Pancreatic Neuroendocrine Neoplasms: CT Spectral Imaging in Grading

Machine Learning Methods Based on CT Features Differentiate G1/G2 From G3 Pancreatic Neuroendocrine Tumors

RATIONALE AND OBJECTIVES

To identify CT features for distinguishing grade 1 (G1)/grade 2 (G2) from grade 3 (G3) pancreatic neuroendocrine tumors (PNETs) using different machine learning (ML) methods.

MATERIALS AND METHODS

A total of 147 patients with 155 lesions confirmed by pathology were retrospectively included. Clinical-demographic and radiological CT features was collected. The entire cohort was separated into training and validation groups at a 7:3 ratio. Least absolute shrinkage and selection operator (LASSO) algorithm and principal component analysis (PCA) were used to select features. Three ML methods, namely logistic regression (LR), support vector machine (SVM), and K-nearest neighbor (KNN) were used to build a differential model. Receiver operating characteristic (ROC) curves and precision-recall curves for each ML method were generated. The area under the curve (AUC), accuracy rate, sensitivity, and specificity were calculated.

RESULTS

G3 PNETs were more likely to present with invasive behaviors and lower enhancement than G1/G2 PNETs. The LR classifier yielded the highest AUC of 0.964 (95% confidence interval [CI]: 0.930, 0.972), with 95.4% accuracy rate, 95.7% sensitivity, and 92.9% specificity, followed by SVM (AUC: 0.957) and KNN (AUC: 0.893) in the training group. In the validation group, the SVM classier reached the highest AUC of 0.952 (95% CI: 0.860, 0.981), with 91.5% accuracy rate, 97.3% sensitivity, and 70% specificity, followed by LR (AUC: 0.949) and KNN (AUC: 0.923).

CONCLUSIONS

The LR and SVM classifiers had the best performance in the training group and validation group, respectively. ML method could be helpful in differentiating between G1/G2 and G3 PNETs.

Dual-layer spectral detector CT: A noninvasive preoperative tool for predicting histopathological differentiation in pancreatic ductal adenocarcinoma

PURPOSE

To predict histopathological differentiation grades in patients with pancreatic ductal adenocarcinoma (PDAC) before surgery with quantitative and qualitative variables obtained from dual-layer spectral detector CT (DLCT).

METHODS

Totally 128 patients with histopathologically confirmed PDAC and preoperative DLCT were retrospectively enrolled and categorized into the low-grade (LG) (well and moderately differentiated, n = 82) and high-grade (HG) (poorly differentiated, n = 46) subgroups. Both conventional and spectral variables for PDAC were measured. The ratio of iodine concentration (IC) values in arterial phase(AP) and venous phase (VP) was defined as iodine enhancement fraction_AP/VP (IEF_AP/VP). Necrosis was visually assessed on both conventional CT images (necrosis_con) and virtual mono-energetic images (VMIs) at 40 keV (necrosis_40keV). Forward stepwise logistic regression method was conducted to perform univariable and multivariable analysis. Receiver operating characteristic (ROC) curves and the DeLong method were used to evaluate and compare the efficiencies of variables in predicting tumor grade.

RESULTS

Necrosis_con (odds ratio [OR] = 2.84, 95% confidence interval [CI]: 1.13-7.13; p < 0.001) was an independent predictor among conventional variables, and necrosis_40keV (OR = 5.82, 95% CI: 1.98-17.11; p = 0.001) and IEF_AP/VP (OR = 1.12, 95% CI:1.07-1.17; p < 0.001) were independent predictors among spectral variables for distinguishing LG PDAC from HG PDAC. IEF_AP/VP (AUC = 0.754, p = 0.016) and combination model (AUC = 0.812, p < 0.001) had better predictive performances than necrosis_con (AUC = 0.580). The combination model yielded the highest sensitivity (72%) and accuracy (79%), while IEF_AP/VP exhibited the highest specificity (89%).

CONCLUSION

Variables derived from DLCT have the potential to preoperatively evaluate PDAC tumor grade. Furthermore, spectral variables and their combination exhibited superior predictive performances than conventional CT variables.

Dual-energy CT improves differentiation of non-hypervascular pancreatic neuroendocrine neoplasms from CA 19-9-negative pancreatic ductal adenocarcinomas

PURPOSE

To evaluate the utility of dual-energy CT (DECT) in differentiating non-hypervascular pancreatic neuroendocrine neoplasms (PNENs) from pancreatic ductal adenocarcinomas (PDACs) with negative carbohydrate antigen 19-9 (CA 19-9).

