Mariantonietta D'Ambrosio was looking at the cells that had survived. Not the ones the drugs had killed. The ones that were still there, weeks after the chemotherapy ended, sitting quietly at the edge of the tumour like something left behind by a retreating army. She had expected them to die. Most researchers had. Instead, they lingered.
The Laboratory After the Treatment Ends
What the Cells Were Doing
D'Ambrosio is a postdoctoral researcher at the MRC Laboratory of Medical Sciences in London, and the lead author of a study published in Nature Cell Biology in April 2026 that the senescence field has been waiting for. The paper doesn't announce a cure. What it does is more precise than that: it identifies, for the first time, a reliable molecular weakness in senescent cancer cells. A pressure point. Something that, when pressed, causes the cells to destroy themselves.
To understand why that matters, you have to understand what these cells actually are. Chemotherapy is designed to stop tumours from growing by halting cell division. For the most part, it does this well. The cancer cells hit by the drugs enter a state called senescence: they stop replicating, stop dividing, stop spreading. Oncologists once considered this a victory condition. The tumour is stable. The patient goes home.
But some of those arrested cells don't die. They enter a state that researchers have started calling "zombie-like" with increasing scientific seriousness. The cells are alive, damaged, and profoundly active. They can't divide. They won't die. And they are doing something that is, on reflection, quite alarming. They're talking.
They talk through a system called the Senescence-Associated Secretory Phenotype, which the field abbreviates to SASP. Every day, these zombie cells secrete a cocktail of inflammatory proteins, growth factors, and signalling molecules into the surrounding tissue. And those signals, D'Ambrosio's lab found, do not just sit there.
"Senescence was considered for a long time to be positive, because senescent cells don't proliferate, which is the core feature of cancer," she told MRC LMS. "Normal chemotherapy induces senescence blocking the proliferation of cancer cells, so the tumour doesn't get bigger. But with time you also see the negative side of the senescent cells, because they secrete a lot of factors that influence neighbouring cells and induce even more proliferation, metastasis, and recruitment of bad parts of the immune system."
The good news from the treatment was real. The problem was what happened after.
What the Cells Left Behind
The Signal in the Tissue
The SASP mechanism is worth sitting with, because it rewrites the story we tell about cancer recurrence. When a patient finishes chemotherapy and their scans show no active tumour, there's an implicit assumption that the threat is contained. What the science now suggests is that a contained threat can still be a communicating one.
Mechanism
What SASP Actually Does
Senescent cells that survive chemotherapy secrete a complex mix of molecules into surrounding tissue. This Senescence-Associated Secretory Phenotype includes pro-inflammatory cytokines (such as IL-6 and IL-8), matrix-remodeling enzymes that degrade the tissue barrier, and growth factors that encourage neighbouring cells to proliferate. Two specific molecules, CXCL12 and CSF1, have been found to suppress the anti-tumour immune response: they block CD8+ T cells (the immune cells that kill cancer) from infiltrating the tumour site, and they recruit macrophages that actively protect the tumour instead of attacking it. The result is a microenvironment that is, paradoxically, more hospitable to cancer after chemotherapy than it was before.
Senescent cells also trigger something called "spreading senescence," converting healthy neighbouring cells into senescent ones. The zombie population grows not by dividing, but by corrupting.
The accumulation of therapy-induced senescent cells has been found to promote tumour invasiveness, drug resistance, angiogenesis (the growth of new blood vessels that feed tumours), and metastasis. In breast cancer, increased expression of a senescence marker called p16INK4a is associated with elevated risk of tumour relapse. Across colorectal, breast, ovarian, lung, and pancreatic cancers, senescent cells created by the chemotherapy itself have been found to worsen outcomes.
This is the dark irony at the heart of the field. The drug that stops the tumour from growing can simultaneously prepare the conditions for it to grow back, and to grow back more aggressively.
The biology is specific enough to feel like a design flaw. Chemotherapy works by inducing oxidative stress in rapidly dividing cells. But that same oxidative stress, it turns out, creates the conditions for a kind of metabolic precariousness in the cells that survive it. Senescent cells accumulate high levels of iron and reactive oxygen species. They are, in a sense, already standing on the edge of a particular form of cell death called ferroptosis. And yet they don't fall.
They don't fall because they overexpress a protein called GPX4. Glutathione peroxidase 4 is an antioxidant enzyme that neutralises the lipid peroxides that would otherwise trigger ferroptosis. The senescent cell balances on a wire above an abyss, and GPX4 is what keeps it there. Cut the wire, and the cell destroys itself.
That is exactly what D'Ambrosio and her colleagues, led by Professor Jesus Gil, head of the Senescence group at the LMS, set out to do.
Figure 1 — Therapy-Induced Senescence Pathway
The Signal in the Noise
What They're Trying to Clear
The field of senolytics has been building for about a decade. The word comes from the Greek lysis, meaning dissolution: drugs that selectively kill senescent cells while leaving healthy ones alone. The challenge is selectivity. Senescent cells don't carry a neon sign. They share surface proteins and metabolic signatures with their neighbours. A drug blunt enough to kill everything with senescence markers will cause collateral damage the patient cannot afford.
The first drugs to show genuine senolytic activity were dasatinib, a cancer drug repurposed from its original use against leukaemia, and quercetin, a natural plant compound. Together they've been called D+Q. The combination disrupts the survival pathways that senescent cells depend on, and in animal models it has cleared substantial populations of zombie cells from aged tissue. Clinical trials are ongoing for conditions ranging from Alzheimer's disease to chronic kidney disease. The approach works. But it's not selective enough for some applications, and the mechanism isn't clean enough to be confident about long-term use.
