LUNG CANCER is the most frequent cause of cancer death in

Transcription

1 Epidermal Growth Factor Receptor in Non Small-Cell Lung Carcinomas: Correlation Between Gene Copy Number and Protein Expression and Impact on Prognosis By Fred R. Hirsch, Marileila Varella-Garcia, Paul A. Bunn Jr, Michael V. Di Maria, Robert Veve, Roy M. Bremnes, Anna E. Barón, Chan Zeng, and Wilbur A. Franklin Purpose: The epidermal growth factor receptor (EGFR) is frequently overexpressed in non small-cell lung carcinoma (NSCLC), and EGFR inhibitors are promising new therapeutic agents. The molecular mechanisms responsible for EGFR overexpression are poorly understood. Materials and Methods: Gene copy number and protein status of EGFR were investigated in microarrayed tumors from 183 NSCLC patients, including squamous cell carcinoma (SCC; 89 patients) and non-scc (94 patients) histologies. Protein expression was assessed by immunohistochemistry on a scale from 0 to 400 (percentage of positive cells staining intensity). Gene and chromosome 7 copy numbers were identified by fluorescent in situ hybridization (FISH). Results: EGFR protein overexpression was observed in 62% of the NSCLC (25% scored 201 to 300; 37% scored 301 to 400), more frequently in SCC than non-scc (82% v 44%; P <.001), and in 80% of the bronchioloalveolar carcinomas. LUNG CANCER is the most frequent cause of cancer death in the Western world 1,2 and has a poor prognosis. Less than 15% of patients survive 5 years. 2 Epidermal growth factor receptor (EGFR; erbb-1) is a member of the erbb family of tyrosine kinase receptor proteins, which also include erb-b2 (HER2/neu), erb-b3, and erb-b4. These receptors play an important role for tumor cell survival and proliferation. 3-7 EGFR overexpression has also been demonstrated in premalignant bronchial epithelium, suggesting that EGFR plays an important role in lung carcinogenesis In lung carcinomas, EGFR is more commonly overexpressed From the Departments of Pathology, Medicine, and Preventive Medicine and Biometrics, University of Colorado Health Sciences Center; and the Tobacco Related Malignancy Program, University of Colorado Comprehensive Cancer Center, Denver, CO. Submitted November 13, 2002; accepted April 15, Supported in part by the National Cancer Institute grants Cancer Center Core Grant 2P30-CA46934, Specialized Program of Research Excellence P01-CA58187, and Early Detection Research Network U01-CA A career development grant was awarded to F.R.H. from the International Association for the Study of Lung Cancer/Cancer Research Foundation of America. Authors disclosures of potential conflicts of interest are found at the end of this article. Address reprint requests to Fred R. Hirsch, MD, PhD, Departments of Medicine and Pathology, Campus Box B188, University of Colorado Health Sciences Center, Denver, CO 80262; Fred.Hirsch@uchsc.edu by American Society of Clinical Oncology X/03/ /$20.00 The prevalent FISH patterns were balanced disomy (40%) and trisomy (38%) for EGFR gene and chromosome 7 (40%), whereas balanced polysomy was seen in 13% and gene amplification was seen in 9% of the patients. Gene copy number correlated with protein expression (r 0.4; P <.001). EGFR overexpression or high gene copy numbers had no significant influence on prognosis. Conclusion: EGFR overexpression is frequent in NSCLC, is most prominent in SCC, and correlates with increased gene copy number per cell. High gene copy numbers per cell showed a trend toward poor prognosis. It will be important to evaluate EGFR gene and EGFR protein status and signal protein expression to properly interpret future clinical trials using EGFR inhibitors. J Clin Oncol 21: by American Society of Clinical Oncology. than HER2/neu. 12,13 The prognostic association of EGFR overexpression in lung cancer, however, is a controversial issue. Several reports indicated that EGFR was associated with a poor prognosis, whereas no prognostic association was reported by other reports The EGFR levels were evaluated by immunohistochemistry (IHC) in most studies, but other methods also have been reported. 23 Different conclusions regarding prognostic significance may reflect differences in detection methods, reagents, assay cutoff points, and population characteristics. The EGFR gene, located on chromosome 7p12, 24 has been found amplified in low frequency by Southern blot 25 and polymerase chain reaction analyses. 26 Inhibition of EGFR by different classes of blocking agents was demonstrated to induce apoptosis, reduce proliferation, and inhibit angiogenesis in lung cancer cells, resulting in suppression of tumor growth in vitro and in vivo in a variety of models. 7,11 More recently, clinical trials with EGFR inhibitors have shown promising results in non small-cell lung cancer (NSCLC). 7 However, the correlation between EGFR expression and treatment response, and the optimal method to determine EGFR levels in the tumors, are undefined. This study was designed to investigate the association between EGFR protein expression evaluated by IHC and gene status evaluated by fluorescence in situ hybridization (FISH) techniques, and to verify whether EGFR expression evaluated by these two methods was associated with prognosis in a large series of tissue microarrayed NSCLC Journal of Clinical Oncology, Vol 21, No 20 (October 15), 2003: pp DOI: /JCO

