br For further characterization of c MET
For further characterization of c-MET we calculated inter-quartiles using the average H-score. Skewness was calculated and found to be 1.41, indicating that the distribution was not normal, with a median H-score of 20. The H-score for 25-, 50- and 75- quartiles was 0.875, 20 and 80, respectively (Suppl. Fig. 1). For the main purpose of the analysis, the median H-score of 20 was used.
2.3. Statistical analysis
Demographical and clinical characteristics were described by an H-score ≥ and < 20, including p-tests calculated using the Chi-square, and the t-test. The primary outcome of interest was overall survival defined as the time from diagnosis to the date of death by any cause, or last date of follow-up (for patients who received surgery from January 1995 to January 2005 last date of follow- up was 28/11/2010 and for patients who received surgery from February 2005 to December 2010 it was 10/10/2015). The secondary outcome was progression-free sur-vival (PFS) which was determined from date of diagnosis to date of relapse or death, whichever occurred first, or to date of last planned visit to the oncology department without any signs of relapse radi-ologically or clinically. Overall survival was assessed by the Kaplan-Meier method. The relative risk of death of all-causes was expressed as hazard ratios (HR) with 95% confidence intervals (CI) using univariate and multivariate Cox regression analyses.
Univariate Cox regression analyses were undertaken with perfor-mance status (PS), age at diagnosis, histology, received adjuvant treatment, surgical stage of disease, smoking status, gender, ALK and c-MET H-score as independent variables. Univariate Cox regression ana-lyses were done in the whole Ferrostatin1 and in the subgroup of lung adenocarcinoma and squamous cell carcinoma patients with stage IIA-IIIB.
The three diﬀerent quartiles were examined separately with Cox regression multivariate analyses in the whole population. All the col-lected variables were included in these analyses, but we performed separate analyses with or without adjuvant treatment included, due to a larger number of missing data regarding the delivery of adjuvant treatment compared with all other variables.
The assumption of proportional hazards in the Cox Regression models was in a first step violated, which was verified visually and tested based on weighted residuals. Therefore, we included time-de-pendent variables in the model such as ALK rearrangement, treatment and age at diagnosis, and then the final models were not violated.
Multivariate OS and PFS analyses were done separately in sub-groups stratified according to delivery of adjuvant treatment, as well as in the subgroup of lung adenocarcinoma and squamous cell carcinoma patients with stage IIA-IIIB.
All tests were two-sided and statistical significance was considered at a 5% level. Statistical analyses were performed using R 9.2.
3.1. Patient characteristics
The total number of patients included in our analyses was 653 after patients with missing information were excluded (Table 1). The de-mographics of the whole cohort and stratified by c-MET H-score 20 are presented in Table 2. We had information about c-MET IHC and ALK FISH for the whole cohort, but KRAS and EGFR mutation status only for 360 patients, this being the reason for excluding KRAS and EGFR from
Demographical and clinical characteristics by cMET score.
Stage at diagnosis
Performance Status (PS)
cMET: cellular Mesenchymal Epithelial Transition factor, ALK: Anaplastic
Lymphoma Kinase, LCLC: Large-Cell Lung Cancer, NOS: Non Otherwise Specified, CT: Chemotherapy, RT: Radiotherapy, sd: standard deviation.
the present analysis. Regarding the adjuvant treatment regimen, data collection was feasible for 148 patients, of which 34.5% received Cis-platin/Vinorelbine, 54.7% Carboplatin/Vinorelbine, 9.4% Carboplatin/ Gemcitabine and 1.4% Carboplatin/Paclitaxel.
3.2. Survival analyses
The OS and PFS analyses for all the c-MET H-score quartiles failed to show any statistical significance (data not shown for 25- and 75-quartiles). Regarding the 50-quartile (c-MET H-score 20), we performed separate Kaplan-Meier OS and PFS analyses in the whole study popu-lation, in the subgroup of patients who received adjuvant treatment and in the subgroup of patients who did not receive any adjuvant treatment (Figs. 1 and Suppl. 2). We observed a crossover of the curves at 7.5 years in the whole study population (Fig. 1a) and in patients not re-ceiving adjuvant treatment (Fig. 1c), something which was not seen for the subgroup of patients who received adjuvant treatment (Fig. 1b). A