Variable phenotype of congenital corneal opacities in biallelic CYP1B1 pathogenic variants

Abstract

Introduction:

To describe the variable phenotype of congenital corneal opacities (CCO) occurring in patients with biallelic CYP1B1 pathogenic variants.

Methods:

A retrospective chart review was conducted to identify patients with CCO and CYP1B1 pathogenic variants seen at UPMC Children’s Hospital of Pittsburgh. Ophthalmic examination, high frequency ultrasound, anterior segment optical coherence tomography, histopathology images and details of genetic testing were reviewed.

Results:

Three children were identified. All presented with raised intraocular pressure (IOP). Two patients showed bilateral limbus-to-limbus avascular corneal opacification which did not resolve with IOP control; one showed unilateral avascular corneal opacity with a crescent of clear cornea, irido-corneal adhesions (ICA), irido-lenticular adhesions (ILA), and classical features of congenital glaucoma in the fellow eye (enlarged corneal diameter, Haab striae, and clearing of the corneal clouding with appropriate IOP control). The first 2 patients were visually rehabilitated with penetrating keratoplasty. Histopathology revealed distinct features: a variably keratinized epithelium; a thick but discontinuous Bowman-like layer with areas of disruption and abnormal cellularity; Descemet’s membrane, when observed, showed reduced endothelial cells; no pathological changes of Haab striae were identified. Two patients had compound heterozygous pathogenic variants in CYP1B1 causing premature stop codons, whilst one was homozygous for a pathogenic missense variant.

Conclusion:

Congenital corneal opacities seen in biallelic CYP1B1 pathogenic variants have a variable phenotype. One is that commonly termed as Peters Anomaly type 1 (with ICA, with or without ILA); the other is a limbus-to-limbus opacity, termed CYP1B1 cytopathy. Clinicians should be aware of this phenotypic variability.

Introduction

Pathogenic variants of CYP1B1 are the commonest molecular cause of primary congenital glaucoma.¹ However, changes in CYP1B1 have also been reported in association with other anterior segment developmental anomalies (ASDA), including congenital corneal opacities (CCO).

Three previous reports have found CYP1B1 pathogenic variants in patients with Peters Anomaly (PA), proposing that primary congenital glaucoma and PA may share a common molecular pathophysiology. Peters Anomaly cases with CYP1B1 variants were not phenotypically different from the ones without involvement of this gene and were namely due to irido-corneal adhesions (ICA) with or without irido-lenticular adhesions (PA type 1).^2–4

CYP1B1 has been also linked to a particular phenotype of CCO, described as “CYP1B1 cytopathy”.^5,6 The term refers to a phenotype characterized by raised intraocular pressure (IOP) and limbus-to-limbus avascular corneal opacity, which do not resolve with glaucoma control. The histopathology was reported as showing central absence of Bowman layer, Descemet membrane, and endothelium, with no Haab striae. This latter feature is important to distinguish this condition from CCO secondary to uncontrolled congenital glaucoma. Corneal transparency in these patients can be obtained with penetrating keratoplasty (PK), but the graft survival may be threatened by the difficult glaucoma management.

We herein report three patients with biallelic CYP1B1 pathogenic variants and a spectrum of phenotypic presentation of ASDA.

Material and methods

A retrospective chart review was conducted to identify patients with CYP1B1 variants and corneal opacity seen in the Division of Pediatric Ophthalmology, Strabismus, and Adult Motility at UPMC Children’s Hospital of Pittsburgh.

Ocular examination under anesthesia was performed in all patients, including high frequency ultrasonography (HFU), and anterior segment optical coherence tomography (AS-OCT). Surgical interventions were performed when indicated and histopathology was reviewed when available.

Results

Three children were included. Details about the ophthalmic examination are reported in Table 1, and clinical pictures in Figure 1.

Table 1.

Clinical data of patients included in the study.

Patient (gender)	Age at presentation	Corneal diameter		Axial length
		RE	LE	RE	LE
1, M	2 months	10	9.5	19.53	19.51
2, M	3 years	13.5	13	28.87	27.62
3, F	3 years	12	12.25	23.31	25.33

Figure 1.

Clinical appearance of CYP1B1 cytopathy (A), with a limbus-to-limbus avascular corneal opacity. AS-OCT shows corneal thickening with subepithelial edema. Clinical appearance of the second phenotype related to CYP1B1 biallelic pathogenic variants (B), with avascular CCO secondary to irido-corneal adhesions.

