HLA Typing for Cord Blood Banking and Transplantation: Ensuring Accurate Matching with NGS-Based Methods

Meta Intent: A practical, research-focused guide to designing NGS-based HLA evidence for cord blood inventory records, match review, confirmatory identity checks, and auditable reporting.

Cord blood HLA typing is easy to oversimplify. A unit can carry a familiar match score, a high-resolution allele assignment, and a bank identifier, yet still be difficult to compare if the report does not state which specimen was typed, which loci were observed, which database release was used, or whether a later result confirms the same cryopreserved unit. For cord blood programs, the useful deliverable is therefore not merely an HLA string. It is a traceable evidence record that survives the transition from collection to inventory listing, from preliminary search to candidate review, and from attached-segment confirmation to final research documentation.

This distinction matters because cord blood is not an adult donor sample that can simply be recollected. The original unit, an attached segment, archived maternal material, and later recipient-side data may all enter a comparison at different times. Their analytical roles are related but not interchangeable. NGS improves allele resolution and can reduce familiar ambiguity, but it does not repair a weak identity chain, a poorly specified match calculation, or a report that hides the boundary between directly observed and inferred information.

This article treats the workflow as four linked evidence layers: identity, locus coverage, match representation, and confirmation logic. It complements the matrix Spoke 1, High-Resolution HLA Typing by NGS, which addresses locus-level resolution and phasing. The broader HLA Typing and Immune Repertoire Sequencing Services Hub remains a planned matrix destination and is deliberately not linked until its page is live.

Figure 1: Cord blood HLA evidence lifecycle. The figure separates collection identity, inventory typing, match review, and attached-segment confirmation so that one HLA result is not mistaken for proof of every stage.

Why Cord Blood Units Need a Different HLA Evidence Model

Cord blood expands the available graft-source inventory partly because matching constraints have historically been less restrictive than in some adult unrelated-donor settings. That does not mean HLA evidence is less important. It means that match information must be interpreted with the unit's other documented attributes and with the specific research or translational question in view. A compact score such as 4/6, 6/8, or 8/8 can be useful as a first filter only if the reader knows the loci, resolution, directionality convention, and data source that produced it.

A match score is not a portable fact. One report may count HLA-A and HLA-B at antigen-level equivalence with DRB1 at allele level; another may use allele-level calls across HLA-A, -B, -C, and -DRB1. Both can describe the same pair of samples, but they encode different biological detail. The first is a screening representation. The second preserves more of the sequence differences that can matter when candidate units look similar by a coarser calculation. A reliable record stores the full allele calls and then derives any score from a declared rule, instead of storing only the score.

The practical implication is to create two distinct fields in every inventory-ready record: the source HLA call and the match representation. The source call identifies the locus, field level, ambiguity notation, database version, and assay boundary. The match representation identifies the comparison rule used for a particular search. This separation prevents a later team from treating a historical shortlist label as if it were an allele-level laboratory result.

Figure 2: Match score versus source evidence. A unit and recipient are shown with complete locus-level calls feeding a declared matching rule; the score is visibly downstream of the underlying sequence evidence.

Start With the Decision Point, Not a Generic Locus Panel

A cord blood program usually faces at least three different HLA questions. At inventory creation, it needs a typed record that makes a unit discoverable and comparable. During candidate review, it needs the exact loci and resolution needed by the predefined matching logic. Before a unit is used in a downstream program, it needs an identity-confirming result from an appropriate linked specimen or segment. Combining these questions in one vague request for "high-resolution HLA typing" produces predictable friction: the initial assay may not support the intended comparison, or a later confirmation may be incorrectly judged against an inventory result reported at another resolution.

Define the evidence request in operational language. For example: "Create an inventory call for HLA-A, -B, -C, and -DRB1 at the reportable field level; retain raw ambiguity information; flag incomplete loci; and retain a linked specimen identifier for later confirmation." This is more useful than saying "type the unit comprehensively." It defines the report, the data fields, and the future handoff before any DNA is extracted.

