Cancer as Evolution
Cancer is one of the most feared words in medicine. It kills approximately 10 million people per year worldwide, affects one in three people at some point in their lives, and despite decades of research and billions in funding, remains incompletely understood and incompletely treatable. Why is it so hard to cure?
The answer has everything to do with what cancer actually is. Most people think of cancer as a disease — something that happens to the body, analogous to an infection. But cancer is not foreign. Cancer cells are your own cells. They carry your DNA, use your metabolic pathways, and exploit your body's systems. What makes them dangerous is not that they're alien — it's that they've undergone evolution. They have acquired, through mutation and selection, the capabilities to grow without restraint, evade death, corrupt their environment, and spread through the body. Cancer is Darwinian evolution operating at the cellular level, inside you, with your life as the selection pressure.
This reframing is not just intellectually satisfying. It has direct consequences for how we think about treatment. The most common approach — hit the tumor hard with cytotoxic therapy to kill every cancer cell — works if you can achieve total eradication. But if even a few cells survive, those survivors are by definition the most resistant. You have just run an evolution experiment that selected for the most drug-resistant variants in the population. Understanding cancer as an evolving ecosystem rather than a static target changes the entire logic of treatment.
Part I — The Darwinian logic of cancer
Darwin's conditions for evolution are simple: a population of entities with heritable variation in fitness-relevant traits, subject to selection pressure. Cancer satisfies all three. A tumor is a population of cells. Cancer cells have high mutation rates — both from the genomic instability that often drives cancer and from the selection pressure of replication — generating heritable genetic variation. And there is constant selection: cells that grow faster, resist apoptosis, evade immune surveillance, or tolerate oxygen deprivation outcompete their neighbors and leave more descendants.
The evolutionary process in a tumor is somatic evolution — evolution within the body of a single individual rather than across generations. It operates by the same rules as organismal evolution, just faster (cell generations are days rather than years) and in a different arena (a tissue rather than an ecosystem). A key difference from evolutionary biology is that the "fitness" being selected for is cell-level fitness — reproductive success of the cell — even when that fitness is lethal to the organism containing the cells. Cancer is evolution in a context where cell-level and organism-level fitness have been decoupled.
Multicellularity can be understood as a cooperative arrangement: cells agree, in evolutionary terms, to suppress their individual replication and defer to the collective. Tumor suppressor genes, cell cycle checkpoints, and apoptosis are the enforcement mechanisms. Cancer is a defector — a cell that breaks the cooperative arrangement and reverts to individual-level selection. This is the "tragedy of the commons" at the cellular level: individual defectors gain in the short run, but their proliferation ultimately destroys the resource (the body) that all cells depend on. The cancer cell "wins" at the cellular level and loses at the organismal level, taking its host with it — an evolutionary dead end, since cancer cells die with the patient.
Tumor heterogeneity — the cancer within the cancer
One of the most important insights from treating cancer as evolution is the recognition of intratumor heterogeneity — the fact that a single tumor contains not one cancer but many. Different cells within the same tumor have different mutation profiles, different gene expression patterns, different metabolic strategies, and different sensitivities to drugs. This heterogeneity has been measured by sampling multiple regions of the same tumor and sequencing each — the result is a family tree of related but distinct clones, each carrying different accumulated mutations.
Heterogeneity develops through branching evolution: a founder cell acquires a set of early driver mutations and divides. Its descendants acquire additional mutations, diverging into distinct subclones. Some subclones acquire mutations that give them a growth advantage over their siblings; others are eliminated. The result is a clonal hierarchy — some mutations are present in all cells (trunk mutations, present in the founding clone), while others are present only in subsets (branch mutations, acquired later). This matters enormously for treatment: a drug targeting a branch mutation will kill some cancer cells but leave the trunk and other branches untouched, which then expand to fill the vacancy.
How long does it take a normal cell to become a metastatic cancer? For colorectal cancer, careful analysis of mutation timing in tumors suggests the first driver mutation occurs, on average, 17 years before the cancer is diagnosed. The progression from first mutation to localized cancer takes about 17 years; from localized cancer to metastatic disease, another 2–3 years. This means that a colorectal cancer diagnosed today likely began its molecular journey around 20 years ago. The clinical relevance: the window for early detection and interception is very long. Colonoscopies detecting pre-cancerous polyps are effective because they catch the evolutionary process before it has progressed to invasive disease — removing the early-stage lesion before natural selection has had time to produce a more dangerous clone.
🤔 Why don't we all get cancer if our cells are constantly mutating?
