Genetic influence on neurodevelopment in non-syndromic craniosynostosis

Abstract

BACKGROUND:

Non-syndromic craniosynostosis (NSCS) is one of the most common anomalies treated by craniofacial surgeons. Despite optimal surgical management, nearly half of children with NSCS have subtle neurocognitive deficits. Whereas the impact of timing and type of surgical intervention on neurodevelopment in NSCS has been studied, the possibility of genetic influence on neurodevelopment remains unexplored.

METHODS:

We performed whole exome sequencing for 404 case-parent trios with sporadic NSCS. Statistical analyses were performed to assess the burden of de novo mutations in cases compared to both expectation and 1,789 healthy control trios. Individuals with and without each mutation class were analyzed, and the presence or absence of various types of neurodevelopmental delay, including motor delays, speech/language delays, and intellectual disability, were recorded alongside demographic information.

RESULTS:

We identified a highly significant burden of damaging de novo mutations in mutation intolerant (pLI>0.9) genes in NSCS probands (P=5.9x10⁻⁶). Children with these mutations had a 2-fold higher incidence of neurodevelopmental delay (P=0.001), >20-fold greater incidence of intellectual disability (P=7.2x10⁻⁷), and were 3.6-fold more likely to have delays that persisted past 5 years of age (P=4.4x10⁻⁴) in comparison to children with NSCS without these mutations. Transmitted loss of function mutations in high pLI genes also conferred a 1.9-fold greater risk of neurodevelopmental delay (P=4.5x10⁻⁴).

CONCLUSIONS:

These findings implicate genetic lesions concurrently impacting neurodevelopment and cranial morphogenesis in the pathoetiology of NSCS, and identify a strong genetic influence on neurodevelopmental outcomes in affected children. These findings may eventually prove useful in determining which children with NSCS are most likely to benefit from surgical intervention.

Introduction

Occurring once in every 2,000 births, craniosynostosis is one of the most frequent congenital anomalies encountered by craniofacial surgeons¹. Children with NSCS have premature fusion of one or more cranial sutures in the absence of other congenital anomalies. Surgery for NSCS is generally performed within the first year of life to prevent elevations in intracranial pressure and thus mitigate risk of extrinsic brain compression, which can contribute to adverse neurodevelopmental outcomes². Despite early and comprehensive surgical management, nearly half of NSCS probands are ultimately diagnosed with some form of neurodevelopmental delay^3,4. While studies have assessed the contribution of both timing of surgery and type of surgical correction performed on neurocognitive outcomes^5,6, the potential role of genetic lesions affecting both cranial suture morphogenesis and intrinsic brain function have not been explored. We sought to determine whether the frequent aberrations in neurodevelopment observed in NSCS are under genetic influence.

The impact of sporadic, or de novo, mutations on neurocognitive outcomes in structural birth defects and other neurodevelopmental disorders has been investigated extensively^7,8. These studies have identified a frequent role of pathogenic mutations in genes with high pLI (probability of loss of function intolerance), a metric of intolerance to heterozygous loss of function based on mutation frequencies in healthy populations compared to expectation^9,10. De novo mutations (DNMs) in the BMP, Wnt, and FGF/MAPK pathways have been implicated in the pathogenesis of NSCS, with de novo and transmitted mutations in these pathways causing at least 10% of cases¹¹. The genetic cause of the majority of NSCS cases, however, remains unknown. We posited that sporadic mutations altering genes under strong evolutionary constraint (high pLI) might contribute to the neurodevelopmental delays frequently present in children with NSCS.

Materials and Methods

Subjects and Samples

Participants in this study were enrolled from the Yale Pediatric Craniofacial Clinic, or by responding to an invitation posted on the Cranio Kids–Craniosynostosis Support and Craniosynostosis–Positional Plagiocephaly Support Facebook pages. All individuals or their parents, if participants were minors at the time of enrollment, provided written informed consent to participate in a study of the genetic causes of CS. Inclusion criteria were diagnosis of NSCS by a craniofacial plastic surgeon or pediatric neurosurgeon. Clinical documentation supporting a diagnosis of NSCS and indicating timing and type of surgery was provided by participants in the form of operative reports and clinic visit notes. Information regarding presence or absence of various forms of physician-diagnosed neurodevelopmental delay was collected from parents in 2020, which was, on average, 5 years after enrollment in the study. The study protocol was approved by the Yale Human Investigation Committee’s Institutional Review Board.

