The Oncologist, Vol. 12, No. 1, 90-98, January 2007; doi:10.1634/theoncologist.12-1-90 © 2007 AlphaMed Press
Response to Treatment and Survival of Patients with Non-Small Cell Lung Cancer Undergoing Somatic EGFR Mutation TestingMassachusetts General Hospital Cancer Center, Harvard Medical School/Partners HealthCare Center for Genetics and Genomics, Massachusetts General Hospital Department of Pathology, Dana-Farber Cancer Institute, Boston, Massachusetts, USA Key Words. Lung cancer • EGFR • Mutation • Genetic screening Correspondence: Lecia V. Sequist, M.D., MPH, MGH Cancer Center, 32 Fruit Street, Yawkey Suite 7B, Boston, Massachusetts 02114, USA. Telephone: 617-724-4000; Fax: 617-726-0453; e-mail: lvsequist{at}partners.org Received August 3, 2006; accepted for publication September 29, 2006.
Access and take the CME test online and receive 1 AMA PRA Category 1 CreditTM at CME.TheOncologist.com
Somatic mutations in the epidermal growth factor receptor (EGFR) gene are associated with clinical response and prolonged survival in patients with non-small cell lung cancer (NSCLC) treated with EGFR tyrosine kinase inhibitors (TKIs). We began screening patients for somatic EGFR mutations by DNA sequencing as part of clinical care in 2004. We performed a retrospective cohort study of 278 patients with NSCLC referred for EGFR testing over a 10-month period. Tumor samples underwent direct DNA sequence analyses of EGFR exons 18 through 24. We determined the clinical characteristics and EGFR mutation status of the patients and analyzed their response to therapy and survival. EGFR somatic mutations were identified in 68 (24%) of patients. A minimal smoking history was the strongest clinical predictor of harboring a mutation. In multivariable analyses, each pack-year of smoking corresponded to a 5% decreased likelihood of having an EGFR mutation. Among 92 patients with unresectable disease undergoing subsequent systemic therapy, EGFR mutations were associated with an increased response rate to EGFR TKIs (p < .0001) but not chemotherapy. Overall survival was significantly prolonged in EGFR mutation-positive patients (p = .001), with a median survival of 3.1 years compared with 1.6 years in mutation-negative patients, after adjusting for age, gender, and stage at diagnosis. Integrating molecular profiling into clinical care is feasible in NSCLC patients and provides useful clinical information.
Non-small cell lung cancer (NSCLC) is the most frequent cause of cancer death in the world, yet progress with chemotherapeutic agents in this disease has reached a plateau [1, 2]. Therefore, new treatment approaches are needed. Targeting the epidermal growth factor receptor (EGFR) has played a central role in advancing NSCLC research initiatives, treatment, and patient outcome over the last several years. Novel targeted therapeutic agents, including the small-molecule tyrosine kinase inhibitors (TKIs) gefitinib and erlotinib and the monoclonal antibody cetuximab, have been developed to interfere with EGFR signaling [3]. Soon after the TKIs were introduced, it became apparent that certain patient characteristics, including a never-smoking history, adenocarcinoma tumor histology, female gender, and East Asian origin, were correlated with improved response to the drugs and improved survival after treatment [48]. One of the explanations for the improved outcome in these patient subsets was elucidated in 2004, when somatic mutations within the EGFR gene itself were discovered [9 11]. EGFR mutations are associated with both TKI response and prolonged survival and occur more frequently in the NSCLC patient subsets mentioned above [916]. The mutations cause EGFR to transduce an intracellular antiapoptotic signal, upon which the cancer cells become "addicted" [17]. Administered EGFR TKIs interrupt the necessary survival signal, thus rendering the cell exquisitely susceptible to apoptosis. The initial observation that patients harboring an EGFR mutation had an increased chance of response to TKI treatment prompted us to offer EGFR mutation testing by DNA sequencing as part of clinical cancer care and protocol therapy, beginning in August 2004. Herein we report clinical and molecular characteristics, response to TKI treatment and chemotherapy, and survival of patients screened within the first 10 months of this program.
