The incidence and mortality rates of prostate cancer are significantly higher in African-American men when compared with European-American men. We tested the hypothesis that differences in tumor biology contribute to this survival health disparity. Using microarray technology, we obtained gene expression profiles of primary prostate tumors resected from 33 African-American and 36 European-American patients. These tumors were matched on clinical variables. We also evaluated 18 nontumor prostate tissues from seven African-American and 11 European-American patients. The resulting datasets were analyzed for expression differences on the gene and pathway level comparing African-American with European-American patients. Our analysis revealed a significant number of genes, e.g., 162 transcripts at a false-discovery rate of ≤5% to be differently expressed between African-American and European-American patients. Using a disease association analysis, we identified a common relationship of these transcripts with autoimmunity and inflammation. These findings were corroborated on the pathway level with numerous differently expressed genes clustering in immune response, stress response, cytokine signaling, and chemotaxis pathways. Several known metastasis-promoting genes, including autocrine mobility factor receptor, chemokine (C-X-C motif) receptor 4, and matrix metalloproteinase 9, were more highly expressed in tumors from African-Americans than European-Americans. Furthermore, a two-gene tumor signature that accurately differentiated between African-American and European-American patients was identified. This finding was confirmed in a blinded analysis of a second sample set. In conclusion, the gene expression profiles of prostate tumors indicate prominent differences in tumor immunobiology between African-American and European-American men. The profiles portray the existence of a distinct tumor microenvironment in these two patient groups. [Cancer Res 2008;68(3):927–36]

Prostate cancer is the most frequently diagnosed malignancy and the second leading cause of cancer mortality in men older than 40 years in the United States (1). The incidence and mortality rates of prostate cancer vary substantially among different geographic areas and ethnic groups (2). Most notably, African-American men have the highest risk of developing prostate cancer, and due to the manifestation of a more aggressive disease, they have over twice the mortality rate of European-American men (3, 4). Whereas socioeconomic factors contribute to this health disparity (4, 5), they do not fully explain the differences in prostate cancer incidence, aggressiveness, and mortality among different race/ethnic groups in the United States (6, 7).

It has been proposed that differences in tumor biology contribute to the health disparity associated with prostate cancer (4). To date, the contribution of genetic factors to the higher prostate cancer incidence and mortality in African-American men remains controversial (8, 9). Whereas some studies have investigated race/ethnic differences in prostate cancer tumor biology, few differences have been identified (10, 11). Recently, the chromosomal region of 8q24 has been strongly implicated as a site of prostate cancer susceptibility (12, 13). In these studies, men of African ancestry were found to carry at-risk alleles at a higher frequency than men of European ancestry. Those results indicate that genetic factors may account for some of the excess prostate cancer risk that African-American men experience when compared with other race/ethnic groups.

We pursued the hypothesis that the gene expression profiles of prostate tumors from African-American and European-American men may reveal biological differences between the two groups that could explain the more aggressive cancer phenotype in African-American men. Thus, we performed genome-wide gene expression profiling in a large set of tumor samples that were matched on selected clinical variables. The resulting datasets were analyzed on gene and pathway levels to reveal the differences in tumor biology between African-American and European-American patients.

Collection of tissue specimens. A total of 69 fresh-frozen prostate tumors were obtained from the National Cancer Institute Cooperative Prostate Cancer Tissue Resource (CPCTR) and the Department of Pathology at the University of Maryland. All tumors were resected adenocarcinomas that had not received any therapy before prostatectomy. The macrodissected CPCTR tumor specimens (n = 59) were reviewed by a CPCTR-associated pathologist, who confirmed the presence of tumor in the specimens. These tissues were collected between 2002 and 2004 at four different sites, with each site providing tissues from both African-American and European-American patients. Information on race/ethnicity (33 African-Americans and 36 European-Americans) was either extracted from medical records (CPCTR) or obtained through an epidemiologic questionnaire in which race/ethnicity was self-reported (University of Maryland). Only one patient, a European-American, was also Hispanic. Surrounding nontumor prostate tissue was collected from 18 of the recruited patients in this study. Of those, seven were African-American men and 11 were European-American men. We also isolated total RNA from 10 needle biopsy specimens collected from patients at the National Naval Medical Center (one African-American and nine European-Americans) who did not have prostate cancer. From those, we prepared two RNA pools, each representing five patients. Clinicopathologic characteristics of the patients, including age at prostatectomy, histology, Gleason score, pathologic stage, PSA at diagnosis, tumor size, extraprostatic extension, margin involvement, and seminal vesicle invasion were obtained from CPCTR. For University of Maryland cases, this information was extracted from the medical and pathology records, if available. Written informed consent was obtained from all donors. Tissue collection and study design were approved by the institutional review boards of the participating institutions.

Second tumor set for validation. For validation of a two-gene tumor signature that distinguished between African-American and European-American prostate cancer patients, we obtained 39 prostate tumor samples that were collected between 1991 and 1994 (14). The characteristics of this sample set were not revealed to the authors to assure a fully blinded analysis. These samples were processed like all other samples in the study.

RNA extraction. Total RNA was isolated using the TRIZOL reagent according to the manufacturer's instructions (Invitrogen). RNA integrity for each sample was confirmed with the Agilent 2100 Bioanalyzer (Agilent Technologies).

Affymetrix microarrays. RNA labeling and hybridization were performed according to Affymetrix standard protocols. Briefly, 5 μg of total RNA were reverse-transcribed with an oligo (dT) primer that has a T7 RNA polymerase promoter at the 5′ end. Second-strand synthesis was followed by cRNA production with incorporation of biotinylated ribonucleotides using the BioArray high-yield RNA transcript labeling kit T3 from Enzo Life Sciences. The labeled cRNA was fragmented and hybridized to Affymetrix GeneChip HG-U133A 2.0 arrays. This array contains 22,283 probe sets that represent ∼13,000 human genes. Hybridization signals were visualized with phycoerythrin-conjugated streptavidin (Invitrogen) and scanned using a GeneChip Scanner 3000 7G (Affymetrix). In accordance with Minimum Information About a Microarray Experiment guidelines, we deposited the CEL files for the microarray data and additional patient information into the GEO repository.7

The GEO submission accession number is GSE6956.

Data normalization and statistical analysis. All chips were normalized using the robust multichip analysis procedure (15). To generate lists of significantly differently expressed genes, the resulting data set was subjected to the significance analysis of microarray procedure (16). We generated gene lists based on both P values from two-sided t tests and intended false discovery rates (FDR). The FDR calculation followed the method described by Storey and Tibshirani (17). Prediction analysis for microarrays (PAM) was used to classify patients as either African-American or European-American (18). In this analysis, the threshold δ was chosen based on the best compensation for both training error rates and coefficient of variation error rates. Cross-validation was performed by leaving out 10% of samples to determine the appropriate threshold variable in PAM.

Pathway and disease association analyses. The pathway analysis was performed with the in-house WPS software (19). Biological processes and pathways were annotated according to Gene Ontology Biological Processes (GOBP; Gene Ontology Consortium8

) and the BioCarta pathway collection.9 These two annotation methods are complementary because GOBP uses substantially more genes for functional annotation than does BioCarta, but the GOBP terms are not as detailed and well-defined as the pathway mappings in BioCarta. Our database had 16,762 human genes annotated for GOBP and 1,429 genes for BioCarta pathways. A one-sided Fisher's exact test was used to determine which biological processes and pathways had a statistically significant enrichment of differently expressed genes. Genes were included into the pathway analysis based on the FDR (≤30%) of their corresponding probesets on the microarray. If several probesets encoded the same gene, the software recognized this and assured that the gene was counted only once for significance testing at the pathway level. In addition, a permutated P value or FDR for enrichment in a biological process/pathway was calculated with a permutation method. To calculate the permutated P value, randomly selected genes were assigned to a biological process/pathway without changing the total number of genes that were originally assigned to this pathway, and a P value for enrichment was calculated. This procedure was iterated 300 times, and the permutated P value for enrichment was calculated as the fraction of random trials resulting in a Fisher's exact test P value less than the original Fisher's exact P value without the permutations. We then compiled the Fisher's exact test results for cluster analyses and displayed the results in color-coded heat maps to reveal the patterns of significantly altered biological processes and pathways. The color coding of the heat maps is related to the enrichment of genes in a pathway [−log(P value)–based] with red indicating a higher enrichment.

The disease association analysis was conducted using both our WPS software and information provided by the genetic association database (20). The WPS software assessed whether differently expressed genes (FDR, ≤30%) between prostate tumors from African-American and European-American patients had previously been associated with other diseases, as indicated by the genetic association database, and whether these disease-associated genes, e.g., systemic lupus erythematosus genes, were statistically significantly enriched in the prostate tumor gene signature using the one-sided Fisher's exact test. In addition, we calculated a permutated P value or FDR for disease association of the differently expressed genes. The permutated P value was calculated following the procedures described in the previous paragraph. The color coding of the heat map that displays the results of the disease association analysis is related to the enrichment of disease genes [−log(P value)–based], e.g., systemic lupus erythematosus genes, among the differently expressed genes between prostate tumors from African-American and European-American patients.

Quantitative real-time PCR. One hundred nanograms of total RNA were reverse-transcribed using the high-capacity cDNA archive kit (Applied Biosystems). Quantitative real-time PCR (qRT-PCR) was subsequently performed in triplicate using TaqMan Gene Expression Assays (Applied Biosystems), which include preoptimized probe and primer sets specific for the genes being validated. The assay ID numbers of the validated genes are as follows: phosphoserine phosphatase homologue (PSPHL), Hs00863464_m1; matrix metallopeptidase 9 (MMP9), Hs00234579_m1; autocrine mobility factor receptor (AMFR), Hs00181609_m1; and chemokine (C-X-C motif) receptor 4 (CXCR4), Hs00607978_s1. Data were collected using the ABI PRISM 7500 sequence detection system. The 18s RNA was used as the internal standard reference. Normalized expression was calculated using the comparative Ct method as described, and fold changes were derived from the 2−ΔΔCt values for each gene (21). Graphs were prepared using relative Ct values that were calculated by subtracting the 18s Ct values from the corresponding Ct values for the gene being measured. For illustration purposes only, an arbitrary value of 20 was uniformly added to all relative Ct values. The statistical analysis of these data was performed with a two-sided t test if the expression data showed normal distribution or with a two-sided Wilcoxon rank-sum test if the expression data did not follow normal distribution.

Tumor selection and gene expression analysis. Prostate tumors were collected from 33 African-American and 36 European-American patients with localized disease. Tumor samples were similar for age at prostatectomy, pathologic stage of the disease, and Gleason score between the African-American and European-American patients (Table 1). African-American patients tended to have higher occurrence of seminal vesicle invasion than European-American patients, which is consistent with the literature (22).

Table 1.

Clinical characteristics of the study population

All cases (n = 69)African-American (n = 33)European-American (n = 36)P* (t test)
Age at prostatectomy [median (range)] n = 69 60 (44–73) 61 (46–72) 60 (44–73) 0.77 
PSA at diagnosis [median (range)] n = 50 6.1 (1.3–47.7) 6.1 (1.3–47.7) 6.1 (4.0–20.0) 0.23 
Largest nodule (g) [median (range)] n = 59 1.6 (0.2–2.9) 1.5 (0.8–2.9) 1.6 (0.2–2.8) 0.97 
     

 
N (%)
 
N (%)
 
N (%)
 
P (Fisher's exact test)
 
Source of tissue     
    National Cancer Institute CPCTR, n (%) 59 (86) 30 (90) 29 (81) 0.31 
    University of Maryland, n (%) 10 (14) 3 (10) 7 (19)  
Pathologic stage     
    pT2, n (%) 38 (55) 18 (55) 20 (55) 1.0 
    ≥pT3, n (%) 31 (44) 15 (45) 16 (42)  
Gleason sum score     
    <7 (5–6), n (%) 18 (26) 9 (27) 9 (25) 1.0 
    ≥7 (7–9), n (%) 51 (74) 24 (73) 27 (75)  
Seminal vesicle invasion     
    No, n (%) 49 (83) 22 (73) 27 (93) 0.08 
    Yes, n (%) 10 (17) 8 (27) 2 (7)  
Surgical margin status     
    Negative, n (%) 35 (59) 18 (60) 17 (59) 1.0 
    Positive, n (%) 24 (41) 12 (40) 12 (41)  
All cases (n = 69)African-American (n = 33)European-American (n = 36)P* (t test)
Age at prostatectomy [median (range)] n = 69 60 (44–73) 61 (46–72) 60 (44–73) 0.77 
PSA at diagnosis [median (range)] n = 50 6.1 (1.3–47.7) 6.1 (1.3–47.7) 6.1 (4.0–20.0) 0.23 
Largest nodule (g) [median (range)] n = 59 1.6 (0.2–2.9) 1.5 (0.8–2.9) 1.6 (0.2–2.8) 0.97 
     

 
N (%)
 
N (%)
 
N (%)
 
P (Fisher's exact test)
 
Source of tissue     
    National Cancer Institute CPCTR, n (%) 59 (86) 30 (90) 29 (81) 0.31 
    University of Maryland, n (%) 10 (14) 3 (10) 7 (19)  
Pathologic stage     
    pT2, n (%) 38 (55) 18 (55) 20 (55) 1.0 
    ≥pT3, n (%) 31 (44) 15 (45) 16 (42)  
Gleason sum score     
    <7 (5–6), n (%) 18 (26) 9 (27) 9 (25) 1.0 
    ≥7 (7–9), n (%) 51 (74) 24 (73) 27 (75)  
Seminal vesicle invasion     
    No, n (%) 49 (83) 22 (73) 27 (93) 0.08 
    Yes, n (%) 10 (17) 8 (27) 2 (7)  
Surgical margin status     
    Negative, n (%) 35 (59) 18 (60) 17 (59) 1.0 
    Positive, n (%) 24 (41) 12 (40) 12 (41)  
*

P value for the difference between African-Americans and European-Americans. All tests were two-sided.

Cases with unknown status are not included.

One European-American patient was staged pT4.

Gene expression profiles from the African-American and European-American prostate tumors were generated using the Affymetrix GeneChip HG-U133A 2.0 array that represents ∼13,000 human genes. In addition, 18 nontumor tissues and two RNA pools from cancer-free prostates were used for genetic comparisons. In an initial analysis of our dataset, we generated the gene expression profiles from tumor and nontumor tissues using the combined dataset (69 tumors and 20 nontumor samples) and obtained the gene signature that differentiated tumor and nontumor tissues. For cross-validation of our dataset, we compared these results with those from a published meta-analysis of four prostate cancer gene expression datasets (23). Despite the use of different technology platforms, we found a very good agreement between our results and the results of the meta-analysis. Of the top 40 overexpressed genes in prostate tumors across the meta-analysis, 21 were also found to be significantly up-regulated (FDR, ≤5%) in our study. Of the top 40 underexpressed genes in prostate tumors across the meta-analysis, six were significantly down-regulated (FDR, ≤5%) in our study (Supplementary Table S1). Specifically, previously validated marker genes for prostate tumors, such as the genes encoding α-methylacyl-CoA racemase, hepsin, and fatty acid synthase, were highly up-regulated in the tumors of our dataset. Signature tumor genes of other studies, such as MUC1 and HOXC6 (24, 25), were also found be overexpressed in these tumors.

The comparison of gene expression profiles from African-American and European-American tumors revealed 162 transcripts, 280 transcripts, and 489 transcripts, to be differently expressed between the two groups at FDRs of ≤5%, ≤ 10%, and ≤ 30%, respectively (Supplementary Table S2). We compared this gene list with the list of genes that were described to be differently expressed between tumor and nontumor tissue in the aforementioned meta-analysis (23). Of the top 80 tumor genes in the meta-analysis, none was found to be differently expressed between African-American patients and European-American patients at the FDR of ≤5% threshold. Only one of those 80 genes, the gene encoding fibrillin 1, was differently expressed at a less stringent FDR of ≤30%. These data suggest that previously identified marker genes for prostate cancer do not significantly differ between African-American and European-American prostate tumors.

Disease and pathway association of transcripts that are differently expressed by race/ethnicity. Bioinformatics enables us to determine common associations between a gene list and disease classes and pathways. We used our in-house software (19) to assess disease and pathway associations of genes that were differently expressed in tumors by race/ethnicity. The disease association analysis revealed associations of these genes with genetic pathways that are linked to autoimmune diseases, allergies, and inflammatory diseases (Supplementary Fig. S1). The enrichment of disease genes among the differently expressed genes was statistically significant, e.g., inflammatory urogenital disease genes (permutated P < 0.001), skin-prick test genes (permutated P = 0.003), systemic lupus erythematosus genes (permutated P = 0.006), and asthma (permutated P = 0.01). Autoimmune disease modulators, such as PTPN22 and components of the HLA complex, and key genes in antigen presentation, such as TAP1 and TAP2, were higher expressed in African-American tumors. In contrast, genes differently expressed between tumor and nontumor tissue, and between tumors with high and low Gleason sum score, showed common associations with other epithelial cancers as one may expect (Supplementary Fig. S1).

These observations were further corroborated by pathway analyses using GO biological processes and BioCarta annotations. Differently expressed genes between tumors of African-Americans and European-Americans clustered uniquely in GO biological processes (Fig. 1) and BioCarta pathways (Supplementary Fig. S2) that are associated with immune response, defense response, antigen presentation, B-cell/T-cell function, cytokine signaling, chemotaxis, and inflammatory response. The clustering of genes in these biological processes was highly significant (Table 2), and immune response genes were enriched at the P < 1.0 × 10−30 significance level. The enrichment of differently expressed genes in the 20 highest-ranked GOBP terms (Table 2) had a low FDR (<1% for each term). Most commonly, the involved genes were more highly expressed in tumors of African-American patients when compared with those of European-American patients, as detailed by an example for the immune response biological process (Supplementary Table S3). An analogue analysis of nontumor tissue did not generate these biological process/pathway associations, suggesting that this gene signature is derived from the tumor microenvironment (Supplementary Fig. S3).

Figure 1.

GOBP terms that are uniquely enriched for differently expressed genes comparing prostate tumors from African-Americans (AA; n = 33) with tumors from European-Americans (EA; n = 36). For comparison, we also show the contrast of tumor (n = 69) versus nontumor (n = 20) and the contrast of high (n = 51) versus low (n = 18) Gleason sum score. Gleason sum score was dichotomized into high (having a sum score of 7–9) and low (having a sum score of 5–6). The results of our analysis are displayed by a heat map with the red color indicating an enrichment of differently expressed genes in a GOBP term.

Figure 1.

GOBP terms that are uniquely enriched for differently expressed genes comparing prostate tumors from African-Americans (AA; n = 33) with tumors from European-Americans (EA; n = 36). For comparison, we also show the contrast of tumor (n = 69) versus nontumor (n = 20) and the contrast of high (n = 51) versus low (n = 18) Gleason sum score. Gleason sum score was dichotomized into high (having a sum score of 7–9) and low (having a sum score of 5–6). The results of our analysis are displayed by a heat map with the red color indicating an enrichment of differently expressed genes in a GOBP term.

Close modal
Table 2.

Twenty highest-ranked GOBP terms enriched for differently expressed genes comparing tumors from African-American and European-American patients

GOBP termTerm hits*All genesAnnotated genes for termAll annotated genes§P (Fisher's exact test)
Immune response 66 217 944 16,762 5.66E-31 
Defense response 68 217 1,045 16,762 4.38E-30 
Response to biotic stimulus 70 217 1,190 16,762 2.44e−28 
Organismal physiologic process 74 217 2,111 16,762 1.46e−16 
Response to stimulus 77 217 2,281 16,762 2.23e−16 
Response to pest/pathogen or parasite 27 217 472 16,762 9.70e−11 
Humoral immune response 16 217 162 16,762 3.38e−10 
Response to external biotic stimulus 27 217 502 16,762 3.83e−10 
Humoral defense mechanism 13 217 114 16,762 2.75e−09 
Response to stress 34 217 929 16,762 3.76e−08 
Antigen processing 217 45 16,762 9.90e−08 
Antigen processing via MHC class I 217 14 16,762 6.32e−07 
Antigen presentation 217 44 16,762 1.41e−06 
Antimicrobial humoral response 217 85 16,762 1.52e−06 
Antimicrobial humoral response 217 88 16,762 2.04e−06 
Antigen presentation 217 13 16,762 1.78e−05 
Cellular defense response 217 93 16,762 2.75e−05 
Signal transduction 67 217 3,465 16,762 2.40e−04 
Cell communication 77 217 4,177 16,762 3.11e−04 
Apoptosis 16 217 458 16,762 3.20e−04 
GOBP termTerm hits*All genesAnnotated genes for termAll annotated genes§P (Fisher's exact test)
Immune response 66 217 944 16,762 5.66E-31 
Defense response 68 217 1,045 16,762 4.38E-30 
Response to biotic stimulus 70 217 1,190 16,762 2.44e−28 
Organismal physiologic process 74 217 2,111 16,762 1.46e−16 
Response to stimulus 77 217 2,281 16,762 2.23e−16 
Response to pest/pathogen or parasite 27 217 472 16,762 9.70e−11 
Humoral immune response 16 217 162 16,762 3.38e−10 
Response to external biotic stimulus 27 217 502 16,762 3.83e−10 
Humoral defense mechanism 13 217 114 16,762 2.75e−09 
Response to stress 34 217 929 16,762 3.76e−08 
Antigen processing 217 45 16,762 9.90e−08 
Antigen processing via MHC class I 217 14 16,762 6.32e−07 
Antigen presentation 217 44 16,762 1.41e−06 
Antimicrobial humoral response 217 85 16,762 1.52e−06 
Antimicrobial humoral response 217 88 16,762 2.04e−06 
Antigen presentation 217 13 16,762 1.78e−05 
Cellular defense response 217 93 16,762 2.75e−05 
Signal transduction 67 217 3,465 16,762 2.40e−04 
Cell communication 77 217 4,177 16,762 3.11e−04 
Apoptosis 16 217 458 16,762 3.20e−04 
*

Annotated genes in a GOBP term that are differently expressed (FDR, ≤30%) comparing tumors from African-American with those from European-American.

All GOBP-annotated genes that are differently expressed in this comparison.

All annotated genes in a GOBP term.

§

All GOBP-annotated genes.

Differences in expression of metastasis-related genes. Next, we directed our attention to specific genes that were most significantly differently expressed in prostate tumors (fold change, ≥1.5; FDR, ≤5%) comparing African-American with European-American patients. This list included 47 genes (Table 3). The two genes that had the greatest expression differences were PSPHL (4.64-fold; P = 1.0 × 10−11), a homologue of a putative phosphoserine phosphatase (26), and CRYBB2 (3.02-fold; P = 1.3 × 10−11). The list also contained a number of genes that encode chemokines and chemokine receptors. Several of those, including the genes encoding CXCR4 and chemokine receptor 7 (CCR7), have been previously linked to cancer metastasis (27, 28). These and other metastasis-associated genes, e.g., AMFR and MMP9, were more highly expressed in tumors of African-American patients than those of European-Americans (Table 3). Three metastasis-associated genes were chosen for validation via qRT-PCR. Consistent with the microarray data, we found a higher expression of AMFR (1.5-fold), CXCR4 (1.8-fold), and MMP9 (2.0-fold) in primary tumors of African-Americans (Fig. 2) when compared with the tumors from European-Americans.

Table 3.

Most differently expressed tumor genes by race/ethnicity

Genes higher expressed in prostate tumors of African-American patients (at least 1.5-fold)
No.Gene symbolGenbank IDGene titleFold change*
ADAMDEC1 NM_014479 ADAM-like, decysin 1 1.81 
AMFR NM_001144 Autocrine motility factor receptor 1.9 
APOLD1 NM_030817 Apolipoprotein L domain containing 1 1.56 
BRDG1 NM_012108 BCR downstream signaling 1 2.06 
C1orf38 AB035482 Chromosome 1 open reading frame 38 1.58 
 C1orf38 NM_004848 Chromosome 1 open reading frame 38 1.51 
C8orf4 NM_020130 Chromosome 8 open reading frame 4 1.71 
CCL5 M21121 Chemokine (C-C motif) ligand 5 1.61 
 CCL5 NM_002985 Chemokine (C-C motif) ligand 5 1.51 
CCR7 NM_001838 Chemokine (C-C motif) receptor 7 1.66 
CD52 N90866 CD52 molecule 1.7 
10 CRYBB2 NM_000496 crystallin, β B2 3.02 
11 CXCR4 AJ224869 Chemokine (C-X-C motif) receptor 4 1.72 
 CXCR4 AF348491 Chemokine (C-X-C motif) receptor 4 1.66 
 CXCR4 L01639 Chemokine (C-X-C motif) receptor 4 1.65 
12 DNAJC15 NM_013238 DnaJ (Hsp40) homologue, subfamily C 1.64 
13 DUSP2 NM_004418 Dual specificity phosphatase 2 1.53 
14 EBI2 NM_004951 EBV-induced gene 2 1.56 
15 GZMA NM_006144 Granzyme A 1.63 
16 IFI44L NM_006820 IFN-induced protein 44-like 1.63 
17 IGH M87789 Immunoglobulin heavy locus 1.78 
18 IGHM BC001872 Immunoglobulin heavy constant mu 2.37 
19 IGJ AV733266 Immunoglobulin J polypeptide 2.02 
20 IGKC L14458 Immunoglobulin κ constant 1.88 
 IGKC BG485135 Immunoglobulin κ constant 1.67 
 IGKC M63438 Immunoglobulin κ constant 1.6 
 IGKC BC005332 Immunoglobulin κ constant 1.59 
21 IGKV1D-13 AW408194 Immunoglobulin κ variable 1D-13 1.73 
22 IGL M87790 Immunoglobulin λ locus 1.82 
 IGL X57812 Immunoglobulin λ locus 1.81 
 IGL AA680302 Immunoglobulin λ locus 1.74 
 IGL AV698647 Immunoglobulin λ locus 1.67 
23 IL7R NM_002185 Interleukin 7 receptor 1.61 
24 INDO M34455 Indoleamine-pyrrole 2,3 dioxygenase 1.64 
25 ISG15 NM_005101 ISG15 ubiquitin-like modifier 1.6 
26 ITK D13720 IL2-inducible T-cell kinase 1.76 
27 LOC283970 AI925734 Hypothetical protein LOC283970 1.53 
28 LOC339562 M20812 Similar to Igκ chain V-I region Walker precursor 1.64 
 LOC651629 AW404894 Similar to Igκ chain V-I region Walker precursor 1.91 
 LOC652745 AJ408433 Similar to Igκ chain V-I region Walker precursor 2.12 
29 LOC96610 X79782 Hypothetical protein similar to KIAA0187 1.57 
30 LYZ AV711904 Lysozyme (renal amyloidosis) 1.61 
31 MMP9 NM_004994 Matrix metallopeptidase 9 1.86 
32 PI15 NM_015886 Peptidase inhibitor 15 1.63 
33 PSCDBP L06633 Pleckstrin homology, Sec7 1.87 
34 PSPHL NM_003832 Phosphoserine phosphatase-like 4.64 
35 PTPRC NM_002838 Protein tyrosine phosphatase, receptor type, C 1.6 
36 RASA4 AB011110 RAS p21 protein activator 4 1.5 
37 SF1 D26121 Splicing factor 1 2.17 
38 STAT1 BC002704 Signal transducer and activator of transcription 1 1.71 
 STAT1 M97935_MB Signal transducer and activator of transcription 1 1.51 
39 TRBV19 M15564 T-cell receptor β variable 19 1.64 
 TRBV19 AL559122 T-cell receptor β variable 19 1.57 
40 TRBV21-1 AF043179 T-cell receptor β variable 21-1 1.58 
     
Genes lower expressed in prostate tumors of African-American patients (0.7-fold or less)
 
    
No.
 
Gene symbol
 
Genbank ID
 
Gene title
 
Fold change
 
ADRB1 NM_000684 Adrenergic receptor, β-1 0.63 
ARL17P1 AF119889 ADP-ribosylation factor-like 17 pseudogene 1 0.67 
EIF2S3 NM_001415 Eukaryotic translation initiation factor 2 0.65 
LGALS8 AF342815 Lectin, galactoside-binding, soluble, 8 0.7 
 LGALS8 AF342816 Lectin, galactoside-binding, soluble, 8 0.58 
PLGLB2 BC005379 Plasminogen-like B2 0.66 
SMA3 NM_006780 SMA3 0.7 
 SMA3 X83301 SMA3 0.61 
ZNF654 NM_018293 Zinc finger protein 654 0.7 
Genes higher expressed in prostate tumors of African-American patients (at least 1.5-fold)
No.Gene symbolGenbank IDGene titleFold change*
ADAMDEC1 NM_014479 ADAM-like, decysin 1 1.81 
AMFR NM_001144 Autocrine motility factor receptor 1.9 
APOLD1 NM_030817 Apolipoprotein L domain containing 1 1.56 
BRDG1 NM_012108 BCR downstream signaling 1 2.06 
C1orf38 AB035482 Chromosome 1 open reading frame 38 1.58 
 C1orf38 NM_004848 Chromosome 1 open reading frame 38 1.51 
C8orf4 NM_020130 Chromosome 8 open reading frame 4 1.71 
CCL5 M21121 Chemokine (C-C motif) ligand 5 1.61 
 CCL5 NM_002985 Chemokine (C-C motif) ligand 5 1.51 
CCR7 NM_001838 Chemokine (C-C motif) receptor 7 1.66 
CD52 N90866 CD52 molecule 1.7 
10 CRYBB2 NM_000496 crystallin, β B2 3.02 
11 CXCR4 AJ224869 Chemokine (C-X-C motif) receptor 4 1.72 
 CXCR4 AF348491 Chemokine (C-X-C motif) receptor 4 1.66 
 CXCR4 L01639 Chemokine (C-X-C motif) receptor 4 1.65 
12 DNAJC15 NM_013238 DnaJ (Hsp40) homologue, subfamily C 1.64 
13 DUSP2 NM_004418 Dual specificity phosphatase 2 1.53 
14 EBI2 NM_004951 EBV-induced gene 2 1.56 
15 GZMA NM_006144 Granzyme A 1.63 
16 IFI44L NM_006820 IFN-induced protein 44-like 1.63 
17 IGH M87789 Immunoglobulin heavy locus 1.78 
18 IGHM BC001872 Immunoglobulin heavy constant mu 2.37 
19 IGJ AV733266 Immunoglobulin J polypeptide 2.02 
20 IGKC L14458 Immunoglobulin κ constant 1.88 
 IGKC BG485135 Immunoglobulin κ constant 1.67 
 IGKC M63438 Immunoglobulin κ constant 1.6 
 IGKC BC005332 Immunoglobulin κ constant 1.59 
21 IGKV1D-13 AW408194 Immunoglobulin κ variable 1D-13 1.73 
22 IGL M87790 Immunoglobulin λ locus 1.82 
 IGL X57812 Immunoglobulin λ locus 1.81 
 IGL AA680302 Immunoglobulin λ locus 1.74 
 IGL AV698647 Immunoglobulin λ locus 1.67 
23 IL7R NM_002185 Interleukin 7 receptor 1.61 
24 INDO M34455 Indoleamine-pyrrole 2,3 dioxygenase 1.64 
25 ISG15 NM_005101 ISG15 ubiquitin-like modifier 1.6 
26 ITK D13720 IL2-inducible T-cell kinase 1.76 
27 LOC283970 AI925734 Hypothetical protein LOC283970 1.53 
28 LOC339562 M20812 Similar to Igκ chain V-I region Walker precursor 1.64 
 LOC651629 AW404894 Similar to Igκ chain V-I region Walker precursor 1.91 
 LOC652745 AJ408433 Similar to Igκ chain V-I region Walker precursor 2.12 
29 LOC96610 X79782 Hypothetical protein similar to KIAA0187 1.57 
30 LYZ AV711904 Lysozyme (renal amyloidosis) 1.61 
31 MMP9 NM_004994 Matrix metallopeptidase 9 1.86 
32 PI15 NM_015886 Peptidase inhibitor 15 1.63 
33 PSCDBP L06633 Pleckstrin homology, Sec7 1.87 
34 PSPHL NM_003832 Phosphoserine phosphatase-like 4.64 
35 PTPRC NM_002838 Protein tyrosine phosphatase, receptor type, C 1.6 
36 RASA4 AB011110 RAS p21 protein activator 4 1.5 
37 SF1 D26121 Splicing factor 1 2.17 
38 STAT1 BC002704 Signal transducer and activator of transcription 1 1.71 
 STAT1 M97935_MB Signal transducer and activator of transcription 1 1.51 
39 TRBV19 M15564 T-cell receptor β variable 19 1.64 
 TRBV19 AL559122 T-cell receptor β variable 19 1.57 
40 TRBV21-1 AF043179 T-cell receptor β variable 21-1 1.58 
     
Genes lower expressed in prostate tumors of African-American patients (0.7-fold or less)
 
    
No.
 
Gene symbol
 
Genbank ID
 
Gene title
 
Fold change
 
ADRB1 NM_000684 Adrenergic receptor, β-1 0.63 
ARL17P1 AF119889 ADP-ribosylation factor-like 17 pseudogene 1 0.67 
EIF2S3 NM_001415 Eukaryotic translation initiation factor 2 0.65 
LGALS8 AF342815 Lectin, galactoside-binding, soluble, 8 0.7 
 LGALS8 AF342816 Lectin, galactoside-binding, soluble, 8 0.58 
PLGLB2 BC005379 Plasminogen-like B2 0.66 
SMA3 NM_006780 SMA3 0.7 
 SMA3 X83301 SMA3 0.61 
ZNF654 NM_018293 Zinc finger protein 654 0.7 

NOTE: FDR ≤ 5% for all genes in the list.

*

Fold change for annotated probesets with European-American patients as the reference.

Figure 2.

qRT-PCR expression analysis of AMFR (A), CXCR4 (B), and MMP9 (C) in prostate tumors of African-Americans and European-Americans. Points, relative Ct values for the individual samples; bars, median value for the sample set. The genes are significantly higher expressed in African-American patients than in European-American patients; fold change = 1.52, P = 0.033 (two-sided t test) for AMFR; fold change = 1.75, P = 0.027 (two-sided t test) for CXCR4; fold change = 2.00, P = 0.044 (two-sided Wilcoxon rank-sum test) for MMP9. Graphs were prepared using relative Ct values that were calculated by subtracting the 18s Ct values from the corresponding Ct values for the gene being measured (see Subjects and Methods). Fold changes were derived from the 2−ΔΔCt values for each gene.

Figure 2.

qRT-PCR expression analysis of AMFR (A), CXCR4 (B), and MMP9 (C) in prostate tumors of African-Americans and European-Americans. Points, relative Ct values for the individual samples; bars, median value for the sample set. The genes are significantly higher expressed in African-American patients than in European-American patients; fold change = 1.52, P = 0.033 (two-sided t test) for AMFR; fold change = 1.75, P = 0.027 (two-sided t test) for CXCR4; fold change = 2.00, P = 0.044 (two-sided Wilcoxon rank-sum test) for MMP9. Graphs were prepared using relative Ct values that were calculated by subtracting the 18s Ct values from the corresponding Ct values for the gene being measured (see Subjects and Methods). Fold changes were derived from the 2−ΔΔCt values for each gene.

Close modal

Further analyses were performed to determine the expression of these genes in tumor and surrounding nontumor tissue. The data showed that CXCR4 was differently expressed between tumor and surrounding nontumor tissue in African-Americans, but not in European-Americans (Supplementary Table S4). None of the other genes followed this pattern. Instead, these genes seemed to be higher expressed in both tumor and surrounding nontumor tissue of African-American patients when compared with the European-American patients (Supplementary Table S4). Additional pathway analysis reiterated the potential significance of CXCR4 in prostate cancer biology among the African-American patients. That analysis identified the CXCR4 pathway as the highest-ranked pathway with the most significant enrichment of differently expressed genes comparing tumor with surrounding nontumor tissue among the African-American patients (Supplementary Table S5).

A two-gene predictor is associated with race/ethnicity. We interrogated our datasets for genes that can differentiate between African-Americans and European-Americans using the gene expression profile of prostate tumors and surrounding nontumor tissue. We identified a two-gene signature that accurately differentiated between these two groups. These two genes were PSPHL (probeset 205048_s_at) and CRYBB2 (probeset 206777_s_at). Based on the expression of these two genes in prostate tumors, PAM classified correctly 30 of 33 African-Americans (91%) and 34 of 36 European-Americans (94%). We could not obtain this accuracy with the two-gene signature using the gene expression profile of the nontumor tissue. To achieve similar prediction accuracy (86% for African-Americans and 100% for European-Americans) with the gene expression profile of nontumor tissue, a classifier consisting of 19 probesets was required (data not shown). We confirmed the ability of PSPHL and CRYBB2 expression levels to predict race using a blinded analysis of 39 additional tumors (14). The two-gene classifier correctly predicted 30 of 34 African-Americans (88%) and 5 of 5 patients from Italy into the European-American category (100%).

Because the microarray data indicated a greater heterogeneity of PSPHL expression in nontumor tissue than tumor tissue, we performed qRT-PCR validation of the PSPHL expression data (Fig. 3). qRT-PCR results revealed a bimodal distribution of the PSPHL expression in prostate tumors resulting in a 160-fold difference in gene expression between African-American and European-American patients. In the surrounding nontumor tissue, this difference was 38-fold. Furthermore, PSPHL was found to be up-regulated 84-fold in tumors of African-Americans when compared with the surrounding nontumor tissues from this population. These fold change values should be considered estimates. Because of the low expression of PSPHL in prostate tissue from European-Americans, accurate determination of PSPHL baseline expression was difficult. Nevertheless, the qRT-PCR results unequivocally validated the microarray data by showing that PSPHL is significantly higher expressed in tumors of African-Americans than in those of European-Americans.

Figure 3.

qRT-PCR expression analysis of PSPHL in prostate tumors (T) and surrounding nontumor tissue (N) of African-Americans and European-Americans. Points, relative Ct-values for the individual samples; bars, median value for the sample set. PSPHL is significantly higher expressed in surrounding nontumor tissue (fold change = 38, P = 0.012; two-sided Wilcoxon rank-sum test) and tumor tissue (fold change = 160, P < 0.001; two-sided Wilcoxon rank-sum test) of African-American patients when compared with European-American patients. Graphs were prepared using relative Ct values that were calculated by subtracting the 18s Ct values from the corresponding Ct values for the gene being measured (see Subjects and Methods). Fold changes were derived from the 2−ΔΔCt values for each gene.

Figure 3.

qRT-PCR expression analysis of PSPHL in prostate tumors (T) and surrounding nontumor tissue (N) of African-Americans and European-Americans. Points, relative Ct-values for the individual samples; bars, median value for the sample set. PSPHL is significantly higher expressed in surrounding nontumor tissue (fold change = 38, P = 0.012; two-sided Wilcoxon rank-sum test) and tumor tissue (fold change = 160, P < 0.001; two-sided Wilcoxon rank-sum test) of African-American patients when compared with European-American patients. Graphs were prepared using relative Ct values that were calculated by subtracting the 18s Ct values from the corresponding Ct values for the gene being measured (see Subjects and Methods). Fold changes were derived from the 2−ΔΔCt values for each gene.

Close modal

The causes of the prostate cancer health disparity that exists between African-American and European-American men remain to be fully elucidated. Much of the research on this disparity has focused on socioeconomic factors. To date, very few studies have examined differences in tumor biology despite suggestive clinical evidence that they exist.

We analyzed the gene expression profiles of 69 tumors from 33 African-American and 36 European-American patients. Highly significant differences in the expression of genes related to immunobiology within the tumor microenvironment were identified between these two groups. This interpretation was supported by both disease association and pathway analyses. Most of the immune-related genes were higher expressed in tumors from African-American patients than in those from European-American patients. Although preliminary, these findings are novel and could have implications for cancer therapy. Clinical trials of immunotherapy for prostate cancer have been conducted (29), and these therapies may enter clinical practice in the near future. Our data suggest that African-American patients and European-American patients might respond differently to these types of therapy.

We observed that previously recognized diagnostic and prognostic marker genes for prostate tumors (2325) were not significantly differently expressed by race/ethnicity. The findings suggest that the most common alterations in gene expression in prostate carcinogenesis do not develop any differently in African-Americans than in European-Americans. This result corroborates published data that showed a similar pattern of chromosomal alterations in prostate tumors of African-American and European-American men (8).

Several metastasis-associated genes were found to be more highly expressed in tumors of African-American patients when compared with European-American patients. This group of genes contained AMFR, CXCR4, CCR7, and MMP9, among others. To the best of our knowledge, none of those genes have been previously recognized as a prostate cancer marker gene by examination of gene expression profiles. Unfortunately, most previous gene expression profiling studies do not disclose the race/ethnic make up of the patients when describing the sample sets used (3033). It is likely that the majority of the studied tumors were obtained from European-American patients. Thus, genes that are dysregulated in prostate tumors of African-American patients may have been overlooked until now. However, multiple studies have shown that both CXCR4 and MMP9 are key metastasis regulators in prostate cancer (27, 3437), and both genes are linked through a common pathway (38). The causal role of CXRC4 and also CCR7 in cancer metastasis is well demonstrated (3941), and therapeutic approaches to target these receptors are being developed (42).

From the present study, we do not know why prostate tumors from African-American patients would have a different immunologic profile than tumors from European-American patients. We hypothesize that the causes of these differences are multifactorial. The nature of the differently expressed transcripts indicates the involvement of various cell types in the generation of the gene signature, e.g., B cells, T cells, and tumor-infiltrating macrophages. An association of those cells with inflammation in prostate carcinogenesis has been recognized (43). To investigate the contribution from B-cell and T-cell infiltration, we evaluated B-cell and T-cell density of 48 prostate tumors (23 African-American and 25 European-American) in a pilot study. No significant difference in tumor B-cell and T-cell infiltration was found between the two race/ethnic groups (data not shown). These preliminary data argue that the infiltrating B-cell and T-cell populations could be functionally different in the African-American and European-American tumors. Possible causes contributing to the different immunologic profiles could be environmental factors, e.g., infections, genetic factors, e.g., genetic variations in immune regulatory genes, or the interactions of both. There is evidence that infectious diseases are associated with prostate carcinogenesis (43, 44), and two genes involved in innate immunity, MSR1 and RNASEL, are candidate susceptibility genes for prostate cancer (45, 46). Notably, the gene signature that was generated in this study comparing prostate tumors from African-American and European-American men revealed many up-regulated IFN-related genes in the African-American prostate tumors. This distinctive signature could be indicative of a possible viral involvement in this African-American population.

Chronic inflammation is also thought to be a causative factor in prostate carcinogenesis (43, 47). Studies have suggested that inflammation is more prevalent in nontumor prostate biopsy specimens from African-American men when compared with European-American men (48). Other investigations have shown that an immune response signature in the liver of cancer patients predicts metastasis and recurrence of hepatocellular carcinoma (49). Thus, future studies should evaluate whether the immunologic profile of prostate tumors in African-American men is a predisposing factor for tumor progression and metastasis.

Our investigations identified a two-gene tumor signature that accurately differentiated between African-American and European-American patients. Little is known about the functions of the two genes, PSPHL and CRYBB2, in this classifier. PSPHL was first isolated from Fanconi anemia fibroblasts in a genetic screen and was identified as a homologue of a putative phosphoserine phosphatase (26). Interestingly, PSPHL is located on chromosome 7q11.2, a chromosomal region known to have gain of function related to advanced tumor stage in prostate cancer (50). Despite its chromosomal location, there are no studies linking PSPHL expression to cancer progression. Our data also showed that PSPHL is higher expressed in prostate tumors than the surrounding nontumor tissue. Thus, the possible role of PSPHL in contributing to prostate cancer susceptibility will have to be tested in future studies.

Whole-genome gene expression analysis experiments are prone to findings that are either unique to a selected patient population or are artificially created by the applied technology. To exclude the possibility of an artifact, three approaches were used to cross-validate our gene expression data. First, we compared our results of the differently expressed genes between tumor and surrounding nontumor tissue with those of other studies and found very good agreement (2325, 33). Second, we validated the expression of several key genes via qRT-PCR and confirmed that these genes are higher expressed in tumors from African-Americans than from European-Americans. Lastly, we used prediction analysis on a blinded dataset and corroborated that a two-gene signature can distinguish between African-American patients and European-American patients. Despite all of the cross-validation approaches used to strengthen our results, we recognize that the findings of the current study are preliminary and will have to be further investigated, including the immunohistochemical analysis of larger sample sets.

In conclusion, the gene expression profile of prostate tumors corresponds to differences in tumor biology between African-American and European-American men. The implications of those differences in disease aggressiveness and response to therapy should be evaluated in future studies.

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

Grant support: Intramural Research Program of NIH National Cancer Institute Center for Cancer Research.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank the National Cancer Institute Cooperative Prostate Cancer Tissue Resource and James Goedert (Viral Epidemiology Branch, National Cancer Institute) for providing tissue specimens and supporting data and Audrey Salabes, John Cottrel, Raymond Jones, and other personnel at University of Maryland and Baltimore Veterans Administration for their contributions.

1
Jemal A, Siegel R, Ward E, Murray T, Xu J, Thun MJ. Cancer statistics, 2007.
CA Cancer J Clin
2007
;
57
:
43
–66.
2
Hsing AW, Tsao L, Devesa SS. International trends and patterns of prostate cancer incidence and mortality.
Int J Cancer
2000
;
85
:
60
–7.
3
Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2006.
CA Cancer J Clin
2006
;
56
:
106
–30.
4
Powell IJ. Epidemiology and pathophysiology of prostate cancer in African-American men.
J Urol
2007
;
177
:
444
–9.
5
Reddy S, Shapiro M, Morton R, Jr., Brawley OW. Prostate cancer in Black and White Americans.
Cancer Metastasis Rev
2003
;
22
:
83
–6.
6
Hoffman RM, Gilliland FD, Eley JW, et al. Racial and ethnic differences in advanced-stage prostate cancer: the Prostate Cancer Outcomes Study.
J Natl Cancer Inst
2001
;
93
:
388
–95.
7
Thompson I, Tangen C, Tolcher A, Crawford E, Eisenberger M, Moinpour C. Association of African-American ethnic background with survival in men with metastatic prostate cancer.
J Natl Cancer Inst
2001
;
93
:
219
–25.
8
Cher ML, Lewis PE, Banerjee M, et al. A similar pattern of chromosomal alterations in prostate cancers from African-Americans and Caucasian Americans.
Clin Cancer Res
1998
;
4
:
1273
–8.
9
Shibata A, Whittemore AS. Genetic predisposition to prostate cancer: possible explanations for ethnic differences in risk.
Prostate
1997
;
32
:
65
–72.
10
Shuch B, Mikhail M, Satagopan J, et al. Racial disparity of epidermal growth factor receptor expression in prostate cancer.
J Clin Oncol
2004
;
22
:
4673
–7.
11
Guo Y, Sigman DB, Borkowski A, Kyprianou N. Racial differences in prostate cancer growth: apoptosis and cell proliferation in Caucasian and African-American patients.
Prostate
2000
;
42
:
130
–6.
12
Haiman CA, Patterson N, Freedman ML, et al. Multiple regions within 8q24 independently affect risk for prostate cancer.
Nat Genet
2007
;
39
:
638
–41.
13
Yeager M, Orr N, Hayes RB, et al. Genome-wide association study of prostate cancer identifies a second risk locus at 8q24.
Nat Genet
2007
;
39
:
645
–9.
14
Strickler HD, Burk R, Shah K, et al. A multifaceted study of human papillomavirus and prostate carcinoma.
Cancer
1998
;
82
:
1118
–25.
15
Gentleman RC, Carey VJ, Bates DM, et al. Bioconductor: open software development for computational biology and bioinformatics.
Genome Biol
2004
;
5
:
R80
.
16
Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response.
Proc Natl Acad Sci U S A
2001
;
98
:
5116
–21.
17
Storey JD, Tibshirani R. Statistical significance for genomewide studies.
Proc Natl Acad Sci U S A
2003
;
100
:
9440
–5.
18
Tibshirani R, Hastie T, Narasimhan B, Chu G. Diagnosis of multiple cancer types by shrunken centroids of gene expression.
Proc Natl Acad Sci U S A
2002
;
99
:
6567
–72.
19
Yi M, Horton JD, Cohen JC, Hobbs HH, Stephens RM. WholePathwayScope: a comprehensive pathway-based analysis tool for high-throughput data.
BMC Bioinformatics
2006
;
7
:
30
.
20
Becker KG, Barnes KC, Bright TJ, Wang SA. The genetic association database.
Nat Genet
2004
;
36
:
431
–2.
21
Bookout AL, Mangelsdorf DJ. Quantitative real-time PCR protocol for analysis of nuclear receptor signaling pathways.
Nucl Recept Signal
2003
;
1
:
e012
.
22
Pettaway CA, Troncoso P, Ramirez EI, Johnston DA, Steelhammer L, Babaian RJ. Prostate specific antigen and pathological features of prostate cancer in Black and White patients: a comparative study based on radical prostatectomy specimens.
J Urol
1998
;
160
:
437
–42.
23
Rhodes DR, Barrette TR, Rubin MA, Ghosh D, Chinnaiyan AM. Meta-analysis of microarrays: interstudy validation of gene expression profiles reveals pathway dysregulation in prostate cancer.
Cancer Res
2002
;
62
:
4427
–33.
24
Singh D, Febbo PG, Ross K, et al. Gene expression correlates of clinical prostate cancer behavior.
Cancer Cell
2002
;
1
:
203
–9.
25
Lapointe J, Li C, Higgins JP, et al. Gene expression profiling identifies clinically relevant subtypes of prostate cancer.
Proc Natl Acad Sci U S A
2004
;
101
:
811
–6.
26
Planitzer SA, Machl AW, Rueckels M, Kubbies M. Identification of a novel c-DNA overexpressed in Fanconi's anemia fibroblasts partially homologous to a putative L-3-phosphoserine-phosphatase.
Gene
1998
;
210
:
297
–306.
27
Taichman RS, Cooper C, Keller ET, Pienta KJ, Taichman NS, McCauley LK. Use of the stromal cell-derived factor-1/CXCR4 pathway in prostate cancer metastasis to bone.
Cancer Res
2002
;
62
:
1832
–7.
28
Muller A, Homey B, Soto H, et al. Involvement of chemokine receptors in breast cancer metastasis.
Nature
2001
;
410
:
50
–6.
29
Small EJ, Schellhammer PF, Higano CS, et al. Placebo-controlled phase III trial of immunologic therapy with sipuleucel-T (APC8015) in patients with metastatic, asymptomatic hormone refractory prostate cancer.
J Clin Oncol
2006
;
24
:
3089
–94.
30
Luo J, Duggan DJ, Chen Y, et al. Human prostate cancer and benign prostatic hyperplasia: molecular dissection by gene expression profiling.
Cancer Res
2001
;
61
:
4683
–8.
31
Magee JA, Araki T, Patil S, et al. Expression profiling reveals hepsin overexpression in prostate cancer.
Cancer Res
2001
;
61
:
5692
–6.
32
Welsh JB, Sapinoso LM, Su AI, et al. Analysis of gene expression identifies candidate markers and pharmacological targets in prostate cancer.
Cancer Res
2001
;
61
:
5974
–8.
33
Dhanasekaran SM, Barrette TR, Ghosh D, et al. Delineation of prognostic biomarkers in prostate cancer.
Nature
2001
;
412
:
822
–6.
34
Sehgal G, Hua J, Bernhard EJ, Sehgal I, Thompson TC, Muschel RJ. Requirement for matrix metalloproteinase-9 (gelatinase B) expression in metastasis by murine prostate carcinoma.
Am J Pathol
1998
;
152
:
591
–6.
35
Dong Z, Bonfil RD, Chinni S, et al. Matrix metalloproteinase activity and osteoclasts in experimental prostate cancer bone metastasis tissue.
Am J Pathol
2005
;
166
:
1173
–86.
36
Sun YX, Wang J, Shelburne CE, et al. Expression of CXCR4 and CXCL12 (SDF-1) in human prostate cancers (PCa) in vivo.
J Cell Biochem
2003
;
89
:
462
–73.
37
Sun YX, Fang M, Wang J, Cooper CR, Pienta KJ, Taichman RS. Expression and activation of α(v)β(3) integrins by SDF-1/CXC12 increases the aggressiveness of prostate cancer cells.
Prostate
2007
;
67
:
61
–73.
38
Chinni SR, Sivalogan S, Dong Z, et al. CXCL12/CXCR4 signaling activates Akt-1 and MMP-9 expression in prostate cancer cells: the role of bone microenvironment-associated CXCL12.
Prostate
2006
;
66
:
32
–48.
39
Orimo A, Gupta PB, Sgroi DC, et al. Stromal fibroblasts present in invasive human breast carcinomas promote tumor growth and angiogenesis through elevated SDF-1/CXCL12 secretion.
Cell
2005
;
121
:
335
–48.
40
Kang Y, Siegel PM, Shu W, et al. A multigenic program mediating breast cancer metastasis to bone.
Cancer Cell
2003
;
3
:
537
–49.
41
Wiley HE, Gonzalez EB, Maki W, Wu MT, Hwang ST. Expression of CC chemokine receptor-7 and regional lymph node metastasis of B16 murine melanoma.
J Natl Cancer Inst
2001
;
93
:
1638
–43.
42
Vaday GG, Hua SB, Peehl DM, et al. CXCR4 and CXCL12 (SDF-1) in prostate cancer: inhibitory effects of human single chain Fv antibodies.
Clin Cancer Res
2004
;
10
:
5630
–9.
43
De Marzo AM, Platz EA, Sutcliffe S, et al. Inflammation in prostate carcinogenesis.
Nat Rev Cancer
2007
;
7
:
256
–69.
44
Hayes RB, Pottern LM, Strickler H, et al. Sexual behaviour, STDs and risks for prostate cancer.
Br J Cancer
2000
;
82
:
718
–25.
45
Xu J, Zheng SL, Komiya A, et al. Germline mutations and sequence variants of the macrophage scavenger receptor 1 gene are associated with prostate cancer risk.
Nat Genet
2002
;
32
:
321
–5.
46
Carpten J, Nupponen N, Isaacs S, et al. Germline mutations in the ribonuclease L gene in families showing linkage with HPC1.
Nat Genet
2002
;
30
:
181
–4.
47
De Marzo AM, Marchi VL, Epstein JI, Nelson WG. Proliferative inflammatory atrophy of the prostate: implications for prostatic carcinogenesis.
Am J Pathol
1999
;
155
:
1985
–92.
48
Eastham JA, May RA, Whatley T, Crow A, Venable DD, Sartor O. Clinical characteristics and biopsy specimen features in African-American and White men without prostate cancer.
J Natl Cancer Inst
1998
;
90
:
756
–60.
49
de Visser KE, Eichten A, Coussens LM. Paradoxical roles of the immune system during cancer development.
Nat Rev Cancer
2006
;
6
:
24
–37.
50
Alers JC, Rochat J, Krijtenburg PJ, et al. Identification of genetic markers for prostatic cancer progression.
Lab Invest
2000
;
80
:
931
–42.