Abstract
Background/Aim: Epidermal growth factor receptor (EGFR) mutation testing is standard-of-care for advanced non-small cell lung cancer (NSCLC). Outcomes of second-/third-line compared to first-line tyrosine kinase inhibitors (TKIs) have shown conflicting results. We investigated utilization of molecular diagnostics and the outcomes of treatment with first-/second-line TKIs in patients with advanced NSCLC. Materials and Methods: Retrospective analysis was carried out of 2,206 patients with stage IIIb/IV NSCLC treated between 2008 and 2014 in four hospitals in the Netherlands. Results: The rate of performing molecular diagnostics increased from 20.8% to 74.4% in the study period. The median overall survival of EGFR mutation-positive patients treated with TKIs was superior compared to EGFR mutation-negative patients treated with chemotherapy (720 vs. 274 days, p<0.0001). No difference in overall survival was found between EGFR mutation-positive patients treated only with TKIs compared to those treated with chemotherapy prior to TKIs, or upon progression under TKIs. Conclusion: The rate of EGFR testing has improved, increasing the number of patients eligible for targeted therapy. Chemotherapy, prior or subsequent to TKIs, for the treatment of EGFR mutation-positive patients, did not result in significantly better overall survival compared to that achieved with TKIs alone.
Non-small cell lung cancer (NSCLC) accounts for 80-85% of all lung cancers, with adenocarcinoma being the most common histological type (1). Treatment of metastasized disease is mostly based on platinum-based chemotherapy, but the prognosis is poor. In recent years, the development of targeted therapies, such as tyrosine kinase inhibitors (TKIs), has broadened the therapeutic options and improved prognosis of advanced lung cancer. Patients with NSCLC that harbors TKI-sensitizing mutation in epidermal growth factor receptor (EGFR, commonly exon 19 in-frame deletions or exon 21 p.L858R) and are treated with TKIs have a longer progression-free survival (PFS) and higher response rates compared to those with EGFR mutation-negative tumors treated with conventional chemotherapy (1-4). In addition, TKI treatment also reduces tumor burden and symptoms, and improves quality of life (5, 6). Furthermore, since TKIs are associated with less toxicity than conventional chemotherapy (7), they are also suitable for use in elderly and disabled patients (5, 8).
By not performing molecular diagnostic tests, treatment options are not fully explored, that can have a potential impact on the outcome. Routine EGFR mutation testing is therefore recommended in all patients with advanced non-squamous lung cancer or NSCLC- not otherwise specified (NOS) (5, 9). Although EGFR testing has been available since 2005, it only became part of the guidelines in 2011 (9). There is a lack of information on utilization of molecular diagnostic testing in patients with metastatic lung cancer. A study performed in England revealed that in 2014 only 62% were tested, showing that despite guideline recommendation, there is a large gap between provision of testing and demand (10).
Despite better response rates and longer PFS, studies so far have failed to show better overall survival (OS) associated with first-line TKI versus chemotherapy in patients with EGFR mutation-positive NSCLC (11), partly because a large fraction of the patients treated initially with chemotherapy crossed over to the EGFR TKI treatment arm, confounding the interpretation of overall survival data (12). Current data on treatment outcomes of second- and third-line EGFR therapy have shown conflicting results, with some suggesting tumor response rate and PFS to be similar to those with first-line therapy (13, 14), while others showed lower tumor response rates (15, 16).
Therefore, we aimed to investigate (i) the utilization of molecular diagnostic testing and (ii) the outcomes of treatment with first- or second-line TKIs compared to chemotherapy in patients with advanced NSCLC in real-world setting.
Materials and Methods
Clinical data were collected from four large teaching hospitals in the Netherlands. These four hospitals work together with two additional hospitals under the name of Santeon, a Dutch nationwide Hospital Group. Santeon serves more than 12.5% of the Dutch lung cancer patient population. The individual hospitals included in this study were: Canisius Wilhelmina Hospital (Nijmegen), Catharina Hospital (Eindhoven), Martini Hospital (Groningen), and St. Antonius Hospital (Nieuwegein/Utrecht).
In 2012, Santeon established the Care for Outcome (CfO) registry that includes outcome data from all patients diagnosed with lung and prostate cancer at any of the six hospitals from 2008 onwards. The CfO registry is built on the general Dutch Cancer Registry (managed by the Netherlands Comprehensive Cancer Organization) and subsequently enriched with clinical data from the hospitals (through extensive medical chart review and nominalizations of data). These clinical data include, amongst others, tumor characteristics, patient co-morbidities, Eastern Cooperative Oncology Group performance status (ECOG PS), treatment planning and clinical outcomes. Trained medical students supervised by a pulmonary physician extracted all items.
