Presented in part at the metastatic/nonsmall cell lung cancer poster discussion session of 47th Annual Meeting of the American Society of Clinical Oncology; June 3-7, 2011; Chicago, IL.
The purpose of this study was to investigate the overall survival (OS) of patients with advanced ALK-positive nonsmall cell lung cancer (NSCLC) who were managed in the pre-ALK inhibitor era and to compare their survival with that of a matched case cohort of ALK wild-type (WT) patients.
Data from 1166 patients who had stage IIIB/IV NSCLC with nonsquamous histology were collected from the NSCLC database of Seoul National University Hospital between 2003 and 2009. ALK fluorescence in situ hybridization (FISH) was used to analyze 262 patients who either had the WT epidermal growth factor receptor (EGFR) or were nonresponders to previous EGFR tyrosine kinase inhibitor (TKI) therapy. Overall survival (OS) was compared between 3 groups: 1) ALK-positive patients, 2) EGFR mutation-positive patients, and 3) ALK-WT/EGFR-WT patients. Progression-free survival (PFS) after first-line chemotherapy and EGFR TKIs also was analyzed.
Twenty-three patients were ALK-positive according to FISH analysis and did not receive ALK inhibitors during follow-up. The median OS for ALK-positive patients, EGFR mutation-positive patients, and WT/WT patients was 12.2 months, 29.6 months, and 19.3 months, respectively (vs EGFR mutation-positive patients, P = .001; vs WT/WT, P = .127). The PFS after first-line chemotherapy for the 3 groups was not different. However, the PFS for patients who received EGFR TKIs was shorter in ALK-positive patients compared with the other 2 groups (vs EGFR mutation-positive patients, P < .001; vs WT/WT, P < .021).
It has become evident that nonsmall cell lung cancer (NSCLC) has distinct genetic alterations that are crucial for tumorigenesis. These molecular changes, called driver mutations, allowed a new way to classify lung cancer into clinically relevant subgroups.1-3 One of these groups is the echinioderm microtubule-associated protein-like 4 (EML4)-anaplastic lymphoma kinase (ALK) gene translocation EML4-ALK, which was identified in 2007.4, 5 A small inversion within chromosome 2p results in the formation of a fusion gene comprising portions of the EML4 gene and the ALK gene.6 The product of this fusion gene works as a driver for proliferation in lung cancer cells that harbor this translocation, demonstrating the phenomenon of oncogene addiction.7, 8 Less than 3 years after identification of the EML4-ALK translocation, a phase 1 trial of crizotinib (PF-02341066; Pfizer, New York, NY), an orally active ALK and MET dual inhibitor, resulted in a significant response in patients who had the EML4-ALK translocation. In a pretreated patient population that generally has a 10% response rate to conventional chemotherapy, treatment with crizotinib yielded an overall response rate (ORR) of 55% and an estimated 6-month progression-free survival (PFS) rate of 72%.9, 10 Furthermore, the mechanism of resistance to crizotinib was identified at the same time.11
In the center of this rapid advance of translational research, there has been an early understanding of the clinical and pathologic characteristics of patients with the EML4-ALK translocation.12 Prevalence of the EML4-ALK translocation in unselected patients with NSCLC ranges from 3% to 5%.13-16 The EML4-ALK translocation is highly correlated with younger age and never-smoking or light-smoking history.15, 17 The pathologic features of ALK-positive tumors also are distinct. Almost all of them are adenocarcinomas; signet-ring cell histology and acinar pattern were commonly identified.13, 18-20 Recent studies have proposed using these clinicopathologic characteristics as screening strategies to enrich for the likelihood of ALK-positive tumors.3, 21-23
The EML4-ALK translocation is now a clear positive predictive marker of ALK-inhibitor therapy.9, 24 However, the prognostic value of ALK translocation is not fully understood. Previous studies tried to analyze overall survival (OS) in patients with the EML4-ALK translocation, but the clinical significance was in question because of small numbers of events in enrolled patients and confounding from the administration of crizotinib in the ALK-positive group.12, 23 Therefore, we present the current study, in which we compare ALK-positive patients and matched ALK-WT patients who were treated in the pre-ALK inhibitor era. Informed consent was obtained from the patients and/or surrogates who participated in this study. This protocol was reviewed and approved by the Institutional Review Board (IRB) of Seoul National University Hospital (IRB No. H-1008-035-326).
