Applying whole exome sequencing in a consanguineous population with autism spectrum disorder

Abstract

This study aimed to systematically assess the impact of clinical and demographic variables on the diagnostic yield of Whole Exome Sequencing (WES) when applied to children with Autism Spectrum Disorder (ASD) from a consanguineous population. Ninety-seven children were included in the analysis, 63% were male and 37% were females. 77.3% had a suspected syndromic aetiology of which 68% had co-existent central nervous system (CNS) clinical features, while 69% had other systems involved. The diagnostic yield of WES in our cohort with ASD was 34%. Children with seizures were more likely to have positive WES results (46% vs. 31%, p = 0.042). Probands with suspected syndromic ASD aetiology showed no significant differential impact on the diagnostic yield of WES.

Introduction

The hallmark feature of Autism Spectrum Disorder (ASD) is social and communication deficits along with repetitive and stereotypic behaviour (American Psychiatric Association 2013). ASD prevalence is estimated to be around 1% globally, with variability between populations largely attributed to screening, diagnosis, service distribution and monitoring of statistical reports (Elsabbagh et al. 2012, Durkin et al. 2015, Al-Mamri et al. 2019, Maenner et al. 2020). Genetic-environmental interactions have been postulated to underly the pathophysiologic mechanisms of ASD which is known to manifest with high levels of heterogeneity (Kim and Leventhal 2015). Notwithstanding the advancement in molecular genetic testing, their performance in neuropsychiatric disorders in general and ASD in specific couldn’t precisely characterize the genetic makeup (Geschwind and Flint 2015). Despite the great advances in standardizing the diagnostic criteria of ASD, the wide phenotypic variability and complexity of the underlying genetic aetiology continue to challenge the efforts to unravel the exact genetic architecture of ASD.

Prevalence of comorbidities in autism is bound to reporting bias, however, its presence can point to a genetic mechanism and help in identifying an underlying aetiology (Geschwind and State 2015). Frequently reported co-morbidities include intellectual disability, reported in around 35% of autistic children (Rivard et al. 2014) and epilepsy, documented in 5–30% of autistic children (Lewis et al. 2018). The exact pathophysiological mechanism that explains the link between epilepsy and autism is not well understood. However, EEG features have been extensively studied as a marker of autistic features and progress (Bosl et al. 2011, Bernier, Aaronson and McPartland 2013). Furthermore, certain epilepsies can be explained by the common etiological trajectories leading to both disorders e.g. tuberous sclerosis (Al-Futaisi et al. 2016, Bozzi, Provenzano and Casarosa 2018).

Advances in molecular genetic technologies have allowed for an extensive study of the potential association between advanced paternal age and ASD as a result of the increased presence of de novo mutations (Anello et al. 2009, Lee and McGrath 2015). However, limited literature is available to assess confounding factors and their effect on predicting etiological factors and/or phenotypic characteristics of patients diagnosed with ASD.

ASD is a genetically heterogeneous disorder with 90% heritability as determined from family and twin studies with cumulative evidence suggesting a strong genetic component (Sandin et al. 2017). Several underlying mechanisms have been implicated in the occurrence of ASD, most importantly: copy number variants (CNVs) and de novo point mutations, but these only explain about 30% of all simplex ASD cases (Iossifov et al. 2014). Whole Exome Sequencing (WES) studies in consanguineous families have revealed a significant number of biallelic mutations in known mendelian disorder genes which have manifested in a hypomorphic pattern (Yu et al. 2013). There are no published studies on the yield of WES in ASD in consanguineous populations.

