TABLE 4.
| Study | Kasiske et al., 2010 (72) | Foucher et al., 2010 (73) | Moore et al., 2011 (74) | Schnitzler et al., 2012 (14) | Shabir et al., 2014 (75); Gonzales et al., 2016 (76) | Gonzales et al., 2016 (76) | Prémaud et al., 2017 (77) | Loupy et al., 2019 (9) |
|---|---|---|---|---|---|---|---|---|
| Parameter | USRDS Risk Prediction Tool | KTFS | LOTESS Composite Risk Score | USRDS Predictive Model | Birmingham Risk Score | Birmingham-Mayo Histology-Based Model | AdGFS | iBox Risk Prediction Score |
| Development set | USRDS registry data (N = 59,091) | Multicentre French registry (N = 2,169) | Multicentre national cohort study (N = 2,763) | USRDS registry data (N = 87,575) | Single-center UK data (N = 651) | Single-center US data (N = 1,465) | Single-center French data (N = 664) | French multicentre cohort (N = 4,000) |
| External validation | No | Yes (N = 317) | Yes (single UK center; N = 731) | No | Yes (2 European centers (N = 736, N = 787) and 1 Canadian center (N = 475); 1 US center N = 1,465) | No | Yes (2 other French centers; N = 896) | Yes; N = 3,557 (2,129 patients in 3 European centers; 1,428 in 3 North American centers) |
| Prediction time point | 12 months post-transplant | 12 months post-transplant | Variable time after 12 months post-transplant | 12 months post-transplant | 12 months post-transplant | 12 months post-transplant | Time adjusted (only for ‘rejection’) | Time adjusted |
| Outcome parameter | Overall graft failure at 5 years after transplantation | Death-censored graft failure at 8 years | Overall graft failure and death-censored graft failure over time; follow-up time not specified | Overall graft failure beyond 1 year post-transplant, up to 9 years | Overall graft failure and death-censored graft failure at 5 years post-transplant | Overall graft failure and death-censored graft failure at 5 years post-transplant | Death-censored graft failure beyond 2 years post-transplant, up to 10 years | Death-censored graft failure over time post-transplant, up to 7 years |
| Pre-transplant factors included in the model | Recipient age | Recipient sex | Recipient age | A large array of donor and recipient demographic factors (N = 20) | Recipient age | Recipient age | Donor age Pre-transplant non-DSA HLA antibodies | Yes, adjusted for all relevant factors |
| Recipient race | Recipient age | Recipient sex | Recipient sex | Recipient sex | ||||
| Insurance | # Previous transplantations | Recipient race | Recipient race | Recipient race | ||||
| Cause of ESRD | Donor creatinine | |||||||
| Post-transplant factors included in the model | eGFR at 12 months Hospitalization | Serum creatinine Acute rejection Creatinine at 3 months 24-h proteinuria | eGFR at 12 months eGFR evolution Acute rejection Serum urea at 12 months Serum albumin | eGFR at 12 months Acute rejection within first year | Acute rejection eGFR Serum albumin UACR | Acute rejection eGFR UACR Black ethnicity Glomerulitis score Tubular atrophy score | Serum creatinine Proteinuria dnDSA Serum creatinine trajectory Acute rejection | Time post-transplant eGFR Proteinuria Histology (IFTA, microcirculation inflammation, TG) DSA-MFI |
| Prognostic accuracy | C-statistic 0.65–0.78 | ROC AUC 0.78 (0.73–0.80) | C-statistic 0.83 for death-censored graft failure; 0.70 for overall graft failure | Not reported | C-statistic 0.78–0.90 for death-censored failure; 0.75–0.81 for overall graft failure | C-statistic 0.90 for death-censored failure; 0.81 for overall graft failure | C-statistic at 10 years post-transplant 0.83 (0.76–0.89) | C-statistic 0.81 in development cohort, 0.81 in European validation cohort, 0.80 in US validation cohort |
| Calibration | Good | Not assessed | Good | Good | Good | Good | Good | Good |
| Limitations | No external validation set No data on DSA No data on proteinuria Prognostic accuracy moderate | Small validation set Validity not tested in other countries No data on DSA No data on rejection phenotype Limited prognostic accuracy | Small validation set Validity not tested in other countries No data on DSA No data on rejection phenotype Prediction time point variable | No external validation set No data on DSA No data on proteinuria No data on rejection phenotype | No data on rejection phenotype No data on DSA | No external validation set Data on DSA did not improve the model | Small validation sets and validity in other countries not tested Not tested in living donors or patients with pre-transplant DSA | Not yet prospectively implemented in an RCT |
| Tested in randomized trial data? | No | No | No | Yes, but calibration and validity as surrogacy for improved outcome by the intervention was not tested | No | No | No | Yes; validation in 3 RCTs; association with improved outcome not confirmed given lack of efficacy of the intervention |
AdGFS, adjustable score for prediction of graft failure; dn, de novo; DSA, donor-specific antibodies; eGFR, estimated glomerular filtration rate; ESRD, end-stage renal disease; IFTA, interstitial fibrosis and tubular atrophy; KTFS, kidney transplantation failure score; LOTESS, long-term efficacy and safety surveillance; MFI, mean fluorescence intensity; RCT, randomized controlled trial; TG, transplant glomerulopathy; UACR, urine albumin to creatinine ratio; USRDS, United States Renal Data System.