Imaging, Diagnosis, Prognosis

The ABCA5 : A Urine Diagnostic Marker for Prostatic Intraepithelial Neoplasia Yo uji Hu , 1Min Wang,1KarenVeverka,2 Fernando U. Garcia,1and Mark E. Stearns1

Abstract Purpose: To develop a urine diagnostic test for preneoplastic intraepithelial neoplasia of the prostate. Experimental Design: We have used a DNA-binding assay and electrophoretic mobility shift assays (EMSA) to screen for novelduplexed DNA-binding sequences, which bind protein(s) overexpressed in crude protein extracts from high-grade prostatic intraepithelial neoplasia (HGPIN). EMSAs, immunohistochemistry, and ELISAs were used to measure expression of the ABCA5 protein identified as a specific marker in prostate tissue and patient urine. Results: Following screening of 4,096 sequences, an 8-bp dsDNA sequence (i.e.,TCCAGCGA) was identified, which binds the ABCA5 protein, a member of the ATP-binding cassette multidrug resistant family. EMSAs showed that ABCA5 was overexpressed in HGPIN tissue (n =11/11)and in the urine of patients with HGPIN (n = 18/18) but was not expressed in prostate cancer, benign prostatic hyperplasia, or stroma. Immunohistochemistry indicated that ABCA5 was overex- pressed in foci of intermediate basalcellsin normalglands and in HGPIN. ABCA5 was faintly expressed in prostate cancer glands. ELISAs showed in‘blinded studies’ that ABCA5 was a highly sensitive (>98% sensitivity) urine diagnostic marker for HGPIN in biopsy-positive patients (n = 107) at a ‘cutoff’of 25 ng/mL. ABCA5 was present at very low levels (i.e., <25 ng/mL) in the urine of patients diagnosed with benign prostatic hyperplasia (n = 79) or prostatitis or kidney and bladder cancer (>86% specificity). Conclusions: The data indicate that ABCA5 might be a specific urine marker for diagnosis of patients with HGPIN.

Franks (1) first proposed that atypical hyperplasia was a development of HGPIN predated the development of clinically precursor of prostatic carcinoma. Helpap (2) showed by detectable cancer by 5 to 10 years, consistent with the concept [3H]thymidine uptake that ‘‘severe atypical primary hyperplasia that HGPIN is a premalignant lesion (18). Other studies have is a precancerous lesion.’’ In a key study, McNeal and Bostwick provided strong support for the association of HGPIN with the (3) showed an association between high-grade prostatic intra- incidence of prostate cancer. For example, Gokden et al. (19) epithelial neoplasia (HGPIN) and prostatic adenocarcinoma. retrospectively identified 190 men with HGPIN and 1677 men Bostwick and Brawer (4) subsequently provided a detailed with only benign prostatic hyperplasia (BPH) in needle biopsy description of the architectural features of HGPIN where they tissue. The cumulative risk of detection of carcinoma on serial pointed out that >75% of all HGPIN exhibits a dome-like sextant follow-up biopsies was 30.5% for those with isolated architecture. Since these initial studies, Epstein et al. (5) HGPIN compared with 26.2% for the control group (P < provided further linkage of HGPIN with organ-confined 0.001). HGPIN found on the first repeat biopsy was associated carcinoma, and multiple studies have now described HGPIN with a 41% risk of subsequent detection of carcinoma and showed a significant association with cancer (6–17). In a compared with an 18% risk if benign prostatic tissue was series of 249 autopsy cases, 77% of prostates with HGPIN found on the first repeat biopsy. The results suggest that the risk harbored invasive adenocarcinoma compared with only 24% of prostate cancer is 30.5% after a diagnosis of isolated HGPIN without HGPIN (10). Autopsy studies also showed that the in a needle biopsy. Likewise, Bishara et al. (20) showed that the multiple core involvement by HGPIN, both on initial and first repeat biopsy, defines a subset of men that is at increased risk of harboring synchronous invasive carcinoma. Studies by Authors’ Affiliations: 1Drexel University College of Medicine, Philadelphia, Bostwick and Brawer (4) and Bostwick (21) have shown that 2 Pennsylvania and GTx, Inc., Memphis,Tennessee HGPIN, patient age, and prostate-specific antigen (PSA) are all Received 7/13/06; revised 9/13/06; accepted 10/10/06. The costs of publication of this article were defrayed in part by the payment of page highly significant predictors of prostate cancer, with HGPIN charges. This article must therefore be hereby marked advertisement in accordance having the highest risk ratio (4, 21). In fact, HGPIN is more with 18 U.S.C. Section 1734 solely to indicate this fact. predictive of prostate cancer in older patients and those with a Requests for reprints: Mark E. Stearns, Department of Pathology, Drexel serum PSA of >4 ng/mL (4, 21). Because HGPIN predates the University College of Medicine, 245 North Broad Street, Philadelphia, PA 19102- 1192. Phone: 215-762-1597; Fax: 215-246-5918; E-mail: [email protected]. appearance of prostate cancer by approximately 5 to 10 years F 2007 American Association for Cancer Research. (4, 21), these reports suggest that patients with HGPIN need to doi:10.1158/1078-0432.CCR-06-1718 be aggressively monitored for the development of cancer.

www.aacrjournals.org 929 Clin Cancer Res 2007;13(3) February 1,2007 Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Imaging, Diagnosis, Prognosis

