Sirolimus Treatment in Sturge-Weber Syndrome

Abstract

Background:

Sturge-Weber syndrome is a rare neurovascular disorder associated with capillary malformation, seizures, cognitive impairments, and stroke-like episodes (SLE), arising from a somatic activating mutation in GNAQ. Studies suggest this mutation may cause hyperactivation of the mTOR pathway. Sirolimus is an mTOR inhibitor studied in other vascular anomalies, and a potentially promising therapy in Sturge-Weber syndrome.

Methods:

10 patients with Sturge-Weber syndrome brain involvement and cognitive impairments were enrolled. Oral sirolimus was taken for 6 months (maximum dose 2 mg/day, target trough level 4–6 ng/mL). Neuropsychological testing, electroencephalogram (EEG), and port-wine score were done at baseline and after 6 months on sirolimus. Neuroquality of life (Neuro-QoL), adverse events, and Sturge-Weber syndrome Neurological Score (neuroscore) were recorded at each visit.

Results:

Sirolimus was generally well tolerated; one subject withdrew early. Adverse events considered related to sirolimus were mostly (15/16) grade 1. A significant increase in processing speed was seen in the overall group (p=0.031); N=5/9 with available data demonstrated statistically rare improvement in processing speed. Improvements were seen in the Neuro-QoL sub-scales measuring anger (p=0.011), cognitive function (p=0.015), and depression (p=0.046). Three subjects experiencing SLE before and during the study reported shortened recovery times while on sirolimus.

Conclusions :

Sirolimus was well tolerated in Sturge-Weber syndrome, and may be beneficial for cognitive impairments, especially in patients with impaired processing speed or a history of SLE. A future, randomized placebo-control trial of sirolimus in Sturge-Weber syndrome patients is needed to further understand these potentially beneficial effects.

Introduction

Sturge-Weber syndrome (SWS) is a vascular malformation disorder, caused by an activating somatic mutation in GNAQ [1] and associated with capillary malformations including a characteristic facial port-wine birthmark (capillary vascular malformation), leptomeningeal angioma (capillary-venous vascular malformation), and glaucoma [2]. Sturge-Weber syndrome is a rare disease, with an estimated incidence of 1 in 20,000 to 1 in 50,000 live births [2]. The leptomeningeal vascular malformation of the brain is associated with seizures in about 83% of patients [3], and SWS can also be associated with headaches [3], attention difficulties [2], cognitive impairments [4], and stroke-like episodes [5], which can result in transient or permanent hemiparesis [6] and neurological deterioration [5].

Cerebral hemodynamic studies of seizures in patients with Sturge-Weber syndrome suggest that capillary-venous malformation in the brain impairs the normal hemodynamic response during seizures, leading to progressive ischemia; this process can result in cognitive decline and perpetuate additional seizures ultimately resulting in a spiral of neurologic deterioration [7]. Increased seizure frequency in SWS is also associated with poorer cognitive outcomes [8]; thus, typical treatment in SWS places emphasis on aggressive seizure management by using anticonvulsants and, at some centers low-dose aspirin [2]. In infants, presymptomatic treatment with antiseizure drugs and/or aspirin is promising in delaying seizure onset [9]. Patients with medically refractory seizures may benefit from newer anti-seizure drugs tested in SWS such as cannabidiol [10]. Hemispherectomy is also an option in patients with unilateral brain involvement and intractable seizures [11]. Treatment with stimulants may ameliorate attention and behavioral issues in some patients [12]. However, disease-specific therapeutic options for the range of cognitive impairments seen in SWS currently remain limited.

Previous studies have demonstrated that cutaneous vascular malformation samples from patients with SWS have increased expression of phosphorylated S6, a downstream target of AKT signaling [13]. It is hypothesized that the somatic mutation in GNAQ hyperactivates the PI3K/AKT/mTOR pathways in Sturge-Weber syndrome; therefore, SWS patients may benefit from treatment with an mTOR inhibitor, such as sirolimus [1], [13]. Animal models provide further supporting rationale for the cognitive sparing potential of mTOR inhibition in SWS. Mouse models have demonstrated improved outcomes after ischemic brain events with mTOR inhibitor treatment [14], [15], [16] and treatment with sirolimus can improve cognitive function in mouse models of tuberous sclerosis complex [17], Alzheimer’s disease [18] and Downs Syndrome [19].

In one infant with bihemispheric leptomeningeal angiomatosis characteristic of Sturge-Weber syndrome, sirolimus was used as a presymptomatic treatment in conjunction with aspirin [20]; this infant with bilateral brain involvement (typically a marker of early onset seizures with increased severity) remained seizure free at 23-month follow-up. A recently published, retrospective study of sirolimus given to 6 SWS patients (maintaining blood levels between 10 and 15 ng/mL) suggested that sirolimus was safe in SWS, with potential lightening of the port-wine stain and benefits for seizure control observed [21]. Sirolimus use has been studied in other conditions involving hyperactivation of the mTOR pathway, including children aged 3 months-10 years with tuberous sclerosis [22], and people with complicated vascular anomalies aged 0–31 years [23]. In both studies, sirolimus was well tolerated, and findings indicated improvement in quality of life with sirolimus use [23]. The purpose of this study was to determine the safety and efficacy of low dose sirolimus when used as a potential treatment of cognitive impairments in Sturge-Weber syndrome.

Methods

This study received Institutional Review Board approval from Johns Hopkins Medicine and Cincinnati Children’s Hospital Medical Center, and written IRB approved consent was obtained from the participant or their parent/guardian (if under the age of 18) before commencing study procedures or enrollment. An investigational new drug (IND) number from the Federal Drug Administration was obtained prior to initiation of the study. Participants were recruited from the Hunter-Nelson Sturge-Weber Center at the Kennedy Krieger Institute research database, and from the Hemangioma and Vascular Malformation Center’s Sturge-Weber syndrome database at Cincinnati Children’s Hospital Medical Center. Information about the study was posted on ClinicalTrials.gov and the National Institutes of Health Brain Vascular Malformation Website, and was offered for posting to relevant advocacy groups. All participants were seen in clinic by Dr. Comi or Dr. Hammill at least once before consenting to the study. Full eligibility requirements are summarized in Table 1.

