Article Text

Download PDFPDF

Outcomes following surgical interventions for hypothalamic hamartomas: protocol for a systematic review and individual patient data meta-analysis
  1. Keshav Goel1,
  2. Farbod Niazi2,
  3. Jia-Shu Chen3,
  4. Aristides Hadjinicolaou4,5,
  5. Mark Keezer5,
  6. Anne Gallagher6,7,
  7. Aria Fallah8,
  8. Alexander G Weil9,10
  1. 1University of California Los Angeles David Geffen School of Medicine, Los Angeles, California, USA
  2. 2Department of Medicine, University of Montreal, Montreal, Quebec, Canada
  3. 3Warren Alpert Medical School of Brown University, Providence, Rhode Island, USA
  4. 4Department of Pediatrics, Sainte-Justine University Hospital Center, Montreal, Quebec, Canada
  5. 5Department of Neuroscience, University of Montreal, Montreal, Quebec, Canada
  6. 6Laboratoire d’Imagerie Optique en Neurodéveloppement (LIONlab), Saint Justine University Hospital Research Centre, Montreal, Quebec, Canada
  7. 7Department of Psychology, University of Montreal, Montreal, Quebec, Canada
  8. 8University of California Los Angeles, Los Angeles, California, USA
  9. 9Brain and Development Research Axis, Saint Justine University Hospital Research Centre, Montreal, Quebec, Canada
  10. 10Department of Surgery, Division of Neurosurgery, Sainte-Justine University Hospital Center, Montreal, Quebec, Canada
  1. Correspondence to Dr Alexander G Weil;{at}


Introduction Hypothalamic hamartomas (HHs) are deep-seated congenital lesions that typically lead to pharmacoresistant epilepsy and a catastrophic encephalopathic syndrome characterised by severe neuropsychological impairment and decline in quality of life. A variety of surgical approaches and technologies are available for the treatment of HH-related pharmacoresistant epilepsy. There remains, however, a paucity of literature directly comparing their relative efficacy and safety. This protocol aims to facilitate a systematic review and meta-analysis that will characterise and compare the probability of seizure freedom and relevant postoperative complications across different surgical techniques performed for the treatment of HH-related pharmacoresistant epilepsy.

Methods and analysis This protocol was developed according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Individual Participant Data guidelines. Three major databases, PubMed, Embase and Scopus, will be systematically searched from database inception and without language restrictions for relevant articles using our predefined search strategy. Title–abstract and full text screening using inclusion and exclusion criteria created a priori will be performed by two independent reviewers to identify eligible articles. Conflicts will be resolved via discussion with a third team member. Following data extraction of both study-level and individual patient data (IPD), a study-level and IPD meta-analysis will be performed. Study-level analysis will focus on assessing the degree of heterogeneity in the data and quantifying overall seizure outcomes for each surgical technique. The IPD analysis will use multivariable regression to determine perioperative predictors of seizure freedom and complications that can guide patient and technique selection.

Ethics and dissemination This work will not require ethics approval as it will be solely based on previously published and available data. The results of this review will be shared via conference presentation and submission to peer-reviewed neurosurgical journals.

PROSPERO registration CRD42022378876.

  • Systematic Review
  • Epilepsy

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • We will use multiple imputation by chained equations to increase the statistical power of our individual patient data meta-analysis.

  • We will use propensity score weighting to account for differences in presurgical characteristics of patients.

  • This meta-analysis might be limited by a significant risk of bias since most evidence on surgical treatment of hypothalamic hamartomas comes from single-arm, retrospective studies.



Hypothalamic hamartomas (HHs) are rare, deep-seated congenital lesions arising from the ventral hypothalamus and the tuber cinereum.1 2 HHs are commonly attached to the mammillary bodies, a location which is densely connected to the mammillothalamic tract and fornix of the Papez circuit, which are thought to be responsible for seizure propagation.1 3 4 A subset of patients harbour a pedunculated HH in the anterior portion of the third ventricle, which can be associated with endocrinopathies such as central precocious puberty.5 6

The classic presentation of posterior-located HH is the onset of epilepsy in infancy or early childhood in the form of gelastic seizures with the majority (>80%) progressing to pharmacoresistant epilepsy.7–9 Some authors attribute long-standing epilepsy to secondary epileptogenesis of extra-HH structures and the onset of a more complex and extensive epileptogenic network with additional seizure types, such as focal motor, focal with impaired awareness, tonic, atonic and bilateral tonic–clonic seizures.7 10–14 Without prompt surgical therapy, half of the patients develop a catastrophic progressive encephalopathic syndrome characterised by cognitive impairment involving multiple domains,9 15 a severe behavioural disorder and a major decline in quality of life.16–20

