The

Oncologist

®

Cancer Diagnostics and Molecular Pathology BAC Clones Related to Prognosis in Patients with Esophageal Squamous Carcinoma: An Array Comparative Genomic Hybridization Study SHIGEO HIRASAKI,a– c TSUYOSHI NOGUCHI,d KOSHI MIMORI,a JUNKO ONUKI,c KEIKO MORITA,c HIROSHI INOUE,a KENICHI SUGIHARA,b MASAKI MORI,a TAKASHI HIRANOc a

Department of Surgery and Molecular Oncology, Medical Institute of Bioregulation, Kyushu University, Beppu, Japan; bDepartment of Surgical Oncology, Graduate School of Medicine, Tokyo Medical and Dental University, Tokyo, Japan; cResearch Institute for Cell Engineering, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Japan; dDepartment of Oncological Science (Surgery II), Faculty of Medicine, Oita University, Yufu, Japan Key Words. Laser microdissection • Genome • Microarray • Clinical samples • Survival curve

ABSTRACT Purpose. The prognosis of patients with esophageal carcinoma is poor. To identify genomic alterations associated with poor patient prognosis, we analyzed whole DNA copy number profiles of esophageal squamous carcinomas (ESCs) using array-based comparative genomic hybridization (aCGH). Materials and Methods. Twenty-one operated and two biopsied cases of esophageal squamous cancer were examined for study. Each sample was laser microdissected to obtain pure cancer cell populations. The extracted DNA was analyzed using aCGH. Results. One of the most representative alterations was a previously reported amplification at 11q13.3. In addition, some novel alterations, such as deletion of 16p13.3, were identified. Of the 19 patients who were reassessed more than 5 years after the operation, nine

were still living and 10 had died from disease recurrence. When aCGH profiles from the surviving group and the deceased group were compared, significant differences were recognized in 68 of 4,030 bacterial artificial chromosome (BAC) clones. Almost half of these clones were present at nine limiting regions in 4q, 13q, 20q, and Xq. For 22 of these 68 BAC clones, there also was a significant difference in the KaplanMeier survival curve, using the log-rank test, when comparing patients who had an alteration in a particular clone with those who did not. Conclusions. aCGH study of esophageal squamous cancer clearly identified BAC clones that are related to the prognosis of patients. These clones give us the opportunity to determine specific genes that are associated with cancer progression. The Oncologist 2007;12:406 – 417

Disclosure of potential conflicts of interest is found at the end of this article.

Correspondence: Masaki Mori, M.D., Ph.D., F.A.C.S., Department of Surgery and Molecular Oncology, Medical Institute of Bioregulation, Kyushu University, 4546 Tsurumibaru, Beppu 874-0838, Japan. Telephone: 81-977-27-1650; Fax: 81-977-27-1651; e-mail: [email protected] or Takashi Hirano, Ph.D., Research Institute for Cell Engineering, National Institute of Advanced Industrial Science and Technology (AIST), 1-1-1 Higashi, Tsukuba, Ibaraki 305-8566, Japan. Telephone: 81-29-861-6152; Fax: 81-29-8616144; e-mail: [email protected] Received December 4, 2006; accepted for publication February 1, 2007. ©AlphaMed Press 1083-7159/2007/$30.00/0 doi: 10.1634/theoncologist.12-4-406

The Oncologist 2007;12:406 – 417 www.TheOncologist.com

Hirasaki, Noguchi, Mimori et al.

INTRODUCTION Comparative genomic hybridization (CGH) has been developed to analyze whole chromosomal aberrations of an entire genome [1]. This method has identified chromosomal aberrations in several kinds of cancer. Recently, DNA microarrays have been applied to CGH (array CGH), and this combined technique has a great potential for comprehensive analysis of both the relative DNA copy number and altered chromosomal regions in cancer [2–3]. In addition, the completion of the human genome sequence has enabled identification of candidate genes within regions of DNA associated with cancers [4]. Esophageal cancer is one of the most prevalent cancers worldwide. Most esophageal cancers seen in Japanese patients show squamous cancer histologically, while adenocarcinomas are frequently seen in Caucasians [5]. The 5-year survival rate of esophageal cancer is ⬍40% despite the development of treatment modalities such as surgery, chemotherapy, and radiotherapy [6]. Thus, in order to develop better treatments, it is desirable to identify genomic alterations associated with esophageal squamous cancers (ESCs). Several previous studies have used chromosomal CGH (cCGH) to identify chromosomal aberrations associated with ESC. Noguchi et al. [7] studied the relationship between a gain in the 3p chromosomal region and tumor progression. Yen et al. [8] studied the relationship between a gain in 5p and 7q and a loss in 4p, 9p, and 11q and prognosis. Shinomiya et al. [9] reported the possible involvement of the DPI gene in the 13q34 amplicon. Adding to these reports, array-based CGH (aCGH) has been used recently in some reports. Ishizuka et al. [10] detected the amplification of eight oncogene loci using aCGH containing 51 oncogenes. Arai et al. [11] also detected the amplification of eight oncogene loci using the same aCGH. However, there are few reports studying whole genome profiles in ESC using aCGH. One of the most important factors in studies such as ours is the preparation of samples. It is necessary to obtain pure samples consisting only of cancer cells to evaluate the genomic changes associated with cancer. Samples obtained from cancer tissue usually contain both cancer cells and stromal cells. The recent development of laser microdissection (LMD) enables us to obtain pure cancer cells from the samples. In this study, we used a human bacterial artificial chromosome (BAC) array containing 4,030 human BAC clones and performed aCGH for 23 cases of ESC and two cases of normal esophageal epithelium. The BAC array does not require moving averages and data manipulation, in comparison with that of an oligoarray, because large clone inserts provide a strong signal as a result of the length of DNA available. In addition, the same clone on the array can be used for other applications, for example, fluorescence in

www.TheOncologist.com

407

situ hybridization (FISH) [12]. DNA samples were prepared through laser microdissected sections. Consequently, we identified 68 BAC clones that were well associated with the prognosis of postoperative patients. For 22 of these 68 BAC clones, there was a significant difference in the Kaplan-Meier survival curve, using the log-rank test, when comparing patients who had an alteration in a particular clone with those who did not.

