Colorectal cancer (CRC) has shifted from a disease typically associated with aging to a pressing crisis for the young. As the second leading cause of cancer-related death in the United States, it now holds the grim distinction of being the primary cause of cancer mortality among adults under 50. Despite the transformative power of immunotherapy in modern oncology, the vast majority of CRC patients remain untouched by these advancements.
Dr. Karin Pelka, an investigator at the Gladstone Institutes and a recipient of the Cancer Research Institute (CRI) Technology Impact Award, is spearheading a high-stakes scientific effort to bridge this gap. By utilizing artificial intelligence to eavesdrop on the "molecular conversations" occurring within tumor microenvironments, her lab aims to turn "cold," resistant tumors into "hot," immune-responsive ones.
The Landscape: Why Colorectal Cancer Defies Modern Treatment
The Immunotherapy Gap
Immunotherapy relies on the immune system’s ability to recognize a tumor as a foreign invader. In cancers such as melanoma or non-small cell lung cancer, the tumor often harbors a high mutational burden. These thousands of DNA mutations act as red flags, allowing the immune system to identify and attack the malignant cells. Checkpoint inhibitors—the current gold standard—function by "releasing the brakes" on these pre-primed immune cells, allowing them to destroy the tumor.
Colorectal cancer, however, is notoriously stealthy. The vast majority of CRC cases do not possess the high mutational load required to trigger this natural immune recognition. "The vast majority of colon cancer falls into this relatively large area of tumors where we haven’t quite figured out yet how to use the immune system to fight them," Dr. Pelka explains.
The dMMR Exception
A small but vital subset of patients—roughly 5% to 10%—carries tumors with a DNA mismatch repair (dMMR) defect. These tumors are "immunologically hot." In a landmark 2022 clinical trial, patients with locally advanced rectal cancer carrying this specific mutation experienced a 100% response rate to immunotherapy alone, entirely bypassing the need for toxic chemotherapy, radiation, or disfiguring surgery. While this is a medical miracle for a few, it underscores the urgent need to unlock similar pathways for the remaining 90% of patients.
A Chronology of Discovery: From Observation to AI Integration
The trajectory of CRC research has undergone a seismic shift in the last decade.
- 2010–2015: The Rise of Checkpoint Inhibitors: The medical community witnessed the initial success of immunotherapy, which primarily focused on cancers with high mutational burdens.
- 2018–2020: The "Cold Tumor" Problem: As clinical data matured, researchers realized that most gastrointestinal cancers were "cold"—they excluded immune cells, effectively hiding from the body’s defenses.
- 2022: The Paradigm Shift: The NEJM trial results for dMMR rectal cancer patients provided a "proof of concept" that if a tumor can be made visible to the immune system, the need for traditional, harsh treatments could be eliminated.
- 2024–2026: The AI Era: Researchers like Dr. Pelka began utilizing spatial transcriptomics and foundational AI models to map the "immune hubs" that determine why some patients respond while others do not.
Supporting Data: Mapping the Immune Hubs
Dr. Pelka’s research centers on the discovery of "immune hubs"—structured clusters of immune activity within the tumor landscape. These hubs serve as a battlefield where killer T cells, cancer cells, and structural tissue engage in continuous, complex signaling.
Predictive Power
The dynamics within these hubs are crucial. When these hubs exist in a specific "activated" state, they act as a biological barometer for immunotherapy success. Crucially, the Pelka Lab has found that these hubs are not unique to colorectal cancer; they appear conserved across melanoma and lung cancer as well. This suggests that the biological machinery for an immune response is present in many patients, but it is currently being suppressed or misdirected. The current research question is simple yet profound: What is preventing these hubs from forming or maintaining their activity in patients who do not respond to existing immunotherapies?
Technological Innovation: Teaching AI to Read the Cell
To decode the complexity of these immune hubs, the Pelka Lab is moving beyond traditional microscopy. They are applying Geneformer, a foundational AI model developed in collaboration with Dr. Christina Theodoris.
How Geneformer Works
Think of Geneformer as a "large language model" for biology. Just as ChatGPT learns the rules of syntax and semantics from millions of sentences, Geneformer was trained on tens of millions of human cell profiles to understand the "language" of gene activity. By analyzing the transcriptome of individual cells, the AI can predict how those cells will behave in different environments.
Identifying the "Master Switches"
Dr. Pelka’s team is using this AI to identify "central regulators"—the master molecular switches that dictate whether a cancer cell remains invisible or reveals itself to the immune system. By layering this AI analysis over spatial transcriptomics (which tracks the exact location of cells within the tumor), the lab is effectively creating a high-definition map of the tumor’s internal social network. This level of detail allows researchers to pinpoint potential drug targets that were previously invisible to human observation.
Implications: The Future of Patient Care
The "Living Therapeutic" Advantage
The urgency of this work is underscored by the rising incidence of CRC in young adults. Conventional targeted therapies, while precise, are often outmaneuvered by cancer’s rapid ability to evolve. Immunotherapy, by contrast, offers the promise of a "living therapeutic." If the immune system can be properly programmed, it acts as a dynamic, adaptive shield that can evolve alongside the cancer, preventing recurrence in a way that static drugs never could.
The Importance of Early Intervention
While Dr. Pelka’s work focuses on the frontiers of immunology, she remains a staunch advocate for established prevention. "Screening matters," she notes. Because colorectal cancer often develops from polyps over a 10-to-15-year period, regular colonoscopies starting at age 45 remain the most effective tool for preventing the disease before it ever requires immunotherapy.
Funding High-Risk, High-Reward Science
The CRI Technology Impact Award has provided the necessary runway for this research. Dr. Pelka acknowledges the difficulty of securing traditional funding for such experimental work. "Having a foundation willing to take a bet early on—when the science is not yet ready to be employed widely—that is often research that is hard to fund," she says. "But those early discoveries are exactly what move the needle."
Conclusion: A New Horizon for Oncology
The mission of the Pelka Lab is not merely to add a few months to a patient’s life, but to fundamentally alter the prognosis for the 90% of colorectal cancer patients currently left behind by immunotherapy. By combining the biological insight of immune hubs with the analytical power of foundational AI, Dr. Pelka is closing in on the "central regulators" that keep cancer hidden.
For the young adults facing an unprecedented rise in CRC diagnoses, this research offers more than just hope; it offers a path toward a future where "terminal" no longer means inevitable. As the molecular conversations inside the tumor are decoded, the immune system may finally be given the tools it needs to finish the fight, transforming the landscape of oncology from one of reactive, toxic treatments to one of precise, long-term immunological control.
