Distribution of port wine birthmarks and glaucoma outcomes in Sturge-Weber Syndrome

Abstract

Purpose:

To identify which features of Sturge-Weber Syndrome (SWS) were most associated with glaucoma onset, severity, and treatment failure at a tertiary care center.

Design:

Retrospective cross-sectional study.

Subjects:

Children who had SWS with and without glaucoma.

Methods:

Electronic health records were reviewed for all children with SWS presenting between 2014–2020. Examination and imaging findings from dermatology, neurology, and ophthalmology were collected. Logistic regression was used to identify factors associated with glaucoma-related outcomes.

Main Outcome Measures:

Primary outcomes included glaucoma development, progression to surgery, and treatment failure. Failure was defined as having a final intraocular pressure >21 mmHg, devastating complication, or ≤20/200 vision.

Results:

Twenty-three of 44 SWS patients (52.3%) developed glaucoma, and 6 of 23 patients (26.1%) had both eyes affected. Sixteen of 29 eyes (55.2%) required surgery, and 29.6% overall met our failure criteria (mean follow-up: 5.1±4.3 years). Glaucoma diagnosis was associated with bilateral port wine birthmarks (PWB; OR 5.9; 95% CI 1.3–43.2), PWB with any lower eyelid involvement (OR 9.7, 95% CI 2.6–44.5), and choroidal hemangiomas (OR 3.8, 95% CI 1.1–13.8), but was not associated with upper eyelid or leptomeningeal angiomas, seizures, prior hemispherectomy or pulsed-dye laser. Eyes that progressed to surgery were more likely to have PWB affecting the lower eyelid (OR 33.7, 95% CI 4.5–728.0). No clinical or demographic factors were associated with treatment failure. In most cases, angle surgery failed (72.7%) but was a temporizing measure before subconjunctival filtering surgery.

Conclusions:

Lower eyelid and choroidal angiomas were associated with glaucoma diagnosis, suggesting a spatial relationship with SWS findings. However, leptomeningeal angiomas were not associated possibly because these are further from the eye.

Introduction

Sturge-Weber Syndrome (SWS), or encephalofacial angiomatosis, is a rare, neurocutaneous syndrome that can result in severe complications such as intellectual disability, seizures, and blindness secondary to glaucoma or choroidal detachments.¹ The condition is notable for the appearance of a facial angioma or “port-wine birthmark” (PWB), which is the preferred terminology recommended by the patient-oriented Sturge-Weber Foundation in order to destigmatize this medical condition. SWS is classified as one of several phakomatoses with a sporadic inheritance pattern in which a somatic mosaic mutation occurs during embryonic development. The incidence is estimated at 1 in 50,000 infants, with no race nor gender predilection.² A gene mutation in GNAQ has been identified as the pathologic cause of SWS and non-syndromic PWBs, resulting in vascular endothelial cell overproliferation.³

Facial PWBs in SWS were initially thought to be distributed along the divisions of the trigeminal nerve, but are now suspected of following embryonic craniofacial patterns.⁴ This is pronounced in cases where the birthmark crosses the vertical midline or bilateral cases. Previous reports have commented on the association between the location of the PWBs (particularly the involvement of the eyelids) and glaucoma onset.^5–9 In the 1990s, one center found that 92% of their SWS patients with glaucoma had a PWB involving the upper and lower eyelids while 8% had a birthmark involving only the upper eyelid.⁶ These birthmarks have been treated with pulsed-dye laser (PDL) photocoagulation for the last few decades.^5,10 One concern of PDL of the eyelid has been that this photocoagulation of facial vessels may block a pathway for episcleral venous flow.⁴ The premise that thermolysis of eyelid birthmarks from PDL may have an effect on secondary glaucoma from increased episcleral outflow has been disputed.^7,8

