Expression data from 27 Asian gastric patient-derived xenograft (PDX) models. Expression data from 27 Asian gastric patient-derived xenograft (PDX) models

There is a strong need to develop patient-derived xenograft (PDX) tumor models for studying new treatment options for gastric cancer (GC). With low engraftment success, few collections of GC PDX have been reported and molecular basis of the model establishment remain largely unknown. Here we established n=27 PDX models from n=100 GC tumors and compared their characteristics to GC patient tumors based on the recent work done by ACRG and TCGA, to evaluate the representativeness and relevance of the collection for drug testing. We show that MSI, CIN and MSS/TP53- tumors were preferentially established as PDX, while MSS/EMT and EBV not and that PDX models retained histology and molecular subtypes of parental tumors. By using synapse database, we identified 48 druggable alterations that could be investigated with the collection. Counting alterations for these 48 genes in PDX compared to TCGA tumors revealed models frequently classified with heavily altered tumors but well preserved genomic alteration patterns specific of each GC subtype. The molecular analysis of n=8/27 tumors and corresponding PDX at passage P1, P2 and P3 revealed variations in somatic alteration content both at single nucleotide and chromosomal level in highly unstable MSI and CIN tumors, with changes occurring mainly at P1. In two cases, we show likely emergence of rare subclones carrying known oncogenic alterations in KRAS and PIK3CA. Significance. This study presents a resource of fully annotated GC PDX models for anticancer agent testing. We show that beside close resemblance of PDX with parental tumors, not all subtypes are established, and that the clonal selection plays a key role the establishment of certain tumors. This may have a bearing on translation of observations into the clinic and underline the need to frequently survey the molecular characteristics of the PDX models. Overall design: total RNA samples were extracted from 4 tumor samples per patient-derived xenograft model using the mirVana kit (Ambio) following the manufacturer's recommendations. After standard quality controls RNA were subjected to Affymetrix HGU133 Plus2.0 microarray hibridization.

开发患者来源异种移植（patient-derived xenograft, PDX）胃癌（gastric cancer, GC）模型以研究新型治疗方案存在迫切需求。由于移植成功率偏低，目前已报道的胃癌PDX模型集合寥寥无几，且此类模型建立的分子基础仍未得到充分阐明。本研究从100例胃癌肿瘤组织中成功构建27例PDX模型，并参照亚洲癌症研究组（Asian Cancer Research Group, ACRG）与癌症基因组图谱（The Cancer Genome Atlas, TCGA）的最新研究成果，将该模型集合的特征与胃癌患者肿瘤进行比对，以评估该模型集合用于药物测试的代表性与相关性。研究发现，微卫星不稳定（microsatellite instability, MSI）、染色体不稳定（chromosomal instability, CIN）以及微卫星稳定伴TP53缺失（MSS/TP53⁻）的肿瘤更易成功构建为PDX模型，而微卫星稳定伴上皮间质转化（MSS/EMT）及EB病毒（Epstein-Barr virus, EBV）阳性的肿瘤则难以建立；同时，PDX模型保留了亲本肿瘤的组织学与分子亚型特征。通过Synapse数据库，我们筛选出48个可靶向干预的基因变异，可依托本模型集合开展相关研究。对比本PDX模型与TCGA数据库中胃癌肿瘤的这48个基因变异情况，发现本模型集合多对应变异负荷较高的肿瘤，但仍保留了各胃癌亚型特有的基因组变异模式。对27例肿瘤中的8例及其传代P1、P2、P3代PDX模型进行分子分析后发现，在高度不稳定的MSI及CIN型肿瘤中，单核苷酸水平与染色体水平的体细胞变异含量均存在变化，且此类变化主要发生在P1传代阶段。在2个案例中，我们观察到携带KRAS与PIK3CA已知致癌变异的罕见亚克隆可能出现富集。研究意义：本研究构建了一套经过完整注释的胃癌PDX模型资源，可用于抗癌药物筛选与测试。研究表明，尽管PDX模型与亲本肿瘤特征高度相似，但并非所有胃癌亚型均可成功建立PDX模型，且克隆选择在部分肿瘤的模型建立过程中发挥关键作用。这一发现对将实验结果转化至临床实践具有指导意义，同时也强调了定期监测PDX模型分子特征的必要性。整体实验设计：按照制造商推荐流程，使用mirVana试剂盒（Ambio）从每株患者来源异种移植模型中提取4份肿瘤组织的总RNA样本。经标准质量质控后，将RNA样本进行Affymetrix HGU133 Plus2.0微阵列杂交实验。