Retinal and Choroidal Changes With Severe Hypertension and Their Association With Visual Outcome

Abstract

Purpose.: To investigate retinal and choroidal changes using spectral-domain optical coherence tomography (SD-OCT) and to evaluate visual outcome in patients with severe hypertension.

Methods.: In 42 eyes with hypertensive retinopathy from 21 patients with severe hypertension (systolic blood pressure [SBP] ≥ 180 mm Hg or diastolic blood pressure [DBP] ≥ 110 mm Hg), SD-OCT was performed on the day blood pressure (BP) was measured. Best-corrected visual acuity (BCVA), fundus features, and SD-OCT morphologic findings were evaluated and OCT features were compared between baseline and final visits. Associations between clinical findings, OCT image features, and Keith-Wagener-Barker (KWB) hypertensive retinopathy grades and baseline and final BCVA were examined.

Results.: Optical coherence tomography findings included macular edema (central macular thickness > 300 μm), irregularly reflective regions, retinal nerve fiber layer thickening, subretinal fluid (SRF), intraretinal fluid, and intraretinal hyperreflective dots. All abnormalities rapidly resolved with BP control except for the hyperreflective dots. Central macular thickness, subfoveal choroidal thickness (SCT), and SRF height significantly decreased following BP control. Subretinal fluid height was significantly correlated with baseline BCVA, final BCVA, and SCT. Based on fundoscopic and OCT features, eyes were classified as showing mild to moderate retinopathy, malignant retinopathy without SRF, and malignant retinopathy with SRF. Unlike KWB grades (P = 0.077), the OCT-based retinopathy grades were significantly correlated to final BCVA, as shown by linear regression analyses (P = 0.025).

Conclusions.: Severe hypertension resulted in exudative retinal and choroidal changes, characterized by SRF accumulation and increased SCT. Our grading system may represent retinopathy severity and predict visual outcome better than the KWB grading system in patients with severe hypertension.

Introduction

Hypertension is one of the most common adult conditions in industrialized countries, with over 65 million Americans having hypertension.^1,2 Hypertension accounts for the largest proportion of cardiovascular deaths in the United States.³ In the seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure, hypertension is defined as a systolic and/or a diastolic blood pressure consistently higher than the accepted norm (systolic: ≤139 mm Hg, diastolic: ≤89 mm Hg).⁴ In the 2007 European Society of Hypertension–European Society of Cardiology guideline⁵ and the report of the fourth working party of the British Hypertension Society,⁶ severe hypertension (stage III hypertension), also termed malignant hypertension, was defined as a systolic blood pressure (SBP) exceeding 179 mm Hg or a diastolic blood pressure (DBP) exceeding 109 mm Hg. 

Hypertension is associated with vascular abnormalities in the brain, heart, kidneys, and eyes. Hypertension may cause retinal hemorrhages, cotton wool spots, intraretinal lipid accumulation, and vessel closure in the retinal capillaries and choriocapillaris. Based on these fundoscopic features, Keith et al.⁷ developed a classification system for hypertensive retinopathy to categorize these signs into four groups of severity. This classification is widely used in current clinical practice, but Wong and Mitchell⁸ most recently proposed a simplified hypertensive retinopathy grading system with the following classifications: none, mild, moderate, and malignant retinopathy (Table 1). 

Table 1

 

Classification Systems for Hypertensive Retinopathy

Classification	Category	Funduscopic Findings	OCT Findings*
Keith-Wagener-Barker	1	Mild generalized arteriolar narrowing or sclerosis	Not applicable
	2	Definite focal narrowing and arteriovenous crossings; moderate to marked sclerosis of the retinal arterioles
	3	Signs of grade 2 retinopathy plus retinal hemorrhages, exudates, and cotton wool spots
	4	Severe grade 3 retinopathy plus papilledema
Wong and Mitchell9	Mild	1 or more of the following signs: generalized arteriolar narrowing, focal arteriolar narrowing, arteriovenous nicking, arteriolar wall opacity	Not applicable
	Moderate	1 or more of the following signs: retinal hemorrhage (blot-, dot-, or flame-shaped), microaneurysm, cotton wool spot, hard exudates
	Malignant	Moderate retinopathy plus optic disc swelling
Present study†	Mild to moderate	KWB grade 1, 2, or 3	With/without SRF
	Malignant without SRF	KWB grade 4	No SRF
	Malignant with SRF	KWB grade 4	SRF

*  OCT performed on the same day as blood pressure measurement (the day when severe hypertension was diagnosed).

†  For patients with severe hypertension (systolic blood pressure ≥ 180 mm Hg or diastolic blood pressure ≥ 110 mm Hg).

In patients with severe hypertension, ocular changes can be striking and include optic neuropathy, choroidopathy, and retinopathy.^9,10 Hypertensive retinopathy in severe hypertension has been investigated in several studies^10–13; but hypertensive choroidopathy has been documented in only a few case reports, and the mechanism by which choroidal damage occurs remains relatively unknown. 

