Mutations differ in normal and cancer cells of the oesophagus

What determines whether genetic mutations lead to cancer? Analyses of healthy cells in the human oesophagus reveal that a high level of genetic alterations arises as people age, yet this doesn’t usually result in cancer.
Francesca D. Ciccarelli is at the Francis Crick Institute, London NW1 1AT, UK, and at the School of Cancer & Pharmaceutical Sciences, King’s College London.

Search for this author in:

Errors in DNA replication can alter a cell’s DNA sequence. If such alterations occur early enough in embryonic development, the changes are inherited by all of an organism’s cells. But if the alterations arise later in adult life, it is more difficult to track such changes in a small number of cells in a specific tissue, so the extent of these alterations in normal tissues is poorly understood. It is thought that cancer is initiated when cells acquire a minimum compendium of genetic alterations needed to trigger tumour formation. Understanding when such initiating mutations occur in normal cells is crucial for enabling reconstruction of the early events that lead to cancer. Yokoyama et al.1, writing in Nature, and Martincorena et al.2, writing in Science, have analysed the extent of mutations in human epithelial tissue from the healthy oesophagus, and how this relates to the processes that drive cancer development.

Martincorena and colleagues sequenced 74 cancer-associated genes in 844 tissue samples taken from the upper oesophagus of 9 healthy donors who differed in gender, age and lifestyle. For 21 of these samples, the authors also determined whole-genome sequences. A previous study3 assessing mutations in healthy skin cells reported between two and six mutations per million nucleotides of DNA. By contrast, Martincorena and colleagues report that the mutations in oesophageal cells arose at a roughly tenfold lower rate than that reported for skin. This difference is unsurprising, because skin cells are exposed to more DNA-damaging agents, such as ultraviolet light, than are oesophageal cells.

Instead, the surprise is that, compared with healthy skin, the healthy oesophagus has more mutations in cancer-associated genes. Moreover, at least a subset of these altered genes was under strong positive selection, meaning that the genetic alterations promoted cell proliferation, leading to the formation of cell clones (Fig. 1). Compared with the samples from younger people, the overall number of mutations, the number of mutations in cancer-associated genes and the size of the clones were all greater in the samples from older people. The authors found that the donors’ samples had an average of about 120 different mutations in NOTCH1, a known cancer-associated gene, per square centimetre of normal oesophageal tissue. Several of these mutations were of the same type as those seen in a cancer of the upper oesophagus called oesophageal squamous cell carcinoma (OSCC).

Figure 1 | Mutations in the human oesophagus. Yokoyama et al.1 and Martincorena et al.2 analysed the mutations that arise in oesophageal tissue during healthy ageing, and compared these with changes that occur in a cancer called oesophageal squamous cell carcinoma (OSCC). a, The authors report that, in oesophageal samples from young, healthy individuals, only a small percentage of tissue has cells with cancer-associated mutations, and that such cells divide to form cell clones (green). b, By comparison, samples from older, healthy individuals contained many more of these mutations and cell clones. The authors report that the most commonly mutated gene in tissue samples from older healthy individuals is NOTCH1. c, The authors found that a mutation in TP53 was the most common genetic alteration in OSCC. It is thought that environmental factors might have a role in driving the genetic alterations that cause OSCC7,8.

Yet despite these similarities, there were striking differences between the expansion of cell clones in the normal oesophageal tissue and in OSCC. Normal and cancerous clones seem to be driven by mutations in different genes. NOTCH1 was the most frequently mutated gene in healthy oesophageal samples, whereas a previous study4 reported that NOTCH1 was mutated in only around 10% of OSCCs. Mutations in the cancer-promoting gene TP53 are found in more than 90% of OSCC cases4, but were present at a much lower frequency in normal oesophageal samples.

In normal cell clones, Martincorena et al. found that the prevalent mutational signatures (the type of nucleotide changes and the DNA context in which these occur) were typical of physiological processes, such as the mutational changes that occur over time owing to errors in DNA replication. By contrast, OSCC is dominated mostly by mutational signatures associated with mutation-causing agents, such as cigarette smoking, alcohol intake or an enzyme called APOBEC that can modify DNA5. Another hallmark of OSCC is chromosomal instability, which causes frequent gene loss or gain4. Martincorena and colleagues observed low levels of chromosomal instability in healthy oesophageal cells.

Yokoyama and colleagues’ results were similar to those of Martincorena and colleagues. Yokoyama et al. analysed 682 samples of healthy and cancerous oesophageal tissue from 139 individuals, who differed in their ages and risk of developing OSCC. To determine DNA sequences, the authors used a combination of approaches, ranging from the sequencing of whole genomes and whole exomes (protein-coding regions) to the resequencing of specific genes. They found mutations in normal tissue samples from people who had cancer and in samples from healthy individuals. The number of mutations in the normal oesophagus increased with age and exposure to known cancer-risk factors.

