Molecular Vision 2007; 13:1154-1160 <http://www.molvis.org/molvis/v13/a126/> Received 6 February 2007 | Accepted 8 July 2007 | Published 13 July 2007

Download Reprint

Two Chinese families with pulverulent congenital cataracts and ΔG91 CRYBA1 mutations

Shasha Lu,^1,2 Chen Zhao,^1,3 Hong Jiao,⁴ Juha Kere,⁴ Xin Tang,³ Feng Zhao,^1,5 Xiumei Zhang,⁶ Kanxing Zhao,³ Catharina Larsson¹
 
(The first two authors contributed equally to this publication)
 
 

¹Department of Molecular Medicine and Surgery, Karolinska Institutet, Karolinska University Hospital-Solna, Stockholm, Sweden; ²JiangSu Province Hospital, Nanjing, People's Republic of China; ³Laboratory of Molecular Genetics, Tianjin Eye Hospital, Tianjin Medical University, Tianjin, P. R. China; ⁴Department of Biosciences and Nutrition, Karolinska Institutet, Stockholm, Sweden; ⁵Department of Molecular Genetics, Tianjin Chest and Heart Hospital, Tianjin, People's Republic of China; ⁶The Health School of Jiaozuo, Jiaozuo, Henan, People's Republic of China

Correspondence to: Chen Zhao, Department of Molecular Medicine and Surgery, Karolinska Institutet, Karolinska University Hospital-Solna, CMM L8:01, SE-171 76, Stockholm, Sweden; Phone: +46 8 51773930; FAX: +46 8 51776180; email: chen.zhao@ki.se

Abstract

Purpose: To characterize the disease-causing mutations and related phenotypes in two Chinese families with autosomal dominant congenital cataract.

Methods: Family members were clinically characterized by a complete eye examination. Genome-wide linkage screening was performed in Family 1 using a 10K single nucleotide polymorphism approach followed by genotyping of microsatellite markers from the regions with highest support for linkage. The candidate gene, βA1-crystallin (CRYBA1), was sequenced in both families.

Results: Lens examinations in three affected phakic members showed bilateral pulverulent nuclear cataracts in two subjects of Family 1 while another subject of Family 2 displayed bilateral pulverulent lamellar cataract. Linkage analysis in 14 individuals (eight affected, three unaffected and three of their spouses) of Family 1 gave a maximum logarithm of odds score of 2.41 for D17S1294 in chromosomal region 17q11.12 that includes the CRYBA1 gene. In both families in-frame deletions of three bp were detected in exon 4 of CRYBA1 leading to loss of a guanine residue (ΔG91). The mutations cosegregated completely with the cataract phenotype in both families but were associated with distinct haplotypes suggesting that they had occurred independently.

Conclusions: The previously described CRYBA1 mutation ΔG91 was demonstrated in two Chinese families with distinct phenotypes of congenital cataract, suggesting a lack of genotype-phenotype correlation. The findings also raise the possibility that the ΔG91 mutation arise in a relatively mutation-prone sequence of the CRYBA1 gene.

Introduction

Congenital or juvenile cataract is a critical diagnosis in pediatric ophthalmology [1]. Its early recognition and surgical intervention are essential to avoid irreversible visual loss, especially in the case of complete cataract [1,2]. If left untreated, normal retinal development will be impaired due to lack of sharp focus of light and sensory deprivation [2].

Congenital cataracts have an overall prevalence of 1 to about 6 per 10,000 live births [2], and comprise a group of clinically and genetically heterogeneous conditions. The patients are generally classified according to the type and location of the observed opacities including cataracts of anterior polar, posterior polar, nuclear, lamellar, pulverulent, aceuliform, cerulean, total, cortical, polymorphic, or sutural types [3]. Approximately one-third of patients with isolated congenital cataracts have a familial form of the disease preferentially with autosomal dominant inheritance (autosomal dominant congenital cataract [ADCC]) [4]. To date, 16 disease genes have been identified for ADCC and 10 additional loci are implicated from family studies [3,5,6]. Nine of the known disease genes are related to the normal formation and function of crystallins including αA-crystallin (CRYAA) [7,8], αB-crystallin (CRYAB) [9], βA1-crystallin (CRYBA1) [10,11], βA4-crystallin (CRYBA4) [6], βB1-crystallin (CRYBB1) [12], βB2-crystallin (CRYBB2) [13-15], γC-crystallin (CRYGC) [16,17], γD-crystallin (CRYGD) [18], and γS-crystallin (CRYGS) [5].