Despite this evidence, a considerable controversy remains cDNA Synthesis kit and ZAP Express cDNA Gigapack III Gold Cloning about the importance of HGPIN as a precursor of prostate kit (Promega). ABCA5 cancer. Thus, more work is required to confirm the relationship Cloning of the . Cloning of the ABCA5 gene was of HGPIN with the onset of prostate cancer. Because HGPIN carried out using a HGPIN-specific cDNA library [made by Dr. Y. Hu (Drexel University) in our laboratory] according to methods modified is not always apparent in biopsies due to the multifocal nature by our laboratory (23) from Singh et al. (22). In brief, the 32P-labeled of the condition, a sensitive and specific marker capable of HGPIN-specific binding sequence (TCCAGCGA) was used as a probe uniformly detecting HGPIN with improved accuracy is to screen the HGPIN cDNA library. The phage was plated at 1 Â 104 required. For these reasons, we have developed methods for plaque-forming units/150-mm plate. The replica filters were incubated identification of a HGPIN diagnostic marker. These efforts have in binding buffer containing 32P-labeled probe and poly(deoxyino- led to the identification of ABCA5 as a potential tissue and sinic-deoxycytidylic acid) for 60 min at room temperature with gentle urine marker for HGPIN. ABCA5 is a member of the ATP- shaking. The filters were washed four times with the binding buffer, binding cassette multidrug resistant family, which is overex- dried, and exposed to Kodak (Rochester, NY) X-OmatAR film À j pressed in prostate tissue samples and urine of patients with overnight at 80 C. The positive clones were picked and resuspended PIN but is not present at detectable levels in the urine of in SM buffer [5.8 g NaCl, 2 g MgSO47H2O, 1 mol/L Tris-Cl (pH 7.5) plus 5 mL 0.2% gelatin] and then put on a shaker overnight at 4jC. patients with BPH or prostatitis. ELISAs were developed for Three positive cDNA clones that expressed ABCA5 were purified by detection and quantitation of ABCA5. A preliminary evaluation three sequential repetitions of the above protocol. Sequencing and of the ELISA data in a blinded study indicated that, at 25 ng/mL blast analysis revealed that the cDNAs (f3,321 bp) were encoded for used as a ‘cutoff’ value, the sensitivity was >98% and the the full-length ABCA5 protein (968 amino acids). A glutathione S- specificity was >86%. The data indicate, therefore, that ABCA5 transferase (pGEX) fusion protein was generated to purify the protein is an excellent urine diagnostic marker for HGPIN in biopsy- according to published methods (24, 25). confirmed patients (n = 107). Reverse transcription-PCR analysis. RNA was isolated from different patient tissues according to methods described above and subsequently prepared for reverse transcription-PCR (RT-PCR). RT-PCR was carried Materials and Methods out with primers for the ABCA5 gene (see below). This work was carried out by Dr. M. Wang (Drexel University) in our laboratory or by Dr. Isolation of protein from prostate tissues. After surgical removal, Celia Chang (Human Genomics Laboratory, Wistar Research Institute, human prostates were collected within 30 min and dissected to remove Philadelphia, PA). cDNA (f100 bp) was extracted from the gels for pieces of tissue containing prostate cancer, HGPIN, stroma, BPH, and sequencing. Blast analysis and comparisons on Genbank were carried seminal vesicles. The diagnosis of these tissues was confirmed by out for each RNA specimen. Dr. F.U. Garcia (Drexel University, Philadelphia, PA) by H&E-stained Real-time quantitative RT-PCR. Total RNA was isolated from tissues sections. In addition, freshly frozen prostate tissues with histologically following a protocol described by RNAgents Total RNA Isolation confirmed HGPIN (n = 10) were purchased from a private vendor System (Promega). Custom-made primers were obtained from Inte- (Genomics Collaborative, Inc., Boston, MA). Protein extracts were grated DNA Technologies (Coralville, IA). Brilliant SYBR Green qPCR prepared from the tissue samples (f100-200 mg) containing at least was used to do quantitative PCR amplification for ABCA5-specific and 80% HGPIN, and there was no evidence of prostate cancer (based on glyceraldehyde-3-phosphate dehydrogenase–specific regions. ABCA5 histologic examinations of tissue sections) using a modification of the gene was amplified using the primers (26) 5¶-GGCTGCTATTCTGAC- method of Singh et al. (22) as previously described by our laboratory CACTCACTATA-3¶ (forward) and 5¶-TTAACTGCCCAGACACCATGAT-3¶ (23). In brief, tissue was washed thrice with PBS and cut to small pieces (reverse). Glyceraldehyde-3-phosphate dehydrogenase was amplified and then suspended in 2 mL cold buffer A [10 mmol/L HEPES (pH using the primers 3055¶-TGTTACCAACTGGGACGACA324 (forward) and 7.9), 10 mmol/L KCl, 0.1 mmol/L EDTA, 0.1 mmol/L EGTA, 1 mmol/L 8585¶-AAGGAAGGCTGGAAAAAGAGC837 (reverse). DTT, 0.5 mmol/L phenylmethylsulfonyl fluoride]. The pellet was Western blot and immunoprecipitation studies. Rabbit polyclonal ¶ sonicated for 30 s and centrifuged (10,000 rpm for 15 min). The antibodies were raised against an NH2-terminal peptide (termed P1), 5 - supernatant was collected and the protein was precipitated with 4 MSTAIREVGVWRQTRTLLLKNY-3¶ (i.e., +1 to +22), and a COOH- volumes of cold acetone (À20jC). The protein was then snap frozen in terminal peptide (termed P2), 5¶-FSGLFSALDSHSNLGVISYGVS-3¶ (i.e., liquid nitrogen and stored at À80jC for <1 year. For DNA-binding +854 to +876), unique to the ABCA5 protein. The antibodies were assays, the samples were thawed and 2 AL of the protein extract (f5 Ag partially purified using a protein A column. Whole-cell lysates were protein) were incubated with 1 Ag of poly(deoxyinosinic-deoxycytidylic prepared and subjected to Western blot analysis as described previously acid) for 15 min followed by incubation with the 32P-labeled DNA (27). Blots were probed with rabbit polyclonal anti-ABCA5 (1:2,000 probe (f100,000 dpm) for 30 min at 4jC. DNA binding to the dilution). A secondary horseradish peroxidase–conjugated IgG anti- protein(s) was then examined by gel electrophoretic mobility shift rabbit antibody (Pierce, Rockford, IL) was applied for 1 h (1:5,000 assays (EMSA; see below). Tissue was obtained by Dr. Garcia in dilution) followed by SuperSignal Western Pico Chemiluminescent accordance with Institutional Review Board–approved protocol at substrate detection (Pierce) and exposure on Chemidoc XRS 5 (Bio-Rad, Drexel University College of Medicine. Richmond, CA) to visualize immunoreactive bands. Immunoprecipita- Identification of the DNA sequence that binds ABCA5. An EMSA (23) tion assays were carried out by adding 1 Ag/mL of the rabbit polyclonal was used to screen 32P-labeled duplexed DNA sequences (8 bp), which antibody (P1) to 1 mL volume of the supernatants of the bacteriophage preferentially bind a protein(s) expressed by HGPIN compared with or an ABCA5 phagemid clone or to 10 mL volumes of human urine prostate cancer, BPH, and stroma tissue. Accordingly, we used EMSAs from patients diagnosed with HGPIN, prostate cancer, or BPH. The and screened f4,096 sequence combinations to identify the tissue- immunoprecipitates were run on 10% SDS-polyacrylamide gels and specific binding sequence (i.e., TCCAGCGA), which binds a single Western blotted with the anti-ABCA5 (P1) polyclonal antibodies protein uniquely expressed by HGPIN. according to methods described above. HGPIN-specific cDNA expression library. mRNA was isolated from Immunolabeling. To determine the subcellular localization of freshly dissected HGPIN tissue according to the protocol described by ABCA5, CD44, and ABCG2 , whole-mount prostate sections PolyATtract System 1000 (Promega, Madison, WI). A phage vector, ZAP and pieces of human brain (as a positive control) were incubated for Expression Vector (Stratagene, La Jolla, CA), was used and construction 45 min at 37jC with the polyclonal rabbit anti-ABCA5 (P1), CD44 of the cDNA library was according to a protocol of the ZAP Express (Abcam, Inc., Cambridge, United Kingdom), or ABCG2 (Chemicon,

