Metadata supporting data files of the related manuscript: LobSig is a multigene predictor of outcome in Invasive Lobular Carcinoma

The related study presents an integrative analysis of gene expression and DNA copy number to identify novel drivers and prognostic biomarkers for Invasive Lobular Carcinoma (ILC). A 194-gene set (LobSig) was derived that is highly prognostic in ILC.<br><br><b>Study design and methodology</b>: <i>In silico</i> integrative analysis of gene expression and DNA copy number was carried out using single nucleotide polymorphism (SNP) array data from ILC tumors of three cohorts: the "in house" UQCCR (n=25), the METABRIC (n=125) and the TCGA (n=146) cohorts. Fresh frozen tumors were accessed from the Brisbane Breast Bank at the University of Queensland Centre for Clinical Research and from the Australian Breast Cancer Tissue Bank. These cases constituted the ‘UQCCR’ cohort. DNA and RNA were extracted from frozen tissue and gene expression profiling of the “in house” UQCCR samples was performed. Please read the related manuscript for more details on the methodology.<br><b>Participant consent:</b> All patients provided written, informed consent to the use of their tissues for research and the study had ethics approval from the Human Research Ethics Committee and Royal Brisbane and Women’s Hospital.<br><b>Data files:</b> The .<b>xlsx </b>file provides a description of the manuscript-related datasets that can be found in data repositories and the persistent links to these datasets.<b>Data supporting figures:</b>Figure 1 shows data in Supplementary Table 2 including gene copy number landscape of 303 ILC tumors and GISTIC significant focal alterations in each chromosome across the genome. 1C (Excel files): heatmap of frequency of recurrent amplifications in ILC tumors. 1D, E, F and G (.jpg files) show FISH analysis using gene-specific probes for <i>FGFR1</i> and <i>CCND1</i> in tumor (D, F, G) and normal cells (E). 1H (Prism file): Boxplot of copy number versus mRNA expression z-scores of <i>FGFR1</i> and <i>CCND1</i>. 1I (Excel file): Spearman genes plotted as rho across chromosomal location and ANOVA genes plotted as -log P Value across chromosomes.Figure 2: 2A (Excel file) corresponds to Supplementary Table 4 and shows a Manhattan plot of the prognostic grade 2 ILC differentially expressed genes across all chromosomes. 2B-E (Prism files): Survival plots where high and low LobSig scores have been associated with patient survival in different types of ILCs. 2F-J (Prism files): survival plots where scores of various gene expression prognostic tests have been associated with survival in patients with grade 2 ILC. 2K (Prism file): Survival plot for the LobSig stratification in cases of the RATHER cohort. ROC curves comparing performances of prognostic gene signatures in ILC tumors (Figure 2L, jpg file) and Grade 2 ILC tumors (Figure 2M, jpg file).Figure 3 (Prism file): Heatmap comparing LobSig risk predictions to the risk scores generated by NPI, GGI, PAM50 ROR and OncotypeDx.Figure 4: 4A (Prism files): Survival plot of the LobSig stratified NPI moderate grade 2 ILC population. 4B (Excel file): Heatmap showing histopathological characteristics of NPI moderate LobSig stratified tumors. 4C (Prism and Excel files): Scatterplot showing the genomic alterations that are enriched in the LobSig stratified NPI moderate ILC cohort. 4D (jpg images): ROC curve comparing the prognostic performance of different prognostic gene signatures in the NPI moderate grade 2 ILC cases.Figure 5: 5A (Prism and Excel files): Genomic alterations and their enrichment in the LobSig high risk group from grade 2 ILC tumors. 5B (GeneGo csv and tiff files): Gene Ontology analysis of the differentially expressed genes between LobSig high and low tumors.<b>Data supporting Supplementary figures:</b>Figure 1 (Excel files): Corresponds to data shown in Supplementary Table 2 and shows global recurrent alterations plotted along the genome across the three different cohorts.Figure 2 (Excel files): Shows breast cancer specific survival (BCSS) for patients in the TCGA cohort and its association with specific genomic alterations.Figure 3 (Excel files): Shows BCSS for patients in the TCGA cohort and its association with specific focal co-amplifications.Figure 4 (Excel and Prism files): 4A: Flow chart showing a summary of the experimental design. 