内镜评估影像学分期为IVA 期的宫颈癌：支持基于证据的分期算法建议的双变量荟萃分析。

Endoscopic assessment of radiological stage IVA cervical cancer: A bivariate meta-analysis supporting an evidence-based staging algorithm proposal.

机构信息

Department of Radiation Oncology, Dan L Duncan Comprehensive Cancer Center, Baylor College of Medicine, Houston, TX, USA.

出版信息

Gynecol Oncol. 2022 Jun;165(3):642-649. doi: 10.1016/j.ygyno.2022.03.026. Epub 2022 Apr 8.

DOI:10.1016/j.ygyno.2022.03.026

PMID:35410732

Abstract

OBJECTIVE

To optimize the use of confirmatory endoscopic exams (cystoscopy/proctoscopy) in the staging of locally advanced cervical cancer (LACC), the present study evaluates the predictive value of radiological exams (CT and MRI) to detect bladder/rectum invasion.

METHODS

A systematic search of databases (PubMed and EMBASE) was performed (CRD42021270329). The inclusion criteria were: a) cervix cancer diagnosis; b) staging CT and/or MRI (index test); c) staging cystoscopy and/or proctoscopy (standard test); and d) numbers of true positives (TP), true negatives (TN), false positives (FP), and false negatives (FN) provided. A random-effects bivariate meta-analysis of positive predictive value (PPV) and negative predictive value (NPV) was performed with moderator analyses by imaging modality (CT and MRI) and prevalence.

RESULTS

Nineteen studies met the inclusion criteria, totaling 3480 and 1641 patients for bladder and rectum analyses, respectively. For bladder invasion (prevalence ranged from 0.9% to 34.5%), the overall PPV was 45% (95% confidence interval, 33%-57%, based on 19 studies). Per subgroup, the PPV was 31% for MRI/prevalence ≤6%, 33% for CT/prevalence ≤6%, and 69% for CT/prevalence >6%. For rectal invasion (prevalence ranged from 0.4% to 20.0%), the overall PPV was 30% (95% confidence interval, 17%-47%, based on 8 studies). Per subgroup, the PPV was 36% for MRI/prevalence ≤1%, 17% for MRI/prevalence >1%, and 38% for CT/prevalence >1%. The overall NPV for bladder invasion and rectal invasion were 98% (95% confidence interval, 97%-99%) and 100% (95% confidence interval, 99%-100%), respectively. Considering prevalence and radiological modality, the point estimate of NPV varied from 95% to 100% for bladder invasion and from 99% to 100% for rectum invasion.

CONCLUSIONS

Due to low PPV (<50%) of radiological staging, endoscopic exams may be necessary to correctly assess radiological stage IVA LACC. However, they are not necessary after negative radiological exam (NPV ≥95%).

摘要

目的

为了优化局部晚期宫颈癌（LACC）分期中确认性内镜检查（膀胱镜/直肠镜检查）的使用，本研究评估了影像学检查（CT 和 MRI）对检测膀胱/直肠侵犯的预测价值。

方法

对数据库（PubMed 和 EMBASE）进行了系统检索（CRD42021270329）。纳入标准为：a）宫颈癌诊断；b）分期 CT 和/或 MRI（指标试验）；c）分期膀胱镜和/或直肠镜检查（标准试验）；d）提供真阳性（TP）、真阴性（TN）、假阳性（FP）和假阴性（FN）的数量。采用随机效应二项式meta 分析对阳性预测值（PPV）和阴性预测值（NPV）进行了分析，并根据影像学检查（CT 和 MRI）和患病率进行了调整。

结果

19 项研究符合纳入标准，膀胱和直肠分析分别纳入 3480 例和 1641 例患者。对于膀胱侵犯（患病率为 0.9%-34.5%），总体 PPV 为 45%（95%置信区间，33%-57%，基于 19 项研究）。按亚组分析，MRI/患病率≤6%时 PPV 为 31%，CT/患病率≤6%时 PPV 为 33%，CT/患病率>6%时 PPV 为 69%。对于直肠侵犯（患病率为 0.4%-20.0%），总体 PPV 为 30%（95%置信区间，17%-47%，基于 8 项研究）。按亚组分析，MRI/患病率≤1%时 PPV 为 36%，MRI/患病率>1%时 PPV 为 17%，CT/患病率>1%时 PPV 为 38%。膀胱侵犯和直肠侵犯的总体 NPV 分别为 98%（95%置信区间，97%-99%）和 100%（95%置信区间，99%-100%）。考虑到患病率和影像学模式，膀胱侵犯的 NPV 点估计值为 95%-100%，直肠侵犯的 NPV 点估计值为 99%-100%。