Department of Thoracic Surgery, Jiangsu Province Hospital, The First Affiliated Hospital of Nanjing Medical University, Nanjing, China.
Cancer Med. 2022 Jul;11(14):2855-2864. doi: 10.1002/cam4.4645. Epub 2022 Mar 14.
This study aims to reveal the serum tumor marker (STM) levels in lung adenocarcinoma (LUAD) histological subtypes and evaluate their values in predicting the solid and micropapillary components (SMC).
We retrospectively analyzed 3100 invasive LUAD patients between January 2017 and December 2020. Associations between preoperative STMs (CEA, CYFRA21-1, CA199, CA724, NSE, AFP) and LUAD subtypes were evaluated. Multivariate regression analyses were used to determine the independent predictors. Predictive models for SMC were constructed and AUC (area under the curve) was calculated.
CEA and CYFRA21-1 levels differed across the LUAD histological subtypes, with the SPA (solid-predominant adenocarcinoma) having the highest level and the LPA (lepidic-predominant adenocarcinoma) harboring the lowest level (p <0.001). Tumors with SMC also had higher CEA and CYFRA21-1 levels than those absence of SMC. Gender, tumor size, CEA, Ki-67, EGFR mutation (solid components only), and tumor differentiation were significantly independently associated with the containing of SMC. Patients were split into two data sets (training set: 2017-2019 and validation set: 2020). The model with gender and tumor size yielded an AUC of 0.723 (training set) and 0.704 (validation set) for the solid component. Combination of CEA, gender, and tumor size led to a significant increase in the predictive accuracy (training set: 0.771, p = 0.009; validation set: 0.747, p = 0.034). The AUC of the model for micropapillary component with only gender and tumor size was 0.699 and 0.711 in the training set and validation set, respectively. Integration of CEA with gender and tumor size significantly improved the predictive performance with an AUC of 0.746 (training set, p = 0.045) and 0.753 (validation set, p <0.001).
Serum CEA and CYFRA21-1 varied considerably according to LUAD histological subtypes. The combination of serum CEA and other factors showed prominent values in predicting the SMC.
本研究旨在揭示肺腺癌（LUAD）组织学亚型的血清肿瘤标志物（STM）水平，并评估其在预测实体和微乳头状成分（SMC）中的价值。
我们回顾性分析了 2017 年 1 月至 2020 年 12 月间的 3100 例浸润性 LUAD 患者。评估了术前 STM（CEA、CYFRA21-1、CA199、CA724、NSE、AFP）与 LUAD 亚型之间的关系。采用多变量回归分析确定独立预测因素。构建 SMC 的预测模型并计算 AUC（曲线下面积）。
CEA 和 CYFRA21-1 水平在 LUAD 组织学亚型中存在差异，其中 SPA（实体为主型腺癌）水平最高，LPA（贴壁为主型腺癌）水平最低（p<0.001）。含有 SMC 的肿瘤的 CEA 和 CYFRA21-1 水平也高于不含有 SMC 的肿瘤。性别、肿瘤大小、CEA、Ki-67、EGFR 突变（仅实体成分）和肿瘤分化与 SMC 的存在显著相关。患者被分为两个数据集（训练集：2017-2019 年；验证集：2020 年）。仅使用性别和肿瘤大小的模型在实体成分的训练集和验证集中的 AUC 分别为 0.723（训练集）和 0.704（验证集）。CEA、性别和肿瘤大小的组合导致预测准确性显著提高（训练集：0.771，p=0.009；验证集：0.747，p=0.034）。仅使用性别和肿瘤大小的模型对微乳头状成分的 AUC 在训练集和验证集中分别为 0.699 和 0.711。CEA 与性别和肿瘤大小的整合显著提高了预测性能，在训练集和验证集中的 AUC 分别为 0.746（p=0.045）和 0.753（p<0.001）。
血清 CEA 和 CYFRA21-1 根据 LUAD 组织学亚型有很大差异。血清 CEA 与其他因素的结合在预测 SMC 方面具有显著价值。
