Choroidal Thickness in Patients With a History of Retinopathy of Prematurity

Importance  The cause of reduced vision in patients with a history of retinopathy of prematurity (ROP) is not yet fully understood. The role of the choroid in ROP remains unknown and existing studies of choroidal thickness in patients with a history of ROP are limited. It might be helpful to understand the association of the choroid with ROP by measuring the choroidal thickness in patients with a history of ROP and correlating these findings with the visual outcome of these patients.

Objective  To examine choroidal thickness by spectral-domain optical coherence tomography in children with a history of ROP and assess the impact of choroidal thickness on visual acuity.

Design  A prospective cross-sectional analysis from August 2011 to September 2012.

Setting  Institutional referral centers.

Participants  Children aged 6 to 14 years with a history of ROP were classified into the following 2 groups: patients with a history of threshold ROP and treatment with laser or cryotherapy (treated group) and those with regressed ROP who had not received any treatment (nontreated group). All of the patients had a normal-appearing posterior pole.

Intervention  Examinations of visual acuity, refractive errors, and optical components and measurement of choroidal thickness.

Main Outcomes and Measures  Best-corrected visual acuity, optical components, and optical coherence tomography findings.

Results  In total, 49 patients were enrolled in the study. Patients in the treated group had a significantly thinner choroidal thickness than the patients in the nontreated group after adjusting for age, axial length, and spherical power. Choroidal thickness was found to be positively associated with spherical power and spherical equivalent and negatively associated with axial length and vitreous depth. In addition, a thin choroidal thickness was associated with a worse best-corrected visual acuity.

Conclusions and Relevance  Choroidal thickness is thinner in patients with threshold ROP compared with the patients with spontaneously regressed ROP. A thinner choroid is associated with worse vision in these patients. This study might imply the association of choroid circulation with ROP.

Retinopathy of prematurity (ROP) is one of the leading causes of childhood blindness. Children with a history of ROP have a lower visual acuity (VA) than full-term children despite a normal-appearing fundus. Our prior study using spectral-domain optical coherence tomography (SD-OCT) in patients with a history of ROP showed a thicker foveal thickness, higher incidence of absence of foveal depression, and retaining of inner retinal layers in patients with ROP compared with age-matched full-term children.¹ These changes are associated with poor VA and high refractive errors in these patients.¹

Existing studies on choroidal thickness in patients with ROP are limited. Prior studies indicate that the choroid is the prominent supplier of oxygen and nutrients to the outer retina, namely the photoreceptor layer.² The choroid begins to develop before the retinal vasculature, although its maturation continues to progress after birth.³ Choroidal vascular involution can cause decreased oxygen and nutrient delivery to the outer retina. The decreased perfusion can lead to the loss of overlying photoreceptors and affects photoreceptor signal transduction. Animal studies of oxygen-induced retinopathy have shown that deficient vascularity occurs in the retinal plexus and the choroid.⁴ This sustained and marked choroidal degeneration is specifically confined to central regions of the retina and presents as persistent photoreceptor loss with corresponding functional deficits.⁴ The depth-enhancing feature of SD-OCT allows the measurement of choroidal thickness. Park and Oh⁵ were the first to use SD-OCT to measure choroidal thickness in patients with ROP. They found that the choroidal thickness 3.0 mm temporal to the fovea in preterm children was significantly less than that in full-term children. In this study, we used SD-OCT to examine choroidal thickness in formerly preterm children with no sequela of regressed ROP. We compared the choroidal thickness between patients with a history of threshold ROP who had laser treatment or cryotherapy (treated group) and those with regressed ROP without any treatment (nontreated group). The relationship between the choroidal thickness and foveal thickness, refractive errors, and optical components was then analyzed. Finally, the effect of variable choroidal thickness on vision was calculated.

This study was approved by the institutional review board at Chang Gung Memorial Hospital and was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from each of the participant’s parents or guardians.

This was a prospective cross-sectional study that included children with a history of ROP who were followed up at our hospital or Mackay Memorial Hospital. All of the patients were Asian. All of the eyes had a normal-appearing fundus. Patients were excluded from the study if they had a history of cerebral damage or severe congenital defects preventing their cooperation in the tests. Patients with residual abnormalities of ROP, including macular dragging, macular fold, and retinal detachment, were excluded. All of the participants were free from other ocular diseases. Both eyes in each participant were examined, but only 1 eye was randomly selected for the data analysis.