Since this was a retrospective study and all material was gathered as a part of routine diagnostic procedure, no written informed patient consent was required. The study was approved by the local Research Ethics Committee of each individual hospital (W16.138).
Patient selection and data collection. From the CfO database, we extracted data from all patients that were diagnosed with NSCLC from 2008 until 2014 at one of the Santeon hospitals, had clinical stage IIIB or IV (including both patients with and without systemic treatment), and had follow-up data (vitality status) available.
Information about the type of NSCLC and molecular diagnostic testing were additionally manually obtained from the electronic medical records. Patients who received chemotherapy were grouped according to the treatment received: platinum-based doublet, platinum-based doublet with pemetrexed maintenance, platinum-based doublet with bevacizumab, and TKIs. Patients were only included if malignancy had been histologically confirmed. Staging was carried out according to the seventh edition of the TNM classification (17).
Statistical analysis. Kaplan-Meier survival curves were created to present the difference in OS between different treatment arms. A Cox proportional hazard model was used to compare survival among the different treatment groups and investigate the influence of age at diagnosis (continuous variable), gender and performance status. A chi-square test for trend was performed to evaluate the utilization of molecular diagnostic tests in the study period. p-Values of 0.05 or less were considered statistically significant. All statistical analysis was performed using SPSS (IBM SPSS Statistics for Windows, Version 22.0; IBM Corp., Armonk, NY, USA).
Results
At the participating Santeon hospitals, 2206 patients were diagnosed with advanced NSCLC (stage IIIb and IV), between Jan 1st 2008 and Dec 31st 2014. The characteristics of the patients are provided in Table I.
Molecular diagnostic tests were performed in 879 patients (39.8% of all cases), of which 853 of patients had NSCLC-NOS or adenocarcinoma, and 26 patients had a squamous-cell carcinoma. Molecular diagnostic tests were performed in 10.5% of all patients diagnosed with adenocarcinoma/NSCLC-NOS in 2008. In 2014, this percentage had increased significantly to 74.5%. There were significant differences between utilization of molecular diagnostic tests between hospitals (p<0.0001; Figure 1).
Outcome of molecular diagnostics. For 92 out of the 879 patients (10.5%), there was not sufficient or suitable material on which to perform molecular diagnostic testing. EGFR mutation was reported in 84 patients with adenocarcinoma or NSCLC-NOS (10.7%) and an activating kirsten rat sarcoma viral oncogene homolog (KRAS) mutation (typically in codons 12, 13 or 61) was found in 290 patients (36.8%). As expected, all tumors harboring a KRAS mutation were wild-type for EGFR. A TKI-sensitizing EGFR mutation was observed for 75 patients, with most commonly an exon 19 in-frame deletion (n=41, 48.8%) or an exon 21 activating mutation (n=31, 36.9%) (Table III). Nine patients (10.7%) had a mutation in exon 20 (p.T790M or in-frame insertion) considered to be resistant/non-responder mutations.
Characteristics of patients with EGFR-mutated lung cancer are shown in Table II.
Treatment of EGFR mutation-positive patients. Of the patients with a TKI-sensitizing EGFR mutation (in exon 18, 19 or 21), the great majority (90.7%) were treated with TKIs (Table III). Thirty-two (42.6%) patients received only TKIs (gefitinib, erlotinib or afatinib), whereas 16 (21.3%) and 20 (26.7%) were treated with second-line TKIs after chemotherapy or chemotherapy after progression under TKIs, respectively.
Seven patients with sensitizing EGFR mutations received no treatment because of their low performance status (n=3), refusal of treatment (n=3) or rapid disease progression (n=1).
Out of the patients with a resistant/non-responder EGFR exon 20 mutation (n=9), four were treated with conventional chemotherapy because of expected poor response to TKIs, two received no treatment, two were treated with chemotherapy before TKIs and one patient was treated with TKIs before chemotherapy (Table III).
