Dosimetric feasibility of stereotactic ablative radiotherapy in pulmonary vein isolation for atrial fibrillation using intensity‐modulated proton therapy

Abstract Purpose To evaluate dosimetric properties of intensity‐modulated proton therapy (IMPT) for simulated treatment planning in patients with atrial fibrillation (AF) targeting left atrial‐pulmonary vein junction (LA‐PVJ), in comparison with volumetric‐modulated arc therapy (VMAT) and helical tomotherapy (TOMO). Methods Ten thoracic 4D‐CT scans with respiratory motion and one with cardiac motion were used for the study. Ten respiratory 4D‐CTs were planned with VMAT, TOMO, and IMPT for simulated AF. Targets at the LA‐PVJ were defined as wide‐area circumferential ablation line. A single fraction of 25 Gy was prescribed to all plans. The interplay effects from cardiac motion were evaluated based on the cardiac 4D‐CT scan. Dose‐volume histograms (DVHs) of the ITV and normal tissues were compared. Statistical analysis was evaluated via one‐way Repeated‐Measures ANOVA and Friedman’s test with Bonferroni’s multiple comparisons test. Results The median volume of ITV was 8.72cc. All plans had adequate target coverage (V23.75Gy ≥ 99%). Compared with VMAT and TOMO, IMPT resulted in significantly lower dose of most normal tissues. For VMAT, TOMO, and IMPT plans, Dmean of the whole heart was 5.52 ± 0.90 Gy, 5.89 ± 0.78 Gy, and 3.01 ± 0.57 Gy (P < 0.001), mean dose of pericardium was 4.74 ± 0.76 Gy, 4.98 ± 0.62 Gy, and 2.59 ± 0.44 Gy (P < 0.001), and D0.03cc of left circumflex artery (LCX) was 13.96 ± 5.45 Gy, 14.34 ± 5.91 Gy, and 8.43 ± 7.24 Gy (P < 0.001), respectively. However, no significant advantage for one technique over the others was observed when examining the D0.03cc of esophagus and main bronchi. Conclusions IMPT targeting LA‐PVJ for patients with AF has high potential to reduce dose to surrounding tissues compared to VMAT or TOMO. Motion mitigation techniques are critical for a particle‐therapy approach.


| INTRODUCTION
Atrial fibrillation (AF) is the most common cardiac arrhythmia 1 mostly caused by ectopic beats initiated mostly from the pulmonary veins (PVs) entering the left atrium. 2 A well-established approach for treatment of symptomatic drug-refractory patients with AF is catheter ablation (CA) by generating electrical isolation of the left atrial-pulmonary vein junction (LA-PVJ) which includes radiofrequency ablation and freeze ablation. 1,3 However, ablation procedures are invasive, complicated, and requires sufficient training and experience.
The risk of congestive heart failure, stroke, bleeding, and complications of CA increase hospitalizations rate. 4 The incidence of major complications with CA, including bleeding, atrioesophageal fistula, pericarditis, cardiac tamponade/perforation, and even death, 1 is approximately 5%. 5-7 In patients with paroxysmal and nonparoxysmal AF, the overall 1-year recurrence rate after single procedure is about 40% or more. 8

