Integrating navigated transcranial magnetic stimulation motor mapping in hypofractionated and single-dose gamma knife radiosurgery: A two-patient case series and a review of literature

Inclusion criteria

Subject to multidisciplinary discussion, as well as institutional and ethical approval, two adult patients with MRI-verified metastatic lesions in or adjacent to cortical motor areas were recruited for the purpose of this study. In both cases, microsurgery was deemed less suitable than radiosurgery due to topoanatomical limitations and the extension of extracranial metastatic disease; other forms of radiotherapy or systemic treatment were assessed as of no benefit to these patients. Patients were required to have a Karnofsky Performance scale (KPS) of at least 70 (recursive partitioning analysis [RPA] 1–2). Previous microsurgery, presence of perilesional edema (resistant to steroids or not) and extracranial tumor activity were not criteria for exclusion. However, future or ongoing oncological treatment and survival expectancy of at least 6 months were mandatory. In this particular setting, one patient was accepted for dose-volume adaptive hypofractionated GKRS while the other was planned for single-dose GKRS.

Pre- and post-GKRS nTMS protocol

Protocols for nTMS motor mapping followed updated institutional guidelines and previously published medical data.[26] In each case, muscles were chosen in accordance with the lesion’s specific location (Table 1 for details of chosen muscles). The eXimia NBS system 4.3 (Nexstim, Helsinki, Finland) equipped with a stereotactic navigation system and a 6-channel electromyography (EMG) recorder was used. Single-pulse TMS was delivered with a focal biphasic figure- of-eight coil (70 mm outer diameter; Nexstim Ltd., Helsinki, Finland) affecting an area of about 1.7 cm² or smaller.[45] Motor evoked potential (MEP) data from the eXimia system (sampling rate 3000 Hz) were obtained, using the surface EMG electrodes (Neuroline 715, Ambu, Ballerup, Denmark), attached to muscles contralateral to the stimulated hemisphere. Subject-specific MRI was used to reconstruct the head model. Structural MRI of the brain, with a suitable navigation sequence (axial 3D BRAVO 1 mm reformatted in three planes, FoV 256 mm, NEX 1, TR 8.16, TE 3.18, TI 450, Flip angle 12, 1.5 T, GE Discovery 450 CE) was acquired earlier. During the TMS session, the patient’s head was first coregistered with the corresponding MRI; then, the standard tracking unit was fixed to the patient’s forehead allowing real-time navigation.[43] Initially, the resting motor threshold (rMT) was estimated and used as a marker for subject- specific cortical excitability. All relevant rMT-estimations were measured for both hemispheres. Motor mapping was done at 110% of rMT.

Table 1:: nTMS data for both cases.

Complementary studies: pre- and post-GKRS electroencephalography (EEG)

Both patients were examined pre- and posttreatment with complementary scalp-EEG studies. Electrodes (21-scalp electrodes) were positioned using the 10–20 system; Fp1, Fp2, F7, F8, T3, T4, T5, T6, O1, O2, F3, F4, C3, C4, P3, P4, Fz, Cz, Pz, A1, and A2. Electrode impedances were kept below 5 kΩ. The recording consisted of 30 min of spontaneous activity with the patient awake in a supine position. The patients were provoked using 3 min of hyperventilation and photostimulation. The pre- and posttreatment EEGs showed no epileptiform activity; however, in Case 2, a focal slowing was demonstrated before and after the radiosurgery in the right central regions.

Integrating nTMS motor assessments to GKRS planning

Optimal cranium fixation was achieved mounting the stereotactic Leksell^® frame. As a result, no margins were required outside the gross tumor volume limits. To attain ideal dynamic target definition, a stereotactic MRI was performed before each GKRS. MRI sequences included pre- and postgadolinium enhanced T1 fast spin echo, postgadolinium 3D T1-weighted volume FSPGR reconstructed in three planes as well as axial T2-weighted propeller, 1.5 T GE Discovery 450w. All positive nTMS points were fused together with the anatomical MRI obtaining identifiable motor points deemed as organs at risk (OAR) ultimately saved as a digital imaging and communication file. The data were then successfully imported to the LGP system and taken into consideration for GKRS planning. The corresponding GKRS treatment plan was set on the preoperative MRI evolution, and the presence of treatment feasibility variables described elsewhere.[17,22] For Case 1, we designed a regimen of three GKRS-interventions during the course of 7 days to adapt radiation doses to expected morphological changes during the course of treatment. Each fraction was to be delivered every 72 h. Further setup characteristics have been explained in prior reports.[17,22] For Case 2, single-dose GKRS-treatment was designed to deliver a peripheral dose of at least 18 Gy at the 50% isodose line due to its relatively smaller volume. To monitor possible radiation- induced plastic changes while adapting to extracranial disease evolution, a follow-up (second) nTMS was to be performed at least 6 months, but no more than 12 months, post treatment on both patients (see discussion).

