Predicting immunotherapy response with living tumor biopsies

Predicting immunotherapy response remains one of the persistent challenges in cancer care. Two patients can have seemingly similar tumors, receive the same treatment, and experience dramatically different outcomes. One tumor may shrink or disappear. Another may continue growing.

Doctors currently rely on molecular biomarkers to estimate who is likely to benefit. But a new approach asks a more direct question: what if a sample of the patient's tumor could be exposed to an immunotherapy drug before treatment even begins?

A study published on October 15, 2026, in Cell Reports Medicine describes a method for doing precisely that, using small, living tumor fragments obtained through routine clinical biopsies.

In a preliminary clinical validation cohort, the test's predictions agreed with clinical outcomes in 26 of 28 patients, a concordance rate of 93%.

The number is striking. The more consequential scientific development, however, is the type of information the researchers were able to collect.

A tumor's molecular profile is only part of the story

Immune checkpoint inhibitors have transformed treatment for several cancers. Rather than attacking malignant cells directly, these drugs interfere with inhibitory signals that restrain the immune system.

Those inhibitory pathways serve a legitimate biological purpose: preventing excessive immune activation and damage to healthy tissues.

Cancer cells can exploit the same mechanisms to escape immune surveillance.

One familiar example is the PD-1/PD-L1 pathway. Blocking this interaction can restore the activity of certain immune cells and help them attack a tumor.

Yet this approach works well in only a subset of patients.

Clinicians use biomarkers such as PD-L1 expression, tumor mutational burden, and abnormalities in DNA mismatch repair to guide treatment decisions.

These measurements offer valuable information, but they largely describe the tumor's molecular characteristics at a particular moment.

They cannot directly show how the tumor's immune microenvironment will respond when exposed to an immunotherapy drug.

A biomarker-positive tumor may fail to respond, while a biomarker-negative tumor may still be sensitive.

The new study explores whether measuring that response directly could provide information that static biomarkers miss.

Keeping a piece of the tumor alive

The researchers, led by David Braun at Yale School of Medicine in collaboration with Elephas Biosciences, developed a functional testing platform based on living tumor fragments.

The process begins with a standard core needle or forceps biopsy.

Fresh tissue is transported to a laboratory, where it is divided into thin fragments, approximately 300 micrometers thick.

Crucially, these fragments retain more than malignant cells.

They also preserve resident immune cells and other elements of the tumor microenvironment. These cellular relationships are central to whether immune checkpoint blockade can generate an effective antitumor response.

The fragments are maintained in culture and exposed sequentially to control antibodies and immune checkpoint inhibitors.

Each treatment phase lasts approximately 24 hours.

The scientists then measure changes in cytokine production.

Cytokines are signaling proteins that immune cells use to communicate. Their changing levels can reveal whether exposure to the drug has activated or altered the immune response within the tissue.

The team developed a classification score called the elive index, based on changes in eight cytokines.

The assay also revealed something biologically interesting: samples classified as laboratory responders did not all display the same pattern of immune activation.

Four distinct cytokine-response profiles emerged, reflecting different immune states rather than a single uniform response mechanism.

The researchers reported an area under the ROC curve of 0.99 for distinguishing laboratory-defined response clusters. This figure should not be confused with clinical predictive accuracy. It measures the algorithm's ability to separate groups defined by ex vivo cytokine patterns, rather than its performance against patient outcomes.

That clinical comparison was evaluated separately.

What the 93% result actually means

The study used biopsy material collected from 228 patients with suspected or confirmed solid tumors.

The specimens represented multiple cancer types, including lung, kidney, head and neck, and bladder cancers.

But the full group of 228 patients was not used to establish the reported 93% clinical concordance.

That finding came from a validation cohort of 28 patients who subsequently received immunotherapy-containing treatment.

Among 17 patients whose tumor fragments were classified as likely responders, all 17 experienced some form of clinical benefit.

Their outcomes included complete responses, pathological complete responses, partial responses, and stable disease.

This distinction matters. Stable disease means a tumor has stopped progressing for a period of time; it is not equivalent to tumor shrinkage.

Of 11 patients classified as predicted nonresponders, nine subsequently experienced progressive disease.

Together, these outcomes produced 26 correct classifications among 28 patients.

The test also identified 10 of the 12 patients who achieved an objective response, meaning complete or partial tumor regression.

These findings are encouraging, but their interpretation depends heavily on the study design.

The validation cohort was small and clinically heterogeneous. It consisted of two groups of 14 patients, reflecting different levels of correspondence between the treatments tested in the laboratory and those administered in clinical practice.

In some cases, patients received immune checkpoint inhibitors together with chemotherapy or other treatments.

Consequently, clinical benefit could not always be attributed specifically to the drug exposure evaluated in the assay.

The study was also observational. Treatment decisions were not randomized or assigned according to the test result.

It therefore establishes an encouraging association between laboratory predictions and clinical outcomes, not evidence that using the test to select treatment improves survival or other patient outcomes.

The work was funded by Elephas Biosciences, the company developing the platform, making independent validation particularly valuable.

The patients existing biomarkers may miss

One especially interesting finding involved tumors that tested negative for established immunotherapy biomarkers.

Some of these samples nevertheless produced a positive functional response when exposed to immune checkpoint inhibitors in the laboratory.

Two biomarker-negative patients with matching laboratory and clinical immunotherapy conditions experienced clinical benefit despite their negative conventional biomarker status.

This suggests that functional testing may capture aspects of tumor biology that existing molecular measurements do not fully represent.

It does not establish that the new assay is superior to every approved biomarker, nor does it demonstrate that biomarker-negative patients should automatically receive immunotherapy.

Instead, the study points toward a potentially complementary source of information.

Traditional biomarkers describe features associated with treatment sensitivity.

A live-tissue assay measures how an intact cellular environment reacts when a treatment is actually applied.

The difference could become clinically useful, particularly when conventional biomarkers provide uncertain or conflicting guidance.

Larger prospective studies will be required to determine whether this strategy reliably changes treatment decisions and improves outcomes. But the underlying achievement is already clear: a functional immune response can be measured using the small tissue samples collected during routine diagnostic procedures.

Cancer treatment usually involves administering a drug and then waiting to see how the tumor responds.

Here, researchers were able to observe part of that response in advance, using living tissue from the patient's own tumor.

That is the central advance behind the 93% figure.