AI-Powered Pan-Cancer Map Reveals Tertiary Lymphoid Structures - Genetic Engineering and Biotechnology News

Researchers at The University of Texas MD Anderson Cancer Center have developed a spatial atlas of specialized immune structures known as tertiary lymphoid structures (TLSs), across multiple cancer types, revealing how key features vary across tumor types and influence patient outcomes. Led by Linghua Wang, MD, PhD, professor of genomic medicine, executive director and head of the Center for Cellular Language Intelligence, associate member of the James P. Allison Institute™, and focus area co-lead with the Institute for Data Science in Oncology at UT MD Anderson, the team developed scalable artificial intelligence (AI) frameworks to detect, profile and classify TLSs from spatial omics data and routine pathology slides.

Tumors can contain TLSs with very different levels of organization, cellular composition and spatial relationships within tumor cells and the researchers’ newly reported study showed that these differences carry important biological and clinical information. The team suggests that their first-of-its-kind atlas indicates that TLS maturation state, spatial location, and composition within tumors may provide clinically meaningful information about cancer prognosis and treatment response. They also created a composite scoring system to more effectively stratify patients by prognosis and treatment response across different cancer types and treatment contexts.

“Prior to this study, most of the focus on TLSs as biomarkers was simply on whether or not they were present and, in some cases, whether they were mature,” Wang said. “Here, we show that we can go much deeper. TLSs in tumor tissues are much more complex than that. Their maturation state, spatial location and composition within tumors can tell us critical information about the tumor immune microenvironment, treatment response and clinical outcomes.”

Wang is senior author of the team’s published paper in Science, Titled “Pan-cancer spatial atlas of tertiary lymphoid structures.” In their paper the team concluded, “Together, this work provides a comprehensive landscape of TLS heterogeneity across cancers and establishes spatially defined TLS features and artificial intelligence (AI)–driven TLS classification as scalable tools for precision immuno-oncology.”

The immune system’s response to a tumor is a highly coordinated effort taking place within the tumor microenvironment (TME), the authors explained. In some tumors, immune cells come together to form organized structures called tertiary lymphoid structures, or TLSs. These structures operate as local immune “hubs,” bringing together B cells, T cells, antigen-presenting cells and other supporting cells that help coordinate antitumor immune responses. “TLSs frequently develop within the tumor microenvironment (TME) and have been observed across a broad range of human solid tumors, where they contribute to lymphocyte activation, B cell immunity, and regulation of antitumor immune responses,” they noted.

Previous studies have shown that TLSs—particularly those that are more mature—are often associated with better patient outcomes and improved responses to immunotherapy across a variety of cancer types. “The presence of TLSs has been linked to favorable responses to immune checkpoint blockade (ICB) and prolonged survival across multiple cancer types, fueling interest in TLSs as predictive biomarkers, prognostic indicators, and potential therapeutic targets. However, the presence of TLSs alone does not tell the whole story,” the scientists noted. “While it is well acknowledged that TLSs are important in cancer, our understanding of their cellular and molecular heterogeneity has remained limited, especially in their natural spatial context across large cohorts of human tumor samples.”

“Although TLS presence has been associated with enhanced immune activity and improved outcomes in several settings, their maturation states, spatial locations relative to tumors, and context-dependent associations have not been systematically characterized at a pan-cancer scale, limiting a unified view of TLS biology and clinical utility,” they stated.

For their reported study the team developed scalable computational frameworks to precisely detect, comprehensively profile and classify TLSs from spatial omics data. Leveraging this framework, the team built a pan-cancer spatial atlas of TLSs across 340 samples from 12 cancer types. This atlas allowed them to examine the TLS landscape in tumor tissues, to define how TLSs vary in key features, and identify transcriptional programs associated with TLS maturation. “By integrating transcriptomic, spatial, histopathological, and clinical data, we systematically characterized TLS abundance, spatial distribution, size, maturation states, and transcriptomic programs in 340 ST samples across 12 cancer types and examined their interactions with tumor cells and the surrounding TME,” they wrote in summary.

The study found that TLSs vary substantially across tissues. As TLSs mature, they become more organized and undergo coordinated changes in immune, stromal, and vascular components. Further, their proximity to tumor cells is associated with spatial gradients of tumor signaling. These findings suggest that TLS maturation and spatial context are linked to distinct tumor signaling environments and may reflect important features of the tumor immune microenvironment.

To make these insights more scalable, the team developed an AI framework to rapidly identify and classify TLSs from hematoxylin and eosin (H&E) whole-slide images (WSIs), pathology images that are routinely used in daily clinical care. Training this AI model makes the process of analyzing TLSs more easily translatable to the clinic, while also making the process significantly faster and more scalable. The AI framework enabled the researchers to go one step further, evaluating 25,088 TLSs from more than 3,000 whole-slide images across 10 independent cohorts and developing a TLS “composition score” for a given patient’s tumor. “By developing a scalable AI-enabled framework to detect and classify TLSs directly from routine H&E WSIs, we have extended TLS analysis from limited spatial datasets to thousands of tumors,” the team noted.

This composition score captures not only the number of TLSs, but also their maturation states within a tumor. This method significantly outperformed conventional TLS measures in stratifying patients by prognosis and treatment response, suggesting that a more detailed view of TLS biology, accounting for maturation state, may provide more clinically meaningful information than TLS presence alone. “… we developed a data-driven, unsupervised TLS-based patient stratification framework that outperformed existing approaches in prognostic evaluation,” they commented.

The TLS composite scoring approach must be validated in prospective clinical trials. If successful, the framework could support broader integration of TLS profiling into routine pathology workflows, since it uses routine pathology images. “Together, this work establishes generalizable and clinically scalable frameworks for TLS profiling and highlights TLS state composition as a key dimension of tumor immune organization with translational relevance. It also provides a foundation for prospective evaluation of TLS-informed biomarkers in clinical settings,” they stated.

The findings raise important biological and therapeutic questions, the researchers suggest. One important observation from the study is that many TLSs in tumor tissues remain immature, and some are located away from tumor regions rather than within or adjacent to tumor cells. This suggests that future studies should investigate how to promote TLSs toward more mature and functional states, and how to enhance their spatial interaction with tumor cells and the broader tumor microenvironment.

These efforts may help identify therapeutic strategies to promote effective TLS formation and maturation and enhance TLS-associated anti-tumor immune responses. In their paper the team concluded, “Prospective studies should test whether TLS composition improves risk and response modelling beyond established clinicopathologic and molecular predictors, and whether TLS-informed stratification can guide clinical trial design or therapeutic modulation strategies.”