Clonality and Prognostic Implications of p53 and Epidermal Growth Factor Receptor Somatic Aberrations in Multiple Primary Lung Cancers

The incidence of synchronous multiple primary lung cancers in reported clinical series ranges from 1% to 7% (1), and up to 10% of patients who survive the first primary lung carcinoma will develop a second primary lung tumor later in their life (2). On the other hand, autopsy studies have revealed more accurately that the incidence of multiple synchronous tumors in the lung ranges from 3.5% to 14% (3). These preneoplastic and neoplastic lesions are usually extensive and multifocal, occurring throughout the entire respiratory tree, a phenomenon due to similar exposure of carcinogens referred to as the concept of “field cancerization” (4). In addition, lung is also the most common site of hematogenous lung cancer metastasis. Therefore, the synchronous multiple primary lung cancers could be either multifocal with different clonality or intrapulmonary metastases that migrated from the same origin. Multiple primary lung cancers are potentially curable; however, the prognosis is usually poor if intrapulmonary metastasis occurred (5). Clinically, distinction of clonality in patients with synchronous multiple primary lung cancer is of therapeutic and prognostic importance.

The criteria proposed by Martini and Melamed (6) in 1975 are still commonly used for the diagnosis of synchronous multiple primary lung cancer for patient management. The diagnosis is primarily based on the histologic characteristics of tumors, such as morphology, location, presence or absence of carcinoma in situ, vascular invasion, metastasis, and other empirical features without biological and molecular basis. Therefore, molecular analysis of clonal relation between different lesions may solve the diagnostic riddle in these patients. In particular, PCR amplification of genetic markers on small amount of tumor DNA isolated from microdissection-purified cancer cells from the same patient allows detailed analysis and determination of tumor clonality.

Recent advances in molecular tumorigenesis indicated that accumulated genetic alterations of cancer genome during multistep tumor progression are potentially useful marker for clonality analysis. Several technologies detecting genetic alterations, including X-chromosome inactivation analysis, loss of heterozygosity analysis by using microsatellite markers, viral integration analysis, and detection of mutation in tumor-associated genes (7), have been widely used with limited success. To develop optimal genetic markers for clonality analysis, highly frequent and independent somatic aberrations that occurred in early stages and are maintained throughout the tumor progression are important for clonality assessment. For instance, the mutation analysis of the p53 tumor-suppressor gene is useful for clonality assessment in lung cancer due to highly frequent, stable, and well-distributed point mutations occurring in exons 5 to 8 (8, 9). Because coding mutations of p53 occur relatively early in the development of lung cancer and are potentially required for maintaining malignant phenotype, the acquired p53 mutations are preserved during tumor progression and metastatic spread (10–12). It is reported that somatic mutations and increased expression of p53 were frequently found in ∼45% and ∼65% of non–small cell lung carcinomas (NSCLC), respectively (13).

Because protein tyrosine kinases play important roles in the pathogenesis of many malignant tumors, development of selective tyrosine kinase inhibitors is currently one of the major efforts in cancer treatment (14, 15). The epidermal growth factor receptor (EGFR), the first receptor protein tyrosine kinase described, is detected with increase of expression by immunohistochemistry in 43% to 89% of NSCLC patients (16, 17). Two molecularly targeted agents, gefitinib (AstraZeneca, Wilmington, DE) and erlotinib (OSI Pharmaceuticals, Inc., Melville, NY), have been approved and have shown greater benefit for the treatment of advanced NSCLC. The better outcomes of tyrosine kinase inhibitor treatment in NSCLC patients are strongly associated with mutations of the tyrosine kinase domain (exons 18-22) in EGFR of tumor tissues (18, 19). Depending on the tumor subtypes and races, the Caucasian NSCLC patients have an EGFR mutation frequency of ∼10% compared with a mutation rate of at least 30% in Asian patients (20, 21).

