A novel, yet simple MLC‐based 3D‐crossfire technique for spatially fractionated GRID therapy treatment of deep‐seated bulky tumors

Abstract Purpose Treating deep‐seated bulky tumors with traditional single‐field Cerrobend GRID‐blocks has many limitations such as suboptimal target coverage and excessive skin toxicity. Heavy traditional GRID‐blocks are a concern for patient safety at various gantry‐angles and dosimetric detail is not always available without a GRID template in user’s treatment planning system. Herein, we propose a simple, yet clinically useful multileaf collimator (MLC)‐based three‐dimensional (3D)‐crossfire technique to provide sufficient target coverage, reduce skin dose, and potentially escalate tumor dose to deep‐seated bulky tumors. Materials/methods Thirteen patients (multiple sites) who underwent conventional single‐field cerrobend GRID‐block therapy (maximum, 15 Gy in 1 fraction) were re‐planned using an MLC‐based 3D‐crossfire method. Gross tumor volume (GTV) was used to generate a lattice pattern of 10 mm diameter and 20 mm center‐to‐center mimicking conventional GRID‐block using an in‐house MATLAB program. For the same prescription, MLC‐based 3D‐crossfire grid plans were generated using 6‐gantry positions (clockwise) at 60° spacing (210°, 270°, 330°, 30°, 90°, 150°, therefore, each gantry angle associated with a complement angle at 180° apart) with differentially‐weighted 6 or 18 MV beams in Eclipse. For each gantry, standard Millenium120 (Varian) 5 mm MLC leaves were fit to the grid‐pattern with 90° collimator rotation, so that the tunneling dose distribution was achieved. Acuros‐based dose was calculated for heterogeneity corrections. Dosimetric parameters evaluated include: mean GTV dose, GTV dose heterogeneities (peak‐to‐valley dose ratio, PVDR), skin dose and dose to other adjacent critical structures. Additionally, planning time and delivery efficiency was recorded. With 3D‐MLC, dose escalation up to 23 Gy was simulated for all patient's plans. Results All 3D‐MLC crossfire GRID plans exhibited excellent target coverage with mean GTV dose of 13.4 ± 0.5 Gy (range: 12.43–14.24 Gy) and mean PVDR of 2.0 ± 0.3 (range: 1.7–2.4). Maximal and dose to 5 cc of skin were 9.7 ± 2.7 Gy (range: 5.4–14.0 Gy) and 6.3 ± 1.8 Gy (range: 4.1–11.1 Gy), on average respectively. Three‐dimensional‐MLC treatment planning time was about an hour or less. Compared to traditional GRID‐block, average beam on time was 20% less, while providing similar overall treatment time. With 3D‐MLC plans, tumor dose can be escalated up to 23 Gy while respecting skin dose tolerances. Conclusion The simple MLC‐based 3D‐crossfire GRID‐therapy technique resulted in enhanced target coverage for de‐bulking deep‐seated bulky tumors, reduced skin toxicity and spare adjacent critical structures. This simple MLC‐based approach can be easily adopted by any radiotherapy center. It provides detailed dosimetry and a safe and effective treatment by eliminating the heavy physical GRID‐block and could potentially provide same day treatment. Prospective clinical trial with higher tumor‐dose to bulky deep‐seated tumors is anticipated.

providing similar overall treatment time. With 3D-MLC plans, tumor dose can be escalated up to 23 Gy while respecting skin dose tolerances.
Conclusion: The simple MLC-based 3D-crossfire GRID-therapy technique resulted in enhanced target coverage for de-bulking deep-seated bulky tumors, reduced skin toxicity and spare adjacent critical structures. This simple MLC-based approach can be easily adopted by any radiotherapy center. It provides detailed dosimetry and a safe and effective treatment by eliminating the heavy physical GRID-block and could potentially provide same day treatment. Prospective clinical trial with higher tumordose to bulky deep-seated tumors is anticipated.
3D-MLC Crossfire, Bulky-tumors, cerrobend GRID-block, dose-escalation 1 | INTRODUCTION Spatially fractionated GRID therapy with megavoltage (MV) x-ray beams has proven to be an effective treatment modality for shrinking bulky (>8 cm, in diameter) malignant tumors. 1 Traditional GRID therapy treatments have shown great tumor response of bulky lesions with an overall response rate increase of 62% to 91% when they were treated with a single-dose of GRID therapy (≥15 Gy) followed by conventional extremal beam radiotherapy. 2 Another study of 71 patients with advanced or bulky tumors of varying histologies demonstrated that 78% response rate for pain palliation and 58.5% and 72.5% objective clinical response rate for mass effect after GRID therapy of 10 to 20 Gy dose with or without additional external beam radiation. 3 All these early clinical studies demonstrated no significant skin toxicity with GRID therapy.
Currently, MV GRID therapy treatments are delivered using a high attenuation GRID-block with divergent holes, with step and shoot multileaf collimator (MLC) control points, and/or Tomotherapy machines. [4][5][6][7][8][9][10][11][12][13][14][15] Although, the treatment planning studies for GRID therapy using tomotherapy and step and shoot MLCs are evolving, these techniques require longer treatment times due to beam modulation and need patient-specific quality assurance (QA). Moreover, noninterdigitating MLCs potentially may not allow an efficient implementation of this method. The commercial availability of the standard traditional GRID-block is very limited in each radiotherapy clinic and it is very difficult to design. Additionally, treating deep-seated bulky-tumors with traditional single-field Cerrobend GRID-blocks could have major limitations such as suboptimal target dose and potentially unwarranted skin toxicity. Heavy traditional GRID-blocks are a concern for patient safety at various slanted gantry-angles and dosimetric detail may not be available readily without GRID-block template in the user's treatment planning system (TPS).
There are several studies suggesting that high doses of radiation (>15 Gy) cause an environment of potential lethal damage making tumor cells more sensitive to further doses of radiation. This is due to the endothelial cells of the tumor microvasculature. [17][18][19][20][21] Therefore, killing endothelial cells or obstructing small capillaries inside of the tumor will result in an avalanche of tumor cell deaths due to bystander killing in cells adjacent to irradiated regions. The GRID therapy approach takes advantage of this bystander effect that can result in de-bulking of large tumors. 21 Our clinical experience is that when using a single-field GRIDblock, deep-seated bulky tumors can receive about 1/3 or less of 15 Gy prescription dose, delivering a sub-optimal treatment to these patients. In this setting, skin toxicity is a major concern when escalating tumor dose. As mentioned above, traditional GRID-block is not readily available to the radiotherapy clinics and designing and mounting this heavily lifted cerrobend GRID-block to the Linac head poses a serious concern for patient safety. Therefore, the MLC-based 3D-crossfire technique can substantially escalate tumor dose to deep-seated bulky masses, deliver a more accurate and faster treatment, reduce dose to skin and other critical structures while avoiding patient safety concerns. Herein, we propose and validate a simple, yet clinically useful 3D-MLC crossfire technique that can be used at any radiotherapy clinic for possible same day treatments of deep-seated bulky tumors with potentially escalated tumor doses.

