使用MasSpec笔在人类内分泌手术中术中识别甲状腺和甲状旁腺组织
这项诊断/预后研究在实验室和手术室中使用MasSpec笔(MSPen)分析184份新鲜冷冻的甲状腺、甲状旁腺和淋巴结组织,并对102例接受甲状腺切除术和甲状旁腺切除术的患者进行术中分析。MSPen数据区分甲状腺与甲状旁腺组织的总体准确率为92.4%(95%CI,87.7-95.4),独立测试集准确率为82.6%(95%CI,76.5-87.4);区分甲状腺与淋巴结、甲状旁腺与淋巴结的训练集准确率分别为97.5%(95%CI,92.8-99.1)和96.1%(95%CI,91.2-98.3)。作者认为该技术可能有助于提供近实时组织类型反馈以辅助手术决策。
要点
- MSPen区分甲状腺与甲状旁腺组织的总体准确率为92.4%,独立测试集为82.6%。
- MSPen区分甲状腺与淋巴结、甲状旁腺与淋巴结的训练集准确率分别为97.5%和96.1%。
- 作者认为MSPen可能有助于提供近实时组织类型反馈以辅助手术决策。
核心结论
- MSPen在术中区分甲状腺、甲状旁腺和淋巴结组织方面表现出高性能,作者认为该技术可能有助于提供近实时组织类型反馈以辅助手术决策。
研究设计与样本
- 这项诊断/预后研究于2017年8月至2020年3月在学术医疗中心的手术室进行,使用MSPen分析184份新鲜冷冻的甲状腺、甲状旁腺和淋巴结组织,并对102例接受甲状腺切除术和甲状旁腺切除术的患者进行术中分析。
- 184份新鲜冷冻组织中131份因信号充足和分析了病理诊断而被纳入;临床测试纳入102例手术患者,共进行1015次术中分析,其中269次分析接受统计分类。
- 统计分类器使用包含实验室和术中数据的训练集生成,并在术中数据的独立测试集上评估;数据分析时间为2022年7月至12月。
主要发现
- 102例术中研究患者中80例为女性(78%),中位(IQR)年龄为52(42-66)岁。
- MSPen数据区分甲状腺与甲状旁腺组织的总体准确率(定义为与病理学一致)为92.4%(95%CI,87.7-95.4),独立测试集准确率为82.6%(95%CI,76.5-87.4)。
- 区分甲状腺与淋巴结、甲状旁腺与淋巴结的训练集总体准确率分别为97.5%(95%CI,92.8-99.1)和96.1%(95%CI,91.2-98.3)。
关键数值与推荐
- MSPen区分甲状腺与甲状旁腺组织的总体准确率为92.4%(95%CI,87.7-95.4),独立测试集准确率为82.6%(95%CI,76.5-87.4)。
- MSPen区分甲状腺与淋巴结的训练集准确率为97.5%(95%CI,92.8-99.1),区分甲状旁腺与淋巴结的训练集准确率为96.1%(95%CI,91.2-98.3)。
临床怎么用
- 作者认为MSPen技术可能有助于提供近实时组织类型反馈以辅助手术决策。
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
IMPORTANCE: Intraoperative identification of tissues through gross inspection during thyroid and parathyroid surgery is challenging yet essential for preserving healthy tissue and improving outcomes for patients. OBJECTIVE: To evaluate the performance and clinical applicability of the MasSpec Pen (MSPen) technology for discriminating thyroid, parathyroid, and lymph node tissues intraoperatively. DESIGN, SETTING, AND PARTICIPANTS: In this diagnostic/prognostic study, the MSPen was used to analyze 184 fresh-frozen thyroid, parathyroid, and lymph node tissues in the laboratory and translated to the operating room to enable in vivo and ex vivo tissue analysis by endocrine surgeons in 102 patients undergoing thyroidectomy and parathyroidectomy procedures. This diagnostic study was conducted between August 2017 and March 2020. Fresh-frozen tissues were analyzed in a laboratory. Clinical analyses occurred in an operating room at an academic medical center. Of the analyses performed on 184 fresh-frozen tissues, 131 were included based on sufficient signal and postanalysis pathologic diagnosis. From clinical tests, 102 patients undergoing surgery were included. A total of 1015 intraoperative analyses were performed, with 269 analyses subject to statistical classification. Statistical classifiers for discriminating thyroid, parathyroid, and lymph node tissues were generated using training sets comprising both laboratory and intraoperative data and evaluated on an independent test set of intraoperative data. Data were analyzed from July to December 2022. MAIN OUTCOMES AND MEASURES: Accuracy for each tissue type was measured for classification models discriminating thyroid, parathyroid, and lymph node tissues using MSPen data compared to gross analysis and final pathology results. RESULTS: Of the 102 patients in the intraoperative study, 80 were female (78%) and the median (IQR) age was 52 (42-66) years. For discriminating thyroid and parathyroid tissues, an overall accuracy, defined as agreement with pathology, of 92.4% (95% CI, 87.7-95.4) was achieved using MSPen data, with 82.6% (95% CI, 76.5-87.4) accuracy achieved for the independent test set. For distinguishing thyroid from lymph node and parathyroid from lymph node, overall training set accuracies of 97.5% (95% CI, 92.8-99.1) and 96.1% (95% CI, 91.2-98.3), respectively, were achieved. CONCLUSIONS AND RELEVANCE: In this study, the MSPen showed high performance for discriminating thyroid, parathyroid, and lymph node tissues intraoperatively, suggesting this technology may be useful for providing near real-time feedback on tissue type to aid in surgical decision-making.
引用
- DeHoog RJ, King ME, Keating MF, et al. Intraoperative Identification of Thyroid and Parathyroid Tissues During Human Endocrine Surgery Using the MasSpec Pen. JAMA Surg. 2023. PMID: 37531134. doi:10.1001/jamasurg.2023.3229
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。