Heart diseases are among the most common congenital diseases, occurring in 1% of live births, with frequent co-existence of ocular sequelae.Reference Mansour, Bitar and Traboulsi 1 These consequences include retinal arteriolar narrowing (focal and diffuse), increased vascular tortuosity, retinal capillary disease – found as oedema or ischaemic and haemorrhagic infarction – thrombotic or embolic occlusion of the retinal artery, choroid and optic disc, and proliferative retinopathy with secondary retinal detachment. Other signs of ocular involvement include cataracts, strabismus, glaucoma, eyelid abnormalities, and amblyopia, with or without associated genetic syndromes.Reference Cheung, Zheng and Hsu 2 – Reference Vilela, Mielke, Tyllman and Stein 7 The prevalence of ocular signs in patients with CHDs ranges from 6.3Reference Fusco, Magli and Pantaleo 8 to 65%Reference Petersen and Rosenthal 9 in literature reports. Such a large variation may be due to sample characteristics, heterogeneity of heart diseases, and techniques of ophthalmic evaluation. In addition, most of the published data include case reports or case series with specific heart lesions, with scarce information from population studies. Our hypothesis is that true prevalence may be closer to the higher values, as the largest and methodologically sound studies report high prevalences, and survival has been increasing progressively in more complex patients.
Thus, the present study aimed to estimate the prevalence of ocular abnormalities in patients from birth to adulthood with CHD, through a systematic review with meta-analysis of observational studies.
Methods
Sources and search strategy
The investigators, who received formal training in systematic review, performed all searches. A systematic search was performed in the following electronic databases: MEDLINE (accessed via PubMed), Cochrane Central Register of Controlled Trials (Cochrane CENTRAL), and EMBASE, from inception until August, 2014. In addition, references of studies published on the patient were manually searched and authors were contacted when necessary. The search strategy used in PubMed is shown in Table 1. There was no language restriction. One non-English paper was translated by an experienced professional.Reference Raczyńska, Potaz and Aleszewicz-Baranowska 10 A manual search was performed, but no unpublished study or conference abstract fulfilled the inclusion criteria. Thus, there was no need to contact authors for further information or to handle unpublished abstracts.
Search strategy used for the PubMed database.
Eligibility criteria
This review included observational studies – cohort, cross-sectional, or case–control studies – baseline results of randomised or non-randomised clinical trials, or observational follow-up after clinical trials assessing the prevalence of ocular findings in children and adolescents aged between 6 and 18 years with CHDs. The primary outcome was the presence of ocular symptoms, including all findings detected in the eyeball and annexes.
Selection of studies and data extraction
The analysis of titles and abstracts of all publications identified in the search strategy, as well as the extraction of the data regarding the methodological characteristics of the studies, patients, and results, were carried out in duplicate by two independent reviewers – L.C.P and M.A.P.V. – through standardised forms. Papers whose abstracts did not contain sufficient information regarding the inclusion and exclusion criteria were selected for full evaluation. In the second phase, the same reviewers independently assessed the complete articles and selected them in accordance with the eligibility criteria. The extracted variables included all signs of structural or functional modifications involving the eyeball and its annexes. Discrepancies between the results from the two reviewers were resolved by consensus and, in cases of persistent differences, by a third assessment (G.S.).
Assessment of risk of bias
For the assessment of risk of bias, two independent and blinded reviewers – L.C.P and M.A.P.V. – assessed the methodological quality of the studies, based on the criteria of Downs and Black,Reference Downs and Black 11 which considers the following 5 items: available information – objectives, outcome, sample description, description of loss, variability of results, and the actual probability of the findings – external validity – representativeness of the sample, sampling procedures, site representativeness, and team involved in the intervention – bias – kind of blinding, prior planning of analysis, duration of follow-up, adequacy of analysis, and accuracy of the tests – confusion – origin of the population, selection period, randomisation, allocation, adjusted analysis and analysis for intention to treat, and critical appraise of losses in the discussion section – and power of the study.
Disagreements between the reviewers were resolved by consensus and, in cases of persistent disagreement, by a third reviewer (G.S.).
Data analysis
The quantitative analysis of the included studies was performed by meta-analysis of single arm, with estimation of an average prevalence weighted by sample size. A random effect model was used. The meta-analysis was performed in the Comprehensive Meta Analysis software. Heterogeneity between studies was explored using the following strategy: we repeated the meta-analyses, removing one paper at a time to check whether an individual study explained heterogeneity.
