Article Text

Download PDFPDF

Short report
Analysis of genotype–phenotype correlations in PAX6-associated aniridia
  1. Tatyana A. Vasilyeva1,
  2. Andrey V. Marakhonov1,
  3. Anna A. Voskresenskaya2,
  4. Vitaly V. Kadyshev1,
  5. Barbara Käsmann-Kellner3,
  6. Natella V. Sukhanova4,
  7. Lyudmila A. Katargina4,
  8. Sergey I. Kutsev1,
  9. Rena A. Zinchenko1
  1. 1 Laboratory of Genetic Epidemiology, Research Centre for Medical Genetics, Moscow, Russian Federation
  2. 2 Cheboksary Branch of S. Fyodorov Eye Microsurgery Federal State Institution, Cheboksary, Russian Federation
  3. 3 Universitätsklinikum des Saarlandes und Medizinische Fakultät der Universität des Saarlandes, Homburg, Saarland, Germany
  4. 4 Moscow Helmholtz Research Institute of Eye Diseases, Moscow, Russian Federation
  1. Correspondence to Dr Andrey V. Marakhonov, Research Centre for Medical Genetics, Moscow, Russian Federation; marakhonov{at}



Aniridia is a severe autosomal dominant panocular disorder associated with pathogenic sequence variants of the PAX6 gene or 11p13 chromosomal aberrations encompassing the coding and/or regulatory regions of the PAX6 gene in a heterozygous state. Patients with aniridia display several ocular anomalies including foveal hypoplasia, cataract, keratopathy, and glaucoma, which can vary in severity and combination.


A cohort of 155 patients from 125 unrelated families with identified point PAX6 pathogenic variants (118 patients) or large chromosomal 11p13 deletions (37 patients) was analyzed. Genetic causes were divided into 6 types. The occurrence of 6 aniridic eye anomalies was analyzed. Fisher’s exact test was applied for 2×2 contingency tables assigning numbers of patients with/without each sign and each type of the PAX6 variants or 11p13 deletions with Benjamini–Hochberg correction. The age of patients with different types of mutation did not differ.


Patients with 3′-cis-regulatory region deletions had a milder aniridia phenotype without keratopathy, nystagmus, or foveal hypoplasia. The phenotypes of the patients with other rearrangements involving 11p13 do not significantly differ from those associated with point pathogenic variants in the PAX6 gene. Missense mutations and genetic variants disrupting splicing are associated with a severe aniridia phenotype and resemble loss-of-function mutations. It is particularly important that in all examined patients, PAX6 mutations were found to be associated with multiple eye malformations. The age of patients with keratopathy, cataract, and glaucoma was significantly higher than the age of patients without these signs.


We got clear statistically significant genotype-phenotype correlations in congenital aniridia and evident that aniridia severity indeed had worsened with age.

  • congenital aniridia
  • PAX6pathogenic variants
  • chromosome region 11p13 deletions
  • 3′-cis-regulatory region deletions, genotype–phenotype correlations

Statistics from


Congenital aniridia [MIM: 106210] is a dominantly inherited Mendelian disorder with prevalence of 1:57143.1 It is predominantly caused by either small intragenic pathogenic variants in the PAX6 gene located on chromosome 11p13 or large chromosomal aberrations affecting the gene locus or its 3′-cis-regulatory region.

Congenital aniridia is always a complex panocular dysgenesis. Diagnostic features include absence of the iris or its hypoplasia in a complex with foveal hypoplasia, low visual acuity and nystagmus. These main signs can be accompanied by congenital or secondary cataract, keratopathy, glaucoma and optic nerve hypoplasia.2 Eye malformations can vary considerably in pattern and severity.3 It is reasonable to assume that diverse types of PAX6 defects would explain differences in the manifestation of aniridia. However, the great phenotypic variability of aniridia, levelled effect of nonsense-mediated decay (NMD) and small number of patients in studied samples impede the search for genotype–phenotype correlations. Nevertheless, some suggestions were made earlier. Taking into account the theory of PAX6 haploinsufficiency and dosage effect, Glaser et al suggested that there could be at least three types of PAX6 mutant alleles—amorphic, hypomorphic and neomorphic—leading to different activity of the resulting PAX6 protein.4 Gupta et al suggested that some aspects of the aniridia phenotype (e.g., cataract and foveal hypoplasia) correlate with an expected effect of point mutations on the function of the proline–serine–threonine-rich (PST) and paired domains of the PAX6 protein.5 This inference was not tested statistically due to the small examined sample.