Clin Cancer Res 2007;13(3) February1,2007 930 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. ABCA5 Is a Urine Diagnostic Marker for PIN

Inc., Temecula, CA) antibodies. Unbound antibodies were removed ABCA5 gene. The deduced amino acid sequence (968 amino with three rinses of PBS followed by incubation for 45 min at 37jC acids) was also >98% homologous to the ABCA5 protein. with horseradish peroxidase–conjugated anti-rabbit antibodies (Jack- EMSA with the 32P-labeled TCCAGCGA probe subsequently son ImmunoResearch Laboratories, West Grove, PA) according to confirmed that ABCA5 was preferentially overexpressed in methods described by our laboratory (28). prostate tissue with HGPIN (n = 11) when compared with ELISAs. ELISAs were carried out using the rabbit polyclonal antibodies raised against P1 of the ABCA5 proteins. ELISAs were normal (n = 12), BPH (n = 12), or prostate cancer (n = 12) according to antibody capture assay technique and methods described tissue. Figure 1D further illustrates that ABCA5 was overex- by Harlow and Lane (29). All protein measurements were according to pressed in the urine of patients with HGPIN (Fig. 1D, lanes 1, 3, methods of Bradford (30). Urine was collected fresh and frozen at 5, 7, and 9) but was not found in the urine of patients À20jC for storage and shipment (31). Patients were not subjected to diagnosed with BPH (Fig. 1D, lanes 2 and 4) or in the urine of digital rectal examination or massage before collection of the urine. patients diagnosed with prostate cancer only (i.e., no indication Urine specimens were thawed and assayed by ELISA. Aliquots were of HGPIN; Fig. 1D, lanes 6, 8, 10, and 11). Again, the specific À j refrozen at 80 C and assayed within 1 year. Consent was obtained affinity of the 32P-labeled TCCAGCGA probe for the urine from each of the patients in accordance with Health Insurance ABCA5 protein was shown by ‘supershift’ assays. EMSAs with Portability and Accountability Act regulations. increased amounts of the ABCA5 antibody produced protein- antibody complexes near the tops of the gels, indicating that the 32 Results P-labeled TCCAGCGA probe specifically binds the ABCA5 protein (data not shown). In addition, ‘cold competition’ We have developed a highly sensitive and consistent experiments with the TCCAGCGA probe (i.e., 20-50 ng) screening assay using EMSAs for identification of duplexed eliminated the ‘bandshift’ normally observed. Cold competi- DNA sequence(s), which binds picogram levels of protein in tion assays where 1 base was changed or with a random crude protein extracts from tissue or urine (23). Using this oligonucleotide sequence failed to block the bandshift observed technique, we screened f4,096 sequence combinations and with urine protein (5 Ag total protein; data not shown). identified a specific 8-mer oligonucleotide sequence (i.e., Western blots. Rabbit polyclonal antibodies were raised TCCAGCGA) that binds a single protein present in crude against two different peptide domains unique to the ABCA5 protein extracts from HGPIN. The example shows the ‘band- protein and partially purified using a protein A column. shift’ observed using the 8-mer sequence and protein extracts Western blots showed that both antibodies specifically recog- from three HGPIN specimens (Fig. 1A, lanes 5-7). A ‘bandshift’ nized recombinant ABCA5 protein and neither antibody cross- was not observed when the 8-mer oligonucleotide was reacted with other multidrug resistant (i.e., ABCG2, incubated with protein extracts from stroma, BPH, or prostate ABCB1, and mdr-1) or other proteins present in recombinant cancer tissue. Attempts to increase the amount of 32P-radio- phagmid extracts or whole-cell extracts of prostate cells (data labeled probe (i.e., to 150,000 dpm) or the amount of crude not shown). Immunoprecipitation assays (combined with SDS protein extract (i.e., to 20 Ag) did not produce a signal in the gel electrophoresis; Fig. 2, top bands) with excess amounts of prostate cancer, BPH, or stroma preparations, indicating that ABCA5 antibody (i.e., 1 Ag/mL) raised against the P1 peptide the protein might be uniquely overexpressed by HGPIN. showed that the 160-kDa ABCA5 protein was present in the Changes in the base sequence of the probe eliminated binding urine of patients with HGPIN (Fig. 2, lanes 1-3) and was faintly to the HGPIN protein (data not shown). Similar studies with expressed in the urine of patients with prostate cancer (Fig. 2, urine (f10 Ag protein) from patients with biopsy-positive lane 6). More importantly, ABCA5 was not found in the urine HGPIN and no indication of cancer showed a similar bandshift of patients with BPH (Fig. 2, lanes 4 and 5) or in crude extracts (Fig. 1B, lanes 1, 2, 3, and 6). The bandshift was not detected of the bacteriophage (Fig. 2, lane 7). ABCA5 was expressed by in the urine of patients with BPH or prostate cancer (Fig. 1B, the ABCA5 phage clone, however (Fig. 2, lane 8). Western blots lanes 4 and 5). However, a faint bandshift was observed with (bottom bands) of the immunoprecipitates confirmed that the urine of patients with biopsy results showing HGPIN ABCA5 was expressed in the urine of patients with HGPIN concurrent with prostate cancer (Fig. 1B, lane 6). In the above (Fig. 2, lanes 1-3), was faintly expressed in the urine of prostate assays, cold competition experiments with excess unlabeled cancer patients (Fig. 2, lane 6), and was present in protein DNA probe (20-50 ng) completely blocked the bandshift extracts of the ABCA5 phage clone (Fig. 2, lane 8). Western observed with HGPIN tissue extracts and HGPIN urine, but a blots confirmed that ABCA5 was not expressed in the urine of random DNA sequence failed to block binding to the DNA patients with BPH (Fig. 2, lanes 4 and 5)andinthe (data not shown). bacteriophage (Fig. 2, lane 7). Cloning of the ABCA5 gene. Cloning of the ABCA5 gene was Immunolabeling studies. Immunolabeling of human pros- carried out using a phage vector, ZAP Expression Vector, tate glands (n = 10 whole-mount prostates) confirmed that according to methods modified by our laboratory (23) from the ABCA5 protein was selectively expressed in glands with Singh et al. (22). Three positive clones were identified by HGPIN lesions (n = 10). In contrast, ABCA5 was not autoradiography. These clones were expanded in suspension expressed in BPH or was faintly expressed in normal glands cultures, and EMSA was carried out with crude protein extracts and prostate cancer (Fig. 3A). Note that, in normal glands, the from each clone alongside an extract from prostate tissue with occasional foci of intermediate basal cells (arrow) were HGPIN. As shown in Fig. 1C, each of the three clones showed a strongly labeled with the ABCA5 antibodies, suggesting that single bandshift in the EMSA (Fig. 1C, lanes 2-4). The bandshift these cells might be the origin of HGPIN (Fig. 3A). Because was identical to that detected with the HGPIN protein extract basal cells are thought to express CD44, as a positive control, (Fig. 1C, lane 1). Sequencing of the gene (3,321 bases) and adjacent sections were labeled with CD44 antibodies. We blast analysis revealed that it was >98% homologous to the found that CD44 antibodies labeled the vast majority of the

