Clinically relevant variants in a large cohort of Indian patients with Marfan syndrome and related disorders identified by next-generation sequencing

Marfan syndrome and related disorders are a group of heritable connective tissue disorders and share many clinical features that involve cardiovascular, skeletal, craniofacial, ocular, and cutaneous abnormalities. The majority of affected individuals have aortopathies associated with early mortality and morbidity. Implementation of targeted gene panel next-generation sequencing in these individuals is a powerful tool to obtain a genetic diagnosis. Here, we report on clinical and genetic spectrum of 53 families from India with a total of 83 patients who had a clinical diagnosis suggestive of Marfan syndrome or related disorders. We obtained a molecular diagnosis in 45/53 (85%) index patients, in which 36/53 (68%) had rare variants in FBN1 (Marfan syndrome; 63 patients in total), seven (13.3%) in TGFBR1/TGFBR2 (Loeys–Dietz syndrome; nine patients in total) and two patients (3.7%) in SKI (Shprintzen–Goldberg syndrome). 21 of 41 rare variants (51.2%) were novel. We did not detect a disease-associated variant in 8 (15%) index patients, and none of them met the Ghent Marfan diagnostic criteria. We found the homozygous FBN1 variant p.(Arg954His) in a boy with typical features of Marfan syndrome. Our study is the first reporting on the spectrum of variants in FBN1, TGFBR1, TGFBR2, and SKI in Indian individuals.


Results
We recruited 83 patients from 53 families with MFS, aortopathy or a related HCTD. The ages of the patients ranged from 3 months to 56 years with a median age of 14 years. The majority were males (51, 61.5%; CI 95% 51-71) and children and adolescents (53, 64% were less than 18 years of age; CI 95% 53-73). Twenty-one families (39% of 53 families; CI 95% 28-53) had more than one affected individual, including a set of monozygotic twins. Echocardiographic information was available for 77/83 individuals that included all index patients. Ophthalmological, skeletal and other information were available for 72/83 individuals.
In a 5-years-8-months-old male (patient 10) we identified the homozygous pathogenic FBN1 missense variant c.2861G > A/p.(Arg954His) (Supplementary Table S1). He is the first child of a third degree consanguineous couple. His measurements were: weight of 20 kg (− 0.30 z), height of 116.5 cm (0.01 z) and head circumference Table 1. Number of affected family members, in silico pathogenicity predictions and ACMG classification for novel variants found in the cohort. The functional impact of the identified variants was predicted by the Combined Annotation Dependent Depletion (CADD) tool, the Rare Exome Variant Ensemble Learner (REVEL) scoring system, and the Mendelian Clinically Applicable Pathogenicity (M-CAP) Score. CADD is a framework that integrates multiple annotations in one metric by contrasting variants that survived natural selection with simulated mutations. Reported CADD scores are phred-like rank scores based on the rank of that variant's score among all possible single nucleotide variants of hg19, with 10 corresponding to the top 10%, 20 at the top 1%, and 30 at the top 0.1%. The larger the score the more likely the variant has deleterious effects; the score range observed here is strongly supportive of pathogenicity, with all observed variants ranking above ~ 99% of all variants in a typical genome and scoring similarly to variants reported in ClinVar as pathogenic (~ 85% of which score > 15) 61 . REVEL is an ensemble method predicting the pathogenicity of missense variants with a strength for distinguishing pathogenic from rare neutral variants with a score ranging from 0 to 1. The higher the score the more likely the variant is pathogenic 62 . M-CAP is a classifier for rare missense variants in the human genome, which combines previous pathogenicity scores (including SIFT, Polyphen-2, and CADD), amino acid conservation features and computed scores trained on mutations linked to Mendelian diseases. The recommended pathogenicity threshold is > 0.025 63 .
Splice site prediction scores were calculated for wild-type and mutated sequences by using the programs Human Splicing Finder 3.1, NetGene2, and the Berkeley Drosophila Genome Project Database [65][66][67][68] . Genetic tolerance at the affected amino acid position in the protein was predicted by MetaDome 64 . www.nature.com/scientificreports/ of 49.5 cm (− 1.94 z). He had a long face, exotropia of the right eye, thin vermilion of the upper lip, high arched palate, bilateral lens subluxation, pes planus, mild distal joint laxity, bicuspid aortic valve, tricuspid and mitral valve prolapse and aortic sinus z-score of 2.86 (Fig. 2). Limited clinical information could be gathered via a video consultation and we specifically noted absence of breathlessness, visual problems, skin striae and chest deformity in parents. Mother however had features suggestive of Leri-Weill dyschondrosteosis (short stature, Madelung deformity, with similarly affected females and mildly affected males in the family).

