Treatment Options of First-Line Tyrosine Kinase Inhibitors and Subsequent Systemic Chemotherapy Agents for Advanced EGFR Mutant Lung Adenocarcinoma Patients: Implications From Taiwan Cancer Registry Cohort

Objectives Large-scale, population-based real-world studies on the treatment outcomes of first-line tyrosine kinase inhibitors (TKIs) and subsequent systemic chemotherapy agents for lung adenocarcinoma (with activating epidermal growth factor receptor [EGFR] mutations) remain limited. Materials and Methods From March 2014 to December 2016, patients with advanced lung adenocarcinoma, identified from the Taiwan Cancer Registry were included in this study if they received any of the three TKIs as first-line treatment. The primary outcome was overall survival (OS). The secondary outcome was time-to-treatment discontinuation (TTD). Results A total of 4,889 patients (median age: 67 years and two-thirds with distant metastasis) were recruited (1,778 gefitinib, 1,599 erlotinib, and 1,512 afatinib users). A 1:1 propensity score (PS)-matched cohorts of 1,228 afatinib/erlotinib and 1054 afatinib/gefitinib was created. After PS matching, it was found that afatinib was not associated with better OS (afatinib vs. erlotinib, HR: 0.96, 95% CI: 0.86–1.07; afatinib vs. gefitinib, HR: 0.91, 95% CI: 0.81–1.02). In the subgroup analysis, afatinib demonstrated a survival benefit in patients with active smoking (afatinib vs. erlotinib, HR: 0.69, 95% CI: 0.51–0.93; afatinib vs. gefitinib, HR: 0.67, 95% CI: 0.48–0.94) and ECOG > 1 (afatinib vs. erlotinib, HR: 0.79, 95% CI: 0.63–0.99; afatinib vs. gefitinib, HR: 0.78, 95% CI: 0.62–0.98). A total of 41.1% (n = 1992) of first-line TKI users received subsequent chemotherapy. Among the three TKI groups, pemetrexed usage was associated with better OS compared with other chemotherapy agents, with the exception of gemcitabine in the afatinib and gefitinib groups. Pemetrexed and gemcitabine had the longest TTD of 3–4 months. Conclusions Among patients with EGFR mutant lung adenocarcinoma, afatinib use may not provide longer OS compared with first-generation TKIs. Afatinib may be preferably considered among patients with active smoking and should not be withheld among those with worse performance status. With 40% of patients receiving subsequent chemotherapy, pemetrexed may be the preferred agent, while gemcitabine can be a reasonable alternative.


INTRODUCTION
Lung cancer is the leading cause of cancer-related deaths in the 21 st century (1,2). Adenocarcinoma is the major histological type of non-small cell lung cancer (NSCLC), but the standard care for patients with metastatic NSCLC has shifted from traditional platinum-based doublets to precision targeted therapy to the driver genes with mutations, such as the epidermal growth factor receptor (EGFR), anaplastic lymphoma kinase (ALK), ROS-1, and BRAF (3,4). Targeting lung adenocarcinoma with EGFR mutations among Asians is important because they have a significantly higher prevalence of the EGFR mutation compared with the Caucasians (5-7). Multiple generations of EGFR tyrosine kinase inhibitors (TKIs) have been effective as first-line therapy for advanced EGFRmutant lung adenocarcinoma patients (8)(9)(10)(11)(12).
Gefitinib, erlotinib, and afatinib are widely prescribed firstline TKIs worldwide. All of them provide robust and similar effects in advanced EGFR-mutant lung adenocarcinoma patients (11,13,14). Although afatinib has minimal clinical significance in progression-free survival (PFS) compared with gefitinib (median PFS 11.0 vs. 10.9 months, respectively) in the first-line setting (15), it did not show improved overall survival (OS) compared with gefitinib (13,14). Recently, several real-world studies have investigated the characteristics and clinical effectiveness of these three EGFR TKIs administered in advanced lung adenocarcinoma patients (16)(17)(18)(19)(20). However, the conclusions from these studies may not provide convincing evidence for clinical practice because of their limited case numbers, discrepant recruitment time, disproportional populations in which TKIs were used, and lack of information on subsequent therapy after first-line EGFR TKI failure (16)(17)(18)(19)(20).
Prolonging cancer patients' OS is a major goal of all cancer treatments, and understanding the optimal treatment sequences is a key factor that allows patients to live longer. p.T790M in the EGFR gene is the most common acquired resistance mechanism following first-line TKI treatment (21), and osimertinib proved to be effective in patients with the EGFR p.T790M mutation as a standard second-line treatment (22). Owing to the unavailability of osimertinib in some situations, for cases without acquired p.T790M or accessible tumor tissues for re-biopsy, chemotherapy remains an important subsequent therapy after first-line TKI (23,24). Furthermore, only, few studies have investigated the optimal regimen of chemotherapy as second-line treatment in patients who are p.T790M negative or have an unknown acquired resistance mechanism after first-line TKI failure (25,26).
This study, therefore, aimed to investigate the treatment sequences and clinical outcomes of treatment-naïve, EGFRmutant advanced lung adenocarcinoma patients receiving TKIs in a real-world, population-based setting. Additionally, we explored the prognostic factors of TKI users and treatment durations of individuals after they underwent subsequent chemotherapy. Our results were informative with respect to clinical decision-making.

