A [68Ga]Ga-DOTANOC PET/CT Radiomic Model for Non-Invasive Prediction of Tumour Grade in Pancreatic Neuroendocrine Tumours

Abstract

Predicting grade 1 (G1) and 2 (G2) primary pancreatic neuroendocrine tumour (panNET) is crucial to foresee panNET clinical behaviour. Fifty-one patients with G1-G2 primary panNET demonstrated by pre-surgical [68Ga]Ga-DOTANOC PET/CT and diagnostic conventional imaging were grouped according to the tumour grade assessment method: histology on the whole excised primary lesion (HS) or biopsy (BS). First-order and second-order radiomic features (RFs) were computed from SUV maps for the whole tumour volume on HS. The RFs showing the lowest p-values and the highest area under the curve (AUC) were selected. Three radiomic models were assessed: A (trained on HS, validated on BS), B (trained on BS, validated on HS), and C (using the cross-validation on the whole dataset). The second-order normalized homogeneity and entropy was the most effective RFs couple predicting G2 and G1. The best performance was achieved by model A (test AUC = 0.90, sensitivity = 0.88, specificity = 0.89), followed by model C (median test AUC = 0.87, sensitivity = 0.83, specificity = 0.82). Model B performed worse. Using HS to train a radiomic model leads to the best prediction, although a "hybrid" (HS+BS) population performs better than biopsy-only. The non-invasive prediction of panNET grading may be especially useful in lesions not amenable to biopsy while [68Ga]Ga-DOTANOC heterogeneity might recommend FDG PET/CT.

Keywords: [68Ga]Ga-DOTANOC; biomarker; machine learning; pancreatic neuroendocrine tumour; standardized uptake value.

Figure 1

Workflow employed to extract radiomic features (RFs). First, segmentation of the primary lesion was performed on the transaxial PET images, then RFs were generated and extracted. The single and coupled RFs showing the lowest p-values and the highest AUC were selected and employed to develop the three predictive radiomic models used in this study. In model C, cross-validation (CV) on the whole mixed population was performed. Tumour grade was assessed either by histology on the whole excised primary lesion (HS) or on its biopsy (BS).

Figure 2

The procedure adopted to build model C starting from a “mixed” population (a) of patients with grade 1 (G1) and 2 (G2) pancreatic neuroendocrine neoplasia assessed through the analysis of histological (HS) or biopsy (BS) samples is described. In particular, repeated cross validation is applied 100 times to extract as many radiomic models trained and tested on randomly selected and balanced groups of patients (b) and exploited to build the final radiomic model (c).

Figure 3

[68Ga]Ga-DOTANOC PET/CT transaxial fused images of two pairs of patients presenting similar SUVmax values of the primary lesion but different tumour grades. In particular, although SUVmax of (a,b) is similar, (a) is G1 while (b) is G2; accordingly, (c,d) present similar SUVmax primary tumor values but (c) is G1 while (d) is G2. On the contrary, the primary lesion’s grade was correctly identified (G1: a,c; G2: b,d) when using the selected RFs.

Figure 4

The ROC curve of the discriminative radiomic model (a). Boxplots of radiomic scores (RS) of grade 1 (G1) and grade 2 (G2) groups, with median values of −2.53 and 2.53, respectively (b). Waterfall plot of the RS of G1 (green bars) and G2 (red bars) pancreatic neuroendocrine neoplasia (panNET) patients (c).

Figure 5

Discrimination of grade 2 (G2, in red) and grade 1 (G1, in green) pancreatic neuroendocrine neoplasia (panNET) patients by second-order normalized homogeneity and entropy, linearly separated by the linear discriminant analysis boundary (in black). Here, patients with grading assessed through histology on the whole excised primary lesion (HS) were well discriminated. This also represents the output of the training stage of model A.

Figure 6

ROC curves for test of model A (a), and training (b) and test (c) of model B are shown. In each figure, the area under the ROC curve (AUC) is reported together with sensitivity (SN) and specificity (SP).