Development and Retrospective Application of Novel Outcome Measure: Sturge–Weber Syndrome Acute Crisis (SWAC) Index

ABSTRACT

Objective

Sturge–Weber syndrome (SWS) symptoms frequently overlap in neurological acute crises (ACs) presenting with seizures, stroke‐like episodes (SLE), and headaches. Recent efforts to develop a prospective drug trial highlighted the need for a new outcome measure capable of quantifying these symptoms during an SWS AC. We aimed to develop such a measure and evaluate its practical application.

Methods

The Sturge Weber Acute Crisis (SWAC) index was developed based on clinical care guidelines and experience. We retrospectively applied the SWAC to an open‐label prospective drug trial published in 2020; SWAC scores were assigned for every AC during 6‐month baseline and treatment phase periods.

Results

Six of 10 subjects experienced 670 ACs; only 78/670 ACs were included in the primary quantitative analyses (one subject was an extreme outlier) and 71/78 events were single‐symptom while 7/78 included multiple symptoms. Three of six subjects experienced improved median seizure total score; all three improved their median SLE total score. Three of five subjects with headache improved their median headache total score. Poisson‐log GLMM models demonstrated significant improvements in composite severity and total, SLE duration and total, and headache severity and total index scores.

Interpretation

The newly developed SWAC index, applied retrospectively, reflected site‐PI clinical experience, quantified the severity and duration of SWS ACs ranging from mild to severe, and better enabled comparison of treatment versus baseline responses. These results support additional studies with the SWAC index to further assess clinical utility and use in future treatment trials in SWS.

1. Introduction

Sturge–Weber syndrome (SWS) is a congenital neurovascular disorder characterized by capillary‐venous malformations of the leptomeninges and brain, face (port‐wine birthmark), and eye (choroidal hemangioma commonly resulting in glaucoma). The characteristic port‐wine birthmark and features of SWS brain involvement are depicted in Figures 1 and 2, respectively. The disease is rare and is estimated to occur in 1 of every 20 000–50 000 births [1]. A facial port‐wine birthmark (PWB) in a high‐risk area (forehead, eyelids, and temple) is typically the first indicator of SWS, and approximately 20% of these individuals receive diagnoses of SWS brain involvement [1, 2, 3, 4]. Diagnosis of SWS brain involvement is made when the leptomeningeal vascular malformation is visualized on contrast‐enhanced magnetic resonance imaging (MRI). Other brain abnormalities associated with SWS include abnormal, tortuous leptomeningeal vessels, impaired blood flow, venous hypertension, and often atrophy and calcification of involved brain regions. The clinical presentation of SWS includes often‐overlapping symptoms of seizures, headaches, and stroke‐like episodes. Neurological signs may include hemiparesis, visual field cuts and other vision deficits, developmental delay, and cognitive impairment; these can develop over time and are strongly associated with seizures and stroke‐like episodes.

Figure 1.

Facial port‐wine birthmark. Photograph of a port‐wine birthmark in a high‐risk area, typically the first indicator of Sturge‐Weber syndrome.

Figure 2.

Neuroimaging depicting features commonly seen in SWS brain involvement. Axial neuroimaging depicting typical features of SWS. Panel A (contrast‐enhanced T1) depicts leptomeningeal enhancement predominantly in the left occipital and parietal regions (bottom arrow, panel A); enhancement of the choroid plexus within the left ventricle is also visible (top arrow, panel A). Left‐sided deep draining medullary veins are readily visualized with SWI (Panel B). Calcifications in the left frontal and temporal lobes may be seen in panel C (head CTA). Abbreviations: CTA, computed tomography angiography; SWI, susceptibility‐weighted imaging; SWS, Sturge‐Weber syndrome.

Epilepsy is estimated to occur in 80%–90% of patients with SWS and an individual's risk of epilepsy is thought to be related to their extent of brain involvement [5, 6]. Due to the neurovascular pathophysiology of SWS, seizures can trigger further impaired cortical blood flow, potentially resulting in venous stroke or stroke‐like episode (SLE) [7, 8, 9, 10, 11]. SLEs are transient focal neurological deficits that can last between 24 h and 3+ months. These transient SLEs are estimated to occur in one‐third of SWS patients and recur in 20% of individuals with an SLE history [12, 13, 14, 15]. SLEs often follow a seizure, headache, or minor head trauma [12, 15]. Headaches are more common in SWS—a recent meta‐analysis placed the prevalence rates between 37% and 71% [16].