Ethics Statement
This study was approved by the Institutional Review Board (IRB) committee of the National Taiwan University Hospital Hsinchu Branch (NTUH-HC REC: 105-040-E). The IRB waived the requirement of informed consent because the utilized data were de-identified in this study.

Study Design and Population
This study used the Taiwan Cancer Registry (TCR), which is a population-based registry system that includes 90% of all cancer patients in Taiwan (27,28). We identified patients with advanced lung adenocarcinoma, including those in stages IIIb and IV (M1a and M1b) from the TCR during March 2014 and December 2016. Patients were included if they received gefitinib, erlotinib, or afatinib as first-line treatment within 60 days after diagnosis. Patients were excluded if they received chemotherapy prior to first-line TKI therapy. In Taiwan, gefitinib, erlotinib, and afatinib have been sequentially reimbursed by the Taiwan National Health Insurance (NHI) as first-line therapy for advanced EGFR-mutant lung adenocarcinoma since June 2011, November 2013, and May 2014, respectively (29). Considering that the study period could be an important confounding variable, which could strongly influence the outcome by improving lung cancer treatment, we truncated our dataset to the date of afatinib approval for use (May 2014) in Taiwan.
During the study period, physicians applied for TKI use prior to TKI initiation, and the application was reviewed by the experts of the NHI committee. EGFR mutation results, clinical images, pathology, and clinical information were provided along with the application for TKIs. For every 3 months, physicians provided the imaging evidence of partial remission or stable disease to allow further TKI use (https://www.nhi.gov.tw/).
After TKI use, the recruited cohort was then followed, and mortality was confirmed using mortality data from the Department of Statistics, Taiwan. Underlying diseases, TKI use, and duration were ascertained from the Taiwan NHI database (28,30,31). Using the linkage between the abovementioned databases, we longitudinally followed our cohort patients till December 31, 2017.

Data Collection and Definition
TNM staging data at diagnosis available in the TCR were made according to the International Association for the Study of Lung Cancer 7th edition lung cancer staging system (32). Accordingly, the metastasis (M) category of stage IV lung cancer was subdivided into M1a for cases with intra-thoracic metastases (including pleural seeding, malignant pleural/pericardial effusion, and contralateral pulmonary nodules) and M1b for cases with distant metastases (32). The patients' performance status was represented as the Eastern Cooperative Oncology Group (ECOG) scores (33). Meanwhile, the Charlson comorbidity index (CCI) was used to assess the patients' comorbidities using the NHI claims data (34), but malignancyrelated score was excluded (cancer-free CCI) as it was previously reported (27). Hospital levels were classified hierarchically into medical centers, regional hospitals, and local hospitals (35). The defining codes for lung adenocarcinoma in the TCR in the NHI are summarized in Supplementary Table S1. We also categorized second-line chemotherapy agents into pemetrexed, gemcitabine, paclitaxel, docetaxel, vinorelbine, and others.

Statistical Analyses
We used proportions or means to describe the demographics and clinical characteristics of the patients. Categorical variables were analyzed using chi-square tests. One-way ANOVA or Student's t-test was applied for continuous variables. The cohort entry date was that of diagnosis. OS, the primary outcome, was defined as the period from the date of diagnosis to death. Participants were censored if they were still alive at the end of the study period (December 31, 2017). The secondary outcome was time-totreatment discontinuation (TTD), which was defined as the interval between the date of TKI treatment or chemotherapy initiation and discontinuation. The BMIs were missing for 8% of the patients, but we still considered it important to include BMIs in the final analysis. We imputed the missing values of BMI by age and sex with linear regression methods.
The propensity score (PS) for the probability of TKI administration was derived using a logistic regression model, which included potential confounders such as, age, sex, ECOG, BMI, cancer staging, smoking, alcoholism, CCI, year of TKI use, and hospital level. A 1:1 matched cohort group of afatinib/ erlotinib and afatinib/gefitinib was created. Variables that remained significantly different after PS matching were further adjusted in the final model. In this study, only the categorical BMI groups were imbalanced among the different TKI groups, while the absolute values of BMIs were not different between the matched groups.
Subgroup analysis was performed among the different age groups, BMI groups, ECOG groups, sexes, smoking habit, and stages (IIIb, M1a, and M1b). We also compared the OS of five common chemotherapy agents, including pemetrexed, gemcitabine, vinorelbine, paclitaxel, and docetaxel, among the three TKI groups using multivariate Cox regression.
We used SAS (version 9.4; SAS Institute Inc., Cary, NC, USA) for data analyses. A p value of < 0.05 on a two-sided test was considered statistically significant.

