Functional avoidance‐based intensity modulated proton therapy with 4DCT derived ventilation imaging for lung cancer

Abstract The primary objective is to evaluate the potential dosimetric gains of performing functional avoidance‐based proton treatment planning using 4DCT derived ventilation imaging. 4DCT data of 31 patients from a prospective functional avoidance clinical trial were evaluated with intensity modulated proton therapy (IMPT) plans and compared with clinical volumetric modulated arc therapy (VMAT) plans. Dosimetric parameters were compared between standard and functional plans with IMPT and VMAT with one‐way analysis of variance and post hoc paired student t‐test. Normal Tissue Complication Probability (NTCP) models were employed to estimate the risk of two toxicity endpoints for healthy lung tissues. Dose degradation due to proton motion interplay effect was evaluated. Functional IMPT plans led to significant dose reduction to functional lung structures when compared with functional VMAT without significant dose increase to Organ at Risk (OAR) structures. When interplay effect is considered, no significant dose degradation was observed for the OARs or the clinical target volume (CTV) volumes for functional IMPT. Using fV20 as the dose metric and Grade 2+ pneumonitis as toxicity endpoint, there is a mean 5.7% reduction in Grade 2+ RP with the functional IMPT and as high as 26% in reduction for individual patient when compared to the standard IMPT planning. Functional IMPT was able to spare healthy lung tissue to avoid excess dose to normal structures while maintaining satisfying target coverage. NTCP calculation also shows that the risk of pulmonary complications can be further reduced with functional based IMPT.

A | Functional avoidance radiation therapy for definitive lung cancer The regional function heterogeneity of lung tissues has long been exploited in the design of external beam radiation therapy treatment planning using single photon emission tomography (SPECT) with radiopharmaceutical aerosols. [1][2][3] Positron emission tomography (PET) paired with a positron emitting inert gas is another reported ventilation imaging method. 4,5 Hyperpolarized xenon or helium magnetic resonance imaging (MRI) techniques have also been reported to generate pulmonary ventilation information. [6][7][8] With the advent of 4DCT-based pulmonary ventilation calculation, the changes in the local air content are converted into pulmonary functional images to guide the placing and optimization of radiation treatment beams. [9][10][11][12][13][14] Currently, there are multiple National Institutes of Health (NIH) funded clinical trials (NCT02528942, NCT02308709, and NCT02843568) that aim to investigate the efficacy of the functional avoidance-based treatment planning techniques.

1.B | Proton beam therapy for lung cancer
The ability of proton beams to stop in tissue within a finite range provides its dosimetric superiority over conventional external beam photon therapy. However, complexity in proton treatment planning increases when considering uncertainties in proton range, patient daily setup, physiological changes of the tumor, CT density conversion, and tumor motion. In a recent large-cohort retrospective comparison study, the National Cancer Database was queried to analyze the outcomes and predictors associated with proton therapy for stage I-IV non-small cell lung cancer (NSCLC) patients; better 5-year overall survival was reported for proton radiation therapy 15 ; better overall survival rates and tolerable toxicities were also reported for stage II-III NSCLC when using proton therapy with concurrent chemotherapy. [16][17][18] More clinical trials are available to investigate the effect of proton beam therapy on local control rates, toxicity, and survival improvement for patients with early or advanced stage lung cancer (NCT00875901, NCT01629498, NCT01993810, NCT03132532).
With more proton centers employing scanning pencil beam methods for treatment delivery and plan optimization, the interplay effect of tumor motion and proton spot arrangement has gone through vigorous numerical simulations, [19][20][21][22][23] measurement validations, 24,25 and methods to manage it have been clinically implemented. [26][27][28][29] The layer or volumetric dose repainting techniques are the most investigated and applied methodologies in mitigating dose degradation caused by the motion interplay effect due to relatively easy clinical implementation within commercial treatment planning systems.
The purpose of this study is to investigate the potential dosimetric gains and toxicity reduction for the 4DCT-based functional avoidance treatment planning technique using robustly optimized intensity modulated proton therapy (IMPT). Clinical data from a phase II photon functional avoidance study were exported to perform an in-silico comparison with functional based proton planning using a structuralbased approach. We report the dosimetric comparison and estimate the reduction in radiation pneumonitis toxicity using a prior published Nor-

2.B | 4DCT-based ventilation calculation
The phase-resolved pre-treatment 4DCT simulation data from each patient was used to generate the pulmonary ventilation. Lung parenchyma was first segmented from the trachea, main bronchi, pulmonary vasculature, and the gross tumor volume on both the peak inhale and peak exhale phase data using intensity-based segmentation algorithm. This segmented lung tissue mask determined the spatial domain in the inhale and exhale CT data where pulmonary ventilation was quantified. Based on the method proposed by Simon, 10 the fractional changes in air content due to pulmonary ventilation in a specific CT voxel was represented by where V in and V ex are the volumes of air within the inhale and exhale CT voxel pair respectively, and HU in and HU ex are the Houns-

2.F | Dosimetric comparison and statistical analysis
A number of the CTV volume and organ at risk (OAR) dose metrics were evaluated for the comparison of the standard and functional VMAT and IMPT plans (see Table 2). For comparison reasons, the standard and functional IMPT plans, and the standard and functional VMAT plans, were all normalized to D99 of the CTV receiving 99% of the prescription. Comparison of the means of the dosimetric parameters was performed with the one-way analysis of variance (ANOVA) with 0.05 selected as the alpha level. Post-hoc paired student t-tests with multiple testing correction were used to determine which groups demonstrated a difference in means.    Gy, the fV10 from 28.78% (11.6%) to 25.4% (10.3%), and the fV20 from 17.59% (8.4%) to 13.48% (6.6%).

3.B | Interplay effect analysis on functional IMPT
plans Figure 3 shows the DVH metric differences between the nominal Monte Carlo plans and the 4D dynamic plans. No significant dose difference was observed for the D 5% -D 95% and D 95% of the CTV volumes, mean lung dose, mean esophagus dose, max cord dose, V20% and V30% of the functional lung, as well as the mean functional lung dose. As a general observation, as the motion amplitude increases, there is increasing differences in the functional V20% and V30% of the lung, as well as the functional mean lung dose. The maximum observed difference in CTV D 5% -D 95% metric between the nominal plan and the 4D dynamic plan was 2.2 Gy. No significant dose degradation was observed for the CTV D95%. Table 3 shows the estimated mean NTCP for Grade 2+ RP on three of the functional dose metrics and the calculated absolute toxicity reduction between functional and standard plans. There was statistically significant difference between the standard and the functional plans. An absolute reduction of 1.8%, 3.3%, and 5.7% was observed for fMLD, fV30, and fV20 dose metrics. For a certain patient, as high as 26% in absolute reduction was observed in Grade 2+ RP when planning with functional IMPT. We also report the Grade 3+ RP NTCP calculation results in Table A1 as an additional reference.

| DISCUSSION
The results presented in this study support the feasibility and the potential dosimetric gains of 4DCT-derived functional avoidance  Table B1.
Recently, a consensus statement has been published by the Particle Therapy Co-operative Group (PTCOG) Thoracic Subcommittee, 42

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