This fits with Thomas Seyfried’s view that metastasis is the macrophages spreading bad mitochondria.
https://www.nature.com/articles/s41586-026-10216-0
chatGPT:
Summary
The paper, “Ageing promotes metastasis via activation of the integrated stress response”, argues that ageing changes the behaviour of KRAS-driven lung adenocarcinoma. In old mice, primary lung tumours grew less, but the cancers became more metastatic. The authors therefore separate two processes that are often assumed to move together: tumour mass expansion and metastatic competence.
Using young and old KrasLSL-G12D/+;Trp53flox/flox lung cancer mice, the authors found that old mice had a 2.5-fold lower primary tumour burden, yet more adenocarcinoma progression, lymph-node involvement, distant metastases and poorer survival. Tumour cultures from old mice showed more EMT markers, more anoikis resistance, larger spheroids, more invasive 3D growth and greater metastatic colonisation after transplantation.
Mechanistically, RNA-seq and ATAC-seq pointed to two enriched themes in old-tumour cultures: epithelial–mesenchymal transition and unfolded protein response / integrated stress response. The key node was PERK–eIF2α–ATF4. Old tumour cells had more open chromatin at the Atf4 locus and reduced accessibility at several UPR-resolving loci, implying stronger ATF4 induction and poorer stress resolution.
The causal claim is quite strong. Inhibiting the ISR with ISRIB, inhibiting PERK, or genetically knocking down/deleting Atf4 reduced anoikis resistance and metastasis in old-tumour cultures. Conversely, ATF4 overexpression in young mouse tumour cells and in human A549 lung adenocarcinoma cells increased EMT-like features and metastatic burden. The paper therefore claims ATF4 is both necessary and sufficient for much of the age-linked metastatic phenotype.
The metabolic section shows that old tumour cells shift away from glucose-derived TCA anaplerosis and towards glutamine-dependent anaplerosis. This appears ATF4-dependent. As a result, old-tumour cultures became more sensitive to glutamine pathway blockade, especially CB-839/telaglenastat, a glutaminase inhibitor. In mouse metastasis models, CB-839 strongly reduced metastasis from old-tumour cells while having little effect on primary tumour growth.
Human relevance was tested using Swedish NSCLC cohorts, TCGA-LUAD and other datasets. Older KRAS-mutant patients were more likely to present with advanced or metastatic disease, and in advanced-stage LUAD, high ATF4 expression was associated with worse survival. In a Swedish KRAS-mutant LUAD tissue cohort, ATF4 nuclear positivity increased with age and high ATF4 was associated with much poorer 5- and 8-year survival.
Novelty
The main novelty is that the paper frames ageing itself as an active tumour-evolutionary force, not just a background risk factor. The surprising point is that ageing can apparently suppress primary tumour expansion while promoting metastatic competence.
A second novel aspect is the proposed ageing → chromatin remodelling → persistent ISR/ATF4 → EMT/metabolic plasticity → metastasis chain. This connects several ageing hallmarks—proteostasis stress, epigenetic change and altered nutrient sensing—to a specific metastatic programme.
A third novelty is therapeutic: the paper suggests that aged metastatic tumours may be especially vulnerable to glutamine metabolism inhibition, not because all KRAS lung cancers are glutamine-addicted, but because ageing-induced ATF4 activation creates a particular metabolic dependency.
A fourth useful contribution is methodological. The authors explicitly criticize the standard use of young mice in lung cancer studies and show that old mice can produce a qualitatively different tumour phenotype, which may explain some failures of preclinical cancer therapies to translate into older human patients.
Critique
The paper is strong because it uses multiple layers of evidence: aged genetically engineered mice, tumour-derived cultures, ATAC-seq/RNA-seq, isotope tracing, pharmacological inhibition, genetic ATF4 manipulation, transplantation models and human cohort validation. The causal experiments around ATF4 are particularly persuasive: loss-of-function suppresses the metastatic phenotype, and gain-of-function can induce it.
The main weakness is that much of the mechanistic depth comes from derived tumour cultures and transplantation assays. These are powerful, but they may select for subclones adapted to culture or transplantation. The autochthonous aged-mouse model is important, but the more detailed ATF4 and glutamine-dependency experiments rely heavily on cell-culture-derived systems.
A second limitation is that the model is centred on KRAS/Trp53 lung adenocarcinoma. The authors show some human evidence that the pattern extends beyond KRAS in some driver-unknown or ALK-positive cases, but not uniformly across EGFR, BRAF, ROS1 or RET-driven tumours. So the result should not yet be generalized to all NSCLC, let alone all cancers.
A third issue is that ATF4 is a stress-response hub, not a cancer-specific target. Inhibiting ISR/ATF4 systemically may have toxicity, especially in older people whose normal tissues may also depend on stress responses. The authors acknowledge that current ISR inhibitors have limited clinical utility and suggest targeting ATF4 or downstream effectors, but that remains a translational gap.
A fourth concern is clinical translation of glutaminase inhibition. CB-839 has already had mixed results in cancer trials, and the paper’s value is partly to suggest better patient stratification. But whether “old, ATF4-high, KRAS-mutant/metastasis-prone LUAD” is sufficient to predict benefit remains unproven. The mouse data are compelling, but clinical proof would require biomarker-selected trials.
A fifth limitation is that the paper mostly treats ageing as a tumour-cell-intrinsic programme. It does discuss the need to study immune surveillance and tumour microenvironment changes, but the current causal chain focuses on cancer cells. In real ageing, immune ageing, stromal remodelling, senescence, extracellular matrix changes, vascular changes and systemic inflammation could all cooperate with or counteract ATF4-driven effects.
Overall assessment
This is a substantial and conceptually interesting paper. Its central claim—that ageing can make lung tumours smaller but more metastatic—is important because it challenges the common preclinical assumption that tumour burden is the main readout of danger. The ATF4 mechanism is well supported experimentally, and the link to glutamine dependency creates a plausible therapeutic angle. The key unresolved question is whether ATF4-high aged NSCLC will prove to be a clinically actionable subgroup in humans, rather than only a strong biological model.