Exome Sequencing

Patients provided DNA via buccal swab or saliva samples according to established manufacturer’s protocols. Exome sequencing was performed using the IDT xGen or Roche V2 capture reagent, followed by either 75 or 99 base paired end sequencing on the Illumina platform, with variants called and analyzed as previously described^11,12. De novo mutations were called using TrioDeNovo¹³ (Table 1). Control trios were sequenced on the same platform at the same institution, with variants called and analyzed using the same protocol.

Table 1:

De novo mutations identified in 404 probands with NSCS

	Observed		Expected		Enrichment	P
All genes	n	Rate	n	Rate
Total	451	1.12	451.50	1.12	1.00	0.52
Synonymous	89	0.22	128.20	0.32	0.69	1.00
Protein altering	357	0.88	323.40	0.80	1.10	0.03
Total missense	305	0.75	283.60	0.70	1.08	0.10
T-mis	233	0.58	230.40	0.57	1.01	0.44
D-mis	72	0.18	53.20	0.13	1.35	8.1 x 10−3
LOF	52	0.13	39.70	0.10	1.31	0.04
Damaging	129	0.32	92.90	0.23	1.39	2.3 x 10−4
High pLI genes
Total	104	0.26	55.55	0.14	1.87	4.3 x 10−9
Synonymous	22	0.05	14.45	0.04	1.52	0.04
Protein altering	82	0.20	41.10	0.10	2.00	1.3 x 10−8
Total missense	71	0.18	35.00	0.09	2.03	6.1 x 10−8
T-mis	53	0.13	27.77	0.07	1.91	7.8 x 10−6
D-mis	18	0.04	7.23	0.02	2.49	5.2 x 10−4
LOF	10	0.02	4.52	0.01	2.21	0.02
Damaging	30	0.07	11.74	0.03	2.56	5.9 x 10−6
High brain expressed, high pLI genes
Total	53	0.13	27.10	0.07	1.96	7.0 x 10−6
Synonymous	11	0.03	8.81	0.02	1.25	0.27
Protein altering	42	0.10	18.29	0.05	2.30	1.4 x 10−6
Total missense	36	0.09	15.58	0.04	2.31	6.7 x 10−6
T-mis	26	0.06	12.2	0.03	2.13	3.9 x 10−4
D-mis	10	0.02	3.38	0.008	2.96	0.003
LOF	6	0.01	2.03	0.005	2.96	0.02
Damaging	16	0.04	5.42	0.013	2.95	1.7 x 10−4

n, number of de novo mutations; Rate, number of de novo mutations per subject; Damaging and tolerated missense called by MetaSVM (D-mis, T-mis respectively); Loss of function denotes premature termination, frameshift, or splice site mutation; Damaging denotes LOF, D-mis, and nonframeshift insertion/deletion mutations. For overall burden of de novo mutations, expected number derived from expectation in DenovolyzeR. For high pLI/HBE gene sets, expectation is derived from observed number of mutations in 1,789 control trios (see Methods). P-values represent the upper tail of the Poisson probability density function. Bold values represent significant findings (P<0.01 after Bonferroni correction)

Comparison of Rates of Neurodevelopmental Delays in cases

Patients and families provided clinical information regarding physician-diagnosed developmental delays, including intellectual disability, motor delay, speech or language delay, and whether or not the diagnosed delays had resolved to date. Demographic information including parental age, income, and education levels was self-reported (Table 2). The number of individuals with and without each mutation class analyzed was tabulated after undergoing sequencing and downstream analyses, and the presence or absence of each type of neurodevelopmental delay was recorded. Values from 2x2 contingency tables were compared using Fisher’s exact test.

