Nature 子刊重磅成果 | 泛生子携顶级专家团队揭秘肺肉瘤样癌分子特征图谱

近期，中国医学科学院肿瘤医院赫捷院士团队与泛生子合作，在Nature Communications（影响因子12.121）上发表了关于肺肉瘤样癌（PSC）的重磅研究成果，依托泛生子全外显子检测（WES）、RNA测序等，是国际上首次从基因组、转录组、甲基化等多组学层面揭示PSC分子特征图谱的研究成果，对PSC的发生及治疗指导提供重要参考依据。

成果亮点概述

1	该研究描绘了PSC的突变全景图谱，发现MET 14外显子跳跃突变在PSC中发生频率较高，在其中的梭形细胞癌中，44% (4/9)的患者携带导致MET 14外显子跳跃的突变。
2	该研究发现了PSC具有高TMB和淋巴细胞浸润程度，提示免疫治疗或能为PSC患者带来新的希望。
3	该研究揭示了PSC中的上皮肿瘤细胞成分与肉瘤样肿瘤细胞成分有着共同的起源，其发生可能与上皮间质转化（EMT）相关，而这一过程可能受甲基化水平改变调控。
4	该研究建立了基于甲基化水平的PSC分子分型，几个亚型在预后和免疫微环境等方面具有明显差异。并且发现PSC分型与肺腺癌/肺鳞癌的个别亚型具有相似的分子特征。

研究背景与研究方法

肺肉瘤样癌（Pulmonary Sarcomatoid Carcinoma，PSC）是一种罕见、预后差、化疗效果差的高级别非小细胞肺癌。已有研究对PSC分析了部分基因的突变，但尚无研究全面地探索靶向治疗和免疫治疗在PSC中的应用潜能。本研究对56例PSC患者的全外显子组、转录组、DNA甲基化进行检测分析，描述PSC的突变全景图，识别PSC患者中更多的靶向治疗靶点，还通过肿瘤免疫微环境和肿瘤免疫原性来评估PSC患者从免疫治疗中获益的潜能。基于多组学数据对PSC患者区分分子亚型，分析不同亚型与治疗、预后的相关性。

PSC存在上皮肿瘤细胞和肉瘤样肿瘤细胞两种成分，使得PSC具有高度肿瘤内异质性和肿瘤间异质性。本研究对14例PSC患者的上皮成分和肉瘤样成分进行全外显子组、转录组、DNA甲基化分析，探究两种组分之间的关系，这有助于了解PSC的发生、指导临床治疗和预后预测。

基因组变化

PSC全外显子组的平均体细胞突变率为7.1个突变/Mb。 

高频突变基因（图1a）分析显示：PSC患者在p53、RTK/RAS、PI3K等通路发生突变。很大比例的PSC中检测到受益于靶向治疗的敏感突变。 

突变特征分析显示，COSMIC signature 4在很大比例的患者中贡献明显（图1b），提示吸烟可能是PSC的风险因素。

(a) Spectrum of the key molecular alterations in PSC. The genomic alterations of smokers and nonsmokers are demonstrated in the left and right side, respectively. The overall number of somatic mutations and clinicopathological data are shown at the top. The type of base-pair substitutions of each sample are displayed in the bottom panel. (b) Mutational signatures in our cohort (n = 56). Samples, annotated for the histological subtype, smoking status and tumor mutation burden, are ordered by the contribution of signature 4. Source data are provided as a Source data file.

上皮成分与肉瘤成分有着共同起源

共有体细胞突变比例、共有CNV、克隆系数等结果都提示，上皮成分与肉瘤成分是同一克隆起源（图2a-c）。TP53、KRAS、EGFR等驱动突变均在进化树的主干上（图2d），提示这些突变是PSC发生的驱动事件。

(a) Columns depict the proportion of shared and specific somatic mutations of the 14 tumors microdissected. (b) Clonality indices for the 14 cases of PSC, suggesting the likelihood of common origin of the two components. Black dotted lines represent the cut-off value of clonality index to define clonal relatedness. (c) The distribution of copy number variations (CNVs) for all PSC samples microdissected. Red indicates CNV gain, and blue CNV loss. White represents the failure in CNV calling. AT, adenocarcinoma component; SCCT, squamous cell carcinoma component; ST, sarcomatoid component. Source data are provided as a Source data file. (d) Phylogenetic trees generated for 4 PSC samples. The length of the trunk (green) and branch (red or blue) represents the number of shared and specific nonsynonymous mutations, respectively. Part of driver mutations is marked. The number of truncal and total nonsynonymous mutations is indicated below. E and S represent epithelial and sarcomatoid component, respectively.

