Qian Yuquan, Itzel Timo, Ebert Matthias, Teufel Andreas
Division of Hepatology, Division of Clinical Bioinformatics, Department of Medicine II, Medical Faculty Mannheim, Heidelberg University, 68167 Mannheim, Germany.
Department of Medicine II, Medical Faculty Mannheim, Heidelberg University, 68167 Mannheim, Germany.
Cancers (Basel). 2022 Sep 3;14(17):4322. doi: 10.3390/cancers14174322.
Gene expression signatures correlate genetic alterations with specific clinical features, providing the potential for clinical usage. A plethora of HCC-dependent gene signatures have been developed in the last two decades. However, none of them has made its way into clinical practice. Thus, we investigated the specificity of public gene signatures to HCC by establishing a comparative transcriptomic analysis, as this may be essential for clinical applications.
We collected 10 public HCC gene signatures and evaluated them by utilizing four different (commercial and non-commercial) gene expression profile comparison tools: Oncomine Premium, SigCom LINCS, ProfileChaser (modified version), and GENEVA, which can assign similar pre-analyzed profiles of patients with tumors or cancer cell lines to our gene signatures of interests. Among the query results of each tool, different cancer entities were screened. In addition, seven breast and colorectal cancer gene signatures were included in order to further challenge tumor specificity of gene expression signatures.
Although the specificity of the evaluated HCC gene signatures varied considerably, none of the gene signatures showed strict specificity to HCC. All gene signatures exhibited potential significant specificity to other cancers, particularly for colorectal and breast cancer. Since signature specificity proved challenging, we furthermore investigated common core genes and overlapping enriched pathways among all gene signatures, which, however, showed no or only very little overlap, respectively.
Our study demonstrates that specificity, independent validation, and clinical use of HCC genetic signatures solely relying on gene expression remains challenging. Furthermore, our work made clear that standards in signature generation and statistical methods but potentially also in tissue preparation are urgently needed.
基因表达特征将基因改变与特定临床特征相关联，为临床应用提供了可能性。在过去二十年中，已经开发出大量依赖于肝癌的基因特征。然而，它们都尚未进入临床实践。因此，我们通过建立比较转录组分析来研究公共基因特征对肝癌的特异性，因为这对于临床应用可能至关重要。
我们收集了10个公共肝癌基因特征，并使用四种不同的（商业和非商业）基因表达谱比较工具进行评估：Oncomine Premium、SigCom LINCS、ProfileChaser（修改版）和GENEVA，这些工具可以将肿瘤患者或癌细胞系的类似预分析谱分配给我们感兴趣的基因特征。在每个工具的查询结果中，筛选不同的癌症实体。此外，还纳入了7个乳腺癌和结直肠癌基因特征，以进一步挑战基因表达特征的肿瘤特异性。
尽管评估的肝癌基因特征的特异性差异很大，但没有一个基因特征对肝癌表现出严格的特异性。所有基因特征对其他癌症都表现出潜在的显著特异性，尤其是对结直肠癌和乳腺癌。由于特征特异性被证明具有挑战性，我们进一步研究了所有基因特征中的共同核心基因和重叠富集途径，然而，它们分别显示没有重叠或只有很少的重叠。
我们的研究表明，仅依赖基因表达的肝癌基因特征的特异性、独立验证和临床应用仍然具有挑战性。此外，我们的工作明确表明，迫切需要在特征生成和统计方法方面制定标准，可能还需要在组织制备方面制定标准。
