Kim Jae Sik, Kim Jin Ho, Chang Ji Hyun, Kim Do Wook, Shin Kyung Hwan
Department of Radiation Oncology, Seoul National University College of Medicine, Seoul, South Korea.
Department of Radiation Oncology, Soonchunhyang University Seoul Hospital, Seoul, South Korea.
Front Oncol. 2022 Oct 26;12:1026043. doi: 10.3389/fonc.2022.1026043. eCollection 2022.
We identified novel clinical and dosimetric prognostic factors affecting breast cancer-related lymphedema after postoperative radiotherapy (RT) and developed a multivariable logistic regression model to predict lymphedema in these patients.
In total, 580 patients with unilateral breast cancer were retrospectively reviewed. All patients underwent breast surgery and postoperative RT with or without systemic treatment in 2015. Among the 580 patients, 532 with available RT plan data were randomly divided into training (n=372) and test (n=160) cohorts at a 7:3 ratio to generate and validate the lymphedema prediction models, respectively. An area under the curve (AUC) value was estimated to compare models.
The median follow-up duration was 5.4 years. In total, 104 (17.9%) patients experienced lymphedema with a cumulative incidence as follows: 1 year, 10.5%; 3 years, 16.4%; and 5 years, 17.6%. Multivariate analysis showed that body mass index ≥25 kg/m (hazard ratio [HR] 1.845), dissected lymph nodes ≥7 (HR 1.789), and taxane-base chemotherapy (HR 4.200) were significantly associated with increased lymphedema risk. Conversely, receipt of RT at least 1 month after surgery reduced the risk of lymphedema (HR 0.638). A multivariable logistic regression model using the above factors, as well as the minimum dose of axillary level I and supraclavicular lymph node, was created with an AUC of 0.761 and 0.794 in the training and test cohorts, respectively.
Our study demonstrated that a shorter interval from surgery to RT and other established clinical factors were associated with increased lymphedema risk. By combining these factors with two dosimetric parameters, we propose a multivariable logistic regression model for breast cancer-related lymphedema prediction after RT.
我们确定了影响术后放疗（RT）后乳腺癌相关淋巴水肿的新的临床和剂量学预后因素，并建立了多变量逻辑回归模型来预测这些患者的淋巴水肿。
总共回顾性分析了580例单侧乳腺癌患者。所有患者均在2015年接受了乳房手术及术后放疗，部分患者接受了或未接受全身治疗。在这580例患者中，532例有可用的放疗计划数据，按7:3的比例随机分为训练队列（n = 372）和测试队列（n = 160），分别用于生成和验证淋巴水肿预测模型。通过估计曲线下面积（AUC）值来比较模型。
中位随访时间为5.4年。共有104例（17.9%）患者发生淋巴水肿，累积发病率如下：1年时为10.5%；3年时为16.4%；5年时为17.6%。多因素分析显示，体重指数≥25 kg/m²（风险比[HR] 1.845）、清扫淋巴结≥7个（HR 1.789）和紫杉类化疗（HR 4.200）与淋巴水肿风险增加显著相关。相反，术后至少1个月接受放疗可降低淋巴水肿风险（HR 0.638）。使用上述因素以及腋窝I级和锁骨上淋巴结的最小剂量建立了多变量逻辑回归模型，在训练队列和测试队列中的AUC分别为0.761和0.794。
我们的研究表明，手术至放疗的间隔时间较短以及其他既定临床因素与淋巴水肿风险增加相关。通过将这些因素与两个剂量学参数相结合，我们提出了一个用于预测放疗后乳腺癌相关淋巴水肿的多变量逻辑回归模型。
