Write a comment

Summary

In this study, the authors use genetically engineered mouse models of tumorigenesis to address what at face value appears to be a paradox: although cancer incidence in humans increases with age, the rate of incidence eventually slows down and even decelerates for many cancers starting around age 70 [1]. The hypothesis they test is whether the molecular changes coincident with aging can reduce tumorigenesis in lung cancer, a type of cancer which in humans has a well-known age bias [1]. By using a method that allows them to control the introduction of a common mutation that causes cancer, including adenocarcinomas of the lung, the authors convincingly show that older mice (equivalent to the human age of 67[2]) have significantly lower tumor burdens. This was determined by assaying the number of cancerous cells, number of tumors, and tumor size after controlling for several important factors such as transduction efficiency, lung weight, and sex.

The authors then ask how molecular changes coincident with aging alters tumorigenesis and growth. Using the same method they used to quantify tumorigenesis, they target 25 known or putative tumor suppressor genes using CRISPR and show that young and old mice respond differently to the inactivation of specific tumor suppressor genes. Although there are genes whose inactivation leads to a greater effect in either young or old mice, two well-known cancer suppressor genes, p53 and Pten, show a greater effect in young mice. While the authors claim that aging reduces the average magnitude of the effect of tumor suppressor gene inactivation, I am curious to see whether this effect is still significant if Pten is not considered in this analysis. Validation of these pooled results in the context of p53 deficiency and Pten convincingly strengthen their case for the genes identified in their pooled screen of tumor suppressor genes.

Finally, the authors use single-cell RNA sequencing to ask what notable transcriptome wide changes they can find when comparing cancerous cells from young and old mice with or without Pten inactivation. Using this method, they are able to highlight the importance of various signaling pathways and cell types in which many of these changes are taking place, leaving many interesting open questions for additional follow up. Additionally, they show that upon Pten inactivation, the transcriptomes of cancerous cells from older mice look more similar to those of younger mice compared to a non-targeting control.

Overall, the study is a high quality effort to better understand the relationship between aging and cancer that includes many important controls and carefully considers its findings. That being said, I believe that there are several important points that require follow up listed below.

Major comments

• My first comment is meant as a compliment: I very rarely see replication of experiments as in this study where the authors often performed the same experiment on batches of mice 3 times and explicitly presented them in this way. While not every lab could afford to do it this way, I commend the authors for using their resources to ensure their observations were replicable and it certainly speaks to the rigor of their science.

• I am somewhat wary of the title of the manuscript and the generality that it presumes. Considering that this study was performed in heavily engineered lab mice and focuses on a particular type of cancer induced by a particular type of mutation I would recommend amending the title to be more specific to the authors findings while extrapolating on their findings in the discussion of the paper.

• Justification for what is a young or old mouse is crucial for this study. Thus, it stood out to me that although the decline of cancer incidence in humans is not noticed until around age 70, old mice are defined in this study as being of “the age at which most molecular phenotypes of aging emerge…”. Although the difference in tumorigenesis and tumor growth between young and old mice in this study is convincing, I wonder if the authors could address more directly how generalizable these results are to humans. That is, if one were to theoretically perform this type of experiment in humans (impossible, of course), would we expect to see a difference in cancer incidence earlier than the epidemiological data suggests? The underlying assumption here being that the molecular hallmarks of aging in humans show up earlier than 70 years. This would suggest that the effects of aging on tumorigenesis are quite strong, but obfuscated by the many confounders that motivated this study in the first place.

• As mentioned above in my summary, I am interested in an analysis similar to what was performed in Figure 2d, but without including Pten. My reasoning for this is simply because it appears to be a strong outlier relative to the rest of the data and skews the narrative towards one where tumor suppression is more skewed in old relative to young mice. If the effect is still significant upon removal of Pten, then I think this can only strengthen the authors’ original claims.

• In Figure 4f the authors show a pathway activation score and highlight the reduction in MAPK signaling in old mice. However, It struck me that JAK/STAT signaling activation was much higher in old versus young mice. Considering that mutations in the JAK/STAT pathway are also implicated in certain types of cancer and play important roles in cellular proliferation, among other things, I wonder if the authors could address why they believe the combination of differential pathway activation that they see would lead to repression of tumorigenesis and tumor growth and not vice versa [3]? It just felt to me as if the authors presented the full results, but only highlighted those that were easy to explain given their model. For example, I wonder if the differential activation of these pathways is in any way specific to the type of cancer or type of cancer-driving mutation being modeled here.

Minor comments

• PTEN is introduced early on in the manuscript, but never properly spelled out.

• This part of the abstract is difficult to read–particularly the last sentence: “Moreover, cancer incidence decreases in the oldest part of the population, suggesting that very old age may reduce carcinogenesis. Here we show that aging represses tumor initiation and growth in genetically engineered mouse models of human lung cancer. Moreover, aging dampens the impact of inactivating many, but not all, tumor suppressor genes with the impact of inactivating PTEN, a negative regulator of the PI3K/AKT pathway, weakened to a disproportionate extent.”

• While I appreciate the authors attempts to make the code and data accessible prior to publication, it’s unclear what this means considering that the preprint is now available for others to review. While making data available ahead of time is difficult, making analysis code accessible through Github at the same time as bioRxiv upload would definitely help in more rigorously evaluating the results of the paper.

• It appears as though in Extended Data Fig. 10d and g the coloring for the “Mean expression in group” part of the figure legend does not look like it rendered properly.

References

1. Radkiewicz C, Järkvik Krönmark J, Adami H-O, Edgren G. Declining Cancer Incidence in the Elderly: Decreasing Diagnostic Intensity or Biology? Cancer Epidemiol Biomarkers Prev. 2022;31: 280–286.

2. Dutta S, Sengupta P. Men and mice: Relating their ages. Life Sci. 2016;152: 244–248.

3. Constantinescu SN, Girardot M, Pecquet C. Mining for JAK–STAT mutations in cancer. Trends Biochem Sci. 2008;33: 122–131.

Competing interests

The authors declare that they have no competing interests.