Crystallins are essential for maintenance of lens transparency and refraction [19]. The super family of crystallins comprises crystallins of α-, β-, or γ-crystallin types [20], among which the β-crystallins are most abundant in the lens. The CRYBA1 gene (also known as CRYBA3/A1) encodes the β-crystallin A3 isoform 1, a 215 aa protein with a molecular weight of 25 kDa. Following linkage in an affected family to chromosomal region 17q11-12 [10], CRYBA1 was first identified as a cause of cataract in a pedigree with autosomal dominant zonular cataract [11]. Consequently, CRYBA1 mutations were reported in several families [21-25]. In addition to three splice mutations at the donor splice site of intron 3 [11,21,22], ΔG91 mutations have been described in three affected families of different ethnic backgrounds [23-25].

Herein, we report two unrelated Chinese families with ADCC in which a ΔG91 mutation of CRYBA1 were demonstrated. Phenotypic studies demonstrate that the ΔG91 mutation was associated with nuclear pulverulent cataract and lamellar pulverulent cataract in the respective families.

Methods

Families and clinical examinations

The two families (Family 1 with eight affected members and three unaffected siblings and Family 2 with four affected members and two unaffected siblings) with ADCC were identified and clinically evaluated in Tianjin Eye Hospital, Tianjin, China (Figure 1). In 17 individuals (12 affected and 5 unaffected members), the ophthalmic investigations included best correct visual acuity, slit-lamp examinations, measurement of intraocular pressure, and direct funduscopy. In three affected phakic individuals, (IV:3 and IV:4 of Family 1 and III:3 of Family 2), the lens was examined by slit-lamp. Peripheral blood samples for DNA analysis were obtained from 22 individuals including 14 members of Family 1 and 8 members of Family 2 (Figure 1). Informed written consent was obtained from each member of the family or their parents for sample collection and molecular analysis and the research was conducted with local ethical approval according to the Declaration of Helsinki.

Genotyping and linkage analysis in Family 1

Genome-wide genotyping of single nucleotide polymorphisms (SNPs) was carried out in 10 members of Family 1 (Figure 1) using Affymetrix GeneChip Mapping 10K Set of microarrays (Affymetrix, Santa Clara, CA). Genomic DNA was extracted from peripheral blood leukocytes using standard methods and 250 ng genomic DNA was assayed according to the recommendations of the manufacturer (GeneChip Mapping 10K 2.0 Assay Manual). The experiments were carried out at the Affymetrix Core Facility BEA at Karolinska Institutet, Stockholm, Sweden. Genotypes were determined using GDAS 2.0 software (Affymetrix). Non-parametric linkage (NPL) analyses were carried out using Merlin software [26] whereby the eight affected family members were included. Parametric linkage analyses were performed with GENEHUNTER version 2.1 and errors were removed using Pedcheck software (version 1.0) [27]. Chromosomal intervals were chosen based on linkage peaks in the non-parametric analyses. Furthermore, in some regions, SNPs were randomly removed to allow analysis by GENEHUNTER and to avoid association between markers. Parametric analyses were performed under an assumption of an autosomal dominant inheritance with a penetrance of 0.999 for heterozygotes and included the 10 SNP typed individuals. A disease allele frequency of 0.001 and a phenocopy rate of 0.001 were used. Allele frequencies were calculated based on the typed individuals. Genetic positions of non-parametric linkage (NPL) or logarithm of odds (LOD) scores on a chromosome were indicated in relation to the deCODE map.

Subsequently, additional microsatellite markers were selected for genotyping in five linkage-peak regions in chromosomes 1, 3, 4, 15, and 17 (Table 1). The aims were to refine the linkage mapping to a single chromosome, to refine the critical interval on a candidate chromosome, and to support cosegregation between a detected mutation and disease-associated haplotype. Amplification and detection of microsatellites was performed as previously described [28]. Multi-point linkage analysis was performed using the LINKAGE software package of SimWalk2, Version 3.35, under the assumption of an autosomal dominant trait with a disease-allele frequency of 0.0001 and a penetrance of 99%. The allele frequencies for each marker were assumed to be equal as well as the recombination frequencies in males and females.

