Seizure Outcomes and Reoperation in Surgical Rasmussen Encephalitis Patients

BACKGROUND:

Rasmussen encephalitis (RE) is a rare inflammatory disease affecting one hemisphere, causing progressive neurological deficits and intractable seizures.

OBJECTIVE:

To report long-term seizure outcomes, reoperations, and functional outcomes in patients with RE who underwent hemispherectomy at our institution.

METHODS:

Retrospective review was performed for all patients with RE who had surgery between 1998 and 2020. We collected seizure history, postoperative outcomes, and functional data. Imaging was independently reviewed in a blinded fashion by 2 neurosurgeons and a neuroradiologist.

RESULTS:

We analyzed 30 patients with RE who underwent 35 hemispherectomies (5 reoperations). Using Kaplan-Meier analysis, seizure-freedom rate was 81.5%, 63.6%, and 55.6% at 1, 5, and 10 years after surgery, respectively. Patients with shorter duration of hemiparesis preoperatively were less likely to be seizure-free at follow-up (P = .011) and more likely to undergo reoperation (P = .004). Shorter duration of epilepsy (P = .026) and preoperative bilateral MRI abnormalities (P = .011) were associated with increased risk of reoperation. Complete disconnection of diseased hemisphere on postoperative MRI after the first operation improved seizure-freedom (P = .021) and resulted in fewer reoperations (P = .034), and reoperation resulted in seizure freedom in every case.

CONCLUSION:

Obtaining complete disconnection is critical for favorable seizure outcomes from hemispherectomy, and neurosurgeons should have a low threshold to reoperate in patients with RE with recurrent seizures. Rapid progression of motor deficits and bilateral MRI abnormalities may indicate a subpopulation of patients with RE with increased risk of needing reoperation. Overall, we believe that hemispherectomy is a curative surgery for the majority of patients with RE, with excellent long-term seizure outcome.

ABBREVIATION:

RE

Rasmussen encephalitis.

Rasmussen encephalitis (RE) is a rare inflammatory neurological disorder in which patients experience drug-resistant focal epilepsy with progressive neurological decline.¹ Most patients will present in childhood or young adulthood, with median age of onset at 6 years.² Without treatment, patients can develop severe neurological deficits, including hemiparesis, hemianopsia, cognitive decline, and dysphagia.^3,4 This acute decline is followed by a residual stage with persistent neurological deficit and medically refractory epilepsy. Although immunomodulatory treatments may slow disease progression, surgery remains the best option to address the epilepsy.⁵ Advancements in surgical technique allow for functional disconnection of the affected hemisphere, a less invasive equivalent to anatomic resection.

Given the morbidity of hemispherectomy, it is important to optimize patient selection through careful weighing of patient characteristics and potential outcomes. Existing RE literature is limited by small sample sizes or meta-analysis methodologies with restricted ability to evaluate the full clinical spectrum and possible predictors of outcomes.^3,6-13 Reported one-year seizure freedom rates are between 68% and 95%, 5-year seizure freedom rates are between 44% and 72%, and 10-year seizure freedom rates are between 22% and 52%.^6,9,10,14-17 We report here our institution's experience in 30 patients with RE who underwent surgery, one of the largest cohorts to date, to propose possible predictors of seizure freedom and reoperation.

METHODS

We studied all patients who underwent surgery for RE at our institution between 1998 and 2020. All patients were diagnosed with RE based on the consensus of our multidisciplinary patient management conference, which includes epileptologists, neurosurgeons, and neuroradiologists. This study was approved by our Institutional Review Board (17-1015). Informed consent was waived. We identified 35 total cases of surgery for RE (Figure 1). All surgeries were complete hemispherectomies (either functional or anatomic) and performed by a single surgeon. Each patient is extensively discussed in our patient management conference for risk stratification before surgery. Patients with previous history of resective epilepsy surgery had neuroimaging and scalp electroencephalography findings evaluated before being offered either functional or anatomic hemispherectomy, depending on the individual residual anatomy.

FIGURE 1.

Flowchart of patients diagnosed with Rasmussen encephalitis at Cleveland Clinic between 1998 and 2020. ¹This grouping included patients with prior brain biopsy (n = 7) or history of multiple subpial transections (n = 1) as, neither procedure was classified as resective epilepsy surgery.

We collected clinical variables from chart review of the patient demographics, imaging findings, epilepsy history, operative details, and overall hospital course. Seizure outcome at the time of last follow-up, defined by Engel classification, was used as the primary outcome.¹⁸ These data were obtained through our institution's internal Epilepsy Center Outcomes Registry. Date of last follow-up and seizure recurrence (if applicable) were collected by chart review. Whether a patient required reoperation was also used as an outcome.

