PS01.30: Domain-Specific c-Met Measurement by Quantitative Immunofluorescence and Mass Spectrometry in Non–Small Cell Lung Cancer

Topic: Pathology


      C-Met plays a key role in tumor proliferation, survival, invasion and metastasis, but c-Met targeted therapeutic antibodies, like onartuzumab, have been unsuccessful so far. Notably, the companion diagnostic test used in that trial relied on an antibody (SP44) that binds to a different domain epitope than that of the c-Met targeted agent. Here, we measured the different c-Met domain expression in non–small cell lung cancer (NSCLC) patient tumors with the goal of determining any domain discordances that could potentially affect the patient stratification efficacy of companion diagnostic tests.


      We measured c-Met expression using quantitative immunofluorescence (QIF) with two antibodies; D1C2 targeting the intracellular domain (ICD) and L6E7 targeting the extracellular domain (ECD). Both antibodies were validated using Mass Spectrometry (MS) in cell lines. These antibodies were also used to assess c-Met domain expression in 3 cohorts of over 600 NSCLC patient cases on tissue microarrays (TMAs). Additionally, c-Met ICD QIF measurement by SP44 (commonly used in c-Met clinical trials) was compared to the other 2 antibodies.


      QIF scores of both ICD and ECD antibodies were strongly correlated with the absolute c-Met concentration measured by MS (in attomoles/microgram) in 15 cell lines (R2=0.77 and R2=0.75 respectively). Additionally, in cell lines, there was a high correlation between the two different c-Met domain antibodies (R2=0.88). However, there was no correlation (R2=0.01) between ICD and ECD in any of the 3 patient cohorts, indicating that cases with high expression of one of the domains may have low expression of the other. Domain specific heterogeneity was minor, with a high correlation between different regions of tissue (TMA build 1 vs build 2) for each antibody (R2>0.8). Domain-specific discordance was further confirmed in whole tissue of selected cases, ruling out the possibility of TMA derived artifacts. Similarly, high correlation was seen between c-Met expression measured by SP44 ICD and D1C2 ICD, but not between SP44 ICD and L6E7 ECD. C-Met ECD expression was statistically significantly lower in patients with KRAS mutations compared to EGFR-mutant and EGFR/KRAS wild type patients, while no correlation was seen between ICD and mutational status.


      Determination of domain-specific c-Met expression in NSCLC tumors suggests biological differences in the tumors that are not seen in cell line models. This high incidence of c-Met domain discordance in patients’ tumors should be considered in the design of companion diagnostic assays.