Table 1:

Eligibility and Exclusion Criteria

Eligibility Criteria:

Participants with Sturge-Weber syndrome brain involvement as defined on neuroimaging and the following:

• Male or female, ages 3 to 31 years.

• Cognitive impairment, defined as a SWS cognitive neuroscore greater than or equal to one.

• Able to participate in direct neuropsychological and developmental testing.

• English as primary language.

• Stable anti-epileptic drugs (no changes in medications, except for dose, for more than 3 months).

• Adequate renal function (glomerular filtration rate greater than 50 ml/min/m2).

• If female and of child bearing potential: documentation of a negative pregnancy test prior to enrollment, determined by a urine test.

○ Use of adequate contraceptive measures, excluding estrogen containing contraceptives and including abstinence, while on sirolimus in sexually active pre-menopausal female patients (and in female partners of male patients).

• International Normalized Ratio (INR) <1.5 (Anticoagulation is allowed if target INR < 1.5 on a stable dose of warfarin or on a stable dose of low-molecular-weight (LMW ) heparin for >2 weeks.)

• Adequate liver function (serum bilirubin <1.5x upper limit of normal (ULN); alanine aminotransferase (ALT) and aspartate aminotransferase(AST) <2.5x ULN)

• Written, informed consent obtained from the patient or the patient’s legal representative, in accordance with local guidelines for consent and assent.

• Stable does of medications affecting the cytochrome P-450 3A4 (CYP3A4) and p-glycoprotein (P-gp) systems for at least 3 months prior to consent.

Eligibility was determined at the initial screening and consenting visit.

Exclusion Criteria:E

• Allergy to sirolimus or other rapamycin analogues.

• Patients with seizures secondary to metabolic, toxic, infectious or psychogenic disorder, drug abuse or current seizures related to an acute medical illness.

• Inability to keep follow-up appointments, maintain close contact with Principal Investigators, and/or complete all necessary studies to maintain safety.

• Patients in need of immediate major surgical intervention.

• Concurrent severe and/or uncontrolled medical disease, which could compromise participation in the pilot study (e.g. uncontrolled diabetes, uncontrolled hypertension, severe infection, severe malnutrition, chronic liver or renal disease, active upper gastrointestinal (GI) tract ulceration, impaired or restrictive pulmonary function, pneumonitis or pulmonary infiltrates).

• Chronic treatment with systemic steroids or another immunosuppressive agent. Patients with endocrine deficiencies are allowed to receive physiologic or stress doses of steroids if necessary. Inhaled steroids are allowed.

• Known history of HIV seropositivity or known immunodeficiency. Testing is not required unless a condition is suspected.

• Impairment of gastrointestinal function or gastrointestinal disease that may significantly alter the absorption of sirolimus (e.g. ulcerative disease, uncontrolled nausea, vomiting, diarrhea, malabsorption syndrome or small bowel resection). A gastric tube or nasogastric tube is allowed.

• Patients with an active, bleeding diathesis.

• Patients with uncontrolled hyperlipidemia: fasting serum cholesterol > 300 mg/dL AND fasting triglycerides > 2.5 x ULN.

• Patients who have had a major surgery or significant traumatic injury within four weeks of study entry. Patients who have not recovered from the side effects of any major surgery (defined as requiring general anesthesia) or patients that may require major surgery during the course of the pilot study.

• Patients with a prior history of organ transplant.

• Patients who have received live attenuated vaccines within one week of start of sirolimus and during the course of the pilot study

• Patients with a prior history of organ transplant.

• Patients who have received live attenuated vaccines within one week of start of sirolimus and during the pilot study.

• Patients who have a history of malignancy.

• Patients who are currently part of or have participated in any clinical investigation with an investigational drug within one month prior to enrollment.

• Patients being treated with felbamate, unless treatment has been continuous for greater than or equal to one year.

• Patients currently receiving anticancer therapies or who have received anticancer therapies within four weeks of study entry (including chemotherapy, radiation therapy, antibody based therapy, etc.).

Subjects were enrolled from March 2017 through December 2018. Subjects received twice daily oral dosing of sirolimus liquid (Rapamycin ®/Rapamune ®- study drug provided by Pfizer), which was studied as an investigational new drug for treatment of cognitive impairments in Sturge-Weber syndrome, for 6 months. The goal trough serum level was 4–6 ng/mL, which was selected as lower dose sirolimus (trough levels of 3–7 ng/mL) has become standard practice for non-life threatening vascular malformations [24]. Each participant started sirolimus at week 0, using a dosage calculated based on their height and weight. Sirolimus levels were drawn 1 hour and 3 hours after the first dose for pharmacokinetic (PK) testing; PK evaluations were performed using Bayesian estimation with the clinical software program MW/Pharm (ver. 3.82, Mediware, Prague, Czech Republic) and as described previously [25], [26]. Based on these results, the sirolimus dosage was adjusted if necessary. Sirolimus doses were measured a week later, at visit 3, and at all subsequent study visits. To ensure safety for this study, maximum dose was limited to 2 mg/day.

The study consisted of 6 visits, with the option to continue on sirolimus clinically if significant improvements were identified. The first visit was a screening visit (week −2). Subsequent visits took place at baseline (week 0, study baseline; to start sirolimus), and at weeks 2, 6, 14 and 26. Prior to starting sirolimus, participants underwent baseline neuropsychological testing and electroencephalogram (EEG). These measures were repeated at week 26, after 6 months on sirolimus. The primary study outcome measures were: 1) safety and tolerability of sirolimus in SWS patients, and 2) change in cognitive function after six months of sirolimus treatment. Figure 1 shows the complete schedule of study events.

