C-reactive protein may be a prognostic factor for the whole gastroenteropancreatic neuroendocrine tumor group

Novel ensemble feature selection techniques applied to high-grade gastroenteropancreatic neuroendocrine neoplasms for the prediction of survival

BACKGROUND AND OBJECTIVE

Determining the most informative features for predicting the overall survival of patients diagnosed with high-grade gastroenteropancreatic neuroendocrine neoplasms is crucial to improve individual treatment plans for patients, as well as the biological understanding of the disease. The main objective of this study is to evaluate the use of modern ensemble feature selection techniques for this purpose with respect to (a) quantitative performance measures such as predictive performance, (b) clinical interpretability, and (c) the effect of integrating prior expert knowledge.

METHODS

The Repeated Elastic Net Technique for Feature Selection (RENT) and the User-Guided Bayesian Framework for Feature Selection (UBayFS) are recently developed ensemble feature selectors investigated in this work. Both allow the user to identify informative features in datasets with low sample sizes and focus on model interpretability. While RENT is purely data-driven, UBayFS can integrate expert knowledge a priori in the feature selection process. In this work, we compare both feature selectors on a dataset comprising 63 patients and 110 features from multiple sources, including baseline patient characteristics, baseline blood values, tumor histology, imaging, and treatment information.

RESULTS

Our experiments involve data-driven and expert-driven setups, as well as combinations of both. In a five-fold cross-validated experiment without expert knowledge, our results demonstrate that both feature selectors allow accurate predictions: A reduction from 110 to approximately 20 features (around 82%) delivers near-optimal predictive performances with minor variations according to the choice of the feature selector, the predictive model, and the fold. Thereafter, we use findings from clinical literature as a source of expert knowledge. In addition, expert knowledge has a stabilizing effect on the feature set (an increase in stability of approximately 40%), while the impact on predictive performance is limited.

CONCLUSIONS

The features WHO Performance Status, Albumin, Platelets, Ki-67, Tumor Morphology, Total MTV, Total TLG, and SUVmax are the most stable and predictive features in our study. Overall, this study demonstrated the practical value of feature selection in medical applications not only to improve quantitative performance but also to deliver potentially new insights to experts.

The assessment of Ki-67 for prognosis of gastroenteropancreatic neuroendocrine neoplasm patients: a systematic review and meta-analysis

Background

Neuroendocrine neoplasm (NEN) is a group of rare tumors. Among which, gastroenteropancreatic neuroendocrine neoplasm (GEP-NEN) is the most common group. The World Health Organization (WHO) classified these tumors into three different grades (G1, G2, and G3) based on Ki-67 and mitotic rate, and updated the classification in 2019. Several previous studies proved that Ki-67 was related to tumor prognosis, but others still reported that Ki-67 had no predictive value for tumor prognosis. There are different conclusions between studies regarding the correlation between Ki-67 and tumor prognosis, and there is a lack of studies about this correlation of GEP-NENs. Further analysis is still needed to evaluate the prognostic value of Ki-67 in GEP-NENs, to provide reference for clinical decisions.

Methods

A total of 303 studies were retrieved that included Ki-67, GEP-NENs, prognosis, survival, and other subject terms and keywords. We excluded studies that did not show complete Ki-67 index, number of patients and 5-year survival data available for meta-analysis, non-cohort studies, articles published before 2000 or not published in English. Fifteen studies were finally included to assess the value of Ki-67 in the prognosis of patients with GEP-NENs using a random-effects model.

Results

The cumulative 5-year survival rate for GEP-NEN G1 (Ki-67 ≤2%), G2 (Ki-67 2-20%) and G3 (Ki-67 >20%) was 86%, 65%, 25% respectively. The 5-year survival rate of GEP-NEN G1 (Ki-67 <3%, first revised in WHO classification 2017, redefined WHO classification 2019) and G1 (Ki-67 ≤2%, WHO classification 2010) was 97% and 84% respectively.

Conclusions

The overall prognosis of GEP-NENs patients showed a decreasing trend with the increase of Ki-67, which confirmed the significance of Ki-67 index as a prognostic marker for the prognosis of GEP-NENs. Increasing the cut-off value of Ki-67 index for G1 grade from ≤2% to <3% according to WHO classification 2019 did not significantly decrease the 5-year survival rate.

C-reactive protein independently predicts survival in pancreatic neuroendocrine neoplasms

Pancreatic neuroendocrine neoplasms (pNEN) are highly variable in their postresection survival. Determination of preoperative risk factors is essential for treatment strategies. C-reactive protein (CRP) has been implicated in the pathogenesis of pNEN and shown to be associated with survival in different tumour entities. Patients undergoing surgery for pNEN were retrospectively analysed. Patients were divided into three subgroups according to preoperative CRP serum levels. Clinicopathological features, overall and disease-free survival were assessed. Uni- and multivariable survival analyses were performed. 517 surgically resected pNEN patients were analysed. CRP levels were significantly associated with relevant clinicopathological parameters and prognosis and were able to stratify subgroups with significant and clinically relevant differences in overall and disease-free survival. In univariable sensitivity analyses CRP was confirmed as a prognostic factor for overall survival in subgroups with G2 differentiation, T1/T2 and T3/T4 tumour stages, patients with node positive disease and with and without distant metastases. By multivariable analysis, preoperative CRP was confirmed as an independent predictor of postresection survival together with patient age and the established postoperative pathological predictors grading, T-stage and metastases. Preoperative serum CRP is a strong predictive biomarker for both overall and disease free survival of surgically resected pNEN. CRP is associated with prognosis independently of grading and tumour stage and may be of additional use for treatment decisions.

