Although human lung adenocarcinoma has diverse histological subtypes, the correlation between histological subtypes and occurrence of the p53 gene mutation has been given less attention. We investigated 145 surgically resected lung adenocarcinomas to search for the incidence of p53 mutations and for record data on survival in each histological subtype, according to the new WHO criteria (1999). The frequency of p53 mutation in bronchioloalveolar carcinoma (BAC; 0% in 17 cases) and BAC with invasive growth component (BAC-invasive; 11% in 27 cases), which is conventionally categorized as the mixed subtype in WHO typing, were apparently significantly lower than in other types (non-BAC including acinar, papillary, solid, or mixed histology with these subtypes; 48% in 101 cases; P < 0.01). Multivariate analysis revealed that the histological subtype including BAC-invasive was a strong, independent, and significant prognostic factor (P < 0.03), as were tumor size and pathological stage (P < 0.001 and 0.002, respectively) for overall survival. However, the occurrence of p53 mutation itself was seen to be significant only in case of the univariate analysis. Therefore, histological subtyping may be a better prognostic indicator than is p53 mutation. These findings suggest that the WHO classification with the BAC and BAC-invasive from other histological subtypes may prove useful to predict the outcome for surgically treated patients with lung adenocarcinoma.
Lung carcinomas are currently the leading cause of cancer-related death in most countries, including Japan. Among non-small cell lung cancers, adenocarcinoma is increasing in frequency and accounts for almost half the number of lung cancers (1). Adenocarcinoma is of diverse subtypes (1, 2); however, little information is available regarding biological characteristics of each subtype. BAC2, a subcategory of lung adenocarcinoma, has a fairly good prognosis for surgically treated patients; however, the prognosis of BAC with fibrotic foci associated with destructive and invasive growth (BAC-invasive), which is categorized in mixed subtypes in the WHO classification published in 1999, is poorer than that of BAC without it (2). Because emerging evidence suggests that accumulation of allelic mutations largely affects the biological and clinical behaviors of neoplasms, this means that histological subtypes may respond to the diverse gene abnormalities. However, there is a paucity in data on the relationship between histological subtypes of human peripheral lung adenocarcinoma and genetic aberrations.
Neoplastic transformation is considered to be the result of a multistep accumulation of genetic abnormalities, including either activation of oncogenes or inactivation of tumor suppressor genes. p53 tumor suppressor gene mutation is common and frequent among genetic abnormalities in various human cancers, suggesting that the occurrence is a fundamentally important step in carcinogenesis, and it may even play a key role in the clinical prognosis. Regarding non-small cell lung cancers, although many workers have investigated the relationship of p53 abnormalities and prognosis, the results have differed; hence, the clinicopathological significance of p53 alteration has remained unknown. The findings conflicted perhaps because of the methodology used in examinations, such as immunohistochemistry and genomic analysis, regardless of histological subtypes. An important issue to be considered is that there is little information regarding if and when p53 mutation occurs during tumor progression. Although recent studies (3, 4, 5, 6, 7) suggest that BAC likely derives from an atypical adenomatous hyperplasia, a putative premalignant lesion, we find no documentation as to when p53 mutation occurs.
In the present study, we examined (a) the frequency of p53 mutation and (b) the clinicopathological background in each histological subtype, classified according both to the current WHO criteria, and the minor modification, including a proposed subgroup of BAC-invasive. On the basis of our data, we propose that BAC-invasive type should be classed independently from the mixed subtype of lung adenocarcinoma, with regard to both p53 mutations and clinical prognosis. We also found that histopathological subtyping with this modification is a more productive and independent prognostic indicator than is p53 mutation.
MATERIALS AND METHODS
Patients and Tissue Preparation.
p53 mutational analysis was made on 145 specimens with primary lung adenocarcinomas. The patients were surgically treated at Kyushu University Hospital, Fukuoka, Japan during the period from 1995 to 1998. These Japanese patients were never treated with chemotherapy or irradiation before the surgery. Table 1 shows the clinicopathological and histopathological background of these patients. The average age was 66.5 ± 9.5 years, and the man/woman ratio was 1.3. According to the tumor-node-metastasis staging system of the Union International Contre le Cancer (8), 82 patients were in pathological stage I, 12 were in stage II, 47 were in stage III, and 4 were in stage IV. Patients with a smoking history were divided into smokers including not only current smokers but also those with a past history of smoking or nonsmokers without any past history of smoking. The resected specimens were fixed with 10% buffered formalin and sliced at 5 mm. Sections (4 μm thick) obtained from the maximal cut surface area of the cancer tissue were stained with H&E, elastica Van Gieson’s stain, and Alcian blue.
