ESTRO 2024 - Abstract Book

S2245

Clinical - Upper GI

ESTRO 2024

Three of the 10 patients were resected in the SBRT arm, while none was resected in the CRT arm. The one-year overall survival (OS) in SBRT arm was 80% while it was 45% in CRT arm. CD4:CD8 counts showed a declining trend, however when compared between SBRT and CRT arms, did not show any statistical significance. There was a declining trend of absolute values of neutrophil and platelets in the SBRT arm. But the absolute lymphocyte count showed increasing trend. Lower NLR and PLR values were associated with better OS and local progression free survival. QOL improved in both the arms with treatment, with more improvement seen in the SBRT arm; however, there was no statistical difference. A total of 851 radiomic features were extracted. Using logistic regression statistics, nine features based on low and high pass wavelet filters were found to be significantly associated with survival.

Conclusion:

NLR and PLR can be used as biomarkers for survival in patients of pancreatic cancer. CT based radiomic features can be used for prognostication and predicting survival in pancreatic cancer patients. SBRT offered better resectability, OS and QOL, however the results need to be validated in a larger patient population.

Keywords: biomarkers, pancreatic cancer, radiomics, SBRT

2329

Digital Poster

A survival prediction model by radiomics in elderly esophageal cancer patients after radiotherapy

Ikuno Nishibuchi 1 , Daisuke Kawahara 1 , Kantatsu Tanaka 2 , Riku Nishioka 1 , Takashi Sadatoki 1 , Tsuyoshi Katsuta 1 , Nobuki Imano 1 , Junichi Hirokawa 1 , Yuji Murakami 1 1 Hiroshima University, Radiation Oncology, Hiroshima, Japan. 2 Hiroshima University, School of Medicine, Hiroshima, Japan

Purpose/Objective:

In the treatment of esophageal cancer (EC), definitive radiotherapy (RT) is less invasive than esophagectomy and plays a crucial role as a treatment option, particularly for elderly patients. Elderly cancer patients have diverse backgrounds, including age-related declines in organ function, physical capabilities, cognitive function, and the presence of various comorbidities. Consequently, multiple factors impact their prognosis. In recent years, sarcopenia has garnered attention as a prognostic factor in cancer patients. Furthermore, late pulmonary toxicity and decreased respiratory function are concerns in RT for EC. Radiomics analysis involves the extraction of an extensive range of features from images, with the expectation of its application for prognosis prediction. While radiomics analysis has traditionally focused on the tumor region, there is the potential for improved prognostic accuracy if it can capture features related to skeletal muscles and lung function from images. This study aimed to construct an artificial intelligence (AI) -based survival prediction model that incorporates lung function and skeletal muscle quality in addition to tumor factors for elderly EC patients treated by definitive RT.