Overall survival. The median OS adjusted for age, gender and performance status was 720 days (95% confidence interval (CI)=641-911 days) in EGFR mutation-positive patients (sensitizing mutations only, excluding exon 20 mutations) treated with TKIs (with or without prior/subsequent chemotherapy) versus 274 days (95% CI=385-452 days) in patients without EGFR mutations treated with conventional chemotherapy (p<0.0001) (Table III, Figures 2 and 3). Further analysis of this patient group indicated no statistically significant differences in OS between the different treatment regimens: median OS of 1034 (95% CI=703-1285) days for those treated with conventional chemotherapy before TKI; 704 (95% CI=581-961) days for those treated with TKIs followed by chemotherapy; and 599 (95% CI=455-760) days for those treated only with TKIs (Table III). There was no evidence that survival was related to age at diagnosis (p=0.14), gender (p=0.43) or ECOG PS (p=0.26).
Re-biopsy after progression of disease. Out of 84 patients with an initial EGFR-sensitizing mutation, 13 patients (15.5%) underwent a biopsy to evaluate the potential development of an exon 20 resistant mutation after progression on TKI. For one patient, no reliable result could be generated because of inappropriate material. The majority of patients with progressive disease (n=8) developed the well-known EGFR p.T790M resistance mutation. Four out of these eight patients with EGFR T790M continued treatment with third-generation TKI (osimertinib). In four patients, only the initial sensitizing mutations (in exon 19 and 21) were found, without any sign of an exon 20 mutation.
Discussion
This study showed that the utilization of molecular testing has improved significantly over the years, leading to more effective targeted therapies. The median OS in patients with EGFR mutations treated with TKIs (with or without prior/subsequent) chemotherapy was longer than that for EGFR mutation-negative patients treated with conventional chemotherapy. Although EGFR mutation-positive patients treated with chemotherapy prior to TKIs tended to have the longest OS, partly because of the small number of patients, no statistically significant survival benefit was shown.
By not performing molecular diagnostics, patients cannot be considered for targeted therapy. Therefore, it is important to assess the extent to which these tests are being utilized. EGFR testing has been available since 2005, however specific recommendations for EGFR mutation analysis in stage IIIb and IV adenocarcinoma and NSCLC-NOS only became a part of the Dutch guidelines in 2011 (9). Underutilization before 2011 was therefore expected. Nevertheless, only 63% of patients were tested in the period of study here, meaning 403 (37%) patients were missing out on molecular diagnostic tests. Surprisingly, despite comparable hospital characteristics, there were also significant institutional differences in utilization of EGFR assays. The incidence of molecular diagnostic testing did however improve significantly between 2011 and 2014, increasing from 52.1% in 2011 to 74.5% in 2014 (Figure 1). A study performed on the frequency of molecular diagnostic testing in England showed an even larger gap between provision of testing and demand, with only 62% being tested in 2014 (10).
EGFR mutations were found for 10.7% of all patients tested in our cohort, which is in line with other studies performed on other European populations, with incidence ranging from 9.1% to 16.6% (13, 18, 19).
Not only can obtaining a tissue sample be challenging, often (in our case 10.5%) these samples contain too few tumor cells to allow reliable molecular testing. (20) Since patients with advanced NSCLC have a limited life expectancy and molecular diagnostic tests are time-consuming, practitioners tend to start treatment before mutation test results are available (21). In 2016, the U.S. Food and Drug Administration approved the first liquid biopsy test for EGFR mutation in NSCLC (cobas EGFR Mutation Test v2) (22), which allows a sample to be obtained in a minimally invasive way and is less time-consuming, which might lead to improvement of utilization of molecular diagnostic tests in the future (23).
Despite high initial response to EGFR-TKIs, most patients with sensitizing EGFR mutation eventually acquire resistance to EGFR TKI therapy, by developing secondary mutations such as the T790M mutation in exon 20 (approximately 50%) or amplification of mesenchymal epithelial transition (MET) (~20%) (24). Since 2015, osimertinib, an irreversible EGFR TKI selective for both EGFR-mutated and T790M resistance mutations (25), has been recommended for treatment of patients with T790M-positive advanced NSCLC after progression with prior EGFR-TKI therapy (5). Due to a variety of resistance mechanisms and availability of targeted therapy for the most commonly (acquired) mutation, post-treatment biopsies at the time that clinical resistance to EGFR TKIs develops are therefore beneficial in selecting the most appropriate treatment (26). Although, our study was performed in the period when osimertinib was not yet recommended for patients with T790M mutation, 13 patients (15.5%) with EGFR mutation underwent rebiopsy after progression to evaluate T790M mutation, and the majority (four out of the six patients who developed EGFRe 20 T790M mutation) were also treated with osimertinib.