Case 1

A 2-month-old boy was referred for CCO and glaucoma, previously managed with trabeculotomies, which failed to resolve the corneal clouding. At examination, he had diffused corneal opacities extending from limbus to limbus with corneal edema (Figure 1 A); in the left eye, the central area was relatively less opaque. The eyes were soft to moderate on palpation. The corresponding AS-OCT findings showed corneal thickening with subepithelial edema in the right eye and a relative central thinning in the left. The child was treated with bilateral PK, and the grafts remained clear at the one-year-follow-up visit. Six weeks after surgery, the cup-to-disc ratios were 0.1 and 0.2 in the right and left eye, respectively, but axial length was increased (20.24 mm and 20.04 mm, respectively), thus cyclodiode laser was performed.

Histopathological examination of the right cornea showed a thickened Bowman’s-like layer extending across the specimen (Figure 2 A, B), with central regions of attenuation associated with abnormal cellularity within the layer (Figure 2 C). The left cornea showed similar fragments of thickened Bowman’s-like tissue at the periphery (Figure 2 D, F) that was nearly absent centrally (Figure 2 E, G). High resolution composite images in comparison to a normal infant cornea showed disruption of the lamellar architecture of the stroma with an anterior condensed region and myxoid expansion of mid-corneal to posterior lamellae (Supplemental Figure S1). Both eyes exhibited corneal edema with keratinization of the corneal epithelium. Descemet’s membrane was thin and prone to disruption but was present across the majority of the left corneal profile. A corneal endothelial layer was not observed.

Figure 2.

The right cornea (A, PAS) exhibits severe hyperkeratosis and a markedly thickened Bowman’s layer (closed arrowheads) that is most striking towards the periphery (B, H&E). In the central cornea (C, H&E), this layer becomes more attenuated with fragmentation/loss and appearance of nuclei within the Bowman’s-like layer (open arrowheads). Descemet’s membrane is present at the periphery (A, arrow). The left cornea (D, PAS) shows keratinization of the corneal epithelium and small subepithelial bullae. Descemet’s layer is present peripherally (arrow) as well as centrally (E, arrow, PAS). The peripheral left cornea exhibits fragmented segments of a Bowman’s-like layer containing nuclei (F, open arrowhead, H&E), similar to the central region of the right eye. Bowman’s is mostly absent in the central region of the left eye (G, H&E). See also high resolution images in Supplemental Figure S1 for additional stromal changes. Scale bars: Panels A, D = 200 μm; Panels B-C, F-G = 50 μm; Panel E = 100 μm.

Of note, neither of patient’s parents have anterior segment abnormalities and there was no known family history of congenital ocular anomalies or glaucoma. The child showed an umbilical hernia, inguinal hernia and hypospadias, therefore genetic testing to rule out Axenfeld-Rieger syndrome were performed. Whole Exome Sequencing (WES) revealed compound heterozygosity for two CYP1B1 pathogenic variants, specifically: c.171G>A (p.Trp57X), which was maternally inherited, and c.1064_1076del (p.Arg355HisfsX69), which was paternally inherited.

CYP1B1 c.171G>A (p.Trp57X) was interpreted as pathogenic (ACMG criteria PVS1, PP5, PM2) and is a well-established variant in the medical literature having been previously detected in a homozygous or compound heterozygous state in individuals with glaucoma and Peter’s anomaly.^3,7–12 CYP1B1 c.1064_1076del (p.Arg355Hisfs*69) was interpreted as pathogenic (ACMG criteria PVS1, PP5, PM2), representing a well-established variant in the medical literature detected in a homozygous or compound heterozygous state in individuals with glaucoma.^7,10,13–15

Case 2

A 3-year-old male with a history of bilateral congenital glaucoma and dense corneal clouding extending to the limbus was referred to our clinic for consideration of PK. He was diagnosed as affected by congenital glaucoma at 10 days of life. The high IOP was initially managed pharmacologically, but it was necessary to proceed with numerous cyclodiode laser procedures and tube placement in both eyes. These procedures were effective for IOP control but did not lead to resolution or even reduction of the CCO.

The ophthalmic examination at our institution was significant for moderate-to-high IOP, bilaterally enlarged corneas with dense avascular opacities extending to the limbus, and bilateral aniridia. AS-OCT and HFU revealed that iris was absent bilaterally and lenses were present and not attached to the cornea. He underwent bilateral PK. Two months after surgery, the cup-to-disc ratio were 0.4-0.5 and 0.1 in the right and left eye, respectively. Axial length remained stable, but there appeared to be some corneal bulging within the sutures in the right eye, and an epithelial defect in the left eye, therefore cyclodiode laser was performed.

