A Simple Clinical–MRI Model Combining Tumor Mobility and ADC for Preoperative Prediction of Malignancy in Parotid Gland Tumors

The parotid glands, located just in front of the ears, are the largest of the salivary glands in the human body. Like many parts of the body, they can develop lumps, most of which are harmless growths that cause no trouble. However, a small portion of these lumps are malignant, meaning they are cancerous and capable of spreading or invading nearby tissues. Distinguishing between a benign lump and a dangerous one before surgery is a critical challenge for doctors. If a tumor is benign, a surgeon might remove it with a simple procedure, preserving the facial nerve that runs right through the gland. If it is malignant, the surgery becomes far more complex, often requiring wider removal of tissue and careful management of that same nerve to prevent paralysis. For decades, doctors have relied on physical exams and magnetic resonance imaging, or MRI, to make this call. MRI is a powerful tool that uses magnetic fields to create detailed pictures of the body's interior, and a specific type of scan called diffusion-weighted imaging can measure how water molecules move inside a tumor. In general, tightly packed cancer cells restrict this movement more than the looser structure of benign tumors. Yet, even with these tools, uncertainty remains, and doctors often need to know more to plan the right treatment.

Researchers at Kansai Medical University in Japan set out to create a simpler, more reliable way to predict whether a parotid tumor is benign or malignant before an operation. They focused on combining two specific pieces of information: a physical finding from a standard doctor's visit and a quantitative measurement from an MRI scan. The first piece of information is tumor mobility. During a physical exam, a doctor gently pushes on the lump to see if it moves freely under the skin or if it feels stuck. A tumor that does not move often suggests it has grown into or stuck to surrounding tissues, a sign of malignancy. The second piece of information is the apparent diffusion coefficient, a number derived from the MRI that quantifies how much water can move within the tumor. The team wanted to see if using just these two factors together could provide a clear answer, avoiding the need for complex computer models or expensive, time-consuming image analysis software that is difficult to use in everyday clinics.

To test this idea, the researchers looked back at the medical records of 146 patients who had surgery for parotid tumors between May 2021 and July 2025. They excluded patients with lymphoma, a type of blood cancer, because it behaves differently from the typical salivary gland tumors they were studying. The group included 125 patients with benign tumors and 21 with malignant ones. The team split the patients into two groups: a larger group of 109 whose MRI scans were done at their own hospital, and a smaller group of 37 whose scans came from other medical centers. They used the first group to build their prediction model and the second group to see if the model would work on data it had never seen before. The researchers measured the tumor mobility based on what was written in the medical records and calculated the diffusion numbers from the MRI scans, making sure to avoid areas of the tumor that were cystic or bleeding, which could skew the results.

The analysis revealed that both tumor mobility and the diffusion number were strongly linked to whether a tumor was cancerous. Patients with malignant tumors were significantly more likely to have lumps that did not move during the physical exam. They also had lower diffusion numbers, indicating that water movement inside the tumor was more restricted. When the researchers combined these two factors into a single model, the results were promising. The model successfully distinguished between benign and malignant tumors in the group used to build it, and it maintained a similar level of accuracy when tested on the patients from other hospitals. This external test was important because it showed that the model could handle differences in MRI machines and scanning settings, suggesting it might work in various clinical environments. The researchers found that adding the diffusion measurement to the simple check for mobility significantly improved the ability to predict malignancy compared to checking mobility alone.

The study also examined other potential clues, such as the brightness of the tumor on different types of MRI images compared to the surrounding muscle and gland tissue. While these brightness ratios were also linked to malignancy in initial tests, they did not add enough new information to justify including them in the final model. The researchers decided that keeping the model simple with just two variables was better, especially since their dataset of cancer cases was relatively small. A more complex model with too many variables risked fitting the data too perfectly to the specific group of patients studied, which would make it less useful for future patients. By sticking to mobility and the diffusion number, they created a tool that is easy to use and interpret. The final result was presented as a nomogram, a visual chart that allows a doctor to simply add points for the mobility status and the diffusion number to get a probability score for malignancy.

The authors acknowledge that their study has limitations. It was a retrospective look at past data from a single hospital, which means there could be biases in how patients were selected. The number of malignant tumors was relatively low, which makes the statistical estimates less precise than they would be with a larger group. Furthermore, the group of patients from other hospitals used for testing was small, so the confidence in how well the model performs outside the original hospital is not absolute. The researchers also noted that diffusion measurements can vary depending on the specific MRI machine used, and they did not apply a strict mathematical correction to standardize these differences across all scanners. Despite these caveats, the study suggests that a straightforward approach combining a physical exam finding with a standard MRI measurement can offer a useful, low-burden way to estimate cancer risk. This method does not replace the need for biopsies or advanced imaging in every case, but it provides a practical tool for doctors to make better-informed decisions about surgical planning and patient counseling before the first incision is made.