Genotyping in Family 2

In Family 2, genotyping was carried out for eight microsatellite markers located close to the CRYBA1 gene in 17q11.2 (D17S921, D17S805, D17S1294, D17S1293, D17S966, D17S1299, D17S1868, and D17S787) with the aim of supporting cosegregation between detected mutation and disease-associated haplotype as well as to evaluate the possibility of a founder effect between Families 1 and 2. Pedigrees and haplotype data were generated using Cyrillic (version 2.1) software and confirmed by inspection.

Mutation screening of CRYBA1 in Families 1 and 2

The CRYBA1 gene was screened for mutations by sequencing of the six coding exons and flanking exon-intron boundaries using newly designed primers (Table 2). After amplification, the PCR products were purified and sequenced using the ABI BigDye^TM Terminator cycle sequencing kit v3.1, according to the manufacturer's instructions. The sequencing products were run and analysed in an ABI 3730 Genetic Analyzer (Perkin Elmer, Forster City, CA). Sequencing in both directions was carried out on DNA samples from three affected (II:1 and IV:4 in Family 1, II:1 in Family 2) and two unaffected (III:1 in Family 1, II:2 in Family 2) individuals. Exons with detected variations were sequenced in all family members, and candidate mutations were sequenced in 100 reference individuals to verify whether they represent disease-associated mutations.

Results

Clinical characteristics of the two autosomal dominant congenital cataract families

Both families demonstrated autosomal dominant inheritance of cataract predisposition with affected members in four successive generations and male to male transmission (Figure 1). According to the medical records, all affected individuals had presented with bilateral and symmetrical pulverulent nuclear cataracts in early childhood but were without progressive development of lens opacities. Taken together with the negative history of other systemic abnormalities, the disease was classified as autosomal dominant congenital cataract (ADCC) in both families. In the three phakic individuals, slit lamp examinations were performed to characterize the lens phenotypes. Individual IV:3 and IV:4 in Family 1 presented powdery opacities with a diameter of approximately 5 mm in the nucleus whereas the proband III:3 in Family 2 showed perinuclear-shaped pulverulent opacities which were restricted to the lamellae with a transparent embryonic nucleus (Figure 2). Furthermore, horizontal and pendular nystagmus was observed in five individuals in Family 1 in four cases together with strabismus (Table 3).

Genome-wide linkage screening

Genome-wide SNP typing using Affymetrix 10 K SNP arrays was carried out in Family 1. Non-parametric linkage analysis identified regions of NPL scores between 4.5 and 4.9 in chromosomes 1, 3, 4, 7, 15, and 17 using an "affected-only" approach. Subsequent parametric linkage analysis revealed maximum LOD score of 1.8 in chromosomes 1, 3, 4, 15, and 17, while only nonsignificant LOD scores were obtained for chromosome 7. The maximum NPL and LOD scores and target intervals of the five significant chromosomal regions are summarized in Table 1. In the next step microsatellite markers representing the five candidate regions were analyzed (Table 2). This excluded chromosomes 1, 3, 4, and 15 while close linkage without recombination was observed between ADCC and markers from chromosome 17 (D17S1294, D17S1923, D17S966, D17S1299, and D17S1868). From multi-point linkage analysis, a maximum LOD score of 2.41 was obtained for D17S1294, which is located close to CRYBA1, a gene known to be involved in ADCC (Figure 3).

Detection of CRYBA1 mutations

In-frame deletions of 3 bp (GAG) were identified in both families in exon 4 of CRYBA1 These deletions affect nucleotides 276-278 or 279-281 and are predicted to result in loss of a glycine residue at amino acid position 91 (ΔG91; Genbank NM_005208). The mutations completely cosegregated with the cataract phenotype in both families but were not found in any unaffected members or in 100 unrelated normal individuals. SNPs were also observed including rs1047790 and rs2286047 in Family 1 and rs1047790 in Family 2. To determine whether the mutations had occurred independently in the two families, or whether they could represent a founder effect, the disease associated haplotypes of the CRYBA1 region were determined and compared. For this purpose, microsatellite markers from the CRYBA1 region were genotyped in Family 2 as well. Taken together with the SNP rs1047790 in exon 5 of CRYBA1, the ΔG91 mutations were shown to be linked to different haplotypes in the two families (Figure 1), which argues against a founder effect.