Histopathology of both eyes (Figure 3) revealed similar changes to the left eye of Case 1, with peripheral disrupted segments of an abnormally thick Bowman’s-like layer associated with abnormal cellularity, and corneal epithelial keratinization. There were regions of anterior condensed stroma with loss of lamellar architecture and myxoid deposits in the posterior lamellae with relatively normal architecture at the mid-level (Figure 3 A, Supplemental Figure S1). Segments of Descemet’s layer were observed in both peripheral and central cornea, with reduced endothelial cell numbers (Figure 3 A, E). There were no histopathological changes to indicate Descemet’s membrane rupture in vivo (Haab striae). The right eye additionally showed superficial chronic lymphocytic keratitis (Figure 3 F).

Figure 3.

Histopathological findings of patient 2. A thickened Bowman’s-like layer (arrowhead) containing extra nuclei (open arrowhead) are observed on the left eye (A, H&E). PAS stain reveals thin, discontinuous segments of Descemet’s membrane lacking endothelial cells spanning both peripheral and central cornea (A, inset, arrow). Higher magnification image of the peripheral cornea reveals mild keratinization of the corneal epithelium and subepithelial bullae (B, H&E, asterisk). Bowman’s layer is absent in the central cornea (C, H&E). The right eye showed short discontinuous segments of a Bowman’s-like layer that was not obviously thickened and extended into the central region (D, PAS, open arrowhead). A thin Descemet’s membrane was present along most of the specimen, showing occasional breaks (E, PAS, arrow). Centrally, there was some intraepithelial and subepithelial lymphocytes (F, leucocyte common antigen, red AEC chromagen). See also high resolution images in Supplemental Figure S1 for additional stromal changes. Scale bars: Panel A = 200 μm, inset A = 100 μm; Panels B, C, D, E, F = 50 μm.

Genetic testing via Next Generation Sequencing (NGS) of 54 genes related to anterior segment dysgenesis revealed homozygosity for a CYP1B1 missense variant (c.1333T>A, p.Phe445Ile) interpreted as pathogenic (ACMG criteria PM5, PM1, PP3, PP5, PM2). This variant is absent from the gnomAD population database but has been reported in the medical literature in a homozygous state in a patient with PCG.¹⁶ The pedigree was significant for consanguinity, explaining the observed homozygosity. A full list of genes analyzed appears in Supplemental Data.

Case 3

A 3-year-old girl was referred for bilateral congenital glaucoma and right CCO. At birth, she was noted to have complete opacification of both corneas. Ophthalmic surgical history included bilateral trabeculectomy and Ahmed tube implantation, with satisfactory IOP control. Her parents reported that whilst the left cornea cleared, the opacification of the right eye did not improve. For this reason, selective endothelial removal was attempted in the right eye, again without improvement of the CCO.

At examination at UMPC Children’s Hospital, she presented with right avascular corneal opacity with a crescent of clear cornea infero-temporally, ICA and irido-lenticular adhesions (Figure 1 B); the HFU highlighted that the tube was in an unusual position touching the lens, which showed a mild anterior lens opacification at that site. The left cornea was clear with one linear Haab Striae. Visual acuity was counting fingers and 20/250 in the right and left eye, respectively. A topical treatment to control IOP was started bilaterally. Parents’ eye examination was unremarkable except for father’ mild iris hypoplasia.

She was treated with right peripheral iridectomy. Histopathology (Figure 4) showed iris stroma with iris pigment epithelium and focal areas of iris sphincter muscle. Much of the stroma was replaced by dense fibrosis, confirmed by Masson trichrome stain. The material did not stain for PAS or Congo red.

Figure 4.

Histopathological findings of patient 3. The iris showed dense eosinophilic tissue filling the stroma (A), which stained positive for collagen (B, Masson Trichrome, blue) and was negative for amyloid (C, Congo Red). Scale bars = 200 μm.

