The Brain Connectome after Gamma Knife Radiosurgery of the Ventro-Intermediate Nucleus for Tremor: Marseille-Lausanne Radiobiology Study Protocol

Stereotactic radiosurgery (SRS) was invented by Lars Leksell at the beginning of the 1950s, originally as a primary tool for treating functional disorders [1, 2]. Later on, in 1968, Leksell created the Gamma Knife (GK), an SRS tool using cobalt-67 sources of radiation [2]. Currently, clinical applications of GK include benign and malignant tumors, vascular malformations, and functional disorders (including psychiatric) [3, 4].

Mechanism of action varies according to the treated condition [5, 6]. In the particular case of functional disorders, GK is used either to target a specific anatomical point, such as the ventro-intermediate nucleus (Vim) of the thalamus [7, 8], the anterior limb of the internal capsule [9], or the trigeminal nerve [9-11] with high doses, or for targeting larger zones (e.g., epileptic focus) [12], with much lower dosage. Thus, the mechanisms of action can further differ.

Essential tremor (ET) is the most common movement disorder [13]. There are several pathophysiological theories with regard to ET. The most common one presumes tremor as generated by abnormal oscillations within the cerebello-thalamo-cortical axis, corresponding to the motor cortex, ipsilateral thalamic area (e.g., Vim), and the contralateral cerebellar dentate nucleus [13, 14]. Drug-resistant ET can benefit from standard deep brain stimulation [15] or stereotactic radiofrequency lesioning [16]. As a minimally invasive alternative, SRS has been successfully used during the past decades [17]. More recently, high-intensity focused ultrasound, producing an MR-controlled thermocoagulation, has emerged [18]. The majority of surgical procedures aim at the same target, the Vim. However, some authors have advocated also for the use of the zona incerta, with good clinical results [19].

The main limitation of Vim GK for tremor is the inability to predict the 1-year MR signature. Importantly, there is a lack of understanding of the radiobiological effect, with a progressive clinical improvement, after a median time of 3 months, going up to 1 year [20]. Thus, a better understanding of the radiobiological effect may improve our capacity to generate better clinical result.

The title of the protocol in French is “Evaluation des prédicteurs de la réponse clinique et radiologique à la radiochirurgie Gamma-Knife dans le traitement par thalamotomie du VIM” and in English is “Gamma Knife response predictability (GARP).” The first patient included in this protocol was treated in CHU Timone, Marseille, France, in September 2014. The protocol is ongoing. It had received formal approval from the CHU Timone Ethics Committee ID-RCB: 2017-A01249–44. Formal approval was also obtained from Paris, France: September 12, 2017, following the reference methodology CNIL-MR-03.

The primary aim of the present protocol is to assess the clinical response after Vim GK for tremor (in particular for ET), in relationship to neuroimaging changes, both at structural and functional level, but also with the genetic analysis, chronobiological details, concomitant therapies, concomitant morbidities, the brain’s state on pretherapeutic MRI, or tremor phenotype. Such an approach allows extracting rich information, with minimal patient’s compliance. The secondary endpoint is the influence of the paraclinical aspects (see below) on the radiological reaction in terms of timing and amplitude.

Our protocol was conceived as prospective, interventional, single group, and bicentric. All SRS treatments have been performed in Marseille University Hospital (CHU Timone; Professor Régis). Clinical and genetic data were analyzed in Marseille, France. Neuroimaging data were blindly analyzed in Lausanne, Switzerland. Enrollment of >100 cases is expected.

Inclusion criteria (Table 1, upper part): patients with tremor (ET, Parkinsonian origin, mixed, Holmes, polyneuropathy context, and multiple sclerosis patients), able to give written informed consent and attend the follow-up visit, with potential medical comorbidities, advanced age, or refusing Vim deep brain stimulation. Exclusion criteria (Table 1, lower part): contraindication to perform brain MRI (cardiac pacemaker), inability to provide written informed consent, patients refusing to enroll in the trial, and previous contralateral Vim surgery for tremor [21].

All targeting procedures were performed unilaterally by the last author using stereotactic imaging (MRI and CT) for all cases. The Guiot’s diagram (see Fig. 1, left) is placed 2.5 mm above the AC-PC line, 11 mm lateral to the wall of the third ventricle. A unique 4-mm isocenter is used. The dose prescribed is 130 Gy in all cases [22]. Beam channel blocking is used to shape the gradient towards the internal capsule (see Fig. 1, right). The “x” coordinate is fine-tuned accordingly to the exact anatomical location of the internal capsule on the nonstereotactic diffusion tensor imaging, acquired the day before GK.

Tremor severity is always assessed by the second author, using standard tremor scores: Bain’s questionnaire, including items such as the activities of daily living [23], Fahn-Tolosa-Marin tremor rating scale (tremor score of the treated hand) [24], and Tremor Research Group Essential Tremor Rating Assessment (TETRAS, head’s tremor). The quality of life questionnaire (QUEST) was fulfilled [25]. The exact timing of tremor improvement is noted. The Mattis dementia score is evaluated [26].

