Ocular findings in 22q11.2 deletion syndrome: A systematic literature review and results of a Dutch multicenter study

Abstract The 22q11.2 deletion syndrome (22q11.2DS) is a multisystem disorder with an estimated prevalence of 1:3000 live births. Manifestations show a marked variability in expression and include speech‐ and language delay, intellectual disability, and neuropsychiatric disorders. We aim to provide an overview of ocular findings in 22q11.2DS in order to optimize recommendations for ophthalmic screening. We combined results from a systematic literature review with results from a multicenter cross‐sectional study of patients with 22q11.2DS who were assessed by an ophthalmologist. Our systematic literature search yielded four articles, describing 270 patients. We included 132 patients in our cross‐sectional study (median age 8.9 [range 0–56] years). Most reported ocular findings were retinal vascular tortuosity (32%–78%), posterior embryotoxon (22%–50%), eye lid hooding (20%–67%), strabismus (12%–36%), amblyopia (2%–11%), ptosis (4%–6%), and refractive errors, of which hyperopia (6%–48%) and astigmatism (3%–23%) were most common. Visual acuity was (near) normal in most patients (91%–94%). Refractive errors, strabismus, and amblyopia are treatable conditions that are frequently present in patients with 22q11.2DS and should be corrected at an early stage. Therefore, in 22q11.2DS, we recommend ophthalmic and orthoptic screening at the age of 3 years or at diagnosis, and a low‐threshold referral in adults.


| INTRODUCTION
The 22q11.2 deletion syndrome (22q11.2DS) is a multisystem disorder with an estimated prevalence of 1 in 3000 live births (McDonald-McGinn et al., 2015). Patients show a marked variability in the clinical expression. Well-known manifestations include speech-language and developmental delay, intellectual disability, and an increased risk of developing psychiatric disorders such as schizophrenia and anxiety disorders (Bassett et al., 2005;Campbell et al., 2018;Schneider et al., 2014). Sensory dysfunction has been described as well. For example, hearing loss is frequently reported in 22q11.2DS and large deficits in olfactory function have been described in several studies (Moberg et al., 2020;Verheij et al., 2017). A number of studies have reported on ocular findings in 22q11.2DS, mainly focusing on children (Casteels et al., 2008;Forbes et al., 2007;Gokturk et al., 2016).
The aim of this study is to provide a systematic review of the literature on ocular findings in patients with 22q11.2DS and to present the results of a Dutch multicenter cross-sectional study of children and adults with 22q11.2DS in order to provide recommendations for ophthalmic screening in 22q11.2DS.

| Search strategy and study selection
On January 14th, 2021, we performed a systematic literature search in PubMed, Embase, and Cochrane medical databases (see Supporting Information Material 1 for details). After removing duplicates, titles and abstracts were independently screened by two reviewers (E.V., M. N., and/or E. V. S.). Subsequently, full-text articles were assessed for eligibility by the three reviewers. All studies that reported on ocular findings, that were assessed by a physician specialized in ophthalmology, in patients with 22q11.2DS were included. We excluded research reporting on patients with a clinical diagnosis of 22q11.2DS, velocardiofacial syndrome, or DiGeorge syndrome, that lacked molecular confirmation. We excluded studies that did not provide prevalences for specific ocular findings. Reviews, case studies, conference abstracts, and non-human studies were excluded. Discrepancies between authors were resolved by discussion. Reference lists of the included studies were hand-searched for additional relevant articles.

| Quality assessment
To assess the relevance and validity of the included articles, we performed a critical appraisal using the Risk of Bias Assessment tool for prevalence studies (Hoy et al., 2012), which was adapted and specified to our research question (see Supporting Information Material 2).
The quality of the studies was assessed independently by two reviewers (E. V., M. N. and/or E. V. S.) and discrepancies were resolved by discussion. In the absence of reference scores, we decided to exclude studies with a very high risk of bias (≥7/10 points) for data extraction. Risk of bias assessment included selection bias, standardization, measurement bias, and nonresponse bias. In case of overlap of populations of the same research group, the study with the lowest risk of bias was included. patients that visited the ophthalmological outpatient clinic of UMCU, MMCV, and/or MUMC+ between January 1992 and January 2021.
All centers are multidisciplinary outpatient clinics for 22q11.2DS.
Ophthalmic screening was carried out as regular screening after diagnosis or referral in all clinics and only in a minority of cases upon clinical indication.

| Study subjects
We included patients with a genetically confirmed 22q11.2 deletion.
Atypical 22q11.2 deletions were excluded, that is not involving the A-B region (McDonald-McGinn et al., 2015).

