Whole-exome screening for primary congenital glaucoma in Lebanon

ABSTRACT Purpose Mutations were previously identified in the CYP1B1 gene in six out of 18 Lebanese families (33%) with primary congenital glaucoma (PCG). The purpose of this study is to determine the frequency and type of pathogenic mutations in other genes and compare to other populations using whole-exome sequencing and perform genotype–phenotype correlations. Methods Twelve PCG patients previously negative for CYP1B1/MYOC mutations were subjected to whole-exome sequencing. Targeted screening for glaucoma-associated genes was performed. Candidate variants were verified by Sanger sequencing and evaluated in family members for segregation analysis and in 100 normal controls. Clinical correlations were established as to severity of disease presentation, course, and visual outcomes. Results Six mutations in known PCG-causing genes were identified in five patients: homozygous mutations in CYP1B1 (p.R368G), LTBP2 (p.E1013G), and TEK (p.T693I), and heterozygous mutations in FOXC1 (p.Q92*), TEK (c.3201-1 G>A), ANGPT1 (p.K186N), and CYP1B1 (p.R368G). Two patients, negative for CYP1B1 in the previous study, were revealed positive in the current study, due to different sets of primers and PCR conditions. Potentially damaging variants were noted in several candidate genes. Except for FOXC1 mutations, all genetic variants described here are novel. Intra-ocular pressure and final optic nerve cup-to-disc ratio were highest in the patient with three mutations in LTBP2/TEK/ANGPT1 genes. Conclusion This study provides new data on the spectrum of mutations of PCG in Lebanon. This highlights the genetic heterogeneity of the Lebanese population, noted for high rates of consanguinity in 50% in this cohort. This study emphasizes the importance of whole-exome sequencing in elucidating new candidate genes for PCG in the Lebanese.


Introduction
Primary congenital glaucoma (PCG) is an inherited disorder with incomplete penetrance, causing a developmental defect in the anterior chamber and trabecular meshwork of the eye (1). This impedes adequate drainage of aqueous humor leading to increased intraocular pressure and optic nerve damage (2). If left untreated, it leads to irreversible visual loss. The disease manifests in early infancy with variable incidence according to geographic region and ethnicity with higher estimates in inbred populations, especially in the Middle East (2). In a study conducted in 2014 at the American University of Beirut Medical Center (AUBMC), 33% of Lebanese families with at least one member affected with PCG had mutations in the CYP1B1 gene (6/18 families included) using Sanger sequencing (2). These mutations were associated with a severe phenotype of the disease. Frequencies of CYP1B1 mutation in PCG reported in Saudi Arabia, Kuwait, and Israel were higher at 72%, 52.9%, and 46%, respectively (3)(4)(5). This led to investigation of mutations in other putative genes, like latent transforming growth factor β-binding protein 2 (LTBP2) and forkhead box C1 (FOXC1) genes. These have been reported in 33.33% (6) and 3.7% (7), respectively, of the Indian PCG population.
Multiple genetic mutations are reported in the literature as pathogenic in PCG (8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20). Studies from Pakistan, Iran, and India report different types of mutations of LTBP2 associated with PCG (8)(9)(10). A case-control study in 2013 failed to uncover any disease-causing mutation in the LTBP2 gene (11). Another report concluded that LTBP2 mutations were not implicated in pathogenesis (6). However, recessive LTBP2 mutations-in spherophakia or ectopia lentis secondary glaucoma-can be related to congenital megalocornea and lens dislocation into the anterior chamber (21). Another study emphasized that recessive mutations in LTBP2 were associated with the phenotypic spectrum of spherophakia, primary megalocornea, and lens-related glaucoma. Although labeled as primary congenital glaucoma, these cases could be secondary to anterior lens movement rather than trabeculodysgenesis (22). FOXC1 gene mutations have been described in anterior segment abnormalities and may affect goniodysgenesis in patients with PCG, and careful re-examination of children with FOXC1 pathogenic variants typically reveals features of Axenfeld-Rieger spectrum (12)(13)(14). This gene was identified as causal for early-onset glaucomas, especially for primary open-angle glaucoma (15). Axenfeld-Rieger syndrome and abnormal development of iris tissue were associated with numerous FOXC1 and pituitary homeobox 2 (PITX2) gene variants (12,16). In addition, studies showed that rare variants in the TEK receptor gene (tunica interna endothelial cell kinase) highly expressed in Schlemm's canal endothelium of the eye, were reported in eight PCG families, supporting a causative role for TEK in PCG (17,18). In 2022, researchers described mutations in CYP1B1, LTBP2, FOXC1 and TEK in 33% of PCG patients in Germany (19). The interaction of TEK and CYP1B1 contributes to PCG pathogenesis, according to another study (20).
This study is the second report on PCG genetics in Lebanon, using whole-exome sequencing to describe the rate and type of genetic mutations encountered in PCG patients that had previously tested negative for CYP1B1 and correlates genotype to phenotype. The ultimate goal is to improve understanding of PCG disease pathology. This will hopefully lead to improved treatment by integrating data from this analysis with similar reports in the literature, hence, drawing conclusions pertaining to early diagnosis, as well as timely and appropriate intervention.