MATERIALS AND METHODS Patients Cancer tissue samples were taken from 27 patients with ESC who underwent esophagectomy with lymph node dissection or endoscopic biopsy between 1999 and 2003 at our hospital. The age of the patients ranged from 36 to 82 years (mean, 63.5 years). Written informed consent was obtained from each patient. We evaluated 23 of the 27 cases that passed a strict DNA quality examination. Twenty-one samples were operated cases and two samples were biopsy cases. Two of the 21 operated patients died of non– cancer related causes 6 months and 7 months after the operation. The other 19 patients were re-evaluated ⬎5 years after the initial operation. Ten of the patients died of cancer recurrence within 3–34 months (mean survival time, 10.1 months). One of the nine surviving patients had local recurrence. Using the tumor node metastasis (TNM) classification system of the International Union Against Cancer [13], the 19 patients who underwent an operation were classified as follows: one with T1 tumors, three with T2 tumors, 13 with T3 tumors, and one with T4 tumors. Pathologically, all the tumors were squamous cancer. Lymph node metastases were present in 16 of the 19 patients (84.2%).

Tissue Samples and Preparation of the Cancer Cell Population by LMD All samples were immediately obtained from the resected or biopsied esophageal tumors and frozen at ⫺80°C. For

Figure 1. Before (A) and after (B) the laser microdissection in tumor samples obtained from case 7. At least 10,000 cancer cells were collected from each sample, and then genomic DNA was extracted. We were able to extract 500 –3,000 ng genomic DNA from each sample.

408

BAC Clone Alteration in Esophageal Squamous Cancer

Figure 2. Array comparative genomic hybridization (CGH) of female-control DNA by male-control DNA, normal squamous epithelial cells obtained from case 14, and squamous cancer cells obtained from cases 14, 6, 12, and 28. In female-control DNA by male-control DNA, the mean value of chromosomes 1–22 was 0.002342 ⫾ 0.052399 (one standard deviation). In the one normal sample (case 14N), 99.6% of the 4,030 BAC clones were within ⫾0.5 of the log2 ratio and the mean value of chromosomes 1–22 was – 0.05977 ⫾ 0.13493. In cancer samples, the regions of gain and loss noted by conventional chromosomal CGH were similarly recognized as clustered aberrations.

the control, a sample of normal esophageal squamous tissue was obtained. Frozen tissues were embedded in Tissue Tek OCT medium (Sakura, Tokyo, Japan), and then 10 –15 serial sections of 7 ␮m thickness were cut using a cryostat (Leica Microsystems, Wetzlar, Germany). Each section was mounted on a glass slide and covered with PEN foil (2.5 ␮m thick; Leica Microsystems). The sections were quickly fixed using a mixture of 100% ethanol and acetic anhydride (19:1). Samples were stored at ⫺80°C until use. Slides were stained with hematoxylin and eosin (H&E) at room temperature, dehydrated in ethanol, and air-dried.

LMD was performed as shown in Figure 1. Ten to 15 slides were dissected for each case. Sections were microdissected using the LMD system with a 337 nm nitrogen UV laser (Leica Laser Microdissection System, Leica Microsystems). Target cells were dropped immediately into a microcentrifuge tube cap filled with 30 ␮l of lysis buffer (Qiagen, Hiden, Germany). At least 10,000 cancer cells were collected from each sample, and then genomic DNA was extracted with the QIAamp DNA Micro Kit (Qiagen) according to the manufacturer’s instructions. The DNA concentration was calculated using a NanoDrop (ND-1000;

Hirasaki, Noguchi, Mimori et al.

409

Table 1. BAC clones with a gain (log2 ratio⬎0.5) or loss (log2 ratio⬍⫺0.5) A: Representative clones with a gain (log2 ratio⬎0.5) n of cases showing a gain BAC BAC size (n ⴝ 23) plate no. Location (bp) Genes contained in the clone

* * * * * *

17 16 16 15 15 15 15 15 14 14 14 13 13 13 13 12 12 12 12 12

314A 2819 243 677 2817 2272 2288 124 4974 755 2846 1562 4775 241 5845 4926 2501 4005 2484 1267

3q26.2 3q27.2 11q13.3–11q13.4 11q13.3 11q13.3 11q13.3 11q13.3 11q13.3 3q26.33 3q29 8q24.21 3q26.2 3q27.1 3q28 3q29 3q25.31 3q26.31 3q26.32 3q26.32 3q27.1

99,946 122,155 112,851 77,017 88,988 116,350 113,952 78,000 96,165 77,018 77,477 115,524 88,154 98,922 83,230 113,757 83,092 127,284 93,897 85,602

12 12 12 12 11 11 11 11 17 16 15 15 15 15 15 15 15 15 14

4009 655 1277 751 2896 2821 2622 1284 1268 920 341 2392 524 4317 4431 678 1436 4944 4229