Figure 1:

Table of Visits

Neuropsychological testing measured change in cognitive function, and consisted of the following National Institutes of Health (NIH) Toolbox measures [27]: the Flanker Inhibitory Control and Attention Test [28] to determine change in executive function and attention, the Dimensional Change Card Sort Test [29] to study change in executive function, the Picture Sequence Memory Test [30], [31] to determine changes in episodic memory, the Oral Reasoning Recognition Test and the Picture Vocabulary Test to determine change in language skills, the Pattern Comparison Processing Speed Test [32] to study change in processing speed, the List Sorting Working Memory Test [33] to determine change in working memory, the 9-hole Peg Test [34] to evaluate change in dexterity, the Grip Strength Test [35] to determine grip strength, and the Patient-Reported Outcomes Measurement Information System (PROMIS) Test [36] to evaluate self-report of emotional functioning. The testing panel was created in consultation with a neuropsychologist experienced in evaluating patients with SWS and familiar with commonly affected domains [4], [37]. The NIH Toolbox cognitive measures have multiple streams of evidence supporting the reliability and validity of the individual subtests used [31], including criterion-related validity studies showing test score differences between neurological groups (e.g., spinal cord injury, traumatic brain injury, and stroke) [38]. Moreover, the NIH Toolbox cognitive measures have been shown to have a “low test floor” allowing precision assessment of children with developmental disorders as well as cognitive and functional impairment at a level similar to children with SWS [39]. A masters-degree trained neuropsychological associate or a neuropsychologist obtained the necessary background information (subject age, native language, subject and parent level of education, handedness) and all testing was administered via an iPad®.

To monitor drug safety and tolerability, all visits consisted of a physical exam, SWS Neurological Score (“neuroscore,” as previously reported, to quantify clinical severity of seizures, visual field cut, hemiparesis, and cognitive impairment [40], Appendix 1), screen for adverse events, updated medical history, interval seizure history, and rescue medication administration, as applicable. Subjects, or their parent/guardian, filled out the Quality of Life in Neurological Disorders (Neuro-QoL) measure (Appendix 2, 3), which includes questions to evaluate physical, emotional, and social functioning of people living with neurological conditions [41]. Patients also provided reports of change in quality of life since the last visit and since starting sirolimus using a Likert like scale at each visit. Vitals and lab work (anticonvulsant levels, sirolimus levels, complete blood count and chemistries, lipid profile, urinalysis, and urine pregnancy test in females Tanner stage 2 and above) were also collected. Between visits, biweekly check-ins with the participants were done by phone or email to determine if there were any adverse events or changes (positive or negative) since the most recent study contact.

Quantitative EEG (qEEG) was a companion metric and consisted of analysis of routine, wakeful EEGs done prior to starting sirolimus and after 6 months on the study drug. Changes in asymmetry of power (indication of neurologic progression, [42], [43]) were chosen as the qEEG measure.

To determine if sirolimus had an effect on patient’s facial vascular malformation, photographs and scoring of the port-wine birthmark, if present, were completed at baseline and after 6 months of treatment. For each subject, the color, hypertrophy, total surface area involvement, and eyelid involvement was assessed using the port-wine score (Appendix 4). Photographs of the front, left, and right facial profiles were also collected.

Statistical Methodology

All data analysis was done in IBM SPSS Statistics 25, unless otherwise stated.

Demographics:

Simple frequencies were generated to determine distribution of sex, race, ethnicity, and age at enrollment of the 10 subjects, and occurrence/severity of adverse events.

Neuropsychological Testing:

Primary Analyses:

All test panels were normed for age and demographic background (e.g., NIH Toolbox Fully Adjusted scores) and utilized published standards to determine change in function. For grip strength and dexterity, right and left measurements were reorganized into dominant and weak hand variables. A one sample, 2-tailed t-test was used to determine if participants’ baseline scores on the NIH-toolbox or PROMIS differed significantly from the reference value (T=50). A paired sample, 2-tailed t-test was used to determine if there was a statistically significant change in any of the NIH toolbox domains, PROMIS domains, dexterity, or strength after 6 months on sirolimus. Corrections of p-values were not made for multiple comparisons, as the focus of this pilot study was on safety and the study was not adequately powered to fully assess efficacy.

Secondary Analyses:

Secondary analysis was planned for any NIH-toolbox domain with a significant mean score increase or decrease. In any domain showing a statistically significant mean score increase or decrease, reliable change index (RCI) methodology [44] was used to identify the number of participants with statistically rare changes in standardized score. This involved creating a 90% confidence interval around each participant’s baseline score to determine if his or her follow-up fell within or outside of this interval. 90% confidence intervals were calculated as follows: 1) Reliability coefficients were identified for each NIH Toolbox task. Given the high frequency of intellectual deficits identified in individuals with SWS [37], RCI was calculated using the reliability coefficients from a NIH Toolbox reliability study involving a cohort of individuals with developmental disabilities [39]. 2) Reliability coefficients (r_xx) from Hessl et al. were used to calculate standard error of measurement (SEM) values, i.e., (SEM=SD(r_xx-1)^1/2) for each NIH Toolbox task, followed by 3) the calculation of the standard error of the difference (SEdiff), i.e., (SEdiff = (2(SEM)²)^1/2) and 4) the calculation of the 90% confidence interval (SEdiff *±1.64). Follow-up scores falling outside of the 90% confidence interval were considered statistically rare. N=3 patients with a history of severe stroke-like episodes with prolonged recovery times had events while on the trial. In light of this, an additional paired sample, 2-tailed t-test was done to determine if a significant change in any neuropsychological testing domain was present among patients with stroke-like episodes only.

EEG and qEEG Analysis:

Primary Analyses:

All EEG readings and qEEG analysis were done at the Kennedy Krieger Institute. Baseline and follow-up clinical EEG interpretations were compared for qualitative improvement. Evidence of epileptiform activity (interictal spikes and/or sharp waves), slowing, asymmetry in EEG activity (asymmetry in amplitude and/or frequency) or other age-related abnormalities were noted. qEEG included a quantification of asymmetry in power analysis as previously described [45], [43]. Values of asymmetry in power analysis were normed so that asymmetry on the side of SWS brain involvement was positive, and asymmetry on the opposite side of SWS brain involvement was negative. One patient had bilateral brain involvement, however their involvement was asymmetric with decreased power on the left side, thus this was considered their side of involvement for norming purposes. Normed values within each brain region (whole hemisphere and occipital lobe, which was chosen based on its frequent involvement in SWS and previous studies [43] ) and frequency band (alpha, beta, theta, delta, and total) were compared using a two-tailed, paired t-test in Microsoft Excel to determine if there was a significant change in asymmetry over the 6-month treatment period.

