Protein expression and gene amplification of epidermal growth factor receptor in thymomas

Abstract

BACKGROUND

Epidermal growth factor receptor (EGFR) overexpression and amplification are important prognostic factors in many solid tumors and anti-EGFR antibody-based therapy is now available as a promising therapeutic modality. There is little information in the literature regarding the biologic role of EGFR in thymomas that are characterized by variable clinical presentations, histologic heterogeneity, and unpredictable behavior.

METHODS

Protein expression and gene amplification of EGFR were investigated in 32 thymomas (9 World Health Organization [WHO] type A, 5 type AB, 7 type B2, 7 type B3, 4 type C) using immunohistochemistry and fluorescence in situ hybridization (FISH). FISH analysis included assessment of the average number of copies of the EGFR gene per cell, the average ratio of the EGFR gene to chromosome 7 copy numbers, and ploidy.

RESULTS

The results of FISH analysis showed statistically significant correlation with WHO histologic type, invasion, advanced clinical stage, but not with tumor size and outcome. Thymomas associated with myasthenia gravis more frequently showed hyperploidy when compared with sporadic tumors, but there was no difference in EGFR gene amplification. EGFR protein expression assessed by immunohistochemistry did not correlate with any studied clinicopathologic variables. There was poor correlation between the protein expression and gene amplification, only 7 of 23 specimens (30%).

CONCLUSIONS

The potential role of EGFR in the pathogenesis of advanced-stage thymomas indicated that evolving anti-EGFR antibody therapy may be considered as a treatment option. Cancer 2005. © 2005 American Cancer Society.

Thymomas are tumors of thymic epithelial cells characterized by marked morphologic heterogeneity. This has resulted in a variety of controversial pathologic classifications.1–5 Traditionally, these were classified on the basis of the lymphocyte-to-epithelial cell ratio and the shape of the epithelial cells. However, this approach did not always correlate with clinical behavior and outcome.6–11 Subsequently, the stage and presence of invasion were considered to be the most important prognostic factors and the significance of morphology was neglected frequently. Several molecular and genetic studies showed that the most recent World Health Organization (WHO) classification reflects histologic, clinical, genetic, and functional heterogeneity of thymomas that can be translated into clinical practice.12–16

When the Masaoka staging system is applied, most type A and AB thymomas are encapsulated (Stage I), and, therefore, are considered to be clinically benign tumors. Usually, these can be cured by surgery alone. In contrast, some type B thymomas, particularly the B3 type, tend to be clinically malignant. The therapeutic strategy for these tumors is controversial.17, 18 Surgical resection is the preferred therapeutic option for all types of thymomas. However, radical surgery may be inadequate in clinically aggressive thymomas and adjuvant therapeutic options need to be considered.19 It is unclear whether local control of invasive localized disease can be improved by radiotherapy. Although many small studies suggest a positive effect of radiotherapy on survival, larger retrospective studies have shown that adjuvant radiotherapy has no impact on survival.20, 21 The role of chemotherapy is even more controversial. The influence of associated autoimmune disorders, such as myasthenia gravis (MG), on survival in patients with thymoma is uncertain.22–26

The expanding knowledge of tumor biology has resulted in a search for novel therapies for the various tumor types. Over the last 20 years, elevated levels of epidermal growth factor receptor (EGFR) have been identified as a common event in many tumors. In many cases, aberrant EGFR activation, mediated primarily through gene amplification and autocrine stimulation, appears to be an important factor in tumorigenesis and one of the predictors of aggressive biologic behavior. Increased EGFR expression in head and neck, ovarian, cervical, bladder, and esophageal carcinomas is a strong prognostic factor and inhibition of its action may produce therapeutic benefits.27 Previous studies established the criteria and methods important in predicting therapeutic response to antiestrogens or to anti-HER-2/neu. However, there are no such criteria for EGFR. It is common for immunohistochemistry to be used to evaluate EGFR protein levels, but without a uniform scoring system, the method remains very subjective and prone to interobserver variability. Furthermore, the use of different antibodies and immunohistochemistry protocols leads to variable results. In many malignancies, except in nonsmall cell lung carcinomas (NSCLC), it is still uncertain whether EGFR protein expression or status of the EGFR gene is the better predictor of therapeutic response, and multiple assessments of the EGFR signaling pathways may be considered in future studies.