Patients were classified into the following 2 groups: those with a history of threshold ROP who had treatment with laser or cryotherapy (treated group) and those with regressed ROP without any treatment (nontreated group).

Bare VA and best-corrected VA (BCVA) were measured. Automatic cycloplegic refraction by an automatic kerato-refractometer (KR-8100A; Topcon) was performed first, and manual refraction was then performed to refine the outcome. Photographs of the fundus were taken with a retinal camera (CR-DGI; Canon Inc). The optical components, including axial length, anterior chamber depth, lens thickness, and vitreous depth, were also measured (Cine Scan; Quantel Medical SA) and compared between the groups of patients. Snellen VA was converted to the logMAR VA for the statistical analysis. Cutoff VA of 20/25 was chosen because VA less than 20/25 was considered to reach the level of amblyopia and required treatment.⁶ Medical data, including the gestational age, birth weight, stage of maximal severity in the acute disease, presence of neurologic events, and significant complications developed in the neonatal stage, were recorded. The type of peripheral ablative procedure, such as laser treatment or cryotherapy, was also recorded. The surgical indication was determined according to the threshold disease defined in a previous Cryotherapy for Retinopathy of Prematurity study.⁷

Retinal and choroidal biometry were measured with an SD-OCT system (RTVue; Optovue Inc) using previously described protocols.⁸ Briefly, the choroid was imaged in the “choroidal mode,” and its thickness was defined as the distance between the outer border of the retinal pigment epithelium to the hyperrefractive line behind the large vessel layers of the choroid, which was presumed to be the choroid-sclera interface. The choroidal thickness under the fovea was measured manually by 2 independent masked observers using the scale supplied by the software. The results were averaged to exclude the possible bias associated with the measurements. The thickness was measured at 1000-µm intervals from the fovea to a distance of 3 mm in the nasal, temporal, superior, and inferior directions.⁸ The average of 14 choroidal thickness readings was recorded as the macular choroidal thickness. Subfoveal choroidal thickness was defined as the choroidal thickness measured at the center of the foveola. In addition, separate calculations of choroidal thickness were made for the nasal, temporal, superior, and inferior directions. The determination of macular and subfoveal choroidal thickness is shown in Figure 1. The foveal thickness of the patients was also measured according to our previous protocol.¹

The receiver operating characteristic (ROC) curve is a plot of sensitivity against 1 minus specificity for the different cutoff points of a parameter.⁹ We performed ROC analysis to reveal the predictive power of choroidal thickness on patients’ BCVA. The line above the reference line was considered to be predictive of the BCVA. The value of the Youden index [maximum (sensitivity + specificity − 1)] to yield optimal sensitivity and specificity on the ROC curve was calculated. Model ability to distinguish between good and poor vision of the patients was tested by using area under the curve (AUC).¹⁰ The power of these variables to accurately predict BCVA was calculated as the AUC and the 95% confidence interval of the AUC. An AUC of 1.0 represented perfect discriminatory performance, whereas an AUC of 0.5 was thought to be due to random assignment. A higher value of an AUC represented a better discriminative performance.

Continuous variables were displayed as means (standard deviations), and categorical variables were displayed as frequencies and percentages. The Shapiro-Wilk test was used to test the normality of the continuous variables. The χ² test was used to compare the differences in categorical variables between the 2 groups. The 2-sample t test and Wilcoxon 2-sample test were used to compare the differences between continuous variables for those who did and did not follow a normal distribution, respectively. Because of multiple comparisons, the P values of comparing choroidal thickness in different locations were adjusted by Bonferroni correction. The correlation between macular and subfoveal choroidal thickness and choroidal thickness and other studied variables was estimated using Pearson correlation or Spearman correlation coefficients. The logistic regressions were used to determine the association between different ranges of macular choroidal thickness or subfoveal choroidal thickness and the BCVA. The results were expressed as odds ratios after adjusting for age, axial length, and spherical power.

The SAS software version 9.2 (SAS Institute Inc) was used for the data analyses. A P value less than .05 was considered to be statistically significant in this study.

