Chromosome 5 harbors two independent deletion hotspots at 5q13 and 5q21 that characterize biologically different subsets of aggressive prostate cancer

Deletion of chromosome 5q is common in prostate cancer and is linked to aggressive disease. Most previous studies focused on 5q21 where CHD1 is located, but deletion of mapping studies has identified a second deletion hotspot at 5q13. To clarify the prevalence and clinical relevance of 5q13 deletions and to determine the relative importance of 5q13 and 5q21 abnormalities, a tissue microarray containing samples from 12 427 prostate cancers was analyzed by fluorescence in situ hybridization. Deletion of 5q13 and 5q21 was found in 13.5% and 10%, respectively, of 7932 successfully analyzed cancers. Deletion was restricted to 5q13 in 49.4% and to 5q21 in 32.0% of cancers with a 5q deletion. Only 18.6% of 5q‐deleted cancers had deletions of both loci. Both 5q13 and 5q21 deletions were significantly linked to advanced tumor stage, high Gleason grade, nodal metastasis and early biochemical recurrence (P < .005 each). Cancers with co‐deletion of 5q13 and 5q21 had a worse prognosis than cancers with isolated 5q13 or 5q21 deletion (P = .0080). Comparison with TMPRSS2:ERG fusion status revealed that 5q21 deletions were tightly linked to ERG negativity (P < .0001) while 5q13 deletions were unrelated to the ERG status. In summary, 5q13 deletion and 5q21 deletion are common, but independent genomic alterations with different functional effects lead to aggressive prostate cancer.

typically large, often extending from centromeric (5q11) to telomeric (5q35) regions, with remarkable variation in size. 9,13,15,16 In addition, apart from the CHD1 locus exists another deletion hot spot located at 5q11-q14. 9,16 It is currently unknown whether this hot spot characterizes a distinct subset of prostate cancers with molecular and/or clinical features that are different from 5q21-deleted cancers.
To learn more about the role of 5q13 deletions, we expanded our previous 5q21 deletion analysis 5 through additional fluorescence in situ hybridization (FISH) analysis of 5q13 and 5q21 to study different 5q deletion patterns and their clinical relevance in prostate cancers.
We made use of our large prostate cancer tissue microarray (TMA), consisting of more than 12 000 prostate cancers linked with a corresponding clinical database. 19,20 2 | MATERIALS AND METHODS

| Patients
Radical prostatectomy specimens were available from 12 427 patients undergoing surgery between 1992 and 2012 at the Department of Urology and the Martini Clinic at the University Medical Center, Hamburg-Eppendorf. Histopathologic data were retrieved from patient files, including tumor stage, nodal stage and status of the resection margin. In addition to the traditional Gleason categories, "quantitative" Gleason grading was performed as previously described. 21 Briefly, for every prostatectomy specimen, the percentage of Gleason 4 patterns was estimated throughout the cancerous tissue during the routine histologic evaluation. This allows the subdivision of Gleason 3 + 4 and 4 + 3 cancers according to their percentage of Gleason 4. For practical use, we subdivided the 3 + 4 and 4 + 3 cancers into eight subgroups: 3 + 4 with ≤5% Gleason 4, 3 + 4 6% to 10%, 3 + 4 11% to 20%, 3 + 4 21% to 30%, 3 + 4 31% to 49%, 4 + 3 50% to 60%, 4 + 3 61% to 80% and 4 + 3 >80% Gleason 4. Additional groups were defined by the presence of a tertiary Gleason 5 pattern, namely 3 + 4 Tert. 5 and 4 + 3 Tert. 5. Follow-up data were available for a total of 11 665 patients with a median follow-up of 36 months (range: 1-241 months; Supplementary Table 1). PSA values were measured following surgery, and PSA recurrence was defined as the time point when postoperative PSA was at least 0.2 ng/mL and increasing at subsequent measurements. All prostate specimens were diagnosed according to a standard procedure, including complete embedding of the entire prostate for histologic analysis. 22 The TMA manufacturing process was described previously in detail. 23

