通讯作者: 张林, stepinghuns2@163.com

DOI：10.12201/bmr.202606.00055

A Novel Real-Time Visual Diagnostic Method for Helicobacter pylori Based on Fluorescence ROSE: A Large-Sample Controlled Study Using Latent Class Analysis Without a Gold

• 摘要：目的 本研究旨在假定无金标准的条件下，比较荧光快速现场评估（Fluorescence Rapid On-Site Evaluation, F-ROSE）快速检测法、传统病理学观察法（苏木精 - 伊红染色 / 吉姆萨染色）、尿素呼气试验（Urea Breath Test, UBT）、快速尿素酶试验（Rapid urease test，RUT）及胃镜白光直视判断法对幽门螺旋杆菌 (Helicobacter pylori, Hp) 感染的诊断效能，结合大样本临床实测数据论证 F-ROSE 的诊断优势，以验证F-ROSE的临床应用价值。 方法 前瞻性纳入317例疑似Hp感染患者，在同次就诊周期内采用盲法同步进行5种方法检测：荧光快速现场评估（Fluorescence Rapid On-Site Evaluation, F-ROSE）快速检测法、传统病理学观察法（苏木精 - 伊红染色 / 吉姆萨染色）、尿素呼气试验（Urea Breath Test, UBT）、快速尿素酶试验（Rapid urease test，RUT）及胃镜白光直视判断法。采用无金标准评价框架进行分析：首先通过Cohen’s Kappa系数评估方法间的一致性；其次运用潜在类别分析（LCA）估算各方法的敏感性（Se）、特异性（Sp）及约登指数（J），结合受试者工作特征（ROC）曲线评价诊断效能，并引入贝叶斯敏感性分析验证LCA模型的稳健性。本研究框架无需预设临床金标准，依托多种非完美检测方法的联合检测反应模式挖掘潜在真实患病状态信息，通过统计模型反向推演真实分类情况，实现对各检测方法诊断效能客观、公平的对比评价。 结果经LCA模型估算，该人群潜在患病率约为34.8%。按综合效能（约登指数）排序，五种方法依次为：F-ROSE（0.8672）> RUT（0.8069）> ¹³C-UBT（0.7889）> 病理（0.6076）> 肉眼观察（0.6013）。其中，F-ROSE敏感性最高（0.9472），漏诊率极低；病理检查特异性最高（0.9843），但敏感性偏低（0.6234）。贝叶斯分析证实了LCA结果的高度稳健性。此外，F-ROSE检测耗时仅25–30分钟，显著优于传统病理（24~72小时）。 结论 综合本研究 317 例患者临床检测数据，F-ROSE 在五种 Hp 检测方法中整体诊断效能最佳，胃镜白光直视观察效能相对欠佳。F-ROSE 兼具高敏感性、短检测时长及优异的弱阳性样本识别能力，曲线下面积（AUC）在潜在类别软标签（riLCA ）框架下显著高于病理检查、¹³C 呼气试验、及肉眼观察，与快速尿素酶试验表现接近、差异处于临界显著性水平；同时保持较高特异性，与常规方法一致性良好，临床落地性与实用性突出，有望成为Hp感染的临床快速检测首选手段。

  关键词： 荧光 ROSE 快速检测; 幽门螺旋杆菌; 传统病理学观察法; 尿素呼气试验法; 快速尿素酶试验; 胃镜白光直视判断; 大样本临床验证

   

  Abstract: Objective This study aimed to compare the diagnostic performance of five detection methods for Helicobacter pylori (Hp) infection—fluorescence rapid on‑site evaluation (F‑ROSE), conventional histopathological examination (hematoxylin‑eosin staining/Giemsa staining), urea breath test (UBT), rapid urease test (RUT), and endoscopic white‑light visual assessment—under the assumption of no gold standard. Using large‑sample clinical data, we sought to demonstrate the diagnostic advantages of F‑ROSE and validate its clinical utility. Methods A total of 317 patients with suspected Hp infection were prospectively enrolled. All five tests were performed synchronously in a blinded manner during the same visit: F‑ROSE, conventional histopathology (HE/Giemsa), UBT, RUT, and endoscopic white‑light visual assessment. A no‑gold‑standard evaluation framework was adopted: (1) inter‑method agreement was assessed using Cohen’s kappa coefficient; (2) latent class analysis (LCA) was employed to estimate the sensitivity (Se), specificity (Sp), and Youden index (J) of each method, with diagnostic performance further evaluated by receiver operating characteristic (ROC) curves; and (3) Bayesian sensitivity analysis was introduced to verify the robustness of the LCA model. This framework did not require a prespecified clinical gold standard but instead leveraged the combined response patterns of multiple imperfect tests to infer the latent true infection status, enabling an objective and fair comparative evaluation of diagnostic performance across methods. Results The LCA model estimated a latent prevalence of approximately 34.8% in this cohort. Ranked by overall efficacy (Youden index), the five methods were: F‑ROSE (0.8672) > RUT (0.8069) > ¹³C‑UBT (0.7889) > histopathology (0.6076) > visual inspection (0.6013). F‑ROSE exhibited the highest sensitivity (0.9472), indicating a very low missed‑diagnosis rate, while histopathology showed the highest specificity (0.9843) but relatively low sensitivity (0.6234). Bayesian analysis confirmed the high robustness of the LCA results. In addition, the turnaround time for F‑ROSE was only 25–30 minutes, significantly shorter than that for conventional histopathology (24–72 hours). Conclusion Based on the clinical data from 317 patients in this study, F‑ROSE demonstrated the best overall diagnostic performance among the five Hp detection methods, while endoscopic white‑light visual assessment exhibited relatively poor efficacy. F‑ROSE combines high sensitivity, short turnaround time, and excellent ability to identify weakly positive samples. Its area under the curve (AUC) under the soft‑label latent class (riLCA) framework was significantly superior to that of histopathology, ¹³C‑UBT, and visual inspection, and was close to that of RUT with a marginally significant difference. It also maintained reasonably high specificity and showed good agreement with conventional methods. With strong clinical feasibility and practicality, F‑ROSE holds promise as the preferred rapid testing method for clinical diagnosis of Hp infection.

  Key words: fluorescence ROSE rapid test; Helicobacter pylori; conventional histopathological examination; urea breath test; rapid urease test; endoscopic white‑light visual assessment; large‑sample clinical validation

  提交时间：2026-06-23

  版权声明：作者本人独立拥有该论文的版权，预印本系统仅拥有论文的永久保存权利。任何人未经允许不得重复使用。
• html

• 图表

• 张林, 侯艳红, 李炳慧, 吴凯, 张静, 杨汨. 基于荧光 ROSE 的幽门螺杆菌实时可视化技术在HP除菌治疗全程动态评估中的应用研究. 2026. doi: 10.12201/bmr.202606.00007

  李伟, 陈景旸, 程明. 幽门螺旋杆菌感染对育龄女性铁储备及缺铁性贫血影响的病例对照研究. 2026. doi: 10.12201/bmr.202605.00103

  , , , , , . . 2026. doi: 10.12201/bmr.202605.00040

  顾艳红. HIV合并结核感染诊断方法的研究与评估：一项网状Meta分析. 2024. doi: 10.12201/bmr.202411.00005

• 序号	提交日期	编号	操作
  1	2026-06-16	10.12201/bmr.202606.00055V1	下载

• 公开评论  匿名评论  仅发给作者 提交评论