What a swap transplant is, what the registry needs from your centre, and exactly what the matching engine does with it.
A patient has a willing living donor, but that donor cannot give to them — the blood groups are incompatible, or the patient carries antibody against the donor's tissue type. A second pair is in the same position. If each donor can give to the other pair's recipient, the two pairs exchange and both patients are transplanted.
Nothing is bought, sold or promised beyond the exchange itself. It is a barter between two families, each of whom already had a donor willing to give.
Because the odds improve with the size of the pool. Two incompatible pairs at one hospital may not resolve each other; the pair that resolves yours may be four hundred kilometres away. A centre searching only its own list is searching the smallest possible pool, and the mismatch that blocks your patient is very often the exact mismatch another centre needs.
Kidney and liver. Both pairs in an exchange must be registered for the same organ — the registry will not propose a cross-organ swap.
No. The registry finds candidate exchanges, scores them, and produces the evidence two consultants need to decide. It does not authorise anything.
Statutory approvals, authorisation committee clearance, donor consent and every other legal and ethical requirement are unchanged and remain the responsibility of the transplanting centres. A match report is a clinical document, not a permission.
The transplant centre registers first. Once it is approved, its consultants and transplant coordinators are enrolled under it, and consultants register their own pairs.
Patients and donors do not register themselves. A pair reaches the registry through the centre that is looking after them.
Four things, and the pair screen tracks all four:
Until all four are present the pair still appears in your list, flagged, and is counted under Awaiting data on the dashboard — but it is skipped by every search. A pair that is registered and not matchable is invisible to other centres, so the registry says so on the front page of your dashboard rather than letting it sit there quietly.
The pair is excluded from matching until they are, and this is deliberate. Every antibody check works by comparing the recipient's antibodies against the donor's alleles. With no donor typing there is nothing to compare against, so every check finds nothing — and "found nothing" is indistinguishable from "found nothing wrong".
An untyped donor would otherwise be reported as a negative crossmatch with zero donor-specific antibody, which is a confident wrong answer about a whole donor. The safe answer to a question you cannot ask is not "acceptable".
Yes. Both identifiers are recorded against recipients and donors
alike — the 14-digit ABHA number and the
ABHA address (someone@abdm) — and the
registration form links straight to the ABDM portal for a patient
who does not yet have one.
The number is validated on its Verhoeff check digit, the same arithmetic the national ID uses, so a transposed pair of digits is caught on the form rather than stored wrong.
It is optional and will stay optional. Most people reaching a swap registry today do not have an ABHA, and a mandatory field would simply block their registration.
To be precise about the limit: the registry records and validates the identifiers, but does not yet exchange records over the ABDM network. Verifying an ABHA against the national registry runs through the patient's own consent flow, which a hospital user cannot complete on their behalf from an admin screen.
Questions about participation terms should go to the Honorary Secretary — see the contact page. We are not going to state a figure here that could be out of date by the time you read it.
Eleven: A, B, C, DRB1, DQB1, DPB1, DPA1, DQA1, DRB3, DRB4 and DRB5 — two alleles each. A, B, DRB1 and DQB1 are mandatory; the rest are recorded when you have them and improve the picture when you do.
Alleles are stored without the locus prefix — 01:01,
not A*01:01 — and a homozygous locus repeats the
allele.
Yes. Blank CSV templates for recipient typing, donor typing and SAB results are on the Downloads screen inside the portal, each with a worked example row. Two vendor layouts are read exactly as they are exported, so a plate does not have to be reshaped by hand:
| Layout | Header row the importer matches |
|---|---|
| Generic | antigen, mfi, serology_eq |
| One Lambda LABScreen | Sample ID, Bead ID, Specificity, Raw Value, Normalized Value, Serology |
| HLA Fusion | PatientID, SampleID, CatalogID, Specificity, NormalValue, Assignment |
The header row is what the importer matches on, so leave it untouched.
It is kept. One panel is marked current and drives matching; the others stay as history, so an earlier plate can be compared rather than overwritten.
A bead that reacts with an HLA allele the recipient carries themselves. Almost always a plate artefact rather than a real antibody — but if it is treated as real it blocks donors it should not, and it does so invisibly.
The SAB screen names them explicitly instead of quietly acting on them.
Six layers, in this order, for each direction separately:
Because compatibility is not symmetric, and this is the single most important thing to understand about swap matching.
Two pairs can have identical blood groups and an identical HLA score, and still one direction is a transplant while the other is a rejection waiting to happen — because the recipient on one side carries an antibody against a donor allele on the other, and the recipient on the other side does not.
When both directions work, the two pairs simply swap. When only one works, the exchange has to continue to a third pair — which is exactly why chains exist.
Points per matched donor allele, weighted by locus. These are the registry's configured weights, carried over from the previous ISOT system so that scores stay comparable:
| Locus | Points per matched allele |
|---|---|
| DRB1 | 15.0 |
| DQB1 | 7.5 |
| A | 4.0 |
| B | 4.0 |
| C | 2.0 |
| DPB1 | 2.0 |
A full A + B + DRB1 match therefore scores 46. Shading in the HLA grid marks a donor allele the recipient also carries — it is not a column-by-column comparison.
| Threshold | Default | Effect |
|---|---|---|
| Unacceptable | 2,500 MFI | the pairing is blocked |
| Positive | 1,500 MFI | the bead is reported as positive |
These are local policy, not a standard. They were inherited from the previous ISOT portal, they are configurable for the registry, and a consultant can override the blocking threshold on any individual search — upward to see what a strict cut-off is hiding, or downward to be more conservative.
