Linac‐based stereotactic radiosurgery (SRS) in the treatment of refractory trigeminal neuralgia: Detailed description of SRS procedure and reported clinical outcomes

Abstract Purpose/Objectives To present our linac‐based SRS procedural technique for medically and/or surgically refractory trigeminal neuralgia (TN) treatment and simultaneously report our clinical outcomes. Materials and Methods Twenty‐seven refractory TN patients who were treated with a single fraction of 80 Gy to TN. Treatment delivery was performed with a 4 mm cone size using 7‐arc arrangement with differential‐weighting for Novalis‐TX with six MV‐SRS (1000 MU/min) beam and minimized dose to the brainstem. Before each treatment, Winston–Lutz quality assurance (QA) with submillimeter accuracy was performed. Clinical treatment response was evaluated using Barrow Neurological Institute (BNI) pain intensity score, rated from I to V. Results Out of 27 patients, 22 (81%) and 5 (19%) suffered from typical and atypical TN, respectively, and had median follow‐up interval of 12.5 months (ranged: 1–53 months). For 80 Gy prescriptions, delivered total average MU was 19440 ± 611. Average beam‐on‐time was 19.4 ± 0.6 min. Maximum dose and dose to 0.5 cc of brainstem were 13.4 ± 2.1 Gy (ranged: 8.4–15.9 Gy) and 3.6 ± 0.4 Gy (ranged: 3.0–4.9 Gy), respectively. With a median follow‐up of 12.5 months (ranged: 1–45 months) in typical TN patients, the proportion of patients achieving overall pain relief was 82%, of which half achieved a complete pain relief with BNI score of I‐II and half demonstrated partial pain reduction with BNI score of IIIA‐IIIB. Four typical TN patients (18%) had no response to radiosurgery treatment. Of the patients who responded to treatment, actuarial pain recurrence free survival rates were approximately 100%, 75%, and 50% at 12 months, 15 months, and 24 months, respectively. Five atypical TN patients were included, who did not respond to treatment (BNI score: IV–V). However, no radiation‐induced cranial‐toxicity was observed in all patients treated. Conclusion Linac‐based SRS for medically and/or surgically refractory TN is a fast, effective, and safe treatment option for patients with typical TN who had excellent response rates. Patients, who achieve response to treatment, often have durable response rates with moderate actuarial pain recurrence free survival. Longer follow‐up interval is anticipated to confirm our clinical observations.

response rates. Patients, who achieve response to treatment, often have durable response rates with moderate actuarial pain recurrence free survival. Longer followup interval is anticipated to confirm our clinical observations. intervention such as microvascular decompression, and stereotactic radiosurgery. [1][2][3][4] Historically, gamma knife-based stereotactic radiosurgery (SRS) has been considered an effective and noninvasive alternative treatment modality associated with minimal toxicityparticularly in patients with medically and surgically refractory TN or those who are not ideal surgical candidates. [5][6][7][8][9] For example, in a multi-intuitional review of 503 patients with TN who had been treated with gamma knife-based SRS, 58% of patients achieved complete pain relief and 36% of patients achieved partial pain relief. 8 Linac-based SRS has become an increasingly popular treatment modality for TN due to technological advancements which have allowed for precise radiation delivery in a fast and effective manner. [10][11][12][13] Recently, many researchers have presented linac-based SRS treatment outcomes for TN patients which are comparable with gamma knife data. [14][15][16][17][18][19] Due to the effectiveness of linac-based SRS for treatment of smaller target such as TN, we sought to present a detailed description of our linac-based SRS technique as well as report our long-term clinical outcomes in patients with medically and/or surgical refractory TN.