Yokoyama et al. also observed most of the same differences between normal and cancerous cell clones as those noted by Martincorena and colleagues. Using two computational approaches, they identified 24 genes that were frequently mutated in the healthy and cancerous samples, but only 6 of these were altered in both. Consistent with the study by Martincorena and colleagues, Yokoyama et al. found that TP53 and NOTCH1 were the most commonly mutated genes in cancerous and healthy clones, respectively. The age-related mutational signature was prevalent in normal oesophageal cells, especially in individuals who were at low risk of developing OSCC. By contrast, mutational signatures associated with APOBEC activity or alcohol intake were prevalent in cancer samples and detectable in normal samples from individuals at higher risk of cancer. The authors detected few chromosomal alterations in normal samples and confirmed the high level of chromosomal instability in OSCC.

Both studies offer insights into the evolution of healthy tissues as people age, and prompt speculation about how this might relate to cancer. The clonal expansion of normal oesophageal cells after cancer-promoting genes have mutated seems to be necessary, but not sufficient, to drive cancer, so something else must happen to the cells for tumours to form. For example, gaining a large-enough number of alterations in cancer-promoting genes might be needed. In both studies, few of the mutations were present in all the cells of the normal clones, and many of the cancer-promoting mutations were often found in spatially distinct subclones. This suggests that none of the normal cells had acquired enough cancer-promoting alterations to start cancer formation.

Other missing factors needed to drive cancer formation might be of environmental origin. The incidence of OSCC is high in Asia and South America, but is low in the Western world6. The reasons underlying this geographic distribution are mainly unknown, but it is thought that lifestyle and environmental factors might have key roles7,8. All donors providing samples for Martincorena and colleagues’ study were from the United Kingdom, and so the mutated cells had probably not been exposed to external factors that can drive the initiation of OSCC. By contrast, the samples analysed by Yokoyama and colleagues were from Japanese individuals, and some of the donors were at high risk of developing OSCC. Yet even in these individuals, several normal clones that seemed to have been present for a long time had not developed into cancer. For example, a clonal expansion in a 70-year-old individual with a high risk of developing OSCC probably started with a mutation in TP53 that the authors estimated occurred when the individual was 13 years old. Over the many decades, this clone expanded to reach an area of 7 square millimetres, but did not develop into cancer. This indicates that other factors needed for that to happen were lacking.

Another tempting speculation concerns the cancer-driving role of NOTCH1 and other cancer genes that were more frequently mutated in normal tissue than in cancer tissue. It seems that alterations in these genes are not necessarily early events in cancer progression. Their relatively high mutation frequency in OSCC might simply be because these genes are often mutated in normal tissue. This challenges the idea that common gene alterations in cancer samples indicate genetic changes that are likely to have a cancer-promoting role. Yet this is still the most-common approach used to identify cancer-associated genes9.

Further investigation is needed into the possible role of non-cancer-associated gene mutations in the clonal expansion of normal tissues. Given the partially independent mechanisms that seem to drive the expansion of normal and cancerous clones, perhaps mutations in genes involved in other processes, such as ageing, might be functionally important in normal clones. Martincorena and colleagues mainly sequenced well-known cancer-promoting genes, and therefore such analysis was not feasible in their study.

Yokoyama and colleagues sequenced all protein-coding regions of the genome, but used only well-established computational approaches to find cancer-promoting mutations in normal samples. Unsurprisingly, they found mostly known cancer-associated genes. Interestingly, they discovered that the gene PAX9 , which encodes a transcription factor, is commonly mutated in normal oesophageal tissue. Mutations in PAX9 have not so far been associated with cancer. This suggests that a less ‘cancer-centric’ analysis might reveal other genes that can drive the expansion of clones in normal tissue.

We are only starting to map the extent of genetic alterations in normal tissues. The next challenge will be to fully understand their role in healthy tissues and in disease states.

Nature 565, 301-303 (2019)

doi: 10.1038/d41586-018-07737-8
Nature Briefing

Sign up for the daily Nature Briefing email newsletter

Stay up to date with what matters in science and why, handpicked from Nature and other publications worldwide.

Sign Up


  1. 1.

    Yokoyama, A. et al. Nature 565, 312–317 (2019).

  2. 2.

    Martincorena, I. et al. Science 362, 911–917 (2018).

  3. 3.

    Martincorena, I. et al. Science 348, 880–886 (2015).

  4. 4.

    The Cancer Genome Atlas Research Network. Nature 541, 169–175 (2017).

  5. 5.

    Chang, J. et al. Nature Commun. 8, 15290 (2017).

  6. 6.

    Arnold, M., Soerjomataram, I., Ferlay, J. & Forman, D. Gut 64, 381–387 (2015).

  7. 7.

    Abnet, C. C., Arnold, M. & Wei, W.-Q. Gastroenterology 154, 360–373 (2018).

  8. 8.

    Coleman, H. G., Xie, S.-H. & Lagergren, J. Gastroenterology 154, 390–405 (2018).

  9. 9.

    Repana, D. et al. Genome Biol. (2019).

Download references