To date, seizures, SLEs, and headaches in SWS have been primarily studied and reported separately, although they are frequently experienced in combination. For instance, individuals with a National Institutes of Health Stroke Scale (NIHSS) score of 0 may still experience headaches—an impairment commonly seen in SWS [7, 17]. The SWS Neuroscore, a clinical scoring system used to quantify the neurological status, is one of few metrics validated for use in SWS; it ranges from 0 to 15, where higher scores indicate worse seizure frequency, cognitive function, and visual field deficits [18, 19, 20, 21, 22, 23]. The SWS Neuroscore, however, evaluates chronic, not acute, neurological status. Furthermore, the SWS Neuroscore does not capture data on SLEs or headaches, two major acute complications of SWS.

As SWS is causatively associated with mutations in GNAQ and its paralogues, drugs influencing the affected molecular pathways have been the target of investigation. SWS most commonly results from a somatic mosaic gain‐of‐function GNAQ R183Q mutation, which encodes guanine nucleotide‐binding protein G(q) subunit alpha (Gαq) [1, 24, 25] and is theorized to produce constitutive hyperactivation of downstream effectors in pathways, such as PI3K‐mTOR. Other mutations in GNAQ, in its paralogue GNA11 [26], and in other related genes [27] have also been associated with atypical SWS; however, R183Q GNAQ mutations account for the vast majority of cases [28]. Evidence for the hyperactivation of the mTOR pathway in SWS comes from histological examination of affected SWS tissue with increased expression of phosphorylated S6, a downstream target of the mTOR pathway, in both facial capillary malformations and leptomeningeal endothelial cells [29, 30]. Consequently, drugs targeting the PI3K‐mTOR pathway, such as sirolimus (rapamycin; an mTOR inhibitor), have been studied as potential therapeutic agents [31, 32, 33, 34, 35, 36]. In 2021, we reported the results of our pilot trial of sirolimus for cognitive impairments in SWS. In the process of developing a larger, multi‐center phase II trial protocol, the planning committee determined that a new, combined outcome measure was needed to capture the nature, duration, and severity of these neurological ACs (seizure, SLE, headache; AC). Thus, we aimed to develop and assess the practical application of an index capable of fully capturing these ACs. Here, we report on the development of such an index, applied retrospectively to the subjects from our prior sirolimus trial.

2. Methods

2.1. Process and Consent

Between 2023 and 2025 during a series of meetings, the authors developed the Sturge‐Weber Acute Crisis (SWAC) index, based on the combined clinical senior author expertise (AMC and AH) in the acute and chronic care of patients with SWS at their respective centers over the past two decades. All subjects were consented under an IRB‐approved protocol for the original (previously published) drug trial (KK: IRB00079722; CCHMC: IRB#2016‐6538/IRB#2024‐0199) and under a second set of IRB‐approved protocols allowing for retrospective data analysis (KK: NA_00043846; CCHMC: IRB#2024‐0199).

2.2. SWAC Index

The SWAC index, attached as Supplemental S1, consists of three parts. Part 1 of the form is used to score both the severity and duration of each of the three main symptoms (seizure, SLE, and HA). The symptom scores from part 1 are then used to calculate Part 2 and Part 3 indices.

Part 2 of the SWAC index calculates (from part 1 scores; see above) an overall SWAC Composite Severity (SCS), SWAC Composite Duration (SCD), and SWAC Composite Total score (SCT) (see Supplemental S1, Part 2) and describes the overall severity (range 0–15) and duration (range 0–12) for the AC; adding these together results in a composite total (range 0–27). Higher SCS indicates worse ACs requiring greater medical intervention; higher SCD reflects longer‐lasting symptoms; and the SCT reflects both severity and duration of the AC.