The most widely utilized guideline for diagnostic genetic testing of ASD is the one published by the American College of Medical Genetics and Genomics (ACMG) adopting Chromosomal Microarray (CMA) and Fragile X syndrome screening as the first tier tests followed by specialized molecular genetic testing as well as brain MRI (Schaefer and Mendelsohn 2013). Recently, the American Academy of paediatrics (AAP) started recommending WES as the third step in the first tier of investigations in case of negative CMA (Voigt and Macias 2018). The concept of dual testing with microarray and WES is gaining popularity. A recent study found that the yield of genetic diagnosis from CMA and Fragile X testing for children with ASD reached 12%. Interestingly, there were no phenotypic differences between subjects with and without pathogenic variants. This finding can be utilized to support the AAP recommendation to subject children with ASD to first-tier testing which includes CMA & Fragile X testing without selecting patients based on their clinical phenotype (Harris et al. 2020). However, the financial burden may prevent physicians from considering further molecular genetic testing. While the diagnostic yield of CMA was found to be 27% in a cohort of 230 Omani children with ASD (Al-Mamari et al. 2015), to our knowledge, data is lacking when it comes to the diagnostic yield of WES. International figures vary, depending on the phenotypic complexity, from 3.1% to 28.6% (Rossi et al. 2017). The primary objective of this study is to evaluate the diagnostic yield of WES among patients with ASD seen at Sultan Qaboos University Hospital (SQUH) in Oman. The study also aimed to evaluate clinical predictors that may potentially impact the diagnostic yield of WES. The latter may serve to inform clinical decisions to rationalize resource allocation and guide cost-effective evidence-based recommendations. This is especially important when considering that cost-effectiveness cannot be judged without addressing diagnostic yield (Dragojlovic et al. 2018).

Methods

This was a retrospective review of electronic records of patients diagnosed with ASD through a genetic clinic in Muscat, Oman, from February 2014 to February 2020. SQUH is located between the Muscat and Batinah governorates, representing the most densely populated areas of the country where more than 40% of the Omani population reside (National Centre of Statistics and Information 2014). The available WES reports, from children 14 years or younger, who had a full clinical phenotype assessed by specialists in Developmental Paediatrics and Medical Genetics, were included in the study. Over the study period, records of 97 children aged between 1.5 years to 14 years were retrieved. This study underwent ethical approval by the Medical Research and Ethics Committee (MREC#1726). The study adheres to the World Medical Association’s Declaration of Helsinki (1964–2008) for ethical human research regarding participant’s confidentiality, privacy, and management of the data.

Clinical assessment

The clinic receives referrals from all over the country and accepts children less than 14 years of age. Patients evaluated through the clinic generally follow the diagnostic flowchart depicted in Figure 1. ASD cases were diagnosed based on the Diagnostic Statistical Manual, Fifth Edition (DSM-5) criteria (American Psychiatric Association 2013). A multidisciplinary team, headed by a senior developmental paediatrician, utilized Autism Diagnostic Interview-Revised and Autism Diagnostic Observation Schedule to confirm the diagnosis of ASD. Patients were further clinically characterized by the contribution of a medical geneticist. Clinical and demographic variables were evaluated for their impact on the diagnostic yield. In our study, multiplex ASD refers to families in which multiple individuals were affected, while simplex ASD refers to families in which only a single individual was affected. Syndromic ASD was assigned to patients with a clinically defined pattern of systemic abnormalities and a neurobehavioral phenotype that could include ASD (Fernandez and Scherer 2017). Consanguinity was defined as the parents of the proband being related as second cousins or closer (Bittles 2001).

Figure 1.

Diagnostic flowchart.

Whole exome sequencing and clinical interpretation

DNA samples collected from patients whose standard diagnostic workup was negative were sent for WES. Samples were sent to one of the following clinical diagnostic laboratories: Breda Genetics srl, CENTOGENE AG, and MNG Laboratories. These laboratories adopt the American College of Medical Genetics and Genomics (ACMG) guidelines for interpretation of sequence variants (Richards et al. 2008, 2015). Furthermore, the abovementioned diagnostic laboratories are conducting deletion/duplication analysis for cases that didn’t have microarray CGH done on their first-tier investigation and providing confirmation for variants as part of their agreement with our institution. Variants of uncertain significance (VUS), identified from the clinical WES reports, were reviewed by two independent local clinical geneticists against a local database and following the availability of further segregation data from informative family members. Discordant re-classification of a VUS between the two clinical geneticists kept the original interpretation unchanged. A positive diagnosis, where a pathogenic or likely pathogenic variant was reported, was only considered as causative if it could explain or likely explain the ASD phenotype in question. Thus only disease-causing variants in genes implicated with neurodevelopmental disorders were considered. All the variants were checked against two genetic databases for ASD, namely; AutDB & SFARI Gene (MindSpec 2021, The Simons Foundation 2021). Furthermore, a literature review was conducted to review the variants which were not identified in either of the databases.