Table 2:

Demographics for all patients (n=256)

	n (%)
Mean age at surgery (months)	6.2 ± 5.3
Frequency of surgery type
Endoscopic procedure with helmeting	74 (28.9%)
Cranial Vault Remodeling	154 (60.2%)
Spring-assisted cranioplasty	10 (3.9%)
None/not sure	18 (7.0%)
Frequency of second surgery
Yes	34 (13.3%)
No	222 (86.7%)
Mean parental income – mean rank	2.8 ± 1.2
<20,000	4 (1.6%)
20,000-50,000	23 (9.0%)
50,000-100,000	85 (33.2%)
100,000-175,000	84 (32.8%)
175,000-250,000	34 (13.3%)
>250,000	26 (10.2%)
Mean father education level – mean rank	2.7 ± 1.1
Some high school	3 (1.2%)
High school graduate	41 (16.0%)
Some college	52 (20.3%)
College Graduate	97 (37.9%)
Grad/Professional school	63 (24.6%)
Mean mother education level – mean rank	3.1 ± .9
Some high school	1 (.4%)
High school graduate	13 (5.1%)
Some college	44 (17.2%)
College Graduate	103 (40.2%)
Grad/Professional school	95 (37.1%)
Geographic region
Northeast	63 (24.6%)
Midwest	69 (27.0%)
South	63 (24.6%)
West	54 (21.1%)
Other	7(2.7%)
Frequency of delays/no delays
Yes	88 (34.4%)
No	168 (65.6%)
Frequency of motor, language, intellectual delays
Motor	48 (18.8%)
Language	70 (27.3%)
Intellectual disability	11 (4.3%)
Frequency of ADD/ADHD	24 (9.4%)
Frequency of emotional disturbance	20 (7.8%)
Persistent Delay
No delay	168 (65.6%)
Delay, not resolved	42 (16.4%)
Delay, resolved	46 (18.0%)

Burden of De Novo Mutations

Statistical analyses assessing the burden of de novo mutations in cases compared to expectation were performed in R using the denovolyzeR package¹⁴. The burden of de novo mutations in high pLI genes across gene classes was assessed by comparing the observed number of de novo mutations in each gene class in cases to the expected rate derived from mutations observed in 1,789 control trios representing unaffected siblings of probands with autism from the Simons Simplex cohort⁸. To determine the expected number of de novo mutations in each gene class, the rate observed in 1,789 control patients was multiplied by 404/1,789 to derive the number expected in a similarly sized cohort. The observed and expected values were compared using the Poisson test.

Results

We recruited a cohort of 404 trios with sporadic NSCS, including 247 with sagittal CS, 133 with metopic CS, 16 with lambdoid CS, and 8 with sagittal and metopic CS (Table 2). We identified, on average, 1.12 coding de novo mutations per proband, as well as a significant excess of both loss of function (LOF) and damaging (as called by MetaSVM) missense mutations in probands (P=2.3x10⁻⁴; Table 1; Figure 1). In order to assess the burden of de novo mutation in high pLI (>0.9) genes, we compared those mutations identified in NSCS probands to expectation based on prevalence in 1,789 healthy controls using the Poisson distribution (Table 1). NSCS probands had a 2.6-fold excess of damaging de novo mutations in high pLI genes (P=5.9x10⁻⁶). Restricting these findings to only those high pLI genes within the highest quartile of expression in the developing brain (HBE)⁷, we identified a 3-fold enrichment of damaging de novo mutations in probands (P=1.7x10⁻⁴; Table 1). Whereas we identified a significant enrichment in ‘tolerated’ high pLI missense mutations in NSCS cases as well (1.9-fold enrichment, P=7.8 x 10⁻⁶, Table 1), the overall impact of these mutations on rates of neurodevelopmental delay was not significant (1.2-fold increase in neurodevelopmental delay prevalence, P=0.4; Table 1). Pathway analysis of genes in which damaging de novo mutations were identified revealed several gene ontology terms with significant enrichment, including regulation of multicellular organismal development (GO:2000026, 4.8-fold enrichment, q=7.1x10⁻⁵), regulation of developmental process (GO:0050793, 4.1-fold enrichment, q=1.5x10⁻⁴), regulation of nervous system development (GO:0051960, 5.1-fold enrichment, q=0.04) and regulation of neuron projection development (GO:0010975, 7.0-fold enrichment, q=0.05) (See Table, Supplemental Digital Content 1, which shows Pathway analysis of genes mutated in NSC probands. Genes with de novo mutations were assessed for enrichment in gene ontology terms. Pathways demonstrating significant enrichment include several processes implicated in neurodevelopment).