Part 3 of the SWAC calculates Symptom Domain indices, again from scores in part 1. The SWAC contains three SWAC Symptom Domains (seizure, stroke‐like episode, and headache), with each symptom domain including a severity, duration, and total (SSD‐SZ, SSD‐SLE, and SSD‐HA). Therefore, the SSD index reflects the contribution (severity + duration) of each symptom (seizure, SLE, and/or HA) to the AC being scored.

2.3. Clinical Data

The original study consisted of a 6‐month baseline and 6‐month treatment period. We reviewed the study case report forms and clinical records from these periods, which were used to complete part 1 of the SWAC form (see above); the SWAC form was completed retrospectively by senior authors (AC and AH). Figure 3 depicts the study schema. Each subject had a SWAC form completed for each individual AC during the baseline and treatment periods. All subjects had typical SWS brain involvement as confirmed by MRI with and without contrast. The 10‐subject cohort consisted of five females, and overall age ranged from five to 20 years; nine subjects identified as white and one as Black/African American. Seven subjects had unilateral left‐sided SWS brain involvement, two unilateral right‐sided, and one bilateral left greater than right‐sided brain involvement. Subject 1 withdrew from the study early, completing 5 of 6 months on treatment. The remaining nine subjects completed the original trial in its entirety. For additional demographics and detail on subjects and the trial itself, see previously published manuscript [31]. Subject 1 did not complete the entire trial period and was an extreme outlier in terms of the frequency of brief, atypical myoclonic seizures experienced. Thus, they were included in descriptive statistics only.

Figure 3.

Study schema. The SWAC index was retrospectively applied to a previous sirolimus trial. The trial enrolled 10 subjects with SWS and consisted of two phases, each lasting 6 months: (1) baseline and (2) treatment. All acute crises that occurred during these two 6‐month periods were scored and formed the study dataset (670 total acute crises). Abbreviations: ACs, acute crises; SWAC, Sturge–Weber Acute Crisis; SWS, Sturge–Weber syndrome.

2.4. Database and Data Analysis

Data management and analysis occurred within Excel and SPSS. Parametric analyses (Mann–Whitney U and pairwise Mood's median tests) compared scores from the treatment phase with those from the baseline phase. For these statistical tests, analyses of SWAC symptom domains (SSD) included only ACs with scores greater than 0. All ACs were included in SWAC composite analyses (SC). Lastly, demographics and SWAC scores are reported via descriptive statistics.

A Poisson‐log generalized linear mixed models (GLMM) was used to assess the effects of treatment on SWAC scores. A random effect was placed on each individual to account for within‐subject clustering. The Poisson model was chosen due to non‐normality and skewedness of the data toward 0. The Poisson model was selected over competing models (e.g., multinomial) as it yielded a significantly better fit based on the Bayesian Information Criteria. Incidence rate ratios (IRR), 95% confidence intervals, and p‐values are reported for each model. Scores of 0 were included in all GLMM analyses as these models account for the characteristics of the data. P‐values less than 0.05 determined statistical significance.

3. Results

3.1. Descriptive Results

In total, 670 ACs were scored, 399 of which occurred during the baseline period (59.6%) and 271 during the treatment phase (40.4%). As summarized in File S2, subjects 6, 7, 9, and 10 experienced no ACs during the trial period. Subject 1 experienced daily 1 s myoclonic jerks, which accounts for 592/670, or 88.4% of these data; their data are included in descriptive but not quantitative results due to them being an extreme outlier. Of the remaining 78 scored events, nine were seizure only (11.5%), zero were SLE only, 62 were headache only (79.5%), and seven were combined ACs in which multiple neurovascular symptoms were experienced (9.0%). All combined episodes were in the subjects who experienced SLEs (subjects 4, 5, and 8).

Six subjects experienced seizures, three experienced SLEs, and five experienced headaches. Three subjects experienced a decrease in seizure frequency (one subject increased in frequency), one subject experienced fewer SLEs (two unchanged), and two subjects saw decreases in the number of headaches experienced (three increased). Frequencies for each symptom are outlined in File S2.

Figure 4 panel A displays the range of SWAC composite (SCS, SCD, and SCT) scores experienced. The composite total score ranged from 0 to 18 at baseline and 0 to 17 on treatment. Three subjects improved their median SCT scores, one worsened, and two experienced no change; median SCS scores improved similarly. Two subjects experienced decreased overall AC duration (SCD), one worsened, and three were unchanged.