Optical coherence tomography (OCT) has revolutionized our ability to visualize structural abnormalities in the retina. In comparison with older time-domain OCT (TD-OCT) technologies, spectral-domain OCT (SD-OCT) provides significantly better image resolution, allowing detailed images of retinal morphologic features to be obtained. Morphologic abnormalities associated with severe hypertension, as identified with TD-OCT, include subretinal fluid (SRF) and macular edema.¹³ However, choroidal abnormalities have not been investigated with conventional OCT due to poor light penetration past the retinal pigment epithelium (RPE). Relatively new enhanced depth imaging (EDI) techniques allow the choroid to be visualized on a microscopic level, so this is now technologically feasible. 

The purpose of this study was to use EDI-OCT to investigate retinal and choroidal features of ocular changes occurring in patients with severe hypertension. We investigated associations between blood pressure (BP) levels and the quantitative/qualitative ocular changes as identified with OCT. By comparing retinal and choroidal images before and after BP control, associations between elevated BP and pathologic findings were examined. Visual outcomes and prognostic factors were also investigated. Finally, we developed a novel classification system based on both fundoscopic and OCT findings to grade hypertensive retinopathy in patients with severe hypertension. The association between retinopathy grading and visual prognosis was also evaluated for both the new and conventional classification systems. 

Methods

Patients

A retrospective chart review was performed on 48 eyes of 24 patients. All patients had hypertensive retinopathy diagnosed between May 1, 2009, and July 30, 2012, and complained of blurred vision, with or without headaches, within 2 weeks of the initial visit. All included patients had normal visual acuities in both eyes before the symptom onset. All patients had a systolic BP ≥ 180 mm Hg or a diastolic BP ≥ 110 mm Hg at the initial visit, at which SD-OCT (Spectralis OCT; Heidelberg Engineering, Inc., Heidelberg, Germany) was performed. Patients with history of ocular trauma, macular disease, diabetic retinopathy (n = 1), ocular surgery, or high myopia (>6 diopters [D], n = 1) were excluded. Patients with a follow-up period shorter than 2 months (n = 1) were also excluded from analyses. 

The study conduct adhered to the tenets of the Declaration of Helsinki, and the study protocol was approved by the Institutional Review Board of Seoul National University Bundang Hospital. Patients had their hypertension medically managed (i.e., oral [n = 3] or intravenous [n = 18] antihypertensive medications) and had their BP and heart monitored in the emergency department. 

Ophthalmic Examinations

All patients underwent a complete ophthalmic examination, which included best-corrected visual acuity (BCVA) measurement, slit-lamp biomicroscopy, indirect ophthalmoscopy, fundus photography (VX-10; Kowa Optimed, Tokyo, Japan), fluorescein angiography (FA), and SD-OCT imaging. Fundus photographs were used to grade hypertensive retinopathy according to the Keith-Wagener-Barker (KWB) classification system.⁷ Based on fundoscopic features and the presence of SRF, eyes were classified into mild to moderate retinopathy (KWB grade 1, 2, or 3), malignant retinopathy without SRF (KWB grade 4 without SRF), and malignant retinopathy with SRF (grade 4 with SRF), and the grades were correlated with baseline and final BCVA. Hypertensive retinopathy classifications are presented in Table 1. 

The SD-OCT scans with eye-tracking system were performed at a scan rate of 40,000 A-scans/s over a 4.5- × 6.0-mm area.¹⁴ The macular thickness protocol acquires a 512 × 128 scan macular cube, which was used for quantitative evaluation of macular thickness. Central macular thickness (CMT) was measured within the innermost central circle (1000 μm in diameter) of the macular thickness scan. The height of the fluid space was manually measured at its thickest point using the OCT system's built-in calipers. Full-thickness choroidal images were obtained using EDI-OCT with eye-tracking and image-averaging systems as previously described.¹⁵ Scan interpretations and measurements were performed by two independent investigators (SJA, SJW), both of whom were masked to clinical information. For quantitative OCT evaluations, the average of each investigator's measurements was used in data analyses. 

Subfoveal choroidal thickness (SCT) was manually measured with calipers as the distance between the outer border of the RPE to the inner surface of the sclera.^16,17 In two eyes, SCT could not be measured due to poorly demarcated posterior border of the choroid. With the exception of these eyes, good interobserver agreement on SCT measurements was obtained (intraclass correlation coefficient [ICC] = 0.952; 95% confidence interval, 0.896–0.978). 

Baseline OCT images were compared to those obtained 1 month later and at the final visit to identify quantitative and qualitative changes in the retina and choroid over time. In addition, OCT images were used to evaluate the structural integrity of the retinal layers and any abnormal retinal or choroidal features. 

Data Analyses

Data for continuous variables are expressed as mean ± standard deviation, where applicable. Visual acuity measurements were converted to the logarithm of the minimum angle of resolution (logMAR) for all analyses. Mean arterial pressure (MAP) was calculated from SBP and DBP measurements. Linear regression analyses were performed to evaluate the association of SBP, DBP, and MAP with CMT, SRF height, and choroidal thickness. For clinical parameters that were significantly associated with choroidal thickness, multiple regression analyses were performed to control the effect of age. This was done because choroidal thickness may be dependent upon subject age.^18,19 Frequency and incidence data were compared using a χ² test or Fisher's exact test. 

