Predicting Individual Renal Allograft Outcomes Using Risk Models with 1-Year Surveillance Biopsy and Alloantibody Data

Challenges in the management of the kidney allograft: from decline to failure: conclusions from a Kidney Disease: Improving Global Outcomes (KDIGO) Controversies Conference

In March 2022, Kidney Disease: Improving Global Outcomes (KDIGO) held a virtual Controversies Conference to address the important but rarely examined phase during which the kidney transplant is failing or has failed. In addition to discussing the definition of a failing allograft, 4 broad areas were considered in the context of a declining functioning graft: prognosis and kidney failure trajectory; immunosuppression strategies; management of medical and psychological complications, and patient factors; and choice of kidney replacement therapy or supportive care following graft loss. Identifying and paying special attention to individuals with failing allografts was felt to be important in order to prepare patients psychologically, manage immunosuppression, address complications, prepare for dialysis and/or retransplantation, and transition to supportive care. Accurate prognostication tools, although not yet widely available, were embraced as necessary to define allograft survival trajectories and the likelihood of allograft failure. The decision of whether to withdraw or continue immunosuppression after allograft failure was deemed to be based most appropriately on risk-benefit analysis and likelihood of retransplantation within a few months. Psychological preparation and support was identified as a critical factor in patient adjustment to graft failure, as was early communication. Several models of care were noted that enabled a medically supportive transition back to dialysis or retransplantation. Emphasis was placed on the importance of dialysis-access readiness before initiation of dialysis, in order to avoid use of central venous catheters. The centrality of the patient to all management decisions and discussions was deemed to be paramount. Patient "activation," which can be defined as engaged agency, was seen as the most effective way to achieve success. Unresolved controversies, gaps in knowledge, and areas for research were also stressed in the conference deliberations.

Plasma protein signatures reflect systemic immunity and allograft function in kidney transplantation

Kidney transplantation causes large perturbations of the immune system. While many studies focus on the allograft, insights into systemic effects are largely missing. Here, we analyzed the systemic immune response in 3 cohorts of kidney transplanted patients. Using serum proteomics, laboratory values, mass cytometry, histological and clinical parameters, inter-patient heterogeneity was leveraged for multi-omic co-variation analysis. We identified circulating immune modules (CIM) that describe extra-renal signatures of co-regulated plasma proteins. CIM are present in nontransplanted controls, in transplant conditions and during rejection. They are enriched in pathways linked to kidney function, extracellular matrix, signaling, and cellular activation. A complex leukocyte response in the blood during allograft quiescence and rejection is associated with CIM activity and CIM-specific cytokines. CIM activity correlates with kidney function including a 2-month prediction. Together, the data suggest a systemic and multi-layered response of transplant immunity that might be insightful for understanding allograft dysfunction and developing translational biomarkers.

Longitudinal estimated glomerular filtration rate (eGFR) modeling in long-term renal function to inform clinical trial design in kidney transplantation

Kidney transplantation is the preferred treatment for individuals with end-stage kidney disease. From a modeling perspective, our understanding of kidney function trajectories after transplantation remains limited. Current modeling of kidney function post-transplantation is focused on linear slopes or percent decline and often excludes the highly variable early timepoints post-transplantation, where kidney function recovers and then stabilizes. Using estimated glomerular filtration rate (eGFR), a well-known biomarker of kidney function, from an aggregated dataset of 4904 kidney transplant patients including both observational studies and clinical trials, we developed a longitudinal model of kidney function trajectories from time of transplant to 6 years post-transplant. Our model is a nonlinear, mixed-effects model built in NONMEM that captured both the recovery phase after kidney transplantation, where the graft recovers function, and the long-term phase of stabilization and slow decline. Model fit was assessed using diagnostic plots and individual fits. Model performance, assessed via visual predictive checks, suggests accurate model predictions of eGFR at the median and lower 95% quantiles of eGFR, ranges which are of critical clinical importance for assessing loss of kidney function. Various clinically relevant covariates were also explored and found to improve the model. For example, transplant recipients of deceased donors recover function more slowly after transplantation and calcineurin inhibitor use promotes faster long-term decay. Our work provides a generalizable, nonlinear model of kidney allograft function that will be useful for estimating eGFR up to 6 years post-transplant in various clinically relevant populations.

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.

Seeking Standardized Definitions for HLA-incompatible Kidney Transplants: A Systematic Review

BACKGROUND

There is no standard definition for "HLA incompatible" transplants. For the first time, we systematically assessed how HLA incompatibility was defined in contemporary peer-reviewed publications and its prognostic implication to transplant outcomes.

METHODS

We combined 2 independent searches of MEDLINE, EMBASE, and the Cochrane Library from 2015 to 2019. Content-expert reviewers screened for original research on outcomes of HLA-incompatible transplants (defined as allele or molecular mismatch and solid-phase or cell-based assays). We ascertained the completeness of reporting on a predefined set of variables assessing HLA incompatibility, therapies, and outcomes. Given significant heterogeneity, we conducted narrative synthesis and assessed risk of bias in studies examining the association between death-censored graft failure and HLA incompatibility.

