Development of a Predictive Model for Lymph Node Metastasis in Endometrial Cancer Based on Integrated Clinical Characteristics, Serum Tumor Biomarkers, and Related Factors

Endometrial cancer is a disease that begins in the lining of the uterus, the womb where a baby grows. It is one of the most common cancers affecting the female reproductive system, and its numbers are rising in many parts of the world. When this cancer spreads, it often travels first to the lymph nodes, which are small, bean-shaped structures scattered throughout the body that act as filters for the immune system. Knowing whether the cancer has reached these nodes before surgery is critical. If the nodes are clear, a surgeon might perform a less invasive operation. If they are involved, the treatment plan changes to a more extensive procedure to ensure all cancerous tissue is removed. However, predicting this spread before the operation is difficult. Doctors currently rely on imaging scans and physical exams, but these tools are not always perfect. They need a way to look deeper, using clues hidden within the patient's own blood and medical history to make a more accurate guess.

In a recent study, researchers at the Affiliated Cancer Hospital of Xinjiang Medical University set out to build a better tool for this exact purpose. They gathered information from 1,099 women who had been diagnosed with endometrial cancer and underwent surgery between 2021 and 2026. The team wanted to see if they could combine standard medical details, such as the patient's age and the stage of the cancer, with specific substances found in the blood to predict lymph node spread. These substances are known as tumor biomarkers. The researchers focused on three specific ones: CA125, HE4, and CA199. These are proteins or sugars that cancer cells sometimes release into the bloodstream, acting like faint signals that something is wrong inside the body. By looking at how these signals behaved alongside the patient's clinical history, the team hoped to create a reliable guide for doctors.

To turn this data into a prediction tool, the researchers used a type of computer intelligence called machine learning. They fed the information from the 1,099 patients into three different computer models to see which one could learn the patterns of cancer spread most effectively. One model worked like a traditional statistical calculator, while the others were more complex systems capable of finding subtle connections between many different factors at once. The team split the data into two groups, using most of it to teach the models and the rest to test them. The results showed that one specific model, known as XGBoost, was the most accurate. It correctly identified the risk of lymph node spread with a high degree of success, outperforming the other two approaches. This model learned that the most powerful single clue was the clinical stage of the cancer, which describes how far the tumor has grown within the uterus and nearby tissues.

When the researchers looked at the blood markers individually, they found that CA125 was the strongest predictor among the three, followed by HE4 and then CA199. However, the study revealed that relying on just one marker was not enough to get the full picture. When the computer model combined the levels of all three biomarkers, its ability to predict lymph node metastasis improved significantly. The combined approach was better at spotting the women who were at risk than any single test could be on its own. The team also determined specific threshold values for each marker. For instance, if a patient's CA125 level was above 24.8 units per milliliter, it suggested a higher risk. Similarly, specific cut-off points were found for HE4 and CA199. These numbers provide doctors with concrete reference points to help stratify risk before a patient even enters the operating room.

The study also clarified what does not seem to matter as much in this specific prediction. The researchers analyzed factors like whether a woman had given birth, her body mass index, and whether she had a history of diabetes or high blood pressure. While these are important for overall health, the data showed they did not significantly change the likelihood of lymph node spread in this group of patients. The most telling signs remained the stage of the tumor and the levels of the three specific blood markers. The researchers noted that while their model performed very well, it was built on data from a single hospital, and the number of women with confirmed lymph node spread was relatively small compared to those without. This means the tool is promising but would benefit from testing on even larger groups of people in the future to confirm its reliability across different populations.

Ultimately, this work offers a new, data-driven way to support clinical decisions. By integrating the stage of the cancer with a simple blood test that measures three specific markers, doctors can now have a more informed view of the risks before surgery begins. This approach does not replace the surgeon's judgment but adds a layer of precision that could help tailor the treatment to the individual. For a woman facing this diagnosis, having a clearer picture of whether the cancer has spread to the lymph nodes means the surgical plan can be optimized from the start, potentially sparing unnecessary procedures or ensuring that the right extent of treatment is applied immediately. The study suggests that the future of cancer care lies in combining human expertise with these intelligent, data-rich tools to make the most accurate predictions possible.