Grades of hypertensive retinopathy were examined for correlations with baseline and final BCVA. To identify associations with visual outcomes, univariate and multivariate stepwise regression analyses were performed. To identify clinical and OCT features significantly and independently associated with incomplete anatomic recovery, multivariate logistic regression analysis was performed with backward elimination. Statistical analyses were performed using SPSS for Windows (Ver. 18.0 Statistical Package for the Social Sciences; SPSS, Inc., Chicago, IL, USA), and P values < 0.05 were considered statistically significant. 

Results

Clinical, Fundoscopic, and Angiographic Characteristics at Baseline

This study included 42 eyes of 21 patients (10 male, 11 female), who were diagnosed with severe hypertension and were examined with SD-OCT on the day of initial BP measurement. Patient clinical characteristics are summarized in Table 2. Mean patient age was 41.8 ± 11.1 years (range, 23–62 years), and patients were followed for an average of 8.8 ± 10.5 months (range, 2–36 months). At baseline, mean systolic and diastolic BP were 223 ± 32 and 133 ± 27 mm Hg, respectively, and average MAP was 163 ± 27 mm Hg. At the 1-month visit, mean SBP and DBP were 134 ± 19 (range, 110–158) and 82 ± 13 (range, 60–92) mm Hg, respectively. At the final visit, mean SBP and DBP were 128 ± 14 (range, 106–146) and 78 ± 10 (range, 60–98) mm Hg, respectively. 

Table 2

 

Clinical Characteristics of Patients With Severe Hypertension

Characteristics	Values
Patients, male:female (%)	10:11 (47.6:52.4)
Age at presentation, y	41.8 ± 11.1; range, 23–62
Follow-up period, mo	8.8 ± 10.5; range, 2–36
Refractive error, diopters	−1.2 ± 1.8; range, −5.5 to 1.25
Blood pressure
Systolic pressure	223 ± 32; range, 180–280
Diastolic pressure	133 ± 27; range, 94–180
Mean arterial pressure*	163 ± 27; range, 130–206
Cause of hypertension, patients (%)
Renovascular	7 (33.3)
Renal parenchymal disease	6 (28.6)
Endocrine disorder	2 (9.5)
Preeclampsia	1 (4.8)
Unknown	5 (23.8)
Comorbid conditions, patients (%)
Renal failure	13 (61.9)
Intracranial hemorrhage	1 (4.8)
Brain infarct	1 (4.8)
Heart failure	1 (4.8)
Ophthalmic findings, eyes (%)
Hypertensive retinopathy	42 (100)
Keith-Wagener-Barker grade 1:2:3:4	0:6:15:21 (0:14.3:35.7:50.0)
Arteriolar narrowing	42 (100)
Hard exudates	19 (45.2)
Cotton wool patch	36 (85.7)
Flame-shaped retinal hemorrhage	22 (52.4)
Optic disc edema	22 (52.4)
Disc hemorrhage	4 (9.5)
Delayed choriocapillaris filling	20 (47.6)
Optic disc fluorescein leakage	22 (52.4)

 Data presented as mean ± standard deviation, where applicable.

*  Calculated as ⅓ × systolic + ⅔ × diastolic.

Figure 1 shows fundus photographs from younger patients (age, <40 years) with severe hypertension. The association between BP level and fundoscopic change severity is demonstrated, with arteriolar narrowing, hard exudates, flame-shaped retinal hemorrhages, cotton wool spots, and disc edema visible. These abnormalities were typically located around the optic nerve head and the larger arterioles. Hard exudates, cotton wool spots, and flame-shaped retinal hemorrhages were noted on fundus examination in 19 (45.2%), 36 (85.7%), and 22 (52.4%) eyes, respectively (Table 3). Optic disc edema and hemorrhage were also observed in 18 (42.9%) and 4 (9.5%) eyes, respectively. The FA showed optic disc fluorescein leakage and delayed choriocapillaris filling in 22 (52.4%) and 20 (47.6%) eyes, respectively. 

Figure 1

 

Fundus photographs from the right eyes of young patients (age, <40 years) with severe hypertension. Peripapillary and periarteriolar retinal changes are apparent, including cotton wool spots, retinal hemorrhages, and exudates. Patients with more severely elevated BP (systolic blood pressure > 200 mm Hg, upper row) show more extensive cotton wool spots, retinal hemorrhages, disc edema, and hard exudates than those with less severely elevated BP (lower row).