Figure 1:

Exome sequencing of case-parent trios with non-syndromic craniosynostosis. We performed whole exome sequencing of 404 case-parent trios with sporadic non-syndromic craniosynostosis. De novo mutations were identified in children with NSC, and the burden of de novo mutations in cases was compared to those identified in 1,789 control trios. Drawing by Erin M. Wolfe

In order to assess the impact of these de novo mutations in high pLI genes on neurodevelopmental outcomes in NSCS, the frequency of neurodevelopmental delays in those with and without mutations were compared. Probands <3 years of age were excluded from these analyses, as screening for neurodevelopmental delays may not have been performed prior to enrollment in preschool. Those with mutations in SMAD6, a known genetic cause of NSCS imparting high neurodevelopmental risk, were excluded¹⁵. No participating individuals were found to have pathogenic mutations in syndromic craniosynostosis genes frequently mutated in NSCS (FGFR3, TWIST1, TCF12)¹. After exclusion of probands assessed at <3 years, those with SMAD6 mutations, and those with incomplete records, 256 kindreds remained for analysis. There were no significant differences in the timing or type of surgical intervention, nor were there significant differences in parental age, income, or education between NSCS patients with delays in comparison to those who were not diagnosed with delays (Table 3). 63% of probands with damaging DNMs in high pLI genes experienced neurodevelopmental delays compared to 31.0% of those who did not harbor these mutations. Probands with damaging DNMs in high pLI genes were 2.5-fold more likely to experience motor delays (P=0.004), 1.9-fold more likely to experience speech or language delays (P=0.01), and >20-fold more likely to have intellectual disability (P=7.2x10⁻⁷) (Figures 2,3) (See Table, Supplemental Digital Content 2, which shows the dataset for figure 2) (See Table, Supplemental. Digital Content 3, which shows the dataset for figure 3). In considering 180 probands who were assessed past the age of 5 years, those with damaging de novo mutations in high pLI genes were 3.6-fold more likely to have persistent neurodevelopmental delays in comparison to NSCS cases who did not (P=4.4x10⁻⁴; Figure 2; Table 4).

Table 3:

Comparison of patient demographics between all patients with any kind of delay vs. without delays (n=256)

	No Delay	Yes Delay	P
Overall Frequency	168 (65.6%)	88 (34.4%)
Mean age at surgery (months)	5.88 ± 4.05	6.82 ± 7.03	0.18
Frequency of surgery type			0.754
Endoscopic procedure with helmeting	52 (31.0%)	22 (25.0%)
Cranial Vault Remodeling	97 (57.7%)	57 (64.8%)
Spring-assisted cranioplasty	7 (4.2%)	3 (3.4%)
None/not sure	12 (7.1%)	6 (6.8%)
Frequency of second surgery			0.699
Yes	21 (12.5%)	13 (14.8%)
No	147 (87.5%)	75 (85.2%)
Mean father age at birth	33.0 ± 5.9	33.2 ± 6.4	0.72
Mean mother age at birth	31.0 ± 4.9	31.3 ± 5.5	0.611
Mean parental income – mean rank	2.86 ± 1.19	2.63 ± 1.08	0.126
Mean father education level – mean rank	2.73 ± 1.05	2.60 ± 1.05	0.348
Mean mother education level – mean rank	3.14 ± .91	2.98 ± .82	0.154
Frequency of state growing up			0.862
Northeast	38 (22.6%)	25 (28.4%)
Midwest	48 (28.6%)	21 (23.9%)
South	42 (25.0%)	21 (23.9%)
West	35 (20.8%)	19 (21.6%)
Other	5 (3.0%)	2 (2.3%)

Note: Persistent delays were defined as those not having resolved to date. Only patients >5 years of age were included in the analysis (n=180). Percentages reflect the proportion of each column indicating either delay or no delay.

Figure 2. Prevalence of neurodevelopmental delays in probands with NSC across mutation classes.

‘No mutation’ refers to cases in which no protein altering high PLI de novo mutation or transmitted high pLI LOF was identified. ‘Protein-altering high PLI de novo’ describes any de novo mutation affecting protein sequence in a gene with PLI>0.9. Protein-damaging refers to damaging missense as called by MetaSVM, nonsense, frameshift, canonical splice sites, inframe insertion/deletion mutations, or whole gene deletions. ‘Transmitted high PLI LOF’ encompasses transmitted nonsense, frameshift, and canonical splice site mutations with ExAC frequency < 2x10⁻⁵ in genes with PLI>0.9 identified in probands. For each statistic, P values represent Fisher’s exact test statistic comparing rate of delay specified in those with and without mutations in the designated class. Percentages reflect proportion of patients with each type of delay within the mutation class categories. Analysis of persistent delay only includes cases assessed after 5 years of age (n=180); n for each mutation class found in the dataset in Supplemental Digital Content 2.

