Seventy-two small-sized (≤2 cm in diameter) lung adenocarcinomas consisting of 15 noninvasive and 57 invasive tumors were subjected to whole genome allelic imbalance (AI) scanning and mutational analysis of the EGFR, KRAS, and TP53 genes to elucidate genetic pathways of early-stage lung adenocarcinomas. The chromosome 13q13 region showed the most frequent AI (58%) and was affected at similar frequencies between noninvasive and invasive tumors (53% and 60%, respectively), as EGFR and KRAS mutations were. The number of AI regions as well as the frequency of TP53 mutations in invasive tumors was significantly higher than those in noninvasive ones [9.8 ± 5.6 versus 4.8 ± 2.8 (P = 0.00002) and 61% versus 13% (P = 0.001), respectively]. In particular, AIs at the chromosome 11p11-p12, 17p12-p13, and 18p11 regions in invasive tumors were significantly more frequent than those in noninvasive ones (P < 0.01). The results indicated that noninvasive tumors were developed by EGFR, KRAS, and 13q alterations and progressed to invasive ones by subsequent alterations of several tumor suppressor genes, including those on 11p11-p12, 17p12-p13, and 18p11 and TP53. AI at 8p21 was significantly more frequent in advanced stages (>IA) and associated with worse prognoses (P = 0.04) and, thus, would be involved in invasion and/or metastasis of adenocarcinoma cells and useful for the prediction of prognosis of patients with small-sized lung adenocarcinoma. [Cancer Res 2009;69(4):1615–23]

Adenocarcinoma is the most common histologic type of lung cancer, and its highly invasive and metastatic phenotypes cause poor prognosis of adenocarcinoma patients (1). The EGFR and KRAS genes are mutually exclusively mutated in lung adenocarcinomas, and inhibition of activities of these mutants led to the inhibition of adenocarcinoma cell growth (25). EGFR and KRAS mutations have been detected both in noninvasive and invasive adenocarcinomas (68); therefore, these alterations are considered to be critical for the development of lung adenocarcinomas, irrespective of their invasiveness. Recently, we showed that p16 homozygous deletions occur at similar frequencies (∼25%) in noninvasive and invasive adenocarcinomas, whereas the frequency of p53 mutations in invasive ones was much higher than that in noninvasive ones (8). However, phenotypic diversities of adenocarcinomas cannot be explained only by accumulation of these genetic alterations. Importantly, lung adenocarcinomas have been shown to carry allelic losses at tens of chromosomal loci (914). Therefore, inactivation of several other tumor suppressor genes could also be involved in the development as well as progression of lung adenocarcinoma. Furthermore, the roles of those allelic losses in the development and/or progression of lung adenocarcinoma are largely unknown.

Small-sized (i.e., ≤2 cm in maximum diameter) lung adenocarcinomas are mainly in early stages and classified into six histologic types, A to F (15). Type A (localized bronchioloalveolar carcinoma) and type B (localized bronchioloalveolar carcinoma with foci of alveolar structural collapse) are noninvasive tumors, whereas type C (localized bronchioloalveolar carcinoma with foci of active fibroblastic proliferation), type D (poorly differentiated adenocarcinoma), type E (tubular adenocarcinoma), and type F (papillary adenocarcinoma with compressive and destructive growth) are invasive ones (1). Types A, B, and C show growth with replacement of bronchioloalveolar cells (replacement growth type); therefore, type A is considered to progress sequentially through type B to type C. In contrast, types D, E, and F show expansive and destructive growth (nonreplacement growth type), and their precursory noninvasive tumors are unknown. The detection of small-sized lung adenocarcinomas is increasing due to recent advances in spiral computed tomography scans (16, 17). However, even after complete resection by surgery, 20% of patients will not survive because of recurrence within 5 years (15). This implies that a subset of small-sized adenocarcinomas have already metastasized to other organs. Therefore, molecular analyses of small-sized adenocarcinomas are important not only to understand the mechanism of multistage lung carcinogenesis but also to identify molecular targets for diagnosis and therapy of patients.

Due to a limited fraction (∼4%; ref. 18) and a small volume, only a limited number of small-sized adenocarcinomas have been examined for a few genetic alterations, including EGFR, KRAS, and TP53 mutations and allelic imbalance (AI) of a few chromosomal loci (68, 19). However, a subset of the alterations, including TP53 mutations, have been detected more frequently in invasive tumors than in noninvasive tumors. Therefore, it is likely that progression from noninvasive to invasive tumors is caused by accumulation of genetic alterations. Thus, a comprehensive analysis of genetic alterations in small-sized lung adenocarcinoma will give us critical information on molecular mechanisms of lung adenocarcinoma progression as well as genes involved in invasion and metastasis of adenocarcinoma cells. In this study, 72 small-sized adenocarcinomas, consisting of 15 noninvasive and 57 invasive tumors, were microdissected and subjected to whole genome AI scanning and mutational analysis of the EGFR, KRAS, and TP53 genes. Based on the results, a genetic model for the development of noninvasive adenocarcinomas and their progression to invasive adenocarcinomas was depicted, and the association of genetic alterations with clinicopathologic factors was investigated.

Patients and tissues. In total, 379 patients with small-sized lung adenocarcinoma underwent curative pulmonary resections from 1993 to 2000 at the National Cancer Center Hospital, Tokyo, Japan. None of them received chemotherapy and radiotherapy before or after surgery. Tumors were pathologically diagnosed according to the tumor-node-metastasis classification (20). In 205 cases, surgical specimens fixed with methanol were available and, thus, were applicable for molecular analyses. Adenocarcinomas were classified into six histologic types: 7 type A, 18 type B, 152 type C, 17 type D, 6 type E, and 5 type F, according to the criteria of small-sized lung adenocarcinoma (15). All type A, B, and D cases and 40 of the 152 type C cases were subjected to DNA extraction. For the validation of association between genetic alterations and prognosis, additional 33 type C and 7 type D tumors were chosen from 380 patients who underwent curative pulmonary resections from 2001 to 2004 at the National Cancer Center Hospital and subjected to DNA extraction.

Cancer cells were obtained by laser capture microdissection using the PixCell Laser Capture Microdissection System (Arcturus Engineering) as previously described (6). Noncancerous lung tissues were obtained from the regions >5 cm from tumors with macroscopically normal morphology in the resected lobes of the lung. Genomic DNAs were extracted as described previously (6). Both the cancerous and noncancerous cell DNAs of sufficient amounts for this study were obtained from 72 cases: 6 type A, 9 type B, 40 type C, and 17 type D cases. This study was undertaken under the approval of the Institutional Review Board of National Cancer Center.

Mutation analysis of the EGFR/KRAS/TP53 genes. The status of EGFR/KRAS/TP53 mutations in 30 cases was previously reported (6, 8). Forty-two cases were newly analyzed in this study for mutations in exons 18 to 21 of the EGFR gene, in exons 1 to 2 of the KRAS gene, and in exons 4 to 8 of the TP53 gene by genomic PCR and direct sequencing as described previously (6).

Detection of AI by SNP array analysis. AI was examined by the GeneChip Human Mapping 10K Array analysis (Affymetrix, Inc.) according to the method optimized for the analysis of a small amount of DNA samples as described previously (21). Genotype calls of tumors and normal lung tissues were obtained in 80.0% to 94.5% (88.3 ± 3.5) and 83.4% to 97.9% (94.5 ± 2.9), respectively, of the 11,037 SNP sites on the array, and 1,799 to 3,101 loci were informative for detection of AI in the tumors. When a SNP locus was called “homozygous” in tumor DNA and “heterozygous” in the corresponding normal tissue DNA, such a locus was judged as AI in the tumor. The fraction of AI for each tumor was calculated as the fraction of loci judged as AI.