Invasive stereo-electroencephalography (SEEG) recordings21 and non-invasive studies22 over the last 30 years have confirmed intrinsic epileptogenicity of HHs,22 23 which are thought to arise from pacemaker-like activity of small gamma-aminobutyric acidergic (GABAergic) interneuron-like cells within the HH.24 The assorted epileptic semiologies have been shown to originate within the HH itself.1 11 21 Gelastic seizures are thought to arise from synchronised ictal discharges among coordinated regions of the limbic system, including both the hypothalamus and functionally connected extrahypothalamic limbic (e.g., cingulate, parahippocampal gyrus) and orbitofrontal cortices.24 Other semiologies, such as focal impaired awareness seizures (FIAS), typically represent the involvement of mesial temporal lobe structures.24 Surgically targeting the HH has been the mainstay of treatment of HH-related pharmacoresistant epilepsy since the primarily intrinsic epileptogenicity of the hamartoma was documented in the 1990s. In 2003, the first surgical series directly targeting the HH through open microsurgical resection and disconnection showed that good rates of seizure freedom could be obtained, although at the expense of a trade-off of significant surgical and neurological morbidity.23 Specifically, although >90% seizure improvement occurred in 84.6% of patients, only 15% were seizure free and surgery came at the expense of morbidity in 54% of patients, including thalamo-capsular strokes among other neurological and endocrine deficits.23 Over the last 20 years, many groups have published single centre experiences with open microsurgical resection and/or disconnection of HHs associated with pharmacoresistant epilepsy through lateral (pterional) or vertical (transcallosal interforniceal) surgical corridors tailored to the morphology of the HH as defined by the Delalande classification.25–31 While open approaches could result in seizure freedom in about 50% of patients and translate into heterogeneous improvements of psychiatric outcome and cognitive performance at the group level,32–34 there remains significant morbidity, including memory impairments, hypothalamic (hyperphagia, obesity) dysfunction and neurological deficits, and a risk of ischaemic diencephalic/capsular strokes in up to a third of patients.23 27 35

Over the last 20 years, several surgical approaches have been added to the armamentarium with the goal of offering surgical targeting of the HH with less morbidity than open surgery. These techniques include the ‘less invasive’ endoscopic transventricular disconnection, non-invasive stereotactic radiosurgery (SRS) and minimally invasive stereotactic ablation strategies in the form of radiofrequency thermocoagulation (RFTC) and MR-guided interstitial thermal therapy (MRgLITT).36–40 The endoscopic approach, first reported and used by several groups in the early 2000s,28 41 42 has been shown to significantly reduce operative blood loss and shorten the length of stay (LOS) compared with open approaches;28 32 36 37 41 43–49 however, it is only feasible for HH with favourable configuration, such as Delalande Type II and III, and is associated with a non-negligible rate of memory and hypothalamic deficits.36 37 50 SRS was first applied to treat HH in the early 2000s as a non-invasive alternative suited for small- and medium-sized HHs, providing good seizure outcomes without long-term neurological sequelae.32 41 51–53 From a technical standpoint, SRS allows for conformational treatment adapted to the shape of the HH. However, due to the need to limit radiation exposure to <10 Gy at the optic apparatus, indications for SRS are limited in size (<15 mm) and disconnections of larger HHs are largely ineffective.54–56 This technique is hindered by an increase in seizure frequency increase in 16% of patients and a delayed therapeutic effect that can take up to 2–3 years.51 52 56 57 In the largest and only prospective trial in 48 patients with principally intrahypothalamic HHs (Delalande Type II and III), SRS demonstrated very good seizure freedom rates (40%), very low temporary deficits (6% poikilothermia) and no permanent neurological, neuropsychological, hypothalamic or endocrine morbidity.56 Because of its delayed nature, it is ideal for patients with high presurgical cognitive function, who will benefit from the lower risk of memory impairment, and an indolent clinical course (e.g., no progressive neuropsychological decline) who will not be disadvantaged by the delayed efficacy.

Both RFTC and MRgLITT have clear ‘minimally invasive’ benefits, including avoidance of ICU stay, and reduced LOS, blood loss and postoperative pain. These minimally invasive approaches have also been shown to reduce collateral damage to critical structures encountered in the approach (e.g., the corpus callosum and fornices) or at the target (e.g., hypothalamus, mammillary bodies/MTT). As a result, although long-term data are currently unavailable for MRgLITT, these techniques have a favourable risk–benefit profile, with a greater (70%) seizure freedom and reduced morbidity compared with open surgery.58–60 Over the last decade, MRgLITT and RFTC have emerged as the first-line treatment for HH in many centres across North America and Asia, respectively.61–63 In the largest study to date, in 58 cases of HH treated by MRgLITT, 81% were free from gelastic seizures, often requiring staged interventions.60 A variant of the standard SEEG-guided RFTC technique, in which one-stage high density RFTC is performed, may further increase the efficacy of this approach at the expense of requiring multiple probe trajectories in a given patient.47 64 While both stereotactic approaches are safe, there is a non-negligeable rate of transient and permanent morbidity, including memory impairment in 20% and hypothalamic injury in 27% in some studies.56 65–68 Very recently, MR-guided focused ultrasound (MRgFUS) has been discussed in the literature for central lesions in the brain.69 MRgFUS is completely non-invasive procedure that consists of directing multiple acoustic beams at a focal intracranial site, creating a coagulation lesion within the targeted tissue while monitoring the ablation process using real-team MR thermography.70 So far, only a total of five cases (four with epilepsy) have been reported in the literature with favourable outcomes and no major complications.69 71