Statistical analysis

The main outcome variable of the study was the yield of the WES pathogenicity which was dichotomized as positive (pathogenic or likely pathogenic variant explaining the neurodevelopmental phenotype) and negative (no pathogenic or likely pathogenic variant that could explain the neurodevelopmental phenotype). A set of clinical phenotypic characteristics along with the parent's and child’s demographic characteristics were evaluated as predictors for positive diagnostic yield. Both descriptive and inferential statistical techniques were used for data analysis. Frequency distribution was used to describe the characteristics of the children with ASD and bivariate analyses with chi-square test were used to examine the significance of differences in diagnostic outcome in relation to individual clinical phenotypic and demographic characteristics of the children. A P-value of <0.05 was considered as statistically significant. Since the outcome variable pathogenicity was a binary variable, binary logistic regression models were employed to evaluate the association between positive pathogenicity and the demographic and clinical phenotypic characteristics. In logistic regression analysis, age variables were considered as continuous variables, while the other demographic and clinical phenotypic characteristic related variables were considered as categorical variables. Adjusted odds ratio (AOR) of positive pathogenicity was estimated with a 95% confidence interval (95% CI), to examine the association between pathogenicity and the demographic and clinical phenotypic characteristics. An odds ratio greater than one for a category of an explanatory variable indicated the increased probability of positive pathogenicity associated with that category of the variable compared to a reference category. An odds ratio less than one indicated a decreased probability of positive pathogenicity for that category of the variable compared to the reference category.

Results

Demographics and history

A total of 1346 confirmed cases of ASD were eligible to be included in the analysis. Among them, 749 underwent first-tier testing and 133 biochemical testing. Another 46 cases had further genetic testing. 321 cases were not tested for other reasons (parents declined testing, financial reasons or no appropriate sample was available for testing) (Figure 2). The remaining 97 children were included in the analysis, 63% were male and 37% were females (Table 1). The majority (84%) of the children were under the age of nine years, with an average age of 5.5 (SD 2.7) years. Most (89%) of the children included in the analysis had mothers within reproductive age (19–34 years), while 11% had mothers aged 35 years and older. The average age of the mothers at the time of birth was 27.6 years. The average age of fathers was 32.4 years. Array comparative genomic hybridization (array CGH) was done for three-quarters (75%) of the children, while more than half (55%) of the children had WES, performed as either solo (proband only sample analyzed) or trio (proband and parent’s sample analyzed). About one-third (34%) of the children had a positive diagnostic WES.

Figure 2.

Breakdown of patients diagnosed with ASD through Developmental Paediatrics and Medical Genetics clinic in SQUH, Oman, from February 2014 to February 2020.

Table 1.

Distribution of demographic data and genetic results of children with ASD.

Characteristics	Number	Percent
Total	97	100
Child’s age at diagnosis
<5	37	38.1
5–8	44	45.4
9+	16	16.5
Mean (SD)	5.5 (2.7)
Gender
Male	61	62.9
Female	36	37.1
Mother’s age at birth of the child
19–24	29	29.9
25–34	57	58.8
35+	11	11.3
Mean (SD)	27.6 (4.9)
Father’s age at birth of the child
<30	25	25.8
30–39	57	58.8
40+	15	15.5
Mean (SD)	32.4 (5.9)
Genetic testing
CGH done
Yes	73	75.3
No	24	24.7
Whole exome sequencing
Solo	53	54.6
Trio	44	45.4
WES Pathogenicity
Positive	33	34.0
Negative	66	66.0

Clinical factors of ASD children

Table 2 presents the distribution of children with ASD with respect to selected familial and clinical factors. 73% of the children with ASD were born to parents with consanguineous marriage. 45.4% of the ASD children had relatives of second degree or closer with a neurodevelopmental disorder. A total of 17 families had two or more affected siblings. Of those, fourteen out of 17 families had syndromic features.