PWBs are flat, pink patches secondary to dilated capillaries and venules in the dermis, not to be confused with hemangiomas, which are hamartomatous lesions from a proliferation of capillary or cavernous vessels.⁴ Leptomeningeal hemangiomas may cause contralateral hemiparesis, hemiplegia, hemianopsia, headaches, stroke-like episodes, developmental delay, intellectual disability, and behavioral problems, with bilateral lesions resulting in more severe symptoms.^1,4,6 Choroidal hemangiomas can be asymptomatic, but may also lead to cystoid macular edema, subretinal hemorrhages, and serous or hemorrhagic choroidal detachments.^1,4 However, secondary glaucoma from SWS is often described as a bimodal-onset condition, in which the congenital variant is caused by trabecular dysgenesis and the juvenile or older variants are secondary to increased episcleral venous resistance, often seen with episcleral angiomas.^1,4,6 Given the mixture, spatial proximity, and pathogenesis of SWS symptoms, we sought to identify which features of SWS were most associated with glaucoma onset, severity, and treatment failure.

Methods

A retrospective chart review was performed for all SWS patients less than 18 years old seen in the Department of Ophthalmology at Boston Children’s Hospital between January 2014 and October 2020. For some patients, data from external chart records were also reviewed, pertaining to the age of diagnosis and prior treatments when available. This study was approved by the Boston Children’s Hospital Institutional Review Board and was conducted in accordance with the Declaration of Helsinki and the United States Health Insurance Portability and Accountability Act.

Demographic data, systemic conditions and treatments, medications, neuroimaging findings, ocular findings, as well as visual and surgical outcomes and complications were collected for each patient. SWS diagnosis was made using the Roach scale, which included presence of facial and leptomeningeal angiomas with or without ocular involvement (Type 1), facial angioma with ocular involvement (Type 2), or leptomeningeal angioma with or without ocular involvement (Type 3).¹¹ As part of the normal workup for SWS, all patients with a facial or leptomeningeal angioma from our hospital were referred for ophthalmology examination to rule out glaucoma. Those who developed secondary glaucoma were diagnosed using the Childhood Glaucoma Research Network classification system.¹² Intraocular pressure (IOP) was obtained by Goldmann or Perkins Applanation Tonometry (Haag-Streit, Koeniz, Switzerland) when possible. If a child was less cooperative, IOP was obtained using the iCare TA01i Rebound Tonometer (Revenio Group, Vantaa, Finland). For pre- or non-verbal children, grating visual acuity was assessed by preferential looking using Teller acuity cards. Optotype acuity was measured by LEA symbols, HOTV, or Snellen letters, based on the patient’s age and developmental capability. In some developmentally delayed patients, visual acuity was not quantifiable.

Primary outcome measures used in this study were the proportion of patients that developed glaucoma, progressed to surgery, and failed treatment, along with their associated factors. Secondary outcomes included best corrected visual acuity (BCVA) and IOP at the final visit. Success was achieved when patients reached the end of the study period with a final IOP ≤21mmHg with or without glaucoma medications, with a final BCVA better than 20/200, and without any devastating complications. Failure was defined as when success criteria were not met. A devastating complication was counted if any major surgical complication occurred such as a suprachoroidal hemorrhage or endophthalmitis or if non-surgical complications from SWS resulted in a final BCVA ≤20/200 vision. For treatment outcomes, patients were excluded from the final analysis if they had less than 3 months of post-operative follow-up after the first surgery or less than 3 months of total follow-up in the absence of any surgery. When poor vision was measured, we categorized the cause based on objective clinical exam findings to determine whether the cause was ocular (e.g. optic nerve, corneal, or retinal changes consistent with poor visual acuity) or non-ocular (e.g. inability to test due to developmental status or diagnosis of cortical visual impairment in the setting of relatively normal or intact ocular structures).

Continuous variables were summarized using means, medians, standard deviations, interquartile ranges (IQR) while categorical variables were reported using frequency and proportions. Wilcoxon rank-sum tests and Fisher exact tests were used for hypothesis testing between groups. Logistic regression was used to identify factors associated with glaucoma development, progression to surgery, and treatment failure. Odds ratios (OR) and 95% confidence intervals (CI) are reported from the regression models. Statistical tests were two-tailed, with significance defined as p < 0.05. All analyses were performed using R version 4.3.2 (R Foundation, 2023).