Results
Description of studies
Of the 2413 studies identified in the search, eight matched the eligibility criteria, yielding a total of 1061 patients with CHDs; one of the studies also had a control group composed of 76 healthy individuals.Reference Fusco, Magli and Pantaleo 8 Figure 1 shows the flowchart of the studies included in this review. The age of the individuals ranged from 6 days to 39 years. All the included articles were cross-sectional studies; four of the studiesReference Mansour, Bitar and Traboulsi 1 , Reference Fusco, Magli and Pantaleo 8 , Reference Alfano 12 , Reference Gardiner and Joseph 13 examined the prevalence of ocular findings in patients with all forms of CHD; threeReference Raczyńska, Potaz and Aleszewicz-Baranowska 10 , Reference Eisalo, Raitta, Kala and Halonem 14 , Reference Johns, Johns and Feman 15 only in cases with coarctation of the aorta; and one studyReference Petersen and Rosenthal 9 only in patients with cyanotic heart disease. The characteristics of included studies are shown in Table 2.
Flow chart of the studies included in this review.
Characteristics of the studies included in this review.
Assessment of risk of bias
To assess the methodological quality of the studies, the criteria for experimental studies were excluded from the score, as no study with this design was selected. Thus, the maximum score for the 17 items was 12 points (minimum 7 and maximum 12, mean 9.3 points). The average score in the included studies was 7.25/12 (SD±1.83) possible points (75% with 6 points, 12.5% with 7, and 12.5% with 8 points) (Table 3).
Risk of bias of included studies.
Prevalence of ocular findings
The lowest prevalence found in the included studies was of 6.3%Reference Fusco, Magli and Pantaleo 8 and the highest was 65%.Reference Petersen and Rosenthal 9 The average sample-size weighted prevalence was 32.5% (CI95% 19.3–49.3%) (Table 2 and Fig 2).
Prevalence of ocular findings in children and adolescents with CHD.
Studies with all forms of CHD
The prevalence of ocular manifestations in patients with CHDs not associated with syndromes was 6.6–32.6%.
AlfanoReference Alfano 12 evaluated 500 sequential cases of children with CHD not associated with rubella, aged between 6 days and 15 years. Excluding patients with genetic syndromes, with Down or Hurler syndrome, ocular manifestations were detected in 52 children (10.4%). Scleral cyanosis was observed in 32% of them. Although the prevalence of cataract (0.50%) and glaucoma (0.40%) was lower than rubella, it was much higher than that in the general population. Strabismus was present in ~1.6% of the patients, microphthalmia in 0.8%, ptosis in 0.4%, coloboma in 0.8%, and obstruction of the lacrimal ducts in 0.4% of the cases. The type of heart disease presented by the patients or the routine exams used was not described in this study.
Gardiner and JosephReference Gardiner and Joseph 13 analysed 85 children with CHDs (age with a range between 6 and 15 years) seen within a period of 12 months. The global prevalence of non-refractive ocular findings was 17.6% (number of cases: strabismus, 12; cataract, 2; coloboma of choroid, 1). Ocular alterations were observed in 19 of 22 cyanotic children (86%), 18 of 28 obstructive patients (64%), and 17 of 35 children with shunts (49%). The frequencies of amblyopia and of ametropia were 23.5 and 36%, respectively.
Fusco et alReference Fusco, Magli and Pantaleo 8 examined 76 patients, aged between 2 and 39 years, with CHD. Fundoscopic alterations were found in 6.3% of the patients, precisely in three cases with cyanosis and in two of the patients with aortic stenosis. Symptoms observed in these five patients included vascular dilatation with bluish-coloured blood vessels and, in only one patient with Fallot (1.3%), optical disc oedema. After surgery to correct heart defects, functional changes were seen in 18.4% of campimetry and 37.7% of chromatic tests.
Mansour et alReference Mansour, Bitar and Traboulsi 1 studied 240 cases (mean age 2.9 years; SD=4.2 years). The most common cardiac abnormalities were atrial or ventricular septal defects (n=62), tetralogy of Fallot (n=39), pulmonary stenosis (n=25), and transposition of the great arteries (n=24). Other forms found were as follows: patent arterial duct (n=13), double right ventricle (n=11), aortic stenosis (n=10), pulmonary atresia (n=10), coarctation of the aorta (n=9), single ventricle (n=6), and atrioventricular canal defect (n=6). Exclusion of patients with genetic syndromes reduced the sample to 135 cases (56.25%), with a prevalence of ocular findings of 32.6%. In this specific group, the ocular findings were as follows: ptosis (0.75%), congenital cataracts (0.75%), strabismus (1.48%), retinal haemorrhages (4.4%), optic nerve hypoplasia (9.6%), and retinal arterial and venous tortuosity (15.5%). The prevalence of ptosis, strabismus, and cataract was significantly higher in cases associated with syndromes, but the distribution of results depending on the specific type of anomaly among patients with CHDs was not described.
Patients with aortic coarctation
In studies evaluating patients with coarctation of the aorta, the prevalence of ocular abnormalities ranged from 29.6 to 45%.
In a cross-sectional study, Eisalo et alReference Eisalo, Raitta, Kala and Halonem 14 evaluated by angiofluoresceinography 25 cases with coarctation of the aorta. The patients aged between 17 and 46 years, and 60% of them were male. Mild or severe abnormalities were recognised in 36% of the patients, and included arterial narrowing (20%), arterial and venous tortuosity (24%), and increased capillary visibility (40%).
In addition, in a cross-sectional study, Johns et alReference Johns, Johns and Feman 15 examined 20 cases with coarctation of the aorta. The patients aged between 9 days and 20 years, and 70% of them were male. The prevalence of ocular abnormalities was 45%, with 5% of retinal haemorrhages, 10% strabismus, 15% retinal venous and arterial tortuosity, and 20% retinal arterial tortuosity only.
Raczyńska et alReference Raczyńska, Potaz and Aleszewicz-Baranowska 10 investigated a series of 54 cases (mean age of 16 years) with coarctation of the aorta and arterial hypertension. Ocular abnormalities were observed in 29.6% of the patients, including arteriovenous crossings in 19%, increased dorsal reflex in 21%, increased venous tortuosity in 26%, vascular narrowing in 28%, and anomalous arterial tortuosity in 35% of the cases.
Patients with cyanotic CHD
Petersen and RosenthalReference Petersen and Rosenthal 9 analysed 83 patients aged between 2 and 26 years with congenital cyanotic heart disease in a cross-sectional study. Moderate-to-severe fundoscopic abnormalities were found in 35% of the patients, with 14.45% of them presenting venular dilatation with disc oedema, and 20.5% presenting isolated retinal venular dilation.
Discussion
In this systematic review with meta-analysis, we found that 32.5% (CI95% 6.3–65%) of patients with CHDs not associated with specific syndromes have ocular abnormalities. These alterations were more prevalent and more severe among patients with cyanotic CHDs.
It is possible that cataract, strabismus, colobomas, retinopathy, and amblyopia have higher prevalence among patients with CHDs compared with the normal population.Reference Cheung, Zheng and Hsu 2 , Reference Friedman, Repka and Katz 16 – Reference Silbert, Matla and Silbert 20 Most of the abnormalities are seen in the retina, including venular dilation, arterial and venous tortuosity, arteriolar narrowing, increased arterial dorsal reflex, anomalous arteriovenous connections, micro-haemorrhage, and optic nerve swelling.Reference Cheung, Zheng and Hsu 2 , Reference Tsui, Shamsa, Perloff, Lee, Wirthlin and Schwartz 4 , Reference Vilela, Mielke, Tyllman and Stein 7 , Reference Gillum 21 – Reference Wong and Mitchell 27
The population prevalence of congenital cataract is 1.2–6 cases/10,000 individuals (0.012–0.06%).Reference Lambert and Drack 18 In the studies included in this review, this prevalence ranged from 0.5 to 2.3%. Clearly, this form of cataract was more incident, but the reason is not known. Strabismus was reported in 1.5–14% of patients with CHDs, whereas in the general population its prevalence is 2.1–3.3%.Reference Friedman, Repka and Katz 16 , Reference Silbert, Matla and Silbert 20 This wide range may suggest a possible trend, but there is no definitive evidence for a relationship. The overall prevalence of ocular coloboma is 0.0024%.Reference Alfano 12 , Reference Gardiner and Joseph 13 , Reference Nakamura, Diehl and Mohney 17 AlfanoReference Alfano 12 and Gardiner and JosephReference Gardiner and Joseph 13 detected this abnormality in 0.8–1.1% of the patients with CHDs, but this relationship was not described in the other studies selected in this systematic review. Owing to the lack of population data or to the small sample size, a relationship between CHDs and other conditions described, such as congenital ptosis, microphthalmia, and obstruction of lacrimal duct, was not possible to investigate in the studies.
