Long-term tolerability of PRRT in 807 patients with neuroendocrine tumours: the value and limitations of clinical factors

Peptide receptor radionuclide therapy (PRRT) with Y-90 and Lu-177 provides objective responses in neuroendocrine tumours, and is well tolerated with moderate toxicity. We aimed to identify clinical parameters predictive of long-term renal and haematological toxicity (myelodysplastic syndrome and acute leukaemia). Of 807 patients studied at IEO-Milan (1997-2013), 793 (98 %) received Lu-177 (278, 34.4 %), Y-90 (358, 44.4 %) or Lu-177 and Y-90 combined (157. 19.5 %), and 14 (2 %) received combinations of PRRT and other agents. Follow-up was 30 months (1-180 months). The parameters evaluated included renal risk factors, bone marrow toxicity and PRRT features. Data analysis included multiple regression, random forest feature selection, and recursive partitioning and regression trees. Treatment with Y-90 and Y-90 + Lu-177 was more likely to result in nephrotoxicity than treatment with Lu-177 alone (33.6 %, 25.5 % and 13.4 % of patients, respectively; p < 0.0001). Nephrotoxicity (any grade), transient and persistent, occurred in 279 patients (34.6 %) and was severe (grade 3 + 4) in 12 (1.5 %). In only 20-27 % of any nephrotoxicity was the disease modelled by risk factors and codependent associations (p < 0.0001). Hypertension and haemoglobin toxicity were the most relevant factors. Persistent toxicity occurred in 197 patients (24.3 %). In only 22-34 % of affected patients was the disease modelled by the clinical data (p < 0.0001). Hypertension (regression coefficient 0.14, p < 0.0001) and haemoglobin toxicity (regression coefficient 0.21, p < 0.0001) were pertinent factors. Persistent toxicity was associated with shorter PRRT duration from the first to the last cycle (mean 387 vs. 658 days, p < 0.004). Myelodysplastic syndrome occurred in 2.35 % of patients (modelled by the clinical data in 30 %, p < 0.0001). Platelet toxicity grade (2.05 +/- 1.2 vs. 0.58 +/- 0.8, p < 0.0001) and longer PRRT duration (22.6 +/- 24 vs. 15.5 +/- 9 months, p = 0.01) were relevant. Acute leukaemia occurred in 1.1 % of patients (modelled by the clinical data in 18 %, p < 0.0001). Identified risk factors provide a limited (< 30 %) risk estimate even with target tissue dosimetry. These data strongly suggest the existence of unidentified individual susceptibilities to radiation-associated disease.