Subjects
This study was a complement to a previous prospective genetic study of patients diagnosed with PCG (and first-degree relatives) conducted at the Ophthalmology Department of AUBMC (2). Investigators secured study protocol approval per the AUBMC Institutional Review Board. informed consent was obtained from each patient/participant (when > or = 18 years old) their parents/legal guardians (<18 years of age). In the original protocol, review of surgical records according to code of surgery (goniotomy/trabeculotomy), since treatment for PCG is essentially surgical, took place. Patients with secondary glaucomas like anterior segment dysgenesis, aniridia, Peters anomaly, Axenfeld-Rieger (or other congenital anomalies of the eye associated with glaucoma), and postoperative glaucoma (aphakic, pseudophakic, post-retina surgery), were excluded. Lebanese patients from 18 families and their first-degree relatives consented to participate. Recruitment included 18 PCG patients (26 eyes). PCG was defined as glaucoma diagnosed by 2 years of age characterized by corneal edema, increased corneal diameter (>12 mm), increased intraocular pressure (>21 mmHg) and optic nerve cupping. Patient demographic information (age at presentation, ethnic group, family history, consanguinity) and parameters pertaining to disease severity (corneal edema, corneal diameter, intraocular pressure, optic nerve cup-to-disc ratio, visual acuity, refraction, gonioscopic abnormalities, number and type of previous glaucoma surgeries and medications) were collected.
In the previous study (2), disease-causing mutations in CYP1B1 gene were detected in 6/18 families. In the current study, the genetic basis of PCG in the remaining 12 families devoid of CYP1B1 gene mutations using whole-exome sequencing was undertaken.

Whole-exome sequencing
Genomic DNA was extracted from peripheral blood using the QIAamp DNA Blood Mini Kit (Qiagen) and 1 ug used for whole-exome sequencing (WES). DNA library construction and WES were outsourced to Macrogen Inc. (Republic of Korea). SureSelect Human All Exon V6 enrichment kit (Agilent Technologies) and NovaSeq 6000 platform (Illumina) were used. Reads were mapped to the human reference genome hg38 using BWA (23), PCR duplicates removed (Picard-tools, http://broadinstitute.github.io/picard) and variant calling and filtering performed using GATK (24). Filtered variants were annotated using SnpEff (25) and in-house programs.

