Background
Gefitinib (ZD1839, Iressa; AstraZeneca) and erlotinib (Tarceva, OSI-774; OSI Pharmaceuticals) are orally active epidermal growth factor receptor tyrosine kinase inhibitors (EGFR-TKIs) used for the treatment of non-small cell lung cancer (NSCLC) patients [
1]. EGFR-TKIs sometimes cause drastic tumor regression in specific subgroups of patients with advanced NSCLC, including women, non-smokers, patients with lung adenocarcinoma (ADC) histology, patients of Asian origin and patients with
EGFR mutations [
2‐
6]. On the other hand, treatment with EGFR-TKIs is associated with serious side effects, such as life-threatening drug-induced interstitial lung disease (ILD), particularly in Japanese populations [
7‐
13]. These previous studies have reported that male gender, smoking history, poor performance status (PS), and preexisting ILD are risk factors for developing EGFR-TKIs induced ILD, however, we questioned whether each of these should be equally considered for the risk-benefit assessment to use EGFR-TKIs for the treatment of NSCLCs in a practical clinical setting. In addition, we also wondered whether we can assess the severity of EGFR-TKIs induced ILD when it develops during EGFR-TKIs treatment.
KL-6 is a mucin-like glycoprotein with a molecular weight of 200kd and has been classified as human MUC1 mucin [
14‐
17]. Previous studies have demonstrated that serum levels of KL-6 are elevated in a variety of ILDs, such as idiopathic pulmonary fibrosis (IPF), collagen vascular disease associated interstitial pneumonitis, radiation pneumonitis, pulmonary sarcoidosis [
18‐
26]. Furthermore, our laboratory has also demonstrated that absolute levels of KL-6 at the onset of drug-induced ILD can predict the clinical outcomes [
27]. Although our previous studies have suggested the usefulness of KL-6 as a tumor marker [
28,
29] and a predictor of survival in NSCLC patients treated with EGFR-TKIs [
30], significance of circulating KL-6 level as a detector of EGFR-TKIs induced ILD or a predictor of clinical outcome in patients with EGFR-TKIs induced ILD has not been determined yet.
In the cohort of the present study, to obtain more information on risk factors for developing EGFR-TKIs induced ILD, the characteristics of NSCLC patients who developed ILD during EGFR-TKIs treatment were analyzed. In addition, to evaluate whether monitoring serum KL-6 levels in NSCLC patients during the treatment is useful to detect the development of EGFR-TKIs induced ILD or predict the clinical outcome of EGFR-TKIs induced ILD, circulating KL-6 levels were measured in NSCLC patients included in the cohort before and during EGFR-TKIs treatment.
Methods
Study subjects
Between August 2002 and August 2010, 341 advanced NSCLC patients treated with gefitinib (250 mg/day) or erlotinib (150 mg/day) at Hiroshima University Hospital (Hiroshima, Japan), Ehime University Hospital (Ehime, Japan), Shimane University Hospital (Shimane, Japan), Kochi University Hospital (Kochi, Japan) and Onomichi General Hospital (Hiroshima, Japan) were consecutively enrolled in the study. The disease staging was carried out using computed tomography (CT) scan of the chest and abdomen, bone scintigraphy or F-18 fluorodeoxyglucose positron emission tomography (FDG-PET/CT), and magnetic resonance imaging (MRI) of the head. To obtain information on both the response of tumor to EGFR-TKIs treatment and the occurrence of EGFR-TKIs induced ILD, chest radiography and/or CT scans were performed at least once a month at each institution, and the patients were followed-up until 12 weeks after the administration of EGFR-TKIs. Informed consent was obtained from all patients. This study complied with the Declaration of Helsinki, and was approved by the individual institutional Ethical Committees.
