Aldave AJ, Sonmez B. Elucidating the Molecular Genetic Basis of the Corneal DystrophiesAre We There Yet?. Arch Ophthalmol. 2007;125(2):177-186. doi:10.1001/archopht.125.2.177
The identification of the genetic basis of approximately half of the corneal dystrophies in the past decade has resulted in significant advances in our understanding of the genetic control of corneal clarity and has provided clinicians with a definitive means to confirm or refute presumptive clinical diagnoses. This article serves as a guide to understanding the genetic basis of the corneal dystrophies and provides a revised anatomically based classification system that is intended for the clinician, who must possess a working knowledge of the molecular genetic basis of the corneal dystrophies to accurately diagnose,counsel, and manage the disease in affected patients.
Significant advances have been made in the past decade in our understanding of the genetic basis of inherited ocular disorders. The genetic basis of approximately half of the corneal dystrophies has been identified (Table 1), and a chromosomal locus has been described for several others. Dystrophies once thought to be distinct entities based on characteristic clinical and histopathologic features are now known to share a common genetic basis and in some cases are even associated with mutations involving the same codon (Table 2). The clinician is now able to use molecular genetic analysis to confirm or refute presumptive clinical diagnoses in cases of suspected dystrophic corneal disorders.1,2
However, with the ability to definitively diagnose various corneal dystrophies has come a realization that many previous assumptions about the corneal dystrophies are no longer valid. For example, although a dominantly inherited corneal dystrophy would have been considered an unlikely diagnosis in the affected offspring of unaffected parents, mutation screening has confirmed the diagnosis of a highly penetrant, dominantly inherited dystrophy in cases of spontaneous pathogenic mutations.3 In addition, the accepted means of differentiating between the corneal dystrophies on the basis of characteristic clinical features has been challenged by the observance of significant interfamilial and even intrafamilial phenotypic variability among affected individuals sharing a common mutation.4- 6 Even the definition of a corneal dystrophy has been challenged in this new era of molecular genetics, as we are now able to confirm the diagnosis of a corneal dystrophy in cases associated with an atypical phenotype, such as unilateral dystrophies,7- 11 dystrophies that involve more than a single layer of the cornea,5,12 and dystrophies that are associated with extraocular involvement.12- 14 Identification of the genetic basis of the corneal dystrophies and characterization of the expressed protein product have led to insights such as the fact that the transforming growth factor β–induced protein (TGFBI) dystrophies, which involve the Bowman layer and stroma, may actually be of an epithelial genesis.15- 17 This, as well as the characterization of several dystrophies that involve multiple layers of the cornea,5,12 highlights the limitations of the current method of classifying the corneal dystrophies by anatomic level of involvement. Thus, although we know much more about the molecular genetic basis of the corneal dystrophies than we did a decade ago,we are now confronted with even more questions about the role of the genetic background and environmental influences in determining the phenotypic expression of the corneal dystrophies, how to categorize and differentiate between the corneal dystrophies, and even how to define a corneal dystrophy. This article will serve as a guide to the current understanding of the molecular genetic basis of the corneal dystrophies, will provide a revised framework for classifying the corneal dystrophies, and will provide information regarding how molecular genetic analysis can be incorporated into everyday clinical practice to enhance the clinician's ability to differentiate between dystrophic and nondystrophic corneal disorders.