METHODS

This retrospective study included 26 and 39 patients with pathologically confirmed non-hypervascular PNENs and CA 19-9-negative PDACs, respectively, who underwent contrast-enhanced DECT before treatment between June 2019 and December 2021. The clinical, conventional CT qualitative, conventional CT quantitative, and DECT quantitative parameters of the two groups were compared using univariate analysis and selected by least absolute shrinkage and selection operator regression (LASSO) analysis. Multivariate logistic regression analyses were performed to build qualitative, conventional CT quantitative, DECT quantitative, and comprehensive models. The areas under the receiver operating characteristic curve (AUCs) of the models were compared using DeLong's test.

RESULTS

The AUCs of the DECT quantitative (based on normalized iodine concentrations [nICs] in the arterial and portal venous phases: 0.918; 95% confidence interval [CI] 0.852-0.985) and comprehensive (based on tumour location and nICs in the arterial and portal venous phases: 0.966; 95% CI 0.889-0.995) models were higher than those of the qualitative (based on tumour location: 0.782; 95% CI 0.665-0.899) and conventional CT quantitative (based on normalized conventional CT attenuation in the arterial phase: 0.665; 95% CI 0.533-0.797; all P < 0.05) models. The DECT quantitative and comprehensive models had comparable performances (P = 0.076).

CONCLUSIONS

Higher nICs in the arterial and portal venous phases were associated with higher blood supply improving the identification of non-hypervascular PNENs.

Progress in the application of dual-energy CT in pancreatic diseases

Pancreatic diseases are difficult to diagnose due to their insidious onset and complex pathophysiological developmental characteristics. In recent years, dual-energy computed tomography (DECT) imaging technology has rapidly advanced. DECT can quantitatively extract and analyze medical imaging features and establish a correlation between these features and clinical results. This feature enables the adoption of more modern and accurate clinical diagnosis and treatment strategies for patients with pancreatic diseases so as to achieve the goal of non-invasive, low-cost, and personalized treatment. The purpose of this review is to elaborate on the application of DECT for the diagnosis, biological characterization, and prediction of the survival of patients with pancreatic diseases (including pancreatitis, pancreatic cancer, pancreatic cystic tumor, pancreatic neuroendocrine tumor, and pancreatic injury) and to summarize its current limitations and future research prospects.

A nomogram for predicting postoperative complications based on tumor spectral CT parameters and visceral fat area in gastric cancer patients

PURPOSE

To construct a nomogram combining tumor spectral CT parameters and visceral fat area (VFA) to predict postoperative complications (POCs) in patients with gastric cancer (GC).

METHOD

This retrospective study included 101 GC patients who underwent preoperative abdominal spectral CT scan and were divided into two groups (37 with POCs and 64 without POCs) according to the Clavien-Dindo classification standard. Logistic regression was used to establish spectral, VFA, and combined models for predicting POCs. The combined prediction model was presented as a nomogram, and the diagnostic performance of each model was evaluated using receiver operating characteristic (ROC) curve analysis.

RESULTS

The AUCs of the VFA and spectral model were 0.71 (95% CI: 0.62-0.80) and 0.81 (95% CI: 0.72-0.88), respectively. VFA, the slope of spectral curve (λ) in venous phase (λ-VP) and tumor Hounsfield units on monoenergetic images 40 keV in VP (MonoE40keV-VP) were independent predictors of POCs in GC. The nomogram yielded an AUC of 0.89 (95% CI: 0.81-0.94). The combined model was superior to the VFA or spectral models by comparing their AUCs (P = 0.000 and 0.022).

CONCLUSIONS

The nomogram based on two tumor spectral parameters (λ-VP, MonoE40keV-VP) and VFA could serve as a convenient tool for predicting the POCs of GC patients.

Spectral CT - a new supplementary method for preoperative assessment of pathological grades of esophageal squamous cell carcinoma

BACKGROUND

Spectral CT imaging parameters have been reported to be useful in the differentiation of pathological grades in different malignancies. This study aims to investigate the value of spectral CT in the quantitative assessment of esophageal squamous cell carcinoma (ESCC) with different degrees of differentiation.

