One HLA Mismatch Triples Severe Transplant Rejection Risk in Cord Blood Recipients

When the Donor's Immune Cells Attack

Stem cell transplantation can cure blood cancers. It can also kill the patient. The mechanism behind that risk -- graft-versus-host disease -- occurs when donor immune cells, once established in the recipient's body, recognize the recipient's tissues as foreign and attack them. Mild forms are manageable; severe grade III-IV disease carries a mortality risk roughly double that of patients without the complication.

Matching donors and recipients by human leukocyte antigen (HLA) type is the primary strategy for preventing graft-versus-host disease. HLA molecules sit on cell surfaces and present protein fragments to the immune system. When donor and recipient HLA molecules differ, donor T cells detect the mismatch and mount an attack. Standard donor selection minimizes the total number of mismatches.

But not all mismatches are equal. Some HLA combinations provoke a disproportionately strong immune response. Others are better tolerated. Identifying which specific pairings carry unusually high risk allows clinicians to make more informed choices when multiple donor units are available -- avoiding high-risk combinations when a safer alternative exists.

7,462 Transplants, One Standout Pairing

A team led by Associate Professor Takakazu Kawase from Fujita Health University's Department of Immune Regenerative Medicine analyzed data from 7,462 Japanese patients aged 16 and older who underwent their first unrelated cord blood transplant. The dataset came from a nationwide registry compiled by the Japanese Society for Transplantation and Cellular Therapy. Cord blood was specifically the focus because it is known for tolerating HLA mismatches better than bone marrow or peripheral blood transplants -- making it a preferred option when matched donors are unavailable.

The analysis used advanced statistical models adjusted for patient age, disease severity, conditioning regimen, graft cell dose, and total mismatch count. Among all possible individual mismatch combinations tested, one stood out with striking clarity: a donor carrying the HLA allele C*03:04 paired with a recipient carrying C*14:02 was associated with a hazard ratio of 3.09 for severe (grade III-IV) graft-versus-host disease. In other words, patients in that pairing faced three times the risk of severe immune attack compared to patients without that mismatch, after controlling for other variables.

The finding survived strict statistical correction for multiple testing -- a critical hurdle, because analyzing hundreds of allele combinations creates many opportunities for spurious results. The effect held up.

A Risk Profile Specific to Cord Blood

An important nuance emerged from the analysis. High-risk mismatch combinations previously identified in unrelated bone marrow transplants did not show the same elevated risk in cord blood transplantation. Risk profiles from bone marrow transplant registries cannot be directly translated to cord blood settings. The immunological dynamics differ, and the data to guide decisions need to come from cord blood-specific analyses.

"This study shows that even in cord blood transplantation, where HLA mismatches are generally better tolerated, specific HLA combinations can provoke very strong immune reactions," said Kawase. "Identifying these high-risk mismatches gives us an opportunity to improve donor selection and reduce life-threatening complications."

The team also examined the broader clinical impact of graft-versus-host disease severity. Moderate (grade II-IV) disease was actually associated with improved overall survival in some analyses -- a counterintuitive finding that likely reflects the "graft-versus-leukemia" effect. Severe (grade III-IV) disease, however, was associated with an 82% increase in mortality risk (hazard ratio 1.82), confirming that preventing severe disease is the priority.

Practical Implications

The real-world application is straightforward in concept: when selecting among available cord blood units for a recipient carrying HLA-C*14:02, units from donors carrying C*03:04 should be deprioritized if comparable alternatives exist. How broadly this finding applies outside the Japanese registry population studied here remains to be determined. The study was conducted entirely within a Japanese cohort, and HLA allele frequencies vary by ancestry. Whether the same risk pattern holds in European, African, or Latin American populations would require separate analysis. Published in Transplantation and Cellular Therapy (Volume 32, Issue 1, January 2026).