Results

Of the 44 SWS patients that were included in our study, 23 patients (60.9% female) had Sturge-Weber associated glaucoma (SWG, 52.3%). Bilateral involvement was noted in 6 patients (26.1%), for a total of 29 eyes included in the study. Twenty out of 29 (69.0%) eyes were diagnosed with glaucoma within the first year of life (Figure 1). There was one patient whose age of diagnosis was unknown due to a transfer of care to our department later in the patient’s glaucoma course. The mean follow-up time was 5.1 ± 4.3 years for the SWG patients while the follow-up time was 6.2 ± 4.3 years for the SWS patients without glaucoma (p = 0.45). The mean age at the last follow-up was 6.8 ± 3.9 years for those with glaucoma and 8.9 ± 4.4 years for those without glaucoma (p = 0.10).

Figure 1 –

Age of glaucoma onset in the glaucoma group, with a bimodal peak of age development. Most eyes (20/29, 69.0%) were found to be diagnosed within the first 12 months of life, with the remaining eyes (8/29, 27.6%) diagnosed in early childhood. One eye had an unknown age of diagnosis.

Table 1 shows the dermatologic, neurologic, and ophthalmologic features of the SWS patients with and without glaucoma. Among 23 SWG patients, 16 (69.6%) had lower lid involvement of their PWB, 12 (57.1%, 2 unknown) had PDL facial treatment prior to any glaucoma surgery, and 9 (40.9%, 1 unknown) had a bilateral birthmark. In addition, 18 (81.8%, 1 unknown) had leptomeningeal involvement, 16 (69.5%) had a history of seizures, and 5 (21.7%) had a prior hemispherectomy. Lastly, 15 (65.2%) had a choroidal hemangioma and 9 (39.1%) had a history of strabismus. The median spherical equivalent for the affected eye in the SWG group was +1.5 (IQR −1.4 – +2.5).

Table 1:

Relationship between clinical findings and glaucoma diagnosis in patients with Sturge-Weber Syndrome

	Group		Odds of glaucoma diagnosis
	No Glaucoma, N = 21	Glaucoma, N = 23	OR	95% CI	p-value
Bilateral PWB (right and left lid)	2 (11%)	9 (41%)	5.88	1.25, 43.2	0.040
PWB on upper lid, lower lid, or both			3.05	1.59, 6.52	0.002
Neither	2 (9.5%)	0 (0%)
Upper lid	15 (71%)	7 (30%)
Lower lid	0 (0%)	1 (4.3%)
Both	4 (19%)	15 (65%)
Any lower lid involvement	4 (19%)	16 (70%)	9.71	2.57, 44.5	0.002
PWB greater on right side	9 (47%)	12 (52%)	1.21	0.36, 4.16	0.8
PDL treatment prior to any glaucoma surgery	13 (65%)	12 (57%)	0.72	0.20, 2.53	0.6
Seizures	14 (67%)	16 (70%)	1.14	0.32, 4.14	0.8
Leptomeningeal Disease on MRI	19 (90%)	18 (82%)	0.47	0.06, 2.74	0.4
Hemispherectomy	5 (24%)	5 (22%)	0.89	0.21, 3.75	0.9
Choroidal Hemangioma	7 (33%)	15 (65%)	3.75	1.11, 13.8	0.038
Strabismus	5 (24%)	9 (39%)	2.06	0.57, 8.09	0.3

n (%); N = total number of patients; OR = Odds Ratio; CI = Confidence Interval; PWB = port wine birthmark; PDL = pulsed-dye laser; MRI = magnetic resonance imaging

