This article is part of the “Innovations in: Kidney disease,” an editorially independent special report that was produced with financial support from Vertex.
Dennis Moledina faced a common problem during her training in nephrology, the specialty dedicated to kidney health. Many of the patients he saw had acute kidney injury that was not caused directly by a disease, but by a prescribed medication. Each meeting triggered a series of questions for Moledina: Should they stop the problem drug? Was there a drug they could add to the regimen to protect the patient’s kidneys? Should they biopsy the organs to find the cause?
Acute kidney injury (AKI) is a sudden change in the kidney’s ability to filter waste products from the blood, and it affects about one in 10 hospitalized patients – a figure that jumps to more than five in 10 for those in intensive care. But the condition often causes no pain or discomfort, and clinicians have few warning signs.
On supporting science journalism
If you like this article, please consider supporting our award-winning journalism by subscribes. By purchasing a subscription, you help secure the future of impactful stories about the discoveries and ideas that shape our world today.
Medicines, including common antibiotics and pain relievers, can treat ailments such as infections, but damage the kidneys and worsen the overall problem. Some of these diseases also impair kidney function. By the time blood levels of creatinine—a protein normally removed by the kidneys—are high enough to catch a clinician’s attention, it’s often too late. Irreversible kidney damage has already begun. Due to the diversity of possible causes, identifying the true cause of drug-induced AKI in hospitalized individuals can be incredibly complex, and the condition remains underrecognized.
Historically, researchers and clinicians assumed that the kidneys healed when someone stopped taking medication and recovered from hospitalization. They would monitor patients during treatment — for example, while taking a limited course of antibiotics — but did not think short-term changes in creatinine resulted in lasting damage, says University of Calgary nephrologist Matthew James. There wasn’t much research into AKI, says James. “We didn’t really think about the long-term health outcomes.”
But long-term consequences are proving to be a real risk, and today AKI and chronic kidney disease are seen as linked. For more than a decade, a number of studies have shown that people who suffer from AKI in the hospital are significantly more likely to experience chronic kidney disease and end-stage kidney disease later in life.
Now that specialists know the importance of early detection, they are looking for ways to reduce the amount of lasting damage. When acute injury is diagnosed quickly, clinicians can remove offending drugs or add protective ones. So researchers are working to highlight the most vulnerable. Some use electronic health data to flag problem medications for the patient’s care team. Others, including Moledina, are seeking more precise biomarkers that can be detected in urine tests. Nephrologist Jennifer Schaub of the University of Michigan says there are many reasons why someone may develop AKI, but medication is one that clinicians can do something about. “It’s an area where there’s potential for immediate clinical impact,” she says. “It’s an under-recognized problem, and it’s also something we can (change) in our clinical management very quickly.”
The first challenge is to determine when damage occurs. White blood cells in the urine, high levels of creatinine and low urine volume are often the only clues that alert clinicians to the problem. But each of these is a non-specific marker, says Schaub, and all can occur in patients who are seriously ill.
Currently, the only way to confirm most causes of acute kidney injury is via biopsy, which can reveal tissue inflammation, cellular damage, or the presence of inflammatory cells. But biopsies are risky in patients who are already very sick because the procedures can cause bleeding, infection and other problems.
In children, it is even more difficult to detect AKI early. Hospitalized children tend to have fewer daily blood tests than adults, says nephrologist Perry Wilson at Yale University. To try to solve this, about 15 years ago, researchers and clinicians led by a team at Cincinnati Children’s Hospital Medical Center developed a system to send an alert when a child is undergoing treatments that could damage the kidneys. The system, called AKI NINJA (Nephrotoxic Injury Negated by Just-in-time Action), alerted a pharmacist when a child used a single kidney-damaging drug for three days or three nephrotoxic drugs at the same time. Children on these regimens were closely monitored. If clinicians noticed a worrisome increase in creatinine levels, they could assess whether the risks to the kidneys outweighed the benefits of the medications prescribed.
The NINJA team found that the system allowed care providers to make better decisions about whether to continue or change prescriptions, ultimately reducing the number of acute kidney injury days by 42 percent.
