基于剂量组学预测肺癌患者放射性肺炎发生的研究

目的 探讨剂量组学在预测肺癌根治性放疗患者放射性肺炎发生中的应用潜能。方法 回顾性收集行根治性放疗的314例肺癌患者的临床资料、放疗剂量文件、定位及随访CT图像,根据临床资料及影像学随访资料对放射性肺炎进行分级,提取全肺的剂量组学特征,构建机器学习模型。应用1000次自助抽样法（bootstrap）的最小绝对值收敛和选择算子嵌套逻辑回归（LASSO‐LR）及1000次bootstrap的赤池信息量准则（AIC）向后法筛选与放射性肺炎相关的剂量组学特征,随机按照7∶3划分为训练集及验证集,应用逻辑回归建立预测模型,并应用ROC曲线及校正曲线评价模型的性能。结果 共提取120个剂量组学特征,经LASSO‐LR降维筛选得到12个特征进入“特征池”,再经过AIC向后法筛选,最终筛选出6个剂量组学特征进行模型构建,训练集AUC为0.77（95%CI为0.65~0.87）,独立验证集AUC为0.72（95%CI为0.64~0.81）。结论 利用剂量组学建立的预测模型具有预测放射性肺炎发生的潜力,但仍需继续纳入多中心数据及前瞻性数据进一步挖掘剂量组学的应用潜能。     

Patient-Reported Quality of Life Before and After Chemoradiation for Intact Pancreas Cancer: A Prospective Registry Study

PurposeOur purpose was to determine the effect of chemoradiotherapy (CRT) on patient-reported quality of life (QOL) for patients with intact pancreas cancer.Methods and MaterialsWe reviewed a prospective QOL registry for patients with intact, clinically localized pancreatic ductal adenocarcinoma treated with CRT between June 2015 and November 2018. QOL was assessed pre-CRT (immediately before CRT, after neoadjuvant chemotherapy) and at the completion of CRT with the Functional Assessment of Cancer Therapy-Hepatobiliary (FACT-Hep) and its component parts: FACT-General (FACT-G) and hepatobiliary cancer subscore (HCS). A minimally important difference from pre-CRT was defined as ≥ 6, 5, and 8 points for FACT-G, HCS, and FACT-Hep, respectively.ResultsOf 157 patients who underwent CRT, 100 completed both pre- and post-CRT surveys and were included in the primary analysis. Median age at diagnosis was 65 years (range, 23-90). National Comprehensive Cancer Network resectability status was resectable (3%), borderline resectable (40%), or locally advanced (57%). Folinic acid, 5-fluorouracil, irinotecan, and oxaliplatin (FOLFIRINOX) (75%) or gemcitabine and nab-paclitaxel (42%) were given for a median of 6 cycles (range, 0-42) before CRT. Radiation therapy techniques included 3-dimensional conformal (22%), intensity modulated photon (55%), and intensity modulated proton (23%) radiation therapy to a median dose of 50 Gy (range, 36-62.5). Concurrent chemotherapy was most commonly capecitabine (82%). Sixty-three patients (63%) had surgery after CRT. The mean decline in FACT-G, HCS subscale, and FACT-Hep from pre- to post-CRT was 3.5 (standard deviation [SD], 13.7), 1.7 (SD 7.8), and 5.2 (SD 19.4), respectively. Each of these changes were statistically significant, but did not meet the minimally important difference threshold. Pancreatic head tumor location was associated with decline in FACT-Hep. Nausea was the toxicity with the greatest increase from pre- to post-CRT by both physician-assessment and patient-reported QOL.ConclusionsFor patients with intact pancreatic adenocarcinoma, modern CRT is well tolerated with minimal decline in QOL during treatment.