Distinct Clinical Outcomes of Non-Small Cell Lung Cancer Patients with Epidermal Growth Factor Receptor (EGFR) Mutations Treated with EGFR Tyrosine Kinase Inhibitors: Non-Responders versus Responders

Introduction Treatment with epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs) has been associated with favorable progression free survival (PFS) in patients with non-small cell lung cancers (NSCLC) harboring EGFR mutations. However, a subset of this population doesn't respond to EGFR-TKI treatment. Therefore, the present study aimed to elucidate survival outcome in NSCLC EGFR-mutant patients who were treated with EGFR TKIs. Methods Among the 580 consecutive NSCLC patients who were treated at our facility between 2008 and 2012, a total of 124 treatment-naïve, advanced NSCLC, EGFR-mutant patients treated with EGFR TKIs were identified and grouped into non-responders and responders for analyses. Results Of 124 patients, 104 (84%) responded to treatment, and 20 (16%) did not; and the overall median PFS was 9.0 months. Notably, the PFS, overall survival (OS) and survival rates were significantly unfavorable in non-responders (1.8 vs. 10.3 months, hazard ratio (HR) = 29.2, 95% confidence interval (CI), 13.48–63.26, P<0.0001; 9.4 vs. 17.3 months, HR = 2.74, 95% CI, 1.52–4.94, P = 0.0008; and 58% vs. 82% in 6, 37% vs. 60% in 12, and 19 vs. 40% at 24 months, respectively). In multivariate analysis, treatment efficacy strongly affected PFS and OS, independent of covariates (HR = 47.22, 95% CI, 17.88–124.73, P<0.001 and HR = 2.74, 95% CI, 1.43–5.24, P = 0.002, respectively). However, none of the covariates except of the presence of EGFR exon 19 deletion in the tumors was significantly associated with better treatment efficacy. Conclusions A subset of NSCLC EGFR-mutant patients displayed unfavorable survival despite EGFR TKI administration. This observation reinforces the urgent need for biomarkers effectively predicting the non-responders and for drug development overcoming primary resistance to EGFR TKIs. In addition, optimal therapeutic strategies to prolong the survival of non-responders need to be investigated.


Introduction
Lung cancer, which is the most common cause of cancer deaths worldwide, is generally associated with poor prognoses. Recently, advances in personalized medicine have modestly improved treatment efficacy, toxicity and survival in subsets of lung cancer patients. Epidermal growth factor receptor (EGFR) mutation status has been shown to be significantly associated with tumor response to EGFR tyrosine kinase inhibitors (TKIs) [1,2], leading to the routine assessment of the presence of EGFR mutations in advanced non-small cell lung cancers (NSCLC), particularly adenocarcinomas [3,4]. Furthermore, EGFR TKIs have been recommended as first-line treatment for patients with advanced NSCLC that contain EGFR mutations due to the clinical benefits of these novel anti-tumor agents.
Prospective clinical trials have clearly demonstrated that EGFR TKIs are effective therapeutics that carry a 60-82% response rate [2,[5][6][7] and improve progression-free survival (PFS) with 7.7-13.3 months in NSCLC EGFR-mutant patients [2,[5][6][7]. However, 20-40% of NSCLC patients do not experience tumor reduction following EGFR TKI administration despite the presence of EGFR mutations in their tumors. This issue has not been well addressed. Specifically, PFS in NSCLC EGFR-mutant patients whose tumors do not significantly shrink after targeted therapy is rarely reported, contributing to the lack of comprehensive information about the treatment outcome of this subset of NSCLC patients.
In the present study, we aimed to determine survival outcome in treatment-naïve NSCLC patients whose tumors harbored EGFR mutations and who were treated with EGFR TKIs as first-line therapy, with a focus on comparing non-responders to responders.

Case Identification
We retrospectively reviewed the medical records of 580 consecutive patients who were histologically or cytologically diagnosed of NSCLC, including adenocarcinoma, squamous cell carcinoma (SCC) or NSCLC not otherwise specified (NOS), and treated at Taipei Medical University Hospital between January 2008 and November 2012, with an approval from the Joint Institutional Review Board (JIRB) of Taipei Medical University, Taipei, Taiwan (Approval number: 201108006). Additionally, the JIRB also waived the need for written informed consent from the patients. Patients with NSCLC that harbored EGFR mutations and who received EGFR TKIs (either gefitinib or erlotinib) as front-line treatment for advanced (stage IIIb or IV) NSCLC were eligible for these analyses. Patients with NSCLC that did not harbor EGFR mutations or NSCLC in which the EGFR mutation status was uncertain were excluded from the analyses. A patient who had NSCLC that contained any mutations in exons 18-21 of the EGFR gene was defined as an EGFR mutant. Patients who had previously received chemotherapy, had taken EGFR TKIs for less than 14 days, did not receive follow-up imaging studies, such as chest tomography (CT) scans or chest films, during the period of EGFR TKI administration, or had more than 1 primary cancer were excluded from the study.