Case description

Case 1

A 53-year-old female patient with metastatic breast cancer, previously treated with both chemo- and immunotherapy, developed focal epileptic seizures involving her right arm. The investigative computed tomography (CT)-scan and MRI of the brain showed four intra-axial metastases, the largest (6.7 cc) located in the left precentral gyrus (M4). Extracranial CT screening showed stable appearances of disease, including her liver metastases. The patient was planned for adaptive hypofractionated GKRS of M4 and for single fraction GKRS treatment for the remaining lesions (M1-M3); due to their “safe” location, M1-M3 were excluded from this study. A preoperative nTMS for perilesional motor- mapping was performed the day before GKRS 1 (November 2016). In this case, a total of three muscles from the upper limb (adjacent to M4) were mapped and identified as OAR [Figure 1a and b]. rMT was measured for both hemispheres, which showed a clear side differential (left > right); MEP amplitudes and latencies were within normal range [Table 1 for detail neurophysiological values]. At GKRS 1, the patient was still suffering from focal seizures despite antiepileptic treatment (equivalent to Engel-score scale 2–3) but with almost immeasurable right-hand weakness; her KPS was assessed as 100.

Close

Figure 1:: Structural image showing lesion and gamma knife radiosurgery (GKRS) planning in Case 1. (a) Yellow line = peripheral prescription isodose line. Outer and inner green lines = dose distribution inside and outside target. White rectangles = navigated transcranial magnetic stimulation (nTMS) motor points overlayed at 20–25 mm depth. (b) Magnetic resonance-images superimposed with pre-GKRS nTMS-determined organs at risk representing the arm and hand motor function areas for Case 1. Red = gross tumor volume (treatment target); Magenta = upper limb; Blue = lower limb.

Export to PPT

Case 2

A previously healthy 73-year-old man developed gradual left- hand paresis (September 2016). A brain CT and subsequent MRI showed a 26 mm metastatic lesion in the right postcentral gyrus and postcentral sulcus, 5 mm from the hand motor cortex (M1), as well as an 18 mm metastasis (M2) located medially in the right superior frontal gyrus. Due to its location, M2 was deemed safe for GKRS and excluded from this study; both lesions had extensive perilesional edema. An investigative CT-scan of the thorax and abdomen revealed a 6 cm lung tumor in the right lower lobe as well as mediastinal lymph node infiltration. The histology of the primary was then confirmed to be a lung adenocarcinoma (October 2016). The patient was accepted for upfront GKRS management of M1 and M2 before starting systemic treatment. As in the first case, the patient underwent a preoperative nTMS for perilesional motor-mapping before starting treatment. Likely secondary to ongoing steroid treatment, the underlying nTMS MRI revealed a slight volume reduction of M1 (24 mm, vol =5 cc) while unchanged volumetry for M2 (1.3 cc). Due to the location of M1 in relation to the motor homunculus, three muscles from the upper limb and one lower limb muscle were mapped and identified as OAR [Figure 2a and b, Tables 2 and 3]. Based on nTMS–based estimates, M1 was treated with a peripheral dose of 20 Gy at the 50% isodose line (one fraction). At GKRS, the patient presented with slight fatigue and a subjective (hardly measurable) left-hand weakness. However, he remained seizure-free (Engel score scale 1) with a KPS of 90.