Several studies have detected increased EGFR protein expression in thymomas using immunohistochemical techniques.28–31 Sasaki et al.32 found elevated levels of EGFR in serum specimens from patients with advanced-stage thymomas. However, there have not been any studies that have explored EGFR expression at the molecular level, either by gene amplification or by other methodologies.

The aims of our study were to assess the frequency of EGFR protein expression by immunohistochemistry and gene amplification by fluorescence in situ hybridization (FISH) and to determine whether EGFR expression assessed by these two methods correlates with WHO histologic type, tumor invasion, stage, tumor size, occurrence of associated autoimmune disorder, and outcome.

MATERIALS AND METHODS

Thirty-two formalin-fixed thymoma specimens were obtained from the paraffin block archives of the University of Pittsburgh Medical Center (Pittsburgh, PA) after we obtained internal review board consent for the current anonymous study.

Hematoxylin and eosin (H&E)-stained sections were reviewed independently by two pathologists (S.D. and M.N) and classified according to the WHO thymic epithelial tumor classification.12 The thymoma specimens were divided into two major types depending on whether the neoplastic epithelial cells had a spindle or oval shape (type A) or a dendritic or epithelioid appearance (type B). Tumor specimens combining these features were designated type AB. Type B thymoma specimens were subdivided further on the basis of increasing epithelial to lymphocyte ratio and cytologic atypia of epithelial cells (Fig. 1). Type C thymoma specimens resembled carcinomas arising outside the thymus.12

Figure 1. Type B2 thymoma specimen characterized by a relatively even proportion of lymphocytes and epithelial cells with subtle nuclear abnormalities (magnification × 40).

The study included 9 WHO type A, 5 type AB, 7 type B2, 7 type B3, and 4 type C thymoma specimens. Twenty tumor specimens were retrieved from patients with invasive disease according to Masaoka staging criteria (Stages II–IV) and 12 were retrieved from patients with noninvasive disease (Stage 1).33 Because of insufficient clinical data, Masaoka stage could not be assessed with certainty for four specimens with microscopically invasive thymoma. For one patient, it was unknown whether the patient had MG. All patients were followed up to determine outcome.

FISH analysis of EGFR amplification was performed using the dual-color EGFR SpectrumOrange/CEP7 SpectrumGreeen probe (Vysis, Downers Grove, IL) and paraffin pretreatment reagent kit (Vysis). In brief, paraffin sections were deparaffinized, dehydrated in ethanol, and air-dried. Sections were digested with protease K (0.5 mg/mL) at 37 °C for 28 minutes. Slides were denatured at 75 °C for 5 minutes and dehydrated in ethanol. The probes were denatured for 5 minutes at 75 °C before hybridization. Slides were hybridized overnight at 37 °C and washed in 2 × sodium citrate/sodium chloride solution (SSC) Tyepal (Sigma, Saint Louis, MO) at 72 °C for 2 minutes. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI)/antifare I (Vysis). Each FISH assay included normal thymic tissue sections as a negative control, and sections of thymomas previously identified as carrying EGFR gene amplification as a positive control. Analyses were performed using a fluorescence microscope Nikon Optiphot-2 (Nikon, Japan), and Quips Genetic Workstation (Vysis, Inc., Downers Grove, IL), Chroma Technology (Rockingham, VT) equipped with Chroma Technology 83000 filter set with single band exitors for Texas Red/Rhodamine, FITC, and DAPI (ultraviolet 360 nm). The histologic areas previously selected on the H&E-stained sections were identified on the FISH-treated slides. Only individual and well delineated cells were scored. Overlapping cells were excluded from the analysis. At least 60 cells were scored for each specimen and control.

