Materials and DNA isolation
DNA of 63 formalin-fixed and paraffin-embedded (FFPE) gastric adenocarcinomas of which 53 were obtained from Leeds (Leeds University Hospital, UK) and 10 were obtained from the Dutch D1/D2 trial , was isolated as previously described  using a commercial available DNA isolation kit (QIAmp DNA microkit, Qiagen, Hilden, Germany). Briefly, areas of at least 70% tumor cells were demarcated on a 4-µm haematoxylin- and eosin-stained tissue section. Adjacent serial sections of 10 µm were cut, and after deparaffination, the tumor tissue was macrodissected using a needle. After an overnight incubation with sodium thiocyanate (1 M) at 37°C, followed by proteinase K treatment, DNA was extracted. DNA concentrations were measured on a Nanodrop ND-1000 spectrophotometer (Isogen, IJsselstein, The Netherlands), and DNA quality was assessed by isothermal amplification as previously described . Only DNA of excellent, good, and intermediate quality was used for further analysis.
DNA labeling and array CGH)
DNA labeling and array CGH were essentially performed as previously described [17, 28]. In short, tumor and normal DNA were differentially labeled using random priming (Bioprime DNA Labeling System; Invitrogen, Breda, The Netherlands) and hybridized on a BAC array containing approximately 5,000 clones printed in triplicate, consisting of the Sanger BAC clone set with an average resolution along the whole genome of 1.0 Mb (http://www.ensembl.org/Homo_sapiens/cytoview), the OncoBac set (http://informa.bio.caltech.edu/Bac_onc.html), containing approximately 600 clones corresponding to 200 cancer-related genes, and selected clones of interest obtained from the Children’s Hospital Oakland Research Institute (CHORI) to fill any gaps larger than 1 Mb on chromosome 6 and to have full-coverage contigs of regions on chromosomes 8, 11, 13, and 20. All clones were printed on Codelink™ slides (Amersham BioSciences, Roosendaal, The Netherlands) at a concentration of 1 μg/μl in 150-mM sodium phosphate, pH 8.5, using an OmniGrid 100 microarrayer (Genomic Solutions, Ann Arbor, MI, USA) equipped with SMP3 pins (TeleChem International, Sunnyvale, CA, USA), and processed according to the manufacturer’s protocol.
After hybridization, images of the arrays were acquired by scanning (Microaray scanner G2505B; Agilent technologies, Palo Alto, USA), and spot analysis and quality control were automatically performed using BlueFuse 3.4 software (BlueGnome, Cambridge, UK). When the BlueFuse quality flag was below one or the confidence value was below 0.1, spots were excluded from further analysis. The log2 tumor to normal fluorescence ratio was calculated for each spot and normalized against the mode of the ratios of all autosomes. The package CGH call was used for data segmentation and defining copy-number gains and losses of each clone in the array CGH profile .
Multiplex ligation-dependent probe amplification
MLPA was performed, as previously described,  using two different probe mixes. One probemix contained 38 probes representing 31 different genes on chromosome 20 and 10 control probes located on chromosomes 2, 3, 4, 5, 12, and 16. The second probemix contained 11 probes on chromosome 8, 12 probes on chromosome 13, 16 probes on chromosome 20 representing 14 different genes, and 8 control probes located on chromosomes 2, 4, 12, and 16. Some genes on chromosome 20 are present in both probe mixes, leaving 35 different genes on chromosome 20 by combining these two probe mixes. DNA of the cell line HT29, showing a gain on chromosomes 8, 13, and 20, was used as positive control. A human pool of DNA isolated from blood of 36 healthy individuals and a pool of DNA isolated of 30 normal gastric and colon mucosa, spleen, liver, and kidney tissue samples (FFPE), were used as normal controls.
Of each sample, approximately 100 ng of DNA in a volume of 5 µl was denaturated at 98°C for 5 min. A mixture of 1.5-µl salsa probes (1–4 fmol of each short synthetic probe oligonucleotide and each phage M13-derived long probe oligonucleotide in TE (10 mM Tris-HCl, pH 8.2, 1 mM ethylenediamine tetraacetic acid (EDTA))) and 1.5 µl of MLPA buffer (1.5 M KCl, 300 mM Tris-HCl, pH 8.5, 1 mM EDTA) was added. The mixture was heated for 1 min at 95°C followed by 16 h of incubation at 60°C to allow the MLPA hemipobes to hybridize. Next, 32 µl of ligase-65 mixture (dilution buffer containing 2.6 mM MgCl2, 5 mM Tris-HCl, pH 8.5, 0.013% non-ionic detergents, 0.2 mM of nicotinamide adenine dinucleotide (NAD), and 1 U of ligase-65 enzyme) was added to each sample for ligation of hybridized hemiprobes during a 10–15 min of incubation at 54°C, followed by a 5 min of incubation at 98°C to inactivate the ligase.
Polymerase chain reaction (PCR) was performed with 10 µl of polymerase mixture containing the PCR primers (10 pmol), dNTPs (2.5 nmol), and 2.5 U Taq polymerase (promega), 4 µl of PCR buffer (2.6 mM MgCl2, 5 mM Tris-HCl, pH 8.5, 0.013% non-ionic detergents, 0.2 mM NAD), 26 µl of water and 10 µl of MLPA ligation reaction.
Multiplex ligation-dependent probe amplification data analysis
Analysis of the MLPA PCR products for each gene was performed on an ABI 3100 capillary sequencer (Applied Biosystems, Warrington, UK) in a mixture of 8.5 µl of deionized formamide (Applied Biosystems, Warrington, UK), 1 µl of PCR product and 0.5 µl marker including a ROX-labeled internal size standard (ROX-500 Genescan; Applied Biosystems, Warrington, UK). Data analysis was performed using the MLPAnalyzer version 8.0 (http://www.mlpa.com/coffalyser/) . For each tumor, peak heights for every probe were derived from the ABI output and median peak heights of at least two different ligation reactions and three different PCR reactions were calculated. For each sample, tumor to normal DNA copy-number ratios was calculated per probe by dividing the median peak heights in the tumor tissue by the median peak heights in the reference DNA. All ratios were normalized by setting the median of the tumor to reference DNA copy-number ratios of the control genes in the probe mixture to 1.0. When multiple probes were present for one gene, the mean value of the probes was calculated and used for further analysis. Tumor to normal ratios below 0.7 and above 1.3 was considered as a loss or gain, respectively. TMEV software 3.1 (http://www.tigr.org/) was used to present descriptive data.
Box and scatter plots were used to present descriptive statistics. Chi-square test was used to evaluate associations of DNA copy-number gain of chromosomes 8q, 13q, and 20q with clinicopathological variables. t Test was used to evaluate differences in DNA copy-number aberrations and age of the patients. Mann-Whitney U test and Kruskal-Wallis H test were used to evaluate differences in DNA copy-number changes of each gene between lymph node status and histological tumor type according to the Laurén classification , respectively. Correlation coefficients between the log2 ratios for the array CGH and MLPA analysis were obtained by Spearman correlation (SPSS 12.0.1 for Windows; SPSS Inc. Chicago, IL, USA). A threshold of 0.05 for significance was used.