Definition of AI region. AI regions were defined as follows by taking the call error into account. The fraction of error calls for each sample, calculated as the fraction of SNP loci for which normal tissue DNA was called as “homozygous” and tumor DNA as “heterozygous”, were 0.0% to 8.9% (1.8 ± 2.3). The appearance of six consecutive “AI” loci by the call error was far less than 1, even for the case with the largest informative loci and the highest probability of call error [i.e., 3,101 informative loci × (0.089)6 = 0.0015]. Thus, AI regions were defined by the criterion of containing at least six consecutive AI loci. Under the same criterion and by combination of SNP array data with spectral karyotyping and array-comparative genomic hybridization data, 1 of 13 (7.7%) trisomic chromosomes was judged as AI, and 14 of 215 (6.5%) AI regions were due to amplification/gain of one allele in our recent study (22). Therefore, one of allelic chromosomal segments was likely to be lost in most (>90%) of AI regions defined in this study.

Microsatellite analysis. Microsatellite markers, D8S1116 and D8S322, were chosen based on the map location and allele frequency in the Japanese population.6

One hundred picograms of DNA were used for PCR with a set of primers labeled with FAM and NED for D8S1116 and D8S322, respectively. PCR products were run through a 3130xl Genetic Analyzer (Applied Biosystems) and analyzed by the GeneMapper software. A reduction >0.5 or an increase >2.0 of an allele in tumor was determined as AI.

Statistical analysis. Fisher's exact test was used to assess the association of two variables. The differences in the values of fraction of AI and in the number of AI regions between two groups were assessed by the unpaired t test. Association of genetic alterations with clinicopathologic characteristics was evaluated by logistic regression analysis (variables with P ≤ 0.2 were selected). Overall survival of patients with and without genetic alterations was compared by Kaplan-Meier curves and the log-rank test. P < 0.05 was considered statistically significant. Statistical analysis was done using JMP software (version 5.1.1, SAS Institute, Inc.).

Clinicopathologic characteristics and the status of EGFR/KRAS/TP53 mutations. Clinicopathologic characteristics of 72 cases with small-sized lung adenocarcinomas are summarized in Table 1. Adenocarcinomas consisted of 15 noninvasive and 57 invasive tumors. These 72 tumors were of 55 replacement growth types (types A, B, and C) and of 17 nonreplacement growth types (type D). All type A, B, and D cases, whose genomic DNAs were available, in our cohort were enrolled for this study. Fourteen of the 15 noninvasive cases showed good prognoses, whereas 12 of the 17 type D cases showed them. To investigate the association of genetic alterations with prognosis, 40 type C cases consisting of all 19 cases with poor prognosis and 21 random cases with good prognosis were selected from the 152 type C cases in our cohort. Accordingly, the postoperative 5-year overall survival rate of 40 type C cases selected (53%) was considerably lower than those of all 152 type C cases (84%), and the population of advanced stages (>IA) in type C cases selected (53%) was higher than that in all type C cases (24%).

Table 1.

Clinicopathologic characteristics and genetic alterations in small-sized lung adenocarcinomas

Clinicopathologic characteristics and genetic alterations*SubsetTotalSubtype
Noninvasive tumors
Invasive tumors

All (%)
A (%)
B (%)
All (%)
C (%)
D (%)
n = 72n = 15n = 6n = 9n = 57n = 40n = 17
Gender Male 35 (49) 8 (53) 5 (83) 3 (33) 27 (47) 17 (42) 10 (59) 
 Female 37 (51) 7 (47) 1 (17) 6 (67) 30 (53) 23 (58) 7 (41) 
Smoking history Nonsmoker 36 (50) 8 (53) 2 (33) 6 (67) 28 (49) 22 (55) 6 (35) 
 Smoker 36 (50) 7 (47) 4 (67) 3 (33) 29 (51) 18 (45) 11 (65) 
Pathologic stage IA 41 (57) 15 (100) 6 (100) 9 (100) 26 (46) 19 (47) 7 (41) 
 >IA 31 (43) 0 (0) 0 (0) 0 (0) 31 (54) 21 (53) 1 (59) 
Prognosis Alive 47 (65) 14 (93) 6 (100) 8 (89) 33 (58) 21 (53) 12 (71) 
 Dead 25 (35) 1 (7) 0 (0) 1 (11) 24 (42) 19 (47) 5 (29) 
EGFR/KRAS§ E(+)/K(−) 40 (56) 10 (67) 3 (50) 7 (78) 30 (53) 27 (68) 3 (18), 
 E(−)/K(+) 8 (11) 2 (13) 2 (33) 0 (0) 6 (11) 3 (8) 3 (18) 
 E(−)/K(−) 24 (33) 3 (20) 1 (17) 2 (22) 21 (37) 10 (25) 11 (65), 
TP53 Mutation(+) 37 (51) 2 (13) 0 (0) 2 (22) 35 (61) 21 (53) 14 (82), 
 Mutation(−) 35 (49) 13 (87) 6 (100) 7 (78) 22 (39) 19 (47) 3 (18) 
         
Fraction of AI (%)  24.2 ± 15.4 13.8 ± 8.3 11.4 ± 11.5 15.5 ± 5.5 27.0 ± 15.8 27.0 ± 15.7 26.9 ± 6.3 
No. of AI regions  8.7 ± 5.5 4.8 ± 2.8 5.0 ± 3.5 4.7 ± 2.5 9.8 ± 5.6 9.9 ± 5.3 9.6 ± 6.3 
Clinicopathologic characteristics and genetic alterations*SubsetTotalSubtype
Noninvasive tumors
Invasive tumors

All (%)
A (%)
B (%)
All (%)
C (%)
D (%)
n = 72n = 15n = 6n = 9n = 57n = 40n = 17
Gender Male 35 (49) 8 (53) 5 (83) 3 (33) 27 (47) 17 (42) 10 (59) 
 Female 37 (51) 7 (47) 1 (17) 6 (67) 30 (53) 23 (58) 7 (41) 
Smoking history Nonsmoker 36 (50) 8 (53) 2 (33) 6 (67) 28 (49) 22 (55) 6 (35) 
 Smoker 36 (50) 7 (47) 4 (67) 3 (33) 29 (51) 18 (45) 11 (65) 
Pathologic stage IA 41 (57) 15 (100) 6 (100) 9 (100) 26 (46) 19 (47) 7 (41) 
 >IA 31 (43) 0 (0) 0 (0) 0 (0) 31 (54) 21 (53) 1 (59) 
Prognosis Alive 47 (65) 14 (93) 6 (100) 8 (89) 33 (58) 21 (53) 12 (71) 
 Dead 25 (35) 1 (7) 0 (0) 1 (11) 24 (42) 19 (47) 5 (29) 
EGFR/KRAS§ E(+)/K(−) 40 (56) 10 (67) 3 (50) 7 (78) 30 (53) 27 (68) 3 (18), 
 E(−)/K(+) 8 (11) 2 (13) 2 (33) 0 (0) 6 (11) 3 (8) 3 (18) 
 E(−)/K(−) 24 (33) 3 (20) 1 (17) 2 (22) 21 (37) 10 (25) 11 (65), 
TP53 Mutation(+) 37 (51) 2 (13) 0 (0) 2 (22) 35 (61) 21 (53) 14 (82), 
 Mutation(−) 35 (49) 13 (87) 6 (100) 7 (78) 22 (39) 19 (47) 3 (18) 
         
Fraction of AI (%)  24.2 ± 15.4 13.8 ± 8.3 11.4 ± 11.5 15.5 ± 5.5 27.0 ± 15.8 27.0 ± 15.7 26.9 ± 6.3 
No. of AI regions  8.7 ± 5.5 4.8 ± 2.8 5.0 ± 3.5 4.7 ± 2.5 9.8 ± 5.6 9.9 ± 5.3 9.6 ± 6.3 
*

Clinicopathologic characteristics and genetic alterations are shown by the number of cases (%), and fraction of AI and the number of AI regions are shown as mean ± SD.

