Germline genomic patterns are associated with cancer risk, oncogenic pathways and clinical outcomes

Xiaowen Feng, Xue Xu, Derek Li, Qinghua Cui, Edwin WangbioRxiv 2019 · DOI

TL;DR

This study introduces the concept of "cancer-associated germline genomic patterns" (CGGPs) beyond individual gene mutations, revealing that these patterns can predict cancer risk, influence tumor development, and impact clinical outcomes. By analyzing germline genomes of thousands of cancer patients and healthy individuals, the researchers identified seven distinct CGGPs that offer a new perspective on inherited cancer predisposition.

What Problem Does This Paper Address?

For decades, our understanding of inherited cancer risk has largely focused on mutations in specific genes (like BRCA1/2). While this has been informative, it only explains a fraction of hereditary cancer cases. The challenge is to identify broader genetic factors that contribute to cancer predisposition, especially those that might not be captured by looking at single genes in isolation. This paper aims to address this gap by exploring whether recurring patterns of genetic variations across the germline genome, rather than just individual mutations, can serve as a more comprehensive measure of cancer risk and influence disease progression.

Methods

The core of this study involved a systematic analysis of germline genomes using Non-negative Matrix Factorization (NMF). This technique was adapted from its use in identifying somatic mutational signatures in tumors. The researchers applied NMF to germline mutational catalogs to extract recurring "genomic patterns" (CGGPs). They also employed hierarchical clustering to group patients based on their CGGP profiles, allowing for the identification of subgroups with distinct characteristics. Statistical analyses, including t-tests and Fisher's tests, were used to assess the associations between CGGPs, cancer types, smoking status, somatic mutations, and clinical outcomes.

Datasets

The study utilized two primary datasets:

  • Cancer Patient Germline Genomes: Whole-exome sequencing data from 9,712 cancer patients across 22 common cancer types, sourced from The Cancer Genome Atlas (TCGA).
  • Non-Cancer Individual Germline Genomes: Whole-exome sequencing data from 16,670 individuals from three independent cohorts, serving as a control group.

Key Insights and Ideas

  • Beyond Single Genes: The most significant insight is that inherited cancer risk isn't solely determined by mutations in individual genes. Recurring patterns of genetic variations (CGGPs) play a crucial role.
  • CGGP_E and Smoking: One specific pattern, CGGP_E, was found to be significantly enriched in smokers and strongly associated with increased cancer risk in 13 cancer types, highlighting how germline predisposition can interact with environmental factors like tobacco smoke.
  • CGGPs as Predictors: The identified CGGPs and their combinations can predict not only cancer risk but also influence tumor histological subtypes, oncogenic pathways, and patient prognosis across various cancer types.
  • Germline Influence on Somatic Mutations: Germline genomic patterns appear to exert "evo-constraints" on somatic mutations in tumors, suggesting they can influence which mutations are selected during tumor development.
  • Tissue-Specific Effects: The impact of CGGPs can be tissue-specific, meaning certain patterns might be more relevant for particular cancer types or organs.

Key Results

  • Seven Distinct Patterns: Seven cancer-associated germline genomic patterns (CGGPs) were identified.
  • Smoking Association: CGGP_E was significantly enriched in smokers across 13 cancer types, showing a strong correlation with smoking-related somatic mutational signatures in tumors.
  • Prognostic Value: CGGP-defined germline subgroups were significantly associated with prognosis in 12 common cancer types. For example, in brain cancer, specific CGGP subgroups showed significantly different survival outcomes.
  • Histological Subtypes: CGGP-defined subgroups correlated with distinct tumor histological subtypes in brain, kidney, and lung cancers.
  • Oncogenic Pathways: Differential gene expression between CGGP-defined subgroups revealed distinct oncogenic pathways, including those related to DNA damage, cell cycle, immune response, and TNF signaling.

Limitations and Future Work

  • Proof of Concept: The study primarily focused on demonstrating the existence and impact of CGGPs, serving as a "proof of concept." Quantifying intrinsic cancer risk accurately would require much larger cohorts and more sophisticated evaluation approaches.
  • Binary Assignment: For some analyses, consecutive weighting factors were translated into binary CGGP assignments, which might simplify the complex nature of these patterns.
  • Lung Cancer Anomaly: The association between CGGP_E and poor prognosis was not observed in lung cancer, possibly because direct exposure to tobacco mutagens in the lungs might overpower the germline genetic factors.
  • Future Directions: The authors suggest further investigation into germline genomic patterns to uncover genetic mechanisms of diseases that go beyond the current gene-centered paradigm. They also propose that CGGPs could be used for classifying cancer patients and exploring their role in other genetic diseases.

Summary updated

Abstract

Summary Germline genetic polymorphism is prevalent and inheritable. So far mutations of a handful of genes have been associated with cancer risks. For example, women who harbor BRCA1/2 germline mutations have a 70% of cumulative breast cancer risk; individuals with congenital germline APC mutations have nearly 100% of cumulative colon cancer by the age of fifty. At present, gene-centered cancer predisposition knowledge explains only a small fraction of the inheritable cancer cases. Here we conducted a systematic analysis of the germline genomes of cancer patients (n=9,712) representing 22 common cancer types along with non-cancer individuals (n=16,670), and showed that seven germline genomic patterns, or significantly repeatedly occurring sequential mutation profiles, could be associated with both carcinogenesis processes and cancer clinical outcomes. One of the genomic patterns was significantly enriched in the germline genomes of patients who smoked than in those of non-smoker patients of 13 common cancer types, suggesting that the germline genomic pattern was likely to confer an elevated carcinogenesis sensitivity to tobacco smoke. Several patterns were also associated with somatic mutations of key oncogenic genes and somatic-mutational signatures which are associated with higher genome instability in tumors. Furthermore, subgroups defined by the germline genomic patterns were significantly associated with distinct oncogenic pathways, tumor histological subtypes and prognosis in 12 common cancer types, suggesting that germline genomic patterns enable to inform treatment and clinical outcomes. These results demonstrated that genetic cancer risk and clinical outcomes could be encoded in germline genomes in the form of not only mutated genes, but also specific germline genomic patterns, which provided a novel perspective for further investigation.
BibTeX
@misc{feng2019germline,
  author = {Xiaowen Feng and Xue Xu and Derek Li and Qinghua Cui and Edwin Wang},
  title = {{Germline genomic patterns are associated with cancer risk, oncogenic pathways and clinical outcomes}},
  year = {2019},
  doi = {10.1101/616268},
  howpublished = {bioRxiv},
}
Figure