2 EGFR IN NON SMALL-CELL LUNG CANCER 3799 Fig 1. Distinct levels of protein expression in non small-cell lung cancer. Panels illustrate specimens graded with scores (A) 100 (intensity pattern 1 in 100% of cells), (B) 180 (pattern 2 in 90% of cells), (C) 300 (pattern 3 in 100% of cells), and (D) 400 (pattern 4 in 100% of cells). Tumor Specimens MATERIALS AND METHODS Anonymized samples from 183 consecutive patients diagnosed with NSCLC at pathologic stages I to III from 1993 through 1999 were obtained from the archives of the University of Colorado Cancer Center (Denver, CO) and the Johns Hopkins Medical Institution (Baltimore, MD), in protocols approved by the respective institutional review boards. Patients whose samples were inadequate for microarray analysis were excluded. Surgical resections were performed in 172 patients. In the other 11 patients, examinations were made on lymph nodes obtained from mediastinoscopy. The tumors were staged according to the tumor-node-metastasis classification, and histologically classified according to the WHO guidelines. 27 All tumor tissues were independently reviewed by three authors (W.A.F., R.V., and F.R.H.). Microarray Construction Formalin-fixed, paraffin-embedded tissue blocks were used for tissue microarray construction according to the previous description. 13 Briefly, blocks were selected after the sections stained with hematoxylin and eosin were evaluated for tumor viability, and the tissue microarrays were assembled using a Beecher tissue-arraying device (Beecher Instruments, Silver Spring, MD). A large 1.5-mm-diameter stylet was used and three separate core samples of tumor were obtained to account for tumor heterogeneity. Normal lung and 15 other control specimens were included in each of the tissue array blocks. Subsequently, 4- m sections were cut from the array blocks with a Leica microtome (Leica Instruments GmbH, Hubloch, Germany), transferred to adhesive-coated slides using the adhesive-coated tape sectioning system (Instrumedics Inc, Hackensack, NJ), and exposed to ultraviolet light for 60 seconds to seal the sections to the slides. EGFR IHC Assay and Grading Microarrayed sections were deparaffinized, hydrated, predigested for 10 minutes with protease 1 (Ventana Medical, Tucson, AZ), and washed in Tris buffer. Peroxide blocking was performed with 3% H 2 O 2 in methanol at 4 C for 20 minutes, followed by biotin blocking (avidin for 10 minutes and biotin for 10 minutes; DAKO Biotin Blocking System, Carpinteria, CA) and Power Block (BioGenex, San Ramon, CA) for 10 minutes at a 1/10 dilution in Tris-bovine serum albumin. The mouse monoclonal EGFR antibody (No ; Zymed Laboratories Inc, San Francisco, CA), was applied at a final concentration of 1.5 g/ml and incubated for 1.5 hours at 37 C. Incubation with the secondary antibody (Biotinylated Multi-Link; DAKO) with 40% human serum was performed for 30 minutes, followed by application of streptavidin horseradish peroxidase enzyme complex for 30 minutes and diaminobenzidine chromogen for 15 minutes. The slides were then counterstained in hematoxylin and topped with a coverslip. The specimens were evaluated by light microscopy, using a 10 magnification objective; the microscopist had no knowledge of the case histories. Only clear staining of the tumor cell membranes was considered positive. Diffuse cytoplasmic or granular staining was diagnosed as negative. A semiquantitative approach was used to generate a score for each tissue core, as follows. The percentage of positive tumor cells per slide (0% to 100%) was multiplied by the dominant intensity pattern of staining (1, negative or trace; 2, weak; 3, moderate; 4, intense); therefore, the overall score ranged from 0 to 400 (Fig 1). Specimens with scores 0 to 200, 201 to 300, and 301 to 400 were respectively classified as having negative or low, intermediate, and high levels of expression. Specimens with low levels of expression were reevaluated for the presence or absence of membranous or cytoplasmic staining using a 40 objective. To estimate the interobserver variability regarding the IHC scoring, 60 randomly selected cores were evaluated by the same three authors (W.A.F., R.V., and F.R.H.). FISH Assay and Analysis Microarrayed sections were incubated at 56 C overnight, deparaffinized in xylene washes, and dehydrated in 100% ethanol. After incubation in 2 saline sodium citrate buffer (SSC; ph 7.0) at 75 C for 20 to 25 minutes, sections were digested with proteinase K (0.25 mg/ml in 2 SSC; ph 7.0) at 37 C for 20 to 25 minutes, rinsed in 2 SSC (ph 7.0) at room temperature for 5 minutes, and dehydrated in an ethanol series. Dual-target, dual-color

