基于倾向得分匹配肺粘液腺癌的临床、影像及病理的相关分析

Clinical, Imaging and Pathological Analysis of Pulmonary Mucinous Adenocarcinoma Based on Propensity Score Matching

摘要: 目的：本研究比较了肺粘液腺癌与非粘液腺癌的临床、影像学及病理特征，同时探讨了肺粘液腺癌的预测价值。方法：本研究收集了自2017年3月至2023年3月期间在青岛大学附属医院胸外科接受手术治疗且临床分期为I~II期的患者976名(术后病理为肺粘液腺癌患者292名，非粘液腺癌患者684名)，回顾性地分析了所有患者的临床基线资料、影像学特征资料、组织病理学特征资料等，并进行倾向得分匹配(propensity score matching, PSM)，通过二元logistic分析进一步探索鉴别肺粘液腺癌与非粘液腺癌的影响因素，并绘制出受试者工作特征(receiver operating characteristic curve, ROC)曲线评价鉴别的性能。所有统计学分析均使用SPSS软件(版本：25.0)进行，所有统计检验均为双侧检验。P < 0.05被认为存在统计学差异。结果：单因素分析显示，一般资料方面，两组在年龄上(P = 0.042)和吸烟史上(P = 0.045)有统计学差异。而两组在年龄方面没有明显的统计学差异(P = 0.482)，术前血液检测中淋巴细胞水平(P < 0.001)与术前NLR (P < 0.001)两组间存在显著差异。影像学资料上提示两组之间在肿瘤位置(P < 0.001)、影像学分类(P < 0.001)、影像学肿瘤最大径(P < 0.001)、肿瘤实性成分最大径(P < 0.001)、CTR (P < 0.001)、支气管充气征(P < 0.001)、空泡征(P < 0.001)、胸膜牵拉征(P < 0.001)上有着显著差异。相较于非粘液腺癌发生于上叶，粘液腺癌组多发生于肺下叶，且更多以实性结节为主；更多的患者出现支气管充气征、空洞征，而胸膜牵拉征在非粘液腺癌患者的影像学图像上更易发现。病理学相关资料结果提示两组患者在是否存在实体、乳头成分(P = 0.90)，是否侵犯脏层胸膜(P = 0.696)，解剖部位(P = 0.125)均无显著统计学差异。除此之外病理T分期(P < 0.001)、病理N分期(P < 0.001)、病理TNM分期(P < 0.001)、手术清扫淋巴结的个数(P < 0.001)、术后病理显示阳性淋巴结的组数(P = 0.008)、阳性淋巴结的总个数(P = 0.009)均有显著统计学差异。PSM后两组均有292名患者纳入研究，两组患者在年龄(P = 0.885)、性别(P = 0.665)、吸烟人数(P = 0.756)等基本资料上均无统计学差异，匹配性能良好。匹配后两组间术前血清CEA (P = 0.014)出现了差异，两组术后病理显示阳性淋巴结的组数(P = 0.114)、阳性淋巴结的总个数(P = 0.096)无明显差异，其余结果较匹配前相似。多因素分析显示实性结节(P = 0.018)、影像学肿瘤最大径(P = 0.004)、CTR (P < 0.001)、胸膜牵拉征(P < 0.001)、磨玻璃结节(P = 0.001)、支气管充气征(P < 0.001)、空洞征(P < 0.001)、Ki-67 (P < 0.001)均是鉴别两者的影响因素。ROC曲线结果表示在区分两组患者方面具有较好的性能，曲线下面积(AUC)为0.86。结论：再去除两组在年龄、吸烟史等混杂因素影响后，与非粘液腺癌相比，肺粘液腺癌更多存在影像学特征且以实性成分为主，影像学最大径更大，CTR更高甚至为1。

Abstract: Objective: This study compared the clinical, imaging and pathological characteristics of pulmonary mucinous adenocarcinoma and non-mucinous adenocarcinoma, and also explored the predictive value of pulmonary mucinous adenocarcinoma. Methods: This study collected 976 patients who underwent surgical treatment in the Thoracic Surgery Department of Qingdao University Affiliated Hospital from March 2017 to March 2023 and were clinically staged I-II (292 patients whose postoperative pathology was pulmonary mucinous adenocarcinoma and 684 patients whose postoperative pathology was pulmonary non-mucinous adenocarcinoma). The clinical baseline data, imaging characteristics data, histopathological characteristics data of all patients were retrospectively analyzed, and propensity score matching (PSM) was performed. The analysis further explored the factors influencing the differentiation of pulmonary mucinous adenocarcinoma and pulmonary non-mucinous adenocarcinoma, and drew a receiver operating characteristic (ROC) curve to evaluate the performance of identification. All statistical analyses were performed using SPSS software (version: 25.0), and all statistical tests were two-sided. P < 0.05 was considered to be statistically different. Results: Univariate analysis showed that in terms of general information, there were statistical differences between the two groups in age (P = 0.042) and smoking history (P = 0.045). There was no significant statistical difference in gender between the two groups (P = 0.482). There were significant differences in lymphocyte levels in preoperative blood testing (P < 0.001) and preoperative NLR (P < 0.001). Imaging data indicated that there are differences between the two groups in terms of tumor location (P < 0.001), imaging classification (P < 0.001), imaging tumor maximum diameter (P < 0.001), tumor solid component maximum diameter (P < 0.001). There were significant differences in CTR (P < 0.001), bronchial air sign (P < 0.001), vacuole sign (P < 0.001), and pleural traction sign (P < 0.001). Compared with pulmonary non-mucinous adenocarcinoma group that occurs in the upper lobe, pulmonary mucinous adenocarcinoma group mostly occurs in the lower lobe of the lung. Besides, more patients had more solid nodules and air bronchus signs, cavitation signs. However, pleural traction signs were more commonly seen on imaging images of patients with pulmonary non-mucinous adenocarcinoma. Pathology-related data showed that there were no significant statistical differences between the two groups of patients in the presence of solid or papillary components (P = 0.90), invasion of the visceral pleura (P = 0.696), and anatomical locations (P = 0.125). In addition, there were significant differences in pathological T stage (P < 0.001), pathological N stage (P < 0.001), pathological TNM stage (P < 0.001), the number of lymph nodes dissected by operation (P < 0.001), the number of positive lymph nodes after operation (P = 0.008) and the total number of positive lymph nodes (P = 0.009). After PSM, 292 patients were included in the study in both groups. There were no statistical differences between the two groups in basic information, such as age (P = 0.885), gender (P = 0.665), and number of smokers (P = 0.756). The matching performance was good. After matching, there was a difference in preoperative serum CEA (P = 0.014) between the two groups. There was no significant difference in the number of positive lymph nodes (P = 0.114) and the total number of positive lymph nodes (P = 0.096) in the postoperative pathology between the two groups. The remaining results were similar than before matching. Multivariate analysis showed solid nodules (P = 0.018), maximum diameter of imaging tumors (P = 0.004), CTR (P < 0.001), pleural traction sign (P < 0.001), ground-glass nodules (P = 0.001), air bronchus sign (P < 0.001), cavitation sign (P < 0.001), and Ki-67 (P < 0.001) were all influencing factors in identifying the two. The ROC curve results indicated good performance in distinguishing the two groups of patients, with an area under the curve (AUC) of 0.86. Conclusion: After removing the influence of age, smoking history and other confounding factors in both groups, compared with pulmonary non-mucinous adenocarcinoma, pulmonary mucinous adenocarcinoma had more imaging features and was mainly solid components, with a larger maximum imaging diameter. The CTR is higher or even 1.