Cancer as Evolution: The Ecosystem

The previous two posts focused on cancer cells: how they form a diverse population, how treatment selects among them. But cancer cells do not evolve in isolation. They evolve inside a tissue, surrounded by other cell types, embedded in a physical structure, bathed in chemical signals. The tumor microenvironment actively participates in cancer’s evolution, shaping which cancer cells thrive, which die, and which paths of adaptation remain open.

The same dynamics that govern ecosystems in nature operate inside a tumor, and understanding them changes how we think about treatment.

A growing tumor actively remodels the tissue around it, constructing a microenvironment tailored to its own survival.

Cancer cells secrete vascular endothelial growth factor (VEGF) and other angiogenic molecules to recruit new blood vessels. The resulting vasculature is disorganized and leaky, nothing like the orderly capillary networks in healthy tissue. It delivers oxygen and nutrients unevenly, creating steep gradients within the tumor. Some regions are well-perfused. Others are hypoxic, acidic, and nutrient-deprived. These gradients impose different selective pressures on different parts of the same tumor, driving the spatial heterogeneity that our second series documented with spatial transcriptomics and multiplexed protein imaging.

Cancer cells reprogram the fibroblasts that normally maintain tissue architecture. Normal fibroblasts become cancer-associated fibroblasts (CAFs), which deposit dense extracellular matrix, secrete growth factors that feed the tumor, and produce cytokines that suppress immune responses. Single-cell sequencing, as described in our second series, revealed that CAFs are not one cell type but several, each playing a different role: some build physical barriers, some suppress immunity, some signal directly to cancer cells.

Cancer cells manipulate the immune system. They express checkpoint ligands like PD-L1 to exhaust T cells. They recruit regulatory T cells and myeloid-derived suppressor cells that actively dampen anti-tumor immunity. They secrete TGF-beta, IL-10, and other immunosuppressive cytokines that convert the local immune environment from hostile to permissive. The immune landscape of a tumor is the patient’s immune system as reshaped by the cancer.

Each of these modifications feeds back into the evolutionary dynamics. The abnormal vasculature creates hypoxic niches that select for metabolically adapted, often more aggressive cancer cell clones. The fibrotic stroma creates physical barriers that block immune infiltration, providing protected zones where cancer cells evolve free from immune pressure. The immunosuppressive milieu relaxes a major selective constraint, allowing clones that would otherwise be eliminated by the immune system to persist and expand.

The cancer cell population and its microenvironment co-evolve. The cancer shapes the environment, and the reshaped environment shapes which cancer cells succeed. This feedback loop is central to why tumors become progressively harder to treat over time: the ecosystem the cancer builds reinforces the very features that make it dangerous.

Multiple populations coexist within a tumor, and their interactions follow the same principles that govern species interactions in any ecosystem: competition, cooperation, exploitation, and niche partitioning.

Competition. Cancer cell subclones compete for space, oxygen, and glucose. A fast-growing clone can physically crowd out slower neighbors. A clone with better access to blood supply starves adjacent competitors. This competition constrains the population. The dominant clone suppresses minority clones not by killing them directly but by outcompeting them for shared resources. The minority clones persist at low frequency, held in check by competitive pressure.

This has a critical therapeutic implication: kill the dominant clone and you release that competitive pressure. The minority clones, previously suppressed, suddenly face an open niche with abundant resources and expand rapidly. If those minority clones happen to carry drug resistance, their expansion fills the tumor with resistant cells faster than if the dominant sensitive clone had been left partially intact. This phenomenon, competitive release, is one of the most important ecological dynamics in cancer treatment and one of the least intuitive.

Cooperation. Not all interactions between clones are competitive. Some cancer cell subpopulations produce growth factors that benefit neighboring cells, including cells of different genotypes. A clone that secretes a diffusible angiogenic signal improves blood supply for all nearby cells, not just itself. A clone that produces immunosuppressive cytokines shields its neighbors from immune attack regardless of their genotype.

This creates a problem for natural selection within the tumor. A cooperating cell bears the metabolic cost of producing the shared resource. A neighboring cell that does not produce it, a “cheater,” receives the benefit without paying the cost. Classic evolutionary game theory predicts that cheaters should outcompete cooperators and drive cooperation to extinction. Yet cooperation persists in tumors. The likely explanation involves spatial structure: cooperating cells tend to cluster together, so the benefits of cooperation flow primarily to nearby relatives who share the cooperating genotype. The spatial architecture documented in our second series is what makes these dynamics possible.

Niche partitioning. Different subclones can coexist stably by occupying different niches within the tumor. One clone thrives in well-oxygenated regions near blood vessels. Another tolerates the hypoxic core. A third colonizes the invasive margin where it encounters different stromal and immune conditions. Rather than competing to extinction, these clones specialize for their respective niches and persist in parallel. The tumor becomes a mosaic of locally adapted subpopulations, each occupying a different microenvironmental niche.

This helps explain a finding from multi-region sequencing studies: subclones that appear to be distinct evolutionary branches sometimes coexist for extended periods without one displacing the other. Rather than competing for the same niche, they partition the ecosystem.

Cancer, viewed through this lens, is a population of cells embedded in a co-evolving ecosystem of stromal cells, immune cells, vasculature, extracellular matrix, and signaling molecules, all interacting through competitive, cooperative, and exploitative relationships that determine the trajectory of the disease.

The spatial and cellular measurements from our second series capture exactly this ecological structure. An immune-excluded zone reveals a protected niche. A tertiary lymphoid structure marks an organized ecological response. A cancer-associated fibroblast population reflects active construction of the ecosystem in which the cancer sustains itself. Each spatial feature those technologies resolve encodes an ecological relationship that shapes evolution.

Treatment intervenes in this ecosystem. Every drug, every radiation field, every immune-modulating agent alters the ecological balance. The previous post described how treatment selects among cancer cell clones. This post adds the second layer: treatment also reshapes the microenvironment in which that selection occurs. Killing the dominant clone releases competitors into open niches. Damaging the vasculature creates new hypoxic zones that favor different cell populations. Immune-modulating agents impose selective pressures that the entire ecosystem must absorb.

The dynamics of this system unfold in time, and the speed at which they unfold, relative to our ability to observe and respond, is the subject of the next post.

Next in the series: The Tempo

The Cancer as Evolution series is published by Galen Health, where we are building cancer superintelligence: autonomous, self-directed AI systems for cancer research and discovery.