Despite the limitations of an anatomically based classification system for the corneal dystrophies, it remains the most useful means of categorizing the corneal dystrophies for the clinician. Given the ease of examining the cornea, in terms of its forming the anterior ocular surface and being a transparent tissue, the clinician is typically able to determine the precise morphological nature and location of the dystrophic deposits. The pattern of dystrophic deposition and the level of anatomic involvement typically determine the affected individual's associated symptoms. Dystrophies of the corneal epithelium and the Bowman layer are commonly associated with recurrent epithelial erosions and decreased visual acuity secondary to epithelial irregularity and scarring of the Bowman layer; stromal dystrophies produce decreased visual acuity by the deposition of various substances in the keratocytes or the extracellular matrix, in a localized manner or in association with systemic accumulation;and the dystrophies of the Descemet membrane and corneal endothelium impair visual acuity by interfering with the endothelial pump function, leading to corneal edema. Although exceptions exist to these generalizations, they illustrate the utility of an anatomically based classification system for the corneal dystrophies. Even when molecular genetic characterization of each of the corneal dystrophies is complete, the primary means of classifying the corneal dystrophies will likely be clinical, although such a classification will certainly incorporate histopathologic and molecular genetic features of the dystrophies. Thus, the classification system that appears herein, in the text and in tabular form (Table 1), presents the corneal dystrophies in a way that we hope will be useful to the clinician, who must possess knowledge of the molecular genetic basis of the corneal dystrophies to accurately diagnose,counsel, and manage the disease in affected patients.
Although several large pedigrees have been described in which patients affected with recurrent corneal erosions demonstrated characteristic clinical features of the dominantly inherited disorder of epithelial basement membrane dystrophy (EBMD; Mendelian Inheritance in Man [MIM] No. 121820), this condition has not been associated with a particular chromosomal locus because linkage analysis has not been performed in these or in other families. Difficulties in identifying the genetic basis of EBMD have arisen from difficulties encountered in defining the affected phenotype because affected patients may be asymptomatic,the often subtle epithelial changes are evanescent in nature, and the estimated phenocopy rate (ie, the number of other disorders of the corneal epithelium and the Bowman layer that may be associated with similar clinical features)is relatively high. However, the identification of the genetic basis of EBMD would be of interest to vision science researchers and refractive surgeons alike because the presence of subclinical EBMD is often associated with the development of a corneal epithelial defect during laser in situ keratomileusis,resulting in an increased risk of developing a number of different complications.Although screening of all individuals undergoing laser in situ keratomileusis,the most commonly performed surgical procedure in the United States, may not be practical, certainly, screening of family members of affected individuals who are interested in the procedure may be of benefit.
Meesmann corneal dystrophy (MIM No. 122100) is a dominantly inherited disorder of the corneal epithelium associated with the development of myriad epithelial microcysts in the first decade of life. Patients may remain asymptomatic for years, until epithelial erosions associated with rupture of the microcysts produce symptoms such as impaired visual acuity, pain, and photophobia. Meesmann corneal dystrophy is associated with mutations in the genes encoding 2 cornea-specific keratins, K3 (KRT3) and K12 (KRT12).18- 27 With a single exception, all causative mutations reported to date have been missense substitutions in the highly conserved helix initiation motif found in exon 1 of KRT12 or in the helix termination motifs found in exon 7 of KRT3 and exon 6 of KRT1218- 27 (Table 3).
