Six-Gene Score Improves Prognosis Prediction in Acute Myeloid Leukemia

Acute myeloid leukemia (AML) is a blood cancer with outcomes that can vary widely from one patient to another. This variability complicates prognosis despite the laboratory information available from bone marrow sampling that patients already undergo. New findings offer a potential way to refine this assessment, demonstrating that the activity of six genes may help stratify AML patients by their survival prospects.

The Center for Genomic Regulation (CRG) in Barcelona developed PHFScore, a six-gene scoring system that combines the activity of CARMIL1, HOPX, ALDH1A1, CALR, STAR, and PHF19. Within this signature, CARMIL1, HOPX, and ALDH1A1 were more active in patients with poor outcomes, while CALR, STAR, and PHF19 were less active.

Image: Spleen tissue from a mouse without PHF19 (left) and a mouse with normal PHF19 function (right). Dense clusters of immune cells sit within blood-rich regions in pink. Without PHF19, cells are altered in their differentiation process (Photo courtesy of Gianni Gamarra/Centro de Regulación Genómica)

The work began by examining PHF19, a gene involved in keeping other genes switched off as cells differentiate from immature to mature states. To investigate its role, researchers introduced a leukemia-causing gene into mouse blood cells with and without PHF19. Cells lacking PHF19 formed more colonies in laboratory experiments, took hold more easily following transplantation into mice, and remained in an immature, stem-cell-like state.

Building on these experiments, the team developed the score by repeatedly comparing laboratory findings in mice with patient records to identify genes that behaved consistently in both. The molecular pattern observed in mice was also present in human AML datasets.

Among younger adults treated for AML, patients with the lowest PHF19 levels had shorter survival than those with the highest levels and were less likely to enter remission after standard chemotherapy. Fewer than half of patients with the lowest levels entered remission, compared with up to 85% of those with the highest levels.

The study took five years to complete and was published in Science Advances on October 23, 2026. It was led at CRG, with single-cell expertise and technological support from CRG’s Genomics Unit and Flow Cytometry Unit. The researchers said PHFScore could contribute to future prognostic test development, and that measuring six genes may be simpler and cheaper than reading a broader genetic profile.

“PHF19 on its own already tells us something about how patients will fare, but combining it with five other genes is a dramatic improvement which gives us more accurate information for AML prognosis,” said ICREA research professor Luciano Di Croce, senior author of the study and researcher at the CRG.

Related Links
 Center for Genomic Regulation