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

Abstract

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.

Joint latent class model: Simulation study of model properties and application to amyotrophic lateral sclerosis disease

“…The real parameters were chosen to mimic the real data, described in Stamenic et al paper [ 15 ], dealing with a prognostic tool for individualized prediction of graft failure risk within ten years after kidney transplantation, using serum creatinine progression as a longitudinal marker. Following Eqs.…”

“…The time points for repeated measures of the longitudinal marker are fixed to 1, 3, 6, 12, 18 and 24 months, following Stamenic et al [ 15 ]. The parameters vector for a 2-classes model, with class common random effect and error variance of mixed sub-model is as follows: …”

“…The real values for the parameters were chosen as follows: High separation framework. This setting is directly derived from Stamenic et al [ 15 ], resulted in β 01 =170, β 02 =100, β 11 =88 by year, β 12 =1.2 by year, and . Low separation framework.…”

“…These papers appeared following a comprehensive methodology review concerning the JLCM [ 7 ] and have different objectives. These objectives can be summarized as follows: 1) to study the relationship between a longitudinal biomarker and the risk of event [ 8 – 11 ]; 2) to identify sub-groups of longitudinal biomarker progression censored by the event occurrence [ 12 ]; 3) to study the impact of different factors on the longitudinal biomarker progression censored by the event occurrence [ 13 ]; 4) to predict the risk of an event based on the longitudinal biomarker progression [ 14 , 15 ]. Different implementations of the model were proposed to achieve a same objective.…”

“…For example, for the first objective, Syrjälä et al [ 8 ] search for the relation between childhood food consumption and the risk of advanced islet autoimmunity using a JLCM without covariates; Brilleman et al [ 9 ] explore the relationship between the changes in body mass index and the risk of death and/or transplant in hemodialysis patients by means of the JLCM for competing risks, including the pre-specified covariates with a common effect on latent classes only in the survival sub-model; Ogata et al [ 10 ] and Portegies et al [ 11 ] analyze the association between fasting plasma glucose progression and the risk of cardiovascular disease and the association between the blood pressure trajectories and the risk of stroke respectively by including the pre-specified covariates with a latent class-specific effect into the linear mixed sub-model and into the survival sub-model. As other examples, for the fourth objective, [ 14 ] search to prevent Alzeihmer disease using MMSE ( Mini-Mental State Examination ) score progression and creating a predictive risk model with class-specific covariates in both linear mixed sub-model and in the survival sub-model; Stamenic et al [ 15 ] defined latent classes to assess the impact of serum creatinine on graft failure risk with no covariates in JLCM, and performed a multivariable multinomial logistic analysis after defining these latent classes in order to analyze the factors associated to the classes.…”