| Data collection
Data on demographic and clinical characteristics included molecular test results, sex, age at most recent ophthalmic screening, reason for referral, congenital heart defects, ophthalmological abnormalities, presence of a headache, prescription of glasses and treatment and/or ocular surgery in the past, and results of most recent ophthalmic screening. Prevalence rates of vascular tortuosity, posterior embryotoxon, and optic disk abnormalities are based on the total number of patients who were examined using fundoscopy and slit lamp.
Best corrected visual acuity measurements were transformed to LogMAR and categorized as described above (WHO, 2018). Spherical refractive errors were divided into six groups comparable to previous studies on ocular findings in 22q11.2DS (Forbes et al., 2007;Gokturk et al., 2016). Refractive errors, myopia and hyperopia, were considered mild in case of more than 0.5 diopters (D) to 2.0D, moderately severe in case of >2.0D and <4.0D, and severe in case of ≥4.0D. Finally, astigmatism with cylindrical errors of ≤À2.0D were extracted and considered high. Astigmatism was classified as withthe-rule, against-the-rule, and oblique as described before (Núñez et al., 2019).

| Statistical analysis
Categorical data are presented as frequencies with percentage (%) and continuous data are presented as median with ranges. For prevalence rates in our cross-sectional study, 95% confidence intervals were calculated. We used Spearman's Rank-Order correlation for studying the degree of association between age and refractive errors, given the asymmetric data distribution. We used χ 2 tests to compare ophthalmic findings, such as retinal vascular tortuosity, between men and women and between those with and without congenital heart defects. All analyses were two tailed, with statistical significance defined as p < 0.05, using IBM SPSS software (Statistics 25; SPSS, Inc.).

| RESULTS
3.1 | Systematic review of the literature

| Study selection
The flow diagram in Figure 1 shows the study selection process. We identified 1213 records through a literature search in PubMed, Embase, and the Cochrane Library. After deduplicating, we screened 871 titles and abstracts for relevance, resulting in 180 articles of which full-text was screened for eligibility. One article was added after manual search of the reference sections (Ryan et al., 1997). Four studies, including a total of 270 patients, were included for dataextraction in this systematic review. Four studies were excluded because of a high risk of bias (Supporting Information Material 3) (Cirillo et al., 2014;Ryan et al., 1997;Veerapandiyan et al., 2011;Vieira et al., 2015).  (Casteels et al., 2008;Forbes et al., 2007;Gokturk et al., 2016), and one was a retrospective cohort study (Midbari Kufert et al., 2016). Median number of included patients was 63 (range 16-128).

| Quality assessment
Substantial heterogeneity was present among studies concerning demographics, methods, definitions, and outcome measures. The methodology of the studies was poorly described or missing in most studies, complicating comparative evaluation.

| Ocular findings
VA was (near) normal in most patients (91%-94%). In one patient, one eye was severely impaired because of a Peters' anomaly (Casteels et al., 2008). One study described that VA was "lower than normal" in two eyes with high hyperopia and one eye with exotropia and high myopia (Gokturk et al., 2016).
Refractive errors were frequently reported in all included studies (Table 1). Hyperopia was the most common refractive error, with a prevalence ranging from 6% to 48% for moderate to severe hyperopia. Moderate to severe myopia (3%-10%) and high astigmatism (3%-12%) were less frequent. One study showed an increase in high astigmatism with age (Forbes et al., 2007).

