Cancer Therapy Reshapes Mutant Cell Selection in Normal Esophagus, Sanger-Led Study Finds
核心洞察
DNA sequencing of normal esophageal tissue from cancer patients shows chemotherapy and radiotherapy alter which mutant cell clones expand in healthy tissue.
Patients who received chemoradiotherapy (CROSS) had double the proportion of tissue carrying TP53 (搜索) and PPM1D (搜索) mutations compared with other treatment groups.
Combination chemotherapy increased normal cells carrying mutations linked to resistance to 5-fluorouracil, potentially protecting healthy tissue from toxicity.
Cancer treatments give a growth advantage to cells carrying particular genetic changes in healthy tissue, according to a study published in Nature Genetics on 11 September by researchers at the Wellcome Sanger Institute (搜索), University of Cambridge, University College London and collaborators.
The team used DNA sequencing to map mutations in normal esophageal epithelium removed from esophageal cancer (搜索) patients after they had received either combination chemotherapy, chemoradiotherapy, or no treatment before surgery. The analysis revealed that different treatment regimens changed the landscape of mutations in normal tissue, with chemoradiotherapy in particular producing significantly more normal cells carrying cancer-related mutations.
Study Design and Sequencing Approach
Normal esophagus was collected from surgically resected material in patients who had received either no treatment (n = 21), or neoadjuvant ECX (n = 17), EOX (n = 14), FLOT (n = 7) or CROSS chemoradiotherapy (n = 11). Normal epithelium was dissected into a gridded array of 2-mm² samples, imaged to confirm the absence of tumor, and then subjected to DNA sequencing.
The team applied nanorate sequencing (NanoSeq), a modification of duplex sequencing that detects somatic mutations in single molecules in polyclonal samples and determines mutational burden and signatures across 30% of the genome. NanoSeq is insensitive to differences in clonality between samples. Targeted sequencing was also performed on 1,995 2-mm² samples across the five groups for 324 cancer-associated genes with median on-target coverage of approximately 500×, identifying a total of 23,692 independent mutations.
Mutational Burden and the Absence of Treatment Signatures
There was no significant difference in mutational burden between the no-treatment, EOX, ECX and FLOT groups, which all had a similar burden to that reported for normal esophagus in age-matched patients without cancer. However, burden in CROSS patients was significantly higher than the other groups. CROSS patients also showed a significantly higher burden of insertions and deletions, specifically deletions, consistent with exposure to ionizing radiation.
De novo signature analysis revealed two signatures, SBS288A and SBS288B. Signature A was a combination of the clock-like signatures SBS1 (10%), SBS5 (49%) and SBS40a (41%), while Signature B combined SBS1 (4%), SBS5 (17%), SBS40a (54%) and SBS16 (24%). SBS16 is correlated with alcohol ingestion in normal esophagus and esophageal squamous cell carcinoma. There was no significant effect of treatment on the number of Signature A mutations per Mb (P = 0.905, Kruskal–Wallis test) or Signature B mutations per Mb (P = 0.1163, Kruskal–Wallis test).
Notably, the mutational signatures of 5-FU, oxaliplatin or carboplatin (SBS25, 31 and 35) were absent. Using mSigAct analysis, the researchers concluded they could not reliably identify treatment-specific signatures in any of the treatment groups. The absence of chemotherapy-specific signatures and the similar density of mutants across all groups argue that most mutant clones detected likely existed before therapy rather than being newly generated by treatment, suggesting that differences between groups reflect alterations in the selection of mutants.
TP53 and PPM1D Clones Expand Under Chemoradiotherapy
The most prevalent mutant genes — TP53 (搜索), PPM1D (搜索), NOTCH1 (搜索), NOTCH2 (搜索) and FAT1 (搜索) — were positively selected in all groups, as they are in normal epithelium from donors without cancer. The proportion of epithelium carrying nonsynonymous mutants of FAT1, NOTCH1 and NOTCH2 was similar across groups. However, the proportion of tissue mutant for TP53 and PPM1D in CROSS patients was double that of the other groups, and the estimated size of TP53 and PPM1D mutant clones was substantially larger in CROSS patients.
Over 90% of PPM1D (搜索) mutations in all groups were nonsense or frameshift mutations within exon 6, generating a stabilized form of the protein that retains phosphatase activity, inhibits TP53 (搜索) function, activates MTORC1 and confers resistance to chemotherapy and ionizing radiation in mutant hematopoietic stem cells. To test whether exon 6 PPM1D mutants are preferentially selected in irradiated esophageal epithelium, the team performed CRISPR–Cas9 editing of Ppm1d in primary murine esophageal epithelioid cultures. Deletion of Ppm1d by guide RNAs targeting exon 1 reduced clonal fitness, whereas selection of gRNAs targeting exon 6 was increased in irradiated epithelioids compared with those treated with 5-FU or untreated cultures. The authors concluded that radiation selectively expands clones carrying truncating, gain-of-function PPM1D mutants in esophageal epithelium, although the experiment did not test the carboplatin, paclitaxel and radiation combination given to CROSS patients.
Copy number analysis revealed an increased fraction of the genome with copy number alterations within the CROSS group, although this did not reach statistical significance (P = 0.069, two-tailed Kruskal–Wallis test). Over 5% of CROSS samples carried a copy number alteration on chromosome 17p covering the TP53 (搜索) locus, a significantly higher proportion than other treatment groups. Whole-genome sequencing of 37 clonal 0.05-mm² samples from 12 patients found that three of 27 CROSS clones had biallelic TP53 disruption, which was not seen in the four FLOT or six no-treatment samples.
