One of the biggest mysteries of the brain is how memories are formed. For over a century, scientists have hypothesized that neurons are the only cells involved in storing and recalling memories. Now, researchers from Baylor College of Medicine have discovered that another cell in the brain may play a crucial role in memory as well. Their study showing that star-shaped brain cells called astrocytes are also involved in forming and recalling memories won the 2025 STAT Madness popular vote.
In 1904, a German zoologist and memory researcher named Richard Semon made a prediction about how memories were stored in the brain. Semon hypothesized that memories were kept in ensembles of neurons he called engrams, which became reactivated when memories were recalled.
It’s only been in the last several decades that scientists have been able to prove that Semon’s theory is correct — partially. Baylor’s team found that astrocytes, cells that are in close physical proximity to neurons and help with neurotransmission, are also an essential part of the engram.
“An engram is now comprised of multiple ensembles, you have an ensemble of neurons and an ensemble of astrocytes,” said senior author Benjamin Deneen, the Dr. Russell J. and Marian K. Blattner Chair at the Center for Cancer Neuroscience at Baylor. “They work hand in hand.”
The 2025 STAT Madness competition, a bracket-style celebration of biomedical research, stacked 64 entries against each other in a month-long competition. The contest garnered 374,302 votes for topics that included a new gene-editing technique that can turn off prions that cause diseases like mad cow, and a light-powered pacemaker thinner than a strand of hair.
Ultimately, the Baylor team, which included co-first authors and postdoctoral associates Michael Williamson and Wookbong Kwon, won with 63.6% of the 92,001 votes in the final round, making this the medical school’s second successive Madness championship. Baylor beat a team from Florida International University that studied the feasibility of functional precision medicine in treating pediatric cancers.
A team from NYU’s School of Global Public Health won the STAT Madness All-Star Award last month, voted on by attendees at the Breakthrough Summit East, for research finding higher rates of deep sedation in Hispanic patients hospitalized with respiratory failure, which may help explain why Hispanic people are twice as likely to die from respiratory failure than other similar patients.
To showcase astrocytes’ role in memory, the Baylor team honed in on a gene called c-Fos, which plays an important role in neuron plasticity. The researchers found that about 4% to 5% of astrocytes express the same c-Fos gene. The researchers dubbed these astrocytes learning-associated astrocytes, or LAAs, and tagged the cells to study them using a viral vector.
Next, the researchers used genetically modified mice and conditioned them to feel fear. When the mice were put in new situations, they didn’t freeze up in fear like they normally did — until the LAAs were activated. “Our interpretation of that was that the mice were having sort of an artificial memory,” Deneen said, that “is stored in part in the astrocytes.”
The study was published in the journal Nature last November, and was supported by funding from the National Institutes of Health, the National Research Foundation of Korea, and the David and Eula Wintermann Foundation.
The study does have limitations, the authors acknowledge. While the gene has been well studied in neurons, there is very limited research on the gene in astrocytes, and the scientists aren’t sure why some astrocytes express the gene, while others don’t. “God only knows what c-Fos means in an astrocyte,” Deneen said.
Since the study’s publication, the authors have continued to research astrocytes’ roles in memory, including trying to understand the changes that occur in astrocytes that exhibit the c-Fos gene.
“Can all astrocytes become engrams?” Deneen said, “Or are there preordained or specialized astrocytes that are more prone or more likely to become an engram? We don’t know the answer to that.”
The team from Florida International University took second place with research showing that functional precision medicine (FPM) can be done quickly and improves outcomes for pediatric cancer patients.
Normally, cancer patients need to rely on a doctor’s expertise to decide what treatment to try. FPM, which combines cancer genomic profiling with drug sensitivity testing, takes some of the guesswork out of finding the best drug to fight each patient’s individual cancer.
The method had already been shown to work in adults, but Diana Azzam, the scientific director of the Center for Advancing Personalized Cancer Treatments at FIU’s Robert Stempel College of Public Health & Social Work, wanted to see if it could be helpful for pediatric patients with relapsed and hard-to-treat cancers for which few treatment options exist.
“We demonstrated that we could generate functional precision medicine data and return results to a tumor board quickly enough to make a decision for the next line of treatment,” Azzam said.
Azzam’s study had two objectives: to see if the FPM approach could be done quickly enough to be useful, and to see if patients who took drugs recommended by the FPM testing would have better clinical outcomes than they did before participating in the study.
The research team found that it could quickly conduct all the testing required for the FPM method, completing drug sensitivity testing and genomic profiling within 10 days and 27 days, respectively. This was fast enough for the team to come up with actionable treatment recommendations before too many of the children’s cancers progressed.
During the drug sensitivity testing, Azzam and her team exposed the children’s cancer cells to 125 FDA-approved drugs, including cancer drugs, allergy medications, and statins. Azzam said that one of the important results of the paper is showing that some of these drugs had anticancer properties and were indicated as treatments against relapsing and hard-to-treat cancers. “Repurposing of these drugs is key,” Azzam said.
After the drug sensitivity and genomic tests, the team was able to make individualized treatment recommendations for 19 of the 25 patients enrolled in the study. Only six patients took the FPM-recommended drug, but those that did had a much higher likelihood of progression-free survival compared to patients that went back to their doctor’s treatment plan.
One of those six patients was Logan Jenner, a young boy with relapsed acute myeloid leukemia. Testing revealed a specific mutation that could be targeted with medication, and he has been cancer-free for more than two years.
The study was published in Nature Medicine last April and was funded by the Florida Department of Health.
There are limitations to the study, Azzam said. The sample size was small, at only 25 patients, and did not include a randomized control group. In order to limit bias, the researchers used each patient as their own control, comparing their cancer progression after their choice to pursue FPM to their progression before the trial. Those that used the FPM-guided treatment substantially improved.
“This is a proof of principle study,” Azzam said, “to really show that this is feasible and we can scale it up.”
Azzam’s ultimate goal is to use FPM testing for rare cancers before they progress and relapse. She imagines a world where children — and adults — diagnosed with cancers such as osteosarcoma or glioblastoma can get quick FPM testing to determine the best treatment option immediately.
She’s kickstarting the process by working to open a CLIA-certified lab in Florida that will take samples from patients nationwide and perform FPM testing. “It’s going to be the first commercial lab to provide functional precision medicine, along with artificial intelligence, to guide treatments for relapsed, refractory patients,” Azzam said, “not only children, but also adults.”