RESULTS

Of 6656 screened articles, 163 evaluated transplant outcomes by HLA incompatibility. Most articles reported on cytotoxic/flow T-cell crossmatches (n = 98). Molecular genotypes were reported for selected loci at the allele-group level. Sixteen articles reported on epitope compatibility. Pretransplant donor-specific HLA antibodies were often considered (n = 143); yet there was heterogeneity in sample handling, assay procedure, and incomplete reporting on donor-specific HLA antibodies assignment. Induction (n = 129) and maintenance immunosuppression (n = 140) were frequently mentioned but less so rejection treatment (n = 72) and desensitization (n = 70). Studies assessing death-censored graft failure risk by HLA incompatibility were vulnerable to bias in the participant, predictor, and analysis domains.

CONCLUSIONS

Optimization of transplant outcomes and personalized care depends on accurate HLA compatibility assessment. Reporting on a standard set of variables will help assess generalizability of research, allow knowledge synthesis, and facilitate international collaboration in clinical trials.

Comparison of artificial intelligence and human-based prediction and stratification of the risk of long-term kidney allograft failure

BACKGROUND

Clinical decisions are mainly driven by the ability of physicians to apply risk stratification to patients. However, this task is difficult as it requires complex integration of numerous parameters and is impacted by patient heterogeneity. We sought to evaluate the ability of transplant physicians to predict the risk of long-term allograft failure and compare them to a validated artificial intelligence (AI) prediction algorithm.

METHODS

We randomly selected 400 kidney transplant recipients from a qualified dataset of 4000 patients. For each patient, 44 features routinely collected during the first-year post-transplant were compiled in an electronic health record (EHR). We enrolled 9 transplant physicians at various career stages. At 1-year post-transplant, they blindly predicted the long-term graft survival with probabilities for each patient. Their predictions were compared with those of a validated prediction system (iBox). We assessed the determinants of each physician's prediction using a random forest survival model.

RESULTS

Among the 400 patients included, 84 graft failures occurred at 7 years post-evaluation. The iBox system demonstrates the best predictive performance with a discrimination of 0.79 and a median calibration error of 5.79%, while physicians tend to overestimate the risk of graft failure. Physicians' risk predictions show wide heterogeneity with a moderate intraclass correlation of 0.58. The determinants of physicians' prediction are disparate, with poor agreement regardless of their clinical experience.

CONCLUSIONS

This study shows the overall limited performance and consistency of physicians to predict the risk of long-term graft failure, demonstrated by the superior performances of the iBox. This study supports the use of a companion tool to help physicians in their prognostic judgement and decision-making in clinical care.

Impact of Glomerulitis on Long-term Outcomes After Kidney Transplantation

The Banff classification scheme provides a framework for interpreting transplant kidney biopsies and has undergone various updates in the past 2 decades especially related to antibody-mediated rejection. The clinical significance of early glomerulitis seen within 4 mo on protocol biopsies has received limited attention. We hypothesized that early glomerulitis seen on protocol biopsies will lead to significant adverse outcomes as assessed by histopathology and allograft outcome.

Dominant predictors of early post-transplant outcomes based on the Korean Organ Transplantation Registry (KOTRY)

Data for Asian kidney transplants are very limited. We investigated the relative importance of prognostic markers in Asian kidney transplants by using Korean Organ Transplantation Registry (KOTRY) cohort. Prediction models were developed by data-driven variable selection approach. The relative importance of the selected predictors was measured by dominance analysis. A total of 4854 kidney transplant donor-recipient pairs were analyzed. Overall patient survival rates were 99.8%, 98.8%, and 91.8% at 1, 3, and 5 years, respectively. Death-censored graft survival rates were 98.4%, 97.0%, and 95.8% at 1, 3, and 5 years. Biopsy-proven acute rejection free survival rates were 90.1%, 87.4%, and 87.03% at 1, 3, and 5 years. The top 3 dominant predictors for recipient mortality within 1 year were recipient cardiovascular disease history, deceased donor, and recipient age. The dominant predictors for death-censored graft loss within 1 year were acute rejection, deceased donor, and desensitization. The dominant predictors to acute rejection within 1 year were donor age, HLA mismatched numbers, and desensitization. We presented clinical characteristics of patients enrolled in KOTRY during the last 5 years and investigated dominant predictors for early post-transplant outcomes, which would be useful for clinical decision-making based on quantitative measures.