Toward developing useful genetic markers for clonality assessments and patient management, we took advantage of highly frequent and early somatic mutations of p53 (5, 22–24) and EGFR (25) genes to define the clonal origins of tumors in 58 multiple primary lung cancers. The results of somatic mutations along with increase of protein expression of p53 and EGFR detected by immunohistochemistry were further characterized for correlation with the clinicopathologic features for potential diagnostic and prognostic applications.

Patient populations. Fifty-five synchronous and three metachronous multiple primary lung cancer specimens were obtained from 1,037 patients who underwent surgical resection at the National Taiwan University Hospital from August 1995 to December 2004. The research ethics committee of the hospital has approved this study. These patients were not treated with neoadjuvant chemotherapy and irradiation therapy.

Extraction of DNA from microdissected cells in paraffin-embedded tissues. Five-micrometer-thick paraffin sections were used for genomic DNA isolation. After deparaffinization with xylene, tissue sections were stained with hematoxylin, and areas were carefully microdissected with Laser Microdissection System (Leica LMD, Wetzlar, Germany) for obtaining >70% of neoplastic and adjacent normal-appearing cells. An estimated ∼1,000 microdissected cells were digested in 50 μL of buffer consisting of 20 mmol/L Tris-HCl (pH 8.0), 1 mmol/L EDTA (pH 8.0), 1% Tween 20, and 1 mg/mL proteinase K for 24 h at 56°C. The protease-treated DNA mixture was heat inactivated after incubation for 10 min at 95°C as described by Sugio et al. (26) and 1 μL of DNA mixture was used for each PCR reaction.

PCR sequencing and mutation detection of p53 and EGFR genes for clonality analysis. Exons 5 to 8 of p53 and exons 18 to 22 of EGFR with their short flanking intronic sequences were PCR amplified using specific primers in a 96-well format followed by nested PCR reactions as described by previous reports (27, 28). Purified PCR products were subjected to cycle sequencing. Multiple sequence alignments of exon sequences were conducted by using Sequencer 4.14 (Gene Codes, Ann Arbor, MI) for mutation identification. The diagnosis of different clonality was made when different mutations were found in two or more tumor clones from one patient. On the other hand, the same clonality was diagnosed when the tumors showed identical DNA mutation and lack of the same mutation in the adjacent normal lung tissue samples. When no mutation was detected in either p53 or EGFR gene in tumor samples from a patient, a diagnosis could not be made.

Immunohistochemical analysis of p53 and EGFR expression. For immunohistochemical analysis of p53 protein in the tumor tissue, deparaffinized 4-μm-thick sections were treated with 0.3% H₂O₂ in methanol, heated in a microwave oven for 20 min for antigen retrieval, and incubated with normal nonimmune goat serum. After blotting the excessive goat serum, the slides were incubated with a specific mouse anti-p53 protein antibody “p53 (Ab-6), pantropic” (diluted 1:50; Oncogene Science, Cambridge, MA) for 1 h at room temperature. After incubation with biotinylated goat anti-mouse antibody, the sections were incubated with peroxidase-conjugated streptavidin. 3,3′-Diaminobenzidine tetrahydrochloride (0.05%) was used as a chromogen.

The sections for immunohistochemical analysis of the EGFR protein expression were autoclaved in 0.01 mol/L phosphate citrate buffer (pH 6.0) at 121°C for 10 min and then treated with 3% H₂O₂-methanol, incubated with normal goat serum, and subsequently subjected to the primary antibody reaction. The antibody for EGFR protein (diluted 1:30; BioGenix, San Ramon, CA) was left to react with the sections overnight at room temperature. Detection of the immunoreactive staining was carried out by the avidin-biotin-peroxidase complex method according to the manufacturer's instructions (DAKO Corporation, Carpinteria, CA).