Variant identification
Variant filtering was based on screening of select genes associated with glaucoma and eye development/disease (listed in Supplementary Table S1). Synonymous, non-splicing intronic and non-coding variants were excluded, as were variants with minor allele frequency (MAF) >1% in the 1000 Genomes Project phase 3 (26), ExAC (27) and gnomAD (28) population databases. Variants were evaluated for potential clinical significance using HGMD (29) and ClinVar (30)

Copy number variation detection
To investigate the presence of large deletions or duplications in probands of families 2 and 4, chromosomal microarray analysis used the Applied Biosystems CytoScan HD suite (Thermo Fisher Scientific). Briefly, digestion of 250 ng of gDNA with NspI enzyme occurred prior to ligation and amplification. After fragmentation and labeling, the cocktail was hybridized to CytoScan HD arrays at 50•C overnight in a GeneChip Hybridization Oven 645, washed and stained with GeneChip Fluidics Station 450 and scanned with GeneChip Scanner 3000 7 G prior to analysis of copy number variation (ChAS Software,Thermo Fisher Scientific).

In silico analysis
To assess variant impact on protein function, 10 in silico prediction tools were used: PolyPhen-2 HumVar (32), SIFT (33), LIST-S2 (34), LRT (35), MutationTaster (36), MutationAssessor (37), Provean (38), M-CAP (39), fathmm-MKL (40) and Eigen (41). Investigation of missense variants scoring less than 5/10 pathogenic predictions was omitted. For splicing variants, HSF Pro (42) analysis predicted the impact of mutations on splicing signals. HSF Pro system combines multiple algorithms (MaxEntScan and the HSF) to predict the disruption of these sites or the activation of nearby cryptic sites. Disruption or creation of a splice site is predicted when ∆CV (variation in consensus values between mutant and wild type) is > ± 10%. To assess the evolutionary conservation of residues affected by missense variants, amino acid sequences evaluation with UniProt (43) and Multiz alignments of 100 vertebrates on the UCSC Genome Browser (http://genome.ucsc.edu.) was applied (44). Assessing the variant's impact on protein structure used DynaMut2 (45) and was in accordance with known crystal structures when available. Effect of a mutation on protein stability and flexibility is reported as variation in Gibbs free energy (ΔΔG) (in Kcal/mol). Destabilizing effect is reported with ΔΔG < 0.0 kcal/mol and stabilizing with ΔΔG > 0.0 kcal/mol. 3D-structures of CYP1B1 (PDB code: 3PM0, resolution: 2.7 Å) and TEK (PDB code: 5UTK, resolution: 2.5 Å) were available for affected sequence domains and used to assess potential structural changes.

Genotype-phenotype correlations
Genotype-phenotype correlations linked clinical presentation and disease severity with the specific mutation identified. Parameters for the genotype-phenotype correlations were: corneal edema and diameter, optic nerve cupping, intraocular pressure, number of medications and glaucoma surgeries, and final visual acuity. These reflect severity of disease and final outcome.

Statistical analysis
Analyses of the clinical parameters used the SPSS statistical package, version 19. Means were computed with standard deviations and medians were calculated with ranges.

