News

Math and medicine join forces to solve a cancer mystery

21 Aug 2026
Math and medicine join forces to solve a cancer mystery

Immunotherapy has transformed outcomes for many cancer patients. Unlike conventional treatments, these therapies do not attack tumours directly.

Instead, they enable the body’s own immune system to identify and kill cancer cells that had previously escaped elimination.

For patients with advanced melanoma – a particularly aggressive form of skin cancer – PD-1 blockade immunotherapy has converted what were once terminal diagnoses into cases of long-term survival.

Patients who respond successfully and remain cancer-free for three years have a greater-than-95 percent probability of living at least 10 more years.

Unfortunately, favourable outcomes of this kind remain the exception. Approximately 7 out of 10 melanoma patients treated with PD-1 blockade immunotherapy experience disease recurrence, even when the treatment initially appeared to be effective.

Determining why immunotherapy loses efficacy and identifying strategies to address that failure has been among the most challenging problems in cancer research.

It has also been among the most resource-intensive to investigate, as the conventional approach is both slow and incremental: Researchers typically design experiments to test one hypothesis at a time, a process that can span years and require substantial funding.

A research team at the University of California, Irvine pursued a different approach.

Rather than testing numerous potential explanations sequentially in the laboratory, the members first constructed a mathematical model capturing the key interactions between tumour and immune cells and then used that model to identify the most probable mechanism of resistance.

Their findings were published in the journal Cancer Research. “We integrated two entirely distinct disciplines mathematics and biology and enabled them to inform one another,” said Francesco Marangoni, a cancer immunologist, assistant professor of physiology and biophysics, and one of the study’s senior investigators.

He collaborated closely with co-senior investigator John Lowengrub, a Distinguished Professor of mathematics, biomedical engineering and systems biology, as well as Rachel Sousa, the study’s first author and a graduate student researcher in the interdisciplinary mathematical, computational and systems biology programme.

Building a tumour inside a computer

Within a tumour, multiple types of immune cells interact. Effector T cells are the immune system’s primary weapon against cancer, actively seeking out and destroying malignant cells.

Regulatory T cells known as Tregs serve a different purpose: They normally act as a restraint on the immune system, preventing it from mistakenly attacking healthy tissue.

In the context of cancer, however, Tregs can end up protecting the tumour, suppressing the effector T cells that should be fighting it.

Tumour cells add another layer of defence by displaying a protein called PD-L1 on their surface. This protein binds to a receptor called PD-1 on effector T cells, sending an inhibitory signal that shuts down their activity.

PD-1 blockade immunotherapy thwarts this interaction, thereby reactivating effector T cells and allowing them to resume their attack on the tumour.

However, the treatment can also have an unintended consequence: It can simultaneously strengthen Treg activity, helping to restore the immunosuppressive environment it was designed to overcome.

The UC Irvine researchers translated these cellular interactions into a system of mathematical equations grounded in decades of published cancer research.

They then validated the model against experimental data from mice with melanoma, refining it until its predictions aligned with observed outcomes.

Once the model demonstrated sufficient accuracy, the team used it to generate 342 virtual mice with melanoma computational simulations in which each behaved somewhat differently, reflecting the natural biological variation observed across a real population.

Identifying the key variable

Using the validated model, the researchers simulated PD-1 blockade immunotherapy across all 342 virtual mice and compared those that responded favourably with those that experienced disease recurrence.

Of more than 30 biological parameters incorporated into the model, one emerged as the most consistent distinguishing factor between the two groups: the speed at which Tregs were entering the tumour.

“The mathematical analysis pointed directly to one variable,” Sousa said. “It indicated that the rate of Treg infiltration into the tumour was the critical factor.”

To validate this model prediction experimentally, the team engineered mice in which Tregs inefficiently migrated into the tumour while the rest of the immune system remained intact, and then they treated these animals with PD-1 blockade immunotherapy.

The combined intervention substantially outperformed PD-1 blockade alone nearly doubling survival duration and slowing tumour growth in mice whose cancer was not fully eliminated.

“The agreement between the model prediction and experiments is exciting because it points toward a promising new strategy for improving PD-1 blockade immunotherapy outcomes,” Lowengrub said.

“It validates our approach and demonstrates that mathematical modelling can not only forecast biological outcomes with precision but also identify critical and previously unrecognised targets for improving cancer treatment.”

Why this matters for patients

This study does not represent a treatment immediately available to patients. What it provides, however, may be of greater long-term value: a faster, more cost-efficient framework for determining which therapeutic strategies merit further development.

Rather than investigating dozens of potential mechanisms of drug resistance sequentially a process that can require several years and millions of dollars researchers now have a tool capable of narrowing the field of candidates before any large-scale experimental studies are initiated.

For melanoma specifically, this research reinforces interest in a therapeutic strategy already under exploration: Rather than focusing exclusively on further enhancing cancer-killing effector T cells, clinicians could simultaneously target Treg infiltration of the tumour and employ PD-1 blockade immunotherapy.

Earlier attempts at Treg inhibition were complicated by off-target effects on beneficial immune cells. This study indicates that more selective interventions those capable of targeting only the tumour-protective Treg population could be a more promising direction for future clinical investigation.

Equally significant, the mathematical model itself is not a single-use tool. It provides researchers with a reusable platform for testing the efficacy of other treatments individually or in combination and can be utilised to explore other mechanisms of therapy resistance.

Article: Mathematical and Mouse Models Identify Regulatory T Cell Influx as A Key Determinant of Acquired Resistance to PD-1 Immunotherapy

Source: University of California - Irvine