Machine learning does not outperform traditional statistical modelling for kidney allograft failure prediction.

Machine learning (ML) models have recently shown potential for predicting kidney allograft outcomes. However, their ability to outperform traditional approaches remains poorly investigated. Therefore, using large cohorts of kidney transplant recipients from 14 centers worldwide, we developed ML-based prediction models for kidney allograft survival and compared their prediction performances to those achieved by a validated Cox-Based Prognostication System (CBPS). In a French derivation cohort of 4000 patients, candidate determinants of allograft failure including donor, recipient and transplant-related parameters were used as predictors to develop tree-based models (RSF, RSF-ERT, CIF), Support Vector Machine models (LK-SVM, AK-SVM) and a gradient boosting model (XGBoost). Models were externally validated with cohorts of 2214 patients from Europe, 1537 from North America, and 671 from South America. Among these 8422 kidney transplant recipients, 1081 (12.84%) lost their grafts after a median post-transplant follow-up time of 6.25 years (Inter Quartile Range 4.33-8.73). At seven years post-risk evaluation, the ML models achieved a C-index of 0.788 (95% bootstrap percentile confidence interval 0.736-0.833), 0.779 (0.724-0.825), 0.786 (0.735-0.832), 0.527 (0.456-0.602), 0.704 (0.648-0.759) and 0.767 (0.711-0.815) for RSF, RSF-ERT, CIF, LK-SVM, AK-SVM and XGBoost respectively, compared with 0.808 (0.792-0.829) for the CBPS. In validation cohorts, ML models' discrimination performances were in a similar range of those of the CBPS. Calibrations of the ML models were similar or less accurate than those of the CBPS. Thus, when using a transparent methodological pipeline in validated international cohorts, ML models, despite overall good performances, do not outperform a traditional CBPS in predicting kidney allograft failure. Hence, our current study supports the continued use of traditional statistical approaches for kidney graft prognostication.

Long-term Efficacy and Safety of Everolimus Versus Mycophenolate in Kidney Transplant Recipients Receiving Tacrolimus.

BACKGROUND: The short-term efficacy and safety of everolimus in combination with tacrolimus have been described in several clinical trials. Yet, detailed long-term data comparing the use of everolimus or mycophenolate in kidney transplant recipients receiving tacrolimus are lacking. METHODS: This is a 5-y follow-up post hoc analysis of a prospective trial including 288 patients who were randomized to receive a single 3-mg/kg dose of rabbit antithymocyte globulin, tacrolimus, everolimus (EVR), and prednisone (rabbit antithymocyte globulin/EVR, n = 85); basiliximab, tacrolimus, everolimus, and prednisone (basiliximab/EVR, n = 102); or basiliximab, tacrolimus, mycophenolate, and prednisone (basiliximab/mycophenolate, n = 101). RESULTS: There were no differences in the incidence of treatment failure (31.8% versus 40.2% versus 34.7%, P = 0.468), de novo donor-specific HLA antibodies (6.5% versus 11.7% versus 4.0%, P = 0.185), patient (92.9% versus 94.1% versus 92.1%, P = 0.854), and death-censored graft (87.1% versus 90.2% versus 85.1%, P = 0.498) survivals. Using a sensitive analysis, the trajectories of estimated glomerular filtration rate were comparable in the intention-to-treat (P = 0.145) and per protocol (P = 0.354) populations. There were no differences in study drug discontinuation rate (22.4% versus 30.4% versus 17.8%, P = 0.103). CONCLUSIONS: In summary, this analysis in a cohort of de novo low/moderate immunologic risk kidney transplant recipients suggests that the use of a single 3 mg/kg rabbit antithymocyte globulin dose followed by EVR combined with reduced tacrolimus concentrations was associated with similar efficacy and renal function compared with the standard of care immunosuppressive regimen.

Dynamic prediction of renal survival among deeply phenotyped kidney transplant recipients using artificial intelligence: an observational, international, multicohort study.