Demographics and Clinical Variables of the Study
Between May 2014 and December 2016, a total of 4,889 advanced lung adenocarcinoma patients with the EGFR mutation receiving TKIs (including 1,778 gefitinib, 1,599 erlotinib, and 1,512 afatinib) as first-line therapy were included in our study ( Figure 1). The baseline characteristics of the enrolled patients are summarized in Table 1. The median age of all patients was 67 years, and the majority was female (n = 3,083, 63.1%). Meanwhile, 4,669 (95.5%) patients had stage IV disease. Most eligible patients had relatively good performance status (ECOG ≦ 1: n = 3,780, 77.3%) and had never smoked (n = 3,684, 75.3%).
Comparing the OS of different second-line chemotherapy regimens, pemetrexed was associated with better OS than was vinorelbine in gefitinib users with advanced EGFR mutant lung adenocarcinoma (Ref: pemetrexed, HR: 1.65, 95% CI: 1.28-2.13,

DISCUSSION
Our study showed that afatinib did not provide the evidence of a survival advantage over gefitinib and erlotinib. In the subgroup analysis, afatinib was associated with better OS among patients with active smoking and poor performance status. While 40% of patients were able to receive second-line chemotherapy agents, pemetrexed was associated with better OS across the three TKI groups. An alternative choice may be gemcitabine.
To the best of our knowledge, this study is the single largest cohort study to investigate the effectiveness of three EGFR TKIs (16,18). We found that compared with first-generation TKIs, patients receiving afatinib were younger, were more likely to be male, had higher BMIs, and had better performance status. Previous studies have shown that afatinib provided longer PFS than first-generation TKIs (15,18,20), but adverse effects in patients receiving afatinib were also more frequently observed (15). In real-world practice, our research showed that the baseline characteristics could significantly influence the clinicians' judgments and preferences while choosing one of the TKIs (gefitinib, erlotinib, or afatinib) as first-line treatment. Afatinib may be preferred among those who are younger, are male, have higher BMIs, are active smokers, and have better performance status. Interestingly, we found that the proportion of afatinib users among all TKI users had increased in the recent years. In 2016, the number of afatinib users surpassed either  erlotinib or gefitinib users. As physicians became experienced in managing patients with afatinib (especially the toxicity profiles), they selected afatinib over erlotinib or gefitinib.
In PS matching analysis, afatinib use was not associated with better OS compared with erlotinib or gefitinib use. In the subgroup analysis, afatinib use was associated with survival benefit among patients with ECOG > 1 and active smoking. While one may argue that failure to demonstrate the clinical evidence of survival benefit may be due to the relatively small sample size in previous randomized control trials and other observational studies, our study may have the current largest cohort, including more than 1,000 participants in each TKI group (14,16,17,36,37). More importantly, we used PS matching, which is a more robust way of controlling confounders in observational studies, and this analysis strategy was not performed in previous observational studies (38). Clinically, afatinib could be an effective treatment for lung adenocarcinoma patients with the EGFR mutation and brain metastasis (39). Subgroup analysis showed that afatinib provided better OS in patients with distant metastases (stage M1b) compared with gefitinib, but this benefit was not observed when compared with the erlotinib group. Active/current smokers usually have lower EGFR mutation rates (especially exon 19 deletion and p.L858R) than never-smoking female patients (8,40). Notably, our study (through subgroup analysis) showed that afatinib could provide significantly longer OS in active smokers than could gefitinib/erlotinib. The above-mentioned benefits may be because afatinib is a member of the pan-ErbB family of inhibitors. It could covalently and irreversibly bind to the intracellular tyrosine kinase domain of the EGFR and effectively treat common (exon 19 deletion and p.L858R) and uncommon EGFR mutations (41)(42)(43). Furthermore, this real-world study provided additional information on the minor population with a poor performance status (ECOG >1), which is often excluded by randomized controlled trials to evaluate the efficacy and safety of the study drugs. None of the patients (among the 310 patients) with ECOG > 1 in the phase IIb LUX-Lung 7 trial (afatinib vs. gefitinib) were enrolled (19), and only 6 (2.3%) patients with ECOG = 2 (among 256 patients) in the phase III CTONG 0901 trial (gefitinib vs. erlotinib) were included (13). In contrast, there were 1,109 patients (up to 22.7% of total 4,889 patients) with ECOG > 1 in real-world practice, and afatinib surprisingly demonstrated superior TTD and OS benefits compared with first-generation TKIs in patients with worse performance status.
