A cancerous tumor is not a single enemy. It is made up of many different cell groups that can behave in very different ways. Among them, even at the time of diagnosis, there may already be cells that after months or years can give rise to metastases or become resistant to treatment. The challenge is that it is very difficult to detect these critical cell populations in a timely manner. In collaboration with Swiss and Swedish partners, researchers in Szeged have made important progress in this field.
The Momentum microscopic image analysis and machine learning research group, led by Peter Horvath, recently appointed Secretary of State for Science, is working to address this challenge. The group, based at the HUN-REN Center for Biological Research in Szeged, is developing artificial intelligence-based methods that link microscopic tissue images with molecular information. Two recent publications in the international scientific journals EMBO Molecular Medicine and npj Precision Oncology provide, for the first time, single-cell-level measurements of the genetic and protein profiles of tumor clones, as well as the genetics of metastasis formation by tumor clones. These findings open new horizons for the future of precision cancer therapy.
Traditional molecular analyzes often examine the entire tissue sample at once. As a result, important differences between different areas of the tumor may be missed. The goal of the new approach is to enable researchers to see not only what a tumor looks like, but also how different cell populations function. In this way, the internal map of the tumor becomes not only visible but also readable at the molecular level.
The group at the HUN-REN Center for Biological Research, in Szeged, is developing digital pathology and spatial technologies in collaboration with Swedish and Swiss partners. Key collaborators include Holger Moch, a molecular pathologist at Zurich University Hospital, and György Marco Varga, a research professor at Lund University. The main technological basis of the work is the Automated Single Cell Research Center established by the Szeged Group. This globally unique system allows AI-selected cells to be isolated with high precision, without human intervention, even around the clock, for subsequent molecular analysis.
This approach is based on deep visual proteomics, or DVP. In this method, AI first analyzes histological images and identifies relevant cell types or cell groups. The researchers then use a thin laser beam to precisely isolate these cells from the sample. Next, proteomic analysis is used to identify proteins present in the selected cells. Proteins are particularly useful because they directly reflect how a cancer cell functions, grows, adapts, or resists treatment.
In this study, the research group advanced DVP in an important way: cancer cell populations selected by AI-based image analysis were examined not only by proteomics, i.e. at the protein level, but also by transcriptome. Transcriptomics shows which genes are active in the cell. This was especially important for groups of cells that looked more dangerous under the microscope. The researchers were able to examine the same cell populations from two complementary perspectives: which genes were turned on in them, and what protein-level functions resulted from this activity.
Using this approach, researchers can more precisely reveal how visually distinct cell populations within a tumor differ from each other biologically. This is important because the aggressiveness of a cancer cell is not explained by a single factor, but by the combined effects of gene activity, protein function, metabolism, and interactions with the immune system.
In another study, the research team demonstrated the clinical potential of this technology through the case of a young patient with recurrent metastatic melanoma. Samples from the patient’s original tumor, as well as subsequent lung and brain metastases, were analyzed using AI-based digital pathology and spatially resolved proteomics.
The AI identified two distinct groups of cancer cells in the patient’s original skin cancer. The cells in later metastases essentially resemble one of these early populations. Protein level analysis confirmed this observation: the molecular pattern of this cell was the closest to that of metastatic lesions. This suggests that traces of subsequent disease spread may already have been present in the original tumor.
“In the case of melanoma, we did not simply see that different parts of the tumor look different,” says Eddy Mieg, a biologist and one of the researchers involved in the study. “Using this method, we can also show that these differences are reflected at the protein level. This may bring us closer to understanding the cell populations that could be responsible for the formation of metastases.”
The interconnected work of the Sijde research group shows that tumors can no longer be understood simply as uniform masses. Different populations of cells can coexist within the same tumor, and some may play a critical role in tumor spread, aggressive growth, or resistance to treatment. In this approach, the AI does not diagnose on its own. Instead, it provides a new level of precision for studying tumors. While traditional pathological assessment classifies larger tissue areas, AI can analyze sample cell by cell. This may reveal what types of cancer cells are present in different areas of the tumor, where they are located, what patterns they form, and in what proportions they occur.
“The more precisely we can see the cells that drive the disease forward, the closer we will be to targeting not just the tumor as a whole, but the most dangerous parts of it,” Peter Horvath said.
Although these findings do not yet represent an immediate new treatment available to all patients, they may help researchers and clinicians map tumor behavior more accurately in the future, better assess risks, and support more personalized treatment decisions. HUN-REN Biological Research Center in Szeged
