PO.IM01.13 · 免疫学
用于ADC MATCH试验的HER2、TROP2和NECTIN4 IHC预测算法的验证
Validation of HER2, TROP2, and NECTIN4 IHC prediction algorithms for the ADC MATCH trial
作者与单位 Authors & Affiliations
摘要 Abstract
中文摘要
引言
抗体药物偶联物（ADC）近来已成为靶向肿瘤治疗的一大主流类别。其使用仍然复杂，获批情况从广泛的、泛肿瘤的免疫组化（IHC）伴随诊断获批，到更狭窄的特定适应症标签不等。正在进行的ADC MATCH临床试验（NCT06311214）考察由HER2、TROP2和NECTIN4基因/蛋白的反射性RNA/IHC检测所定义的治疗算法，是否能实现对晚期实体瘤成功的生物标志物导向治疗。本研究描述了ADC MATCH研究所要求的、用于识别可能IHC阳性患者的RNA-seq算法的验证。
方法
收集了三个回顾性、去标识化的RNA+IHC数据集，以支持对HER2 IHC预测、TROP2 IHC预测和NECTIN4 IHC预测的CAP/CLIA实验室自建检测的准确性研究。RNA-seq由Tempus AI, Inc（Tempus xR）执行，而IHC由Neogenomics, Inc（HER2；克隆4B5）、Tempus AI, Inc.（TROP2；克隆SP294）或Histologix（NECTIN4；AB192033）执行。IHC阳性定义为2+/3+评分组（即约为H-score > 100）。高表达者定义为使用log2基因tpm值的二元阈值：7.9（HER2）、4.9（TROP2）和6.8（NECTIN4）。
结果
三项准确性研究的样本量分别为2,018（HER2）、184（TROP2）和478（NECTIN4）。使用阳性符合率（PPA）和阴性符合率（NPA）评估了这三个靶点的IHC预测性能。发现各靶点的PPA分别为47%（HER2；n=545）、93%（TROP2；n=146）和35%（NECTIN4；n=71）。各靶点相应的NPA分别为89%（HER2；n=1473）、55%（TROP2；n=38）、92%（NECTIN4；n=407）。PPA和NPA在不同癌症类型间有所不同；例如，TROP2在结直肠癌中的PPA显著低于其他癌症（62% vs 95%；p < .05）。
结论
使用RNA-seq数据进行IHC预测是一种有前景的方法，可识别可能对特定蛋白质生物标志物检测呈阳性的患者，有望助力临床试验筛选和治疗决策。未来的研究，如ADC MATCH临床试验，将评估该方法的临床价值和结局。
查看英文原文 English abstract
Introduction
Antibody-drug conjugates (ADC) have recently emerged as a leading class of targeted oncology therapies. Their usage remains complex, with approvals ranging from broad, tumor-agnostic immunohistochemistry (IHC) companion diagnostic approvals to narrower indication-specific labels. The ongoing ADC MATCH clinical trial (NCT06311214) examines whether a treatment algorithm defined by reflex RNA/IHC testing of the HER2, TROP2, and NECTIN4 gene/proteins can lead to successful biomarker-directed treatment of advanced solid tumors. This study describes the validation of an RNA-seq algorithm for identifying likely IHC-positive patients as required by the ADC MATCH study.
Methods
Three retrospective, de-identified RNA+IHC datasets were collected to enable accuracy studies of CAP/CLIA lab-developed tests for HER2 IHC prediction, TROP2 IHC prediction, and NECTIN4 IHC prediction. RNA-seq was performed by Tempus AI, Inc (Tempus xR), while IHC was performed by Neogenomics, Inc (HER2; clone 4B5), Tempus AI, Inc. (TROP2; clone SP294), or Histologix (NECTIN4; AB192033). IHC positivity was defined as 2+/3+ score groups (i.e., approximately an H-score > 100). High expressors were defined as binary thresholds using log2 gene tpm values of 7.9 (HER2), 4.9 (TROP2), and 6.8 (NECTIN4).
Results
The sample sizes for the three accuracy studies were 2,018 (HER2), 184 (TROP2), and 478 (NECTIN4), respectively. The performance of IHC prediction was assessed for these three targets using positive percent agreement (PPA) and negative percent agreement (NPA). The PPA of each target was found to be 47% (HER2; n=545), 93% (TROP2; n=146), and 35% (NECTIN4; n=71). The corresponding NPA of each target was found to be 89% (HER2; n=1473), 55% (TROP2; n = 38), 92% (NECTIN4; n=407). PPA and NPA varied among cancer types; for example, the TROP2 PPA in colorectal cancer was substantially lower than other cancers (62% vs 95%; p < .05).
Conclusion
IHC prediction using RNA-seq data is a promising approach to identify patients who are likely to test positive for specific protein biomarkers, potentially aiding in clinical trial screening and treatment decisions. Future studies, such as the ADC MATCH clinical trial, will assess the clinical value and outcomes of this approach.
利益披露 Disclosure
K. A. Beauchamp,
Tempus AI Employment, Stock, Patent.
E. Mauer,
Tempus AI Employment, Stock.
K. Nadhamuni,
Tempus AI Employment, Stock.
E. morency,
Tempus AI Employment, Stock.
A. Zander,
Tempus AI Employment, Stock, Patent.
X. Zheng,
Tempus AI Employment, Stock.
K. Mclean,
Tempus AI Employment, Stock.
S. Mukhopadhyay,
Tempus AI Employment, Stock.
S. Hyun,
Tempus AI Employment, Stock.
C. Sangli,
Tempus AI Employment, Stock.
K. Sasser,
Tempus AI Employment, g., Board of Directors, non-salaried role), Stock.
H. Nimeiri,
Tempus AI Employment, g., Board of Directors, non-salaried role), Stock.
C. Koyias,
Tempus AI Employment, Stock.
AbbVie Employment, Stock.
M. Ting-Lin,
Tempus AI Employment, Stock.
F. Meric-Bernstam,
MULTIPLE Other, Funda Meric-Bernstam reports personal fees from AstraZeneca Pharmaceuticals, Becton Dickinson, Biocartis NV, Calibr (a division of Scripps Research), Daiichi Sankyo, Dava Oncology, Debiopharm, EcoR1 Capital, eFFECTOR Therapeutics, Elevation Oncology, Exelixis, GT Aperion, Incyte, Jazz Pharmaceuticals, LigaChem Biosciences, Menarini Group, Molecular Templates, Protai Bio, Ribometrix, SystImmune, TE.
Becton Dickinson ).
Biocartis NV ).
Calibr ).
Daiichi Sankyo ).
Dava Oncology ).
Debiopharm ).
EcoR1 Capital ).
eFFECTOR Therapeutics ).
Elevation Oncology ).
Exelixis ).
GT Aperion ).
Incyte ).
Jazz Pharmaceuticals ).
LigaChem Biosciences ).
Menarini Group ).
Molecular Templates ).
Protai Bio ).
Ribometrix ).
TEMPUS ).