Secondary Analyses:

Additional analysis was done to determine if there was an interaction effect of qEEG and a history of SLE using a repeated measure ANOVA. In all frequency bands where a significant interaction was present, post hoc independent-sample, 2-tailed t-tests were used to determine if there was a significant difference in asymmetry among patients with and without a history of SLE at baseline and follow-up. Similar ANOVA and post-hoc t-test analyses were also done to examine the interaction effect of qEEG and statistically rare improvement on the NIH-toolbox domains (only within domains that were significant in the overall group).

Quality of Life in Neurological Disorders:

Change in Neuro-QoL between screening and 6 month follow up were determined for all children (N=8 participants). Raw scores for each domain were calculated based on the National Institute of Neurological Disorders and Stroke (NINDS) manual [46]; prorated scores were calculated as outlined in the manual for patients who left a question blank. These scores were converted to t-scores according to NINDS guidelines. A paired, two-tailed t-test was done to determine if there was a significant change in Neuro-QoL domain between screening and follow-up. One subject did not complete enough questions in the visit 6 anger domain; for this subject, visit 5 data for the anger domain was used; the analysis was also run with this subject’s data excluded. One patient came off of sirolimus early (after visit 5); the analysis was run using Visit 5 data for this subject and also excluding this subject. A one sample t-test was also done to determine if scores were significantly different in SWS pediatric patients than the normal population. Change in Neuro-QoL was also determined for the adult subjects using the adult form (N=2), however numbers were small.

Neuroscore:

To determine if there was a significant change in neuroscore, a 2-tailed Wilcoxon-signed rank test was used to compare seizure, hemiparesis, visual field cut, cognitive impairment, and composite neuroscores at screening and after six months on sirolimus. One patient came off of sirolimus early (after visit 5); the visit 5 neuroscore was used for the analysis; however this patient’s neuroscore was unchanged between visit 5 and 6.

Change in Port-wine Birthmark:

To determine if there was a change in the port-wine birthmark, baseline and follow-up scores were compared using a 2-tailed Wilcoxon-signed rank tests. To corroborate these results, the principal investigators at each center rated the photographs taken at each visit in a blinded, retrospective fashion.

Patient Reported Quality of Life:

At all visits after starting sirolimus, participants and their parents were asked if their quality of life was better (mildly, moderately, a lot), worse (mildly, moderately, a lot) or unchanged since the last visit, and since starting sirolimus. Response at visit 6 was used to determine if there was a change in quality of life.

Results

Demographics

Ten patients (Table 2) were recruited (9 Caucasian, 5 Female, mean age=12 years, 7 months ± 5 years, 2 months). One patient withdrew from the study early due to a series of mild adverse events, which included seizures/pseudoseizures, behavioral issues, aggression, and a mild mouth sore. Their follow-up neuropsychological testing and EEG was done after visit 5. At the end of the study, 4 of the 9 subjects who completed the study elected to stay on sirolimus clinically. Two of these patients stayed on the drug for a short period of time (N=1 about 6 weeks, N=1 about 6 months), and 2 patients remain on sirolimus clinically at the time of manuscript preparation.

Table 2:

Participant Demographics

Subject	Center†	Sex^	Race	Ethnicity	Age*	Side of Brain Involvement	Stroke-Like Episode	Improved Cognitive Function▴
1	KKI	F	White	Not Hispanic/Latino	8	Left	No	No
2	KKI	F	White	Not Hispanic/Latino	16	Left	No	Yes
3	KKI	F	White	Not Hispanic/Latino	12	Right	Yes	Yes
4	KKI	M	White	Not Hispanic/Latino	18	Left	Yes	No
5	KKI	M	White	Not Hispanic/Latino	20	Left	No	No
6	KKI	F	White	Not Hispanic/Latino	5	Left	No	Yes
7	CCHMC	M	Black/African American	Not Hispanic/Latino	12	Bilateral (Left>Right)	No	Data not available
8	CCHMC	M	White	Not Hispanic/Latino	12	Left	Yes	Yes
9	CCHMC	M	White	Not Hispanic/Latino	17	Right	No	No
10	CCHMC	F	White	Not Hispanic/Latino	6	Left	No	Yes

^†KKI- Kennedy Krieger Institute, CCHMC- Cincinnati Children’s Hospital Medical Center;

^{^}F- Female; M-Male;

^*Age at registration, in years;

^▴Statistically Rare Improvement.

Sirolimus Doses and Adverse Events

The target sirolimus serum level was 4–6 ng/mL, however doses were not to exceed 2 mg/day. 6 of the 10 subjects achieved the target trough level at some point during the study; 4 subjects were below target for the entire study. Adverse events considered to be related to the study drug were all grade 1 (mild), with the exception of one upper respiratory infection (grade 2, possibly related). 9 of the 10 subjects had an adverse event that was considered to be possibly or probably related to sirolimus. All related adverse events can be found in Table 3. Adverse events deemed unrelated to sirolimus are shown in Supplemental Table 1.