Close

Figure 2:: Structural image showing lesion and GKRS planning in Case 2. (a) Yellow line = peripheral prescription isodose line. Outer green line = dose distribution outside target (10 Gy-isodose line). White rectangles = navigated transcranial magnetic stimulation motor points overlayed at 20–25 mm depth. (b) Magnetic resonance-images superimposed with transcranial magnetic stimulation for Case 2. Gross tumor volumes in red (most anterior target included in the study) and organs at risk in magenta (upper limb) and blue (lower limb).

Export to PPT

Table 2:: Case 2 – Dose distribution outside target to upper and lower limb muscles (=OAR).

Table 3:: Case 2 – Total 10 Gy-volume and selectivity estimates (illustrating dose dissipation outside target) and tumor bed coverage with and without integrating nTMS to single-dose GKRS treatment planning (applying a peripheral dose of 20 Gy).

Case 1

The treatment was well tolerated. M4 was 7.4 cc at GKRS 1; 16% of volume reduction was achieved between GKRS 1 and GKRS 3 [Table 4]. The reduction of dose dissipation to the OAR based on n TMS data was around 50% at each fraction without compromising intratumoral dose deployment [Tables 5 and 6]. Motor symptoms and epileptic seizures subsided within weeks after treatment completion. Follow-up MRI at 1 month and subsequent imaging showed considerable tumor volume reduction of M4 compared to GKRS 1 (>95% at 12 months follow-up MRI) without evidence of ARE [Table 7, Figure 3]. The remaining lesions had also responded to treatment. The postsurgical nTMS at 12 months was performed using similar stimulation guidelines, registering from the same muscles. Like the preoperative nTMS, rMT showed a clear side difference (left > right). However, MEP amplitudes and latencies were within normal range [Table 1]. The corresponding representation of the hand motor area at this stage showed no identifiable distortions [Figure 3]. GKRS had neither affected nor distorted the functional configuration mapped on the nTMS examination at pre and post-GKRS, hence successfully managing to preserve/ salvage local neurological function. At 12 months follow- up, the patient had ongoing treatment for her extracranial disease (KPS 70). No major neurological deficit, cognitive impairment or epileptic activity was reported at this stage.

Table 4:: Case 1 – Tumor volume dynamics between GKRS 1 and GKRS 3 (above) and corresponding peripheral prescription doses at each GKRS (below).

Table 5:: Case 1 – Radiation dose dissipation to OAR (three muscle groups of the upper limb) without nTMS and with nTMS at each GKRS. Substantial sparing of healthy tissues was achieved by integrating nTMS.

Table 6:: Case 1 – Dose distribution inside the tumor at 10 Gy, 12 Gy, 15 Gy. Dose distributions required to achieve tumor ablation remained nearly unaffected when integrating nTMS in treatment planning.

Table 7:: Case 1 – MRI- and LGP-verified tumor volume dynamics during treatment (GKRS 1 to GKRS 3) and up to last follow-up.

Close

Figure 3:: Tumor volume dynamics post-GKRS (see table 8). Case 1: optimal tumor ablation. Case 2: inital tumor volume reduction followed by ARE-evolvement.

Export to PPT

Case 2

As in Case 1, the treatment was well tolerated. Follow-up MRI at 1 month showed slight volume reduction of M1 and M2 as well as reduced edema. The follow-up CT scan at 3 months showed further volume reduction of M1 and M2 but progression of the perilesional edema around M1 subsequently assessed as an ARE [Table 8]. At this point, the patient was suffering from intermittent confusion, expressive dysphasia and further weakness of his left hand; the patient’s condition improved with steroids shortly thereafter. A complementary MRI 3 weeks later demonstrated a decrease in perilesional edema around M1. Despite an Engel score of 1, the patient started prophylactic antiepileptic treatment at this time. Unlike Case 1, the patient underwent a follow-up nTMS 6 months after treatment completion (April 2017) due to MRI-confirmed ARE-persistence around M1 [Table 8, Figures 3 and 4] and progressive intrathoracic disease. As in Case 1, the stimulation protocol and the muscles chosen for mapping in the postsurgical session were similar to that of the preoperative assessment. Postoperative rMT was found to be lower than the preoperative assessment [Table 1]; however, there was no substantial side difference. MEP latencies and amplitudes were within normal limits, though lower than the presurgical values. Postsurgical nTMS mapping showed no identifiable motor mapping distortion [Figure 4]. As in Case 1, GKRS did not affect or distort the original functional configuration found in the pre-GKRS nTMS examination. No motor deficit or other neurological disturbances were reported at this point, and the patient was kept on low dose corticosteroids. The corresponding structural MRI at this stage (6 months post-GKRS) revealed a limited increase in the diameter of M1 [Table 8, Figure 4], and persistent signs of ARE; M2 proved essentially unchanged at this stage. Regrettably, this last examination also revealed seven new metastases. Despite further general deterioration in the days that followed (KPS <70/RPA 3), the patient’s neurological condition remained stable. No motor or sensory deficits, cognitive impairment or seizure activities were reported before succumbing to his cancer 10 months post GKRS- treatment.