Functional status was determined by assessment of hemiparesis, fine finger movements, ambulatory status, visual function, and language ability. These data were collected from a combination of chart review and surveys that were administered to patient parents and guardians. Postoperative functional status was subsequently classified as being unchanged, declined, or improved at last follow-up. Cognitive status data were available for less than half of the patients and formal neurocognitive testing was only available for few patients, so this was excluded from final analysis.

MRI scans were reviewed independently by a neuroradiologist and 2 neurosurgeons, who were blinded to associated outcomes data. All patients had preoperative imaging available for review of whether bilateral abnormalities were present. The MRI scans were specifically reviewed for whether the contralateral hemisphere also had evidence of atrophy and/or T2 hyperintense signal. Postoperative MRI scans were reviewed to determine whether disconnection was properly completed after hemispherectomy and whether the insula was entirely resected. Examples of incomplete disconnection on postoperative MRI are demonstrated in Figure 2.

FIGURE 2.

A-F, MRI scans demonstrating incomplete hemispheric disconnection (A-C, sagittal views; D-F, coronal views, residual posterior basal frontal lobe noted by arrows).

Statistical Analysis

We evaluated each individual surgery as a data point (where n = 35). This was to assess the outcomes of functional vs anatomic hemispherectomy. For some analyses, using each individual patient as a data point (where n = 30) made more clinical sense. For example, when assessing reoperation as an outcome, using the reoperations themselves as additional data points would not be appropriate.

In the univariate analysis, we used 2 separate subgroup comparisons to identify possible predictors of seizure freedom and reoperation. In the first comparison, the 35 surgical cases were separated into 2 groups, Engel I and Engel II-IV, depending on seizure freedom at the time of last follow-up. In the second analysis, the 30 individual patients were separated into 2 groups depending on whether they underwent reoperation. Categorical variables were analyzed using Fisher's exact test, whereas continuous variables were analyzed using Student's t-test. Variables with P-value <.05 were considered statistically significant. Kaplan-Meier survival curves were constructed to assess the time to postoperative seizure recurrence in the overall group, and to compare outcomes between functional and anatomic hemispherectomy. All statistical analyses were performed using GraphPAD Prism. The data set was poorly suited for proper multivariate analysis because of the small sample size.

RESULTS

Our study included 30 patients who underwent a total of 35 surgeries between 1998 and 2020 for a diagnosis of RE. Ages ranged between 2 and 28 years at the time of surgery, with an average of 8.6 years. Overall, 74.3% of patients were Engel class I after surgery. Most patients (71.4%) had functional hemispherectomy and the remainder had anatomic hemispherectomy (28.6%). Engel class 1 outcome was 68% after functional hemispherectomy and 90% after anatomic hemispherectomy (Table 1A). Among patients who had reoperation because of seizure recurrence, 100% were Engel class I postoperatively. In summary, 86.7% of patients were Engel class I after either 1 or 2 surgeries (Table 1B). Average follow-up time was 4.6 years after surgery (Table 1C).

TABLE 1A.

Engel Class After Surgery of All Cases (n = 35)

Surgery	Total number	Engel I	Engel II-IV (n = 9)
Total	35	26 (74.3%)	9 (25.7%)
Functional hemispherectomy	25 (71.4%)	17 (68%)	8 (32%)
Anatomic hemispherectomy	10 (28.6%)	9 (90%)	1 (10%)
Reoperation of our functional hemispherectomy	5 (14.3%)	5 (100%)	0 (0%)

TABLE 1B.

Engel Class After Final Surgery (n = 30)

Surgery	Total number	Engel I	Engel II-IV (n = 9)
Total	30	26 (86.7%)	4 (13.3%)
Functional hemispherectomy	20	17 (85%)	3 (15%)
Anatomic hemispherectomy	10	9 (90%)	1 (10%)

TABLE 1C.

