Posted September 4, 2013
Aaron Newman, Ph.D., Stanford University
Ash Alizadeh, M.D., Ph.D., Stanford University
In Fiscal Year 2011, Dr Aaron Newman, under the mentorship of Dr. Ash Alizadeh, proposed a project to understand and predict patients' response to anti-cancer therapies to the Peer Reviewed Cancer Research Program Visionary Postdoctoral Fellowship Award. This award is intended to fund research of the best and the brightest postdoctoral fellows in laboratories of early-career mentors. Drs. Newman and Alizadeh exemplified the intention of the award mechanism to fund the best and the brightest while investing in excellent research to further our understanding of cancer and treatment. Dr. Newman discusses his study and his outstanding results thus far.
What is the most important thing that stakeholders should know about your research?
The overarching goal of our research is to develop robust models that predict which patients will respond well to current and emerging anti-cancer therapies and which patients will not. To do this, we are designing new computational methods to determine how differences in tumor genetics and the immune system underlie complex variation in clinical outcomes. We are particularly interested in predictive models that combine diverse types of patient-specific data, such as gene expression profiles (GEPs) generated from bulk tumor samples and genetic variants. Such models increase the chances of identifying clinically relevant variables.
Tumors are generally composed of diverse cell types, including infiltrating immune cells derived from the circulation and/or adjacent tissues. In modeling heterogeneity across patient outcomes, we sought to test whether specific tumor-infiltrating immune cells (TICs) are associated with differential response to therapy. As such, we developed a novel computational approach to quantify distinct types of immune cells from bulk tumor GEPs (see Figure, left - below). Unlike strategies that generally require living cells as input (e.g., flow cytometry), our method allows both fresh and archival tumor specimens to be analyzed with relative ease and high-throughput. With this new approach, we obtained preliminary results suggesting that certain immune cell types are significantly associated with improved survival in follicular lymphoma (FL) patients receiving personalized vaccination therapy (see Figure, right - below). Since the same TIC signatures in patients receiving control therapy had no association with survival, this finding is consistent with a predictive (and not simply a prognostic) relationship.
How did you arrive at this information or approach?
A recently completed phase III clinical trial testing antitumor vaccination for FL patients failed to meet its primary endpoint (NCT00017290). However, those patients that received a vaccine and mounted a specific immune response had significantly better outcomes than patients in the control arm. Using data from this trial, we originally proposed to develop new tools of genetic analysis to better understand the basis for heterogeneity in patient outcomes and to build predictive models of response to personalized antitumor vaccination. Since the immune system determines patient response to vaccination, we hypothesized that levels of specific tumor immune cells may be predictive of therapeutic outcomes. To address this hypothesis and take advantage of GEPs already generated from a subset of these patients' tumors, we devised a new computational method to estimate the fractions of specific TICs using bulk tumor GEPs.
What is the next step to bringing your research closer to helping cancer patients?
To confirm and extend our initial findings, we will first apply our cell type "unmixing" method to GEPs from a validation cohort of FL patients from the clinical trial. We will then integrate these data with patient-specific genetic variants predictive of therapeutic response. Related to the latter, we recently identified a molecular feature of patient immunoglobulins that is significantly associated with progression-free survival in the vaccination arm, but not the control arm. If validated, we will translate our findings to FL patients, including members of the military and their families, by initiating a new clinical trial to test the effectiveness of personalized vaccination on those most likely to benefit from it. In addition, we will apply our analytical framework to the identification of predictive biomarkers of therapeutic response in other malignancies (e.g., chemotherapy and immunotherapy in other blood cancers as well as in solid tumors).
Is there any other information that you think readers should know about your work?
To our knowledge, our work is unique in its approach to quantitate levels of immune cells in tumor GEPs and relate them to therapeutic outcomes. Moreover, models that robustly predict response to therapy for FL patients have not yet been reported.