Anterior thalamic nucleus stimulation for refractory epilepsy

Patient or population: adults with IQ > 70 with refractory focal epilepsy

Settings: epilepsy centres in the USA

Intervention: anterior thalamic nucleus stimulation

Comparison: sham stimulation

Seizure freedom

(3‐month blinded evaluation period)

OR 0.33 (0.01 to 8.36)

109
(1)

⊕⊕⊕⊝
moderate²

Responder rate

(3‐month blinded evaluation period)

OR 1.20 (0.52 to 2.80)

108
(1)

⊕⊕⊕⊝
moderate²

Seizure frequency reduction (%)

(3‐month blinded evaluation period)

108 (1)

⊕⊕⊕⊕
high³

A trend for increasing efficacy over time was observed during the blinded evaluation period and could result into an underestimation of the treatment effect (treatment effect of month 3: ‐29%).

Adverse events

109 (1)

⊕⊕⊕⊝
moderate²

Stimulation‐related adverse events during the blinded evaluation period include (stimulation versus control): depression (14.8 versus 1.8%, P = 0.02), subjective memory impairment (13.8 versus 1.8%, P = 0.03) and epilepsy‐related injuries (7.4 versus 25.5%, P = 0.01). Standard stimulation parameters could be inappropriate and increase seizure frequency in a small minority of patients.⁴

Asymptomatic intracranial haemorrhages occurred in 3.7% of participants after the initial implant procedure. In 8.2% of participants leads had to be replaced after initial implantation outside the target. Postoperative implant site infections occurred in 4.5% of participants, increasing to 12.7% after 5 years of follow‐up urging (temporary) hardware removal in 8.2% of participants. Implant site pain was not uncommon (year 1: 10.9%, year 5: 20.9%). SUDEP rate during long‐term (including open‐label) follow‐up was 2.9 per 1000 p‐y which is comparable to rates reported in refractory epilepsy populations (2.2‐10 per 1000 p‐y) (Tellez‐Zenteno 2005; Tomson 2008).

Neuropsychological outcome

(3 months)

96‐100 (1)

⊕⊕⊕⊝
moderate⁵

Changes in neuropsychological test scores for cognition and mood were very similar in the treatment and control group and not significantly different. Individual patient data show worsening (> 1 SD) of Profile of Mood States Depression subscale (POMS‐D) in 3/8 stimulated participants with self‐reported depression and 0/7 patients with subjective memory impairment showed worsening (> 1 SD) of verbal or visual memory scores.

Quality of life

(QOLIE‐31)

(3 months)

105 (1)

⊕⊕⊕⊕
high

Positive changes in QOLIE‐31 (quality of life in epilepsy 31) scores indicate improvement. Changes of 5‐11.7 have been defined in literature as being clinically meaningful (Borghs 2012; Cramer 2004; Wiebe 2002).

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; OR: odds ratio; SUDEP: sudden unexpected death in epilepsy patients; p‐y: patient‐years; SD: standard deviation

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Centromedian thalamic nucleus stimulation for refractory epilepsy

Patient or population: patients with refractory (multi)focal or generalized epilepsy

Settings: epilepsy centres in the USA and in Mexico

Intervention: centromedian thalamic nucleus stimulation

Comparison: sham stimulation

Seizure freedom

(3‐month blinded evaluation period)

OR 1.00 (0.11 to 9.39)

6 (1)²

⊕⊝⊝⊝
very low^3,4

Responder rate

(3‐month blinded evaluation period)

OR 1.00

(0.27 to 3.69)

6 (1)²

⊕⊝⊝⊝
very low^3,4,5

Seizure frequency reduction

(3‐month blinded evaluation period)

6 (1)²

⊕⊝⊝⊝
very low^3,4,5

Also another trial (Velasco 2000a) (n = 13) could not demonstrate significant differences between stimulation ON and OFF periods. However, its cross‐over design without any washout period could mask a possible treatment effect.

Adverse events

19 (2)²

21 (2)²

⊕⊕⊝⊝
low^4,6

Stimulation‐related adverse events did not occur.

Postoperative CT revealed an asymptomatic and minimal haemorrhage in one patient, 1 patient required repair of the connection to the pulse generator and skin erosion urged device explantation in 3 other patients (including 2 young children).

Neuropsychological outcome

(3 months)

6 (1)²

⊕⊝⊝⊝
very low^3,4

There were no significant differences in any of the neuropsychological tests between baseline, stimulation ON and OFF periods.

Quality of life

0 (0)

See comment

Impact of centromedian thalamic nucleus stimulation on quality of life has not been studied yet.