What's new?
Deletion of chromosome 5q is common in prostate cancer and is linked to aggressive disease. However, most studies have focused on 5q21 where CHD1, a gene required for maintenance of androgen receptor signaling, is located. Chromosome 5q13 has also been shown to harbor a small interstitial deletion, whose clinical relevance in prostate cancer remains unknown. The hereby presented gene copy number analysis of more than 12,000 prostate cancers reveals that 5q13 deletion is distinct from the well-known 5q21 deletion, associates with TMPRSS2:ERG fusion type cancers, and characterises a patient set with aggressive tumor features and poor prognosis.
Des Plaines). 4 μL of FISH probe mix in 70% formamide 2× SSC solution was applied to the slides and co-denatured with the cellular DNA in a Hybrite hybridization oven for 10 minutes at 72 C, followed by overnight hybridization at 37 C in a humidified chamber. The FISH probe mix consisted of a spectrum-orange labeled 5q13 (CDK7 locus) probe (made from BACs RP11-815B01 and RP11-443M17) and a spectrum green-labeled commercial centromere 10 probe (#06J36-090; Abbott, Wiesbaden, Germany) as a reference-a specific centromere 5 probe is not available. It is unlikely that using the centromere 10 probe caused a significant fraction of false deletion calls, because in our previous FISH studies about 80% of all tumors showed two copies of the respective chromosome. [6][7][8]10,14,[24][25][26] Analysis of 5q21 deletion was performed as previously described. 5 After hybridization, slides were subjected to serial stringent washings (2× SSC solution with 0.3% NP40 at 72 C for 2 minutes) and counterstained

| 5q copy number data sources and analysis
Raw data were obtained from four large studies employing array comparative genomic hybridization (aCGH) or single-nucleotide polymorphism (SNP) array analysis in a total of 442 prostate cancers. [13][14][15][16] Data were imported into the FISH Oracle browser 27,28 and visualized in different tracks corresponding to each study. A global threshold of −0.3 was applied to all four data sets to display deletions.

| Statistics
For statistical analysis, the JMP 14.0 software (SAS Institute Inc., NC) was used. Contingency tables were calculated to study the association between 5q deletion and clinicopathologic variables, and the chisquare (likelihood) test was used to find significant relationships.
Kaplan-Meier curves were generated for PSA recurrence free survival.
The log-rank test was applied to test the significance of differences between stratified survival functions. Cox proportional hazards regression analysis was performed to test the statistical independence and significance between pathological, molecular and clinical variables.

| Architecture of 5q deletions
A re-analysis of own Reference 14 and published References 13,15, and 16 microarray-based 5q copy number data from 442 prostate cancers using the FISH-Oracle browser 27,28 is shown in Figure 1. This analysis suggests the existence of two deletion hotspots located at 5q13 and 5q21, both of which occurred at comparable frequency (about 10% to 12% of the 442 cancers). Only about 3% of these tumors showed unequivocal large deletions including both 5q13 and 5q21.

| Patterns of 5q13 and 5q21 co-deletion
Data on both 5q loci were available for 5510 cancers, including 1137 (20.6%) cancers harboring deletions of 5q13 and/or 5q21.

| 5q-Deletion and TMPRSS2:ERG fusion
If 5q13 and 5q21 deletions were considered separately (irrespective of co-deletions), there was a link for both 5q13 and 5q21 deletions to ERG-negative cancers, which was strong for 5q21 (P < .0001 for ERG-immunohistochemistry [IHC] and ERG-FISH analysis) but only marginal for 5q13 (P = .0018 for ERG-IHC and P = .0026 for ERG-  Figure 3B). To better understand the impact of 5q21 deletion on the association between 5q13 deletion and ERG status, we performed additional subset analysis. It showed that the overall weak association between 5q13-deleted and ERG-negative cancers was driven solely by the strong link between deletions of 5q21 and absence of F I G U R E 1 Re-analysis of array-based copy number analysis data ERG fusion: deletion of 5q13 alone (ie, no 5q21 co-deletion) occurred in 49.9% of 256 ERG-negative and 50.1% of 287 ERGpositive cancers, whereas co-deletion of 5q13 and 5q21 was strongly linked to ERG-negative cancers. All data are summarized in Figure 3C.

| 5q13 and 5q21 co-deletion and PSA recurrence
The combined analysis of both deletions revealed that co-deletions of 5q13 and 5q21 had a higher risk of early biochemical recurrence than T A B L E 1 5q13 deletion and cancer phenotype deletion of 5q13 or 5q21 alone in all cancers (P ≤ .0090, Figure 4A) as well as in the subset of ERG-negative cancers (P ≤ .0090, Figure 4B).
In ERG-positive cancers, no significant additional prognostic impacts over deletions of only one locus were found (P ≥ .4, Figure 4C).

| Multivariate analyses
The prognostic relevance of 5q13 deletion, both with and without 5q21 co-deletion, was further assessed in four different multivariate ERG-positive cancers F I G U R E 2 5q deletion and PSA recurrence grade obtained in the prostatectomy specimen. It is of note that the postoperative determination of Gleason grade is "better" than the preoperatively determined Gleason grade (subject to sampling error and consequent undergrading in more than one third of cases 29 ). Finally, in Scenario 4, the preoperative Gleason grade obtained on the original biopsy was combined with preoperative PSA value, cT and 5q deletion status. In these analyses, 5q13 deletion (either alone or in combination with 5q21 deletion) predicted PSA recurrence independent of the preoperative parameters (Scenario 4, P ≤ .0020, Table 2). In addition, in ERG-negative cancers, 5q21 deletion alone or in combination with 5q13 deletion predicted PSA recurrence independent of the preoperative and postoperative parameters (Scenarios 1-4, P < .05). The results of our study identify 5q13 as a deletion hotspot that develops independent of 5q21 deletion in a relevant subset of prostate cancers.
Successful FISH analysis of more than 6800 prostate cancers in our study revealed 5q13 deletion in 13.5% and 5q21 deletion in 10.0% of tumors. These frequencies are in the range of data from earlier studies employing aCGH for the analysis of copy number changes.
These studies on 72 to 504 primary prostate cancers reported 5q13 and/or 5q21 deletions in 6% to 19% (www.cbioportal.org [11][12][13][14][15][16] ). In a previous FISH study on a subset of our prostate cancer TMA, we found 5q21 deletions in 8.7% of 2093 cancers. 5 FISH is the most precise method for deletion analysis as it allows to determine exact gene copy numbers on a single cell level and is not disturbed by contaminating non-neoplastic cells that are inevitably present in cancer tissues. Scoring criteria used for our deletion analyses in tissue sections had earlier been validated by comparison of FISH and aCGH data. 8 The choice of our FISH probes was based on the known role of CHD1 (5q21) in prostate cancer biology 30 and because of the position of CDK7 in the center of the 5q13 core deletion region, rather than because of a possible biological role of CDK7.
Our study demonstrates that both 5q13 and 5q21 deletions are strongly linked to adverse tumor phenotype and early poor patient outcome. These findings are in line with previous studies from others and us showing a link between 5q deletion and unfavorable tumor parameters 5,9 and poor patient prognosis. 5,16 In contrast, one study identified a link between 5q21 deletion and a favorable prognosis in a cohort of 55 patients 31 and one other study failed to find associations of CHD1 (5q21) deletions and expression with relevant clinicopathologic features in 86 cancers. 13 The fact that deletions of 5q13 and of 5q21 are linked to poor patient outcome is not surprising. Using our TMA resource, our group had earlier studied deletions of 3p13, 14 6q15, 6 8p21, 10 PTEN 8 , 12p13, 24 13q14, 32 16q23, 7 TP53 26 and 18q24 25 and found that all of these deletions were linked to an adverse tumor phenotype and poor prognosis.
Earlier studies analyzing 5q deletions in prostate cancer consistently focused on the 5q21 deletion hotspot harboring the CHD1 tumor suppressor gene. 5,16,18,31,33 The results of our deletion mapping and the observation that about 80% of the 5q deletions identified in our cohort included either 5q13 or 5q21 but not both loci demonstrate that these deletions constitute two separate, unrelated recurrent genomic alterations in prostate cancer. Despite minor overlap, 5q13 and 5q21 deletions include largely different sets of genes with a potential role in cancer. 5q13 and 5q21 deletions will thus induce different biologic changes in affected cells. LOX (5q23), 44 DIAPH1 (5q31) 36 and RPS14 (5q33). 45 In our multivariate analyses, deletions of 5q13 and 5q21 were sig- should not only be independent of currently established factors but also better reproducible and thus more reliable. In principle, FISH analysis is optimally suited for diagnostic testing as it provides unequivocal yes/no answers.
There are two major limitations pertaining to our study. First, we did not perform functional analysis to identify the gene(s) responsible for the observed prognostic differences. Second, it cannot be excluded that the adverse prognostic impact of large deletions is related to the tumor's overall burden of copy number changes, which might generally be higher in tumors with larger deletions. Further studies are required to clarify these issues.
In summary, our study demonstrates that chromosome 5 harbors two distinct deletion hotspots at 5q13 and 5q21, which typically occur independently from one another. Although these deletions have different biologic implications, both are associated with aggressive disease.