Band-shaped, whorled microcystic corneal dystrophy (Lisch dystrophy)is unusual in that it is 1 of only 3 corneal dystrophies to have been linked to the X chromosome.28 Linkage analysis performed in a single affected family excluded linkage to the KRT3 and KRT12 loci and mapped the disease gene to the short arm of the X chromosome (Xp22.3).28 Affected patients demonstrate bilateral gray intraepithelial opacifications in band-shaped,spokelike, and whorled configurations that appear as epithelial microcysts on retroillumination. Affected patients may experience reduced visual acuity when the central cornea is involved, but recurrent epithelial erosions have not been reported.28- 32
Reis33 published a report in 1917 of a dominantly inherited condition associated with recurrent corneal erosions and superficial corneal scarring occurring in the first 2 decades of life.More than 30 years later, Bücklers34 described a similar condition, found in members of the same family described by Reis.33 The clinical characteristics of what has become known as Reis-Bücklers dystrophy are similar to those of another dystrophy of the Bowman layer that was described by Thiel and Behnke35 in 1967. This has unfortunately led to confusion in the literature, which prompted Küchle et al36 to propose that classification be based not only on the clinical manifestations but also on the distinctive electron microscopic features of each. Küchle et al36 have proposed the terms corneal dystrophy of the Bowman layer type I (CDB I) (MIM No. 608470)for what had been previously known as Reis-Bücklers dystrophy or a superficial variant of granular dystrophy, and corneal dystrophy of the Bowman layer type II (CDB II) (MIM No. 602082) for what had been previously known as Thiel-Behnke or honeycomb-shaped dystrophy. As will be discussed in more detail in the section on the stromal dystrophies, the dominantly inherited CDB I and CDB II are associated with the Arg124Leu and Arg555Gln mutations, respectively,in the TGFBI gene.37 Screening of TGFBI in patients with Bowman layer dystrophies has demonstrated that most of the reported cases of CDB I are actually descriptions of CDB II, which appears to be much more common than CDB I.36 Although most of the individuals reported as having a Bowman layer dystrophy in whom TGFBI screening has been performed demonstrate one of the aforementioned mutations, a similar phenotype has been reported with other TGFBI mutations. In addition, 2 cases have been reported of individuals with CDB I in whom the classic Arg124Leu mutation was identified,but whose parents were unaffected.3 Because confirmatory maternity and paternity testing was performed in each case, these represent the first reports of spontaneous pathogenic mutations in TGFBI. These cases also highlight the clinical utility of molecular genetic analysis in confirming the diagnosis of a suspected corneal dystrophy,which was questioned, given the absence of a family history in each case.
A second genetic locus on chromosome 10 (10q23-q24) has been reported for the Bowman layer dystrophies, although linkage has been demonstrated in only 1 family to date.38- 40
The demonstration of linkage of lattice (LCD1) (MIM No. 122200), granular (GCD) (MIM No. 121900), and combined granular-lattice (CGLCD) (MIM No. 607541)corneal dystrophies to chromosome 5q31 by Stone and colleagues41 in 1994, followed 3 years later by the demonstration of pathogenic mutations in TGFBI in each of these dystrophies and in CDB I,42 ushered in a new era of understanding of the relationship between the various corneal dystrophies. Because each of the TGFBI dystrophies is most commonly associated with a mutation in 1of only 2 codons in TGFBI (Table 2), screening of these 2 codons will detect most of the pathogenic mutations in TGFBI. However, a large number of TGFBI mutations have been subsequently described in patients with a variety of phenotypes, most commonly as variants of LCD (Table 4).
Rather than create a new type of LCD, GCD, and CGLCD with the discovery of each novel mutation, it is simple and useful to classify the TGFBI dystrophies as classic and variant. Using this classification system, CDB, LCD, GCD, and CGLCD are diagnosed as they have been traditionally, ie, based on the clinical and histologic features. Molecular genetic analysis would be used to determine whether the dystrophy was classic, that is, associated with the most commonly identified mutations given in Table 2,or variant, that is, associated with 1 of the less commonly encountered mutations given in Table 4 or a novel mutation.Even with the aggregation of all of the variant forms of GCD, LCD, and CGLCD into a single category, most of the patients with a TGFBI dystrophy will still demonstrate one of the classic mutations.
In 2000, Akama and colleagues77 reported the identification of mutations in a newly recognized carbohydrate sulfotransferase gene (CHST6) on chromosome 16 (16q22) in patients with macular corneal dystrophy (MCD; MIM No. 217800). Corneal N-acetylglucosamine-6-sulfotransferase, the product of the CHST6 gene, has been demonstrated to catalyze the sulfation of N-acetylglucosamine in keratan sulfate, and identified mutations have been shown to result in a loss of enzymatic activity.78 This loss of activity in turn is thought to result in the formation of unsulfated keratan sulfate, leading to a loss of transparency in the corneas of affected patients. Akama and colleagues,77 as well as our group and others,79- 91 have described coding region mutations in CHST6 in patients with type I MCD (differentiation between types I and IA was not performed),as well as deletions and rearrangements 5′ of CHST6 in patients with type II MCD. Although more than 100 mutations have been reported in CHST6 in individuals from a number of different ethnic groups, including patients with types I and II MCD, no association between the type and location of the mutation and the clinical features has been observed.77,79- 91 This is in contrast to the TGFBI dystrophies, in which such a phenotype-genotype correlation is observed, and indicates that diagnostic mutation screening necessitates screening the entire CHST6 coding region because no mutation hot spots such as those noted in TGFBI, have been observed.