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; OR: Odds ratio

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Cerebellar stimulation for refractory epilepsy

Patient or population: patients with refractory (multi)focal or generalized epilepsy

Settings: epilepsy centres in the USA and in Mexico

Intervention: stimulation of the superomedial surface of the cerebellum

Comparison: sham stimulation

Seizure freedom

(1‐ to 3‐month blinded evaluation period)

OR 0.96

(0.22 to 4.12)

22 (3)²

⊕⊕⊕⊝
moderate³

Responder rate

(1‐ to 3‐month blinded evaluation period)

OR 2.43

(0.46 to 12.84)

19 (3)²

⊕⊕⊝⊝
low^3,4

Seizure frequency reduction

(1‐ to 3‐month blinded evaluation period)

19 (3)²

⊕⊕⊝⊝
low^3,4

Adverse events

22 (3)²

⊕⊕⊝⊝
low^3,5

Stimulation‐related adverse events were not reported in any of the trials.

In contrast, about half of the patients in every trial required repeated surgery due to electrode migration (n = 6), leakage of cerebrospinal fluid (n = 3), wound infection (n = 1), skin erosion (n = 2), lead problems (n = 1), subcutaneous seroma drainage (n = 1) and defective hardware (n = 1). Wound infections were solved with antibiotics only in 2 additional patients. In particular, electrode migration remains of specific concern, even in the most recent trial (Velasco 2005) (occurring in 3/5 patients).

Neuropsychological outcome

(1 to 2 months)

16 (2)²

⊕⊝⊝⊝
very low^3,4,6

'Psychometry' did not reveal any major change in any patient in any phase of the Wright 1984 trial. Comparing ON to OFF stimulation full scale intelligence and memory scores in Van Buren 1978 showed very similar results in two participants, a moderate increase in one patient and a moderate decrease in another.

Quality of life

(2 months)

12 (1)⁷

⊕⊝⊝⊝
very low^3,4,8

Eleven out of 12 patients in Wright 1984 felt better for cerebellar stimulation, but only 5 chose one phase as being different from the others, being either the continuous (n = 2), contingent (n = 1) or no‐stimulation (n = 2) phase.

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; OR: odds ratio

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Hippocampal stimulation for refractory epilepsy

Patient or population: patients with refractory medial temporal lobe epilepsy

Settings: epilepsy centres in Canada and in Mexico

Intervention: hippocampal deep brain stimulation

Comparison: sham stimulation

Seizure freedom

(1‐ to 3‐month blinded evaluation periods)

OR 1.03
(0.21 to 5.15)

15 (3)²

⊕⊕⊕⊝
moderate³

Also in Wiebe 2013⁴ no single patient achieved seizure freedom after six months of hippocampal active or sham stimulation.

Responder rate

(1‐ to 3‐month blinded evaluation periods)

OR 1.20

(0.36 to 4.01)

15 (3)²

⊕⊕⊝⊝
low^3,5

In Wiebe 2013⁴there was one responder in the stimulation group (n = 2) compared to none in the sham group (n = 4) after six months of follow‐up.

Seizure frequency

(1‐ to 3‐month blinded evaluation periods)

15 (3)²

⊕⊕⊕⊝
moderate³

One trial (Tellez‐Zenteno 2006) has a cross‐over design without any washout period which could result into an underestimation of the true treatment effect.

In Wiebe 2013⁴ the sham stimulation group reported a median seizure frequency increase of 60% compared to a 45% decrease in the stimulation group after 6 months of follow‐up.

Adverse events

15 (3)²

⊕⊕⊝⊝
low⁶

There were neither stimulation‐related adverse events, nor early surgical complications. Skin erosion and local infection required explantation after >2 years in 3/9 patients in Velasco 2007.

Wiebe 2013⁴ also did not report any adverse event after 6 months of follow‐up.

Neuropsychological outcome

(1‐ to 3‐month periods)

6 (2)²

⊕⊝⊝⊝
very low^5,6

Neuropsychological test results were the same or very similar during stimulation ON and OFF periods in Tellez‐Zenteno 2006 (n = 4) and in one patient in McLachlan 2010. The other patient in McLachlan 2010 showed worse verbal and visuospatial memory scores when stimulated, notwithstanding that he reported subjective memory improvement during the same period.

At seven months in Wiebe 2013⁴, scores of cognitive scales assessing recall (Rey Auditory Verbal Learning Test, Rey Complex Figure Test) were generally lower in the active stimulation compared to the sham group (p>0.05).

Quality of life

(QOLIE‐89)

(1‐ to 3‐month periods)

3 (1)⁷

⊕⊝⊝⊝
very low^5,6

Positive changes in QOLIE‐89 (quality of life in epilepsy 89) scores indicate improvement. Changes of 5‐11.7 have been defined in literature as being clinically meaningful (Borghs 2012; Cramer 2004; Wiebe 2002).

The overall QOLIE‐89 score at seven months in Wiebe 2013⁴ worsened by 13 points with sham stimulation compared to an improvement of 3 points with active stimulation (p>0.05), and there was a trend for increased QOLIE‐89 subjective memory and attention/concentration scores.

