文献速递妇科产科

利用捕获细胞基因识别算法评估盆腔肿块女性的恶性肿瘤

Malignancy Assessment Using Gene Identification in Captured Cells Algorithm for the Prediction of Malignancy in Women With a Pelvic Mass.
PubMed 文献 · Obstet Gynecol · 2022-09-07

这项IRB批准的前瞻性临床研究纳入183例盆腔肿块患者,术前用微流控装置从外周血捕获稀有细胞并分析72个基因转录本表达,结合血清生物标志物构建MAGIC算法。结果显示,8个基因联合4个血清生物标志物的多变量模型区分良恶性(排除低度恶性潜能肿瘤)的AUC最高,达95.1%(95% CI 92.0-98.2%),显著优于所有单个基因及14个基因联合模型。

要点

  • MAGIC算法(8基因+4血清生物标志物)鉴别盆腔肿块良恶性的AUC为95.1%(95% CI 92.0-98.2%)。
  • 该算法显著优于所有单个基因(AUC 50.2-65.2%,均P<.001)及14基因联合模型(AUC 88.0%,P=.005)。
  • 上皮性卵巢癌I-II期AUC为89.5%(95% CI 81.3-97.8%),III-IV期为98.9%(95% CI 96.7-100%)。

核心结论

  • 对从血液中捕获的细胞进行多重基因表达评估,联合或不联合血清生物标志物水平,可准确检测盆腔肿块女性的恶性肿瘤。

研究设计与样本

  • 本研究为IRB批准的前瞻性临床研究,纳入183例可评估的盆腔肿块患者(平均年龄56岁,范围19-91岁),术前采集外周血并用微流控细胞捕获装置捕获稀有细胞。
  • 从捕获细胞中分离mRNA,分析72个不同基因转录本的表达,同时检测多种常用血清生物标志物水平,所有患者均有组织学诊断。
  • 采用单变量和多变量逻辑回归分析基因表达及血清生物标志物水平对恶性肿瘤的预测价值,构建并比较受试者工作特征曲线。

主要发现

  • 183例患者中,良性肿瘤104例、低度恶性潜能肿瘤17例、恶性肿瘤62例。
  • 区分良性病变(排除低度恶性潜能肿瘤)与恶性肿瘤时,8个基因联合4个血清生物标志物的多变量模型AUC最高,为95.1%(95% CI 92.0-98.2%)。
  • MAGIC算法显著优于所有单个基因(AUC 50.2-65.2%,均P<.001)及14个不同基因联合的多变量模型(AUC 88.0%,95% CI 82.9-93.0%,P=.005)。
  • 上皮性卵巢癌患者中,MAGIC算法对I-II期的AUC为89.5%(95% CI 81.3-97.8%),对III-IV期的AUC为98.9%(95% CI 96.7-100%)。

关键数值与推荐

  • MAGIC算法区分良性病变(排除低度恶性潜能肿瘤)与恶性肿瘤的AUC为95.1%(95% CI 92.0-98.2%)。
  • MAGIC算法对上皮性卵巢癌I-II期的AUC为89.5%(95% CI 81.3-97.8%),对III-IV期的AUC为98.9%(95% CI 96.7-100%)。

局限性

  • 原文未报告局限性。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

OBJECTIVE: To evaluate the detection of malignancy in women with a pelvic mass by using multiplexed gene expression analysis of cells captured from peripheral blood. METHODS: This was an IRB-approved, prospective clinical study. Eligible patients had a pelvic mass and were scheduled for surgery or biopsy. Rare cells were captured from peripheral blood obtained preoperatively by using a microfluidic cell capture device. Isolated mRNA from the captured cells was analyzed for expression of 72 different gene transcripts. Serum levels for several commonly assayed biomarkers were measured. All patients had a tissue diagnosis. Univariate and multivariate logistic regression analyses for the prediction of malignancy using gene expression and serum biomarker levels were performed, and receiver operating characteristic curves were constructed and compared. RESULTS: A total of 183 evaluable patients were enrolled (average age 56 years, range 19-91 years). There were 104 benign tumors, 17 low malignant potential tumors, and 62 malignant tumors. Comparison of the area under the receiver operating characteristic curve for individual genes and various combinations of genes with or without serum biomarkers to differentiate between benign conditions (excluding low malignant potential tumors) and malignant tumors showed that a multivariate model combining the expression levels of eight genes and four serum biomarkers achieved the highest area under the curve (AUC) (95.1%, 95% CI 92.0-98.2%). The MAGIC (Malignancy Assessment using Gene Identification in Captured Cells) algorithm significantly outperformed all individual genes (AUC 50.2-65.2%; all P <.001) and a multivariate model combining 14 different genes (AUC 88.0%, 95% CI 82.9-93.0%; P =.005). Further, the MAGIC algorithm achieved an AUC of 89.5% (95% CI 81.3-97.8%) for stage I-II and 98.9% (95% CI 96.7-100%) for stage III-IV patients with epithelial ovarian cancer. CONCLUSION: Multiplexed gene expression evaluation of cells captured from blood, with or without serum biomarker levels, accurately detects malignancy in women with a pelvic mass. CLINICAL TRIAL REGISTRATION: ClinicalTrials.gov, NCT02781272. FUNDING SOURCE: This study was funded by ANGLE Europe Limited (Surrey Research Park, Guildford, Surrey, United Kingdom).

来源:Obstet Gynecol,查看原文。版权归原作者所有。

引用

  1. Moore RG, Khazan N, Coulter MA, et al. Malignancy Assessment Using Gene Identification in Captured Cells Algorithm for the Prediction of Malignancy in Women With a Pelvic Mass. Obstet Gynecol. 2022. PMID: 36075062. doi:10.1097/AOG.0000000000004927

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。