Logistic regression models were used to analyze the association between ophthalmologic, dermatologic, and neurologic manifestations of SWS with glaucoma development, progression to glaucoma surgery, and treatment failure. There was a higher odds of developing a glaucoma diagnosis in SWS patients if there was any lower eyelid involvement of the PWB (OR 9.7, 95% CI 2.6 – 44.5, p = 0.002), a bilateral birthmark (OR 5.9, 95% CI 1.3 – 43.2, p = 0.040), or presence of a choroidal hemangioma (OR 3.8, 95% CI 1.1 – 13.8, p = 0.38; Table 1). We did not find any significant associations between glaucoma development and the following variables: PDL facial treatment, seizures, leptomeningeal disease, hemispherectomy surgery, or strabismus. Sixteen (55.2%) out of 29 SWG eyes required glaucoma surgery. Only lower eyelid involvement of the birthmark was associated with higher odds of surgical intervention (OR 33.7, 95% CI 4.3 – 728.0, p =0.003; Table 2). Lastly, there were 8 (29.6%) out of 27 eyes that had treatment failure using our definitions (2 SWG eyes had inadequate follow-up and were excluded from the treatment analysis). In the logistic regression analysis for treatment failure, none of our variables of interest were statistically associated with treatment outcome (Table 3).

Table 2:

Relationship between clinical findings and receiving glaucoma surgery in eyes with Sturge-Weber associated glaucoma

	Group		Odds of surgical intervention
	Topical, N = 13	Surgery, N = 16	OR	95% CI	p-value
Bilateral PWB (right and left lid)	8 (62%)	7 (47%)	0.55	0.11, 2.44	0.4
PWB on upper lid, lower lid, or both			5.11	1.96, 18.8	0.003
Upper lid	9 (69%)	1 (6.2%)
Lower lid	0 (0%)	1 (6.2%)
Both	4 (31%)	14 (88%)
Any lower lid involvement	4 (31%)	15 (94%)	33.7	4.52, 728	0.003
PWB greater on right side	6 (46%)	7 (44%)	0.91	0.20, 4.02	0.9
PDL treatment prior to any glaucoma surgery	8 (67%)	7 (47%)	0.44	0.08, 2.05	0.3
Seizures	10 (77%)	12 (75%)	0.90	0.15, 5.05	>0.9
Leptomeningeal Disease on MRI	10 (77%)	14 (93%)	4.20	0.46, 92.1	0.2
Hemispherectomy	3 (23%)	3 (19%)	0.77	0.12, 4.95	0.8
Choroidal Hemangioma	6 (46%)	13 (81%)	5.06	1.03, 30.7	0.056
Strabismus	2 (22%)	7 (50%)	3.50	0.59, 29.5	0.2

n (%); N = total number of eyes; OR = Odds Ratio; CI = Confidence Interval; PWB = port wine birthmark; PDL = pulsed-dye laser; MRI = magnetic resonance imaging

Table 3:

Relationship between clinical findings and glaucoma treatment failure in eyes with Sturge-Weber associated glaucoma

	Group		Odds of treatment failure
	Success, N = 19	Failure, N = 8	OR	95% CI	p-value
Bilateral PWB (right and left lid)	10 (53%)	4 (57%)	1.20	0.21, 7.53	0.8
PWB on upper lid, lower lid, or both			28,395,023	0.00, NA	>0.9
Upper lid	9 (47%)	0 (0%)
Lower lid	1 (5.3%)	0 (0%)
Both	9 (47%)	8 (100%)
Any lower lid involvement	10 (53%)	8 (100%)	251,492,812	0.00, NA	>0.9
PWB greater on right side	9 (47%)	3 (38%)	0.67	0.11, 3.54	0.6
PDL treatment prior to any glaucoma surgery	11 (61%)	4 (50%)	0.64	0.11, 3.50	0.6
Seizures	13 (68%)	7 (88%)	3.23	0.42, 67.5	0.3
Leptomeningeal Disease on MRI	14 (78%)	8 (100%)	66,085,024	0.00, NA	>0.9
Hemispherectomy	3 (16%)	2 (25%)	1.78	0.20, 13.6	0.6
Choroidal Hemangioma	14 (74%)	5 (63%)	0.60	0.10, 3.75	0.6
Strabismus	4 (29%)	4 (57%)	3.33	0.51, 24.8	0.2

n (%); N = total number of eyes; OR = Odds Ratio; CI = Confidence Interval; PWB = port wine birthmark; PDL = pulsed-dye laser; MRI = magnetic resonance imaging

