Abstract: TH-PO1013

Development of a New Prediction Model for Graft Function in Living Donor Kidney Transplantation

Session Information

Category: Transplantation

  • 1702 Transplantation: Clinical and Translational

Authors

  • Matsukuma, Yuta, Department of Medicine and Clinical Science, Fukuoka, Japan
  • Masutani, Kosuke, Department of Medicine and Clinical Science, Fukuoka, Japan
  • Tanaka, Shigeru, Fukuoka Dental College, Fukuoka, Japan
  • Tsuchimoto, Akihiro, Department of Medicine and Clinical Science, Fukuoka, Japan
  • Unagami, Kohei, Tokyo Women's Medical University, Tokyo, Japan
  • Okumi, Masayoshi, Tokyo Women's Medical University, Tokyo, Japan
  • Tanabe, Kazunari, Tokyo Women's Medical University, Tokyo, Japan
  • Tsuruya, Kazuhiko, Department of Medicine and Clinical Science, Fukuoka, Japan
  • Kitazono, Takanari, Department of Medicine and Clinical Science, Fukuoka, Japan
Background

In recent years, there has been an increase in usage of grafts from marginal donors in living donor kidney transplantation. Such donors have several co-existing atherosclerotic factors such as aging, hypertension, dyslipidemia, and relatively low renal function. A simple prediction model for post-operative graft function may help to determine whether marginal donors are suitable in clinical settings of living donor kidney transplantation.

Methods

We conducted a single-center retrospective cohort study using clinical and laboratory measurements to construct a prediction model for graft function at 1 year. Low graft function was defined as estimated glomerular filtration rate (eGFR) ≤ 45 mL/min/1.73m2 at 1 year. A risk prediction model for low graft function was developed using multivariable logistic regression model with a stepwise backward elimination method.

Results

A total of 343 living donor kidney transplantation procedures were performed between January 2006 and May 2016. 123 patients had eGFR ≤ 45 mL/min/1.73m2 at 1 year after transplantation. Risk prediction model for low graft function was developed using donor age, pre-operative eGFR, hypertension, and donor/recipient body weight mismatch. The incidence of low graft function increased linearly with increase in total risk scores (p for trend <0.01). This model demonstrated modest discrimination (c-statistics = 0.75) and good calibration (Hosmer–Lemeshow test: p=0.77).

Conclusion

A new prediction model computed from four pre-operative variables created in this study is simple and useful system in clinical settings for prediction of graft function at 1 year in living donor kidney transplantation.