3575 - Estimating RT-Attributable Risk of Pulmonary Hospitalization in NSCLC Using Dosimetric and Personalized Clinical Predictors
Presenter(s)
J. D. Nieto1, W. Wang2, M. Schipper2, D. P. Bergsma1, H. Yin3, M. M. Dominello4, M. Zaki5, A. F. Dragovic6, L. L. Kestin7, P. A. Paximadis8, M. M. Matuszak1, S. R. Miller II9, and S. Jolly3; 1Department of Radiation Oncology, University of Michigan, Ann Arbor, MI, 2University of Michigan, Ann Arbor, MI, 3Michigan Radiation Oncology Quality Consortium Coordinating Center, Ann Arbor, MI, 4Department of Radiation Oncology, Karmanos Cancer Center, Detroit, MI, 5Covenant HealthCare, Saginaw, MI, 6Department of Radiation Oncology, Brighton Center for Specialty Care, Brighton, MI, 7Michigan Healthcare Professionals/GenesisCare USA, Farmington Hills, MI, 8Department of Radiation Oncology, Corewell Health South, St. Joseph, MI, 9Wayne State University, Detroit, MI
Purpose/Objective(s):
Definitive radiation therapy (RT) for non-small-cell lung cancer (NSCLC) can cause morbidity. We investigated clinical and dose metrics that may predict pneumonitis and other lung-related hospitalizations (LRH).Materials/Methods:
Stage I-III NSCLC (2018-2024) patients treated with definitive RT ± concurrent chemotherapy were included from a prospective statewide quality consortium database. Cumulative incidence of pneumonitis-, COPD-, pneumonia-related and overall LRH was estimated at 1-, 2-, and 3-years post-RT. LASSO Fine-Gray models identified clinical predictors of any LRH. Lung dose metrics (mean lung dose, V5–V60) were added to build multivariable models (MVM). Sensitivity to stereotactic body-RT (SBRT) was assessed by refitting MVMs excluding SBRT patients. RT-attributable hospitalization risk was calculated by subtracting the MVM estimated risk when the dose metric was set to 0.Results:
Among 2,084 patients, 381 experienced =1 LRH. Cumulative incidence of any LRH 1-, 2-, and 3-years post-RT was 15%, 23%, and 30%, respectively. Two-year cumulative incidence of pneumonitis-, pneumonia-. and COPD-related LRHs was 39 (1.9%), 170 (8.2%) and 140 (6.7%), respectively. In 1,834 patients with complete data, lung-GTV/IGTV/ITV V40% (V40) was the most significant dosimetric predictor for any LRH when added to significant clinical factors. Median V40 was 5.39%, 2.14%, and 7.91% for all, SBRT, and non-SBRT fractionation, respectively. The final MVM for any LRH included age, supplemental oxygen (O2) use, ECOG G2+, COPD, and V40. The overall estimated 2-year risk increased from 21% at V40=0%, to 25% at V40=10%. Results were qualitatively unchanged after excluding SBRT. Table 1 shows clinical and dosimetric predictors of any LRH and by subtype. Across patients, the median 2-year RT attributable risk (AR) comparing V40 = 10% versus 0% was 6% (10th, 90th percentiles: 4%, 9%). AR values were higher in patients at higher baseline risk: at V40=10%, AR for any LRH was 11% for a 72-year-old ECOG 2 patient with COPD on O2 versus 4% for a 72-year-old patient ECOG 0/1 without COPD and not on O2.Conclusion:
Lung dose-volume metrics, particularly V40, independently inform LRH risk after RT for NSCLC patients at risk of hospitalization. Integrating personalized clinical and dosimetric predictors enhances toxicity prediction, guides RT planning, and supports risk-adaptive optimization. Table 1:| Type of Lung Related Hospitalization | ||||||||
| Any | Pneumonitis | Pneumonia | COPD | |||||
| HR (95% CI) | P-Value | HR (95% CI) | P-Value | HR (95% CI) | P-Value | HR (95% CI) | P-Value | |
| Age | 1.01(0.99, 1.02) | 0.078 | ||||||
| COPD (Y) | 1.58(1.24, 2.02) | <.001 | 1.36(0.96, 1.93) | 0.082 | 2.64(1.65, 4.25) | <.001 | ||
| O2 (Y) | 1.89(1.47, 2.44) | <.001 | 1.52(1.02, 2.27) | 0.041 | 2.03(1.37, 3.01) | <.001 | ||
| ECOG (2+) | 1.78(1.36, 2.34) | <.001 | 1.32(0.85, 2.03) | 0.220 | 1.55(1.01, 2.38) | 0.047 | ||
| V40 | 1.03(1.00, 1.06) | 0.039 | 1.08(1.02, 1.15) | 0.024 | 1.04(1.00, 1.08) | 0.069 | ||