Diagnosis of preexisting pulmonary disorder and EGFR-TKIs induced ILD
The presence of preexisting pulmonary fibrosis was determined according to the diagnostic criteria set by the ATS/ERS on the basis of clinical characteristic and/or chest CT findings, and the types of preexisting pulmonary fibrosis were classified into idiopathic pulmonary fibrosis (IPF) pattern and non-IPF pattern [
31‐
33]. In addition, the presence of preexisting pulmonary emphysema was determined by chest CT findings that show low attenuation areas occupying more than 25% of the entire lung field in at least one slice [
34]. The diagnosis of EGFR-TKIs induced ILD was made using the diagnostic algorithm described elsewhere [
11,
35]. We defined EGFR-TKIs induced ILD as diffuse pulmonary infiltrates newly developed during EGFR-TKIs treatment with lack of evidence for alternative diseases such as infection, tumor progression, heart failure and pulmonary embolism. When the occurrence of EGFR-TKIs induced ILDs was suspected, chest CT scans were performed, levels of brain natriuretic peptide (BNP) and D-dimer in blood were measured, the sputum culture, blood culture, urine antigen test for
Legionella pneumophila and
Streptococcus pneumoniae, cytomegalovirus antigen test, and polymerase chain reaction test for
Pneumocystis jiroveci were conducted. When possible, bronchoalveolar lavage or lung biopsy was carried out. Tumor progression was carefully excluded on the basis of the clinical information including chest CT findings, physical examinations, and tumor markers. The final diagnosis of EGFR-TKIs induced ILD was made by the consensus of at least two independent pulmonologists. We collected the clinical information of all 341 patients, such as patient age, sex, histologic type, disease stage, performance status, prior chemotherapy and thoracic radiation therapy, preexisting pulmonary fibrosis, preexisting pulmonary emphysema,
EGFR mutation status, types of EGFR-TKIs, duration of EGFR-TKIs treatment and laboratory data.
Subclassification of EGFR-TKIs induced ILD
The chest radiography and CT of the patients who developed EGFR-TKIs induced ILD were reviewed separately by two independent observers who were not aware of the patients' profiles, and were categorized into four patterns as previously described [
27,
36]: (1) acute interstitial pneumonia (AIP) pattern characterized by extensive bilateral ground glass attenuation or airspace consolidations with traction bronchiectasis, (2) chronic interstitial pneumonia (CIP) pattern characterized by fibrosis and/or consolidation, (3) cryptogenic organizing pneumonia/eosinophilic pneumonia (COP/EP) pattern showing peribronchial or subpleural consolidation without fibrosis, and (4) hypersensitivity pneumonitis (HP) pattern with diffuse ground glass opacities without fibrosis.
EGFR mutation status
In 148 out of 341 NSCLC patients included in the study,
EGFR mutation statuses were assessed using paraffin-embedded biopsy samples or surgically resected tumor tissues. To evaluate
EGFR mutations, the peptide nucleic acid-locked nucleic acid polymerase chain reaction (PNA-LNA PCR) clamp test that can detect G719C, G719S, G719A, L858R, L861Q, T790M and 7 different exon 19 deletions [
37] was used.
Electrochemiluminescence immunoassay (ECLIA) to determine circulating levels of KL-6
At least one serum sample was obtained before the EGFR-TKIs treatment from each patient included in the study. From 15 out of 20 patients who developed EGFR-TKIs induced ILD, a total of 2-5 serum samples per patient were also collected weekly after the occurrence of EGFR-TKIs induced ILD, and stored at -80°C. Serum KL-6 levels were measured by sandwich-type electrochemiluminescence immunoassay (ECLIA) using a Picolumi 8220 Analyzer (Eidia, Tokyo, Japan), as previously described [
29,
30].