Approximately 40%-60% of acquired EGFR p.T790M develops after the patients receive first-line TKIs (44,45), and osimertinib was approved as second-line treatment by the US Food and Drug Administration (FDA) and Taiwan Food and Drug Administration in November 2015 and November 2016, respectively. In the FLAURA trial, 14.1% (39 of 277) of patients received chemotherapy and 30.7% (85 of 277) of patients in the gefitinib/erlotinib group received osimertinib as second-line treatment (46). Osimertinib, however, was not reimbursed by the NHI during the study period. Shifting to osimertinib treatment after first-line TKI failure, therefore, was not widely used during our study period in Taiwan, and platinum-based doublets remained the standard for second-line treatment. Our real-world study indicated that 41.1% of all TKI patients in Taiwan could have subsequent cytotoxic chemotherapy as an effective treatment. While chemotherapy may be the most important and preferred systemic therapy after the failure of first-line osimertinib treatment, only 32.3% (90 of 279) of patients in the osimertinib group of the FLAURA trial received chemotherapy as a subsequent systemic therapy (46). In our study, pemetrexed (54.6%) and vinorelbine (21.7%) were the most common subsequent chemotherapy agents. Pemetrexed was the most preferred subsequent therapy in clinical practice owing to its efficacy, tolerability, and convenience in administration (47), and vinorelbine was also frequently prescribed because of the oral route of administration and less toxicity in elderly patients (48). In this real-world study, only 56.2% of patients used platinum-based doublets as subsequent chemotherapy agents after TKI failure. Interestingly, pemetrexed and gemcitabine were found to be the most common partners for platinum drugs. Meanwhile, pemetrexed as a subsequent therapy could provide the best TTD benefit among all agents in erlotinib users. Furthermore, pemetrexed and gemcitabine demonstrated similar effectiveness in TTD among gefitinib and afatinib users. These findings from our claims database epidemiological studies could provide personalized guidance in clinical practice, complementary to biomarker and genetic risk factor studies for oncological patients.
There were some limitations in our study. First, detailed information on EGFR mutation sites was not available in the TCR database. Therefore, the effectiveness of different generation TKIs could not be compared with common or uncommon mutations. Meanwhile, the causes of TTD and TKI-related toxicity profiles could not be readily clarified. In the subsequent treatment analysis, osimertinib was not reimbursed by the Taiwan NHI. Therefore, self-financed or clinical trial osimertinib users could not be identified in this study. Finally, the FLAURA trial demonstrated the superior efficacy and safety of osimertinib compared with gefitinib and erlotinib as first-line TKIs in EGFR mutant NSCLC patients, and osimertinib is therefore currently considered the standard for first-line therapy (46). However, data regarding the activity of osimertinib in patients harboring rare EGFR mutations are limited. Economic issues, such as high cost and the lack of insurance reimbursement may preclude osimertinib use in realworld. Meanwhile, the optimal therapeutic strategy for disease progression after osimertinib administration may still be ambiguous for physicians because of the lack of large-scale realworld data. Gefitinib, erlotinib, and afatinib are, therefore, still used as first-line treatment in many EGFR mutant NSCLC patients. Our study indicates that despite the increasing use of afatinib as first-line TKI for EGFR mutant, late-stage adenocarcinoma patients, afatinib use was not associated with longer OS than were first-generation TKIs. Afatinib administration, however, may be considered among active smokers. Additionally, for patients with poor performance status, afatinib administration may also lead to survival benefits and should not be withheld due to the fear of toxicity. For second-line chemotherapy, pemetrexed may be the preferred agent, and gemcitabine can also be considered as a reasonable alternative.

DATA AVAILABILITY STATEMENT
All datasets presented in this study are included in the article/ Supplementary Material.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Institutional Review Board (IRB) committee of National Taiwan University Hospital Hsinchu Branch. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.