Counting Molecular Mismatches: HLA Eplet Analysis Moves Kidney Transplant Care Toward Precision Immune Risk Assessment

Kidney transplantation has made remarkable strides over the past half-century, yet the graft a patient receives today is still too often lost a decade or more later. A new narrative review by Christine Jia Yu Wong and Emmett Tsz Yeung Wong of the National University Hospital in Singapore, published in Frontiers in Medicine, argues that a concept called HLA molecular mismatch could reshape how clinicians judge the immunological danger posed by a particular donor–recipient pair. The review, which synthesized literature published through August 2026, maps the journey from crude antigen counting toward quantifying the actual structural differences between donor and recipient HLA molecules — and finds growing, though still incomplete, evidence that this finer measurement predicts who will develop the antibodies that destroy transplanted kidneys.

The clinical stakes are considerable. Although short- and medium-term transplant outcomes have improved dramatically through better immunosuppressive drugs, matching, organ preservation and perioperative care, late graft attrition persists. In a cohort of 1,477 adult kidney transplants, Mayrdorfer and colleagues found that death with a functioning graft was driven mainly by cardiovascular disease, malignancy and infection, whereas death-censored graft failure tracked alloimmune injury, particularly T-cell and antibody-mediated rejection. Mulley and colleagues went further, identifying rejection as the leading cause of death-censored kidney allograft failure in 47.5 percent of 642 cases, with chronic antibody-mediated rejection responsible for 37.4 percent. In a retrospective analysis of 5,752 recipients, Merzkani et al. found that alloimmunity accounted for roughly half of graft failures occurring beyond the first year in younger non-diabetic patients. The picture that emerges is one of competing risks: too little immunosuppression invites rejection, while too much invites infection and malignancy.

Conventional histocompatibility assessment has evolved through distinct generations of technology. After Joseph Murray’s successful identical-twin kidney transplant in 1954 demonstrated that grafts survive indefinitely when genetic disparity is absent, HLA typing and antigen-level matching became standard. Patel’s landmark study of 225 transplants showed that a positive complement-dependent cytotoxicity crossmatch — evidence of preformed cytotoxic antibodies against donor lymphocytes — predicted immediate graft failure, cementing prospective crossmatching as a core pre-transplant safeguard. The panel reactive antibody test extended this logic to population screening, and during the 1980s and 1990s flow cytometry crossmatching detected lower-level donor-reactive antibodies invisible to the older method. Solid-phase Luminex single-antigen bead assays then enabled characterization of individual anti-HLA antibody specificities, making the virtual crossmatch possible without ever mixing recipient serum with donor cells.

Yet even these sensitive tools share a blind spot: antigen-level matching treats all mismatches of the same HLA antigen as equivalent, when in reality the molecular composition and immunogenic potential of individual mismatches vary substantially. Alloantibodies do not recognize HLA antigens as whole units; they recognize discrete structural configurations of polymorphic residues on the molecular surface. This insight underlies HLAMatchmaker, developed by Duquesnoy, which originally quantified mismatched surface-exposed amino acid triplets and later evolved to count eplets — patches of non-self, surface-exposed residues within a 3 Ångström radius capable of forming antibody paratope–HLA epitope interactions. The review stresses that an eplet is a computationally defined structural unit, not necessarily an experimentally verified antibody epitope. A rival approach, the HLA Epitope Mismatch Algorithm, or HLA-EMMA, instead counts directly solvent-accessible polymorphic amino acid mismatches; the two measures overlap but are not interchangeable, a point of real consequence when comparing risk thresholds across studies.

The prognostic evidence for molecular mismatch has accumulated steadily, with the strongest findings centered on HLA-DR and HLA-DQ. In a retrospective cohort of 664 recipients followed for a median of 7.5 years, Wiebe and colleagues showed that HLA-DR/DQ single-molecule epitope mismatch was associated with de novo donor-specific antibody (dnDSA) development, antibody-mediated rejection and Banff grade 1A or higher T-cell mediated rejection, and discriminated risk better than conventional antigen-level mismatch. Senev and colleagues, studying 926 recipients, demonstrated a linear relationship between the number of HLA-DQ antibody-verified eplet mismatches and dnDSA risk, finding no threshold below which risk vanished. Beyond acute events, a case-control study found a 25 percent increase in the odds of transplant glomerulopathy for every 10 additional HLA-DR/DQ eplet mismatches, and Sapir-Pichhadze and colleagues independently linked eplet mismatch burden to death-censored graft failure.

How should clinicians translate these findings into risk categories? Two influential studies embody competing philosophies. Wiebe’s group identified clinically informative thresholds — HLA-DR at 7 or more, or HLA-DQ at 9 or more eplet mismatches — and stratified recipients into low, intermediate and high risk. Senev’s team, by contrast, treated risk as a continuum, deliberately avoiding any implication of a biologically safe mismatch burden. The review emphasizes that reported thresholds are method-specific, varying with the mismatch algorithm, typing resolution, HLA locus, population, immunosuppressive regimen and outcome measured, and should be regarded as risk-stratification tools requiring external validation rather than universal biological cut-offs. More recent work by Schawalder and colleagues added a further wrinkle: individual HLA-DQ eplet mismatches show substantial heterogeneity in their propensity to provoke antibody formation, meaning mismatch quantity alone incompletely captures alloimmune risk.