Surrogate Endpoints for Late Kidney Transplantation Failure

In kidney transplant recipients, late graft failure is often multifactorial. In addition, primary endpoints in kidney transplantation studies seek to demonstrate the short-term efficacy and safety of clinical interventions. Although such endpoints might demonstrate short-term improvement in specific aspects of graft function or incidence of rejection, such findings do not automatically translate into meaningful long-term graft survival benefits. Combining many factors into a well-validated model is therefore more likely to predict long-term outcome and better reflect the complexity of late graft failure than using single endpoints. If conditional marketing authorization could be considered for therapies that aim to improve long-term outcomes following kidney transplantation, then the surrogate endpoint for graft failure in clinical trial settings needs clearer definition. This Consensus Report considers the potential benefits and drawbacks of several candidate surrogate endpoints (including estimated glomerular filtration rate, proteinuria, histological lesions, and donor-specific anti-human leukocyte antigen antibodies) and composite scoring systems. The content was created from information prepared by a working group within the European Society for Organ Transplantation (ESOT). The group submitted a Broad Scientific Advice request to the European Medicines Agency (EMA), June 2020: the request focused on clinical trial design and endpoints in kidney transplantation. Following discussion and refinement, the EMA made final recommendations to ESOT in December 2020 regarding the potential to use surrogate endpoints in clinical studies that aim to improving late graft failure.

Progressive decline of function in renal allografts with normal 1-year biopsies: Gene expression studies fail to identify a classifier

Histologic findings on 1-year biopsies such as inflammation with fibrosis and transplant glomerulopathy predict renal allograft loss by 5 years. However, almost half of the patients with graft loss have a 1-year biopsy that is either normal or has only interstitial fibrosis. The goal of this study was to determine if there was a gene expression profile in these relatively normal 1-year biopsies that predicted subsequent decline in renal function. Using transcriptome microarrays we measured intragraft mRNA levels in a retrospective Discovery cohort (170 patients with a normal/minimal fibrosis 1-year biopsy, 54 with progressive decline in function/graft loss and 116 with stable function) and developed a nested 10-fold cross-validated gene classifier that predicted progressive decline in renal function (positive predictive value = 38 ± 34%%; negative predictive value = 73 ± 30%, c-statistic = .59). In a prospective, multicenter Validation cohort (270 patients with Normal/Interstitial Fibrosis [IF]), the classifier had a 20% positive predictive value, 85% negative predictive value and .58 c-statistic. Importantly, the majority of patients with graft loss in the prospective study had 1-year biopsies scored as Normal or IF. We conclude predicting graft loss in many renal allograft recipients (i.e., those with a relatively normal 1-year biopsy and eGFR > 40) remains difficult.

Application of the iBox prognostication system as a surrogate endpoint in the TRANSFORM randomised controlled trial: proof-of-concept study

OBJECTIVES

Development of pharmaceutical agents in transplantation is currently limited by long waits for hard endpoints. We applied a validated integrative risk-prognostication system integrative Box (iBox) as a surrogate endpoint to the TRANSFORM Study, a large randomised controlled trial, to project individual patient long-term kidney allograft survival from 1 year to 11 years after randomisation.

DESIGN

Post-hoc analysis of a randomised open-label controlled trial.

SETTING

Multicentre study including 186 centres in 42 countries worldwide.

PARTICIPANTS

2037 de novo kidney transplant recipients.

INTERVENTION

Participants were randomised (1:1) to receive everolimus with reduced-exposure calcineurin inhibitor (EVR+rCNI) or mycophenolic acid with standard-exposure CNI (MPA+sCNI).

PRIMARY OUTCOME MEASURE

The iBox scores were computed for each participant at 1 year after randomisation using functional, immunological and histological parameters. Individual long-term death-censored allograft survival over 4, 6 and 11 years after randomisation was projected with the iBox risk-prognostication system.

RESULTS

Overall, 940 patients receiving EVR+rCNI and 932 receiving MPA+sCNI completed the 1-year visit. iBox scores generated at 1 year yielded graft survival prediction rates of 90.9% vs 92.1%, 87.9% vs 89.5%, and 80.0% vs 82.4% in the EVR+rCNI versus MPA+sCNI arms at 4, 6, and 11 years post-randomisation, respectively (all differences below the 10% non-inferiority margin defined by study protocol). Inclusion of immunological and histological Banff diagnoses parameters in iBox scores resulted in comparable and non-inferior predicted graft survival for both treatments.

CONCLUSIONS

This proof-of-concept study provides the first application of a validated prognostication system as a surrogate endpoint in the field of transplantation. The iBox system, by projecting kidney allograft survival up to 11 years post-randomisation, confirms the non-inferiority of EVR+rCNI versus MPA+sCNI regimen. Given the current process engaged for surrogate endpoints qualification, this study illustrates the potential to fast track development of pharmaceutical agents.

TRIAL REGISTRATION NUMBER

TRANSFORM trial: NCT01950819.iBox prognostication system: NCT03474003.

Predicting acute rejection in children, adolescents, and young adults with a kidney transplant by assessing barriers to taking medication

BACKGROUND

Nonadherence to immunosuppression is common among pediatric, adolescent, and young adult kidney transplant recipients and a leading cause of graft loss. Assessing barriers to medication adherence in clinical practice may identify patients at risk for rejection and provide therapeutic targets.