Genetic testing via NGS of 71 genes related to anterior segment dysgenesis revealed two CYP1B1 pathogenic variants, specifically c.1064_1076del (p.Arg355Hisfs*69) and c.868dup (p.Arg290Profs*37). CYP1B1 c.1064_1076del (p.Arg355Hisfs*69) was interpreted as pathogenic for the reasons previously described in Case 1. CYP1B1 c.868dup (p.Arg290Profs*37) was interpreted as pathogenic (ACMG criteria PVS1, PP5, PM2) as it has been previously detected in a homozygous or compound heterozygous state in individuals with glaucoma.^{7,9,11,17–20} Due to the genomic distance of the variants, the NGS testing method was unable to determine whether the variants are in cis or in trans. Segregation studies were declined by the parents and therefore not performed. A full list of genes analyzed appears in Supplemental Data.

Discussion

Autosomal recessive pathogenic variants of CYP1B1 are the main known cause of primary congenital glaucoma (PCG), being responsible for up to 85% of cases depending on the population studied.^1,21,22 Rarely, biallelic variants of CYP1B1 have been reported in association with corneal opacities unrelated to the elevated IOP. ^2–4 Vincent et al described 3 patients with PA, underlying the possible role of CYP1B1 in anterior segment developmental anomalies beyond PCG, since one of their patients showed CCO without glaucoma. The authors described an ocular phenotype characterized by CCO with ICA and irido-lenticular adhesions.⁴ Edward et al found CYP1B1 variants in 6 out of 11 patients with PA, without finding neither phenotypic nor histopathologic differences between the subjects with and without CYP1B1 involvement.²

Kelberman et al previously reported 3 patients with glaucoma and corneal opacification that persisted despite early intervention and IOP control, suggesting that diffuse avascular limbus-to-limbus corneal opacity with no evidence of Haab’s striae and unique corneal histopathologic findings can occur in patients with CYP1B1 variants, albeit rarely. The outcomes of PK in these patients are good, whilst the IOP control may be difficult and can affect graft survival.⁵ Clinical features of this phenotype named “CYP1B1 cytopathy” included corneal diameter not exceeding 12 mm, diffuse corneal clouding without neovascularization and absence of ICA or kerato-lenticular adhesion (KLA). In particular, there were no signs of DM rupture related to Haab striae nor DM thickening as in corneal endothelial dystrophy (CHED). These histopathological features are important in the diagnosis of exclusion process since both two aforementioned conditions can present with total avascular corneal clouding and elevated IOP. Other authors reported a case with phenotype and histopathology consistent with CYP1B1 cytopathy. Since the outcomes of PK were poor in this report, they proposed that different pathogenic variants of CYP1B1 may have a distinct prognostic value.⁶ Edward et al also suggested that modifiers of the ocular phenotype can either mitigate or worsen the effects of CYP1B1 variants, thus explaining the co-occurrence of PCG, PA, and a normal ocular phenotype in relatives with homozygous CYP1B1 variants.²

In our case series, whilst the first patient fulfilled the diagnostic criteria for CYP1B1 cytopathy as previously defined,⁵ patients 2 and 3 showed a certain grade of phenotypic variability. First, they both presented increased corneal diameter. A horizontal corneal diameter greater than 13 mm has been reported as a risk factor for the development of Haab striae.²³ Despite this, in patient 2 (who presented the greater corneal diameters in our series), Haab Striae were absent clinically and pathologically, thus excluding the diagnosis of CCO secondary to uncontrolled raised IOP due to congenital glaucoma. The same patient presented with aniridia and was initially suspected as having Axenfeld-Rieger syndrome. Even though CYP1B1 can be associated with Rieger’s anomaly,²⁴ histopathological images were more consistent with CYP1B1 cytopathy. Of note, ASDA, including aniridia, are also associated with CYP1B1 pathogenic variants, both in the literature and in our clinical experience.^24,25

Haab striae exhibit tight coiling or thickening from healing over the ends that indicate tissue reaction as evidence that they formed in a living eye.²⁶ In patient 1, there was clearly a normal DM on one profile that stretches across most of the length of the specimen, with artifactual separation from posterior cornea. In the other eye, segments of DM were present at the periphery and in the mid-periphery; they were uniform in thickness with no coiling or deposition to suggest healing reactions. In patient 2, a thin DM was observed in segments separated by short gaps across the entire posterior surface. Higher magnification examination of the gap areas suggests that a deeper cut into the block would have shown a more continuous DM. Again, there was no evidence of healing reaction to support multiple ruptures in vivo.

