Efficacy and safety of first-line osimertinib in Chinese patients with EGFR-mutated advanced non-small cell lung cancer: a prospective, multicenter, non-interventional study (FLOURISH)

Introduction

Lung cancer is the leading cause of cancer-related mortality in China¹. More than 50% of Asian patients with non-small cell lung cancer (NSCLC) exhibit epidermal growth factor receptor (EGFR) mutations^2,3. Osimertinib, a third-generation EGFR-TKI, was approved as a first-line (1L) treatment for patients with EGFRm NSCLC, according to results from the FLAURA study. In that study, osimertinib, compared with erlotinib or gefitinib, significantly prolonged the median progression-free survival (PFS) [18.9 vs. 10.2 months; hazard ratio (HR), 0.46; 95% confidence interval (CI), 0.37–0.57; P < 0.001]⁴. Similarly, in the FLAURA China study enrolling exclusively Chinese patients, osimertinib demonstrated a longer median PFS (mPFS) than first-generation EGFR-TKIs (17.8 vs. 9.8 months; HR, 0.56; 95% CI, 0.37–0.85)⁵.

However, clinical trials differ from real-world practice because the latter often involves older patients or individuals with more comorbidities. For example, in a real-world study in the United States, 55% of patients were classified as FLAURA-ineligible and experienced shorter median osimertinib time to treatment discontinuation (TTD) than FLAURA-eligible populations (12.2 vs. 31.1 months)⁶. Therefore, prospective real-world studies with larger sample sizes are needed to validate the effectiveness and safety of osimertinib among Chinese patients with NSCLC with EGFR mutations in real-world settings.

In recent years, real-world studies have preliminarily confirmed the effectiveness and safety of osimertinib initiated in patients with EGFRm NSCLC^7–10. Although existing real-world investigations have provided preliminary insights, several limitations remain. First, the predominance of retrospective study designs introduced inherent selection bias. Second, the limited sample sizes in the current published prospective real-world evidence have left unresolved questions, particularly regarding clinical efficacy in subgroups such as patients with EGFR co-mutations or those not meeting FLAURA criteria. These gaps further underscore the need for prospective, multicenter investigations.

Materials and methods

Next-generation sequencing

Tumor tissue samples were collected at baseline. Next-generation sequencing (NGS) was performed on FFPE tissue and plasma-derived cfDNA samples by using panels from Burning Rock Biotech (Guangzhou, China), the central laboratory in this study. Tissue samples were analyzed with the 518-gene OncoScreen™ Plus system at a sequencing depth of 1,000×. Variant detection and annotation were performed with specialized bioinformatics pipelines.

• Download figure
• Open in new tab
• Download powerpoint

Study flowchart

The FLOURISH study (NCT04391283) was a prospective, multicenter, non-interventional study evaluating first-line osimertinib in 481 treatment-naïve Chinese patients with EGFR-mutated advanced NSCLC across 22 sites. First, the study background and rationale are presented, highlighting the limited real-world evidence of first-line osimertinib in Chinese populations. Subsequently, the study design, patient flow, endpoint definitions, and subgroup analysis framework are described, including analyses by EGFR mutation type, CNS metastasis status, FLAURA eligibility, and co-mutation status. Next, key efficacy outcomes are summarized for the FAS population. Consistent clinical benefit was also observed in FLAURA-ineligible patients, whereas patients bearing co-mutations exhibited diminished clinical benefit. Finally, the key findings and clinical implications are highlighted. CNS, central nervous system; DCR, disease control rate; EGFR, epidermal growth factor receptor; FAS, full analysis set; NGS, next-generation sequencing; NSCLC, non-small cell lung cancer; ORR, objective response rate; OS, overall survival; rwPFS, real-world progression-free survival; TTD, time to treatment discontinuation.

Results

From July 27, 2020, to April 27, 2022, a total of 507 patients were screened at 22 sites across China, among whom 481 eligible patients [denoted the full analysis set (FAS)] received at least 1 dose of osimertinib monotherapy as the 1L treatment. At the data cut-off (October 30, 2024), the median follow-up duration was 31.3 months (95% CI: 30.2–32.3).