Figure 3. Odds ratios for neurodevelopmental delays in probands with NSCS across mutation classes.

The vertical, dashed line represents the point at which the odds ratio is 1. Odds ratios with associated confidence intervals are found below in the table. LOF; loss of function mutation, including nonsense, frameshift, or canonical splice site mutations. ‘No mutation’ refers to cases in which no protein altering high PLI de novo mutation or transmitted high pLI LOF was identified. ‘Protein-damaging’ refers to damaging missense as called by MetaSVM, inframe insertion/deletion mutations, or LOFs. OR= odds ratio, derived from Fisher’s exact test statistic comparing rates of delay in those with and without specified mutations. Persistent delays refer to those still present when assessed past age 5 (See dataset in Table, Supplemental Digital Content 3).

Table 4:

Comparison of frequency of persistent delay between various types of genetic mutations

	Frequency		No persistent delay		Yes persistent delay
	n	%	n	%	n	%	P
Damaging de novo							0.003
Absence of mutation	130	72.2%	114	87.7%	16	12.3%
Presence of mutation	50	27.8%	34	68.0%	16	32.0%
Damaging de novo with high PLI							4.4x10−4
Absence of mutation	160	88.9%	138	86.3%	22	13.8%
Presence of mutation	20	11.1%	10	50.0%	10	50.0%
Either damaging de novo high PLI or transmitted high PLI							1.7x10−7
Absence of mutation	119	66.1%	111	93.3%	8	6.7%
Presence of mutation	61	33.9%	37	60.7%	24	39.3%
Protein Altering High PLI							0.062
Absence of mutation	141	78.3%	120	85.1%	21	14.9%
Presence of mutation	39	21.7%	28	71.8%	11	28.2%
Transmitted High PLI LOF							3.2x10−5
Absence of mutation	130	72.2%	117	90.0%	13	10.0%
Presence of mutation	50	27.8%	31	62.0%	19	38.0%
Any high PLI							7.7x10−5
Absence of mutation	102	56.7%	94	92.2%	8	7.8%
Presence of mutation	78	43.3%	54	69.2%	24	30.8%

Note: Persistent delays were defined as those not having resolved to date. Only patients >5 were included in the analysis (n=180). Percentages reflect the proportion of each row indicating either presence or absence of mutation.

Considering only families in which transmitted loss of function (LOF) mutations in high pLI genes were identified, probands with these transmitted LOFs were 1.9-fold more likely to have developmental delays (P=4.4x10⁻⁴: Figures 2,3), and 3.8-fold more likely to have delays that persisted past age 5 (P=3.2x10⁻⁵; Figures 2,3). Finally, we compared the rate of neurodevelopmental delay (NDD) in probands with either damaging DNMs or transmitted LOFs in high pLI genes to those who had neither; probands with these mutations were 1.9-fold more likely to have any NDD (P=1.1x10⁻⁴), and 5.9-fold more likely to have delays that persisted past age 5 (39.3% vs 6.7%; P=1.7x10⁻⁷; Figures 2,3; Table 4).

Discussion

Taken together, our results support a role for damaging de novo mutations and transmitted LOF mutations in highly constrained genes in affecting neurodevelopment in NSCS patients. Interestingly, these results also suggest that those harboring these mutations are significantly more likely to have sustained delays after early, comprehensive surgical intervention. Limiting our analyses to only those high pLI genes within the highest quartile of expression in the developing brain (HBE)⁷, our findings demonstrated significant enrichment, implicating de novo mutations in evolutionarily constrained genes involved in neurodevelopment in the pathoetiology of NSCS.