Finally, patient 3 presented with unilateral CCO due to ICA, analogous to the one previously reported by Vincent et al (PA type 1).⁴

In the first report of 2011, all CYP1B1 cytopathy cases correlated with truncating variants with a presumed loss of protein function. In our series, patients 1 and 3 also had loss-of-function nonsense and frameshift variants, likely resulting in protein truncation or nonsense-mediated decay.⁷ Of note, the c.1064_1076del (p.Arg355Hisfs*69) frameshift variant was identified in both cases and prior functional studies have demonstrated that this variant results in an absence of CYP1B1 enzymatic activity.²⁷ While a homozygous missense variant was detected in patient 2, prior in vitro functional data of another variant affecting the same residue (c.1334C>G, p.Phe445Cys) supports a significant reduction in CYP1B1 activity and suggests that this residue is functionally important.²⁸ The variants found in our patients have been previously reported in association with primary congenital glaucoma and several anterior segment abnormalities, such as corneal opacities and Rieger anomaly.^11,24 It appears that there is not yet to be a specific phenotype-genotype correlation for CYP1B1 cytopathy.^3,11,12

Histopathologic findings in our patients were similar to the first report.⁵ Neither Edward nor Biénzobas specified Bowman’s appearance in the periphery of the cornea, so it’s not possible to know if a thick Bowman-like layer was also present in their specimens. They reported a central absence of Bowman’s layer and DM, with a concave appearance of the posterior corneal surface, which suggested the initial definition of “internal ulcer” by Von Hippel in 1897.²⁹ While our cases confirm central absence of Bowman’s layer, DM was thin and sometimes discontinuous, but was present centrally in 3 of the 4 eyes we examined. Given the loss or reduction in corneal endothelial cells observed histologically in patients 1 and 2, preoperative assessments of corneal thickness would have been helpful. Unfortunately, no pachymetry data was available for these patients and the Integrated Operative OCT instrumentation used did not include a measuring reticle.

Histopathologic analysis of the iris specimen of our patient 3 revealed abnormal findings. Since we did not analyze iris tissue from the remaining 2 patients, we can’t conclude that these abnormalities were related to the disease; however, Doshi et al suggested that since CYP1B1 can be detected very early in the iris (57 days after conception), the enzyme may have a role in the development of the iris itself. In the same way, CYP1B1 immunoreactivity can be found in fetal corneal epithelium and keratocytes, but not in adult’s ones, thus suggesting a possible role of CYP1B1 expression in the normal development of the anterior segment structures.³⁰ Interestingly, changes in anterior segment structures have also been described in CYP1B1 knockout mice, consisting primarily of focal iridocorneal synechiae.³¹

Although patients with PCG are often thought to have cloudy corneas only secondary to the raised IOP, the cases presented here indicate that CYP1B1 pathogenic variants show a greater range of phenotypic variability than previously reported. In particular, there were abnormalities of corneal stromal architecture observed in Cases 1 and 2. In cases where glaucoma control has been achieved but corneal clouding remains, CYP1B1 pathogenic variants should be suspected.

In conclusion, CYP1B1 is associated with a range of anterior segment developmental anomalies, including CCO which have a variable phenotype: CYP1B1 cytopathy and PA type 1 (due to ICA). Whether different pathogenic variants or external modifiers contribute to determine the phenotypic variability remains unclear and requires further investigation.

Supplementary Material

Supplemental Data

Supplemental Data. List of genes analyzed by next generation sequencing for case 2 and case 3.

Supplemental Data File (.doc, .tif, pdf, etc.)

Supplemental Figure S1. High resolution images of hematoxylin & eosin-stained corneal sections were acquired at 200x magnification as overlapping tiles and stitched together using Adobe Photoshop v. 22.4.3. Stromal features from the right peripheral and left central cornea of Case 1 and the central corneal of Case 2 were compared to the cornea of a normal 9-week old infant. The right cornea of Case 1 was thicker than the others. There were abnormalities in Bowman’s layer (vertical bars), with thickened Bowman’s at the periphery (see also Figures 2 A, B and 3 A, B) and marked attenuation and gaps in this layer centrally in the patient corneas. In the control cornea, Bowman’s layer is acellular and of uniform thickness across the cornea. The central cornea of Case 1 and 2 also show regions of abnormal stromal condensation (a) suggesting loss of the normal lamellar architecture, and regions of deeper stroma with a myxoid (blue-tinged) appearance (b). Descemet’s membrane (arrows) in the control cornea is lined by intact corneal endothelium, whereas Descemet’s in the patient corneas was disrupted or missing without a clear endothelial layer. Scale bar: 100 microns.