Single-Isocenter, Multiple Metastasis Treatment Planning

Abstract

Treatment of multiple brain metastases with radiosurgery has historically been performed with an isocenter for each individual target. On the Gamma Knife™ (Elekta, Stockholm, Sweden) platform, treatment was and still is performed with one more “shots” for each target; on the linear accelerator, treatment was historically executed with an array of cone- or MLC-based conformal arcs for each target. Clinical evidence and interest have continued to increase in treatment of larger numbers of metastases. For multiple-isocenter radiosurgery, the amount of treatment time required for radiosurgery delivery is positively correlated with the number of targets and each target’s geometric complexity. Single-isocenter treatment utilizes modulated MLC plans to simultaneously treat all targets at once. This approach renders treatment time largely insensitive to the number of targets. However, prior to 2010, treatment planning software platforms were not mature enough to reliably deliver a modulated single-isocenter plan of comparable quality to multiple-isocenter plans.

However, this has changed. There are now multiple treatment planning platforms on which excellent single-isocenter plans can be generated with equivalent quality to their multiple-isocenter plans. This allows treatment of five, ten, twenty, and even thirty metastases in a single radiosurgery plan, in less than 20 minutes. The efficiency benefits to clinical workflow of this approach are evident. So attractive is the process that multiple treatment planning vendors have begun selling packages that automate the technique.

This chapter details the origins and progress of single-isocenter treatment methodology, as well as the current state of the field. Finally, this chapter provides a step-by-step walkthrough and case example of a single-isocenter, multiple metastasis treatment in order that the reader may implement the technique within their own clinic.

Appendix 17.1

Updated Institutional Systematic Treatment Planning Technique for Single-Isocenter VMAT Radiosurgery at the University of Alabama at Birmingham

(v. 2019)

• 1. Contouring

  • 1.1. Contour all normal structures, including the brain, optic nerves, and brainstem.

  • 1.2. Contour each GTV and label as GTV1, GTV2, GTV3, etc.

  • 1.3. Using Boolean operations, create a GTV_total; for example, for three targets “GTV_total” = “GTV1” OR (“GTV2” OR “GTV3”).

    • 1.3.1. ∗Note: previous versions of this guide erroneously used the Boolean operator AND instead of OR.

  • 1.4. Using Boolean operations, create a structure for healthy brain tissue, i.e., “BRAIN” SUBTRACT “GTV_TOTAL.”

  • 1.5. Create control structures as volumetric shells surrounding the target. In a stepwise fashion, create structures, and use the “margin for structure” function as follows:

    • 1.5.1.1. “inner control” = GTV_total + 0.2 cm.

    • 1.5.1.2. “middle control” = GTV_total + 0.5 cm.

    • 1.5.1.3. “outer control” = GTV_total + 1.5 cm.

    • 1.5.1.4. Shell sizes in original Technique are listed below, but we have found the smaller shells in general superior for all plans.

      • 1.5.1.4.1. “inner control” = GTV_total + 0.5 cm.

      • 1.5.1.4.2. “middle control” = GTV_total + 1.0 cm.

      • 1.5.1.4.3. “outer control” = GTV_total + 3.0 cm.

    • 1.5.2. Boolean operation: “outer control” = “outer control” SUB “middle control.”

    • 1.5.3. Boolean operation: “middle control” = “middle control” SUB “inner control.”

    • 1.5.4. Boolean operation: “inner control” = “inner control” SUB “GTV_total.”

    • 1.5.5. Crop any control structures outside the body using Boolean operations.

  • 1.6. If a plan contains multiple prescriptions, for optimal recipe performance, there should be a separate GTV_total and set of shell structures for each prescription level (e.g., GTV_total_18, Outer18, Middle18, Inner18, etc.).

  • 1.7. If plan contains targets in close proximity, dose “bridging” may occur. This is when the dose cloud at a given isodose level connects two targets. This is only problematic at moderate- to high-isodose levels (e.g., >9 Gy in an 18Gy single-fraction plan).

    • 1.7.1. To mitigate this, create an arbitrary structure in between the two “bridged” targets, and add an optimization criterion that does not permit any of the structure to receive > the dose level of concern less 1 Gy (e.g., if 9Gy level is of concern: upper constraint, 0% of structure receiving 8Gy). If this does not alleviate bridging, increase priority of this structure or lower control dose in constraint.

