Pharma, MedCity News eNewsletter

Can a math model help predict an immune response to cancer?

A new Nature paper has described a first-of-its-kind mathematical model for predicting how a cancer patient will benefit from checkpoint inhibitors, a class of drugs that includes Merck’s Keytruda and Bristol Myers-Squibb’s Opdivo.

Why do some immunotherapies work in certain patients and not others? It’s a multi-million dollar question, which researchers at the Icahn School of Medicine at Mount Sinai are hoping to solve with some custom computer modeling.

Their work appears in a new Nature paper, published Wednesday, titled “A neoantigen fitness model predicts tumor response to checkpoint blockade immunotherapy.”

The authors describe a first-of-its-kind mathematical model for predicting how a cancer patient will benefit from checkpoint inhibitors, a class of drugs that includes Merck’s Keytruda and Bristol Myers-Squibb’s Opdivo.

Checkpoint inhibitors target PD-L1 or PD-1 mutations, which cancers use to shield themselves from an immune response. But along the way, clinicians observed that the drugs sometimes work in the absence of these target proteins. Tumors began shrinking in patients that wouldn’t be expected to respond.

Several other variables have since been identified. Some, such as the role of certain inflammatory signals, are still being parsed out. Others are more clear-cut. Earlier this year the FDA approved Keytruda for certain solid tumors that are microsatellite instability-high (MSI-H) or mismatch repair deficient (dMMR). Both are abnormalities that affect the proper repair of DNA inside the cell (aka mismatch repair genes).

It has also been noted that tumors with a broadly high mutational burden are more likely to respond to this class of drugs. In other words, the more the cancer cells mutate, the greater the chance the immune system will recognize them as a threat and work to eliminate them. These novel mutations are known as neoantigens.

Benjamin Greenbaum, assistant professor of medicine, hematology and medical oncology, pathology, and oncological sciences at The Tisch Cancer Institute at the Icahn School of Medicine at Mount Sinai and lead author of the Nature paper, said his team’s work builds off those observations and data.

“The initial goal was to try and provide more predictive value from mutations than mutational burden alone, by capturing the underlying process of immune recognition of neoantigens,” Greenbaum said in an email forwarded by a Mount Sinai representative.However, the mathematical framework is meant to capture other biomarkers, such as the state of inflammation in the immune environment.”

The researchers analyzed data from melanoma and lung cancer patients, but the principles could help inform treatment for a range of tumor types. Greenbaum said the eventual aim is to deliver an “overall score” for how likely a patient is to respond to a checkpoint inhibitor.

“It is still in the early stages and requires further research, though of course, we hope this will ultimately be translated into patient care,” he said. “Our work shows a proof of concept that this approach can be valuable, and our aim is for it to benefit patients and ultimately be incorporated as a quantity that captures many relevant features of response, and can be expanded as new findings occur.”

Jill O’Donnell-Tormey, CEO and director of scientific affairs at the Cancer Research Institute, a nonprofit organization that invests in promising areas of cancer immunotherapy, said the Mount Sinai research addresses a major challenge in the field.

“For those not likely to respond to off-the-shelf immunotherapies — who at present comprise the majority of cancer patients — there is an urgent a need to identify cancer targets that are likely to elicit an immune response in these patients,” O’Donnell-Tormey said in an email forwarded by the nonprofit’s representative. “Methodologies such as this may prove to be useful in guiding treatment choices for patients in regard to predicting response to off-the-shelf and personalized immunotherapies as well as potential adverse events.”

A second Nature study, also published Wednesday, showed how a similar model can be used to understand the immune response in patients with pancreatic cancer who survive longer than others. This research was led by a team at Memorial Sloan Kettering Cancer Center (MSK) with support from Greenbaum and first author of the checkpoint paper Marta Luksza of the Simons Center for Systems Biology at the Institute for Advanced Study. 

Along with the modeling, the common theme in both studies is the importance of understanding when the immune system will lead to productive recognition of a tumor.

“This research represents a big step forward in understanding why some tumors are more aggressive than others and being able to predict rationally which neoantigens will be the most effective at stimulating an immune response,” said Vinod P. Balachandran, a member of the David M. Rubenstein Center for Pancreatic Cancer Research at MSK, and corresponding author of the companion study in Nature, in a news release.

Given the huge cost of cancer care in the United States, there will be a lot of interest in moving the research from proof-of-concept to application, if it is possible to fill in these oncological blind spots. To do so, Greenbaum believes more data and knowledge of the immune system will be needed

“It is in the early stages so it is hard to tell yet what the scope and limitations are,” he stated. “In that sense, the limited datasets available are the biggest caveats – we will only know the true scope of this approach when we can apply this process to more tumor types and therapies. Moreover, the model is built on the underlying processes of antigen presentation and recognition, which will benefit from continued improvement in computational tools in the coming years.”

Photo: mathisworks, Getty Images