Argentina Builds Its Own Transplant Matching Calculator to Reflect Local Donors

For a sensitized kidney transplant candidate, one number can shape the entire waiting-list experience: the calculated panel reactive antibody, or cPRA. This statistic estimates the percentage of donors in a given pool who would be immunologically incompatible with a particular patient, based on the HLA antigens the patient’s immune system has learned to recognize and reject. Yet the number is only as meaningful as the population behind it. A tool calibrated on North American or European donors may quietly misrepresent the odds facing a patient in Buenos Aires, where the genetic makeup of the donor pool differs in ways that matter. A new web-based application called cPRArgentina, described in Frontiers in Genetics, tackles this problem by grounding cPRA estimation directly in Argentine deceased donor data, and by offering something most national calculators still treat as optional: an explicit accounting of ABO blood group compatibility.

The logic of cPRA is straightforward in principle. When a transplant candidate carries antibodies against specific HLA antigens, those antigens are registered as unacceptable, meaning that any donor expressing them would be excluded from consideration. The cPRA then asks a simple question of the reference donor population: what fraction of actual donors carries at least one of these unacceptable antigens? A patient whose antibodies are common in the population will have a high cPRA and a correspondingly slim chance of a compatible offer, while a patient with rare antibodies may have a low cPRA despite substantial sensitization. The catch is that the answer depends entirely on the HLA frequencies of the donor dataset used. Commercial antibody screening panels and international calculators were built around populations whose HLA distributions do not necessarily mirror those of Latin America, a region known for substantial genetic diversity shaped by admixture among Indigenous American, European, and African ancestries.

In Argentina, sensitization assessment has traditionally relied on commercial Luminex-based PRA assays using manufacturer-defined antigen panels. These assays offer standardization, but their panels may not capture the actual distribution of HLA antigens among Argentine donors, and they say nothing about ABO compatibility, the blood-group barrier that independently governs who can receive whose kidney. To close this gap, a team led by Manuel Quirno Costa of the Oxford Transplant Centre, working with colleagues at the Hospital General de Agudos Carlos G. Durand and the Centro Provincial de Histocompatibilidad of CUCAIBA in La Plata, built cPRArgentina as a web application that compares a patient’s unacceptable antigens directly against a curated dataset of local donors.

The dataset behind the calculator is substantial: HLA and ABO information from 4,159 deceased kidney donors in the Province of Buenos Aires, collected during routine clinical practice between 1998 and 2026. Serologic HLA-A, -B, and -DR assignments are available for every donor, while HLA-DQ typing, incorporated into routine practice from 2019 onward, exists for 1,156 more recently typed donors. This uneven coverage forced a design decision that reflects a real-world tension in immunogenetics. If a user enters a DQ-directed antibody but the denominator included all donors, many of whom lack DQ typing, the resulting cPRA would be artificially low, since donors with unknown DQ types would be wrongly counted as compatible. The application therefore adjusts its denominator: with no DQ antigens entered, all 4,159 donors count; once a DQ antigen appears, the calculation restricts itself to the 1,156 fully typed donors.

The calculation pipeline itself is carefully guarded against input errors. User-supplied unacceptable antigens undergo nomenclature normalization and validation before any computation. Only supported split antigens contribute to the result; broad antigens are recognized but excluded rather than being expanded into assumed split equivalents, and immunogenetically valid antigens outside the application’s current scope are reported to the user without entering the calculation. Invalid entries block the calculation entirely until corrected. Once validated, a donor is flagged as incompatible if at least one unacceptable antigen is present, and the cPRA is simply the proportion of incompatible donors, expressed as a percentage rounded to one decimal place. When the optional ABO mode is enabled, donors are considered incompatible if they fail either the HLA or the ABO criterion, so the resulting figure represents the proportion of the donor pool unavailable to that candidate for any immunological or blood-group reason.

