第一部分

背景：肺肉瘤样癌是一种罕见的肺恶性肿瘤，其恶性程度高；临床多见的为病案个案报道或者是数据库分析，国内单中心的大规模回顾性研究较少。本文通过回顾性分析于我院就诊的肺肉瘤样癌患者的临床资料及生存情况来归纳总结肺肉瘤样癌患者的临床特点和预后影响因素。

方法：本文回顾性分析了在2006年1月1日-2015年12月31日期间于中国医学科学院肿瘤医院接受治疗的119名肺肉瘤样癌患者；通过医院信息系统的病历资料采集获得患者信息包括：性别、年龄、身体质量指数、症状、恶性肿瘤家族史、吸烟史、肿瘤大小、肿瘤部位、肿瘤最大径、肿瘤TNM分期、淋巴结状态、病理组织学类型、手术方式、术后化疗与否等信息；电话询问患者的生存期及存活状态。对变量进行单因素、多因素分析，使用Kaplan-Meier计算出上述因素对患者的1年、3年、5年及总生存期等影响。

结果：肺肉瘤样癌好发于有吸烟史的男性（94，79%)患者，中位年龄为60.67±10.50岁（26-89岁）。患者并没有特异性的临床症状，绝对大数的患者表现为呼吸道症状（n=82/119)。肺肉瘤样癌较难通过活检穿刺或术中冰冻病理得以明确，多形性癌是常见的病理类型。肺肉瘤样癌多出现在双肺上叶，绝大多数患者接受了包含有手术在内的治疗，肺叶切除术及袖式肺叶切除术是主要的手术方式。对于中晚期患者在术后是否需要接受辅助化疗，本研究显示术后辅助化疗并没有对患者的远期预后带来明显改善。研究中，肺肉瘤样癌患者的中位OS为17.47个月（Q1–Q3:7.93–62.53个月），1年、3年、5年生存率分别为61.3%、34.5%和 31.9%。通过生存曲线分析可以看出TNM分期、肿瘤的T分期、N分期和M分期对患者的总生存期、1年、3年、5年生存期均有显著影响（p<0.05）。有淋巴结转移较无淋巴结转移预后差，其中多站淋巴结转移较单站淋巴结转移的预后更差，淋巴结转移数量、百分比越多预后越差（p<0.05）。其他的临床特征、肿瘤特性及治疗方式并没有明显影响患者的预后（p>0.05）。

结论：肺肉瘤样癌患者整体预后较差，肿瘤的T分期、N分期、M分期都是影响患者预后的主要因素；目前对于肺肉瘤样癌的临床和基础研究均较少，因而造成临床医生对于肺肉瘤样癌的认识十分不足；本文通过单中心大样本病例的回归性分析，得出了国内PSC患者的临床特征和预后影响因素，可以起到一定的临床指导作用。

第二部分

研究背景

肺肉瘤样癌（Pulmonary Sarcomatoid Carcinoma，PSC）是一种罕见的肺恶性肿瘤。因其罕见，大多数的研究都是个案报道或者是数据库文章，较少有大规模的单中心研究来分析其临床特征及预后影响因素。对于恶性肿瘤的预后分析及预测有多种方式，包括使用公式、网页、列线图、评分系统及颜色打分卡等，其中列线图是一种比较直观且方便的预测评价方式，也是较常使用于临床预测模型展示的一种方式。本项研究是以中国医学科学院肿瘤医院（以下称本中心）就诊的PSC患者为基础来构建肺肉瘤样癌患者的预后预测模型。

研究目的

本研究通过分析本中心曾就诊过的肺肉瘤样癌患者的临床特征及预后情况，来构建预后预测模型，并使用列线图进行展示。

研究方法

    回顾性分析了2006年1月1日至2015年12月31日就诊于本中心的119名肺肉瘤样癌患者的临床资料，根据可重复抽样的方式获得训练队列，并使用原样本作为验证队列。使用训练队列进行PSC患者预后预测模型的构建，使用验证队列进行预测模型的验证。使用逐步回归法（step AIC）、最小绝对收缩和选择算法（Lasso）来筛选变量，并根据训练队列的单因素分析和多因素分析，选择有意义的变量来进行预后预测模型的构建。使用验证队列对模型进行检验，随后使用校准曲线、模型决策曲线分析等来对预测模型进行评价。

研究结果

    筛选出来对预后有影响因素的变量包括身体质量指数（Body Mass Index，BMI）、肿瘤临床分期（tumor-lymph nodes-metastasis,TNM）、肿瘤T分期等三个变量。使用校准曲线和决策曲线对构建出来的预测模型进行检验，显示出了模型与验证队列中的一致性，也显示除了在验证队列中的可行性。

研究结论

    本文在基于我中心接诊患者的数据构建的关于肺肉瘤样癌患者预后预测模型，也是首次基于非欧美人种数据库构建的关于PSC预后的预测模型。通过构建的模型我们可以看出对PSC患者生存期有明显影响的是BMI指数、肿瘤TNM分期和肿瘤的T分期。该模型通过内部验证验证了其预测能力，该模型可以作为一个关于PSC患者良好预后预测的模型应用于临床。

Part one 

Background: Pulmonary sarcomatoid carcinoma is a rare malignant tumor of the lung with a high degree of malignancy. The most common clinical researches are case reports or database analysis, and there are few large-scale retrospective studies of single centers in China. By retrospective analysis of the clinical data and survival of patients with pulmonary sarcomatoid carcinoma in our hospital, the clinical characteristics and prognostic factors of patients with pulmonary sarcomatoid carcinoma were summarized.

