Clinical efficacy of osimertinib in EGFR-mutant non-small cell lung cancer with distant metastasis

This retrospective cohort study was conducted with the approval of the ethical review committee of Nagoya University Hospital (approval number:2018–017) and in accordance with the guidelines of the Declaration of Helsinki [20, 21]. We retrospectively reviewed the medical records of patients from five facilities, including Nagoya University Hospital, Konan Kosei Hospital, Kariya Toyota General Hospital, Tosei General Hospital, and Japanese Red Cross Aichi Medical Center Nagoya Daiichi Hospital. Patients enrolled in this study were selected based on the following eligibility criteria: (1) diagnosed with stage III/IV or recurrent non-squamous NSCLC, as confirmed by histological or cytological examination from January 2015 to December 2020; (2) presented with a positive EGFR mutation (exon 19 deletion or L858R point mutation); (3) were receiving 1^st-generation EGFR-TKI (gefitinib or erlotinib), or 2^nd-generation EGFR-TKI (afatinib), or 3^rd-generation EGFR-TKI (osimertinib) for 1^st line therapy. We excluded patients with no available data and non-target regions, and the time of data cut-off was August 2021. The clinical information of eligible patients, including age, sex, smoking history, histological subtype, clinical stage, performance status, treatment outcome, metastatic site, and EGFR mutation status, were retrospectively obtained from medical records. Clinical stages were assigned according to the eighth edition of the American Joint Committee on Cancer. Objective tumor responses were evaluated according to the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 [22].

We retrospectively reviewed the clinical data of 388 eligible patients with advanced non-squamous NSCLC harboring EGFR mutations treated with EGFR-TKIs as 1^st-line therapy, which were collected from five medical institutions. The flowchart of patient selection from our medical records is shown in Supplementary Fig. 1, and the enrolled patient characteristics are summarized in Table 1. The median age was 72.0 years (range 26–92), 61.6% were female, and 62.1% did not have a history of smoking. In the majority of patients, the Eastern Cooperative Oncology Group (ECOG) performance status (PS) was 0 (n = 230, 59.3%), and the clinical stage was IV (n = 270, 69.6%). Exon 19 deletions were observed in 188 patients (48.6%), and L858R point mutations were observed in 199 patients (51.4%). The number of patients with other metastatic organs included 92 (23.7%) for the contralateral lung, 160 (41.2%) for the bone, 118 (30.4%) for the brain, 34 (8.8%) for the liver, and 26 (6.7%) for the adrenal, which are consistent with the frequencies of EGFR-mutant NSCLC metastases in previous reports [3‐5]. In this cohort, 183, 55, and 150 patients were treated with gefitinib, erlotinib, afatinib, and osimertinib, respectively. There were no statistically significant differences in sex, smoking status, performance status (PS), stage, and metastatic organs between gefitinib/erlotinib, afatinib, and osimertinib groups. The proportion of exon 19 deletion mutations in the afatinib group was higher than that in the other TKI groups (P < 0.0001), and the patients treated with afatinib were significantly younger than those in the other groups (P < 0.0001).

First, we performed univariate and multivariate analyses of PFS in each of the three groups. In addition to brain, bone, and liver metastases, the characteristic variables with P < 0.15 in univariate analysis of the osimertinib group were used in the multivariate analysis. In the osimertinib group, male sex, poor PS, L858R mutation, and liver metastasis were independently associated with shorter PFS (male: HR, 2.06; 95% CI, 1.25–3.38; P = 0.0045, PS1: HR, 2.70; 95% CI, 1.47–4.94; P = 0.0013, PS≧2: HR, 2.43; 95% CI, 1.16–5.11; P = 0.0193, L858R mutation: HR, 2.03; 95% CI, 1.17–3.53; P = 0.0120; liver metastasis: HR, 6.20; 95% CI, 2.87–13.38; P < 0.0001, Table 2). Among them, the PFS in patients with liver metastases was significantly shorter than that in patients without liver metastases (7.4 vs. 19.7 months; Wilcoxon P < 0.0001 and log-rank P < 0.0001; Fig. 1A), while the other distant metastatic sites were not associated with shorter PFS (Table 2). In addition, the OS in patients with liver metastases was remarkably shorter than in patients without liver metastases (12.1 months vs not reached; Wilcoxon P < 0.0001 and log-rank P < 0.0001; Fig. 1B). In the gefitinib/erlotinib group, poor PS (PS ≥ 2) was an independent poor prognostic factor (HR, 1.83; 95% CI, 1.14–2.94; P = 0.0124), although brain, bone, and liver metastases were not associated with shorter PFS (brain: HR, 1.15; 95% CI, 0.79–1.66; P = 0.4663; Bone: HR, 1.17; 95% CI, 0.83–1.66; P = 0.3712; Liver: HR, 1.44; 95% CI, 0.81–2.57; P = 0.2108, Supplementary Table 1A). Similarly, in the afatinib group, metastasis sites were not statistically associated with PFS (brain, HR, 1.42; 95% CI, 0.74–2.73; P = 0.2950; Bone, HR, 1.19; 95% CI, 0.59–2.42; P = 0.6258; Liver, HR, 2.09; 95% CI, 0.50–8.71; P = 0.3132; Supplementary Table 1B).

