Purpose
To compare the severity of macular vascular changes in children with sickle cell disease (SCD) vs age- and race-matched controls.
Design
Cross-sectional study.
Methods
Children (<18 years old) with HbSS and HbS variant (HbSC and HbS thalassemia) genotypes, and their age- and race-matched controls, were recruited between January 2017 and December 2018. All subjects underwent optical coherence tomography angiography (OCTA) scans centered on the fovea and temporal macula. Retinal thickness, superficial capillary plexus (SCP) and deep capillary plexus (DCP) vessel density (VD), and foveal avascular zone (FAZ) size were measured and compared between HbSS and HbS variant vs controls.
Results
Thirty-four HbSS, 34 HbS variant (Goldberg staging 0-3 for SCD eyes), and 24 control eyes (total 48 children, aged 5-17 years) were included. Total VD (3-mm ETDRS circle) was lower in HbS variant eyes than in controls for both the SCP (42.9% vs 47.7%, P = .02) and DCP (47.4% vs 52.6%, P = .01). In HbSS eyes, VD was lower in the DCP (47.7%, P = .008) but not in the SCP (45.5%, P = .5), compared to controls. A higher proportion of HbSS (n = 18, 55%) than HbS variant eyes (n = 9, 26%) had pathologic areas of retinal thinning associated with SCP and DCP flow loss ( P = .03). However, retinal thickness measurements and FAZ size did not differ between either HbSS or HbS variant group vs controls.
Conclusions
Children with SCD have similar retinal thickness but less dense vasculature on OCTA compared to age and race-matched controls, suggesting that microvascular insult may precede structural thinning.
Sickle cell disease is an inherited hematologic disorder in which deoxygenated erythrocytes assume an elongated, rigid form. These abnormal, sickle-shaped erythrocytes can disrupt blood flow in small vessels, potentially causing vascular occlusions. As a result of repeated cycles of infarction and inflammation, a broad range of acute and chronic complications involving multiple organs can occur in sickle cell disease. In the eye, microvascular insults result in characteristic retinal changes referred to as sickle cell retinopathy (SCR). In SCR, vaso-occlusions of retinal vessels lead to ischemia and pathologic neovascularization, which can cause vision loss due to vitreous hemorrhage, traction retinal detachment, retinal vascular occlusions, or neovascular glaucoma.
While SCR in adult patients is well described, less is known about the retinal microvascular changes that occur in pediatric patients. Currently, screening guidelines for SCR include dilated eye examinations starting at age 10, with follow-up screening visits every 1- 2 years if no clinically apparent SCR is seen. However, peripheral retinal vascular occlusions have been identified in sickle cell patients as young as 20 months old. A limitation of current screening guidelines includes reliance solely on expert opinion and observational evidence given the lack of randomized clinical trials in pediatric SCR.
Fluorescein angiography (FA) is the current gold standard for evaluation of retinal vascular abnormalities. However, in children, general anesthesia is often required to obtain intravenous access to perform FA. Although ultra-wide-field FA with oral dye administration has been proposed to circumvent the need for intravenous access, FA with oral dye shares some of the same risks as intravenous FA, including nausea, urticaria, and, rarely, anaphylaxis. In comparison, optical coherence tomography angiography (OCTA) is a noncontact and noninvasive imaging technology that allows a detailed 3-dimensional view of retinal and choroidal vascular anatomy without the need for fluorescein dye. These properties make OCTA an appealing imaging modality for children who can cooperate for the tabletop examination.
Our group and others have previously demonstrated that OCTA can detect macular vascular abnormalities in patients with sickle cell disease and that vascular density loss on OCTA correlates to quantitative measures of peripheral ischemia on ultra-wide-field fluorescein angiography. However, most of these studies were performed in adults with sickle cell disease, and there is limited literature on OCTA detection of microvascular abnormalities in children with sickle cell disease. ,
In this study, we report qualitative and quantitative OCTA findings in 67 eyes of 34 children with sickle cell disease (age range 5-17 years). To our knowledge, this represents the largest number of children with sickle cell disease examined with OCTA reported in the literature to date. We evaluate the feasibility of obtaining OCTA images in children as young as 5 years old and compare measures of retinal thickness, vessel density, and foveal avascular zone area in sickle cell patients with age- and race-matched unaffected controls.
Methods
This prospective, observational case series was approved by the Institutional Review Board at Johns Hopkins University and was performed in compliance with the Declaration of Helsinki and the Health Insurance Portability and Accountability Act. Informed written consent was obtained from the parents or legal guardians of all children enrolled in the study.
Consecutive patients younger than age 18 with known electrophoretic confirmation of sickle cell disease (HbSS, HbSC, HbS alpha thalassemia, HbS beta thalassemia) and their age- and race-matched unaffected controls were prospectively recruited at the Wilmer Eye Institute at Johns Hopkins University from January 2017 to December 2018. Exclusion criteria include history of retinopathy from any causes (other than sickle cell disease), high myopia >-6 diopters, lens or media opacity precluding imaging, any history of amblyopia, and any history of other systemic conditions that can cause retinal vascular disease such as prematurity, diabetes, hypertension, infection, neoplasm, or autoimmune disease.
