← Back to Home

Why Does Lung Cancer Metastasize? An Atlas of 2.5 Million Cell Nuclei Reveals Risk Clues

A German team linked tumor cell profiles from 454 patients to their long-term disease courses, finding that combinations of cancer cells and their surrounding environment may help identify risks not fully captured by conventional clinical assessments. But more validation is needed before predictions of disease course can guide treatment.

By SURL BioNews

After lung cancer surgery, which patients may still face the spread of cancer cells? Tumor size and lymph node involvement provide important clues, but cannot fully capture differences within a tumor. A new study compared the molecular characteristics of cells from hundreds of patients with their subsequent disease courses, seeking a more detailed picture of metastatic risk from cancer cells and their surrounding environment.

The non-small cell lung cancer atlas, called LuCaS+, was developed by teams including those at Ludwig Maximilian University of Munich, Charité in Berlin, and the BIFOLD institute at Technische Universität Berlin. The findings were published in Nature Cancer. The study covered 454 patients at three German university hospitals, analyzing more than 2.5 million cell nuclei and linking them to detailed clinical data and long-term follow-up records.

A key foundation of the study was formalin-fixed, paraffin-embedded tumor tissue stored in pathology archives. The team extracted information on the activity of more than 2,000 genes in each nucleus from these specimens, allowing molecular analyses to be compared retrospectively with patients’ disease courses over many years. The atlas platform specifies that 364 patients from Berlin and Cologne formed the discovery cohort, while 90 patients from Munich, sampled and analyzed independently, were used for validation.

Machine learning was used to identify patterns in the extensive gene activity data, divide cancer cells into distinct populations, and then analyze the associations of each population with metastasis and clinical outcomes. One goal was to identify tumors that appear to be at an early stage but may carry a higher risk of metastasis—differences that conventional staging alone may not fully reveal.

Beyond cancer cells, the team also incorporated immune cells, connective tissue cells, and vascular cells into the model. According to an announcement from Ludwig Maximilian University of Munich, combining the metastatic potential of cancer cells with features of the tumor microenvironment allowed the model to predict clinical outcomes better than conventional clinical parameters. The findings were also validated in an independent patient cohort. This suggests that assessing the course of disease may require understanding both cancer cells themselves and their cellular environment.

However, validation in an independent cohort does not yet mean the model can be used to determine treatment for individual patients. The public announcement did not provide the full prediction performance metrics needed to assess clinical utility, and cellular patterns associated with metastasis cannot directly be regarded as mechanisms that cause metastasis. Whether the model can maintain its performance in other healthcare systems, and whether adjusting treatment based on its results improves patient outcomes, remain questions that must be answered before clinical use.

LuCaS+ is now available for researchers to explore, providing sample-level sequencing outputs, clinical data, cell classifications, and tables of differences in gene expression. Its near-term value lies in enabling researchers to investigate which cellular states accompany an unfavorable disease course and use those findings to propose testable treatment hypotheses. The team also explicitly noted that new treatment approaches must first be tested in cell cultures and animal models.

References

  1. Medical Xpress / Ludwig Maximilian University of Munich
  2. Ludwig Maximilian University of Munich
  3. LMU Institute of Pathology — LuCaS+ Explorer