Patient age patterns offer clues to distinguish selection in normal tissues from the cancer-driving effects of mutations
Common cancer mutations are not necessarily strong cancer drivers
Mutations estimated to have stronger cancer-promoting effects tended to be more concentrated in relatively younger cancer patients. In acute myeloid leukemia (AML), mutations in genes that undergo positive selection in normal blood were more common among older patients. Published in Nature Genetics, this study reanalyzed existing public genomic datasets and presented a statistical framework for distinguishing selection in normal tissues from cancer-driving effects. The findings suggest that patient age distribution can provide an additional clue when assessing a mutation’s potential to drive cancer.
In cancer genomics, mutations observed more often than neutral mutations have been classified as cancer-promoting drivers. Yet positive selection also occurs in healthy tissues: cells carrying certain mutations may survive or proliferate better than neighboring cells. As these cell populations expand with age, mutations that do not directly promote cancer can still be common in tumors. This makes it difficult to infer cancer-promoting effects from frequency in tumor tissue alone.
Comparisons with normal tissue and age analysis reveal differences
The researchers compared normal tissues from blood, the esophagus, and the colon with cancer genomes. They defined the cancer-promoting effect as the ratio of cancer-initiation rates between two cells that are statistically alike in other respects, one carrying the mutation and the other not. In the actual analysis, they used an odds ratio calculated from the proportion of mutation-carrying cells in normal tissue and mutation frequency in cancer samples as an estimate of this effect. This value differs from an individual’s cancer risk ratio.
The esophageal analysis used paired samples from 63 patients, excluding five patients whose normal and cancer tissues shared mutations. The estimated cancer-promoting effects for TP53 and NFE2L2 were in the hundreds, while NOTCH1 mutations were estimated to have an effect in the direction of cancer suppression. This indicates that mutations showing positive-selection signals in tumors do not all have the same cancer-promoting effect. For blood, the researchers compared 2,410 normal blood samples and 2,340 AML samples spanning the same age range, 36–103 years; unlike the esophageal analysis, these were not paired samples from individual patients.
In a separate age analysis, the researchers examined 4,736 AML patients aged 0–100 years across four combined datasets. The mean age of patients carrying mutations varied by gene: 36 years for KIT and 63 for TET2. Mutations in genes estimated to have stronger cancer-promoting effects tended to occur in younger patients, while mutations in genes positively selected in normal blood were skewed toward older patients. Although mutation frequencies differed substantially between younger and older groups, the relative ranking of cancer-promoting effects was fairly consistent. The researchers interpreted the findings to mean that age-related changes in normal blood should also be considered when explaining age-related differences in mutations in AML.
Age adds evidence about cancer-driving potential but does not establish it
The researchers also assessed mutation frequency alongside age distribution in cancers with limited normal-tissue data. Their reasoning was that mutations more frequent in cancer than neutral mutations and skewed toward younger patients would not fit the model they had specified that explains the pattern solely through positive selection in normal tissues. However, the absence of a significant bias toward younger ages does not mean a mutation lacks cancer-driving potential.
This is a statistical and modeling study, not a clinical intervention trial. Other mutations, genetic backgrounds, and exposure factors may affect the estimates, and the analysis was also limited by grouping different mutations within the same gene. In particular, because data on normal blood from younger AML patients were limited, mutation frequency was approximated using gene length. Age information can complement assessments of cancer-driving potential, but individual risk prediction or improved screening performance has not been demonstrated. The analysis and its limitations are described in the full paper.
Nature Genetics, Published online: 05 May 2026; doi:10.1038/s41588-026-02593-z Mutations may be enriched in tumor samples because they promote carcinogenesis or because they promote clonal expansions in healthy tissue. This study mathematically disentangles these two possibilities by analyzing tumor and normal tissue sequencing datasets.
When mutations commonly found in cancer are assessed as priority targets for treatment or prevention research, effects that expand cell populations in normal tissue must be distinguished from contributions to cancer development. This study presents a framework for evaluating both by considering normal-tissue genomes and patient age together. For example, in cancer types where sufficient normal-tissue controls are difficult to obtain, mutation frequency and a bias toward younger ages could serve as supporting evidence to narrow candidates for follow-up functional experiments. Mutations without an age bias should not, however, be excluded wholesale. Before applying these findings to preventive interventions or treatment-target selection, the function and clinical utility of each mutation must be validated separately.