Follow-up Time for All Cases (n = 35)

Surgery	Average follow-up (y)	Standard deviation of follow-up (y)
Total	4.6	4.2
Functional hemispherectomy	4.5	3.9
Anatomic hemispherectomy	4.9	5.0
Reoperation of our functional hemispherectomy	5.4	5.8

Univariate Analysis of Preoperative Variables

Table 2A presents the results of univariate analysis of preoperative variables correlating to seizure outcome. Patients with less than 6 months of hemiparesis before surgery were less likely to achieve seizure freedom (P = .011). Patients with Engel class I outcome after surgery had a greater number of failed antiepileptic drugs, but this finding did not meet statistical significance (P = .072). Univariate analysis was also performed to correlate to whether patients underwent reoperation (Table 2B). Patients were more likely to undergo reoperation if they were age less than 5 years at the time of initial surgery (P = .031), experienced a shorter duration of epileptic seizures by time of surgery (P = .026), have fewer number of failed antiepileptic drugs before surgery (P = .041), had a shorter duration of hemiparesis before surgery (P = .004), or displayed bilateral abnormalities on preoperative MRI (P = .011). Other preoperative variables' correlation with seizure freedom or reoperation was not significant.

TABLE 2A.

Univariate Analysis of Preoperative Variables Associated With Seizure Freedom

Preoperative variables	All cases (n = 35)	Engel I (n = 26)	Engel II-IV (n = 9)	P value
Sexa				1.0
Male	18 (51.4%)	13 (50%)	5 (55.6%)
Female	17 (48.6%)	13 (50%)	4 (44.4%)
Handednessa,b				1.0
Left	4 (11.8%)	3 (12.0%)	1 (11.1%)
Right	30 (88.2%)	22 (88.0%)	8 (88.9%)
Mean age at the time of surgery (y)c	8.6 ± 5.3	9.0 ± 5.6	7.3 ± 4.1	.36
Mean age at first nonfebrile seizure (y)c	4.7 ± 2.8	4.7 ± 3.1	4.8 ± 2.2	.92
Mean number of failed antiepileptic drugsc	6.5 ± 3.0	6.8 ± 3.3	5.3 ± 1.5	.072
Historical use of nonpharmaceutical therapy for epilepsya	30 (85.7%)	22 (84.6%)	8 (88.9%)	1.0
Ketogenic dieta	23 (65.7%)	17 (65.4%)	6 (66.7%)	1.0
IVIGa	15 (42.9%)	10 (38.5%)	5 (55.6%)	.45
Steroidsa	13 (37.1%)	10 (38.5%)	3 (33.3%)	1.0
Plasma exchangea	11 (31.4%)	8 (30.8%)	3 (33.3%)	1.0
History of epilepsy surgerya,d	22 (62.9%)	18 (69.2%)	4 (44.4%)	.24
Biopsya	7 (20.0%)	5 (19.2%)	2 (22.2%)	1.0
Craniotomy for resection (not including hemispherectomy)a	6 (17.1%)	5 (19.2%)	1 (11.1%)	1.0
Hemispherectomya	8 (22.9%)	7 (26.9%)	1 (11.1%)	.65
Multiple subpial resectionsa	2 (5.7%)	2 (7.7%)	0 (0.0%)	1.0
History of epilepsia partialis continua before surgerya	18 (51.4%)	15 (57.7%)	3 (33.3%)	.26
History of hemiparesis before hemispherectomya	32 (91.4%)	24 (92.3%)	8 (88.9%)	1.0
Mean time with hemiparesis (mo)c	20.6 ± 27.9	25.1 ± 30.8	10.1 ± 16.2	.11
≤6 moa,e	7 (25.9%)	2 (10.5%)	5 (62.5%)	.011
>6 moa,e	20 (74.1%)	17 (89.2%)	3 (37.5%)
Mean percentage time of heímiparesisc,f	52.9% ± 28.6%	61.1% ± 27.8%	33.4% ± 20.9%	.011
Bilateral MRI abnormalities notedd,g	12 (40%)g	8 (38.1%)g	4 (44.4%)g	1.0

^aCategorical variables analyzed using the Fisher's exact test.

^b Only 34 patients had handedness data available. Percentages calculated out of the sample sizes: total cases (n = 34), Engel class I (n = 25), and Engel class II-IV (n = 9).

^cContinuous variables analyzed using the Student's t-test (2-tailed), presented as mean ± SD.

^dReoperations were not included. Percentages calculated out of the sample sizes: total patients (n = 30), Engel class I (n = 21), and Engel class II-IV (n = 9).

^eReoperations were not included. Percentages calculated out of the sample sizes: total patients with hemiparesis (n = 27), Engel class I (n = 19), and Engel class II-IV (n = 8).

^fReoperations were not included. Mean percentage calculated as 100 × (duration of hemiparesis in months)/(duration of epilepsy in months).

^gReoperations were not included. Percentages calculated out of the sample sizes: total patients (n = 30), Engel class I (n = 21), Engel class II-IV (n = 9).