MFI is semi-quantitative. Prozone, bead density and lot-to-lot drift all move it. It is not a titre, and no threshold on it is a fact of nature.
Every bead in the recipient's current SAB panel is matched against the donor's typed alleles. If none reacts, the virtual crossmatch is negative and the report says why in those words.
Serological beads — A2 rather than
A*02:01 — are resolved to molecular donor alleles
through the OPTN equivalence table, not only by exact string
match. A crossmatch that compared the two as text would call a
sensitised recipient negative, which is precisely the failure this
is built to avoid.
A virtual crossmatch is a screen. It does not replace the physical crossmatch your laboratory performs before transplant.
Eplets are the small patches of an HLA molecule that an antibody actually contacts. The report lists the eplets carried by the donor's mismatched alleles and not by the recipient's, and — separately — eplet sharing: antibodies that name an allele the donor does not carry, but which the donor's graft nonetheless presents through a shared eplet.
CREG is the older serological idea of cross-reactive antigen groups. It is reported because a serologist reads it quickly; where it disagrees with the eplet analysis, the eplet answer is finer.
Neither excludes anybody. They are used to rank candidates. There is no evidence base for eplet-guided paired exchange in any population, and none in an Indian one, and the registry is not going to imply otherwise. Eplet load is also summed across loci here, while published risk thresholds are per molecule — the two numbers must not be compared.
Candidates whose recipient has an antibody panel are listed first — a crossmatch that cannot be assessed is a different kind of unknown from one that has been assessed and came back clean. Within each of those groups, by a combined score:
HLA score − (worst DSA MFI ÷ 1000) − (eplet mismatches × 0.25)
So a slightly lower tissue match with a clean crossmatch outranks a slightly better one with antibody against it.
Your own pairs, your own hospital, a named hospital, a city or district, a state, or the whole registry — set per search, along with the MFI threshold and whether to include pairs whose recipient has no antibody panel yet.
Between two and fifteen. India has run chains of twelve; fifteen is the registry's configured ceiling.
A chain is a ring: each pair's donor gives to the next pair's recipient, and the last closes back to the first, so everybody who gives also receives.
No — and the total score is the most misleading number on the screen, because it simply adds up more hops. What compares across lengths is the weakest link: the lowest-scoring hop in the cycle. A fifteen-way chain is only as good as its worst transplant, and the report names that hop on its first page.
Length also has a practical cost. Every operation in a cycle has to happen at once. Fifteen pairs is fifteen consents, fifteen theatre slots and fifteen chances for the whole cycle to collapse, which is why the search returns shorter cycles ahead of longer ones.
That the engine enumerated every cycle of that length rather than pruning to keep the search tractable. When it has had to prune, the result says so.
This matters because it is the difference between "there is no chain" and "we did not find one". The registry never claims the first when only the second is true.
One document, in three parts:
A fifteen-way chain is therefore fifteen pair summaries in a single file.
No. Until a swap request between your pair and theirs is accepted, another centre sees the registry number, organ, blood group, readiness and the clinical data needed to judge the exchange — not the patient's name.
Masking is decided for each viewer at the moment the data is sent, never stored, so a pair that later becomes visible needs no migration. It applies to the pair list as well as the report: a name that has already appeared in a dropdown has leaked, and hiding it two screens later protects nobody.
No, and that is deliberate. Deciding whether an exchange is viable is the entire purpose of the report, and it cannot be done against redacted HLA typing or hidden MFI values. What is protected is identity. What is shared is immunology.
Only after a swap request between the two pairs is accepted. Lab reports, surgical summaries, scans and clearances attached to a pair belong to that pair's centre until then.
Every read and every write, including document downloads: who, what and when. Passwords are stored with Argon2 hashing, traffic is TLS-only, and access is role-based.
You send it from a match report. The receiving centre accepts or declines; the sending centre can withdraw it. All three outcomes notify the other side and can carry a note.
Only one open request can link the same two pairs at a time, in either direction. The HLA score and worst DSA at the moment of sending are stored on the request, so the decision stays reproducible even if a later SAB import changes the numbers.
On acceptance, both pairs move to matched, identities are revealed, and documents are shared.
The portal runs in any modern browser, and the same application is built for Android, iOS, Windows and macOS from a single codebase — so a coordinator on a ward sees the same registry as a consultant at a desk.
Yes. Pairs, swaps, consultants and coordinators export as CSV, Excel or PDF, filtered exactly as they are on screen. Match reports and chain reports print as PDF.
Please tell us — a wrong answer on this page is worth reporting. Contact the registry, and keep patient identifiers out of email.
Transplant consultants register here. Your centre is enrolled as part of approving you, and your transplant coordinators are added under you once you are in.