Study Design We performed a cohort study of all NSCLC patients referred for somatic EGFR kinase domain sequencing as part of standard and clinical trial care between August 2004 and June 2005 at Massachusetts General Hospital (MGH), Dana-Farber Cancer Institute (DFCI), and Brigham and Womens Hospital (BWH), and of all available patients from other institutions that were referred through the Laboratory of Molecular Medicine (LMM) at the Harvard Medical School/Partners HealthCare Center for Genetics and Genomics (CLIA no. 22D1005307). The study period was chosen to allow a minimum of 1 year of potential follow-up time. Clinicians had selected which patients to refer for testing; however, patients needed to have sufficient and appropriate tumor specimens available with tumor cells comprising at least 50% of the specimen based on histologic examination by MGH and BWH reference pathologists. Samples could be either paraffin-embedded or frozen tissue. Due to the low frequency of EGFR mutations in squamous cell tumors, patients with this diagnosis were not eligible for DNA sequencing of EGFR [18]. We identified the first 267 consecutive NSCLC patients referred for EGFR DNA sequencing from MGH, DFCI, and BWH using the EGFR case log maintained at the LMM. This cohort included all patients with NSCLC referred for testing during the study period. Patient demographics, cancer history, EGFR test results, subsequent treatments, response to treatments, and survival were documented using structured physician chart review. Response was assessed from medical record review of radiology reports and physician office notes. A prospective independent assessment of films was not performed. Response rate was defined as the number of patients with partial response to a therapy divided by the total number of patients receiving that therapy. Survival was defined as the interval from the date of diagnosis until the date of death. Patients without a known date of death were censored for survival analyses at the last date they were known to be alive. All reported deaths were independently confirmed using the Social Security Death Index. Detailed smoking histories were collected via prospective collection and chart review. Patients were categorized using standard criteria that included former smokers, defined as patients who had quit smoking at least 1 year before their diagnosis of lung cancer, and never-smokers, defined as patients who had smoked less than 100 cigarettes in their lifetime [19]. Pack-years of smoking were calculated by multiplying the number of packs smoked per day by the number of years of smoking. This study was approved by the Institutional Review Board at Dana-Farber/Harvard Cancer Center. In addition, all clinicians from other institutions that had referred patients to the LMM for EGFR testing during the study period were contacted via questionnaire to provide the equivalent follow-up information about their patients. Sixty-three patient-specific questionnaires were sent to 53 individual referring physicians; 15 physicians responded regarding 17 (27%) tested patients, and 11 (17%) questionnaires supplied adequate information for inclusion. The questionnaire was approved by the Institutional Review Board at Partners Health Care. Complete data were available for age, gender, tumor histology, and EGFR mutation status. There were missing data for race (13%), smoking status (4%), and subsequent treatment information (18%). A subset of the patients included in this cohort has been reported in other series from our institution [20, 21].
EGFR Kinase Domain Sequencing
Statistical Analysis
Patient Characteristics Among the 278 patients in the study cohort, the mean age was 63 years, and 69% were female (Table 1
Mutations Identified Among 278 patients screened, 68 (24%) were found to have at least one mutation in the EGFR kinase domain, with six of these patients demonstrating two point mutations each, and one patient with three point mutations, for a total of 76 mutations identified (Table 2
The most common mutations observed were the in-frame exon 19 deletions in 30 patients (39%) and the exon 21 point mutation L858R in 25 patients (33%). Other mutations identified included six point mutations in exon 18 (8%), five insertion/duplications in exon 20 (7%), eight point mutations in exon 20 (10%), and two point mutations at exon 21 codons other than 858 (3%). One of the somatic exon 20 point mutations identified did not result in an amino acid change (L778L). The significance of this synonymous nucleotide change is not known. Four patients (5%) had the point mutation T790M in exon 20, which has been associated with acquired resistance to gefitinib and erlotinib [22, 23]. Three of these four patients also had concurrent L858R-sensitizing mutations. One such L858R/T790M patient had been known to harbor only the L858R mutation initially and had enjoyed a response to gefitinib for 6 months before developing clinically resistant disease that was shown to harbor both L858R and T790M. The other three patients with T790M had not received any EGFR TKI at the time their biopsy revealed the T790M mutation. Of the seven patients with multiple point mutations, three had the above-mentioned combination L858R plus T790M. The other four patients with multiple mutations all harbored well-described sensitizing-substitution mutations at codon 719 in exon 18, together with infrequently described or not previously described point mutation(s) in exons 18 (n = 1) or 20 (n = 3) [9, 10, 13, 18, 24]. None of these four patients had been exposed to an EGFR TKI prior to EGFR mutation testing.