Clinical characteristics
Eighteen unrelated families with primary congenital glaucoma were originally recruited at AUBMC and underwent screening for mutations in CYP1B1 and MYOC genes by direct Sanger sequencing. Six of 18 families were positive for CYP1B1, and none for MYOC. In the current study, probands of 12 remaining families underwent wholeexome sequencing to investigate the genetic background of the disease. The goal was screening for select genes associated with glaucoma or eye development. Table 1 summarizes patient clinical and molecular data. Parental consanguinity was present in 50% (6/12) of the screened families. Family history was present in 8.3% (1/12) of subjects. Four subjects had bilateral PCG and eight had unilateral disease (five left eyes and three right eyes). Mean follow-up time was 16.4 ± 7.6 years. Gender ratio was 1:1, with a mean age at presentation of 5.4 ± 5.1 months and a median of 3.5 months (age ranging from 0.25 to 18 months). Corneal edema was severe in 3/ 12 (25%).
Proband in Family 1 had consanguineous parents and presented at 1.5 months of age. She underwent three surgeries (trabeculotomy in both eyes, trabeculectomy in her left eye) and required a single medication for good control with a final visual acuity of 20/20 and 20/500 in her right and left eye, respectively, after a follow-up period of 30 years. Patient in Family 2 had unilateral disease with severe corneal edema and a later presentation at 18 months of age. He underwent trabeculotomy, endocyclophotocoagulation, complicated cataract surgery and removal of anterior chamber intraocular lens, with a final IOP of 20 mmHg requiring 3 glaucoma medications. His cup/disc ratio has advanced to 0.9 since presentation. Patient in Family 3 presented at 2 months of age with moderate corneal edema in both eyes, underwent trabeculotomy and Ahmed valve insertion in both eyes, cyclophotocoagulation twice for the right eye and once for his left eye. He subsequently required three glaucoma medications. His final visual acuity was poor with counting fingers for both eyes with a last measured IOP of 26 mm Hg and 30 mm Hg in right and left eye, respectively and a C/D ratio of 0.9. He was the product of a consanguineous union but no family history of PCG. Patient was lost to follow up and no further surgeries occurred.
Proband in Family 4 presented with unilateral PCG and mild corneal edema. She underwent a trabeculectomy with glaucoma controlled off medications. There was no family history of PCG, but parents were consanguineous. Her final visual acuity was 20/100 with a C/D ratio of 0.4.
The patient in Family 5 was a child of non-consanguineous parents but had positive family history for PCG. He presented at a young age with mild corneal edema and underwent early trabeculectomy in both eyes with well-controlled glaucoma with one medication and a final visual acuity of 20/20 in both eyes. His optic nerves were healthy with a C/D ratio of 0.4 in both eyes on his final visit. He did not have any signs of Axenfeld-Rieger spectrum. However, his cousin had posterior embryotoxon and corectopia.

Identification of mutations in PCG causal genes
Average number of SNPs/indels obtained by WES on 12 PCG subjects was 114,537. The current analysis was focused on screening of only 44 genes (listed in Supplementary Table S1), selected for association with glaucoma and/or eye development/disease. After a series of filtering, 11 novel or rare variants (MAF < 0.001) emerged as potentially deleterious based on prediction of up to 10 in silico tools. No variant detection occurred in 200 alleles of healthy Lebanese controls, with absence in Middle Eastern subjects in the gnomAD database. Six variants in known PCG-causing genes were identified in five families (Table 2 and Figure 1).
In Family 1, a homozygous missense mutation NM_000104.3:c.1102C > G in exon 3 of the CYP1B1 gene was detected in the proband and was absent in gnomAD. It results in substitution of the amino acid at position 368 from arginine to glycine (p.R368G). The arginine residue is highly conserved (Figure 2), and its substitution to glycine is predicted to negatively affect protein stability ( Figure 3). Codon 368 is a hot spot in the CYP1B1 gene. The substitution to glycine here is novel. This mutation occurred in a heterozygous state in the proband of Family 2 ( Figure 4). Pathogenic mutation in the second allele of the CYP1B1 gene remains unidentified. Presence of large deletions or duplications was excluded by chromosomal microarray analysis.
In In Family 4, a homozygous missense mutation in exon 13 of TEK (NM_000459.4:c.2078C>T, p.T693I) was detected in the proband. An unaffected sister also carried the heterozygous mutation. Exclusion of homozygosity due to large deletions on one of the alleles occurred by chromosomal microarray analysis. In silico analysis of the mutation, including evolutionary conservation ( Figure 2) and protein stability predictors (Figure 3), supports a damaging impact on protein structure and/or function.
In Family 5, the only familial case in this study, a heterozygous nonsense mutation in FOXC1 (NM_001453.2:c.274C>T, p.Q92*) was identified in the proband and the affected mother, confirming an autosomal dominant inheritance pattern. DNA of the affected brother with PCG and a cousin with anterior segment dysgenesis (ASD) were not available for testing.