Table 3

 

Blood Pressure Parameters and Optical Coherence Tomography Findings at Baseline and During Follow-up

	Baseline	1 Month	Final Visit	P Value, Baseline vs. Final Visit
Systolic blood pressure, mm Hg	223 ± 32	134 ± 19	128 ± 14	<0.001
Diastolic blood pressure, mm Hg	133 ± 27	82 ± 13	78 ± 10	<0.001
Mean arterial pressure,* mm Hg	163 ± 27	100 ± 14	95 ± 11	<0.001
Qualitative features, eyes (%)
Macular edema†	16 (35.7)	0	0	<0.001
Irregular reflection/thickening of RNFL	14 (33.3)	6 (14.3)	0	<0.001
Subretinal fluid	21 (50.0)	6 (14.3)	0	<0.001
Inner retinal fluid	10 (23.8)	2 (4.8)	0	0.001
Hyperreflective dots within the retina	26 (61.9)	17 (40.5)	7 (16.7)	<0.001
Quantitative features
Central macular thickness, μm	359 ± 194	245 ± 30	243 ± 29	<0.001
Subfoveal choroidal thickness, μm	299 ± 97	242 ± 78	227 ± 65	<0.001
Subretinal fluid height, μm	101 ± 176	8.3 ± 22	0	0.002

 Means compared with paired t-tests and incidences compared with χ² or Fisher's exact test. RNFL = retinal nerve fiber layer.

*  Calculated as ⅓ × systolic + ⅔ × diastolic.

†  Defined as central macular thickness > 300 μm.

The most common KWB hypertensive retinopathy grade assigned was grade 4 (n = 18 eyes, 42.9%), followed by grades 3 (n = 17, 40.5%) and 2 (n = 7, 16.7%). However, Figure 2 shows fundus photographs from eyes of KWB grade 3 or 4, in which more aggressive fundoscopic features (Fig. 2A compared to Fig. 2B) do not correspond to worse visual acuities. Figure 3 shows the association between KWB grade and baseline and final visual acuities. Mean logMAR visual acuity at baseline for each KWB grade was 0.20 ± 0.34, 0.18 ± 0.25, and 0.44 ± 0.41 for grades 2, 3, and 4, respectively. Differences among the groups were statistically significant (P = 0.045, Kruskal-Wallis test). In particular, baseline visual acuity was significantly different between grades 3 and 4 (P = 0.024, Mann-Whitney U test). Linear regression also revealed a significant association between KWB grade and baseline visual acuity (correlation coefficient [r] = 0.305, P = 0.049). Mean final logMAR visual acuity for each KWB grade was 0.02 ± 0.06, 0.02 ± 0.19, and 0.13 ± 0.17 for grades 2, 3, and 4, respectively. Differences among the groups were not statistically significant (P = 0.088, Kruskal-Wallis test). Additionally, there was no significant linear relationship between KWB grade and final visual acuity (r = 0.276, P = 0.077). Therefore, this conventional grading system, based on fundoscopic features of hypertensive retinopathy, may not be a good indicator of visual outcomes in patients with severe hypertension. 

Figure 2

 

Fundus photographs and optical coherence tomography (OCT) images taken at baseline in (A) a 30-year-old man with BP of 186/120 mm Hg (systolic/diastolic) and (B) a 57-year-old woman with BP of 206/106 mm Hg. Keith-Wagener-Barker (KWB) grades are indicated in the upper right corner of each fundus photograph, and best-corrected visual acuities (VA) are noted in the lower right corner. Right-column OCT images demonstrate retinal changes, including irregular thickening/reflectance of the retinal nerve fiber layer (white arrows) and intraretinal hyperreflective dots (black arrows). The eyes shown in (B) had greater subretinal fluid height (white numbers), thicker subfoveal choroids, and worse visual acuities than those shown in (A), although KWB grades and fundoscopic features were less severe.

Figure 3

 

Correlations between visual acuity and conventional Keith-Wagener-Barker (KWB) grading (A) and optical coherence tomography measurements of central macular thickness (B) and subretinal fluid height (C). Subretinal fluid height at baseline was significantly correlated with both baseline and final visual acuity, but conventional KWB grading, based on fundus appearance, was significantly associated only with baseline visual acuity. Error bars represent the upper bound of the 95% confidence interval.

Abnormal Ocular Morphology in Patients With Acute Severe Hypertension

The images in Figure 2 and Supplementary Figure S1 (point-by-point correlation of fundus photograph, infrared reflectance, and OCT images) demonstrate how macular edema, SRF, irregular reflection, retinal nerve fiber layer thickening (white arrowhead), and intraretinal hyperreflective dots (black arrowheads) are represented on OCT. Intraretinal hyperreflective dots correspond to hard exudates in fundus photographs. Table 3 shows the type and frequency of retinal and choroidal morphologic abnormalities as imaged by EDI-OCT at baseline. Macular edema (CMT > 300 μm), irregular reflection and retinal nerve fiber layer thickening, SRF, intraretinal fluid, and intraretinal hyperreflective dots were noted in 16 (35.7%), 14 (33.3%), 21 (50.0%), 10 (23.8%), and 26 (61.9%) eyes, respectively. All 21 patients who had SRF had subfoveal SRF. Hyperreflective dots within the retina were most often found in the outer nuclear layer, but their location ranged from the subretinal space to the ganglion cell layer. Intraretinal fluid was most often found in the outer nuclear layer. 