Each tumor specimen was assessed by the average number of copies of the EGFR gene per cell, the average ratio of the EGFR gene to chromosome 7 copy numbers, and ploidy. Amplification was defined as a ratio of EGFR signals to chromosome 7 centromere signals of ≥ 1.30. Polyploidy was defined as a mean of > 2 chromosome 7 signals per cell.

The immunohistochemical study was performed and graded using anti-EGFR mouse monoclonal antibody at a dilution of 1:200 (DakoCytomation, Carpenter, CA) according to the manufacturer's instructions. Only clear staining of the tumor cell membrane was considered positive. Diffuse cytoplasmic or granular staining was interpreted as negative. Membrane expression was graded from 0 to 3 (in the style of DakoScore for HER-2/neu). Specimens with a score > 1 were regarded as showing overexpression.34 Sections of adenocarcinoma of lung previously identified as positive for EGFR protein served as a positive control. Negative controls included omitting the primary antibody and its substitution with normal serum.

Statistical analysis was performed by Student t test and the Fisher exact probability test. Significance was defined as P < 0.05.

RESULTS

Of 32 patients with thymic epithelial tumors, 17 were women and 15 were men, with a mean age of 58 years (range, 29–78 years). The mean size of the tumor specimens was 7.8 cm (range, 5–12.5 cm). Complete follow-up lasted until April 2003 or until death. The mean follow-up time was 78.6 months (6.5 years; range, 20–159 months). Only one patient died of disease. The cause of death was unknown in five patients. There were 12 specimens associated with MG.

EGFR protein overexpression was observed in 22 specimens (69%; Fig. 2). It was not associated with histologic type (P = 0.74), stage or invasion (P = 1.00), or the presence of autoimmune disease (P = 0.46).

Figure 2. Membranous expression of epidermal growth factor receptor immunostain in a B3 type thymoma specimen (magnification × 40).

Table 1 summarizes the results of FISH analysis. Type B3 thymoma specimens showed a statistically significant higher ratio of the EGFR gene to chromosome 7 copy number than type A, AB, and B2 thymoma specimens (mean ratio, 1.42 vs. 1.08, 1.05, and 1.08, respectively; P = 0.01; Fig. 3). The average number of EGFR gene signals per cell was significantly higher in type B3 thymoma specimens (mean, 3.10) than in type A (mean, 2.11; P = 0.023) and type B2 specimens (mean, 1.08; P = 0.018; Fig. 3). Although type AB thymoma specimens showed a lower average number of EGFR gene copies per cell than B3 thymoma specimens, this difference did not reach statistical significance. The significance of FISH results in type C thymoma specimens is uncertain, because of the limited number of specimens included in our study. Type C thymoma specimens showed a higher average number of EGFR gene copies per cell than other histologic types, but that difference was not significant. EGFR gene amplification did not show any difference between histologic types.

Table 1. Summary of EGFR FISH Analysis of Thymomas

WHO Type	FISH Analysis
	Ratio (mean ± SD)a	Average (mean ± SD)b	Hyperploidy (%) (mean ± SD)c	No. of amplification (n)
EGFR: epidermal growth factor receptor; FISH: fluorescence in situ hybridization; MG: myasthenia gravis; SD: standard deviation. a ratio of EGFR to chromosome 7. b ▿ average no. of EGFR gene signals per cell. c Percent of cells with hyperploidy (60 cells count).
A (n = 9)	1.08 ± 0.11	2.11 ± 0.73	10.78 ± 18.53	1
AB (n = 5)	1.05 ± 0.02	2.32 ± 0.52	24.6 ± 28.64	0
B2 (n = 7)	1.08 ± 0.11	2.12 ± 0.17	11.34 ± 9.59	1
B3 (n = 7)	1.42 ± 0.30	3.10 ± 0.83	26.78 ± 22.25	4
C (n = 4)	1.23 ± 0.43	4.72 ± 5.92	2.90 ± 2.09	1
Stage
1 (n = 10)	1.04 ± 0.04	1.92 ± 0.17	6.07 ± 4.91	0
2 (n = 13)	1.13 ± 0.19	2.36 ± 0.73	17.64 ± 22.92	3
3 and 4 (n = 5)	1.63 ± 0.22	4.52 ± 2.02	28.42 ± 26.66	4
Invasion
Present (n = 20)	1.24 ± 0.29	3.15 ± 2.60	19.86 ± 22.95	7
Absent (n = 12)	1.04 ± 0.04	199 ± 0.23	8.43 ± 8.72	0
Myasthenia Gravis
Present (n = 12)	1.18 ± 0.23	2.63 ± 0.88	25.23 ± 22.12	3
Absent (n = 19)	1.17 ± 0.27	2.76 ± 2.68	9.86 ± 16.07	4