TABLE 2B.

Univariate Analysis of Preoperative Variables Associated With Reoperation

Preoperative variables	All patients (n = 30)	Reoperation (n = 5)	No reoperation (n = 25)	P-value
Sexa				1.0
Male	15 (50.0%)	3 (60.0%)	12 (48.0%)
Female	15 (50.0%)	2 (40.0%)	13 (52.0%)
Handednessa,b				1.0
Left	4 (13.8%)	0 (0.0%)	4 (16.7%)
Right	25 (86.2%)	5 (100.0%)	20 (83.3%)
Mean age at the time of surgery (y)	8.6 ± 5.5	5.6 ± 3.2	9.2 ± 5.7	.077
≤5 ya	10 (33.3%)	4 (80.0%)	6 (24.0%)	.031
>5 y	20 (66.7%)	1 (20.0%)	19 (76.0%)
Mean age at first nonfebrile seizure (y)c	4.8 ± 3.0	4.1 ± 1.7	5.0 ± 3.2	.39
Duration of epilepsy at the time of surgery (mo)c	45.1 ± 52.5	17.4 ± 18.5	50.6 ± 55.6	.026
≤12 moa	3 (10.0%)	2 (40.0%)	1 (4.0%)	.064
>12 mo	27 (90.0%)	3 (60.0%)	24 (96.0%)
Mean number of failed antiepileptic drugsc	6.7 ± 3.1	5.0 ± 1.4	7.0 ± 3.2	.041
Historical use of nonpharmaceutical therapy for epilepsya	25 (83.3%)	5 (100.0%)	20 (80.0%)	.56
Ketogenic dieta	18 (60.0%)	5 (100.0%)	13 (52.0%)	.066
IVIGa	12 (40.0%)	3 (60.0%)	9 (36.0%)	.36
Steroidsa	12 (40.0%)	2 (40.0%)	10 (40.0%)	1.0
Plasma exchangea	8 (26.7%)	3 (60.0%)	5 (20.0%)	.10
History of epilepsy surgerya	17 (56.7%)	1 (20.0%)	16 (64.0%)	.14
Biopsya	7 (23.3%)	1 (20.0%)	7 (28.0%)	1.0
Craniotomy for resection (not including hemispherectomy)a	6 (20.0%)	0 (0.0%)	6 (24.0%)	.55
Hemispherectomya	3 (10.0%)	0 (0.0%)	3 (12.0%)	1.0
Multiple subpial resectionsa	2 (6.7%)	0 (0.0%)	2 (8.0%)	1.0
History of epilepsia partialis continua before surgerya	17 (56.7%)	2 (40.0%)	15 (60.0%)	.63
History of hemiparesis before hemispherectomya	27 (90.0%)	4 (80.0%)	23 (92.0%)	.43
Mean time with hemiparesis (mo)c,d	20.6 ± 27.9	3.75 ± 3.1	23.6 ± 29.3	.004
≤6 moa	7 (25.9%)	3 (75.0%)	4 (17.4%)	.042
>6 mo	20 (74.1%)	1 (25.0%)	19 (82.6%)
Mean percentage time of hemiparesisc	0.53 ± 0.3	0.36 ± 0.1	0.56 ± 0.3	.055
Bilateral MRI abnormalities noteda	8 (26.7%)	4 (80.0%)	4 (16.0%)	.011

^aCategorical variables analyzed using the Fisher's exact test.

^bHandedness data only available for 29 patients. Percentages out of the following sample sizes: total patients (n = 29), reoperation (n = 5), and no reoperation (n = 24).

^cContinuous variables analyzed using the Student's t-test (2-tailed).

^dMean time with hemiparesis only calculated for the patients with hemiparesis before initial surgery: total patients (n = 27), reoperation (n = 4), and no reoperation (n = 23).

Univariate Analysis of Operative and Postoperative Variables

Adequate disconnection visualized on postoperative MRI was associated with seizure freedom (P = .021) and not requiring reoperation (P = .034). There were no other operative variables that were associated with seizure outcome (Table 3A) or with reoperation (Table 3B). Patients who underwent reoperation were significantly more likely to experience readmission because of surgical complication (P = .022) and more specifically, postoperative hydrocephalus (P = .023) and shunt placement after their initial hemispherectomy (P = .023). There were no other postoperative variables associated with seizure outcome or reoperation (Tables 4A and 4B).

TABLE 3A.