Variables
Demographic and clinical characteristics, including gender, age at diagnosis of lung cancer diagnosis or recurrence (cutoff at 60 years), smoking status (never vs. former or current), subtype of NSCLC histology (adenocarcinoma, SCC, NSCLC-NOS), stage (3b vs. 4b), and subtype of EGFR exon 18-21 mutations were collected. Additionally, Eastern Cooperative Oncology Group (ECOG) performance status (PS) at EGFR TKI administration, and response to EGFR TKI treatment (responder vs. nonresponder) were also collected. In this study, follow-up time, PFS and overall survival (OS) were calculated from the date of EGFR TKI administration to the last follow-up, to the date of disease progression, and the date of death or the last follow-up, respectively. Patients whose NSCLC did not progress at the last follow-up were censored at the date of their last contact with our institution.

Assessment of Response (Efficacy)
Treatment effectiveness and disease progression were determined using RECIST criteria [8]. Patients who were either in complete remission or who displayed a partial response were categorized as responders, and those with either stable disease or disease that had progressed were categorized as non-responders.

Statistical Analyses
Frequencies and descriptive statistics on demographic and clinical characteristics were obtained. PFS and OS were estimated using the Kaplan-Meier method and the difference in survival between the subgroups was compared using log-rank test. The association of demographic and clinical characteristics with PFS and OS was evaluated using univariate and multivariate Cox regression. The associated factors with treatment efficacy of EGFR TKI were identified by univariate and multivariate logistic regression. The result was presented as odds ratio (OR) for logistic regression or hazard ratio (HR) for Cox regression with their corresponding 95% confidence intervals (CI). All of the data analyses were conducted using SPSS software version 18 (SPSS Inc, Chicago, Illinois).

Results
A total of 124 NSCLC, including 121 adenocarcinoma, 2 NSCLC-NOS and 1 SCC, patients who received EGFR TKIs as the front-line treatment for their advanced NSCLC with EGFR mutations were identified for the analyses (Figure 1), with a mean age: 68.2613.0 years, and median follow-up time: 9.8 months (inter-quartile rage = 4.8-16.1 months). Sixty-three (50.8%) patients were alive at the last follow-up. The patient characteristics are listed in Table 1. Female gender, never smokers, and young age ,60 years represented 66%, 79% and 23% of the patients, respectively. Moreover, the majority of eligible patients was at stage IV and displayed good performance status (ECOG PS 0 -2). In addition, 92% of 124 patients possessed either gene deletion in EGFR exon 19 or point mutation in EGFR exon 21.
The overall treatment efficacy of EGFR TKIs was 84% (104/ 124), and twenty patients (16%) did not experience significant tumor shrinkage despite EGFR TKI administration, as shown in Table 1. This observation implied that individuals with EGFR mutations in their tumors may not consistently experience clinical benefits, such as longer survival, following EGFR TKI treatment. Therefore, we compared PFS in non-responders to PFS in responders. The overall median PFS of 124 eligible patients was 9.0 months (25 th and 75 th percentiles were 16.5 and 5.2 months, respectively) ( Figure 2A). Notably, PFS was significantly shorter in nonresponders (1.8 vs. 10.3 months, HR = 29.2, 95%CI = 13.48-63.26, P,0.0001) ( Figure 2B). We further analyzed 101 patients with ECOG PS of 0-2 and found that the difference between the two groups remained significant (2.0 months in non-responders and 11.5 months in responders, P,0.0001). This analysis clearly demonstrated that a proportion of NSCLC EGFR-mutant patients did not display favorable PFS in spite of EGFR TKI administration.
To weigh the impact of baseline variables and treatment response on the subsequent survival outcome, multivariate analyses were conducted and revealed that response to EGFR TKIs was significantly associated with favorable PFS and OS, independent of age, performance status and subtype of EGFR mutation (HR = 47.22, 95% CI = 17.88-124.73, P,0.001, and HR = 2.74, 95% CI = 1.43-5.24, P = 0.002, respectively) as shown in Tables 2 and 3. These findings indicated that the treatment response could be translated into subsequent survival outcome, and suggested the importance of predictors of treatment efficacy for EGFR TKIs in NSCLC EGFR-mutant patients.
To investigate the baseline clinical variables that could predict treatment response to EGFR TKIs, logistic regression models were used to determine the relationships between these factors and treatment response. In multivariate analyses (Table 4), the presence of EGFR exon 19 deletion in tumors was significantly associated with better treatment response (OR = 5.58, 95% CI = 1.30-23.93, P = 0.021), independent of the remaining clinical variables, including age, gender, history of smoking, stage, and performance status.