4B: Correlation of <i>CCND1</i> copy number state and gene expression data; the Spearman analysis. 4C: Relationship between <i>CCND1</i> gene expression and gene copy number state; ANOVA analysis.Figure 5 (Excel and Prism files): Boxplots showing the relationship between gene expression and gene copy number state for the top 6 ANOVA genes.Figure 6 (Excel and Prism files): Scatterplots showing the correlation of copy number state and mRNA expression data for the top 6 Spearman genes.Figure 7 (Prism files): Survival plots of LobSig stratified groups in the three different cohorts, with the Logrank p-value, hazard ratio and confidence intervals reported.<b>Data supporting supplementary tables:</b>Table 1 (Excel files): Table of clinical characteristics of all the ILC cases in all 4 cohorts (METABRIC, TCGA, CCR and RATHER).Table 2 (Excel files): Putative driver genes identified by GISTIC analysis of the ILC cases in TCGA and METABRIC cohorts, respectively.Table 3 (Excel files): GISTIC focal alterations associated with BCSS data in regions that are highly prognostic in ILC tumors.Table 4 (Excel files): Supervised analysis of differential gene expression profiling of ‘good‘, and ‘poor’ BCSS outcome groups.Table 5 (Prism files): Clinical characteristics of two differential gene expression clusters. Chi-squared analysis was performed to determine whether the two sample subgroups were significantly associated with clinical/histopathological characteristics.Table 6 (csv files): Panel of genes that were analyzed using the MetaCore which led to the identification of molecular pathways in the poor outcome and good outcome groups, respectively.Table 7 (R files and Excel files): List of genes that were identified from the ANOVA analysis of ILC tumors from the three cohorts. Tables 8 (R and Excel files) and 9 (R, Excel and Venn text files): Genes that were identified from the Spearman analysis and genes that were common between the two methods (Spearman analysis and ANOVA), respectively, in ILC tumors from the three cohorts.Table 10 (R and Excel files): List of the significant, prognostic ILC genes, with Logrank p&lt;0.05.Table 11 (Excel files): Commonality of the 194-gene signature of LobSig with other prognostic gene signatures.Tables 12, 13 and 14 (R and Excel files) show the univariate and multivariate models to assess the prognostic significance of LobSig grade 2 ILCs, LobSig grade 1, 2 and 3 ILCs and LobSig grade 3 ILCs, respectively, compared to other clinical and experimental indicators.Table 15 (Excel files): Unique molecular subgroups (that were prevalent among LobSig stratified tumors.Table 16 (Excel files): The table shows a list of pathways that were identified using GeneGo for both LobSig High and LobSig Low stratified tumors.Table 17 (Excel files): The table shows a list of pathways that were identified using GeneGo in LobSig High and LobSig Low stratified tumors.<br><b>Data access and terms of use:</b> The datasets generated during the current study are available in the GEO repository (accession GSE98528), or are included in this published article (and its supplementary information files). Additional datasets analysed during this study are available from TCGA data portal (http://cancergenome.nih.gov/), data status as at May 15, 2014)15; n= 125 ILC from the METABRIC cohort (EGAS00000000083). Specific datasets (such as those in Prism, R and Venn txt file formats) used to generate figures 1-5, supplementary figures 2-7 and supplementary tables 1-10 and 12-17, are available upon request from the corresponding author. Access to datasets will be provided for non-commercial use only and will be granted upon the completion of a Data Usage Agreement (DUA).<br>Requests for data access should be directed to Dr Peter Simpson PhD, FFSc RCPA Senior Lecturer:<b> </b>Discipline of Molecular &amp; Cellular Pathology, Faculty of Medicine and Group Leader:<b> </b>UQ Centre for Clinical Research, at the University of Queensland | Building 71/918 | Royal Brisbane &amp; Women's Hospital | Herston | Qld 4029 | Australia, email: [e-mail]<br>