In a December 2024 analysis, University of Iowa nephrologist Benjamin Griffin and his colleagues used hospital data and computational models to test whether the NINJA system would be effective in adults. The problem he faced was not that the system didn’t work, but that it wasn’t specific enough. Because hospitalized adults often use more medications for existing conditions than children, the system produced a large number of alerts. Instead of the 10 or so that clinicians received each month at the pediatric hospital, the system sent hospital staff 30 alerts each day, many of which were not a real cause for concern.
To try to recalibrate the system’s sensitivity, Griffin’s team tested machine learning models that incorporate a patient’s medical history, vital signs and other clinical data to improve the model’s ability to predict drug-induced kidney damage in adults. Currently, the model can measure that risk correctly 60 percent of the time, Griffin says.
ONEHowever, an electronic notification can only do so much. Once someone has been flagged, doctors must make complex decisions about the care of seriously ill patients. “It’s not always as simple as stopping a medication that can be bad for the kidneys,” says Wilson. In addition to blood and urine tests, doctors need better tools to understand how the kidneys are damaged. “We need more than just electronic data now.”
Drugs can damage the kidneys in a number of ways, says Wilson. Research and clinical trials have revealed some of the many different mechanisms through which drugs affect kidney function. This understanding is a critical first step in reducing drug-induced harm.
Some, such as NSAIDs, damage the kidney’s glomeruli, which perform the first step of filtration and keep blood cells and large proteins out of the urine. If we imagine the glomerulus as a colander, Wilson says, “these substances cause the holes in the colander to get bigger” in some people. Other drugs act like poison, killing kidney tubule cells, which help filter waste and reabsorb nutrients. Still others trigger an immune response similar to an allergic reaction.
If clinicians know which of these mechanisms is responsible for someone’s kidney damage, they can figure out how to best combat it, Moledina says. Creatinine peaks, for example, can have a number of explanations: acute interstitial nephritis, spread of tumors or side effects of a completely different drug. Once a spike is detected, clinicians face a diagnostic maze. They can change the treatment, or add steroids that can reduce the problematic kidney inflammation, or just go ahead and hope that the creatinine rise is not related to the medication. Each of these choices presents both risks and benefits. “There are real-world implications, and you don’t want to waste time,” says Moledina.
To go beyond generic alerts, Moledina has used data from the Kidney Precision Medicine Project to create a confirmatory clinical test. He has identified two key proteins, called TNF-alpha and CXCL9, which appear to be closely linked to the type of acute kidney injury triggered by immune reactions. He and his colleagues have begun work on commercializing tests for these biomarkers.
Although biomarkers do not reduce drug-induced kidney damage, they can help with early detection, which is critical to reducing the chances of long-term damage, says Schaub. The longer acute kidney injury goes undetected, the greater the chances of scarring and fibrous tissue blocking kidney function. “The more delayed the diagnosis or the longer it takes to implement treatment, the worse the outcomes for the patient,” she says.
Non-invasive urine and blood tests can also be used to detect other promising biomarkers, such as the protein KIM-1, which can indicate acute kidney injury and tubular injury, and NGAL, a protein biomarker that can help clinicians identify those at risk for AKI as early as 48 to 72 hours into their ICU stay. Such efforts will not only help to find the cause of the problem, but also lead to more precise solutions, says Wilson.
Schaub sees a growing need for these biomarker-based tests as new drugs become available for cancer, heart disease and other conditions. “There are new things to treat people with, which is good,” she says, “but the kidney is sometimes an innocent bystander with all these therapies that are being developed.”
In the long term, more precise, non-invasive biomarkers could be used to develop better drugs and kidney-protective drugs, as well as to determine whether new drugs for other conditions pose a kidney risk. Ultimately, these advances will change AKI from a condition that is difficult to detect and diagnose to one that can be stopped. New biomarkers can be translated into powerful tools for doctors to help patients, says Moledina. “Nephrotoxic damage is something you can do something about,” he adds. “It’s actionable.”
