Background: Pre-transplant quality assessments of the isolated islets are critical to predicting transplantation outcomes for type 1 diabetes. Although islet viability, morphological scoring, and oxygen consumption rate are reported as potential markers for successful human islet transplantations into diabetic immunodeficient mice [1, 2], identifying additional biomarkers is needed for accurate prediction. Since islets are susceptible to hypoxia, we hypothesized that the expressions of hypoxia-related genes in pre-transplant isolated islets, reflecting the intra-islet oxygen microenvironment, could be the potential predictive markers for the transplantation outcomes.

Methods: An overview of the study is shown in Figure 1. Human islets from deceased donors were isolated at City of Hope (n = 86 batches). Islets were transplanted under the kidney capsule of streptozotocin-induced diabetic NOD-scid mice (1200 IEQ per graft, 327 mice). Post-transplant glycemic control was assessed quantitatively using the area under the curve (AUC) of blood glucose from days 0–28 (AUC_0-28 [mg/dL*day]) in individual mice [1] and averaged per islet batch. Note that low AUC_0-28 indicates controlled post-transplant blood glucose level, and low AUC_0-28 is highly correlated to the diabetes reversal in mice [1]. Microfluidic qPCR of 40 hypoxia-related genes was employed using RNA from pre-transplant islets, and gene expressions were normalized to a housekeeping gene. Correlations between gene expressions and AUC_0-28 were calculated using linear regression. Multiple regression analysis was further employed to generate an equation for predicting the transplantation outcomes.

Results: Among 40 genes examined, 11 genes were significantly correlated to the transplantation outcome, (Figure 2A). Of note, DDIT4 expression demonstrated the highest correlation to the transplantation outcomes (R = 0.4971, P <0.0001). Further, we generated the prediction formula using multiple regression model with 6 genes (Prediction value = 7111 + 62*CD18 -116*EDN1 + 313*HK + 5059*Tie-2 - 1295*VHL + 1177*DDIT4), which demonstrated the improved correlation to transplantation outcome (R = 0.5580, Figure 2B). Lastly, we applied the prediction formula to predict diabetes reversal (qualitative factor), which calculated a high AUC of Receiver Operating Characteristic (0.81992, Figure 3A), with high sensitivity at 70.4% and specificity at 90.0% in predicting transplantation success (Figure 3B).Conclusions: Hypoxia-related genes of pre-transplant human islets are promising prediction markers for islet transplantation outcomes in diabetic mice. Notably, a multiple regression model of hypoxia-related genes was highly predictive of transplantation outcomes.