A total of 49 patients met the screening criteria and completed the examinations. The numbers of patients in the treated and nontreated groups were 35 and 14, respectively. The mean (SD) gestational age in the treated and nontreated groups was 26.0 (1.8) and 28.7 (2.0) weeks, respectively. The mean (SD) birth weight in the treated and nontreated groups was 913.0 (231.2) and 1135.0 (266.0) g, respectively. The mean gestational age and mean birth weight for the treated group were significantly less than the nontreated group (P < .001 and P = .006). The mean (SD) age of the patients was 9.5 (2.9) years and 10.1 (2.9) years in the treated and nontreated groups, respectively; this difference was not statistically significant (P = .51). There was no statistically significant difference in the sex distribution between the groups (P = .32).

The BCVA for patients in the treated group was significantly worse than the BCVA for patients in the nontreated group (P = .03). The ratio of a BCVA worse than 20/25 in the treated group was higher than the nontreated group, but this did not reach statistical significance (P = .20). The mean (SD) spherical power of the patients in the treated and nontreated groups was −3.5 (4.8) and −1.1 (2.2) diopters, respectively (P = .11). The mean (SD) cylindrical power of the patients from the treated and nontreated groups was −2.2 (1.5) and −1.8 (1.3) diopters, respectively (P = .32). These data showed that a greater spherical equivalent was observed in the treated group (−4.7 [5.0] diopters) compared with the nontreated group (−2.0 [2.3] diopters), but this difference did not reach statistical significance (P = .08). In the optical component analysis, there was no significant difference in the corneal curvature between these 2 groups of patients. The anterior chamber depth was significantly shallower for the patients in the treated group (P = .007). The lens thickness was also significantly thicker in patients in the treated group than patients in the nontreated group (P < .001). However, no significant difference was found between the groups for vitreous depth (P = .85) or axial length (P = .59). The demographic and optical component data of the patients are shown in Table 1.

Both eyes in each participant were examined by SD-OCT, the results were found to be similar between the right and left eyes (data not shown), and therefore, only 1 eye was randomly selected for the data analysis. In the SD-OCT analysis, macular choroidal thickness was significantly lower in patients in the treated group after adjusting for age, axial length, and spherical power (mean [SD], 196.5 [61.5] vs 238.3 [46.1] μm for the treated and nontreated groups, respectively; P = .04) (Figure 2 and Table 2). Similarly, subfoveal choroidal thickness was significantly lower in patients in the treated group after adjusting for age, axial length, and spherical power (mean [SD], 210.0 [74.3] and 261.1 [62.0] μm for the treated and nontreated groups, respectively; P = .047). Foveal thickness was found to be thicker in patients in the treated group (mean [SD], 281.7 [21.5] and 270.4 [14.3] μm for the treated and nontreated groups, respectively), but the difference did not reach statistical significance (P = .08). Patients in the treated group did not show a significant difference in parafoveal thickness compared with the patients in the nontreated group; the difference in perifoveal thickness was marginally significant (P = .054). The results are shown in Table 2. In further analysis, choroidal thicknesses in the superior and inferior quadrants of the treated group were significantly thinner than the nontreated group even after adjusting for Bonferroni correction (P = .04 and .02, respectively). The choroidal thicknesses in other quadrants were not significantly different between the treated and nontreated groups. The results are shown in Table 3. Twenty-nine patients received laser treatment and 6 patients received cryotherapy, but no significant difference in these SD-OCT measurements was found in the patients with laser treatment or cryotherapy (data not shown).

The macular choroidal thickness was found to be positively associated with spherical power (r = 0.66; P < .001) and spherical equivalent (r = 0.65; P < .001) and negatively associated with axial length (r = 0.72; P < .001) and vitreous depth (r = 0.68; P < .001) (Figure 3). Similarly, subfoveal choroidal thickness was found to be positively associated with spherical power (r = 0.64; P < .001) and spherical equivalent (r = 0.64; P < .001) and negatively associated with axial length (r = 0.69; P < .001) and vitreous depth (r = 0.63; P < .001). Age was not found to be associated with choroidal thickness, and other factors in the optical components or the foveal thickness were also not found to be related to choroidal thickness. The correlation between the macular and subfoveal choroidal thickness was highly significant (r = 0.98; P < .001).