2.A | Patient imaging and frame placement
After obtaining approval from our institutional review board, a retrospective review was conducted consisting of a total of 27 TN patients who had been treated at our institution from 2009 to 2016 using frame-based, linac-based SRS. All patients underwent a high-resolution magnetic resonance imaging (MRI) scan consisting of 1 mm thin slices with T1-weighted, T2-weighted, and 3D-fast imaging employing steady state acquisition (FIESTA) sequences prior to treatment. On the day of radiosurgery treatment, an experienced neurosurgeon placed a BrainLAB stereotactic frame on the patient's head after application of a local anesthetic. Depth Helmet bobble 20 measurement was performed for quality assurance of the frame placement and, immediately thereafter, the patient was set up for the planning computerized tomography (CT) simulation which was performed on a 16 slice Phillips Brilliance Big Bore CT Scanner (Phillips, Cleveland, OH) and BrainLAB CT localizer (BrainLab Head&Neck Localization Inc., Heimstetten, Germany). CT simulation images were acquired with 512 9 512 pixels at 0.75 mm slice thickness and 0.75 mm slice spacing following departmental SRS scanning protocol.

2.B | Target delineation and SRS treatment planning
The MRI was co-registered with the planning CT image set and an experienced neurosurgeon and radiation oncologist delineated the trigeminal nerve root (TNR), for isocenter placement, using the 3D-FIESTA MRI sequence. The target was localized to the base of the trigeminal nerve at the junction of nerve entry into Meckel's Cave and exit from the brainstem. Organs at risk (OAR) were delineated on the co-registered MRI and consisted of the following structures: brainstem, optic apparatus (optic chiasm and bilateral optic nerves), eyes and lenses, and temporal lobe of the brain.
For each treatment, a seven-arc plan was devised in iPlan Brain-LAB to deliver the single-fraction prescription dose to the 100% isodose line (IDL), using six MV-SRS beams (1000 MU/min), and a 4 mm diameter cone size. The treatment plans were optimized in order to minimize brainstem dose as well as avoided beam entry through the eyes. All treatment plans were performed using heterogeneity corrected pencil-beam algorithm with 1.

2.E | Machine quality assurance and patient setup
For the given collimator, couch, and gantry rotations, daily Winston-Lutz (WL) QA tests 10 were performed using a 7.5 mm circular cone and a couch mount with a 5 mm diameter mechanical bearing ball (BB). In our clinic, due to the integration of WL QA procedure with ExacTrac system, ExacTrac system was calibrated before the WL QA F I G . 2. Dose distribution for a 62-yr-old male with refractory right trigeminal neuralgia. An 80 Gy point dose to the isocenter was prescribed. The IDLs for 40 Gy (light green) and 16 Gy (blue) are clearly shown in conjunction with contours for brainstem (green) and TNR (red). The isocenter was localized by identifying the midpoint between the trigeminal eminence where the dorsal root merges with the lateral pons (brainstem) and entry into Meckel's cave (see plus sign -Coord 1 in all 3-view). A 4 mm diameter circular cone and seven noncoplanar differentially weighted arcs were used to minimize brainstem dose. A total of 19,140 MU was delivered with a total beam-on-time of 19.14 min (not including couch kick time). In this particular case, max-dose to brainstem was 14.9 Gy, dose to 0.5 cc of brainstem was 3.8 Gy, max-dose to optic apparatus was less than 1.5 Gy, and max-dose to eyes and lenses were 0.6 Gy and 0.1 Gy, respectively. Follow-up at 13 months demonstrated that this patient had achieved complete pain relief (no pain, no medication, and BNI score of I). with a pair of oblique kilo-voltage x-ray images of the BB was acquired and automatic 2D-to-3D image registration was performed.
The WL QA results were considered acceptable if the 5 mm diameter mechanical BB was conformally encompassed by the 7.5 mm radiation field for every gantry, couch, and collimator angle. On a single strip of Gafchromic film, eight static fields (5 mm BB, with 7.5 mm cone size) with following gantry and couch angles were shot for daily WL QA test (G0, G90, G180, G270; with Couch 0) and (C270, C315, C45, C90; with Gantry 0), respectively. A total of 700 MU/beam was used for WL QA. On each shot, submillimeter coincidence of radiation and mechanical isocenters was maintained at all the times with the use of daily WL QA. In addition to the WL QA, a daily QA check of kilovoltage to megavoltage imaging isocenter coincidence was performed prior to patient setup for TN SRS. All QA procedures were in compliance for radiosurgery treatment delivery including QA for frame placement verification using the depth Helmet bobble measurement. 20 It was ensured that the originality of the frame placement before CT simulation and prior to treatment was within AE 1 mm of reproducibility. respectively. The mean value of angular couch correction discrepancy was 0.1 AE 0.4°(ranged, À0.7-0.6°). These couch correction discrepancies, however, were not applied for the actual treatment considering that these errors were within the range of uncertainty for CBCT image reconstruction and OBI gantry rotation (within AE 1 mm for translational and AE 0.7°for rotational shifts). Overall purpose of verification CBCT was to conform that if there was any unanticipated huge shifts (≥AE 2 mm/2°) have been observed, therefore, patient setup could be reconsidered.  Table 1