本相关研究整合分析基因表达与DNA拷贝数特征，旨在发掘浸润性小叶癌（Invasive Lobular Carcinoma, ILC）的新型驱动基因与预后生物标志物。研究构建了一套含194个基因的标签（LobSig），该标签对ILC具有优异的预后预测能力。

**研究设计与方法学**：*计算机模拟（in silico）*整合分析了三个队列的ILC肿瘤的单核苷酸多态性（SNP）芯片数据：自有队列UQCCR（n=25）、METABRIC队列（n=125）与TCGA队列（n=146）。本研究的新鲜冰冻肿瘤样本取自昆士兰大学临床研究中心布里斯班乳腺银行与澳大利亚乳腺癌组织银行，上述样本构成“UQCCR”队列。研究人员从冰冻组织中提取DNA与RNA，并完成自有UQCCR样本的基因表达谱分析。有关方法学的详细信息，请参阅相关研究论文。

**受试者知情同意**：所有患者均签署书面知情同意书，同意将其组织用于本研究；本研究已获得人类研究伦理委员会与皇家布里斯班妇女医院的伦理审批。

**数据文件**：本.xlsx文件提供了可在数据仓储中获取的与论文相关的数据集说明，以及这些数据集的永久链接。

**支撑正图的数据**：
图1展示了补充表2中的数据，包括303例ILC肿瘤的基因拷贝数全景图，以及全基因组各染色体上GISTIC识别的显著局灶性改变。1C（Excel文件）：ILC肿瘤中复发性扩增频率的热图。1D、E、F与G（.jpg文件）展示了使用基因特异性探针对肿瘤（D、F、G）与正常细胞（E）中的*FGFR1*与*CCND1*进行的荧光原位杂交（FISH）分析结果。1H（Prism文件）：*FGFR1*与*CCND1*的拷贝数与mRNA表达z评分的箱线图。1I（Excel文件）：以rho值沿染色体位置绘制的Spearman分析基因，以及以-log P值沿染色体分布的ANOVA分析基因。
图2：2A（Excel文件）对应补充表4，展示了全基因组范围内与2级ILC预后相关的差异表达基因的曼哈顿图。2B-E（Prism文件）：生存曲线图，展示了不同类型ILC中高、低LobSig评分与患者生存的关联。2F-J（Prism文件）：生存曲线图，展示了多种基因表达预后检测评分与2级ILC患者生存的关联。2K（Prism文件）：RATHER队列中LobSig分层的生存曲线图。ROC曲线对比了ILC肿瘤（图2L，jpg文件）与2级ILC肿瘤（图2M，jpg文件）中各预后基因标签的预测性能。
图3（Prism文件）：对比LobSig风险预测值与NPI、GGI、PAM50 ROR及OncotypeDx生成的风险评分的热图。
图4：4A（Prism文件）：2级NPI中等风险且经LobSig分层的ILC人群的生存曲线图。4B（Excel文件）：展示NPI中等风险且经LobSig分层的肿瘤的组织病理学特征的热图。4C（Prism与Excel文件）：展示在LobSig分层的NPI中等风险ILC队列中富集的基因组改变的散点图。4D（jpg图像）：对比不同预后基因标签在NPI中等风险2级ILC病例中的预测性能的ROC曲线。
图5：5A（Prism与Excel文件）：2级ILC肿瘤中LobSig高风险组的基因组改变及其富集情况。5B（GeneGo csv与tiff文件）：LobSig高表达与低表达肿瘤间差异表达基因的基因本体论（Gene Ontology）分析结果。