Univariate Analysis of Operative Variables Associated With Seizure Freedom

Operative variables	All cases (n = 35)	Engel I (n = 26)	Engel II-IV (n = 9)	P value
Side of surgerya				1.0
Left hemisphere	17 (48.6%)	12 (46.2%)	4 (44.4%)
Right hemisphere	18 (51.4%)	14 (53.8%)	5 (55.6%)
Type of hemispherectomya				.23
Functional	25 (71.4%)	17 (65.4%)	8 (88.9%)
Anatomic	10 (28.6%)	9 (34.6%)	1 (11.1%)
Estimated blood loss (mL)b	306 ± 187	307.7 ± 176.0	300 ± 229.1	.93
Intraoperative complicationsa
Yes	1 (2.9%)	0 (0.0%)	1 (11.1%)	.26
Admitted before surgery for seizures or worsening of statusa	11 (31.4%)	8 (30.8%)	3 (33.3%)	1.0
Length of hospital stay (d)b	10 ± 5.4	10.1 ± 5.5	9.8 ± 5.4	.89
Postoperative MRI findingsa,c
Resection of insula	27 (96.4%)	21 (100%)	6 (85.7%)	.25
Adequate disconnection	22 (78.6%)	19 (90.5%)	3 (42.9%)	.021

^aCategorical variables analyzed using the Fisher's exact test.

^bContinuous variables analyzed using the Student's t-test (2-tailed).

^cOnly 28 cases had postoperative MRI scans available. Percentages and P-values were calculated using the sample sizes: all cases (n = 28), Engel I (n = 21), and Engel II-IV (n = 7).

TABLE 3B.

Univariate Analysis of Operative Variables Associated With Reoperation

Operative variables	All patients (n = 30)	Reoperation (n = 5)c	No reoperation (n = 25)	P value
Side of surgerya				1.0
Left hemisphere	15 (50.0%)	2 (40.0%)	13 (52.0%)
Right hemisphere	15 (50.0%)	3 (60.0%)	12 (48.0%)
Type of hemispherectomya				.56
Functional	25 (83.3%)	5 (100.0%)	20 (80.0%)
Anatomic	5 (16.7%)	0 (0.0%)	5 (20.0%)
Estimated blood loss (mL)a	275.0 ± 142.0	280.0 ± 148.3	274.0 ± 143.9	.93
Intraoperative complicationsa
Yes	1 (3.3%)	0 (0.0%)	1 (4.0%)	1.0
Admitted before surgery for seizures or worsening of statusa	11 (36.7%)	1 (20.0%)	10 (40.0%)	.63
Length of hospital stay (days)b	9.4 ± 3.8	7.4 ± 1.1	9.8 ± 4.0	.20
Postoperative MRI findingsa,d
Resection of insula	23 (95.8%)	3 (75.0%)	20 (100.0%)	.17
Adequate disconnection	18 (75.0%)	1 (25.0%)	17 (85.0%)	.034

^aCategorical variables analyzed using the Fisher's exact test.

^bContinuous variables analyzed using the Student's t-test (2-tailed).

^cFor reoperation patients, information from their first hemispherectomy was used.

^dOnly 24 patients had postoperative MRIs available. Percentages were calculated using the sample sizes: all patients (n = 24), reoperation (n = 4), and no reoperation (n = 20).

TABLE 4A.

Univariate Analysis of Postoperative Variables Associated With Seizure Freedom

Postoperative variables	All cases (n = 35)	Engel I (n = 26)	Engel II-IV (n = 9)	P value
Inpatient complications	15(42.9%)	9 (34.6%)	6 (66.7%)	.13
Fever	6(17.1%)	4 (15.4%)	3 (33.3%)	.34
Aseptic meningitis	7(20.0%)	5 (19.2%)	2 (22.2%)	1.0
Hydrocephalus	1(2.9%)	1 (3.8%)	0 (0.0%)	1.0
Cerebellar hemorrhage	1(2.9%)	1 (3.8%)	0 (0.0%)	1.0
Blood transfusion	1(2.9%)	1 (3.8%)	0 (0.0%)	1.0
Infection (bacteremia, urinary tract infection)	2(5.7%)	1 (3.8%)	1 (11.1%)	.45
Pneumothorax	1(2.9%)	1 (3.8%)	0 (0.0%)	1.0
Deep vein thrombosis	1(2.9%)	0 (0.0%)	1 (11.1%)	.26
Mean emergency department visits within 60 d	3(8.6%)	2 (7.7%)	1 (11.1%)	1.0
Mean emergency department visits within 1 y	7(20.0%)	6 (23.1%)	1 (11.1%)	.65
Readmission for surgical complications	7 (20.0%)	4 (15.4%)	3 (33.3%)	.34
Hydrocephalus	4(11.4%)	2 (7.7%)	2 (22.2%)	.27
Wound care	1(2.9%)	0 (0.0%)	1 (11.1%)	.26
Fevers	2(5.7%)	2 (7.7%)	0 (0.0%)	1.0
Postoperative shunt placement	4(11.4%)	2 (7.7%)	2 (22.2%)	.27

TABLE 4B.

