Impact of angiogenesis inhibitor eligibility on the prognosis of patients with non‐small cell lung cancer harboring EGFR mutation

Abstract Background Epidermal growth factor receptor tyrosine kinase inhibitors (EGFR‐TKIs) are currently the primary treatment option for patients with EGFR‐mutant non‐small cell lung cancer (NSCLC). However, the effect of EGFR‐TKIs are eventually weakened due to resistance, and there is also a differential efficacy based on EGFR mutation subtypes. The combination of angiogenesis inhibitor (AI) with EGFR‐TKI has shown better efficacy than EGFR‐TKI monotherapy, regardless of the mutation subtypes. Nevertheless, the effect of AI eligibility on overall survival (OS) and progression‐free survival (PFS) remains to be elucidated. Thus, we assessed this impact on patients with NSCLC harboring EGFR mutation. Methods In this study, the data for 450 patients with EGFR‐mutant NSCLC, who were treated with EGFR‐TKI monotherapy, were retrospectively analyzed for AI eligibility. The patients were categorized into AI‐eligible (AI fit) and ineligible groups (AI unfit). Results The median PFS of the AI fit group was 12.9 months, compared to 9.6 months in the unfit group (p = 0.007), and OS was also significantly longer in the AI fit group (median OS = 33.0 months) compared to that in the unfit group (18.5 months, p < 0.001). Multivariate analysis indicated that AI ineligibility was associated with shorter PFS and poor prognosis. Also, in the AI fit group, there was no significant difference in the PFS between EGFR L858R mutation and EGFR exon 19 deletion (median PFS = 11.5 months vs. 13.8 months; p = 0.17). Conclusions From our study, AI eligibility resulted in longer OS and PFS, and also had different effects on patients with EGFR L858R and exon 19 deletion. Since this selection bias may have affected previous clinical trial data on the efficacy of AI combination therapy, their results should be carefully considered henceforth.


| INTRODUCTION
Lung cancer is one of the major causes of cancer-related deaths in the world. Although the prognosis remains limited compared to that of other types of cancers, the evolution of molecular targeted therapy has drastically improved the prognosis, especially in driver mutationpositive non-small cell lung cancer (NSCLC). Epidermal growth factor receptor (EGFR) is one of the most essential oncogenes considered when developing a treatment for EGFR-mutant NSCLC. Treatment with EGFR-tyrosine kinase inhibitors (TKIs) has enabled significantly prolonged outcomes for patients compared to conventional platinum-based combination chemotherapy. EGFR-TKIs became the standard first-line regimen for patients with NSCLC harboring an EGFR mutation. Yet, the benefits of EGFR-TKIs eventually fade because of acquired resistance to TKI treatment. Additionally, previous EGFR-TKI monotherapy studies 1-5 have described a difference in the efficacy of EGFR-TKI usage between different EGFR mutation subtypes and have suggested that tumors with exon 19 deletion (19 del) exhibit a higher sensitivity to EGFR-TKI treatment compared to those with EGFR an exon 21 point mutation (L858R). Thus, novel treatment regimens for NSCLC harboring EGFR mutations are needed to prolong the survival time and reduce the imbalance in the effectiveness between EGFR subtypes.
Recently, the EGFR-TKI plus angiogenesis inhibitors (AIs) combination therapy has demonstrated better efficacy than EGFR-TKI monotherapy in patients with EGFRmutant NSCLC. In the recent NEJ026 study, a randomized open-label phase 3 study, EGFR mutation-positive NSCLC patients who received bevacizumab (a humanized monoclonal antibody to vascular endothelial growth factor [VEGF]) plus erlotinib combination therapy showed significantly prolonged progression-free survival (PFS) compared with those who received erlotinib monotherapy. 6 In addition, administering erlotinib plus ramucirumab (a human monoclonal IgG1 antibody that targets VEGFR-2) shows improved PFS compared with administering erlotinib alone. 7 Although these randomized clinical trials have demonstrated the improvement of PFS in AI combination therapy, these studies have failed to show a survival benefit. Besides their efficacy, AI combination therapies also show a comparable PFS in tumors with an L858R mutation and exon 19 deletion. Based on these results, AI combination therapies are expected to become a compelling treatment option for patients with NSCLC harboring an EGFR mutation, especially the L858R subtype.
However, because of the unique eligibility criteria for AIs, some scientists express concerns about a selection bias in AI combination clinical studies and believe that a better prognosis can be achieved even without the administration of AIs. 8,9 Moreover, whether the eligibility criteria for AI affect the prognosis of NSCLC patients with activating EGFR mutation is also unclear. Thus, in this study, we aimed to assess the prognostic impact of AI eligibility criteria on patients with NSCLC with EGFR mutation and compare the effectiveness of EGFR-TKI monotherapy in patients with an L858R mutation to that in those with exon 19 deletion in the AI-eligible population.

