3449 - Does the Heart Know When It Needs Sparing? Cardiac Positional Variability as an Implicit Trigger for Adaptive Plan Selection In Lung Radiotherapy

Purpose/Objective(s):

Improved lung cancer survival has elevated cardiac toxicity as a critical consideration. Adverse cardiac events are associated with cardiac radiation dose, and residual setup errors toward the heart have been linked to poorer survival (HR 1.1 for conventional fractionation; HR 1.26 per mm mediastinal shift for SBRT), both assessed using standard cone-beam CT (CBCT). To our knowledge, no prospective trial has used CT-guided online adaptive RT (ART) to spare both heart and left anterior descending artery (LAD). We hypothesized that (a) cardiac variability on next-generation iterative CBCT would be consistent with published data, and (b) adaptive selection driven by dosimetric scorecards alone would implicitly correlate with heart and LAD proximity to PTV.

Materials/Methods:

Eleven lung cancer patients (5 SBRT, 6 conventional; median age 68, 36% concurrent chemotherapy) were enrolled on a prospective IRB-approved adaptive heart and LAD sparing protocol (16830-34). Treatment used a CT-guided adaptive platform with iterative CBCT providing improved soft-tissue contrast. At each fraction, scheduled and adapted plans were generated; selection used dosimetric scorecard evaluation without geometric criteria. Across 205 total fractions, Euclidean heart-PTV and LAD-PTV centroid distances were computed. Inter-fractional centroid shifts relative to simulation (FX-sim) were quantified. Fractions were grouped by adaptation decision and compared using t-tests and Mann-Whitney U (MW-U) tests.

Results: Observed inter-fractional heart centroid variability (mean shifts 1-4 mm, SD 3 mm) was consistent with published CBCT-based data, confirming the first hypothesis. Across pooled fractions, adapted treatments had significantly shorter heart-PTV and LAD-PTV centroid distances (both p<0.001, Table 1), confirming the second hypothesis. Heart FX-sim shifts differed significantly between groups (p=0.002); LAD FX-sim shifts did not reach significance.

Conclusion:

In this first prospective CT-guided online adaptive lung RT trial targeting heart and LAD sparing, inter-fractional cardiac variability was consistent with prior literature and adaptive plan selection based on dosimetric criteria alone was significantly associated with cardiac proximity to PTV. These findings suggest cardiac positional variability acts as an implicit geometric trigger for adaptation. Correlation with dosimetric sparing of heart and cardiac substructures is under investigation.

Table 1. Pooled fractions: Adapted vs Not Adapted, mean (SD)

Variable	Not Adapted	Adapted	t-test p	MW-U p
Heart-PTV (cm)	10.23 (3.30)	7.29 (1.13)	<0.001	<0.001
LAD-PTV (cm)	13.87 (2.72)	11.14 (2.01)	<0.001	<0.001
FX-sim Heart (cm)	0.14 (0.30)	0.09 (0.29)	0.002	0.004
FX-sim LAD (cm)	-0.01 (0.31)	0.06 (0.19)	0.457	0.372