Univariate Analysis of Postoperative Variables Associated With Reoperation

Postoperative variables	All patients (n = 30)	Reoperation (n = 5)a	No reoperation (n = 25)	P value
Inpatient complications	12(40.0%)	3 (60.0%)	9 (36.0%)	.36
Fever	6(20.0%)	1 (20.0%)	5 (20.0%)	1.0
Aseptic meningitis	7(23.3%)	2 (40.0%)	5 (20.0%)	.57
Hydrocephalus	0(0.0%)	0 (0.0%)	0 (0.0%)	1.0
Cerebellar hemorrhage	1(3.3%)	0 (0.0%)	1 (4.0%)	1.0
Blood transfusion	1(3.3%)	0 (0.0%)	1 (4.0%)	1.0
Infection (bacteremia, urinary tract infection)	1(3.3%)	0 (0.0%)	1 (4.0%)	1.0
Pneumothorax	1(3.3%)	0 (0.0%)	1 (4.0%)	1.0
Deep vein thrombosis	1(3.3%)	0 (0.0%)	1 (4.0%)	1.0
Mean emergency department visits within 60 d	2(6.7%)	1 (20.0%)	1 (4.0%)	.31
Mean emergency department visits within 1 y	4(13.3%)	1 (20.0%)	3 (12.0%)	.54
Readmission for surgical complications	5(16.7%)	3 (60.0%)	2 (8.0%)	.022
Hydrocephalus	2(6.7%)	2 (40.0%)	0 (0.0%)	.023
Wound care	1(3.3%)	1 (20.0%)	0 (0.0%)	.17
Fevers	2(6.7%)	0 (0.0%)	2 (8.0%)	1.0
Postoperative shunt placement	2(6.7%)	2 (40.0%)	0 (0.0%)	.023

^aAlthough questionnaires were collected for all 30 patients, the surveys were not always fully completed.

Kaplan-Meier Analysis of Seizure-Freedom

For all cases, we found a 1-year seizure-freedom rate of 81.5%, a 2-year seizure-freedom rate of 68.9%, a 5-year seizure-freedom rate of 63.6%, and a 10-year seizure-freedom rate of 55.6% (Figure 3A). Kaplan-Meier analysis was also used to compare functional and anatomic hemispherectomies (Figure 3B), where average follow-up times were 4.5 and 4.9 years, respectively (Table 1C). Overall, anatomic hemispherectomies resulted in greater seizure-freedom rates than functional hemispherectomies at 1-year (83.3% vs 80.8%), 2-year (83.3% vs 70.8%), 5-year (83.3% vs 57.7%), and 10-year (83.3% vs 46.2%) follow-up.

FIGURE 3.

A, Kaplan-Meier analysis of all hemispherectomies for RE. Reused under the CC BY license from S Sundar. et al. Seizure outcomes and reoperation in surgical Rasumssen encephalitis patients. 2021; American Epilepsy Society 2021 Annual Meeting Abstract Database: AESnet.org. B, Kaplan-Meier analysis for functional and anatomic hemispherectomies for RE. RE, Rasmussen encephalitis.

Functional Status

The details of preoperative functional status are presented in Table 5. Generally, patients were more likely to experience postoperative improvement in functional status than worsening. Regarding hemiparesis, 46.4% experienced improvement, whereas 17.9% experienced worsening. This was also observed for patient ambulatory status (34.6% improved and 3.8% worsened) and language ability (37.9% improved and 6.9% worsened). The exception was fine finger movements, 13.3% of patients had worsening of their impairments, whereas 10% had improvement.

TABLE 5.