www.aacrjournals.org 931 Clin Cancer Res 2007;13(3) February 1,2007 Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Imaging, Diagnosis, Prognosis

Fig. 1. A to D, EMSAs showing the bandshift (arrow) observed following incubation of the radiolabeled probe (i.e., TCCAGCGA) with crude protein extracts. The 32P-labeled duplexed DNA (100,000 dpm) was incubated for 10 min with the crude protein extract (5 Ag protein/well) before electrophoresis. A, protein extracts from stroma (lanes 1 and 2), BPH (lanes 3 and 4), HGPIN (lanes 5 -7), and prostate cancer (lanes 8 -10 ) tissues, respectively. B, protein from the urine (1mg protein) of patients diagnosed with HGPIN (lanes 1, 2, 3,and6) and BPH and prostate cancer (lanes 4 and 5). Lane 6, the patient had HGPIN and prostate cancer. C, protein extracts from HGPIN tissue (lane 1)and recombinant proteins (lanes 2-4) isolated from three different clones, respectively. D, protein (1mg) from the urine of patients diagnosed with HGPIN (lanes 1, 3, 5, 7,and 9), BPH (lanes 2 and 4), and prostate cancer (lanes 6, 8, 10,and11).

Clin Cancer Res 2007;13(3) February1,2007 932 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. ABCA5 Is a Urine Diagnostic Marker for PIN

ELISAs. Using the P1 polyclonal antibodies, we initially developed an ELISA with the purified recombinant ABCA5 protein. The standard curve (Fig. 5) showed that the limit of detection of ABCA5 was f1 ng/0.1 mL. A linear increase in absorbance was observed for increased amounts of ABCA5 ranging from 1 to 23 ng/0.1 mL with a correlation coefficient of 0.54. Note that an absorbance (450 nm) of 0.2 nm corresponded to f2.0 ng/0.1 mL ABCA5 P1 peptide. The addition of increased amounts of P1 peptide to the negative urine yielded a standard curve identical to that observed with PBS (data not shown). In Fig. 6A, we assayed the levels of ABCA5 protein in freshly collected/frozen urine specimens (by Bioreclamation, Inc., Wilmington, DE) from patients with PSA of <1.5 ng/mL (n = 30; Fig. 6A, filled square) compared with patients with serum PSA of >10 ng/mL (n = 30; Fig. 6A, filled inverted triangle) without the benefit of biopsy. The data showed that >97% of the patients with high PSA levels also had elevated ABCA5 levels in their urine (i.e., >25 ng/mL). In comparison, >90% Fig. 2. Immunoprecipitation assays showing that the 160-kDa ABCA5 protein was present in the urine of patients with HGPIN (lanes1-3), was faintly expressed in of the patients with low PSA levels (<1.5 ng/mL) also had low the urine of patients with prostate cancer (lane 6), but was not found in the urine levels of ABCA5 in their urine (i.e., <25 ng/mL; Fig. 6A). One of patients with BPH (lanes 4 and 5) or the bacteriophage (lane 7). IgG band at interpretation of this result is that urine from patients with 55 kDa. Arrow, ABCA5 was expressed by the ABCA5 phagemid clone (lane 8). Western blots of the immunoprecipitates confirmed that the 160-kDa ABCA5 levels of PSA of <1.5 ng/mL, considered to be putatively protein (bottom band) was expressed in the urine of patients with HGPIN normal patients (i.e., no prostate cancer or HGPIN), also (lanes1-3), prostate cancer (lane 6), and the ABCA5 phagemid clone (lane 8). ABCA5 was not expressed in the urine of patients with BPH (lanes 4 and 5) exhibited low levels of ABCA5. However, patients with and the bacteriophage (lane 7). elevated levels of PSA who would be suspected to have carcinoma of the prostate were found to have elevated ABCA5 levels. Patients with prostate cancer often have HGPIN (10), basal cells and intermediate basal cells in normal glands and and thus, we expect that the patients with elevated PSA levels strongly labeled the epithelial cells in HGPIN. Like ABCA5, probably have high-grade prostate cancer and HGPIN. In this the CD44 antibodies faintly labeled the luminal epithelial connection, of the 107 HGPIN patients evaluated here, we cells of prostate cancer. In comparison, ABCG2, a homologue found that the PSA values ranged from 0.3 to 42.8 ng/mL, of ABCA5, failed to label the basal cells or intermediate basal with the majority of the patients exhibiting PSA values of >4 cells or HGPIN and prostate cancer cells in any of the ng/mL. Thus, there was no correlation of the PSA values with specimens examined (n = 10). the incidence of HGPIN or the ABCA5 levels in urine (R2 = PCR and quantitative PCR assays. PCR analysis to amplify 0.998). Finally, assays of urine from patients diagnosed with mRNA isolated from different types of prostate tissue (n =5; BPH having prostatitis (n = 12) or bladder cancer (n =6)or i.e., normal, BPH, HGPIN, and prostate cancer) combined with kidney cancer (n = 8) indicated that there was very low sequencing of the 100-base fragments revealed that ABCA5 was background levels of ABCA5 present in the urine (Fig. 6A, highly expressed in HGPIN, whereas it was only faintly open triangle). expressed in tissue specimens of prostate cancer. Quantitative Figure 6B showed data from a ‘blinded study’ of patient urine RT-PCR and comparisons of the relative levels of expression of (n = 195 total; obtained from GTx, Inc., Memphis, TN). Based ABCA5 in different tissues clearly showed that the ABCA5 gene on 10 to 12 sextant biopsies, patients had been previously was overexpressed in HGPIN but was not expressed in BPH, diagnosed biopsy negative for HGPIN (n = 79; Fig. 6B, filled prostate cancer, normal prostate, or prostate stroma (Fig. 4A). square) or biopsy positive for prostate cancer and HGPIN Likewise, ABCA5 was not expressed in bladder, kidney cancer, (n = 9; Fig. 6B, filled triangle) or biopsy positive for HGPIN only and lung cancer (Fig. 4A) or in prostate cancer cells collected by (n = 107; Fig. 6B, filled inverted triangle). The sensitivity was centrifugation (10,000 Â g, 10 min) from 10 mL of fresh urine >98.1% and the specificity was >86% at a ‘cutoff’ of 25 ng/mL. (n = 5 specimens; ref. 31). However, ABCA5 was overexpressed That is, the levels of ABCA5 were above the ‘cutoff’ of 25 ng/mL in seminal vesicle, kidney, liver, and testis tissue (but not in in the biopsy-positive HGPIN patients and below 25 ng/mL in testis cancer; Fig. 4A), in agreement with a previous report of the BPH patients biopsy negative for HGPIN. The range of Langmann et al. (26). In addition, assays of RNA extracts from detection was from approximately 25 to 230 ng/mL in the BPH tissue of patients with prostatitis (n = 5) plus bladder HGPIN patients. The patients with prostate cancer and HGPIN (n = 4), kidney (n = 4), and testis (n = 2) cancer specimens all exhibited elevated levels of ABCA5 in the urine. The assay revealed that ABCA5 mRNA was not expressed in any of these conditions were identical to the methods described for the pathologies (Fig. 4B). The positive controls confirmed that standard curve (Fig. 5). ABCA5 was overexpressed in HGPIN and normal testis speci- Accuracyand precision. Repeat assays with 10 different mens (Fig. 4B). Note that, in all the above studies, the specimens positive for HGPIN and 10 BPH specimens (i.e., specimens were freshly collected and stored frozen at À80jC negative for HGPIN) have shown that all the HGPIN were for <1 year before extraction of RNA for quantitative RT-PCR positive for ABCA5 and all the BPH were negative for assays. ABCA5. More importantly, the measurements from repeat

