Multidimensional System to Dynamically Predict Graft Survival After Kidney Transplantation

The incidence of end stage renal disease (ESRD) is rapidly increasing, now affecting an estimated 7.4 million people worldwide. Numerous parameters such as demographic, clinical and functional factors drive the deterioration of the kidney, ultimately leading to ESRD. Although some ESRD prediction models have been derived in the past years, none of these models are dynamic: they do not integrate the repeated measurements recorded throughout individuals' follow-up. As highlighted in several studies, kidney function repeated measurements (i.e., trajectories) are highly associated with graft survival after kidney transplantation. The investigators made the hypothesis that these trajectories may bring relevant information in the context of graft survival risk prediction model. Hence, combining these trajectories with standard graft survival risk factors may enhance prediction performance. This could permit to derive a robust tool that could be updated over time by continuously capturing patient' personal evolution.

850 million individuals suffer from chronic kidney disease (CKD), while diabetes, cancer, and HIV/AIDS affect 422, 42, and 37 million individuals, respectively. End stage renal disease (ESRD) hence places a heavy burden on health systems worldwide. Linked to that, the kidney-disease-associated mortality rate worldwide has risen over the past decade, now causing the death of 5 to 10 million individuals every year. In kidney transplantation, numerous parameters such as demographic, clinical and functional factors drive the deterioration of the kidney, sometimes leading to graft failure. Current approaches for investigating the relationship between these factors and graft failure have been limited by standard statistical approaches and by registries with an overall lack on granular data, including infrequent kidney function measurements for a single patient and convenience clinical samples. Identifying the determinants of graft failure with a dynamic approach may bring an original perspective to the traditional graft survival risk prediction model that are impeded by their reliance on low-granularity datasets, cross-sectional parameters, and limited follow-up.