P < 0.05 for the difference against types A + B by Fischer's exact test.

Postoperative 5-y overall survival.

§

E(+), EGFR mutation (+); E(−), EGFR mutation (−); K(+), KRAS mutation (+); K(−), KRAS mutation (−).

P < 0.05 for the difference against type C by Fischer's exact test.

P < 0.05 for the difference against types A + B by unpaired t test.

EGFR and KRAS mutations were detected in a mutually exclusive manner, being consistent with previous studies (58, 23, 24). Frequencies of EGFR and KRAS mutations were not significantly different between noninvasive tumors and invasive tumors (67%/53% and 13%/11%, respectively; P > 0.05). In replacement growth types (types A, B, and C), EGFR mutations were detected in ≥50% of the cases, whereas KRAS mutations were detected with lower frequencies (≤33%). EGFR mutations were detected at a lower frequency in nonreplacement growth types (18%) than in replacement growth types (67%), whereas KRAS mutations were detected at similar frequencies in both types (18%/9%). The frequency of TP53 mutations in invasive tumors was significantly higher than that in noninvasive tumors (61%/13%; P = 0.001). TP53 mutations were detected in none of type A (0%) and in a small subset of type B (22%) but in the majority of type C (53%) and type D (82%) tumors.

Frequency of AI on each chromosome arm. All 72 cases were subjected to a whole-genome AI scanning using a Human Mapping 10K Array covering 11,560 SNP sites placed at a mean interval of 210 kb. Fraction of AI in total cases was 24.2 ± 15.4% and was twice higher in invasive tumors than in noninvasive tumors (P = 0.00008; Table 1). The number of AI regions ranged from 0 to 21 (mean ± SD, 8.7 ± 5.5) and was also twice higher in invasive tumors than in noninvasive tumors (P = 0.00002).

At least two AI regions were mapped in all chromosome arms except five acrocentric arms; therefore, common regions of AIs among the 72 cases were defined on each chromosome arm. In total, 52 regions on 39 chromosome arms were defined as common, with two or more regions on 1p, 4q, 5p, 7p, 7q, 8p, 10p, 11p, and 16q (Supplementary Table S1). Frequencies of AI ranged from 8% to 58% (mean ± SD, 25.6 ± 11.8; Table 2). Therefore, regions with frequencies of AI more than mean + SD (37.4) were considered as being “hotspots” of AIs and those more than mean + 2SD (49.2) as being “critical regions” of AIs. There were nine hotspots of AIs, and two critical regions of AIs that were mapped to chromosomes 13q13 and 17p12-p13. The frequency of AI at 13q13 was 58% whereas that at 17p12-p13 was 56%.

Table 2.

Frequency of AI on each chromosome arm in small-sized lung adenocarcinomas

Chromosomal regions with AI
No. of cases (%)
TotalSubtype
Noninvasive
Invasive


A + B
All
C
D
n = 72n = 15n = 57n = 40n = 17
1p21-p22 13 (18) (20) 10 (18) (20) (12) 
1p13-p21 13 (18) (20) 10 (18) (20) (12) 
1q21 13 (18) (20) 10 (18) (18) (18) 
2p22-p24 11 (15) (13) (16) (15) (18) 
2q35-q37 12 (17) (13) 10 (18) (23) (6) 
3p14 24 (33) (27) 20 (35) 14 (35) (35) 
3q11 23 (32) (27) 19 (33) 13 (33) (35) 
4p12-p13 11 (15) (13) (16) (18) (12) 
4q13 17 (24) (20) 14 (25) 12 (30) (12) 
4q22 17 (24) (20) 14 (25) 11 (28) (18) 
5p15 10 (14) (0) 10 (18) (23) (6) 
5p12-p13 10 (14) (13) (14) (18) (6) 
5q23 23 (32) (20) 20 (35) 16 (40) (24) 
6p11 14 (19) (0) 14 (25)*, (20) (35)*, 
6q22 26 (36) (27) 22 (39) 14 (35) (47) 
7p21-p22 (8) (7) (9) (8) (12) 
7p15 (8) (7) (9) (8) (12) 
7q11 (11) (20) (9) (8) (12) 
7q31-q32 (11) (20) (9) (5) (18) 
7q33-q35 (11) (20) (9) (5) (18) 
7q36 (11) (20) (9) (5) (18) 
8p22-p23 31 (43) (27) 27 (47) 19 (48) (47) 
8p21 31 (43) (27) 27 (47) 20 (50) (41) 
8q11 27 (38) (20) 24 (42) 17 (43) (41) 
9p22 33 (46) (27) 29 (51) 19 (48) 10 (59) 
9q12-q13 28 (39) (20) 25 (44) 16 (40) (53) 
10p13-p14 12 (17) (13) 10 (18) (15) (24) 
10p12-p13 12 (17) (20) (16) (15) (18) 
10p12 12 (17) (20) (16) (18) (12) 
10p11 12 (17) (20) (16) (18) (12) 
10q25-q26 16 (22) (27) 12 (21) (23) (18) 
11p12 16 (22) (0) 16 (28)*, 10 (25)*, (35)* 
11p11-p12 16 (22) (0) 16 (28)*, (23) (41) 
11q24-q25 21 (29) (7) 20 (35)*, 15 (38)*, (29) 
12p12-p13 24 (33) (13) 22 (39) 16 (40) (35) 
12q11-q12 21 (29) (7) 20 (35) 14 (35)*, (35) 
13q13 42 (58) (53) 34 (60) 26 (65) (47) 
14q21 14 (19) (7) 13 (23) (18) (35) 
15q14-q21 23 (32) (7) 22 (39)*, 16 (40)*, (35) 
16p11 11 (15) (0) 11 (19) (23) (12) 
16q22 18 (25) (7) 17 (30) 13 (33) (24) 
16q23-q24 18 (25) (7) 17 (30) 13 (33) (24) 
17p12-p13 40 (56) (13) 38 (67) 25 (63) 13 (76) 
17q11 32 (44) (13) 30 (53) 21 (53)* (53)* 
18p11 24 (33) (7) 23 (40)* 18 (45) (29) 
18q21 30 (42) (27) 26 (46) 19 (48) (41) 
19p12-p13 20 (28) (27) 16 (28) 10 (25) (35) 
19q12 20 (28) (27) 16 (28) 10 (25) (35) 
20p11 20 (28) (7) 19 (33) 15 (38)* (24) 
20q11 19 (26) (7) 18 (32) 14 (35)* (24) 
21q11-q21 13 (18) (0) 13 (23) (23) (24) 
22q12 18 (25) (7) 17 (30) 10 (25) (41)* 
Chromosomal regions with AI
No. of cases (%)
TotalSubtype
Noninvasive
Invasive