EMT在PSC发生中的作用

转录组分析发现，在PSC的上皮组分到肉瘤组分的转化过程中，没有发生大范围的转录水平变化，差异表达基因多富集到上皮或间质相关的特征以及EMT相关条目。 

分析EMT特征基因的表达水平发现，上皮成分与肉瘤成分的EMT状态不同（图3a, b）。甲基化分析显示，两个成分具有不同的甲基化图谱（图3c），差异甲基化位点的功能也多富集在EMT相关的条目。进一步分析发现，34% EMT特征基因的表达水平与甲基化水平呈现为负相关（图3d）。这些结果提示，在PSC发生过程中，DNA甲基化参与EMT的调控，促进肿瘤增殖和转移。

(a) The heatmap of hierarchical clustering with the expression level of a 76-gene EMT signature, annotated for the histological type and the patients. AT, adenocarcinoma component; SCCT, squamous cell carcinoma component; ST, sarcomatoid component. Source data are provided as a Source data file. (b) EMT scores are plotted for adenocarcinoma (AT), squamous cell carcinoma (SCCT), and sarcomatoid (ST) components as boxplots. Center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range; colored points, outliers. Two-sided Wilcoxon rank-sum test was used for statistical analysis. No P-value adjustment was applied. n = 7, 7, and 14 for AT, SCCT, and ST group, respectively. Source data are provided as a Source data file. (c) The heatmap of hierarchical clustering with DNA methylation data of the epithelial and sarcomatoid components, annotated for the histological type and the patients. AT, adenocarcinoma component; SCCT, squamous cell carcinoma component; ST, sarcomatoid component. (d) The mean expression levels and DNA methylation levels of the 26 genes in the epithelial and sarcomatoid components. M (mesenchymal) markers and E (epithelial) markers are displayed in the upper and lower panel, respectively. S and E in the bottom represent sarcomatoid and epithelial components. Source data are provided as a Source data file.

PSC的肿瘤微环境

泛癌种比较分析发现，PSC的TMB在前五位，淋巴细胞浸润纯度（LF）排第一位，提示PSC存在T细胞炎症微环境（图4a, b）。高TMB和LF提示PSC患者可能会从免疫治疗获益。此外，多重免疫荧光结果显示，上皮成分与肉瘤成分的免疫微环境无显著差异（图4c, d），进一步提示免疫治疗应用于PSC的潜力。

(a) Tumor mutation burden (TMB) of pulmonary sarcomatoid carcinoma (PSC) and The Cancer Genome Atlas (TCGA) tumor types, ordered by median. Somatic single-nucleotide variants and small indels of PSC were called using MuTect (version 3.1-0-g72492bb) and Strelka (version 1.0.14) via an in-house computational pipeline, and the mutation list of TCGA cancer types was released by the Pan-Cancer Atlas consortium (https://gdc./about-data/publications/pancanatlas, mc3.v0.2.8.PUBLIC.maf.gz). (b) Leukocyte fraction (LF) of PSC and TCGA tumor types, ordered by median. The LF of all TCGA samples was collected from a previous study, and we estimated the LF of PSC using the algorithm provided by the same study58. Center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range; points, outliers. (c) Spearman correlation of the fluorescence intensity of CD4, CD8, and CD68 between epithelial and sarcomatoid components. Source data are provided as a Source data file. (d) Representative images of fluorescent multiplex immunohistochemical analysis of P46 and P48. E and S represent epithelial and sarcomatoid components, respectively. Scale bar: 50 μm. The fluorescent multiplex immunohistochemical analysis was performed on the epithelial and sarcomatoid components of 14 patients. CD4, green; CD8, cyan; CD68, red; PD-L1, orange; FOXP3, magenta. CD4, cluster of differentiation 4; CD8, cluster of differentiation 8; CD68, cluster of differentiation 68.