Screening for candidate variants
Five candidate genes were mutated in five families (Table 3 and Figure 4). All variants were missense mutations, with only one being homozygous. The latter was in the COL6A3 gene. The variant COL6A3 p.R544W (NM_004369.3:c.1630C>T) was detected in proband of Family 6, a consanguineous family. The unaffected parents and sister who also carried the heterozygous mutation, suggesting an autosomal recessive mode of inheritance. The variant was only found in 1/152248 alleles in gnomAD, with no homozygotes observed. The mutation occurred in a highly conserved residue (Figure 2), and 10 mutation prediction tools suggested pathogenicity. This study uncovered another mutation in the collagen-encoding gene, COL1A1. The heterozygous variant COL1A1 p.R528H (NM_000088.3:c.1583 G>A) was detected in the proband of Family 2 who also carried the heterozygous CYP1B1 p.R368G mutation. The unaffected mother and siblings also harbored the COL1A1 p.R528H variant. Detection of heterozygous variants in WDR36, FOXC2 and MAF genes transpired in Family 7, 8 and 9, respectively. Finding of variants in these genes in at least one unaffected member of the three families suggested they were less likely to be pathogenic.

Genotype-phenotype correlation
Of the twelve patients/families studied, four bore potentially pathogenic mutations, and a fifth patient carried a single heterozygous mutation in the CYP1B1 gene. The non-mutation group included mostly unilateral disease with only one in seven families having bilateral disease (14.3%). The number of medications and surgeries were the highest in the LTBP2/TEK/ANGPT1 patient compared to patients with other mutations or no mutations. A positive family history of PCG was noted in the FOXC1 family alone, with PCG present in a brother and mother, also a cousin of the proband had anterior segment dysgenesis. The initial measured IOP was much higher in the LTBP2/TEK/ANGPT1 triple mutation patient, as was the last IOP when compared to others. The discovery of an additional number of future cases with similar mutations and clinical presentations will allow more meaningful genotype-phenotype descriptions.