Discussion

Our two families with distinctive ADCC phenotypes harbored ΔG91 mutations of the CRYBA1 gene. The cosegregation of mutation and cataract phenotype together with its absence in reference individuals support its pathogenic importance. By analyzing allele sharing for SNPs between affected members in Family 1, chromosome 17 was correctly identified as a candidate location. However, several additional chromosomes were also suggested and supported by NPL scores close to five. Similar results were obtained in parametric linkage analysis, which gave maximum LOD scores of 1.8 in five of these chromosomes. Subsequent analysis of regional microsatellites permitted assignment to the CRYBA1 region in 17q11 with exclusion of the other suggested chromosomes. The correct assignment was confirmed by demonstration of a ΔG91 mutation.

Taken together with the present study, CRYBA1 mutations have been reported in eight ADCC families. These CRYBA1 mutations were found in two locations, the splice donor site of intron 3 and nucleotides 276-281 in exon 4 [11,23-25]. In general, mutational clustering could reflect founder effects, mutational hot spot sequences, or a functional importance for the ADCC phenotype. In this study, ΔG91 was linked to different haplotypes of the CRYBA1 region, suggesting that the mutations in Family 1 and 2 have occurred independently. This finding is supported by the three published kindreds with ΔG91 that are from different geographical regions, which suggests their independent occurrence. Furthermore, the ΔG91 mutation can result from a 3 bp deletion affecting either the GAG at nucleotide 276-278 or the GAG at nt 279-281, while kindreds with ΔG91 do not necessarily share the same mutational event.

The functional consequences of ΔG91 have been described by Reddy et al. [25]. In experimental systems, the mutant protein was found to cause defective folding and reduced solubility [25]. βA1-crystallin consists of four "Greek-Key" β-sheet motifs with four β-strands termed a, b, c, and d [29]. The impaired folding is suggested to result from broken hydrogen bonds between c2 and d1 strands following the loss of glycine 91 next to a tyrosine corner that stabilize the protein at the connection between b and c β-strands [24].

Studies of affected families segregating the same mutations allow for genotype-phenotype comparisons. Although such comparisons in ADCC will be hampered because of the limited numbers of studied individuals, some observations deserve mentioning. A ΔG91 mutation of CRYBA1 has so far been identified in one Swiss [23], one British [25], and one Chinese family [24]. In our study, the two phakic individuals in Family 1 showed pulverulent nuclear cataract with powdery and white opacities while the proband of Family 2 had pulverulent lamellar cataract without involvement of embryonic nucleus or surrounding cortex. In the previously reported British family, bilateral dense opacities of lamellae were observed in seven phakic individuals [25]. In the Swiss family, one phakic subject presented with a symmetrical nuclear opacity (radial diameter of 5 mm) but was without involvement of the anterior or posterior Y-sutures [23]. Similarly, bilateral nuclear cataracts with a well-defined and dense opacity in the embryonic nucleus were observed in the Chinese family [24]. According to the most recent classification [3], patients with pulverulent cataracts have powdery opacities in the lens. In pulverulent nuclear or pulverulent lamellar cataracts, the opacities are restricted to the nucleus or lamellae, respectively [3]. The opacity observed in Family 1 had a size of 5 mm, indicating that both fetal and embryonic nuclei were involved. The phenotype of Family 1 partly overlaps with that of the previously-reported Swiss and Chinese families. Both of these families showed nuclear cataracts however, pulverulent opacities were not reported. Family 2 has pulverulent opacities restricted to the lamellae and are perinuclear-shaped with a transparent embryonic nucleus (Figure 2). In addition, a few dot-like opacities deposited in anterior lamellae were observed. In contrast to the British family reported by Reddy et al [25], Family 2 did not show dense opacities of lamellae.

Five affected members of Family 1 developed horizontal and pendular nystagmus at early ages before cataract surgery although the exact ages of onset are unknown due to delayed ophthalmic examination. It has been shown that children who lose central vision in both eyes before the age of two years develop nystagmus and that its severity depends on the extent of visual loss [30]. Given the central location of the opacity, the nystagmus is more likely secondary to the ADCC.