Figure 4.

Range of SWAC scores assigned for all acute crises. The SWAC composite severity (SCS) and each SWAC symptom domain severity (SSD) score is plotted against their respective duration scores (composite [SCS and SCD], panel A; seizure [SSD‐SZ], panel B; stroke‐like episode [SSD‐SLE], panel C; headache [SSD‐HA], panel D). All 670 scored acute crises are included. Total scores are the sum of the duration and severity scores. Subjects experienced acute crises ranging from mild to severe, all of which were captured and quantified by the SWAC index. Abbreviations: SCS, SWAC composite severity; SSD, SWAC symptom domain; SSD‐HA, SWAC symptom domain‐headache; SSD‐SLE, SWAC symptom domain‐stroke‐like‐episode; SSD‐SZ, SWAC symptom domain‐seizure; SWAC; Sturge‐Weber Acute Crisis.

Figure 4 panels B–D shows the range of SWAC symptom domain scores (seizure, panel B; SLE, panel C; headache, panel D). SSD‐SZ total scores ranged from 0 to 9, SSD‐SLE scores ranged from 0 to 9, and SSD‐HA scores ranged from 0 to 8. Three subjects experienced an improvement in their median SSD‐SZ total scores (3/6, 50%), two worsened, and one was unchanged. All subjects with SLE experienced improvements in median SSD‐SLE total scores (3/3, 100%). Four of five subjects' median SSD‐HA total scores improved (4/5, 80%) while one worsened. The point change in median SSD scores is shown in Figure 5.

Figure 5.

Point change in median subdomain total scores from baseline to treatment. Positive values indicate an improvement (decrease in SWAC score) whereas negative values indicate worsening (increased SWAC score). Subject 1 experienced no SLE or headaches, subjects 2 and 5 experienced no SLE, and subjects 6, 7, 9, and 10 experienced no acute crises during the trial period. Abbreviations: AC, acute crises; HA, headache; SLE, stroke‐like episode; SWAC, Sturge‐Weber Acute Crisis; SZ, seizure.

3.2. Quantitative Results

3.2.1. SWAC Compound Severity, Duration, and Total Scores

From baseline to treatment, subjects experienced significantly improved SCS (baseline mean rank = 55.9, treatment mean rank = 31.2, U = 394, n = 86, p < 0.001), SCD (baseline mean rank = 53.0, treatment mean rank = 34.0, U = 514, n = 86, p < 0.001), and SCT scores (baseline mean rank = 54.9, treatment mean rank = 32.2, U = 437, n = 86, p < 0.001). Similarly, significantly improved medians were noted in subjects' SCS (baseline median = 3, treatment median = 1, p < 0.001) and SCT scores (baseline median = 5, treatment median = 2, p < 0.001). As shown in Table 1, the results from GLMM revealed significant improvements in composite SCS and SCT scores only. The expected SCS score was 1.8 times greater in the baseline phase compared with the treatment phase (95% CI = [1.2, 2.6], p = 0.004). Similarly, the expected composite total score was 1.6 times greater in the baseline phase compared with the treatment phase (95% CI = [1.2, 2.1], p < 0.001) (Table 1).

Table 1.

Generalized linear mixed modeling (GLMM).

SWAC index	β (coefficient)	t	p	expβ(IRR)	95%CI lower (expβ)	95%CI upper (expβ)
Composite Severity (SCS)	0.571	2.959	0.004*	1.770	1.206	2.265
Composite Duration (SCD)	0.377	1.821	0.072	1.457	.966	2.199
Composite Total (SCT)	0.483	3.413	< 0.001**	1.621	1.223	2.148
Seizure Severity (SSD‐SZ Severity)	−0.590	−1.348	0.181	0.555	0.232	1.323
Seizure Duration (SSD‐SZ Duration)	0.067	0.280	0.780	1.069	0.665	1.718
Seizure Total (SSD‐SZ Total)	−0.584	−1.864	0.066	0.558	0.299	1.040
Stroke‐like Episode Severity (SSD‐SLE Severity)	0.228	1.380	0.171	1.256	0.904	1.744
Stroke‐like Episode Duration (SSD‐SLE Duration)	0.635	3.114	0.003*	1.887	1.258	2.831
Stroke‐like Episode Total (SSD‐SLE Total)	0.401	3.133	0.002*	1.494	1.158	1.927
Headache Severity (SSD‐HA Severity)	0.818	3.475	< 0.001**	2.265	1.419	3.617
Headache Duration (SSD‐HA Duration)	0.420	1.708	0.091	1.522	0.933	2.481
Headache Total (SSD‐HA Total)	0.652	3.782	< 0.001**	1.920	1.362	2.705

Note: GLMM models demonstrated significant improvements in composite severity and total, SLE duration and total, and headache severity and total index scores.

Abbreviations: β, beta coefficient; expβ, exponentiated beta coefficient; IRR; incidence rate ratio; SCD, SWAC composite duration; SCS, SWAC composite severity; SCT, SWAC composite total; SSD, SWAC symptom domain; SSD‐HA, SWAC symptom domain‐headache; SSD‐SLE, SWAC symptom domain‐stroke‐like episode; SSD‐SZ, SWAC symptom domain‐seizure; SWAC; Sturge‐Weber Acute Crisis.

^* P‐values less than 0.05

^** P‐values less than 0.001.

3.2.2. SWAC Symptom Domain Scores

No statistically significant changes were detected in SSD‐SZ scores, as shown in Table 1. The expected SSD‐SLE duration score was 1.9 times greater in the baseline phase compared with the treatment phase (95% CI = [1.3, 2.8], p = 0.003); the expected SSD‐SLE total score was 1.5 times greater (95% CI = [1.2, 1.9], p = 0.002) (Table 1). GLMM (Table 1) also demonstrated improved SSD‐HA severity (IRR = 2.3, 95% CI = [1.4, 3.6], p < 0.001) and total scores (IRR = 1.9, 95% CI = [1.4, 2.7], p < 0.001).

4. Discussion

Seizures, stroke or SLEs, and headaches are the most common causes of acute and transient neurological deterioration in SWS [12]. In this study, they are termed acute crises (ACs). ACs negatively affect quality of life and can result in extended hospitalizations, rehabilitation needs, extended pain and suffering, or even death. As such, future SWS research must target these ACs, and a comprehensive index for these transient events represents important progress. The data summarized by the SWAC help to summarize, both quantitatively and qualitatively, their clinical experience with each subject and overall clinical assessment of the trial results.

Treatment required to address an AC may reasonably approximate the severity of an AC. Thus, required treatment formed one basis for assessing AC severity; in addition, for SLE and headaches, degree of impairment (are activities constrained, or can normal activities continue) was also incorporated into part 1 severity responses. The least severe ACs require no medical intervention and are minimally impairing (scores of 1 or 2), while more severe ACs require medication, hospitalizations, or ICU stays and are more impairing (scores of 3, 4, and 5). AC duration in part 1 reflected symptom duration intervals appropriate to each AC symptom as determined by the extensive clinical SWS experience of the two senior authors.

Combined symptom episodes occurred for three subjects; all had histories of prolonged SLE beginning in childhood. These subjects experienced impactful and meaningful reductions in SLE duration after 6 months on sirolimus. A decrease in SLE duration score from 4 to 1, for example, as was the case for subject 8, translates to a reduction in time lived with symptoms from greater than 1 month (score of 4) to less than 24 h (score of 1). This change reflects a markedly improved quality of life and illustrates how the SWAC may be leveraged with descriptive statistics to describe changes in complex acute events.

Within SWS, there is a large degree of variability in how individual patients present across time, and it is therefore difficult to discern the therapeutic effect that sirolimus may have had on any one episode. The authors considered both the variable symptoms and the need to capture individuals on both ends of a spectrum of AC, from those with brief and low acuity ACs to those experiencing high‐acuity episodes with extended rehabilitation needs and deficits. This is well shown in Figure 3, which illustrates the range of ACs experienced across nearly all severity and duration score levels.