A + B
All
C
D
n = 72n = 15n = 57n = 40n = 17
1p21-p22 13 (18) (20) 10 (18) (20) (12) 
1p13-p21 13 (18) (20) 10 (18) (20) (12) 
1q21 13 (18) (20) 10 (18) (18) (18) 
2p22-p24 11 (15) (13) (16) (15) (18) 
2q35-q37 12 (17) (13) 10 (18) (23) (6) 
3p14 24 (33) (27) 20 (35) 14 (35) (35) 
3q11 23 (32) (27) 19 (33) 13 (33) (35) 
4p12-p13 11 (15) (13) (16) (18) (12) 
4q13 17 (24) (20) 14 (25) 12 (30) (12) 
4q22 17 (24) (20) 14 (25) 11 (28) (18) 
5p15 10 (14) (0) 10 (18) (23) (6) 
5p12-p13 10 (14) (13) (14) (18) (6) 
5q23 23 (32) (20) 20 (35) 16 (40) (24) 
6p11 14 (19) (0) 14 (25)*, (20) (35)*, 
6q22 26 (36) (27) 22 (39) 14 (35) (47) 
7p21-p22 (8) (7) (9) (8) (12) 
7p15 (8) (7) (9) (8) (12) 
7q11 (11) (20) (9) (8) (12) 
7q31-q32 (11) (20) (9) (5) (18) 
7q33-q35 (11) (20) (9) (5) (18) 
7q36 (11) (20) (9) (5) (18) 
8p22-p23 31 (43) (27) 27 (47) 19 (48) (47) 
8p21 31 (43) (27) 27 (47) 20 (50) (41) 
8q11 27 (38) (20) 24 (42) 17 (43) (41) 
9p22 33 (46) (27) 29 (51) 19 (48) 10 (59) 
9q12-q13 28 (39) (20) 25 (44) 16 (40) (53) 
10p13-p14 12 (17) (13) 10 (18) (15) (24) 
10p12-p13 12 (17) (20) (16) (15) (18) 
10p12 12 (17) (20) (16) (18) (12) 
10p11 12 (17) (20) (16) (18) (12) 
10q25-q26 16 (22) (27) 12 (21) (23) (18) 
11p12 16 (22) (0) 16 (28)*, 10 (25)*, (35)* 
11p11-p12 16 (22) (0) 16 (28)*, (23) (41) 
11q24-q25 21 (29) (7) 20 (35)*, 15 (38)*, (29) 
12p12-p13 24 (33) (13) 22 (39) 16 (40) (35) 
12q11-q12 21 (29) (7) 20 (35) 14 (35)*, (35) 
13q13 42 (58) (53) 34 (60) 26 (65) (47) 
14q21 14 (19) (7) 13 (23) (18) (35) 
15q14-q21 23 (32) (7) 22 (39)*, 16 (40)*, (35) 
16p11 11 (15) (0) 11 (19) (23) (12) 
16q22 18 (25) (7) 17 (30) 13 (33) (24) 
16q23-q24 18 (25) (7) 17 (30) 13 (33) (24) 
17p12-p13 40 (56) (13) 38 (67) 25 (63) 13 (76) 
17q11 32 (44) (13) 30 (53) 21 (53)* (53)* 
18p11 24 (33) (7) 23 (40)* 18 (45) (29) 
18q21 30 (42) (27) 26 (46) 19 (48) (41) 
19p12-p13 20 (28) (27) 16 (28) 10 (25) (35) 
19q12 20 (28) (27) 16 (28) 10 (25) (35) 
20p11 20 (28) (7) 19 (33) 15 (38)* (24) 
20q11 19 (26) (7) 18 (32) 14 (35)* (24) 
21q11-q21 13 (18) (0) 13 (23) (23) (24) 
22q12 18 (25) (7) 17 (30) 10 (25) (41)* 

NOTE: Frequency > mean + SD is marked by yellow, and frequency > mean + 2SD is marked by red.

*

P < 0.05 for the difference against noninvasive tumors by Fischer's exact test.

P < 0.01 for the difference against noninvasive tumors by Fischer's exact test.

In noninvasive tumors and invasive tumors, hotspots as well as critical regions of AIs were also defined based on the value of mean ± SD for the frequencies of AIs. In invasive tumors, hotspots and critical regions were further defined separately in types C and D. In noninvasive tumors, there were 11 hotspots distributed on several chromosome arms and was only one critical region at 13q13. In invasive tumors, there were nine hotspots and two critical regions mapped to 13q13 and 17p12-p13. Similarly, there were eight and seven hotspots and two and two critical regions, respectively, in types C and D. As a whole, 16 of either hotspots or critical regions were identified in total of or in each group of small-sized adenocarcinomas.

Common and differential AIs between noninvasive and invasive tumors. We next searched for regions of AI commonly and differentially affected among noninvasive, invasive, type C, and type D tumors (Table 2). Forty of the 52 regions did not show any significant differences in the frequency of AI among the subtypes. In particular, frequencies of AI in the 13q13 region were similar between noninvasive and invasive tumors (53% and 60%, respectively); therefore, 13q13 was commonly and frequently affected both in noninvasive and invasive tumors. The remaining 12 regions showed significant differences in the frequency of AI among the subtypes. For instance, frequencies of AI at 17p12-p13 in all invasive, type C, and type D tumors were significantly higher than that in noninvasive tumors (P < 0.01). Other 11 regions were also significantly more frequently affected by AI in invasive, type C, and/or type D tumors than in noninvasive tumors (P < 0.05). Particularly, differences in the frequency of AI at 18p11 and 11p11-p12 regions in type C and type D, respectively, against that in noninvasive tumors were highly significant (P < 0.01). In addition, two regions, 11p12 and 17q11, were significantly more frequently affected in both types of invasive tumors than in noninvasive tumors (P < 0.05). The other seven regions were significantly more frequently affected in either type C or type D than in noninvasive tumors (P < 0.05). There were no regions that were significantly more frequently affected in noninvasive tumors than in invasive tumors or in either type C or type D tumors. In addition, regions whose frequencies of AI were significantly different between type C and type D tumors were not observed either.

Association of AIs with pathologic stage and prognosis. Nine regions on five different chromosomes showed AIs significantly more frequently in stage >IA tumors than in stage IA tumors (P < 0.05; Table 3). In particular, AI at 8p21 was also significantly more frequently detected in cases with poor prognosis than in cases with good prognosis (P = 0.046; Table 3). AIs of the other 51 regions as well as EGFR, KRAS, and TP53 mutations were not associated with prognosis. The log-rank test also indicated that overall survival of cases with AI at 8p21 is worse than that without (Fig. 1; P = 0.036). Because 8p21 was defined as a hotspot of AI in both noninvasive and invasive tumors, AI at 8p21 was suggested to occur early in the development of adenocarcinoma. Therefore, it is possible that adenocarcinomas with AI at 8p21 are more aggressive than those without, even if the sizes of tumors are small. The result may indicate that AI at 8p21 could be a prognostic marker of small-sized adenocarcinomas.

Table 3.