Mean CMT and SRF height was 359 ± 194 (range, 227–1089 μm) and 101 ± 176 μm (range, 0–748 μm), respectively. Mean SCT was 299 ± 97 μm (range, 162–573 μm). Figure 3B shows the weak, and insignificant, association between CMT and baseline (r = 0.274, P = 0.083) and final (r = 0.251, P = 0.113) visual acuity. In contrast, SRF height was significantly associated with both baseline (r = 0.440, P = 0.004) and final (r = 0.316, P = 0.042) visual acuity (Fig. 3C). 

A modified classification system using both OCT and fundoscopic findings divides hypertensive retinopathy into mild to moderate retinopathy, malignant retinopathy without SRF, and malignant retinopathy with SRF. Among the grades, baseline and final BCVA were significantly different (P = 0.036 and 0.049 for baseline and final BCVA, respectively, Kruskal-Wallis test). As presented in Figure 4, the grades were significantly correlated with both baseline (r = 0.347, P = 0.023) and final (r = 0.345, P = 0.025) BCVA. Compared to KWB disease severity grades (Fig. 3A), our novel, modified classification system may better predict visual outcomes in patients with retinopathy resulting from severe hypertension. 

Figure 4

 

A modified three-step classification of hypertensive retinopathy in patients with severe hypertension. The new system was based on both fundoscopic features and optical coherence tomography findings. (A) Photographic examples of mild to moderate, malignant without subretinal fluid (SRF), and malignant with SRF. (B) Associations between hypertensive retinopathy grades and baseline and final visual acuity as tested by the Kruskal-Wallis test (upper) and linear regression analysis (lower).

One month after BP was brought under control, many abnormal OCT features resolved. By the final visit, the proportion of eyes having each type of abnormal feature significantly decreased (all P < 0.05). Hyperreflective dots within the retina gradually decreased over the follow-up period, but the abnormality persisted in seven eyes (16.7%) through the final visit. Retinal and choroidal thickness began to decrease once BP was controlled, and plateaued by the 1-month visit (Supplementary Fig. S2). The CMT decreased from 359 ± 194 μm at baseline to 245 ± 30 μm at 1 month and 243 ± 29 μm at the final visit (P < 0.001, paired t-test). Subfoveal choroidal thickness also decreased from 299 ± 97 μm at baseline to 242 ± 78 μm at the 1-month visit (P < 0.001) and to 227 ± 65 μm at the final visit (P < 0.001, paired t-test). The EDI-OCT images in Figure 5 show how choroidal thickness decreased following BP control. 

Figure 5

 

Choroidal changes identified with enhanced depth imaging optical coherence tomography in two patients with severe hypertension before and after BP control. Left: One month after BP control, subfoveal choroidal thickness decreased from 282 μm at baseline to 222 μm. Right: One month after BP control, choroidal thickness decreased from 241 μm at baseline to 166 μm. In both cases, subretinal fluid height also remarkably decreased following BP control.

Relationship Between Blood Pressure and Retinal/Choroidal Thickness Parameters, Visual Acuity, and Hypertensive Retinopathy Grade

Table 4 presents the relationship between BP measures and retinal/choroidal thickness parameters. Subfoveal choroidal thickness was significantly correlated with all BP measures (P = 0.007, 0.032, and 0.005 for SBP, DBP, and MAP, respectively). However, CMT was only weakly and insignificantly correlated with BP parameters (all P > 0.05). In contrast, SRF height was significantly correlated with SBP (r = 0.367, P = 0.035) and with SCT (r = 0.646, P < 0.001). After controlling for the effect of age, SCT was also significantly associated with all BP measures (SBP: regression coefficient [B] = 1.979, P = 0.004; DBP: B = 1.723, P = 0.043; MAP: B = 2.037, P = 0.014) and SRF height (B = 0.353, P = 0.002). 

Table 4

 

Associations Between Blood Pressure and Optical Coherence Tomography Thickness Parameters and Conventional Fundoscopic Grading

	Systolic Blood Pressure		Diastolic Blood Pressure		Mean Arterial Pressure
	R	P Value	r	P Value	r	P Value
OCT measurement
Central macular thickness	0.329	0.066	0.185	0.310	0.253	0.162
Subretinal fluid height	0.367	0.035	0.217	0.226	0.307	0.093
Subfoveal choroid thickness	0.486	0.007	0.393	0.032	0.512	0.005
Fundoscopic grading, KWB	0.307	0.048	0.149	0.346	0.220	0.162
Novel OCT and fundoscopic grading*	0.312	0.044	0.245	0.118	0.285	0.067

*  Based on the presence of subretinal fluid on OCT and fundoscopic features. Hypertensive retinopathy groups were as follows: mild to moderate retinopathy (Keith-Wagener-Barker grade 2 or 3; n = 24 eyes), malignant retinopathy (Keith-Wagener-Barker grade 4) without SRF (n = 7 eyes), and malignant retinopathy with SRF (n = 11 eyes).

The SBP was significantly correlated with baseline visual acuity (r = 0.316, P = 0.042). Additionally, SBP (r = 0.470, P = 0.002), DBP (r = 0.306, P = 0.049), and MAP (r = 0.388, P = 0.011) were all significantly associated with final visual acuity. 

