AI-Based CT Scan Analysis Predicts Early-Stage Kidney Damage Due to Cancer Treatments

However, one of the potential side effects of this therapy is kidney function decline over the course of treatment. Researchers have now developed a method to predict early-stage kidney damage caused by certain cancer therapies. This damage can be detected by observing a slight shrinkage of the kidneys, which occurs months before any measurable decline in kidney function. The team identified this trend through the analysis of CT scans using an artificial intelligence (AI)-powered algorithm. They also noticed similar changes in the spleen. These findings could eventually lead to early adjustments in treatment to prevent further organ damage.

In their latest research, a team from the Technical University of Munich (TUM, Munich, Germany) studied data from 121 patients undergoing treatment for prostate cancer with lutetium-177 PSMA. In a prior study, they discovered that patients who experienced a decline in kidney function following lutetium-177 PSMA therapy showed structural changes in their kidneys. Since obtaining tissue samples routinely is not practical, the team set out to determine if these changes could be detected using less invasive methods. The approach chosen aimed to avoid any additional burden on the patients, as CT scans and blood tests are already a standard part of cancer care to monitor treatment progress. The researchers analyzed various indicators from these routinely collected data to detect early signs of kidney damage.

Although factors such as kidney length or patient age did not provide reliable predictions, changes in kidney volume emerged as a strong signal. Specifically, when kidney volume decreased by 10% or more within the first six months of treatment, there was a high likelihood that kidney function would decline significantly within the following six months. These changes in kidney volume are subtle and can easily be overlooked during routine image assessments, where clinicians are typically focused on tracking tumors and other critical findings. However, with properly trained image analysis algorithms, even these minor changes can be reliably detected.

“If it becomes clear that a patient is at increased risk of kidney impairment after six months of treatment, both the number of therapy cycles and the dosage can be individually adjusted,” said lead author Dr. Lisa Steinhelfer.

Related Links:
Technical University of Munich