Pancreatic neuroendocrine tumors G3 and pancreatic neuroendocrine carcinomas: Differences in basic biology and treatment

In 2017 the World Health Organization revised the criteria for classification of pancreatic neuroendocrine neoplasms (pNENs) after a consensus conference at the International Agency for Research on Cancer. The major change in the new classification was to subclassify the original G3 group into well-differentiated pancreatic neuroendocrine tumors G3 (pNETs G3) and poorly differentiated pancreatic neuroendocrine carcinomas (pNECs), which have been gradually proven to be completely different in biological behavior and clinical manifestations in recent years. In 2019 this major change subsequently extended to NENs involving the entire digestive tract. The updated version of the pNENs grading system marks a growing awareness of these heterogeneous tumors. This review discusses the clinicopathological, genetic and therapeutic features of poorly differentiated pNECs and compare them to those of well-differentiated pNETs G3. For pNETs G3 and pNECs (due to their lower incidence), there are still many problems to be investigated. Previous studies under the new grading classification also need to be reinterpreted. This review summarizes the relevant literature from the perspective of the differences between pNETs G3 and pNECs in order to deepen understanding of these diseases and discuss future research directions.

A Preoperative Clinical Risk Score Including C-Reactive Protein Predicts Histological Tumor Characteristics and Patient Survival after Surgery for Sporadic Non-Functional Pancreatic Neuroendocrine Neoplasms: An International Multicenter Cohort Study

Background: Oncological survival after resection of pancreatic neuroendocrine neoplasms (panNEN) is highly variable depending on various factors. Risk stratification with preoperatively available parameters could guide decision-making in multidisciplinary treatment concepts. C-reactive Protein (CRP) is linked to inferior survival in several malignancies. This study assesses CRP within a novel risk score predicting histology and outcome after surgery for sporadic non-functional panNENs. Methods: A retrospective multicenter study with national exploration and international validation. CRP and other factors associated with overall survival (OS) were evaluated by multivariable cox-regression to create a clinical risk score (CRS). Predictive values regarding OS, disease-specific survival (DSS), and recurrence-free survival (RFS) were assessed by time-dependent receiver-operating characteristics. Results: Overall, 364 patients were included. Median CRP was significantly higher in patients >60 years, G3, and large tumors. In multivariable analysis, CRP was the strongest preoperative factor for OS in both cohorts. In the combined cohort, CRP (cut-off ≥0.2 mg/dL; hazard-ratio (HR):3.87), metastases (HR:2.80), and primary tumor size ≥3.0 cm (HR:1.83) showed a significant association with OS. A CRS incorporating these variables was associated with postoperative histological grading, T category, nodal positivity, and 90-day morbidity/mortality. Time-dependent area-under-the-curve at 60 months for OS, DSS, and RFS was 69%, 77%, and 67%, respectively (all p < 0.001), and the inclusion of grading further improved the predictive potential (75%, 84%, and 78%, respectively). Conclusions: CRP is a significant marker of unfavorable oncological characteristics in panNENs. The proposed internationally validated CRS predicts histological features and patient survival.

The Diagnostic Value of Chromogranin A in Neuroendocrine Neoplasms is Potentiated by Clinical Factors and Inflammatory Markers

Objective: Neuroendocrine neoplasms (NENs) are a heterogenous group of indolent tumors, with variable clinical behavior and steadily rising incidence. The aim of this study is to investigate the clinical and laboratory factors that contribute in predicting the aggressiveness and invasiveness of NENs. Special focus is given to clinical parameters that would enhance the diagnostic value of chromogranin A (CgA), via formalizing an integrated probability model, which would contribute to the timely and accurate identification of patients at high risk for metastatic disease at initial diagnosis. Designs and Methods: We identified a total of 93 patients with NENs, recruited at a specialized academic center in Athens, Greece. Anthropometric, clinical, laboratory, and pathological data were obtained from every patient before any therapeutic intervention. Results: Age over 50 years and male gender were accompanied by increased risk for metastases at the time of initial diagnosis. Additionally, when these parameters were combined with CgA levels, they were shown to enhance the predictive capacity of CgA. Different patient scenarios combining age, gender, and CgA levels are associated with different probabilities for metastatic disease, demonstrated schematically in a gradually escalating model, as age and CgA levels increase in both males and females. The lowest risk is observed in women aged <50 years old with CgA levels <200 ng/dl (6.5%), while the highest one is in males over 50 years old with CgA > 200 ng/dl (62.9%). Finally, it was shown that c-reactive protein (CRP) can predict disease extent at the time of diagnosis. Conclusions: CgA levels can not only be used as a direct predictor of tumor load in patients with NENs, but also, when interpolated with the effects of age and gender, cumulatively predict whether a NEN would be metastatic or not at the time of initial diagnosis, via a risk-escalating probability model.