Table 2.

Distribution of family and clinical factors of children with ASD.

Characteristics	Number	Percent
Family history
Consanguinity
Yes	71	73.2
No	26	26.8
Second degree or closer with neurodevelopmental disorder
Yes	44	45.4
No	53	54.6
Syndromic Versus Non-syndromic
Non-syndromic	22	22.7
Syndromic	75	77.3
Simplex Versus Multiplex families
ASD with a positive family history (Multiplex)	17	17.5
ASD with a negative family history (Simplex)	80	82.5
presence of additional factors
Intellectual Disability
Yes	83	85.6
No	14	14.4
Global Developmental Delay
Yes	59	60.8
No	38	39.2
Brain MRI abnormalities
Positive	27	27.8
Negative	30	30.9
ADHD
Yes	17	17.5
No	80	82.5
Seizures
Yes	22	22.7
No	75	77.3
Suspected syndromic etiology
Other CNS features
Yes	66	68.0
No	31	32.0
Other systemic involvement
Yes	67	69.1
No	30	30.9

The majority (77%) of the children had syndromic ASD, 83% were ASD with negative family history and 17% with positive family history cases. 86% of children had an intellectual disability, 61% had global developmental delay (GDD), 28% had brain MRI abnormalities, while 18% had ADHD and another 23% had seizures (Table 2). In our cohort, 68% had a coexistent central nervous system (CNS) clinical features, while 69% had other systems involved.

Correlates of diagnostic yield of whole exome sequencing

Table 3 shows the results of the bivariate analysis of the diagnostic yield of WES according to the characteristics of ASD children. There is a positive association between the age of children and the positive pathogenicity, as the rate of positive pathogenicity increased with age, but the effect was not statistically significant. Females were more likely to have positive pathogenicity compared to males, however, that difference was found insignificant (36% Vs. 33%, p = 0.74). Positive pathogenicity increased with increasing fathers’ age while it decreased with the age of mothers. Probands with consanguineous parents, syndromic ASD and ASD with negative family history had a higher chance of having a positive WES yield than their counterparts, but the differences were not statistically significant. In addition to the ID associated genes, presence of seizures also showed significant association with positive WES diagnostic yield. Children with seizures were more likely to have positive WES results than those who had no seizures (46% vs. 31%, p = 0.042). Probands with suspected syndromic ASD aetiology also showed no significant differential impact on the diagnostic yield of WES. No significant difference, in the yield of positive cases from solo WES (20 disease causing variants) versus Trio WES (13 disease causing variants) was identified (p-value = 0.396).

Table 3.

Percentage of children with ASD having WES pathogenicity positive or negative by their background characteristics.

Characteristics	WES Pathogenicity			Number	p-value*
	Positive n = 33	Negative n = 64				Total
Child’s age at diagnosis					0.624
<5	32.4	67.6	100.0	37
5–8	31.8	68.2	100.0	44
9+	43.8	56.3	100.0	16
Gender					0.738
Male	32.8	67.2	100.0	61
Female	36.1	63.9	100.0	36
Mother’s age at birth of the child					0.027
19–24	44.8	55.2	100.0	29
25–34	35.1	64.9	100.0	57
35+		100.0	100.0	11
Father’s age at birth of the child					0.009
<30	36.0	64.0	100.0	25
30–39	42.1	57.9	100.0	57
40+		100.0	100.0	15
Family History
Consanguinity					0.940
Yes	33.8	66.2	100.0	71
No	34.6	65.4	100.0	26
Second degree or closer with neurodevelopmental disorder					0.201
Yes	27.3	72.7	100.0	44
No	39.6	60.4	100.0	53
Syndromic Versus Non-syndromic					0.804
Non-syndromic	31.8	68.2	100.0	22
Syndromic	34.7	65.3	100.0	75
Simplex Versus Multiplex families					0.903
ASD with positive family history (Multiplex)	35.3	64.7	100.0	17
ASD with negative family history (Simplex)	33.8	66.3	100.0	80
Neurodevelopmental genes involvement
Intellectual Disability					0.641
Yes	34.9	65.1	100.0	83
No	28.6	71.4	100.0	14
Global Developmental Delay					0.684
Yes	35.6	64.4	100.0	59
No	31.6	68.4	100.0	38
Brain MRI abnormalities					0.386
Positive	40.7	59.3	100.0	27
Negative	31.4	68.6	100.0	70
ADHD					0.659
Yes	29.4	70.6	100.0	17
No	35.0	65.0	100.0	80
Seizure					0.042
Yes	45.5	54.5	100.0	22
No	30.7	69.3	100.0	75
Suspected syndromic etiology
Other CNS features					0.477
Yes	36.4	63.6	100.0	66
No	29.0	71.0	100.0	31
Other systemic involvement					0.576
Yes	35.8	64.2	100.0	67
No	30.0	70.0	100.0	30

