常规超声特征鉴别胆囊肿瘤性与非肿瘤性息肉样病变的临床价值

Clinical Value of Conventional Ultrasound Features for Differentiating Neoplastic from Non-Neoplastic Gallbladder Polypoid Lesions

摘要: 目的：探讨常规超声特征鉴别胆囊肿瘤性与非肿瘤性息肉样病变的临床价值，比较纯超声模型与临床–超声联合模型的判别效能，并进行内部验证。方法：回顾性纳入合肥市第二人民医院2023年1月至2025年12月经术前常规超声检查并获手术病理证实的胆囊息肉样病变患者236例。按病理结果分为非肿瘤性组175例和肿瘤性组61例。记录临床资料、二维灰阶及彩色多普勒超声特征，以病理诊断为参照标准，采用单因素分析、多因素Logistic回归和受试者工作特征曲线分析。以年龄、性别构建临床模型，以7项核心超声征象构建纯超声模型，并进一步构建临床–超声联合模型；采用Bootstrap重抽样1000次进行内部验证，报告乐观度校正后的AUC及95% CI。结果：两组基线资料及主要超声特征差异均有统计学意义(P < 0.05)。纯超声模型中，病灶最大径 ≥ 10 mm、广基底/无蒂、CDFI可见血流信号、邻近胆囊壁增厚、高回声点/彗星尾缺如、单发病灶及合并胆泥/结石均与肿瘤性病变独立相关(P < 0.05)。纯超声模型的表观AUC为0.902 (95% CI: 0.858~0.946)，Bootstrap校正AUC为0.887 (95% CI: 0.842~0.935)；临床模型和临床–超声联合模型的校正AUC分别为0.842 (95% CI: 0.793~0.893)和0.907 (95% CI: 0.873~0.943)。联合模型较纯超声模型的表观AUC增加0.025，差异无统计学意义(DeLong检验P = 0.064)。结论：常规超声征象结构化整合具有较好的判别能力，Bootstrap校正后仍保持稳定。年龄和性别在纯超声模型基础上仅带来小幅增量，其差异未达到统计学意义。由于本研究仅纳入手术病理病例，结果不宜直接外推至无症状筛查或非手术随访人群。

Abstract: Objective: To investigate the clinical value of conventional ultrasound features for differentiating neoplastic from non-neoplastic gallbladder polypoid lesions, compare an ultrasound-only model with a clinical-ultrasound model, and perform internal validation. Methods: A total of 236 patients with gallbladder polypoid lesions who underwent preoperative conventional ultrasound and had postoperative pathological confirmation at Hefei Second People’s Hospital from January 2023 to December 2025 were retrospectively enrolled. According to pathological results, 175 lesions were classified as non-neoplastic and 61 as neoplastic. Clinical data, two-dimensional grayscale findings, and color Doppler flow imaging features were collected. Pathology was used as the reference standard. Univariate analysis, multivariable logistic regression, and receiver operating characteristic analysis were performed. A clinical model was constructed using age and sex, an ultrasound-only model using seven core sonographic findings, and a combined clinical-ultrasound model using both sets of variables. Internal validation was performed with 1000 bootstrap resamples, and optimism-corrected AUCs with 95% CIs were reported. Results: Baseline variables and key ultrasound features differed significantly between the two groups (P < 0.05). In the ultrasound-only model, lesion size ≥10 mm, sessile morphology, Doppler vascularity, adjacent gallbladder wall thickening, absence of hyperechoic spots/comet-tail artifact, solitary lesion, and concomitant sludge/stones were independently associated with neoplastic lesions (P < 0.05). The apparent AUC of the ultrasound-only model was 0.902 (95% CI: 0.858~0.946), and the bootstrap-corrected AUC was 0.887 (95% CI: 0.842~0.935). The corrected AUCs of the clinical model and the combined clinical-ultrasound model were 0.842 (95% CI: 0.793~0.893) and 0.907 (95% CI: 0.873~0.943), respectively. The apparent AUC increased by 0.025 after adding age and sex, but the difference was not statistically significant (DeLong P = 0.064). Conclusion: Structured integration of conventional ultrasound findings showed good discrimination and remained stable after bootstrap correction. Age and sex provided only a small, non-significant incremental value beyond the ultrasound-only model. Because only surgically treated patients with pathological confirmation were included, direct extrapolation to asymptomatic screening or non-surgical surveillance populations is inappropriate.