Understanding how normal tissues give rise to cancer is crucial for improving prevention and early detection of this deadly disease. Over the past two decades, the genomic profiles of most types of invasive cancer have been catalogued; however, similar profiling of normal tissues presents a unique set of challenges. Cancer tissues are often abundantly available from biopsies or surgery, but samples from normal tissues tend to be much smaller, and specimen-collection practices are less well established, making it hard to gather high-quality material. Writing in Nature, Moore et al.1 overcome these challenges and successfully catalogue cancer-driving mutations in normal endometrial glands.
Endometrial glands are abundant in the lining of the uterus, where they secrete hormones and other substances that are essential for normal menstruation and embryonic development. Endometrial cancer is the sixth most common cancer in women worldwide, with more than 382,000 cases annually2. The mortality rate has increased over the past decade3, heightening the need for prevention and early detection of this disease.
Moore et al. obtained 257 normal endometrial glands from 28 women of various ages. In each case, the authors meticulously isolated the glands using a technique called laser-capture microdissection to separate the epithelial tissue, which lines the gland, from the surrounding stromal cells that make up the gland’s connective tissue. They then performed whole-genome sequencing of the epithelial samples and various other normal tissues from the same women, using a protocol they had developed that is tailored to the analysis of small amounts of DNA.
The group analysed these sequences to identify mutations that are unique to the normal endometrial glands, as well as endometrium-specific changes in the number of copies of any genetic region (caused by duplication or deletion of DNA). They found that, in almost 90% of individuals, the normal endometrial tissue contained driver mutations — which give cells a selective advantage over non-mutated counterparts, and so are thought to promote cancer development. Nearly 60% of the endometrial glands in these women contained one or more drivers.
The authors found 12 genes that contained driver mutations with statistically increased prevalence in normal endometrial tissue compared with that in other tissues. These genes are all known to be frequently mutated in cancer, and, collectively, these mutations have the potential to affect many cellular processes. However, isolated mutations in the individual genes, as was typically the case in Moore and colleagues’ samples, are probably insufficient to make a tissue become cancerous4.
A remarkable finding is that each endometrial gland seems to be clonal — that is, all the cells in the gland are derived from a single epithelial progenitor cell. It might be expected that each gland could develop multiple independent mutations, but the authors’ discovery of clonality indicates that there is instead a uniformity to the mutational process.
As would be expected, the number of mutations increased with age, at the rate of about 29 nucleotide substitutions per gland per year during adult life. Moore et al. reconstructed the phylogeny (the evolutionary development and diversification) of individual glands to document the initial presentation and spread of driver mutations through the tissue over time. They report that many glands that were located in close physical proximity in the uterine wall displayed distant phylogeny. This suggests that the cellular populations in each gland remain genetically isolated, providing many separate opportunities for cancer to develop. The researchers also provide evidence that driver mutations can arise at any time, occurring in some women in their first decade of life and in others at various stages of adulthood. This insight is important because the typical timeline between developing driver mutations and cancer is not yet well defined.
The group’s rigorous methods for sample isolation and sequencing, coupled with their well-developed bioinformatics algorithms, mean that the results of this study should be highly reliable and reproducible. But one caveat is that the authors isolated endometrial glands from a select population of women: most samples were obtained from people undergoing evaluation for infertility, from organ donors, or from women who had died of non-gynaecological causes. Both infertility and nulliparity (having never given birth) are known independent risk factors for endometrial cancer5. And samples collected from women who had died of non-gynaecological causes might be more likely than the average endometrial gland to contain low-risk driver mutations that have less potential to trigger cancer, given that these women died without having developed endometrial cancer.
Future studies would benefit from a more-representative cross-sectional population. The inclusion of women who have conditions that are well-known precursors to cancer, such as atypical endometrial hyperplasia (in which the lining of the uterus becomes abnormally thick) could help in this regard. Researchers might then be able to define a robust landscape of changes that occur during the progression from normal to precancerous tissue to invasive disease. This approach might also help to define the pathogenicity of, and possible necessity for, individual driver mutations that lead to the development of cancer.
Another caveat is the discrepancy between driver mutations identified by Moore et al. and those from other cancer-genome projects, including The Cancer Genome Atlas6. Although the most frequently mutated genes identified in the current study have been previously reported in endometrial cancers, several of the most commonly mutated genes in this cancer are notably not mutated in Moore and colleagues’ samples. The group found mutations in these well-known drivers in less than 2% of the normal endometrial glands that they studied — a surprisingly low frequency, because one would expect that the drivers present in all cancer cells would be the first to arise in normal tissue. This discrepancy probably hints at unknown aspects of the multistep process of tumour initiation, in which certain mutations must arise before others. Determining when and how gatekeeper mutations occur and permit the next step in tumour development will require further analyses of benign, premalignant and invasive tissues.
Knowing that the compilation of driver mutations in normal endometrial glands is different from those found in established endometrial cancers might change the approach for further research into the prevention and early detection of this disease. Determining the role of these mutations in concert with other known risk factors, such as nulliparity, obesity, race and genetic predisposition, will help to better identify women who are at risk of endometrial cancer. Even before we obtain this information, Moore and colleagues’ findings should be useful for ongoing research to detect endometrial cancer at early stages, which includes analyses of cell-free DNA circulating in blood, tampon-based collection of vaginal secretions and liquid-based examination of cervical tissues7–9. More broadly, a better overall understanding of the normal mutational spectra in tissues will refine our knowledge of the consequences of specific cancer drivers for many solid tumours.
Nature 580, 595-596 (2020)