  • 1.8. For DVH calculation of conformity indices, create evaluation structures using the “margin for structure” function as before:

    $$ {\displaystyle \begin{array}{c}{}^{``}\mathrm{GTV}1\_{\mathrm{eval}}^{"}=\mathrm{GTV}1+1.0\ \mathrm{cm}{,}^{``}\mathrm{GTV}2\_{\mathrm{eval}}^{"}\\ {}=\mathrm{GTV}2+1.0\ \mathrm{cm},\mathrm{etc}.\end{array}} $$

    • 1.8.1. For a single target, this is not necessary; the prescription isodose volume of the BODY structure may be used for conformity calculation, or Eclipse will calculate the RTOG CI automatically in the dose statistics tab.

  • 1.9. For maximum accuracy (and possibly for increased plan quality for plans with very small targets), right click on each target structure and select “high-resolution segment.”

    • 1.9.1. In the External Beam Planning window, select Calculation Models, click Edit next to the appropriate Volume Dose Algorithm, and change the dose grid resolution to 0.1 cm from 0.25 cm. (Note that this will significantly increase the time required for calculation – using Acuros algorithm will mitigate time increase.)

• 2. Isocenter and Field Geometry

  • 2.1. Consider arc geometries as follows (Fig. 17.17) (they are ordered for optimal delivery):

Arc number	Field	Arc length (°)	Table rotation (°)	Collimator angle (°)	Arc direction	Gantry angle (°)	Stop angle (°)
1	1	360	0	45	CW	181	179
2	1	360	0	45	CW	181	179
	2	180	315/45/90a	45	CCW	180	0
3	1	360	0	45	CW	181	179
	2	180	315	45	CCW	180	0
	3	180	45	315	CCW	0	180
4	1	360	0	45	CW	181	179
	2	180	315	45	CCW	179	0
	3	180	45	315	CCW	0	181
	4	180	90	45	CW	181	0

1. ^aChoosing a 315° or 45° instead of a 90° couch kick for this arc forces some of the exit dose to spill out the sides of the head instead of into the body. Please note some Linacs may be configured with default table rotation to be 180° instead of 0°

• 2.2. Place isocenter at the geometric center of GTV_total by right-clicking a field and selecting “Align grouped fields to structure [GTV_total].”

  • 2.2.1. If this places the isocenter too close to a large target such that a small target is not within space covered by leaves, the isocenter can be manually adjusted so that all targets are within the field.

• 3. Plan Optimization and Normalization

  • 3.1. If available, enable jaw tracking.

  • 3.2. Each individual GTV (not the composite) receives the lower objective: 100% of the target to receive 100% of the prescription dose; default priority = 50 (this can be adjusted to 100 according to planner preference, but if done, adjust other priorities in similar proportion).

  • 3.3. What you prioritize most in the plan should get the highest optimization priority; normalization will scale the dose so that adequate target coverage occurs.

  • 3.4. For a plan with no critical OARs and in which low-dose spill to healthy tissue is to be minimized, each control structure receives the following upper objective:

    • 3.4.1. Inner control: 0% of the structure to receive 98% of the prescription dose; priority = 50.

    • 3.4.2. Middle control: 0% of the structure to receive 50% of the prescription dose; priority = 50.

    • 3.4.3. Outer control: 0% of the structure to receive 40% of the prescription dose; priority = 50.

    • 3.4.4. Healthy brain: 1% of the structure to receive one-sixth of the prescription dose; priority = 125 (in our experience, this is typically about the point where the healthy tissue DVH curve’s inflection point should be, which corresponds to the modal dose on the healthy brain’s DDH curve).

      • 3.4.4.1. Note that weighting this priority to such a high value will lead to a lower plan normalization value (and thus a higher degree of normalization). If the value becomes such that you are uncomfortable with the level of post-calculation MU scaling, then reduce the priority of this criterion.

      • 3.4.4.2. Further note that these priority values are surrogates for weighting coefficients in the mathematical expression being optimized. They are proportional.

  • 3.5. Additional dose constraint objectives may be needed if sensitive adjacent normal structures are within the region of the control structures (e.g., brainstem, optic nerves, chiasm, etc.). If not nearby, control structures will adequately serve as constraints on limiting dose to these organs.

    • 3.5.1. For example, if the brainstem overlaps the inner control shell, consider the following additional constraint – brainstem: 0% of the structure to receive 800 cGy; priority = 75. Priority must be ≥ priority of control structure with which organ at risk overlaps or optimization algorithm will not take objective into consideration as desired.

