Clinical Cancer Research Versailles No Abst Advances in Breast Cancer
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Chapman, E. J.
Right arrow Articles by Knowles, M. A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Chapman, E. J.
Right arrow Articles by Knowles, M. A.
Clinical Cancer Research Vol. 11, 5740-5747, August 15, 2005
© 2005 American Association for Cancer Research

Human Cancer Biology

Comprehensive Analysis of CDKN2A Status in Microdissected Urothelial Cell Carcinoma Reveals Potential Haploinsufficiency, a High Frequency of Homozygous Co-deletion and Associations with Clinical Phenotype

Emma J. Chapman1, Patricia Harnden1, Philip Chambers2, Colin Johnston1 and Margaret A. Knowles1

Authors' Affiliations: 1 Cancer Research UK Clinical Centre and 2 Cancer Research UK Mutation Detection Facility, St. James's University Hospital, Leeds, United Kingdom

Requests for reprints: Margaret A. Knowles, Cancer Research UK Clinical Centre, St. James's University Hospital, Beckett Street, Leeds, LS97TF, United Kingdom. Phone: 44-113-206-4913; Fax: 44-113-242-9886; E-mail: margaret.knowles{at}cancer.org.uk.


   Abstract
Top
 Abstract
Materials and Methods
 Results
 Discussion
 References
 
Purpose: There are significant differences in reported frequencies, modes of inactivation, and clinical significance of CDKN2A in urothelial cell carcinoma (UCC). We aimed to address these issues by investigating all possible modes of inactivation and clinicopathologic variables in a single tumor panel.

Experimental Design: Fifty microdissected UCCs were examined. CDKN2A gene dosage (quantitative real-time PCR), allelic status (microsatellite analysis), hypermethylation (methylation-specific PCR), mutation status (denaturing high-performance liquid chromatography and sequencing), protein expression (immunohistochemistry), and clinicopathologic variables (stage, grade, and disease recurrence during follow-up) were assessed.

Results: Exon 2 was underrepresented in 20 of 46 (43%) and exon 1ß in 21 of 46 (46%) of cases. Underrepresentation of exon 2 was accompanied by loss of heterozygosity (LOH) of 9p in 6 of 18 (30%) and of exon 1ß in 11 of 19 assessable cases (58%). Overall, LOH of 9p was identified in 15/41 (37%). Homozygous deletion of exons 2 and 1ß was detected in 16 of 46 (35%) and 10 of 46 tumors (22%), respectively. Co-deletion was most common, but exon 2–specific homozygous deletion was also detected. In tumors without homozygous deletion, p16 promoter hypermethylation was detected in 1 of 18 (6%). Hypermethylation of the p14ARF promoter or mutations in CDKN2A were not observed. Homozygous deletion of exon 2 or LOH on 9p were associated with invasion. Homozygous deletion of exon 2 or exon 1ß was associated with recurrent disease.

Conclusions: These results confirm CDKN2A as a clinically relevant target for inactivation in UCC and show that the true frequency of alteration is only revealed by comprehensive analysis. Our results suggest that CDKN2A may be haploinsufficient in human cancer.

Loss of heterozygosity (LOH) of chromosome 9 is one of the most common events in urothelial cell carcinoma (UCC) of the bladder. Using multiple polymorphic markers, key regions of deletion have been identified on 9p and 9q. On 9p, the critical region includes the CDKN2A locus at 9p21 (1). CDKN2A encodes two proteins, p16 and p14ARF (p19ARF in mice), which are translated from alternatively spliced mRNA and each regulated by a unique promoter. p16 is encoded by exons 1{alpha}, 2, and 3 and p14ARF by an alternative exon 1, 1ß, and the same exons 2 and 3 in an alternative reading frame (Fig. 1). Inactivation of this locus is one of the most frequent genetic changes seen in human solid tumors and can occur by deletion, methylation, or mutation.



View larger version (13K):
[in this window]
[in a new window]
[Download PPT slide]
  		Fig. 1. CDKN2A locus. Exons are represented as boxes and coding regions are shaded. p16 is encoded by exons 1{alpha}, 2, and 3. p14ARF is encoded by an alternative first exon, 1ß, and the same exons 2 and 3 in an alternative reading frame. Sequences encoding p16 (black) and sequencing encoding p14ARF (gray).

 
The importance of inactivation of CDKN2A in cancer development stems from the fact that its products are involved in the Rb and p53 tumor suppressor pathways. As its products are encoded by shared exons, inactivation of the locus has the potential to inactivate both pathways simultaneously. p16 inhibits the phosphorylation of Rb by cyclin D-CDK4/6 complexes and thus inhibits cell cycle progression in G1. p16 has a role in mediating senescence in human urothelial cells (2) and its loss is necessary for immortalization of some urothelial cancer cells in culture (3). In addition, p16 has other potential tumor suppressor effects, in modulating anchorage-independent growth (4), anoikis (5), angiogenesis (6), and tumor cell invasion (7). p14ARF inhibits the degradation of p53 by binding to hdm2 to help mediate p53-dependent G1 or G2 arrest or apoptosis. p14ARF also has p53-independent actions such as the inhibition of HIF-1 (8), interaction with topoisomerase (9), and the promotion of caspase-9-induced apoptosis (10).

Other members of the INK4 family of cyclin-dependent kinase (CDK) inhibitors such as p27 show haploinsufficiency in mice (11). Mouse models also suggest that partial down-regulation of p16 and/or p19ARF expression alone (12) may yield a phenotypic effect. In vitro data showing that an incomplete reduction in p19ARF or p16 expression leads to an extended lifespan in mouse embryonic fibroblasts (13) supports this concept. There is some evidence to suggest that a partial inactivation of CDKN2A may be of clinical significance in particular tumor types (14). However, to date, haploinsufficiency of CDKN2A has not been fully considered in human cancer.

The potential for the inactivation of CDKN2A to play a vital role in tumorigenesis has prompted widespread investigation of its status in bladder cancer. The mode of CDKN2A inactivation seems to be tumor-type specific. In UCC, homozygous deletion is the major mode of inactivation (15, 16). LOH at 9p21 is reported in up to 59% of cases (17), although how this may relate to gene dosage has not been considered. Hypermethylation of CpG islands in the p16 and p14ARF promoters has been reported in 18% and 56% of UCC, respectively (16, 18). Finally, point mutation is a rare event and has been detected in 0% to 9% of cases (16, 19).