Stereotactic ablative radiotherapy (SABR) is a matured treatment
technique that delivers precise and intense radiation dose in one or a few fractions to tumors while minimizing damage to surrounding tissues. 9 Successful adaptation of SABR to target cardiac sarcoma has been reported. 10 Preclinical studies exploring SABR using various forms of particle therapy for cardiac ablation have demonstrated histological changes and electrophysiological effects (voltage/potential amplitude or bidirectional block). [11][12][13][14][15][16][17][18] Using heavy ion-based radioablation for healthy pigs, Lehmann and Rapp et al. demonstrated radiation-induced fibrosis, microvascular damage, and chronic inflammation. 12,17 Successful clinical treatments for malignant ventricular tachycardia or recurrent paroxysmal AF have also been reported. [19][20][21][22][23][24][25][26][27][28][29][30][31][32] Therefore, SABR may be a novel and noninvasive approach for AF. However, delivering one concentrated dose to the cardiac region can be challenging due to cardiac motion and its close proximity to critical structures. Compared to ventricular tachycardia, previous studies showed that SABR was considered more challenging for AF patients, and the reasons are multifold: (a) creating electrical isolation of the LA-PVJ requires a significantly complex target; (b) it is unclear whether single fraction 25 Gy is adequate; and (c) the close proximity of esophagus and bronchus to the target may be a limiting factor. [33][34][35] In addition, long-term follow-up in some clinical trials for patients with breast cancer has shown that radiotherapy can increase the subsequent rate of ischemic heart disease, presumably through incidental exposure of the heart to ionizing radiation. 36 Several studies revealed that radiation dose to the heart in lung cancer patients was associated with the risk of cardiac events 37-41 and overall survival. 37,40,[42][43][44][45] While cardiotoxicity is usually considered as a latent event after 10 years, acute toxicity has also been observed early after radiotherapy. 36,44 Previous reports of non-invasive ablative radiation for ventricular tachycardia or recurrent paroxysmal AF used photon-based treatment plans, i.e., volumetric-modulated arc therapy (VMAT) or Cyberknife. [19][20][21][22]32 It has been shown that single-fraction ablative radiotherapy significantly reduced burden of AF and ventricular tachycardia, but some cardiac complications, i.e., pericarditis and heart failure exacerbation, have also occurred early upon treatment completion. 20 As opposed to photon-delivery techniques, intensitymodulated proton therapy (IMPT) takes advantage of proton beam's unique Bragg Peak property and high-dose conformity with pencil beam modulation, thus may be a promising alternative for AF patients. However, IMPT is highly sensitive to tissue motion, resulting in interplay effects between treatment beams and intrafractional target motion, which could produce dosimetric cold or hot spots. 46 Hohmann et al. reported possible decrease in left ventricular function after proton beam therapy in 20 swine and the adverse effect was dose dependent. 47 This study aimed to investigate dosimetric properties of IMPT for non-invasive AF ablation in comparison with photon-base treatment techniques (including VMAT and TOMO) and the interplay effects in IMPT plan from cardiac motion.

2.A | CT simulation
The study has been reviewed and approved by the institutional review board (No. 2019[240]). We retrospectively collected 11 thoracic 4D computed tomography (CT) scans from our institution, of which 10 were stamped with respiratory motion phases (patients 1-10) and 1 was stamped with cardiac motion phases (patient 11). All had received 2 thoracic 4D-CT scans, first without and with intra-

2.C | Treatment planning
The AIP dataset was used to create all photon and proton plans for 10 patients about respiratory motion. A single fraction of 25 Gy was used as the prescription for all plans. For the photon plans, PTV was expanded uniformly 3 mm from ITV and physical dose of 25 Gy was prescribed to the PTV; while a total dose of 25 GyE (relative biological effectiveness 1.1) was prescribed to ITV for the proton plans. Normal tissue constraints include 0.03 cc of spinal cord, esophagus, and main bronchi receiving no more than 10 Gy, 16 Gy, and 20.2 Gy, respectively, and total lung V 20 ≤ 10%, V 2.5 ≤ 50%, and V 7 Gy < 1500 cc. The dose to the sensitive structures was minimized, with the following priorities settings (high to low): 1 = spinal cord, 2 = esophagus, 3 = main bronchi, and 4 = total lung. Despite no hard constraints for cardiac dose exposures, we try the best to minimize dose of whole heart, pericardium, and all cardiac substructures.
All VMAT, TOMO, and IMPT plans were created blindly without prior knowledge to the quality of others. All three plans were reviewed by senior physicists and approved by one attending physician based on target goals and normal tissue constraints. VMAT plans were generated using double-full arcs on Eclipse treatment planning system (Eclipse, Version 13.5, Varian Medical Systems, Palo Alto, CA). For the TOMO plans, a field width of 1.048 cm, pitch of 0.287, and modulation factor of 3.0 were used on the HD TM treatment planning system (Version 5.1.1.6, Accuray, Sunnyvale, CA).
IMPT plans were created using four beams based on IBA Proteus ® PLUS machine model using pencil beam scanning. The dose distribution was calculated by the Monte Carlo algorithm on RayStation treatment planning system (Version 7.0, RaySearch Laboratories, Stockholm, Sweden), with robust optimization to cover 99% of ITV.
Set-up error was determined at 3 mm, and the range uncertainty at 3.5% in addition to a distal margin of 2 mm. 51 The spot size sigma varies between 2.6 and 7.2 mm in the energy range 70-230 MeV.
When evaluating interplay effects in IMPT plans, we calculated an IMPT nominal plan with the same optimized approach as mentioned above. A "dynamic dose" was calculated using in-house Python scripts in RayStation for evaluating interplay effects from cardiac motion. The calculation parameters used for the IMPT nomi-