The histopathological classification was essentially done based on the criteria of WHO (Ref. 1; Figs. 1,2,3) but with some modifications, as indicated below, and was determined by four pathologists (T. K., S. M., Y. M., K. Sue.). An adenocarcinoma showing a pure bronchioloalveolar growth pattern and neither evidence of stromal, vascular, nor pleural invasion is categorized in BAC according to the classification of WHO (Fig. 1). Concerning adenocarcinoma of mixed subtypes, a predominant subtype was evident in this study (Fig. 2 and 3). An adenocarcinoma, predominantly with a bronchioloalveolar component but with stromal invasion, was classified as an independent group, i.e., BAC with invasive growth (Fig. 2). In this category, we defined the area of fibrosis with invasive growth as less than 50% of the tumor area. The invasive growth of cancer cells was confirmed using elastica Van Gieson’s stain.
DNA Extraction and PCR of Exon 2 to 9 of the p53 Gene.
The 4-μm-thick sections, excised from the paraffin-embedded blocks of the resected lung cancers, were dewaxed and lightly stained with hematoxylin. The cancer cells, dissected under a light microscope, were collected into 0.6-ml siliconized microcentrifuge tubes. The precipitates were digested with 0.02% proteinase K in 200 μl of 20 mm Tris-HCl (pH8.0) containing 1 mm EDTA and 0.5% Tween 20 at 37°C for 42 h and then were heated at 95°C for 15 min to inactivate proteinase K activity. One μl of digested sample was then used for the following PCR reactions.
Exons 2–9 of the p53 gene were analyzed using PCR-SSCP and p53-specific oligonucleotide primers (Table 2). Thereafter, semi-nested or nested PCR was done as described (9), but with some modifications. The first PCR reaction was performed in a 20-μl reaction mixture containing 20 mm Tris (pH 8.4), 50 mm KCl, 1.5 mm MgCl2, 200 μm of deoxyribonucleotide triphosphate, 0.2 μm of outer primer pairs, 1 unit of Taq DNA polymerase (Takara Shuzo Co., Kyoto, Japan), and 1 μl of template DNA. The first PCR product was diluted to a ratio of 1:50 in distilled water, and 1 μl of the dilution, as a template, was applied to the second PCR. The PCR reaction conditions for the first PCR were 95°C for 20 s, annealing temperature 59–60°C for 20 s, and 72°C for 20 s (35 cycles), whereas the second PCR was run at 95°C for 20 s, 59°C for 20 s, and 72°C for 20 s (30 cycles). A Perkin-Elmer 9600 Thermal Cycler (Perkin-Elmer Co., Norwalk, CT) was used.
Mutations in exons 2 through 9 of the p53 gene were identified by SSCP, as described (9), but with minor modifications. Each secondary PCR product (0.6 μl) was mixed with 4.5 μl of loading buffer composed of 83% formamide containing 8.3 mm EDTA and 0.05% methyl violet. After denaturing DNA at 80°C for 5 min, 2 μl of the mixture was electrophoresed at 800 v for 2 h in a 5% acrylamide gel with 5% glycerol. PCR-SSCP procedures were repeated at least twice to confirm the reproducibility for the same mobility shift of the bands.
All of the mutations detected by PCR-SSCP were confirmed by direct DNA sequencing, as described (9), but with minor modifications. Any band showing any aberrant mobility shifts was excised from the gel and eluted into water at 80°C for 5 min, followed by reamplification with the same inner primers and conditions, as described above. The samples were subjected to subsequent direct DNA sequencing, using a Thermo Sequenase core sequencing kit (Amersham, CA) and SQ-5500 DNA sequencer (Hitachi, Japan).
Immunohistochemical Analysis of P53 Protein.
An immunohistochemical study was done using monoclonal antihuman P53 antibody (DO7; Novocastra, Newcastle, United Kingdom). For the antigenic retrieval of the antibody, the sections were autoclaved for 5 min at 121°C in 0.1 m citrate buffer solution (pH 6.0). After treating the sections with 1.5% milk solution to reduce the nonspecific absorption of antibody, the sections were reacted with the primary monoclonal antibody diluted to 1:100 overnight at 4°C. The tissue sections were treated with biotin-labeled antimouse antibody and then with 0.1% H2O2-methanol solution, followed by the streptavidin-biotin-peroxidase complex method (10).
The P53-labeling index of cancer cells in each cancer tissue was determined by counting the number of P53-positive cells among at least 300 cancer cells.