In UCC, homozygous deletions at CDKN2A typically encompass both the p16 and p14ARF genes. Selective inactivation of p14ARF without p16 has been reported in other sporadic tumors (20) and a p14ARF-specific germline mutation has been described in familial melanoma-neural tumor syndrome (21). However, in UCC, specific inactivation of p14ARF occurs only rarely (22); thus, in this type of tumor, it seems that either inactivation of both genes is necessary for tumorigenesis or that inactivation of p14ARF alongside p16 is simply a coincidence of their proximity and unusual exon sharing arrangement. Despite numerous studies, there are significant differences in reported frequencies and modes of inactivation of CDKN2A in UCC. Although geographic variation and patient selection cannot be excluded, technical differences are the most likely explanation. Some studies have not used microdissected tumors, and therefore the detection of deletions, particularly homozygous deletion, is masked by the presence of contaminating normal cell DNA. In PCR-based assays, designation of a cutoff point for homozygous deletion based upon an estimation of the amount of product attributable to normal contamination is likely to be inaccurate unless each tumor is individually assessed for contamination. Current knowledge is also limited by the fact that some studies were done before the discovery of p14ARF and not all studies since have considered the alternative transcript.

Generally no link between CDKN2A status and clinicopathologic variables has been found in UCC, although associations of LOH/homozygous deletion with larger tumor size and lower recurrence-free survival (15) have been described. As some modes of inactivation can coexist, an incomplete picture of gene status is achieved from studies concentrating on only a single mode of inactivation. We set out to do a comprehensive study of CDKN2A status, investigating gene dosage, allelic status, hypermethylation, mutation, protein expression, and clinicopathologic variables in the same panel of microdissected UCCs.


   Materials and Methods
Top
 Abstract
 Materials and Methods
Results
 Discussion
 References
 
Tissue samples. Tissue was obtained with consent and the approval of the Local Research Ethics Committee from patients undergoing transurethral resection for UCC at St. James's University Hospital, Leeds, United Kingdom, in 1999 to 2001. A blood sample was collected into EDTA tubes and high molecular weight DNA was extracted using the Nucleon Genomic DNA extraction kit (Tepnel, Manchester, United Kingdom). Fifty tumors were fixed in 10% formalin and paraffin embedded. If possible, an unfixed piece of the same tumor was frozen in liquid nitrogen. Tumors were graded according to the 1973 WHO recommendations and staged according to tumor-node-metastasis classification. Ten-micrometer paraffin-embedded sections were stained with H&E and tumor cells were isolated from contaminating stroma, normal tissue, or immune infiltrate using laser capture microdissection (10,000 pulses of 30-µm beam; Arcturus, Mountain View, CA). Clinicopathologic data (tumor stage, grade, whether the tumor was a primary or recurrent tumor, and recurrence during follow-up) was collected after completion of laboratory analyses, a minimum of 3 years after collection of the initial sample. Tumor DNA was extracted using a QIAamp DNA mini kit (Qiagen, Sussex, United Kingdom) according to the manufacturer's instructions except that 60 µL of elution buffer were left on the column for 5 minutes before centrifugation for 1 minute. For methylation and mutation analyses, DNA was extracted from nonmicrodissected, frozen tumor sections and eluted into 60 µL Tris-EDTA. If frozen tissue was not available, DNA for mutation analyses was extracted from paraffin-embedded sections. DNA was quantified with Picogreen double-stranded DNA quantization reagents (Invitrogen, Paisley, United Kingdom) and a FLUOstar Galaxy machine (BMG Labtech Ltd., Aylesbury, United Kingdom).

Quantitative real-time PCR. Quantitative real-time PCR was done using 5 ng of tumor DNA as previously described (22). Primers and probes were designed to target p16 exon 2, p14ARF exon 1ß, and the reference gene liver phosphofructokinase (PFKL). PFKL at 21q22.3 was chosen as the reference gene because chromosome 21 is rarely altered numerically in UCC. For this study, the standard curve was optimized within the range 0.1 to 20 ng DNA so that when using ~5 ng of template DNA from microdissected UCC, the concentration of tumor DNA fell midway within the range of the standard curve, with several standards close to the template concentration. Normalized gene dosage ratios were interpreted as follows: 0 to 0.2 = homozygous deletion, 0.21 to 0.69 = underrepresentation (of the test gene relative to the reference gene), 0.7 to 1.2 = retention of copy number, and >1.2 = overrepresentation. These ratios were used upon the assumption that because the UCC samples had been microdissected, normal tissue contamination of the specimen was <20%.

Detection of loss of heterozygosity by fluorescent microsatellite analysis. Monoplex fluorescent PCR was done for microsatellite markers D9S1748 and D9S1749 on 9p and D9S176 and D9S272 on 9q. Primer sequences, sizes of PCR products and locations are available at The Genome Database (http://gdbwww.gdb.org). Forward primers were labeled with tetrachlorofluorescein or 6-carboxyfluorescein. Reactions (25 µL) contained 10 µmol/L of each primer, 1.5 mmol/L MgCl2, 1 unit Amplitaq Gold (Applied Biosystems, Warrington, United Kingdom), 0.2 mmol/L deoxynucleotide triphosphates (Amersham Biosciences International, Berkshire, United Kingdom), 1x PCR buffer, and 5 ng template. Thermocycler conditions were 95°C for 10 minutes followed by 33 cycles of 95°C for 30 seconds, 55°C for 30 seconds, 72°C for 30 seconds, then 72°C for 7 minutes. PCR products were diluted one in four with H20 and 1 µL was added to 10 µL Hi-Di formamide and 0.25 µL Rox 500 standards (Applied Biosystems). Samples were denatured at 95°C for 2 minutes and electrophoresis was done using a 3100 Genetic Analyser (Applied Biosystems), with a 36-cm capillary array and denaturing POP4 polymer (Applied Biosystems). Data was exported into Genescan software (v 3.1.2). In many cases, a simple visualization of the electropherograms allowed identification of tumors with retention of heterozygosity or LOH (Fig. 2). Where the classification was unclear, allelic balance was calculated using the equation: allelic ratio = (peak height blood allele 1 / peak height blood allele 2) / (peak height tumor allele 1 / peak height tumor allele 2), where allele 1 is the smaller sized allele and allele 2 is the larger one. As samples had been microdissected, normal contamination was minimal. Therefore, if one allele was absent in the tumor (LOH), the peak height produced from contamination was <20% the expected allele height, compared with the allele in the blood. A reduction of >80% of allele height ratio in tumor compared with blood DNA was scored as LOH. Reductions in ratios between 0% and 79% were scored as retention of heterozygosity. A reduction of >80% in height of both alleles relative to the blood was scored as homozygous deletion. PCR reactions that seemed to show homozygous deletion were repeated. To discriminate between a failed PCR and homozygous deletion, the DNA was required to generate product from at least one allele of another marker.



