Supplementary Material for: Systematic Review and Meta-Analysis of the Diagnostic Accuracy of the International Academy of Cytology Yokohama System for Reporting Breast Fine-Needle Aspiration Biopsy in Diagnosing Breast Cancer

<b><i>Objectives:</i></b> The primary objective is to determine the accuracy of fine-needle aspiration biopsy (FNAB) in breast lesions reported according to the International Academy of Cytology (IAC) Yokohama system for reporting breast FNAB. The participants include any patient presenting with any breast lesion found suitable for FNAB. The target condition was breast cancer. The secondary objective was to study the proportion of inadequate FNAB in the selected studies. <b><i>Methods:</i></b> PubMed/MEDLINE and Embase were searched for studies having all the following key search terms: Breast AND FNAB AND Diagnostic Accuracy published in the time frame of 2017 to May 16, 2022. The Cochrane and PROSPERO databases, citations of selected articles and articles citing the selected articles were also searched. Studies assessing the diagnostic accuracy of breast FNAB in diagnosing breast cancer, which had at least 75 subjects (and at least 20 subjects each in the benign and malignant FNAB groups), were selected. The reference standard was histopathology (or adequate clinical follow-up for benign disease). Studies were screened independently by two researchers, with a consensus reached among the authors in cases of conflict. The risk of bias and applicability were assessed using the QUADAS-2 tool. Sensitivity and specificity at each diagnostic cut-off were assessed by bivariate generalized linear mixed-model meta-analysis. The area under the receiver operating characteristics curve (AUC) and inadequacy rate were assessed by random-effects meta-analysis. The confidence intervals of sensitivity, specificity, and AUC were examined against a value of 0.95. <b><i>Results:</i></b> Twenty-two studies, all of which were cross-sectional single-gate studies, were selected with a total of 10,886 subjects with a primary breast lesion having concurrent FNAB and reference standard reports. Sensitivity and specificity, with 95% confidence intervals, were 0.978 [0.968, 0.985] and 0.832 [0.76, 0.886] for the diagnostic cut-off of “Atypical considered positive for malignancy,” 0.916 [0.892, 0.935] and 0.983 [0.97, 0.99] for the cut-off of “Suspicious of Malignancy considered positive,” and 0.763 [0.706, 0.812] and 0.999 [0.994, 1] for the cut-off of “Malignant considered positive.” The overall AUC was 0.975 [0.962, 0.984]. FNAB sampling without imaging guidance was associated with lower inadequacy. <b><i>Discussion:</i></b> There is strong evidence that the overall accuracy, sensitivity for “Atypical category considered positive” and specificity when “Suspicious or Malignant categories are considered positive” of FNAB are high when using the categories of the IAC Yokohama Reporting System, demonstrating the usefulness of FNAB in diagnosing breast cancer.

**研究目的**：本研究首要目标为明确按照国际细胞学学会（International Academy of Cytology, IAC）横滨乳腺细针穿刺活检报告系统进行报告的乳腺病变细针吸取活检（fine-needle aspiration biopsy, FNAB）的诊断准确性。研究对象为所有存在适合行FNAB的乳腺病变的就诊患者，目标疾病为乳腺癌。次要研究目的为分析纳入研究中FNAB标本不合格的比例。

**研究方法**：检索PubMed/MEDLINE及Embase数据库，检索时限为2017年至2022年5月16日，检索关键词包含「Breast AND FNAB AND Diagnostic Accuracy」。同时检索Cochrane、PROSPERO数据库，以及纳入文献的参考文献与施引文献。纳入标准为：评估乳腺FNAB诊断乳腺癌的诊断准确性、受试者总例数至少75例（且良性、恶性FNAB组各至少20例）的研究。本研究的参考标准为组织病理学（或良性病变的充分临床随访）。由2名研究者独立完成文献筛选，意见分歧时由作者团队协商达成一致。采用诊断准确性研究质量评估工具2（Quality Assessment of Diagnostic Accuracy Studies-2, QUADAS-2）评估偏倚风险与研究适用性。采用双变量广义线性混合模型Meta分析，评估各诊断截断值下的灵敏度与特异度；采用随机效应模型Meta分析，评估受试者工作特征曲线下面积（area under the receiver operating characteristics curve, AUC）与标本不合格率。以0.95为置信水准，分析灵敏度、特异度及AUC的置信区间。

**研究结果**：共纳入22项横断面单中心研究，合计纳入10886例同时具备FNAB结果与参考标准报告的原发性乳腺病变患者。以「非典型性结果判定为恶性阳性」为诊断截断值时，灵敏度与特异度（95%置信区间）分别为0.978[0.968, 0.985]与0.832[0.76, 0.886]；以「高度怀疑恶性判定为阳性」为截断值时，对应数值分别为0.916[0.892, 0.935]与0.983[0.97, 0.99]；以「恶性结果判定为阳性」为截断值时，对应数值分别为0.763[0.706, 0.812]与0.999[0.994, 1]。总体AUC为0.975[0.962, 0.984]。未采用影像学引导的FNAB采样标本不合格率更低。

**讨论**：现有强有力证据表明，采用国际细胞学学会横滨报告系统的分类标准时，FNAB的总体准确性、「非典型性类别判定为阳性」时的灵敏度，以及「高度怀疑或恶性类别判定为阳性」时的特异度均处于较高水平，证实了FNAB在乳腺癌诊断中的应用价值。