Table 3:

Adverse Events Considered Possibly or Probably Related to Sirolimus

Related Adverse Event	Number of Subjects Experienced	Number of Events	Grade	Relatedness (# of instances)
Abdominal pain	1	1	1	Possibly Related (1)
Mucositis oral/Mouth Ulcer	3	7	1	Probably Related (5) Possibly Related (2)
Nausea	1	1	1	Possibly Related
Nightmares	1	1	1	Possibly Related(1)
Upper respiratory infection	1	1		Possibly Related (1)
Increased Alanine Aminotransferase	1	1	1	Possibly Related (1)
Increased Aspartate Aminotransferase	2	3	1	Possible Related (3)
Cholesterol high	1	4	1	Possibly Related (4)
Hypertriglyceridemia	4	9	1	Possibly Related (9)
Decreased HDL cholesterol	2	9	1	Possibly Related (5)
Abnormal Urine Labs (Elevated urobilinogen, increased protein in urine)	1	2	1	Possibly Related (2)
Headache	2	4	1	Possibly Related (4)
Seizure	1	2	1	Possibly Related (2)
Somnolence	1	1	1	Possibly Related
Personality change	2	2	1	Possibly Related (2)
Behavioral Issues/Aggression	1	1	1	Possibly Related (1)

Three subjects had a history of prominent stroke-like episodes associated with seizures, ultimately requiring prolonged hospitalizations followed by increased rehabilitative needs. All three of these subjects had at least one stroke-like episode while on sirolimus, but were reported by the parents and the investigators to have a significant reduction in the severity and duration of the stroke-like episode symptoms, and in one case the frequency of the episodes was also significantly reduced. These events were not considered to be triggered by the study drug considering past history in all three cases, however the marked improvement in severity of stroke-like episodes while on sirolimus was noted with interest.

Cognitive Impairment

Participants

NIH-toolbox data was available for nine subjects, with the exceptions of the sequential memory domain (available for eight subjects, due to an app malfunction) and the working memory domain (available in seven subjects; due to app/keyboard malfunction). PROMIS testing data was available in five subjects; missing data was due to poor comprehension (n=1), fatigue (n=1), the inability to complete (n=1), young age (n=1), and missing demographic information needed for norming scores (n=1). Hand strength data was available in five subjects (four subjects were missing the necessary equipment; n=1 testing was not done for one subject). Dexterity data was available for eight subjects in the dominant hand, and seven in the weak hand (one subject could not complete in their weak hand). Missing data was due to patient fatigue in one subject, testing was not done in one subject.

Baseline Measures

At baseline on the NIH Toolbox cognitive tasks, mean t-scores were significantly below the reference value of t=50, suggesting a pattern of cognitive deficit in all domains except Picture Vocabulary (Supplemental Table 2). Despite significant cognitive deficits at baseline, the individuals with SWS did not differ significantly from the reference value of t=50 on any of the PROMIS measures.

Neuropsychological Testing Results

At six-month follow-up, statistically significant improvement was noted on the Pattern Comparison Processing Speed Test (p=0.031). Changes in other domains measured by the NIH-toolbox (executive function, attention, episodic memory, language skills, working memory), or on the PROMIS, did not reach statistical significance; all p-values and mean scores at baseline and follow-up can be found in Table 4. Stability of hand strength and dexterity were maintained even when patients who experienced stroke-like episodes were excluded.

Table 4:

Neuropsychological Testing Outcomes Before and After Sirolimus Use

Test	Variable	Average ± Standard Deviation at Baseline	Average ± Standard Deviation After 6 Months on Sirolimus	p-value
NIH Toolbox	Vocabulary	44.2±12.5	42.3±12.3	0.196
	Attention	34.1±10.5	35.2±14.1	0.666
	Working Memory	26.3±9.8	28.1±3.9	0.540
	Executive Function	33.1±10.2	35.7±15.4	0.375
	Processing Speed	27.0±14.5	36.4±17.6	0.031
	Sequential Memory	39.3±12.3	40.9±8.6	0.451
	Recognition	37.7±12.8	36.0±10.4	0.429
PROMIS	Anxiety	43.5±8.3	35.4±20.5	0.372
	Depression	45.9±7.9	35.7±20.4	0.317
	Fatigue	53.8±5.1	36.7±15.3	0.099
	Pain Interference	49.2±11.1	36.1±18.8	0.095
	Physical Function	46.9±10.9	41.9±15.7	0.577
	Peer Relations	49.8±10.9	44.1±17.5	0.339
Grip Strength	Dominant	39±14.4	42±5.1	0.630
	Non-Dominant	26.1±28.2	20.6±28.9	0.440
Dexterity	Dominant	30.6±16.8	31.6±14.3	0.824
	Non-Dominant	32.3±19.5	30.1±19.9	0.306

Bold type denotes a significant p-value.

Secondary Analysis

When patients with stroke-like episodes were studied (N=3), improvement in processing speed at follow-up was significant (p=0.041). RCI analysis revealed that five of the nine subjects with available data demonstrated a statistically rare improvement on the Pattern Comparison Processing Speed task. Although interesting, the small sample size (n=3) is a limitation that must be underscored.

EEG

Initial Analyses

Clinical EEG readings for 5/10 subjects found no difference in EEG before and after starting sirolimus One subject demonstrated asymmetry of the waking posterior background rhythm at baseline that was less asymmetric at follow-up. One subject showed decreased activity over the left hemisphere at baseline, with a normal EEG reading at follow-up. One subject showed an asymmetry in the posterior background rhythm at follow-up that was not present at baseline, another subject showed left posterior temporal slowing at follow-up that was not present at baseline. One subject had spikes independently seen on the right and left at follow-up, as compared to just the left at baseline. These findings are summarized in Supplemental Table 3. When qEEG analysis of all 10 subjects was studied, change in mean power after using sirolimus was not statistically significant in any of the bands (Supplemental Table 4).

Stroke-Like Episodes

In comparing baseline qEEG in patients with (N=3) and without stroke-like episodes (N=7), the repeated measures ANOVA demonstrated a significant interaction effect of stroke-like episodes and qEEG in the alpha frequency band of the occipital lobe (p=0.041, partial eta squared=0.424). Post hoc analysis by t-test demonstrated that the occipital alpha band was significantly different at baseline (p=0.001), but not at follow-up (p=0.113). A decrease in power asymmetry in patients with stroke-like episodes was observed in these frequency bands between baseline and follow-up (0.69 to 0.43; 40% change, see Figure 2). The beta, theta, delta and total frequency bands of the occipital lobe and alpha, beta, theta, delta and total frequency bands of the whole hemisphere did not demonstrate a significant interaction effect of group (±SLE)*qEEG time-point (Tables 5 and 6).

