METHOD

The notes were reviewed of all cases of congenital corneal opacification presenting for the first time between December 1997 and October 1999. All cases had undergone a full clinical evaluation with or without examination under anaesthetic including anterior segment photography, high frequency UBM, and relevant serology and, in those cases where penetrating keratoplasty was performed, histopathology. In all cases a clinical diagnosis had been made before UBM was performed and then again after UBM was performed. The decision to proceed to penetrating keratoplasty was made by the principal surgeon in each centre (DSR, Toronto and KKN, London). The technique of penetrating keratoplasty was the same as that described by Ehrlich and colleagues in 1991.¹⁶ UBM was performed using the ultrasound biomicroscope (Allergan-Humphrey, San Leandro, CA, USA). Scans were performed in all cases except three, during examination under anaesthetic. A lid speculum was used to keep the eyelids open and Teargel or Viscotears (Ciba Vision) were used as the coupling agent between the transducer head and the patient's cornea. Scans were performed by two of the authors (LDM and KKN) using the same protocol. This protocol consisted of a minimum of four scans radial to the limbus and four scans parallel to the limbus at positions 12, 3, 6, and 9 o'clock. At least one scan axial to the estimated position of the pupil was also performed. Scan images were saved onto hard disc and hard copies were also made.

RESULTS

In total, 13 cases were seen with a mean postnatal age of 32.1 days (range 3–225 days). Nine cases presented to the Hospital for Sick Children, Toronto, between December 1997 and March 1999, while four presented to Great Ormond Street Hospital for Children, London, between March 1999 and September 1999.

The clinical cases and results are summarised in Tables 1, 2, and 3.

DISCUSSION

There have been four isolated reports of the UBM findings in sclerocornea and Peters' anomaly^7,8,10,11 and while one⁸ had histology, it was in an adult. We have reported the UBM findings in 13 cases (22 eyes) of congenital corneal opacification, 12 of which presented within 3 weeks of birth. More importantly, we have correlated the UBM findings with the clinical features in all cases and the histological findings in nine cases. The UBM findings changed the clinical diagnosis in five cases (see Table 3). Two cases of suspected congenital glaucoma (1 and 8) were thought to be cases of corneal dystrophies (posterior polymorphous dystrophy and congenital hereditary endothelial dystrophy) while two cases of suspected sclerocornea and the one case of corneal ectasia were found to be Peters' anomaly after UBM examination. The UBM findings in posterior polymorphous dystrophy and in congenital hereditary endothelial dystrophy have not previously been reported (see Table 2 and Fig 1). The final diagnosis in our series was Peters' anomaly in nine cases (70%), corneal dystrophy in two cases (15%), and sclerocornea in two cases (15%).

Ultrasound biomicroscopy (UBM) allows visualisation of the anterior segment to a depth of 5 mm with a resolution of approximately 50 μm ^1–3 and gives excellent differentiation of ocular tissues because different tissues produce different amounts of beam backscatter.¹ Using an 80 MHz transducer Pavlin has demonstrated that UBM will differentiate corneal epithelium, Bowman's membrane and stroma but not between Descemet's membrane and the endothelial layer.^1,2 In our study, using a 50 MHz transducer which is the commonest commercially available UBM, the corneal epithelium and Bowman's layer appear as a single hyperechoic layer. In order to record corneal epithelium and Bowman's layer as two separate layers the transducer must be exactly perpendicular to the corneal surface and the gain must be low. Since the intracorneal structures were not the focus of our examination we did not stringently try to meet these criteria which may explain the discrepancy between our findings and those of Pavlin.^1,2 Beneath the corneal epithelium/Bowman complex we found the relatively hypoechoic but heterogeneous stroma and finally Descemet's membrane (DM) and endothelial layer again as a single hyperechoic layer (see Fig 2).

