The registry types nearly every donor for HLA — and almost never by DNA.
Coverage is not the limitation. Resolution is. STAR records broad and split antigens, not alleles, and the donor side has barely moved in twenty years. That single fact decides which matching questions the registry can answer.
Source: OPTN/UNOS STAR, Dec 2025 release · 163,764 HLA typing records; 60,093 joined to a lung or heart-lung transplant.
3.3–3.8%donor molecular typing, 2020–25, PIRCHE core loci
Recipient molecular typing at HLA-A rose from 0.0% before 2005 to 15.0% today. Donor typing went to 3.3%. Epitope matching needs both sides.
01 · The trend
Molecular typing is rising — on one side of the pair
Recipient typing is elective and can be done by DNA at leisure. Donor typing happens under allocation time pressure, where a serologic result in hours beats a molecular result tomorrow. The registry records the consequence.
Molecular share at HLA-A (%)
donorrecipient
Donor molecular typing by locus, 2020–2025 (%)
DPB1 is the exception that proves the rule. It is 96.1% molecular on the donor side — because HLA-DP has no practical serologic assay. When DP is typed at all it is typed by DNA. Every locus with a serologic alternative stays serologic. DQA1 is the opposite pole: typed for about half the cohort and 0.0% molecular throughout.
02 · What it rules out
Epitope matching is an imputation problem, not a selection problem
PIRCHE-II and comparable algorithms need molecular typing at HLA-A, -B, -C, -DRB1 and -DQB1 on both sides of the pair. Applying that to this cohort leaves 59 pairs — 0.1%.
Restricting to recent transplants does not rescue it. That is the obvious next move and it fails: across the five core loci, donor molecular typing in 2020–2025 runs 3.3–3.8%. The donor is the binding constraint, and the donor is still typed serologically. Epitope analysis on this registry means imputing alleles from serologic antigens against a reference panel — a defensible method that inserts a second model between the data and the result, with error that is larger in under-represented ancestries.
03 · What it supports
Antigen mismatch — computable, and heavily skewed
The classical ABDR 0–6 count resolves for 84.3% of pairs (62% before 2005, 91% today). Mean mismatch is 4.64 of a possible 6.
Share of pairs by ABDR mismatch count (%)
This is what allocation without an HLA term looks like. 59.8% of pairs sit at 5 or 6 mismatches and just 50 pairs in the entire cohort are fully matched. The variable has good spread at the poorly-matched end and almost no observations at the well-matched end — so the comparison that would matter clinically rests on a few hundred pairs.
04 · The outcome test
Mismatch does not predict early graft dysfunction
Severe PGD at 72 hours across the full mismatch range moves 2.1 percentage points, with intervals overlapping throughout.
Severe PGD at 72 h by ABDR mismatch (%)
ABDR mismatches
Recipients graded
Severe PGD at 72 h
0-3 (better)
2,328
31.2% [29.4–33.1]
4
4,180
31.6% [30.2–33.0]
5
5,980
33.4% [32.2–34.6]
6 (full)
3,663
33.1% [31.6–34.6]
Report this as a positive result, not a disappointing null. Severe PGD is ischemia-reperfusion injury expressed within 72 hours, before adaptive alloimmunity plausibly contributes. A large antigen-mismatch effect would have been surprising; its absence fits the mechanism. It also means HLA matching has nothing to offer a PGD-prediction model — a useful negative for feature selection.
The question this data could answer is the long-horizon one — chronic rejection, bronchiolitis obliterans, graft survival at five and ten years — where alloimmunity is the mechanism and the same 84%-computable mismatch applies. That analysis has not been run.
One instance of a pattern. Across this series of briefs, five clinically decisive quantities turned out not to be collected at all: implant sequence in multi-organ procedures; the offer sequence a candidate was reached at; the lung perfusion device; and molecular rejection surveillance. Each is recorded somewhere in the transplant system — by a centre, a laboratory, a manufacturer, or OPTN itself — and none reaches the research release. The registry records the organ’s disposition but not its journey, the graft’s dysfunction grade but not the operative sequence that produced it. Distinguishing a question that is underpowered from one that is unanswerable in principle is the discipline that separates the two.