3 3800 HIRSCH ET AL FISH assays were performed using the LSI EGFR SpectrumOrange/CEP 7 SpectrumGreen probe (Vysis, Downers Grove, IL). The probe set was applied to the selected area on each slide, and the hybridization area was covered with a glass coverslip and sealed with rubber cement. The slides were incubated at 80 C for 10 minutes for codenaturation of chromosomal and probe DNA, and hybridization occurred in a humidified chamber at 37 C for 20 hours. Posthybridization washes were performed in 1.5 M urea and 0.1 SSC (ph 7.0 to 7.5) at 45 C for 30 minutes and in 2 SSC for 2 minutes at room temperature. After dehydration in an ethanol series, 4 6-diamidino-2-phenylindole (DAPI; 0.15 mg/ml in Vectashield mounting medium, Vector Laboratories, Burlingame, CA) was applied for chromatin counterstaining. Analysis was performed on an BX60 brightfield and epifluorescence microscope (Olympus B 60, Olympus America, Lake Success, NY) equipped with the Quips XL genetic workstation (Applied Imaging, Santa Clara, CA). The EGFR sequence was visualized with a Texas red filter, the chromosome 7 centromere sequence was visualized with a fluorescein isothiocyanate (FITC) filter, and the nuclei were identified with a DAPI filter. Double (FITC and Texas red) and triple band pass filters (DAPI, FITC, and Texas red; Chroma Technology, Brattleboro, VT) were also used. Representative images of each specimen were acquired with a SenSys cooled CCD camera (Photometrics, Tucson, AZ) in monochromatic layers that were subsequently merged by the SmartCapture software (Vysis). At least 200 nonoverlapping interphase nuclei per core were scored by two independent observers (M.D. and M.V.G.), who followed strict scoring guidelines and used constant adjustment of microscope focus because signals were located at different focal planes. In each nucleus, the number of copies of EGFR and chromosome 7 probes was assessed independently. Four major FISH patterns were identified: disomy for both the EGFR gene and chromosome 7 (balanced disomy); similar gains in copy numbers for gene and chromosome in low (balanced trisomy) and high levels (balanced polysomy); and a clustered unbalanced gain of the EGFR gene (gene amplification). Gene amplification was classified at a low level when the gene-to-chromosome ratio ranged between 2.1 and 3.0, and at a high level when the gene-to-chromosome ratio was more than 3. In rare instances in which different patterns were found among the cores, the patient was classified by the most severe FISH pattern. Statistical Analysis Three cores were taken from each tumor to account for tumor heterogeneity, and the analyses were done per core and per patient (IHC used the average of the three cores; FISH used the pattern showing the highest gene copy number). The interobserver reliability was tested by a two-way intraclass correlation coefficient. FISH and IHC results were compared using the exact linear-by-linear association test of Mantel-Haenszel and results were compared across histologies using 2 methods with exact P values. Correlation was estimated using the Spearman rank method. Univariate analyses associating FISH, IHC, and clinical parameters with lung cancer survival were performed using the Kaplan-Meier method and log-rank tests. Where appropriate, continuous variables were categorized before analysis. Tests for linear trend were used for ordinal variables with more than two categories. To assess the impact of EGFR by FISH and IHC on lung cancer survival in the presence of all other variables, multivariate analysis was performed using the Cox proportional hazards model. Only variables from the univariate analyses that were significant at the.05 level were entered into the Cox regression analysis. We performed manual backward elimination using likelihood ratio tests with a significance level of.05. Data analysis was performed using SAS Version 8.1 (SAS Institute Inc, 1999 to 2000, Cary, NC). RESULTS We studied 183 NSCLC patients with an average of 2.6 assessable tissue cores per patient. Table 1 summarizes the demographic and clinical data. There were 107 males and 76 females with a median age of 65 years (range, 34 to 86 years). Smoking status was available for 140 patients: 66 former smokers (36%), 62 current smokers (34%), and 11 never smokers (6%). Most patients (89 patients; 49%) had squamous cell carcinoma (SCC), whereas 69 patients (38%) had adenocarcinoma (ADC), 10 patients (5%) had bronchioloalveolar carcinoma (BAC), and 15 patients (8%) had large-cell carcinoma (LCC). The majority of patients had poorly differentiated histology (52%), pathologic stage I (53%), and negative surgical margins (92%). The mean follow-up was 51 months from surgery. At the last follow-up, 60 patients had died of lung cancer and 123 patients were alive. The median survival time was not reached. The 75th percentile survival time (21 months; 95% CI, 13 to 39 months) was considered in the Kaplan-Meier method. EGFR Protein Expression The reproducibility of the IHC evaluation in 60 randomly selected specimens analyzed by three observers was high, with an intraclass correlation coefficient of The IHC score was estimated as the average of all tissue cores analyzed and the results are summarized according to clinicopathologic data in Table 1. EGFR expression was not associated with age, sex, smoking status, pathologic stage, or tumor or node status. Significant differences were associated with histologic differentiation, with well-differentiated tumors expressing higher levels of EGFR than the poorly differentiated tumors. A statistically significant difference in the EGFR expression was observed across the histological subtypes (P.001; Table 2). The EGFR expression was highest in SCC (mean, 299; range, 184 to 400) and BAC (mean, 281; range, 148 to 366), and lower in ADC (mean, 182; range, 32 to 304) and LCC (mean, 150; range, 6 to 284). The majority (114 patients; 62%) had either intermediate (score 201 to 300; 46 patients; 25%) or high levels (score 301 to 400; 68 patients; 37%) of EGFR expression. Among the 89 SCC patients, 82% had intermediate (31%) or high (51%) EGFR expression, which was significantly different from expression