ABT-263, also known as navitoclax, took a different route. It targets the BCL-family proteins that keep senescent cells from undergoing apoptosis (programmed cell death). In mice, it has cleared therapy-induced senescent tumour cells and reduced cancer burden. The problem is that navitoclax also depletes platelets, which creates bleeding risks that complicate clinical translation.
The GPX4 approach discovered by D'Ambrosio and Gil is structurally different from both. Ferroptosis is not apoptosis. It's a distinct form of cell death, triggered by iron-dependent lipid peroxidation. The senescent cells that survive chemotherapy are already, metabolically speaking, primed for it. They have elevated iron. They have elevated reactive oxygen species. GPX4 is the only thing standing between them and that outcome. And crucially, this vulnerability appears to be specific to senescent cells in a way that their neighbours don't share to the same degree.
"Recent papers have shown this predisposition of senescent cells to ferroptosis, but it's a new senescence vulnerability. That creates an opportunity for us to exploit. So now there is research to find senolytic drugs to kill cells through ferroptosis."Mariantonietta D'Ambrosio, MRC Laboratory of Medical Sciences, London — Nature Cell Biology, April 2026
The team screened nearly 10,000 covalent compounds, searching for ones that could inhibit GPX4 selectively. They found four with meaningful activity, three of which targeted GPX4 directly. In mouse models of cancer, these compounds reduced tumour size and improved survival rates. The paper's title is exact in its reach: "Electrophilic compound screening identifies GPX4-dependent ferroptosis as a senescence vulnerability."
This is pre-clinical research. No human trials have been announced. What the paper establishes is a mechanism and a proof of concept: senescent cells that survive chemotherapy rely on GPX4 to stay alive, and GPX4 can be targeted. That's an important distinction from prior senolytic work, which targeted survival pathways that aren't unique to senescent cells. The GPX4 dependency is, at least in current evidence, a more specific handle.
Professor Jesus Gil's broader research agenda gives context to where this fits. His lab has also investigated whether the immune system can be turned against senescent cells: whether inhibiting SMARCA4, a chromatin-remodelling protein, might make senescent cancer cells visible to the immune system's natural killer cells. These are complementary approaches to the same problem. None of them, individually, is a finished therapy. Together, they suggest that the field is now past the point of debating whether senescent cells matter and into the harder work of figuring out how to clear them without causing new problems.
This connects to something larger. The senolytic field has developed almost entirely within the longevity space, where the goal is to clear senescent cells that accumulate with age and drive chronic inflammation, tissue dysfunction, and age-related disease. The work Lisa has covered previously on the longevity revolution documented that trajectory: from the discovery that cellular senescence contributes to the aging phenotype, to the first demonstrations in mice that clearing senescent cells could extend healthspan. That framework is now being applied, with new urgency, to a much more immediate clinical problem. Cancer patients are not waiting to grow old. They're finishing chemotherapy this week, and the cells left behind are already making decisions.
What They're Trying to Clear
After the Last Infusion
D'Ambrosio talks about the problem the way someone does who has been staring at it from close range for a long time. The conventional picture of chemotherapy success is that the drugs work, the tumour shrinks, and the patient gets better. What she has spent her research career examining is the gap between the end of treatment and the return of the disease. That gap is not empty. It's populated by cells that survived something that was supposed to kill them, and that are doing something consequential with that survival.
The clinical question that follows from this research is whether cancer patients could, in the future, receive a course of senolytics after their chemotherapy ends. The sequencing matters. You use chemotherapy first to stop the tumour growing. The senescent cells appear. Then you use a senolytic compound to clear those cells before their SASP signals can rebuild the tumour-friendly environment. Two phases of treatment, targeting two different populations of cells at two different moments in the disease.
This isn't a protocol anyone is running yet. The experiments in mice are promising, but the translation to human cancer biology is always where the hard work begins. The specificity of GPX4 targeting needs to be confirmed in human cancer cell lines, then in organoids, then in early safety trials. Professor Gil has already flagged the next question: whether the drugs that clear senescent cells also improve the anti-tumour immune response, activating the T cells and natural killer cells that are part of the body's own defences. If clearing zombie cells also unmutes the immune system, the therapeutic window gets considerably more interesting.
There is also the question of timing and patient selection. Not all chemotherapy regimens induce senescence equally. Not all tumours develop the same SASP profile. The GPX4 vulnerability may be more pronounced in some cancer types than others, or in patients whose tumours express certain genetic signatures. Gil's lab has noted that if a patient undergoing chemotherapy overexpresses GPX4 in their residual cancer cells, that specific subset might be the most responsive to this approach.
What the April 2026 paper accomplishes is making the case that the zombie cell problem is not just a biological curiosity. It's a mechanism with a target. The field has spent ten years demonstrating that senescent cells matter; it is now beginning the work of doing something specific about them in a cancer context. That's a different kind of progress. It doesn't have the headline simplicity of a drug that kills tumours directly. But for the patients who finish treatment and wait, scan after scan, to see whether the cancer stays gone, a therapy that clears the cells creating the conditions for recurrence might matter just as much.
D'Ambrosio is still looking at the cells that survived. The question she and her colleagues are now pursuing is whether a drug can be designed that finds those cells specifically, strips away the protein that keeps them alive, and lets the biochemistry they've been holding back finally run its course. In the mice, it worked. The cells fell. The tumours shrank.
Whether that translates is the only question left that matters.
Buy me a coffee