Evolution of Metastases in Space and Time under Immune Selection
来自Cell的一篇文章。
作者发现两个存活时间异常长的病人:P210 and P45
所以从他们的系列手术过程中取到了31个转移瘤的样品进行测序。
image.png
文章亮点:
第一:将tumor clones 分为persistent,Non-recurrent,Eliminated三组
第二:利用改良的算法进行Immunoscore-Editing评分将标本分为Lo-No,Hi-No,Hi-Yes三个组别。immunoediting scores (Yes/No), metastasis size (Hi = T3/T4, Lo = T1/T2)。转移灶大小
image.png | image.png |
---|
采用的实验方法:简简单单四总方法。
1. DNA测序
image.png做了全外显子测序,但是对3个sample做了50个基因的靶向测序,并且把这50肿瘤相关基因定义为Driver mutation
In the case of the three samples in P210 for which exome-seq could not be performed, their origin was inferred from the shared hotspot mutations detected with targeted sequencing.
2.肿瘤周边和中心进行immunoscore评分(根据MI(周边)CT(中心)的免疫细胞浸润数量进行评分)
image.pngimage.png
参考文章 Towards the introduction of the ‘Immunoscore’ in the classification of malignant tumours
3. multispectral imaging.很酷炫,主要是用于后续计算肿瘤细胞和免疫细胞空间位置
image.pngimage.png
4. 770个免疫相关基因表达谱
image.png数据分析方法:
1. SNV:Mutect2 and VarScan2. 还有常规的mutation signature的分析
normalized the number of mutations by the number of bases with sufficient coverage(> 10X) per megabase on the reference genome.突变只有覆盖大于10X才算,是筛选标准。
2. Driver mutations:并且把这50肿瘤相关基因定义为Driver mutation
image.png3.CNV:肿瘤纯度是HE染色判断,涉及三款软件:Sequenza,CNVkit, TitanCNA
image.pngTumor ploidy, tumor purity, and allele-specific copy number quantification were called with Sequenza (v1.12)
CNVkit, TitanCNA计算拷贝数目变异情况
4. Metastatic evolvogram:
4.1 通过somatic mutation 用Phylogenetic tree:R package phangorn这R包做的事情,也可以用MEDICC通过CNV做这个事情
image.png image.png4.2转移演化图也就是克隆演进图可以用 PyClone,SCHISM v1.1.2,FishPlot鱼型绘图可以做到;
image.png PyClone 就可以做到4.3 瘤内异质性的评估:IMH index
image.png5 HLA type 和新抗原预测
分为四步1.HLA typing,2.mutation detection in HLA-I genes,3.Loss of heterozygosity in HLA class I genes ,4.HLA-binding epitope prediction
HLA 基因中的high fidelity mutations筛查 POLYSOLVER
LOHHLA分析Loss of heterozygosity in HLA class I genes
netMHCpan (v3.0)预测结合力计算新抗原
6 Metastasis and clonal immunoediting
作者使用了新的方法,也是文章的highlight,我目前还接受不了;
immunoediting score
来个图,留着以后研究
7 T cell and B cell repertoire 免疫组学研究,使用Exome-seq reads
image.png其实条件允许可以用TCR测序
8 RNA-seq immune related gene 得到以下结论:可以作为重要参考,但是感觉就是出了下面这个和各个指标之间相关性的图以外,也没有别的应用了。
image.png9 这是文章的技术难点:immunoediting score 其实就是immunogenic mutations免疫原性突变/非同义突变。可以这样简单理解。
The immunoediting score represents the ratio of expected to observed immunogenic mutations per nonsilent mutation
image.png
要理解上述问题,需要理解概念Bpred
The expected fraction of immunogenic mutations, Bpred, for sample i with a set of coding mutations Mi was calculated as: image.png
需要理解概念Npred
Given a set of coding mutations, Mi and their corresponding mutation context, sðmÞ, in the sample i, the expected number of nonsynonymous
mutations, was calculated as:
image.png
10 使用Dn/Ds score 评价肿瘤突变克隆进化时免疫选择的压力
We compared positive selection of nonsynonymous mutations to selection of immunogenic mutations, using the Dn/Ds score.
image.png
11 使用the package spatstat 对细胞于细胞之间的空间距离进行计算。
12 通多数据库的资料,构建了预测构建预测模型。The covariates were selected for the model using the Akaike Information Criterion (AIC) and one-sample-out approach. 很高大上的样子。Akaike Information Criterion (AIC)
文章的方法学部分就此结束。 immunoediting score有点难以理解,但其实应该是一个简单的概念。真正的难点是文章对这个immunoediting score进行了数据库的验证工作。此外预测模型的构建也是很难理解,同意也做了模型数据库验证工作。
Results 来看几个图
图一:所有的患者信息都在这个图中:转移灶直径、免疫评分、location、treatment
image.png图二:tumor clone的比例 + clone演进的鱼形图
鱼型图图三:肿瘤WES的结果信息
image.png图四:免疫细胞在肿瘤周边肿瘤内部浸润的计数
图五:TCRβ clone数量
图六:immunoediting socre 与临床各指标间的关系
image.png图七:肿瘤与免疫细胞之间的空间距离
image.png图八:immunoediting score 与临床相关性分析,ROC曲线计算edting sore 对tumor clone status进行预判能力
图九:generated a predictive time-to-event model of metastatic cancer evolution to estimate the probability of recurrence based on the selected covariates。复发预测模型建立
image.png#########
未完待续。。。
网友评论