Correlation of genetic alterations with pathologic stage and prognosis in small-sized lung adenocarcinomas

Genetic alterationNo. of cases (%)
Pathologic stage
Prognosis*
IA
>IA
Alive
Dead
(n = 41)(n = 31)(n = 47)(n = 25)
Allelic imbalance         
    1p21-p22 (17) (19) (17) (20) 
    1p13-p21 (17) (19) (17) (20) 
    1q21 (20) (16) (17) (20) 
    2p22-p24 (15) (16) (17) (12) 
    2q35-q37 (15) (19) (15) (20) 
    3p14 10 (24) 14 (45) 17 (36) (28) 
    3q11 12 (29) 10 (32) 18 (38) (20) 
    4p12-p13 (12) (19) (15) (16) 
    4q13 10 (24) (23) 11 (23) (24) 
    4q22 11 (27) (19) 11 (23) (24) 
    5p15 (10) (19) (9) (24) 
    5p12-p13 (12) (16) (13) (16) 
    5q23 12 (29) 11 (35) 14 (30) (36) 
    6p11 (7) 11 (35) (15) (28) 
    6q22 12 (29) 14 (45) 15 (32) 11 (44) 
    7p21-p22 (2) (16) (9) (8) 
    7p15 (2) (16) (9) (8) 
    7q11 (7) (16) (13) (8) 
    7q31-q32 (7) (16) (15) (4) 
    7q33-q35 (7) (16) (15) (4) 
    7q36 (10) (13) (13) (8) 
    8p22-p23 12 (29) 19 (61) 17 (36) 14 (56) 
    8p21 12 (29) 19 (61) 16 (34) 15 (60) 
    8q11 11 (27) 16 (52) 15 (32) 12 (48) 
    9p22 16 (39) 17 (55) 19 (40) 14 (56) 
    9q12-q13 14 (34) 14 (45) 16 (34) 12 (48) 
    10p13-p14 (15) (19) (19) (12) 
    10p12-p13 (15) (19) (19) (12) 
    10p12 (12) (23) (19) (12) 
    10p11 (12) (23) (19) (12) 
    10q25-q26 (15) 10 (32) 12 (26) (16) 
    11p12 (10) 12 (39) (15) (36) 
    11p11-p12 (12) 11 (35) (17) (32) 
    11q24-q25 (20) 13 (42) 13 (28) (32) 
    12p12-p13 10 (24) 14 (45) 12 (26) 12 (48) 
    12q11-q12 (22) 12 (39) 10 (21) 11 (44) 
    13q13 23 (56) 19 (61) 26 (55) 16 (64) 
    14q21 (15) (26) 11 (23) (12) 
    15q14-q21 10 (24) 13 (42) 13 (28) 10 (40) 
    16p11 (10) (23) (11) (24) 
    16q22 (15) 12 (39) (19) (36) 
    16q23-q24 (15) 12 (39) (19) (36) 
    17p12-p13 17 (41) 23 (74) 25 (53) 15 (60) 
    17q11 15 (37) 17 (55) 19 (40) 13 (52) 
    18p11 10 (24) 14 (45) 13 (28) 11 (44) 
    18q21 14 (34) 16 (52) 19 (40) 11 (44) 
    19p12-p13 11 (27) (29) 12 (26) (32) 
    19q12 11 (27) (29) 13 (28) (28) 
    20p11 (20) 12 (39) 10 (21) 10 (40) 
    20q11 (20) 11 (35) 10 (21) (36) 
    21q11-q21 (15) (23) (13) (28) 
    22q12 (17) 11 (35) (19) (36) 
Mutation         
    EGFR 23 (56) 17 (55) 24 (51) 16 (64) 
    KRAS (15) (6) (13) (8) 
    TP53 18 (44) 19 (61) 21 (45) 16 (64) 
Genetic alterationNo. of cases (%)
Pathologic stage
Prognosis*
IA
>IA
Alive
Dead
(n = 41)(n = 31)(n = 47)(n = 25)
Allelic imbalance         
    1p21-p22 (17) (19) (17) (20) 
    1p13-p21 (17) (19) (17) (20) 
    1q21 (20) (16) (17) (20) 
    2p22-p24 (15) (16) (17) (12) 
    2q35-q37 (15) (19) (15) (20) 
    3p14 10 (24) 14 (45) 17 (36) (28) 
    3q11 12 (29) 10 (32) 18 (38) (20) 
    4p12-p13 (12) (19) (15) (16) 
    4q13 10 (24) (23) 11 (23) (24) 
    4q22 11 (27) (19) 11 (23) (24) 
    5p15 (10) (19) (9) (24) 
    5p12-p13 (12) (16) (13) (16) 
    5q23 12 (29) 11 (35) 14 (30) (36) 
    6p11 (7) 11 (35) (15) (28) 
    6q22 12 (29) 14 (45) 15 (32) 11 (44) 
    7p21-p22 (2) (16) (9) (8) 
    7p15 (2) (16) (9) (8) 
    7q11 (7) (16) (13) (8) 
    7q31-q32 (7) (16) (15) (4) 
    7q33-q35 (7) (16) (15) (4) 
    7q36 (10) (13) (13) (8) 
    8p22-p23 12 (29) 19 (61) 17 (36) 14 (56) 
    8p21 12 (29) 19 (61) 16 (34) 15 (60) 
    8q11 11 (27) 16 (52) 15 (32) 12 (48) 
    9p22 16 (39) 17 (55) 19 (40) 14 (56) 
    9q12-q13 14 (34) 14 (45) 16 (34) 12 (48) 
    10p13-p14 (15) (19) (19) (12) 
    10p12-p13 (15) (19) (19) (12) 
    10p12 (12) (23) (19) (12) 
    10p11 (12) (23) (19) (12) 
    10q25-q26 (15) 10 (32) 12 (26) (16) 
    11p12 (10) 12 (39) (15) (36) 
    11p11-p12 (12) 11 (35) (17) (32) 
    11q24-q25 (20) 13 (42) 13 (28) (32) 
    12p12-p13 10 (24) 14 (45) 12 (26) 12 (48) 
    12q11-q12 (22) 12 (39) 10 (21) 11 (44) 
    13q13 23 (56) 19 (61) 26 (55) 16 (64) 
    14q21 (15) (26) 11 (23) (12) 
    15q14-q21 10 (24) 13 (42) 13 (28) 10 (40) 
    16p11 (10) (23) (11) (24) 
    16q22 (15) 12 (39) (19) (36) 
    16q23-q24 (15) 12 (39) (19) (36) 
    17p12-p13 17 (41) 23 (74) 25 (53) 15 (60) 
    17q11 15 (37) 17 (55) 19 (40) 13 (52) 
    18p11 10 (24) 14 (45) 13 (28) 11 (44) 
    18q21 14 (34) 16 (52) 19 (40) 11 (44) 
    19p12-p13 11 (27) (29) 12 (26) (32) 
    19q12 11 (27) (29) 13 (28) (28) 
    20p11 (20) 12 (39) 10 (21) 10 (40) 
    20q11 (20) 11 (35) 10 (21) (36) 
    21q11-q21 (15) (23) (13) (28) 
    22q12 (17) 11 (35) (19) (36) 
Mutation         
    EGFR 23 (56) 17 (55) 24 (51) 16 (64) 
    KRAS (15) (6) (13) (8) 
    TP53 18 (44) 19 (61) 21 (45) 16 (64) 
*

Postoperative 5-y overall survival.

P < 0.05 by Fischer's exact test.

Figure 1.

Prognostic significance of AI at 8p21 in patients with small-sized lung adenocarcinoma. Overall survival of patients with and without AI were compared by Kaplan-Meier curves and the log-rank test (P = 0.036).