Figure 2:

qEEG Stroke-Like Episodes Figure made using IBM SPSS Statistics 25

Table 5 :

Results of Within-Subjects Contrasts, EEG^* Stroke-Like Episode

	Band	p-value	Effect Size*
Whole Hemisphere	Delta	0.644	0.028
	Theta	0.897	0.002
	Alpha	0.089	0.318
	Beta	0.337	0.115
	Total	0.295	0.136
Occipital Lobe	Delta	0.943	0.001
	Theta	0.437	0.077
	Alpha	0.041	0.424
	Beta	0.256	0.158
	Total	0.150	0.240

Bold type denotes a significant p-value.

^*partial eta squared

Table 6:

Difference in Mean Power Asymmetry of the Occipital Alpha Frequency Band at Baseline and Follow-up

EEG	Patients with stroke-like episodes	Patients without stroke-like episodes	p-value
Baseline	0.686±0.03	0.167±0.23	0.001
Follow-up	0.426±0.19	0.247±0.13	0.113

Bold type denotes a significant p-value

Statistically Rare Improvement in Cognitive Function

There was no significant effect of the interaction of statistically rare improvement in processing speed and qEEG asymmetry (Supplemental Table 5).

Quality of Life in Neurological Disorders

Eight participants were children (under age 18), and completed the pediatric Neuro-QoL form as previously described [41]. Compared to the population norm, this sample of pediatric SWS patients had significantly lower scores in the pain (p=0.035) and depression (p=0.002) domains. SWS patient scores in anger, anxiety, social relations, stigma, cognitive function, and fatigue did not differ significantly from the rest of the population (Supplemental Table 6).

Compared to baseline evaluations, there were significant improvements in the anger (p=0.011, mean 50.7±11.6, 41.8±9.1 at baseline and 6-month follow-up, respectively), depression (p=0.046, mean 41.7±5.0, 38.1±3.2 at baseline and 6-month follow-up, respectively), and cognitive function (p=0.0146, mean 46.3±6.1, 51.1± 5.9 at baseline and 6-month follow-up, respectively) domains after 6 months on sirolimus. The other domains generally showed improvements as well, though change was not statistically significant; p-values and mean t-scores are shown in Table 7.

Table 7:

Pediatric Neurologic Quality of Life Scores at Baseline and After 6 Months on Sirolimus

Neurologic Quality of Life Domain	Average T-score ± Standard Deviation at Baseline	Average T-score ± Standard Deviation After 6 Months on Sirolimus†	p-value
Anger	50.7±11.6	41.8±9.1*	0.011
Anxiety	50.1±9.7	47.8±6.1	0.32
Pain	44.0±6.5	42.4±7.7	0.539
Social Relations	55.9±8.1	55.2±10.9	0.797
Stigma	45.7±8.7	42.4±7.0	0.090
Depression	41.7±5.0	38.1±3.2	0.046
Cognitive Function	46.3±6.1	51.1±5.9	0.015
Fatigue	46.5±7.1	43.1±7.3	0.185

Bold type denotes a significant p-value;

^*For one subject, visit 6 data was incomplete; visit 5 data was used instead.

^†One subject stopped sirolimus after visit 5, their visit 5 Neurologic Quality of Life data was used.

Improvements remained significant when the visit 6 data was used for the subject who discontinued early (p=0.013 anger; p=0.046 depression; p=0.013 cognitive function); when the subject who discontinued early was removed entirely, improvements in the anger (p=0.03) and cognitive function (p=0.034) domains remained significant, improvement in the depression domain (p=0.10) did not; when the subject with incomplete visit 6 anger data was excluded, p=0.024 anger. Two patients were over the age of 18, and completed the adult Neuro-QoL (NIH Adult NeuroQoL) as previously described [47]. There were no significant changes in any domains on the adult form; p-values and mean t-scores are shown in Supplemental Table 7.

Reported Quality of Life

In addition to the formal scoring, subjects were asked to rate their quality of life using a Likert type scale. At visit six, 6 of the 9 subjects who completed the trial reported that their quality of life had improved since starting sirolimus (N=4 mildly better, N=2 a lot better); 2 patients reported their quality of life was worse (N=1 mildly, N=1 moderately), and 1 patient reported no change in their quality of life since starting sirolimus. The patient who reported moderate worsening was experiencing high levels of anxiety due to external stressors (unrelated to the study drug) around the time of visit 6. One subject stopped sirolimus after visit 5 due to side effects; at visit 5, this subject reported that their quality of life was moderately worse since starting sirolimus, but reported no change in their quality of life since starting sirolimus when they returned for their visit 6 follow up. These patient reports of change in quality of life since starting sirolimus are summarized in Table 8; changes in quality of life since the last visit are reported in Supplemental Table 8.

Table 8:

Patient Reported Outcomes Since Starting Sirolimus

Change in Quality of Life Since Starting Sirolimus	Number of Subjects
A lot Better	2
Moderately Better	0
Mildly Better	4
No Change	1
Mildly Worse	1
Moderately Worse	2*
A lot Worse	0

^*One of these subjects stopped the trial early (after visit 5); moderately worsened quality of life reflects their response at visit 5, however at their visit 6 follow up (off sirolimus), the subject reported no change in their quality in life since starting sirolimus.

Neuroscores

There was no significant change in seizure, hemiparesis, visual field cut, cognitive function, or composite neuroscores between the baseline evaluation and after 6 months on sirolimus (Table 9). However, four patients had an improvement in their composite neuroscore between the first and final visit. Four subjects showed no change, and two subjects showed worsening. Supplemental Table 9 shows neuroscores at the screening visit and after 6 months on sirolimus.

