Beyond Cell StateWhy Biological Context Shapes Therapeutic Response

Cell state is conditional

Molecular profiles help us describe a cell at a particular moment. They do not, by themselves, reveal which environmental signals produced that state, how stable it is, or whether it will persist during treatment.

Single-cell analysis of glioblastoma identified recurring malignant states and substantial plasticity among them. Genetic alterations influenced the distribution of those states, while the tumor microenvironment also shaped which programs were expressed [1]. A study of pancreatic cancer reached a related conclusion by comparing patient tumors with matched organoid models. Cell-state heterogeneity changed outside the native environment, and experimentally changing culture conditions shifted both state and drug sensitivity [2].

Genotype and lineage constrain the range of states a cell can occupy. The surrounding biology helps determine which state is present when treatment begins and how readily the cell moves into another one.

The extracellular matrix is part of the experiment

The phrase extracellular matrix can make the surrounding material sound like scenery. For a cell, it is an active source of biochemical and mechanical information. Cells sense matrix composition, stiffness, degradability, and time-dependent behavior, then adjust adhesion, cytoskeletal tension, signaling, metabolism, migration, and gene expression.

A 2026 study using engineered 3D glioblastoma models illustrates why apparently similar matrices can produce different biology. In hydrogels with matched stiffness and different stress-relaxation behavior, slow-relaxing matrices promoted cell clustering and were associated with increased P-selectin expression, nascent extracellular-matrix production, lipid-droplet storage, and altered cytokine secretion [3]. Matching one familiar mechanical parameter did not equalize the biology because the cells also responded to how the material relaxed over time.

A separate study connected matrix context directly to treatment response. Patient-derived pancreatic cancer organoids developed greater resistance to clinically relevant chemotherapies when cultured in engineered matrices with higher, tumor-relevant stiffness. Transferring the organoids to a softer matrix reduced that resistance, indicating that the surrounding matrix helped maintain the response [4].

These studies reach beyond the familiar observation that cells behave differently in 3D. They show that specific material properties can help create, maintain, or reverse a disease-relevant response. Matrix design is therefore part of the therapeutic experiment.

Neighboring cells can redirect disease state

Tumor cells also interpret treatment inside cellular communities. Immune cells, neurons, glia, stromal cells, and vascular cells exchange soluble signals with them, establish direct contacts, and remodel a shared environment. Removing these relationships from a model can remove a source of disease behavior and therapeutic resistance.

Glioblastoma makes this influence difficult to ignore because its malignant cell states are plastic and spatially heterogeneous. Analysis of patient tumors and model systems showed that macrophage-derived signals can induce transitions toward mesenchymal-like states in glioblastoma cells [5]. The state observed in the malignant population reflected its interaction with the immune neighborhood as well as tumor-cell-intrinsic programs.

Spatial arrangement further changes which interactions can occur. In a spatial biofabrication study, human neural organoids and patient-derived diffuse intrinsic pontine glioma organoids were positioned in controlled arrangements to form assembloids. These models made it possible to study tumor infiltration at neural–tumor interfaces. In a proof-of-principle experiment, the apoptotic response to panobinostat varied with the tumor’s metastatic state and the regional identity of the neural organoid incorporated into the assembloid [6].

Composition and architecture should follow the mechanism under study. An experiment centered on immune-mediated state transitions needs a different cellular neighborhood from one designed to isolate a tumor-cell-intrinsic response.

Apparent sensitivity depends on how it is measured

Biological context can change the underlying response. Experimental conditions can also change its apparent magnitude. This distinction matters when drug effects are compared across models with different baseline growth rates.

Conventional viability measures can make a slowly dividing population appear less drug-sensitive than a rapidly dividing one, even when the drug produces a comparable effect on proliferation. Hafner and colleagues showed that growth-rate inhibition metrics reduce this confounding by relating treated-cell expansion to the growth of matched controls [7]. Seeding density, assay duration, and baseline proliferation are consequently part of the interpretation, rather than minor technical details.

The issue becomes more pronounced in complex human models, where cell populations may grow at different rates or respond through changes in morphology, invasion, or differentiation without immediately changing total viability.

A drug-response value is meaningful only in relation to the biological and experimental conditions under which it was produced.

Complexity should be used as an experimental variable

A more elaborate model is not automatically a more informative one. Added matrix components, cell populations, or spatial features are useful when they represent a biological hypothesis that can be tested.

The strongest contextual experiments make the comparison explicit. A mechanical property can be changed while ligand presentation is held constant. Tumor cells can be studied with and without a defined neighboring population. The position of two tissues can be controlled while their cellular composition remains the same. These comparisons help separate a contextual effect from ordinary experimental variation.

This approach also makes model limitations easier to describe. A system can be appropriate for studying matrix-mediated resistance without claiming to reproduce an entire tumor, immune system, or patient.

Context changes which therapeutic questions can be answered

A contextual model can reveal whether a target remains relevant in the presence of stromal or immune signals, whether a compound suppresses growth while leaving an invasive population intact, or whether an apparently resistant state depends on a reversible feature of the environment.

These questions support different decisions from a conventional rank order of compound potency. They can identify environmental dependencies, distinguish cytostatic effects from broader phenotypic changes, and expose responses that disappear when a model is simplified.

Computational analysis can compare these multidimensional outcomes across experiments. Its conclusions remain bounded by the biology represented in the models and by the quality of the comparisons used to generate the data.