For focused, pre-specified loci, Targeted Region Sequencing can be aligned to a declared reportability boundary. Amplicon Sequencing Services may support high-depth interrogation when the unit record clearly states the amplified regions. The key is not the label on the assay; it is whether its observed intervals and phase information support the fields that the inventory intends to publish.

Specify the Match Rule as Data, Not Narrative

Terms such as "matched," "permissive," and "acceptable mismatch" are not analytical outputs by themselves. They are conclusions produced after a rule is applied to calls. A reproducible match rule should name: the loci included; the resolution used at each locus; whether an antigen, allele, G-group, or P-group comparison is made; how unresolved ambiguity is handled; whether a mismatch direction is recorded; and whether any optional loci are used only for tie-breaking or exploratory analysis.

For example, a research database may first rank units with a predeclared four-locus allele-level comparison, then show DQB1, DPB1, maternal typing, or KIR-ligand data as additional contextual fields rather than silently adding them into the primary score. That approach lets a user see why a rank changed and prevents an optional biological annotation from being misreported as a core matching criterion.

Do not hard-code a universal "best" algorithm into an article or inventory. Institution-specific procedures, current standards, and the exact research protocol govern how any candidate is reviewed. The transferable practice is to make the algorithm inspectable. A match calculation should be reproducible from the stored calls without relying on memory, a spreadsheet note, or a deprecated export.

Figure 3: Declared HLA match-rule architecture. A modular computation view shows locus set, resolution, ambiguity handling, optional context fields, and an auditable output rather than a black-box score.

Build Inventory Typing Around Unit Identity

HLA typing cannot compensate for an identity problem. Every result should be associated with a durable unit identifier, the exact material typed, the collection or processing relationship of that material to the cryopreserved unit, extraction batch, library batch, and the report version. This does not require turning an HLA report into a full manufacturing dossier. It does require enough lineage information for a reviewer to answer a basic question: "What molecule was tested, and how is it linked to the listed unit?"

Attached segments are particularly valuable because they allow later testing while preserving the main cryopreserved unit. But a segment is not simply a technical convenience. It is a specimen relationship that must be documented. A confirmation result from a segment is persuasive only when the segment-to-unit linkage is retained, the specimen label is unambiguous, and the comparison uses compatible nomenclature and database releases.

A useful inventory schema therefore keeps an identity block separate from the analytical block. The identity block includes unit ID, segment ID, source-material type, collection-to-processing linkage, and versioned chain-of-custody references. The analytical block includes loci, assay boundaries, database release, caller version, allele calls, ambiguity flags, coverage exceptions, and review status. This structure is more resistant to handoff errors than a single free-text result field.

Figure 4: Unit-to-segment identity chain. A cryopreserved cord blood unit, attached segment, extracted DNA aliquot, library, and final HLA report are connected by persistent identifiers and QC checkpoints.

Choose an NGS Design That Can Defend the Inventory Call

An NGS method should be selected from the report backward. A short, high-depth targeted design may be appropriate for a defined set of polymorphic regions and a specific group- or allele-level output. A broader amplicon architecture may add differentiating positions. A long contiguous design can be valuable when the practical question is phase across a locus or distinction among candidates that differ outside conventional exon targets. None of these options is automatically "more accurate" in the abstract; each makes a different type of evidence available.

For cord blood records, the most important assay specification is not only the nominal resolution. It is the combination of locus inclusion, observed intervals, allele balance, phasing capability, and no-call policy. A report that formats every locus as a four-field allele but does not identify unobserved discriminating regions can overstate what the underlying sequence proves. Conversely, a precisely documented two-field call may be fully adequate for a narrow, predeclared comparison.

Long Amplicon Analysis is relevant when continuous locus-scale evidence is needed to resolve phase or to examine regions beyond a short target. Nanopore Target Sequencing can be considered when long molecular spans are central to the stated evidence question. Both still require an explicit coverage and calling contract; a long read is not a substitute for validating which loci, alleles, and quality rules the workflow can report.

Do Not Treat DNA Yield as the Only Pre-Analytical Variable

Cord blood material is finite, and that changes the pre-analytical conversation. The right question is not "How little DNA can the assay accept?" It is "What molecular quality and specimen relationship are needed for the claim we plan to make?" A short amplicon design can tolerate different DNA fragmentation than a long-range design. A low-input workflow may be useful for a limited confirmatory question but may not provide equivalent locus continuity. An extraction result should be assessed against the intended library architecture, not against a generic concentration threshold alone.