MATERIALS AND METHODS
In total, 1166 patients who had stage IIIB/IV NSCLC with nonsquamous histology were collected from the NSCLC database of Seoul National University Hospital, Seoul, Korea, between January 1, 2003 and August 31, 2009. To enrich for ALK-positive patients, we excluded patients who harbored EGFR mutations, because the EML4-ALK translocation is rarely coincident with EGFR mutation.14, 25 Among the patients with unknown EGFR mutation status, we excluded patients who achieved an objective response with gefitinib or erlotinib, using this as a proxy for the likelihood of an EGFR mutation-positive patient (Fig. 1).12 Patients who had insufficient tissue for pathologic examination or whose tissue produced an inconclusive result in EML4-ALK fluorescence in situ hybridization (FISH) also were excluded. Therefore, in total, 262 patients with examinable tissue were enrolled. Patients who had received crizotinib were not included in the analysis.
The medical records of enrolled patients were reviewed to collect demographic, clinical, and pathologic information. We examined chemotherapeutic regimens, responses, sites of metastases, and clinical outcomes. We also recorded the EGFR mutation status of patients, which had been determined using a direct sequencing method of EGFR exons 18 to 21. Radiologic responses were evaluated according to version 1.0 of Response Evaluation Criteria in Solid Tumors.26 OS was calculated from the date metastatic disease was diagnosed to the date of death. PFS was measured from the first day of chemotherapy until radiologic or clinical disease progression.
Pathologic Examination and Anaplastic Lymphoma Kinase Testing
In total, 262 patients with testable formalin-fixed, paraffin-embedded tissue were included in this study. Two hundred six samples were biopsied by percutaneous technique, and 56 samples were surgically resected. All histologic diagnoses were reviewed based on the latest World Health Organization classification.27
ALK FISH analysis was performed using a dual-color break-apart probe (Abbott Molecular, Abbott Park, Ill), which hybridizes the 2p23 band (red signal) and the ALK gene breakpoint (green signal). All procedures were conducted according to the manufacturer's instructions. Three micron-sectioned, formalin-fixed, paraffin-embedded tissues were deparaffinized, dehydrated, immersed in 0.2 N HCl, and incubated in 1 M NaSCN at 80°C for 30 minutes. A pepsin solution was added to treated sections; then, dual-probe hybridization for ALK was performed. After application of the probe mixture, the slides were treated with protease and then incubated in a humidified atmosphere with HYBrite (Abbott Molecular) at 77°C for 5 minutes for denaturation. Subsequently, the slides were incubated at 37°C for 16 hours for hybridization. Slides were then immersed in 0.3% NP-40 (Abbott Molecular)/0.4 × saline sodium citrate (SSC) for 5 minutes at room temperature, followed by 0.3% NP-40 of 0.4 × SSC for 5 minutes at 72°C. For the counterstaining of nuclei, 4,6-diamidino-2-phenylindole was used. FISH was regarded as positive when break-apart signals or 5′-deletions were observed in >15% of ≥50 tumor cells. All specimens from the FISH assay were examined by 1 trained pathologist (H.S.P.) in a blinded manner.
Case-Case Matching and Statistical Analysis
To control for known prognostic variables in lung cancer survival, each ALK-positive patient was matched to 2 EGFR mutation-positive patients and 2 WT/WT patients. All patients in the matched cohort also were restricted to nonsquamous histology. Matching variables were age at diagnosis, sex, disease stage, and smoking status. The cutoff point for survival analysis was January 13, 2011.
Statistical analyses of categorical variables were performed using Pearson chi-square tests or Fisher exact tests, as appropriate. The t test was used to compare continuous variables between groups. The median durations of OS and PFS were calculated using the Kaplan-Meier method. Comparisons between groups were done using the log-rank test. Multivariate analysis was performed using a Cox proportional hazards model. Two-sided P values < .05 were considered statistically significant. All analyses were performed using the PASW Statistics software package (version 18.0; SPSS Inc. Chicago, Ill).
Among 262 examined tumors, we identified 23 ALK-positive cases by FISH. One ALK-positive case was matched to 2 EGFR mutation-positive patients and to 2 WT/WT patients, as mentioned above (Table 1). All 3 groups included patients with stage IV disease or recurrent tumors except for 1 EGFR mutation-positive patient who had stage IIIB disease. Consequently, a total of 115 patients (23 ALK-positive, 46 EGFR mutation-positive, and 46 WT/WT patients) were included in the survival analysis. On pathologic examination, the ALK-positive group included more unspecified nonsmall cell carcinomas (30%) than the EGFR mutation-positive and WT/WT groups (11% and 20%, respectively). Three ALK-positive patients (13%) and 1 WT/WT patient (2%) had signet ring cell carcinoma. In terms of metastatic site, 30% of both ALK-positive and WT/WT patients had central nervous system (CNS) metastases proven in radiologic or cerebrospinal fluid cytopathologic examinations during treatment. However, EGFR mutation-positive patients had a higher rate of CNS metastasis (63%; P < .001) (Table 2). Fewer liver metastases were observed in the WT/WT group (P = .035).