Despite over 20 years of published reports of these techniques for the treatment of HH-related pharmacoresistant epilepsy, there is a paucity of well-conducted studies that directly compare efficacy and safety, and identify the optimal technique for the entire population or specific subgroups of HHs.32 51 72 The literature is almost exclusively limited to small case series and case reports describing novel techniques and is methodologically hampered by sample size and cohorts with a single treatment arm and limited follow-up. There are also several meta-analyses investigating developmental, endocrine, neuropsychological and seizure outcomes in surgically managed HH; however, to our knowledge, no study investigates the full landscape of surgical approaches to treat HH-related pharmacoresistant epilepsy.32 51 The current study fills an important knowledge gap and can help inform surgical management and decision making for both clinicians and patients with HH undergoing surgical interventions.


We aim to conduct a systematic review and meta-analysis of individual patient data (IPDMA) to investigate the safety and efficacy of the surgical techniques used for the treatment of pharmacoresistant epilepsy related to HH. We also aim to better understand the effect of these surgical approaches on developmental, endocrine and neuropsychological outcomes in patients with HH.

Methods and analysis

This protocol is registered on PROSPERO (CRD42022378876). This protocol was informed by, and the meta-analysis will be reported according to, the Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Individual Participant Data guidelines.73 Any future amendments to the protocol will be reflected in the PROSPERO record.

Information sources

Literature search will first be performed on three databases, namely, PubMed, Embase and Scopus from database inception without language restrictions. The references of included manuscripts will be reviewed to cross-reference any missing relevant articles. Eligible articles will be included for title–abstract and full-text screening as listed below. If IPD is not provided in an eligible article, the corresponding author will be contacted for data.

Search strategy

The following general search strategy will be used in each database: (hypothalamic hamartoma) AND (laser OR radiofrequency ablation OR radiosurgery OR electrocoagulation OR thermocoagulation OR Gamma Knife OR Cyber Knife OR Neurosurgical Procedures). Table 1 displays the specific search terms used for each database. The search strategy was developed with input and verification from our institution’s biomedical research librarian team.

Table 1

Search strategy by database

Data management

Search results will first be sorted in Zotero (V.6.0.21) where automatic filters will identify and remove any duplicate articles. Manual verification of duplicates will be performed, and remaining articles will be stored and organised in Rayyan (, a collaborative online software for meta-analysis title–abstract and full-text screening.

Study screening process

Study screening will be split into two phases, namely title–abstract and full-text screening. Title–abstract screening will involve two independent reviewers with content expertise who will independently screen articles for relevance to the topic of study. The selected articles will then proceed to the full-text screening phase, where each article is carefully reviewed for eligibility based on the inclusion criteria listed below by two independent reviewers. Conflicts will be resolved by a third, blinded study reviewer. Once included, articles will move on to data extraction and analysis.

Eligibility criteria

To be included in our systematic review and met-analysis, studies should (1) have a case series, case cohort, case control or randomised controlled trial design and (2) report data on patients with HH-related pharmacoresistant epilepsy treated with open microsurgical approaches, endoscopic transventricular approach, MRgLITT, RFTC or SRS. Studies will be excluded if they (1) report data on patients with HH without pharmacoresistant epilepsy, (2) include data only on patients managed non-surgically, (3) are case reports and 4) do not report seizure outcomes at last follow-up.

Data collection and extraction

When available, IPD and study-level data will be extracted. For studies missing IPD, the corresponding authors will be contacted and IPD requested. Following data collection, data will be independently extracted by two team members (KG, FN) and stored in Microsoft Excel 2020. Conflicts will be resolved after discussion with two senior expert team members (AH, AGW). Variables of interest will be standard demographic variables (sex, age at surgery/seizure onset), HH size and volume, Delalande classification, preoperative seizure types (gelastic and non-gelastic) and frequency, preoperative neuropsychological, psychiatric and endocrine status, history and type of prior surgery, type of surgical treatment employed along with intraoperative variables such as extent of resection and disconnection, postoperative seizure freedom for various seizure types, length of follow-up, surgical complications, postoperative neuropsychological, psychiatric and endocrine status and stability, mortality, and requirement of revision surgery with type of revision surgery and seizure outcome following all revision surgeries.