All neuroimaging is obtained on a 3-T MRI scanner Skyra TrioTrim (Siemens, Munich, Germany). Structural imaging includes the following: T1-weighted imaging (T1, TR/TE = 2,300/2.03, inversion time 900 ms, and isotropic voxel of 1 mm³ and 192 slices) [27, 28] and diffusion-weighted imaging (DWI, 64 gradient directions and b = 1000 s/mm) [29, 30]. Functional neuroimaging by resting-state functional MRI is further acquired: T2*-weighted fast echo planar imaging (BOLD contrast, before the Gadolinium injection, TR/TE = 3.3 s/30 ms/90°, voxel size 4 × 4 × 4 mm³, and 300 volumes acquired per subject), during 10 min of acquisition, in absence of any explicit task [31, 32]. Several methodologies are being used, including independent component analysis (Fig. 2a) and seed-to-voxel connectivity (Fig. 2b).

¹⁸F-FDG PET scans are performed with an integrated PET/CT camera. ¹⁸F-FDG is injected intravenously at the activity of 150 MBq, in an awake and resting state with eyes closed in a quiet environment, as previously reported [33]. Image acquisition starts 30 min after injection and ended 15 min thereafter. Images are reconstructed using the ordered subsets expectation maximization algorithm and corrected for attenuation using a CT transmission scan. In detail, ¹⁸F-FDG PET scans were performed before and after GK for each patient, under the same conditions as those for the healthy subjects acquired as the control group. A PET/CT camera (Discovery ST; GE Healthcare, Waukesha, WI, USA) was used with an axial resolution of 6.2 mm, allowing 47 3.27-mm thick contiguous transverse sections of the brain. ¹⁸F-FDG (150 MBq) was injected intravenously during both awake and resting states with each patient’s eyes closed in a quiet environment. Thirty minutes later, images were acquired over a 15-min period. Reconstruction parameters were 5 iterations and 32 subsets using the ordered subsets expectation maximization algorithm and with a CT transmission scan for attenuation correction. The FDG PET utility is to characterize PET changes in the metabolism of glucose and metabolic connectivity in patients with ET treated by GK through the observational study (comparing the pre- and posttherapeutic state, but also with healthy subjects).

Besides neuroimaging evaluation, several other investigations are performed. We always acquire blood sampling for genetic analysis, the chronobiological details (the moment of treatment during the day), the concomitant therapies (at the time of SRS and during the following months), concomitant morbidities (diabetes mellitus and vascular problems), and the brain’s state on pretherapeutic MRI (lacunar infarcts and brain atrophy).

Clinical and radiological evaluation was performed at 6 and 12 months and on yearly basis after.

The main neuroimaging analysis includes voxel-based morphometry (for structural imaging) and resting-state fMRI and metabolic PET methodologies, so as to be able to correlate clinical improvement with neuroimaging changes.

The trial is currently ongoing. More than 100 cases have been included, up-to-date, of various tremors (ET and Parkinsonian).

The first part focused on ET cases, to have a uniform disease. This part is also, currently, enrolling patients, to upgrade the number of cases.

The main challenge is the use of complex computational approaches, so as to be able to extract the relevant structural and functional connectomics, in relationship to clinical effect. Of major importance is the uniform assessment, on the same MRI machine. Other aspect, especially in tremor patients, is to further evaluate motion, as a potential confounding factor. Moreover, establishing neuroimaging metrics to evaluate their correlations with standard tremor scores is of interest.

A major aspect is that the relevant functional data are considered at baseline and 1-year follow-up, to account for the delayed clinical effect. The main expected result is how functional connectomics will change after Vim GK. Moreover, of particular interest is how this relates to clinical effect. The translational impact is of relevance. Such a trial could improve patient selection in the future, based on a specific functional and/or structural connectivity profile.

Dr. Tuleasca gratefully acknowledges the University of Lausanne (Unil), Faculty of Biology and Medicine (FBM), and Lausanne University Hospital (CHUV) for the research grant “Jeune Chercheur en Recherche Clinique.” Special acknowledgments are due to Axelle Cretol, who keeps the database up-to-date and thus plays a major role in this trial.

This study had received formal approval from the CHU Timone Ethics Committee ID-RCB: 2017-A01249-44. Formal approval was also obtained from Paris, France: September 12, 2017, following the reference methodology CNIL-MR-03.

Constantin Tuleasca was a scientific advisor for Elekta Instruments, AB, Sweden, without any relevance, nor funding for the present work; Jean Régis is a consultant and received research grants from Elekta Instruments, AB, Sweden, and from Medtronic.

This study was funded by Lausanne and Marseille University Hospital (CHUV and CHU Timone).

Constantin Tuleasca: conception and design, study supervision, and acquisition and interpretation of data; Tatiana Witjas: conception and design, study supervision, and acquisition and interpretation of data; Marc Levivier: conception and design and data acquisition; Nadine Girard: conception and design and data acquisition; Axelle Cretol: data acquisition and keeping the database up-to-date; Nicolas Levy: conception and design and interpretation of data; Jean-Philippe Thiran: interpretation of data; Eric Guedj: conception and design and acquisition and interpretation of data; Dimitri Van De Ville: conception and design, study supervision, and acquisition and interpretation of data; Jean Régis: conception and design, study supervision, and acquisition and interpretation of data.

Dr. Tuleasca and Dr. Witjas equally contributed as the first author.Professor Van De Ville and Professor Régis equally contributed as senior authors.