Table 1. Clinicopathologic Characteristics of Matched Cohorts
No. of Patients (%)
ALK Positive, N = 23
EGFR Mutation Positive, N = 46
WT/WT, N = 46
ALK vs EGFR
ALK vs WT/WT
Abbreviations: ALK, anaplastic lymphoma kinase; ECOG PS, Eastern Cooperative Oncology Group performance status; EGFR, epidermal growth factor receptor; NOS, not otherwise specified; SD, standard deviation; TKI, tyrosine kinase inhibitor.
Heavy smoker means smoker who have smoked ≥10 pack years.
We examined patient responses to and clinical outcomes after chemotherapy and EGFR TKI treatment documented in the medical records (Tables 3 and 4). Among 115 patients, 92 patients (80%) received cytotoxic chemotherapy as first-line treatment. All of those patients received a platinum-based doublet regimen, except for 2 patients who received gemcitabine/vinorelbine and docetaxel. Various doublet combinations were identified; the most common regimen was paclitaxel/carboplatin, which was received by 33 patients, followed by gemcitabine/cisplatin (32 patients). Response rates to cytotoxic chemotherapy did not differ between the 3 groups (ALK-positive patients, 27%; EGFR mutation-positive patients, 32%; WT/WT patients, 35%). Although PFS for EGFR mutation-positive patients (4.93 months) was longer than for the other 2 groups (ALK-positive patients, 3.87 months; WT/WT patients, 3.73 months) (Fig. 2), the difference was not statistically significant (Table 4). In total, 73 patients (63%) received subsequent cytotoxic chemotherapy. The proportion of patients who received second-line chemotherapy was well balanced between the groups (ALK-positive patients, 70%; EGFR mutation-positive patients, 59%; WT/WT patients, 65%,). Pemetrexed32 was the most commonly used agent as second-line therapy followed by combined gemcitabine/vinorelbine18 and docetaxel.8 PFS after second-line chemotherapy was did not differ between the 3 groups (ALK-positive patients, 2.07 months; EGFR mutation-positive patients, 1.63 months; WT/WT patients, 2.93 months; P = .353).
Table 3. Treatment Responses by Molecular Subtypes
Eighty-six patients were received an EGFR TKI. Specifically, 57 patients received gefitinib, 29 patients received erlotinib, and 4 patients received both agents. For an appropriate comparison of PFS after EGFR TKI treatment, we excluded ALK-positive patients who were enrolled because of their nonresponses to EGFR TKI treatment. Because all 9 ALK-positive patients already were preselected on the basis of their nonresponse to EGFR TKI, it was inappropriate to measure their PFS because this was a selection criterion. If they had been included, then it would have biased the group based on nonresponse to TKI. The EGFR mutation-positive group had a much higher response rate (81%) than the 2 other groups (ALK-positive patients, 0%; WT/WT patients, and 15%). ALK-positive patients who received an EGFR TKI had rapid disease progression and did not respond to EGFR TKI treatment. The median PFS for ALK-positive patients (1.37 months) was shorter than that for the other 2 groups (EGFR mutation-positive patients, 9.80 months; WT/WT patients, 2.07 months; P = .037 vs WT/WT) (Fig. 2).
The median OS of ALK-positive patients was 12.2 months compared with 29.6 months for EGFR mutation-positive patients (P = .001) and 19.3 months for WT/WT patients (P = .127). In a multivariate analysis that included age, sex, stage, smoking status, and histology with a Cox-proportional hazards model, the calculated hazard ratio was 0.446 for EGFR mutation-positive patients and 0.631 for WT/WT patients. Other variables, including histology, did not significantly affect OS. Finally, ALK-positive patients had the shortest (albeit statistically nonsignificant) median OS in the pre-ALK inhibitor era. They did not differ in their response to conventional cytotoxic chemotherapy compared with ALK-WT patients. However, they were more resistant to EGFR TKI treatment, even compared with WT/WT patients.
In this study, we used FISH to identify an historic cohort of ALK inhibitor-naive patients and examined the possible prognostic role of the EML4-ALK translocation in the clinical outcomes of patients with NSCLC. We demonstrated that the OS of ALK-positive patients did not differ statistically from their WT/WT matched comparators, although their survival was numerically shorter. Shaw et al12 evaluated survival outcomes in 17 patients with metastatic, ALK-positive disease by determining PFS and OS. In that study, ALK-positive patients had inferior clinical outcomes compared with EGFR mutation-positive patients, resembling the survival of WT/WT patients. However, the number of events was small within the ALK-positive patient group, the follow-up duration was relatively short, and there were differences in age and smoking status between comparator groups in our study. Moreover, 7 of 17 ALK-positive patients enrolled in the phase 1 crizotinib clinical trial, which may have, as acknowledged by the author, influenced the OS outcome of the study. To minimize imbalances in potential prognostic clinicopathologic characteristics, we performed a 2:1 case matching of ALK-WT patients to ALK-positive patients. This matching took into account age at diagnosis, sex, disease stage, and smoking status. The follow-up period of our study was relatively long, with a median follow-up of 26 months. In addition, ALK inhibitor-related effects on survival fundamentally were ruled out in our study. Although our study had the limitations of a single-center, retrospective design and restricted statistical power because of the small sample size, we carefully controlled for confounding factors in our analyses in an effort to present the comparative clinical course of ALK-positive patients (treated without ALK inhibitors) and ALK-WT patients.