Figure 1.

Prognostic significance of AI at 8p21 in patients with small-sized lung adenocarcinoma. Overall survival of patients with and without AI were compared by Kaplan-Meier curves and the log-rank test (P = 0.036).

Close modal

Association of AIs with EGFR/KRAS mutations, gender, and smoking history. It is known that EGFR mutations are frequently detected in adenocarcinomas in female nonsmokers, whereas KRAS mutations are in adenocarcinomas in male smokers (25, 26). Such an association was also detected in this study (Supplementary Table S3). Therefore, it is important to examine whether any AIs occur in association with EGFR/KRAS mutations, gender, or smoking. The 72 tumors were subdivided into three groups according to the status of EGFR/KRAS mutations: 40 (56%) tumors with EGFR mutations, 8 (11%) tumors with KRAS mutations, and 24 (33%) tumors without EGFR/KRAS mutations (Table 1). Among the 52 AI regions, only 2 regions were significantly differentially affected among the three groups. 4q13 and 4q22 were more frequently affected in tumors with EGFR mutations than in those without EGFR/KRAS mutations (Supplementary Table S2). We also evaluated the association of genetic alterations with gender and smoking history. AI at 1p21-p22, 1p13-p21, and 9p22 was detected predominantly in female patients, and AI at 19p12-p13 was in smokers (Supplementary Table S3).

Genetic model for the development of lung adenocarcinoma. Based on the results, a genetic model for the development of lung adenocarcinoma was constructed (Fig. 2). Mutually exclusive EGFR or KRAS mutations were detected in the majority of type A and type B tumors, and the frequency of EGFR/KRAS mutations in type A and type B tumors was similar to that in type C tumors. Therefore, it is likely that type A and type B noninvasive tumors with EGFR or KRAS mutations progress to type C invasive tumors by acquisition of additional genetic alterations. A subset of type A and type B tumors had neither EGFR nor KRAS mutation and, thus, may also progress to type C invasive tumors without acquiring EGFR and KRAS mutations. The frequency of KRAS mutations in type D tumors was similar to those in replacement growth type (types A, B, and C) tumors, whereas the frequency of EGFR mutations in type D tumors was much lower than those in replacement growth type tumors. These results strongly indicate that the majority of type D tumors are generated through a pathway(s) distinct from replacement growth type tumors without EGFR/KRAS mutations (14), and a subset of type D tumors are progressed from type C tumors, in particular, with KRAS mutations.

Figure 2.

A genetic model for progression of small-sized lung adenocarcinoma. EGFR or KRAS mutations, AI of 13q13, and p16 inactivation are genetic alterations for the development of type A and type B tumors, and they progress to type C tumors by acquiring TP53 mutations and AIs of 17p12-p13/18p11. Because KRAS mutations were infrequent in type B tumors, tumors with KRAS mutations could rapidly progress from type A to type C tumors. A subset of type A, B, or C tumors do not have EGFR/KRAS mutations and, thus, are developed through a pathway other than the EGFR/KRAS pathways. Type D, nonreplacement growth type tumor, can either arise de novo or progress from some replacement growth type tumors by acquiring p16 inactivation, TP53 mutations, and AIs of 13q13, 17p12-p13, and 11p11-p12.

Figure 2.

A genetic model for progression of small-sized lung adenocarcinoma. EGFR or KRAS mutations, AI of 13q13, and p16 inactivation are genetic alterations for the development of type A and type B tumors, and they progress to type C tumors by acquiring TP53 mutations and AIs of 17p12-p13/18p11. Because KRAS mutations were infrequent in type B tumors, tumors with KRAS mutations could rapidly progress from type A to type C tumors. A subset of type A, B, or C tumors do not have EGFR/KRAS mutations and, thus, are developed through a pathway other than the EGFR/KRAS pathways. Type D, nonreplacement growth type tumor, can either arise de novo or progress from some replacement growth type tumors by acquiring p16 inactivation, TP53 mutations, and AIs of 13q13, 17p12-p13, and 11p11-p12.

Close modal

AI of 13q13 was observed frequently in both noninvasive and invasive tumors, and with similar frequencies in replacement and nonreplacement growth types and among three groups according to the status of EGFR/KRAS mutations (Supplementary Table S2). Therefore, 13q13 alterations were suggested to contribute to the formation of noninvasive tumors irrespective of EGFR/KRAS mutation. We recently reported that p16 homozygous deletions also occur with similar frequencies (∼25%) in noninvasive and invasive tumors, in replacement and nonreplacement growth types, and among three groups according to the status of EGFR/KRAS mutations (8). The p16 gene is located on chromosome 9p, which was a hotspot/critical region of AI in both noninvasive and invasive tumors. Therefore, p16 inactivation would also be an early genetic event as 13q13 alterations are.

AI of 17p12-p13 as well as TP53 mutations was observed frequently in invasive tumors, particularly in type D tumors, and commonly among three groups for EGFR/KRAS mutations (Table 1; Supplementary Table S2). There was a significant association between TP53 mutations and AI of 17p12-p13 (P = 0.0007) in the tumors. Therefore, the TP53 gene was inactivated by these two-hit alterations in the majority of tumors analyzed. The result indicated that TP53 inactivation contributes to the progression of noninvasive tumors to invasive ones, irrespective of EGFR/KRAS mutation. Because frequencies of AI of 18p11 and 11p11-p12 in type C and type D tumors, respectively, against those in noninvasive tumors were significantly high (P < 0.01; Table 2), AI of 18p11 is likely to be involved in the progression of replacement type growth tumors, whereas AI of 11p11-p12 may contribute to the development of type D tumors.

Confirmation of AI at 8p21 by microsatellite analysis. To confirm the presence of AI at 8p21 detected by 10K array analysis, microsatellite analysis was done against two microsatellite markers, D8S1116 and D8S322, mapped in the common region of AI at 8p21 (Supplementary Fig. S1). Among 72 cases subjected to 10K array analysis, DNA was available in 61 cases for the microsatellite analysis. Fifty-one of the 61 cases were informative (heterozygous for either D8S1116 or D8S322), and the frequency of AI by microsatellite analysis (29 of 51, 57%) was slightly higher than that by 10K array analysis (23 of 51, 45%). However, the concordance of the results between microsatellite analysis and 10K array analysis was highly significant (P < 0.0001), and AI at 8p21 determined by microsatellite analysis was also significantly more frequently detected in cases with poor prognosis than cases with good prognosis (10 of 12 versus 19 of 39, P = 0.0475). Therefore, AI detected by 10K array analysis was confirmed by microsatellite analysis. To validate the prognostic significance of AI at 8p21, newly microdissected 40 cases were further subjected to microsatellite analysis. Thirty-eight of them were informative, and, in total of 89 informative cases, AI at 8p21 in cases with poor prognosis (22 of 29, 76%) was significantly higher than that in cases with good prognosis (30 of 60, 50%; P = 0.0234).

