis, and 7 (16.3%) patients underwent chorionic villus sampling. Among the 24 AR cases, 6 fetuses were found to carry two pathogenic variants that originated from parents who were healthy carriers, 13 fetuses harbored one variant, and 5 fetuses did not have any variants. Among the 14 AD cases, 12 fetuses did not have the variants, while 2 fetuses carried the same variants as the proband in the GATAD2B gene. Of the 6 XL cases, only 1 fetus harbored the pathogenic variant. All variants carried by fetuses were veri ed after birth or induction of labor. The appropriate time for genetic counseling is before the next pregnancy, owing to the additional procedure to con rm original molecular tests. In this study, 20 (46.5%) families had been pregnant before referral to genetic counseling and prenatal diagnosis, which might in uence further management. It has been suggested that, for most families in China in which a proband with a rare monogenic GDD/ID, referral to genetic counseling is usually delayed and re ects a shortage of related resources. Therefore, timely genetic counseling after index patients obtain a genetic diagnosis, should be emphasized to families who plan to have additional children.

Discussion
In this article, we analyzed the diagnostic courses and clinical and genetic characteristics of 108 individuals with rare monogenic GDD/ID. It often took 0.5-4 years and 3-5 referrals to obtain a genetic diagnosis after disease onset, re ecting the di culty of diagnosis. Many factors are associated with this di culty, including genetic heterogeneity, phenotype and penetrance variability, and shared signs and symptoms. Despite the great variability, when treated as a group of diseases, some features are noteworthy. The empirical ndings regarding onset age, severity and coexisting symptoms can be summarized as follows.
One of the distinguishing features is the early age of onset. In our study, all individuals presented developmental delay before 6 years of age, and 80% of them showed abnormal symptoms in the rst year of life. Nearly 10% of patients were abnormal during the prenatal stage. This nding is in accordance with previous studies showing that in monogenic forms of ID, the time of onset ranges from the 12th week after conception to early childhood[8] [46]. It also implies that future efforts should be made using NGS in the prenatal stage to detect abnormal prenatal ultrasound ndings available and affordable [47,48].
The severity of GDD/ID ranged from mild to profound in our study, and 80% of patients had severe to profound disability. This nding is consistent with previous reports that GDD/ID caused by genetic factors could be more severe than those resulting from environmental factors, as the latter are usually mild [49,50]. Previous studies concluded that de novo variants in ADID genes are the major causes of severe ID, and ARID and XLID are rare in outbred European or Korean populations [40][41][42][43], and ARID with homozygous variants is most prevalent in consanguineous populations [44]. However, in the outbred Chinese population, ARID, ADID and XLID had similar rates of severe cases. In addition, ARID with compound heterozygous variants accounted for approximately 50% of severe cases. Our results suggested that ARID with compound heterozygous variants plays an important role in monogenic GDD/ID. The inconsistency could be partially explained by the difference in study design and population.
Another possible explanation is that each person carries 100 to 200 heterozygous private variants that are potentially deleterious [51], and when asymptomatic and unrelated parents carry such a variant in the same ARID gene, their offspring have a 25% chance of illness.
Nearly 90% of individuals in this cohort had manifestations other than GDD/ID. A large proportion of patients, approximately 70% (74/108), encompassed epilepsy and ASD. This nding supports the theory that GDD/IDs share a common etiology with other cognitive and neurological disorders including ASD and seizures [49]. In addition, 32.4% (36/108) of patients manifested abnormalities in appearance, such as short stature, microcephalus, macrocephalus, facial dysmorphism and changes in skin or hair. The percentage of patients in our cohort who had involvement in other organs, including the heart, kidney and skin, was under 10% (11/108), which might be lower than the rate in GDD/ID caused by aneuploid and chromosome structural abnormalities.  [52,53].
The second aim of our study was to analyze the prenatal diagnostic situation of these groups of patients.
We have helped 46 families with prenatal diagnosis, and the recurrence rates are 25%, 15.4% and 16.7% for ARID, ADID, and XLID, respectively. For GDD/ID caused by variants in autosomal genes, the recurrence rate is determined by whether the origin of the variants is inherited from parents or occurring de novo. For XLID, the recurrence rate is related not only to the originality of the variants but also to the sex of the fetus, which should be taken into consideration in speci c situations. The necessity for prenatal diagnosis of variants inherited from parents has reached consensus in ARID and recessive XLID.
However, with an estimated recurrence rate less than 1%, the question of whether it is necessary to perform prenatal diagnosis for de novo variants in ADID and dominant XLID remains controversial. In this cohort, 2 families with ADID tested positive in prenatal diagnosis, suggesting the possibility of parental mosaicism in these cases. However, it was con rmed in only one AD family via Sanger sequencing with peripheral blood. A previous study suggested that both Sanger sequencing and exome sequencing have the ability to detect somatic mosaicism [43,50]. However, due to the limitation of sequencing depth and di culty in specimen acquisition, the existence of mosaicism in asymptomatic parents, which results in an increasing recurrence rate, is usually undetectable at present. Multiple lines of evidence suggest that the occurrence of parental germline mosaicism is underestimated. Studies using deep amplicon sequencing and digital PCR methods to detect multiple samples found that the proportion of parental mosaicism in some AD or XLID genes reached 5-20% [54][55][56]. In addition, nearly half of the ADID and XLID genes in this study have related case reports on germline mosaicism (Supplement 5). Therefore, whether con rming parental mosaicism, we recommend that prenatal diagnosis is also necessary for ADID and dominant XLID. In conclusion, genetic counseling and prenatal diagnostic services are important for families with any ARID, ADID and XLID probands.
Another observation is that at present, genetic counseling and prenatal diagnostic services are not timely for nearly 50% of families, who are rst referred for genetic counseling only after conceiving again. This might result from the lack of awareness and limitation of resources in this eld. With the improvement and availability of genetic testing technology, an increasing number of individuals will obtain accurate genetic diagnoses. Additionally, with the implementation of a universal two-child policy [57], the need for genetic counseling and prenatal diagnosis is bound to increase. Therefore, more attention should be paid to this area.
There are several limitations in our study. First, rare monogenic GDD/ID consists of a group of different disorders, and while analyzing it in a cohort, we failed to perform genotype-phenotype correlation of a single syndrome or gene. However, to delineate the relationship in detail, a group of individuals with variants in the same gene or diagnosed with the same syndrome are needed. This kind of study is restricted by sporadic cases. Second, our understanding of these rare monogenic diseases is insu cient, and regular follow-up observations of such patients might provide additional clinical information.

Conclusion
In summary, individuals with rare monogenic GDD/ID are characterized by early onset, relatively severe phenotype as well as great clinical variability and genetic heterogeneity. Patients or pedigrees with such features should be considered to undergo appropriate NGS as early as possible. The spectrum of causal genes and pathogenic variants has not yet been fully discovered. Therefore, clinicians, genetic counselors and genetic laboratories should collaborate tightly to address the problems of diagnosis posed by the bewildering clinical and genetic heterogeneity. Moreover, obtaining clinical and genetic diagnosis is not the nal step; timely referral to genetic counseling and prenatal diagnostic laboratories are important for families that plan to have additional children.

Declarations
Ethics approval and consent to participate: The Ethics Committee of Peking University First Hospital approved the study (2020-333). Informed consent was obtained from all participants.

Consent for publication:
Liling Lin collected data, performed statistical analysis and wrote the rst manuscript. Ying Zhang performed the molecular diagnostic experiments in this study. Jingmin Wang included the participants.
Yinan Ma conceived and designed the study and critically revised the manuscript. Hong Pan and Yu Qi supervised the study and critically revised the manuscript.