observed in non-scc histologies (P.001). Conversely, ADC and LCC had mostly negative or low levels of expression (59% and 67%, respectively). Among BAC tumors, 80% had intermediate or high levels of EGFR expression; however, the level of expression in BAC was not significantly different from expression levels in other tumors (P.40). EGFR Gene Copy Numbers The four major FISH patterns (balanced disomy, balanced trisomy, balanced polysomy, and gene amplification) are illustrated in Figure 2. The patients FISH patterns identified by two independent observers showed a high correlation (r 0.95), suggesting that the selected criteria for scoring were reproducible. The average gene copy number per cell and the EGFR-tochromosome 7 ratio for the major FISH patterns are listed in Table 3. Seventy-four patients (40%) showed balanced disomy, with the averages of EGFR and chromosome 7 signals per cell ranging between 1.6 and 2.0 and the gene-tochromosome ratio ranging from 0.8 to 1.2. Seventy patients (38%) had balanced trisomy, with the averages of gene and chromosome copies per cell ranging from 2.2 to 3.0 and the chromosome-to-gene ratio ranging from 0.9 to 1.3. Twenty-

4 EGFR IN NON SMALL-CELL LUNG CANCER 3801 Table 1. Characteristics of the 183 NSCLC Patients, Average Scores for Protein Expression, and P Values for Group Comparison Characteristic Patients EGFR Protein Expression EGFR Gene Status No. % Mean SD P* P Age, years Sex Male Female Smoking status Never Former Current Histology SCC ADC BAC LCC Histologic differentiation Well differentiated Moderate Poorly differentiated Stage I II III Tumor status T T T T Node metastasis N N N Lung cancer death Yes No Abbreviations: NSCLC, non small-cell lung cancer; EGFR, epidermal growth factor receptor; SD, standard deviation; SCC, squamous cell carcinoma; ADC, adenocarcinoma; BAC, bronchoalveolar carcinoma; LCC, large-cell carcinoma; IHC, immunohistochemistry; ANOVA, analysis of variance; FISH, fluorescent in situ hybridization. *Average IHC score compared by t test (age, gender, lung cancer death), one-way ANOVA (smoking status, histology), or linear association (stage, differentiation, tumor status, node metastasis). Distribution of patients among the four FISH patterns compared by Mantel-Haenszel linear test. three patients (13%) had balanced polysomy, with the averages per cell for the gene and chromosome 7 copies ranging from 3.1 to 4.4 and the ratio of gene-to-chromosome from 0.8 to 1.4. Sixteen patients (9%) showed EGFR gene amplification, of whom five (3%) had a low level of amplification (gene-to-chromosome ratio ranging between 2.1 and 2.5), and 11 patients (6%) had a high level of amplification, mostly with multiple overlapped, unscorable signals (gene-to-chromosome ratio 3). The distribution of the FISH patterns was not associated with age, sex, smoking status, histologic differentiation, stage, tumor Table 2. EGFR Protein Expression (mean score and levels of expression) According to the NSCLC Histologic Subtypes Histologic Subtype No. of Patients Score Negative or Low (score 0-200) Intermediate (score ) High (score ) Mean SD No. % No. % No. % SCC ADC BAC LCC Abbreviations: EGFR, epidermal growth factor receptor; NSCLC, non small-cell lung cancer; SD, standard deviation; SCC, squamous cell carcinoma; ADC, adenocarcinoma; BAC, bronchoalveolar carcinoma; LCC, large-cell carcinoma.

5 3802 HIRSCH ET AL Fig 2. Dual-color fluorescent in situ hybridization assays with probes for epidermal growth factor receptor (EGFR; red) and chromosome 7 (CEP7; green). (A) Balanced disomy; (B) balanced trisomy; (C) balanced polysomy; (D) gene amplification. Ideogram shows probe location. Balanced patterns display ratio of gene to chromosome signals ranging from 0.8 to 1.4. (Bottom right) Disomy for chromosome 7 and EGFR gene amplification (ratio > 3). status, or presence of node metastasis (P values in Table 1). Table 4 summarizes the analysis according to the histologic subtypes. High gene copy numbers represented by balanced polysomy and gene amplification were more frequent in SCC (18% and 11% v 7% and 6%, respectively, in non-scc), whereas low gene copy numbers (balanced disomy and balanced trisomy) were more common in non-scc (44% and 42% v 37% and 34%, respectively, in SCC). The distribution of high and low gene copy numbers was statistically different between SCC and non-scc (P.01). Correlation Between EGFR Protein Expression and Gene and Chromosome (FISH) Patterns The association of EGFR protein expression and the FISH patterns is listed in Table 5 and shown in Figure 3. Among the 69 patients with a negative or low level of EGFR expression, 35 patients (51%) had disomic FISH pattern and 33 patients (48%) had balanced trisomy, whereas one patient showed balanced polysomy. In contrast, among the 46 patients with an intermediate level of protein expression, eight patients (17%) had either Table 3. Average Number of Copies per Cell of the EGFR Gene and the Chromosome 7 Centromere FISH Probes, and Ratio of Gene to Chromosome in the NSCLC Specimens Grouped by FISH Pattern FISH Pattern Patients No. % Average EGFR/Cell Average Chromosome 7/Cell Ratio EGFR to Chromosome 7 Balanced disomy Balanced trisomy Balanced polysomy GA Low level High level Abbreviations: EGFR, epidermal growth factor receptor; FISH, fluorescent in situ hybridization; NSCLC, non small-cell lung cancer; GA, gene amplification.