The SBP was significantly correlated with retinopathy grades assigned using our three-step classification system (r = 0.312, P = 0.044), but neither DBP (r = 0.245, P = 0.118) nor MAP (r = 0.285, P = 0.067) was. This result was similar to that obtained with the KWB grades, which showed a statistically significant association only with SBP (r = 0.307, P = 0.048). 

Visual Outcomes and Predictors of Visual Outcomes

Mean logMAR BCVA was 0.29 ± 0.36 at baseline, which gradually improved over the follow-up period. At the 1-month and final visits, mean BCVA was 0.16 ± 0.29 and 0.08 ± 0.17, respectively. The number of eyes having a BCVA worse than 20/40 was 15 (35.7%), 8 (19.0%), and 3 (7.1%) at diagnosis, 1 month, and the final visit, respectively (Supplementary Fig. S3). 

The serial OCT images in Figure 6 show how the qualitative abnormalities, visible on OCT, gradually disappeared following BP control. The three patients with a final BCVA worse than 20/40 showed incomplete recovery of foveal photoreceptor defects, visible in OCT images obtained at the final visit (Fig. 6A). In contrast, all photoreceptor layers had recovered by the final visit in 36 of 39 eyes with a final BCVA equal to or better than 20/40 (Fig. 6B). This difference in the proportion of eyes with a final BCVA equal to or better than 20/40 was statistically significant (36 of 36 eyes with complete photoreceptor recovery versus three of six eyes without complete photoreceptor recovery, P = 0.002, Fisher's exact test). 

Figure 6

 

Spectral-domain optical coherence tomography (SD-OCT) images comparing the clinical courses of malignant hypertensive retinopathy between patients with poor (A) and good (B) visual outcomes. Both eyes showed resolution of subretinal fluid, intraretinal fluid, and intraretinal hyperreflective dots. However, the eye with poor final visual acuity (20/50) had incomplete photoreceptor recovery, as indicated by interdigitation zone (IZ) loss and focal inner segment ellipsoid zone (EZ) loss, at the final visit. In contrast, the eye with good final visual acuity (20/15) had complete photoreceptor recovery by the final visit. In the eye with a poor visual outcome (left), the posterior choroidal border was indefinite and choroidal thickness could not be measured. In the eye with a good visual outcome (right), baseline choroidal thickness at week 1, week 2, month 1, month 2, and the final visit was 182, 170, 163, 154, 157, and 158 μm, respectively. ELM, external limiting membrane; VA, visual acuity.

Univariate and multivariate regression analyses for identifying factors associated with final BCVA were performed using baseline clinical and OCT features (Table 5). Univariate regression analyses showed that baseline BCVA (r = 0.663, P < 0.001), SBP (r = 0.470, P = 0.002), DBP (r = 0.306, P = 0.049), baseline choroid thickness (r = 0.367, P = 0.033), OCT-based retinopathy grades (r = 0.345, P = 0.025), and SRF height (r = 0.316, P = 0.042) were significantly associated with final BCVA. Among these factors, baseline BCVA (B = 0.633, P < 0.001) and DBP (B = 0.339, P = 0.010) were significantly associated with final BCVA also in multivariate stepwise regression analysis. Additionally, logistic analyses revealed that large SRF height (μm) was significantly associated with incomplete photoreceptor recovery (adjusted odds ratio = 1.017, 95% confidence interval 1.002–1.032, P = 0.029). 

Table 5

 

Univariate and Multivariate Regression Analyses for Identifying Factors Associated With Final Best-Corrected Visual Acuity in Patients With Severe Hypertension

	Univariate Analysis		Multivariate Analysis
	Correlation Coefficient	P Value	Regression Coefficient	P Value
Age, y	0.028	0.905
Baseline visual acuity, logMAR	0.663	<0.001	0.633	<0.001
Systolic blood pressure, mm Hg	0.470	0.002	0.125	0.602
Diastolic blood pressure, mm Hg	0.306	0.049	0.339	0.010
Baseline choroid thickness, μm	0.367	0.033	−0.051	0.736
Central macular thickness, μm	0.251	0.113
Subretinal fluid height, μm	0.316	0.042	−0.181	0.288
Fundoscopic (Keith-Wagener-Barker) grading	0.236	0.137
New combination, OCT and fundoscopic grading*	0.345	0.025	0.078	0.607

*  Based on the presence of subretinal fluid on OCT and fundoscopic features. Hypertensive retinopathy groups were as follows: mild to moderate retinopathy, malignant retinopathy without SRF, and malignant retinopathy with SRF.

Long-Term Sequelae of Retinal and Choroidal Changes

At the final visit, several retinal sequelae persisted. As shown in Figure 7A, retinal nerve fiber layer defects were observed in 32 eyes (76.2%) and were topographically associated with cotton wool spots, as seen in temporal-superior-nasal-inferior-temporal thickness profiles (Fig. 7). Additionally, a significant association between retinal nerve fiber layer defects and cotton wool spots was found (P = 0.002, Fisher's exact test). Specifically, we compared retinal nerve fiber layer defect prevalence at the final visit in eyes with (31 of 36 eyes, 86.1%) and without (1 of 6, 16.7%) cotton wool spots. At the final visit, seven eyes (16.7%) had persistent hard exudates (Fig. 7B), which corresponded in location to hyperreflective dots around the macula on OCT images. Other changes included foveal photoreceptor defects (6 eyes [14.3%]), the presence of which was associated with poor visual outcome (BCVA worse than 20/40, Fig. 6A). 