3q27.1 3q27.3–3q28 3q28 11q13.3 3q27.1 3q27.3 3q29 8q24.3 3p22.3 11q22.3 3p14.2 3p14.2 4q33–4q34.1 5q34 11q25 13q13.2 18q22.3 18q23 3p24.3

129,275 137,406 105,270 106,169 95,439 61,903 84,185 86,161 101,937 97,746 100,213 78,749 91,032 79,457 87,111 88,589 82,839 97,979 133,553

MDS1 LOC344887, ETV5 PPFIA1, CTTN, SHANK2 CCND1, FLJ42258, ORAOV1 FGF4 FGF4, FGF3

LOC401102 LOC401110, LOC440995, MFI2, LOC391609, DLG1 MYC, PVT1 SKIL, CLDN11, KIAA1613 MCF2L2, B3GNT5 LRCH3, IQCG KCNAB1 TNFSF10 PIK3CA, KCNMB3 AP2M1, ABCF3, LOC90113, ALG3, MGC2408, ECE2, CAM-KIIN EPHB3 LPP LPP CPT1A THPO, CHRD, LOC285248 MASP1 LOC442100, BDH ZC3HDC3, GSDMDC1, PP3856, EEF1D LOC389105, ARPP-21 OR2AL1P FHIT PTPRG, C3orf14

OPCML

LOC400662 (Continued)

www.TheOncologist.com

BAC Clone Alteration in Esophageal Squamous Cancer

410

Table 1. (Continued) B: Representative clones with a loss (log2 ratio⬍⫺0.5) n of cases showing BAC a loss (n ⴝ 23) plate no. Location 15 15 15 15 15 15 15 15 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14

341 2392 524 4317 4431 678 1436 4944 4229 2757 4988 4434 4416 1366 4995 5026 5025 689 450 404 1093 2424 4379 5022 915

3p14.2 3p14.2 4q33–4q34.1 5q34 11q25 13q13.2 18q22.3 18q23 3p24.3 3p14.2 3p12.2 3p12.1 4q12 4q32.3 4q34.3 5q23.1 5q23.3 5q23.3 5q23.3 5q33.2–5q33.3 5q35.1 11q23.1 13q21.1 18q22.3 4p14

BAC size (bp)

Genes contained in the clone

100,213 78,749 91,032 79,457 87,111 88,589 82,839 97,979 133,553 82,602 112,750 108,101 81,905 79,732 106,527 108,125 87,281 92,005 76,590 104,685 90,446 72,669 98,603 62,424 101,650

FHIT PTPRG, C3orf14

OPCML

LOC400662 FHIT LOC391555

LOC402191

ACSL6 IL3, CSF2 KCNIP1 NCAM1 NETO1

* Indicates a BAC clone showing a higher level of gain (log2 ratio ⬎1.0). Abbreviation: BAC, bacterial artificial chromosome.

NanoDrop Technologies, Wilmington, DE) according to the manufacturer’s instructions. We were able to extract 500 –3,000 ng genomic DNA from each sample.

Microarray Hybridization and Scanning The MACarray Karyo 4000 array slide (Macrogen Inc., Seoul, Korea, http://www.macrogen.com) was used for this study. It contains 8,060 elements, representing 4,030 human BAC clones selected from the original BAC library (96,768 clones) established by Macrogen Inc. in duplicate and randomly distributed throughout one microarray slide. The BACs, 4,030 clones, employed in this array were selected as fulfilling the following criteria: (a) the end-sequencing of the clones was complete and (b) the chromosomal locations of the clones were determined by FISH physical mapping. The entire genome was covered by BAC clones with a resolution of ⬍1 M base pairs.

Test DNA and human female DNA (0.5 ␮g each) were labeled with Cyanine3- or Cyanine5-dCTP (Perkin Elmer, Wellesley, MA) using the random primer method (BioPrime DNA labeling kit, Invitrogen, Carlsbad, CA). Labeled probes were purified by spin column (QiaQuick PCR Purification Column, Qiagen) and dissolved in 140 ␮l of hybridization solution (Macrogen) containing 100 ␮l of Cot-1 DNA solution (Macrogen) and 4 ␮l of yeast tRNA solution (Macrogen). The probe solution was heated to 70°C for 10 minutes to denature the DNA, then incubated for 60 minutes at 37°C to block repetitive sequences. The array slide was pretreated with salmon sperm DNA for 30 minutes at room temperature and then mounted on the slide processor (GeneMachines HybStation, Genomic Solutions, Ann Arbor, MI). After injection of 120 ␮l of probe, hybridization was performed for 72 hours on the HybStation with continuous agitation, followed by posthybridization

Hirasaki, Noguchi, Mimori et al.

411

Figure 3. A 1.1-Mb amplification corresponding to six consecutive clones (arrow) of 11q13.3 was the most frequently identified alteration. Fifteen of 23 cases showed this change, and six representative cases are shown.

washes as follows: 50% formamide/2⫻ standard saline citrate (SSC) for 15 minutes at 46°C, 0.1% SDS/2⫻ SSC for 30 minutes at 46°C, PN buffer (0.1M Na2PO4/0.1% NonDiet P-40, Nacalai Tesque, Kyoto, Japan) for 15 minutes at 46°C, and 2⫻ SSC for 5 minutes at 46°C. After washing, the slides were dehydrated in ethanol. Following hybridization, the slides were scanned at 532 and 635 nm by a GenePix 4000A (Axon Instruments, Sunnyvale, CA). Next, MacViewer software (http://www. macrogen.com/eng/biochip/macviewer.jsp/), an analytical software program developed by Macrogen, was used to locate spots automatically on the Cy3 and Cy5 image acquisitions and to calculate fluorescence ratios. The MacViewer software automatically analyzed and summarized the results as follows: (a) averaged the ratios of the replicates and calculated the standard deviation, (b) rejected individual spot data based on several criteria (including weak fluorescent signals), (c) adjusted the Cy5/Cy3 ratios such that ratios of the normal genomic regions were always equal to 0, despite variations in dye labeling efficiency, and (d) plotted data relative to the position of the clones in the human genome (according to July 2003 University of California– Santa Cruz cartography). Data from the aCGH are summarized in online supplementary Table 1.

⬎0.5 and a loss when this ratio was below ⫺0.5 with reference to previous reports [14 –17]. To elucidate the loci that correlate with postoperative prognosis, we counted the number of cases representing a gain or loss (deviation from the log2 ratio of ⫾0.5) in each of the 4,030 BAC clones, and with this number, a ␹2 test was performed between the nine survivors and 10 nonsurvivors. For clones showing differences with a p ⬍ .05 by the ␹2 test, the survival curve was analyzed between patients who had an alteration in a particular clone and those who did not, using the Kaplan-Meier method and the log-rank test. Clones that contain regions known to have genomic variants, according to the Database of Genomic Variants (The Centre for Applied Genetics, Toronto, Canada, http://projects.tcag.ca/variation/), were omitted. Prognostic factors for clinicopathological variables were examined by univariate and multivariate analyses (Cox proportional hazards regression model). A p-value ⬍ .05 was considered to be statistically significant. All statistical analyses were carried out with JMP 5.0.1J software for Windows (SAS Institute Inc. Cary, NC).

Statistical Analysis

To confirm the profile of normal epithelium, we performed aCGH of normal epithelial DNA for two cases (case 14 and 24). As shown in Figure 2 (cases 14N and 14T), few alter-

Chromosomal aberrations were classified as a gain when the normalized log2-transformed fluorescence ratio was

www.TheOncologist.com

RESULTS aCGH of Normal Esophageal Squamous Epithelium: Quality Controls

BAC Clone Alteration in Esophageal Squamous Cancer

412

ations were seen in genomic DNA extracted from normal esophageal squamous cells (case 14N) compared with those in the cancer cell sample (case 14T). In cases 14N and 24N, only 18 (0.4%) and 31 (1%) loci of the 4,030 BAC clones deviated from the ⫾0.5 range of the log2 ratio, respectively. These regions were omitted for further analyses in this study.

BAC Clones Showing a Nonrandom Gain or Loss in Cancer Samples Figure 2 shows log2 ratio plots for the 4,030 BAC clones from four (cases 14T, 6T, 12T, and 28T) of the 23 ESC cases. Gains in 3q, 8q, and 11q and losses in 3p, 4p, 9p, and 13q have been reported in previous cCGH studies, and we also verified those alterations using our methods. Table 1A and 1B list the BAC clones showing a gain (⬎0.5 of the log2 ratio) or a loss (⬍⫺0.5 of the log2 ratio), according to the number of cases with a gain or a loss. The clones containing known oncogenes, such as CCND1, MYC, and PIK3CA,

showed gains in 15, 14, and 12 cases, respectively. In addition, the clones containing the known tumor suppressor gene FHIT showed losses in 15 cases. The BAC clones showing a higher gain (⬎1.0 of the log2 ratio) are indicated with an asterisk in Table 1A. Interestingly, a region at 11q13.3, corresponding to six consecutive clones, frequently showed a gain, in ⬎15 of 23 cases. A gain in 11q13.3 has been reported in cancers of the esophagus, oral cavity, and nasopharynx; however, our aCGH data clearly demonstrate this region as a 1.1-Mb amplicon corresponding to six consecutive clones, as shown in Figure 3.

Correlation Between a Copy Number Gain or Loss and Patient Prognosis In 19 patients who underwent a curative operation and were followed for ⬎5 years after the operation, nine survived and 10 died from recurrent disease (nonsurvivors). To elucidate the clones that correlate with postoperative prognosis, we

Table 2. BAC clones with a significant difference using a ␹2 test between survivors and nonsurvivors A: Clones with a normal ratio in survivors but a gain in nonsurvivors BAC plate no. Location BAC size (bp) Genes contained in the clone 1027 3q26.2 104,097 * 5622 8q24.3 150,891 B: Clones with a normal ratio in survivors but a loss in nonsurvivors BAC Location BAC size (bp) plate no. 697 1p13.2 90,044 2134 1p13.2 89,971 341 3p14.2 100,213 * 4135 3p13– 75,975 3p12.3 * 4214 9p21.3 130,760 1584 11p15.4 125,339 * 4369 11q14.1 114,822 2195 11q23.1 76,206 2573 13q14.11 75,884 4116 13q14.11 95,540 1183 13q14.13 98,146 * 2979 13q14.2 89,466 1130 13q14.3 117,874 * 84 13q14.3 111,777 * 1605 Xq12 89,996 * 199 Xq12 108,521 70 Xq13.1 126,385 * 1400 Xq13.1 82,380

LOC389174 ZNF251, LOC441382, ZNF34, RPL8, ZNF517 Genes contained in the clone WNT2B, ST7L CAPZA1, MOV10, RHOC, PPP2CZ FHIT[B] LOC442083 MTAP SMPD1, APBB1, HPX POU2AF1 FOXO1A FLJ10094 COG3, FLJ32682

VSIG4, LOC392485, LOC441498 AR LOC389866, EFNB1 PJA1, LOC439943 (Continued)

Hirasaki, Noguchi, Mimori et al.

413

Table 2. B: Clones with a normal ratio in survivors but a loss in nonsurvivors (Continued) BAC plate no. Location BAC size (bp)

*

5192 5219 5194 5195 5096 4185 5221

Xq13.3 Xq21.1 Xq21.1 Xq21.33 Xq21.33 Xq21.33 Xq21.33

134,421 82,560 75,926 98,270 88,990 84,274 109,434

Genes contained in the clone LOC286495, MAGEE2

C: Clones with a gain in survivors but a normal ratio in nonsurvivors

* *

BAC plate no.

Location

BAC size (bp)

Genes contained in the clone

200 4009 2622 4280 5637 5653 5510 5860 840

1q32.1 3q27.1 3q29 5p15.2 7q36.3 18p11.32 18p11.32 18p11.32 20q11.21

128,832 129,275 84,185 108,721 122,100 99,515 120,580 76,279 116,351

PKP1, TNNT2, LAD1, TNNI1 EPHB3 LOC442100, BDH MARCH6 PTPRN2 USP14, THOC1 COLEC12 LOC441805, LOC441806 LOC400842, LOC149950

D: Clones with a loss in survivors but a normal ratio in nonsurvivors

*

*

*

* * *

BAC plate no.

Location

BAC size (bp)

4001 4157 4023 5282

4q13.3 4q21.1 4q21.21 4q21.21

79,487 107,816 109,909 85,596

324 807 2639 2784 1567

4q21.21 4q22.1 4q23 4q25 4q25

81,715 114,004 90,171 83,807 100,602

106 4975 2091

4q25 4q26 4q27

106,065 101,900 105,940

1330 229 4111 1233

4q27 4q31.1 5q11.2 5q12.3

82,605 117,680 113,544 90,841

827 732 1054

5q14.2 5q15 5q33.2

153,856 108,355 97,524

Genes contained in the clone

MGC10646 FGF5, MRPS25P1, MGC35043 MGC35043 TIGD2 RAP1GDS1 LEF1, LOC441033 FLJ20647, CASP6, PLA2G12A, IF ENPEP, PITX2 SEC24D EXOSC9, CCNA2, BBS7, TRPC3 FLJ35630 MAML3 KIAA0052, PPAP2A LOC133993, FLJ36754, SDCCAG10 RPS23, LOC92270 DKFZP564D172 GALNT10, FLJ38109, SAP30L, HAND1, LOC389339 (Continued)

www.TheOncologist.com

BAC Clone Alteration in Esophageal Squamous Cancer

414

Table 2. D: Clones with a loss in survivors but a normal ratio in nonsurvivors (Continued) BAC plate no. Location BAC size (bp) Genes contained in the clone * * * *

91 164 1047 1127 4297 150 740 1162 171 566 247 4096 1252

8p21.3 10q11.23 10q25.2 10q26.2–3 14q21.3 20q12 20q12 20q12 20q12 20q12 20q13.12 20q13.2 20q13.32

100,400 109,864 86,657 102,446 111,877 171,384 91,326 76,972 114,459 103,066 86,265 83,546 81,946

FLJ10569 PRKG1

LOC339568 MAFB PTPRT PTPRT C20orf100 LOC343629, C20orf174

*Indicates a BAC clone showing a significant difference using a log-rank test. Abbreviation: BAC, bacterial artificial chromosome.

counted the number of cases representing a gain or loss (deviation from the log2 ratio of ⫾0.5) in each of the 4,030 BAC clones, and with this number, a ␹2 test was performed between the nine survivors and the 10 nonsurvivors. A significant difference (p ⬍ .05) was detected in 92 of the 4,030 clones. Twenty-four of these 92 clones were omitted because they contain a region known to have genomic variants, according to the Database of Genomic Variants, so 68 clones were used. Two clones showed a normal ratio in survivors but a gain in nonsurvivors, 25 showed a normal ratio in survivors but a loss in nonsurvivors, nine showed a gain in survivors but a normal ratio in nonsurvivors, and 32 showed a loss in survivors but a normal ratio in nonsurvivors (Table 2). Almost half of these clones (31 of the 68 BAC clones) were located at nine limiting regions: 5.3 Mb of 4q13.3– 4q21.21 (76.4 – 81.7 Mb, 5 clones), 2.5 Mb of 4q25 (109.4 –111.9 Mb, 3 clones), 4.1 Mb of 4q26 – 4q27 (119.9 –124.0 Mb, 3 clones), 5.1 Mb of 13q14.11–13 (40.0 – 45.3 Mb, 3 clones), 0.5 Mb of 13q14.2–3 (49.6 Mb50.2 Mb, 3 clones), 1.6 Mb of 20q12 (37.1–38.7 Mb, 3 clones), 1.9 Mb of 20q12–13.12 (40.2– 42.1 Mb, 3 clones), 3.3 Mb of Xq12–13.1 (64.9 – 68.2 Mb, 4 clones), and 1.4 Mb of Xq21.33 (93.7–95.1 Mb, 4 clones). Furthermore, we plotted Kaplan-Meier survival curves for each of these 68 BAC clones, and used a log-rank test to distinguish between patients who had an alteration in a specific clone and those who did not. Consequently, 22 of these 68 BAC clones showed a significant difference (p ⬍ .05) using the log-rank test: one clone showed a normal ratio in survivors and a gain in nonsurvivors (Table 2A), nine showed a normal ratio in survivors and a loss in nonsurvivors (Table 2B), two

showed a gain in survivors and a normal ratio in nonsurvivors (Table 2C), and 10 showed a loss in survivors and a normal ratio in nonsurvivors (Table 2D). Figure 4 shows the survival curves of six representative BAC clones of the 22 total. Although the number of cases in this study is small, we performed univariate and multivariate analyses with copy number changes in these six representative BAC clones and with the clinicopathological factors of each case. Univariate analysis showed that the following factors were significantly related to postoperative survival: lymphatic invasion, venous invasion, and copy number changes in each of the six BAC clones (online supplementary Table 2A). Multivariate regression analysis, using lymphatic invasion, venous invasion, and alteration in each of the six BAC clones, indicated that the copy number changes of four of the clones (BAC no. 164, 2622, 5282, and 5622) were independent prognostic factors (online supplementary Table 2B).

DISCUSSION Previous cCGH studies of esophageal cancer identified several characteristic chromosomal aberrations, such as gains in 3q, 8q, and 11q and losses in 3p, 4p, 9p, and 13q [7–9, 18, 19]. These changes were similarly identified in our current study. The two previous aCGH studies disclosed that amplification of the regions of 11q13 containing the Cyclin D1, FGF3/4, and EMS1 genes were each identified in 9 of 32 cases [10] and in 5, 4, and 7 of 20 cases [11], respectively. In our study, we detected this amplification as a 1.1-Mb amplicon corresponding to six BAC clones in more cases (Table 1A and Fig. 3). The higher detectability in our

Hirasaki, Noguchi, Mimori et al.

415

Figure 4. Kaplan-Meier curves showing a significant difference using the log-rank test in the region of 22 BAC clones when comparing patients who had a gain (red line) or loss (blue line) in that particular clone with those who did not (green line). Six representative regions are shown in this figure. Abbreviation: BAC, bacterial artificial chromosome.

study may be partly a result of strict sample preparation with LMD [20]. On the other hand, the remaining 8 of the 23 cases did not show a gain in this region, suggesting that this change may not always occur in ESC. Amplification at 11q13 has been reported previously not only in ESC but also in oral cancer, head and neck cancer, nasopharyngeal cancer, and several other types of cancer [21–25]. For example, Hui et al. [21] reported this amplification as a 5.3-Mb amplicon at 11q13.1–13.3 as the most frequently detected gain in 26 nasopharyngeal cancers. They found that CCND1 is a target oncogene of nasopharyngeal cancer by analyzing six genes—MEN1, CCND1, FGF3, EMS1, GARP, and PAK1—with high-resolution aCGH. Other alterations in the clones at 17q12, containing ERBB2, or 8p11, containing FGFR1, were also identified in two and four cases, respectively, similar to previous reports [10]. However, the alterations in these clones were demonstrated as a part of widely gained regions, and thus are much different from the 11q13.3 amplifications, with a clear-cut, narrow change. Also, one clone (BAC no. 5359) with a novel loss (indicating homozygous deletion) was recognized at 16p13.3, containing the genes ZNF434, ZNF174, ZNF597, FLJ14154, LOC390671, and CLUAP1, which has not been reported previously in ESC (online supplementary Table 1). In addition, one clone (BAC no. 5308) with a novel loss was recognized at 11p15.4 (online supplemen-

www.TheOncologist.com

tary Table 1) containing the gene CDKN1C, known as a tumor suppressor gene [26 –29]. However this clone was omitted from our analysis because it contains a region known to have genomic variants according to the Database of Genomic Variants. With respect to the relationship between patient prognosis and chromosomal gain or loss, cCGH analysis has previously demonstrated that gains in 5p, 7q, and 12p and losses in 4p, 9p, and 11q are associated with poor prognosis in ESC patients [8, 18]. However, in our study, no such correlations were demonstrated using aCGH analysis. A chromosomal gain in 7q and losses in 4p and 11q were observed in both survivors and nonsurvivors. However, there were no significant differences. There were no chromosomal gains or losses in 5p, 12p, or 9p. Next, we studied the correlation between patient prognosis and DNA copy number change for each clone. Using a ␹2 test, we detected that the occurrence of a gain or loss showed a significant difference between patients with a good prognosis and those with a poor prognosis in 68 of the 4,030 BAC clones. Almost half of these clones (31 of 68 BAC clones) were present at nine limiting regions of 4q, 13q, 20q, and Xq. Further study is necessary to discern why many clones located in these nine regions show an alteration from unity. In 22 of 68 BAC clones, there was a significant difference in the KaplanMeier survival curves, using the log-rank test, when comparing patients who had an alteration in a particular clone

BAC Clone Alteration in Esophageal Squamous Cancer

416

with those who did not. All but two of the loci of clones in 11q (11q14.1, 11q23.1) were not reported in previous cCGH studies. Among the genes located in the selected BAC clones in our study, the MTAP gene, located at 9p21.3, has been reported as a tumor suppressor in several cancer types [30 –35]. In those studies, homozygous deletions of MTAP and p16/CDKN2A in pancreatic tumors and breast cancer cell lines were reported [30 –32]. In addition, a lack of MTAP protein expression was associated with poor prognosis in mantle cell lymphoma [33]. We previously reported that the FHIT gene, located at 3p14.2, could be a tumor suppressor gene [36]. In this study, BAC clones no. 341 and 2757, which include this gene, were frequently lost in patients who died of recurrent ESC. These findings strongly support the fidelity of our CGH study. Many reports have described that DNA copy number alterations are usual phenomena in poor prognosis patients with many kinds of cancer, including esophageal cancer [8] [18]. However, this study clarified that alterations are frequently seen in patients with a good prognosis, compared with those with a poor prognosis, in a large number of clones. This is a novel finding, and needs further study to disclose its significance. Amplifications and deletions in genomic DNA are associated with high and low expression levels of the corresponding mRNA, respectively, for the most part [37, 38]. Prediction of the prognosis of cancer patients can be performed with DNA examination, as well as with RNA expression examination, protein expression examination, and so on. This study clearly demonstrates that some selected loci of BAC clones are very useful to predict the prognosis of patients with ESC. Studies using RNA expression re-

quire more strict sample preparation than those using DNA, thus, we consider DNA examination to be more practical than the use of RNA expression. Indeed Kyomoto et al. [39] reported that genomic amplification of the region that contains the Cyclin D1 gene is a more potent prognostic factor than its protein overexpression in human head and neck squamous cancer. It is desirable to develop methods that can easily identify gains or losses in specific genomic regions for practical examination. To our knowledge, our study is the first to demonstrate the correlation between alterations in specific loci of BAC clones and patient prognosis using aCGH. This study is very precise, because we used LMD samples obtained from frozen tissue. Of course, LMD is difficult to use in routine examination in its present form; however, if much easier methods were developed, this would be very useful in practice. In addition, the BAC clones identified in this study give us the opportunity to determine which specific genes are associated with cancer progression or metastasis.

ACKNOWLEDGMENTS We thank Y. Nakagawa for her technical assistance. Financial support came from Grant-in-Aid for Scientific Research, Japan Society for the Promotion of Science, No. 17109013 (S), and Ministry of Education, Culture, Sports, Science and Technology, No. 18790964(B). The authors Masaki Mori and Takashi Hirano contributed equally.

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST The authors indicate no potential conflicts of interest.

REFERENCES 1

2

Kallioniemi A, Kallioniemi OP, Sudar D et al. Comparative genomic hybridization for molecular cytogenetic analysis of solid tumors. Science 1999;258:818 – 821. Solinas-Toldo S, Lampel S, Stilgenbauer S et al. Matrix-based comparative hybridization: Biochips to screen for genomic imbalances. Genes Chromosomes Cancer 1997;20:399 – 407.

ageal squamous cell carcinoma : 3q gain correlates with tumor progression but not prognostic significance. Oncol Rep 2003;10:1393–1400. 8

Yen CC, Chen YJ, Chen JT et al. Comparative genomic hybridization of esophageal squamous cell carcinoma: Correlations between chromosomal aberrations and disease progression/prognosis. Cancer 2001;92:2769 – 2777.

9

Shinomiya T, Mori T, Ariyama Y et al. Comparative genomic hybridization of squamous cell carcinoma of the esophagus: The possible involvement of the DPI gene in the 13q34 amplicon. Genes Chromosomes Cancer 1999; 24:337–344.

3

Pinkel D, Segraves R, Sudar D et al. High resolution analysis of DNA copy number variation using comparative genomic hybridization to microarrays. Nat Genet 1998;20:207–211.

4

International Human Genome Sequencing Consortium. Finishing the euchromatic sequence of human genome. Nature 2004;431:931–945.

5

Sarbia M, Becker KF, Hofler H. Pathology of upper gastrointestinal malignancies. Semin Oncol 2004;31:465– 475.

6

Ando N, Ozawa S, Kitagawa Y et al. Improvement in the results of surgical treatment of advanced squamous esophageal carcinoma during 15 consecutive years. Ann Surg 2000;232:225–232.

11 Arai H, Ueno T, Tangoku A et al. Detection of amplified oncogenes by genome DNA microarrays in human primary esophageal squamous cell carcinoma: Comparison with conventional comparative genomic hybridization analysis. Cancer Genet Cytogenet 2003;146:16 –21.

7

Noguchi T, Kimura Y, Takeno S et al. Chromosomal imbalance in esoph-

12 Smith LT, Mayerson J, Nowak NJ et al. 20q11.1 amplification in giant-cell

10 Ishizuka T, Tanabe C, Sakamoto H et al. Gene amplification profiling of esophageal squamous cell carcinomas by DNA array CGH. Biochem Biophys Res Commun 2002;296:152–155.