6 EGFR IN NON SMALL-CELL LUNG CANCER 3803 Table 4. EGFR Gene and Chromosome 7 FISH Patterns According to Tumor Histology FISH Pattern Histology Balanced Disomy Balanced Trisomy Balanced Polysomy Gene Amplification Total No. % No. % No. % No. % No. % SCC Non-SCC ADC BAC LCC Total Total Abbreviations: EGFR, epidermal growth factor receptor; FISH, fluorescent in situ hybridization; SCC, squamous cell carcinoma; ADC, adenocarcinoma; BAC, bronchoalveolar carcinoma; LCC, large-cell carcinoma. balanced polysomy or gene amplification, and among 68 patients with high levels of protein expression, 30 patients (44%) had balanced polysomy or gene amplification. Levels of protein expression significantly correlated with copy number of genes per cell on the basis of patient (r 0.39; P.001) and on tissue core results (r 0.40; P.001). Patients with balanced disomy and balanced trisomy showed low mean IHC scores (206 and 213, respectively), whereas the mean IHC score increased when FISH abnormalities became more severe. The mean score was 355 for patients with balanced polysomy and 367 for patients with gene amplification. These results suggested that a low number of EGFR gene copies per cell, including the disomic and trisomic patterns, had not influenced the level of protein expression, whereas a high number of gene copies, both as clustered amplification or balanced polysomy, had an impact on the protein level. The association between EGFR expression and gene levels was also investigated across histologies. The high number of gene copies per cell was closely associated with high protein expression independent of histology (P.05). Balanced disomy and balanced trisomy were more closely associated with low levels of protein expression in ADC and LCC than in SCC (Fig 4). BAC showed a different distribution compared with the other non-sccs, with a strong EGFR protein overexpression and low gene copy number status. Among the 10 BACs, 80% had intermediate or high protein expression, whereas 40% had disomy and 50% had balanced trisomy by FISH. The lack of a statistically significant difference in EGFR expression between BAC and the other NSCLC tumors might be due to the relatively small number of BAC patients included. Prognostic Implication of EGFR Protein Expression Levels and Copy Number of Genes In univariate analysis, age, sex, histology, tumor differentiation, EGFR expression, and gene pattern were not significantly associated with survival. Clinical stage, tumor size, presence of metastatic nodes, and negative surgical margins showed a statistically significant impact on survival (P.05). The Kaplan-Meier survival curves for patients in the different categories of EGFR expression or gene numbers are shown in Figure 5. The 75th percentile survival was 19 months for low EGFR expression, 15 months for intermediate, and 39 months for high expression. The 75th percentile survival was 22 months for disomy or trisomy and 15 months for balanced polysomy or gene amplification. There was no difference in survival among the three IHC score groups (trend test, P.22) or the dichotomized FISH groups (low and high copy numbers of the EGFR gene; log-rank test, P.13). The hazard ratio was calculated for the different combination of FISH results and IHC scores as shown in Table 6, after adjusting for age, sex, stage, and surgical margins. High gene copy numbers combined with low EGFR score by IHC was associated with a significantly increased hazard ratio of 3.17 (95% CI, 1.09 to 9.20) compared with the patients with tumors having low gene copy numbers and low IHC score. Survival curves were also compared for patients grouped according to the protein and gene levels combined (Fig 5C). For Table 5. Levels of EGFR Protein Expression (IHC score), and EGFR Gene and Chromosome 7 FISH Patterns FISH Pattern Balanced Disomy Balanced Trisomy Balanced Polysomy Gene Amplification Total IHC Score No. % No. % No. % No. % No. % Total Average SD Abbreviations: EGFR, epidermal growth factor receptor; IHC, immunohistochemistry; FISH, fluorescent in situ hybridization; SD, standard deviation.