We analyzed the predictive power of choroidal thickness on VA using the ROC curve. The AUCs of macular choroidal thickness and subfoveal choroidal thickness were 0.69 (95% CI, 0.54-0.85) and 0.71 (95% CI, 0.55-0.86), respectively. The data are shown in Figure 4. The Youden index was determined to be 168.7 µm and 176.9 µm for macular choroidal thickness and subfoveal choroidal thickness, respectively. Multivariate analysis further determined the association between different ranges of macular choroidal thickness and the BCVA. The value of the Youden index calculated from the ROC curve was 168.7 µm for macular choroidal thickness. Therefore, the range of macular choroidal thickness was divided into 2 groups: 168.7 µm or less and greater than 168.7 µm. Comparing the patients with choroidal thickness more than 168.7 µm, the odds ratio of BCVA worse than 20/25 in patients with macular choroidal thickness of 168.7 µm or less was 38.3 (95% CI, 3.1-468.6) in multivariate analysis. It reached statistical significance even after adjusting for age, axial length, and spherical power (P = .004). This result showed correlation of patients’ BCVA with macular choroidal thickness. Similarly, comparing patients with subfoveal choroidal thickness more than 176.9 µm, the odds ratio of BCVA worse than 0.8 in patients with subfoveal choroidal thickness of 176.9 µm or less was 22.1 (95% CI, 2.2-220.8) in multivariate analysis. The results were statistically significant (P = .008). These data are shown in Table 4.

The study of choroid in patients with ROP has been limited, and its association with ROP remains unclear. In this prospective cross-sectional study, we found that patients with a history of threshold ROP had a significantly thinner choroidal thickness than the patients with a history of ROP and spontaneous regression after adjusting for age, axial length, and spherical power. Choroidal thicknesses in the superior and inferior quadrants were especially thinner in patients with a history of threshold ROP. The choroidal thickness was found to be positively associated with spherical power and spherical equivalent and negatively associated with vitreous depth and axial length. The predictive power of macular and subfoveal choroidal thickness on vision was moderate (the AUC: 0.69 and 0.71, respectively). Finally, a thin choroidal thickness was negatively associated with BCVA on both univariate and multivariate analyses. These results highlight the association of choroidal thickness, which translates into choroidal circulation, and ROP. Prior studies have shown that prematurity not only affects the anterior segment of the patients¹ but it also affects the retina,^1,11 optic nerve,¹² and visual cortex.¹³ The current study highlights the association of ROP with a thinner choroid.

Vrabec et al¹⁴ reported extensive atrophy of the choroidal vasculature in a child with threshold ROP who received cryotherapy. Park and Oh⁵ showed that only the area 3.0 mm temporal to the fovea in preterm children had a choroidal thickness that was significantly less than full-term children. The choroidal thickness in the other areas was not significantly different between the children with ROP and full-term children.⁵ Our study focused on patients with a history of ROP only. Patients with a history of threshold ROP had a significantly thinner choroidal thickness than the patients with a history of ROP and spontaneous regression. We also found an association between poor vision and a thinner choroid in a cohort of patients with ROP.

The mechanism of choroid involution in ROP remains to be investigated. Nitric oxide has been shown to play a role in the autoregulation of choroidal blood flow through its interactions with prostaglandins.³ Shao et al⁴ suggested that 15-deoxy-△12,14-prostaglandin J2 (15d-PGJ2; a nonenzymatic product of prostaglandin D₂) might be a prominent contributor to the choroidal decay in ROP. This key mediator might cause choroidal vaso-obliteration, thereby affecting choroidal thickness through the induction of oxidant stress in ROP.⁴ In addition, the application of laser treatment or cryotherapy to the eyes could affect choroid circulation.^15-17

Choroidal thickness has been found to be a good indicator of the development and severity of lacquer cracks in patients with high myopia.⁸ In addition, macular choroidal thickness has been suggested to be a better predictor of BCVA than axial length or refractive error in patients with dry-type myopic maculopathy.⁸ The changes in the choroidal thickness also help determine the pathologic processes of various retinal diseases, such as central serious chorioretinopathy,^18-20 uveal effusion syndrome,²¹ Vogt-Koyanagi-Harada disease,^22,23 and age-related macular degeneration.^24,25Although an indocyanine green angiogram could be used as a choroidal vasculature study, it is an invasive study for these patients.^26,27 Spectral-domain optical coherence tomography enables us to study the choroid noninvasively. At this point, the determination of choroidal thickness remains a time-consuming manual process because commercially available SD-OCT equipment cannot perform this task automatically. Therefore, some variation might exist in each measurement. To exclude the possible bias associated with the measurements, the choroidal thickness was measured by 2 masked observers and the results were averaged.

