Limitations and suggestions for type 2 diabetes mellitus and pancreatic neuroendocrine tumors based on meta-analysis

Abstract

We read with great interest the systematic review and meta-analysis by Cigrovski Berkovic et al published recently, which evaluated the association between type 2 diabetes mellitus (T2DM) and pancreatic neuroendocrine tumors (pNETs). The study identified T2DM as a risk factor for the development of pNETs and linked it to poor tumor-free survival. However, due to the limited number of studies and high heterogeneity, the role of metformin in the diagnosis and prognosis of pNETs remained inconclusive. We believe the study has some limitations regarding literature search, risk of bias assessment, and analysis of heterogeneity and publication bias. Expanding the search to more databases, applying more appropriate bias assessment tools, and using better statistical methods to evaluate heterogeneity and publication bias would strengthen the study’s conclusions. Addressing these concerns could provide more robust evidence for understanding the diagnostic and prognostic impact of T2DM in pNETs.

Key Words: Pancreatic neuroendocrine tumors; Type 2 diabetes mellitus; Tumor-free survival; Letter; Systematic review and meta-analysis

Core Tip: In the study, we can see in the type 2 diabetes mellitus results and pooled analyses, the authors said that “the low heterogeneity was graphically observable, also looking at the Galbraith and Funnel plots (Figure 3A and B)”. This is mistake and funnel plot should only be used to check for the publication bias. And in order to test for publication bias for this study, quantitative Begg’s test and Egger test were recommended, because the number of inclusion studies are less than 10.

TO THE EDITOR

We read with great interest the article by Cigrovski Berkovic et al[1] published recently. This systematic review and meta-analysis aimed to verify the diagnostic and prognostic significance of type 2 diabetes mellitus (T2DM) and metformin in pancreatic neuroendocrine tumors (pNETs). The study discovered T2DM as a risk factor for the onset of pNETs and an important predictor of poor prognosis (tumor-free survival) of pNETs. However, a limited number of heterogeneous studies hindered the exploration of metformin’s diagnostic and prognostic values in pNETs. We aimed to highlight some of these shortcomings of the aforementioned study.

First, we found that only the PubMed database was searched in the study by Cigrovski Berkovic et al[1]. We believe that other databases, such as Web of Science, Cochrane Library, EMBASE, MEDLINE and Scopus, should also be searched to ensure the inclusion of other studies meeting the eligibility criteria. Second, regarding the risk-of-bias assessment of nonrandomized studies, the Risk of Bias in Non-randomized Studies of Interventions is recommended instead of the Newcastle-Ottawa Scale[2]. There did not exist the certainty of evidence assessment in this study. Based on the Grades of Recommendations Assessment, Development, and Evaluation to assess the quality of evidence and the strength of recommendations[3]. Overall strength of evidence was divided into very low, low, moderate, and high levels[3]. Finally, as evident in the T2DM results and pooled analyses, the authors stated that “the low heterogeneity was graphically observable, also looking at the Galbraith and Funnel plots (Figure 3A and 3B)”. However, this is inappropriate, as a funnel plot should only be used to detect publication bias. To evaluate heterogeneity of studies, in quantitative method, the I² statistic and Cochran’s Q statistic were utilized. Also, quantitative methods such as Begg’s and Egger’s tests were recommended to assess publication bias in this meta-analysis because the number of publications included is less than 10[4]. In summary, this study found that T2DM is a risk factor for pNETs. Focusing on modifiable disease risk factors and susceptible populations is essential to develop strategies for early detection, prevention and management pNETs. We appreciate the authors’ efforts in demonstrating the diagnostic and prognostic values of T2DM in pNETs. If the aforementioned limitations are addressed, more convincing conclusions can be drawn.