As of 2009, Hingorani et al had investigated the largest group of patients (n=43).3 According to the predicted consequence of different kinds of mutations at the mRNA and protein levels, they categorised PAX6 mutant alleles into three main classes: (1) loss of gene expression; (2) amino acid substitution; (3) predicted C-terminal extension (CTE). They observed that individuals with loss-of-function (LoF) and CTE mutations were severely affected, whereas individuals with missense mutations had the mildest phenotypes. These observations were also not tested statistically.

Analysis of the Leiden Open Variation Database (LOVD) PAX6 pathogenic variants by Tzoulaki et al suggested some main contemporary ideas on correlations in aniridia.6 These can be summarised as follows: all pathogenic variants leading to premature termination codon (PTC) formation, irrespective of their localisation and type, should trigger mRNA degradation through the NMD mechanism and should result in similar phenotypes; PTC-causing pathogenic variants affecting the most 3′-sequence of the penultimate and the last exons encoding a PST domain could have dominant negative effect and thus could be associated with very severe phenotypes, rather than a congenital aniridia; missense substitutions in PAX6 are predominantly associated with non-aniridia phenotypes.

However, a statistically significant result was obtained only for the absence of late PTC pathogenic variants affecting the last 50 nucleotides of PAX6 exon 12 and the whole exon 13. These pathogenic variants have never been identified in humans.

Despite all these efforts and a great number of published aniridia cases with identified genetic causes, no significant results supporting association between diverse clinical signs of congenital aniridia and the type of PAX6 damage have been established to date.

Based on the results of detailed clinical examination and occurrence of clinical features in a large cohort of patients with aniridia, we undertook statistical analysis and revealed significant relationships between particular types of PAX6 mutation and aniridic eye phenotype.

Materials and methods


Clinical and molecular genetic studies were performed in accordance with the Declaration of Helsinki and written informed consent was obtained from each participant and/or their legal representative as appropriate.

In total, 155 patients from 125 unrelated families were included in the study. Of these, 75 (48.4%) are male and 80 (51.6%) are female. The mean age at examination was 16.8, median 9.0 years (25%–75%, range 3–28, minimum 0.2, maximum 65.0). Of the 155 patients, 87 (56.1%) are sporadic and 68 (43.9%) have a positive family history of aniridia; of the 125 probands, 87 (69.6%) were sporadic and 38 (30.4%) had a family history. The patients were examined by ophthalmologists of four clinics: Research Centre for Medical Genetics, Moscow; Cheboksary Branch of S. Fyodorov Eye Microsurgery Federal State Institution; Moscow Helmholtz Research Institute of Eye Diseases; National Medical Research Centre for Children’s Health, Moscow. Not all individuals were able to be examined for all ocular features.

PAX6 pathogenic variant analysis

Earlier screening for pathogenic variants in PAX6 was carried out by Sanger sequencing analysis and Multiplex Ligation-dependent Probe Amplification (MLPA) as described previously.7 All intragenic pathogenic variants were named based on PAX6 transcript variant 1 (NM_000280.4) and deposited in the PAX6 LOVD8 and ClinVar9 databases (online supplementary table s1). Pathogenic status of single nucleotide variation was established using the consensus recommendations of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology for interpretation of sequence variants.10 Deletion borders are designated according to location of MLPA probes with reduced signals according to the NCBI36/hg18 assembly of the human genome (online supplementary table s2).

Supplemental material

Supplemental material

Statistical analysis

To ensure a uniformly examined cohort of affected individuals, only those who met the following criteria were included in the study: clinical diagnosis of congenital aniridia; detailed clinical examination; identified molecular genetic cause differing from the Wilms tumour-aniridia syndrome (WAGR) region deletions (those patients might have WAGR syndrome as a possible diagnosis).

A total of 95 patients with congenital aniridia (from 78 unrelated families) from a recently published analysis were included in the study.7 In addition, 60 new patients (from 51 unrelated families) extended the cohort. Among the new patients, we found 47 intragenic PAX6 pathogenic variants (18 novel and 29 previously reported) as well as 2 patients with 3′-cis-regulatory PAX6 region deletions and 11 with other 11p13 deletions (tables S1 and S2). In total, 155 patients met the required criteria. Of these, 118 patients had point PAX6 pathogenic variants and 37 patients had large chromosomal 11p13 deletions.