Wilbaut and Van Went92 published a description of the dominantly inherited central crystalline dystrophy of Schnyder (SCCD)(MIM No. 121800) in 1924 that today bears the name of the man who further described its characteristic clinical features in 2 studies published in 1929and 1939.93,94 An important distinguishing feature of SCCD is that it is associated with a number of nonocular manifestations,the most important of which is hypercholesterolemia, found in approximately 40% of affected patients.14 Cultured fibroblasts obtained from skin biopsy specimens demonstrate evidence of abnormal intracellular cholesterol metabolism, providing evidence of a generalized metabolic disorder and a means of confirming the diagnosis.95
Ten years ago, Shearman and colleagues96 mapped the genetic locus for SCCD in 2 large Swedish-Finnish families to the short arm of chromosome 1 (originally reported to be 1p34.1-1p36, but revised to 1p36.2-1p36.3). After performing fine mapping in these and an additional 11families, the authors refined the candidate interval to a 2.32-megabase region that contains 30 genes, including 24 genes with a known gene function (http://www.ncbi.nlm.nih.gov; build 36.1). However, our laboratory has recently completed screening the coding region of each of these positional candidate genes in 2 families with SCCD and has not identified a presumed pathogenic coding region mutation.97 The possibility remains that the genetic basis of SCCD in the families that we screened links to another chromosomal region (ie, locus heterogeneity exists for SCCD), that pathogenic mutations are present in the promoter region or the 5′ or 3′ untranslated region of one of the positional candidate genes, or that an unidentified gene within the candidate interval is involved.
The genetic basis of rare, dominantly inherited, visually insignificant Fleck corneal dystrophy (MIM No. 121850) was recently elucidated with the identification of pathogenic mutations in the PIP5K3 gene in affected individuals from 8 families.98 This gene is a member of the phosphoinositide 3–kinase family and is involved in generating and relaying phosphorylation signals that regulate the formation and localization of intracellular lipid products.98,99 Fleck corneal dystrophy is characterized by discrete gray-white stromal flecks,representing keratocytes containing excess glycosaminoglycan and lipids100- 103 and is not associated with any systemic disorders involving mucopolysaccharide or lipid deposition.
Congenital hereditary stromal dystrophy (MIM No. 610048), an uncommon cause of congenital corneal opacification, has been reported in only 4 families in the English-language ophthalmic literature since it was first described in 1939.104- 107 In 2005, Bredrup and colleagues105 reported a mutation in the decorin gene in a newly described pedigree with congenital hereditary stromal dystrophy. Those authors hypothesized that the identified frameshift mutation in the decorin gene results in premature truncation of the protein product, a dermatan sulfate proteoglycan. This in turn results in abnormal binding of the mutant protein to collagen, disrupting collagen fibril spacing and thereby resulting in loss of corneal clarity.105
In 2001, Biswas and colleagues108 reported the results of a genome-wide linkage analysis in a large family with early-onset Fuchs endothelial corneal dystrophy (FECD) (MIM No. 136800), demonstrating linkage to a 6- to 7-centimorgan region on chromosome 1p34.3-1p32. The gene that encodes the α2 chain of type VIII collagen, COL8A2, was previously mapped to this region109 and was selected for screening because it is highly expressed in the Descemet membrane.110 A missense mutation, Gln455Lys,was identified that segregated with the disease in the family.108 Gottsch and colleagues111 have also identified a missense mutation, Leu450Trp, in another family with early-onset FECD, indicating that mutations in COL8A2 are associated with an early-onset variant of FECD.
