Molecular and phenotypic spectrum of cardio-facio-cutaneous syndrome in Chinese patients

Background Cardio-facio-cutaneous (CFC) syndrome is a RASopathy subtype that presents with unique craniofacial dysmorphology, congenital heart disease, dermatologic abnormalities, growth retardation, and intellectual disability. This study describes the phenotypic spectrum of CFC in China and its association with CFC syndrome gene variants. Results Twenty Chinese CFC patients, aged 0.6–9.5 years old, were included in this study and their clinical phenotypic spectrum was compared with that of 186 patients with CFC from non-Chinese ethnicities. All 20 Chinese patients with CFC carried de novo heterozygous BRAF, MAP2K1, and MAP2K2 variants. Two novel variants were detected and consistently predicted to be deleterious using bioinformatic tools. The clinical features of CFC in the Chinese patients included hypertrophic cardiomyopathy (2/20, 10%), pulmonary valve stenosis (2/20, 10%), curly or sparse hair (7/20, 35%), epilepsy (1/20, 5%), and hypotonia (10/20, 50%); these features were less frequently observed in Chinese patients than non-Chinese patients (p < 0.05). In contrast, feeding difficulties (19/20, 95%) were more frequently observed in the Chinese patients. Absent eyebrows and severe short stature were more common in patients with BRAF variants than in those with MAP2K1/2 variants. Facial recognition software was used to recognize most CFC patients using artificial intelligence. Conclusion This study identified novel and common variants in our cohort of 20 Chinese patients with CFC. We uncovered differences in clinical features between Chinese and non-Chinese patients and detected genotype–phenotype correlations among the BRAF and MAP2K1/2 variant subgroups. This is the largest cohort of Chinese CFC patients to our knowledge, providing new insights into a subtype of RASopathy. Supplementary Information The online version contains supplementary material available at 10.1186/s13023-023-02878-0.


Background
Cardio-facio-cutaneous syndrome (CFC; MIM#115150) was first reported by Reynolds et al. in 1986 [1].CFC is an autosomal dominant disorder characterized by dysmorphic craniofacial features, congenital heart disease, dermatologic abnormalities, failure to thrive, gastrointestinal dysfunction, neurocognitive delay, and epilepsy [1].The prevalence of CFC in Japan is estimated at 1 in 810,000 individuals.At present, there are few case reports of CFC patients in China; large-scale cohort studies of Chinese patients are still lacking, making it difficult to summarize the common features of CFC in China.
The diagnosis of CFC is based on clinical suspicion and genetic testing, including next-generation sequencing (NGS) [1].The CFC index, as a clinical and objective method, was first proposed in 2002 to identify and differentiate CFC from other phenotypically similar genetic syndromes; however, it is not widely used [6].In 2014, an international group of experts developed international consensus diagnostic criteria for CFC pediatricians and other care providers.Artificial intelligence (AI) tools, such as DeepGestalt (Face2Gene, FDNA Inc., Boston, MA), have improved facial recognition of genetic syndromes and can achieve > 90% top-10 diagnostic accuracy in correctly identifying a genetic syndrome [7].
Here, we report the clinical and genetic spectrum of 20 Chinese patients with CFC, which, to our knowledge, is the largest CFC cohort studied in China.Furthermore, genotype-phenotype correlations were analyzed, and the phenotypes of CFC patients in China were compared to those of patients in Canada.Finally, to explore new diagnostic tools, we used AI tools to improve the diagnostic rate.

Clinical data
Twenty patients aged from 0.6 to 9.5 years old (11 male and 9 female) genetically diagnosed with CFC were included in this study.One patient was previously reported in 2019 due to cardiovascular traits, while the other 19 patients have not been reported [8].Facial pictures are shown in Fig. 1.
Phenotypic information is summarized in Table 1 and covered multisystem abnormalities, including craniofacial, dermatological, cardiovascular, skeletal, neurological/developmental, endocrine, gastrointestinal, tumorous, and prenatal manifestations.Outstanding craniofacial features are wide nasal base/anteverted nares, hypoplasia of the eyebrows, macrocephaly, epicanthus and low-set ears, which were found in more than half of the patients.Two-thirds of the patients have a history of congenital heart disease (13/20, 65%), including atrial septal defect, patent foramen ovale, hypertrophic cardiomyopathy, and pulmonary valve stenosis.The most prominent dermatological abnormalities were melanocytic nevi (60%; 12/20), followed by abnormal nail growth, hemangioma, keratosis pilaris, and eczema.Short stature was observed in 70% (14/20) of patients and the average standard deviation (SD) for height was − 3.0 SD, using standard Chinese growth curves.All patients exhibited delayed motor development and intellectual disability.A total of 65% (13/20) of patients suffered from sleep disturbance, with brain MRI abnormalities present in 55% (11/20) of patients, while only one patient had a history of epilepsy.Perinatal findings were polyhydramnios, preterm birth, and increased nuchal translucency.Other clinical findings included pectus excavatum/carinatum, scoliosis, strabismus, hearing impairment, and cryptorchidism.Only one patient had a tumor, which was an inflammatory myeloid neoplasm called Langerhans cell histocytosis.Four patients in our study were treated with growth hormone (GH).One of the patients was GH deficient (patient 4) and two of them (patient 4, patient 6) achieved a satisfactory result in height.None of them suffered side effects following GH treatment.