2.D | Evaluation
For evaluating target displacements of cardiac motion, the maximal displacements for centroids of CTVL and CTVR at all phases of cardiac-gated 4D-CT were recorded in medial-lateral (ML), anterior-posterior (AP), and superior-inferior (SI) direction, in reference to phase T 00 .
For dosimetric evaluation, dose-volume histograms (DVHs) and dose distributions were calculated for ITV and other critical structures, such as the whole heart, pericardium, cardiac substructures, total lung, esophagus, main bronchi, and spinal cord. For pair comparison, all plans were normalized to 100% of prescription dose (25GyE) covering 99% ITV. The dosimetric parameters for the rest of unconventional critical structures (including whole heart, pericardium, individual cardiac chambers, and four main coronary arteries (LMC, LAD, LCX, and RCA)) are shown in Table 2. Compared with VMAT or TOMO, IMPT improved D mean , V 2 , V 5 , and V 10 of the whole heart and pericardium.

2.E | Statistical analysis
Compared with VMAT and TOMO plans, IMPT reduced D mean and V 5 of the whole heart by 2.51 Gy and 2.88 Gy, 23.65% and 23.15%, respectively, which also reduced D mean and V 5 of pericardium by 2.15 Gy and 2.39 Gy, 18.94% and 18.91%, respectively. In addition, For IMPT plans, D mean of four cardiac chambers was significantly lower than VMAT and TOMO plans.
Dose-volume histograms analysis of V 2 , V 5 , V 10 , V 15 , and V 20 of four cardiac chambers and four main coronary arteries was shown in Stereotactic ablative radiotherapy may be a viable option to overcome some shortcomings of previous treatment for AF ablation.
In the past decades, it is very difficult to balance the safety and efficacy of radiofrequency ablation for AF. High recurrence risk of AF after CA is usually caused by inaccessible atrial myocardium and discontinuous ablation lines due to insufficient energy to generate transmural myocardial scar considering severe complications such as atrioesophageal fistula. 52 Compared with conventional radiofrequency ablation, stereotactic arrhythmia radioablation could provide continuous ablation lines and adequate energy to generate transmural lesions for isolating electrical conduction between pulmonary veins and left atrium.
Intensity-modulated proton therapy can reduce the dose of heart and cardiac substructures and may reduce potential adverse effects.
The recent phase I/II trial reported that two patients (10.5%) who received stereotactic ablative radiotherapy (IMRT or VMAT) for malignant ventricular tachycardia experienced a grade 3 treatmentrelated severe adverse effect. 20 One patient experienced heart failure exacerbation (grade 3), and another patient had pericarditis and was treated with prednisone. These adverse events could be associated with dose exposure to the normal heart and pericardium, but it is necessary to notice that the health condition of these patients is very poor. The following secondary analysis for this trial found that the correlation between larger target volume and shorter overall survival. 19 Compared with the radioablation of ventricular arrhythmias, which has been carried out in clinics, the targets in AF have high complexity, and close to critical structures. Hence, special attention should be paid to the dose of whole heart, pericardium, and cardiac substructures. A recent research found that rates of major coronary events (MCE) increased linearly with the mean dose to the heart by 7.4% per gray, with no apparent threshold. 36 A validation and modification of a prediction model for acute cardiac events in patients with breast cancer treated with radiotherapy showed that left ventricle volume receiving 5 Gy (LV-V 5 ) was the most important prognostic dose-volume parameter. 