METHODS

Kidney transplant patients and/or their caregivers were assessed for 14 barriers to medication adherence using the barriers assessment tool. We compared rejection rates between patients with at least one reported adherence barrier to those without reported adherence barriers using a Kaplan-Meier estimator and Cox proportional hazard models to adjust for other mediators of acute rejection at 2 years following barriers assessment.

RESULTS

Ninety-eight patients were assessed for barriers to adherence. Over the 2-year observation period, 22 patients developed biopsy-proven acute rejection (BPAR). Kaplan-Meier estimates show that patients with an identified barrier to adherence were more likely to have BPAR (p = 0.02) than patients without an identified barrier in the 24 months following barriers assessment. The median time to rejection for patients who experienced acute rejection was 175.5 days (IQR 63-276 days) from the time of barriers assessment. An identified barrier to adherence remained the only statistically significant predictor of BPAR with Cox modeling (HR 2.6, p = 0.04), after accounting for age, sex, and race.

CONCLUSIONS

Pediatric and adolescent kidney transplant recipients with identified adherence barriers are at increased risk for acute rejection. Barriers to adherence provide a potentially modifiable therapeutic target that can be assessed in clinic to guide targeted interventions.

Simultaneous bilateral laparoscopic nephrectomy with kidney transplantation in patients with ESRD due to ADPKD: A single-center experience

Patients with autosomal dominant polycystic disease (ADPKD) may require bilateral nephrectomy (BN) in addition to kidney transplantation (KT) for symptom control. This study aims to compare simultaneous BNKT to contemporaneous controls by reviewing our cohort of ADPKD patients who underwent KT from a living donor from January 2014 to October 2019. Symptomatic patients who underwent laparoscopic BNKT were compared to KT alone. Clinical differences related to undertaking bilateral nephrectomies showed increased total kidney volumes (P < .001). We assessed operative parameters, complications, and clinical outcomes. The complications were classified according to the Clavien-Dindo system. In 148 transplant recipients, 51 underwent BNKT, and 97 KT alone. There was no difference in baseline demographics. BNKT recipients had longer cold ischemia time, required more ICU care, increased blood transfusions and longer hospital stays. The kidney function was similar in the first year in both groups, with no difference in delayed graft function, readmissions or severe grade III and IV complications within 3 months after surgery. Laparoscopic BNKT is safe and feasible at the time of living donor KT. Although higher acuity care is needed with a longer initial hospital stay, there are comparable posttransplant patient and allograft outcomes.

Clinical impact of serum bilirubin levels on kidney transplant outcomes

Serum bilirubin, a potent endogenous antioxidant, has been associated with decreased risks of cardiovascular disease, diabetes, and kidney disease. However, the effects of serum bilirubin on kidney transplant outcomes remain undetermined. We analyzed 1628 patients who underwent kidney transplantations between 2003 and 2017. Patients were grouped into sex-specific quartiles according to mean serum bilirubin levels, 3–12 months post-transplantation. Median bilirubin levels were 0.66 mg/dL in males and 0.60 mg/dL in females. The intra-individual variability of serum bilirubin levels was low (9%). Serum bilirubin levels were inversely associated with graft loss, death-censored graft failure, and all-cause mortality, independent of renal function, donor status, and transplant characteristics. Multivariable analysis revealed that the lowest serum bilirubin quartile was associated with increased risk of graft loss (HR 2.64, 95% CI 1.67–4.18, P < 0.001), death-censored graft failure (HR 2.97, 95% CI 1.63–5.42, P < 0.001), and all-cause mortality (HR 2.07, 95% CI 1.01–4.22, P = 0.046). Patients with lower serum bilirubin were also at greater risk of rejection and exhibited consistently lower glomerular filtration rates than those with higher serum bilirubin. Serum bilirubin levels were significantly associated with transplantation outcomes, suggesting that bilirubin could represent a therapeutic target for improving long-term transplant outcomes.

Hypertensive nephrosclerosis: wider kidney biopsy indications may be needed to improve diagnostics

BACKGROUND

Hypertensive nephrosclerosis is the presumed underlying cause in many end-stage kidney disease (ESKD) patients, but the diagnosis is disputed and based on clinical criteria with low diagnostic accuracy.

OBJECTIVE

To evaluate and improve the diagnostic process for nephrosclerosis patients.

METHODS

We included adults from the population-based HUNT study (n = 50 552), Norwegian CKD patients referred for kidney biopsy 1988-2012 (n = 7261), and unselected nephrology clinic patients (n = 193) used for matching. Decision tree analysis and ROC curve-based methods of optimal cut-offs were used to improve clinical nephrosclerosis criteria.