Table 8:: Case 2 – Tumor volume dynamics from the time of GKRS up to last follow-up (at 10 months).

Close

Figure 4:: nTMS motor mapping – pre- and post-GKRS for Case 1 and 2. nTMS motor mapping of three upper limb muscles (mAPB - green, mFDI - orange, mEDC - yellow) contralateral to the side of the stimulation. (a) Case 1 pre-GKRS nTMS. (b) Case 1 – 12 months post-GKRS nTMS. (c) Case 2 pre-GKRS nTMS. (d) Case 2 – 6 months post-GKRS nTMS.

Export to PPT

Rationale of the study

The present paper illustrates the complexity of metastatic brain disease in motor eloquent cortex. Patients with such lesions are generally at increased risk of postsurgical neurological deterioration with reduced gross and fine motor performance in the arm and hand. Furthermore, tumors located in the rolandic and perirolandic regions remain a source of concern due to the ensuing risk of developing epileptic seizures. These patients may benefit from focal therapy such as single fraction and hypofractionated GKRS;[23,33,35,49] however, the level of risk of radiation-induced injury to the eloquent cortex (cortical and subcortical), remains a subject of debate, particularly in the face of larger target volumes, extensive edema, and histology-surrogate factors such as radioresistance. In a prospective study by Williams et al.,[58] the authors reviewed 273 patients with 316 metastatic brain lesions and showed that 32% of lesions were associated with new complications. In their multivariate analysis, eloquent brain cortex location was found to be significantly associated with this type of neurological adverse event. Therefore, in the case of brain radiosurgery, careful treatment planning, aiming to protect adjacent (perilesional) functional structures, could decrease the risks of postsurgical functional deficiencies. In this two-patient case series, we successfully integrated the noninvasive neurophysiological method, nTMS, in presurgical modified dose planning. Furthermore, we used nTMS as a tool for postsurgical assessment, where the evaluation did not show any change in the location of the primary motor area of the hand suggesting the benefits of tailored dose distribution to the cortical regions. The latter was particularly the case in subject nr 2, where the patient developed an ARE post-GKRS yet without any serious neurological repercussions at long-term.

Nevertheless, the application of nTMS in radiosurgery is fairly new. Historically, the use of TMS in presurgical mapping remained limited until the availability of MRI- based neuronavigation assisted surgical series. In this context, intraoperative mapping of motor eloquent cortex under general anesthesia and speech mapping during awake surgery have generally been performed applying direct cortical stimulation (DCS); although this method is widely considered the gold standard of cortex functional mapping, its use in GKRS remains precluded by invasive requirements.

nTMS and radiosurgery: what is the evidence?

Functional mapping modalities, for example, functional MRI (fMRI),[15,52,56] magnetoencephalography,[3] and diffusion tensor imaging (DTI)[13,36] have been used as presurgical mapping tools; however, as formulated in several comparative studies, the latter modalities have method-specific limitations when compared to gold standard DCS.[14,19,28] Nonetheless, there is extensive evidence demonstrating the advantages of nTMS mapping in terms of high spatial resolution and accuracy;[27] indeed, nTMS based function mapping modalities, including nTMS-based DTI-fiber tracking, have been increasingly investigated for the purpose of providing more accurate topographic data of the perilesional OAR while optimizing the extent of resection and the outcome of microsurgery in terms of neurofunction.[7,34,40-42,51,50] Surprisingly, the advantages of the above mentioned nTMS- based modalities have not been equally utilized in the field of radiosurgery. Only a limited number of studies have described the feasibility and ulterior outcome of integrating nTMS data onto radiosurgery planning.[6,9,20,38,46,55] In this context, we must assume that both DTI and fMRI have their inherent methodological limitations in the field of stereotactic radiotherapy; nevertheless, it is our view that nTMS data can at least be utilized as a seeding point for safe volume delineation; the latter observation has previously been discussed elsewhere.[57]