View larger version (11K):
[in this window]
[in a new window]
[Download PPT slide]
  		Fig. 2. Electropherograms for microsatellite marker D9S1748. Allelic balance is determined by comparing electropherograms of products generated from tumor DNA (bottom row) and blood DNA (top row) derived from the same patient. Examples show retention of heterozygosity (UCC 448), LOH (UCC 484), and homozygous deletion (HD, UCC 498). *, two allele peaks.

 
Methylation-specific PCR. One microgram of tumor DNA plus 10 µg salmon sperm DNA (Sigma, St. Louis, MO) in a total volume of 100 µL H20 was treated with the CpGenome DNA modification kit (Intergen, Oxford, United Kingdom) and eluted into 60 µL Tris-EDTA. Methylation-specific PCR was done for p16 and p14ARFpromoter regions on tumors without homozygous deletion, as described previously (23). PCR products were separated in 2% agarose gels and visualized under UV. Modified DNA from the DLD1 cell line, was a positive control for methylated sequences and the negative control for unmethylated sequences. Negative control for methylation was modified human placental DNA (Sigma). A no-template control was also included.

Mutation analyses. In tumors without homozygous deletion, all of exons 1ß and 2 and the translated regions of exons 1{alpha} and 3 were amplified. Reactions contained 1 unit Amplitaq Gold polymerase (Applied Biosystems), PCR buffer (Applied Biosystems), 1.5 mmol/L MgCl2, 200 µmol/L of each deoxynucleotide triphosphate, 0.5 pmol µL–1 of each primer, 5% DMSO and 2.5 ng µL–1 genomic DNA from nonmicrodissected tumors. Primer sequences (forward then reverse) were exon 1ß-CACCTCTGGTGCCAAAGGGC/CCTAGCCTGGGCTAGAGACG, exon 1{alpha}-CAGCACCGGAGGAAGAAAG/ GCGCTACCTGATTCCAATTC, exon 2-GGAAATTGGAAACTGGAAGC/ TCTGAGCTTTGGAAGCTC, exon 3-CCATTGCGAGAACTTTATCC/TGGACATTTACGGTAGTGGG/. Cycling conditions were 95°C for 12 minutes and then 35 cycles of 95°C for 30 seconds, 55°C for 30 seconds, 72°C for 30 seconds followed by incubation at 72°C for 10 minutes with a final hold at 15°C. If frozen tumor was not available, exon 2 was amplified in three fragments using DNA extracted from paraffin-embedded tumor. In these cases, PCR was done as described above but with a MgCl2 concentration of 3.0 mmol/L and an annealing temperature of 62°C for 30 seconds. Primer sequences were, as follows: fragment A-AGCTTCCTTTCCGTCATGC/ GCAGCACCACCAGCGTG, fragment B-AGCCCAACTGCGCCGAC/ CCAGGTCCACGGGCAGA, and fragment C-TGGACGTGCGCGATGC/GGAAGCTCTCAGGGTACAAATTC. PCR products were denatured at 95°C for 5 minutes and then cooled to 65°C. Denaturing high-performance liquid chromatography was carried out at the temperatures determined by the "DHPLC Melt" program (24). A Transgenomic WAVE Nucleic Acid Fragment Analysis system and DNASep column (Transgenomic, Cheshire, United Kingdom) were used. Data analysis was by visual inspection of chromatograms by two independent observers. Cases with possible polymorphisms or mutations were sequenced.

Immunohistochemistry. For detection of p16, 5-µm deparaffinized and rehydrated sections were treated with 3% hydrogen peroxide (Sigma), Avidin Biotin blocking kit (Vector Laboratories, Peterborough, United Kingdom), and then the catalyzed signal amplification system (DakoCytomation, Buckinghamshire, United Kingdom). Primary antibody was mouse anti-p16 (Ab-7; Labvision, Suffolk, United Kingdom) at 1:1,600 for 1 hour. Slides were counterstained with hematoxylin, dehydrated, and mounted. Positive control for p16 expression was cervical carcinoma. p16 was scored as negative, heterogeneous, or strong expression in >50% of tumor cells (25) by three independent observers without knowledge of tumor identity or genotype.

Statistical analyses. Statistical analyses of homozygous deletion of CDKN2A exon 2 (p16) or 1ß (p14ARF) with tumor presentation, stage, grade, or recurrence were done using a two-tailed Fisher's exact test and SAS (version 8.0) computer software. The stages and grades of the 45 tumors, for which gene dosage and clinicopathologic information was available were pTa G1 (six cases), pTa G2 (19 cases), pT1 G2 (five cases), pT1 G3 (eight cases), pT2 G2 (two cases), and pT2 G3 (five cases). pT1 and pT2 tumors were grouped together as invasive tumors. Statistical analyses were also done in relation to LOH on 9p or 9q or on both 9p and 9q. P ≤ 0.05 was taken to indicate a statistically significant relationship. P values were not adjusted for multiple comparisons.