Interestingly, of 82 genes in which protein altering de novo mutations were identified in high pLI genes, 56(68%) have no known disease association, and nearly all of the 26 with known disease associations are associated with neurologic or neurodevelopmental disorders (Table 5). With the exception of Marfan syndrome (FBN1), none of the associated disease phenotypes are known to involve craniosynostosis. Many of the mutations identified in this cohort are missense mutations in genes in which heterozygous LOF mutations cause the associated neurodevelopmental syndromes (Table 5), suggesting that hypomorphic alleles or allele specific effects of these mutations could result in craniosynostosis with a concomitant impact on neurodevelopment. We identified protein altering high pLI DNMs in chromatin modifiers (KMT5B, KMT2B, EP300), transcription factors (MYRF, MYT1L, ETV6, FOXO1) transcriptional regulators (ATF7IP, BAZ2A, BRD3, NCOR2, PWWP2A, ZNF503, TAF3, TAF4, CBX4, WAC), regulators of RNA processing (SON, ZCCHC11, SETX), and components of the ubiquitin-proteasome system (PSMC2, PSMC5, USP34, HERC2); each of these processes has been implicated in the pathoetiology of human neurodevelopmental disorders, supporting a role for such mutations in NSCS-associated neurodevelopmental delay.

Table 5:

High pLI genes with protein-altering de novo mutations with known OMIM associations

Gene	Mutation	pLI	Brain Expression	OMIM association
LRP2	R3236X	1.00	57.56%	222448: Donnai-Barrow syndrome
KMT5B	T97fs	1.00	89.25%	617788: Mental retardation, autosomal dominant 51
ETV6	gene deletion	0.97	75.95%	616216: dominant thrombocytopenia
MACF1	IVS89+1G>A	1.00	72.82%	618325: lisencephaly with complex brain stem malformation
PIK3R1	IVS4-2A>G	1.00	51.12%	269880: short syndrome
PTCH1	S403R	1.00	93.22%	109400: Gorlin syndrome; 610828: holoprosencephaly 7
HERC2	R3102W	1.00	84.47%	615516: Mental retardation, autosomal recessive 38
TSC1	H711R	1.00	68.63%	191100: tuberous sclerosis
COL11A1	G1162S	1.00	64.87%	154780: Marshall syndrome 604841: Stickler syndrome type 2
PTPRC	Y882C	1.00	22.56%	608971: Severe combined immunodeficiency, T cell-negative, B-cell/natural killer-cell positive (recessive)
SETX	G2423R	0.95	62.67%	602433: Amyotrophic lateral sclerosis 4, juvenile 606002: Spinocerebellar ataxia, autosomal recessive, with axonal neuropathy 2
SON	L600F	1.00	97.49%	617140: ZTTK syndrome
WAC	R47G	1.00	90.14%	616708: Desanto-Shinawi syndrome
KMT2B	R974H	1.00	87.43%	617284: Dystonia 28, childhood-onset
EP300	Q1043H	1.00	87.21%	613684: Rubinstein-Taybi syndrome 2 618333: Menke-Hennekam syndrome 2
KIAA1549	R1603W	0.99	84.38%	618613: Retinitis pigmentosa (recessive)
PKD1	R2186C	1.00	71.09%	173900: Polycystic kidney disease 1
FBN1	I2812V	1.00	60.84%	154700: Marfan syndrome 604308: MASS syndrome 184900: Stiff skin syndrome 608328: Weill-Marchesani syndrome 2, dominant 102370: Acromicric dysplasia 129600: Ectopia lentis, familial 614185: Geleophysic dysplasia 2 616914: Marfan lipodystrophy syndrome
DMXL2	C2088R	1.00	43.80%	618663: Epileptic encephalopathy, early infantile, 81
KCNB1	P775H	1.00	36.27%	616056: Epileptic encephalopathy, early infantile, 26
ADCY5	K166Q	1.00	35.07%	606703: Dyskinesia, familial, with facial myokymia
MYT1L	R179Q	1.00	24.14%	616521: Mental retardation, autosomal dominant 39
DOCK2	K636Q	1.00	21.89%	616433: Immunodeficiency 40 (recessive)
SPTBN4	R256W	1.00	18.58%	617519: Neurodevelopmental disorder with hypotonia, neuropathy, and deafness (recessive)
GRIN2A	K1339E	1.00	14.75%	245570: Epilepsy, focal, with speech disorder and with or without mental retardation
MYRF	P906T	1.00	56.54%	618280: Cardiac-urogenital syndrome; 618113: Encephalitis/encephalopathy, mild, with reversible myelin vacuolization

pLI, probability of loss of function intolerance; Brain expression, percentile in comparison to all protein-coding genes expressed in developing brain; OMIM, Online Mendelian Inheritance in Man

