Fangfang Yan, MD, discusses the place of CAR T-cell therapy in the MCL treatment paradigm and how AI can address drivers of CAR T-cell therapy resistance.
Artificial intelligence (AI)–based approaches could be used to identify drivers of resistance to CAR T-cell therapies in patients with mantle cell lymphoma (MCL), according to Fangfang Yan, MD.1
During the
AI-Based Model Identifies Potential Resistance Drivers to CAR T-Cell Therapy in MCL
“Our key findings revealed a list of candidate targets,” Yan explained in an interview with OncLive®. “By inhibiting these targets, we may be able to reverse or overcome CAR T-[cell therapy] resistance. We also found that these targets are related to pathways that support tumor cell growth. That helps explain, at least in part, why some patients develop resistance.”
In the interview, Yan, a postdoctoral fellow at The University of Texas MD Anderson Cancer Center in Houston, discussed the place of CAR T-cell therapy in the MCL treatment paradigm, how AI can be used to address drivers of CAR T-cell therapy resistance, and the details of the analysis that she presented during ASH.
Yan: CAR T-cell therapy is an important treatment option for patients with relapsed or refractory MCL, especially for those who have progressed on BTK inhibitors. In the clinic, patients typically receive BTK inhibitors [after frontline therapy]. If those treatments stop working or if the patient relapses, CAR T-cell therapy becomes an option.
[CAR T-cell therapy is] important, and although the response rate [with this class of agents] is high, some patients do not respond, or they eventually relapse after an initial response.2 That’s why we need to understand why some patients develop resistance and what happens after progression on CAR T-cell therapy.
There have been huge advances recently in the AI field. When AI is applied to biology and cancer research, it allows us to analyze complex, high-dimensional genomic datasets in an efficient way. Instead of focusing on single genes or individual pathways, AI models help us uncover hidden connections and patterns within patient samples. This is useful when we’re trying to understand resistance mechanisms.
Resistance is not caused by a single gene; it’s driven by multiple genes. If we want to understand resistance mechanisms and ultimately overcome cancer, we need to identify and take down those [genes], and that’s where AI comes in. In this study, we used several AI-based approaches. Our model can handle patient heterogeneity, since every patient is different, and it can also learn how genes interact with one another in resistant tumors. We trained the model on millions of cells to recognize these hidden connections.
One core component of our approach was running virtual experiments. We computationally knocked out each gene one by one and observed what happened. If knocking out a gene caused resistant cells to behave more like sensitive ones, then we identified that gene as a key driver. We could then follow up with experimental studies to inhibit that gene. In this way, the AI model helped us identify critical targets.
We’re currently conducting laboratory experiments to validate whether these targets truly work. If they do, the clinical effect could be significant. We could design compounds to overcome CAR T-[cell therapy] resistance, giving patients who become resistant to CAR T-cell therapy another option to improve survival.