Discussion
In the present study, we reconfirmed that a subset of NSCLC patients who were treated with EGFR TKIs did not experience marked tumor shrinkage despite the presence of EGFR mutations and clearly demonstrated that PS and OS strikingly differed in non-responders and responders (1.8 vs. 10.3 months, P,0.0001 and 9.4 vs. 17.3 months, P = 0.0008, respectively). These differences in PFS and OS mostly resulted from tumor response to treatment. However, pre-treatment clinical variables except the subtype of EGFR mutations failed to predict treatment efficacy. These findings emphasize the need for further studies investigating novel biomarkers that can predict EGFR TKI treatment efficacy, particularly the non-responders, and the primary mechanisms underlying resistance in NSCLC containing EGFR mutations. The introduction of EGFR TKIs as first-line treatment has been generally accepted to improve tumor response rates and PFS in patients with advanced NSCLC harboring EGFR mutations compared to standard chemotherapy. Historically, PFS in this population has been reported to be approximately 7.7-13.3 months [2,[5][6][7]. In the present study, median PFS was 9.0 months for all eligible patients. It is reasonable to expect that the patients who did not experience significant tumor shrinkage after treatment would have unfavorable prognoses. However, to the best of our knowledge, PFS in non-responders, which accounted for a proportion (20-40%) of all of the enrolled patients in these breakthrough clinical trials [2,[5][6][7], has not been widely reported, contributing to the lack of information about the survival outcome of this subset who did not respond to EGFR TKI administration. In the present study, we did not intend to re-emphasize the overall advantages from EGFR TKI treatment in NSCLC EGFR-mutant patients compared to conventional chemotherapy. Instead, we focused on elucidating the distinct survival outcomes of nonresponders and responders who were treated with EGFR TKIs as first-line treatment for their NSCLC with EGFR mutations.
In the present study, we demonstrated that PFS strikingly differed in non-responders and responders (1.8 vs. 10.3 months, P,0.0001) despite the fact that both groups possessed EGFR mutations and received EGFR TKIs as their first-line treatment. Moreover, our analyses revealed that OS differed significantly in both groups (9.4 vs. 17.3 months, P = 0.0008). These distinct survival outcomes in non-responders and responders prompted us to consider several critical issues that are encountered in clinical practice. For example, the process that should be used to select optimal candidates who possess EGFR mutations and will experience favorable tumor response to EGFR TKIs remains unclear. We do not know whether non-responders can be successfully identified prior to EGFR TKI administration or soon after treatment. Moreover, which regimen of cytotoxic chemotherapy is superior in non-responders when they are recognized before or after the start of EGFR TKI treatment, remain unknown. The discrepancies in survival benefits following EGFR TKI administration among NSCLC EGFR-mutant patients may result from both differences in pre-treatment variables and tumor response to treatment. In our analyses, however, none of the baseline clinical factors was identified to be significantly associated with PFS and OS, with the exception of tumor response to TKIs, performance status and subtype of EGFR mutation. These findings repeatedly highlighted the importance of factors that can predict therapeutic efficacy. Disappointingly, neither demographic data nor baseline pre-treatment clinical factors was demonstrated to effectively discriminate non-responders from responders prior to treatment in the current analyses. Collectively, our findings revealed that clinical outcomes in NSCLC EGFR-mutant patients who were treated with EGFR TKIs were distinct, and imply that reproducible and reliable biomarkers that clearly predict treatment efficacy are crucial for the treatment of NSCLC EGFR-mutant patients. The development of novel drugs that can overcome primary EGFR TKI resistance is of equal importance.
Some pilot studies have reported that intrinsic factors in EGFRmutant lung cancer cells, such as mutations in T790M, PI3CA, or KRAS and activation of the intracellular Fas/NF-kB signaling pathway, may confer primary resistance to EGFR TKIs [9][10][11]. In addition, exogenous factors from the tumor microenvironment, including hepatocyte growth factor and interleukin-6, may also play a role in primary EGFR TKI resistance [12,13]. Furthermore, suboptimal drug exposure, which may result from dose-escalation due to toxicities, increased metabolism of EGFR TKIs by cytochrome P450 3A4, or cigarette smoke-induced upregulation of cytochrome P450 1A1, has been shown to be associated with primary drug resistance [14]. The results of these studies are promising, and may prove to be helpful in predicting therapeutic efficacy and overcoming resistance to TKIs in patients whose NSCLC harbors EGFR mutations in the near future.
There were several limitations of current study that may have influenced the analyses that were conducted. First, the number of non-responders was small (n = 20). However, the differences in PFS between responders and non-responders were marked. Thus, we believe that the major findings of this study will not be biased by the issue of patient number. Second, approximately 30% of the 124 enrolled patients displayed poor performance statuses (ECOG PS = 3-4), which may have influenced treatment efficacy. In an effort to exclude this possibility, multivariate analyses were conducted and revealed that performance status was not significantly associated with treatment efficacy, as shown in Table 4 (P = 0.411). We further focused on the 101 patients with ECOG PS of 0-2, and re-analyses indicated that the survival results were consistent with those in all eligible patients. In addition, 2 nd -line treatment may have affected OS in responders and non-responders in this retrospective study, and we will address this issue by attempting to enroll more non-responders in the future.
In conclusion, we demonstrated that a subset of NSCLC EGFRmutant patients did not experience tumor shrinkage, leading to unfavorable PFS and OS. Moreover, none of the clinical variables that we assessed could be successfully used to predict EGFR TKI   treatment efficacy in this population with the exception of subtype of EGFR mutations. These observations highlight the need for a novel biomarker that effectively discriminates non-responders from responders prior to or earlier after EGFR TKI administration. In addition, optimal therapeutic strategies to prolong the survival of non-responders need to be investigated, and the development of novel drugs that can overcome primary EGFR TKI resistance is of equal importance.