In cases 1 and 8 the usually contiguous reflectivity of the DM/endothelium layer seen on UBM¹ was found to be irregular (Fig 1A) which was confirmed on histology to be due to focal absences of Descemet's membrane with multilayering of the endothelium case 1 (Fig 1B) and absences of endothelium in case 8; a feature not previously reported to our knowledge.

In seven cases (4–7, 10, 12, 13) where the normal hyperreflectivity of the DM/endothelium was not seen in association with a central posterior corneal defect (Fig 3B, 4A), absence of DM and endothelium was confirmed histologically where available (Fig 3C, 4B).

In four cases (4, 5, 7, 12) an unusual hypoechoic region was seen in the anterior stroma on UBM examination (Fig 3B, 4A). In all these four cases (eight eyes) histology revealed an absence of Bowman's layer with oedema in the region adjacent to where Bowman's layer should have been, together with absence of Descemet's membrane (Fig 3C, 4B). In case 9 Bowman's layer was also found to be histologically absent but no hypoechoic region was seen in the anterior stroma. One possible explanation may be that there was also vascularisation of the stroma and this case had presented after 7 months while all the other cases had presented within 2 weeks of birth.

In case 6 histology revealed a thickened Bowman's layer but there was no obvious UBM correlate. However, hyperplasia or thickened Bowman's layer has previously been described.^17–19

In the only clinicopathological correlation to date of clinically diagnosed sclerocornea,⁸ the authors described a flattened cornea, diffuse scleralisation of the cornea indicated by hyperreflectivity, abnormal Bowman's layer, thickening of the peripheral cornea, with central posterior excavations involving the posterior stroma, Descemet's membrane and endothelium. Histologically all the UBM findings were confirmed and in addition Bowman's layer was noted to be absent and replaced by a few irregular patches of hyaline material. It is noteworthy that despite the presence of a central posterior corneal defect both on UBM and histologically, a feature consistent with Peters' anomaly,^17,20–29 the authors failed to comment on this. Additionally no comment was made about the state of the anterior stroma on UBM⁸ but if the UBM figure is perused in the report there is a clear area of hypoechogenicity in the anterior stroma; this is identical to our findings on UBM in cases 4, 5, 7, and 12. In this report Bowman's layer was reported to be absent also histologically.

We suggest that the presence of a hypoechoic layer in the anterior stroma just below the epithelial hyperechoic layer may be indicative of absent Bowman's layer with concomitant oedema as evidenced by the histology of our cases and that of the only other clinicopathological report.⁸ To our knowledge, this has never been previously reported.

Avitabile et al⁵ have studied acquired corneal oedemas using UBM; however, all of their studies were at least 30 days after the initial insult, at which stage the opacity of the cornea seen was most probably related to scarring rather than true acute corneal oedema. This would explain why they describe increased hyperreflectivity within the stroma.

The description of UBM in Peters' anomaly has been reported in three papers ^10,11,13 but none had any correlation with histology. Azuara-Blanco et al¹³ described three eyes of two patients who had had a clinical diagnosis of Peters' anomaly made without histological confirmation. Their UBM findings were similar to ours with the central posterior corneal defect described as an excavation. We agree with their description of central keratolenticular and iridocorneal adhesions as seen in cases 2, 3, 4, 6, 7, and 12 (Fig 5B). Although case 13 showed iridocorneal adhesions, these were peripheral.

As early as 1867³⁰ the clinical condition of defect in Descemet's membrane giving rise to a central corneal opacification was attributed to defective separation of the lens from surface ectoderm. Peters³¹ in 1906 emphasised this aetiology and in so doing gave the condition its eponymous name. There is a substantial volume of literature regarding the histology of Peters' anomaly^{20–29,32,33} and less so for sclerocornea.^26,34 Regardless of the author, the hallmark of Peters' anomaly histologically is the central deficiency of the posterior stroma, Descemet's membrane, and endothelium with or without keratolenticular and/or iridocorneal adhesions^17,20–29 with a corresponding central corneal opacity clinically. Interestingly, absence of Bowman's membrane is also alluded to in a number of reports^{21,22,34–37} but some of these reports clinically describe sclerocornea with a rudimentary presence of DM.^34,36