Immunostaining was classified in the following two groups according to both intensity and extent: (a) negative, when no staining or positive staining was detected in ≤50% of the cells; (b) positive, when immunostaining was present in >50% of the cells. Two independent pathologists (Y.-L.C. and C.-T.W.) were involved in the assessment of the expression.

Statistical analysis. The correlation between various clinicopathologic variables and the somatic aberrations of p53 and EGFR were analyzed by Fisher's exact test. Survival curves were estimated by using the Kaplan-Meier method. The log-rank test was used to compare survival curves. The Cox proportional hazards model was used to carry out the survival analysis without any adjustment or adjusted for age of 50 years, sex, and smoking status. Odds ratios (OR) and 95% confidence intervals (95% CI) were also calculated. The multiple logistic regression was used to assess the effect of the factor of somatic aberrations on lymph node metastasis without any adjustment or with adjustment for age of 50 years, sex, and smoking status. ORs of tumors with the same clonality versus tumors with different clonality and 95% CIs were also calculated. All tests were two-tailed, and P < 0.050 was considered significant. All statistical analyses were done using SAS statistical software (version 8.2, SAS Institute, Inc., Cary, NC).

Patient characteristics. Of the 58 multiple primary lung cancer patients (Table 1), 33 cases with multiple tumors were located in the same lobe; 17 cases were unilaterally located in different lobes; and 8 cases were bilaterally located in different lobes. There were 36 cases with three or more tumors and 22 cases with less than three tumors in the same patient. The size of the largest tumor >3 cm in a patient was observed in 34 cases and ≤3 cm was detected in 24 cases. The histologic types of multiple tumors in each patient were the same in all cases, including 87.9% of adenocarcinomas (51 of 58), 5.2% of bronchioloalveolar carcinomas (3 of 58), 3.4% of squamous cell carcinomas (2 of 58), and 3.4% of adenosquamous carcinomas (2 of 58). Immunohistochemical staining indicated that 34.5% (20 of 58) and 29.3% (17 of 58) of cases showed up-regulation of p53 and EGFR proteins, respectively.

Mutations of p53 for correlations with clinicopathologic features and clonality assessment. The p53 mutations were detected in 33 of 58 cases (56.9%) distributed in all subtypes of multiple primary lung cancers with predominant exon 5 mutations in 23 of 58 (39.7%) cases (Fig. 1). The mutations of p53 were observed more frequently in male patients (68.6%) than in female patients (39.1%; P = 0.033; Table 1). Because somatic mutation occurred in early stages and is stable throughout the progression of tumorigenesis, it is an important characteristic for a genetic marker in clonality analysis; we found that p53 mutations were detected in 4 of 58 cases (6.9%) of the histologic normal lung parenchyma.

For clonality assessments, we classified p53 mutation status into four different distributing patterns for categorization of multiple tumors: group A, only one tumor with a mutation; group B, two or more tumors with identical mutations; group C, two or more tumors with different mutations; and group D no mutations were detected in tumors. Fifteen (25.9%) patients were categorized into group B for a conclusion of the same clonality of multiple tumors that potentially originated from intrapulmonary metastasis. A total of 18 (31.0%) patients were categorized into either group A or C and were concluded as having different clonal origins in developing multifocal tumors of lung (Table 2). For 25 (43.1%) patients without mutation in exons 5 to 8 of p53 (group D), clonality of tumors was not evaluated. Our results indicated that the tumors of the same clonal origin with p53 mutations were associated with high incidence of lymph node metastasis (P = 0.005; Table 2). The OR (Table 3) of tumors having the same clonality versus tumors having different clonality, with p53 mutations for lymph node metastasis with adjustment, was 16.46 (95% CI, 2.45-110.78, P = 0.004).

