Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Article
  • Published:

An automated histological classification system for precision diagnostics of kidney allografts

Abstract

For three decades, the international Banff classification has been the gold standard for kidney allograft rejection diagnosis, but this system has become complex over time with the integration of multimodal data and rules, leading to misclassifications that can have deleterious therapeutic consequences for patients. To improve diagnosis, we developed a decision-support system, based on an algorithm covering all classification rules and diagnostic scenarios, that automatically assigns kidney allograft diagnoses. We then tested its ability to reclassify rejection diagnoses for adult and pediatric kidney transplant recipients in three international multicentric cohorts and two large prospective clinical trials, including 4,409 biopsies from 3,054 patients (62.05% male and 37.95% female) followed in 20 transplant referral centers in Europe and North America. In the adult kidney transplant population, the Banff Automation System reclassified 83 out of 279 (29.75%) antibody-mediated rejection cases and 57 out of 105 (54.29%) T cell-mediated rejection cases, whereas 237 out of 3,239 (7.32%) biopsies diagnosed as non-rejection by pathologists were reclassified as rejection. In the pediatric population, the reclassification rates were 8 out of 26 (30.77%) for antibody-mediated rejection and 12 out of 39 (30.77%) for T cell-mediated rejection. Finally, we found that reclassification of the initial diagnoses by the Banff Automation System was associated with an improved risk stratification of long-term allograft outcomes. This study demonstrates the potential of an automated histological classification to improve transplant patient care by correcting diagnostic errors and standardizing allograft rejection diagnoses.

ClinicalTrials.gov registration: NCT05306795.

This is a preview of subscription content, access via your institution

Access options

Rent or buy this article

Prices vary by article type

from$1.95

to$39.95

Prices may be subject to local taxes which are calculated during checkout

Fig. 1: Development of the comprehensive Banff Automation System for kidney allograft precision diagnostics.
Fig. 2: Example of the output of the online Banff Automation System application.
Fig. 3: Reclassification of rejection-related diagnoses by the Banff Automation System.
Fig. 4: Risk stratification of allograft outcomes by the Banff Automation System.

Similar content being viewed by others

Data availability

All minimum data to reproduce the Banff Automation System and the figures are deposited into the synapse public repository32 and are freely available. A sign-in process is required to access the data. Full source data are available from the corresponding author.

Code availability

Complete code to reproduce the Banff Automation System and the figures are freely available in the synapse public repository32. A sign-in process is required to access the code.

References

  1. Solez, K. et al. International standardization of criteria for the histologic diagnosis of renal allograft rejection: the Banff working classification of kidney transplant pathology. Kidney Int. 44, 411–422 (1993).

    Article  CAS  PubMed  Google Scholar 

  2. Racusen, L. C. et al. Antibody-mediated rejection criteria – an addition to the Banff ′97 classification of renal allograft rejection. Am. J. Transplant. 3, 708–714 (2003).

    Article  PubMed  Google Scholar 

  3. Racusen, L. C., Halloran, P. F. & Solez, K. Banff 2003 meeting report: new diagnostic insights and standards. Am. J. Transplant. 4, 1562–1566 (2004).

    Article  PubMed  Google Scholar 

  4. Mengel, M. et al. Banff 2011 meeting report: new concepts in antibody-mediated rejection. Am. J. Transplant. 12, 563–570 (2012).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  5. Mengel, M. et al. Banff 2019 meeting report: molecular diagnostics in solid organ transplantation–consensus for the Banff Human Organ Transplant (B-HOT) gene panel and open source multicenter validation. Am. J. Transplant. 20, 2305–2317 (2020).

    Article  PubMed  PubMed Central  Google Scholar 

  6. Farris, A. B. et al. Banff digital pathology working group: going digital in transplant pathology. Am. J. Transplant. 20, 2392–2399 (2020).

    Article  PubMed  PubMed Central  Google Scholar 

  7. El-Zoghby, Z. M. et al. Identifying specific causes of kidney allograft loss. Am. J. Transplant. 9, 527–535 (2009).

    Article  CAS  PubMed  Google Scholar 

  8. Solez, K. et al. Clinical validation and reproducibility of the Banff schema for renal allograft pathology. Transplant. Proc. 27, 1009–1011 (1995).