Discussion
In this study, 12 families affected with PCG and testing previously negative for CYP1B1 underwent whole-exome sequencing. The rate of consanguinity was 50% and age at presentation ranged from 0.25 to 18 months (median 3.5 months). This study uncovered several genetic mutations incriminated in PCG. Except for mutations in FOXC1, five novel genetic variants emerged in Lebanese PCG patients. Homozygous mutations in CYP1B1 (p. R368G), LTBP2 (p.E1013G) and TEK (p.T693I), and heterozygous mutations in FOXC1 (p.Q92*), TEK (c.3201-1 G>A), ANGPT1 (p.K186N) and CYP1B1 (p.R368G) were uncovered. The corneal edema was most severe in the patient with associated mutations in CYP1B1 and COL1A1. The patient with multiple LTBP2/TEK/ ANGPT1 mutations had the highest initial IOP, last IOP, and cup/disc ratio in the mutation group. Sixty percent (3/5) of patients with identified likely pathogenic variants had consanguineous parents and only 20% (1/5) had a family history of PCG.
For the distribution of PCG mutations in Lebanon and the entire Middle East and North Africa (MENA region) see Table 4. In the previous report from this center in Lebanon, the rate of CYP1B1 mutations was 33% (2). The current study, using whole-exome sequencing, revealed one more patient testing positive for CYP1B1. This may have been missed earlier as it was based on PCR amplification and Sanger sequencing whereby errors are possible in 1/1000 base pairs (62), also possibly due to the use of different sets of primers/PCR conditions as well as re-sampling of subjects, since the previous work was undertaken in a different laboratory. Whole-exome sequencing helped uncover this. CYP1B1 mutations were detected in 42.5% of Lebanese PCG alleles similar to Turkish (42%) and Egyptian (43.6%) rates ( Table 4). The second most common mutated gene in Lebanon was TEK (7.5%), followed by LTBP2 and FOXC1 genes (5% each), then ANGPT1 (2.5%). No MYOC mutations were identified in Lebanon, nor in neighboring countries (Table 4). LTBP2 and FOXC1 mutations were also scarce in Iran (9) and Turkey (58) ( Table 4).
Of the CYP1B1 mutations, p.G61E was the most prevalent (23.5%) in Lebanon, followed by p.E229K and p.R368G (17.6% each). Whereas p.R368G and p.S464* are unique in the Lebanese, p.G61E is the most prevalent CYP1B1 mutation in the region, reaching 72.2%, 73.9%, and 85.7% in Saudi Arabia, Kuwait, and Turkey, respectively. The p.E229K mutation is present in Iranians, Israeli Arabs, Iraqis, and Omanis (data not shown). The p.E229K mutation reported in Saudi Arabia belonged to a patient originating from Burma. The p.R469W mutation observed as homozygous in this cohort, was the most prevalent mutation in Israeli Muslim Arabs and Druze, and the second most prevalent mutation in Saudi Arabians. This mutation is present in combination with another homozygous mutation, p.E229K, in a single patient. The p.M1T mutation occurred only in Israeli Druze. The p.A443G mutation is Reported in abnormality of the cardiovascular system (46). present in one Lebanese patient in Germany and reported in one Saudi Arabian patient. Finally, the p.R444Q mutation, identified in the Lebanese cohort discussed here, was absent in other MENA countries ( Table 4). The latent transforming growth factor β-binding protein 2 (LTBP2) mutations segregated with PCG in many studies. In 2009, four LTBP2 missense variants and small deletions in four consanguineous families from Pakistan and in persons of Roma extraction with PCG were reported (8). In the same year, two LTBP2 loss-of-function pathogenic variants were identified in two Iranian families, and a compound heterozygous mutation was identified in a third family and was classified as potentially but not conclusively pathogenic (9). Moreover, a nonsense mutation (c.2421 G>A, p.W807×) in  LTBP2 was found in eight Indian patients (10). Studies have shown that LTBP2 was associated with a complex anterior segment phenotype rather than primary congenital glaucoma: patients with spherophakia or ectopia lentis had megalocornea or lens dislocation, associated with recessive mutations in LTBP2. A secondary, rather than a primary, autosomal recessive congenital or infantile glaucoma often occurs in these cases, as a result of anterior crystalline lens movement (21,22). In this cohort, one patient with a homozygous missense mutation in LTBP2 was associated with a heterozygous missense mutation in the ANGPT1 gene and a heterozygous splicing site mutation in the TEK gene. For this patient, functional studies are required to ascertain variant pathogenicity or lack thereof (as for all variants identified in this study) and to  determine a causative or modifier role of each variant. The LTBP2 p.E1013G variant was only found in one European subject in gnomAD (1/152202 alleles) and no homozygotes were reported. The transition G>A at position c.3201-1 in TEK is predicted to have a significant impact on splicing, due to a broken acceptor site (Suplementary Table S3). This may result in exon skipping or the creation of a new cryptic acceptor site. The three PCG mutations mentioned here are novel in PCG. This was a consanguineous family without history of PCG. The patient had poor final visual acuity (counting fingers in both eyes). This is in contrast to many studies stating that LTBP2 mutations do not cause PCG (6,11). Furthermore, in two Marfan-like