Table 9:

Change in Median Neuroscore Between Baseline and Follow-up

Neuroscore Domain	Baseline	Six Month Follow-up	p-value
Seizure	1	1	1.000
Hemiparesis	1	1	1.000
Visual Field Cut	0.5	0	0.312
Cognitive Function	2.5	2.5	1.000
Composite	4.5	4.5	0.625

Port-wine Birthmark

Change in port-wine birthmark score did not approach statistical significance. Table 10 shows the change in port-wine birthmark score, and the results of retrospective blinded review by the principal investigators; Supplemental Table 10 shows all scoring.

Table 10:

Change in Port-wine Birthmark Score Between Visits 2 and 6

	Port-Wine Score	Blinded Review of Photographs
Improved	4	2
No Change	2	6
Worsened	2	0
No Port-Wine Birthmark	2	2

Other Reported Findings

Anecdotal improvements mentioned by the participant or the participant’s parents were noted during clinic visits and at biweekly check-ins. The most frequently reported improvements were improved school reports (N=6), improved behavior (N=5), increased vocabulary, improved reading, and increased independence (all reported in 3 subjects). A full list of noted improvements can be found in Supplemental Table 11.

Discussion

Overall, our findings in this pilot study indicate that low dose sirolimus is generally safe and well tolerated in patients with Sturge-Weber syndrome. Improvements in cognitive impairment were remarkable, with rare improvements in over half of the subjects. While the sample size (n=3) was small, notable changes were also demonstrated in the three patients with a history of stroke-like episodes; these three patients demonstrated improved processing speed when studied separately from the entire group, hastened SLE recovery, and improved qEEG findings, indicating these changes had neuropathophysiological implications as well.

Safety of Sirolimus in Sturge-Weber syndrome

Adverse events considered to be related to sirolimus were generally grade 1 and mild. The most frequently reported related adverse events were hypertriglyceridemia (4 subjects), which was managed with diet, and mouth sores (3 subjects). These events were expected and previously reported in other studies of sirolimus [23], [22]. Other events experienced by 2 or more participants included headaches, increased liver function tests (N=1 dosage was decreased, N=1 not clinically relevant), decreased high density lipoprotein (HDL) cholesterol, and personality change (N=1 childlike behavior after a seizure, N=1 mild worsening of behavior). Aside from one subject who withdrew from the study early due to a cluster of mild adverse events, sirolimus was well tolerated. It is important to note that 9 of 10 subjects had unilateral brain involvement; thus, the safety of sirolimus is not necessarily generalizable to all patients. The safety of sirolimus in bilaterally affected patients should be studied further.

Improvement in Cognitive Functioning

At baseline, most patients’ test scores were well below age and demographic normative levels for executive functioning, attention, episodic memory, working memory, and processing speed. These findings suggest a cross-domain pattern of cognitive deficit in individuals with SWS. After six months on sirolimus, mean processing speed scores improved for the group, and five of the nine subjects showed a statistically rare improvement in individual processing speed scores relative to the expected pattern of test score stability seen in a disabled comparison population [39]. Given the 90% confidence interval, only 1/10 subjects would have been expected to see this change. Moreover, anecdotal improvements made by the participants included six subjects who reported improved school reports, three subjects who reported improved reading, and three subjects who reported improved vocabulary. These quantitative and qualitative findings support our hypothesis of cognitive improvement associated with sirolimus treatment in individuals with SWS. Since n=9 of the 10 subjects had unilateral brain involvement, however, these findings may not be generalizable to bilateral SWS cases, who tend to be the more severely affected. Further studies are needed to understand the efficacy of sirolimus in attenuating cognitive impairments in patients with bilateral SWS.

Increased processing speed correlates with increased working memory and enhanced fluid intelligence [48]. Furthermore, increased processing speed mediates an increase in general intelligence occurring during adolescence [49] thus, preserving or improving processing speed is important for cognitive development in children, especially those with SWS who may already have lower processing speed than their neurotypical counterparts. Supporting development of working memory and general intelligence may improve outcomes for patients with Sturge-Weber syndrome in school and work later in life.

Cognitive function in unilateral SWS is often stable, and may improve over time even without epilepsy surgery. One study, which followed SWS patients over a median of 2 years, found that in a non-surgical cohort of 33 children with unilateral SWS brain involvement, the majority of patients had stable or improved cognitive function, [50], possibly due to prolonged seizure control and/or functional reorganization in the unaffected hemisphere that may lead to partial reversal of some cognitive deficits. Given the relatively narrow time window of the study (6 mo), and the stability of seizure severity in the cohort (7/10 subjects show no change in seizure severity on neuroscore), time-linked functional reorganization of cortex or improved seizure control seem unlikely to explain the positive effects observed To further elucidate the effects of sirolimus on cognitive impairment in SWS, a prospective, randomized placebo controlled trial will be needed.

EEG

No clear trends were noted in the clinical EEG readings as a group. Three patients possibly became worse, 2 possibly became better, and the remaining 5 subjects did not show a change. Patients with SLE showed a different response to sirolimus than other patients. Recognizing the limitations of our small sample size on the robustness of the statistical conclusions, they started from a different qEEG starting point (greater asymmetry) and “normalized” after treatment. Change in qEEG was not seen among patients who displayed a statistically rare improvement in processing speed, suggesting that qEEG laterality scores may be sensitive to changes in brain function induced by vascular pathophysiology, but changes in processing speed may be due to a different pathophysiology not detectable by qEEG. . Analyses were run on the group level (comparing all patients with stroke-like episodes to all patients without stroke-like episodes) rather than on the individual level, and further work will need to be done to determine whether these qEEG changes may be informative for a single patient [51], [52].

Improvement in Quality of Life

Overall, sirolimus use was associated with improved quality of life. On the Neuro-QoL, there was a significant decrease in the anger and depression domains, and a significant increase in the cognitive function domain after 6 months on sirolimus (pediatric subjects). Improvement in the cognitive function domain was supported by findings of improved processing speed in the neuropsychological testing and aforementioned anecdotal reports of improved school and academic performance. Additionally, anecdotal improvements in behavior (N=5) and independence (N=3) were reported by the participants or their families, and were possibly reflective of improvements in mood. Finally, 6 of the 9 subjects who remained on sirolimus for the full six months reported improvement in their overall quality of life since starting the drug. Previous studies have found that children with SWS have a significantly lower cognitive function quality of life on the Neuro-QoL assessment compared to a control population [53] and that people with Sturge-Weber syndrome may be at greater risk of suicidal thoughts or actions compared to neurologically involved controls [54]. Thus, improvement in these quality of life measures with sirolimus use is an important finding.