In comparison with the clinical efficacy of EGFR-TKIs, the PFS in the patients treated by osimertinib was statistically significantly longer than in the patients treated by gefitinib/erlotinib or afatinib (17.1 vs. 10.1 months; Wilcoxon P < 0.0001, log-rank P < 0.0001; HR, 0.52; 95% CI, 0.39–0.69; and 17.1 vs. 13.4 months; Wilcoxon P = 0.0541, log-rank P = 0.0250; HR, 0.65; 95% CI, 0.45–0.95; Fig. 2A). Furthermore, the patients of the osimertinib treatment group showed longer OS than the patients of the gefitinib/erlotinib treatment group, although not statistically significant (not reached vs. 34.1 months; Wilcoxon P = 0.2150 and log-rank P = 0.1818; HR, 0.77; 95% CI, 0.52–1.13; Fig. 2B), and there were no significant differences of OS in between osimertinib and afatinib groups (Fig. 2B).

Patient characteristics with brain, bone, and liver metastases are summarized in Supplementary Table 2A, 2B, and 2C, respectively. The number of patients with brain metastases treated with stereotactic radiosurgery was significantly lower in the osimertinib group than in the other TKI groups (P = 0.0239; Supplementary Table 2A). The number of bone metastatic sites, treatment with bone-modifying agents, or radiation therapy was not significantly different between the gefitinib/erlotinib, afatinib, and osimertinib groups (Supplementary Table 2B), and the number of liver metastatic sites were also not significantly different between the gefitinib/erlotinib, afatinib, and osimertinib groups (Supplementary Table 2C). In comparison with gefitinib/erlotinib, the forest plots of PFS showed that osimertinib was associated with a significant survival benefit in brain metastases (HR, 0.55; 95% CI, 0.34–0.88; P = 0.0137), bone (HR, 0.41; 95% CI, 0.27–0.63; P < 0.0001) and pleura (HR, 0.52; 95% CI, 0.33–0.80; P = 0.0034) (Fig. 3A). On the contrary, the largest numerical differences in the hazard ratio between osimertinib and the gefitinib/erlotinib group were observed in the patients with and without liver metastases (HR, 1.40; 95% CI, 0.63–3.11; P = 0.4054); however, there were no statistical differences between osimertinib and gefitinib/erlotinib in the subgroup with contralateral lung metastases and adrenal metastases (Fig. 3A). The second largest numerical difference was observed between exon 19 deletion and L858R mutations (Fig. 3A). Similar to previous studies [17], osimertinib was associated with a significant survival benefit in the exon 19 deletion subgroup (HR, 0.36; 95% CI, 0.22–0.59; P < 0.0001) compared to the L858R subgroup (HR, 0.67; 95% CI, 0.47–0.95; P = 0.0253).

Similarly, in comparison with afatinib, osimertinib was associated with better survival benefit in the brain and bone metastases subgroup, although the difference was not statistically significant (brain: HR, 0.59; 95% CI, 0.33–1.09; P = 0.0888; Bone: HR, 0.66; 95% CI, 0.39–1.11; P = 0.1175) (Fig. 3B). In contrast, the largest numerical differences were observed between patients with and without liver or adrenal metastases, indicating that the clinical benefits of osimertinib are weak in patients with liver metastases and adrenal metastases (liver: HR, 1.83; 95% CI, 0.41–8.30; P = 0.4309, adrenal; HR, 2.22; 95% CI, 0.58–8.54; P = 0.2448; Fig. 3B). Additionally, large numerical differences were observed between the exon 19 deletion and L858R mutation, indicating that osimertinib was associated with a significant survival benefit in the exon 19 deletion subgroup (HR, 0.41; 95% CI, 0.24–0.70; P = 0.0010), but not in the L858R subgroup (HR, 0.94; 95% CI, 0.46–1.94; P = 0.8763) (Fig. 3B).

Subsequently, we analyzed the clinical efficacy of EGFR-TKIs in patients with distant organ metastasis using the Kaplan–Meier estimator. The PFS of the patients with brain metastasis treated by osimertinib was significantly longer than those of the patients of the gefitinib/erlotinib group (16.3 vs. 7.9 months; Wilcoxon P = 0.0075 and log-rank P = 0.0120), and the afatinib group (16.3 vs. 8.3 months; Wilcoxon P = 0.0347 and log-rank P = 0.0845) (Fig. 4A). Furthermore, the OS in the patients with brain metastasis treated by osimertinib was significantly longer than in the patients treated by gefitinib/erlotinib (not reached vs. 20.9 months; Wilcoxon P = 0.0725 and log-rank P = 0.0326), while there was not significantly difference between the osimertinib and the afatinib group (not reached vs. 53.5 months; Wilcoxon P = 0.6219 and log-rank P = 0.8118) (Fig. 4B). These results indicate that the clinical efficacy of osimertinib for patients with EGFR-mutant NSCLC with brain metastases is equal to or greater than that of the other EGFR-TKIs, and our analyzed data were consistent with the previous reports of the FLAURA trial [17, 18]. Furthermore, in patients with bone metastasis, the PFS of the osimertinib group was significantly longer than that of the gefitinib/erlotinib group (17.0 vs. 8.6 months; Wilcoxon P < 0.0001 and log-rank P < 0.0001; Fig. 4C) and showed a better trend compared with those of the patients in the afatinib group (17.0 vs. 12.9 months; Wilcoxon P = 0.1884 and log-rank P = 0.1144; Fig. 4C), although the OS demonstrated no significant differences among the three EGFR-TKI groups (Fig. 4D). However, the PFS in the patients with liver metastasis of osimertinib group showed no superiority to the patients of the gefitinib/erlotinib group (7.4 vs. 7.1 months; Wilcoxon P = 0.7406 and log-rank P = 0.3997; Fig. 4E) and the afatinib group (7.4 vs. 5.6 months; Wilcoxon P = 0.8674 and log-rank P = 0.4247; Fig. 4E). Similar to the PFS analyses, the OS of patients with liver metastasis treated with osimertinib showed no significant difference compared to the patients treated with 1^st- or 2^nd-generation EGFR-TKIs (Fig. 4F). However, the PFS in patients without liver metastasis was significantly better in the osimertinib group than in the other EGFR-TKI groups (Supplementary Fig. 2A). The OS in the patients without liver metastasis in the osimertinib group was significantly better than that of the patients in the gefitinib/erlotinib group, while there was no significant difference between the osimertinib and afatinib groups (Supplementary Fig. 2B). These results indicate that liver metastasis critically affects the clinical efficacy of osimertinib in patients with EGFR-mutant NSCLC.

In patients treated with osimertinib, the PFS and OS of patients with exon 19 deletions were significantly longer than those of patients with the L858R mutation (Fig. 5A and Fig. 5B). As expected, the PFS in the patients with exon 19 deletion mutation in the osimertinib group was significantly better than that in the patients in the gefitinib/erlotinib group and the afatinib group (not reached vs. 9.8 months; Wilcoxon P = 0.0001 and log-rank P < 0.0001, and not reached vs. 13.2 months; Wilcoxon P = 0.0085 and log-rank P = 0.0007; Fig. 5C). However, the PFS of the patients with L858R mutation did not show a significant difference between the osimertinib group and the afatinib group (13.6 vs. 14.9 months; Wilcoxon P = 0.6038 and log-rank P = 0.8761; Fig. 5C); though, osimertinib showed superior median PFS to gefitinib/erlotinib (13.6 vs. 10.2 months; Wilcoxon P = 0.0176 and log-rank P = 0.0239; Fig. 5D). On the other hand, the OS of patients with both exon 19 deletion and L858 mutation in the osimertinib group were not significantly different from those in the other EGFR-TKI groups (Supplementary Fig. 3A and 3B).

To assess the response rate to osimertinib, we compared the ORRs in patients treated with each EGFR-TKI. The ORRs were 68.2% (116/170) in the gefitinib/erlotinib group, 59.6% (31/52) in the afatinib group, and 73.9% (99/134) in the osimertinib group, indicating that osimertinib demonstrated a slightly better response rate than the other EGFR-TKIs (Table 3). However, the response rate in patients with liver metastasis treated with osimertinib was remarkably reduced, and the ORRs were 53.3% (8/15), while those of the gefitinib/erlotinib and afatinib groups were 57.1% (8/14) and 66.7% (2/3), respectively (Table 4 and Supplementary Fig. 4A). Consistent with the results of PFS and OS, the clinical response to osimertinib in patients with liver metastases was significantly attenuated, and the effectiveness was similar to 1^st- or 2^nd-generation EGFR-TKIs.

In addition, in patients with exon 19 deletion mutations, the ORR of the osimertinib group was 76.2% (48/63), whereas that of the gefitinib/erlotinib and afatinib groups was 66.7% (46/69) and 57.1% (24/42), respectively (Table 5 and Supplementary Fig. 4B). However, in patients with L858R mutations, the ORRs of the osimertinib group were 71.8% (51/71), whereas those of the gefitinib/erlotinib and afatinib groups were 70.0% (70/100) and 70.0% (7/10), respectively (Table 5 and Supplementary Fig. 4B). In addition to the analyses of PFS and OS, the ORRs in patients with exon 19 deletions treated with osimertinib were superior to those in patients with other EGFR-TKIs; while those in patients with L858R did not show remarkable differences among the three groups.