Comparison with published literature
We used statistical tools to analyze the differences in phenotypes between our 20 Chinese patients and a cohort of 186 non-Chinese patients (Table 2).The results indicated that pulmonary valve stenosis, curly/sparse hair, and epilepsy were less likely to occur in Chinese CFC patients with BRAF variants, whereas curly/sparse hair and hypotonia were rare in Chinese patients with MAP2K1/2 variants.Conversely, feeding difficulties were more prominent in the Chinese patients than the non-Chinese patients.It is worth noting that epilepsy may be underestimated because it can also occur in childhood, late adolescence and even early adulthood [9].

Spectrum of variants in Chinese CFC patients
A schematic diagram of the variant distribution in this study is shown in Fig. 2. All variants were classified as either likely pathogenic or pathogenic following the American College of Medical Genetics and Genomics (ACMG) guidelines, except the p.Asp140His variant in the MAP2K2 gene (patient 9), which was classified as having uncertain significance.Two novel variants were identified in this study: one de novo variant (p.Leu54Pro) in MAP2K1 and one de novo variant (p.Asp140His) in MAP2K2.Both variants were located in the functional domain of the protein.All variants, including the two novel variants, were consistently predicted to be deleterious using three bioinformatics tools (Table 3).

Genotype-phenotype correlations in CFC
Genotype-phenotype analysis was performed in CFC patients with variants in the BRAF or MAP2K1/2 genes (Fig. 3).Individuals with BRAF variants exhibited a significantly higher rate of hypoplasia of eyebrows (100%) than those with variants in MAP2K1/2 genes (56%; p = 0.03).Severe short stature occurred more frequently in patients with BRAF variants, although the difference was not statistically significant.Furthermore, eczema (p = 0.07), hearing impairment (p = 0.07), cryptorchidism (p = 0.13), and pulmonary valve stenosis (p = 0.19) were commonly found in patients with MAP2K1/2 variants, while macrocephaly (p = 0.13) was often observed in patients with BRAF variants.

Facial gestalt analysis of CFC
Pictures of the 20 Chinese CFC patients were collected and analyzed using Face2Gene CLINIC tools.For the majority of patients (18/20), CFC was listed in the TOP 5 diagnosis list produced by Face2Gene CLINIC tools.A composite photo was computed through Face-2Gene RESEARCH (Fig. 4A).Furthermore, to assess patient facial patterns among different genotypes, we added two additional facial pictures of Chinese patients with MAP2K1 variants [10,11].The receiver operating characteristic (ROC) curves showed that facial patterns were similar in CFC patients with different genotypes, with an area under the curve (AUC) of 0.595 (p = 0.297) (Fig. 4B).This suggests that facial recognition analysis cannot distinguish between different genotypes in patients with CFC.