To estimate the correlation between the frequency of p53 mutation and clinicopathological data, including histological subtypes, tumor size, smoking status, and pathological stage, χ2 test, Student’s t test, and the Mann-Whitney U test were used. All of the Ps were based on two-hypothesis testing, and statistical significance was assumed at a level of P < 0.05. Survival curves were obtained using the Kaplan-Meier method, and the statistical significance of differences was calculated using the log-rank test. Multivariate analysis was performed to identify independent prognostic factors and to assess the hazard ratio with the Cox proportional hazards model, using the Statistical Package for Social Science (SAS). In this model, seven factors potentially related to survival (age at surgery, gender, histological subtype, p53 gene status, tumor size, smoking history, and pathological stage) were included, and the model selection for identifying the subset of significant variables was based on the stepwise method for background selection. Discriminant analysis was also examined between an independent prognostic factor obtained from multivariate analysis and other factors. In this analysis, a stepwise method for background selection was used.
The 145 cases consisted of 17 nonmucinous and mucinous BACs, 27 BACs with invasive growth (Figs. 1 and 2, respectively; Table 1), 4 acinar adenocarcinomas, 59 papillary adenocarcinomas, and 38 solid adenocarcinomas (Fig. 3, a–c, respectively; Table 1). A mean of the area of lepidic growth in BAC with invasion was 90.4 ± 10.4% (mean ± SD).
The Frequency of p53 Mutation in Each Histological Subtype.
Of 145 cases, 51 (35%) had a mutation in the p53 gene in exons 2–9 (Tables 1 and 3), 5 (9.8%) in exon 4, 26 (51%) in exon 5, 4 (7.8%) in exon 6, 8 (15.7%) in exon 7, 7 (13.7%) in exon 8, and 1 (2%) in exon 9. Cases 7 and 29 had two different mutations (Table 3). No mutation was found in exons 2 and 3. Regarding the relation of p53 mutation to histological subtypes, p53 mutations were found in 3 of 27 (11%) BACs with invasion, 1 (25%) of acinar adenocarcinomas, 27 (46%) of papillary adenocarcinomas, and 20 (53%) of solid adenocarcinomas (Table 1), but no mutation was found in the case of BAC alone (P < 0.01; Table 1).
Relationship between p53 Mutation and Clinicopathological Data.
There was no difference in mean age, mean tumor size, and pathological stage with regard to p53 gene status (Table 1). The frequency of p53 mutation was significantly higher in smokers than that in nonsmokers, and the correlation coefficient was apparently significant (P = 0.012; r = 0.22; Table 1). There was a tendency toward a more frequent occurrence of p53 mutation in men than of that in women but with no statistical significance (P = 0.07; Table 1).
Immunohistochemical Overexpression of P53 Protein and the Relation to p53 Gene Mutation.
In Table 4, 61 of the 145 cases (42%) showed overexpression of P53 protein. The concordance rate of overall cases between P53 immunohistochemistry and p53 gene status was 67% with a statistical significance (P < 0.001). However, the concordance rates were higher in BAC without and with invasion (94% and 85%, respectively; P < 0.05) than that of the acinar, papillary, and solid adenocarcinomas ranging from 50 to 58%.
Relationship between Histological Subtypes and Smoking History and p53 Mutational Pattern.
The frequency of smoking in non-BAC cases was significantly higher than that of BACs and BACs with invasion (71% and 30%, respectively; P < 0.001). G:C → T:A transversions were observed in 9 (18.8%) of 48 non-BAC types, but in none of 3 BACs with invasion.
Survival Rate, Histological Subtype, and p53 Gene Status.
Analyses for the relationship between histological subtypes, p53 gene status, and survival rates revealed that all of the patients with BAC with invasion, acinar, papillary, or solid type had a poorer prognosis than in cases with BAC alone, which was statistically significant (P = 0.001; Fig. 4,A). We also studied the relationship between p53 gene status and overall survival. Of the 145 cases, patients with p53 mutations had a shorter survival period than did those without any p53 mutation (Fig. 4,B; P < 0.05); however, among patients with either acinar, papillary, or solid type, no statistical significance was noted with regard to p53 gene status (Fig. 4 C). Examining the relationship between the type of p53 mutation and survival by stage, no statistical significance was apparent regarding survival time between patients with p53 null mutations and missense mutations in any stage (data not shown).
Univariate and Multivariate Analyses for Survival.