We undertook a genome-wide comparative analysis of AI as well as mutational analysis of the EGFR/KRAS/TP53 genes between noninvasive and invasive small-sized lung adenocarcinomas. EGFR and KRAS mutations were detected at similar frequencies between noninvasive and invasive tumors, whereas the frequency of TP53 mutations in invasive tumors was significantly higher than that in noninvasive ones (Table 1). The number of AI regions in invasive tumors was significantly higher than that in noninvasive ones. There were 12 regions whose frequencies of AI were significantly higher in invasive tumors than in noninvasive ones, whereas there were no regions whose frequencies of AI were significantly higher vice versa (Table 2). The results indicated that noninvasive tumors progress to invasive ones by further inactivation of several tumor suppressor genes. A genetic model for the progression of small-sized lung adenocarcinoma constructed based on the results of this study went along well with the progression model deduced from the histopathologic findings (15); type A tumors progress sequentially through type B to type C tumors, whereas the majority of type D tumors are generated through a pathway distinct from type C tumors. EGFR or KRAS mutations and AI at 13q13, in addition to p16 inactivation, contribute to the development of type A and type B noninvasive tumors, whereas p53 mutation and AIs of 17p12-p13 and 18p11 contribute to their progression to type C invasive tumors. On the other hand, most type D tumors do not have EGFR or KRAS mutations and, therefore, are generated through a pathway distinct from that of type C tumors. A small subset of type D tumors carried EGFR or KRAS mutations and, therefore, could be developed from type A and type B noninvasive tumors, but these tumors rapidly progressed to invasive ones.

Importantly, AI at 13q13 was frequent not only in replacement growth types but also in nonreplacement growth types. Therefore, it is a common event in adenocarcinoma irrespective of growth types. AI at 13q has been defined as a frequent genetic alteration in non–small-cell lung cancer (9, 10). RB1 is a tumor suppressor gene located at 13q14 that is frequently inactivated in small-cell lung carcinoma. However, RB1 inactivation is infrequent in non–small-cell lung cancer (27, 28) and, therefore, RB1 is unlikely to be a target for AI in adenocarcinoma. Consistently, the 13q13 region defined as a common region of AI in this study did not include the RB1 gene. Thus, a gene(s) other than RB1 is likely to function as a tumor suppressor.

17p12-p13 is the most common region of AIs in invasive tumors, in particular in type D tumors. It was shown in our present and previous studies that the TP53 gene is preferentially mutated in invasive tumors among small-sized adenocarcinomas (6, 8), and the TP53 gene was mapped in the common region of AIs on 17p in this study. Therefore, TP53 inactivation due to a mutation of one allele and a loss of the other allele is likely to contribute to the progression of noninvasive tumors to invasive ones. The frequency of TP53 mutations in invasive tumors, particularly in type D (82%), was extremely high in this study. A major reason of such a high frequency would be the use of microdissected materials for the analysis to avoid overlooking of mutations. In addition, type C cases with poor prognosis and poorly differentiated type D cases were intentionally selected as invasive tumors. Therefore, a majority of the cases were males, smokers, and in advanced stages (>IA). TP53 mutations were more frequently detected in these subsets, although the difference in each subset was not statistically significant (Table 3; Supplementary Table S3). The results strongly indicate that TP53 mutations are predominantly accumulated in advanced stage invasive tumors of male smokers.

The 11p12-p13 region possibly contains a tumor suppressor gene involved in the formation of type D tumors. We recently reported that AI at 11p11-p12 is frequently accumulated in brain metastases of lung cancer (21); therefore, this region might contain a gene(s) involved in the invasion and metastasis of adenocarcinoma cells. The 18p11 region possibly contains a tumor suppressor gene involved in the formation of type C tumors. In addition, AIs at 4q13 and 4q22 were significantly more frequent in EGFR type tumors than in non-EGFR/KRAS type tumors. Therefore, it is likely that these AIs are also involved in the development and/or progression of adenocarcinoma in genetic pathway– and progression stage–specific manners.

Patients with AI at 8p21 in tumors showed worse prognoses than those without. Therefore, AI at 8p21 would also be involved in invasion and metastasis of adenocarcinoma cells and be a useful marker for the prognosis of patients with small-sized lung adenocarcinoma. It was reported that AI at 8p is present in premalignant epithelium of the lung as well as in adenocarcinomas (29, 30), and these findings were confirmed in this study. 8p21 was a hotspot of AIs both in noninvasive and invasive tumors; therefore, AI at 8p21 might enhance metastatic potentials of tumor cells at any developmental stages. AI on 8p was reported to occur frequently in several types of cancers, including hepatocellular carcinoma, prostate cancer, and breast cancer, and its correlation with poor prognosis was also observed in these cancers (3134). Thus, AI on 8p could be useful for the prediction of prognosis of patients with small-sized lung adenocarcinoma as well as several other cancers. Although a target gene(s) on 8p is unknown at present, 37 genes, including TNFRSF10B/TRAIL-R2 and LZTS1/FEZ1 candidate tumor suppressors (3537), were mapped in the 8p21 AI regions (Supplementary Table S1). Therefore, mutational and expression analyses of these genes are in progress.

AIs associated with other clinicopathologic factors were also identified at several chromosomal regions. There were nine regions more frequently affected in stage >IA tumors than in stage IA tumors (P < 0.05; Table 3). Therefore, genes in these regions might also contribute to invasion and/or metastasis of adenocarcinoma cells. AI at 19p12-p13 was detected more frequently in smokers than in nonsmokers (Supplementary Table S3). This result was consistent with our recent report that the LKB1 gene at 19p13 is inactivated preferentially in lung cancers of smokers (38) and, therefore, indicates that LKB1 is a target tumor suppressor for 19p AI. In addition, AIs of 1p21-p22, 1p13-p21, and 9p22 were detected predominantly in female patients; therefore, these regions might contain tumor suppressors whose inactivation preferentially contributes to the development of adenocarcinomas in females that are different from those in males in several aspects (25, 26). Further molecular analyses with more cases of small-sized adenocarcinoma will be necessary to validate the roles of these genetic alterations. Identification of target genes will also facilitate the understanding of multistep carcinogenic pathways of lung adenocarcinoma.

No potential conflicts of interest were disclosed.

Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).

H. Nakanishi and S. Matsumoto contributed equally to this work.

Grant support: Grants-in-Aid from the Ministry of Health, Labor and Welfare for the 3rd-Term Comprehensive 10-Year Strategy for Cancer Control and for Cancer Research (16-1) and from the Program for Promotion of Fundamental Studies in Health Sciences of the National Institute of Biomedical Innovation (NiBio).

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.

