1 INTRODUCTION

Costello syndrome (CS) is a rare, autosomal dominant multisystem disorder caused by pathogenic variants in HRAS (Gripp et al., 2019). CS belongs to a group of conditions called RASopathies, which are caused by pathogenic variants in genes of the RAS/MAPK pathway. Currently, there are more than 20 known genes in which pathogenic variants can cause RASopathy syndromes including Noonan syndrome (NS), cardiofaciocutaneous syndrome (CFCS), Noonan syndrome with multiple lentigines (NSML), and neurofibromatosis (NF1) (Tajan et al., 2018). Hyperactivation of the RAS pathway has been well established as the mechanism of pathogenicity in RASopathy phenotypes including CS (Tidyman & Rauen, 2016). Individuals with CS have craniofacial characteristics similar to individuals with other RASopathies, including down-slanting palpebral fissures, ptosis, coarse facial features, macrocephaly, and low-set and posteriorly rotated ears (Cao et al., 2017).

Craniosynostosis, the premature fusion of cranial sutures which often leads to abnormal head shape, is associated with a monogenic cause in approximately 25% of cases. Pathogenic, activating variants in genes of the fibroblast growth receptor (FGFR) pathway are the major cause of syndromic craniosynostoses, with variants in FGFR2 and FGFR3 causing almost half of these cases (Cunningham et al., 2007; Wilkie et al., 2017). However, craniosynostosis has recently been recognized to be associated with multiple RASopathies including CFCS, NS, NSML, and Noonan-like with loose anagen hair (Bertola et al., 2017; McDonald et al., 2018; Takenouchi et al., 2014; Ueda et al., 2017). The etiology of RASopathy-associated craniosynostosis is hypothesized to be related to the intersection of the RAS/MAPK and FGFR signaling pathways (Addissie et al., 2015; Takenouchi et al., 2014). FGF receptor signaling occurs upstream of RAS/MAPK, and pharmacologic and genetic inhibition of RAS/MAPK signaling inhibits premature suture fusion in two mouse models of FGFR craniosynostosis syndromes (Eswarakumar et al., 2006; Shukla et al., 2007). Therefore, the mechanism for RASopathy associated craniosynostosis is presumed to be secondary to dysregulated RAS signaling. Despite genetic and phenotypic heterogeneity of RASopathies, they are all associated with activation of the RAS pathway. We describe five individuals with CS and craniosynostosis and summarize previously reported individuals with RASopathies and craniosynostosis. We propose that all RASopathies except NF1 can be associated with craniosynostosis, and clinicians should be aware of this possibility when evaluating children with established RASopathy diagnoses, or children presenting with concern for craniosynostosis.

2 METHODS

Individuals 1 and 2 were identified during the course of their clinical care at Cincinnati Children's Hospital, and parents signed consent for publication. Individuals 2–5 are included in an ongoing study of the natural history of Costello syndrome (KWG, Nemours IRB 2005-051). Parents/guardians signed consent for use of photographs for publication. Genetic testing and imaging studies were reviewed retrospectively; all were performed as part of clinical care.

3 CLINICAL REPORT

3.1 Case 1

Individual 1 is a 6-year-old white male born at 34 weeks gestation to a 31-year-old G3P2 mother after a pregnancy complicated by polyhydramnios. Birthweight was 2.27 kg (50th centile). He required neonatal intensive care unit (NICU) admission for 45 days to manage respiratory and feeding problems. During this time, he had multiple episodes of apnea and cyanosis, ultimately requiring intubation. At age 2 months, he had tracheostomy and gastrostomy tubes placed. He became ventilator dependent at 16 months. He was first evaluated at our institution at age 23 months due to a longstanding history of apnea and ventilator dependence. Abnormal head shape was noted, and head computed tomography (CT) scan demonstrated sagittal craniosynostosis (Figure 1a, b) and enlarged lateral and third ventricles. Brain magnetic resonance imaging (MRI) scan confirmed ventriculomegaly and the presence of a Chiari I anomaly with severe crowding (Figure 1c). Of note, Chiari I and ventriculomegaly were not present at the initial brain MRI at 2 months of age. While awaiting results of genetic testing for RASopathies, he was diagnosed with a neuroblastoma. This was identified as a retroperitoneal mass on high-resolution chest CT performed at 24 months due to ventilator dependence. Tumor testing identified HRAS c.34G>A (p.Gly12Ser), which was then confirmed to be present in blood. A ventriculoperitoneal shunt was placed at age 2.5 years to normalize his intracranial pressure, followed by surgical repair of the craniosynostosis in two stages: bicoronal craniotomy for cranial vault remodeling and expansion at 34 months of age and bicoronal craniotomy with fronto-orbital remodeling at 41 months of age. At 6 years of age, he remains tracheostomy dependent with ventilator used during sleep.