Discussion
In our study, NGS-based genetic was used to diagnose the majority of patients (19/20).NGS-based genetic testing is a rapid and inexpensive technique that provides a genetic, evidence-based diagnosis for CFC.Additionally, direct measurement of the head and face of CFC patients has been used to objectively enhance diagnostic accuracy [12].In patients with moderate clinical features, artificial intelligence tools (Face2Gene) may help with facial recognition and improve diagnostic accuracy [13]; indeed, it has been reported that the mean accuracy computed by automated image analysis (Face2Gene RESEARCH) can help distinguish genotypes in NS patients [7,13].However, facial images could not differentiate between patients with variants in BRAF or MAP2K1/2 genes in our study.
By comparing the phenotypes of 20 Chinese patients with CFC to those of a cohort of 186 non-Chinese CFC patients, we concluded that the frequencies of pulmonary valve stenosis, curly/sparse hair, epilepsy, and hypotonia were lower in Chinese patients (Table 2).The .81 a popularization of prenatal practice, ethnic differences, and the lifetime risk of epilepsy may explain these discrepancies.In contrast, the Chinese patients in our study more often experienced feeding difficulties, and patient 2 has had to receive prolonged supplemental feeding via nasogastric tube feeding since birth.This is likely because the majority of our patients (16/20) were diagnosed before the age of 3 (Additional file 1: Table S1).Feeding problems, including gastroesophageal reflux, suck/ swallow dysfunction, and oral aversion, are commonly observed in CFC [1].
After analyzing the phenotypic diversity between patients with BRAF or MAP2K1/2 variants, we found that the occurrence rates of hypoplasia of the eyebrows and macrocephaly were higher in patients with BRAF variants, whereas eczema, hearing impairment, cryptorchidism, and pulmonary valve stenosis were more common in patients with MAP2K1/2 variants.Hypoplasia of the eyebrows was found to be of important significance in differentiating between BRAF and MAP2K1/1 mutant patients (p = 0.03).In addition, severe short stature (height SD < − 3.0) was more common in individuals with BRAF variants than in those with MAP2K1/2 variants.Retrospective studies have identified several other genotype-phenotype correlations.For example, the frequency and severity of epilepsies were higher in individuals with BRAF or MAP2K1 variants (especially variants in the catalytic protein kinase domain of BRAF and p.Y130 site of MAP2K1) than in those with MAP2K2 variants [14].BRAF variants are also associated with distinct dermatological manifestations [15].MAP2K1 variants are correlated with skeletal deformities and motor dysfunction, and MAP2K2 and KRAS variants are correlated with mild intellectual disability [1,14].
Precise identification of subgroups among RASopathies has been an area of high interest and a bottleneck problem in previous studies.A recent cohort study found that CFC patients tend to have prominent problems in the nervous system, psychomotor development, and dermatologic manifestations, including severe intellectual disability, intractable epilepsy, spare/curly hair, hypoplasia of the eyebrow, and melanocytic nevi [16].Neurological, dermatologic, and gastrointestinal issues and severe short stature tend to be less common in patients with NS than in those with CS or CFC [17].Outstanding perinatal problems, such as hypoglycemia, arrhythmia, coarse face, intractable feeding difficulty, and a higher risk of tumors, are more common in patients with CS [17].
Short stature is a common feature in all RASopathies, and two-thirds of patients with CFC have short stature.This is likely because the RAS/MAPK pathway contributes to intracellular insulin-like growth factor (IGF-1) signaling, and MAPK activation is important for the synthesis and release of GH [1].Feeding difficulties and gastrointestinal abnormalities can also cause short stature [19].In a recent study, seven patients with CFC underwent GH treatment and height SDS increased by more than 1 SD after two years of GH therapy without side effects [16].Four of the 20 Chinese patients in our study were treated with GH; patient 4 (GH deficiency) and patient 6 achieved a satisfactory result following GH treatment, with a 10-11 cm increase in height during first 6 months.Patient 18 underwent three months of GH therapy with no change in height.GH therapy data for patient 13 were not available.None of the patients experienced side effects from GH treatment, demonstrating that CFC should be screened for GH deficiency firstly and then be assessed the possibility of GH treatment.
RAS is the most frequently mutated oncogene in human cancers, with variants occurring in approximately 30% of all cancers [20].Somatic mutations of BRAF have been associated with various malignancies including malignant melanoma, thyroid, lung, ovarian and colorectal cancers [21].Conversely, germline variants in the BRAF gene rarely cause cancer, except in a few cases of acute lymphoblastic leukemia and non-Hodgkin's lymphoma.For patients carrying both germline and somatic variants in the BRAF gene, regular monitoring of the tumor is necessary [16].In our cohort, patient 12 was diagnosed with Langerhans cell histocytosis when he was 2.4 years old.He was cured after chemotherapy in our hospital, and his tumor indicators were normal during long-term follow-up.Among RASopathies, CS patients were reported to carry the highest risk of malignancy (10.8%), followed by NS patients (3.9%), and CFC patients (3.5%) [21,22].Even so, tumor risk should be monitored, especially when GH is administered.
Our study was limited by the lack of CFC patients with KRAS variants in our cohort and the low number of patients with MAP2K2 variants (2/20), which constrained our assessment of genotype-phenotype correlations.Additionally, the categories of phenotypes selected in our study were not completely consistent with those of the 186 non-Chinese patients; therefore, we only analyzed the phenotypes observed in both

Conclusions
Herein, we reported the molecular and phenotypic spectra of 20 Chinese patients with CFC, including the discovery of two novel variants.The CFC phenotype in Chinese populations overlaps with that in non-Chinese populations, although some phenotypic differences exist.Diversity of phenotype was observed between patients with BRAF and MAP2K1/2 variants, and automated facial image analysis tools (Face2Gene) failed to distinguish CFC patients by genotype.Overall, we conclude that the diagnosis of CFC can be facilitated using typical clinical features and facial recognition software tools and recommend genetic testing using NGS-based approaches for final diagnosis.counseling was provided to all patients and their parents to obtain informed consent.