In sclerocornea there is extension of opaque scleral tissue and fine vascular conjunctival and episcleral tissue into the peripheral cornea obscuring the limbus.²⁶ The severity of scleralisation varies from mild to complete but is usually bilateral in 90% of cases.^26,38–40 Histologically the corneal epithelium shows secondary changes with Bowman's layer absent in the affected areas^26,34 with interstitial vascularisation without inflammation and the stromal collagen fibrils are comparable to scleral collagen in size and organisation. There may be irregular absence of both endothelium and Descemet's membrane or an abnormally thinned Descemet's membrane composed of multilaminar basement membrane.^26,34

It appears that Peters' anomaly and sclerocornea are most likely conditions in the same spectrum of anterior segment dysgenesis.

UBM was useful in evaluating both the cornea itself as shown above and very useful in revealing associated ocular anomalies as demonstrated most clearly by cases 5, 9, and 13.

In case 9 no evidence of a lens could be found either on UBM or posterior segment ultrasound and we feel confident that this is bilateral primary congenital aphakia. Congenital aphakia is extremely rare^41–44 and when associated with Peters' anomaly even rarer.^41,42 Controversy exists as to whether primary aphakia (failure of any lens formation as opposed to secondary type where lens forms but subsequently is resorbed) can occur with an otherwise normal anterior segment or not.^43–46 Clinically, sclerocornea precluded a thorough examination of the anterior segment, which could only be done using the UBM and this revealed the absence of a lens and aniridia bilaterally (Table 2). This is the first reported case of congenital aphakia in association with clinically diagnosed sclerocornea to our knowledge.

Case 13 demonstrated the presence of Axenfeld-Rieger anomaly with Peters' anomaly. The scarcity of previous reports may be due to the fact that young infants are difficult to examine and gonioscopy to look for the iridocorneal adhesions is particularly difficult if the only such findings are in the affected eye with the corneal opacification.^{17,42,47–49} The use of UBM demonstrated the peripheral iridocorneal adhesions very clearly in our case although the fellow eye also demonstrated Axenfeld-Rieger anomaly very clearly.

Both cases 5 and 9 demonstrated the presence of aniridia which was not readily obvious prior to UBM because of the corneal opacity. Cases of aniridia and Peter's anomaly have been previously been reported^35,50 and so has one case of Peters' anomaly with Wilm's tumour.⁵¹ The iris in this latter report was described as being very hypoplastic.⁵¹

At least three developmental genes, PAX 6, REIG 1, and PITX 3, are involved in the development of the anterior segment of the eye.^52,53 There has been much investigation into the genetics of Peters' anomaly^{48,52,54–56} with controversy over the role of PAX 6.^52,54,56,57 PAX 6 (OMIM 106210) is a homeobox gene responsible for the control of ocular embryogenesis.^52,53,58 Mutations in PAX 6 are responsible for human aniridia and it has been suggested that no locus other than chromosome 11p13 has been implicated in aniridia and that PAX 6 may be the only gene responsible.⁵⁷

Doward et al have reported a case of Peters' anomaly in which a mutation of REIG 1 gene was found.⁴⁸ Mutations in the REIG 1 homeobox gene (OMIM 180500) on chromosome 4q25 have been reported in association with Rieger syndrome.⁴⁸ A mutation in the PITX 3 gene (OMIM 602669) on chromosome 10 has been associated with autosomal dominant Peters' anomaly, congenital cataract and other anterior segment malformations.⁵⁹

We believe that the accurate description of the phenotype of congenital corneal opacification is crucial in the evolution of phenotype-genotype correlation. In our series three of five cases clinically diagnosed as sclerocornea were found on UBM and, in some cases, histologically to have Peters' anomaly. This suggests that the clinical definition of phenotype in such cases is unreliable and the water is further muddied by the fact that sclerocornea and Peters' anomaly appear to be conditions whose histological features overlap suggesting they are part of the same spectrum of disease.^{17,20–29,32–36} In a series of articles in 1974 Townsend et al^17,20,24 tried to move away from the eponymous designations for developmental congenital corneal opacifications classifying them histologically according to the position and presence of defects in DM.