Effectiveness evaluation

Overall effectiveness

At the data cut-off, 426 (88.6%) of 481 patients had tumor response evaluations. ORR and DCR were calculated for both the FAS and the evaluable population. The evaluable population analysis served as a sensitivity analysis to assess the robustness of response outcomes. For the FAS, the ORR was 56.3% (271/481; 95% CI: 51.8–60.8), and the DCR was 84.6% (407/481; 95% CI: 81.1–87.7). Among patients with available tumor response data (n = 426), the ORR was 63.6% (271/426; 95% CI: 58.8–68.2), and the DCR was 95.5% (407/426; 95% CI: 93.1–97.3) (Table S4). After a median follow-up of 31.3 months, 267 patients (55.5%) discontinued osimertinib therapy. The median TTD was 24.6 months (95% CI: 22.4–26.7) (Figure 1). Reasons for discontinuation included disease progression (33.3%), death (35.2%), AEs (7.9%), and other factors (23.6%) (Table S2). Progression events occurred in 274 patients (57.0%), and 169 patients (35.1%) died. The median rwPFS was 19.4 months (95% CI: 16.2–20.4), and the mOS was 41.0 months (95% CI: 39.7 to NR) (Figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Kaplan–Meier estimates of TTD, PFS, and OS in the FAS population. (A) Median TTD. (B) Median rwPFS. (C) mOS. FAS, full analysis set; TTD, time to treatment discontinuation; rwPFS, real-world progression-free survival; OS, overall survival; CI, confidence interval.

Subgroup analysis of effectiveness

For the 245 patients (50.9%) with exon 19 deletion mutations, the median TTD, rwPFS, and OS were 28.6 months (95% CI: 25.2–32.7), 20.6 months (95% CI: 17.7–22.1), and 41.0 months (95% CI: 39.7 to NR), respectively (Figures 2 and S1). Among the 221 patients (45.9%) with exon 21 L858R mutations, the median TTD was 20.5 months (95% CI: 18.3–23.5), the median rwPFS was 15.4 months (95% CI: 13.2–19.5), and the mOS was 37.2 months (95% CI: 30.0 to NR) (Figures 2 and S1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Kaplan–Meier estimates of TTD in patient subgroups. (A) Median TTD in the exon 19 deletion (blue curve) and exon 21 L858R mutation (red curve) subgroups. (B) Median TTD in the CNS metastasis (blue curve) and non-CNS metastasis (red curve) subgroups. (C) Median TTD in the FLAURA-ineligible (blue curve) and FLAURA-eligible (red curve) subgroups. (D) Median TTD in patients with co-mutations (blue curve) or without co-mutations (red curve). (E) Median TTD in patients with TP53 co-mutations (blue curve) or without TP53 co-mutations (red curve). TTD, time to treatment discontinuation; CNS, central nervous system; EGFR, epidermal growth factor receptor; CDG, cancer driver genes; TSG, tumor suppressor genes; CI, confidence interval.

At baseline, CNS metastases were present in 165 patients (34.3%). For these patients, the median TTD was 22.2 months (95% CI: 19.6–25.7), the median rwPFS was 15.6 months (95% CI: 14.1–20.3), and the mOS was 40.1 months (95% CI: 39.7 to NR) (Figures 2 and S1). For the 207 patients (43.0%) without CNS metastases, the median TTD, rwPFS, and OS were 25.3 months (95% CI: 22.4–29.3), 19.7 months (95% CI: 15.7–21.4), and 41.0 months (95% CI: 36.8 to NR), respectively (Figures 2 and S1).

Additionally, patients in the FAS were categorized as FLAURA-eligible or FLAURA-ineligible according to the inclusion/exclusion criteria outlined in the FLAURA trial protocol⁴ (Table S3). Of these, 159 patients (33.1%) were classified as FLAURA-ineligible (Table S1), and their median TTD, rwPFS, and OS were 21.4 months (95% CI: 17.7–26.8), 15.1 months (95% CI: 12.3–19.4), and 33.5 months (95% CI: 26.8 to NR), respectively (Figures 2 and S1). In the FLAURA-eligible group (n = 266, 55.3%), the median TTD, rwPFS, and OS were 25.6 months (95% CI: 23.3–28.8), 20.4 months (95% CI: 17.6–21.5), and 43.1 months (95% CI: 39.7 to NR), respectively (Figures 2 and S1). The ORR and DCR of each subgroup are shown in Table S5.