^*P-values are based on Chi-square test.

Clinical variables and diagnostic yield of WES

To examine the predictive validity of the clinical phenotype in explaining the variations of the diagnostic yield of WES, a multiple logistic regression model was employed after controlling for the effect of demographic factors. The results indicate that the mother’s age and the presence of seizures are the two significant predictors of the positive WES diagnostic outcome. Positive diagnostic yield of WES decreased as the maternal age increased in this cohort. Probands with no seizures had 55% less risk of pathogenic WES outcome than those who had seizures (AOR = 0.453; 95% CI 0.145 − 986, p = 0.048). Although the bivariate analysis showed that the father’s age has a significant positive association with a positive diagnostic yield of WES, in multivariate analysis, after controlling for the effect of other factors, it became insignificant, indicating that it has no independent effect on positive WES yield (Tables 4 and 5).

Table 4.

Logistic regression analysis of factors affecting the positive diagnostic yield of WES pathogenicity.

Factors	B	SE of B	p-value	AOR	95% CI for AOR
					Lower	Upper
Age of child	−0.009	0.100	0.925	0.991	0.814	1.206
Mother’s age at the time of delivery	−0.168	0.078	0.032	0.845	0.725	0.986
Father’s age at the time of delivery	0.027	0.058	0.646	1.027	0.916	1.151
Gender of Child
Male	−0.087	0.542	0.873	0.917	0.317	2.654
Female (ref.)				1.000
Consanguinity of parents
Yes (ref.)				1.000
No	−0.219	0.652	0.737	0.803	0.224	2.882
Second degree or closer with neurodevelopmental disorder
Yes	−0.455	0.527	0.387	0.634	0.226	1.781
No (ref)				1.000
Syndromic versus Non-syndromic
Syndromic (ref.)				1.000
Non-syndromic	−0.537	0.738	0.467	0.585	0.138	2.486
Simplex versus Multiplex families
Simplex	0.187	0.717	0.794	1.206	0.296	4.917
Multiplex (ref.)				1.000
Intellectual Disability
Yes (ref.)				1.000
No	−0.264	0.737	0.720	0.768	0.181	3.253
Global Developmental Delay
Yes	0.399	0.566	0.480	1.491	0.492	4.519
No (ref.)				1.000
ADHD
Yes (ref.)				1.000
No	0.215	0.744	0.773	1.240	0.289	5.326
Seizure
Yes (ref.)				1.000
No	−0.792	0.082	0.048	0.453	0.145	0.986
Other CNS features
Yes	0.743	1.080	0.492	2.101	0.253	17.460
No (ref.)				1.000
Other System Involved
Yes	0.711	1.103	0.519	2.036	0.234	17.684
No (ref.)				1.000
Constant	3.762	2.791	0.178	43.039

Note: ref.: reference category.

Table 5.

Summary of genes with (pathogenic/likely pathogenic) variants detected by WES.

