Is the Living Donor Kidney Profile Index Valuable in Japanese Patients?

Live-Donor Kidney Transplant Outcome Prediction (L-TOP) using artificial intelligence

BACKGROUND

Outcome prediction for live-donor kidney transplantation improves clinical and patient decisions and donor selection. However, the currently used models are of limited discriminative or calibration power and there is a critical need to improve the selection process. We aimed to assess the value of various artificial intelligence (AI) algorithms to improve the risk stratification index.

METHODS

We evaluated pre-transplant variables among 66 914 live-donor kidney transplants (performed between 1 December 2007 and 1 June 2021) from the United Network of Organ Sharing database, randomized into training (80%) and test (20%) sets. The primary outcome measure was death-censored graft survival. We tested four machine learning models for discrimination [time-dependent concordance index (CTD) and area under the receiver operating characteristic curve (AUC)] and calibration [integrated Brier score (IBS)]. We used decision-curve analysis to assess the potential clinical utility.

RESULTS

Among the models, the deep Cox mixture model showed the best discriminative performance (AUC = 0.70, 0.68 and 0.68 at 5, 10 and 13 years post-transplant, respectively). CTD reached 0.70, 0.67 and 0.66 at 5, 10 and 13 years post-transplant. The IBS score was 0.09, indicating good calibration. In comparison, applying the Living Kidney Donor Profile Index (LKDPI) on the same cohort produced a CTD of 0.56 and an AUC of 0.55-0.58 only. Decision-curve analysis showed an additional net benefit compared with the LKDPI 'treat all' and 'treat none' approaches.

CONCLUSION

Our AI-based deep Cox mixture model, termed Live-Donor Kidney Transplant Outcome Prediction, outperforms existing prediction models, including the LKDPI, with the potential to improve decisions for optimum live-donor selection by ranking potential transplant pairs based on graft survival. This model could be adopted to improve the outcomes of paired exchange programs.

Clinical performance of the iPREDICTLIVING tool for the prediction of the post-transplant recipient and living donor outcomes in a European cohort

A living donor kidney transplant (LDKT) is the best treatment for ESRD. A prediction tool based on clinical and demographic data available pre-KT was developed in a Norwegian cohort with three different models to predict graft loss, recipient death, and donor candidate's risk of death, the iPREDICTLIVING tool. No external validations are yet available. We sought to evaluate its predictive performance in our cohort of 352 pairs LKDT submitted to KT from 1998 to 2019. The model for censored graft failure (CGF) showed the worse discriminative performance with Harrell's C of .665 and a time-dependent AUC of .566, with a calibration slope of .998. For recipient death, at 10 years, the model had a Harrell's C of .776, a time-dependent AUC of .773, and a calibration slope of 1.003. The models for donor death were reasonably discriminative, although with a poor calibration, particularly for 20 years of death, with a Harrell's C of .712 and AUC of .694 with a calibration slope of .955. These models have moderate discriminative and calibration performance in our population. The tool was validated in this Northern Portuguese cohort, Caucasian, with a low incidence of diabetes and other comorbidities. It can improve the informed decision-making process at the living donor consultation joining clinical and other relevant information.

Clinical relevance of the living kidney donor profile index in Korean kidney transplant recipients

BACKGROUND

The Living Kidney Donor Profile Index (LKDPI) was developed in the United States to predict graft outcomes based on donor characteristics. However, there are significant differences in donor demographics, access to transplantation, proportion of ABO incompatibility, and posttransplant mortality in Asian countries compared with the United States.

METHODS

We evaluated the clinical relevance of the LKDPI score in a Korean kidney transplant cohort by analyzing 1860 patients who underwent kidney transplantation between 2000 and 2019. Patients were divided into three groups according to LKDPI score: <0, 1-19.9, and ≥20.

RESULTS

During a median follow-up of 119 months, 232 recipients (12.5%) experienced death-censored graft loss, and 98 recipients (5.3%) died. High LKDPI scores were significantly associated with increased risk of death-censored graft loss independent of recipient characteristics (LKDPI 1-19.9: HR 1.389, 95% CI 1.036-1.863; LKDPI ≥20: HR 2.121, 95% CI 1.50-2.998). High LKDPI score was also significantly associated with increased risk of biopsy-proven acute rejection and impaired graft renal function. By contrast, overall patient survival rates were comparable among the LKDPI groups.

CONCLUSION

High LKDPI scores were associated with an increased risk of death-censored graft loss, biopsy-proven acute rejection, and impaired graft renal function among a Korean kidney transplant cohort.

Comprehensive Analysis of Donor Factors for Allograft Survival in Living Kidney Transplantation: A Single-Center Study in Japan

BACKGROUND

Various donor characteristics have been reported as predictive factors for graft survival in kidney transplantations. The living kidney donor profile index (LKDPI) was established in 2016 to evaluate the quality of living donor kidneys. Herein, we verified whether the index score was associated with graft survival and analyzed various donor factors to identify predictors of graft survival in living donor kidney transplantations.

METHODS

This retrospective study included 130 patients who received a living donor kidney between 2006 and 2019 at our hospital. Clinical and laboratory data were obtained from the medical records. Living donor kidneys were categorized into 3 groups by LKDPI score, and the death-censored graft survival and predictors of graft survival were evaluated.

RESULTS

The median LKDPI score was 35 (IQR: 17-53). The index scores of the living donor kidneys in this study were higher than in previous studies. The groups with the highest scores (LKDPI >40) had significantly shorter death-censored graft survival compared with the group with the lowest scores (LKDPI <20; hazard ratio = 4.0, P = .005). There were no significant differences between the group with the middle scores (LKDPI, 20-40) and the other 2 groups. Donor/recipient weight ratio <0.9, ABO incompatibility, and 2 HLA-DR mismatches were identified as independent predictive factors for shorter graft survival.

CONCLUSION

The LKDPI was correlated with death-censored graft survival in this study. However, more studies are required to establish a modified index that is more accurate for Japanese patients.

The living kidney donor profile index fails to discriminate allograft survival: implications for its use in kidney paired donation programs

The inclusion of blood group- and human leukocyte antigen-compatible donor and recipient pairs (CPs) in kidney paired donation (KPD) programs is a novel strategy to increase living donor (LD) transplantation. Transplantation from a donor with a better Living Donor Kidney Profile Index (LKDPI) may encourage CP participation in KPD programs. We undertook parallel analyses using data from the Scientific Registry of Transplant Recipients and the Australia and New Zealand Dialysis and Transplant Registry to determine whether the LKDPI discriminates death-censored graft survival (DCGS) between LDs. Discrimination was assessed by the following: (1) the change in the Harrell C statistic with the sequential addition of variables in the LKDPI equation to reference models that included only recipient factors and (2) whether the LKDPI discriminated DCGS among pairs of prognosis-matched LD recipients. The addition of the LKDPI to reference models based on recipient variables increased the C statistic by only 0.02. Among prognosis-matched pairs, the C statistic in Cox models to determine the association of the LKDPI with DCGS was no better than chance alone (0.51 in the Scientific Registry of Transplant Recipient and 0.54 in the Australia and New Zealand Dialysis and Transplant Registry cohorts). We conclude that the LKDPI does not discriminate DCGS and should not be used to promote CP participation in KPD programs.