Interindividual HLA Evolutionary Divergence in Single HLA-Mismatched Unrelated Donor Hematopoietic Cell Transplantation for Malignant Hematological Disorders: A Report on Behalf of the Cellular Therapy and Immunobiology Working Party of the EBMT

This study demonstrates that quantifying HLA evolutionary divergence (HED) in 9/10 mismatched unrelated donor transplants reveals significant within-mismatch heterogeneity and complex, locus-specific interactions that independently predict clinical outcomes such as relapse-free survival, graft-versus-host disease, and mortality, thereby complementing conventional mismatch classification.

The Gist

Imagine you are trying to send a very specific, delicate package (a new immune system) from a donor to a recipient to cure a serious blood disease. This is what happens in a bone marrow transplant.

The biggest challenge is that the recipient's body might reject the package (Graft-versus-Host Disease), or the package might not be strong enough to fight the remaining cancer (Relapse). To prevent this, doctors try to find a donor whose "ID card" (called HLA) matches the recipient's as closely as possible.

However, sometimes a perfect match isn't available. Doctors have to use a donor who is a "9 out of 10" match. For a long time, doctors thought that any "9 out of 10" match was roughly the same risk. If you missed one letter on the ID card, it was just a mismatch.

This paper says: "Not all mismatches are created equal."

Here is the breakdown using simple analogies:

1. The Old Way: Counting Typos

Imagine the HLA ID card is a sentence made of letters.

• The Old View: If the donor's sentence has one different letter than the recipient's, it's a "mismatch." Doctors treated all single-letter typos the same.
• The Problem: A typo of the letter "A" changing to "B" is very different from "A" changing to "Z." One might be a harmless spelling error; the other might completely change the meaning of the sentence.

2. The New Idea: "Evolutionary Distance" (HED)

The researchers introduced a new tool called HED (HLA Evolutionary Divergence).

• The Analogy: Think of the HLA molecules as locks and the immune system's keys as keys.
• The "peptide-binding groove" is the keyhole.
• HED measures how different the shape of the keyhole is between the donor and the recipient.
  • If the donor's keyhole is slightly different, the recipient's immune system might get confused.
  • If the donor's keyhole is wildly different (evolutionarily distant), the immune system might go into overdrive, causing a severe reaction.

3. What They Found

The researchers looked at nearly 4,700 patients who had a "9 out of 10" match. They didn't just count the mismatch; they measured how different the mismatched parts were.

• The "Class I" vs. "Class II" Difference:

  • Class I (HLA-A, B, C): These are like the main doors of a house. If the main door is mismatched, it's usually bad news. The study confirmed that mismatches here are generally riskier for survival.
  • Class II (HLA-DR, DQ): These are like the side windows. Historically, doctors thought a mismatch here was very dangerous. But this study found that on average, a mismatch in the "window" (DRB1) is actually less dangerous than a mismatch in the "door" (Class I).

• The "Quality" of the Mismatch Matters:

  • Even within the "window" (DRB1) group, some mismatches are worse than others.
  • The Analogy: Imagine two people have a mismatched window.
    • Person A: The window is slightly cracked (low evolutionary distance). The immune system ignores it.
    • Person B: The window is a completely different shape and size (high evolutionary distance). The immune system panics, thinking it's an invader, leading to complications.
  • The study found that for patients with a DRB1 mismatch, if the "shape difference" (HED) was high, they had a higher risk of the cancer coming back later.

• Cross-Talk (The "Domino Effect"):

  • The study found that the "shape" of the mismatched part affects the other parts of the ID card, too.
  • Analogy: If you change the lock on the front door, it might change how the back door works. For example, if a patient has a mismatch in the "B" gene, the diversity of their "DR" gene (even if it matches!) can change the outcome. It's a complex dance where every part of the immune system is connected.

4. The "DPB1" Surprise

There is a specific gene called DPB1 that doctors often worry about.

• The Finding: In this specific group of patients (who already had one other mismatch), the "shape difference" in DPB1 didn't seem to matter much. It was like having a slightly different keyhole on a side door when the main door was already broken; the side door didn't add much new risk.

Why Does This Matter?

This paper is like upgrading from a black-and-white map to a 3D topographical map.

• Before: "You have a mismatch. It's risky."
• Now: "You have a mismatch. But let's measure how different it is. If the difference is small, it might be safe. If the difference is huge, we need to be very careful."

The Takeaway:
Doctors can now use this "Evolutionary Distance" math to pick the best "9 out of 10" donor. Instead of just picking any donor with one mismatch, they can pick the one whose mismatch is the "least shocking" to the recipient's immune system. This could mean fewer complications, less chance of the cancer returning, and more lives saved.

In short: It's not just about how many differences there are; it's about how different those differences are.

Technical Summary

1. Problem Statement

Allogeneic hematopoietic cell transplantation (allo-HCT) relies on a delicate balance between the graft-versus-tumor (GvT) effect and graft-versus-host disease (GvHD). While donor selection typically prioritizes allele-level HLA matching, single-mismatched unrelated donors (9/10 MMUD) are often necessary. Current clinical practice treats HLA mismatches as binary events (matched vs. mismatched) or relies on locus-specific hierarchies (e.g., Class I vs. Class II). However, this approach fails to capture the functional heterogeneity within mismatch categories. Two patients with a mismatch at the same locus (e.g., HLA-DRB1) may experience vastly different outcomes because the specific alleles involved differ in their peptide-binding capabilities. There is a need for metrics that quantify the "functional distance" or immunopeptidome divergence between donor and recipient alleles to better predict post-transplant outcomes.