Preoperative and Postoperative Functional Status for All Patients^a

Preoperative functional status		Postoperative functional changes
hemiparesis		hemiparesis
None	1/30 (3.3%)	Worsened	5/28 (17.9%)
Mild to moderate	17/30 (56.7%)	Unchanged	10/28 (35.7%)
Severe	12/30 (40%)	Improved	13/28 (46.4%)
Fine finger movements		Fine finger movements
Normal	3/29 (10.3%)	Worsened	4/30 (13.3%)
Fair dexterity	5/29 (17.2%)	Unchanged	13/30 (43.3%)
Minimal function, helper hand	9/29 (31.0%)	Improved	3/30 (10%)
None	12/29 (41.4%)
Ambulatory status		Ambulatory status
Walks unaided	14/30 (46.7%)	Worsened	1/26 (3.8%)
Uses aides/orthoses to walk	8/30 (26.7%)	Unchanged	16/26 (61.5%)
Few steps with assistance	2/30 (6.7%)	Improved	9/26 (34.6%)
Nonambulatory	6/30 (20.0%)
Language		Language
Normal	13/29 (44.8%)	Worsened	2/29 (6.9%)
Delayed/nonfluent	14/29 (48.3%)	Unchanged	16/29 (55.2%)
Nonverbal	2/29 (6.9%)	Improved	11/29 (37.9%)

^aAlthough questionnaires were collected for all 30 patients, the surveys were not always fully completed. Denominators reflect total responses per functional category.

DISCUSSION

Epilepsy surgery is believed to be the most effective method of seizure control in patients with RE. The existing literature primarily consists of retrospective studies with small sample sizes.^6-8,10-13 Patients commonly have limited clinical data, become lost to follow-up, or studies feature multiple surgery types (hemispherectomy, hemispherotomy, or single lobar resection). Except for a single study examining 5 total patients who all underwent functional hemispherectomy between 1978 and 1981, no existing literature assesses outcomes of patients with RE who all received hemispherectomy.¹³ The largest RE cohort studied is 45 patients, with 29 undergoing hemispherectomy.¹⁹ Meanwhile, pediatric epilepsy studies with patients who all received hemispherectomies include multiple etiologies, with RE comprising only a small minority.^15,17,20-23 This makes conclusions from a heterogeneous population challenging to generalize.

The method used to calculate seizure-freedom rates also varies across the literature. Some studies report whether patients are seizure-free at last follow-up and length of average follow-up, whereas others use a Kaplan-Meier analysis to censor patients as they are either lost to follow-up or have seizure recurrence. Kaplan-Meier analysis helps to account for variable follow-up times, including seizure recurrence, patient age and functional status, and whether patients receive medical care at multiple institutions.^24-26 Patients with RE can have delayed seizure recurrence years after surgical intervention, demonstrated by higher 1- or 2-year seizure freedom rates than 5- or 10-year rates, so comparing a seizure-free percentage without knowing the context of follow-up can be misleading.

Seizure-freedom rates from exclusive RE studies for 1 year, 5 years, and 10 years are 68% to 95%, 44% to 72%, and 22% to 52%, respectively.^1,3,6-11 By contrast, exclusive pediatric hemispherectomy studies show seizure freedom rates for 1 year, 5 years, and 10 years as 67% to 86%, 44% to 81%, and 42% to 63%, respectively.^{14-17,20,27-29} As a pure hemispherectomy RE series, our results fall within the reported ranges for both RE and hemispherectomy for 1-year and 5-year outcomes. Our longer term 10-year seizure freedom rate of 55.6% is notably higher than the 10-year rate recently published.⁶ This may be because we exclusively performed functional or anatomic hemispherectomies. Another difference may be in the length of follow-up; our study had an average follow-up of 4.6 years after surgery (Table 1C), which is lower than the average of 8.1 years reported by Bellamkonda et al.⁶ Our 10-year seizure freedom rate of 55.6% was comparable with that published in a 2019 meta-analysis, featuring 19 studies with similar follow-up (average of 4.3 years after surgery).¹⁰

We found significant correlation between incomplete disconnection on postoperative MRI scan and epilepsy requiring reoperation. Because disconnection can be difficult to confirm on immediate postoperative imaging, we assessed this based on the 6-month postoperative MRI scan. In practice, however, patients may present with recurrence of seizure earlier than 6 months. Although patients may not have a clearly visualized persistent connection between the diseased and the normal hemisphere, they do well after revision hemispherectomy, implying that residual “connected” tissue can be difficult to confirm. Scalp electroencephalography is also difficult to interpret in postoperative patients. Therefore, the decision to convert to anatomic hemispherectomy may be necessary to “guarantee” that all hemispheric connections have been adequately addressed and is often made with a low threshold as other treatment options are limited. All 5 of our patients who underwent reoperation remain seizure-free at the time of last follow-up.