Altogether, the variation in cataract phenotype resulting from a ΔG91 mutation in CRYBA1 argues against a genotype-phenotype association. Instead, environmental factors or genetic variations in modifier genes could influence the phenotypic expression through inactivation or elimination of the mutant protein. Additionaly, the different phenotypes observed in ΔG91-carrying patients could result from mutation expression at different time-points during lens development [25]. A transient impairment might lead to a lamellar cataract [3] while nuclear cataract would develop over a longer period and at earlier stages.

In summary, we demonstrated ΔG91 mutations associated with distinct phenotypes in two Chinese families with ADCC. The findings suggest that ΔG91 occurs in a mutation prone sequence of CRYBA1 and that the phenotypic presentation of ADCC involves additional factors to the exact CRYBA1 mutation.

Acknowledgements

The study was financially supported by Swedish Research Council, Göran Gustafsson Foundation for Research in Natural Sciences and Medicine, Stockholm County Council, and China National Natural Science Foundation Awards (grant number 30070805).

References

1. Vijaya R, Gupta R, Panda G, Ravishankar K, Kumaramanickavel G. Genetic analysis of adult-onset cataract in a city-based ophthalmic hospital. Clin Genet 1997; 52:427-31.

2. Lambert SR, Drack AV. Infantile cataracts. Surv Ophthalmol 1996; 40:427-58.

3. Reddy MA, Francis PJ, Berry V, Bhattacharya SS, Moore AT. Molecular genetic basis of inherited cataract and associated phenotypes. Surv Ophthalmol 2004; 49:300-15.

4. Rahi JS, Dezateux C. Congenital and infantile cataract in the United Kingdom: underlying or associated factors. British Congenital Cataract Interest Group. Invest Ophthalmol Vis Sci 2000; 41:2108-14.

5. Sun H, Ma Z, Li Y, Liu B, Li Z, Ding X, Gao Y, Ma W, Tang X, Li X, Shen Y. Gamma-S crystallin gene (CRYGS) mutation causes dominant progressive cortical cataract in humans. J Med Genet 2005; 42:706-10.

6. Billingsley G, Santhiya ST, Paterson AD, Ogata K, Wodak S, Hosseini SM, Manisastry SM, Vijayalakshmi P, Gopinath PM, Graw J, Heon E. CRYBA4, a novel human cataract gene, is also involved in microphthalmia. Am J Hum Genet 2006; 79:702-9.

7. Litt M, Kramer P, LaMorticella DM, Murphey W, Lovrien EW, Weleber RG. Autosomal dominant congenital cataract associated with a missense mutation in the human alpha crystallin gene CRYAA. Hum Mol Genet 1998; 7:471-4.

8. Mackay DS, Andley UP, Shiels A. Cell death triggered by a novel mutation in the alphaA-crystallin gene underlies autosomal dominant cataract linked to chromosome 21q. Eur J Hum Genet 2003; 11:784-93.

9. Berry V, Francis P, Reddy MA, Collyer D, Vithana E, MacKay I, Dawson G, Carey AH, Moore A, Bhattacharya SS, Quinlan RA. Alpha-B crystallin gene (CRYAB) mutation causes dominant congenital posterior polar cataract in humans. Am J Hum Genet 2001; 69:1141-5.

10. Padma T, Ayyagari R, Murty JS, Basti S, Fletcher T, Rao GN, Kaiser-Kupfer M, Hejtmancik JF. Autosomal dominant zonular cataract with sutural opacities localized to chromosome 17q11-12. Am J Hum Genet 1995; 57:840-5.

11. Kannabiran C, Rogan PK, Olmos L, Basti S, Rao GN, Kaiser-Kupfer M, Hejtmancik JF. Autosomal dominant zonular cataract with sutural opacities is associated with a splice mutation in the betaA3/A1-crystallin gene. Mol Vis 1998; 4:21 <http://www.molvis.org/molvis/v4/a21/>.

12. Mackay DS, Boskovska OB, Knopf HL, Lampi KJ, Shiels A. A nonsense mutation in CRYBB1 associated with autosomal dominant cataract linked to human chromosome 22q. Am J Hum Genet 2002; 71:1216-21.

13. Gill D, Klose R, Munier FL, McFadden M, Priston M, Billingsley G, Ducrey N, Schorderet DF, Heon E. Genetic heterogeneity of the Coppock-like cataract: a mutation in CRYBB2 on chromosome 22q11.2. Invest Ophthalmol Vis Sci 2000; 41:159-65.

