Objective Given the rising popularity of antibody-drug conjugate (ADC), this study was performed to assess the expression of human epidermal growth factor receptor 2 (HER-2) in metastatic colorectal cancer (mCRC) based on the Hofmann and HERACLES standards and to investigate its clinical significance.

Methods We evaluated 105 metastatic lesions and corresponding primary tumor tissues from 70 patients diagnosed with mCRC at Tianjin Medical University Cancer Institute & Hospital from January 2011 to December 2013. We assigned each patient to one of the three HER-2 expression classifications (HER-2 absent, HER-2 low, and HER-2 positive) separately using Hofmann and HERACLES criteria. We compared the consistency of the two evaluation criteria and the results between matched metastatic lesions and primary tumors. We also analyzed the correlation between HER-2 expression and clinicopathological factors in the patients.

Results HER-2 positivity rates based on Hofmann and HERACLES criteria were 3.8% (4/105) for metastatic lesions of mCRC and 7.1% (5/70) and 5.7% (4/70) for primary tumors, respectively. The detection rates of low HER-2 expression in both primary and metastatic lesions were higher with the Hofmann criteria (28.6% vs. 25.7%) than with the HERACLES criteria (20.0% vs. 11.4%) . The consistency of HER-2 expression assessment between the two criteria was good (kappa coefficient: 0.828). An inconsistency was observed in HER-2 expression between primary tumors and metastatic lesions (inconsistency rate, 28.6%).

Conclusions The incidences of HER-2 positivity and low HER-2 expression were inconsistent between metastatic and primary lesions in mCRC, suggesting that clinical testing for both should be conducted simultaneously to improve the detection rates. The Hofmann criteria showed a higher detection rate for low HER-2 expression than that by the HERACLES criteria. Given the lack of a unified interpretation standard for HER-2 expression in CRC, the relationship between the clinical efficacy of ADC and HER-2 expression under different interpretation standards needs to be further evaluated in large-sample, multicenter studies.