Loeys-Dietz syndrome (LDS).
Nine patients from seven families (7/53, 13%; CI 95% 7-25) tested positive for a TGFBR1 or TGFBR2 rare variant and clinically presented with LDS. Phenotypic features are summarized in Table 2. Cardiac manifestation was observed in all individuals with aortic root dilatation in six individuals (6/8, 75%; CI 95% 41-93). One proband had a dissection of the aorta at 34 years of age. Typical facial features such as long and narrow face, hypertelorism, downslanted palpebral fissures and micro/retrognathia were seen in all eight patients (Table 2, Fig. 3). Three (37%; CI 95% 14-69) individuals had cleft palate/bifid uvula. Additionally, we noted developmental delay or motor delay (four patients), craniosysostosis (two patients), atopic dermatitis and anemia (in monozygotic twins), platybasia with basilar invagination and atlantoaxial subluxation with retroflexion (one individual) and joint dislocation (one individual) in patients with LDS.  (Fig. 4a,b). He also had pectus carinatum, kyphoscoliosis, long and narrow fingers, decreased palmar creases, long and narrow feet, camptodactyly of fingers and toes, pes planus, metatarsus adductus, recurrent or incisional hernia and decreased muscle mass (Fig. 4c,d). Echocardiography revealed myxomatous prolapsing atrioventricular valves with tricuspid and mitral regurgitation and aortic root dilatation (z-score 6.7). Computed tomography of skull showed calvarial thickening in fronto-parietal bones with partially fused coronal and sagittal sutures. Anterior displacement of the atlas from the occipital condyle and atlanto-occiptal assimilation, mild levoscoliosis of the cerivothoracic vertebra and mild dextroscoliosis of the thoracic vertebra (T9) were observed on computed tomography of the spine. The second patient (patient 38), at 3 months of age, weighed 4.3 kg (− 2.29 z), had a length of 61 cm (− 0.29 z) and head circumference of 39.5 cm (− 1.58 z). His mother had gestational diabetes mellitus. He had wide anterior fontanel, long ears, hairy pinnae, short and downslanted palpebral fissures, hypertelorism, depressed nasal bridge, inverted V-shaped upper lip with thin vermilion, high arched palate, bifid uvula, excess scalp skin, long and narrow fingers, bilateral talipes equinovarus and Mongolian spots (Fig. 4e-h). Mild pectus excavatum, skin laxity, cutis marmorata and umbilical hernia were also noted in him. Echocardiography was normal with z-score of 0.2 for the aortic root size.  Table 3. None of them met the revised Ghent criteria. However, cardiac abnormalities were noted in all except one, and systemic score ≥ 7 was observed in two individuals. About half of them (4/9) fit criteria of the MASS (Mitral valve, Aorta, Skin, and Skeletal features) phenotype (MIM#604308), followed by Mitral valve prolapse syndrome (MIM%157700; #607829; %610840) (2/9), aortopathy (2/9) and an individual with Marfan-like disorder.

Discussion
We describe the clinical spectrum and genetic findings in 83 individuals from 53 Indian families with MFS, LDS and SGS. This is so far the largest cohort of Indian patients with a definitive molecular diagnosis for an aortopathy in a total of 45 index patients. The identification of clinically significant variants in MFS and related disorders reduces the uncertainty in diagnosis in individuals with a suspected diagnosis and guides appropriate www.nature.com/scientificreports/ www.nature.com/scientificreports/ management of their cardiovascular and ocular complications. Our study also provides the mutation spectrum in Indian patients with these three types of HCTDs and adds 21 novel rare variants. We could not find any published reports on Indian patients with LDS or SGS with a molecular diagnosis. Our report now adds seven patients with LDS, two novel TGFBR2 variants, two patients with SGS and a novel disease-causing variant in the R-SMAD binding domain of SKI to the literature. We did not note any unusual clinical features in our small cohort of individuals with LDS and SGS.
Previously, only two publications have reported pathogenic variants in FBN1 in individuals from India 28,29 . One reported a fetus with arthrogryposis, multiple joint dislocations, scoliosis and facial dysmorphism who carried the variant p. Several large cohorts on MFS and related disorders have been published previously with patients originating from European countries or China 23,30-36 . They report a definitive molecular diagnosis in 40-95% of individuals, depending on the inclusion criteria and the testing strategy. We obtained a molecular diagnosis in 45/53 (85%; CI 95% 73-92) families by NGS, including targeted multiple gene panel and whole-exome sequencing, and multiplex ligation-dependent probe amplification.