  • 3.6. For multiple target plans with identical prescriptions, low-isodose spill is minimized when prescription isodose coverage is homogeneous across all targets. This is because normalization occurs for the entire plan and not for each target independently, as in a Gamma Knife plan.

    • 3.6.1. On the DVH, homogenous target coverage will appear as all target DVH lines superimposing one another at the prescription isodose point (Fig. 17.18).

      • 3.6.1.1. Note: homogenous target coverage is not to be confused with plan homogeneity (minimizing hot spots within tumor volume).

    • 3.6.2. We have found that plan quality can be increased by freezing the optimization on the first step of the first level and waiting for the optimization to stabilize (indicated by a leveling out of the line for each structure in the optimization line graph). Increase or decrease the priority of each structure’s cost function as needed to achieve homogeneous target coverage. Once reached, unfreeze the optimization from the first level.

      • 3.6.2.1. If the plan finishes and DVH shows heterogeneous coverage, one can reinitiate optimization and “Continue the previous optimization.” The optimization will start frozen in the final level. Cost function priorities can also be adjusted here to achieve the desired homogeneity of coverage.

  • 3.7. Normalize plan for desired coverage. We utilize a normalization that delivers 100% of the prescription to 99–100% of the GTV_total, and ensure that each target has received appropriate coverage.

  • 3.8. For plans with targets of differing prescriptions, ensuring each target receives sufficient coverage but not excessive coverage can sometimes require additional effort.

    • 3.8.1. During the initial optimization, freeze the optimization at the first or second level of the multilevel resolution optimization and pick one target as an anchor.

    • 3.8.2. For each other target, adjust the priority of its optimization constraint such that its DVH curve is covering a dose the same percentage difference from the anchor curve as the percentage difference between the two targets’ respective prescription doses.

      • 3.8.2.1. For example, prescription dose for target A is 18 Gy and is 24 Gy for target B (∆ = 6 Gy or 25%). Target A is designated as the anchor target. If, during the initial phase of the optimization, target A’s DVH curve indicates 100% coverage at 12 Gy, then adjust target B’s optimization constraint priority such that its curve indicates 100% coverage at 12 + 25% or 15Gy. Thus, when the post-optimization plan-wide normalization is performed, each target should have roughly 100% coverage at its respective prescription dose.

Fig. 17.17

One-, two-, three-, and four-arc geometries

Fig. 17.18

Homogeneous (a) and heterogeneous (b) target coverage with 100% prescription dose

• 4. Plan Evaluation

  • 4.1. Calculate the conformity index (CI) of each target by using the evaluation structures.

    • 4.1.1. Ensure that 100% isodose line is not extending outside of evaluation structures (if it does, you do not have a very good plan) or overlapping with another target’s 100% isodose line.

    • 4.1.2. RTOG CI = 100% isodose line of eval structure divided by the target volume; for example: CI for GTV1 = 100% isodose line for “GTV1_eval” divided by the volume of GTV1.

      • 4.1.2.1. CI can also be calculated by the Eclipse treatment planning software, but the current software version will not accurately compute the CI if the plan contains more than one target; in such cases, the individual target 100% isodose volume can be obtained by extracting the prescription volume from each evaluation structure.

  • 4.2. Evaluate the high- to moderate-dose falloff.

    • 4.2.1. Paddick gradient index (GI) or the R50% of the plan

      • 4.2.1.1. GI = volume of 50% isodose line (V_50%) divided by the volume of the 100% prescription isodose line (PIV).

      • 4.2.1.2. R50% = volume of 50% isodose line (V_50%) divided by the total target volume (GTV_total).

      • 4.2.1.3. Note that Eclipse will calculate a gradient measure which is a different definition than the above calculated GI.

      • 4.2.1.4. Also note that the Paddick GI cannot be used to accurately compare the dose falloff between plans with different conformity indices. This is because different conformity indices indicate different 100% prescription isodose volumes.

    • 4.2.2. For comparing the high- to moderate-isodose falloff between two radiosurgery plans of differing conformities, we recommend comparing either:

      • 4.2.2.1. AUC-DVH – the area under the DVH curve in the region of interest (e.g., 9 to 18Gy)

        • 4.2.2.1.1. This can be done easily in by exporting the DVH at a sufficiently fine-dose resolution (e.g., 1cGy) to excel and utilizing the trapezoidal rule for numerical integration.

      • 4.2.2.2. R50% – volume of 50% isodose line (V_50%) divided by the total target volume (GTV_total)