Development of a Nomogram Model Based on Contrast-Enhanced CT Histogram Analysis for Preoperative Prediction of Ki-67 Expression Status in Patients with Clear Cell Renal Cell Carcinoma

Cancer is not a single disease but a collection of disorders defined by how quickly cells divide and how aggressively they spread. In the kidney, the most common form of cancer is clear cell renal cell carcinoma, a tumor that can vary wildly in its behavior from patient to patient. Some grow slowly and remain contained, while others are fast-moving and dangerous. To understand which is which, doctors often look for a specific protein inside the tumor cells called Ki-67. This protein acts like a molecular switch that is turned on only when a cell is actively preparing to divide. The more Ki-67 present, the more frantic the cell's activity, and the more likely the tumor is to be aggressive. However, finding out how much of this protein exists usually requires surgery to remove the tumor and then analyzing the tissue under a microscope. This invasive process means doctors often cannot know the tumor's true nature until after the operation is already done.

A team of researchers at Lanzhou University Second Hospital has developed a way to peek at this hidden information before a patient ever goes under the knife. By using standard CT scans, which are like detailed three-dimensional X-rays, they created a mathematical tool that can predict the level of Ki-67 activity with surprising accuracy. The study focused on 213 patients who had already undergone surgery for this type of kidney cancer. The researchers looked back at the CT images taken before the operations and combined them with simple blood test results and the size of the tumors. They discovered that the internal texture of the tumor, as seen on the scan, holds clues about how fast the cells are multiplying. By blending these visual details with the patient's blood work, they built a guide that helps doctors estimate the tumor's aggressiveness in advance, allowing for more informed decisions about treatment before the first incision is made.

The researchers started by gathering data from patients treated between January 2022 and December 2024. They split the group into two sets: a larger group to build their model and a smaller group to test it. For every patient, they examined the CT scans taken during two specific moments when contrast dye was flowing through the kidneys. They also looked at the patients' blood tests, specifically counting the number of monocytes, a type of white blood cell, and calculating the ratio between neutrophils and lymphocytes, two other types of immune cells. They measured the tumor's size and carefully analyzed the gray-scale patterns inside the tumor on the scan. This pattern analysis, known as histogram analysis, looks at how the brightness of every tiny pixel within the tumor is distributed. Instead of just looking at the average brightness, they measured how uneven or "skewed" the distribution was, a detail that can reveal hidden variations in the tissue's structure.

Using a computer algorithm, the team sifted through dozens of potential clues to find the four that mattered most. They found that the tumor's size, the patient's monocyte count, the ratio of neutrophils to lymphocytes, and the specific skewness of the tumor's brightness on the scan were the strongest predictors. These four factors were combined into a single visual chart, called a nomogram. This chart allows a doctor to plug in a patient's specific numbers and see a calculated probability of whether the tumor has high or low Ki-67 expression. When they tested this tool on the group of patients used to build it, it achieved an AUC of 0.86. When they tested it on the separate group of patients used for validation, the AUC remained just as high, at roughly 0.86.

The study showed that this combined approach was far better than looking at any single factor alone. For instance, looking only at the tumor size or only at the blood counts was not enough to make a reliable prediction. The power came from the combination. The researchers also checked how well their predictions matched reality using statistical tests, and the results showed a strong agreement between what the model predicted and what was actually found in the pathology reports after surgery. They further demonstrated that using this tool would provide a better clinical outcome than simply treating every patient the same way or ignoring the risk entirely. The model successfully highlighted that tumors with higher Ki-67 levels tended to be larger, had more uneven internal textures on the scan, and were associated with specific patterns of inflammation in the blood.

While the results are promising, the researchers are careful to note that this was a retrospective study, meaning they looked at past data rather than following patients forward in time. The data came from a single hospital, and the method relies on specific types of CT scans, which means the tool needs to be tested in other settings and with different equipment to ensure it works universally. Additionally, the study did not track how long patients lived after treatment, so the tool's ability to predict long-term survival remains to be seen. However, the work offers a significant step forward in non-invasive medicine. By turning a standard CT scan and a routine blood test into a window on the tumor's biology, this model gives doctors a new way to understand the enemy before the battle begins, potentially sparing patients from unnecessary procedures or ensuring they receive the right intensity of care from the very start.