Figure 3. Epidermal growth factor receptor (EGFR) gene amplification in a type B2 thymoma specimen using fluorescence in situ hybridization with an EGFR (red) and chromosome 7 (green)-specific probe (magnification × 630).

FISH analysis revealed that invasive thymomas were associated with a higher ratio of the EGFR gene to chromosome 7 copy (mean, 1.24 invasive vs. 1.04 noninvasive; P = 0.0073), and with frequent hyperploidy (mean, 19.86 invasive vs. 8.43 noninvasive; P = 0.05). Although invasive thymoma specimens showed a higher average number of EGFR signals per cell (mean, 3.12 invasive vs. 1.99 noninvasive), this difference was not statistically significant (P = 0.07). EGFR amplification was identified in 7 of 20 specimens (35%) of invasive thymoma and in none of the encapsulated tumors (P = 0.028). Specimens from patients with advanced-stage thymomas (Stage II–IV) showed a statistically significant higher mean ratio of the EGFR gene to chromosome 7 copy than specimens from patients with Stage I thymomas (1.275 vs. 1.047 respectively; P = 0.005), specimens with a higher mean average number of EGFR gene signals per cell (3.25 vs. 1.93; P = 0.05), and specimens with a higher mean percentage of cells with hyperploidy (20.64 vs. 6.071; P = 0.02). EGFR gene amplification was absent in all specimens from patients with Stage I thymomas.

There was no correlation between tumor size and any FISH measured parameter. Hyperploidy was observed more frequently in MG-associated thymomas than in the absence of autoimmune disease (25.23 vs. 9.87; P = 0.03). However, there was no difference in EGFR gene amplification, gene ratio, and number of gene copies per cell. Statistical analysis failed to show any significant correlation between outcome and results of FISH analysis.

As shown in Table 2, the results of EGFR protein expression determined by immunohistochemistry did not correlate with the results of FISH analysis (P = 0.149). Of 23 specimens with protein overexpression, only 7 (30%) showed EGFR gene amplification by FISH.

Table 2. Comparison of EGFR Protein Expression by Immunohistochemistry and EGFR Gene Amplification by FISH

Characteristics		Gene amplification
		Negative	Positive
EGFR: epidermal growth factor receptor; FISH: fluorescence in situ hybridization.
Protein expression	Negative	9	0
	Positive	16	7

DISCUSSION

EGFR is a very important growth factor in many epithelial neoplasms including tumors of the head and neck, lung, colon, and breast.35–40 The prognostic significance of EGFR expression in various malignancies is controversial, but most reports indicate that EGFR overexpression is associated with poor prognosis.41–43 This observation has resulted in the development of different classes of blocking agents that cause inhibition of EGFR and subsequent suppression of tumor growth. More recently, clinical trials with EGFR inhibitors have shown promising results in colon carcinoma and NSCLC.44, 45

The role of EGFR in thymomas is relatively unexplored. We were able to demonstrate protein overexpression by immunohistochemistry in 69% of the thymoma specimens included in our study. Reported EGFR overexpression in thymomas ranges from 43% to 100%.28–30 The reason for such variation is uncertain, but may reflect differences in detection methods, reagents, scoring systems, and interpretation. A potential association between EGFR protein overexpression and histologic type and stage is controversial. Hayashi et al.28 used the Levine and Rosai and Müller–Hermelink classification of thymomas in their study and were able to demonstrate an association between histologic type and advanced-stage thymoma with EGFR protein overexpression. In our study, protein overexpression did not correlate with histologic type of the tumor or stage and this observation is similar to results reported by Gilhus et al.29 and Pescarmona et al.30 who used the Müller–Hermelink classification.