Although 3 other presumed pathogenic COL8A2 sequence variants were identified by Biswas and colleagues108 in families with classic late-onset FECD, the identification of 2 of these variants in unaffected control individuals112,113 and the identification of the third variant in only a single sporadic case of FECD of more than 200 unrelated affected individuals screened indicates that it is likely a rare polymorphism.108,112,113 Thus,insufficient evidence exists to implicate COL8A2 mutations in the pathogenesis of classic FECD.
Several genome-wide linkage studies performed recently in families with the classic late-onset form of FECD have demonstrated evidence of linkage to a number of different chromosomal regions. Sundin and colleagues114 have demonstrated linkage to a 26.4-megabase region on chromosome 13 (13pTel-13q12.13) in a large kindred with FECD, although screening of the positional candidate genes has not identified any presumed pathogenic mutations. Recently, Sundin and colleagues115 reported linkage of 3 unrelated families with FECD to a 6.9-megabase region on chromosome 18 (18q21.2-18q21.32), although no pathogenic mutations have been identified in the positional candidate genes screened to date. Thus, it appears that locus heterogeneity may exist for FECD, in which mutations in several genes on different chromosomes may produce a common disease phenotype.
A National Institutes of Health–sponsored multicenter study to identify the genetic basis of classic FECD, organized in 2005 by investigators at Case Western Reserve University, Cleveland, Ohio, is currently recruiting affected families with the goal of enrolling 500 families from the 24 study sites. Given the delayed onset of the affected phenotype in FECD, single families with classic FECD typically do not have a sufficient number of individuals in multiple generations to successfully perform genome-wide association studies.Therefore, this multicenter study is recruiting hundreds of affected sibling pairs for use in performing genome-wide linkage analysis, which may replicate linkage to a previously identified locus for FECD or demonstrate linkage to other chromosomal loci.
Congenital hereditary endothelial dystrophy (CHED) (MIM No. 121700)is a rare, dominantly inherited cause of congenital corneal opacification.Autosomal dominant (CHED1) and autosomal recessive (CHED2) forms of CHED have been described, differing in the time of onset and characteristic clinical features. Both forms have been mapped to genetically distinct loci on chromosome 20, with CHED1 having been mapped to a region that lies within the region linked to posterior polymorphous dystrophy,116- 119 features of which are noted with increased frequency in relatives of patients with the dominant form of CHED. Recently, Vithana and colleagues120 have reported that mutations in the SLC4A11 gene, a member of the SLC4 family of bicarbonate transporter proteins, are associated with CHED2; missense and nonsense mutations in this gene segregated with the disease phenotype in 8 affected pedigrees.
Although mutations in the visual system homeobox gene 1 (VSX1) (MIM No. 605020), COL8A2, and the gene for transcription factor 8 (TCF8) have been reported to play a role in the pathogenesis of posterior polymorphous corneal dystrophy (PPCD) (MIM No. 122000), convincing evidence exists to support the role of only TCF8 mutations in this autosomal dominant corneal endothelial dystrophy. In 1995, Heon and colleagues121 reported the results of a genome-wide linkage analysis in a large family with PPCD,demonstrating evidence of linkage to the long arm of chromosome 20 (20q11).The VSX1 gene was mapped to the PPCD candidate interval in 2000 and was selected for screening by Heon and colleagues122 2years later as a positional and functional candidate gene for PPCD because it is a homeodomain-containing transcription factor gene that regulates the expression of cellular differentiation during development in various tissues,including the eye.123 Although 2 presumed pathogenic mutations were identified (Asp144Glu and Gly160Asp), our laboratory's identification of the Asp144Glu missense change in an unaffected control subject and the identification of the Gly160Asp variant in individuals without PPCD and in only 1 of 49 affected probands in whom VSX1 screening has been reported have led to doubts about a potential role of VSX1 in the pathogenesis of PPCD.13,122,124- 126