Mutations of EGFR for correlations with clinicopathologic features and clonality assessment. The mutations of the tyrosine kinase domain of EGFR were detected in 44 of 58 cases (75.9%) with a majority of exon 21 mutations in 24 of 57 (42.1%) cases (Fig. 1). Similarly, EGFR mutations were also detected in nearly all subtypes of multiple primary lung cancers except squamous cell carcinomas cases. Unlike p53 mutations, which are preferentially observed in male patients, there was no gender preference for EGFR mutations. The early-stage mutations of EGFR tyrosine kinase domain observed in histologic normal lung cancer cells were 12 of 58 cases (20.7%). EGFR mutation was noted more frequently in tumors with vascular invasion (P = 0.025; Table 1).

For clonality assessment with EGFR mutations, 19 (32.8%) patients were classified into the same clonality (group B); 25 (43.1%) patients belonged to either group A or C with conclusion of different clonality; and 14 patients (24.1%) without EGFR mutation (group D) were not classified according to clonality. Patients with EGFR mutations classified into the same clonality were also statistically associated with lymph node metastasis (P = 0.017; Table 2). The OR (Table 3) of tumors having the same clonality versus tumors having different clonality, with EGFR mutations for lymph node metastasis with adjustment, was 6.89 (95% CI, 1.56-30.45, P = 0.011).

Clonality analysis by using mutations in p53 and/or EGFR. Because 24% to 43% of patients show no detectable somatic mutations in either p53 or EGFR, we therefore combined mutations in both genes for clonality assessment. The tumors were classified into different clonality whenever either p53 or EGFR mutation in one patient belonged to either group A or C. In the rest of the patients, the tumors were categorized into the same clonality whenever either p53 or EGFR mutation in one patient belonged to group B. Indeed, our strategy increased the patient number for clonality analysis to 50 cases (50 of 58, 86.2%) with only eight patients showing no mutations. Among 50 cases accessible for clonality, 22 cases were classified as having the same clonality (37.9%) and 28 cases were categorized into different clonality (48.3%; Table 2). At least three advantages were immediately revealed by our strategy. First, there was no gender preference by combining mutation data from both genes for clonality analysis. Second, there was no conflict in the result of clonality assessment among 58 cases by using mutations that occurred in both genes. Finally, multiple primary lung cancers classified as the same clonality either by p53 or EGFR genes were correlated with lymph node metastasis (P = 0.005 and P = 0.017, respectively; Table 2). The multiple primary lung cancers that were classified as having the same clonality, determined by p53 and/or EGFR mutations, remained statistically associated with lymph node metastasis (P = 0.045). The OR (Table 3) of the tumors having the same clonality versus tumors having different clonality, with p53 and/or EGFR mutations for lymph node metastasis with adjustment, was 4.57 (95% CI, 1.30-16.01, P = 0.018).

Correlations of p53 and EGFR somatic aberrations with clinicopathologic features. Our results indicated that the overexpression of either p53 or EGFR occurred more frequently in the tumors of the same clonal origin (11 of 22, 50.0%) compared with tumors of different clonal origin (6 of 28; 21.4%; P = 0.042; Table 2) determined by p53 and/or EGFR mutations. The overexpression of p53 was marginally associated with poor survival adjusted for age of 50 years, sex, and smoking status (P = 0.066; Table 4). Unlike p53 mutations in exons 5 to 8 showing no correlation with protein expression in tumor tissues, EGFR overexpression was statistically associated with EGFR tyrosine kinase domain mutations (P = 0.046; Table 1). The vascular invasion of multiple primary lung cancers was also correlated with overexpression of EGFR (P = 0.012, data not shown). Although there was no significant correlation between clonality and survival (P = 0.630; Table 4, Fig. 2A), the tumors having the same clonality were associated with lymph node metastasis and the 5-year survival rates in 44.2% and 83.0% of patients with and without lymph node metastasis, respectively (P = 0.001; Fig. 2B; Table 4). The OR (Table 4) for survival of lymph node metastasis over no lymph node metastasis with adjustment was 7.83 (95% CI, 2.29-26.72, P = 0.001).