   Results
Top
 Abstract
 Materials and Methods
 Results
Discussion
 References
 
CDKN2A status. Gene dosage at CDKN2A exon 2 and exon 1ß was successfully measured in 46 of 50 tumors. Three samples were excluded as DNA yield or quality was not sufficient and one was excluded because it was found to be a biopsy of a previous resection site. Homozygous deletion of exons 2 and 1ß was detected in 16 of 46 (35%) and 10 of 46 (22%) of tumors, respectively. Specific deletion of p14ARF exon 1ß was not detected. Exon 2–specific deletions were identified in six cases. In five of these, homozygous deletion was accompanied by underrepresentation of p14ARF exon 1ß and in one case, with retention of exon 1ß gene dosage (Table 1; Fig. 3). A reduction in gene dosage was the most common event at CDKN2A. Exon 2 was underrepresented in 20 of 46 (43%) of tumors and exon 1ß in 21 of 46 (46%). We were interested to determine whether frequent reduction in CDKN2A gene dosage corresponded to LOH in the 9p21 region. Microsatellite marker-based LOH analysis was successfully done in 41 of 46 tumors (Figs. 2 and 3). Five cases were either noninformative for the markers investigated, constitutive DNA was not available, or the PCR reaction repeatedly failed. LOH of at least one marker on 9p was detected in 15 of 46 tumors (33%). Underrepresentation of exon 2 was accompanied by 9p LOH in 6 of 18 (30%) and of exon 1ß in 11 of 20 assessable cases (55%). Allelic balance at a 9q marker (D9S972 or D9S176) was measured to obtain information on the relative status of 9q to 9p in the same tumors. LOH on 9q was detected in 24 of 44 tumors (55%). Predicted loss of an entire parental homologue (i.e., LOH of at least one marker on both q and p arms) was detected in 12 of 40 tumors (30%). Overall, CDKN2A was altered by a reduction in gene dosage and/or LOH in 36 of 46 tumors (78%) indicating that alteration of this locus in involved in the genesis of the vast majority of UCC.


View this table:
[in this window]
[in a new window]

  		Table 1. Patterns of CDKN2A gene dosage as determined by quantitative real-time PCR

 


View larger version (20K):
[in this window]
[in a new window]
[Download PPT slide]
  		Fig. 3. Gene dosage and allelic balance at CDKN2A and 9q. Boxes represent gene dosage relative to PFKL, determined by quantitative real-time PCR: retention of copy number (white), underrepresentation (gray), homozygous deletion (HD, black). Allelic balance determined by microsatellite marker analysis: retention (R), LOH, and noninformative/no data (–). UCC 372 had hypermethylation of p16 promoter region.

 
Mutation and methylation analyses. No mutations were detected in CDKN2A exons 1{alpha}, 1ß, 2, or 3. Previously reported polymorphisms (26, 27) were detected in exon 3 in tumors 355, 463, and 467 and in exon 2 in tumor 393. For tumors without homozygous deletion, PCR was done with primers specific for either the nonmethylated or methylated p16 or p14ARF promoter sequence. As these tumors were not microdissected, all should have produced a product with at least one primer set. DNA from UCC 364 failed to produce a PCR product for either methylated or unmethylated p16 and therefore was not included in the analysis. In contrast to the 56% previously described (18), hypermethylation of p14ARF promoter was not seen in any tumor but was detected in positive control DNA (Fig. 4). Hypermethylation of the p16 promoter was detected in 1 of 18 tumors examined (UCC 372) and was always detected in the positive control DNA (data not shown). All other tumors produced a product for the nonmethylated p16 sequence. No product was produced from negative or no template controls with either primer pair.



View larger version (21K):
[in this window]
[in a new window]
[Download PPT slide]
  		Fig. 4. Methylation-specific PCR for p14ARF CpG island. Unmethylated sequence produced a band of 132 bp and methylated a band of 122 bp. No methylated sequence was detected in any UCC investigated. m, primers for methylated sequence; u, unmethylated sequence; nt, no template control; +, positive; –, negative controls.

 
Immunohistochemistry. Immunohistochemistry was done on paraffin sections of 43 of 46 tumors and was concordant with CDKN2A status in that homozygous deletion of exon 2 correlated with absence of p16 expression in tumor cells (Fig. 5A). Twelve of 15 cases with homozygous deletion showed a complete lack of p16 staining in tumor cells yet positive internal control cells in the stroma. Only three tumors had staining patterns potentially discrepant with gene dosage. One case (387) was negative in its major papillary component but did contain a small invasive component with positive staining. UCCs 446 and 360 (gene dosages indicating homozygous deletion) had rare positive cells. It is possible that these areas were not captured during microdissection and consequently the tumors were scored as having homozygous deletion by genetic analysis. p16 expression was detected in all tumors without homozygous deletion of exon 2, but there was no clear correlation between gene dosage and level of protein expression. The pattern of p16 expression ranged from homogeneous to heterogeneous (Fig. 5B). Tumor 447 (underrepresentation of exon 2 with LOH) contained distinct areas of negative and positive staining that seemed to be two tumor clones (Fig. 5C).



View larger version (49K):
[in this window]
[in a new window]
[Download PPT slide]
  		Fig. 5. p16 expression by immunohistochemistry (CDKN2A status), original magnification x 100. A, absence of expression in UCC 498 (homozygous deletion) but positive stromal cells. B, heterogeneous expression in UCC 452 (underrepresentation without LOH). C, two distinct tumor clones in UCC 447 (underrepresentation with LOH).

 
Correlation of CDKNA status with clinicopathologic variables. Clinicopathologic information was obtained in retrospect. Eleven tumors were disease recurrences and the remainder were first presentations. Clinicopathologic data was not available for UCC 500. Thirty-two of 45 patients (71%) experienced recurrent disease during the follow-up period. Homozygous deletion of exon 2 or LOH on 9p, 9q, or both 9p and 9q were associated with invasion (Table 2). Homozygous deletion of exon 2 or exon 1ß was not predictive of disease recurrence but was more common in recurrences rather than initial presentations. Similarly, LOH on 9q was also more common in recurrences but when present in the initial tumor was not predictive of disease recurrence.


View this table:
[in this window]
[in a new window]

  		Table 2. Statistically significant (P ≥ 0.05) associations between CDKN2A status and clinicopathologic variables

 

   Discussion
Top
 Abstract
 Materials and Methods
 Results
 Discussion
References
 
We have confirmed that chromosome 9 is a major target for alteration in UCC. Thirty-four of 40 (85%) assessable cases (those with gene dosage information for CDKN2A plus allelic balance information for at least one marker on 9p and 9q) had some alteration of the chromosome. CDKN2A is likely to be the key target of these chromosome 9 alterations as only 10 of 46 UCC (22%) retained gene dosage of both p16 and p14ARF. One of these tumors (UCC 372) potentially had silencing of the p16 gene by promoter hypermethylation. As the CDKN2A locus is commonly inactivated in tumors by wide deletions targeting both genes, it has been suggested that deletion of exon 1ß may be coincidental. Our findings support this suggestion, as specific targeting of exon 2 but not exon 1ß was detected and all homozygous deletions of exon 1ß were accompanied by homozygous deletion of exon 2. However, it could be argued that because homozygous deletion of exon 2 could prevent p14ARF expression due to exon sharing, deletion of exon 1ß would have no additional benefit. If this is the case, homozygous deletion of exon 1ß may be irrelevant unless p14ARF is the major 9p21 target. Whether the transcript encoded by exon 1ß alone is able to encode a protein with all the functions of p14ARF is unclear. It is reported that the first 25 amino acids of the protein are sufficient to perform the key roles of p14ARFor p19ARF in inhibiting the hdm2/mdm2-mediated degradation of p53 (28) and in inducing cell cycle arrest. Indeed, adenovirus-mediated transduction with the exon 1ß fragment can suppress growth and viability of human cells in vitro (29). However, the exon 2–encoded sequence is required for some functions of p14/p19ARF such as interaction with topoisomerase I (30) and nucleolar localization of the human protein (31). Whether p14ARF can exert its tumor suppressive activity without nucleolar localization is not known, but mutations in exon 2 affect the proper function of both p16 and p14ARF in melanoma cells (32). Thus, exon 1ß could encode a functional protein with at least some tumor suppressor activities and this could explain the frequent codeletion with p16 exon 2 in UCC. However, as we identified exon 2–specific deletions in six tumors, deletion of exon 1ß cannot be essential. Homozygous deletion of exon 2 was detected in 16 of 46 tumors (35%). Homozygous deletion of p16 has been reported previously in 14.1% of undissected bladder tumors by multiplex PCR (15) and in 29% by fluorescence in situ hybridization (33). Here, homozygous deletion of p14ARF exon 1ß was detected in 10 of 46 microdissected tumors (22%). This frequency is higher than reported recently in a study of undissected tumors using the same method (34). We believe that our use of a sensitive quantitative real-time PCR method with a carefully selected reference gene and microdissected tumor material has allowed accurate measurement of CDKN2A gene dosage. The high frequency of deletion was confirmed by homozygous deletion of the microsatellite marker, D9S1748, between exons 2 and 1ß in tumors where both genes were deleted. The frequency of homozygous deletion detected here was similar to that previously detected by us using the same method on a separate panel of nonmicrodissected UCC (22). However, a clear advantage of the use of microdissected material and consequently a higher gene dosage ratio threshold for retention of copy number was the ability to detect a higher frequency of underrepresentation than previously reported.

Of the six tumors with p16 exon 2–specific deletions, four were accompanied by a reduction in gene dosage of p14ARF and LOH at flanking markers. This may represent LOH at 9p21, with a targeted p16-specific homozygous deletion, consistent with the classic two-hit hypothesis of tumor suppressor gene inactivation. A reduction in gene dosage of p14ARF exon 1ß accompanied by homozygous deletion of p16 exon 2 is likely to be of phenotypic significance. Mice null for p16 and heterozygous for p19ARF are tumor prone but those null for p16 only are not (35). Similarly, Carnero et al. (13) showed that incomplete inhibition of p19ARF expression in mouse embryonic fibroblasts was sufficient to cause a phenotypic effect in vitro. These studies indicate that a reduction in p19ARF gene dosage may have an independent role in mouse tumorigenesis. Whether this is also the case in human cancer is not yet clear. A reduction in gene dosage (underrepresentation) was the most frequent event at CDKN2A, indicating that the locus may be haploinsufficient. Indeed, a statistically significant association of LOH in the CDKN2A region with invasive tumors and high grade was observed and we were unable to find evidence for inactivation of the retained allele by any other mechanism. LOH was seen at flanking markers and so is the likely cause of underrepresentation in approximately one third of cases. LOH of one or more markers on 9p21 was seen in 33% of tumors, but it is likely that we have underestimated the frequency of 9p LOH due to the frequent homozygous deletion in our tumor panel. Certainly, in samples 498, 387, 446, and 351, a large homozygous deletion spanning the 9p21 region encompassed all the markers we investigated. Up to 40% of stage T1 UCC and virtually all higher stage tumors are aneuploid not diploid (36). Therefore, aneuploidy or small hemizygous deletions could explain the 11 cases where exon 2 and the 12 cases where exon 1ß were underrepresented in the absence of detectable LOH. These tumors had retained at least one copy of both alleles in the CDKN2A region but had fewer copies of 9p than of PFKL. It is also possible that apparent underrepresentation is a reflection of a heterogeneous tumor containing some cells with homozygous deletion and some with retention of copy number. Indeed, "heterogeneous" and "clonal" patterns of p16 expression were visible in some tumors (Fig. 5B and C). Alternatively, tumors showing a reduction in gene dosage but no LOH may contain small hemizygous deletions targeting the exons but not flanking markers.

The frequency of p16 hypermethylation (6%) detected in our study was lower than previously reported (18) and methylation of the p14ARF promoter was not seen. The reason for the discrepancy with the study of Dominguez et al. is unclear but may be because of the small size of our tumor panel. The only other study that has examined p14ARF and p16 promoter methylation in bladder tumors, detected hypermethylation of p14ARF in 5% of cases and of p16 in 9% (37).

Other studies have concluded that mutations at CDKN2A are rare in UCC. Accordingly, we found no evidence of point mutations. Three common single nucleotide polymorphisms were identified. As these occurred in the 3' untranslated region, they do not cause an amino acid change but may affect expression of p16 or p14ARF (38). Neither single nucleotide polymorphism is reported to affect bladder cancer risk, but both have been associated with shorter survival of bladder cancer patients and one with disease progression (27).

A key clinical problem in UCC is disease recurrence after initial resection and currently all patients are followed by regular cystoscopies to treat recurrences early and to prevent progression. However, up to 30% of patients remain tumor free and therefore undergo unnecessary investigations. Current clinical and pathologic variables cannot reliably assess recurrence risk or identify the smaller group of patients with tumors that are destined to progress to muscle invasion and who would benefit from more aggressive initial therapy. Many studies have considered the relationship between the status of p16 and/or p14ARF genes and have found no relationship with recurrence (39) or stage and grade (34). However, LOH of 9p21 has been linked with a more aggressive phenotype (40) and an association of LOH/homozygous deletion with reduced recurrence-free survival and larger tumor size has been reported (15). Here we have found statistically significant associations between CDKN2A status with invasion beyond the basement membrane and the presence of recurrent disease. As the tumor panel was small, we feel that this highlights potential relationships of interest that merit further investigation rather than providing robust correlations. However, high accuracy of detection of homozygous deletion may render such associations clear even in a small population.

In agreement with previous studies, homozygous deletion of exon 2 or exon 1ß was not predictive of tumor recurrence. However, a higher frequency of homozygous deletion was detected in tumors that were recurrent disease compared with initial presentations. This suggests that although homozygous deletion of CDKN2A is not predictive of recurrence, it may be an event involved in the development of the recurrent phenotype. If so, this raises the question of whether some primary UCC are already "preprogrammed" to recur or whether this phenotype is acquired later. If homozygous deletion of p16 is necessary for recurrence, post-resection gene therapy to restore p16 expression in residual tumor cells or the induction of p16 expression by pharmacologic means may be a way to overcome this key clinical problem. CDKN2A status in primary and recurrent tumors from the same patient has not yet been examined and this will be an interesting area of investigation.

An association of LOH on 9q but not on 9p with recurrent tumors was also observed suggesting that loss of function of a 9q gene is independently involved in recurrence. Indeed, TSC1 and DBC1, tumor suppressor genes implicated in UCC, reside on 9q (41, 42). Our present results therefore support the notion that genes on both arms of this chromosome play a key role in UCC development. LOH on both 9p and 9q simultaneously was seen in approximately one third of tumors. LOH spanning both p and q arms has been associated with an increased risk of recurrence (43), which has been attributed to loss of the TSC1 gene at 9q34 rather than the 9p21 region. Here, this association was not observed but LOH of both chromosome arms was associated with invasion indicating a contribution to tumor phenotype. An association of p16 loss with invasion in UCC and other tumors has been suggested (44, 45). However, here we report the first significant association of homozygous deletion of exon 2 with invasion. p16 has cell cycle–independent functions in promoting anoikis (4), matrix-dependent cell spreading (46), and directional motility (7), all of which could contribute to invasion. An association of LOH on 9p21 with invasive tumors (pT1 and above) was also identified and was more significant than that with homozygous deletion. Overall, LOH of 9p21 was the most frequent event at CDKN2A and was always accompanied by either a reduction in gene dosage or homozygous deletion of exon 2 or both exon 2 and exon 1ß. As LOH was not always accompanied by homozygous deletion, this raises the possibility that underrepresentation of CDKN2A rather than complete deletion may be sufficient to contribute to the invasive phenotype and that subsequent homozygous deletion is not of any further benefit to invasion but may contribute to other characteristics. LOH on 9q was not associated with invasion. Thus, LOH of CDKN2A seems of greatest significance to the invasive phenotype.

The status of p14ARF exon 1ß was not associated with stage or grade. However, this does not rule out a functional role for inactivation of this gene in UCC. The role of p14ARF as a tumor suppressor in human cancer remains unclear. Its loss can cause inappropriate cell cycling. However, cell cycle–independent functions have also been described, including an interaction with topoisomerase 1 (9), which is involved in DNA supercoiling during transcription and replication. Interestingly, unlike the p53-dependent functions, this function required exon 2– but not exon 1ß–encoded sequence (30). p14ARF is also reported to inhibit the transcriptional ability of HIF-1 (8), and p19ARF has effects on angiogenesis in mouse models (47, 48), a role which may be important in papillary UCC, which are highly vascular. Thus, p53-independent functions of p14ARF have the potential to be tumor suppressive and might be associated with clinicopathologic variables such as microvessel density, which were not investigated in this study.

In conclusion, the use of laser capture microdissection and quantitative real-time PCR has allowed us to make accurate measurements of CDKN2A gene dosage and to uncover potential phenotype-genotype relationships that now require confirmation in a larger tumor panel. Complementation of gene dosage data with LOH, methylation, and mutations analyses has provided the most comprehensive study of the CDKN2A in UCC to date. A key question is whether underrepresentation of CDKN2A is sufficient for a phenotypic effect and whether homozygous deletion of either or both genes confers some additional advantage. The frequent but not compulsory coinactivation of the two genes suggests that both have important and discrete tumor suppressor activities, as expected from their different physiologic roles. With this knowledge, we are better placed to investigate genotype-phenotype relationships and clinical significance of inactivation of these genes in UCC. As a model for the role of deletion of CDKN2A in UCC is developed, it will be important to consider that the two pleiotropic proteins it encodes can potentially interact with factors involved in cell cycle control, angiogenesis, apoptosis, and cell migration.


   Acknowledgments
 
We thank Claire Taylor for her assistance in scoring of denaturing high-performance liquid chromatography chromatograms and the Urology team and patients of St. James's University Hospital for their donations of tissue.


   Footnotes
 
Grant support: Cancer Research UK.

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.

Received 2/24/05; revised 5/ 9/05; accepted 5/18/05.


   References
Top
 Abstract
 Materials and Methods
 Results
 Discussion
 References
 

  1. Devin J, Keen AJ, Knowles MA. Homozygous deletion mapping at 9p21 in bladder carcinoma defines a critical region within 2cM of IFNA. Oncogene 1994;9:2757–60.[Medline]
  2. Schwarze SR, Shi Y, Fu VX, Watson PA, Jarrard DF. Role of cyclin-dependent kinase inhibitors in the growth arrest at senescence in human prostate epithelial and uroepithelial cells. Oncogene 2001;20:8184–92.[CrossRef][Medline]
  3. Yeager TR, DeVries S, Jarrard DF, et al. Overcoming cellular senescence in human cancer pathogenesis. Genes Dev 1998;12:163–74.[Abstract/Free Full Text]
  4. Plath T, Detjen K, Welzel M, et al. A novel function for the tumor suppressor p16(INK4a): induction of anoikis via upregulation of the {alpha}(5)ß(1) fibronectin receptor. J Cell Biol 2000;150:1467–78.[Abstract/Free Full Text]
  5. Rocco JW, Sidransky D. p16(MTS-1/CDKN2/INK4a) in cancer progression. Exp Cell Res 2001;264:42–55.[CrossRef][Medline]
  6. Harada H, Nakagawa K, Iwata S, et al. Restoration of wild-type p16 down-regulates vascular endothelial growth factor expression and inhibits angiogenesis in human gliomas. Cancer Res 1999;59:3783–9.[Abstract/Free Full Text]
  7. Natarajan E, Saeb M, Crum CP, Woo SB, McKee PH, Rheinwald JG. Co-expression of p16(INK4A) and laminin 5 {gamma}2 by microinvasive and superficial squamous cell carcinomas in vivo and by migrating wound and senescent keratinocytes in culture. Am J Pathol 2003;163:477–91.[Abstract/Free Full Text]
  8. Fatyol K, Szalay AA. The p14{ARF} tumor suppressor protein facilitates nucleolar sequestration of HIF-1{{alpha}} and inhibits HIF-1 mediated transcription. J Biol Chem 2001;276:28421–9.[Abstract/Free Full Text]
  9. Karayan L, Riou JF, Seite P, Migeon J, Cantereau A, Larsen CJ. Human ARF protein interacts with topoisomerase I and stimulates its activity. Oncogene 2001;19:836–48.
  10. Nakazawa Y, Kamijo T, Koike K, Noda T. ARF tumor suppressor induces mitochondria-dependent apoptosis by modulation of mitochondrial Bcl-2 family proteins. J Biol Chem 2003;278:27888–95.[Abstract/Free Full Text]
  11. Fero ML, Randel E, Gurley KE, Roberts JM, Kemp CJ. The murine gene p27Kip1 is haplo-insufficient for tumour suppression. Nature 1998;396:177–80.[CrossRef][Medline]
  12. Serrano M. The INK4a/ARF locus in murine tumorigenesis. Carcinogenesis 2000;21:865–9.[Abstract/Free Full Text]
  13. Carnero A, Hudson JD, Price CM, Beach DH. p16INK4A and p19ARF act in overlapping pathways in cellular immortalization. Nat Cell Biol 2000;2:148–55.[CrossRef][Medline]
  14. Carter TL, Watt PM, Kumar R, et al. Hemizygous p16(INK4A) deletion in pediatric acute lymphoblastic leukemia predicts independent risk of relapse. Blood 2001;97:572–4.[Abstract/Free Full Text]
  15. Orlow I, LaRue H, Osman I, et al. Deletions of the INK4A gene in superficial bladder tumors. Association with recurrence. Am J Pathol 1999;155:105–13.[Abstract/Free Full Text]
  16. Florl AR, Franke KH, Niederacher D, Gerharz CD, Seifert HH, Schulz WA. DNA methylation and the mechanisms of CDKN2A inactivation in transitional cell carcinoma of the urinary bladder. Lab Invest 2000;80:1513–22.[Medline]
  17. Cairns P, Shaw ME, Knowles MA. Preliminary mapping of the deleted region of chromosome 9 in bladder cancer. Cancer Res 1993;53:1230–2.[Abstract/Free Full Text]
  18. Dominguez G, Carballido J, Silva J, et al. p14ARF promoter hypermethylation in plasma DNA as an indicator of disease recurrence in bladder cancer patients. Clin Cancer Res 2002;8:980–5.[Abstract/Free Full Text]
  19. Asamoto M, Iwahori Y, Okamura T, Shirai T, Tsuda H. Decreased expression of the p16/MTS1 gene without mutation is frequent in human urinary bladder carcinomas. Jpn J Clin Oncol 1997;27:22–5.[Abstract/Free Full Text]
  20. Viswanathan M. Selective deletion of p14 (ARF) exon 1B of the INK4a locus in oral squamous cell carcinoma of Indians. Oral Oncol 2001;37:341–4.[CrossRef][Medline]
  21. Randerson-Moor JA, Harland M, Williams S, et al. A germline deletion of p14(ARF) but not CDKN2A in a melanoma-neural system tumour syndrome family. Hum Mol Genet 2001;10:55–62.[Abstract/Free Full Text]
  22. Aveyard JS, Knowles MA. Measurement of relative copy number of CDKN2A/ARF and CDKN2B in bladder cancer by real-time quantitative PCR and multiplex ligation-dependent probe amplification. J Mol Diagn 2004;6:356–65.[Abstract/Free Full Text]
  23. Esteller M, Tortola S, Toyota M, et al. Hypermethylation-associated inactivation of p14(ARF) is independent of p16(INK4a) methylation and p53 mutational status. Cancer Res 2000;60:129–33.[Abstract/Free Full Text]
  24. Jones AC, Austin J, Hansen N, et al. Optimal temperature selection for mutation detection by denaturing HPLC and comparison to single-stranded conformation polymorphism and heteroduplex analysis. Clin Chem 1999;45:1133–40.[Abstract/Free Full Text]
  25. Benedict WF, Lerner SP, Zhou J, Shen X, Tokunaga H, Czerniak B. Level of retinoblastoma protein expression correlates with p16 (MTS-1/INK4A/CDKN2) status in bladder cancer. Oncogene 1999;18:1197–203.[CrossRef][Medline]
  26. FitzGerald MG, Harkin DP, Silva-Arrieta S, et al. Prevalence of germ-line mutations in p16, p19ARF, and CDK4 in familial melanoma: analysis of a clinic-based population. Proc Natl Acad Sci U S A 1996;93:8541–5.[Abstract/Free Full Text]