7 3804 HIRSCH ET AL Fig 3. Protein expression levels plotted according to the fluorescent in situ hybridization (FISH) patterns. Immunohistochemistry (IHC) scores increased from the FISH patterns with low copy numbers (balanced disomy and trisomy) to the FISH patterns with high copy numbers (balanced polysomy and gene amplification). this analysis, two FISH groups (low and high gene copy numbers) and two IHC groups (score 300 and score 300) were considered. Survival curves were significantly different among these groups (log-rank test, P.01), with the worse survival rates related to the FISH pattern with high gene copy numbers per cell. However, gene amplification per se did not represent a statistically significant adverse prognosis (P.36). DISCUSSION This study was undertaken to explore the relationship between expression of the EGFR protein assessed by IHC, gene copy Fig 4. Protein expression in the non small-cell lung cancer histologic subtypes according to the fluorescent in situ hybridization (FISH) patterns. Low gene copy numbers were more frequent in adenocarcinoma (ADC) and large-cell carcinoma (LCC), whereas squamous cell carcinoma (SCC) and bronchoalveolar carcinoma (BAC) showed an important fraction of specimens with low gene copy numbers and high level of protein expression. Bal., balanced; Dis., disomy; Tris., trisomy; Polys., polysomy; Amplif., amplification; IHC, immunohistochemistry. number assessed by FISH, and their association to the prognosis. We demonstrated that the majority of NSCLC tumors exhibited either intermediate (25%) or high (37%) levels of EGFR protein expression, and there was a significant correlation between EGFR expression and histologic subtypes, with the highest expression in SCC and the lowest expression in LCC. Furthermore, the study showed that EGFR protein expression was more prominent in well-differentiated than in the poorly differentiated histology. To our knowledge, this is the first study of EGFR gene copy number in lung carcinomas using FISH technology. Four distinct categories of FISH patterns were identified, according to EGFR gene and chromosome 7 copy numbers present per cell. Sixty percent of the patients had tumors with increased number of the EGFR gene copies per cell, most of them with balanced trisomy or polysomy. Gene amplification was seen in only 9% of patients. EGFR protein was overexpressed in all tumors with gene amplification, implicating gene amplification as one mechanism for overexpression. This is consistent with the results of Reinmuth et al 26 using double-differential polymerase chain reaction technology. Furthermore, the correlation between increasing levels of EGFR protein expression and increased gene copy number suggests that the additive effect of gene copies is an important mechanism for EGFR protein expression. We found a similar result regarding the HER2/neu expression in NSCLC patients. 13 However, a fraction of NSCLC tumors, mostly SCC and BAC, showed balanced disomy for chromosome 7 and EGFR gene while expressing high EGFR protein levels, suggesting that high protein expression in SCC might also be controlled by other mechanisms unrelated to the number of genes present per cell. One such mechanism might be alternative splicing of EGFR RNA resulting in an inframe deletion that may be expressed at the cell membrane and was first described in glial tumors that were amplified at the EGFR locus. 28 Preliminary reports utilizing Western blots or IHC have indicated that mutant protein

8 EGFR IN NON SMALL-CELL LUNG CANCER 3805 Fig 5. Actuarial survival curves for non small-cell lung cancer patients stratified for (A) levels of protein expression, (B) levels of copy numbers of the epidermal growth factor receptor (EGFR) gene per cell, and (C) combination of gene (low or high gene copy number) and protein expression (scores < 300 and > 300). IHC, immunohistochemistry; FISH, fluorescent in situ hybridization; pt, patients; Bal, balanced; Dis, disomy; Tris, trisomy; Polys, polysomy; Ampl, amplification. may be present in approximately 15% of NSCLC tumors. 29,30 However, actual splice variant mrna is not reported in lung cancer cell lines 31 and the relevance of mutant or alternatively spliced EGFR mrna remains to be determined. We also sought to determine whether EGFR gene or EGFR protein expression affected survival. The literature contains conflicting data on the relationship between EGFR expression and survival in lung cancer. This variability may be due to heterogeneity of study populations or lack of a standardized assay for determining EGFR status. 12 After adjusting for prognostic factors as age, sex, stage, and surgical margins, patients with high gene copy numbers had a tendency to

9 3806 HIRSCH ET AL Table 6. Multivariate Analysis Adjusted for Age, Sex, Stage, and Surgical Margins Groups No. of Patients Hazard Ratio 95% CI Low FISH/low IHC Low FISH/high IHC to 1.05 High FISH/low IHC to 9.20 High FISH/high IHC to 2.31 NOTE. Patients were grouped according to the dichotomized FISH patterns (low gene copy number is balanced disomy and trisomy; high gene copy number is balanced polysomy and gene amplification) and IHC scores (low level of expression is a score 300; high level of expression is a score 300). Abbreviations: FISH, fluorescent in situ hybridization; IHC, immunohistochemistry. experience shorter survival times, suggesting that the survival was more associated with the gene status than with the protein levels. The question whether the molecular profile of EGFR correlates with the response to therapy with EGFR inhibitors in lung carcinomas was not the focus of this study. A lack of prognostic significance for EGFR expression does not preclude benefit from EGFR inhibitor therapy. Recent clinical studies have shown promising results in patients with NSCLC. 