After the identification of COL8A2 mutations in families with FECD, Biswas and colleagues108 screened COL8A2 in 15 patients with PPCD, assuming that the 2 endothelial dystrophies might share a common genetic basis. Although a presumed pathogenic mutation was identified in 2 related individuals, our laboratory and others have not identified this previously reported mutation or any novel presumed pathogenic mutations in COL8A2 in a large number of individuals affected with PPCD, raising questions regarding the role of COL8A2 in PPCD.13,112,127
Krafchak and colleagues13 have recently reported frameshift and nonsense mutations in TCF8 in 5 of 11 families with PPCD. Each mutation was demonstrated to segregate with the affected phenotype in each family examined with the exception of 2 unaffected individuals from 1 large family. This exception was attributed by the authors to incomplete penetrance. Our laboratory has also identified presumed pathogenic frameshift and nonsense mutations in TCF8 in 25%of the affected probands screened; each TCF8 mutation identified by Krafchak et al13 and our laboratory has been identified in only a single family. Although the mechanism by which TCF8 mutations result in the ocular and histopathologic findings associated with PPCD (characterized by abnormal epithelial cell–like corneal endothelial cells128,129)has not been clearly elucidated, TCF8 has been demonstrated to be involved in the repression of the epithelial cell phenotype.130,131
The most recently described corneal dystrophy, X-linked endothelial corneal dystrophy (XECD), has been identified in a single large Austrian pedigree in which an X-linked inheritance pattern is suggested by the transmission of the dystrophy from affected fathers to all of their daughters but to none of their sons.132 However, the corneal endothelial changes, described as resembling moon craters, were observed in both male and female patients. Corneal opacification, in the form of congenital corneal edema and subepithelial band keratopathy, was identified only in male patients,however.132 To identify the genetic basis of XECD, the authors performed linkage analysis of the X chromosome and demonstrated evidence of linkage to a 4.73-centimorgan region on Xq25. Because this candidate gene region contains less than 100 known and predicted genes, the authors are in the process of prioritizing and screening the positional candidate genes.132
Identification of the genetic basis of many of the corneal dystrophies has led to the identification and characterization of many of the encoded proteins responsible for maintaining corneal transparency and thus has opened many new avenues of investigation for vision scientists. Clinicians have also benefited from the ability to definitively differentiate between dystrophic and nondystrophic causes of corneal opacification1,2,5,133,134 and will continue to use molecular diagnostic testing with increasing frequency in their practices as the genetic basis of more of the corneal dystrophies is elucidated and efficient screening methods for the most common pathogenic mutations continue to be refined. Diagnostic laboratories such as The John and Marcia Carver Nonprofit Genetic Testing Laboratory at the University of Iowa, Iowa City (http://www.carverlab.org) already offer clinicians the ability to screen for mutations in genes associated with a variety of inherited ocular disorders, including the TGBFI dystrophies. As mentioned previously, the TGBFI gene is ideally suited for mutation screening because most of the pathogenic mutations occur at 1 of 2 mutation hot spots (Table 2). Commercially available buccal swabs and saliva specimen containers that greatly simplify the collection, packaging, and shipping of DNA specimens for analysis also have eliminated nearly all of the practical constraints that a clinician might encounter in relying on peripheral blood specimens as a source of DNA for genetic analysis.1,2
Beyond providing the clinician with a definitive diagnostic tool, elucidation of the molecular genetic basis of the corneal dystrophies also has led to the ability to develop animal models of corneal disorders,135- 138 the next step in developing and testing novel treatment strategies for the corneal dystrophies.139
Correspondence: Anthony J. Aldave, MD, Jules Stein Eye Institute, 100 Stein Plaza, University of California–Los Angeles,Los Angeles, CA 90095 (firstname.lastname@example.org).
Submitted for Publication: June 13, 2006; final revision received August 26, 2006; accepted August 28, 2006.
Financial Disclosure: None reported.