| Study participants
Patients with an EGFR-mutant (L858R or 19 del) NSCLC, who started first-line EGFR-TKI treatment at Shizuoka Cancer Center between 2002 and 2019 were collected retrospectively. We excluded patients whose Eastern Cooperative Oncology Group (ECOG) performance status (PS) was 3 or 4, who had been treated by any AIs, and patients with symptomatic brain metastasis. We defined the enrolled patients as the ALL group. In the ALL group, we defined patients as AI-ineligible if they met at least one of the following conditions: (1) A history of tumor exposure in the bronchus or of producing bloody sputum; (2) a major vessel infiltration (MVI) by the tumor, diagnosed by a radiologist; (3) a history of cardiovascular disease (CVD), including thrombotic diseases, ischemic heart diseases, or congestive heart failure; (4) treatment by chemoradiotherapy before initiation of EGFR-TKI monotherapy. Since there were no cases with active peptic ulcer disease before treatment, we did not include active peptic ulcer disease as the classifying condition. Next, we evaluated efficacy of AI combination therapy, their results should be carefully considered henceforth.

K E Y W O R D S
angiogenesis inhibitor, epidermal growth factor tyrosine kinase inhibitor, non-small cell lung cancer, vascular endothelial growth factor the effects of first-line EGFR-TKI monotherapy in both AI-eligible (AI fit) and ineligible (AI unfit) groups. We also evaluated the efficacies of EGFR-TKI monotherapy in patients with exon 19 deletion or L858R mutation in the AI fit and unfit groups. PFS was defined as the time from the start of EGFR-TKI monotherapy to death, disease progression, or censoring at the last follow-up examination. A considerable number of cases showed several months or years of a treatment-free period due to EGFR-TKI toxicity before disease progression or initiation of sequential treatment. Therefore, patients who started new therapy without confirmation of tumor progression were censored at the time of the latest tumor assessment before the new therapy was initiated. We defined overall survival (OS) as the interval between the initiation of EGFR-TKI therapy, and death from any cause or censoring at the last followup. EGFR mutations in tumor tissues were detected using cobas® EGFR mutation test v2 (Roche Molecular Systems), CycleavePCR ™ Assay (TAKARA, Co., Ltd.), or scorpion arms assay (DxS). Tumor response was evaluated using Response Evaluation Criteria in Solid Tumor criteria version 1.1. 10 This study was authorized by the institutional review board of Shizuoka Cancer Center.

| Statistical analysis
Patient characteristics were compared between the AI fit and unfit groups and patients with L858R and exon 19 deletion, using the Mann-Whitney U test for all continuous variables, and Fisher's exact test for the categorical data. Clinical evaluation of PFS and OS was conducted using the Kaplan-Meier method. We used the Log-rank test to compare the cumulative survival in each group and Cox's proportional hazards analysis for the multivariate analysis.
All p values were two-sided, and p < 0.05 was considered significant. Statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University), which is a converted version that added frequently-used biostatistical functions to an original R commander (version 1.6-3). 11

| Patient characteristics
In total, 535 EGFR major mutation-positive patients had been treated with first-line EGFR-TKIs monotherapy during the study term ( Figure 1). We excluded 85 patients, including 50 patients with PS 3, four with PS 4, seven without PS data, and 24 with symptomatic brain metastasis. Overall, 450 patients were included in the study (ALL group), and the background information is shown in Table 1.
This study included 293 female patients (65.1%), and the median age at the initiation of EGFR-TKI therapy was 71 years (range 31-92). Most patients were diagnosed with stage IV disease (69.8%), and 28.0% relapsed after the surgery. Around 50% of the enrolled patients had a history of smoking, and 45.8% had an L858R mutation. Most patients underwent gefitinib treatment (60.7%), followed by erlotinib (18.9%), osimertinib (13.6%), and afatinib (6.9%). As per the four factors of AI eligibility, 113 patients were defined as AI-ineligible, of which 51 patients had a history of tumor exposure in the bronchus or of producing bloody sputum, 17 had MVI, 45 had a history of CVD, and 16 had been treated with chemoradiotherapy before starting the EGFR-TKI treatment. Both groups (AI fit group and AI unfit group) showed similar patient characteristics regarding sex, smoking history, EGFR subtype, and first EGFR-TKI drug. However, there was a significant disproportion in the two groups based on ECOG PS (p = 0.03) and age (p = 0.019).

F I G U R E 1
Flow chart of all the patients. AI, angiogenesis inhibitor; EGFR, epidermal growth factor receptor; NSCLC, non-small cell lung cancer; TKI, tyrosine kinase inhibitor
In the multivariate analysis of the OS, AI eligibility (HR = 0.60, 95% CI: 0.47-0.77, p < 0.  T A B L E 1 Characteristics of all patients and those in the AI fit and unfit groups four AI eligibility criteria, a history of tumor exposure in the bronchus or producing bloody sputum was related to a shorter PFS (HR = 1.43, 95% CI: 1.04-1.97, p = 0.027), and was indicated to be a significant prognostic factor for the OS (HR = 1.61, 95% CI: 1.14-2.28, p = 0.007).