Statistical analysis
The data were analyzed with a statistical software package (JMP, version 7.0.1; SAS Institute Inc.; Cary, North Carolina) and p < 0.05 indicated a significant difference. Data are shown as the mean ± SEM. Differences between patients with and without preexisting pulmonary fibrosis, survivors and non-survivors, and patients with AIP pattern and the other patterns of EGFR-TKIs induced ILD were analyzed using the Mann-Whitney U-test. We analyzed differences between patients with preexisting pulmonary fibrosis who developed EGFR-TKIs induced ILD or not using the Fisher's exact test. In order to test differences among the variables evaluated prior to and at the diagnosis of EGFR-TKIs induced ILD, Wilcoxon test was used. The risk factors associated with EGFR-TKIs induced ILD were evaluated using multiple logistic regression analysis. The criterion for removing a variable was the likelihood ratio statistic, which was based on the maximum partial likelihood estimate (default p-value of 0.05 for removal from the model).
Discussion
In this large multi-institutional study, we investigated the incidence and risk factors for developing ILD in patients treated with EGFR-TKIs until 12 weeks after the start of EGFR-TKIs therapy. Univariate analyses revealed that preexisting pulmonary fibrosis at baseline was the only risk factor for EGFR-TKIs induced ILD. Although absolute serum KL-6 levels at neither baseline nor the onset of ILD could discriminate between life-threatening and non-life-threatening EGFR-TKIs induced ILDs, the ratio of serum KL-6 level at the occurrence of EGFR-TKIs induced ILD to that at baseline was found to quite precisely do so. These findings suggest the significance of serum KL-6 level for the detection of life threatening EGFR-TKIs induced ILD.
The development of molecular targeted agents has been a key factor in recent advances in cancer therapy, and some of these agents have been applied in clinical practice. EGFR-TKIs are one of the representative molecular target agents and, at first, were considered to be safe agents with mild side effects in comparison to cytotoxic agents. However, following the increase in usage of EGFR-TKIs in lung cancer therapy, a significantly higher incidence of life-threatening drug induced ILD in Japanese patients than that of patients in the rest of the world was reported [
38,
39]. In the present study, out of 341 NSCLC patients treated with EGFR-TKIs, 20 patients (5.9%) developed ILD and 9 patients (2.6%) died from ILD. The incidence and mortality of EGFR-TKIs induced ILD were relatively higher than those reported in previous studies from Japan [
7‐
13,
39]. This result might be due to the high incidence of preexisting pulmonary fibrosis in this study. In this study, the manifestations of chest CT scans in 20 patients who developed EGFR-TKIs induced ILD were classified as AIP pattern for 5 patients, COP/EP pattern for 9 patients and HP pattern for 6 patients. Interestingly, CIP pattern was not observed as was the case in a previous study [
36]. All the patients who demonstrated the AIP pattern died, whereas the majority of patients with other patterns recovered from EGFR-TKIs induced ILD. In this study, the postmortem examination of three patients with AIP pattern revealed that DAD was the main cause of death and observations similar to ours have been reported previously [
7,
8]. In this study, univariate analysis revealed that preexisting pulmonary fibrosis was the only risk factor for developing EGFR-TKIs induced ILD. Although previous studies reported that male gender, smoking history and poor PS were also independent risk factors for developing EGFR-TKIs induced ILD [
7‐
13,
39], neither of them correlated with incidence or mortality of EGFR-TKIs induced ILD in the present study. This may be due to the small sample size and high incidence of preexisting pulmonary fibrosis in our studied patients.