Plan sample allocation before testing. Define which material supports inventory typing, which linked material is retained for confirmation, which aliquot is reserved for a failed-library investigation, and how duplicate extractions will be labeled. This helps a bank or research program avoid spending the only traceable material on an assay that does not answer the future comparison question.

Where a project needs wider genomic context in addition to HLA loci, Whole Genome Sequencing or Whole Exome Sequencing should be described as complementary data sources with locus-specific coverage limits, not as automatic replacements for a targeted HLA design. Existing WES/WGS data may support a screening inference only when coverage, reference construction, and caller validation are demonstrably suited to the intended HLA report level.

Figure 5: Pre-analytical design for finite cord blood material. The illustration allocates linked material to inventory typing, confirmation reserve, re-extraction reserve, and long-span escalation without implying that all aliquots support identical claims.

Use Confirmation to Test Identity and Reportability

Confirmatory typing should answer a defined question. It may test whether a linked segment and listed unit are concordant at the required loci, whether an earlier ambiguous call can be resolved with a broader design, or whether a data-handling discrepancy is analytical rather than biological. "Repeat the test" is not a sufficient confirmation plan because repeating the same limited assay on the same source material may reproduce the same information gap.

Write escalation rules before the first inventory run. Typical triggers include a locus no-call, strongly imbalanced allele support, unresolved competing alleles, a phase boundary outside the assay span, a mismatch between inventory and confirmation material, or a mismatch calculation that changes after database revision. For each trigger, specify the next evidence type: re-extraction, redesigned target, extended amplicon, long-span sequencing, or a local orthogonal check.

Sanger Sequencing can be useful for a defined local sequence question, but it should not be presented as a universal answer to full-locus phase or complex allele ambiguity. The confirmation report should state exactly what the method tested and whether it confirms a base position, a limited interval, a locus assignment, or a phased allele pair.

Figure 6: Confirmation escalation map. A quality-trigger framework routes no-call, low balance, interval gap, phase gap, and identity discordance to distinct follow-up evidence rather than a generic re-test.

Keep the Analysis Versioned and Reviewable

HLA analysis is not a one-time software event. Allele databases expand, naming conventions are maintained, and callers may change their handling of multimapping reads or ambiguity. A cord blood inventory should preserve the versioned basis of its original report and make reanalysis traceable when it is performed. Re-running a historical sample against a later database can be scientifically valuable, but the new result should not silently overwrite the original listing record.

Store the database release, caller version, locus configuration, filters, and report-generation date alongside each call. When reanalysis changes a result, record whether the change reflects a newly cataloged allele, a database-name revision, a different ambiguity policy, or a newly available sequence interval. This creates a meaningful change log and stops a later user from interpreting two versioned results as a specimen identity failure.

Variant Calling workflows can contribute to a wider sequencing analysis, but their HLA use must remain conditional on the data geometry and validated caller assumptions. The same principle applies to Gene Panel Sequencing Service designs: reportability comes from demonstrated locus coverage and calling behavior, not from the presence of an HLA gene in a panel list.

Design for Data Exchange, Not Just a Local Report

Cord blood inventories frequently move across laboratories, registries, and review teams. The greatest practical risk is not that a future user cannot open the original PDF; it is that they can open it but cannot safely compare it with another result. For that reason, an NGS HLA record should have a machine-readable core that survives software, language, and staff changes. At minimum, retain the canonical locus names, allele strings exactly as issued, the nomenclature/database release, ambiguity or no-call status, specimen relationship, analysis version, and the date on which the record was generated. Free-text interpretation can sit alongside these fields, but it must not be the only representation.

Normalize before comparing. Strip no suffixes, collapse no fields, and silently convert no historical name to a newer name without retaining a mapping record. A reviewer should be able to see whether two outputs are genuinely discordant or merely expressed with different field depths, G/P-group notation, or database releases. If a registry export uses a coarser representation than the internal NGS report, store the transformation rule and retain the original source call. This is especially important when a cross-site review starts from an inventory file rather than the sequencing analysis workspace.