| DISCUSSION
In this study, we assessed the effect of AI eligibility in patients with an EGFR-mutant NSCLC who had been treated with EGFR-TKI monotherapy and evaluated the impact of AI eligibility for different EGFR mutation subtypes. As far as we know, this is the first study to assess the impact of AI eligibility between different EGFR subtypes. Several factors, such as sex, history of smoking, EGFR subtypes (common/uncommon), ECOG PS, and clinical stages are related to EGFR mutation-positive NSCLC prognosis. 12,13 Studies have shown that a selection bias in bevacizumab eligibility in patients with NSCLC treated with chemotherapy, 8 and that eligibility for bevacizumab use is independently associated with OS in patients with EGFR-mutant NSCLC. 14 In accordance with these studies, we also observed significant differences in both PFS and OS of patients in AI fit and unfit groups, indicating that AI eligibility is related to prolonged PFS and OS in patients with NSCLC harboring an EGFR mutation. Therefore, we believe that there is a possibility that the results described in previous AI combination studies may have overestimated median PFS data or median OS data because of the selection bias due to AI eligibility. Our results also explain why erlotinib monotherapy in the NEJ026 study resulted in a relatively longer median PFS (13.3 months) and OS (46.2 months) than other erlotinib monotherapy studies (PFS = 10-13 months, OS = 23 months), 15,16 and even a longer OS than osimertinib monotherapy in the FLAURA study 17 (38.6 months). Since several studies evaluating AI F I G U R E 3 Kaplan-Meier curves for (A) progression-free survival and (B) overall survival in the AI fit group, and (C) progression-free survival and (D) overall survival in the AI unfit group. AI, angiogenesis inhibitor; CI, confidence interval; HR, hazard ratio; OS, overall survival; PFS, progression-free survival plus EGFR-TKI combination therapies are ongoing, it is necessary to carefully interpret the PFS and OS data in these studies.
Previous phase III studies and meta-analyses have shown that patients with an exon 19 deletion benefit more from EGFR-TKI treatment than those with an L858R mutation. Accordingly, EGFR mutation subtype has been considered as an important stratified factor in recent clinical trials for patients with NSCLC harboring EGFR mutations. [1][2][3][4][5] There are several rationales for this difference, including the prevalence of de novo T790M resistance mutation, tumor heterogeneity, difference in protein structure, and EGF-induced tyrosine phosphorylation patterns. [18][19][20] However, the multivariate analysis of the AI fit group showed that the EGFR subtype was not an independent factor for both PFS and OS. Our results imply that the prognosis of EGFR subtypes is comparable for patients that are potentially eligible for AI, irrespective of the AI used. Similar results can be observed in the NEJ 026 and RELAY studies, which demonstrated an equivalent median PFS in subgroups of EGFR mutations in the erlotinib monotherapy group. 6,7 In accordance with these results, it may be necessary to re-evaluate the actual benefit of AIs and determine if the combination of AI with erlotinib could increase the rate of severe adverse events and mortality. 21,22 In this study, we also assessed the effects of all four factors of an AI fit condition. The multivariate analysis demonstrated that among the four factors, a history of tumor exposure in the bronchus or producing bloody sputum significantly affected both PFS and OS. This factor was likewise associated with PFS and OS in the L858R subgroup, but not in the 19 del subgroup. The presence of bloody sputum before treatment is reported to be a strong prognostic factor for advanced non-squamous NSCLC. 8 However, why the presence of bloody sputum and tumor exposure in the bronchus has a different impact on the L858R mutation and exon 19 deletion is yet to be elucidated. Some reports indicate that the location of the tumor, which could be related to the proportion of tumor exposure in the bronchus, is associated with PFS and OS in adenocarcinomas. 23 Although our analysis showed that a central tumor causes a poor PFS and OS, the proportion of central or peripheral tumors did not differ between the L858R and exon 19 deletion groups. Thus, further considerations are required to clarify the mechanism underlying the difference in sensitivity between the EGFR subtypes.
The limitations of this study must be acknowledged. First, since this is a single-center, retrospective study, there is a possibility of an unintentional selection bias. Further, unlike previous phase III studies and meta-analyses of Abbreviations: Bloody sputum, tumor exposure in the bronchus or producing bloody sputum; CI, confidence interval; CRT, chemoradiotherapy; CVD, cardiovascular disease; first-, second-gen, first-and secondgeneration EGFR-TKI; HR, hazard ratio; MVI, major vessel infiltration; PFS, progression-free survival; PS, performance status; TKI, tyrosine kinase inhibitor.
Bold values are p <0.05, which is statistically significant.
EGFR-TKI treatment, the PFS in the ALL group was equivalent between 19 del and L858R subgroups, which might have affected the result for the AI fit and unfit groups.

| CONCLUSION
Our study showed that the eligibility for AIs resulted in a longer PFS and OS and suggested that selection bias for AI eligibility could impact the selection of patients with NSCLC harboring an EGFR mutation. Moreover, the impact of AI eligibility may differ based on the EGFR subtype, which could, in turn, result in an unexpected selection bias in clinical trials and an overestimation of results. Since there is a possibility that such a selection bias affected previous trials evaluating the efficacy of EGFR-TKI in combination with AI, the results of these trials should be carefully evaluated henceforth.