Influencing Factors and Prediction Model of False-negative Central Lymph Nodes in Thyroid Cancer Patients

Overview

The problem of false-negative lymph nodes in the central region (CLN) refers to the inability to detect lymph nodes (LN) metastasis during preoperative ultrasound examination but is confirmed by surgical and postoperative pathological examination. According to this study, the false-negative rate of CLN of patients with TC was relatively high at approximately 71.7%. The high proportion of false-negatives indicates limitations in the sensitivity of ultrasound examination, especially for detecting small LN or small metastases. The presence of false-negative CLN directly affects clinical practice. First, it may lead to insufficient selection of treatment strategies for patients with TC, thereby affecting their prognosis. Second, such false-negative results may lead surgeons to overly trust ultrasound examination during preoperative evaluation, neglecting the importance of comprehensive clinical information, and thus affecting the formulation of treatment plans. This study aimed to explore the factors influencing false-negative CLN through a cohort study and to establish a predictive model.

Setting and Population This study collected 6369 patients admitted to the Breast and Thyroid Surgery Department of Shandong Provincial Hospital between January 2013 and December 2023, who were diagnosed with TC through postoperative pathology. (SDC, Figure 1) All patients included in the study signed a written informed consent form before surgery. Ethical approval for this study (NO.SWYX2024-225) was provided by the Ethical Committee of Shandong Provincial Hospital, Jinan, China on 15 April 2024. According to the relevant literature, the criteria for identifying abnormal cervical LN under ultrasound are: disappearance of the LN hilum structure or abnormalities in the skin and medulla, LN becoming round or having an increased aspect ratio, microcalcifications in the LN, cystic changes present in LN, and abnormal blood flow signals in LN.17 The inclusion and exclusion criteria are: all cases are newly diagnosed TC cases; all cases underwent preoperative thyroid and neck LN examinations by the Ultrasound Department of Shandong Provincial Hospital; all cases underwent thyroid surgery for the first time; all cases were diagnosed with TC through postoperative pathological examination; excluding cases of other combined tumors or major diseases. This study used the postoperative paraffin pathology results as the gold standard for diagnostic testing. All ultrasound examinations were performed by two ultrasound physicians. When two ultrasound physicians had different opinions, a third senior ultrasound physician was requested to assist in the evaluation. Data Collection After obtaining written consent, we recorded the patient\'s relevant information in detail and followed up on the patient\'s postoperative paraffin pathology results. The collected and organized items included: 1) sex, age, and past medical history; 2) preoperative ultrasound information; 3) preoperative fine-needle biopsy information; 4) surgical information; 5) postoperative pathological information; and 6) immunohistochemical information. Statistical Analysis This study first analyzed the baseline data of the included patients and obtained descriptive data on the current status of patients with TC. (Table 1) This study also analyzed the efficacy of ultrasound and fine-needle aspiration (FNA) for the evaluation of cervical LN. (Table 2) Subsequently, analysis was conducted on 52 factors that may cause false-negative CLN of TC. (Table 3) A total of 16 statistically significant influencing factors were identified. (Table 4) To construct a predictive model for the occurrence of false-negative CLN of the thyroid gland, we selected five preoperative influencing factors with predictive significance from eight preoperative influencing factors and drew a forest chart for data visualization and outcome prediction. This study was analyzed using IBM SPSS 29.0 software. Quantitative data are represented by mean ± standard deviation when it follows a normal distribution and homogeneity of variance, and an independent sample t-test is used for comparison between the two groups; the median (P25, P75) is used to represent non-normal distributions, and a non-parametric Mann-Whitney test is used for comparison between the two groups. Qualitative data were expressed as percentages (%), and intergroup comparisons were conducted using the chi-square or Fisher\'s exact test. Variables with statistical significance in the univariate analysis and those professionally considered to have an impact on the outcome were included in the logistic regression model to explore the independent factors influencing the outcome. The test level P was set at 0.05. A forest map was drawn using Graphpad Prism 10.12 software, and a column chart and credibility analysis were drawn using R Studio 4.3 software.