Use a discrepancy taxonomy rather than one generic "mismatch" label. A nomenclature discrepancy may result from a renamed allele. A coverage discrepancy may arise when one assay did not observe a differentiating region. A phase discrepancy may reflect unlinked short fragments. An identity discrepancy requires immediate specimen-linkage investigation. These categories lead to different actions and should never be collapsed into a single unresolved status. The right escalation is evidence-specific: compare versions, inspect coverage, obtain longer molecular linkage, or audit the unit-to-segment chain.

For a bank creating records for multi-site use, add an exchange-readiness gate before release of the report: all required loci are present; field level is declared; the result can be parsed without local abbreviations; ambiguity flags are carried through; and the source-unit relationship is intact. This gate is not an accreditation claim. It is a practical data-quality discipline that makes HLA information usable after the original analyst, instrument batch, and local file path are gone.

Figure 7: Exchange-ready HLA record. A portable data record preserves canonical calls, database release, ambiguity state, specimen relationship, and version history as separate fields that can be compared across sites.

Separate Inventory Qualification From Candidate Selection

Public and directed banking models may differ in when a detailed HLA record is generated and how it is used, but both benefit from separating inventory qualification from downstream candidate selection. Qualification asks whether the unit record is complete, linked to the correct material, and fit for listing under the program's defined data model. Candidate selection applies a specific comparison rule to a defined recipient-side dataset and incorporates non-HLA unit information within the authorized workflow. These are different decisions and should produce different outputs.

The distinction prevents a common reporting error: calling an inventory record "suitable" merely because it has a typed HLA profile. A typed unit is a discoverable record. Whether it is selected for any specific downstream use depends on a comparison context that is outside the standalone laboratory call. This article does not prescribe selection thresholds or medical decisions; it describes how an NGS HLA report can remain technically interpretable when used in a controlled review process.

Figure 8: Inventory qualification versus candidate selection. The left lane validates unit record completeness and traceability; the right lane applies a declared comparison rule and authorized non-HLA review without collapsing the two decisions.

A One-Page Cord Blood HLA Evidence Worksheet

Before sending samples, use a one-page worksheet to align the laboratory request with the inventory record. It should contain four blocks:

• Identity block: unit ID, linked segment ID, material typed, collection-to-processing relationship, and retained confirmation material.
• HLA evidence block: required loci, minimum reportable resolution, acceptable group notation, phase requirement, and positions or intervals that must be observed.
• Match representation block: named matching rule, comparison level by locus, ambiguity handling, optional context fields, and output format.
• Escalation block: no-call policy, discordance definition, re-extraction criteria, orthogonal method, and versioned reanalysis policy.

This worksheet is deliberately method-neutral. It can guide an HLA Typing request, a targeted follow-up, or a long-span escalation. Its value is that it forces a program to define what the evidence must prove before selecting a technology.

Figure 9: One-page cord blood HLA evidence worksheet. An isometric laboratory worksheet is divided into identity, HLA evidence, match representation, and escalation blocks with concise data labels.

Frequently Asked Questions

Research Use Only. This content supports research, inventory-design, and translational workflow discussion; it does not direct individual care decisions.

References:

1. Gupta AO, Wagner JE. Umbilical Cord Blood Transplants: Current Status and Evolving Therapies. Frontiers in Pediatrics. 2020;8:570282. DOI: 10.3389/fped.2020.570282.
2. Cornaby C, Montgomery MC, Liu C, Weimer ET. Unique Molecular Identifier-Based High-Resolution HLA Typing and Transcript Quantitation Using Long-Read Sequencing. Frontiers in Genetics. 2022;13:901377. DOI: 10.3389/fgene.2022.901377.
3. Lee M, Seo J-H, Song S, et al. A New Human Leukocyte Antigen Typing Algorithm Combined With Currently Available Genotyping Tools Based on Next-Generation Sequencing Data. Frontiers in Immunology. 2021;12:688183. DOI: 10.3389/fimmu.2021.688183.