RESULTS

Nephrosclerosis prevalence was 2.7% in the general population, and eGFR decline and risk for kidney-related hospital admissions and ESKD were comparable to patients with diabetic kidney disease. In the biopsy cohort, current clinical criteria had very low sensitivity (0.13) but high specificity (0.94) for biopsy-verified arterionephrosclerosis. A new optimized diagnostic algorithm based on proteinuria (<0.75 g d^-1 ), systolic blood pressure (>155 mm Hg) and age (>75 years) only marginally improved diagnostic accuracy (sensitivity 0.19, specificity 0.96). Likewise, there were still false-positive cases with treatable diagnoses like glomerulonephritis, interstitial nephritis and others (40% of all test positive). Decision curve analysis showed that the new criteria can lead to higher clinical utility, especially for patients considering the potential harms to be close to the potential benefits, while the more risk-tolerant ones (harm:benefit ratio < 1:4) should consider kidney biopsy.

CONCLUSION

Further improvements of the current clinical criteria seem difficult, so risks and benefits of kidney biopsy could be more actively discussed with selected patients to reduce misclassification and direct treatment.

Significance of revised criteria for chronic active T cell-mediated rejection in the 2017 Banff classification: Surveillance by 1-year protocol biopsies for kidney transplantation

Diagnostic criteria for chronic active T cell-mediated rejection (CA-TCMR) were revised in the Banff 2017 consensus, but it is unknown whether the new criteria predict graft prognosis of kidney transplantation. We enrolled 406 kidney allograft recipients who underwent a 1-year protocol biopsy (PB) and investigated the diagnostic significance of Banff 2017. Interobserver reproducibility of the 3 diagnosticians showed a substantial agreement rate of 0.68 in Fleiss's kappa coefficient. Thirty-three patients (8%) were classified as CA-TCMR according to Banff 2017, and 6 were previously diagnosed as normal, 12 as acute TCMR, 10 with borderline changes, and 5 as CA-TCMR according to Banff 2015 criteria. Determinant factors of CA-TCMR were cyclosporine use (vs tacrolimus), previous acute rejection, and BK polyomavirus-associated nephropathy. In survival analysis, the new diagnosis of CA-TCMR predicted a composite graft endpoint defined as doubling serum creatinine or death-censored graft loss (log-rank test, P < .001). In multivariate analysis, CA-TCMR was associated with the second highest risk of the composite endpoint (hazard ratio: 5.42; 95% confidence interval, 2.02-14.61; P < .001 vs normal) behind antibody-mediated rejection. In conclusion, diagnosis of CA-TCMR in Banff 2017 may facilitate detecting an unfavorable prognosis of kidney allograft recipients who undergo a 1-year PB.

The Kidney Failure Risk Equation for Prediction of Allograft Loss in Kidney Transplant Recipients

Rationale & Objective

The Kidney Failure Risk Equation (KFRE) is a simple widely validated prediction model using age, sex, estimated glomerular filtration rate, and urinary albumin-creatinine ratio to predict the risk for end-stage kidney disease. Data are limited for its applicability to kidney transplant recipients.

Study Design

Validation study of the KFRE as a post hoc analysis of the Folic Acid for Vascular Outcomes Reduction in Transplantation (FAVORIT) Trial.

Setting & Participants

Adult kidney transplant recipients with functioning kidney allografts at least 6 months posttransplantation from 30 centers in the United States, Canada, and Brazil. Participants with estimated glomerular filtration rates < 60 mL/min/1.73 m² at study entry were included.

Predictor

2- and 5-year kidney failure risk predicted by the KFRE using variables at study entry.

Outcome

Graft loss, defined by initiation of dialysis.

Analytical Approach

Discrimination of the KFRE was assessed using C statistics; calibration was assessed by plotting predicted risk against observed cumulative incidence of graft loss.

Results

2,889 participants were included. Within 2 years, 98 participants developed graft loss, 107 participants died with a functioning graft, and 129 participants were lost to follow-up, and by 5 years, 252 had developed graft loss, 265 died with a functioning graft, and 1,543 were lost to follow-up. The KFRE demonstrated accurate calibration and discrimination (C statistic, 0.85 [95% CI, 0.81-0.88] at 2 years and 0.81 [95% CI, 0.78-0.84] at 5 years); performance was similar regardless of donor type (living vs deceased) and graft vintage, with the noted exception of poorer calibration for graft vintage less than 2 years.

Limitations

Unavailable cause of graft loss.

Conclusions

The KFRE accurately predicted the risk for graft loss among adult kidney transplant recipients with graft vintage longer than 2 years and may be a useful prognostic tool for nephrologists caring for kidney transplant recipients.

Prediction scores for risk of allograft loss in patients receiving kidney transplants: nil satis nisi optimum

Long-term graft survival is the main concern of kidney transplantation. Some strategies have been tested to predict graft survival using estimated glomerular filtration rate or proteinuria at different time points, histologic assessment, non-invasive biomarkers or even machine-learning methods. However, the 'magical formulae' for allograft survival prediction does not exist yet.