Methods: 119 patients with sarcomatoid carcinoma of lung who received treatment in Cancer Hospital of Chinese Academy of Medical Sciences from January 1, 2006 to December 31, 2015 were retrospectively analyzed. Patient information was obtained through medical records collected from the hospital information system, including: gender, age, body mass index, symptoms, family history of malignant tumor, smoking history, tumor size, tumor site, tumor maximum diameter, tumor TNM stage, lymph node status, histopathological type, surgical method, postoperative chemotherapy or not, etc. Patients were asked about their survival and survival status by telephone. Univariate and multivariate analyses were performed on the variables, and Kaplan-Meier was used to calculate the effects of the above factors on 1-year, 3-year, 5-year and overall survival of the patients.

Results: Pulmonary Sarcomatoid carcinoma was most common in male patients (94,79 %) with a history of smoking. The median age was 60.67 ± 10.50 years old(26-89 years old). Patients did not have specific clinical symptoms and the absolute majority of patients presented with respiratory symptoms (n=82/119). Pulmonary sarcomatoid carcinoma is difficult to be identified by biopsy puncture or intraoperative freezing pathology. Pleomorphic carcinoma is the common pathological type. Pulmonary Sarcomatoid carcinoma mostly presents in the upper lobes of both lungs, and the majority of patients undergo surgical treatment, with lobectomy and sleeve lobectomy being the main surgical procedures. As for whether patients with advanced stage need to receive adjuvant chemotherapy after surgery, this study showed that postoperative adjuvant chemotherapy did not significantly improve the long-term prognosis of patients. In the study, the median overall survival time of patients with pulmonary sarcomatoid carcinoma was 17.47 months (7.93–62.53 months), with 1-, 3-, and 5-year survival rates of 61.3%, 34.5%, and 31.9%, respectively. The survival curve analysis showed that TNM staging, T-staging, N-staging and M-staging of the tumor had a significant effect on the overall survival, 1-year, 3-year and 5-year survival of the patients (p<0.05). The prognosis with lymph node metastasis was worse than that without lymph node metastasis, where the prognosis was worse with multiple lymph node metastasis than with single lymph node metastasis, and the more the number and percentage of lymph node metastasis, the worse the prognosis (p<0.05). Other clinical features, tumor characteristics and treatment did not significantly affect the prognosis of patients (p>0.05).

Conclusion: The overall prognosis of patients with pulmonary sarcomatoid carcinoma was poor, and the T-stage, N-stage, and M-stage of the tumor were the main factors affecting the prognosis of patients. At present, there are few clinical and basic studies on PSC, which leads to the lack of understanding of clinicians about PSC. In this paper, the clinical characteristics and prognostic factors of domestic PSC patients were obtained through regression analysis of a large sample of single center cases, which can play a certain clinical guiding role.

part two

Background

Pulmonary Sarcomatoid Carcinoma (PSC) is a rare malignant tumor of lung. Due to its rarity, most studies are case reports or database articles, and there are fewer large-scale single-center studies to analyze its clinical features and prognostic influences. There are various ways to analyze and predict the prognosis of malignant tumors, including the use of formulas, web pages, nomogram, scoring systems and color scoring cards, among which nomogram is a more intuitive and convenient way of prediction and evaluation, and is also a way commonly used in clinical prediction model display. This study was based on PSC patients admitted to Chinese Academy of Medical Sciences, Cancer Hospital (hereinafter referred to as our institution) to construct a prognostic model for patients with pulmonary sarcomatoid carcinoma.

Purpose

This study was conducted to construct a prognostic prediction model by analyzing the clinical characteristics and prognosis of patients with pulmonary sarcomatoid carcinoma who had been seen at our institution, and to present it using nomogram.

Method

The clinical data of 119 patients with pulmonary sarcomatoid carcinoma admitted to our institution from January 1, 2006 to December 31, 2015 were retrospectively analyzed. The training cohort was obtained by means of repeatable sampling, and the original sample was used as the validation cohort. The training cohort was used to construct the prognostic model for patients with PSC, and the validation cohort was used to verify the prognostic model. Stepwise regression and Least absolute shrinkage and selection(Lasso) were used to screen variables, and meaningful variables were selected to construct prognostic models based on univariate and multivariate analysis of the training cohort. The prediction models were subsequently evaluated using calibration curves and model decision curve analysis.

Results

The variables that were screened out to have an impact on prognosis included body mass index (BMI), tumor-lymph nodes-metastasis (TNM) staging system and tumor T-stage. The constructed prediction models were tested using calibration curves and decision curves, showing the consistency of the models with the validation cohort, and also showing the feasibility of the models except in the validation cohort.

Conclusion

In this paper, a prognostic model for patients with pulmonary sarcomatoid cancer was constructed based on the data of patients admitted to our center, and it was also the first prognostic model for patients with PSC based on the database of non-European and American populations. The constructed model showed that BMI, TNM stage of tumor and T stage of tumor had significant influence on the survival of patients with PSC. The model was validated by internal validation for its predictive ability. This model can be used as a good prognostic model for patients with PSC.