Improved Recovery After Stroke-Like Episodes

Three patients experienced a stroke-like episode while enrolled in the study; all three subjects had a history of severe stroke-like episodes with prolonged recovery times prior to starting the trial. Compared to previous episodes, these patients experienced significantly shortened recovery times while using sirolimus; two of the three subjects continued on sirolimus clinically after the trial for this reason. Patients with stroke-like episodes also had significantly improved processing speed after 6 months on sirolimus. Though this groups is so small as to preclude statistical inferences, these findings are nonetheless interesting and could inform future studies of sirolimus in SWS.

In one study of 277 SWS patients, about 37% reported stroke-like episodes and about 11% experienced symptomatic stroke (defined as “a period of weakness or other deficit on one side of the body lasting longer than 24 hours” and “an abrupt onset of a new neurological deficit which does not fully resolve within a period of one month.”, respectively) [55]. Studies in patients with Sturge-Weber syndrome demonstrate hypoperfusion in the affected cerebral areas on single photon emission computed tomography (SPECT) studies [56]; other cerebral hemodynamic studies of seizures in SWS suggest impaired hemodynamic response to seizures and abnormal venous drainage during seizures may lead to ischemia, further compromising the cortex and compounding deficits with each seizure episode [7]. The effects of mTOR inhibitors on stroke and ischemic events are an evolving area of research [57]. Studies on sirolimus-eluting stents have shown suppression of neointimal proliferation [58], potentially by inhibiting abnormal smooth muscle cell proliferation and migration [59]. In mice, preconditioning with sirolimus before a focal cerebral ischemic event with reperfusion improved survival, decreased neurological deficit score (24 hours post event), and decreased the area of infarction [14], [15]; it is proposed that increased autophagy caused by the mTOR inhibitor has a neuroprotective effect [15], [16]. A future, multi-centered study is needed to understand how sirolimus may affect recovery from stroke-like episodes in SWS patients.

Sirolimus Use in Tuberous Sclerosis and Other Vascular Malformations

Sirolimus use investigated for medically refractory seizures in tuberous sclerosis [22] did not show improvement in seizure control or cognitive function in a cohort of tuberus sclerosis patients. However, the patient population with tuberous sclerosis was severely intellectually disabled and suffered intractable epilepsy, and thus differed significantly from the SWS patients in the study reported here. In contrast to the tuberous sclerosis cohort, our cohort had varying severity of epilepsy and mild-moderate cognitive impairment (ability to complete neuropsychological testing was required). Interestingly, the target trough level in tuberous sclerosis patients was slightly higher, at 5–10 ng/mL, although only 14 of the 23 participants were able to achieve this [22].

In a study of sirolimus use in patients with complicated vascular anomalies, 85% had a partial response to treatment by the end of the of the 12 courses (about 336 days), with a significant improvement in quality of life measurements overall and 80% showing partial response to functional impairments by the end of the 12 courses [23]. Trough levels in this study were greater than ours (10–15 ng/mL), and there were more events of toxicity [23]. This study included a wide range of phenotypes, and was specific to severely affected patients most of whom had failed previous therapies and interventions [23]; thus, the SWS sample differed from this patient cohort in several ways. [17–19]

Limitations and Conclusions

Limitations in this pilot study of safety of sirolimus for cognitive impairments in Sturge-Weber syndrome included a small sample size (N=10) with mostly Caucasian participants. Small sample sizes are unavoidably a limitation in rare disease research. In the future, multi-centered studies will be needed to further study sirolimus in patients with Sturge-Weber syndrome. Statistical analyses were not corrected for multiple comparisons, as this was a pilot study underpowered to fully evaluate drug efficacy. Patients with SWS and stroke-like episodes will be of particular interest. Quantitative EEG analysis and port-wine birthmark scoring (from pictures) were done in a blinded fashion; however, clinicians, of necessity, were not blinded to study participation, and all participants were on drug, therefore physicians could be susceptible to response and/or reporting biases. In addition, dosages on this study were not to exceed 2 mg/day; this level was chosen for its known safety and tolerability, but some patients were never able to achieve the target trough level, so it is possible that these subjects could have shown additional improvement on a higher dose. The Neuro-QoL is normed for children ages 8–17, however N=2 of our subjects were younger than 8 while on the trial; family assisted with answering these questions for the child. Finally, neuropsychological data was not available for all subjects for each domain. In conclusion, these results suggest that sirolimus treatment may be promising in abating impairments in processing speed, and shortening recovery time for SWS patients with stroke-like episodes.

APPENDIX

Appendix 1: Neurological exam (also known as the “Neuroscore”)

Seizures
0	No seizures
1	1+, but controlled for the last 6 months
2	Breakthrough seizures during the last 6 months, but not monthly
3	Monthly
4	Weekly +
Hemiparesis
0	Normal
1	Intermittent postures
2	Fine motor impairment
3	Fine and gross motor impairment
4	Severe fine and gross motor impairment, poor helper arm function, walks poorly/not at all
Visual Field Cut
0	None
1	Partial hemi-field cut
2	Full hemi-field cut
Cognitive Function (Child*; 5–18 years)
0	None
1	School difficulties, regular classes
2	Resource help needed in school
3	Special education required
4	Trainable for activities of daily living
5	Full care
Cognitive Function (Adult; 18+ years)
0	None
1	Lives and works independently
2	Works in community with parental support
3	Significant difficulty maintaining employment or satisfactory social relationships
4	Trainable (group home, supervised work setting)
5	Full care

^*The child delineation in age for Cognitive Function is meant to coincide with when the child begins school.

Appendix 2: Pediatric Neuro-QoL Form

Appendix 3: Adult Neuro-QoL Form

Appendix 4: Port Wine Score