Figure 7

 

Sequelae of acute retinal and choroidal changes caused by severe hypertension. The retinal nerve fiber layer defect in the vicinity of cotton wool patches (A, B) and persistent hard exudates around the macula (B). The time at which photographs were taken is indicated in the lower right corner. Blue, black, and white arrows are topographically matched between photographs and optical coherence tomography analyses of peripapillary nerve fiber layer thickness.

Discussion

This study investigated morphologic changes of the retina and choroid in patients with severe hypertension. These changes were correlated with several BP parameters. Using EDI-OCT imaging, we identified, in vivo, quasi-histologic retinal and choroidal abnormalities in patients with retinal and choroidal parenchymal end organ damage caused by severe hypertension. We also evaluated visual outcomes in these patients and explored anatomical changes in the retina and choroid associated with persistent vision loss. Although retinal and choroidal changes occurring during severely elevated BP can be resolved within a short period of time following BP control, they can lead to visual loss caused by incomplete photoreceptor recovery and/or nerve fiber layer defects. As an indicator of visual prognosis, a modified three-step grading system, based on fundoscopic and OCT image features, was developed. Retinopathy grades assigned with this novel classification system, but not with the fundoscopic KWB classification system, were associated with both baseline and final visual function in patients with severe hypertension. 

Retinal changes observed in patients with severe (malignant) hypertension included intraretinal transudate, multiple cotton wool spots, and retinal hemorrhages. Additionally, OCT revealed intraretinal fluid and SRF in approximately one-third of patients with severe hypertension. Intraretinal transudate results from breakdown of the retinal arteriole blood–retinal barrier caused by highly elevated BP.¹¹ Retinal changes were generally observed in the peripapillary and periarteriolar areas, as shown in Figure 1. Additionally, patients with more severely elevated BP (i.e., >240 mm Hg) had more extensive areas of retinal exudates. This indicates that large arterioles in the peripapillary area had remarkably high intravascular pressure, high enough to break the blood–retinal barrier. However, the intravascular pressure in smaller arterioles may not have been quite as elevated because pressure is attenuated across the peripapillary arterioles. The peripapillary and periarteriolar distribution of retinal abnormalities is a unique feature of hypertensive retinopathy and may aid in distinguishing it from other conditions causing retinal hemorrhage, edema, and exudates. 

Cotton wool spots were very common in our patients with severe hypertension. Malignant hypertension is characterized by arteriolar fibrinoid necrosis and retinal nerve fiber layer ischemia,^20,21 which lead to the spots. Using SD-OCT, irregular reflection and swelling of the retinal nerve fiber layer were identified and are believed to result from ischemic damage to the layer in which the spot is located. Irregular reflections in OCT images can also be caused by flame-shaped retinal hemorrhages located in the nerve fiber layer. Cotton wool spots, although generally resolved within 1 month of BP control, are clinically important because they can represent permanent nerve fiber layer defects. 

In contrast to other quickly resolving (within 1 month) OCT findings, intraretinal hyperreflective dots persisted for more than 6 months in some patients. Outer retinal hyperreflective spots have been previously reported in eyes with macular telangiectasia.²² It remains unclear whether these extracellular deposits are caused by extravasations or by retinal cell degeneration.²² However, as shown in Figure 6, no retinal structural abnormalities around the hyperreflective spots were observed following their disappearance. Interestingly, spot location corresponded to hard exudates on fundus photographs. Therefore, our OCT images support the idea that spots are derived from extracellular deposits in patients with severe hypertension. 

Hypertensive choroidopathy has been described in eyes of patients with pre-eclampsia, systemic lupus erythematosus, and thrombocytic thrombocytopenic purpura.^23,24 Highly elevated BP can lead to choroidal fibrinoid necrosis, choriocapillaris nonperfusion, RPE ischemic necrosis, outer blood–retinal barrier, and SRF accumulation.^12,25 Our study demonstrates that choroidal thickness is also significantly increased during severe hypertension. In patients followed over the long term (>12 months), choroidal thickness began to decrease as BP was normalized (Supplementary Fig. S2), and the timing to reach a plateau in choroidal thickness corresponded to the timing to reach a plateau in BP. This finding further supports the association between BP elevation and increased choroidal thickness. Accumulation of SRF was particularly important in eyes examined here because it was associated with vision loss at both the baseline and final visits. It is generally believed that SRF accumulation is caused by choroidal permeability changes. These changes lead to an increase in choroidal interstitial fluid, which eventually extends into the subretinal space. Our findings support this theory because choroidal thickness increased with severely elevated BP, and choroidal thickness was significantly correlated with SRF height. Increased choroidal thickness may have resulted from interstitial fluid accumulation in the choroid, which led to a subsequent increase in hydrostatic pressure. This increased pressure may have led to fluid movement across the defective RPE and into the subretinal space. This sequence of events explains the association between SRF accumulation and hypertensive choroidopathy.¹¹ Eyes with central serous chorioretinopathy and polypoidal choroidal vasculopathy have been shown to have SRF accumulation from choroidal permeability changes.^26–29 This mechanism, common to these conditions and hypertensive retinopathy, suggests a pathogenic association between choroidal permeability changes and SRF accumulation. Another mechanism of SRF in severe hypertension might be from the accumulation of intraretinal exudative fluid which leaked through the retinal vasculature. The exact origin of SRF in severe hypertension requires further research.  