7 Studies with the EGFR tyrosine kinase inhibitors have shown objective responses in 10% to 20% of advanced NSCLC patients who experienced treatment failure after one or more chemotherapy regimens Furthermore, more than 40% of these patients had symptomatic responses. 35 These data created excitement regarding the potential role of such therapies for the lung cancer patients. Nevertheless, a minority of lung cancer patients had objective responses, and there is little information on how to select patients who would benefit from therapy. Previous studies established the critical role of receptor expression levels for predicting response to antiestrogens or to anti-her2/neu antibodies such as trastuzumab in breast cancer. 36 EGFR receptor expression was expected to be the best predictor of response and prognosis because it reflects genetic, epigenetic, and transcriptional effects. Nevertheless, in breast cancer the HER2/neu gene status defined by FISH provided a more accurate prediction of trastuzumab effects than the assessment of HER2/neu receptor by HercepTest (DAKO Corp, Carpenteria, CA). 37 Multiple assessment of the EGFR and the erbb family signaling pathways should be considered in future clinical trials of EGFR inhibitors. If receptor expression is a major determinant of response, our data suggest that EGFR inhibitors might be more effective in SCC, BAC, and in welldifferentiated lung cancers than in ADC and LCC histologies and poorly differentiated tumors. Indeed, patients with BAC tumors have shown good response to treatment with EGFR tyrosine kinase inhibitors in preliminary clinical trials. 38 Although the difference in EGFR expression between stages I to II and stage III was not statistically significant, we found a higher expression of EGFR in earlier stages compared with that in stage III. This might indicate that less advanced disease could be a better target for EGFR inhibitors. Additional clinical studies are needed to verify this hypothesis. We conclude that most of the NSCLC tumors have high EGFR expression, especially in the SCC and BAC subtypes. There was a correlation between increased EGFR gene copy number per cell and protein expression, but a complex interaction between gene and protein levels seems to occur. Increased understanding about molecular characteristics coupled with more comprehensive data on the EGFR cognate ligands EGF and transforming growth factor alpha, and the heterodimerization partners HER-2, HER-3, and HER-4, may provide a better prognostic indicator of response to therapy with EGFR inhibitors. Histologic and pathologic variables might play a role in the identification of the group of patients most likely to benefit from EGFR inhibitor therapy, and future clinical studies need to include pathologic and biologic correlations to identify the optimal target population. ACKNOWLEDGMENT We thank Edward Gabrielson, MD, and Angelo DeMarzo, MD, for helping to establish the lung tissue microarray; Norma Aumen and Anne Drabkin for technical assistance with the IHC and FISH assays; and Vysis Inc (Downers Grove, IL) for providing the dual-color LSI EGFR SpectrumOrange/CEP 7 SpectrumGreen FISH probe set. AUTHORS DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST The following authors or their immediate family members have indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. Owns stock (not including shares held through a public mutual fund): Paul A. Bunn Jr, Carcinex; acted as a consultant within the last 2 years: Paul A. Bunn Jr, Allos Therapeutics, American Association of Cancer Institutes, American Society of Clinical Oncology, Amgen, Astra-Zeneca, Bristol-Myers Squibb, Eli Lilly & Company, FeRx, Fox Chase Cancer Center, MD Anderson Cancer Center, National Cancer Institute, Novartis, Pharmacia/Upjohn, SG Madison/Center Biomedical Continuing Education, US Government (Food and Drug Administration, Veterans Administration, National Institutes of Health), University of Arizona Cancer Center, Vanderbilt University Cancer Center, Cancer Research and Biostatistics, Millennium Pharmaceutical Inc; Fred R. Hirsch, Astra-Zeneca, Johnson & Johnson. Received more than $2,000 a year from a company for either of the last 2 years: Paul A. Bunn Jr, Memorial Sloan-Kettering Cancer Center, GlaxoSmith Kline Beecham, Puget Sound Oncology Consortium, Physicians Education Resource. 1. Maxwell PD: Global cancer statistics in the year Lancet Oncol 2: , Jemal A, Thomas A, Murray T, et al: Cancer statistics CA Cancer J Clin 52:23-47, Yarden Y, Sliwkowski M: Untangling the ErbB signaling network. Nat Rev Mol Cell Biol 2: , Stoscheck CM, King LE: The role of epidermal growth factor in carcinogenesis. Cancer Res 46: , 1986 REFERENCES 5. Sobol RE, Astarita RW, Hofeditz C, et al: Epidermal growth factor receptor expression in human lung carcinomas defined by a monoclonal antibody. J Natl Cancer Inst 79: , Berger MS, Gullick WJ, Greenfield C, et al: Epidermal growth factor receptors in lung tumours. J Pathol 152: , Ciardiello F, Tortora G: A novel approach in the treatment of cancer: Targeting epidermal growth factor receptor. Clin Cancer Res 7: , 2001

