AI maps tumor cells that may seed later metastases

A cancerous tumor is not a single entity, but rather a complex collection of various cell types that can exhibit different behaviors. Even at the time of diagnosis, there may be cells within the tumor that could potentially lead to metastases or develop resistance to treatment in the future. Detecting these critical cell populations early on is challenging.

However, a research group led by Péter Horváth, a state secretary for science, is developing artificial intelligence-based methods to address this issue. This group, based at the HUN-REN Biological Research Center in Szeged, Hungary, collaborates with experts from Sweden and Switzerland.

The group's innovative approach combines microscopic tissue imaging with molecular information to create a detailed "digital map" of tumors. By leveraging artificial intelligence (AI) and deep visual proteomics (DVP), the researchers can identify and isolate specific tumor cell populations at the single-cell level. This allows for a more comprehensive understanding of how different cell populations function within the tumor.

In their studies, the researchers have provided new insights into the genetic and protein profiles of tumor clones. These findings open up possibilities for precision cancer therapy by enabling researchers to target specific cell populations rather than treating the tumor as a uniform mass. Traditional molecular analyses often ignore important differences between regions of a tumor, but this new approach provides a more accurate and detailed view of the molecular landscape within the tumor.

The researchers' work has been applied to real-world cases, such as a young patient with recurrent metastatic melanoma. By analyzing tumor samples from the patient's original tumor, as well as later metastases in the lungs and brain, the AI-based digital pathology and spatially resolved proteomics technology identified two distinct tumor cell populations.

The protein-level analysis confirmed that the molecular pattern of these populations matched that of the metastatic lesions. This suggests that traces of metastasis may have already been present in the original tumor, highlighting the importance of early detection and targeted treatment.

While this research does not yet translate into immediate therapeutic options for all patients, it represents a significant step forward in understanding the molecular basis of tumor heterogeneity and metastasis. By employing AI to analyze tumors at the single-cell level, researchers can uncover the specific cell populations responsible for aggressive growth, metastasis, or resistance to treatment.

This knowledge may ultimately lead to more effective, personalized cancer therapies that target the most dangerous components of the disease.