Univariate analysis of age, gender, histological subtype, p53 gene status, tumor size, smoking history, and pathological stage for survival significance was made. Tumor size, pathological stage, histological subtype, smoking, and p53 mutation (as the lower P) were significant prognostic factors (Table 5). However, multivariate analysis of the same variables revealed that tumor size and patients’ stage were statistically significant, but smoking and p53 mutation were not significant (Table 5). In these analyses, histological subtypes were arranged into three categories in accordance to the prognostic ranks. Categories 1, 2, and 3 were BAC alone, BAC with invasion, and combined group with acinar, papillary, and solid type, respectively, and the survival rates of these groups showed a statistical significance in the log-rank test with the Kaplan-Meier method (Fig. 4 A). In all of the cases, p53 mutation was not an independent prognostic factor, whereas histological subtype was an independent prognostic factor with a statistical significance (P = 0.028).
Discriminant Analysis for Histological Subtype.
Our study clearly elucidated that: (a) the frequency of p53 gene mutation varies with the histological type of peripheral human adenocarcinomas. This mutation was rare both in BAC (0 of 17 cases; 0%) and BAC-invasive (3 of 27; 11%) but relatively frequent in non-BAC (48 of 101; 48%); and (b) the concordance between P53 immunohistochemistry and p53 gene status is much higher in BAC and BAC-invasive (average, 89%; P < 0.002) than that in other cases (average, 57%; P = 0.12). To our knowledge, this is the first molecular evidence related to possible biological properties in each histological subtype of human lung adenocarcinoma.
Although a significantly poor prognosis was observed in patients with p53 mutation, the frequency of p53 mutation did not significantly affect the prognosis in patients with non-BAC. Furthermore, multivariate analysis for overall survival revealed that p53 mutation was not a significant prognostic factor, whereas the histological subtype is an independent and significant indicator. Together with the discordance between p53 mutation and P53 immunoreactivity shown in Table 4, these findings suggest that the enhanced expression of P53 is likely attributable to mechanisms other than p53 mutation in non-BAC, and there is a lack of significant correlation between p53 mutation and the prognosis in patients with non-BAC. Inversely, BAC and BAC-invasive showed strong concordance between p53 mutation and P53 immunoreactivity, which suggests that the prolonged half-life of P53 seemed largely attributable to the mutated p53 allele.
The discordance between p53 gene status and P53 immunoreactivity within each subtype of lung adenocarcinoma has been unclear. In vitro studies suggested that the type of mutation, including null mutation, other cellular proteins associated with P53, including murine double minute 2 or viral oncoproteins such as simian virus 40 large T antigen, and DNA damage may have an effect.
Many studies (11, 12, 13) revealed that p53 mutations are more frequent in squamous cell carcinoma, well-known smoking-related malignancies, than in adenocarcinomas. In the present study, the frequency of p53 mutation in papillary and solid adenocarcinoma was relatively higher than in BAC groups and close to findings in the case of the squamous cell carcinoma, where the reported rate was from 60 to 68% (11, 12, 13). G:C → T:A transversion in the p53 gene is a smoking-related mutation (13, 14) and is assumed to arise as a direct consequence of benzopyrene diol epoxide-DNA adducts (15). In the present study, the relationship of smoking history with p53 mutation in all of the patients showed a correlation coefficient, and the ratio of patients who used tobacco was significantly higher in patients with non-BAC. These results suggest that a history of tobacco smoking was strongly associated with p53 mutation in these subtypes, similar to findings with squamous cell carcinoma of the lung. Conversely, BAC and BAC-invasive showed a poorer correlation to smoking than did non-BAC.
Histological subtypes reflecting p53 status are useful prognostic markers for determining survival time after surgical intervention. p53 mutation was not a significant prognostic factor, yet discriminant analysis revealed that p53 mutation was a significant factor for discriminating histological subtype.
Our findings also suggest that BAC-invasive should be classed independently from the mixed subtype in the WHO criteria, because this type of adenocarcinoma has a better prognosis than other mixed types composed mainly of non-BAC (Fig. 4 A). There are reports that showed the size of the scar or the area of lepidic growth component is prognostically important in lung adenocarcinoma (16, 17); however, in the present study, no statistical significance was found between the area of lepidic growth component and patients’ survival (data not shown). This conflict may be related to the fact that we did not limit the tumor size of subjects examined in this study, which has been suggested to be an important prognostic factor, within 3 cm in diameter.
In conclusion, BAC is independent from other histological subtypes from the point of clinicopathological and molecular evidence. Evaluation of the histological subtype based on the current WHO classification can predict the clinical prognosis more so than analyzes of the p53 gene mutation. We propose that patients with non-BAC type adenocarcinoma should be prescribed adjuvant therapies regardless of the p53 gene status.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The abbreviations used are: BAC, bronchioloalveolar carcinoma; SSCP, single-strand conformation polymorphism.
We thank Hiroshi Fujii for preparing tissue sections and Mariko Ohara for providing language assistance.