BACKGROUND: Kidney allograft failure is a common cause of end-stage renal disease. We aimed to develop a dynamic artificial intelligence approach to enhance risk stratification for kidney transplant recipients by generating continuously refined predictions of survival using updates of clinical data. METHODS: In this observational study, we used data from adult recipients of kidney transplants from 18 academic transplant centres in Europe, the USA, and South America, and a cohort of patients from six randomised controlled trials. The development cohort comprised patients from four centres in France, with all other patients included in external validation cohorts. To build deeply phenotyped cohorts of transplant recipients, the following data were collected in the development cohort: clinical, histological, immunological variables, and repeated measurements of estimated glomerular filtration rate (eGFR) and proteinuria (measured using the proteinuria to creatininuria ratio). To develop a dynamic prediction system based on these clinical assessments and repeated measurements, we used a Bayesian joint models-an artificial intelligence approach. The prediction performances of the model were assessed via discrimination, through calculation of the area under the receiver operator curve (AUC), and calibration. This study is registered with ClinicalTrials.gov, NCT04258891. FINDINGS: 13 608 patients were included (3774 in the development cohort and 9834 in the external validation cohorts) and contributed 89 328 patient-years of data, and 416 510 eGFR and proteinuria measurements. Bayesian joint models showed that recipient immunological profile, allograft interstitial fibrosis and tubular atrophy, allograft inflammation, and repeated measurements of eGFR and proteinuria were independent risk factors for allograft survival. The final model showed accurate calibration and very high discrimination in the development cohort (overall dynamic AUC 0·857 [95% CI 0·847-0·866]) with a persistent improvement in AUCs for each new repeated measurement (from 0·780 [0·768-0·794] to 0·926 [0·917-0·932]; p<0·0001). The predictive performance was confirmed in the external validation cohorts from Europe (overall AUC 0·845 [0·837-0·854]), the USA (overall AUC 0·820 [0·808-0·831]), South America (overall AUC 0·868 [0·856-0·880]), and the cohort of patients from randomised controlled trials (overall AUC 0·857 [0·840-0·875]). INTERPRETATION: Because of its dynamic design, this model can be continuously updated and holds value as a bedside tool that could refine the prognostic judgements of clinicians in everyday practice, hence enhancing precision medicine in the transplant setting. FUNDING: MSD Avenir, French National Institute for Health and Medical Research, and Bettencourt Schueller Foundation.

The influence of the antithymocyte globulin dose on clinical outcomes of patients undergoing kidney retransplantation.

Optimizing antithymocyte globulin (rATG) dosage is critical for high immunological risk patients undergoing a repeat kidney transplant. This natural retrospective cohort study compared clinical outcomes of two successive cohorts of consecutive recipients of retransplants receiving 5 x 1 mg/kg (rATG-5, n = 100) or a single 3 mg/kg (rATG-3, n = 110) dose of rATG induction therapy. All patients had negative complement-dependent cytotoxicity crossmatch and no anti-HLA A, B, DR donor-specific antibodies (DSA). The primary endpoint was efficacy failure (first biopsy-proven acute rejection, graft loss, or death) at 12 months. There was no difference in the cumulative incidence of efficacy failure (18.0% vs. 21.8%, HR = 1.22, 95% CI 0.66-2.25), respectively. There were no differences in 3-years freedom from biopsy proven acute rejection, and patient, graft, and death-censored graft survivals. There were no differences in the incidence of surgical complications (25.0% vs. 18.2%; p 0.151), early hospital readmission (27.8% vs. 29.5%; p = 0.877) and CMV infections (49% vs. 40%; p = 0.190). There were also no differences in the incidence (59.6% vs. 58.7%, p = 0.897) and duration of delayed graft function but a stable difference in estimate glomerular filtration rate was observed from month 1 (54.7±28.8 vs. 44.1±25.3 ml/min/1.73 m2, p = 0.005) to month 36 (51.1±27.7 vs. 42.5±24.5, p = 0.019). Mean urinary protein concentration (month 36: 0.38±0.81 vs. 0.70±2.40 g/ml, p = 0.008) and mean chronic glomerular Banff score in for cause biopsies (months 4-36: 0.0±0.0 vs. 0.04±0.26, p = 0.044) were higher in the rATG-3 group. This cohort analysis did not detect differences in the incidence of efficacy failure and in safety outcomes at 12 months among recipients of kidney retransplants without A, B, and DR DSA, receiving induction therapy with a single 3 mg/kg rATG dose or the traditional 5 mg/kg rATG.