Of the SWG eyes, the 13 eyes that were controlled with medications alone throughout the follow-up period were older at the time of glaucoma diagnosis than the 16 eyes that required surgery. The mean age at diagnosis of the surgery group was 0.4 ± 0.7 years while it was 2.0 ± 2.4 years in the medication group (p = 0.034). Eight (50%) of 16 surgical eyes did require more than one surgery. Figure 2 shows the Kaplan-Meier survival time between the first and second surgeries. The median time between first and second surgeries was 0.6 years for initial goniotomies (5 eyes, median age at first surgery 0.6 years) and 2.8 years for initial ab externo trabeculotomy (6 eyes, median age 2.4 years). The median follow-up time after initial surgery was 4.2 years for tube shunts (3 eyes, median age 2.5 years) and 5.2 years for trabeculectomy with Mitomycin-C (2 eyes, median age 3.3 years) since these eyes did not require a second surgery. At the final follow-up, the mean number of medications was 2.1 ± 1.6 in the surgery group and 1.6 ± 1.0 in the medication group (p = 0.20).

Figure 2 –

Kaplan Meier survival time between first and second glaucoma surgeries. Among 16 eyes that received surgery, 6 patients received ab externo trabeculotomy (median age at surgery 2.4 years), 5 received goniotomy (0.6 years), 3 received a tube shunt (2.5 years) and 2 received a trabeculectomy with Mitomycin-C (MMC; 3.3 years) as their first surgery. Eight (50%) patients underwent additional surgeries.

At the final follow-up, 9/29 SWG eyes (31.0%) achieved 20/40 vision or better, with another 9 eyes (31.0%) achieving a visual acuity between 20/50 and 20/180 (Figure 3). Two eyes (6.9%) measured 20/200 or worse, while the remaining patients had either fix-and-follow or blink-to-light vision, which in some cases were due to preverbal age, developmental status, or cortical visual impairment. One eye had light perception vision due to end-stage glaucoma. Twenty-two (81.4%) of 27 eyes had a final IOP ≤ 21mm Hg. There were no major surgical complications in this cohort. Two patients developed serous choroidal detachments, one of which was initially treated with oral propranolol. Both were resolved with proton beam therapy, but only one had a BCVA that met failure criteria.

Figure 3 –

Visual acuity at final follow-up for the glaucoma group. Nine out of 29 eyes (31.0%) achieved 20/40 or greater vision, while another 8 eyes (27.6%) achieved visual acuity between 20/50 and 20/60. Some patients had limited visual acuity measurements or outcomes due to preverbal age, developmental status, or cortical visual impairment. F+F = fix-and-follow; BTL = blink-to-light; CVI = cortical visual impairment.

Discussion

In this cohort, about half of the patients with SWS developed glaucoma, and about half of the glaucomatous eyes needed glaucoma surgery during the follow-up period. There was a higher likelihood of SWS patients developing glaucoma and SWG eyes progressing to surgery if their port wine birthmark involved the lower eyelid. Of note, bilateral birthmarks and choroidal hemangiomas were also associated with a higher likelihood of glaucoma diagnosis. In contrast, neurologic features such as seizures and neurosurgical treatments for SWS such as hemispherectomy were not associated with glaucoma diagnosis, possibly because these elements were further away from the eye. Given that SWS expression seems to follow a somatic mosaicism pattern, the proximity of the variables associated with glaucoma onset in this study (lower eyelid angioma and choroidal hemangioma) suggests that they are a biomarker for glaucoma.

Glaucoma prevalence among our SWS cohort (52.3%) was similar to rates found in the literature. Khaier et al. found that 66.7% of patients from their SWS cohort had glaucoma, while Sujansky et al. found 48%, and Sharan et al. found 58.5%.^6,7,13 In contrast, glaucoma prevalence in patients with facial PWBs have varied between 18–53%.^9,13,14 The wide prevalence may depend on the proportion of SWS in each study. For example, Khaier et al. included 13.3% of patients with SWS while Ha et al. included 56% with SWS.^9,13 In the study by Ha et al., the authors also found that lower eyelid PWB involvement was a strong predictor of glaucoma development like in our study.⁹ While lower lid PWB involvement was associated with higher odds of glaucoma diagnosis and needing glaucoma surgery, we did not find any association between PDL treatment for facial PWBs and glaucoma. One previous study also demonstrated lack of correlation between PDL treatment and glaucoma diagnosis.^7,8 Previous studies have also found an association between glaucoma development and bilateral PWBs.^13,14

More children with SWG developed glaucoma during their infancy than those who developed it later, which is consistent with other studies.^6,7 The group that required only medical glaucoma treatment was older than the surgery group at the time of diagnosis. This also may suggest that SWS patients who present with glaucoma later may be more likely to achieve good IOP control from medications alone. Patients in the surgery group and medication group both received medications first (sometimes even as a temporizing measure) which is what we and other groups recommend given the high angle surgery failure rate.^15,16

Similar to other studies, we found that angle surgery was not effective in this group as many patients required additional surgery.^16,17 Subconjunctival filtering procedures such as trabeculectomy and tube shunts may have offered longer IOP control, but these procedures are associated with higher risks, especially in children with SWG.^15,17–19 Since there were fewer numbers of eyes when the surgical group was broken down by surgery type, we could not perform any meaningful risk analysis using our variables of interest. Since goniotomy and ab externo trabeculotomy can be offered with none to minimal subconjunctival scarring, we still recommend angle surgery after initial medical therapy particularly in early onset cases as a means to delay longer-term treatments such as a trabeculectomy or tube shunt in this population.^16,18

Two patients experienced serous choroidal detachments in our study that resolved with proton beam therapy. Neither occurred during the early postoperative period (1.9 and 3.5 years after surgery), and neither occurred in the setting of hypotony. One patient developed an effusion after starting latanoprost, and theoretically increasing uveoscleral outflow could reduce the IOP enough to induce an effusion, which has been previously reported.²⁰ In SWG, post-operative choroidal effusions have been reported in 18–42% of eyes, some of which have resulted in serous retinal detachments.^17–19 Both oral propranolol and proton beam therapy have had reported success in treating choroidal effusions in SWG patients.^19,20 While oral propranolol risks include bradycardia and hypotension, proton beam therapy may cause radiation retinopathy, optic neuropathy, or neovascular glaucoma.²¹ Visual acuities were generally good in our SWG cohort despite half requiring one or more glaucoma surgeries. When excluding those who could not reliably perform visual acuity testing, only two eyes had 20/200 vision or worse. In these two eyes, the presumed causes of poor vision were from end stage glaucoma due to uncontrolled IOP and maculopathy after choroidal effusion.

There were several limitations in this study due to its retrospective nature. Some patients were diagnosed or had prior treatment at other institutions including previous PDL or glaucoma surgery, but we had access to previous records for a majority of patients in this study. Since this was a retrospective review, there was no one protocol for initiating treatment or escalating treatment in these patients. As such, it was difficult to perform a surgical analysis in this cohort due to the small sample size, different ages at presentation, and tendency to need more than one surgery in this disease. Furthermore, the presence of episcleral angiomas, which have been associated with glaucoma in previous studies, was not completely documented in the electronic record, and so we were not able to reliably report on this clinical feature.¹³ In contrast, facial angiomas and choroidal hemangiomas could be reviewed by digital photos in a majority of cases when not documented in the record. Lastly, while several providers managed these patients which may improve its generalizability, a future multicenter study evaluating practice patterns of SWG treatment may better help determine the effectiveness of various surgical techniques for this condition.

In conclusion, we found significant associations between a few vascular features of SWS including bilateral eyelid, lower eyelid, and choroidal angiomas and glaucoma diagnosis, which may be due to their spatial relationship. The biomarkers found in this study may help improve diagnosis and recognition of greater disease severity. Future prospective studies based on this work may help with earlier glaucoma diagnosis and management.