30 January 2018. Seven research teams are receiving grants to study complex problems in cancer immunotherapy with large-scale data sets, and assistance with machine learning and artificial intelligence provided by Microsoft Corp. The teams will divide $11 million in funds from the cancer foundation Stand Up to Cancer, with additional financial support provided by the Lustgarten Foundation for Pancreatic Cancer Research and Society for Immunotherapy of Cancer.
The 7 groups are taking part in Stand Up to Cancer’s Convergence 2.0 program that aims to discover more about highly variable responses by individuals to cancer immunotherapies. The work involves analysis over 3 years of many terabytes of data in multiple data sets from genomic sequencing, imaging studies, medical and medication records, among other sources. The analyses are expected to provide answers to questions about an individual’s response to specific immunotherapies, covering factors such as DNA mismatch repair, tumor-specific proteins from mutated genes, cytokine function, and natural killer cells.
Microsoft is providing its expertise in machine learning, a type of artificial intelligence using algorithms that read and analyze data, then making and adjusting their inferences as new data are encountered. “Through the collaboration with Microsoft and using well characterized, de-identified patient data,” says Arnold Levine, chair of the Convergence Initiative, “we will look at how individuals vary in their immune responses to an array of therapies. And we will create standardized measurement protocols that may speed the development of cancer therapies and ultimately save lives.”
The 7 research teams are …
Computational Deconstruction of Neoantigen-TCR Degeneracy for Cancer Immunotherapy. Leaders: Benjamin Greenbaum, Icahn School of Medicine at Mount Sinai Hospital and Vinod Balachandran, Memorial Sloan Kettering Cancer Center. By looking at the few individuals who survive pancreatic cancer for long periods of time, the team already identified an initial set of high-quality neoantigens, or protein tags, on cancer cells that the immune system recognizes. This project will continue the team’s work to understand what makes a high-quality neoantigen and how the microbiome influences how the immune system recognizes it, with the goal of developing a method for creating vaccines to treat pancreatic cancers.
Single-Cell Functional Multi-Omics to Characterize and Monitor CAR T Therapy. Leader: Rong Fan, Yale University. Immunotoxicity and autoimmune-like response is a significant problem hindering widespread use of cancer immunotherapies. To understand this response, the full spectrum of cytokine functions in a pre-infusion setting will be assessed and matched with patient responses to treatment with the chimeric antigen receptor (CAR) T-cell therapy, with infused CAR T-cells assessed to determine the mechanisms of efficacy or immune toxicity. The team also proposes identifying molecular characteristics underlying therapeutic efficacy and toxicity of CAR T therapy, and biomarker discovery by using computational models and machine learning that look at the data.
Integrating Experimental and Computational Pipelines to Develop Biomarkers of Tumor Cell Resistance to Natural Killer Cells. Leader: Constantine Mitsiades, Dana-Farber Cancer Institute. Natural killer cells are white blood cells that play a role in viral response, and have potent anti-tumor cell killing properties, but are not inhibited by many of the strategies cancer cells use to evade the immune system. The team plans to use computational methods to identify biomarkers responsible for tumor cell sensitivity or resistance to natural killer cells, and Crispr genome editing to devise screens to explore interaction patterns in responses by natural killer cells. Biomarkers will be validated against organoids derived from patients and quantified tumor cell responses.
Correlating Immunological Health to Cancer Susceptibility. Leader: Mark Davis, Stanford University. The team plans to build on a preliminary analysis of 450 adults showing the immune response to influenza vaccine may be correlated with cancer susceptibility, including some elderly participants receiving the influenza vaccine and a less robust immune response more likely to develop and succumb to cancer. The researchers expect to correlate repertoires of T- and B-cells from the immune system to cancer susceptibility in individual patients, and compare the results to a separate group of immune-deficient patients.
Connecting Immune Health and Tumor Biology in Gynecologic Cancers. Leader: John Wherry, University of Pennsylvania. Mismatch repair-deficiency is the inability to repair base pairing within DNA, which can weaken the DNA structure and lead to accumulation of mutations. The response to immune checkpoint inhibitors has been varied in gynecologic cancers, possibly due to the number of mutations carried by each tumor cell, known as mutational burden. The researchers plan to initiate two clinical trials that will test if (a) tumor-intrinsic factors affect the response to checkpoint inhibition; (b) baseline immune function and quality affects response to checkpoint inhibition; and (c) on-treatment blood markers may reflect the tumor-immune interaction.
Responders and Non-Responders to Endometrial Cancers with Mismatch Repair. Leader: Alessandro Santin, Yale University. The researchers will study reasons for only half of patients with mismatch repair deficient metastatic or recurrent tumors responding to anti-PD1, or checkpoint, treatments. The team plans to devise computational methods to assess DNA damage and tumor mutational burden using next-generation sequencing data, and once validated, apply the technique to a current clinical trial of the immunotherapy drug pembrolizumab (Keytruda) in patients with endometrial cancer, affecting the lining of the uterus.
Using Artificial Intelligence to Predict Molecular Pathways and Clinical Outcomes. Leader: Ernest Fraenkel, MIT. Despite recent successes and rising hopes for cancer immunotherapy, many serious challenges confront these techniques before they can be expanded to more types of cancer and patients. In addition, there are few established methods for determining which patients are most likely to benefit from specific cancer immunotherapies. The researchers will propose solutions for integrating data from diverse sources such as lab results and text in electronic medical records with techniques such as machine learning and natural language processing.
More from Science & Enterprise:
- Biotechs Partner on HPV Cancer Immunotherapies
- CAR T-Cells Shown Effective in Long-Term Studies
- Genome Editing Enhances CAR T-Cells to Fight Cancer
- Alliance Exploring Microbiome on Cancer Immunotherapy
- Cancer Center, Analytics Company Partner on Precision Care
* * *
You must be logged in to post a comment.