METHODS

There were 191 patients with proven ESCC who underwent enhanced spectral CT from June 2018 to March 2020 retrospectively enrolled. These patients were divided into three groups based on pathological results: well differentiated ESCC, moderately differentiated ESCC, and poorly differentiated ESCC. Virtual monoenergetic 40 keV-equivalent image (VMI40keV), iodine concentration (IC), water concentration (WC), effective atomic number (Eff-Z), and the slope of the spectral curve(λHU) of the arterial phase (AP) and venous phase (VP) were measured or calculated. The quantitative parameters of the three groups were compared by using one-way ANOVA and pairwise comparisons were performed with LSD. Receiver operating characteristic (ROC) analysis was used to evaluate the diagnostic performance of these parameters in poorly differentiated groups and non-poorly differentiated groups.

RESULTS

There were significant differences in VMI40keV, IC, Eff-Z, and λHU in AP and VP among the three groups (all p < 0.05) except for WC (p > 0.05). The VMI40keV, IC, Eff-Z, and λHU in the poorly differentiated group were significantly higher than those in the other groups both in AP and VP (all p < 0.05). In the ROC analysis, IC performed the best in the identification of the poorly differentiated group and non-poorly differentiated group in VP (AUC = 0.729, Sensitivity = 0.829, and Specificity = 0.569 under the threshold of 21.08 mg/ml).

CONCLUSIONS

Quantitative parameters of spectral CT could offer supplemental information for the preoperative differential diagnosis of ESCC with different degrees of differentiation.

Spectral Computed Tomography-Derived Iodine Content and Tumor Response in the Follow-Up of Neuroendocrine Tumors-A Single-Center Experience

Spectral computed tomography (SCT) allows iodine content (IC) calculation for characterization of hypervascularized neoplasms and thus might help in the staging of neuroendocrine tumors (NETs). This single-center prospective study analyzed the association between SCT-derived IC and tumor response in the follow-up of metastasized NETs. Twenty-six patients with a median age of 70 years (range 51-85) with histologically proven NETs and a total of 78 lesions underwent SCT for staging. Because NETS are rare, no primary NET types were excluded. Lesions and intralesional hotspots were measured in virtual images and iodine maps. Tumor response was classified as progressive or nonprogressive at study endpoint. Generalized estimating equations served to estimate associations between IC and tumor response, additionally stratified by lesion location. Most commonly affected sites were the lymph nodes, liver, pancreas, and bones. Median time between SCT and endpoint was 64 weeks (range 5-260). Despite statistical imprecision in the estimate, patients with higher IC in lymphonodular metastases had lower odds for disease progression (adjusted OR = 0.21, 95% CI: 0.02-2.02). Opposite tendencies were observed in hepatic and pancreatic metastases in unadjusted analyses, which vanished after adjusting for therapy and primary tumor grade.

Qualitative imaging features of pancreatic neuroendocrine neoplasms predict histopathologic characteristics including tumor grade and patient outcome

OBJECTIVES

To identify PanNEN imaging features associated with tumor grade and aggressive histopathological features.

METHODS

Associations between histopathological and imaging features of resected PanNEN were retrospectively tested. Histopathologic features included WHO grade, lymphovascular invasion (LVI), growth pattern (infiltrative, circumscribed), and intratumoral fibrosis (mature, immature). Imaging features included size, degree/uniformity of enhancement, progressive enhancement, contour, infiltrative appearance (infiltrative_im), calcifications, cystic components, tumor thrombus, vascular occlusion (VO), duct dilatation, and atrophy. Multinomial logistic regression analyses evaluated the magnitude of associations. Association of variables with outcome was assessed using Cox-proportional hazards regression.

RESULTS

133 patients were included. 3 imaging features (infiltrative_im, ill-defined contour [contour_ill], and VO) were associated with all histopathologic parameters and poor outcome. Increase in grade increased odds of contour_ill by 15.6 times (p = 0.0001, 95% CI 3.8-64.4). PanNEN with VO were 51.1 times (p = 0.0002, 6.5-398.6) more likely to demonstrate LVI. For PanNEN with contour_ill, infiltrative growth pattern was 51.3 times (p < 0.0001, 9.1-288.4), and fibrosis was 14 times (p = 0.0065, 2.1-93.7) more likely. Contour_ill was associated with decreased recurrence-free survival (p = 0.0003, HR 18.29, 3.83-87.3) and VO (p = 0.0004, HR6.08, 2.22-16.68) with decreased overall survival.

CONCLUSIONS

Infiltrative_im, contour_ill, and VO on imaging are associated with higher grade/histopathological parameters linked to tumor aggression, and poor outcome.