Treatment-Specific Selection of HLA-B, SPOP, NFE2L2, RAC1 and MTOR
Beyond shared drivers, the dN-to-dS ratio and nonrandom mutational clustering analyses identified genes positively selected only in specific treatment groups. In CROSS patients these included HLA-B (搜索), with five of 11 patients carrying at least one nonsynonymous mutation. Thirteen of 22 (59%, equating to 0.91 mutations cm⁻²) HLA-B nonsynonymous mutations were frameshift, splice or nonsense mutations likely to result in no functional protein, compared with 0.12 predicted HLA-B protein loss mutations cm⁻² across all other treatment groups. The HLA locus on chromosome 6p also carried copy number alteration events in 8.7% of 2-mm² samples from CROSS patients.
Nonsynonymous mutants of SPOP (搜索), encoding an adapter protein involved in the DNA damage response and cell death, were also selected only in CROSS patients. Eighteen of the 21 SPOP mutants in the CROSS group were missense and significantly clustered in one-dimensional analysis. Structural modeling showed that mutations in the CROSS group that increased the binding energy of SPOP to its ligand were significantly enriched compared with the neutral distribution, suggesting a functional impact. SPOP mutations in CROSS patients showed similar positional enrichment, 3D proximity to the interface and ligand binding energy to loss-of-function mutations seen in prostate cancer.
In the FLOT group, mutant NFE2L2 (搜索), RAC1 (搜索) and MTOR (搜索) showed clustering in multiple groups but were only under positive selection in FLOT patients. NFE2L2 is a transcription factor regulating cellular responses to oxidative stress and has been linked to chemoresistance in a range of cancers; its levels are tightly regulated by KEAP1 (搜索). Nearly all NFE2L2 mutations observed across the cohort showed a predicted increase in ligand binding energy, decreasing the likelihood of KEAP1 binding. The FLOT group showed the most significant clustering with the highest density of mutations cm⁻², with 78% of missense mutations occurring within the KEAP1 binding region (amino acids 17–82), although this did not reach statistical significance.
For RAC1 (搜索), a member of the RHO family of GTPases, elastic network modeling and molecular dynamics simulations of common FLOT-group mutants (I21V, P29L, P34S and D57H) showed that the flexibility of the Switch domains in all four mutations was significantly increased compared with wild-type protein. P29L is a known gain-of-function mutation linked to cancer progression. The authors suggest that normal epithelium harbors a pool of RAC1 mutations involving the hinge region, but that FLOT treatment selects those that may act through increased affinity of GTP for the binding site.
MTOR (搜索) mutations were significantly closer to the kinase domain than the neutral distribution, most significantly in the FLOT group, and a significant proportion of MTOR mutations in all treatment groups overlapped with known, experimentally validated gain-of-function mutations. Clones carrying kinase domain mutants were significantly larger than those with other missense mutations. Binding energy calculations indicated that FLOT group mutants were not predicted to alter the binding affinity of rapamycin, suggesting the selected clones would likely remain sensitive to inhibition by rapalogs.
Clinical Implications and Next Steps
The researchers suggest the findings begin to uncover the genes and protein domains that make normal cells sensitive or resistant to treatment and could help shape cancer treatment to minimize damage to normal tissues, as well as inform the development of targeted treatments that destroy cancer cells while leaving normal tissue unharmed. By identifying mutant cells in normal tissue that are selected for by cancer treatment, the work may provide a catalog of potential genetic targets that modify how cells respond to treatment and lead to further research into how tumors become drug resistant.
"Cancer treatments not only kill cancer cells but also affect normal tissues. To investigate this, we used genome sequencing to study the normal lining of the esophagus from esophageal cancer (搜索) patients before and after treatment. In particular, in patients who received chemoradiotherapy, we found there were significantly more cells with TP53 (搜索) mutations — a well-known cancer mutation," said Dr. Joanna Fowler, first author at the Wellcome Sanger Institute (搜索).
Dr Phil Jones, co-senior author at the Wellcome Sanger Institute (搜索) and University of Cambridge, said: "Our bodies are a Darwinian battleground, where cells are constantly evolving, expanding and fighting for space in our normal tissues. If you change the rules of this competition by introducing a drug, different genetic mutations are going to enable cells to win or lose. We were surprised to find that only a few weeks of cancer treatment can drastically change decades of evolution in our cells."
Dr Hayley Brown, research information manager at Cancer Research UK, said: "People with esophageal cancer (搜索) often need intensive treatment, but we still have much to learn about how these therapies affect the rest of the body. This study gives us an unusual opportunity to see how healthy cells change during treatment, helping us understand what happens elsewhere in the body, not just in the tumor."
Around 9,500 people are diagnosed with esophageal cancer (搜索) in the UK each year, with almost half of new cases in people aged 75 and over. The disease is treated with surgery, chemotherapy, radiotherapy, chemoradiotherapy and immunotherapy. By age 60 to 70, almost all cells in the esophagus will be mutated, and while the majority of these mutations do not lead to cancer, tumors that do form can be hard to treat because symptoms often appear once the cancer has started to spread.
The team is now conducting a pilot study to investigate these effects in other tissues, taking cheek swabs, blood and urine samples from patients before and after treatment for skin, head and neck cancers, to investigate on a larger scale whether there is further evidence of genetic mutations in normal cells being selected for by cancer treatment.