| Factor | Patients n | OS months | Univariate analysis | Multivariate analysis# | |
| p-value | HR (95% CI) | p-value | |||
| Sex | |||||
| Female | 44 | 12.3 | |||
| Male | 41 | 12.4 | 0.770 | ||
| Age yrs | |||||
| <65 | 54 | 11.5 | |||
| ≥65 | 31 | 13.4 | 0.893 | ||
| Smoking | |||||
| Nonsmokers | 63 | 14.6 | |||
| Current/former smokers | 22 | 8.3 | 0.226 | ||
| ECOG PS | |||||
| 0–1 | 63 | 14.1 | |||
| 2–4 | 22 | 6.8 | 0.165 | ||
| Tumour type | |||||
| Nonadenocarcinoma | 4 | 3.5 | 1 | ||
| Adenocarcinoma | 81 | 13.1 | 0.049 | 0.16 (0.03–0.96) | 0.0448 |
| Extrapulmonary metastasis | |||||
| Yes | 70 | 10.4 | 1 | ||
| No | 15 | 23.5 | 0.032 | 0.45 (0.20–1.01) | 0.0514 |
| EGFR | |||||
| Wild type | 18 | 10.6 | 1 | ||
| Mutation | 42 | 17.8 | 0.008 | 0.19 (0.08–0.44) | 0.0001 |
| TKI use | |||||
| No | 14 | 4.8 | |||
| Yes | 71 | 16.2 | <0.001 | ||
| Treatment sequence¶ | |||||
| Best supportive care | 5 | 0.3 | 106.19 (18.24–618.11) | <0.0001 | |
| TKI only+ | 18 | 6.9 | 4.46 (2.00–9.95) | 0.0003 | |
| Chemotherapy only+ | 5 | 6.3 | |||
| TKI then chemotherapy | 25 | 20.9 | |||
| Chemotherapy then TKI | 15 | 16.2 | |||
| Chemotherapy then chemotherapy | 17 | 19.7 | |||
HR: hazard ratio; ECOG PS: Eastern Cooperative Oncology Group performance status; EGFR: epidermal growth factor receptor, TKI: tyrosine kinase inhibitor. #: before performing multivariate analysis, we performed logistic regression to identify potential factors associated with the 25 patients lacking EGFR sequencing data. Potential factors mainly included male sex, poor performance status and chemotherapy as the first-line treatment. After weighting the missing data of the 25 patients lacking EGFR sequencing data by logistic regression, we used the Cox proportional hazard method for multivariate analysis of OS. ¶: the sequence of patient treatment courses was classified according to the first- and second-line medications. +: TKI-only and chemotherapy-only patients received only first-line systemic treatment without second-line treatment.