Whereas the type of surgical correction and age at surgery’s impact on neurodevelopment have been the predominant focus of investigation in recent years, we identify a substantial genetic contribution to neurodevelopmental outcomes. It is likely that each of these factors is independently influential. The finding that children with damaging de novo mutations in high pLI genes were more than twice as likely to have some form of neurodevelopmental delay and >20-fold more likely to be diagnosed with intellectual disability when compared to mutation-negative children who underwent early surgical management suggests that certain adverse neurodevelopmental outcomes are genetically intrinsic. Although this is the largest and most comprehensive analysis to date of the effects of genetic abnormalities in NSCS, and it provides much new information, it should be taken in the context of the overall discussion of how and when to treat patients presenting with an apparent typical single suture vault synostosis. One particular limitation of this study is the the lack of uniformity in neurodevelopmental testing. This study relied on patient-reported diagnoses, with no standardized battery of neurocognitive tests employed uniformly across all probands. Further validation studies will employ formal neurocognitive testing in children with genetic data to provide further insight into role of genetics in neurodevelopment in NSCS. Studies of outcomes to date on this topic have also focused on patients with a lower rate of developmental delays ^5,6. Consistent with many developmental studies, a range of IQ is set considering what is “typical” and possibly recoverable, and what is attributable to “primary” neurologic dysfunction, which is unlikely to be improved with surgery. That standard is usually an IQ <70. A number of patients in this study have an IQ that is likely to be less than that standard, given a diagnosis of intellectual disability (n=11). Ten of these eleven children were found to have high risk genotypes. Each of these eleven children underwent surgical management for apparent NSCS before one year of age, at which point it would not have been possible to diagnose intellectual disability. The question then is whether it is appropriate to do a genetic analysis on every patient presenting for initial surgical consultation. This, in our point of view, could ultimately inform preoperative assessment, and provide insight into a clearer pathway for appropriate treatment, whether that be surgical or non-surgical. The identification of high risk genotypes in early infancy could also be used to identify those at greatest risk of experiencing delays, and potentially guide these individuals towards enrollment in early intervention programs sooner to mitigate risk.

These findings provide insight into the prevalence of subtle yet persistent neurodevelopmental delays in NSCS as well. The overall observation that those with high risk genotypes had a 5.9-fold greater risk of having any delay that was still evident after 5 years of age (39.3% vs 6.7%; Table 4) despite optimal surgical management suggests that children with these mutations may be less likely to benefit from surgical intervention than children who do not harbor these mutations. Strikingly, 34% of children with NSCS were found to have these high risk genotypes; it is of course possible that surgical intervention ameliorates the degree of neurodevelopmental delay in those with high risk genotypes, however further studies will be necessary to elucidate the degree of benefit that surgery might provide. While genetics alone would never exclude a child with NSCS from undergoing operative intervention, further studies aimed at identifying those children with NSCS who are most or least likely to benefit from surgery could potentially inform preoperative consultations.

Although the most comprehensive to date, it remains to be determined whether the population studied is sufficient in number to provide generalizable results to all patients who present for surgical evaluation. Specifically, are the patients studied representative of all patients throughout the country? Could there be an unintended selection bias as a feature of the recruitment process? We may not be able to answer this question definitively at this point, however the data produced in this study is sufficiently compelling, in our view, to begin considering how results of genetic analyses could be used to inform discussions with patients’ families. These findings provide a framework for further investigations aimed at incorporating precision medicine in craniofacial surgery, and identify several genes that merit further investigation into their role in both neurodevelopment and cranial suture morphogenesis.

Supplementary Material

Supplemental Digital Content 1

Supplemental Digital Content 1. See Table, which shows Pathway analysis of genes mutated in NSC probands. Genes with de novo mutations were assessed for enrichment in gene ontology terms. Pathways demonstrating significant enrichment include several processes implicated in neurodevelopment.

SDC 3

Supplemental Digital Content 3. See Table, which shows the dataset for figure 3.

SDC 2

Supplemental Digital Content 2. See Table, which shows the dataset for figure 2.