Retinal changes related to CHDs include increased dorsal arterial reflex, pathologic arteriovenous anastomosis, vascular narrowing or dilatation, increased arterial and/or venous tortuosity, and retinal haemorrhages. Vascular tortuosity was the most commonly described of these findings, ranging from 15.5 to 35% of the patients. An important correlation was observed between presence of coarctation of the aorta and tortuosity (average 26.3%). It is known that this abnormal change in vascular path increases with age, with a relationship with the diameter of the vessel and its transmural pressure.Reference Cheung, Zheng and Hsu 2 , Reference Laste, Tyllmann, Amin and Vilela 28 The pressure elevation initially modifies the vessel’s diameter, but if critical levels are reached the route of the vessel may be affected. It is probable that the retinal findings are directly related to factors such as hypoxia, blood hyperviscosity, and high blood pressure.Reference Gillum 21 , Reference Liu, Mak and Stewart 22 , Reference McQuillen, Goff and Licht 29 Individually or as a group, these factors can induce these changes, and may even affect the capillary network, with transudation and/or occlusion. Many of these signs are biased by the evaluator’s subjectivity and have low diagnostic sensitivity (3–21%). In case of systemic arterial hypertension, however, the findings have high specificity (88–98%). Therefore, hypertensive retinopathy may not be common in patients with chronic systemic arterial hypertension, but is rare in normotensive individuals. The presence of hypertensive retinopathy increases significantly the risk for left ventricular hypertrophy (odds ratio=2.2), which is typical in patients with coarctation. Furthermore, cerebrovascular accident is the only cardiovascular event strongly linked to hypertensive retinopathy.Reference van den Born, Hulsman, Hoekstra, Schlingemann and van Montfrans 24
The information obtained from studies on neurological damage associated with CHDs shows evidence of a delay in the volumetric and functional development, myelination, and formation of convolutions and glial bands in the central nervous system in 20% of the patients.Reference McQuillen, Goff and Licht 29 As the retina has the same embryological origin, and therefore shares many common aspects with the central nervous system, it is possible that in addition to anatomical changes, functional ocular abnormalities are a consequence of CHDs.
Some limitations of this systematic review merit discussion. The small number of studies limited sub-group or sensitivity analyses. As in all systematic reviews, one major concern is the publication bias. As there is no registry of observational studies that could serve as a source of unpublished work in the same format as the clinical trials or systematic reviews registries, there is always a possibility of remaining bias. On the other hand, some methodological strengths could have served as strategies to decrease the possibility of this bias. First, we selected key words and conducted the strategy search to get most sensitive citations selection. Second, the overall references screening and the inclusion/exclusion paper criteria were carried out with no language restrictions, as there is a trend of publishing positive results in English and this may increase publication bias (or English bias). Third, we used all the efforts to locate unpublished studies, performing manual searches in key periodicals and references.
Another possible limitation is the high heterogeneity between studies. Several methodological reasons may help in explaining this heterogeneity. They are related mainly to the type of examination (full or only partial) and resources used for the evaluation. The description of these results depends on the different equipment or techniques used: in the case of fundoscopy, for example, studies with and without mydriasis, using direct or indirect binocular ophthalmoscope, or even through electroretinogram with different fields. Combining all the studies included in this review, a total of 1061 patients were included; however, most samples were small and were collected in very different times relating to surgical or percutaneous correction. This may result in an overestimation of ocular abnormalities in the earliest studies, as associated syndromes may not have been well diagnosed or excluded. The high heterogeneity of results reported in these studies may be related to differences in evaluation methods, as already mentioned, but also to the heterogeneity of CHDs themselves, which form a group of very different entities, both in pathophysiology and in severity.
In summary, different cardiovascular conditions described in the literature are associated with ocular effects. Strong evidence shows that hypertension modifies the state of the retinal vessels from narrowing to occlusion and haemorrhages or severe oedema.Reference Cheung, Zheng and Hsu 2 , Reference van den Born, Hulsman, Hoekstra, Schlingemann and van Montfrans 24 , Reference Wang, Xu and Jonas 25 Heart failure, ischaemic heart disease, cardiac valve disorders, aortic arch disease, and carotid stenosis are conditions where different retinal signals are described.Reference Gillum 21 – Reference Wong, Klein and Sharrett 26 Another possible link is the presence of genetic abnormalities that are associated with cardiac and ophthalmic alterations, even if not resulting in specific syndromes. As CHD is associated with many of these mechanisms, including modifications in perfusion pressure, secondary hypoxia, and thromboembolic events, the presence of ocular signs is to be expected and maybe actively investigated.
In this way, the present systematic review with meta-analysis allows the conclusion that at least one-third of patients with CHDs, not linked to genetic syndromes, present ocular abnormalities. These findings were more prevalent, although not pathognomonic, in patients with cyanotic CHDs. Cataract, strabismus, and retinopathy were the most frequently observed abnormalities, with a prevalence higher than that in the normal population. The identification of these ocular manifestations has diagnostic and prognostic significance, allowing the implementation of more effective strategies to prevent morbidities in adulthood.
Acknowledgement
None.
Financial Support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Conflicts of Interest
None.
Ethical Standards
The authors assert that all work reported complies with the ethical standards of the Helsinki convention, and Ethics Board of the Institute of Cardiology, Porto Alegre, Brazil.