1
Travis WD, Brambilla E, Muller-Hermelink HK, Harris CC. World Health Organization classification of tumors: pathology and genetics, tumours of lung, pleura, thymus and heart. Lyon: IARC Press; 2004.
2
Adjei AA. Blocking oncogenic Ras signaling for cancer therapy.
J Natl Cancer Inst
2001
;
93
:
1062
–74.
3
Lynch TJ, Bell DW, Sordella R, et al. Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small-cell lung cancer to gefitinib.
N Engl J Med
2004
;
350
:
2129
–39.
4
Paez JG, Janne PA, Lee JC, et al. EGFR mutations in lung cancer: correlation with clinical response to gefitinib therapy.
Science
2004
;
304
:
1497
–500.
5
Shigematsu H, Lin L, Takahashi T, et al. Clinical and biological features associated with epidermal growth factor receptor gene mutations in lung cancers.
J Natl Cancer Inst
2005
;
97
:
339
–46.
6
Matsumoto S, Iwakawa R, Kohno T, et al. Frequent EGFR mutations in noninvasive bronchioloalveolar carcinoma.
Int J Cancer
2006
;
118
:
2498
–504.
7
Sakamoto H, Shimizu J, Horio Y, et al. Disproportionate representation of KRAS gene mutation in atypical adenomatous hyperplasia, but even distribution of EGFR gene mutation from preinvasive to invasive adenocarcinomas.
J Pathol
2007
;
212
:
287
–94.
8
Iwakawa R, Kohno T, Anami Y, et al. Association of p16 homozygous deletions with clinicopathological characteristics and EGFR/KRAS/p53 mutations in lung adenocarcinoma.
Clin Cancer Res
2008
;
14
:
3746
–53.
9
Shiseki M, Kohno T, Adachi J, et al. Comparative allelotype of early and advanced stage non-small cell lung carcinomas.
Genes Chromosomes Cancer
1996
;
17
:
71
–7.
10
Shiseki M, Kohno T, Nishikawa R, Sameshima Y, Mizoguchi H, Yokota J. Frequent allelic losses on chromosomes 2q, 18q, and 22q in advanced non-small cell lung carcinoma.
Cancer Res
1994
;
54
:
5643
–8.
11
Tsuchiya E, Nakamura Y, Weng SY, et al. Allelotype of non-small cell lung carcinoma—comparison between loss of heterozygosity in squamous cell carcinoma and adenocarcinoma.
Cancer Res
1992
;
52
:
2478
–81.
12
Zhao X, Li C, Paez JG, et al. An integrated view of copy number and allelic alterations in the cancer genome using single nucleotide polymorphism arrays.
Cancer Res
2004
;
64
:
3060
–71.
13
Virmani AK, Fong KM, Kodagoda D, et al. Allelotyping demonstrates common and distinct patterns of chromosomal loss in human lung cancer types.
Genes Chromosomes Cancer
1998
;
21
:
308
–19.
14
Girard L, Zochbauer-Muller S, Virmani AK, Gazdar AF, Minna JD. Genome-wide allelotyping of lung cancer identifies new regions of allelic loss, differences between small cell lung cancer and non-small cell lung cancer, and loci clustering.
Cancer Res
2000
;
60
:
4894
–906.
15
Noguchi M, Morikawa A, Kawasaki M, et al. Small adenocarcinoma of the lung. Histologic characteristics and prognosis.
Cancer
1995
;
75
:
2844
–52.
16
Henschke CI, Yankelevitz DF, Libby DM, Pasmantier MW, Smith JP, Miettinen OS. Survival of patients with stage I lung cancer detected on CT screening.
N Engl J Med
2006
;
355
:
1763
–71.
17
Travis WD, Garg K, Franklin WA, et al. Bronchioloalveolar carcinoma and lung adenocarcinoma: the clinical importance and research relevance of the 2004 World Health Organization pathologic criteria.
J Thorac Oncol
2006
;
1
:
S13
–9.
18
Goldstraw P, Crowley J, Chansky K, et al. The IASLC Lung Cancer Staging Project: proposals for the revision of the TNM stage groupings in the forthcoming (seventh) edition of the TNM Classification of malignant tumours.
J Thorac Oncol
2007
;
2
:
706
–14.
19
Aoyagi Y, Yokose T, Minami Y, et al. Accumulation of losses of heterozygosity and multistep carcinogenesis in pulmonary adenocarcinoma.
Cancer Res
2001
;
61
:
7950
–4.
20
Soblin LHW, editor. TNM classification of malignant tumors. 6th ed. New York: Wiley-Liss; 2002. p. 99–103.
21
Takahashi K, Kohno T, Matsumoto S, et al. Clonal and parallel evolution of primary lung cancers and their metastases revealed by molecular dissection of cancer cells.
Clin Cancer Res
2007
;
13
:
111
–20.
22
Ogiwara H, Kohno T, Nakanishi H, Nagayama K, Sato M, Yokota J. Unbalanced translocation, a major chromosome alteration causing loss of heterozygosity in human lung cancer.
Oncogene
2008
;
27
:
4788
–97.
23
Kosaka T, Yatabe Y, Endoh H, Kuwano H, Takahashi T, Mitsudomi T. Mutations of the epidermal growth factor receptor gene in lung cancer: biological and clinical implications.
Cancer Res
2004
;
64
:
8919
–23.
24
Eberhard DA, Johnson BE, Amler LC, et al. Mutations in the epidermal growth factor receptor and in KRAS are predictive and prognostic indicators in patients with non-small-cell lung cancer treated with chemotherapy alone and in combination with erlotinib.
J Clin Oncol
2005
;
23
:
5900
–9.
25
Sun S, Schiller JH, Gazdar AF. Lung cancer in never smokers—a different disease.
Nat Rev Cancer
2007
;
7
:
778
–90.
26
Toyooka S, Tokumo M, Shigematsu H, et al. Mutational and epigenetic evidence for independent pathways for lung adenocarcinomas arising in smokers and never smokers.
Cancer Res
2006
;
66
:
1371
–5.
27
Yokota J, Kohno T. Molecular footprints of human lung cancer progression.
Cancer Sci
2004
;
95
:
197
–204.
28
Tamura K, Zhang X, Murakami Y, et al. Deletion of three distinct regions on chromosome 13q in human non-small-cell lung cancer.
Int J Cancer
1997
;
74
:
45
–9.
29
Wistuba II, Behrens C, Virmani AK, et al. Allelic losses at chromosome 8p21-23 are early and frequent events in the pathogenesis of lung cancer.
Cancer Res
1999
;
59
:
1973
–9.
30
Emi M, Fujiwara Y, Nakajima T, et al. Frequent loss of heterozygosity for loci on chromosome 8p in hepatocellular carcinoma, colorectal cancer, and lung cancer.
Cancer Res
1992
;
52
:
5368
–72.
31
Emi M, Fujiwara Y, Ohata H, et al. Allelic loss at chromosome band 8p21.3-p22 is associated with progression of hepatocellular carcinoma.
Genes Chromosomes Cancer
1993
;
7
:
152
–7.
32
Jenkins R, Takahashi S, DeLacey K, Bergstralh E, Lieber M. Prognostic significance of allelic imbalance of chromosome arms 7q, 8p, 16q, and 18q in stage T3N0M0 prostate cancer.
Genes Chromosomes Cancer
1998
;
21
:
131
–43.
33
Sigbjornsdottir BI, Ragnarsson G, Agnarsson BA, et al. Chromosome 8p alterations in sporadic and BRCA2 999del5 linked breast cancer.
J Med Genet
2000
;
37
:
342
–7.
34
Zhou W, Goodman M, Lyles RH, et al. Surgical margin and Gleason score as predictors of postoperative recurrence in prostate cancer with or without chromosome 8p allelic imbalance.
Prostate
2004
;
61
:
81
–91.
35
Lee SH, Shin MS, Kim HS, et al. Alterations of the DR5/TRAIL receptor 2 gene in non-small cell lung cancers.
Cancer Res
1999
;
59
:
5683
–6.
36
Toyooka S, Fukuyama Y, Wistuba II, Tockman MS, Minna JD, Gazdar AF. Differential expression of FEZ1/LZTS1 gene in lung cancers and their cell cultures.
Clin Cancer Res
2002
;
8
:
2292
–7.
37
Nonaka D, Fabbri A, Roz L, et al. Reduced FEZ1/LZTS1 expression and outcome prediction in lung cancer.
Cancer Res
2005
;
65
:
1207
–12.
38
Matsumoto S, Iwakawa R, Takahashi K, et al. Prevalence and specificity of LKB1 genetic alterations in lung cancers.
Oncogene
2007
;
26
:
5911
–8.

Supplementary data