To evaluate how the tool behaves, the team generated 100 synthetic recipient profiles spanning all ABO blood groups and a full range of HLA sensitization levels, calculating each profile’s cPRA with and without ABO compatibility. The results were striking in their pattern. Adding ABO compatibility raised cPRA in every blood group except AB, whose recipients can accept organs from any donor type and therefore saw no change. The increases were largest for blood group O recipients, at 16.9 percentage points on average, followed by group B at 15.1 points, with group A gaining only 4.6 points. The asymmetry follows directly from the arithmetic of the donor pool: in the Buenos Aires dataset, 57.8 percent of donors are group O, and O recipients can only receive O organs, so the ABO filter removes more than half the pool for them. Group B recipients, facing a compatible pool of just 65.3 percent, fare nearly as poorly, while A recipients retain access to 90.4 percent of donors.

The interaction with HLA sensitization proved equally informative. The mean increase from ABO adjustment was greatest among profiles with low HLA-only cPRA, at 17.2 percentage points, diminishing to 13.4 points for intermediate sensitization and 6.5 points for highly sensitized profiles. In other words, ABO compatibility matters most when HLA antibodies are not already eliminating most of the donor pool; once a patient’s antibody profile alone renders the vast majority of donors unacceptable, the additional exclusion imposed by blood group adds comparatively little. For patients and clinicians, this means that a low HLA-based cPRA can still conceal a sobering reality for a group O candidate, whose effective donor access is halved by blood type before any antibody consideration enters the picture.

How do the Argentine estimates compare with established international tools? The team ran the same 100 profiles through the Eurotransplant and OPTN calculators. HLA-only estimates correlated strongly across platforms, with Pearson correlations of 0.927 against Eurotransplant and 0.958 against OPTN, but mean absolute differences of 6.71 and 4.36 percentage points respectively revealed that the same antibody profile can yield meaningfully different numbers depending on the reference population. Notably, these differences shrank among highly sensitized profiles, where nearly all calculators converge toward high values, and widened among those with low and intermediate sensitization, precisely where population-specific HLA frequencies exert their greatest influence. A parallel pattern held when ABO compatibility was included, with correlations of 0.926 against Eurotransplant’s ABO-compatible donor frequency data and 0.957 against the ABO-adjusted cPRA approach published by Gragert and colleagues, and mean absolute differences of 8.98 and 5.47 percentage points.

The software engineering behind the project emphasizes transparency and reproducibility. cPRArgentina was built in Python using the FastAPI framework, with donor data stored in an SQLite database that supports real-time calculation from user-defined inputs. Automated tests written with pytest cover 75 percent of the core backend code and, crucially, 100 percent of the central cPRA calculation module. Version 1.0.0 is archived on Zenodo, the source code is publicly available on GitHub, and a reproducibility script encodes the entire simulated cohort, allowing anyone to regenerate the published results. Repeating the analysis with the archived version reproduced identical cPRA values for all 100 profiles. The application is freely accessible for research, educational, and non-commercial purposes.

The authors are candid about the limitations. The donor dataset covers the Province of Buenos Aires, which contributes roughly 31 percent of the country’s deceased donation processes, the largest share of any jurisdiction, but leaves much of the national donor pool unrepresented; expanding coverage could reveal whether regional adjustments, as used in the Canadian calculator, are needed or whether a single national pool suffices. The predominance of serologic typing limits resolution and excludes newer loci, though the architecture is designed to absorb additional regions, higher-resolution data, and updated allele-to-serotype assignments as they emerge. And because the evaluation used synthetic rather than real transplant candidates, clinical validation in larger cohorts remains a necessary next step. Even so, the work marks a meaningful shift: from imported estimates built on distant populations toward a locally grounded measure of donor availability, one that acknowledges that for a group O patient in Argentina, the waiting list looks different than the international defaults would suggest.

Subject of Research: A population-specific cPRA calculator for kidney transplant candidates in Argentina incorporating HLA and optional ABO compatibility

Article Title: cPRArgentina: a population-specific cPRA calculator incorporating optional ABO compatibility

Article References: Quirno Costa, M., Balekjian, M. B., Paulos, M. S., & Bianco, M. E. (2026). cPRArgentina: a population-specific cPRA calculator incorporating optional ABO compatibility. Frontiers in Genetics, 17, Article 1971911. https://doi.org/10.3389/fgene.2026.1971911

Image Credits: AI Generated

DOI: 10.3389/fgene.2026.1971911

Keywords: cPRA, HLA, ABO compatibility, kidney transplantation, Argentina, organ allocation, immunogenetics, sensitization, deceased donors, transplantation, population genetics, web application