Immunotherapy is a cutting-edge cancer treatment that works by recruiting and activating a patient’s own immune system in a way that enables the immune system to recognize and kill cancer cells and tumors. While this exciting treatment significantly benefits some patients, it often fails in others, creating a substantial challenge for front-line physicians who are tasked with identifying which patients are likely to receive benefit from this life-saving therapy. This is a serious problem that must be overcome in order to maximize the therapeutic benefit in as many patients as possible, ideally in a way that also suggests treatment strategies that can prime the immune system in patients in which the therapy is expected to fail, in order to increase the likelihood of therapeutic success in these more challenging cases.
In an article recently published in Science Advances, Houston Methodist researchers Professor Vittorio Cristini, director of the Mathematics in Medicine Program, lead author Dr. Joseph Butner, and Professor Zhihui Wang, in collaboration with other researchers from the Houston Methodist Cancer Center and MD Anderson Cancer Center, describe a recently-developed mechanistic mathematical model for predicting the likelihood of patient-specific tumor response to checkpoint inhibitor immunotherapy.
Generally speaking, mathematical modeling uses known laws of physics, chemistry, and mathematics to design equations to describe complex biological systems. Often these important parameters cannot be directly measured in patients, but mathematical modeling overcomes this limitation by creating ways they can be calculated from other measurable quantities. Dr. Cristini and colleagues have designed this model such that all needed measurable inputs are already measured in cancer patients, in an effort to deliver a readily implementable quantitative tool to clinical practice. Moreover, because these are measured uniquely in each individual patient, the model is then able to calculate a predicted response on a per-patient basis, providing a framework for engineered individual treatment strategies; this makes their model extremely unique.
To test their model’s ability to accurately and reliably predict patient response to immunotherapy, the researchers obtained CT-scan imaging data of tumors from before, during, and after immunotherapy in 121 patients treated with checkpoint inhibitor immunotherapy, that were then analyzed using the model to quantify patient-specific indicators of therapeutic response. They found that two model-derived measures (i.e., “mathematical markers” that correlate in a non-linear form with one or more biologically or clinically relevant parameters, such as anti-tumor immune state and tumor kill rates) were able to identify patient response and survival with high accuracy.
These mathematical markers cannot be simply determined through current (more traditional) statistical, data mining, or machine learning approaches, making this concept highly innovative, bold, and practical in the field of cancer immunotherapy. They then further validated these results with data from 124 additional patients treated with either anti-CTLA4 or anti-PD1/PDL1 monotherapies (these are the most common checkpoint inhibitor immunotherapies). These results demonstrate convincing evidence that the model and its parameters may be broadly applicable to many cancer and immunotherapy combinations.
The team is currently investigating other ways the model may be implemented into the clinic by using data that may be obtained from microscopic analysis of cells obtained from needle biopsies of tumors. These may be obtained at or before the start of immunotherapy treatment, potentially allowing for outcome prediction at times before the treatment is even started. The model is now being prepared for inclusion in an upcoming clinical trial at MD Anderson Cancer Center (with Dr. James Welsh) to provide clinicians additional prognostic and decision-making information, a collaborative effort with clinicians at both the Houston Methodist Cancer Center and MD Anderson Cancer Center, thereby demonstrating a potential for direct clinical translation not often achieved in the field of interdisciplinary quantitative cancer research.
Written by: Joseph D Butner, Zhihui Wang, Vittorio Cristini