Gene	Zygosity	Type of mutation	Inheritance	Condition	Autism database		Literature report *
					SFARI Gene	AutDB
SCN1A	Hetero	c.4946T > C (p.Leu1649Pro)	AD	Epilepsy related disorders	Yes	Yes
PTEN	Hetero	c.344A > T (p.Asp115Val)	AD	Autism associated gene Other disorders: cancer	Yes	Yes
CNTNAP2	Homo	c.450dup (p.Gln151ThrfsTer27)	AR	Autism associated gene	Yes	Yes
DEAF1	Homo	c.997 + 4A > C	AD	Dyskinesia, seizures, and intellectual developmental disorder, Mental retardation, autosomal dominant 24, autistic features and other behavioral abnormalities.	Yes	Yes
C12orf57	Homo	c.43C > T (p.Gln15Ter)	AR	Temtamy syndrome	Yes	Yes
WDR45	Hetero	c.831-1G > A	X-linked	Beta-propeller protein-associated neurodegeneration	No	No
TDP2	Homo	c.400C > T (p.Arg134Ter)	AR	Spinocerebellar ataxia, autosomal recessive 23	No	No
MTOR	Hetero	c.5395G > A (p.Glu1799Lys)	AD	Smith-Kingsmore Syndrome	Yes	Yes
PAK3	Hemizygous	c.1274A > T (p.Lys425Met)	X-linked	Mental retardation 30, X-linked	No	No
NSUN2	Homo	c.1020del (p.Gly341ValfsTer15)	AR	Mental retardation, autosomal recessive 5	No	No
KCNQ2	Hetero	c.629G > A (p.Arg210His)		Benign familial neonatal seizures Other: early onset encephalopathy and intellectual disability.	Yes	Yes
CTNNB1	Hetero	c.893C > G (p.Thr298Arg)	AD	Tumor related gene (Wilm's tumor, ovarian cancer and others) Also associated with ASD	Yes	Yes
PGAP2	Homo	c.985G > C (p.Gly329Arg)	AR	Mabry Syndrome (associated with ID)	No	No
PPP2R5D	Hetero	c.592G > A (p.Glu198Lys)	AD	PPP2R5D-related intellectual disability	Yes	Yes
ALG13	Hetero	c.320A > G (p.Asn107Ser)	X-linked	Epilepsy related disorders	No	No
GJC2	Homo	c.1A > G (p.Met1Val)	AR	Pelizaeus-Merzbacher-like disease type 1	No	No
SCN2A	Hetero	c.304C > T (p.Arg102Ter)	AD	Autism and Epilepsy related gene	Yes	No
LARP7	Homo	c.667 + 3_667 + 6del	AR	Alazami syndrome	No	No	(Schneeberger et al. 2019)
NIPBL	Hetero	c.3575-2A > G	AD	Cornelia de Lange syndrome	Yes	Yes
CHST14	Homo	c.391_411dup (p.Val131_ Leu137dup)	AR	Ehlers-Danlos syndrome, musculocontractural type	No	No	(Casanova et al. 2020)
TMEM94	Homo	c.2734C > T (p.Arg912Ter)	AR	Intellectual Developmental Disorder with cardiac defects and dysmorphism	No	No	(Stephen et al. 2018)
VPS13B	Homo	c.8515C > T (p.Arg2839Ter)	AR	Cohen Syndrome	Yes	Yes
DYRK1A	Hetero	c.613C > T (p.Arg205Ter)	AD	Autism spectrum disorder AD mental retardation type 7	Yes	Yes
TBR1	Hetero	c.1784C > T (p.Pro595Leu)	AD	Autism spectrum disorder intellectual developmental disorder with autism and speech delay	Yes	Yes
RERE	Hetero	c.3867G > T (p.Met1289Ile)	AD	neurodevelopmental disorder with or without anomalies of the brain, eye, or heart	Yes	Yes
NAA15	Hetero	c.1720A > G (p.Thr574Ala)	AD	autosomal dominant mental retardation 50 Autistic spectrum disorders	Yes	Yes
SLC45A1	Hetero	c.526C > T (p.Arg176Trp)	AR	Intellectual developmental disorder with neuropsychiatric features	Yes	Yes
KDM5B	Homo	c.1534C > G (p.His512Asp)	AR	Autism spectrum disorder Mental retardation, autosomal recessive 65	Yes	Yes
TUSC3	Homo	c.222delA (p.Lys75AsnfsTer3)	AR	Intellectual Disability	No	No	(Al-Amri et al. 2016)
ASPM	Homo	c.1729_1730del (p.Ser577ArgfsTer33)	AR	AR primary microcephaly type 5	Yes	Yes
VWA3B	Hetero	c.655G > T (p.Glu219Ter)	AR	Spinocerebellar ataxia, autosomal recessive 22	No	No	(Kawarai et al. 2016)

^*Literature reports were obtained for variants which are not currently added in ASD databases.

