A multicenter, non-interventional, descriptive study to assess assay concordance of HER2 IHC testing in Chinese pan-tumor patients - Pan HER2

This is a multicenter, non-interventional, descriptive study to retrospectively collect solid tumor samples from focused tumor types (NSCLC, GYN cancers, BTC and UC) and assess both assay concordance and interpretation concordance.

This is a multicenter, non-interventional, descriptive study which included about 2100 patients, a mixture of NSCLC, BTC, GYN cancers and UC, from 12 sites with confirmed diagnosis between 2023 Jan and 2025 Sep. This study includes two phases, which are enrollment phase and assessment phase. In the enrollment phase, all sites will retrospectively collect about 2100 patients (NSCLC ~1,000; BTC ~500; GYN ~450; and UC ~150). Every patient will be tested for HER2 status by 4B5 in local labs. All HER2 results will be reviewed and aligned by a committee of three pathologists using GC algorithms. Based on HER2 results by HercepTest at central lab, patient will be selected for further evaluation in assessment phase. In assessment phase, a total of 320 patients will be included and selected from enrollment phase to meet sample size requirement which is 80 patients with different HER2 expression level (each 20 cases for IHC 0/1+/2+/3+) in NSCLC, BTC, GYN cancers and UC cohorts. In Part I Assay performance evaluation, tissue sample of every patient will be sectioned at least 15 slides and send to central labs for assays evaluation. Interpretation results will be reviewed and aligned by a committee of three pathologists using GC algorithms. Assays and performed platform details are listed below: Roche 4B5 in Ventana platform, MXB\Zhongshan\An Biping\Amoy in Ventana (or Leica) platform (5 assays) v.s. HercepTest in Dako platform. Part II Interpretation concordance evaluation: inter-obsever agreement will be evaluated based on scanned images. Specifically, the first 40 slides (each 10 for 0/1+/2+/3+, each 10 for four disease cohorts) will be selected and digitalized. 36 pathologists, 3 from each site, will be trained for GC interpretation algorithm before joining in the assessment. All slides will be interpreted by all pathologists.