Choroidal thickness is known to be inversely associated with axial length, spherical power, and age.^28-30 Therefore, in this study, choroidal thickness was adjusted for axial length, spherical power, and the age of the patients to avoid the contributing effect of these factors. However, age was not found to be closely related to choroidal thickness in the current study because all of our patients were in a similar age range. In our study, we measured both macular and subfoveal choroidal thickness. The macular choroidal thickness, which was an averaged value from 14 choroidal thickness readings made at 1000-µm intervals from the fovea to a distance of 3 mm in the nasal, temporal, superior, and inferior directions, might better represent the generalized choroidal thickness in the macula area because of the regional variation in the choroidal thickness.^29,31,32Subfoveal choroidal thickness, on the other hand, measures only the subfoveal choroidal thickness. These 2 values show a statistically significant correlation and yield similar results in our study. Therefore, if the measurement of macular choroidal thickness is not easy to obtain, subfoveal choroidal thickness, which is easier to perform, could be used. In addition, recent studies suggest choroidal thickness measurements obtained with different SD-OCTs were highly correlated and could be used interchangeably.^33-35

To our knowledge, this is the first study that compares choroidal thickness in patients with ROP requiring treatment and patients with ROP with spontaneous regression. Our study has shown the involvement of the choroid in ROP and its impact on the vision of patients with ROP. However, this study is limited by a small case number and the fact that the measurement of choroidal thickness is manual and may contain slight errors. Nevertheless, this is the best clinical method currently available for measuring the choroidal thickness noninvasively. A larger study with more patients with ROP and long-term follow-up is necessary in this emerging field.

In conclusion, patients with a history of threshold ROP had a significantly thinner choroidal thickness than the patients with a history of ROP and spontaneous regression. The choroidal thickness was positively associated with spherical power and spherical equivalent and negatively associated with vitreous depth and axial length. Thinner choroidal thickness is associated with worse vision in patients with ROP. These results have shown the impact of the choroidal circulation on ROP. Photoreceptors or retinal abnormalities in ROP^1,36-38 could be secondary changes after the choroidal alteration.

Corresponding Author: Wei-Chi Wu, MD, PhD, Department of Ophthalmology, Chang Gung Memorial Hospital, No. 5, Fu-Hsin Road, Taoyuan, Taiwan 333 (weichi666@gmail.com).

Submitted for Publication: February 23, 2013; final revision received May 2, 2013; accepted May 7, 2013.

Published Online: September 26, 2013. doi:10.1001/jamaophthalmol.2013.5052.

Author Contributions: Drs Wu and Shih had full access to all of the data in the study and take responsibility for the integrity of the data and accuracy of the data analysis.

Study concept and design: Wu, Lien, Chao, T.-L. Chen, Hwang, Lai, Huang.

Acquisition of data: Wu, Wang, Lien, Y.-P. Chen, Chao, K.-J. Chen, Hwang, Lai, Tsai.

Analysis and interpretation of data: Wu, Shih, Chao, Hwang, Lai.

Drafting of the manuscript: Wu, Chao, Hwang, Lai, Huang, Tsai.

Critical revision of the manuscript for important intellectual content: Wu, Shih, Wang, Lien, Y.-P. Chen, Chao, K.-J. Chen, T.-L. Chen, Hwang, Lai.

Statistical analysis: Wu, Shih, Chao, Hwang, Lai.

Obtained funding: Wu, Chao, Hwang, Lai.

Administrative, technical, or material support: Wang, Y.-P. Chen, K.-J. Chen, T.-L. Chen, Hwang, Huang.

Study supervision: Wu, Lien, Chao, Hwang, Lai.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was partially supported by Chang Gung Memorial Hospital Research Grant CMRPG3A0391 and National Science Council Research Grant NSC98-231-B-182A-001-MY2.

Role of the Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.