Hirasaki, Noguchi, Mimori et al. tumor of bone: Array CGH, FISH, and association with outcome. Genes Chromosomes Cancer 2006;45:957–966. 13 Sobin LH, Fleming ID. TNM Classification of Malignant Tumors, fifth edition (1997). Union Internationale Contre le Cancer and the American Joint Committee on Cancer. Cancer 1997;80:1803–1804. 14 Albertson DG, Ylstra B, Segraves R et al. Quantitative mapping of amplicon structure by array CGH identifies CYP24 as a candidate oncogene. Nat Genet 2000;25:144 –146. 15 Pollack JR, Sorlie T, Perou CM et al. Microarray analysis reveals a major direct role of DNA copy number alteration in the transcriptional program of human breast tumors. Proc Natl Acad Sci U S A 2002;99:12963–12968. 16 Wilhelm M, Veltman JA, Olshen AB et al. Array-based comparative genomic hybridization for the differential diagnosis of renal cell cancer. Cancer Res 2002;62:957–960. 17 Hui AB, Takano H, Lo KW et al. Identification of a novel homozygous deletion at 6q23.1 in medulloblastomas using high-resolution array comparative genomic hybridization analysis. Clin Cancer Res 2005;11:4707– 4716. 18 Kwong D, Lam A, Guan X et al. Chromosomal aberrations in esophageal squamous cell carcinoma among Chinese: Gain of 12p predicts poor prognosis after surgery. Hum Pathol 2004;35:309 –316. 19 Su M, Chin SF, Li XY et al. Comparative genomic hybridization of esophageal adenocarcinoma and squamous cell carcinoma cell lines. Dis Esophagus 2006;19:10 –14. 20 Cardoso J, Molenaar L, de Menezes RX et al. Genomic profiling by DNA amplification of laser capture microdissected tissues and array CGH. Nucleic Acids Res 2004;32:e146.

417 26 Henry I, Grandjouan S, Couillin P et al. Tumor-specific loss of 11p15.5 alleles in del11p13 Wilms tumor and familial adrenocortical carcinoma. Proc Natl Acad Sci U S A 1989;86:3247–3251. 27 Bepler G, Garcia-Blanco MA. Three tumor-suppressor regions on chromosome 11p identified by high-resolution deletion mapping in human nonsmall-cell lung cancer. Proc Natl Acad Sci U S A 1994;91:5513–5517. 28 Shibagaki I, Shimada Y, Wagata T et al. Allelotype analysis of esophageal squamous cell carcinoma. Cancer Res 1994;54:2996 –3000. 29 Baffa R, Negrini M, Mandes B et al. Loss of heterozygosity for chromosome 11 in adenocarcinoma of the stomach. Cancer Res 1996;56:268 –272. 30 Hustinx SR, Leoni LM, Yeo CJ et al. Concordant loss of MTAP and p16/ CDKN2A expression in pancreatic intraepithelial neoplasia: Evidence of homozygous deletion in a noninvasive precursor lesion. Mod Pathol 2005; 18:959 –963. 31 Hustinx SR, Hruban RH, Leoni LM et al. Homozygous deletion of the MTAP gene in invasive adenocarcinoma of pancreas and in periampullary cancer: A potential new target for therapy. Cancer Biol Ther 2005;4:83– 86. 32 Christopher SA, Diegelman P, Porter CW et al. Methyladenosine Phosphorylase, a gene frequently codeleted with p16(cdkN2a/ARF), acts as a tumor suppressor in a breast cancer cell line. Cancer Res 2002;62:6639 – 6644. 33 Marce S, Balague O, Colomo L et al. Lack of methylthioadenosine phosphorylase expression in mantle cell lymphoma is associated with shorter survival: Implications for a potential targeted therapy. Clin Cancer Res 2006;12:3754 –3761. 34 Berasain C, Hevia H, Fernandez-Irigoyen J et al. Methylthioadenosine phosphorylase gene expression is impaired in human liver cirrhosis and hepatocarcinoma. Biochim Biophys Acta 2004;1690:276 –284.

21 Hui AB, Or YY, Takano H et al. Array-based comparative genomic hybridization analysis identified cyclin D1 as a target oncogene at 11q13.3 in nasopharyngeal carcinoma. Cancer Res 2005;65:8125– 8133.

35 Behrmann I, Wallner S, Komyod W et al. Characterization of Methylthioadenosin phosphorylase (MTAP) expression in malignant melanoma. Am J Pathol 2003;163:683– 690.

22 Huang X, Gollin SM, Raja S et al. High-resolution mapping of the 11q13 amplicon and identification of a gene, TAOS1, that is amplified and overexpressed in oral cancer cells. Proc Natl Acad Sci U S A 2002;99:11369 – 11374.

36 Ohta M, Inoue H, Cotticelli MG et al. The FHIT gene, spanning the chromosome 3p14.2 fragile site and renal carcinoma-associated t(3;8) breakpoint, is abnormal in digestive tract cancers. Cell 1996;84:587–597.

23 Bergsagel PL, Kuehl WM. Molecular pathogenesis and a consequent classification of multiple myeloma. J Clin Oncol 2005;23:6333– 6338.

37 Tsafrir D, Bacolod M, Selvanayagam Z et al. Relationship of gene expression and chromosomal abnormalities in colorectal cancer. Cancer Res 2006;66:2129 –2137.

24 Nimeus E, Baldetorp B, Bendahl PO et al. Amplification of the cyclin D1 gene is associated with tumour subsite, DNA non-diploidy and high Sphase fraction in squamous cell carcinoma of the head and neck. Oral Oncol 2004;40:624 – 629.

38 Jarvinen AK, Autio R, Haapa-Paananen S et al. Identification of target genes in laryngeal squamous cell carcinoma by high-resolution copy number and gene expression microarray analyses. Oncogene 2006;25:6997– 7008.

25 Juhasz A, Balazs M, Sziklay I et al. Chromosomal imbalances in laryngeal and hypopharyngeal cancers detected by comparative genomic hybridization. Cytometry 2005;67:151–160.

39 Kyomoto R, Kumazawa H, Toda Y et al. Cyclin-D1-gene amplification is a more potent prognostic factor than its protein over-expression in human head-and-neck squamous-cell carcinoma. Int J Cancer 1997;74:576 –581.

www.TheOncologist.com