Perhaps the most clinically provocative evidence concerns the interaction between molecular mismatch and tacrolimus exposure. In 596 recipients, Wiebe’s team found that both eplet mismatch and tacrolimus exposure independently predicted dnDSA, but the greatest risk occurred among recipients with more than 11 HLA-DR or -DQ eplet mismatches and trough levels below 5 ng/mL; recipients with low mismatch tolerated lower drug exposure with substantially less antibody formation. Davis and colleagues, in 444 donor–recipient pairs, confirmed an interaction whereby intermediate- and high-risk recipients faced elevated dnDSA risk at mean tacrolimus levels below 6 ng/mL compared with above 8 ng/mL. Medication adherence tells a parallel story: among 195 recipients, non-adherence carried the gravest consequences for those with 17 or more HLA-DQ eplet mismatches, with graft loss of 33 percent versus 10 percent. These are observational associations, the review cautions — exposure was not randomly assigned, and trough concentrations imperfectly capture variability and interruptions — but they suggest mismatch burden modifies the danger of under-immunosuppression.

The failures of past minimization trials sharpen the argument. The CTOT-09 randomized trial enrolled recipients meeting every conventional criterion of low immunological risk — panel reactive antibody under 30 percent, no donor-specific antibody, negative flow crossmatch, stable graft function and biopsies — yet tacrolimus withdrawal produced acute rejection and/or dnDSA in 8 of 14 recipients, or 57 percent, forcing premature termination. A post-hoc analysis associated HLA-DQ molecular mismatch with these events. Similarly, in a TRITON sub-study of 29 recipients undergoing tacrolimus minimization, 52 percent of those with 11 or more HLA-DQ eplet mismatches developed dnDSA, compared with none below that threshold, though the sample was tiny. Generalizability is supported by a predominantly Asian cohort of 234 recipients, mostly on cyclosporine, in whom only 1 percent of low molecular mismatch recipients developed dnDSA. Still, the review is unambiguous: no prospective trial has yet assigned immunosuppression intensity by molecular mismatch, so it remains a potential tool for risk-adapted therapy, not an established one.

Risk-adapted surveillance may arrive sooner than treatment guidance. De novo donor-specific antibodies carry sobering prognostic weight: a meta-analysis of 10,535 recipients across 14 countries found dnDSA associated with antibody-mediated rejection at a relative risk of 9.66, chronic antibody-mediated injury and transplant glomerulopathy at 6.78, and graft loss at 4.95. The 2023 European Society for Organ Transplantation consensus recommended antibody monitoring around three to six months post-transplant and annually, with biopsy guiding any treatment escalation. Wiebe’s prospective cohort of 949 recipients went further, combining recipient age with HLA-DR/DQ molecular mismatch into RAMM risk categories: roughly 52 percent fell into the low-risk group with 97 percent subclinical dnDSA-free survival at 10 years, while risk rose more than sixfold in the highest category. The authors estimated RAMM-based surveillance could cut antibody testing by more than half while concentrating monitoring where it matters.

The field is also moving beyond any single score. The PIRCHE-II algorithm estimates donor-derived HLA peptides presentable by recipient class II molecules to CD4 T cells; in 2,787 recipients, higher scores predicted dnDSA and poorer graft survival. A registry analysis of 400,935 transplants by Niemann and colleagues found that multivariable molecular compatibility models outperformed individual metrics and added incremental predictive value to clinical variables. Non-HLA genomic mismatch contributes independently too — Kammer’s analysis of 1,187 pairs found a composite HLA and non-HLA score associated with graft loss at a hazard ratio of 1.76, with the two measures essentially uncorrelated. Dynamic biomarkers such as donor-derived cell-free DNA and prognostic systems like iBox could eventually update a static pre-transplant estimate into a longitudinal risk model. The review’s bottom line is measured: molecular mismatch is now a reproducible prognostic biomarker, particularly for dnDSA and antibody-mediated injury, but demonstrating that mismatch-guided intervention improves clinically meaningful outcomes — without disproportionate infectious, malignant or access-related costs — is the next, still unmet, step for precision transplantation.

Subject of Research: HLA molecular mismatch as a prognostic biomarker for alloimmune risk and precision immunosuppression in kidney transplantation

Article Title: HLA molecular mismatch – A step towards precision medicine in transplantation for individualized immune risk assessment and immunosuppression

Article References: Wong, C. J. Y., & Wong, E. T. Y. (2026). HLA molecular mismatch – A step towards precision medicine in transplantation for individualized immune risk assessment and immunosuppression. Frontiers in Medicine, 13, Article 1990225. https://doi.org/10.3389/fmed.2026.1990225

Image Credits: AI Generated

DOI: 10.3389/fmed.2026.1990225

Keywords: HLA molecular mismatch, eplet mismatch, kidney transplantation, donor-specific antibodies, antibody-mediated rejection, immunosuppression, tacrolimus, precision medicine, PIRCHE, transplant immunology, graft survival, risk stratification