In our cohort, 36 of the 45 (80.0%; CI 95% 66-89) patients carried a rare variant in FBN1, and the majority of them were missense (21, 58.3%; CI 95% 42-73) in concordance with the literature 34,37 , and 13 substituted or introduced a cysteine residue. Also, we observed splicing variants and multi-exon deletions (five each or 13.9% each; CI 95% 6-29 each), small deletion/insertion (three, 8.3%; CI 95% 3-22) and nonsense (two, 5.6%; CI 95% 2-18) variants in FBN1. In the literature only ~ 5% of probands with an FBN1 pathogenic variant have been reported to carry a deletion or duplication 2,33 , which is 2.8-fold lower in our relatively small cohort of Indian patients. We noted missense variants in four (8.9%; CI 95% 4-21) patients in TGFBR1, three (6.7%; CI 95% 2-18) in TGFBR2, and two (4.4%; CI 95% 1-15) in SKI. Overall, there were 41 rare variants in four genes, with four variants identified in more than one family indicating the private nature of the remaining variants 37   www.nature.com/scientificreports/ Disease-causing variants in exon 24-32 have been associated with early-onset and rapidly progressive MFS 2,30,38 . Seven (7/36) index patients of our cohort have pathogenic variants in this region of the FBN1 gene. Their age at diagnosis ranged from 4 months to 7 years except one (23-years-old). Four of them had de novo variants whereas three were familial. We also report on a proband (patient 11) and his paternal half-sister with the FBN1 nonsense variant c.3012C > A/p.(Tyr1004*). The father did not carry this variant in leukocyte-derived DNA indicating germline mosaicism in him. Five of the seven (71%; CI 95% 36-92) had ocular, cardiac and skeletal manifestations. We also had a 4-months-old infant with early onset MFS in our cohort. She had atrioventricular valve prolapse with severe mitral regurgitation, ostium secundum type of atrial septal defect measuring 11 mm, dilated chambers of heart with dilated aortic root, scoliosis, skin laxity and long and narrow fingers. She succumbed to cardiac failure at 6 months of age.
Bi-allelic FBN1 variants have been reported in 16 families (eight with homozygous and eight with compound heterozygous variants) with MFS [39][40][41][42][43][44][45] . We also document a patient (patient 10) with the homozygous FBN1 missense variant c.2861G > A/p.(Arg954His). Similar to the present individual, all the families reported in the literature with homozygous variants were consanguineous, except one (7/8, 87.5%; CI 95% 53-98). The initially described seven patients with bi-allelic variants had a severe clinical course with early age of onset ranging from day 7 to 22 years [39][40][41][42]44,45 . However, Arnaud et al. reported nine families with bi-allelic variants in FBN1 with classical and mild clinical features with age at diagnosis ranging from 8 to 53 years 43 . In the 16 reported families with bi-allelic FBN1 variants 17 missense variants, two frameshift and one nonsense variant have been identified 43 . Together with the p.(Arg954His) variant detected in patient 10 reported here, the vast majority of bi-allelic variants represent amino acid substitutions (18/21; 85.7%; CI 95% 65-95). Patient 10 at 5-years-8-months presented with typical facial features, bilateral ectopia lentis, bicuspid aortic valve with z-score of ≥ 2 and atrioventricular valve prolapse, the classical form of MFS. Although we were unable to perform a detailed clinical examination of patient 10's parents who are heterozygous carriers of the p.(Arg954His) variant, the same heterozygous variant has been previously reported in a 58-year-old female with skeletal features, ectopia lentis but no cardiovascular abnormalities 27 . In the gnomAD browser, the variant was listed in 1 out of 251,154 alleles. Heterozygous carriers of the p.(Arg2726Trp) variant, who have a second pathogenic FBN1 variant on the other allele in three families, only had isolated skeletal features typical of MFS and/or high stature 43 . In addition, incomplete penetrance has been reported for individuals carrying the p.(Arg2726Trp) variant in the heterozygous state 46 , which is in line with a worldwide minor allele frequency (MAF) of 0.067% for this FBN1 variant (gnomAD browser). Interestingly, a worldwide MAF of 0.02% and 0.12% (gnomAD browser) for the FBN1 alterations p.(Pro1424Ala) and p.(Ala986Thr) 43 , respectively, also suggests incomplete penetrance in individuals carrying either of the variants in the heterozygous state and full penetrance in individuals with one of the two aforementioned FBN1 variants   www.nature.com/scientificreports/ in trans with a second pathogenic variant. Although further studies are needed to study the effect of recessive FBN1 missense variants on fibrillin-1 function, several of the 18 missense variants identified in a homozygous or compound heterozygous state may act as hypomorphic alleles. Eight families (15%; CI 95% 8-27) did not have a clinically significant variant in genes known to cause MFS or associated with HCTD (62 genes on NGS panel), and similar observations were reported in the literature 33,34 . Targeted panel NGS testing has considerable limitations in the detection of single-and multi-exon deletions/ duplications and structural variants as well as non-coding and regulatory variants. Thus, clinically relevant variants might have been missed in one or several of the eight index patients. None of the eight patients met the revised Ghent criteria. The majority of the negative patients have atrio-ventricular valve prolapse with regurgitation. We observed poor scholastic performance (P47)/developmental delay (P52), ectopic and horseshoe kidney with polycystic ovaries (P50) and microtia and pre-auricular tag (P53) in some of them. Whole-exome or whole-genome sequencing will be performed in the eight families to identify the genetic cause underlying the disease in the index patients.