Patients
We used the human phenotype ontology (HPO, https:// hpo.jax.org/ app/) to label the phenotype of CFC patients with a unique HPO number.The phenotypes we collected from each patient were compared to those of 186 non-Chinese patients with CFC [5].

Molecular genetic analysis
All patients underwent whole exome sequencing, except for patient 6, who was tested for a panel of hereditary liver diseases.Two milliliters of peripheral blood were collected from patients and their parents.Blood samples were stored in ethylenediaminetetraacetic acid anticoagulant tubes for examination.Exons were captured using the Agilent Sure Select method.High-throughput sequencing was performed using the Illumina sequencing platform, sequencing data were matched and analyzed using NextGENe ® software, and variation filtering and interpretation were performed using the Ingenuity online software system.All detected variants were confirmed by Sanger sequencing in both parents.The method recommended by the ACMG was used to classify the detected variants.To determine the potential pathogenicity of the novel missense variants, three in silico prediction methods were used: Mutation Taster (http:// www.mutat ionta ster.org), Sorting Tolerant from Intolerant (SIFT; http:// sift.jcvi.org/), and Polymorphism Phenotyping v2 (Poly-Phen-2; http:// genet ics.bwh.harva rd.edu/ pph2/).

Automated image analysis
An AI approach was used to analyze facial images of patients with CFC.We use Face2Gene (F2G Inc., Boston; http:// www.fdna.com/ face2 gene/) to evaluate the identification accuracy of CFC patients, which analyzes phenotypic traits based on pattern recognition of frontal photographs of the patients.Face2Gene CLINIC tools were used to analyze the diagnosis accuracy of CFC from only facial pictures, and the diversity of facial patterns between different genotypes was computed using RESEARCH tools.

Statistics analysis
The prevalence of each phenotype was analyzed using the chi-squared (χ 2 ) test.Fisher's exact tests were performed to compare the incidences of the various phenotypes between Chinese and non-Chinese patients.Phenotypic differences between the genotypes were also analyzed by chi-squared (χ 2 ) test.p values < 0.05 were considered statistically significant.All analyses were conducted using the statistical software package IBM SPSS Statistics version 26.
• fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year

•
At BMC, research is always in progress.

Learn more biomedcentral.com/submissions
Ready to submit your research Ready to submit your research ?Choose BMC and benefit from: ? Choose BMC and benefit from:

Fig. 2 3
Fig. 2 CFC variants identified in this study.Novel variants are shown with *. (A) The schematic diagram of the distribution of BRAF mutations in this study.(B, C) The schematic diagram of the distribution of MAP2K1/2 mutations in this study

Fig. 3
Fig. 3 Genotype-phenotype correlation of CFC among BRAF and MAP2K1/2 mutation subgroup This study was approved by the ethics committee of the Shanghai Children's Medical Center, Shanghai Jiaotong University School of Medicine (Shanghai, China).Patients were recruited from the Shanghai Children's Medical Center and other domestic third-class hospitals between 2016 and 2022.CFC patients with a genetic diagnosis of MAP2K2 variant were directly included in the study, whereas those who exhibited other RASopathy manifestations with a genetic diagnosis of BRAF, MAP2K1, or KRAS variants were excluded.Facial pictures, medical history, and clinical features were collected through electronic surveys and confirmed by telephone interviews and face-to-face diagnosis.All physicians were pediatricians with additional subspecialty training in endocrinology.Careful

Fig. 4
Fig. 4 Facial analysis of patients.(A) Composite photos computed from images of BRAF and MAP2K1/2 mutation subgroup.(B) Score distribution and ROC curves of the comparison results of BRAF and MAP2K1/2 mutation subgroup

Table 2
Comparison of CFC syndrome phenotype between Chinese patients and non-Chinese patients reported in the literatureA Chinese patients, B non-Chinese patients, ID intellectual disability, ASD atrial septal defect, VSD ventricular septal defect, PVS pulmonary valve stenosis, HCM hypertrophic cardiomyopathy a p value, per chi-square test