www.aacrjournals.org 933 Clin Cancer Res 2007;13(3) February 1,2007 Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Imaging, Diagnosis, Prognosis

Fig. 3. A, immunolabeling of human prostate cancer (PCA), HGPIN, and normal glands with rabbit polyclonal antibodies specific forABCA5 and antibodies specific for CD44 and ABCG2. ABCA5 was selectively expressed in intermediate basal cells of normal glands (arrow)andfaintly expressed in an irregular fashion in prostate cancer (arrow). ABCA5 was overexpressed in HGPIN (arrow). CD4 4 was highly expressed in normalbasalcellsand HGPIN and prostate cancer (arrow). ABCG2, an analogue of ABCA5, was not expressed in benign, HGPIN, or prostate cancer (arrow). B, immunolabeling with ABCA5 antibody of glial cells (arrows) in a human brain section.

assays were consistent (SE, <0.5 ng/0.1 mL and <0.1 ng/0.1 ABCA5 in accordance with dilution effects of the added urine, mL, respectively). indicating that these urines did not contain an interference Antigen characteristics. Preliminary studies have been car- factor. ried out to evaluate antigen stability and the precision of the assay for ABCA5 in the urine. The data from repeated Discussion measurements of urine specimens showed that ABCA5 is a very stable protein and that the signal is not diminished >0.5 Development of a diagnostic marker for HGPIN is important, ng/0.1 mL if urine is kept at 4jC for 1 week or frozen for 1 year as several major studies have shown that HGPIN is a precursor or frozen and thawed thrice. Dilution of the urine with PBS of prostate cancer. It is expected that f35% of patients (1:1) slightly diminished the signal by f10%. Repeated assays diagnosed with HGPIN will develop prostate cancer within the on the sample indicated that f70% of the ABCA5 protein was first year after diagnosis and f50% will develop cancer by year captured in the first assay and that <5% remained after two 2. In this article, we have identified the ABCA5 gene and assays. Finally, the dilution of positive HGPIN urine with ABCA5 protein as being overexpressed in prostate glands with negative urine specimens from prostate cancer or BPH patients HGPIN. In contrast, the ABCA5 protein was absent or only did not eradicate the signal but simply reduced the level of faintly expressed in prostate cancer, BPH, BPH with prostatitis,

Clin Cancer Res 2007;13(3) February1,2007 934 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. ABCA5 Is a Urine Diagnostic Marker for PIN and stromal tissue of the prostate. Interestingly, immunolabel- ing indicated that ABCA5 was highly expressed in histologically defined intermediate basal cell foci found in normal glands and was overexpressed in all the cells of HGPIN glands. It was faintly expressed in prostate cancer as well. Although further work is required at this juncture, we speculate that the ABCA5- positive intermediate basal cells might constitute a highly proliferative cell type that is the precursor of HGPIN. In this study, we have also identified the ABCA5 protein as a urine biomarker for HGPIN, presumably as a direct result of the proliferation and sloughing of HGPIN cells in the lumen of glands and subsequently shedding in the urine. The ABCA5 protein was found in urine at levels ranging from 25 to 230 ng/ mL, presumably as a direct result of the lysis of HGPIN cells in the hypotonic urine. Further, we found in a ‘blinded study’ of

Fig. 5. ELISA standard curve with the ABCA5 peptide (P1) using a rabbit polyclonal antibody raised against P1 (1:5,000) and a goat anti-rabbit horseradish peroxidase (1:10,000) secondary antibody.