The patients were allocated into seven groups according to the type of pathogenic variants they possess: (1) nonsense (n=40 patients); (2) missense (n=7); (3) splice site changes (n=32); (4) frame-shifting small indels (n=27); (5) CTE pathogenic variants (n=6)—and two groups with large chromosomal rearrangements: (6) 3′-cis-regulatory PAX6 region deletions (n=15); (7) other identified deletions (n=22). Two more small groups had pathogenic variants: (8) in the 5′-untranslated region (5′-UTR) sequence (n=3); (9) with start codon changes (n=3). Only the groups with more than five patients were analysed, but patients of the last two groups were included in the general cohort for comparison (table 1). Splice site changes group 3 included variants affecting canonical splicing site dinucleotides and also deep intronic variants which had been functionally tested previously.11

Table 1

Distribution of the numbers of patients into nine groups depending on different types of PAX6 pathogenic variants or 11p13 chromosomal deletions

Six characteristic features of aniridic eye were referred as complete or partial aniridia, and the presence or absence of nystagmus, keratopathy, cataract, glaucoma and foveal hypoplasia. Partial aniridia was considered when iris tissue is well visualised on biomicroscopy without using additional methods of examination (gonioscopy, optical coherence tomography, ultrasound biomicroscopy). Patients with asymmetric colobomatous defects of the iris, sectorial hypoplasia and pseudopolycoria were also referred as milder partial aniridia (online supplementary figure s1). Clinical features were set to have two simple grades: yes or no, present or absent (table 1). Detailed proportions of individuals with that features among individuals examined for that feature only in each group could be found in online supplementary table s3.

Supplemental material

Supplemental material

Comparison of patients’ age in groups with and without particular clinical signs was performed using Mann–Whitney U test with Benjamini–Hochberg correction for multiple testing. Kruskal–Wallis H test was used for comparison of ages in nine groups of patients with different mutation types. Analysis of genotypephenotype correlations was performed using two-sided Fisher’s exact test for 2×2 contingency tables assigning each of the congenital aniridia clinical signs and each type of PAX6 intragenic variants or 11p13 deletions with Benjamini–Hochberg correction for multiple testing using an in-house R script.12 Probability value less than p=0.05 allows rejection of the null hypothesis, which means that a specific phenotypic sign is significantly more frequently observed in patients with a particular type of PAX6 mutation.


All 155 examined patients demonstrated complex eye phenotypes. Complete aniridia was observed in 78% patients while partial aniridia was found in 22%; nystagmus, in 78%; cataract, in 80%; keratopathy, in 58%; foveal hypoplasia, in 85%; and glaucoma, in 26%. These frequencies are in accordance with other reports.13 14

First of all, we assessed the dependence of manifestations of different clinical signs and age of the patients and revealed that age of the patients with keratopathy, cataract and glaucoma was significantly higher than age of the patients without these signs (p=8.6604×10−7, p=1.0605×10−4 and p=0.0023, respectively). Presence of complete or partial aniridia, nystagmus and foveal hypoplasia did not depend on the age (p=0.9043, p=0.9043 and p=0.8300, respectively). Then, in order to exclude possible influence of patients’ age on the genotype–phenotype correlations, we compared the ages in nine groups of patients with different mutation types (according to table 1) using Kruskal–Wallis H test. The analysis demonstrated that all groups are identical. This conclusion allowed to be confident of independence of inferred genotype–phenotype correlations and age-related manifestation of clinical signs.

The analysis of genotype–phenotype correlations revealed that patients with 3′-cis-regulatory region deletions usually develop a milder aniridia phenotype without nystagmus, keratopathy or macula hypoplasia (p=0.0081, p=0.0003 and p=0.0007, respectively) (table 2). Phenotypes of the patients with all types of point pathogenic variants in the PAX6 gene and large chromosomal rearrangements excluding 3′-cis-regulatory region deletions do not differ significantly (p>0.05) (table 2).

Table 2

Values of adjusted probabilities of the exact Fisher’s test for 2×2 contingency tables created to study the relationship between each of six pathogenic variant types and each of six aniridia clinical traits after Benjamini–Hochberg correction

After that, we tested the popular hypothesis about dependence of phenotype on mutations in a particular domain of PAX6. We did the statistical analysis of localisation of mutations and specific phenotypic signs in our aniridia cohort and revealed no significant associations (online supplementary table s4).

Supplemental material

Our results in some respects support an existing view of genotype–phenotype correlations in aniridia, but they also establish new patterns.

