Decreasing NTCP Toxicity in Lung Cancer RT by Reducing PTV Margins Using a Combined Deep Learning Auto-Segmentation and Automated Treatment Planning Approach
M Thor*, H Veeraraghavan, J Jiang, M Zarepisheh, A Rimner, J Deasy, Memorial Sloan Kettering Cancer Center, New York, NY
Presentations
PO-GePV-M-242 (Sunday, 7/25/2021) [Eastern Time (GMT-4)]
Purpose: Large planning target volumes (PTVs) in locally-advanced non-small cell lung cancer (LA-NSCLC) frequently result in high doses to nearby normal tissues that can consequently lead to severe complications. We reduced current PTV margins and studied the impact on normal tissue complication probabilities (NTCPs). Tumors and considered normal tissues were derived from deep learning (DL) auto-segmentation and treatment was simulated using automated planning with the ambition of using an overall unbiased approach.
Methods: The cord, esophagus, heart, lungs and tumors (GTV) of 8 randomly selected patients with LA-NSCLC previously treated with definitive concurrent chemo-radiation to 60 Gy in 30 fractions were segmented using a multiple residual network DL auto-segmentation algorithm. The CTV was created via an added 7mm isotropic margin. Another 2mm and 5mm isotropic margin were used to generate a study-specific small PTV (PTVS) and a clinical practice large PTV (PTVL), respectively. One PTVS plan and one PTVL plan was generated for all patients using automated hierarchical optimization applying all currently used dose-volume criteria. Plans were compared by means of NTCP using published models (Esophagus: Esophagitis; Lung: Pneumonitis) and by clinically relevant dose metrics: max dose (cord), mean dose (heart) and D95% (PTV).
Results: For all patients, PTV coverage was similar in the PTVS and in the PTVL plans (D95%=60Gy in all plans) while dose to all normal tissues was systematically lower in the PTVs plans: The risk of esophagitis was on average reduced from 60% to 57%, the risk of pneumonitis from 8% to 6%, cord max dose from 24Gy to 17Gy and heart mean dose from 9Gy to 8Gy.
Conclusion: This modestly reduced PTV margin study has demonstrated considerable dose sparing for the key normal tissues in lung cancer RT by means of an unbiased automated segmentation and automated planning approach.
Funding Support, Disclosures, and Conflict of Interest: This work was supported by the MSK Cancer Center support grant/core grantP30 CA008748 and NCI 5 R01 CA198121-04
ePosters
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email