We saw that patients with shorter duration of hemiparesis before surgery were less likely to obtain seizure freedom and more likely to undergo reoperation. One possible explanation is that patients with a more aggressive clinical course may experience rapid progression of permanent neurological deficits, prompting earlier surgical intervention. This theory correlates with our observations that reoperation patients also had a shorter duration of epilepsy and younger age at the time of initial hemispherectomy.

Patients who required reoperation were more likely to have bilateral MRI abnormalities on preoperative scans. Other studies have similarly shown that bihemispheric MRI abnormalities increase the risk of early seizure recurrence or predict worse seizure outcomes.^28,30-32 Larionov et al suggested that these contralateral findings may be due to secondary Wallerian degeneration of commissural fibers because of RE's inflammatory process.³³ These subtle contralateral MRI findings may be a symptom of a more severe subset of RE that progresses to affect the contralateral cortex.

The diagnosis of RE can be quite challenging, especially at smaller hospitals who see very few cases annually. Our cohort included 5 patients whose initial pathology diagnosed non-RE etiologies and were only confirmed to have RE after rereview of specimen by our institution's neuropathologist. RE remains a clinical diagnosis based on progression of symptoms, imaging/electroencephalography findings, and histology findings. Several patients were suspected at outside hospitals to have diagnoses other than RE, often undergoing temporal lobectomy, and ultimately were referred to us. Patients are reviewed in a multidisciplinary conference and treatment options are discussed, including immunotherapy, anticonvulsants, and hemispherectomy. The timing of surgery remains controversial as earlier surgery has been shown to lead to better cognitive outcome.^34-36 Barriers to early surgery include lack of definitive diagnosis and preserved neurological status. We recommend disconnective hemispherectomy as a first option as it reduces the risk of hydrocephalus and shunting compared with anatomic hemispherectomy.³⁷

Most patients experienced stable functional outcomes after hemispherectomy and were, overall, more likely to have improvement rather than worsening of functional status, which is consistent with pre-existing literature.^21,38 Given that our data were collected via parent-completed surveys, functional status may be based on family perception rather than objective measurements.

Limitations

Because of sample size, we were unable to perform multivariable analysis and draw stronger conclusions about predictive factors. Our objective was to report a single-center experience treating RE with resective surgery and analyze long-term outcomes. All our surgical RE patients had hemispherectomies performed by a single surgeon, which allowed us to control for surgeon variability. Our center also does not perform “less than” hemispherectomies that aim to minimize motor cortex disturbances, so our patient population may be more homogeneous and generalizable compared with existing literature.

CONCLUSION

RE is a rare pathology with diverse pathology and presentation, and studies about long-term outcomes are necessary to help counsel families and guide patient selection. Our results demonstrate that hemispherectomy may be a curative surgery for RE and has excellent long-term seizure outcomes. We found that patients with shorter duration of hemiparesis were less likely to achieve seizure freedom. In addition, the small subset of patients who needed reoperation tended to be younger at the time of surgery, with a shorter duration of epilepsy, and bilateral MRI abnormalities on preoperative imaging.

Obtaining a complete disconnection during hemispherectomy is challenging, even in the hands of experienced surgeons. Identifying residual connected tissue on imaging postoperatively can be very difficult, and medical options are limited for these challenging patients. For this reason, anatomic hemispherectomy is offered when the disconnective technique fails, and our very limited experience with 5 patients has been positive.

Funding

This study did not receive any funding or financial support. Dr Jehi is funded by NIH grant #R01 NS097719.

Disclosures

The authors have no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article. Dr Gupta serves on the editorial board of Pediatric Neurology and is a consultant for Mallinckrodt.

COMMENT

This manuscript is a report of a retrospective analysis of 30 patients with Rasmussen encephalitis treated at this center over 22 years with hemispherectomy. The authors are to be congratulated on this well written and clear paper, which describes an impressive number of patients treated in a uniform fashion, despite the rarity of this condition. The significance of this paper is that it clearly demonstrates the important role of hemispherectomy as the best treatment option for these children. Interestingly, they observed that a subset of patients may have a more aggressive clinical course and worse outcome after surgery. Most importantly for those of us who counsel these patients and families was the authors' finding that, although half of their cohort underwent dominant hemisphere disconnection, fewer than 7% of children experienced worsening of their language function. Finally, as in other situations in pediatric epilepsy, reoperation, in this case with anatomic hemispherectomy, can lead to seizure freedom when the initial surgery did not.

Howard L. Weiner

Houston, Texas, USA