Although all eyes with SRF had complete SRF resolution, six eyes had persistent foveal photoreceptor defects at the final visit, which was associated with persistent visual loss. Photoreceptor defects following SRF resorption have been observed in many retinal diseases, including central serous chorioretinopathy and rhegmatogenous retinal detachment.^30,31 Therefore, we believe that photoreceptor defects resulted from SRF, rather than direct damage to the photoreceptor layer caused by hypertension. 

Keith et al.⁷ described a widely used classification system to grade the severity of hypertensive retinopathy. This KWB grading system categorized the condition into four subgroups based on fundoscopic findings (Table 1). Some authors have questioned the usefulness of this classification system because clinicians can have difficulty distinguishing between grades 1 and 2, and KWB grades are not closely correlated with the hypertension severity.⁸ Although KWB grades are still widely used, modifications have been suggested several times.^8,32 Population-based data have shown a significant association between some hypertensive retinopathy grades (i.e., moderate or malignant) and risk of stroke, coronary artery disease, and death.^8,33 However, there has been little effort to associate the retinopathy grades with visual outcome. Our study found that the traditional KWB grading system for hypertensive retinopathy was significantly associated with baseline SBP and BCVA, but not with final BCVA. In contrast, SRF height, as measured on OCT images, was significantly associated with final BCVA. This finding suggests that final visual function may be better predicted using OCT measurements rather than the conventional grading system, which is based only on fundoscopic features. It is difficult to identify and impossible to quantify SRF on fundoscopic examination. This may explain the insignificant association between visual outcomes and the conventional KWB grading system. Thus, for patients with severe hypertension, we developed a new grading system (Fig. 4), based on both fundoscopic signs and OCT findings, that was modified from the KWB grading system and the most recent classifications by Wong and Mitchell.⁹ In our three-group classification system, approximately half of the eyes were assigned the malignant retinopathy grade, which was divided into two categories on the basis of the presence of SRF. The mild and moderate retinopathy categories in the Wong and Mitchell classification system were lumped into one grade (mild or moderate retinopathy) because the proportion of eyes with mild retinopathy was small among the patients with severe hypertension. Additionally, the difference in visual outcome between mild and moderate retinopathy was not remarkable. Applying our novel three-step grading system to our patients revealed a linear relationship between retinopathy grade and baseline and final visual acuities. Therefore, we believe our classification system is an easier, more practical approach to determining hypertensive retinopathy severity in patients with severe hypertension. It provides useful information on both visual outcome and the severity of hypertensive retinopathy in the patients. 

Our study had several limitations. First, the retrospective design results in intrinsic drawbacks, namely, selection bias. Additionally, the short and variable follow-up periods limited results, and prospective studies with longer, uniform follow-up periods are needed to draw more definite conclusions regarding visual outcome. In particular, longer follow-up periods may reveal a lower percentage of eyes with intraretinal lipid deposits and persistent foveal photoreceptor defects. Moreover, our study included a relatively small number of patients, and the derived classification system has not been tested prospectively or applied to an independent cohort. Most importantly, diurnal variation may have affected our results. Using Spectralis OCT, Tan et al.³⁴ showed that normal, healthy subjects had an average diurnal variation in SCT of 31.6 μm. However, the differences in average choroid thickness between baseline and the 1-month visit (57 μm) and between baseline and the final visit (72 μm) were larger than this diurnal variation. Moreover, we suspect that SRF volume, not SRF height, is the more accurate way to quantify SRF. Subretinal fluid volume measurements may provide different results for the association between SRF quantity and choroidal thickness and visual acuity. 

In conclusion, our study showed that severe hypertension resulted in characteristic peripapillary and periarteriolar hypertensive retinopathy features, SRF accumulation, and increased choroid thickness. The presence of SRF was associated with choroidal thickening and with poor visual outcome in patients with severe hypertension. Thus, OCT may be useful to document hypertensive retinopathy and choroidopathy severity. Our OCT-based classification system for patients with severe hypertension may be useful in predicting visual outcomes and in evaluating systemic hypertension severity. Further prospective studies including a larger sample size and incorporating longer follow-up periods are needed to confirm our findings. 

Acknowledgments

Supported by a grant from the National Research Foundation of Korea (Grant 2012R1A2A2A02012821). 

Disclosure: S.J. Ahn, None; S.J. Woo, None; K.H. Park, None