Gender differences in pancreatic neuroendocrine neoplasms: A retrospective study based on the population of Hubei Province, China

OBJECTIVE

The aims of the present study were to investigate gender differences in the clinicopathological features, distant metastasis and prognosis of pancreatic neuroendocrine neoplasms (pNENs) in a Chinese population, and to identify any important gaps in the classification and management of pNENs relative to gender.

METHODS

Retrospective collection of the clinicopathological data of 193 patients with pathologically confirmed pNENs were analyzed and follow up was extended to observe the prognosis of the disease. Differences between genders in basic characteristics, clinical symptoms, comorbidities, and tumor parameters were analyzed.

RESULTS

There was no significant difference in females and males, however, moderately higher for females (52.8% vs. 47.2%), with the largest subgroup being 40~60 years of age (54.9%). Age at onset (P=0.002) and age at diagnosis (P=0.005) were both younger in females compared to males. Males lived more in urban areas and females lived more in rural areas (P=0.047). The proportion of smokers and alcohol drinkers was significantly higher in males than in females (P < 0.001). Non-functional pNENs were more frequent in males and functional pNENs in females (P=0.032). In women, functional status of the tumor was significantly associated with metastatic outcome (P=0.007) and functional tumors proved to be a protective factor compared to non-functional tumors (OR=0.090,95% CI: 0.011~ 0.752). There were no gender differences in tumor size, location, grade, stage or prognosis.

CONCLUSIONS

Gender differences in some clinicopathological features, and distant metastasis in patients with pNENs were identified, which suggested certain management details that justified emphasis based on gender.

Spectral CT in clinical routine imaging of neuroendocrine neoplasms

AIM

To evaluate the potential of new spectral computed tomography (SCT)-based tools in patients with neuroendocrine neoplasms (NEN).

MATERIAL AND METHODS

Eighty-eight consecutive patients with NENs were included prospectively. The patients underwent multiphase CT with spectral and standard mode. The signal-to-noise ratio (SNR)/contrast-to-noise-ratio (CNR)_{tumour-to-liver}, iodine concentrations (ICs, total tumour/hotspot) and attenuation slopes in virtual monochromatic images (VMIs) were used to assess NEN-specific SCT values in primary tumours and metastatic lesions and investigate a possible lesion contrast improvement as well as possible correlations of SCT parameters to primary tumour location and tumour grade. Furthermore, the usability of SCT parameters to differentiate between the primary tumour and metastatic lesions, and to predict tumour response after 6-months follow-up was analyzed. The applied dose of spectral and standard mode was compared intra-individually.

RESULTS

SNR/CNR_{tumour-to-liver} significantly increased in low-energy VMIs. NENs showed significant differences in ICs between primary and metastatic lesions for both absolute and normalised values (p<0.001) regardless of whether the total tumour or the hotspot was measured. There was also a significant difference in the attenuation slope (p<0.001). No significant correlations were found between SCT and tumour grade. A tumour response prediction by SCT parameters was not possible. The applied dose was comparable between the scan modes.

CONCLUSION

SCT was comparable regarding applied dose, improved tumour contrast, and contributed to differentiation between primary NEN and metastasis.

Pancreatic neuroendocrine tumours: spectrum of imaging findings

Pancreatic neuroendocrine tumours (pNETs) are rare and heterogeneous group of neoplasms presenting with a wide variety of symptoms and biological behaviour, from indolent to aggressive ones. pNETs are stratified into functional or non-functional, because of their ability to produce metabolically active hormones. pNETs can be an isolate phenomenon or a part of a hereditary syndrome like von Hippel-Lindau syndrome or neurofibromatosis-1. The incidence has increased in the last years, also because of the improvement of cross-sectional imaging. Computed tomography (CT), magnetic resonance imaging (MRI) and functional imaging are the mainstay imaging modalities used for tumour detection and disease extension assessment, due to easy availability and better contrast/spatial resolution. Radiological imaging plays a fundamental role in detection, characterization and surveillance of pNETs and is involved in almost every stage of patients' management. Moreover, with specific indications and techniques, interventional radiology can also play a role in therapeutic management. Surgery is the treatment of choice, consisting of either partial pancreatectomy or enucleation of the primary tumour. This article reviews the radiologic features of different pNETs as well as imaging mimics, in order to help radiologists to avoid potential pitfalls, to reach the correct diagnosis and to support the multidisciplinary team in establishing the right treatment.