Predictors of Mutations
The most prominent predictor of somatic mutations in EGFR was a lack of cigarette smoking. Never-smokers were 5.6-fold more likely to have an EGFR mutation than ever-smokers (p < .0001). When examining pack-years of smoking as a continuous variable, the mean number of pack-years smoked was significantly lower among EGFR mutation-positive patients (5.9 pack-years) compared with EGFR mutation-negative patients (30.8 pack-years, p < .0001). There was no obvious inflection point in the chance of harboring a mutation over the range of number of pack-years smoked, but rather for each additional pack-year smoked there was a 5% decrease in the likelihood of having a mutation (p < .0001). The number of pack-years of smoking remained a significant predictor of mutation status in a multivariable logistic regression model after controlling for gender, race, and tumor histology (p < .0001). None of the clinical characteristics examined were associated with one type of mutation compared with another, that is, deletions versus point mutations.
Response to Treatment and Survival
Among the seven patients with multiple somatic EGFR mutations, four had metastatic disease and were treated with an EGFR TKI after mutation screening. Two patients had concurrent L858R/T790M and were refractory to subsequent TKI treatment (one had erlotinib and one had an experimental EGFR TKI as part of a clinical trial). Two patients had other multiple EGFR mutations, including a TKI-sensitizing point mutation at codon 719, and both of these had stable disease with gefitinib treatment.
Median follow-up was 13.2 months (range, 0.1137 months). There were 78 deaths, including 68 in the wild-type patients and 10 in the patients with an EGFR mutation. Median survival was significantly longer for EGFR mutation-positive patients (not yet reached) compared with EGFR mutation-negative patients (3.6 years, log-rank p = .001; Fig. 1
We found somatic EGFR mutations in 68 (24%) of the 278 patients with NSCLC screened as part of clinical cancer care at our institutions over a 10-month period. Smoking status was the strongest predictor of harboring an EGFR mutation, with each pack-year of increase in smoking history associated with a 5% decreased likelihood of being mutation-positive, after adjusting for gender, race, and histologic subtype. Asian origin was also associated with harboring a mutation in a univariate model, and tumors with BAC features demonstrated a trend toward mutations compared with adenocarcinoma without BAC features. For patients with advanced disease, EGFR mutations were associated with an improved response to treatment with EGFR TKI agents, but there was no significant difference in response to chemotherapy by mutation status. Among all patients, EGFR mutations were associated with prolonged survival after adjusting for age, gender, and stage of disease. The EGFR mutation rate among our cohort was 24%, which is substantially higher than rates documented by our and other U.S. groups that tested unselected available NSCLC tumor samples [911, 25] and suggests that preselecting patients for molecular screening based on clinical characteristics associated with EGFR mutations is a useful strategy to increase the chance of identifying patients with mutations. Indeed, our cohort demonstrated an increased prevalence of such characteristics. For example, 95% of our patients had adenocarcinoma or BAC tumor histology compared with 61% in the general NSCLC population [26]. Similarly, although never-smokers made up 24% of our population (27% of our women and 19% of our men), the incidence of never-smokers in the general NSCLC population has been reported as 2%10%, although it may be as high as 27% in women with NSCLC [2731]. Although our clinically selected population had a relatively high prevalence of mutations compared with the overall population, a rate of 24% is still modest and reminds us that molecular diagnostics greatly increase the accuracy of clinical selection criteria alone in identifying patients that carry somatic EGFR mutations. Our finding that smoking history is the most robust predictor for mutations is consistent with other series [11, 15, 25, 32]. However, in our population, there was no clear cutoff for pack-years of smoking history above which the incidence of mutations fell sharply, in contrast to data from Pham et al., who have reported that EGFR mutations are much less likely with more than a 15-pack-year smoking history [32]. The distribution of EGFR mutation types identified in our cohort was consistent with other series, with most mutations being in-frame exon 19 deletions (39%) and the L858R point mutation in exon 21 (33%). These two most frequent EGFR mutations have been consistently associated with TKI-responsiveness [9 13]. We also observed substitutions at codon 719 in exon 18 and the point mutation L861Q in exon 21, both of which have been previously described and associated with TKI-responsiveness [9, 13]. The T790M point mutation in exon 20 was initially described as a secondary resistance mutation, acquired after prior exposure to a TKI agent [22, 23]. In our cohort, one patient had concurrent L858R and T790M mutations and matched this clinical scenario, yet there were three other patients who lacked prior TKI exposure and were found to harbor T790M, either alone or in combination with L858R. Identification of T790M in TKI-naïve patients has been described in other series but