The main objective of this study is to quantify and compare the clinical efficacy of the five main surgical techniques used to treat HH. If available, two additional approaches will also be compared, namely, interstitial radiosurgery and stereotactic interstitial radiosurgery.74 75 Clinical efficacy is defined in this review by the probability of seizure freedom (Engel I) at 1-year follow-up. To account for staged treatment plans used for multiple included modalities (e.g., SRS, MRgLITT, RFTC), analysis of seizure freedom after all planned procedures will also be performed. Effects of clinical efficacy will be reported as overall seizure freedom (Engel class I76 77). Outcome of individual seizure types (e.g., gelastic seizures, FIAS) will be reported and discussed if enough data are available. Since it is estimated that a significant number of patients from large centres will be lost if only IPD analysis is performed, study level analysis is preferred for the analysis of overall seizure freedom. The primary outcome will be the pooled proportion of seizure freedom, which will be visualised using forest plots. The secondary objective is to assess safety and compare the morbidity between the approaches. Complications will be defined as new-onset or worsening of comorbidity following surgery by comparing presurgical and postsurgical outcomes if available. Severity of complications will be graded as minor or major depending on the impact on life and the duration of symptoms. A complication will be considered as major if it persists for more than 3 months and affects the patient’s activities of daily living.78 Additional outcomes will include comparing changes in preoperative comorbidities such as neuropsychological, psychiatric and endocrine outcomes and stability between different surgical approaches if enough data are available.

Risk of bias and quality grade

Risk of bias will be assessed by two experienced independent reviewers and any uncertainty will be resolved by a senior team member. The assessment will be done per the ROBINS I tool or the Joanna Briggs Institute case series Critical Appraisal Tool.79 80 We will evaluate our confidence in our conclusions using the BMJ GRADE tool.81

Study-level and individual patient data analysis

All data analysis will be performed using RStudio (RStudio, V.1.2.1335). Cohen’s Kappa scores will be calculated to determine the strength of agreement for full-text review. Cohen’s Kappa will be interpreted according to McHugh et al, where values of 0.01–0.2 indicate slight, 0.21–0.40 as fair, 0.41–0.60 as moderate, 0.61–0.80 as substantial and 0.81–1.00 as near-perfect inter-rater agreement.82 All patients will be stratified into one of the five surgical techniques: open resection, endoscopic resection/disconnection, SRS, RFTC or MRgLITT. A two-stage IPDMA will be performed. Using study-level data, pooled proportions of seizure freedom and major complications at last follow-up, as well as the effect size for each study, will be calculated using the inverse-variance model. A Freeman-Tukey double arcsine transformation or an arcsine transformation will be used to stabilise variances.83 These results will then be visualised using forest plots. Publication bias will be assessed through visual inspection of funnel plots. Heterogeneity will be evaluated using Cochrane’s Q test and the I2 statistic. Subgroup analysis will be performed based on treatment type. Further subgroup analysis will be performed within individual cohorts of open microsurgery and RFTC based on the technique used (lateral vs vertical approach in open microsurgery and SEEG-guided vs stereotactic in the RFTC subgroup). If substantial heterogeneity (I2>50%) exists within subgroups, meta regression will be performed to further study sources of heterogeneity.

A one-stage meta-analysis approach will be used for IPD analysis. Multiple imputations by chained equations will be performed on variables with less than 40% missing data as previously performed by our group.84 Depending on their clinical relevance, variables with more than 40% missing data will either be excluded from the final analysis or will be included in a complete case analysis. Using inverse propensity treatment weighting, propensity weights will be calculated based on the Delalande classification and age at surgery to account for presurgical differences among patients.85 86 Surgical subgroups will be compared with open resection cohort primarily. Predictors of favourable seizure outcome (Engel I) and at >1 year, >2 years and >5 years follow-up will be identified via multivariate mixed-effect logistic regression with the study set as the random covariate. A multivariate mixed-effect logistic regression model, with study set as the random covariate, will also be constructed to identify predictors of major postoperative complications. A two-tailed p value of 0.05 will be the threshold for statistical significance. Finally, long-term change in preoperative comorbidities such as neuropsychological, psychiatric or endocrine impairments will be reported and compared between surgical techniques using descriptive statistics and two-sided comparative tests if enough data are available.

Patient and public involvement


Ethics and dissemination

This work will not require ethics approval as it will be solely based on previously published and available data. The results of this review will be shared via conference presentation and submission to peer-reviewed neurosurgical journals.

Ethics statements

Patient consent for publication



  • Twitter @keshav_goel98

  • Contributors KG, FN and J-SC: study design/planning, drafting of manuscript, critical revision of manuscript. AH, MK and AG: study design/planning, critical revision of manuscript. AF and AGW: study design/planning, critical revision of manuscript, administrative and technical support, study supervision.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.