AUTHOR=Jin Lingli , Zheng Danni , Mo Danni , Guan Yaoyao , Wen Jialiang , Zhang Xiaohua , Chen Chengze 

TITLE=Glucose-to-Lymphocyte Ratio (GLR) as a Predictor of Preoperative Central Lymph Node Metastasis in Papillary Thyroid Cancer Patients With Type 2 Diabetes Mellitus and Construction of the Nomogram

JOURNAL=Frontiers in Endocrinology

VOLUME=Volume 13 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/endocrinology/articles/10.3389/fendo.2022.829009

DOI=10.3389/fendo.2022.829009

ISSN=1664-2392

ABSTRACT=Abstract
Background：The central lymph node metastasis (CLNM) of papillary thyroid cancer (PTC) is difficult to detect before surgery, and the role of routine or preventive central lymph node dissection (CLND) in the management of PTC is still controversial. Moreover, glucose metabolism and systemic inflammation are related to the cancer aggressiveness of various malignant tumors and the prognosis of patients. This article aims to create a nomogram that can predict the occurrence of central lymph node metastasis in patients with thyroid cancer and diabetes using ready-made preoperative clinical features.
Methods：419 patients were enrolled. We used the ROC curve to get the best cutoff value and convert continuous variables into categorical variables. Immediately afterwards, we performed a single-factor logistic analysis of the independent variables, and further performed a multivariate regression analysis on the selected significant risk factors. Ensue a nomogram chart was drawn. The nomogram was verified by external data, and the existing data was compared with the original model.
Result：According to ROC curve analysis, the best cutoff values for GLR and tumor size were 4.23 and 0.95 (cm), respectively. Multivariate logistic regression analysis results hint that age, bilateral tumors, maximum tumor size, and the ratio of glucose to lymphocytes (GLR) were independent risk factors for CLNM. The C-index in the training data set was 0.733 and the external validation data set was 0.664, respectively. Both the calibration curve and the Hosmer-Lemeshow test reflected that the model was well calibrated. Through DCA, it is estimated that the predictive model had strong clinical applicability and more benefits. In order to compare the performance of the new model we built with the original model, we conducted the analysis of the NRI  and the IDI, both of which indicated that the new model had better predictive capabilities.
Conclusion：In patients with T2DM and PTC, high preoperative GLR level was an independent predictor of CLNM. The GLR cutoff value for this article was 4.23. This study aims to build a model to better predict the possibility of CLNM in patients with diabetes and PTC, so as to help clinicians determine appropriate treatment strategies for patients.