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; OR: odds ratio

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Nucleus accumbens stimulation for refractory epilepsy

Patient or population: adults with IQ >70 with refractory focal epilepsy

Settings: epilepsy centre in Germany

Intervention: nucleus accumbens stimulation

Comparison: sham stimulation

Seizure freedom

(3‐month blinded evaluation period)

OR 1.00

(0.07 to 13.64)

4 (1)²

⊕⊕⊝⊝
low³

Responder rate

(3‐month blinded evaluation period)

OR 10.0

(0.53 to 189.15)

4 (1)²

⊕⊕⊝⊝
low³

Seizure frequency reduction

(3‐month blinded evaluation period)

4 (1)²

⊕⊕⊝⊝
low³

When focusing on 'disabling seizures' only and excluding simple partial seizures (occurring in one patient), the mean change in seizure frequency in the control group was +8.2% with a ‐22.9% lower seizure frequency in the intervention group (‐100 lower to +94.0 higher)

Adverse events

4 (1)²

⊕⊕⊝⊝
low³

Except for one patient feeling sad for two weeks during the active stimulation period after a close relative had died, there were no adverse events that were exclusively linked to the active stimulation period (although various adverse events were reported in the sham and the active stimulation group, see text).

One patient developed a local subcutaneous infection with colonization of the pulse generator and the leads 2 weeks post‐surgery urging antibiotic therapy and temporary hardware removal.

Neuropsychological outcome

(3 months)

4 (1)²

⊕⊕⊝⊝
low³

Neurocognitive test scores were similar and not statistically significantly different during sham and active stimulation in this small trial. There were no categorical changes in Beck‐Depression‐Inventory scores during the BEP. However, the Mini International Neuropsychiatric Interview revealed a new‐onset major depression under nucleus accumbens stimulation in one patient, besides an ongoing low suicidal risk following one suicide attempt 10 years before the trial in another patient.

Quality of Life

(QOLIE‐31‐P)

(3 months)

4 (1)²

⊕⊕⊝⊝
low³

The QOLIE‐31‐P is a (slightly) modified version of the QOLIE‐31 questionnaire for which changes of 5 to 11.7 have been defined in the literature (Cramer 2004; Wiebe 2002; Borghs 2012) as being clinically meaningful; positive scores indicate improvement.

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; OR: Odds Ratio; BEP: blinded evaluation period

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Closed‐loop stimulation of the ictal onset zone for refractory epilepsy

Patient or population: adults with refractory focal epilepsy (1 or 2 epileptogenic regions)

Settings: epilepsy centres in the USA

Intervention: responsive stimulation of the ictal onset zone(s)

Comparison: sham stimulation

Seizure freedom

(3‐month blinded evaluation period)

OR 4.95

(0.23 to 104.44)

191 (1)

⊕⊕⊕⊝
moderate²

Responder rate

(3‐month blinded evaluation period)

OR 1.12

(0.59 to 2.11)

191 (1)

⊕⊕⊕⊝
moderate²

Seizure frequency reduction

(3‐month blinded evaluation period)

191 (1)

⊕⊕⊕⊕
high³

A trend for increasing efficacy over time was observed during the blinded evaluation period and could result into an underestimation of the treatment effect (treatment effect of month 3: ‐32%).

Adverse events

191 (1)

256 (2)

⊕⊕⊕⊝
moderate²

Adverse events during the blinded evaluation period were rare and there were no significant differences between the treatment and control group.

Asymptomatic intracranial haemorrhages considered as serious adverse event were found postoperatively in 1.6% of participants. Postoperative implant or incision site infection occurred in 2.0% of participants, increasing to 9.4% of participants after 5 years of follow‐up (additional cases mainly upon battery replacement; urge for (temporary) explantation in the majority of cases). Cranial implantation of the neurostimulator was the probable cause of most adverse events, which include: implant site pain (16% during the first year of the trial), headache (11%), procedural headache (9%) and dysaesthesia (6%). Although the SUDEP rate (4 SUDEPs over 340 patient‐years = 11.8 per 1000 patient‐years) reported in the initial manuscript was slightly higher than those usually reported in refractory epilepsy patients (2.2 to 10 per 1000 p‐y) (Tellez‐Zenteno 2005; Tomson 2008), long‐term open‐label follow‐up has now reported reassuring figures (SUDEP rates of 3.5 per 1000 implant p‐y or 2.6 per 1000 stimulation p‐y).

Neuropsychological outcome

(3 months)

160‐177
(1)

⊕⊕⊕⊕
high

Changes in neuropsychological testing results were very similar in both groups and 95% confidence intervals did not include clinically meaningful differences.

Quality of life

(QOLIE‐89)

(3 months)

180
(1)

⊕⊕⊕⊕
high

Positive changes in QOLIE‐89 (quality of life in epilepsy 89) scores indicate improvement. Changes of 5‐11.7 have been defined in literature as being clinically meaningful (Borghs 2012; Cramer 2004; Wiebe 2002).

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; OR: odds ratio; SUDEP: sudden unexpected death in epilepsy patients; p‐y: patient‐years

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.