53 For lung cancer patients underwent radiotherapy, a study showed that the rates of grade ≥3 cardiac events increased with mean heart dose by 7% per Gy. 40 In our study, IMPT had the reduction of mean dose of whole heart by approximately 2.7 Gy and the LV-V 5 by at least 10% compared with those photon plans. Another study found that the risk of pericardial effusion (PCE) was related to several cardiac parameters (e.g., mean dose of whole heart / pericardium, heart / pericardium V5), and PCE rate increased with mean dose of pericardium by 5% per Gy. 37 Our study showed decreased mean dose of pericardium, heart V 5 , and pericardium V 5 by about 2.3 Gy, 23% and 19%, respectively, with IMPT.
Previous study found that mean dose, V 5 and V 30 to LAD were related to acute coronary syndrome or congestive heart failure after conventional-dose chemoradiation therapy for lung cancer. 54 Another study showed that coronary artery-based model, including LAD-V 5 and LCX-V 20 , was superior to the whole heart when analyzing late ischemic cardiotoxicity. 55   V m (%) means the percentage of volume receiving at least m (Gy) doses. D n (Gy) means the maximum dose received by n % volume. Abbreviations: IMPT, intensity-modulated proton therapy; ITV, internal target volume; TOMO, helical tomotherapy; VMAT, volumetric-modulated arc therapy. aThere was no significant difference of D 0.03cc to esophagus and main brochi among three groups, so we did not do paired test.
constraints for coronary arteries, partly due to limited studies on the relationship between dose of coronary arteries and ischemic heart diseases. Therefore, IMPT has its own safety advantages in the application of atrial arrhythmia ablation, which needs further preclinical and clinical evidence to support.
For CA in AF patients, the esophageal injury is also one of the most important complications and even lethal. 1 This study also showed that esophageal toxicity might still need special attention with IMPT. The average D 5cc of esophagus was 2.5 Gy and D 0.03cc was about 15 Gy, both of which met tolerance thresholds in thoracic single-fraction SABR. 56,57 However, D 0.03cc of esophagus was not significantly improved by IMPT, due to its close proximity to targets.
In addition, motion management is challenging, especially for IMPT.
When considering that patients would be treated under respiratory gated or free breathing, we used the respiratory-phased binned 4D-CT for dosimetric comparison among three modalities in our study, which is the same as previous clinical studies. [19][20][21] In our study, we found that the median ratio of CTV to ITV is 32%. Therefore, ITV might be reduced by 68% if respiratory-gated treatment is applied. The previous work found that displacements of LA-PVJ were mainly affected by respiratory motion, and the mean displacements were 1-4 mm caused by cardiac contraction under breath-hold conditions. 58 A recent publication reported that the motion of cardiac substructures is not uniform and patient specific. 59 Even with appropriately compensated respiration motion using various techniques, including gating, tracking, and/or breath-holding, cardiac motion-induced anatomy deformation is complex, which could be partially compensated using personalized margin determined based on cardiac 4D-CT images. Interplay effects of scanned particle beams and moving targets can severely impact dose distributions. 60

| CONCLUSIONS
IMPT has high potential to reduce dose to surrounding tissues due to its sharp dose gradients but motion mitigation techniques are essential for a particle therapy approach for AF. This study sheds light on the use of proton therapy for noninvasive radiotherapy for AF.

ACKNOWLEDGMENTS
This research was supported by Beijing Natural Science Foundation (Grant No. 7182164).

CONF LICTS OF INTEREST
None.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.