2. Methodology

Study Design and Cohort:

• Type: Retrospective registry-based study.
• Data Source: European Society for Blood and Marrow Transplantation (EBMT) database.
• Cohort: 4,695 adult patients (≥18 years) with acute leukemia, MDS, or MPN undergoing first allo-HCT from a 9/10 MMUD between 2010 and 2019.
• Exclusion Criteria: Patients receiving post-transplant cyclophosphamide (PTCy) were excluded to isolate the immunogenetic effects of HLA mismatching.
• Inclusion Criteria: High-resolution (2-field) HLA typing for loci A, B, C, DRB1, and DQB1.

Key Metrics (HLA Evolutionary Divergence - HED):
The study utilized HED, a metric based on Grantham distance, to quantify amino acid divergence at peptide-binding regions. Two categories were analyzed:

1. Individual HED (HED-R and HED-D): The divergence between the two alleles within a single individual (recipient or donor) at a specific locus. This reflects the individual's intrinsic immunopeptidome diversity.
2. Interindividual HED-Mismatch (HED-MM): The divergence specifically between the non-shared donor and recipient alleles at the mismatched locus. This quantifies the "functional distance" introduced by the mismatch.

Statistical Analysis:

• Endpoints: Overall Survival (OS), Relapse-Free Survival (RFS), Non-Relapse Mortality (NRM), Relapse, Acute GvHD (aGvHD), and Chronic GvHD (cGvHD).
• Models: Cause-specific Cox proportional hazards models were used.
• Adjustments: Models were adjusted for clinically relevant covariates (diagnosis, Disease Risk Index, age, CMV status, conditioning intensity, graft source, T-cell depletion, GvHD prophylaxis, TBI, sex mismatch, KPS).
• Stratification: Analyses were performed within specific mismatch subgroups (e.g., A-mismatch, B-mismatch, DRB1-mismatch) to assess locus-specific and cross-locus interactions.

3. Key Contributions

• Refinement of Mismatch Classification: The study moves beyond binary mismatch classification by introducing HED metrics that capture qualitative differences in donor-recipient immunopeptidome interactions.
• Locus-Specific Heterogeneity: It demonstrates that the clinical impact of a mismatch is not uniform even within the same locus; the degree of evolutionary divergence (HED) significantly modulates outcomes.
• Cross-Locus Interactions: The research identifies complex, non-linear interplays where recipient HED at one locus (e.g., Class II) influences outcomes in the context of a mismatch at a different locus (e.g., Class I).
• Differentiation of DRB1 Impact: It challenges the traditional view that DRB1 mismatches are uniformly deleterious, showing they are generally less detrimental than Class I mismatches but harbor significant internal heterogeneity driven by HED-MM.

4. Key Results

General Mismatch Impact:

• Class I vs. Class II: Class I mismatches (HLA-A, -B, -C) were associated with significantly worse OS, RFS, higher NRM, and increased GvHD compared to Class II (DRB1, DQB1) mismatches. HLA-A mismatches conferred the highest risk.
• Relapse: Interestingly, the cumulative incidence of relapse did not differ significantly across mismatch groups, suggesting HLA mismatching primarily drives NRM and GvHD rather than GvT efficacy in this cohort.

HED Metric Findings:

• DRB1-Mismatched Subgroup:
  • HED-MM: Higher interindividual HED-MM at DRB1 independently predicted inferior RFS immediately post-transplant (HR 1.14), though this effect attenuated over time.
  • HED-R: Higher recipient HED at HLA-A and HLA-C in this subgroup was associated with increased risks of acute GvHD and NRM, respectively.
• HLA-B-Mismatched Subgroup:
  • Cross-Locus Effect: Higher recipient HED at DRB1 (Class II) was associated with worse OS, RFS, and increased relapse risk. Conversely, higher recipient HED at DQB1 appeared protective.
• HLA-A-Mismatched Subgroup:
  • Higher recipient HED at HLA-A was associated with an increased risk of chronic GvHD.
• HLA-DPB1:
  • In the subset of patients with a concurrent DPB1 mismatch, neither recipient HED-R nor HED-MM at DPB1 showed significant associations with any clinical endpoint. This suggests that in the presence of a primary Class I/II mismatch, DPB1 divergence adds limited incremental predictive value.

5. Significance and Clinical Implications

• Risk Stratification: HED metrics provide a tool to refine risk stratification among patients with the same "9/10 mismatch" status. For example, a DRB1 mismatch with high HED-MM carries a higher risk of early relapse-free survival failure than one with low HED-MM.
• Donor Selection: In non-PTCy platforms (common in Europe), donor selection algorithms could incorporate HED to identify "permissive" mismatches (low functional divergence) versus high-risk mismatches, potentially improving survival outcomes.
• Post-Transplant Management: Patients with high HED metrics (indicating high immunopeptidome divergence) might benefit from tailored immunosuppression or closer monitoring for GvHD and relapse in the early post-transplant period.
• Biological Insight: The findings support a model where alloreactivity is driven by the global landscape of HLA divergence and immunopeptidome presentation, rather than isolated allele mismatches. The time-dependent attenuation of HED-MM effects in DRB1 suggests that early immune reconstitution dynamics are particularly sensitive to these functional disparities.

Limitations:

• Retrospective design limits causal inference.
• Exclusion of PTCy limits generalizability to modern transplant protocols where PTCy is standard.
• HED is a sequence-based proxy and does not directly measure peptide processing, expression levels, or T-cell receptor recognition.
• Homozygous genotypes are assigned a value of zero, potentially collapsing biologically distinct states.

In conclusion, this study establishes that HLA evolutionary divergence is a critical, independent predictor of transplant outcomes, offering a more nuanced, biologically grounded approach to donor selection and risk management in mismatched unrelated donor transplants.