  27. Sakano S, Berggren P, Kumar R, et al. Clinical course of bladder neoplasms and single nucleotide polymorphisms in the CDKN2A gene. Int J Cancer 2003;104:98–103.[CrossRef][Medline]
  28. Zhang Y, Xiong Y, Yarbrough WG. ARF promotes MDM2 degradation and stabilizes p53: ARF-INK4a locus deletion impairs both the Rb and p53 tumor suppression pathways. Cell 1998;92:725–34.[CrossRef][Medline]
  29. Saadatmandi N, Tyler T, Huang Y, et al. Growth suppression by a p14(ARF) exon 1ß adenovirus in human tumor cell lines of varying p53 and Rb status. Cancer Gene Ther 2002;9:830–9.[CrossRef][Medline]
  30. Olivier A, Lucie K, Jean-Francois R, Christian-Jacques L, Paule S. Delineation of the domains required for physical and functional interaction of p14ARF with human topoisomerase I. Oncogene 2003;22:1945–54.[CrossRef][Medline]
  31. Lindstrom MS, Klangby U, Inoue R, Pisa P, Wiman KG, Asker CE. Immunolocalization of human p14(ARF) to the granular component of the interphase nucleolus. Exp Cell Res 2000;256:400–10.[CrossRef][Medline]
  32. Rizos H, Darmanian AP, Holland EA, Mann GJ, Kefford RF. Mutations in the INK4a/ARF melanoma susceptibility locus functionally impair p14ARF. J Biol Chem 2001;276:41424–34.[Abstract/Free Full Text]
  33. Balazs M, Carroll P, Kerschmann R, Sauter G, Waldman FM. Frequent homozygous deletion of cyclin-dependent kinase inhibitor 2 (MTS1, p16) in superficial bladder cancer detected by fluorescence in situ hybridization. Genes Chromosomes Cancer 1997;19:84–9.[CrossRef][Medline]
  34. Berggren P, Kumar R, Sakano S, et al. Detecting homozygous deletions in the CDKN2A(p16(INK4a))/ARF(p14(ARF)) gene in urinary bladder cancer using real-time quantitative PCR. Clin Cancer Res 2003;9:235–42.[Abstract/Free Full Text]
  35. Krimpenfort P, Quon KC, Mooi WJ, Loonstra A, Berns A. Loss of p16Ink4a confers susceptibility to metastatic melanoma in mice. Nature 2001;413:83–6.[CrossRef][Medline]
  36. Tribukait B, Gustafson H, Esposti PL. The significance of ploidy and proliferation in the clinical and biological evaluation of bladder tumours: a study of 100 untreated cases. Br J Urol 1982;54:130–5.[Medline]
  37. Esteller M. A gene hypermethylation profile of human cancer. Cancer Res 2001;61:3225–9.[Abstract/Free Full Text]
  38. Aitken J, Welch J, Duffy D, et al. CDKN2A variants in a population-based sample of Queensland families with melanoma. J Natl Cancer Inst 1999;91:446–52.[Abstract/Free Full Text]
  39. Droller MJ. Frequent p16/MTS1 inactivation in early stages of urothelial carcinoma of the bladder is not associated with tumor recurrence. J Urol 2002;168:852.[CrossRef][Medline]
  40. Orlow I, Lianes P, Lacombe L, Dalbagni G, Reuter VE, Cordon-Cardo C. Chromosome 9 allelic losses and microsatellite alterations in human bladder tumors. Cancer Res 1994;54:2848–51.[Abstract/Free Full Text]
  41. Habuchi T, Yoshida O, Knowles MA. A novel candidate tumour suppressor locus at 9q32-33 in bladder cancer: localization of the candidate region within a single 840 kb YAC. Hum Mol Genet 1997;6:913–9.[Abstract/Free Full Text]
  42. Hornigold N, Devlin J, Davies AM, Aveyard JS, Habuchi T, Knowles MA. Mutation of the 9q34 gene TSC1 in sporadic bladder cancer. Oncogene 1999;18:2657–61.[CrossRef][Medline]
  43. Tsukamoto M, Matsuyama H, Oba K, Yoshihiro S, Takahashi M, Naito K. Numerical aberrations of chromosome 9 in bladder cancer. A possible prognostic marker for early tumor recurrence. Cancer Genet Cytogenet 2002;134:41–5.[CrossRef][Medline]
  44. Primdahl H, von der MH, Sorensen FB, Wolf H, Orntoft TF. Immunohistochemical study of the expression of cell cycle regulating proteins at different stages of bladder cancer. J Cancer Res Clin Oncol 2002;128:295–301.[CrossRef][Medline]
  45. Maesawa C, Tamura G, Nishizuka S, et al. Inactivation of the CDKN2 gene by homozygous deletion and de novo methylation is associated with advanced stage esophageal squamous cell carcinoma. Cancer Res 1996;56:3875–8.[Abstract/Free Full Text]
  46. Fahraeus R, Lane DP. The p16(INK4a) tumour suppressor protein inhibits {alpha}vß3 integrin-mediated cell spreading on vitronectin by blocking PKC-dependent localization of {alpha}vß3 to focal contacts. EMBO J 1999;18:2106–18.[CrossRef][Medline]
  47. Gibson SL, Dai CY, Lee HW, et al. Inhibition of colon tumor progression and angiogenesis by the Ink4a/Arf locus. Cancer Res 2003;63:742–6.[Abstract/Free Full Text]
  48. McKeller RN, Fowler JL, Cunningham JJ, et al. The Arf tumor suppressor gene promotes hyaloid vascular regression during mouse eye development. Proc Natl Acad Sci U S A 2002;99:3848–53.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 BloodHome page
S. Sulong, A. V. Moorman, J. A. E. Irving, J. C. Strefford, Z. J. Konn, M. C. Case, L. Minto, K. E. Barber, H. Parker, S. L. Wright, et al.
A comprehensive analysis of the CDKN2A gene in childhood acute lymphoblastic leukemia reveals genomic deletion, copy number neutral loss of heterozygosity, and association with specific cytogenetic subgroups
Blood, January 1, 2009; 113(1): 100 - 107.
[Abstract] [Full Text] [PDF]

Home page
 CarcinogenesisHome page
M. A. Knowles
Molecular subtypes of bladder cancer: Jekyll and Hyde or chalk and cheese?
Carcinogenesis, March 1, 2006; 27(3): 361 - 373.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Chapman, E. J.
Right arrow Articles by Knowles, M. A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Chapman, E. J.
Right arrow Articles by Knowles, M. A.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online