    CAS  PubMed  Google Scholar 

  9. Schinstock, C. A. et al. Banff survey on antibody-mediated rejection clinical practices in kidney transplantation: diagnostic misinterpretation has potential therapeutic implications. Am. J. Transplant. 19, 123–131 (2019).

    Article  PubMed  Google Scholar 

  10. Loupy, A., Mengel, M. & Haas, M. 30 years of the International Banff Classification for Allograft Pathology: the past, present and future of kidney transplant diagnostics. Kidney Int. https://doi.org/10.1016/j.kint.2021.11.028 (2022).

  11. Furness, P. N. & Taub, N. Convergence of European Renal Transplant Pathology Assessment Procedures (CERTPAP) Project. International variation in the interpretation of renal transplant biopsies: report of the CERTPAP Project. Kidney Int. 60, 1998–2012 (2001).

    Article  CAS  PubMed  Google Scholar 

  12. Mengel, M., Sis, B. & Halloran, P. F. SWOT analysis of Banff: strengths, weaknesses, opportunities and threats of the international Banff consensus process and classification system for renal allograft pathology. Am. J. Transplant. 7, 2221–2226 (2007).

    Article  CAS  PubMed  Google Scholar 

  13. Schold, J. D. & Kaplan, B. The elephant in the room: failings of current clinical endpoints in kidney transplantation. Am. J. Transplant. 10, 1163–1166 (2010).

    Article  CAS  PubMed  Google Scholar 

  14. US FDA. Surrogate endpoints for clinical trials in kidney transplantation; public workshop. Federal Register https://www.federalregister.gov/documents/2015/08/03/2015-18957/surrogate-endpoints-for-clinical-trials-in-kidney-transplantation-public-workshop (2015).

  15. Loupy, A. et al. The Banff 2019 Kidney Meeting Report (I): updates on and clarification of criteria for T cell- and antibody-mediated rejection. Am. J. Transplant. 20, 2318–2331 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  16. Loupy, A. & Lefaucheur, C. Antibody-mediated rejection of solid-organ allografts. N. Engl. J. Med. 379, 1150–1160 (2018).

    Article  CAS  PubMed  Google Scholar 

  17. Bouatou, Y. et al. Response to treatment and long-term outcomes in kidney transplant recipients with acute T cell-mediated rejection. Am. J. Transplant. 19, 1972–1988 (2019).

    Article  CAS  PubMed  Google Scholar 

  18. Maggiore, U., Leventhal, J. & Cravedi, P. Rethinking clinical endpoints in kidney transplant trials. Curr. Opin. Organ Transplant. 25, 1–7 (2020).

    Article  PubMed  Google Scholar 

  19. Becker, J. U., Seron, D., Rabant, M., Roufosse, C. & Naesens, M. Evolution of the definition of rejection in kidney transplantation and its use as an endpoint in clinical trials. Transpl. Int. https://doi.org/10.3389/ti.2022.10141 (2022).

  20. Cook, T. S. Human versus machine in medicine: can scientific literature answer the question? Lancet Digit. Health 1, e246–e247 (2019).

    Article  PubMed  Google Scholar 

  21. Ekberg, H. et al. Reduced exposure to calcineurin inhibitors in renal transplantation. N. Engl. J. Med. 357, 2562–2575 (2007).

    Article  CAS  PubMed  Google Scholar 

  22. Cooper, J. E. Evaluation and treatment of acute rejection in kidney allografts. Clin. J. Am. Soc. Nephrol. 15, 430–438 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  23. Solez, K. et al. Report of the third banff conference on allograft pathology (July 20–24, 1995) on classification and lesion scoring in renal allograft pathology. Transplant. Proc. 28, 441–444 (1996).

    CAS  PubMed  Google Scholar 

  24. Racusen, L. C. et al. The Banff 97 working classification of renal allograft pathology. Kidney Int. 55, 713–723 (1999).

    Article  CAS  PubMed  Google Scholar 

  25. Solez, K. et al. Banff ’05 meeting report: differential diagnosis of chronic allograft injury and elimination of Chronic Allograft Nephropathy (‘CAN’). Am. J. Transplant. 7, 518–526 (2007).

    Article  CAS  PubMed  Google Scholar 

  26. Solez, K. et al. Banff 07 classification of renal allograft pathology: updates and future directions. Am. J. Transplant. 8, 753–760 (2008).