For assessing an acute neurological patient event, there is no available, universally recognized “Acute Medical Crisis Index” applicable to all possible patient situations. Rather, there are various measures developed to assess and quantify the severity of different neurological events and to predict outcomes. Commonly used examples of such indices include the Glasgow Coma Scale and the NIHSS; these scales do not, however, capture the unique aspects of ACs in SWS. Like other disease‐specific indices and scales [37], the eventual goal of the SWAC is (1) to guide treatment decisions for patients with SWS and as an outcome measure for clinical trials and (2) to offer a common language to track and understand a patient's condition. Nevertheless, efforts to adapt the SWAC index to other neurological or medical conditions with ACs (particularly with multiple or combination symptoms) may also prove useful in the future. Examples of such neurological disorders could include some channelopathies and metabolic disorders.

The process of developing a new neurological index or scale has several important steps [38]. The first is to define the concept under study, identify the indicators of the phenomenon, evaluate the utility (or lack thereof) of current measures, define the population in which the measure will be applied, and then evaluate the properties of the measure upon data collection. These steps have been taken for the SWAC, and it has been validated to the extent that the results reflect the intended phenomenon. Further validation is needed, including evaluation of reproducibility and reliability, for future multi‐centered trial use.

4.1. Strengths and Limitations

The SWAC index utilizes clinical data that does not rely on imaging such as MRI or electroencephalography, often not done in the context of a mild or short AC, and can thereby be used to capture the full range of ACs; this provides a practical and equitable advantage over these high‐resource tools for both drug trials and clinical monitoring. The data used for SWAC development were high‐quality, relevant, and accurate. However, the population recruited for this study consisted of relatively stable subjects with SWS, and we did not focus enrollment efforts on those actively experiencing ACs. Low sample sizes further restricted analysis for the SLE subdomain, of which only seven ACs occurred across three subjects. Nevertheless, the descriptive and quantitative statistics produced accurately reflect the observation data from the clinical trial. The utility of the SWAC as a clinical monitoring tool or research outcome measure requires future confirmation by additional users at other sites; interrater reliability also requires future study.

The full SWAC score best captures ACs from patients capable of communicating their symptoms and whose events are reliably reported; patients having suspected behavioral rather than ictal events will be more difficult to quantify. Headaches are difficult to identify in young children, and a version of the SWAC used in young children could be adapted by removing the headache section. The questions and answers provided in part 1 were written for completion by a neurologist or sufficiently trained clinician/research assistant. In the future, the SWAC may also be adapted to allow for parent/patient completion at home or while hospitalized at a local hospital. To enable the SWAC to more fully capture changes in quality of life across a broader spectrum, future efforts could focus on systematically gathering and quantifying family‐reported factors that impact their child's quality of life during and after an acute neurological crisis. The authors envision the use of the SWAC as complementary to, rather than replacing, other single‐symptom scales (e.g., NIH stroke scale) for assessing these patients.

5. Conclusions

We developed and investigated the practical application of a novel AC clinical tool and research outcome measure, the SWAC index. The results support the ability of the SWAC to capture the wide range of AC neurological symptoms experienced by patients with SWS; however, further research on the reliability and validity of the SWAC is required. The current SWAC index is scored rapidly by clinicians and is best suited for children and adults with SWS able to describe their symptoms. Clinically, the SWAC index can quantify the severity of acute episodes or gauge treatment response in a patient over an extended period by comparing subdomain or composite scores over time. Thinking through the complexity of issues, as has been done in the development and initial application of the SWAC index, represents an important next step for future clinical trials to improve the treatment and quality of life for these patients.

Author Contributions

Kieran D. McKenney: conceptualization, data curation, formal analysis, investigation, methodology, project administration, validation, visualization, writing – original draft, writing – review and editing. Luther G. Kalb: formal analysis, methodology, validation, writing – review and editing. Adrienne M. Hammill: conceptualization, data curation, funding acquisition, investigation, methodology, project administration, resources, writing – review and editing. Anne M. Comi: conceptualization, data curation, funding acquisition, investigation, methodology, project administration, resources, supervision, writing – review and editing.

Conflicts of Interest

Anne M. Comi is on the editorial board of the Annals of the Child Neurology Society. The other authors declare no conflicts of interest.

Supporting information

Supplemental 1: Sturge‐Weber Acute Crisis Index (SWAC).

Supplemental 2: Median SWAC scores and frequency of ACs.