Multiple primary lung cancers remain a difficult problem for both clinicians and pathologists. For better patient management, it is important to determine tumor clonality for therapeutic options based on both histologic and molecular genetic classifications. However, the majority of current methods used for clonality assessments gained limited success in clinical applications due to lack of biological mechanisms, gender restriction, or unstable somatic alterations throughout the progression of tumorigenesis. In our results, we have taken several advantages of somatic mutations of p53 and EGFR, including early stage, high frequency, and well-distributed mutations for clonality analysis. More importantly, the high clonality assessment rate is potentially applicable in clinical managements due to the use of frequent mutated and limited number of exons on both p53 (exons 5-8) and EGFR (exons 18-22) genes.

For the first time, we revealed that high frequent somatic aberrations of EGFR existed in multiple primary lung cancers are proven to be a good marker for distinguishing multifocal tumors from intrapulmonary metastatic cancers. In agreement with previous studies (29), EGFR protein expression in tissue specimen detected by immunohistochemical staining had good concordance in EGFR mutation, suggesting that EGFR may play an important role in tumorigenesis of NSCLCs and multiple primary lung cancers. The correlation between EGFR mutations and various clinicopathologic factors further revealed that EGFR gene mutations are tightly associated with vascular invasion and lymph node metastasis. Lymph node metastasis occurred more frequently in the tumors with the same clonal origin than in tumors with different clonal origin, suggesting that EGFR mutation might be a prognostic indicator for multiple primary lung cancer with highly invasive, metastatic potentials, and reduced patient survival (15, 30). Because the response rate of the EGFR tyrosine kinase inhibitors is high (69%) in EGFR mutation patients and the EGFR mutation status is strongly associated with the good response to gefitinib (18, 19), our current data suggest that gefitinib and other EGFR tyrosine kinase inhibitors may be particularly effective for treating the inoperable multiple primary lung cancer patients.

With the increase of clonality assessment rate to 86.2% (50 of 58), the revelation of different clonality of multiple primary lung cancers in our results supports the field cancerization concept (4) that multiple respiratory epithelia exposed to carcinogens and that undergo neoplastic transformation independently resulted in tumors with similar morphology but distinct clonality. Indeed, >70% of the second lung cancer was shown to have the same histologic subtype as observed in the first (6, 31). Our results implicated that carcinogenic insults affect many different susceptible cells in the respiratory tract and create various mutations in the same subtype of cancers. Our finding that p53 and/or EGFR mutations had higher incidence of lymph node metastasis in intrapulmonary metastatic cancers than that in carcinomas with different clonality further showed the clinical applications of using genetic markers in clonality assessment of multiple primary lung cancers.

In conclusion, this study is the first to show that the concurrent detection of p53 and EGFR mutations can increase the diagnostic usefulness in multiple primary lung cancers by direct sequencing of functional domains in limited exons. The current examination allowed not only clear diagnosis of multifocal lung cancers in the majority of patients despite similarities in histopathologic features but also genetically supports the independent field cancerization theory. Our results of no statistical association between tumor clonality to patient survival implicated that other factors, in addition to tumor clonality, should have contributed to poor patient outcomes. Nevertheless, the correlations between lymph node metastasis and intrapulmonary metastatic tumors, from the tumors with the same clonality, as well as poor patient survival support the clinical applications of clonality assessment in multiple primary lung cancers. In addition, our studies may provide the possibility for EGFR target therapy by using tyrosine kinase inhibitors in selected inoperable multiple primary lung cancer patients. The recent initiation of cancer genome sequencing projects of prevalent cancers are expected to further reveal additional frequently mutated cancer genes as therapeutic targets (32). Our strategy of combining frequent somatic mutations of newly discovered cancer genes in assessment of tumor clonality will be important to improve patient management of other multiple primary cancers.

We thank Chih-Hsin Chen for her skillful technical support and Wen-Chen Wu for his help in preparation of the manuscript and artwork.