Systemic health information at the time of clinic visit was collected for all sickle cell patients, including average crisis frequency in the previous 2 years as defined by Duan and associates : none, rare (1 episode per year), occasional (2 episodes per year), and frequent (3 or more episodes per year); most recent hemoglobin and fetal hemoglobin values taken in relation to the time of OCTA imaging; current utilization of hydroxyurea, chronic transfusion, or anticoagulation; and any history of pulmonary hypertension (determined by ultrasound), cerebrovascular accident, chronic kidney disease, acute chest syndrome, acute vascular necrosis, or bone marrow transplantation.
All patients underwent best-corrected visual acuity testing using the Early Treatment Diabetic Retinopathy Study chart and complete ophthalmologic examination including dilated ophthalmoscopy by a fellowship-trained retina specialist. At the discretion of the senior investigator (A.W.S.), some of the patients also received ultra-wide-field fluorescein angiograms (Optos 200Tx, Optos PLC, Dunfermline, Scotland, United Kingdom). Each eye was staged based on the available clinical examination and imaging according to Goldberg classification: stage 1 (peripheral arteriolar occlusions), 2 (peripheral arteriovenous anastomoses), 3 (preretinal neovascularization), 4 (vitreous hemorrhage), and 5 (retinal detachment). For the purposes of analysis, eyes were then categorized as having no retinopathy, nonproliferative retinopathy (ie, Goldberg stages 1-2), or proliferative retinopathy (Goldberg stages 3-5).
Macular and temporal paramacular spectral-domain OCT and OCTA (Optovue RTVue XR Avanti; Optovue Inc, Fremont, California, USA) were obtained for all patients. Macular scan protocols included 3 x 3-mm and 6 x 6-mm scans centered on the fovea. Owing to the predilection of temporal paramacular involvement in sickle cell disease, , dedicated 6 x 6-mm scans through the temporal macula with the nasal edge of the scan at or near the fovea were also performed. For eyes where multiple scans were repeated, the images were reviewed by S.S.O., and the best-quality scan was selected for analysis. Poor-quality scans owing to excessive eye movement, blinking, or poor fixation were excluded from analysis. Automated segmentations using the intrinsic OCTA software (RTVue 2017.1.0.151) to analyze retinal microvasculature at the superficial capillary plexus and deep capillary plexus were checked for accuracy and manually corrected when necessary. The automated projection artifact removal software was also applied to every image prior to analysis.
Scans were then qualitatively graded for areas of retinal thinning (defined as focal disruption of retinal contour with undulations of retinal layers) and associated flow loss in the superficial and deep capillary plexus by 2 retina fellowship–trained graders (S.S.O. and I.C.H.). The subfields affected by retinal thinning were also noted. In cases where the 2 graders disagreed, the senior author (A.W.S.) reviewed the scans and arbitrated the grading.
For quantitative OCTA analysis, only images with scan quality 5 (out of 10) or better were included based on previous studies evaluating the effect of signal strength on quantitative metrics. Quantitative measurements were only obtained using 3 x 3-mm and 6 x 6-mm scans centered on the fovea. For the included scans, automated measurements of retinal thickness were obtained for total retina (internal limiting membrane [ILM] to Bruch membrane), as well as inner retina (ILM to inner plexiform layer) and middle retina (inner plexiform layer to outer plexiform layer). Average retinal thickness and vessel density measurements for the superficial capillary plexus (SCP) and deep capillary plexus (DCP) were recorded for the central 1-mm circle; temporal parafovea, superior parafovea, nasal parafovea, and inferior parafovea regions; and total 3-mm ETDRS circle from 3 x 3-mm scans. Foveal avascular zone size was recorded based on measurements taken on 3 x 3-mm scans. Average retinal thickness and vessel density measurements from the temporal perifovea, superior perifovea, nasal perifovea, and inferior perifovea and total 6-mm ETDRS circle were taken from the 6 x 6-mm scans. Figure 1 demonstrates the location where these subfield measurements based on the ETDRS grid are taken. Measurements from the inner ring (central and parafoveal regions) are not collected from the 6 x 6-mm scans, as these regions had been analyzed in the 3 x 3-mm scans and were previously shown to be more reproducibly measured on 3 x 3-mm scans owing to better lateral resolution.
For analyses of demographic, clinical, and ocular characteristics, and qualitative grading of focal thinning and flow loss, the Kruskal-Wallis test was used to compare continuous data while the Fisher exact test was used to compare categorical data, with P < .05 considered statistically significant. For analyses of quantitative measurements, generalized linear mixed models adjusting for age and sex with patients as random effect were used to compare the sickle cell genotypes with their unaffected controls. Statistical analyses were completed using JMP version 13.0 (JMP Statistical Discovery; SAS, Cary, North Carolina, USA) and R.3.5.0. Generalized linear mixed models were performed using R package nlme. The P values were corrected using the Benjamini-Hochberg procedure at a false discovery rate of .05 for multiple comparisons.