The predictive role of the EML4-ALK translocation in response to EGFR TKI therapy, which has been described in several studies,12, 22, 28, 29 was affirmed in the current study. ALK-positive patients were more resistant to EGFR TKI treatment, even compared with WT/WT patients (P = .037). This result also is in concordance with the laboratory data published in 2008, which demonstrated the resistance of an ALK-positive lung cancer cell line to erlotinib.8 Recent studies repeatedly have reported similar data on the resistance of ALK-positive tumors to EGFR TKI, and a screening strategy for ALK positivity has been proposed based on this characteristic.12, 22, 23
The report by Shaw et al and a previous report from our group also demonstrated an objective response rate to conventional chemotherapy that was numerically smaller in ALK-positive patients versus ALK-WT patients.12, 22 However, this finding was not statistically significant in those studies, either as a true result or as a function of the limited sample size of ALK-positive patients in each study. Larger sample sizes or pooled analyses may address this issue more definitively.
Recent retrospective analyses have indicated that ALK-positive patients were more sensitive to pemetrexed compared with ALK-WT comparators.29, 30 In the current study, the percentage of pemetrexed exposure in any line of ALK-positive, EGFR mutation-positive, and WT/WT groups was 52%, 43%, and 48%, respectively. Despite the relatively high use of pemetrexed in the ALK-positive patients, this group had the shortest OS estimate.
ALK-positive patients had a lower rate of CNS metastases (30%) during the follow-up period compared with EGFR mutation-positive patients (63%), and this rate was identical to that of the WT/WT patients (30%). This result may have been caused by bias, because EGFR mutation-positive patients had longer survival compared with the other 2 groups. Six ALK-positive patients (26%), 15 EGFR mutation-positive patients (33%), and 9 WT/WT patients (20%) had CNS metastases confirmed at their initial staging workup. To compare the rate of CNS metastasis in terms of a survival-related effect, a prospective survey would be helpful.
A significant portion of ALK-positive patients (22%) identified in this study had a smoking history, although 3 patients in the heavy-smoker group had 10 pack-year smoking history, which is a borderline value for heavy smoker (defined as a smoking history of ≥10 pack-years). Similarly, a recent study by our group in a different patient population also reported a large number of smokers (31%) in the ALK-positive group.22 These finding suggest that smoking status is not appropriate for patient selection in ALK testing. Smoking history should be approached and interpreted with caution, because it can vary in different cultural and social contexts. Although we excluded squamous cell histology in selecting our patients, our cohort had a significant portion of nonsmall cell carcinoma, not otherwise specified (NOS). We identified 7 ALK-positive patients with NOS histology, including 3 who had an adenocarcinoma immunophenotype that demonstrated expression of thyroid transcription factor-1 and cytokeratin 7. Several previous studies also have reported a small number of ALK-positive patients with nonadenocarcinoma histology.12, 29-31 In addition, misclassification in histology can occur in tumors that are difficult to specify; especially in patients who have small specimens harvested by needle biopsy or aspiration.32 Therefore, we should take care to restrict ALK testing in patients with adenocarcinoma histology, because we can miss a small number of ALK-positive patients who have large cell, NOS, or other minor histology.
In our current study, ALK positivity was suggestive of a poor prognosis (although the finding not statistically significant) and was predictive of poor EGFR TKI outcomes. With the historically dismal survival observed across the unselected NSCLC patient population, this finding may signify an even greater unmet medical need within the ALK-positive subset of patients with NSCLC, and these patients are in need of effective therapy and do not benefit from currently available targeted agents. With the development of molecularly targeted therapy, such as crizotinib, positive ALK status may prove to be a positive predictive marker for ALK inhibitor therapy in the near future.
This study was supported by a grant of the Korean Health Technology R&D Project, Ministry of Health and Welfare, Republic of Korea (A110057), and was funded in part by an outcomes research agreement with Pfizer.
CONFLICT OF INTEREST DISCLOSURES
Dong-Wan Kim has acted as a consultant and has received honoraria from Pfizer. Kimary Kulig is an employee of Pfizer. Yung-Jue Bang has acted as a consultant, has received honoraria, and has received research funding from Pfizer.