In conclusion, we describe the first and largest cohort of patients with MFS or related disorders from India and provide a base for further genetic testing in this large population. About half of them harbored a novel variant, which has expanded the mutation spectrum of these disorders. Biallelic Patient cohort and data collection. We recruited pediatric, adolescent and adult patients referred for genetics counseling at Kasturba Hospital, Manipal, India and Narayana Hrudayalaya Hospitals, Bangalore, India with features suggestive of MFS, aortopathy or related HCTDs over a period of 5 years. Clinical data and samples for all individuals were obtained with informed consent of patients' parents/legal guardians or the patients themselves, including written consent to use photographs in this report. Clinical data that included a three-generation pedigree and family history of similarly affected individuals (specifically for the presence of tall stature, ocular abnormalities or visual defects and cardiac surgeries) were noted. We performed physical examination and recorded anthropometry for all patients. We collected echocardiographic information and calculated z-score for the aortic root measurements. Ophthalmological evaluation comprised a slit-lamp examination. We performed radiographic assessment and other imaging whenever necessary. Revised Ghent criteria was used for the diagnosis of MFS 47,48 . We collected two millilitres of blood samples from patients and their available family members including parents and siblings for genomic DNA isolation.
The lower and upper limits of the 95% confidence interval (CI 95%) for a proportion were calculated with the VassarStats tool (http://vassa rstat s.net/index .html) according to the method previously described 49 . 2)) related to syndromic and non-syndromic forms of aortopathies and connective tissue disorders. Enrichment of the regions of interest (ROI) was performed with the Illumina Rapid Capture Custom Enrichment kit or the Illumina Nextera Flex for Enrichment kit according to the manufacturer's instructions. Briefly, following fragmentation of genomic DNA, fragmented DNA was amplified and patient-specific (index) adapters were added by PCR. Samples from 12 patients were combined into one single hybridization mix containing target-specific capture probes. The DNA-probe hybrids were then captured with streptavidin beads, and non-targeted DNA fragments as well as unspecific binding were removed by heated washes. Next, the captured DNA library was eluted from the beads, purified and amplified by PCR. For generation of clusters and subsequent sequencing of the targeted DNA samples on a flow cell, a sequencing reagent kit from Illumina was used. High-throughput NGS data were generated on an Illumina sequencing platform 26 .
Whole-exome sequencing (WES) in patients 4 and 16 was performed either with Nextera Rapid Capture Exomes (Illumina) or Agilent SureSelect V6 (Agilent Technologies) kit. Massively parallel sequencing was done on an Illumina NextSeq Platform. There was an average coverage depth of 110×, with ~ 94% of bases covered at > 20× and the data was analysed using an in-house pipeline based on Burrows-Wheeler Aligner (v0.7.15) 50 and Genome Analysis Toolkit Best Practices pipeline (v3.6) 51 . We used ANNOVAR to annotate the variant call format (vcf) files 52,53 . We integrated annotated data with phenotypes catalogued in Online Mendelian Inheritance in Man, human phenotype ontology (HPO) terms, and allele frequency details from in-house variant database of 870 exomes of Indians. Rare variants were retrieved with minor allele frequency of < 1% in population databases [Exome Aggregation Consortium (ExAC) and gnomAD 54,55 ] and our in-house data. Variants were prioritized for the phenotypes 56 .
Sanger sequencing was performed for validation of pathogenic, likely pathogenic sequence variants and VUS identified by NGS and for regions of interest covered by less than 20 reads. Segregation analysis of pathogenic and likely pathogenic variants in affected and/or healthy family members of the index patient was performed by Sanger sequencing using an automated capillary DNA sequencer (ABI 3500; Applied Biosystems). Sequence electropherograms were analysed using the Sequence Pilot module SeqPatient software (JSI Medical Systems).
All novel variants were deposited in the LOVD Database, where they are available under the DB-ID numbers 0000667876 to 0000667897, 0000708485 and 0000708486.

Data availability
All data generated or analysed during this study are included in this published article (and its "Supplementary Information File").