The majority of the assessed families in our cohort, with positive WES results, were due to mutations associated with an autosomal recessive mode of inheritance (55%), followed by autosomal dominant inheritance (36%) and X-linked inheritance (9%). Among those with an autosomal dominant mode of inheritance, the majority were due to de novo mutations in seizure or ID-causing genes. Seventeen families had two or more affected children with ASD, most of which were associated with a syndromic form of ASD.

Discussion

The advances in genomic technologies have led to a remarkable improvement in the understanding of genetic contributions to complex disorders such as ASD (Ku et al. 2011). However, the diagnostic yield for a given condition remains a significant factor when calculating the cost-effectiveness of introducing a particular test to the flow of the diagnostic work-up (Dragojlovic et al. 2018). This is particularly evident in centres with limited financial resources where despite the declining cost of WES, it is still usually considered as the last option in the diagnostic workup for ASD. To the best of our knowledge, this is the first study that has systematically assessed the impact of clinical and demographic variables on the diagnostic yield of WES when applied to children with ASD from a consanguineous population.

The variability in the diagnostic yield of WES among children affected with ASD probably reflects the complexity of the ASD phenotype, variability in clinical ascertainment or selection for testing, and/or the changes in genetic discoveries over time (Tammimies et al. 2015, Rossi et al. 2017, Chérot et al. 2018, Du et al. 2018, Srivastava et al. 2019, Stefanski et al. 2021) In this cohort, 97 WES tests were performed in a clinically and demographically heterogeneous group of Omani children with ASD. The cohort was enriched with consanguinity with a rate of 73%. This figure is significantly higher than the national estimate of 52% in 2012 (Islam 2012), as well as the figure reported by Al-Mamari et al. (2015) from their study on Omani children with ASD (31%). Consanguinity is a known contributor to the increased risk for autosomal recessive conditions, however its impact on ASD risk is not yet well characterized. A recent study in a large ASD cohort however has shown that biallelic loss-of-function and damaging missense mutations were found in 5% of total cases, including 10% of females with ASD (Doan et al. 2019). Furthermore, a large cohort in India revealed that consanguinity increases risks for ASD with an odds ratio of 3.22 (Mamidala et al. 2015). In another Lebanese cohort, consanguinity was found to increase the risk of ASD with an odds ratio of 2.5, 95% CI [1.0, 6.0] (Guisso et al. 2018). In our cohort, parental consanguinity was identified among 73% of cases. However, 33.8% of children with consanguineous parents had pathogenic WES results in comparison to 34.6% of children with non-consanguineous parents with pathogenic WES results. The difference in WES pathogenicity was not statistically significant, p-value= 0.940. On the other hand, this could be the underlying cause of abundance of recessive mutations in our cohort which is considered higher than other cohorts. This may be in part due to the small sample size and selection bias, or it could indicate that rare recessive alleles causing ASD are yet to be discovered. It is also worth noting that this study did not consider candidate genes as a positive diagnostic outcome.

The WES diagnostic yield in our study was found to be 34%. This is comparable to figures reported from a meta-analysis for the diagnostic yield of WES among individuals with neurodevelopmental disorders (Srivastava et al. 2019). The male gender predominance among children affected with ASD was also identified within our study with 63% of enrolled patients being male (Halladay et al. 2015).

Despite the number of clinical and demographic variables evaluated to predict a positive WES diagnostic outcome, only seizures and parental age reached statistical significance. In contrast to our findings, Rossi and colleagues reported that the presence of seizures did not significantly impact the diagnostic yield in their study (Rossi et al 2017). Recent data has shown that the yield of WES for children with ASD and epilepsy has reached 9% and this can be increased to 33% by widening the phenotype to include ASD children with abnormal EEG (Shillington and Capal 2021).