**支撑补充图的数据**：
图1（Excel文件）：对应补充表2中的数据，展示了三个不同队列中沿全基因组分布的全局复发性改变。
图2（Excel文件）：展示了TCGA队列中患者的乳腺癌特异性生存（breast cancer specific survival, BCSS）及其与特定基因组改变的关联。
图3（Excel文件）：展示了TCGA队列中患者的BCSS及其与特定局灶共扩增的关联。
图4（Excel与Prism文件）：4A：展示实验设计概要的流程图。4B：*CCND1*拷贝数状态与基因表达数据的相关性；Spearman分析结果。4C：*CCND1*基因表达与基因拷贝数状态的关系；ANOVA分析结果。
图5（Excel与Prism文件）：展示前6个ANOVA分析基因的基因表达与基因拷贝数状态关系的箱线图。
图6（Excel与Prism文件）：展示前6个Spearman分析基因的拷贝数状态与mRNA表达数据相关性的散点图。
图7（Prism文件）：三个不同队列中经LobSig分层的各组生存曲线图，附Logrank检验P值、风险比及置信区间。

**支撑补充表的数据**：
表1（Excel文件）：所有4个队列（METABRIC、TCGA、CCR与RATHER）中所有ILC病例的临床特征表。
表2（Excel文件）：分别对TCGA与METABRIC队列中的ILC病例进行GISTIC分析所识别的潜在驱动基因。
表3（Excel文件）：与BCSS数据相关的GISTIC局灶性改变，这些区域在ILC肿瘤中具有高度预后价值。
表4（Excel文件）：对“良好”与“不良”BCSS结局组的差异表达基因进行的监督分析结果。
表5（Prism文件）：两个差异表达基因簇的临床特征。采用卡方检验分析两个样本亚组是否与临床/组织病理学特征显著相关。
表6（csv文件）：使用MetaCore分析的基因面板，分别鉴定出不良结局组与良好结局组中的分子通路。
表7（R文件与Excel文件）：从三个队列的ILC肿瘤的ANOVA分析中鉴定出的基因列表。表8（R与Excel文件）与表9（R、Excel与Venn文本文件）：分别从三个队列的ILC肿瘤的Spearman分析中鉴定出的基因，以及两种方法（Spearman分析与ANOVA）共有的基因。
表10（R与Excel文件）：具有显著预后价值的ILC基因列表，Logrank检验P<0.05。
表11（Excel文件）：LobSig的194基因标签与其他预后基因标签的重叠情况。
表12、13与14（R与Excel文件）分别展示了用于评估2级ILC的LobSig、1/2/3级ILC的LobSig以及3级ILC的LobSig的预后意义的单变量与多变量模型，并与其他临床及实验指标进行对比。
表15（Excel文件）：LobSig分层肿瘤中普遍存在的独特分子亚群。
表16（Excel文件）：该表列出了使用GeneGo为LobSig高表达与低表达分层肿瘤鉴定出的通路列表。
表17（Excel文件）：该表列出了在LobSig高表达与低表达分层肿瘤中使用GeneGo鉴定出的通路列表。

**数据获取与使用条款**：
本研究中生成的数据集可在GEO数据库中获取（登录号GSE98528），或包含在本已发表论文及其补充信息文件中。本研究中分析的其他数据集可从TCGA数据门户（http://cancergenome.nih.gov/，数据状态截至2014年5月15日）15获取；METABRIC队列中的125例ILC样本（EGAS00000000083）。用于生成图1-5、补充图2-7及补充表1-10与12-17的特定数据集（如Prism、R与Venn文本文件格式的数据集）可向通讯作者申请获取。数据集仅可用于非商业用途，且需签署数据使用协议（DUA）后方可获取。
数据获取申请请联系：Peter Simpson博士，FFSc RCPA，高级讲师：分子与细胞病理学学科，医学院；昆士兰大学临床研究中心负责人：昆士兰大学 | 71号楼/918室 | 皇家布里斯班妇女医院 | Herston | Qld 4029 | 澳大利亚，邮箱：[e-mail]