14. Litt M, Carrero-Valenzuela R, LaMorticella DM, Schultz DW, Mitchell TN, Kramer P, Maumenee IH. Autosomal dominant cerulean cataract is associated with a chain termination mutation in the human beta-crystallin gene CRYBB2. Hum Mol Genet 1997; 6:665-8.

15. Vanita, Sarhadi V, Reis A, Jung M, Singh D, Sperling K, Singh JR, Burger J. A unique form of autosomal dominant cataract explained by gene conversion between beta-crystallin B2 and its pseudogene. J Med Genet 2001; 38:392-6.

16. Heon E, Priston M, Schorderet DF, Billingsley GD, Girard PO, Lubsen N, Munier FL. The gamma-crystallins and human cataracts: a puzzle made clearer. Am J Hum Genet 1999; 65:1261-7.

17. Ren Z, Li A, Shastry BS, Padma T, Ayyagari R, Scott MH, Parks MM, Kaiser-Kupfer MI, Hejtmancik JF. A 5-base insertion in the gammaC-crystallin gene is associated with autosomal dominant variable zonular pulverulent cataract. Hum Genet 2000; 106:531-7.

18. Stephan DA, Gillanders E, Vanderveen D, Freas-Lutz D, Wistow G, Baxevanis AD, Robbins CM, VanAuken A, Quesenberry MI, Bailey-Wilson J, Juo SH, Trent JM, Smith L, Brownstein MJ. Progressive juvenile-onset punctate cataracts caused by mutation of the gammaD-crystallin gene. Proc Natl Acad Sci U S A 1999; 96:1008-12.

19. Delaye M, Tardieu A. Short-range order of crystallin proteins accounts for eye lens transparency. Nature 1983; 302:415-7.

20. Graw J. The crystallins: genes, proteins and diseases. Biol Chem 1997; 378:1331-48.

21. Bateman JB, Geyer DD, Flodman P, Johannes M, Sikela J, Walter N, Moreira AT, Clancy K, Spence MA. A new betaA1-crystallin splice junction mutation in autosomal dominant cataract. Invest Ophthalmol Vis Sci 2000; 41:3278-85.

22. Burdon KP, Wirth MG, Mackey DA, Russell-Eggitt IM, Craig JE, Elder JE, Dickinson JL, Sale MM. Investigation of crystallin genes in familial cataract, and report of two disease associated mutations. Br J Ophthalmol 2004; 88:79-83.

23. Ferrini W, Schorderet DF, Othenin-Girard P, Uffer S, Heon E, Munier FL. CRYBA3/A1 gene mutation associated with suture-sparing autosomal dominant congenital nuclear cataract: a novel phenotype. Invest Ophthalmol Vis Sci 2004; 45:1436-41.

24. Qi Y, Jia H, Huang S, Lin H, Gu J, Su H, Zhang T, Gao Y, Qu L, Li D, Li Y. A deletion mutation in the betaA1/A3 crystallin gene (CRYBA1/A3) is associated with autosomal dominant congenital nuclear cataract in a Chinese family. Hum Genet 2004; 114:192-7.

25. Reddy MA, Bateman OA, Chakarova C, Ferris J, Berry V, Lomas E, Sarra R, Smith MA, Moore AT, Bhattacharya SS, Slingsby C. Characterization of the G91del CRYBA1/3-crystallin protein: a cause of human inherited cataract. Hum Mol Genet 2004; 13:945-53.

26. Abecasis GR, Cherny SS, Cookson WO, Cardon LR. Merlin--rapid analysis of dense genetic maps using sparse gene flow trees. Nat Genet 2002; 30:97-101.

27. O'Connell JR, Weeks DE. PedCheck: a program for identification of genotype incompatibilities in linkage analysis. Am J Hum Genet 1998; 63:259-66.

28. Zhao C, Lu S, Zhou X, Zhang X, Zhao K, Larsson C. A novel locus (RP33) for autosomal dominant retinitis pigmentosa mapping to chromosomal region 2cen-q12.1. Hum Genet 2006; 119:617-23.

29. Jaenicke R, Slingsby C. Lens crystallins and their microbial homologs: structure, stability, and function. Crit Rev Biochem Mol Biol 2001; 36:435-99.

30. Kanski JJ. Clinical Ophthalmology. In: Neuro-ophthalmology. 4the ed. Woburn: Butterworth-Heinemann; 1999. p. 613-15.