分子分型

DNA甲基化显示PSC分为三个亚型（图5a）。亚型C1的甲基化水平最低，50%患者携带靶向治疗靶点突变。亚型C3的EMT得分比另外两个亚型低，CDH1等上皮细胞特征基因和鳞癌特征基因TP63的表达量更高，提示该亚型保持了上皮细胞的特征，亚型C3具有更长OS。亚型C2的甲基化水平最高，部分患者检测到MET和BRAF靶点突变。在肿瘤微环境方面，淋巴细胞浸润程度、PD-L1表达、CD8 T细胞浸润程度、CTLA4表达等结果显示，亚型C1和亚型C2存在T细胞炎症微环境（图5a, d），为PSC免疫治疗的临床研究提供线索。

(a) PSC tumors are classified into subgroups based on DNA methylation patterns, annotated for the mutation status (Mut, mutated; WT, wild-type) of important genes, TNM stage, smoking status, and histological subtype in the top panel and for the immunohistochemical analysis of PD-L1 and CD8 in the bottom panel. (b) Kaplan–Meier analysis of C1, C2, and C3. The significance is determined by log-rank test. (c) Boxplot shows the expression level of TP63 for three clusters. (d) Leukocyte fraction of each sample for three clusters are plotted as boxplots. Center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range; colored points, outliers. Two-sided Wilcoxon rank-sum test was used for statistical analysis of (c) and (d). No P-value adjustment was applied. PD-L1, programmed death ligand 1; CD8, cluster of differentiation 8; TP63, tumor protein 63.

PSC与TCGA LUAD和LUSC整合分析

基于DNA甲基化数据的聚类分析，PSC、LUAD、LUSC聚成两类，分别以LUAD和LUSC为主，基于此，将PSC患者分为PSC_AD和PSC_SC两组（图6a）。PSC的亚型C3中大多数患者都属于PSC_SC，亚型C3的表达谱与LUSC的classical亚型高度相似（图6b），提示PSC亚型C3与LUSC的classical亚型具有生物学相似性。比较PSC_AD与LUAD的DNA甲基化发现，PSC_AD可分为两组，分别对应亚型C1和亚型C2。亚型C1与LUAD的低甲基化亚型（CIMP-low）更相似，而亚型C2与LUAD的中甲基化亚型（CIMP-intermediate）更相似，这可能与亚型C1和亚型C2的致癌机制有关（图6c,d）。

(a) The percentage of PSC, LUSC, or LUAD in the two major clusters yielded by unsupervised hierarchical clustering of DNA methylation data. (b) The proportion of the four LUSC subtypes in three clusters of PSC. Source data are provided as a Source data file. (c) Boxplots show the Euclidean distance-based similarity between the DNA methylation profiles of C1 and three LUAD methylation subtypes. Source data are provided as a Source data file. d Boxplots show the Euclidean distance-based similarity between the DNA methylation profiles of C2 and three LUAD methylation subtypes. Center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range; points, outliers. Two-sided Wilcoxon rank-sum test was used for statistical analysis of (c) and (d). No P-value adjustment was applied. Source data are provided as a Source data file.

研究结论

本研究针对56例PSC患者，从基因组、转录组、DNA甲基化进行多组学分析，描述PSC的分子特征图谱，将PSC分为三个分子亚型，三个亚型具有不同的突变特征、甲基化水平、EMT状态、免疫微环境，因而在靶向治疗和免疫治疗方面具有不同的潜能。此外，基于上皮成分与肉瘤成分的比较分析，该研究发现上皮成分与肉瘤成分有着共同的起源，在PSC的发生中，会经历上皮组分到肉瘤样组分的转化过程，DNA甲基化对EMT的调控起着重要作用。最后，通过与TCGA的LUAD和LUSC进行比较，提示PSC的三个分子亚型分别与经典肺癌的某些亚型具有相似性。