Whether the absence of Bowman's layer in cases of sclerocornea and Peters' anomaly^{21,22,34–37} is a primary event or secondary to an absent Descemet's membrane and endothelium, is unclear. If it were a primary event then elucidating a genetic association would be significant. The embryogenesis of Bowman's layer occurs late (4–5 months). It is thought to be produced by both the epithelium and the anterior stroma.¹⁸ If the lens is removed in the chick embryo on day 3 of gestation there is a resultant failure of the corneal stroma, DM and endothelium to develop and a greatly decreased density of Bowman's layer.¹⁸ Other authors have suggested that in Peters' anomaly the epithelium may be abnormal with an absent Bowman's layer.²² The central posterior defect of the cornea seen in Peters' anomaly may be as a result of failure of lens separation or due to apposition of the lens to the cornea.^17,20,24,26 Townsend has suggested that the posterior defect could be a passive effect of pressure by a forwardly displaced lens against the cornea at a time in development when the DM was absent or still a delicate structure.¹⁷ This suggests that the central corneal opacity of Peters' anomaly could be the final pathway for a number of varied pathologies, much like pulmonary fibrosis is the final pathway for conditions as varied as sarcoid, TB and cystic fibrosis.

Under these circumstances any phenotype-genotype correlation must be undertaken only with the most accurate phenotypic description available. We suggest that in the absence of histological diagnosis, the use of high frequency ultrasound should be mandatory in the description of phenotype where the anterior segment cannot be visualised. It is reasonable to suggest that the presence of Peters' anomaly with aniridia is most likely associated with a PAX 6 mutation according to Prosser and van Heyningen⁵⁶ while Peters' anomaly with Axenfeld-Rieger anomaly may be associated with RIEG 1 mutations.

We performed penetrating keratoplasty in nine cases (16 eyes) and one case had autologous rotational keratoplasty elsewhere. Penetrating keratoplasty for such cases is well described,^{16,37,60–69} while there are fewer reports of optical iridectomy and rotational keratoplasty.^70,71 It is noteworthy that one group of authors³⁷ named absence of Bowman's layer and, separately, absence of DM histologically as poor prognosticators; our UBM findings suggest that both these features could be determined preoperatively, thus giving the parents more information before consenting to surgical intervention. Furthermore, other authors⁶⁵ make the point that in most cases of Peters' anomaly the clinician has difficulty detecting keratolenticular adhesions hidden behind the dense corneal opacity and that for proper graft centration and wound entry site retroillumination must be employed.⁶⁰ By using UBM all surgical planning can be done before the eye is opened.

In summary then we have described the first series of clinico-ultrasonico-pathological descriptions of congenital corneal opacification. We have demonstrated that the clinical description of phenotype may be unreliable, by showing that the clinical diagnosis was changed in five out of 13 cases (38%) by the UBM findings and that in every case but one the UBM finding was confirmed histologically

In so doing we have described a new sign in high frequency ultrasound of hypoechogenecity of the anterior stroma (subepithelium) which has been shown histologically to be due to absent Bowman's layer with associated oedema. It is necessary to emphasise that sclerocornea and Peters' anomaly are part of the same spectrum of pathology. The importance of preoperative assessment and diagnosis in cases of corneal opacity cannot be overstated and is easily undertaken with UBM even in the awake infant.

Finally, in the present climate of increasing emphasis on studies of phenotype-genotype correlation we feel we have shown that UBM examination is an invaluable adjunct in accurately defining the phenotype.