    Article  CAS  PubMed  Google Scholar 

  27. Sis, B. et al. Banff ’09 meeting report: antibody mediated graft deterioration and implementation of Banff working groups. Am. J. Transplant. 10, 464–471 (2010).

    Article  CAS  PubMed  Google Scholar 

  28. Haas, M. et al. Banff 2013 meeting report: inclusion of C4d-negative antibody-mediated rejection and antibody-associated arterial lesions. Am. J. Transplant. 14, 272–283 (2014).

    Article  CAS  PubMed  Google Scholar 

  29. Loupy, A. et al. The Banff 2015 kidney meeting report: current challenges in rejection classification and prospects for adopting molecular pathology. Am. J. Transplant. 17, 28–41 (2017).

    Article  CAS  PubMed  Google Scholar 

  30. Haas, M. et al. The Banff 2017 Kidney Meeting Report: revised diagnostic criteria for chronic active T cell-mediated rejection, antibody-mediated rejection, and prospects for integrative endpoints for next-generation clinical trials. Am. J. Transplant. 18, 293–307 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  31. von Elm, E. et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann. Intern. Med. 147, 573–577 (2007).

    Article  Google Scholar 

  32. Loupy, A. et al. Development, application, and validation of a histological classification automation system for kidney allograft precision diagnostics. Synapse https://doi.org/10.7303/syn33342930 (2022).

Download references

Acknowledgements

We thank I. Louis Boudhabhay, G. Morin and C. Ursule-Dufait for testing the Banff Automation System at Necker Hospital, O. Nekachtali and M. Kamel for data acquisition, B. Robin for his help to design Fig. 1 and M. Mengel, C. Roufosse and J. Ulrich Becker for their critical review of the Banff classification criteria. INSERM-Action thématique incitative sur programme Avenir provided financial support. V.G. received a grant from the French Foundation for Medical Research and the French-Speaking Society of Transplantation. G.D. received a grant from the French Foundation for Medical Research. O.A. received a grant from the Fondation Bettencourt Schueller. The KTD-Innov study was funded by the French government, with financial support managed by the National Research Agency under the program ‘Investissements d’avenir’ under grant agreement no. ANR-17-RHUS-0010. The EU-TRAIN study was funded by the European Union’s Horizon 2020 research and innovation program under grant agreement no. 754995. Academic grant support was provided by the non-profit organizations MSD Avenir and OrganX.

Author information

Authors and Affiliations

Authors

Contributions

D.Y., V.G., G.D., J.G. and A.L. designed the study. D.Y., V.G., G.D., J.G., M. Rabant and A.L. developed the online application. D.Y., V.G., G.D., J.G., B.A., O.A., M. Raynaud, Z.D., J.H., P.W., J.S., R.G., B.W., R.Z., M.S., K.T., L.C., T.B., O. Boyer, J.P.D.V.H., M.G., A.A., P.A.G., E.M., M.L.Q., S.B., C. Legendre, D.A., J.V., W.Z., N.K., O. Bestard, A.D., K.B., M.H., C. Lefaucheur, M. Rabant and A.L. contributed to data acquisition. D.Y., V.G., G.D., M. Raynaud, J.G. and A.L. performed data analysis. D.Y., V.G., G.D., M. Raynaud, M.H., M. Rabant and A.L. performed data interpretation. D.Y., V.G., G.D., J.G., M. Raynaud, M. Rabant and A.L. wrote the manuscript. All authors revised and critically reviewed the manuscript.

Corresponding author

Correspondence to Alexandre Loupy.

Ethics declarations

Competing interests

A.L. holds shares in Cibiltech, a software company that is not involved in the present research. The other authors declare no competing interests.

Peer review

Peer review information

Nature Medicine thanks Nada Alachkar, Leonardo Riella and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editor: Michael Basson, in collaboration with the Nature Medicine team.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Banff Automation System’s reclassification of rejection-related diagnoses in observational studies of adult kidney transplant recipients, after excluding the discrepant diagnoses explained by classification changes over time.

