A genetic marker already used to flag risk of a life-threatening reaction to the antibiotic vancomycin also identifies patients at risk of that reaction to lamotrigine — and it is missing from the gene panels clinicians order before prescribing the drug, Vanderbilt Health-led researchers report Sept. 2 in JAMA Network Open .
The two variants those panels do test for showed no association with lamotrigine reaction in a U.S. population.
Drug reaction with eosinophilia and systemic symptoms, or DRESS, can progress from rash and fever to organ damage, carries a 3%-10% mortality rate, and can leave survivors with long-term complications. Lamotrigine, a first-line treatment for bipolar and seizure disorders prescribed to an estimated 2.5 million Americans in 2024, ranks among the top five causes of DRESS worldwide, affecting an estimated 1 in 1,000 to 1 in 10,000 patients who take it.
Psychiatrists and neurologists now routinely order pharmacogenomic panels before starting aromatic antiseizure drugs. Those panels test HLA-B*15:02 and HLA-A*31:01, genetic variants validated for reactions to the drug carbamazepine — B*15:02 for Stevens-Johnson syndrome in patients of Southeast Asian ancestry and A*31:01 for DRESS across ancestries. Neither had ever been validated for lamotrigine in a U.S. population.
"The test being ordered was built for a different drug, yet for lamotrigine a negative result can literally be misinterpreted as a green light, offering reassurance that the evidence simply does not support," said corresponding author Elizabeth Phillips , MD, Professor of Medicine in the Division of Infectious Diseases and Director of the Center for Drug Safety and Immunology.
The matched case-control study compared 29 patients with lamotrigine-induced DRESS, enrolled prospectively from 2016 to 2025 at Vanderbilt Health and Mass General Brigham, with 290 lamotrigine-tolerant controls from Vanderbilt's BioVU biobank, matched on age, sex and self-reported race.
HLA-A*32:01 was carried by 12 of the 29 patients who developed DRESS (41.4%) versus 12 of 290 controls (4.1%), a 16.4-fold higher odds , which is the strongest genetic signal reported for lamotrigine-induced DRESS in any population to date. HLA-A*31:01 showed no significant difference, and HLA-B*15:02 was carried by no one, consistent with its rarity outside Southeast Asian ancestry.
The authors are direct about their study's limitations. With HLA-A*32:01 having been present in fewer than half the cases, it's neither necessary nor sufficient; a negative result does not rule out risk. They estimate that roughly 2,400 patients would need testing to prevent one case — figures that closely parallel HLA-A*31:01 and carbamazepine-induced DRESS.
"That performance is in the same range as a test already in routine use for carbamazepine, so it's not a hypothetical standard — it's the one the field has accepted," said first author Matthew Krantz , MD, Assistant Professor of Medicine in the Division of Allergy, Pulmonary and Critical Care Medicine.
With a rapid PCR assay for HLA-A*32:01 already developed for vancomycin-induced DRESS, the marker is inexpensive and immediately testable. Lamotrigine becomes the second drug tied to DRESS through the allele, though docking simulations reported in the study suggest the two drugs bind it differently.
The cohort is small and 79.3% white, and the study was powered only for large effects — the authors caution that the absence of other associations should not be read as excluding them.
Phillips holds the John A. Oates Chair in Clinical Research. Other VH authors are Alexis Yu, MPH, Rama Gangula, MS, and Simon Mallal, MBBS. They're joined by researchers from Mass General Brigham, the Mayo Clinic, the University of Florida and Murdoch University.
The study was supported in part by the National Institutes of Health under grants K08AI185260, R21AI199163 and P50GM115305, among others.
JAMA Network Open
2-Sep-2026