2.A | Patient setup and CT simulation
This retrospective study included 13 patients with deep-seated bulky lesions. Each patient had different primary diseases and treatment sites as shown in Table 1. Each patient was immobilized using a  reporting. The skin contour was generated within 5 mm of the patient body contour. In Eclipse, the skin to tumor center was estimated by using the tumor radius and its proximity to the skin contour. The maximal dose to 2 cm away in any directions from the GTV (D2cm) was calculated for plan evaluation. On the treatment day, these patients were setup using 100 cm SSD followed by a verification port film with GRID-block setup before treatment. The treatment was delivered once the GRID-block setup was verified by the treating physician.

2.C | 3D-MLC crossfire plans
All 13 patients (multiple primary disease sites, see Table 1) who underwent conventional single-field GRID-block therapy as described above were re-planned using a simple MLC-based 3Dcrossfire method. For each patient, a GTV contour was used to generate a 10 mm diameter and 20 mm center-to-center distance grid-pattern mimicking the conventional GRID-block using an inhouse MATLAB program. The program read the 3D-CT images and structure set (GTV contour) in DICOM format. A voxel mask of the grid lattice structure was created inside the GTV structure using MATLAB's boundaries function in DICOM format. The lattice structure was then imported into Eclipse for 3D-MLC based cross-

2.G | Simulating dose escalated plans
It has been reported in the GRID therapy literature that therapeutic  Fig. 2. In this case, the internal critical structures such as large bowel, liver, and right kidney were also spared in addition to skin dose tolerances while using the 3D-MLC crossfire technique. In Table 2, the skin to tumor center distance is shown for all patients.  Table 2).
In the physics second check, the independent MU calculation     13 However, in their study the GRID therapy plan was inversely-optimized with a simultaneous integrated boost (SIB) for many spheres generated inside the target.
Therefore, the plan needed additional treatment planning and optimization time as well as patient-specific QA due to MLC modulation. A major difference of our study from the previous two-dimensional-GRID therapy approach, 1-3,16,22 tomotherapy or MLC-based studies [4][5][6][7][8][9][10][11][12][13][14][15] was that our treatment planning approach uses an MLCbased, 3D-conformal forward planning technique with no beam modulation. Therefore, this MLC cross-firing procedure preserves the characteristics of 3D-conformal radiation therapy and provides all dosimetry information without the need for patient-specific QA.
Clinical and biological data suggest that the success of GRID therapy in shrinking large tumors depends on the high PVDR. One potential concern is the dose blurring due to tumor motion in GRID therapy. 26 Even with relatively shorter beam on times, but similar overall treatment time compared to traditional GRID-block, our 3D-MLC crossfire plans could potentially be delivered using image-guidance procedure. Furthermore, this time can be reduced by using recently adopted flattening filter free (FFF) beams 31 31,32 in the management of tumor motion for the MLC-based GRID therapy patients. Moreover, the potential use of the 3D-MLC crossfire approach for highly irregular GRID targets will be explored.

| CONCLUSION S
A simple yet clinically useful 3D-conformal MLC-based crossfire GRID-therapy technique resulted in enhanced target coverage for the deep-seated bulky tumors with reduced skin toxicity and other internal critical structures. This simple MLC-based approach can be easily adopted by any radiotherapy clinic. It provides detailed F I G . 4. Calculation of predicted average skin doses (maximal and dose to 5 cc of skin) as a function of escalated prescription doses (Dp) for all 13 GRID therapy patients. A simple three-dimensionalmultileaf collimator crossfire GRID planning technique allowed for escalation of tumor doses up to 23 Gy while maintaining the skin toxicity.
| 73 dosimetry and a safe, effective treatment modality by eliminating the heavy physical GRID-block without beam modulation. Moreover, using the 3D-MLC approach, our simulation study suggests that tumor dose can be escalated up to 23 Gy while avoiding skin toxicity. A prospective clinical trial is underway to evaluate the tumor local control rates and treatment related toxicity with an escalateddose for patients with 3D-MLC GRID therapy.

CONF LICT OF I NTEREST
No conflict of interest.