Sankey diagrams depicting the flow of the reassessments of rejection-related diagnoses by the Banff Automation System in the observational studies of adult kidney transplant recipients, after excluding the discrepant diagnoses explained by classification changes over time. The proportions of the diagnoses by pathologists and the Banff Automation System are shown on the left and right side of each diagram, respectively. The category ‘equivocal AMR’ refers to biopsies with lesions evocative of AMR but not fulfilling all the criteria for a diagnosis of AMR (moderate microvascular inflammation (g + ptc ≥ 2 in the absence of recurrent or de novo glomerulonephritis and at least g ≥ 1 in the presence of acute TCMR, borderline infiltrate or infection) without circulating DSA or a substitute (complement C4d staining or expression of validated molecular markers) and C4d+ acute tubular injury without circulating DSA). The category ‘other diagnoses’ consists of interstitial fibrosis and tubular atrophy, acute tubular injury, thrombotic microangiopathy, calcineurin inhibitor toxicity and biopsies with no specific diagnosis. Diagnoses polyomavirus nephropathies, recurrent disease, de novo glomerulonephritis, post-transplant lymphoproliferative disorder, pyelonephritis and drug-induced interstitial nephritis were excluded because their assignment relied on the user input in the application. Abbreviations: AMR, antibody-mediated rejection; TCMR, T-cell mediated rejection; g, glomerulitis; ptc, peritubular capillaritis; DSA, donor-specific antibody.

Extended Data Fig. 2 Probability of death-censored allograft survival according to antibody-mediated rejection reclassifications by the Banff Automation System.

These Kaplan–Meier curves show the probability of death-censored allograft survival according to antibody-mediated rejection reclassifications by the Banff Automation System in patients from observational cohorts. Log-rank test was used to compare survival curves and two-sided p values were calculated. The overall exact p value indicated by ‘Log-rank p value< 0.0001’ is ‘4.071e-17’. Abbreviations: HR, hazard ratio; CI, confidence interval.

Extended Data Fig. 3 Probability of death-censored allograft survival according to T cell-mediated rejection reclassifications by the Banff Automation System.

These Kaplan–Meier curves show the probability of death-censored allograft survival according to T cell-mediated reclassifications by the Banff Automation System in patients from observational cohorts. Log-rank test was used to compare survival curves and two-sided p values were calculated. The overall exact p value indicated by ‘Log-rank p value< 0.0001’ is ‘3.186e-10’. Abbreviations: HR, hazard ratio; CI, confidence interval.

Extended Data Fig. 4 Probability of death-censored allograft survival according to Banff Automation System’s diagnostic reclassifications, after excluding the discrepant diagnoses explained by classification changes over time.

a: Risk stratification according to rejection reclassifications b: Risk stratification according to antibody-mediated rejection reclassifications. c: Risk stratification according to T cell-mediated rejection reclassifications. Log-rank test was used to compare survival curves, and two-sided p values were calculated. The overall exact p values indicated by ‘Log-rank p value< 0.0001’ for extended data Fig. 4a–c are ‘6.836e-16’, ‘2.223e-17’, and ‘1.822e-10’, respectively. Abbreviations: HR, hazard ratio; CI, confidence interval.

Extended Data Table 1 Banff Automation System suggestions and notes for equivocal diagnoses and missing data in real-life situations
Extended Data Table 2 Characteristics of the patients and biopsies performed from the overall cohort of adult kidney transplant recipients
Extended Data Table 3 Explanation of the 393 discrepancies between pathologists’ diagnoses and application’s assignments in the Paris Transplant Group and KTD-Innov’s databases
Extended Data Table 4 Detailed comparison between pathologists and Banff Automation System diagnoses in the biopsies from the observational studies of adult kidney transplant recipients, after excluding the discrepant diagnoses explained by classification changes over time
Extended Data Table 5 Detailed characteristics of the patients and biopsies from the multicentric cohort of pediatric kidney transplant recipients included in the analyses
Extended Data Table 6 Detailed comparison between pathologists and Banff Automation System’s diagnoses in the 514 biopsies from the multicentric cohort of pediatric kidney transplant recipients

Supplementary information

Supplementary Information

Supplementary Figs. 1–4 and Supplementary Tables 1–6.

Reporting Summary

Supplementary Video 1

Introduction of the online application.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Yoo, D., Goutaudier, V., Divard, G. et al. An automated histological classification system for precision diagnostics of kidney allografts. Nat Med 29, 1211–1220 (2023). https://doi.org/10.1038/s41591-023-02323-6

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s41591-023-02323-6

This article is cited by

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing