Int. J. Cancer: 117, 241–247 (2005) ' 2005 Wiley-Liss, Inc.
Genomewide loss of heterozygosity and its clinical associations in non small cell lung cancer Ruo-Chia Tseng1, Jer-Wei Chang1, Feng-Jen Hsien1, Ya-Hui Chang2, Chin-Fu Hsiao3, Jung-Ta Chen4, Chih-Yi Chen5, Yuh-Shan Jou2* and Yi-Ching Wang1* 1 Department of Life Sciences, National Taiwan Normal University, Taipei, Taiwan 2 Division of Molecular and Genomic Medicine, National Health Research Institutes, Taipei, Taiwan 3 Division of Biostatistics and Bioinformatics, National Health Research Institutes, Taipei, Taiwan 4 Department of Pathology, Veterans General Hospital, Taichung, Taiwan 5 Department of Thoracic Surgery, Veterans General Hospital, Taichung, Taiwan We extensively allelotyped a panel of 71 microdissected primary surgically resected non small cell lung cancer (NSCLC) tumors to identify chromosomal regions that are likely to contain tumor suppressor genes (TSGs) or associated with clinicopathologic and prognostic effects. Loss of heterozygosity (LOH) was detected by genotyping of 177 microsatellite markers and correlation of LOH with clinicopathologic parameters and prognosis was analyzed. Twenty markers showed an LOH frequency greater than 48%, and 8 of them (2p23.3, 2p24.3, 2q35, 6p22.2, 7p14.3, 7p22.2, 17q24.3 and 21q22.3) were novel in NSCLC. The high LOH regions were confirmed by further aligning continuous LOH regions from another set of 24 NSCLC tissues and defining 7 minimal deletion regions ranging from 1.29 to 12.26 cM. The aberrations of 8 markers showed a significant correlation with alteration of p16 and Rb proteins, suggesting the gene(s) located in the chromosomal loss that may interact with p16/Rb pathway. In addition, markers specifically associated with smoking, histology types and tumor stages were identified and the linked candidate TSGs were suggested. For example, marker D1S1612 closely linked with Mig-6 gene was associated with smoking patients, squamous cell carcinoma patients and late-stage patients. Furthermore, 3 markers, D2S2968, D6S2439 and D7S1818, were significantly associated with poor prognosis of NSCLC patients using both univariate and multivariate Cox’s regression analyses (p 5 0.035, 0.022 and 0.006, respectively). These markers can potentially be used for early lung cancer detection, outcome measurement and the positional cloning of new TSGs whose loss of function contributes to NSCLC tumorigenesis. ' 2005 Wiley-Liss, Inc. Key words: non small cell lung cancer; loss of heterozygosity; tumor suppressor gene; prognosis
Lung cancer is one of the most common malignancies in the world and is the leading cause of cancer deaths in industrialized countries.1 A number of factors, such as tumor staging based on the results of physical and surgical examinations, are known to influence prognosis.2 However, patients with the same stage often show varied prognoses. Furthermore, lung cancer classification based primarily on morphologic appearance of the tumor has limitations, because cancer is characterized by genotypic and phenotypic changes that result in a tremendous variability in clinical behavior. Therefore, genotypic detection methods could be used to identify patients at risk for disease metastasis and potentially improve survival of lung cancer. In addition, identification and characterization of the genetic changes that drive lung cancer development can shed light on the molecular mechanism involved in lung tumorigenesis. Human cancers develop through multistep processes involving mutations in several types of genes, including the activation of dominant oncogenes and the inactivating of recessive tumor suppressor genes (TSGs) and genes involved in DNA repair or replication.3 One allele of TSGs may be inactivated by point mutation, methylation changes, or small deletions. The other allele is frequently inactivated by a large deletion involving the gene of interest as well as adjacent stretches of DNA.4 Therefore, genomewide loss of heterozygosity (LOH) allelotyping, which assays the frequency and extent of lost regions on all chromosomal arms, Publication of the International Union Against Cancer
could help to identify or confirm the chromosomal regions associated with tumorigenesis. For example, Shiseki et al.5 investigated LOH using 44 microsatellite markers in 23 stage I and 22 metastatic non small cell lung cancer (NSCLC) patients. Their data indicated that LOH on chromosome arms 3p, 13q and 17p was detected frequently in both stage I lung tumors and brain metastases, whereas the incidence of LOH on chromosome arms 2q, 5q, 9p, 12q, 18q and 22q was higher in brain metastases. SanchezCespedes et al.6 studied LOH in 27 smoking and 18 nonsmoking lung adenocarcinoma (AD) patients using 54 microsatellite markers. They demonstrated that LOH occurs more frequently in smoking lung AD than did in nonsmoking patients.6 In addition, Field et al.7 analyzed LOH in 45 NSCLC tumors using 92 microsatellite markers, and Virmani et al.8 investigated LOH in 29 lung cancer cell lines using 85 markers. Previously, Girard et al.9 conducted a genomewide search of LOH in 36 lung cancer cell lines (14 SCLC and 22 NSCLC). They found that 22 regions, located in 3p, 4q, 8p, 9q, 13q, 17p, Xp and Xq, showed very frequent LOH. In addition, SCLC and NSCLC had different regions of frequent LOH.9 The last study provides valuable information that can be used for the position cloning of new TSGs. However, the correlation of LOH with the clinical characteristics and prognoses of the lung cancer patients was not examined. We have established and extensively allelotyped a panel of 71 microdissected primary surgically resected NSCLC tumors and their matched control DNAs using a high-resolution genomewide LOH search of 177 microsatellite markers. The clinicopathologic and prognostic data of patients were also collected to determine the association of LOH with the clinical and biologic behavior of cancer patients. Some identified LOH markers had been reported previously in other lung cancer studies, validating our results. In addition, 7 minimal deletion regions (MDRs) that were defined as regions with a high frequency of LOH in consecutively locating markers confirmed their importance in lung cancer. Importantly, we also identified novel deletion regions in NSCLC and regions specifically associated with clinicopathologic parameters and prognoses. The data generated in this study provided new markers that may be used for early lung cancer detection, monitoring of tumor progression and the positional cloning of new TSGs. Grant sponsor: The National Science Council, Executive Yuan, the Republic of China; Grant number: NSC91-3112-P-400-004, NHRI92A1NSCLC06-5, NHRI93A1-NSCLC06-5. *Correspondence to: Department of Life Sciences, National Taiwan Normal University, No. 88, Section 4, Tingchou Road, Taipei 116, Taiwan. Fax: 1886-2-29312904. E-mail:
[email protected]; Institute of Biomedical Sciences, Academia Sinica, No. 128, Yen-Chin-Yuan Rd., Soc. 2, Taipei 115, Taiwan. Fax: 886-2-27858594. E-mail:
[email protected] Received 28 December 2004; Accepted after revision 11 March 2005 DOI 10.1002/ijc.21178 Published online 17 May 2005 in Wiley InterScience (www.interscience. wiley.com).
242
TSENG ET AL. TABLE I – CLINICOPATHOLOGIC DATA OF 71 NSCLC PATIENTS Characteristic
Age1 < 65 65 Sex Female Male Tumor type AD SQ Tumor stage Early (I, II) Late (III, IV) Smoker Yes No Follow-up2 Survived Died
Patient numbers
25 46 12 59 37 34 41 30 44 27 24 47
1 The median age is 68 years old (range, 41–86 years old).–2The median follow-up period was 26 months (range, 0.5–121 months).
Material and methods Tumor specimens and clinical characteristics of patients The study group consisted of 71 patients diagnosed with primary NSCLC, who were admitted to Veterans General Hospital, Taichung, Taiwan, between 1993 and 2000. Table I includes the complete clinicopathologic and follow-up data of 71 patients. These patients were selected because both of their paraffin-embedded materials and resected normal lung tissues were available. The histologies of the tumor types and stages were determined according to the WHO classification method and the TNM system, respectively. The duration and amount of smoking prior to the diagnosis of lung cancer in all patients were obtained from hospital records. The patients were classified into smoking and nonsmoking groups; the former included both current smokers and ex-smokers. Follow-up of the 71 studied patients was performed at 2-month intervals in the first year after surgery, and at 3-month intervals thereafter, at outpatient clinics or by routine phone calls. The end of the follow-up period was defined as November 2004 for all 71 patients. For the 24 patients who survived the follow-up period (censored patients), the median follow-up time was 93 months (range, 59–121 months). For the 47 patients who died during the follow-up period, the median follow-up time was 13 months (range, 0.5–66 months). The test group for the MDR assay consisted of 24 patients admitted to Veterans General Hospital, Taichung, Taiwan, between 2002 and 2003. Microdissection and DNA extraction Serial 5 lm sections were cut from formalin-fixed, paraffinembedded tumor tissues. The tumor cells were microdissected from up to 3 sections. Sections were examined microscopically at 403 magnifications, and neoplastic cells were microdissected manually with a 20-gauge needle. Following de-waxing in xylene, the genomic DNA of the recovered tumor cells was prepared using proteinase K digestion and phenol/chloroform extraction followed by ethanol precipitation. Normal lung tissues at the resection margin were immediately snap-frozen and subsequently stored in liquid nitrogen and extracted for the genomic DNA as described above. Microsatellite polymorphic markers and PCR-LOH analysis A total of 177 informative microsatellite polymorphic markers spanning the 39 nonacrocentric autosomal arms, with an average interval of 20 cM, were used for genomewide allelotyping. In addition, 18 more microsatellite markers locating at the 7 high LOH regions were used for MDR analysis. These 7 high LOH regions studying for MDR were selected based on their being representative marker of mean plus 1 standard deviation (SD),
plus 2 SD, minus 0.5 SD, or minus 1 SD. All the fluorescentlabeled dinucleotide markers were based on the Marshfield Map and purchased from PE Applied Biosystems (Foster City, CA). The cytogenetic locations of the microsatellite markers were based on the National Center of Biotechnology Information (http:// www.ncbi.nlm.nih.gov/genome/guide/human/). Five ng of genomic DNA from adjacent noncancerous lung tissues, microdissected tumor cells, or sputum samples were used for each PCR analysis. Multiplex-PCR reactions were conducted for markers with different amplification sizes. PCR products with different fluorescent labels and fragment sizes were pooled and mixed with internal fluorescent molecular weight markers for subsequent electrophoresis in ABI 377 automated fluorescent DNA sequencer (PE Applied Biosystems). The allelic ratio was calculated as (T1/T2)/(N1/N2) for the ratio of area values of the tumor (T) versus the normal (N) alleles. The LOH was defined as the allelic ratio above 2 or below 0.5. Therefore, LOH was defined as a reduction of 50% or more in the peak intensity of one of the tumor sample alleles when compared with a heterozygous normal tissue control. Statistical analysis To examine the association between LOH and the clinical characteristics, we only included markers and patients with informative frequencies higher than the median in all studied markers and patients for statistical analysis. The survival probabilities were estimated using the Kaplan-Meier method. The Cox proportional hazards model was used to assess the impact of LOH on mortality without or with adjustment for age, sex, type, stage and smoking. We used the 2-tailed Fisher’s exact test to examine the null hypothesis of no association between LOH with the classification of clinical characteristics at those specified loci. The correlation between markers within the same MDR region was assessed using both Pearson correlation coefficient and covariance. A p < 0.05 was considered to be statistically significant. Cluster analysis was also applied to evaluate potential interactions of LOH and the histologic stages of patients. We established a cross-tabulation showing the number of cases in each classification with or without LOH. All of the 12,576 segments of data (177 markers in 71 NSCLC cases) were each assigned a number and a color code as follows: the existence of LOH in a marker of NSCLC was assigned the number 21 and red color; the retention of a chromosomal region was assigned the number 11 and green color; and the noninformative data were assigned black color; no data were assigned gray color. The color-coding system provides interactive graphic analysis of clustering results in the TreeView program.10 Based on a previously described clustering algorithm,9,10 the similarity of LOH profiles of a group of tumor samples (such as a subgroup of NSCLC features) was calculated. Results Frequent LOH regions in NSCLC The percentage of markers displaying LOH was calculated. The mean heterozygosity rate in 177 microsatellite markers was 71%. The mean LOH frequency was 35%. The percentage of LOH per informative case ranged from 8% (marker D6S1027) to 64% (marker D2S164). When the numbers of markers were plotted with the frequencies of LOH, we observed a biphasic distribution in which 2 peaks were separated at a frequency of 48% (data not shown). According to previous LOH studies,9,11 markers appearing in the peak of low LOH frequencies detect mostly random changes reflecting the genetic instability of tumors, whereas markers in another peak are likely to show LOHs associated with cancer-specific phenotypes. Therefore, a significant percentage of LOH was chosen to be a value above 48% of LOH. A high percentage of LOH (> 48%) was observed at 20 loci. Their LOH frequency and chromosomal location are summarized in Figure 1. Note that chromosome regions at 2p23.3, 2p24.3, 2q35, 6p22.2, 7p14.3, 7p22.2, 17q24.3 and 21q22.3 were novel frequent LOH regions in NSCLC.
243
GENOMEWIDE LOH ANALYSIS IN NSCLC
FIGURE 1 – Chromosomal location of 20 markers (squared) showing the high percentage of LOH (> 48%) along with the markers analyzed at the respective chromosome. The frequency of LOH is noted on the side of the marker and the novel sites of frequent LOH are indicated by an asterisk. TABLE II – REFINED MAP OF MDR REGIONS IN LUNG CANCER PATIENTS Loci
2p23.3~p24.3 2q35 3p21.32~p22 4q34.1 6p21.1~p22.2 17p13.1~p13.2
21q22.3
Marker
Correlation marker2
LOH 1
D2S1360 D2S4051 D2S2382 D2S1641 D2S4341 D3S3567 D3S17681 D3S3685 D4S1595 D4S24311 D6S1660 D6S24391 D17S1537 D17S1791 D17S9741 D17S1303 D21S1897 D21S14461
53% 51% 63% 50% 44% 58% 53% 50% 58% 50% 43% 82% 56% 50% 47% 50% 58% 69%
Candidate genes3
CGREF1, XAB1, ALK, RHOB30 XRCC531
D3S36850.009 D3S17680.009
PDCD61P, hMLH132 FBXO8, SAP30 TTRAP
D17S17910.004, D17S13030.004 D17S15370.004, D17S9740.001 D17S17910.001, D17S13030.036 D17S15370.004, D17S9740.036 D21S14460.062 D21S18970.062
GAS7
DNMT3L
1 High LOH marker in the genomewide study.–2One marker is linked to other marker(s) indicated in the same MDR region with a correlation coefficient greater than 0.69. The p-values with significance were shown as superscripts.–3Candidate growth control genes locating between 0.24 and 2.4 Mb to the markers are shown. Genes in bold represent those TSGs that were previously shown to be abnormal in lung cancer.
Verification of high LOH marker by defining minimal deletion regions in another set of NSCLC samples To verify the high LOH markers and to determine the maximal contiguous LOH loci in the chromosomes of each patient, we assigned MDRs based on overlapping contiguous LOH loci in another set of 24 NSCLC tissues using 18 more markers, which covers the selected 7 regions, including the highest and lowest markers among the 20 high LOH markers. After aligning LOH data of all cancer tissues on each chromosome, 7 MDRs were defined. The genetic interval of these MDRs ranged from 1.29 to 12.26 cM (Table II). To determine whether one marker is linked to other markers within the same MDR, i.e., to identify the markers with high frequency of codeletion in the same patients, we employed Pearson correlation analysis of MDR profiles of the 24 NSCLC tissues. The markers locating within 3 MDRs at chromosomes 3p21-22, 17p13.1-13.2 and 21q22-22.3 showed a concordance score greater than 0.69 (Table II), suggesting that markers were frequently codeleted in patients, and that candidate TSGs may be located in these MDRs.
Correlation of LOH with alteration in TSGs These 71 patients had been analyzed for the protein expression of p16INK4a and Rb in our laboratory previously.12,13 Therefore, the LOH of the 177 microsatellite markers were tested for their association with the alterations in p16INK4a and RB to reveal the existence of correlation of LOH with p16/Rb pathway (Table III). The correlation of loss of expression of the p16 and RB with their intragenic markers, D9S942 and RB, respectively, indicated their being involved in NSCLC tumorigenesis through mechanisms of allelic deletion. Note that some markers located at different chromosomes from the loci of p16INK4a and RB were significantly deleted in the patients without p16 and/or RB expression, suggesting that genetic imbalance of these regions may be associated with loss of p16 and RB expression. Correlation of LOH with clinicopathologic parameters Frequency of LOH by sexes, smoking habits and tumor types. To examine whether there were associations between LOH and the clinical characteristics of the patients, the occurrence of LOH at
244
TSENG ET AL. TABLE III – ASSOCIATION BETWEEN LOH AND PROTEIN EXPRESSION OF P16 AND RB TSGS
p16
RB
Candidate targeted protein in the marker region
Associated clinical parameter
Site
Categories
No LOH
LOH
p-value
D2S405
2p23.3
0.031
Poor survival
D2S2968
2q37.3
0.049
Poor survival
D4S3335
4q35.1
D10S2481
10p12.1
D10S1423
1012.31
RB
13q14.11
D18S858
18q21.31
15 (65%) 9 (35%) 11 (44%) 4 (16%) 6 (55%) 4 (20%) 9 (53%) 5 (21%) 12 (67%) 4 (19%) 9 (64%) 4 (24%) 18 (95%) 12 (67%) 8 (53%) 1 (10%)
Early stage
9p21
8 (35%) 17 (65%) 14 (56%) 21 (84%) 5 (45%) 16 (80%) 8 (47%) 19 (79%) 6 (33%) 17 (81%) 5 (36%) 13 (76%) 1 (5%) 6 (33%) 7 (47%) 9 (90%)
0.032
D9S942 (p16)
p16(2) p16(1) p16(2) p16(1) RB(2) RB(1) RB(2) RB(1) RB(2) RB(1) RB(2) RB(1) RB(2) RB(1) RB(2) RB(1)
Variable
Markers
0.033
CASP3
0.003 0.022
VIM
0.029 0.027
BCL2
TABLE IV – ASSOCIATION BETWEEN LOH AND PATIENTS’ CLINICOPATHOLOGIC PARAMETERS Variable
Smoke
Type
Markers
Site
Marker associated with smoker patients D1S1612 1p36.23 D2S1788
2p22.3
D3S1768
3p22.2
D9S925
9p21.3
GATA121A08
10q22.1
D10S2327
10q22.3
D13S796
13q33.3
D17S938
17p13.3
D21S1446
21q22.3
Marker associated with AD patients D14S1426 14q32.32 D20S186
20p12.1
Marker associated with SQ patients D1S1612 1p36.23 D3S3050
3p26.2
D3S2432
3p22.2-p22.3
D8S1179
8q24.13
D9S925
9p22.1
D13S894
13q12.3
D17S938
17p13.3
D19S221
19p13.2
each marker was compared with the patients’ clinicopathologic parameters, including sex, smoking history and tumor type (Table IV). There were 9 markers specifically associated with patients who smoked (p 5 0.016, 0.035 and < 0.001, respectively, using the Fisher’s exact test). The markers D9S925 and D17S938 were also associated with squamous carcinoma (SQ) patients (p 5 0.042 and 0.013, respectively). Of interest, the markers GATA121A08 and D10S2327, which were closely located in
Categories
No LOH
LOH
p-value
Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No
16 (48%) 16 (80%) 14 (48%) 14 (88%) 8 (32%) 10 (77%) 17 (53%) 16 (84%) 11 (50%) 15 (88%) 10 (43%) 13 (93%) 14 (42%) 16 (73%) 4 (20%) 11 (85%) 10 (37%) 11 (79%)
17 (52%) 4 (20%) 15 (52%) 4 (12%) 17 (68%) 3 (23%) 15 (47%) 3 (16%) 11 (50%) 2 (12%) 13 (57%) 1 (7%) 19 (58%) 6 (27%) 16 (80%) 2 (15%) 17 (63%) 3 (21%)
0.041
AD SQ AD SQ
13 (42%) 16 (70%) 19 (59%) 22 (92%)
18 (58%) 7 (30%) 13 (41%) 2 (8%)
0.057
AD SQ AD SQ AD SQ AD SQ AD SQ AD SQ AD SQ AD SQ
20 (74%) 12 (46%) 14 (82%) 9 (47%) 21 (88%) 14 (61%) 20 (87%) 10 (53%) 19 (79%) 14 (52%) 13 (81%) 9 (38%) 12 (67%) 3 (20%) 16 (94%) 7 (54%)
7 (26%) 14 (54%) 3 (18%) 10 (53%) 3 (12%) 9 (39%) 3 (13%) 9 (47%) 5 (21%) 13 (48%) 3 (19%) 15 (62%) 6 (33%) 12 (80%) 1 (6%) 6 (46%)
0.045 0.016 0.035 0.017 0.004 0.032 0.000 0.020
0.013 0.051 0.041 0.049 0.020 0.042 0.010 0.013 0.025
10q22, were both frequently deleted in smokers (p 5 0.017 and 0.004, respectively). Furthermore, marker D1S1612 at 1p36.23, which associated with SQ and smoking patients, was also specifically associated with late-stage patients (Fig. 2). Of note is that different markers were specifically associated with either AD or SQ lung cancer. For example, markers D14S1426 and D20S186 were specifically related to AD (p 5 0.044 and 0.013, respectively).
245
GENOMEWIDE LOH ANALYSIS IN NSCLC
FIGURE 2 – Cluster analysis of markers associated with tumor stages of NSCLC patients. Selected patients with informative frequency more than median in studied marker (as defined in text) are shown. Tumors are ordered in such a way that markers with similar LOH patterns tend to be grouped (clustered) together in columns, marker data and LOH frequencies are in rows. Color codes: red box, LOH; green box, retention of heterozygosity; black box, noninformative. TABLE V – COX PROPORTIONAL-HAZARDS MODELS FOR ASSOCIATION OF LOH WITH PROGNOSIS OF NSCLC PATIENTS Markers
Hazard ratio
95% confidence interval
Unadjusted analysis D2S2968 2.650 (1.071–6.528) D6S2439 3.956 (1.154–18.117) D7S1818 3.531 (1.060–10.512) D22S445 2.137 (0.884–4.908) Adjusted for age, sex, type, stage, and smoking D1S3721 3.992 (1.225–14.032) D1S423 17.369 (1.465–418.520) D2S2972 2.258 (1.043–4.897) D2S2968 2.550 (1.015–6.358) D3S1234 5.790 (1.039–32.362) D5S421 4.188 (1.424–13.578) D6S474 5.042 (1.401–17.367) D6S2439 4.220 (1.192–19.816) D7S1818 13.961 (1.307–304.015) D7S3070 10.013 (1.604–77.851) RH28094 12.837 (1.599–263.230) D8S1132 5.271 (1.155–26.890) D10S583 3.684 (1.147–12.347) GATA49D12 2.554 (1.301–5.033) D13S894 5.880 (1.352–25.516) D19S586 12.837 (1.599–263.230) D21S2055 8.399 (2.248–40.103) D22S686 11.515 (1.413–237.296) D22S445 8.379 (1.291–164.234)
p-value
Association with other parameter1
0.025 0.027 0.018 0.067
RB High LOH Stage II
0.014 0.032 0.025 0.035 0.018 0.003 0.004 0.022 0.006 0.003 0.024 0.030 0.034 0.049 0.006 0.004 0.002 0.023 0.015
High LOH, stage III RB FHIT High LOH Stage II
Stage II SQ
1
The parameters include tumor stage, type, TSG and high LOH marker.
Frequency of LOH by tumor stages. To identify chromosomal regions that were associated with the steps in NSCLC tumorigenesis, the occurrence of LOH at each marker was compared with the tumor stage of patients. Early steps are defined as genetic alterations that are present in both early stages (I and II) and late stages (III and IV), whereas later steps are detected solely in the later stages. A representative of the data based on a one-dimensional cluster analysis in patients with informative rate higher than the median in studied markers is given in Figure 2. Eight markers, i.e., D2S405, D3S1285, D3S4545, D4S2361, D5S1456, D8S1130, D12S375 and D17S974, showed deletion in more than 45% for patients in both early and late stages (Fig. 2). Six markers, i.e., D1S1612, D4S1625, D7S3047, D8S1477, D18S877 and D19S591, were frequently deleted in late-stage patients but not in early-stage
patients, indicating their association with cancer progression (Fig. 2). Association of LOH with prognosis. The relationship between LOH and postoperative survival was analyzed for all 71 patients using Cox’s univariate and multivariate regression analyses (Table V). The Cox’s proportional hazard analyses indicated that 3 markers were significantly correlated to the prognosis of NSCLC patients and all of them, D2S2968, D6S2439 and D7S1818, were still significantly associated with a poor prognosis after adjusting by the additional covariates of tumor stage, histology, smoking and age at diagnosis (p 5 0.035, 0.022 and 0.006, respectively). Seven markers associated with poor prognosis were also associated with some other parameters such as TSG alterations, early
246
TSENG ET AL.
FIGURE 3 – Kaplan-Meier survival curves with respect to LOH at D2S2968 (a) and GATA49D12 (b). The patients with LOH had worse prognoses than the patients without LOH (p 5 0.035 and 0.049, respectively).
tumor stage and high LOH markers (Table V). Figure 3 is the representative figure for the survival curves of the 2 markers. Discussion The present study describes a comprehensive genomewide allelotyping of 71 microdissected NSCLC tumors using 177 fluorescentlabeled microsatellite markers. Previous LOH allelotyping analyses of lung cancer by other researchers were mostly conducted on particular chromosomal regions, used relatively low densities of markers, or utilized nonmicrodissected tumors or lung cancer cell lines for the study. We identified novel deletion regions in NSCLC and regions specifically associated with clinicopathologic parameters and prognoses. In addition, the present study identified interactions of chromosomal aberrations at markers in different regions and with known TSG alterations. This may facilitate the dissection of genetic pathways in NSCLC tumorigenesis. Twelve high-LOH loci located on chromosome 1p34.1,14 3p14.3,8 3p22.2,8 3p25.1,8 4p15.2,15 4q34.1,16 5q14.3,17 5q35.1,18 9q21.32,19 12q14.3,5,20 17p13.121 and 17p13.26 regions have been confirmed in other investigations in lung cancer, validating our LOH analysis. Note that we reported for the first time 8 highly frequent loss regions on chromosomes 2p23.3, 2p24.3, 2q35, 6p22.2, 7p14.3, 7p22.2, 17q24.3 and 21q22.3. Although deletion in the 2p23-24 region was not reported in the cell line studies,8,9 the high LOH in the 2p23-24 region was confirmed by aligning continuous LOH regions from another set of NSCLC tissues (Table II). Some of the previously reported high LOH regions such as chromosome 19p138,9 were not recognized as a frequent deletion region in our study. This discrepancy may be due to the differences in patients’ characteristics in the contrasting studies. There were more patients who were NSCLC and early-stages lung cancer patients in our study compared to that of Virmani et al.8 and Girard et al.9 In addition, ethnicity may also be an important confounding factor in genotyping studies involving hereditary factors. The present study showed frequent LOH at 3p14.1, 3p22.2, and 3p25.1 regions. Considerable attention has been given to the identification of the 3p genes involved in lung cancer pathogenesis.22–24 Interestingly, the region 3p14.1 identified by marker D3S1285 was also deleted early in NSCLC tumorigenesis (Fig. 2).
We therefore examined the LOH of D3S1285 and the other marker, D5S1456, in cytologically negative sputum samples taken before surgery and their matched NSCLC tumors in another set of 56 NSCLC patients. The data indicated a significant concordance and specificity rates with a feasible sensitivity rate in sputum (data not shown), suggesting their potential use as diagnostic biomarkers and an early involvement of TSGs locating in these loci in lung tumorigenesis. To identify the gene(s) located in the chromosomal loss that may interact with p16/Rb pathway, the aberrations of chromosome regions were analyzed for their correlation with alteration of p16 and Rb proteins. The high significance of these associated markers located at regions other than the p16/Rb loci might represent the existence of genes in these regions involved in the p16/Rb pathway. For example, 3 markers specifically associated with patients with loss of RB expression, and they were located near the genes, i.e., caspase-3, CASP3 (D4S3335), vimentin, VIM (D10S1423) and BCL2 (D18S858). The caspase-3 is the primary executioner caspase in the apoptosis system, necessary for cytochrome c/dATP-inducible cleavage of several substrate proteins, including RB and VIM.25 In addition, RB activates BCL2 gene by binding directly to the respective promoter sequences.26 The synergistic loss of possible different TSGs would provide a growth advantage during NSCLC development. With regard to the markers associated with clinicopathologic parameters such as smoking history, tumor types and the tumor stages of the patients, we identified specific markers using the 2-tailed Fisher’s exact test and cluster analysis. It has been shown previously that there is a significant association between 3p21 LOH and increasing polycyclic aromatic hydrocarbons-DNA adducts levels in SQ of the lung.27 We also identified that D1S1612, D9S925 and D13S796 were specifically associated with smoking NSCLC. The marker D1S1612 is particularly interesting because it was also specifically deleted in SQ (Table IV) and late-stage patients (Fig. 1). It spans the region 1p36.2-36.3 and contains several growth control genes such as Mig-6 (mitogen-inducible gene 6)28 and CTNNBIP1 (catenin, beta interacting protein 1).29 In addition, our results showed that 2 markers (D7S1818 and GATA49D12), which deleted in 50% and 69% of stage II patients, were worse prognostic markers in both univariate and multivariate Cox analyses (Table V). Furthermore, the worse prognostic marker D1S3721 was deleted in 67% of stage III patients, suggesting that D1S3721 may be associated with metastasis of NSCLC. Allelotyping studies with more microsatellite markers in these regions and candidate gene approaches at these nearby putative TSGs warrant further investigation. Seven MDRs were identified in the present study. The markers within the MDR showed high correlation with each other, manifesting the deletion of TSGs locating in these regions involved in lung tumorigenesis. These MDRs allow a more accurate determination of chromosomal losses and should facilitate positional candidate cloning of TSGs of NSCLC. Refined mapping at the regions with high LOH frequency and at the regions with a correlation of clinical parameters and prognoses may reveal sites for putative TSGs. The genetic and biologic characterization of these candidate TSGs in lung tumors should also provide information that will help to give us a better understanding of the molecular mechanism of NSCLC tumorigenesis; they can be tested further in population-based screening and evaluated as possible targets for the detection and diagnosis of NSCLC. Our study also provides evidence to use microsatellite markers as biomarkers for NSCLC prognosis, which will be important for NSCLC patient management.
References 1. 2. 3.
Minna JD, Roth JA, Gazdar AF. Focus on lung cancer. Cancer Cell 2002;1:49–52. Mountain CF. The international system for staging lung cancer. Semin Surg Oncol 2000;18:106–15. Fearon ER. Human cancer syndromes: clues to the origin and nature of cancer. Science 1997;278:1043–50.
4. 5.
Kohno T, Yokota J. How many tumor suppressor genes are involved in human lung carcinogenesis? Carcinogenesis 1999;20:1403–10. Shiseki MT, Kohno J, Adachi J, Okazaki T, Otsuka T, Mizoguchi H, Noguchi M, Hirohashi S, Yokota J. Comparative allelotype of early and advanced stage non-small cell lung carcinomas. Genes Chromosomes Cancer 1996;17:71–7.
GENOMEWIDE LOH ANALYSIS IN NSCLC
6.
7.
8.
9.
10. 11.
12. 13. 14.
15.
16.
17. 18. 19.
20.
Sanchez-Cespedes M, Ahrendt SA, Piantadosi S, Rosell R, Monzo M, Wu L, Westra WH, Yang SC, Jen J, Sidransky D. Chromosomal alterations in lung adenocarcinoma from smokers and nonsmokers. Cancer Res 2001;61:1309–13. Field ,JK, Neville EM, Stewart MP, Swift A, Liloglou T, Risk JM, Ross H, Gosney JR, Donnelly RJ. Fractional allele loss data indicate distinct genetic populations in the development of non-small-cell lung cancer. Br J Cancer 1996;74:1968–74. Virmani AK, Fong KM, Kodagoda D, McIntire D, Hung J, Tonk V, Minna JD, Gazdar AF. Allelotyping demonstrates common and distinct patterns of chromosomal loss in human lung cancer types. Genes Chromosomes Cancer 1998;21:308–19. Girard L, Zochbauer-Muller S, Virmani AK, Gazdar AF, Minna JD. Genomewide 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. Eisen MB, Brown PO. DNA arrays for analysis of gene expression. Methods Enzymol 1999;303:179–205. Jou YS, Lee CS, Chang YH, Hsiao CF, Chen CF, Chao CC, Wu LS, Yeh SH, Chen DS, Chen PJ. Clustering of minimal deleted regions reveals distinct genetic pathways of human hepatocellular carcinoma. Cancer Res 2004;64:3030–6. Chen JT, Chen YC, Chen CY, Wang YC. Loss of p16 and/or pRb protein expression in NSCLC: an immunohistochemical and prognostic study. Lung Cancer 2001;31:163–70. Chen JT, Chen YC, Wang YC, Tseng RC, Chen CY, Wang YC. Alterations of the p16(ink4a) gene in resected nonsmall cell lung tumors and exfoliated cells within sputum. Int J Cancer 2002;98:724–31. Gasparian AV, Laktionov KK, Belialova MS, Pirogova NA, Tatosyan AG, Zborovskaya IB. Allelic imbalance and instability of microsatellite loci on chromosome 1p in human non-small-cell lung cancer. Br J Cancer 1998;77:1604–11. Petersen S, Aninat-Meyer M, Schluns K, Gellert K, Dietel M, Petersen I. Chromosomal alterations in the clonal evolution to the metastatic stage of squamous cell carcinomas of the lung. Br J Cancer 2000;82:65–73. Shivapurkar N, Virmani AK, Wistuba II, Milchgrub S, Mackay B, Minna JD, Gazdar AF. Deletions of chromosome 4 at multiple sites are frequent in malignant mesothelioma and small cell lung carcinoma. Clin Cancer Res 1999;5:17–23. Wu X, Zhao Y, Kemp BL, Amos CI, Siciliano MJ, Spitz MR. Chromosome 5 aberrations and genetic predisposition to lung cancer. Int J Cancer 1998;79:490–3. Mendes-da-Silva P, Moreira A, Duro-da-Costa J, Matias D, Monteiro C. Frequent loss of heterozygosity on chromosome 5 in non-small cell lung carcinoma. Mol Pathol 2000;53:184–7. Petersen I, Bujard M, Petersen S, Wolf G., Goeze A, Schwendel A, Langreck H, Gellert K, Reichel M, Just K, du Manoir S, Cremer T, et al. Patterns of chromosomal imbalances in adenocarcinoma and squamous cell carcinoma of the lung. Cancer Res 1997;57:2331–5. Tommasi S, Dammann R, Jin SG, Zhang Xf XF, Avruch J, Pfeifer GP. RASSF3 and NORE1: identification and cloning of two human homologues of the putative tumor suppressor gene RASSF1. Oncogene 2002;21:2713–20.
247
21. Tsuchiya E, Tanigami A, Ishikawa Y, Nishida K, Hayashi M, Tokuchi Y, Hashimoto T, Okumura S, Tsuchiya S, Nakagawa K. Three new regions on chromosome 17p13.3 distal to p53 with possible tumor suppressor gene involvement in lung cancer. Jpn J Cancer Res 2000; 91:589–96. 22. Sozzi G, Veronese ML, Negrini M, Baffa R, Cotticelli MG, Inoue H, Tornielli S, Pilotti S, De Gregorio L, Pastorino U, Pierotti MA, Ohta M, et al. The FHIT gene 3p14.2 is abnormal in lung cancer. Cell 1996;85:17–26. 23. Fong KM, Biesterveld EJ, Virmani A, Wistuba I, Sekido Y, Bader SA, Ahmadian M, Ong ST, Rassool FV, Zimmerman PV, Giaccone G, Gazdar AF, et al. FHIT and FRA3B 3p14.2 allele loss are common in lung cancer and preneoplastic bronchial lesions and are associated with cancer-related FHIT cDNA splicing aberrations. Cancer Res 1997;57:2256–67. 24. Wang YC, Lu YP, Tseng RC, Lin RK, Chang JW, Chen JT, Shih CM, Chen CY. Inactivation of hMLH1 and hMSH2 by promoter methylation in primary non-small cell lung tumors and matched sputum samples. J Clin Invest 2003;111:887–95. 25. Slee EA, Adrain C, Martin SJ. Executioner caspase-3, -6, and -7 perform distinct, non-redundant roles during the demolition phase of apoptosis. J Biol Chem 2001;276:7320–6. 26. Decary S, Decesse JT, Ogryzko V, Reed JC, Naguibneva I, Harel-Bellan A, Cremisi CE. The retinoblastoma protein binds the promoter of the survival gene bcl-2 and regulates its transcription in epithelial cells through transcription factor AP-2. Mol Cell Biol 2002;22: 7877–88. 27. Hirao T, Nelson HH, Ashok TD, Wain JC, Mark EJ, Christiani DC, Wiencke JK, Kelsey KT. Tobacco smoke-induced DNA damage and an early age of smoking initiation induce chromosome loss at 3p21 in lung cancer. Cancer Res 2001;61:612–5. 28. Makkinje A, Quinn DA, Chen A, Cadilla CL, Force T, Bonventre JV, Kyriakis JM. Gene 33/Mig-6, a transcriptionally inducible adapter protein that binds GTP-Cdc42 and activates SAPK/JNK: a potential marker transcript for chronic pathologic conditions, such as diabetic nephropathy—possible role in the response to persistent stress. J Biol Chem 2000;275:17838–47. 29. Daniels DL, Weis WI. ICAT inhibits beta-catenin binding to Tcf/Leffamily transcription factors and the general coactivator p300 using independent structural modules. Mol Cell 2002;10:573–84. 30. Delarue FL, Taylor BS, Sebti SM. Ras and RhoA suppress whereas RhoB enhances cytokine-induced transcription of nitric oxide synthase-2 in human normal liver AKN-1 cells and lung cancer A-549 cells. Oncogene 2001;20:6531–7. 31. Price EA, Bourne SL, Radbourne R, Lawton PA, Lamerdin J, Thompson LH, Arrand JE. Rare microsatellite polymorphisms in the DNA repair genes XRCC1, XRCC3 and XRCC5 associated with cancer in patients of varying radiosensitivity. Somat Cell Mol Genet 1997; 23:237–47. 32. Wang YC, Lu YP, Tseng RC, Lin RK, Chang JW, Chen JT, Shih CM, Chen CY. Inactivation of hMLH1 and hMSH2 by promoter methylation in primary non-small cell lung tumors and matched sputum samples. J Clin Invest 2003;111:887–95.