phenotype Roma patients, a homozygous variant of LTBP2 (c.895C > T, p.R299×) was identified in 2019 (63). Lens luxation with a high IOP, a frequent presentation in Marfan syndrome, can develop into glaucoma (64). Lens luxation or Marfan-like phenotype were absent in this patient. The ocular drainage system derives from the mesenchyme where the forkhead box C1 gene (FOXC1) is expressed (12). In fact, previous descriptions of FOXC1 gene mutations in PCG patients with anterior segment abnormalities exist, and features of Axenfeld-Rieger spectrum were typically revealed in children with FOXC1 pathogenic variants (12) The p.Q92* mutation of the FOXC1 gene was identified in one patient from this cohort. A Mexican patient with Axenfeld-Rieger syndrome (ARS) (65) and a Vietnamese patient with ASD and congenital glaucoma carried this mutation (16). This mutation leads to a premature termination codon at position 92 of the protein. The truncated region affects the DNAbinding domain and the Nuclear Localization Signal 1 (NLS1) motif of the fork head domain of the FOXC1 protein.
An analysis of FOXC1 variants in 133 pedigrees with PCG in 2016 demonstrated that FOXC1 mutations may effect goniodysgenesis in PCG (14). FOXC1 variants were associated with PCG frequently complicated with Axenfeld-Rieger syndrome (hearing loss, heart murmur and developmental delay). Two point mutations of FOXC1 and PITX2 genes were identified in individuals with abnormal development of iris tissue (16). This cohort included a patient who presented at 1 week of age; agreeing with finding mutations in the FOXC1 gene causative of early-onset glaucomas (15). He also has a younger affected brother with PCG and a cousin with signs of Axenfeld-Rieger spectrum: posterior embryotoxon and corectopia.
The angiopoietin receptor TEK (tunica interna endothelial cell kinase) is a receptor tyrosine kinase that regulates vascular homeostasis through phosphorylation (17). In the eye, this receptor is highly expressed in the endothelium of Schlemm's Canal (17). Heterozygous novel/rare protein-altering mutations in TEK occurred in 10 of 189 families with PCG. It was evident that TEK gene dosage was critical for proper development of aqueous humor outflow pathways and Schlemm's canal (17). A different disease-causing variant of TEK discovered in an African patient with PCG highlights the value of whole-exome sequencing in detecting these variants (66). In 2021, TEK mutations causing PCG in Chinese patients, with one variant (p.R1003H) of the tyrosine kinase domain characterized as an activating mutation were described. Most cases were males, and the affection was mostly bilateral (67). This correlates with LTBP2/ANGPT1/TEK mutations in a male Interaction between these genes in PCG pathogenesis can occur. Twelve rare heterozygous missense mutations in TEK identified with four of them co-occurring with three heterozygous mutations in CYP1B1 exist (20). In this study, associated LTBP2/ANGPT1/TEK mutations were present in one patient with the highest initial and last IOP among the group, highest optic nerve cup-to-disk ratio and highest number of medications/surgeries needed. Adding to this, and in Family 2, a heterozygous missense mutation of CYP1B1 occurred in association with a COL1A1 mutation. The heterozygous missense CYP1B1 variant does not explain the glaucoma phenotype. The detection of a COL1A1 variant was suggested as a modifier gene and not a causative one, and other genes may be implicated in the pathogenesis of the PCG in this patient. He presented at 18 months of age with severe corneal edema and cup/disc ratio of 0.9.
Five candidate genes bore mutations in five families. The COL6A3 gene, previously described in a Peters' anomaly patient (68), a rare form of ASD, and associated with elevated IOP (69) and primary open-angle glaucoma (70), is reported here in a PCG patient for the first time. A COL1A1 mutation was uncovered in this study. Although the variant is less likely to be pathogenic, a modifier role of the COL1A1 gene remains possible. Mutations in the COL1A1 gene, primarily causing Osteogenesis Imperfecta (OI), and previously reported in two PCG patients not manifesting OI symptoms exist (71). Detection of variants in WDR36, FOXC2, and MAF genes occurred in three families in this study. WDR36 was previously associated with juvenile open-angle glaucoma (JOAG) (72), MAF associated with congenital cataract and glaucoma (73), and FOXC2 may be a modifying factor in congenital glaucoma (74). Many mutations in any given family may or may not be causative of the specific phenotype in this given family. Variants in these genes detected in at least one other member of these families makes them less likely pathogenic. Incomplete penetrance may occur as is often implicated in PCG genes. A search for candidate genes outside the screened gene list in Supplementary  Table S1 is planned in future analysis. Further analysis of all variants, notably expression and protein functional studies are necessary before ascertainment or lack thereof in pathogenicity is proven.
In conclusion, the studied cohort provides new data about PCG in Lebanon and its spectrum of genetic mutations. This study highlights the genetic heterogeneity of the Lebanese population with a notable high rate of inbreeding  and demonstrates the valuable role of whole-exome sequen-