Although a previous study from our laboratory reported that serum KL-6 levels at diagnosis increased only in the life-threatening types, such as the DAD and CIP patterns, of drug induced ILDs [
27], absolute serum KL-6 levels at the onset of EGFR-TKIs induced ILD did not correlate with clinical outcomes in the present study. The immunohistochemical analysis of KL-6 using three postmortem autopsy specimens showed that KL-6 was expressed at tumor cells in the primary lesions as well as alveolar epithelial cells in the EGFR-TKIs induced ILDs (data not shown). Therefore, we speculate that the origin of serum KL-6 at the onset of EGFR-TKIs induced ILD might be associated with both NSCLCs and EGFR-TKIs induced ILDs. On the other hand, we found that the ratios of serum KL-6 levels just after the onset of ILD to those at baseline could quite precisely discriminate life-threatening ILD from non-life-threatening ILD, and correlate well with the disease progression. We can speculate that a drastic increase in serum KL-6 levels after the administration of EGFR-TKIs might be due to severe lung injury accompanied with both alveolar-capillary destruction and enhancement of alveolar-capillary permeability which allow KL-6 to leak into the circulation from the alveolar space [
40]. Based on these observations, KL-6 can be regarded as a good serum biomarker to assess the severity of alveolar epithelium injury and the clinical outcome of EGFR-related ILD. Regarding the association between KL-6 and other serum biomarkers for ILD such as surfactant protein (SP)-A and SP-D in EGFR-TKIs induced ILD, we do not have data to discuss. Previous studies, which measured serum SP-A, SP-D, and KL-6 levels in 4 patients with EGFR-TKIs induced ILD, demonstrate that serum SP-A and SP-D levels increased in all studied patients whereas KL-6 levels only elevated in patients with life-threatening EGFR-TKIs induced ILD [
8,
41]. This observation is compatible with the findings of the present study.
In addition to its ability to detect patients who develop life-threatening ILD, the monitoring of serum KL-6 levels is also useful to predict survival and progressive disease in NSCLC patients treated with EGFR-TKIs [
30]. As measurement of serum KL-6 level is more rapid, inexpensive, reproducible, and easier to perform than CT scans, its monitoring could be quite useful to assess the condition of NSCLC patients receiving EGFR-TKIs. The development of EGFR-TKIs induced ILD is reported to mostly occur within the first 4 weeks after the start of EGFR-TKIs [
11]. In the present study, 5 cases developed ILD within the first 2 weeks (ranged from 5 to 14 days) after the start of EGF-TKIs. Therefore, based on the results of the present study, once a week monitoring of serum KL-6 levels in addition to chest radiography could be recommended for NSCLC patients receiving EGFR-TKIs particularly for the first 4 weeks after the start of treatment.
Although these promising results were obtained, we are aware that this study has a number of limitations. First, the number of EGFR-TKIs induced ILD patients included in the study was not sufficient for a valid statistical analysis. Second, this study was conducted in a retrospective manner. Therefore, the information on EGFR mutation statuses in cancer tissue was not obtained from all the studied patients. Furthermore, multiple measurements of serum KL-6 levels were not achieved in all patients who developed EGFR-TKIs induced ILD. Third, the enrolled NSCLC patients might be biased compared with general advanced NSCLC population. We believe that this was caused by our trend to use EGFR-TKIs for specific subgroups of NSCLC patients such as women, non-smokers, and patients with EGFR mutations. Finally, the studied patients were only Japanese. Considering ethnic differences in the efficacy of EGFR-TKIs treatment and/or the occurrence of adverse side effects related by EGFR-TKIs, we should carefully interpret the results when this monitoring system is applied to non-Japanese patients. A large and prospective study to measure serum KL-6 levels serially before and after EGFR-TKIs treatment, also including non-Japanese patients, will be required to evaluate the utility of monitoring KL-6 in EGFR-TKIs induced ILDs.
Competing interests
Nobuoki Kohno has a personal royalty of KL-6 from a Japanese pharmaceutical company, Eisai Co., Ltd. The remaining authors have no conflict of interest.
Authors' contributions
SK performed part of the statistical analysis and drafted the manuscript. NH conceived the study, and participated in its design and coordination and helped to draft the manuscript. NI conceived the study, and participated in patient recruitment and helped to draft the manuscript. YH performed part of the statistical analysis and participated in creating the figures. KF, OF, TI, SM, HH and TY participated in the selection and collection of patient material. AY conceived the study, and participated in its design and coordination. NK conceived the study, and participated in its design and coordination and supervised the study. All authors read and approved the final manuscript.