Effects of baseline co-mutations on osimertinib treatment efficacy

Baseline NGS data obtained from 118 patients revealed 2 distinct mutation profiles: 23.7% (28/118) patients had only EGFR mutations, whereas 76.3% (90/118) had additional gene co-mutations (in HER2, KRAS, BRAF, ROS1, ALK, TP53, RB1, or PTEN) (Figure 3). In these subgroups, the median TTD was 32.8 months (95% CI: 21.8 to NR) for patients with EGFR mutations alone and 24.6 months (95% CI: 20.2–26.6) for those with co-mutations (Figure 2). The median rwPFS was 20.8 months (95% CI: 7.4–25.1) in patients with only EGFR mutations, compared with 16.7 months (95% CI: 14.6–20.6) in those with co-mutations. Notably, the mOS was NR (95% CI: 32.9 to NR) in the EGFR-only group but was 29.3 months (95% CI: 25.4 to NR) in the co-mutation group (Figures 2 and S1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Distribution of gene aberrancies detected by baseline next-generation sequencing. Landscape of genetic alterations identified in 118 patients who underwent baseline NGS analysis. Each column represents an individual patient, and each row represents a specific gene. NGS, next-generation sequencing.

Among the 80 (67.8%) patients bearing TP53 co-mutations, the median TTD, rwPFS, and OS were 23.6 months (95% CI: 19.1–26.4) (Figure 2), 16.7 months (95% CI: 14.6–20.6), and 28.5 months (95% CI: 25.3 to NR), respectively (Figure S1). In contrast, for the 38 (32.2%) patients without TP53 co-mutations, the median TTD, rwPFS, and OS were 32.8 months (95% CI: 24.0 to NR) (Figure 2), 20.8 months (95% CI: 9.6–23.7), and NR (95% CI: 32.9 to NR), respectively (Figures 2 and S1). The ORR and DCR of each subgroup are shown in Table S5.

Discussion

This study was the first large-scale, prospective, real-world investigation, to our knowledge, examining the outcomes of 1L osimertinib in Chinese patients with EGFRm NSCLC. Patients with poor PS, various comorbidities, and other exclusion criteria from the FLAURA trial were included, thereby confirming the effectiveness and safety of 1L osimertinib in a broader patient population. Furthermore, our findings suggested that patients with co-mutations might experience diminished clinical benefits from osimertinib monotherapy; therefore, intensified treatment strategies might be necessary in these patients.

Our real-world study indicated a median rwPFS of 19.4 months and an mOS of 41.0 months, values comparable to those in the FLAURA trial (mPFS 18.9 months, OS 38.6 months), despite our cohort’s higher baseline CNS metastasis prevalence (34.3% vs. 20.8%) and higher ECOG ≥2 proportion (7.5% vs. 0%)¹¹. These results aligned with those from previous real-world studies. For example, the prospective Reiwa study⁹, in Japanese patients treated with 1L osimertinib, has reported an mPFS of 20 months and mOS of 41 months. Similarly, the retrospective CAPTRA-Lung study¹², in Chinese patients receiving 1L osimertinib, has reported an mPFS of 19.4 months and an mOS of 40.5 months. The prospective FLOWER study¹³, evaluating osimertinib in Italy, has reported an mPFS of 18.9 months. In addition, our real-world study indicated an ORR of 56.3% (95% CI, 51.8–60.8) and a DCR of 84.6% (95% CI, 81.1–87.7) in the FAS. Notably, a substantial proportion of patients lacked tumor response evaluations, thus potentially influencing our estimates. To enhance interpretability, we also calculated ORR and DCR in the evaluable population. Our findings highlight the importance of systematically capturing tumor response assessments to minimize missingness in future real-world studies.

Exploratory subgroup analyses described variations across different patient groups. In our study, the median rwPFS was 19.7 months in patients without CNS metastases and 15.6 months in those with CNS metastases. Additionally, the median rwPFS was 20.6 months in patients bearing exon 19 deletion mutations and 15.4 months in those with exon 21 L858R mutations. These results were consistent with findings from the FLAURA study⁴ and other real-world evidence¹⁴. Collectively, our findings reinforce osimertinib’s position as a 1L therapy for EGFRm NSCLC.

Notably, inherent differences exist between patients in clinical trials vs. real-world practice. Therefore, our exploration of whether patients not meeting FLAURA criteria might still benefit from osimertinib was essential. In a study by Viray et al.⁶, 55% of patients were classified as FLAURA-ineligible, and the mOS was 16.5 months in the FLAURA-ineligible group and NR in the FLAURA-eligible group. In our study, 55.3% (266/481) of patients were classified as FLAURA-eligible, whereas 33.1% (159/481) were classified as FLAURA-ineligible, predominantly because of comorbidities (71.7%) and ECOG ≥2 (22.6%). In the FLAURA-ineligible group, the median TTD, rwPFS, and OS were 21.4 months, 15.1 months, and 33.5 months, respectively. In the FLAURA-eligible group, these values were 25.6 months, 20.4 months, and 43.1 months, respectively. However, the mOS in our FLAURA-ineligible group was longer than that in the FLAURA-ineligible group reported by Viray et al.⁶ This discrepancy might be due to differences in classification criteria between studies. The FLAURA-ineligible subgroup in the retrospective study by Viray et al.⁶ had more severe clinical characteristics and did not include active infections as a criterion. Our study included more patients with active infections in the FLAURA-ineligible group, who might have met FLAURA criteria after antimicrobial treatment, thus potentially leading to improved outcomes. Nonetheless, our findings were consistent with those of Viray et al.⁶ in indicating that patients classified as FLAURA-ineligible can still benefit from osimertinib, although the benefits are greater in FLAURA-eligible patients.

Although osimertinib has demonstrated superior efficacy, the considerable variation in the duration of responses among populations emphasized the need to identify subsets of patients with poorer outcomes who might benefit from more aggressive treatments. Previous studies on first-generation EGFR-TKIs have shown that concomitant gene mutations significantly compromise therapeutic efficacy in EGFRm NSCLC. The BENEFIT study¹⁵ has demonstrated shorter PFS in patients bearing concurrent driver mutations (HR = 1.78) or tumor suppressor gene alterations (HR = 2.66) than in patients with EGFR mutations alone. Hu et al.¹⁶ have also reported inferior outcomes in patients with EGFRm NSCLC with co-mutations in PIK3CA, ALK, HER2, ROS1, or KRAS (mPFS: 6.0 vs. 10.9 months; mOS: 17.6 vs. 21.0 months).

Although the predictive value of co-mutations has been well characterized for first-generation TKIs, the effects of co-mutation profiles on osimertinib efficacy remain inadequately defined. In a previous prospective study, patients with concurrent TP53 mutations or MET amplification experienced worse clinical outcomes than patients without these co-mutations when treated with 1L osimertinib in a real-world setting¹⁷. However, whether other co-mutations might affect osimertinib monotherapy outcomes has not been explored prospectively in real-world studies. To our knowledge, our study is the first prospective real-world analysis exploring the relationship between osimertinib efficacy and co-mutations involving other driver and tumor suppressor genes. Baseline NGS data from 118 patients revealed that 90 (76.3%) had additional gene co-mutations (in HER2, KRAS, BRAF, ROS1, ALK, TP53, RB1, or PTEN). Patients with these co-mutations had shorter median TTD and OS than patients with EGFR mutations alone (mTTD: 24.6 months vs. 32.8 months; mOS: 29.3 months vs. NR). Furthermore, TP53 mutations, which are frequently observed in NSCLC, were detected in 80 (67.8%) of the patients who had baseline NGS data. Patients with TP53 co-mutations had shorter median TTD and OS than those without TP53 mutations. Collectively, our findings suggested that co-mutations might limit clinical benefit from osimertinib monotherapy.

The variability in treatment effectiveness among subgroups might arise from distinct molecular mechanisms across different co-mutation subtypes. Tumor suppressor genes such as TP53, RB1, and PTEN lose their normal function, thus resulting in inhibition of apoptosis and acceleration of cell cycle progression, and ultimately promoting tumor growth^18–20. In contrast, driver gene co-mutations, such as MET/HER2 amplifications and KRAS/BRAF mutations, enhance tumor invasiveness by activating alternative signaling pathways^21–24. These findings underscore the importance of exploring alternative therapeutic strategies for patients bearing co-mutations.

Recent studies, including FLAURA-2²⁵ and MARIPOSA²⁶, have evaluated combination therapies for advanced EGFRm NSCLC. These studies have demonstrated that osimertinib combined with chemotherapy or lazertinib combined with amivantamab achieves superior outcomes to osimertinib monotherapy, regardless of patients’ co-mutation status. Specifically, the FLAURA-2 study²⁵ analyzed baseline tissue samples from 141 patients, of whom 79 had TP53 mutations. Patients with baseline TP53 co-mutations who received osimertinib plus chemotherapy had longer mPFS than those receiving osimertinib alone (HR = 0.57). Similarly, in the MARIPOSA study²⁶, amivantamab combined with lazertinib, compared with osimertinib alone, significantly prolonged mPFS in patients with TP53 co-mutations (18.2 vs. 12.9 months; HR = 0.65, P = 0.003). However, because of limited sample sizes, larger prospective studies remain necessary. To address this gap, the ongoing TOP study²⁷ is aimed at evaluating the efficacy and safety of osimertinib plus chemotherapy vs. osimertinib alone as a 1L treatment in patients with EGFR-sensitive NSCLC with concurrent TP53 mutations. The results from the TOP study are highly anticipated. Our exploratory analysis highlighted a possible association between co-mutations and diminished osimertinib efficacy, thus underscoring the need for further evaluation through larger prospective trials.

Previous real-world studies have confirmed the safety profile of osimertinib. In the Reiwa study, the incidence of grade ≥3 AEs was 23.3%. In the FLOWER study¹³, 87.3% of patients experienced AEs. Additionally, a real-world study by Kishikawa et al.²⁸ has reported an AE incidence of 89.3%, with grade ≥3 AEs occurring in 26.8% of patients. Our safety results aligned with these prior reports and identified no new safety signals. Additionally, previous studies did not specifically evaluate safety in FLAURA-ineligible populations. To our knowledge, our study is the first to demonstrate the safety and efficacy of osimertinib in this subgroup. Notably, patients with PS ≥2 exhibited an AE incidence comparable to that in the general population, thus supporting continued use of osimertinib in these patients in real-world clinical practice. The safety profile in the FLAURA-ineligible subgroup was similar to that observed in the FAS population, and no additional safety concerns were identified.

This study has several limitations. First, because of the study’s real-world nature, data were primarily derived from medical records and investigator assessments, thus resulting in missing baseline characteristics for some patients. This aspect reflected routine clinical practice, in which comprehensive documentation was not always performed. The potential for selection bias from such missing data should be acknowledged. Second, in this observational study, subgroup analyses were descriptive, and were neither based on pre-specified statistical hypotheses nor adjusted for multiplicity. Furthermore, certain subgroups had relatively small sample sizes. Therefore, our results provide only preliminary signals and directions and should be interpreted with caution. Further validation is warranted.

In conclusion, osimertinib demonstrated robust efficacy and manageable safety profiles in treatment-naïve patients with EGFRm advanced NSCLC in real-world settings. Patients classified as FLAURA-ineligible also benefited clinically and exhibited tolerable safety profiles. Our exploratory analysis indicated inferior real-world clinical outcomes of 1L osimertinib in patients with EGFRm bearing driver and/or tumor suppressor genes co-mutations, thus suggesting a potential need for more aggressive treatment strategies for this population. However, further prospective studies are required to validate this finding.

Author contributions

Conceived and designed the analysis: Lisha Shen, Dongqing Lv, Ke-Jing Tang, Jianya Zhou, Jianying Zhou.

Collected the data: Lisha Shen, Dongqing Lv, Ke-Jing Tang, Jun Zhao, Dan Zhu, Yanqiu Zhao, Ke Wang, Yan Wang, Zhigang Cai, Ligang Xing, Ke Xie, Jiuwei Cui, Lin Mu, Chao Cao, Liren Ding, Peifeng Chen, Jun Liang, Yongmin Ding, Liqin Lu, Jiliang Zhang, Xinmin Yu, Li Chen, Jing Zheng, Jianya Zhou.

Contributed data or analysis tools: Lisha Shen, Dongqing Lv, Ke-Jing Tang, Jing Zheng, Jianying Zhou.

Performed the analysis: Lisha Shen, Jing Zheng.

Wrote the article: Lisha Shen, Dongqing Lv, Ke-Jing Tang, Jianya Zhou, Jianying Zhou.