is rare, and the clinical implications are not yet elucidated [3335]. We have previously described a family with a germline T790M mutation that displayed a high incidence of BAC tumors, suggesting that this mutation may play a role in genetic susceptibility to lung cancer [36]. Further clinical studies examining the implications of T790M are ongoing, including trials of second-generation EGFR TKIs that may be able to circumvent the T790M steric hindrance resistance mechanism [37]. In addition, other rare mutations were identified in our cohort, including point mutations in exon 18, 20, and 21 and five cases with insertion mutations in exon 20. Our group previously reported that exon 20 insertion mutations are transformative but do not confer gefitinib or erlotinib sensitivity in cell culture models [38]. None of the patients in this series with rare point mutations or exon 20 insertions responded if treated with EGFR TKI agents. In future studies, information should be collected and reported separately for these mutations and other novel mutations, as their significance is not yet clear. We found that EGFR mutations were associated with increased response to EGFR TKI agents in patients with metastatic disease, as has been previously reported [916, 24, 39]. Furthermore, response to chemotherapy did not differ by mutation status, in line with the molecular subgroup analysis of TRIBUTE, a randomized study of chemotherapy plus either erlotinib or placebo that showed mutation status affected response rate only when treatment included erlotinib [40]. Patients with deletion mutations have been reported to have an increased response rate, longer time to progression, and longer survival following treatment with TKI compared with the L858R point mutation [21, 41]. This trend was observed in the response rate to TKI treatment in our cohort but was not statistically significant, perhaps because of insufficient power. Patients with EGFR mutations had significantly prolonged survival compared with mutation-negative patients, with a hazard ratio for death of 0.68 and a doubling of the median survival time after adjusting for age, gender, and stage at diagnosis. This was a heterogeneous population receiving many different treatment modalities and therapeutic regimens; therefore, this finding suggests that EGFR mutation is a favorable prognostic factor regardless of treatment. This conclusion has been drawn by other investigators, including the TRIBUTE investigators, where mutation-positive patients had a prolonged survival regardless of treatment arm (8 months vs. 5 months in the mutation-positive and -negative groups, respectively, p < 0.001) [40, 42]. However, a prospective randomized study of an EGFR-targeted agent compared with another type of treatment, ideally in EGFR mutation-positive patients exclusively, is required to determine whether EGFR mutations may be both prognostic and predictive of increased survival. Such a trial has not yet been performed. Our study is unique in its description of a large series of patients undergoing EGFR mutation testing as part of standard clinical practice, but it is subject to some limitations. Because we could only evaluate patients whose tumor samples were logged in to our system for EGFR mutation testing, we do not have a method of determining the characteristics and the number of patients whose clinicians wanted to refer them for testing but lacked sufficient and appropriate tumor specimens. Since patients in our population were not prospectively enrolled in a clinical trial, we do not have complete information about clinical characteristics and follow-up for all patients studied. Our sample size was limited to 68 mutation-positive patients, and we therefore lacked the power to perform some analyses, such as examining differences in outcome between various subtypes of EGFR mutations. Finally, increased EGFR gene copy number has been reported to correlate with increased survival in TKI-treated NSCLC patients in some series [16, 43, 44]. It would be interesting to compare the outcome of patients with EGFR mutations and those with increased EGFR gene copy number in this cohort; however, analysis of EGFR gene copy number was not performed for most of these patients. In conclusion, we have demonstrated that EGFR mutation screening can be incorporated into the clinical care of NSCLC patients. In a clinically selected population, the mutation-positive rate is approximately 25%. This implies that designing prospective clinical trials that require molecular profiling is feasible, especially if using clinical selection criteria prior to molecular diagnostics. We have confirmed that patients identified as harboring a mutation have an increased response to EGFR TKI therapy and a prolonged survival that may be independent of treatment. Most importantly, we believe that molecular profiling will likely be an integral component of the future clinical care of cancer patients. Across our medical system, it is of paramount importance to learn how to efficiently and effectively provide individual genetic information to a patient and their oncologist, both at the time of diagnosis and at the time of failure of a particular treatment.
D.A.H., B.E.J., P.A.J., T.J.L., and M.M. are part of a pending patent application on EGFR mutation testing.
This research was supported in part by National Service Research Award Grant T32 CA 09001.
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||