Prognostic value for long-term graft survival of estimated glomerular filtration rate and proteinuria quantified at 3 months after kidney transplantation

Background

The estimated glomerular filtration rate (eGFR) measured at 1 year is the usual benchmark applied in kidney transplantation (KT). However, acting on earlier eGFR values could help in managing KT during the first post-operative year. We aimed to assess the prognostic value for long-term graft survival of the early (3 months) quantification of eGFR and proteinuria following KT.

Methods

The 3-, 6- and 12-month eGFR using the Modified Diet in Renal Disease equation (eGFR_MDRD) was determined and proteinuria was measured in 754 patients who underwent their first KT between 2000 and 2010 (with a mean follow-up of 8.3 years) in our centre. Adjusted associations with graft survival were estimated using a multivariable Cox model. The predictive accuracy was estimated using the C-index and net reclassification index. These same analyses were measured in a multicentre validation cohort of 1936 patients.

Results

Both 3-month eGFR_MDRD and proteinuria were independent predictors of return to dialysis (all P < 0.05) and there was a strong correlation between eGFR at 3 and 12 months (Spearman’s ρ = 0.76). The predictive accuracy of the 3-month eGFR was within a similar range and did not differ significantly from the 12-month eGFR in either the derivation cohort [C-index 62.6 (range 57.2–68.1) versus 66.0 (range 60.1–71.9), P = 0.41] or the validation cohort [C-index 69.3 (range 66.4–72.1) versus 71.7 (range 68.7–74.6), P = 0.25].

Conclusion

The 3-month eGFR was a valuable predictor of the long-term return to dialysis whose predictive accuracy was not significantly less than that of the 12-month eGFR in multicentre cohorts totalling >2500 patients. Three-month outcomes may be useful in randomized controlled trials targeting early therapeutic interventions.

Prediction system for risk of allograft loss in patients receiving kidney transplants: international derivation and validation study

OBJECTIVE

To develop and validate an integrative system to predict long term kidney allograft failure.

DESIGN

International cohort study.

SETTING

Three cohorts including kidney transplant recipients from 10 academic medical centres from Europe and the United States.

PARTICIPANTS

Derivation cohort: 4000 consecutive kidney recipients prospectively recruited in four French centres between 2005 and 2014. Validation cohorts: 2129 kidney recipients from three centres in Europe and 1428 from three centres in North America, recruited between 2002 and 2014. Additional validation in three randomised controlled trials (NCT01079143, EudraCT 2007-003213-13, and NCT01873157).

MAIN OUTCOME MEASURE

Allograft failure (return to dialysis or pre-emptive retransplantation). 32 candidate prognostic factors for kidney allograft survival were assessed.

RESULTS

Among the 7557 kidney transplant recipients included, 1067 (14.1%) allografts failed after a median post-transplant follow-up time of 7.12 (interquartile range 3.51-8.77) years. In the derivation cohort, eight functional, histological, and immunological prognostic factors were independently associated with allograft failure and were then combined into a risk prediction score (iBox). This score showed accurate calibration and discrimination (C index 0.81, 95% confidence interval 0.79 to 0.83). The performance of the iBox was also confirmed in the validation cohorts from Europe (C index 0.81, 0.78 to 0.84) and the US (0.80, 0.76 to 0.84). The iBox system showed accuracy when assessed at different times of evaluation post-transplant, was validated in different clinical scenarios including type of immunosuppressive regimen used and response to rejection therapy, and outperformed previous risk prediction scores as well as a risk score based solely on functional parameters including estimated glomerular filtration rate and proteinuria. Finally, the accuracy of the iBox risk score in predicting long term allograft loss was confirmed in the three randomised controlled trials.

CONCLUSION

An integrative, accurate, and readily implementable risk prediction score for kidney allograft failure has been developed, which shows generalisability across centres worldwide and common clinical scenarios. The iBox risk prediction score may help to guide monitoring of patients and further improve the design and development of a valid and early surrogate endpoint for clinical trials.

TRIAL REGISTRATION

Clinicaltrials.gov NCT03474003.

Modeling graft loss in patients with donor-specific antibody at baseline using the Birmingham-Mayo (BirMay) predictor: Implications for clinical trials

Predicting which renal allografts will fail and the likely cause of failure is important in clinical trial design to either enrich patient populations to be or as surrogate efficacy endpoints for trials aimed at improving long-term graft survival. This study tests our previous Birmingham-Mayo model (termed the BirMay Predictor) developed in a low-risk kidney transplant population in order to predict the outcome of patients with donor specific alloantibody (DSA) at the time of transplantation and identify new factors to improve graft loss prediction in DSA+ patients. We wanted define ways to enrich the population for future therapeutic intervention trials. The discovery set included 147 patients from Mayo Cohort and the validation set included 111 patients from the Paris Cohort-all of whom had DSA at the time of transplantation. The BirMay predictor performed well predicting 5-year outcome well in DSA+ patients (Mayo C statistic = 0.784 and Paris C statistic = 0.860). Developing a new model did not improve on this performance. A high negative predictive value of greater than 90% in both cohorts excluded allografts not destined to fail within 5 years. We conclude that graft-survival models including histology predict graft loss well, both in DSA+ cohorts as well as DSA- patients.

