PET-CT imaging of neuroendocrine tumours - Beyond diagnostics

Background: Neuroendocrine tumours (NETs) typically overexpress somatostatin receptors. By radiolabelling somatostatin analogues with a positron-emitting radionuclide (68Ga-DOTATOC or 68Ga-DOTATATE), these tumours can be detected with high sensitivity and specificity using somatostatin receptor positron emission tomography-computed tomography (PET-CT).
Aim: To enhance knowledge about NETs imaged with PET-CT, beyond the diagnostic work-up.
Methods: Paper I, based on a clinical trial (Gapetto), evaluated whether treatment with long-acting somatostatin analogues affected the uptake of 68Ga-DOTATATE in the normal liver or tumours. Changes in uptake, measured as SUVmax, were assessed in patients before and after treatment initiation. Paper II, based on a cohort study, explored whether the total somatostatin receptor-expressing tumour volume measured by PET-CT imaging, correlated with health-related quality of life or specific NET symptoms in patients with metastatic gastroenteropancreatic NET (GEP-NET). Paper III was a developmental study aimed at constructing an Artificial intelligence (AI) model to automatically detect and quantify somatostatin receptor-expressing tumour volume using a UNet3D convolutional neural network. The AI model’s tumour segmentation was compared with that of two reference physicians. Paper IV, a retrospective study, assessed whether the total somatostatin receptor-expressing tumour volume at baseline PET-CT could predict treatment outcomes in GEP-NET patients post 177Lu-DOTATATE treatment. Tumour volumes were quantified from baseline and follow-up PET-CT images for these patients.
Results: Paper I revealed that treatment with long-acting somatostatin analogues significantly reduced the uptake of 68Ga-DOTATATE in normal liver tissue, while the tumour uptake remained unchanged. Paper II did not find a correlation between total tumour volume and health-related quality of life, although a weak positive correlation was observed between specific NET-associated symptoms (such as dyspnoea, diarrhoea, and flushing) and larger tumour volume. Paper III demonstrated that an AI model for tumour segmentation could be developed, displaying a strong correlation with the physicians’ reference segmentation. Paper IV indicated that baseline tumour volume did not predict treatment outcomes, but an increase in tumour volume at the first follow-up predicted worse outcomes.
Conclusions: Evaluating NETs with somatostatin receptor PET-CT is feasible after initiating treatment with long-acting somatostatin analogues. Factors other than tumour volume likely have a greater impact on the health-related quality of life in patients with metastasised GEP-NET. AI models can be developed to segment tumour volume from somatostatin receptor PET-CT. Baseline tumour volume was not a predictive factor for outcomes following treatment with 177Lu-DOTATATE.