br An optimal cut o value for c MET protein
An optimal cut-oﬀ value for c-MET protein expression with clinical impact on prognosis or prediction has not yet been defined. In our study we analysed diﬀerent quartile cut-oﬀ values using the H-scoring system from to 300 and identified H-score 20, being the 50-quartile, to be a clinically significant cut-oﬀ value. We believe that the H-scoring system is a good method taking into account both the intensity and extent of protein expression. Other studies have identified diﬀerent clinically Lung Cancer 133 (2019) 69–74
significant cut-oﬀ values, including H-score 60 in stage IV NSCLC pa-tients  and ≥50% of cells with moderate or strong staining in the MetMAb trial . These diﬀerences may be due to several factors that diﬀer between the studies, including the use of diﬀerent antibodies, scoring methods, disease stage and treatments. Regarding diﬀerent antibodies, patient selection through immunohistochemical evaluation of c-MET expression in formalin-fixed and paraﬃn-embedded (FFPE) samples is problematic, because only a limited number of validated STA-21 to c-MET that work in FFPE are available. Most studies in diﬀerent types of cancer demonstrating a correlation between Met and disease progression have used antibodies directed against the COOH-terminus of Met. However, in one study on lymph node negative breast carcinoma, antibodies against the intracellular domain of Met were predictive of worse outcome whereas antibodies against the extra-cellular domain were not . Furthermore, the cellular distribution of Met may vary and influence the IHC results; it has been shown that Met may be expressed not only in the cytoplasmic and membranous com-partments but also in the nucleus, especially in the invasive front of tumors . A harmonization between diﬀerent studies would be needed to decide on an optimal cut-oﬀ value.
Adjuvant chemotherapy has proved to be of benefit in patients with stage IIA-IIIB NSCLC , and though its use in stage IB patients re-mains controversial, many centers oﬀer adjuvant chemotherapy to pa-tients with tumors ≥4 cm . In our cohort, 20.4% with stage IA disease surprisingly received adjuvant chemotherapy while the ma-jority of patients with stage IB disease did not receive any adjuvant therapy. A subgroup analysis in stage IIA-IIIB patients with adeno-carcinoma or squamous cell carcinoma was undertaken and since the majority of NSCLC patients in our cohort had stage I disease, this de-cision resulted in less statistical power (n = 202). Nevertheless, uni-variate and multivariate Cox regression analysis showed a statistically significant positive impact on OS for c-MET H score ≥20. The PFS benefit was more profound in the multivariate analyses in this subgroup of patients. c-MET has been implicated in resistance to chemotherapy, including platinum , which would be in contrast to our finding that a higher c-MET score is beneficial in patients treated with adjuvant platinum-based chemotherapy. On the other hand, c-MET has been implicated in the process of metastasis and our results may be explained by the fact that patients with a higher c-MET expression derive a greater benefit from adjuvant chemotherapy than those with a lower expres-sion.
The major limitation of our study is its retrospective nature making it prone to selection bias. There is also a risk of information bias re-garding missing data mostly on patients receiving adjuvant therapy. Information bias regarding c-MET H-scoring is limited due to the methodology used (see methods), but can not be avoided Tumour fixation is important for the reliability of IHC staining, because over-fixation or underfixation can cause false positive or false negative re-sults, respectively.Interobserver variability, a major problem in inter-preting IHC results, has been reduced in our study by considering the H-score as a continuous variable. Another consideration is that TMAs may not be representative of the entire tumour, because of intratumoral heterogeneity, this being a limitation in all published trials with IHC or H-scoring. Missing data regarding EGFR-mutation is not a major lim-itation since there is no evidence suggesting that it could be a con-founder in this context . However, there was a large number of patients included in our trial and we had available data regarding the use of adjuvant therapy for 444 patients. The real-life character of this cohort trial adds strength to the external validity of our results. Future trials should probably not be based on IHC alone. However, the use of MET kinase inhibitors for IHC positive or MET amplified tumors has not been fruitful. The exception might be patients with MET exon 14 skipping mutation positive tumors, where new treatments are currently under development . The importance of MET amplification or MET exon 14 skipping mutation as oncogenic drivers is still unclear. The development of conjugate drugs (c-MET antibodies with cytotoxic drug