4 DISCUSSION

We report for the first time that CS, like other RASopathies, can be associated with multi-suture craniosynostosis. The majority of reported individuals with RASopathy and craniosynostosis have had pathogenic variants in KRAS (9), followed by BRAF (7), PTPN11 (5), and now HRAS (4) (Brasil et al., 2012; Kratz et al., 2009; Schubbert et al., 2006) (Table 1). Unlike CFCS, NS, and NSML, CS is uniquely caused by pathogenic variants in a single gene, HRAS. Four of the five new individuals we report have the common Gly12Ser variant in HRAS, while one has the rare Gln22Lys variant. Given that p.Gly12Ser is the most common variant identified in CS individuals (found in 80%), and it is present in 80% of these five with CS and craniosynostosis, there is no evidence at this time that this specific HRAS protein variant confers increased risk for craniosynostosis.

There have been 24 previously reported individuals with craniosynostosis and RASopathy (Table 1). Overall, characteristics of the combined cohort (n = 29, 24 NS/CFCS and five CS) are similar to other cohorts of individuals with craniosynostosis. Isolated sagittal synostosis is the most common type (16/29, 55%), and there is a predominance of males affected (18/29 male). This is similar to general trends in craniosynostosis: Sagittal is the most common suture affected (in 40%–60%), and is more frequently seen in males than females (Proctor & Meara, 2019). Age of diagnosis of craniosynostosis and treatment description were available for 22 of the 24 previously reported patients and all five of the patients with CS. The average age of diagnosis of craniosynostosis was 3.8 years (range 2 weeks–12 years). Fifteen of 27 were treated with surgery, and the average age of diagnosis of surgically treated individuals was 2.5 years, versus 6 years for those who did not require surgery.

In addition to craniosynostosis and CS, the individuals we describe all have a Chiari I and two have shunted hydrocephalus. Individuals with CS are known to have an increased incidence of Chiari I with potential for post-natal progression, presumably due to cerebellar overgrowth, as well as increased incidence of hydrocephalus (Calandrelli et al., 2015; Gripp et al., 2010). In one series of 28 individuals with CS, 25% had hydrocephalus requiring shunting or ventriculostomy, and 32% had a Chiari I (Gripp et al., 2010). Nonsyndromic craniosynostosis is also associated with Chiari I and shunted hydrocephalus: Davis et al. (2019) reviewed 400 patients with isolated sagittal synostosis and noted that Chiari I was more commonly found in patients diagnosed with synostosis after age 1 than in those diagnosed prior to age 1 (9% vs. 1.2%) (Davis et al., 2019). Furthermore, patients diagnosed with craniosynostosis after age 1 who required operative repair more often had a Chiari I than patients diagnosed prior to age 1 (22% vs. 2%) (Davis et al., 2019). Four of the five individuals in our series were diagnosed with craniosynostosis after the age of 1 and required surgical treatment. Davis et al. (2019) also noted that 23 of the 400 patients with isolated sagittal synostosis had synostosis that was attributed to shunting. In our series, individuals 1 and 4 both required shunting for hydrocephalus. Individual 1 was diagnosed with hydrocephalus and craniosynostosis on the same scan, prior to shunt placement, therefore, the synostosis cannot be attributed to shunting. Individual 4 was diagnosed with Chiari I and hydrocephalus and shunt placed prior to craniosynostosis diagnosis therefore shunt-associated synostosis cannot be excluded.

Ventriculomegaly, Chiari I, and macrocephaly are also known features of NS and CFCS, but less commonly reported than in CS (Gripp et al., 2010; Gripp et al., 2016; Reinker et al., 2011; Yoon et al., 2007; Zarate et al., 2014). Interestingly, only one of the previously reported individuals with RASopathy and craniosynostosis had a Chiari I and ventriculomegaly, and two others had ventriculomegaly (Kratz et al., 2009; Ueda et al., 2017). None are reported to have required a shunt. However, the simultaneous association between CS and Chiari, Chiari and craniosynostosis, and now CS and craniosynostosis raises a question about the relationship between Chiari I and craniosynostosis in individuals with CS: is one of these the primary phenotype that leads to the other, or do they occur independently?

5 CONCLUSION

We report five individuals with CS and craniosynostosis and conclude that CS, CFCS, and NS may all be associated with this diagnosis. Abnormal head shape or growth in children with CS, CFCS, or NS should prompt evaluation for craniosynostosis. Further work is needed to determine if shunting of ventriculomegaly in children with CS is associated with an increased risk for shunt-associated synostosis.

AUTHOR CONTRIBUTIONS

K. Nicole Weaver and Carlos E. Prada conceived the study. K. Nicole Weaver wrote the manuscript with input from all authors. K. Nicole Weaver, Emily Wakefield, Jesse Skoch, Yuri A. Zarate, Karen W. Gripp, and Carlos E. Prada collected and reviewed clinical information. Marguerite Care reviewed and interpreted imaging studies for all patients. All authors provided critical feedback on the data, conclusions, and manuscript text.