Introduction: Islet transplantation success is influenced by the mass of infused insulin producing tissue, resulting in many islet transplant programs employing a minimum dose of 5,000 IEQ/Kg for first transplants. Donor-specific factors (such as age, BMI, and past medical history) affect islet function. Similarly, recipient factors (such as pre-transplant insulin dose and duration of diabetes) may influence islet transplant success. However, current islet dosing strategies do not consider the impact of such factors when determining the islet mass that is required to achieve a successful transplant outcome. Here, we aim to determine if transplant dose can be tailored depending on specific donor and recipient factors.

Method: UK data were analysed from 75 first islet transplant recipients (2010-2021), at 3 months post-transplant. The Beta-2 Score (β2) was calculated using fasting C-peptide and glucose, insulin dose, and HbA1c. β2 score was modelled against a combined binary measure with 90min C-peptide ≥0.17nmol/L, insulin dose <0.5units/Kg, and HbA1c <7% represented as transplant success, and <0.17nmol/L, >0.5units/Kg, or >7% represented as transplant failure. The β2 score associated with a successful transplant outcome under these criteria was ≥3.8 (Accuracy 0.82, Precision 0.80, Specificity 0.77, Sensitivity 0.93). The impact of donor, recipient, and processing factors on transplant success (β2≥3.8) was determined by univariate modelling. Based on this univariate analysis, stepwise backwards elimination acquired a multivariate model from variables with p<0.25.

Results: From univariate modelling, IEQ/Kg was the only variable associated with transplant success (p=0.011), reinforced by point-biserial correlation (p=0.007). Using only IEQ/Kg, transplant success was predicted with 0.76 accuracy, 0.77 precision, 0.80 specificity, and 0.93 sensitivity. A subsequent multivariate model predicted transplant success with 0.82 accuracy, 0.82 precision, 0.80 specificity, and 0.95 sensitivity. Positively associated variables were IEQ/Kg (p=0.042) and donor girth (p=0.060), negatively associated variables were recipient pre-transplant insulin dose (p=0.075) and islet purity (p=0.135). Using the median of all other variables, the model predicted that a successful transplant required a minimum IEQ/Kg of 3,800. When increasing the recipient insulin dose from the median (0.5) to 0.7, the IEQ/Kg required increased to 5,000, highlighting how our algorithm can be used to tailor islet dose to specific donor and recipient factors.

Conclusion: These data demonstrate that IEQ/Kg is the best univariate indicator for transplant success. The introduction of additional donor, recipient, and processing factors increases prediction accuracy. This predictive algorithm allows an islet ‘dosing’ strategy to be applied based on individual cases, which could reduce the number of discarded islet preparations, ultimately increasing the effectiveness and availability of islets for transplant.