In the first randomized clinical trial, the Iressa Pan-Asia Study or IPASS, conducted on an East Asian population, comparing gefitinib with carboplatin/paclitaxel in EGFR mutation-positive patients, showed significantly longer PFS in patients treated with gefitinib versus those treated with carboplatin/paclitaxel (9.5 vs. 6.3 months, respectively). Median OS was 18.6 months among patients receiving gefitinib and 17.3 months among patients receiving carboplatin–paclitaxel. (12). Studies that followed (NEJ002, WJTOG3405, EURTAC, OPTIMAL, LUX LuNG3, LUX-LUNG6) confirmed the positive effect of TKIs on EGFR-mutated tumors for PFS (1-4, 8, 27). These studies, however, failed to show better OS, partly because a large fraction of the patients with EGFR-mutated tumors treated initially with chemotherapy crossed over to the EGFR TKI treatment arm, confounding the interpretation of overall survival data (12). This raises the question whether chemotherapy prior to TKIs or subsequently affects the treatment outcome. It is not yet well understood if systemic therapy influences the expression of different biomarkers such as EGFR mutation in tumor (28).
Studies comparing first- and second-line EGFR therapy are limited. Chemotherapy may affect EGFR mutations status by reducing the frequency of EGFR mutations (29), suggesting reduced efficacy of second-line treatment with TKIs. A meta-analysis of 13 randomized trials showed that the response rate of TKI treatment in the first-line trials was 70%, while in in second-line trials, it was 47.4%, it did, however, failed to improve OS (30).
Studies advocating second- or third-line EGFR TKIs showed inconsistent results. A single-arm phase II study conducted in Spain suggested that second- or third-line EGFR TKI could attain tumor response rates and PFS similar to those of first-line therapy (31, 32). A randomized phase 3 trial on chemo-naive East Asian patients with advanced pulmonary adenocarcinoma, comparing first-line pemetrexed plus cisplatin followed by gefitinib maintenance therapy with gefitinib monotherapy, failed to show any significant difference in PFS in the EGFR-mutation subgroup (33).
In our cohort, EGFR mutation-positive patients treated with TKIs (without additional chemotherapy) had a significant OS benefit compared to EGFR mutation-negative patients treated with chemotherapy (599 vs. 274 days). The OS of EGFR mutation-positive patients treated with TKIs (Table III, Figure 2) was comparable to those of other studies performed on European cohorts, such as the FIELT study with a median OS of 23 months (34), and the EURTAC study with a median OS of 19.3 months (1). Despite potential toxicity related to chemotherapy possibly resulting in taking away the opportunity to receive second-line therapy (32), our patients treated with TKIs in the second line had the longest OS (34 months). OS in patients treated with second-line chemotherapy after progression under TKIs (23.1 months) and those only treated with TKIs (19.7 months) seemed inferior but no statistical significance was shown. The number of patients with EGFR-mutated tumors was probably too small to draw any significant conclusions. Furthermore, EGFR mutation heterogeneity could perhaps explain the mixed response leading to a wide CI.
Since obtaining the appropriate material for molecular diagnostics can be challenging, results suggest that starting chemotherapy while waiting for results of the EGFR mutation testing could result in promptness of therapy without compromising the OS.
Strengths and limitations. To our knowledge, no previous studies gave insight into the treatment outcomes of first- and second-line TKI compared to chemotherapy in such a long study period in real-word setting in an European cohort. While previous studies failed to show any significant OS benefit of TKIs when compared to chemotherapy, this study did manage to show that OS of patients with EGFR mutation-positive disease treated with TKIs (even without subsequent or prior chemotherapy) patients was longer than OS of EGFR mutation-negative patients treated with chemotherapy.
Since most studies have been performed in predominantly Asian cohorts, this study helps to further clarify outcomes of treatment in European patients.
The low number of patients harboring EGFR mutation is a limitation of this study, and was partly responsible for a large CI for the first- and second-line treatment groups.
Unfortunately, we had no information available about the PFS and were unable to analyze the effect of first- or second-line treatment with TKI on PFS.
Conclusion
Incidence of molecular diagnostic testing has improved significantly, leading to more effective targeted therapy. Treatment with TKIs in patients with EGFR-mutated tumors with or without prior/subsequent chemotherapy led to better OS than that of patients with EGFR mutation-negative tumors treated with chemotherapy. Chemotherapy, prior to TKIs or subsequently, did not result in significantly better OS compared to therapy with TKIs alone.
- Received October 1, 2017.
- Revision received October 23, 2017.
- Accepted October 26, 2017.
- Copyright© 2018, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved