第一部分 超声引导下细针抽吸活检对于甲状腺乳头状微小癌的诊断效能

摘要

目的：比较细胞学、细胞块及二者联合对于甲状腺乳头状微小癌（papillary thyroid microcarcinoma, PTMC）的诊断效能。

方法：纳入2020年6月至2023年6月于北京协和医院行甲状腺微小结节FNA，以获取细胞学及细胞块进行穿刺病理诊断，并于北京协和医院接受手术治疗患者，共纳入结节154例（PTMC 125例，良性病变29例）。首先，计算术前细胞学及细胞块各诊断类型分布差异。其次，将细胞学、细胞块及二者联合诊断分别与金标准进行一致性检验分析。此外，分别计算细胞学、细胞块及二者联合在PTMC中的诊断效能，并用DeLong检验比较三种诊断方法曲线下面积（Area Under the Curve, AUC）的统计学差异。

结果：细胞学与细胞块诊断一致102例（102/154，66.23%），二者诊断一致且符合术后病理94例（94/102，92.16%）。细胞学、细胞块及二者联合与金标准均基本一致（Kappa值分别为0.639、0.710、0.795）。对于三种诊断方法的敏感度、特异度、阳性预测值及AUC，细胞学分别为88.80%、82.76%、99.14%及0.858，细胞块分别为93.60%、79.31%、95.12%及0.865，二者联合分别为98.40%、75.86%、94.62%及0.871，三种诊断方法AUC比较，差异均无统计学意义（P ＞ 0.05）。

结论：细胞学、细胞块及二者联合诊断对于PTMC均有较好的诊断效能，三种诊断方法诊断效能差异没有显著统计学意义。

第二部分 基于灰阶及血流超声特征建立列线图鉴别甲状腺滤泡性良恶性肿瘤

摘要

目的：旨在基于灰阶及超声血流特征建立列线图以鉴别甲状腺滤泡癌(follicular thyroid carcinoma, FTC)和滤泡性良性肿瘤(follicular benign neoplasms, FBN) 。

方法：本研究共纳入153例患者。首先，评估所有结节各种超声特征，并分析不同血流模式对于FTC的诊断效能。其次，使用多因素逻辑回归筛选FTC的独立预测因子。最后，基于独立预测因子建立列线图以预测FTC，并验证该列线图的诊断效能。

结果：FTC多呈现“烟花征”和“中心聚集征”血流模式（P<0.001， P=0.007）， 而FBN多出现“环征”血流模式（P<0.001）。相比于其他血流模式，“烟花征”血流模式的准确率（71.2%）和阴性预测值（73.4%）最高。此外，低回声厚晕、“烟花征”、“中心聚集征”和“环征”血流模式是构建列线图的四个预测因子。bootstrap 曲线下面积为0.814 (95% CI: 0.743 ~ 0.884)，表明该列线图具有较好的预测精度。

结论：“烟花征”和“中心聚集征”为FTC的血流特征，而“环征”为FBN的血流特征。基于灰阶及血流特征建立的列线图有助于术前鉴别FTC和FBN。

Part One The diagnostic efficiency of ultrasound-guidance fine-needle aspiration for papillary thyroid microcarcinoma

Abstract

Objective: To compare cytology, cell block and their combination in the diagnosis of papillary thyroid microcarcinoma (PTMC).

Methods: Between June 2020 and June 2023, a total of 154 patients (PTMC/Benign thyroid micro-lesion = 125/29) confirmed by postoperative pathology at our center (Peking Union Medical College Hospital) were included. All the patients underwent ultrasound-guidance FNA and obtained both cytology and cell block samples. First, preoperative cytology and cell block distribution were calculated. Next, to test whether cytology, cell block and combined diagnosis were consistent with the gold standard. Besides, the diagnostic efficiency of cytology, cell block and their combination in PTMC was calculated, and the statistical difference of the area under the curve (AUC) between the three diagnostic methods was compared by DeLong test. 

Results: Cytology and cell block diagnosis were consistent in 102 cases (102/154, 66.23%) and 94 cases (94/102, 92.16%) were consistent with postoperative pathological diagnosis. Cytology, cell blocks and their combinations were basically consistent with the gold standard (Kappa values were 0.639, 0.710 and 0.795, respectively). The sensitivity, specificity, positive predictive value and AUC of the three diagnostic methods were 88.80%, 82.76%, 99.14% and 0.858 in cytology, 93.60%, 79.31%, 95.12% and 0.865 in cell block, and 98.40%, 75.86%, 94.62% and 0.871 in the combination of cytology and cell block, respectively. There was no significant difference among the three methods (P > 0.05).

Conclusion: Although cytology, cell block and their combination showed good performance in diagnosing PTMC, there was no statistically significant difference among the three diagnostic methods.

Key words: Cell block; Thyroid papillary microcarcinoma; Cytology; Diagnosis; Fine-needle aspiration

Part Two Establish a nomogram based on gray-scale and vascularity ultrasound characteristics to differentiate

follicular thyroid carcinoma from follicular benign neoplasms

Abstract

Purpose: This study aimed to construct a nomogram based on gray-scale and vascularity ultrasound features to differentiate follicular thyroid carcinoma (FTC) from follicular benign neoplasms (FBN) .

Methods: A total of 153 patients were included. We evaluated various sonographic features of all nodules and analyzed diagnostic efficiency of different vascularity patterns for FTC. Furthermore, independent predictors for FTC were screened using multivariable logistic regression. A nomogram was established to predict FTC and various methods were used to evaluate the performance of the nomogram.

Results: The firework-like pattern and central-grasp pattern were more likely to appear in the FTC (P<0.001, P=0.007), while the ring-like pattern were more likely to appear in the FBN (P<0.001). The firework-like pattern showed the best performance in accuracy (71.2%) and NPV (73.4%) among other vascularity patterns. Besides, thick hypo-halo, central-grasp pattern and firework-like pattern were four predictors used to construct the nomogram. The bootstrap AUC was 0.814 (95% CI: 0.743-0.884) which revealed the nomogram had an accurate prediction accuracy.

Conclusion: The firework-like pattern and central-grasp pattern were vascularity characteristics for FTC, while the ring-like pattern was for FBN. Besides, our nomogram based on gray-scale and novel vascularity characteristics can help distinguish FTC and FBN before operation.

Keywords: follicular thyroid carcinoma, follicular benign neoplasms, ultrasound, vascularity pattern; nomogram