The underlying mechanisms of EGFR protein overexpression are not understood completely. Gene amplification was proposed as a possible mechanism for many years, but recent studies indicate that some other mechanisms may play a role, such as posttranslational changes or the length of polymporphisms in intron 1 of the EGFR gene.34, 46 In contrast to EGFR protein overexpression, the status of EGFR gene amplification in thymomas is unknown. To our knowledge, this is the first study that investigated the relationship of EGFR gene expression as assessed by FISH, histology, and clinicopathologic characteristics of thymic epithelial tumors. There are several ways to quantitate gene expression by FISH including average or maximum gene copy numbers, the ratio of the gene copy number to the chromosome number, or the percentage of cells with unbalanced gene copy numbers. Our study showed a significant correlation between WHO histologic type of thymoma and the number of EGFR gene copies and ratio of the EGFR gene to chromosome 7, with the highest average gene copy number occurring in type B3 and C thymomas. However, our results failed to show a correlation between the histologic appearance of thymoma and amplification of the EGFR gene when the WHO classification system was used. Our results suggest that this new classification system may be helpful in clinical practice for the assessment and treatment of patients with thymomas. Chromosomal imbalance studies using comparative genomic hybridization, FISH, and loss of heterozygosity (LOH) analysis also support the idea that the current histologic classification of the thymic epithelial tumors mirrors the differences in the biology of the histologic subtypes of thymomas.47, 48

We also sought to determine whether EGFR gene amplification correlates with stage and invasion by thymomas. Invasive and advanced-stage thymomas more frequently showed amplification of the EGFR gene and hyperploidy when compared with lower-stage tumors. Similar to previously published studies in breast carcinoma, we showed that FISH analysis of gene amplification correlates better with tumor biologic behavior than with immunohistochemical analysis of protein expression.49–51

Our study showed a striking discrepancy between EGFR protein expression and EGFR whole gene amplification. Although most specimens with EGFR gene amplification showed protein overexpression, 69% of specimens with protein overexpression did not show gene amplification. Similar findings were encountered in invasive breast carcinoma.34 It is possible that recently described modulation of transcription by secondary structures and polymorphic short sequence repeats residing within the EGFR gene that cannot be detected by FISH represent an underlying mechanism for protein overexpression in thymomas. Further studies using more precise polymerase chain reaction assays targeting the region of interest may provide a clue for the underlying mechanism.

The significance of more frequent hyperploidy in thymomas associated with MG observed in our study is uncertain, particularly because we were unable to demonstrate a relationship between EGFR protein expression or gene amplification and survival in thymomas. Larger series of specimens with known clinical outcome and longer follow-up are needed to better address the possible impact of EGFR on the survival of patients with thymomas.

In summary, we have demonstrated that amplification of the EGFR gene in thymomas is associated with invasive and advanced-stage disease. A significant discrepancy between protein expression and gene amplification strongly suggests that other molecular mechanisms play a role in EGFR overexpression in thymomas. Our results also suggest that more advanced thymomas could be a good target for EGFR inhibitors as adjuvant therapy. A lack of prognostic significance of EGFR gene amplification and protein expression does not preclude potential benefit from EGFR inhibitor therapy. Further large studies are needed to assess clinicopathologic variables and to determine the underlying molecular mechanisms of EGFR expression in thymic tumors that might play a role in the identification of patients most likely to benefit from intensive adjuvant therapy.