To further substantiate the above-mentioned presumptions, we feel worth discussing three pioneer studies applying nTMS alone in radiosurgical planning [Table 9]. Conti and colleagues[6] conducted a retrospective analysis on a group of patients with brain lesions in critical areas who had radiosurgery (n=25); pre-treatment investigations included functional neuroimaging (fMRI, DTI) (n = 13) or navigated TMS motor mapping (n = 12). While measuring the volume of the functional motor cortex defined by either fMRI or nTMS, the study showed that the functional relevance of the cortical surface was 30% smaller than that of fMRI, leading to greater dose reduction (25%) to cortical functional areas. This indicated the clear advantage of the use of nTMS mapping over other functional imaging modalities. Later studies showed similar encouraging results. Pitch et al.[38] investigated 11 patients with eloquent tumors who were eligible for radiosurgery treatment. Unlike our study, the group investigated the utility of speech mapping data (n = 3) in addition to nTMS motor mapping (n = 8) and integrated both in radiosurgical planning. In a similar environment, Kato et al.[20] investigated 10 patients with arteriovenous malformation using nTMS motor and speech mapping. Successful delineation of the the affected motor cortex was reported, including the cortical areas of hypervascularization. Of note, unlike Kato et al., the first two studies included patients with variable types of lesions; it is more likely that a clearer clinical advantage is demonstrated when investigating a more homogenous group of patients. In view of the underlying neurofunctional symbiosis between the corticospinal tract and the motor cortex, we propose that adding DTI to nTMS mapping could provide substantial benefits in terms of identifying a “broader” deployment of OAR, hence redirecting and optimizing dose distributions outside the target. Of note, since this feasibility study was specifically conceived to assess the technical aspects of integrating nTMS cortex motor mapping into GKRS treatment planning, we decided not to include DTI as a measure to decrease the risk of bias. However, based on the results of this study and the above drawn conclusions, combining DTI and nTMS to radiosurgery planning will potentially be the subject of a next case series.

Table 9:: Review of literature.

Further aspects

As reported in this case series, both patients had metastatic brain lesions assessed as radiosensitive, yet treated with two different GKRS modalities due to different tumor- surrogate and perilesional variables. Unlike previously published data,[6,20,38] we integrated nTMS mapping in post-GKRS follow-up with the aim to (i) identify possible plasticity-associated relocation of adjacent motor cortex in relation to clinical motor functional recovery and (ii) plot the degree of motor fiber damage postradiation. Although our reported cases did not show any significant plastic changes, several other studies on patients undergoing tumor resection have indeed reported postsurgical plastic relocation of the motor cortex after 3–42 months,[8] >7 months,[21] >6 months in younger patients[31] and even after multistage surgery.[54] When investigated preoperatively, Kato et al. reported potential plastic changes of speech eloquent cortex in the form of right hemispheric involvement of speech function; however, posttreatment changes were not investigated in this study.[20] In the case of radiosurgery, it can be debated whether plastic changes mimicking the rates described at microsurgery occur in the adult patient.[30] In our cases, the postsurgical nTMS performed 6 months postsingle-fraction GKRS (Case 2) and at 12 months posthypofractionated GKRS (Case 1) showed largely unchanged motor mapping parameters. Most importantly, despite nTMS assisted modified dose planning, none of the patients showed any considerable motor functional disability at short- or long-term follow-up. Since different studies seem to suggest a decreased risk of ARE in hypofractionated treatments compared to single-dose GKRS,[10,17,18,22,32,47] a series of questions arise in this context: is there any correlation between the time and degree of ARE evolvement and possible focal plastic redeployment? How could customized radiation schedules and potential dose-modulated immune response influence the outcome of these plastic changes? Based on our observations, we suggest that GKRS-based hypofractionated treatments (as compared to single-dose GKRS) could reduce the risk for focal plastic changes in definite settings. More studies with larger groups are warranted to confirm the latter.