Besides the constraint of the smaller sample size in our cohort, it remains possible that the diagnostic impact of seizures on the outcome of WES becomes more pronounced in consanguineous populations. Rossi and colleagues found psychiatric conditions, ataxia and/or paraplegia and multiple congenital anomalies being associated with a positive diagnostic outcome. These clinical variables have been under-represented in our cohort as next-generation sequencing panels were preferred by our clinical geneticists as the starting diagnostic tool in our unit. We found no statistically significant association between multiple congenital anomalies and a positive diagnostic outcome in our cohort.

This study is among the first studies to evaluate the impact of parental age on the diagnostic outcome of WES. While the effect of paternal age may be a statistical artefact, the effect of maternal age is rather interesting. A positive diagnostic outcome of WES was more likely to be encountered with decreasing maternal age. Younger women are more represented in our society, as women tend to have children at a young age. Unless reproduced in a larger study, one cannot exclude that this finding was in part an effect of selection bias. In our clinic, children born to women with advanced maternal age are more likely to undergo chromosomal analysis and array CGH without necessarily proceeding to WES as the reproductive counselling value then becomes of less significance.

Tammimies et al. (2015) demonstrated that differences in the yield of WES was related to morphological stratification of ASD probands based on clinical examination. In their study, a total morphology score was used in which each child was classified as essential, equivocal, or complex (Miles et al. 2005). Using this classification, they identified the WES yield to be higher in the equivocal and complex phenotypic presentations in comparison to the group with the essential phenotype. Interestingly, the middle group ‘equivocal’ had a higher yield (28.6% OR 11.5) than that of the complex phenotype group (16.7% OR 6.0) (Tammimies et al. 2015). Additionally, by combining the analysis of CMA & WES, they found that 35% of ASD children with additional medical and dysmorphology features had a molecular diagnosis in contrast to 6.0% of ASD children without syndromic features. Our classification system was inspired by those results as we opted to classify children into a dichotomous; syndromic and non-syndromic to avoid the middle group effect. The classification of ASD cases into morphologically syndromic versus non-syndromic is subject to the geneticist’s judgment based on their level of clinical expertise (Fernandez and Scherer 2017). Two-thirds of this cohort of children with ASD were classified clinically as ‘syndromic.’ This relatively high percentage is likely due to clinical selection bias and patients being seen in a teaching hospital with highly specialized expertise. Considering the retrospective nature of this study, it would be expected that patients were already under follow up awaiting the attainment of a clinical genetic diagnosis. Undoubtedly, ASD with a syndromic phenotype would tempt geneticists to proceed with WES despite negative first-tier results as per ACMG 2013 recommendations (Schaefer and Mendelsohn 2013). It is worth mentioning that 75% of this cohort underwent first-tier investigations with either negative or inconclusive results that failed to provide sufficient explanation of the clinical phenotype.

Despite the progressive decline in WES costs, it is still considered to pose a financial burden on the health system and is usually kept as a last resort diagnostic test. In addition, its diagnostic yield for a given condition is highly pivotal for calculating its cost-effectiveness to inform the decision about the flow of clinical diagnostic odyssey (Dragojlovic et al. 2018).

This study has a few limitations largely stemming from its retrospective nature, the representation of a single centre experience, and the relatively small sample size. In addition, it would not be possible to ignore the inherent clinical selection bias when making decisions about the diagnostic workup of patients with ASD in our clinic.

Conclusion

The diagnostic yield of WES in our cohort of Omani children with ASD is 34%, which is considered higher than other populations. Despite the complexity of our sample and the high percentage of syndromic cases, there was no clinical prediction model that could predict the positive WES results, however, seizure disorder was statistically significant when evaluating its association with a positive WES result. These results confirm that WES is a powerful diagnostic tool for patients with ASD enabling further recurrence risk estimates for couples with ASD children. Our results might help in estimating the cost-effectiveness of involving WES at an earlier step in the diagnostic process of children with ASD in the Omani population.

Funding Statement

The authors declare that no funding was received from any agency for the research.

Conflict of interest

The authors declare no conflict of interest with respect to this research, authorship, and/or publication of this article.