Results
Sixty-eight eyes from 34 pediatric sickle cell patients and 24 eyes from 14 unaffected controls met inclusion criteria and were enrolled in the study. Of the sickle cell patients, 17 had been diagnosed with HbSS, 11 with HbSC, 4 with HbS alpha thalassemia, and 2 with HbS beta plus thalassemia. The HbSC, HbS alpha thalassemia, and HbS beta plus thalassemia patients were grouped together as HbS variant (n = 17) for all subsequent analyses. Two of the control subjects were being followed for unilateral idiopathic and traumatic choroidal neovascular membranes, respectively, and only the unaffected, contralateral eyes were included as control eyes. For another 2 control subjects, only 1 eye was included in the study each owing to poor-quality scans from the other eye. There was no difference in age, sex, ethnicity, or mean visual acuity ( Table 1 ) between the HbSS, HbS variant, and control groups.
| HbSS | HbS Variant | Control | P Values | |
|---|---|---|---|---|
| Patients (eyes), n | 17 (34) | 17 (34) | 14 (24) | – |
| Age in years, mean ± SD (range) | 13.24 ± 3.90 (5-17) | 12.76 ± 2.88 (7-17) | 11.43 ± 3.25 (6-16) | .19 |
| Male, n (%) | 8 (47.1) | 5 (29.4) | 8 (57.1) | .30 |
| Black or African descent, n (%) a | 16 (94) | 17 (100) | 14 (100) | 1.0 |
| BCVA, logMAR ± SD | 0.052 ± 0.117 | 0.075 ± 0.122 | 0.011 ± 0.033 | .2 |
Systemic Health Comparisons Between HbSS and HbS Variant Genotypes
The mean hemoglobin in HbSS patients (8.76 ± 1.23 g/dL) was lower than in HbS variant (10.88 ± 1.52 g/dL) patients ( P = .0003). For reference, the lower hemoglobin limit defining anemia for African-American children ranges from 10.6 to 12.9 g/dL in male and female subjects aged 5-17 years. Chronic red blood cell exchange transfusion was more common in HbSS patients than HbS variant (30% vs 0%, P = .04). Similarly, cerebrovascular accidents were more common in HbSS patients than HbS variant (53% vs 0%, P = .0009). There was no difference in frequency of vaso-occlusive crisis, mean fetal hemoglobin, use of hydroxyurea, history of echocardiogram-defined pulmonary hypertension, acute chest syndrome, and bone marrow transplantation between the HbSS and HbS variant groups ( Table 2 ). No patient in either group was on anticoagulation or had a history of diabetes, chronic kidney disease, or avascular necrosis.
| HbSS | HbS Variant | P Values | |
|---|---|---|---|
| Systemic characteristics a | |||
| Crisis frequency, b n (%) | .72 | ||
| None | 7 (41.2) | 4 (23.5) | |
| Rare | 4 (23.5) | 7 (41.2) | |
| Occasional | 1 (5.9) | 1 (5.9) | |
| Frequent | 5 (29.4) | 5 (29.4) | |
| Hemoglobin in g/dL, mean ± SD | 8.76 ± 1.23 | 10.88 ± 1.52 | .0003 |
| Hemoglobin F in g/dL, mean ± SD c | 9.76 ± 6.16 | 14.7 ± 10.66 | .38 |
| Hydroxyurea, n (%) | 10 (58.8) | 6 (35.3) | .30 |
| Chronic transfusion, n (%) | 5 (29.4) | 0 | .04 |
| Anticoagulation, n (%) | 0 | 0 | – |
| EDPHTN, n (%) | 1 (5.9) | 0 | 1.0 |
| CVA, n (%) | 9 (52.9) | 0 | .0009 |
| DM, n (%) | 0 | 0 | – |
| CKD, n (%) | 0 | 0 | – |
| ACS, n (%) | 12 (70.6) | 8 (47.1) | .30 |
| AVN, n (%) | 0 | 0 | – |
| BMT, n (%) | 1 (5.9) | 0 | 1.0 |
| Ocular characteristics d | |||
| Sickle cell retinopathy, n (%) | 0.06 | ||
| None | 18 (53) | 9 (26) | |
| Nonproliferative | 15 (44) | 24 (71) | |
| Proliferative | 1 (3) | 1 (3) | |
| Goldberg staging, e n (%) | 1.0 | ||
| 1 | 2 (13) | 3 (13) | |
| 2 | 13 (81) | 20 (83) | |
| 3 | 1 (6) | 1 (4) | |
| 4 | 0 | 0 | |
| 5 | 0 | 0 | |
| FA performed, n (%) | 16 (47) | 22 (65) | 0.2 |
| Iridescent spots, n (%) | 4 (12) | 3 (9) | 1.0 |
| Black sunbursts, n (%) | 9 (26) | 3 (9) | 0.1 |
| Salmon patch hemorrhage, n (%) | 3 (9) | 4 (12) | 1.0 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