A Prognostic Tool for Individualized Prediction of Graft Failure Risk within Ten Years after Kidney Transplantation

Identification of patients at risk of kidney graft loss relies on early individual prediction of graft failure. Data from 616 kidney transplant recipients with a follow-up of at least one year were retrospectively studied. A joint latent class model investigating the impact of serum creatinine (Scr) time-trajectories and onset of de novo donor-specific anti-HLA antibody (dnDSA) on graft survival was developed. The capacity of the model to calculate individual predicted probabilities of graft failure over time was evaluated in 80 independent patients. The model classified the patients in three latent classes with significantly different Scr time profiles and different graft survivals. Donor age contributed to explaining latent class membership. In addition to the SCr classes, the other variables retained in the survival model were proteinuria measured one-year after transplantation (HR=2.4, p=0.01), pretransplant non-donor-specific antibodies (HR=3.3, p<0.001), and dnDSA in patient who experienced acute rejection (HR=15.9, p=0.02). In the validation dataset, individual predictions of graft failure risk provided good predictive performances (sensitivity, specificity, and overall accuracy of graft failure prediction at ten years were 77.7%, 95.8%, and 85%, resp.) for the 60 patients who had not developed dnDSA. For patients with dnDSA individual risk of graft failure was not predicted with a so good performance.

Subclinical inflammation phenotypes and long-term outcomes after pediatric kidney transplantation

The implementation of surveillance biopsies in pediatric kidney transplantation remains controversial. Surveillance biopsies detect subclinical injury prior to clinical dysfunction, which could allow for early interventions that prolong allograft survival. We conducted a single-center retrospective cohort study of 120 consecutive pediatric kidney recipients, of whom 103 had surveillance biopsies ≤6 months posttransplant. We tested the hypothesis that subclinical inflammation (borderline or T cell-mediated rejection without clinical dysfunction) is associated with a 5-year composite endpoint of acute rejection and allograft failure. Overall, 36% of subjects had subclinical inflammation, which was associated with increased hazard for the composite endpoint (adjusted hazard ratio 2.89 [1.27, 6.57]; P < .01). Subjects with treated vs untreated subclinical borderline rejection had a lower incidence of the composite endpoint (41% vs 67%; P < .001). Subclinical vascular injury (subclinical inflammation with Banff arteritis score > 0) had a 78% incidence of the composite endpoint vs 11% in subjects with no major surveillance abnormalities (P < .001). In summary, we showed that subclinical inflammation phenotypes were prevalent in pediatric kidney recipients without clinical dysfunction and were associated with increased acute rejection and allograft failure. Once prospectively validated, our data would support implementation of surveillance biopsies as standard of care in pediatric kidney transplantation.

[Multidimensional approaches for risk stratification in transplantation]

Despite considerable progress in the short-term outcomes of renal transplantation, there has been little improvement over the last 15years on long-term survival. The main limitation is the lack of precise knowledge of the determinants of renal allograft loss and robust prognostic systems providing an individual prediction. Kidney transplantation must address a pressing clinical need to accurately define the determinants of kidney renal allograft survival in order to improve risk stratification. To achieve this goal, four steps need to be considered in the development of prognostic tools: the characterization and identification of the phenotype of the pathology, the assessment of prognostic factors of the event of interest in the population, the assessment of the additional value provided by a newly identified prognostic factor to conventional factors already known in clinical practice as well as the construction of prognostic tools, on the basis of multidimensional integrative models allowing a precise stratification of the risk, at individual and population scale.

Human Kidney Tubule-Specific Gene Expression Based Dissection of Chronic Kidney Disease Traits

Chronic kidney disease (CKD) has diverse phenotypic manifestations including structural (such as fibrosis) and functional (such as glomerular filtration rate and albuminuria) alterations. Gene expression profiling has recently gained popularity as an important new tool for precision medicine approaches. Here we used unbiased and directed approaches to understand how gene expression captures different CKD manifestations in patients with diabetic and hypertensive CKD.

Transcriptome data from ninety-five microdissected human kidney samples with a range of demographics, functional and structural changes were used for the primary analysis. Data obtained from 41 samples were available for validation. Using the unbiased Weighted Gene Co-Expression Network Analysis (WGCNA) we identified 16 co-expressed gene modules.

We found that modules that strongly correlated with eGFR primarily encoded genes with metabolic functions. Gene groups that mainly encoded T-cell receptor and collagen pathways, showed the strongest correlation with fibrosis level, suggesting that these two phenotypic manifestations might have different underlying mechanisms. Linear regression models were then used to identify genes whose expression showed significant correlation with either structural (fibrosis) or functional (eGFR) manifestation and mostly corroborated the WGCNA findings.