| DISCUSSION
The results of our study indicate that ocular findings are frequently present in patients with 22q11.2DS. We report on ocular findings in the largest cohort of 22q11.2DS patients to date.
Importantly, VA was (near) normal in almost all patients. Severe visual impairment was reported for two children, one with Peters' anomaly and one with keratoconus, posterior subcapsular cataract, and tapetoretinal degeneration with a suspected second genetic hit. It is important to detect visual impairment because of its impact on language and communication development and for its negative effect on psychiatric illness such as depression or anxiety (Demmin & Silverstein, 2020;Mosca et al., 2015). Patients with intellectual disabilities and visual impairment may have an atypical presentation, such as self-injurious behavior or functional deterioration (de Winter et al., 2011). Also, fatigue and headaches are common in 22q11.2DS (Vergaelen et al., 2017), and may be caused by visual impairment in some cases. When measuring VA of patients with 22q11.2DS, cognitive abilities should be taken into account.
We found a high prevalence of moderate to severe hyperopia, especially in children with 22q11.2DS aged 6 years and older, compared to children and adults in the general population and also children with intellectual disabilities (Akinci et al., 2008;Hashemi et al., 2018;Williams et al., 2015). Studies in the general population have shown that emmetropisation takes place during early T A B L E 2 Refractive errors found in 212 eyes of 106 patients with 22q11.2DS <6y 6-11.9 y 12-17.9 y ≥18 y Total development resulting in a reduction and stabilization of refractive errors in early teenage years (Harb & Wildsoet, 2019;Read et al., 2007), which was not the case in our cohort. A possible reason for the high prevalence of hyperopia may be that the axial length of the eye is too short relative to the refractive power of the lens or cornea because of a delay in growth. Also, lag in accommodation has been found in children with severe hyperopia in the general population and may have contributed to the high prevalence of moderate to severe hyperopia in our study (Tarczy-Hornoch, 2012). In addition, the prevalence of astigmatism in children and adults with 22q11.2DS was much higher compared to the general population and compared to adults with intellectual disabilities (Akinci et al., 2008;van Splunder et al., 2003). Also, in our cross-sectional study, high astigmatism was more frequently present compared to previous studies in 22q11.2DS.
This may be explained by a higher inclusion rate of adults in our study, in whom astigmatism was found more often. Against-the-rule astigmatism was most common in all age groups in our cohort and can be influenced by a reduction in lid pressure (Read et al., 2007), which may have contributed to the disturbed emmetropisation in our cohort.
Eye lid hooding and ptosis were reported in a substantial number of patients with 22q11.2DS (20%-67% and 4%-6% respectively). Myopia was less frequently reported in children with 22q11.2DS compared to the general population, but a similar prevalence was found for adults (Hashemi et al., 2018). Correction of refractive errors in patients with 22q11.2DS at an early stage is important because it can improve reading abilities (Crewther et al., 1998). Also, high refractive errors and anisometropia have been associated with amblyopia (Brown et al., 2000;Mocanu & Horhat, 2018). Strabismus and amblyopia were frequently reported in 22q11.2DS and may have direct clinical consequences. The prevalence of strabismus and amblyopia is higher compared to the general population (12%-36% versus 1%-3% and 2%-11% versus 1%-4%, respectively) (Hashemi et al., 2019;Webber & Wood, 2005;Yekta et al., 2017), but comparable to what has been reported in children with intellectual disabilities (14% for strabismus) (Akinci et al., 2008). This may suggest that these results are not specific for a 22q11.2 deletion. Clinicians treating patients with 22q11.2DS should be aware of the increased prevalence of refractive errors, strabismus, and amblyopia and their influence on VA and language and communication development if not treated correctly (Blair et al., 2021;Catalano, 1990;Pascolini & Mariotti, 2012). Management of amblyopia includes correction of refractive errors or occlusion therapy and intervention preferably takes place as young as possible because of reduced plasticity of the visual cortex after the age of 7 years (Sengpiel, 2014). The management of strabismus also depends upon the etiology and includes surgical and nonsurgical strategies.
The most common ocular finding, though without clinical consequences, in both the systematic review studies and our cross-sectional study, was retinal vascular tortuosity (32%-78%). There was one study that reported a prevalence of 4% but did not provide additional information regarding measurement method or an explanation for this very low prevalence compared to other 22q11.2DS studies (Midbari Kufert et al., 2016). Retinal vascular tortuosity has a prevalence of 6% in the general population and therefore may be considered as a typical finding in patients with 22q11.2DS (Henkind & Walsh, 1980). Importantly, retinal vascular tortuosity has been associated with other disorders including obstructive sleep apnea (Mohsenin et al., 2013), diabetes mellitus (Sasongko et al., 2011), and schizophrenia (Appaji et al., 2019) in non-22q11.2DS populations. These disorders are also frequently reported in patients with 22q11.2DS (Kennedy et al., 2014;Schneider et al., 2014;Van et al., 2020;Zinkstok et al., 2019). In accordance with previous studies in 22q11.2DS, we did not find a correlation between retinal vascular tortuosity and cardiac anomalies (Casteels et al., 2008;Gokturk et al., 2016).
Another common finding in 22q11.2DS was posterior embryotoxon (22%-50%), that also has a higher prevalence compared to the general population (7%) (Rennie et al., 2005). As proposed by others, posterior embryotoxon and other anterior segment abnormalities may be a result of defects in migration, proliferation, and differentiation of neural crest cells in an early embryologic stage in 22q11.2DS (Casteels et al., 2008;Gokturk et al., 2016;Mansour et al., 1987). Anterior segment dysgenesis may increase the risk of glaucoma, which was reported only once in the included review studies (Gokturk et al., 2016) and in not a single patient in our cohort.
Other findings supporting a role of a 22q11.2 deletion in anterior segment dysgenesis were scarce, including Peter's anomaly, iris remnants, and lens opacities (Casteels et al., 2008;Gokturk et al., 2016).  (Evenhuis, 1998;van Splunder et al., 2006). Recommendations for monitoring are provided in Table 4. Third, there is a risk of selection bias since most participants in our cross-sectional study and studies included in this review were assessed in tertiary 22q11.2 centers. However, most participants are referred to these tertiary centers for congenital heart defects, speech and language disorders (including velopharyngeal insufficiency), and/or developmental, psychological, or psychiatric problems. Therefore, we do not expect overestimated ophthalmologic prevalences.

| CONCLUSION
Refractive errors, strabismus, and amblyopia are common, clinically relevant, and treatable ocular findings in patients with 22q11.2DS. Clinicians should be aware of these manifestations and the beneficial result of detection and correction at an early age. Therefore, we would recommend standardized ophthalmic and orthoptic screening in children with 22q11.2DS at the age of 3 years or at diagnosis, and a low-threshold for referral in adults. Note: Each ✓ refers to a single assessment with follow-up as needed.

ACKNOWLEDGMENT
Ophthalmic screening should be done at least once by an ophthalmologist.