• 177Lu-oxodotreotide (Lutathera®)
• G1
• G2
• GEP-NETs
• Ki 67
• PRRT treatment
• radiomic features
• [18F]-FDG PET /CT
• [68Ga]Ga-DOTATOC PET/CT

GEP-NETs (Gastroenteropancreatic neuroendocrine tumours) are the most important NET subtype, ranging between 55% and 70% of all NETs, whose overall incidence is less than 6 cases per year per 100000 individuals. Therefore, they are considered as rare neoplastic entities, but their incidence and prevalence do continue to rise all over the world.
In this study we selected patients affected by GEP-NET (G1, G2) in disease progression and subsequently undergone [68Ga]Ga-DOTATOC PET/CT before and after PRRT (Peptide Receptor Radionuclide Therapy) with 177Lu-oxodotreotide (Lutathera®) between May 2019 and June 2022. We first aimed to evaluate the predictive value of [68Ga]Ga-DOTATOC PET/CT radiomic features at baseline. Then, we assessed the integration of [68Ga]Ga-DOTATOC PET/CT radiomic features with a new diagnostic histopathological approach to the evaluation of Ki-67 in GEP-NET, recently developed by our group as a potential harbinger of a new risk stratification and prediction of PRRT response.
The study cohort included 32 patients, 10 G1/22 G2, of which 14 women and 18 men aged between 38 and 84 yrs, whose mean, median and standard deviation were 64,13, 66,5 and 12,28 yrs, respectively. At the time of the baseline [68Ga]Ga-DOTATOC PET/CT, all patients were under SA (Somatostatin) treatment and their CgA (Chromogranin) mean values were 966,88 ± 278,21 SEM.
A total of 51 [68Ga]Ga-DOTATOC PET/CT examinations were collected: 32 baseline and 19 follow-up, these last all performed after the end of PRRT. PRRT was administered according to the guidelines in 4 cycles 8 weeks apart (cumulativity activity 26,3 +/- 5,93 Gbq - min 7,17 max 29,52 GBq-). 13 patients did not have a follow-up [68Ga] Ga-DOTATOC PET/CT because 8 of them finished the treatment less than 3 months before data collection, 3 patients had a clinical progression, 1 died before the end of the treatment and 1 patient is currently under therapy.
A subgroup of patients (20/32) had a histopathological re-evaluation of the expression of Ki-67, based on staining intensity, in order to compare if and how there was a different clinical impact between the old and the new Ki-67.
All [68Ga]Ga-DOTATOC PET/CT scans were analysed by using EANM guidelines, , in accordance to which image acquisition, comprising a whole-body scan (from the head to the middle of the upper leg), was obtained about 30/40 minutes after [68 Ga]Ga-DOTATOC administration (2MBq per kg). Image analysis and radiomic features extraction were have been previously detailed. Briefly, the LIFEx software (http://www.lifexsoft.org) was used to segment all patients’ lesions by a semi-automatic method with a 40% threshold for volume of interest (VOI) delineation. Texture protocol was then dedicated to the extraction of advanced image features and a total of 158 mainly divided in first and second order features were computed.
573 SSTR-2 (Somatostatin Receptor-2) positive lesions were identified, in detail, 371 in basal scan and 202 in follow-up, articulated in 561 metastases (324 liver, 125 bone, 85 lymph node, 12 peritoneum, 6 lung, and 9 other site) and 12 primitive tumours.
In the 20 patients with the reassessment of Ki-67, in 13 the same Grading was highlighted, whereas in 7 a different was revealed, whereof 6 exhibited a new Grading from G2 to G1 and 1 from G1 to G2.
Radiomic features extracted from [68 Ga] DOTATOC PET/CT were able to discriminate lesions on the basis of primitive and metastasis site and to characterise/discriminate lesions presenting partial response in follow-up scans (AUC > 75%). In terms of discriminated capacity for PRRT response, the new histopathological grading system performed better as compared to the standard one in terms of downgrading to G1 patients responding to PRRT. By the same token, CgA, although just after the third and the fourth PRRT cycle, showed the ability to identify patients in predicting progression. The capacity of new Ki67 (p=0,0279) and then of New Grading (p=0,0312), in comparison to the old, and Chromogranin (p=0,0293 at third PRRT cycle; p=0,0175 at the fourth) in predicting progression was revealed by performing a Kruskal-Wallis test. In addition, a difference PET response emerged as a function of primitive sites: specifically, lesions in ileum appeared to be more stable as respect to in occult site (progression) and in pancreas where progression and partial response prevailed.
The models here developed showed that radiomic analysis was, by itself, able to identify patients with PR, thus highlighting its fundamental role in patient selection for PRRT. When clinical and histopathological findings were added to the model - specifically the new histopathological grading system - the same features better perform in predicting PRRT response to treatment, in particular identifying patients with either CR or PD. Therefore, the combination of radiomic and clinical and histopathological features (macro and micro-imaging) helps to build a better prediction of disease stability or partial response to PRRT, underlying the need of further research in NET subtype G2 field to better phenotype patients to be candidate to PRRT.