The novelty is evidence of a very distinct and mild phenotype associated with 3′-cis-regulatory region deletions. Earlier studies hypothesised PAX6 3′-cis-region deletions to be associated with either milder15 or severe16 phenotypes. There are two possible reasons why observations on phenotype severity associated with PAX6 3′-cis-regulatory region deletions in different studies are disparate. Perhaps deletion formation mechanisms correlate with varying severity of a patient’s phenotype in case of either complex chromosome rearrangements or simple interstitial deletions. Also, the actual size of PAX6 3′-cis-regulatory region deletions could vary widely. Evidently, here we have studied a homogeneous sample of simple interstitial deletions without complex genomic rearrangements and dysregulation, which leads to a similar and less severe phenotype. It is probable that 3′-cis-regulatory region deletions do not result in the complete absence of PAX6 expression from the mutant allele, and there is a residual level of PAX6 expression from intact promotor. This in turn could result in more preserved phenotype. Kleinjan et al reported that the deletion of conserved SIMO element led to a decrease of Pax6 gene transcription level in mouse lens but not to its total absence, as the main Pax6 lens enhancer located 3.5 kb upstream to P0 promotor remains intact and maintains a basal transcription level.17 In addition, Favor et al showed that mouse bearing the deletion of Pax6 locus along with upstream gene Rcn1 resulted in more severe eye phenotypes than the deletion encompassing Pax6 and its downstream region.18 While comparing the degree of reduction of visual acuity in patients with aniridia carrying different 11p13 chromosome deletions, Zhang et al concluded that deletion encompassing the PAX6 and RCN1 genes is associated with significantly lower visual acuity than PAX6 3′-cis-regulatory region deletions.15

Missense pathogenic variants in accordance with Tzoulaki could generate proteins with residual functions and could lead to both aniridia as well as non-aniridia phenotype.6 Non-synonymous substitutions revealed in our study lead to classical aniridia phenotype. In silico analysis of possible molecular outcomes of PAX6 missense mutations demonstrates that aniridia-associated non-synonymous substitutions might completely disrupt protein stability and/or capability or specificity of DNA binding, and thus they could be classified as LoF (loss-of-function) mutations.19 According to this last point of view, missense mutations identified in our cohort lead to LoF at the PAX6 protein level and are associated with severe phenotypes like those caused by ‘classical’ LoF mutations. Besides, some formally missense mutations should be reassessed as LoF variants at the PAX6 transcript level, as they could disrupt splicing and lead to NMD. Functional analysis of one of the identified missense mutations, c.140A>G, p.(Q47R), by in vitro minigene assay showed that it damaged the normal splicing pattern.11 The applied statistical analysis does not support the idea that PAX6 missense mutations in aniridia are associated with a milder phenotype. Perhaps, such dependence will be revealed in patients with no aniridia with different missense mutations with non-LoF effect because of their other mode of action and varying effect of missense substitutions in different domains.

Finally, a finding concerns pathogenic variants disrupting splicing. Regardless of their location in or out of the consensus splice site dinucleotides, these sequence variants are associated with a severe phenotype. Here, we described 32 patients with pathogenic variants in consensus splice sites, in intronic and exonic splicing regulatory regions of the PAX6 gene. Previously conducted functional analysis of several sequence variants identified in the study by minigene splicing assay showed that all of them resulted in an open reading frame shift and thus produced null alleles leading to haploinsufficiency.11 So, we have reason to allocate identified and tested intronic variants into a joint group of disrupting splicing mutations.

Thus, all PAX6 LoF pathogenic variants including nonsense, frame shifting, pathogenic splicing variants, and some missense mutations and large chromosome deletions (except for 3′-cis-regulatory region deletions) identified in our study are undistinguished in terms of severity.


Statistical analysis of this large cohort of patients is a powerful tool for studying and establishing new patterns of genotype–phenotype correlations in congenital aniridia. We have determined a mild aniridia phenotype associated with 3′-cis-regulatory region deletions, which, perhaps, just decrease PAX6 expression to a basic level from the intact promotor. Our findings support the idea of the importance of a cis-regulatory mechanism of PAX6 gene expression and a role of its disruption in development of aniridia phenotype. To the best of our knowledge, this is the first report on statistically significant associations between PAX6 pathogenic variant types and ocular peculiarities in congenital aniridia. Finally, our analysis revealed that all PAX6 mutations found in patients with aniridia are LoF in terms of pathogenic effect except for the 3′-cis-regulatory region deletions associated with a milder phenotype.


We thank Dr. Richard H. Lozier for interest in our work and useful assistance.



  • TAV and AVM are joint first authors.

  • TAV and AVM contributed equally.

  • Contributors TAV, AVM, BK-K, LAK, SIK and RAZ contributed to the conception and design of the study. TAV, AVM, NVS, AAV, VVK, BK-K and RAZ contributed to the acquisition and analysis of data. TAV and AVM drafted the text.

  • Funding This work was supported in part by the Russian Science Foundation grant 17-15-01051 and the state assignment of Ministry of Science and Higher Education of the Russian Federation. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Ethics approval Institutional Review Board of the Research Centre for Medical Genetics, Moscow, Russia (approval no. 2017-4/1 dated 4 May 2017).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.