Histopathologic subtype-specific genomic profiles of renal cell carcinomas identified by high-resolution whole-genome single nucleotide polymorphism array analysis

To elucidate the novel and common genetic changes in histopathologic subtype-specific profiles of renal cell carcinomas (RCCs), a newly developed high-resolution wholegenome array analysis was applied. Human CNV370-Duo DNA Analysis BeadChip (genotype 370K) was used in a panel of 22 primary clear cell RCCs (ccRCCs), seven papillary RCCs (PRCCs) (six type II and one type I) and eight chromophobe RCCs (ChRCCs). In ccRCC, a chromosome 3p loss was identified in 95% of the carcinomas, suggesting that 3p loss is the first stage in ccRCC carcinogenesis. Other frequent changes included losses of 1p (23%), 3q (46%) and 8p (32%), and gains of 5q (32%), 7p (27%), 7q (27%) and 1q (23%). The most frequent chromosomal losses in PRCC (43%) were noted in 3p and 3q, followed by 29% of losses of 1p, 1q, 11q, 18q, 22p and 22q, and gains of 20q (57%), 20p (43%), 8q (43%) and 12q (43%). Loss of the entire chromosomes 1, 2, 6, 8, 10, 13 or 17 was noted in patients with ChRCC. A high-density single nucleotide polymorphism array analysis confirmed that partial chromosomal changes rarely occur in ChRCC. Additionally, 32 microdeletions and 10 microamplifications of less than 1 Mb were detected, which may represent potential candidate tumor suppressor genes and oncogenes, respectively.


Introduction
Renal cell carcinomas (RCCs) constitute 2-3% of adult cancers and include clear cell RCC (ccRCC) (75%), papillary RCC (PRCC) (10%) and chromophobe RCC (ChRCC) (5%) subtypes (1,2).Accurate classification of RCCs is crucial, since different histopathologic subtypes require specific therapeutic management strategies due to their markedly different prognoses and responses to therapy (3).The von Hippel-Lindau (VHL) tumor suppressor gene at chromosome 3p25 is inactivated in over 70% of sporadic ccRCCs, and constitutional VHL mutation carriers have a high lifetime risk of developing ccRCC (4).Molecular-targeting drugs for advanced RCC were designed based on the VHL-hypoxia inducible factor (HIF), vascular endothelial growth factor receptor (VEGFR), plateletderived growth factor (PDGF) and transforming growth factor transduction pathways (5), as well as the mammalian target of the rapamycin pathway (6).However, the therapeutic effect of molecular-targeting drugs in PRCC and ChRCC is less clear, since fewer studies are available on the genetic alterations in these subtypes than those in ccRCC.Therefore, newly developed high-density single nucleotide polymorphism (SNP) array was applied to investigate novel and common genetic change in the three RCC subtypes.
A number of cytogenetic studies have identified the

Materials and methods
Tumor samples, normal control and DNA extraction.Primary RCC tumors were surgically resected at Kyushu university Hospital.Samples of carcinoma tissue were frozen in liquid nitrogen immediately after surgery and stored in deep freeze at -80˚C until use.The RCC specimens were histopathologically diagnosed as pure ccRCC, PRCC or ChRCC by a single pathologist.Mixed histopathologic types of RCC were excluded from this study.The proportion of tumor cells in a tissue section was confirmed to be >70% in all of the tumor tissues.The frozen blocks were then subjected to DNA extraction.DNA from peripheral blood was used as the normal control DNA for each patient.DNA extraction was performed according to standard protocols (10).written informed consent was obtained from the patients.The study was approved by the Institutional Review Board.
Whole-genome SNP array analysis.The Human CNV370-Duo DNA Analysis BeadChip system (Illumina, San Diego, CA, uSA) was used, as previously described (11).The allele balance and log 2 ratio (tumor/normal) were visualized using the Genome Browser on BeadStudio ver3 (Illumina).
Statistical analysis.Two-tailed Student's t-tests were used to evaluate the allelic changes and clinical factors, such as stage, tumor grade and prognosis (data not shown).

Results
Fig. 1 shows a representative result using the Human CNV370-Duo DNA Analysis BeadChip system and the allele frequency of each SNP typing from (A) normal DNA, (B) tumor DNA and (C) log 2 ratio (tumor/normal).A comparison of the three lines showed that a focal deletion and partial chromosomal amplification of 5q were identical.Since the chromosomal position of analyzed SNP is available on the public database (http://www.ncbi.nlm.nih.gov/SNP/),genes involved in observed chromosomal aberrations are precisely identified.For example, the 1.45 Mb focal deletion shown in Fig. 1 is flanked by rs104114206 and rs105568833.The deletion contains only one protein-coding gene as a candidate tumor suppressor.

Loss Gain
Copy-neutral LOH, which was undetectable in previous LOH studies, was identified in a total of 37 chromosomal lesions (32 complete and 5 partial losses) (Table I), which is a strong advantage in this SNP array analysis.However, no significant correlations were noted between the allelic changes and clinical factors, such as stage, tumor grade and prognosis (data not shown).

Discussion
The identification of effective multiple tyrosine kinase inhibitors, including sunitinib and sorafenib, is a crucial development in the treatment of metastatic RCC (5).These drugs inactivate the VHL gene, leading to an accumulation of HIF-1α, followed by the activation of VEGF, PDGF and epidermal growth factors (5).However, the effects of these drugs on PRCC and ChRCC are not well known, since the molecular differences among the different histopathological subtypes of RCC have not been intensively studied and recent trials have mostly been restricted to ccRCC patients (12).
In this study, high-resolution SNP array analysis was used to identify the detailed genomic alterations in the three different histopathologic subtypes of RCC.The results demonstrated that the genetic profiles of the three subtypes were inherently different from each other, while 3p loss was the most frequent change in ccRCC, as well as PRCC.This change suggests that the effectiveness of multiple tyrosine kinase inhibitors should be limited to ccRCC.Notably, the treatment effects of sunitinib and sorafenib in a limited number of patients with PRCC and ChRCC showed poorer clinical responses compared to ccRCC (12).The results of this clinical trial were consistent with the genetic backgrounds identified in the present study.

Loss Gain
LOH was 3p (69 cases), followed by chromosome losses at 8p, 6q and 14q, while the most frequent chromosome gains were at 5q (32 cases), including 10 entire 5q amplifications and 21 large amplifications.The results were similar to those of the present study.However, Chen et al only analyzed ccRCCs, and not PRCCs nor ChRCCs.
Few studies have used high-density whole-genomic analysis in PRCC.In a study by Klatte et al (15) a cytogenetic analysis was performed to distinguish between the tumor profiles of type I and type II PRCCs.The authors found that loss of chromosome 1p and 3p, and gain of 5q were exclusively observed in type II PRCC, whereas trisomy 17 was more frequent in type I (15).Analysis of the genetic profiles of the PRCC cases in the present study, which comprised six cases of type II and one case of type I, were compatible with this previous report, and the single case of type I PCRR showing trisomy 17 (Table I).
we analyzed the patterns of regional gain and loss in the eight samples of ChRCC.The patterns of genomic alterations noted in these ChRCCs differed from those in the ccRCCs and PRCCs.Recurrent genomic losses were detected on chromosomes 1, 2, 6, 8, 10, 13, 17 and 21.Previous studies reported a loss of the entire chromosome arm of chromosomes 1, 2, 6 or 10; and 1, 2, 6, 8, 10, 13, 17 or 21 in ChRCC, using conventional CGH (16) and array-based CGH (7), respectively.Our results confirmed these findings for ChRCC using high-density analysis.we also observed few partial chromosomal changes (Table I) or micro-genetic alterations (<1 Mb) (Table III), suggesting that genetic alterations in small regions, rather than entire chromosomal changes, rarely occur in ChRCC.This entire chromosomal loss may be an initial event in the carcinogenesis of ChRCC.
Contamination with normal cells is occasionally a source of error in the genetic analysis of tumor samples.Peiffer et al Table III.The common hemi-chromosomal losses specifically observed in ChRCC.
addressed the effects of tumor heterogeneity and mosaicism on the detection limits and showed that Illumina's SNP array assay was able to detect LOH in tumor samples combined with 67% normal stroma (8).we confirmed that the proportion of tumor cells in a tissue section was over 70%, and therefore postulate that this system provided a reliable means of determining the detailed genetic profile of RCCs.

Figure 1 .
Figure 1.Representative result using the Human CNV-Duo DNA Analysis BeadChip system.SNP typing of (A) normal DNA, (B) tumor DNA and (C) log 2 ratio (tumor/normal) is shown.A focal deletion (↑) and partial chromosomal amplification of 5q (underlined section) are indicated.
only one candidate tumor suppressor gene (data not shown).

Table II .
The chromosomal locations of microamplifications (G) and microdeletions (L).