>195 patients (of which 107 patients were categorized as ‘‘HGPIN positive’’ by biopsy) that the ABCA5 protein was present in the urine at elevated levels compared with the biopsy-negative patients (n = 79). In the latter population, we found that the ABCA5 protein was either not present in the urine or present at very low levels (i.e., <25 ng/mL). The overall sensitivity and specificity of the ABCA5 marker was >98.1% and 86%, respectively, with a limit of detection of 25 ng/mL at an absorbance of 450 nm. Additional control studies showed by Western blots and ELISAs confirmed that the ABCA5 protein was not present in the urine of patients with BPH and prostatitis and in the urine of patients with other cancers, including testis, kidney, and bladder cancer. In summary, by identifying a marker that is specific for HGPIN, not only for the prostate, the assay for the ABCA5 protein offers a means to improve on the early detection of premalignant prostate cancer (i.e., HGPIN), thereby reducing biopsies when used as an adjunct to current methods. Moreover, a urine assay, such as ABCA5, is convenient and noninvasive because a needle puncture is not required to collect urine and therefore can readily be added to existing diagnostic tools. It is important to emphasize that, although the ABCA5 protein was overexpressed in all the cells of the HGPIN lesions, ABCA5 was not only

Fig. 4. A and B, quantitative RT-PCR assays for quantitative detection of ABCA5 expressed by the HGPIN glands. That is, the ABCA5 protein was mRNA in different human prostate tissue DNA preparations. cDNA equivalent to overexpressed by unique foci of cells, termed intermediate basal C 50 ng of totalRNA was used for each PCR assay. t values obtained with different cells, in normal glands. It was not expressed by other basal cells tissues were used to calculate the input amount of template by the calibration curve method. After normalization to the endogenous control glyceraldehyde-3- or luminal epithelial cells of normal glands, however. In phosphate dehydrogenase (data not shown), these values were expressed relative addition, ABCA5 was faintly expressed by the luminal epithelial to the calibrator (tissue with lowest expression) and displayed as relative levels of cells of prostate cancer (see Fig. 3A), indicating that the HGPIN expression. Columns, average of three independent assays; bars, SD. A, tissues included five different HGPIN (columns 1-5 ), prostate cancer (columns 6 and 7), cells, which express ABCA5, may cause prostate cancer. This BPH (columns 8 and 9), seminalvesicle( column 10), kidney (column 11), testis means that, although the urine levels of ABCA5 are significantly (column 12 ), testis cancer (column 13), liver (column 14), bladder (column 15), kidney cancer (column 16 ), lung cancer (column 17), prostate stroma (column 18), elevated in patients with HGPIN, ABCA5 may be found in the and normalprostate (column 19). B, tissues included HGPIN (column 1), BPH with urine of patients with prostate cancer. Moreover, we cannot rule prostatitis inflammation (columns 2-6), bladder cancer (columns 7-10 ), kidney out that patients with biopsy-confirmed HGPIN might harbor cancer (columns 11-14), testis cancer (columns 15 and 16 ), and normaltestis (column 17 ). ABCA5 primers were 5¶-GGCTGCTATTCTGACCACTCACTATA-3¶ undetected prostate cancer, which also contributes to the (forward) and 5¶-TTAACTGCCCAGACACCATGAT-3¶ (reverse). presence of ABCA5 in the urine.

www.aacrjournals.org 935 Clin Cancer Res 2007;13(3) February 1,2007 Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Imaging, Diagnosis, Prognosis

That is, the majority of prostates with HGPIN harbor invasive (23). In these studies, a novel duplexed probe was identified adenocarcinoma (i.e., 77%; ref. 10), and because epithelial cells that binds a leucine zipper domain of a mutant S2 ribosomal of prostate cancer express ABCA5, it is possible that cancer protein (23). Here, we identified the duplexed TCCAGCGA glands might account, in part, for the elevated levels of ABCA5 probe that presumably binds a specific consensus domain of in the urine specimens of HGPIN patients. Thus, before the ABCA5 protein. Although the ABCA5 protein might drawing firm conclusions, more exhaustive biopsy evaluation selectively bind DNA or RNA in vivo, we believe that the of prostate cancer patients is required to assess their HGPIN DNA-protein binding assays in vitro may reflect the specific status (and vice a versa). This information in combination with buffer conditions that do not normally occur in vivo. ELISAs of the urine specimens would help establish whether The ATP-binding cassette transporter superfamily is a large elevated urine levels of ABCA5 are specifically associated with gene family comprising at least 48 genes and encodes a HGPIN or are released by prostate cancer glands as well. In functionally diverse group of multispan membrane proteins addition, differentials of ABCA5 urine levels to serum PSA and/ involved in energy-dependent transport of a wide variety of or digital rectal examination data plus patient age may be substrates across membranes (32–34). Some multiple drug developed to better stage patients with premalignant or early- resistance phenotypes in tumor cells have been associated with stage cancer, for example. the gene encoding the multidrug resistance protein, which also The cloning protocol used to identify the ABCA5 gene was has an ATP-binding cassette transporter structure. Several ATP- based on a procedure previously developed to isolate a binding cassette proteins in the human system are responsible diagnostic marker for prostate cancer, termed PCADM-1 for drug exclusion in compound-treated tumor cells, providing

Fig. 6. A and B, ELISAs of ABCA5 levels in the urine from patients. See Fig. 5 for antibody dilutions. A, patients with serum PSA levels of <1.5 ng/mL (n)and >10 ng/mL (!). D, BPH with prostatitis (n = 12) or bladder cancer (n =6)orkidney cancer (n =8).B, patients diagnosed as normal( n =79;n) and with prostate cancer (n =9;E)andHGPIN(n =107;!) patients.

Clin Cancer Res 2007;13(3) February1,2007 936 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. ABCA5 Is a Urine Diagnostic Marker for PIN cellular mechanisms for the development of multidrug modulators. Thus, a noninvasive assay, such as the ABCA5 resistance (33–35). One group contains five genes (ABCA5, test, would be of tremendous benefit. This is especially true ABCA6, ABCA8, ABCA9, and ABCA10) arranged in a cluster on in light of the ongoing debate in the clinical diagnostic 17q24. Two members of this subfamily, the community as to whether HGPIN can be routinely detected and ABCA1 and ABCA4 (ABCR) proteins, have been studied diagnosed by pathologists. Detecting HGPIN by biopsy is extensively. The ABCA1 protein is involved in disorders of extremely difficult, and this problem is compounded by the fact cholesterol transport and high-density lipoproteins biosynthe- that HGPIN lesions tend to be small and scattered throughout sis (see below). The ABCA4 protein transports vitamin A the prostate gland. If HGPIN proves to be a precursor to prostate derivatives in the outer segments of photoreceptor cells and cancer as some clinicians believe, a specific HGPIN assay would therefore does a crucial step in the visual cycle (35). be invaluable not only for early detection of cancer but also for Human ABCA5 and rat Abca5 represent recently identified monitoring the progression from HGPIN to prostate cancer. subfamily members (36). Petry et al. (36) reported cloning Likewise, the assay (i.e., for ABCA5) would be of potential use in of cDNA sequences encoding rat and human ABCA5. Up- monitoring the response of promising novel therapies aimed at regulation of Abca5 mRNA expression was observed during preventing the conversion of HGPIN to prostate cancer (i.e., culture of primary rat hepatocytes (37). Quantitative RT-PCR toremifene citrate; ref. 38). showed that an ABCA5 splice variant was expressed in In an ongoing trial, toremifene citrate, which is a selective numerous tissues (including testis, brain, and lung; ref. 26). estrogen receptor modulator compound, was originally devel- The substrate spectrum remains to be defined and its normal oped as an antiestrogen and is currently marketed as Fareston function in different cell types and tissues is unknown. It has for advanced breast cancer in women. Toremifene has been been shown to colocalize with a marker protein for the Golgi convincingly shown to prevent HGPIN-like lesions and prevent apparatus in Leydig cells (37), and it has been suggested that or delay the onset of prostate tumors in the mouse prostate it may play a role in intracellular sterol/steroid trafficking. transgenic adenocarcinoma of mouse prostate model.3 In Whether it also plays a role in drug resistance is not known. addition, a placebo-controlled phase IIb clinical trial was We are the first to clone the ABCA5 gene from a human conducted by GTx for the treatment of HGPIN using prostate prostate HGPIN cDNA expression library (patent pending). cancer on a follow-up biopsy as a primary end point (38). The Because ATP-binding cassette genes are prone to be involved in study included 514 cancer-free patients with affirmed HGPIN. cancer progression and drug resistance, we believe that the Results showed that the risk of prostate cancer incidence was overexpression of ABCA5 in cells of HGPIN lesions may 24.4% in the 20 mg toremifene treatment group (n = 114) ultimately reveal a role of this protein in drug or sterol compared with 31.2% in the placebo group (n = 109), trafficking in the cells or another as of yet unidentified role of representing a 21.8% (P = 0.048) cumulative risk reduction this protein in these premalignant cells. Thus, ABCA5 may be a in prostate cancer. The phase IIb trial was the largest study of promising target molecule for the development of a therapeutic the natural history of prostate cancer to date and included the for the treatment of HGPIN (34–37) and possibly reduce collection of specimens, both blood and urine, for the purpose HGPIN progression to prostate cancer. of biomarker discovery to improve screening and detection of Because HGPIN is more predictive of prostate cancer in older HGPIN. Based on the promising results of the phase IIb study, patients and those with a serum PSA of >4 ng/mL (4, 21), a a double-blind, randomized, multicenter, phase III trial is therapeutic approach for treating HGPIN may be of tremen- ongoing with f1,260 patients to confirm the efficacy of 20 dous benefit to these individuals at high risk for prostate cancer mg/d toremifene versus placebo for the prevention of invasive (i.e., older men, those with first-degree relatives having prostate prostatic adenocarcinoma in high-risk men. We will participate cancer, and men of African decent). in this trial to determine whether the ABCA5 protein is a Currently, there is great interest in developing treatments diagnostic marker that can predict HGPIN and/or predict the for precancerous lesions of prostate. Due to the proposed failure of HGPIN to progress to prostate cancer in toremifene- role of HGPIN in the direct pathway to prostate cancer, treated patients. several researchers believe that it is a suitable target for putative chemopreventive agents, such as green tea poly- phenols, soy isoflavones, and selective estrogen receptor 3 GTx, Inc., unpublished data.

References 1. Franks LM. Atrophy and hyperplasia in the prostate 6. Kovi J, Mostofi FK, Heshmat MY, Enterline JP. Large prostatic intraepithelial neoplasia (HGPIN) and pros- proper. J PatholBacteriol1954;98:617^21. acinar atypicalhyperplasia and carcinoma of the pros- tatic adenocarcinoma between the ages of 20-69: an 2. Helpap B. The biologic significance of atypical hy- tate. Cancer 1988;61:555^ 61. autopsy study of 249 cases. InVivo1994;8:439 ^43. perplasia of the prostate. Virchows Arch 1980;387: 7. Quinn BD, Cho KR, Epstein JI. Relationship of severe 11. Bostwick DG. Prostatic intraepithelial neoplasia: the 307^31. dysplasia to stage A adenocarcinoma of the prostate. most likely precursor of prostate cancer. Cancer 3. McNealJE, Bostwick DG. Intraductaldysplasia: a Cancer 1990;65:2321 ^ 7. 1995(a);75:1823^36. premalignant lesion of the prostate. Hum Pathol 8. Brawer MK, Bigler SA, Sohlberg OE, Nagle RB, 12. Bostwick DG, Qian J. Atypicaladenomatous hyper- 1986;17:64^71. Lange PH. Significance of prostatic intraepithelial plasia of the prostate: relationship with carcinoma in 4. Bostwick DG, Brawer MK. Prostatic intraepithelial neoplasia on prostate needle biopsy. Urology 1991; 217 whole mount radical prostatectomies. Am J Surg neoplasia and early invasion in prostate cancer. Can- 38:103^7. Pathol1995(b);19:506^18. cer 1987;59:788^94. 9. Weinstein MH, Epstein JI. Significance of high grade 13. Davidson D, Bostwick DG, Qian J, et al. Prostatic 5. Epstein JI, Cho KR, Quinn BD. Relationship of severe prostatic intraepithelial neoplasia on needle biopsy. intraepithelial neoplasia is a risk factor for adenocarci- dysplasia to stage A (incidental) adenocarcinoma of Hum Pathol1993;24:624^ 9. noma predictive accuracy in needle biopsies. J Urol the prostate. Cancer 1990;65:2321 ^7. 10. SakrWA,GrignonDJ,CrissmanJD,etal.Highgrade 1995;154:1295 ^ 9.

www.aacrjournals.org 937 Clin Cancer Res 2007;13(3) February 1,2007 Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Imaging, Diagnosis, Prognosis

14. Arakawa A, Song S, Scardino PT, Wheeler TM. 22. Singh H, Lebowitz JH, Baldwin AS, Sharp PA. 30. Bradford MM. A rapid and sensitive method for the High grade prostatic intraepithelial neoplasia in pros- Molecular cloning of an enhancer binding protein: quantitation of microgram quantities of protein utiliz- tates removed following irradiation failure in the treat- isolation by screening of an expression library with a ing the principles of protein-dye binding. Anal Bio- ment of prostatic carcinoma. PatholRes Pract 1995; recognition site DNA. Cell 1988;52:415^ 23. chem 1976;72:248 ^ 54. 191:8 6 8 ^ 72. 23. Ohkia A, Hu Y,Wang M, Garcia FU, Stearns ME. 31. Scholler A, Freibauer C,Wang M, HuY, Stearns ME. 15. Haggman MJ, Macoska JA, Wojno KJ, Oesterling Evidence for a prostate cancer associated diagnostic Austrian AUA Mttg 2006;:36a. JE. The relationship between prostatic intraepithelial marker-1, PCADM-1: immunohistochemistry and 32. Dean M, Hamon Y, Chimini G. The human ATP- neoplasia and prostate cancer: critical issues. J Urol in situ hybridization studies. Clin Cancer Res 2004; binding cassette (ABC) transporter superfamily. J Lipid 1997;158:12^ 22. 10 :2452 ^ 8. Res 2001;42:1007 ^ 17. 16. Qian J,Wollan P, Bostwick DG. The extent and mul- 24. Frangioni JV, Neel BG. Solubilization and purifica- 33. Litman T, Druley TE, Stein WD, Bates SE. From S ticentricity of high-grade prostatic intraepithelial neo- tion of enzymatically active glutathione -transferase MDR to MXR: new understanding of multidrug resis- plasia in clinically localized prostatic adenocarcinoma. (pGEX) fusion proteins. AnalBiochem 1993;210: tance systems, their properties and clinical signifi- Hum Pathol1997;28:143 ^ 8. 17 9 ^ 8 7. cance. CellMolLife Sci 2001;58:931 ^ 59. 25. Sambrook J, Fritsch ED, Maniatis T. Molecular 17. SakrWA. Prostatic intraepithelial neoplasia: a marker 34. Schmitz G, Drobnik W. ATP-binding cassette trans- cloning: a laboratory manual. 2nd ed. Cold Spring for high-risk groups and a potentialtarget for chemo- porters in macrophages: promising drug targets for Harbor (NY): Cold Spring Harbor Laboratory Press; prevention. Eur Urol1999;35:474^8. treatment of cardiovascular disease. Curr Opin Inves- 1989. p. 2.82 ^ 2.108. tig Drugs 2002;3:853 ^ 9. 18. Orozco R, O’Dowd G, KunnelB, Miller MC, Veltri 26. LangmannT,MauererR,ZahnA,etal.Real-time 35. Dean M, Rzhetsky A, Allikmets R. The human ATP- RW. Observations on pathology trends in 62,537 reverse transcription-PCR expression profiling of the binding cassette (ABC) transporter superfamily. Ge- prostate biopsies obtained from urology private prac- complete human ATP-binding cassette transporter nome Res 2001;11:1156 ^ 66. tices in the United States. Urology 1998;51:186 ^ 95. superfamily in various tissues. Clin Chem 2003;49: 36. Petry F, Kotthaus A, Hirsch-Ernst KI. Cloning of hu- 19. Gokden N, RoehlKA, CatalonaWJ, Humphrey PA. 230 ^ 8. man and rat ABCA5/Abca5 and detection of a human High-grade prostatic intraepithelial neoplasia in needle 27. Towbin H, Staehelin T, Gordon J. Electrophoretic splice variant. Biochem Biophys Res Commun 2003; biopsy as risk factor for detection of adenocarcinoma: transfer of proteins from polyacrylamide gels to nitro- 300:343^50. current level of risk in screening population. AmJ Surg cellulose sheets: procedures and some applications. 37. PetryF,RitzV,MeinekeC,etal.Cellularlocalization Pathol2004;28:629 ^ 33. Proc NatlAcad Sci U S A1979;76:4350 ^ 4. of rAbca5, a rat ATP-binding cassette transporter 20. BisharaT, Ramnani DM, EpsteinJI. High-grade pros- 28. Stearns ME,Wang M. Immunoassays of the metal- expressed in Leydig cells, and characterization of its tatic intraepithelial neoplasia on needle biopsy: risk of loproteinase (MMP-2) and tissue inhibitor of metallo- splice variant apparently encoding a half-transporter. cancer on repeat biopsy related to number of involved proteinase (TIMP 1 and 2) levels in noninvasive and Biochem J 2006;393:79 ^ 87. cores and morphologic pattern. Am J Surg Pathol metastatic PC-3 clones: effects of taxol. Oncol Res 38. Price D, Stein B, Sieber P, et al. Toremifene for the 2004;28:629 ^ 33. 1994;6:195^201. prevention of prostate cancer in men with high grade 21. Bostwick DG. High grade prostatic intraepithelial 29. Harlow E, Lane D. Antibodies: a laboratory manual, prostatic intraepithelial neoplasia: results of a double- neoplasia: the most likely precursor of prostate cancer. vol. 1. Cold Spring Harbor (NY): Cold Spring Harbor blind, placebo controlled, phase IIB clinical trial. J Urol Cancer 1995;75:1823^ 36. Laboratory; 1989. p. 563^6. 2006;176:965 ^ 71.

Clin Cancer Res 2007;13(3) February1,2007 938 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. The ABCA5 Protein: A Urine Diagnostic Marker for Prostatic Intraepithelial Neoplasia

Youji Hu, Min Wang, Karen Veverka, et al.

Clin Cancer Res 2007;13:929-938.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/13/3/929

Cited articles This article cites 33 articles, 5 of which you can access for free at: http://clincancerres.aacrjournals.org/content/13/3/929.full#ref-list-1

Citing articles This article has been cited by 1 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/13/3/929.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/13/3/929. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research.