▼Because the body has multiple redundant defense systems that detect and eliminate pre-cancerous cells. Cell cycle checkpoints halt division when DNA damage is detected. The p53 tumor suppressor monitors DNA integrity and triggers apoptosis when damage exceeds repair capacity. The immune system — particularly cytotoxic T cells and NK cells — continuously patrols for cells displaying abnormal surface markers (neoantigens from cancer-specific mutations). Oncogene activation triggers a response called oncogene-induced senescence — the very mutation that would drive proliferation causes the cell to permanently arrest instead, as a safety mechanism. For cancer to develop, typically 5–10 of these defense systems must fail simultaneously — which is why cancer usually requires decades and multiple accumulated mutations. We are not preventing cancer from starting. We are suppressing it continuously, throughout our lives. Cancer appears when that suppression eventually fails.
Part II — The tumor as an ecosystem
A tumor is not just a mass of cancer cells — it is an ecosystem. Embedded within the cancer cells is a diverse community of non-cancerous cells: fibroblasts that remodel the structural scaffolding (extracellular matrix), endothelial cells forming the tumor's blood vessels, immune cells that have been recruited and co-opted, adipocytes supplying fatty acids, and various other cell types. Together, these constitute the tumor microenvironment (TME), and they are as important to tumor survival and progression as the cancer cells themselves.
The cancer cells don't passively coexist with their microenvironment — they actively shape it. They secrete signals that recruit blood vessel formation (angiogenesis) to supply oxygen and nutrients. They secrete proteases that degrade the extracellular matrix, clearing paths for invasion. They produce immunosuppressive signals that convert what should be anti-tumor immune responses into tumor-supporting ones. Tumor-associated macrophages, recruited in large numbers by cancer cells, can be reprogrammed from a tumor-killing state (M1) to a tumor-supporting state (M2) that promotes angiogenesis, matrix remodeling, and metastasis.
This ecosystem framing explains a phenomenon that has frustrated oncologists for decades: acquired drug resistance. A tumor treated with a targeted therapy initially responds dramatically — 90% of cancer cells die. Then, months later, the tumor comes back, now completely resistant to the drug. What happened? Darwinian selection: the 10% of cells that carried pre-existing mutations conferring resistance survived and expanded. The drug, rather than curing the cancer, selected for the most dangerous subclone in the tumor. This happens with virtually every targeted therapy if used as a single agent. The solution is combination therapy — hitting multiple targets simultaneously, making it statistically impossible for any single cell to carry resistance mutations to all of them.
Adaptive therapy — exploiting the evolutionary tradeoff
Here is where the evolutionary framing generates a genuinely novel treatment strategy. The standard approach is to kill as many cancer cells as possible — maximum tolerated dose chemotherapy. But if the goal is to maximize killing, and resistant cells are selected by that very killing, perhaps the goal should be different: not maximal killing but suppressing resistance evolution.
Adaptive therapy, proposed by Robert Gatenby at the Moffitt Cancer Center, exploits a key evolutionary insight: resistance is usually costly. Resistant cells typically grow more slowly than drug-sensitive cells in the absence of drug, because the resistance mutations consume resources that would otherwise go to growth. In a tumor with both sensitive and resistant cells, the sensitive cells normally outcompete the resistant ones — the resistant subclone is kept at low frequency by competition. Treatment that kills sensitive cells removes this competition, allowing resistant cells to expand.
Adaptive therapy intentionally uses lower doses of therapy, targeting not maximum killing but a stable tumor burden — maintaining enough sensitive cells to suppress the resistant subclone through competition. Clinical trials of adaptive therapy in metastatic castrate-sensitive prostate cancer have shown that patients receiving adaptive therapy live longer with fewer side effects than those receiving continuous maximum-dose therapy — the first clinical proof of concept that treating cancer as an evolving ecosystem, and managing evolution rather than trying to eliminate it, can produce better outcomes.
"We have been trying to cure cancer by treating it like an infection — hit it hard and kill everything. But cancer is not an infection. It's us. And it fights back the way evolution always fights back: by selecting for resistance."
🤔 Could immunotherapy cure cancer?
▼For some cancers in some patients, it already has — producing complete, durable remissions that appear to be cures in patients with advanced melanoma, Hodgkin lymphoma, and certain lung cancers. Checkpoint inhibitors have produced 5-year survival in ~40% of advanced melanoma patients — a disease where previously almost no one survived five years. But immunotherapy doesn't work for everyone: roughly 20–40% of patients with responsive cancer types respond durably, while the rest don't respond or relapse. The reason is tumor heterogeneity — tumors with many mutations (high tumor mutational burden) provide more neoantigens for T cells to recognize, while tumors with few mutations (like pancreatic cancer or microsatellite-stable colorectal cancer) are largely invisible to the immune system. Combination approaches — checkpoint inhibitors plus vaccines targeting tumor-specific neoantigens, plus other immune strategies — are being tested and showing early promise. Whether immunotherapy will eventually produce cures across a wide range of cancers is the defining question of the next decade in oncology.