We concluded that gene expression is a very sensitive sensor of fibrosis, as the expression of 1654 genes correlated with fibrosis even after adjusting to eGFR and other clinical parameters. The association between GFR and gene expression was mostly mediated by fibrosis. In conclusion, our transcriptome-based CKD trait dissection analysis suggests that the association between gene expression and renal function is mediated by structural changes and that there may be differences in pathways that lead to decline in kidney function and the development of fibrosis, respectively.

•

Gene expression analysis of kidney samples shows the relationship between gene expression and eGFR is mediated by fibrosis

•

Immune related pathways show the strongest correlation with fibrosis development

•

Metabolic pathways show a strong correlation with eGFR

Chronic kidney disease is characterized by functional changes (glomerular filtration rate, eGFR) and structural changes (mainly renal fibrosis). Gene expression profiles of human kidney samples were analyzed to understand the relationship between these two manifestations. We found that the association between gene expression and eGFR is mediated by fibrosis, suggesting that fibrosis is a crucial determinant of functional kidney decline, and a potential therapeutic target. Gene expression analysis also indicates that fibrosis strongly correlates with immune pathways, and eGFR with metabolic pathways, highlighting potential mechanistic differences between structural and functional manifestations of kidney disease.

A Machine Learning Approach Using Survival Statistics to Predict Graft Survival in Kidney Transplant Recipients: A Multicenter Cohort Study

Accurate prediction of graft survival after kidney transplant is limited by the complexity and heterogeneity of risk factors influencing allograft survival. In this study, we applied machine learning methods, in combination with survival statistics, to build new prediction models of graft survival that included immunological factors, as well as known recipient and donor variables. Graft survival was estimated from a retrospective analysis of the data from a multicenter cohort of 3,117 kidney transplant recipients. We evaluated the predictive power of ensemble learning algorithms (survival decision tree, bagging, random forest, and ridge and lasso) and compared outcomes to those of conventional models (decision tree and Cox regression). Using a conventional decision tree model, the 3-month serum creatinine level post-transplant (cut-off, 1.65 mg/dl) predicted a graft failure rate of 77.8% (index of concordance, 0.71). Using a survival decision tree model increased the index of concordance to 0.80, with the episode of acute rejection during the first year post-transplant being associated with a 4.27-fold increase in the risk of graft failure. Our study revealed that early acute rejection in the first year is associated with a substantially increased risk of graft failure. Machine learning methods may provide versatile and feasible tools for forecasting graft survival.

An adjustable predictive score of graft survival in kidney transplant patients and the levels of risk linked to de novo donor-specific anti-HLA antibodies

Most predictive models and scores of graft survival in renal transplantation include factors known before transplant or at the end of the first year. They cannot be updated thereafter, even in patients developing donor-specific anti-HLA antibodies and acute rejection.We developed a conditional and adjustable score for prediction of graft failure (AdGFS) up to 10 years post-transplantation in 664 kidney transplant patients. AdGFS was externally validated and calibrated in 896 kidney transplant patients.The final model included five baseline factors (pretransplant non donor-specific anti-HLA antibodies, donor age, serum creatinine measured at 1 year, longitudinal serum creatinine clusters during the first year, proteinuria measured at 1 year), and two predictors updated over time (de novo donor-specific anti-HLA antibodies and first acute rejection). AdGFS was able to stratify patients into four risk-groups, at different post-transplantation times. It showed good discrimination (time-dependent ROC curve at ten years: 0.83 (CI95% 0.76–0.89).

Big Data, Predictive Analytics, and Quality Improvement in Kidney Transplantation: A Proof of Concept

We sought proof of concept of a Big Data Solution incorporating longitudinal structured and unstructured patient-level data from electronic health records (EHR) to predict graft loss (GL) and mortality. For a quality improvement initiative, GL and mortality prediction models were constructed using baseline and follow-up data (0-90 days posttransplant; structured and unstructured for 1-year models; data up to 1 year for 3-year models) on adult solitary kidney transplant recipients transplanted during 2007-2015 as follows: Model 1: United Network for Organ Sharing (UNOS) data; Model 2: UNOS & Transplant Database (Tx Database) data; Model 3: UNOS, Tx Database & EHR comorbidity data; and Model 4: UNOS, Tx Database, EHR data, Posttransplant trajectory data, and unstructured data. A 10% 3-year GL rate was observed among 891 patients (2007-2015). Layering of data sources improved model performance; Model 1: area under the curve (AUC), 0.66; (95% confidence interval [CI]: 0.60, 0.72); Model 2: AUC, 0.68; (95% CI: 0.61-0.74); Model 3: AUC, 0.72; (95% CI: 0.66-077); Model 4: AUC, 0.84, (95 % CI: 0.79-0.89). One-year GL (AUC, 0.87; Model 4) and 3-year mortality (AUC, 0.84; Model 4) models performed similarly. A Big Data approach significantly adds efficacy to GL and mortality prediction models and is EHR deployable to optimize outcomes.