3.A | Patient characteristics
The detailed descriptions of patient characteristics are listed in

3.B | Dosimetric and treatment delivery parameters
On a per-patient basis, the total number of delivered MU for all 27 patients who underwent TN SRS is shown in Fig. 4. In our experience, the mean MU was 19,500 and was fairly standard for all TN patients treated with 80 Gy prescription doses. Knowledge of the average total number of MU is advantageous in that it allows for quick identification of some major errors related to dose calculation as would be suggested by a calculated total MU which is well above or below the average value.
In  pain recurrence free survival rates were approximately 100%, 75%, and 50% at 12 months, 15 months, and 24 months, respectively (see Fig. 8). These results were generated using the Kaplan-Meier The safety, efficacy, and localization accuracy of linac-based TN SRS has been studied by several researchers. [10][11][12][13] In our clinical implementation of linac-based TN SRS, we adhered with those standard clinical protocols and guidelines. Treatment planning procedures and patient outcomes for linac-based TN SRS has also been reported by many investigators. [14][15][16][17][18][19] For instance, using a seven-arc geometry with a 4 mm circular cone size, Richards et al. 17  Utilization of CyberKnife SRS for the treatment of TN has also been reported by many investigators. [21][22][23] Although the treatment outcomes were similar to linac-based SRS treatment, the reported beam-on-times for CyberKnife treatment were relatively longer (~45-60 min) compared to the average beam-on-time in our study (~20 min). Our prescription dose was 80 Gy to all patients. By identifying the optimal arc arrangement with a 4 mm circular cone size and differential arc weighting, the dose to the brainstem was minimized such that the average maximum brainstem point dose was less than 16 Gy (20% IDL). There was no posttreatment toxicity such as cranial nerve deficit, numbness, or brain necrosis observed in our study.
In summary, we have presented our faster and robust technique for the treatment of TN utilizing linac-based SRS. Our overall clinical outcomes suggest that this technique is both safe (less radiationinduced toxicity) and effective for patients with typical TN SRS treatment. However, the lack of success rates for those patients who underwent for atypical TN SRS (similar results were presented by Smith et al. 19 ) needs further investigations. Due to the advanced of image-guidance system, linac-based TN SRS with frameless radiosurgery setup 24-26 merits further investigation.

| CONCLUSION
In this paper, we have presented our linac-based SRS treatment procedure for TN and the corresponding clinical outcomes. Our overall response rate was 82% in patients with typical TN with half of those patients achieving complete pain relief. Of the patients who responded to treatment, actuarial pain recurrence free survival rates were approximately 100%, 75%, and 50% at 12 months, 15 months, and 24 months, respectively. None of the atypical TN patients included in this study had a response to treatment. However, there was no treatment-related neurological toxicity observed in this study. Longer follow-up of these patients is anticipated to confirm our clinical observations.

CONFLI CT OF INTEREST
No conflict of interest.