10 EGFR IN NON SMALL-CELL LUNG CANCER 8. Rusch V, Baselga J, Cordon-Cardo C, et al: Differential expression of the epidermal growth factor receptor and its ligands in primary non-small cell lung cancer and adjacent benign lung. Cancer Res 53: , Kurie JM, Sjon HJC, Lee JS, et al: Increased epidermal growth factor receptor expression in metaplastic bronchial epithelium. Clin Cancer Res 2: , Piyathilake CJ, Frost AR, Manne U, et al: Differential expression of growth factors in squamous cell carcinoma and precancerous lesions of the lung. Clin Cancer Res 8: , Lonardo F, Dragnev KH, Freemantle SJ, et al: Evidence for the epidermal growth factor receptor as a target for lung cancer prevention. Clin Cancer Res 8:54-60, Franklin WA, Veve R, Hirsch FR, et al: Epidermal growth factor receptor family in lung cancer and premalignancy. Semin Oncol 29:3-14, 2002 (suppl 4) 13. Hirsch FR, Varella-Garcia M, Franklin WA, et al: Evaluation of HER2/ neu in non-small cell lung cancer by immunohistochemistry and fluorescence in-situ hybridization (FISH) techniques. Br J Cancer 86: , Volm M, Rittgen W, Drings P: Prognostic value of ERBB-1, VEGF, cyclin A, FOS, JUN and MYC in patients with squamous cell lung carcinomas. Br J Cancer 77: , Phsaki Y, Tanno S, Fujita Y, et al: Epidermal growth factor receptor expression correlates with poor prognosis in non-small cell lung cancer patients with p53 overexpression. Oncol Rep 7: , Cox G, Jones JL, O Byrne KJ: Matrix metalloproteinase 9 and the epidermal growth factor signal pathway in operable non-small cell lung cancer. Clin Cancer Res 6: , Rusch V, Klimstra D, Venkatraman E, et al: Overexpression of the epidermal growth factor receptor and its ligand transforming growth factor alpha is frequent in respectable non-small cell lung cancer but does not predict tumor progression. Clin Cancer Res 3: , Pfeiffer P, Clausen PP, Andersen K, et al: Lack of prognostic significance of epidermal growth factor receptor and the oncoprotein p185her-2 in patients with systemically untreated non-small-cell lung cancer: An immunohistochemical study on cryosections. Br J Cancer 74:86-91, Fontanini G, De Laurentis M, Vignati S, et al: Evaluation of epidermal growth factor-related growth factors and receptors and of neoangiogenesis in completely resected stage I-IIIA non-small-cell lung cancer: Amphiregulin and microvessel count are independent prognostic indicators of survival. Clin Cancer Res 4: , Greatens TM, Niehans GA, Rubins JB, et al: Do molecular markers predict survival in non-small-cell lung cancer? Am J Respir Crit Care Med 157: , D Amico TA, Massey M, Herndon JE II, et al: A biologic risk model for stage I lung cancer: Immunohistochemical analysis of 408 patients with the use of ten molecular markers. J Thorac Cardiovasc Surg 117: , Pastorino U, Andreola S, Tagliabue E, et al: Immunocytochemical markers in stage I lung cancers: Relevance to prognosis. J Clin Oncol 15: , Brabender J, Danenberg KD, Metzger R, et al: Epidermal growth factor receptor and HER2-neu mrna expression in non-small cell lung cancer is correlated with survival. Clin Cancer Res 7: , Wang Y, Minoshima S, Shimizu N: Precise mapping of the EGF receptor gene on the human chromosome 7p12 using an improved FISH technique. Jpn J Hum Genet 38: , Reissmann PT, Koga H, Figlin RA, et al: Amplification and overexpression of the cyclin D1 and epidermal growth factor receptor genes in non-small-cell lung cancer. J Cancer Res Clin Oncol 125:61-70, Reinmuth N, Brandt B, Kunze WP, et al: Ploidy, expression of erbb1, erbb2, P53 and amplification of erbb1, erbb2 and erbb3 in non-small cell lung cancer. Eur Respir J 16: , Travis WD, Colby TV, Corrin B, et al: Histologic Typing of Tumours of Lung and Pleura: World Health Organization International Classification of Tumors (ed 3). New York, NY, Springer Verlag, Ekstrand AJ, Sugawa N, James CD, et al: Amplified and rearranged epidermal growth factor receptor genes in human glioblastomas reveal deletions of sequences encoding portions of the N- and/or C-terminal tails. Proc Natl Acad Sci U S A 89: , Moscatello DK, Holgado-Madruga M, Godwin AK, et al: Frequent expression of a mutant epidermal growth factor receptor in multiple human tumors. Cancer Res 55: , Wikstrand CJ, Hale LP, Batra SK, et al: Monoclonal antibodies against EGFRvIII are tumor specific and react with breast and lung carcinomas and malignant gliomas. Cancer Res 55: , Garcia de Palazzo IE, Adams GP, Sundareshan P, et al: Expression of mutated epidermal growth factor receptor by non-small cell lung carcinomas. Cancer Res 53: , Fukuoka M, Yano S, Giaccone G, et al: Multi-institutional randomized phase II trial of gefitinib for previously treated patients with advanced non small-cell lung cancer. J Clin Oncol 21: , Kris MG, Natale RB, Herbst RS, et al: A phase II trial of ZD1839 (Iressa) in advanced non-small cell lung cancer (NSCLC) patients who had failed platinum- and docetaxel-based regimens IDEAL2. Proc Am Soc Clin Oncol 21:292a, 2002 (abstr 1166) 34. Perez-Soler R, Chachoua A, Huberman M, et al: A phase II trial of the epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor OSI-774, following platinum-based chemotherapy in patients with advanced EGFR expressing non-small cell lung cancer. Proc Am Soc Clin Oncol 20:310a, 2001 (abstr 1235) 35. Herbst R, Natale R, Giaccone G, et al: ZD1839 improves diseaserelated symptoms and quality of life of patients with locally advanced or metastatic non-small cell lung cancer in two randomized, double-blind studies (IDEAL 1 and IDEAL 2). Chest 122:166S, 2002 (abstr) 36. Slamon DJ, Clark GM, Wong SG, et al: Human breast cancer: Correlation of relapse and survival with amplification of the HER2/neu oncogene. Science 235: , Pauletti G, Dandekar S, Rong HM, et al: Assessment of methods for tissue-based detection of the HER2/neu alteration in human breast cancer: A direct comparison of fluorescence in situ hybridization and immunohistochemistry. J Clin Oncol 18: , Ruckdeschel JC, Simon G, Antonia S, et al: ZD1839 (Iressa) as a single agent for the treatment of metastatic non-small-cell lung cancer. Proc Am Soc Clin Oncol 21:318a, 2002 (abstr 1269)