TMPRSS2-ERG gene fusions occur in over 50% of prostate cancers, but their impact on clinical outcomes is not well understood. Retention of interstitial genes between TMPRSS2 and ERG has been reported to influence tumor progression in an animal model. In this study, we analyzed the status of TMPRSS2-ERG fusion genes and interstitial genes in tumors from a large cohort of men treated surgically for prostate cancer, associating alterations with biochemical progression. Through whole-genome mate pair sequencing, we mapped and classified rearrangements driving ETS family gene fusions in 133 cases of very low-, low-, intermediate-, and high-risk prostate cancer from radical prostatectomy specimens. TMPRSS2-ERG gene fusions were observed in 44% of cases, and over 90% of these fusions occurred in ERG exons 3 or 4. ERG fusions retaining interstitial sequences occurred more frequently in very low-risk tumors. These tumors also frequently displayed ERG gene fusions involving alternative 5′-partners to TMPRSS2, specifically SLC45A3 and NDRG1 and other ETS family genes, which retained interstitial TMPRSS2/ERG sequences. Lastly, tumors displaying TMPRSS2-ERG fusions that retained interstitial genes were less likely to be associated with biochemical recurrence (P = 0.028). Our results point to more favorable clinical outcomes in patients with ETS family fusion-positive prostate cancers, which retain potential tumor-suppressor genes in the interstitial regions between TMPRSS2 and ERG. Identifying these patients at biopsy might improve patient management, particularly with regard to active surveillance. Cancer Res; 77(22); 6157–67. ©2017 AACR.

While large advances have been made in the understanding of the biology of prostate cancer, there remain significant gaps in our understanding of prostate cancer initiation and progression, and this affects the management of prostate cancer patients. Clinically, biomarkers are needed to better stratify patients that have prostate cancer with a low risk of progression from those at higher risk. This is critically important as many patients are choosing active surveillance with presumptive low-risk disease based on limited sampling of their prostate cancer. Even though more than a decade has passed since the discovery of TMPRSS2-ERG gene fusion (1, 2), the impact of this fusion in the clinical management of patients with prostate cancer has remained limited. In the largest cohorts of prostate cancer patients tested to date involving a combined total of over 5,500 cases, ERG overexpression was not prognostic for biochemical recurrence (BCR) or disease-specific mortality following radical prostatectomy (3–6). In addition, the presence of TMPRSS-ERG fusion has not been predictive of improved response to radiotherapy, although a link between ETS fusions and altered DNA repair mechanisms has been proposed (7, 8). However, clinical associations have been predicted with ETS fusion status and response to hormonal therapies (3, 9). In several cohorts of active surveillance patients, men whose tumors were ERG fusion positive at diagnosis were more than twice as likely to require treatment compared with men with ERG-negative tumors. (10).

The ETS-family of transcription factors currently comprises 29 unique genes in humans, including ERG(21q22), ETV1(7p21), and ETV4(17q21) (11). Conserved ETS DNA-binding domains define the family members, with additional basic helix-loop-helix pointed domains, transcriptional activation and/or inactivation domain variably present, instilling specific cellular functions (12). Overexpression of ETS family proteins in prostate cancer is commonly initiated through the positioning of androgen-responsive promoters in frame with ETS-family member genes, resulting in disease-driven oncogenic functions. The androgen-responsive gene, TMPRSS2 (21q22), is principally observed fused near its first exon, placing its promoter in frame with foremost 5′-exons of ETS-family transcription factors, retaining their characteristic functional domains (1). Although less numerous, additional 5′ fusion partners include other prostate-responsive genes, SLC45A3, NDRG1, HERV-K 22q11.23, and C15orf21, as well as a strongly expressed housekeeping gene (HNRNPA2B1) with no prostate specificity or androgen-responsiveness (13, 14).

The prevalence of ETS-family fusions in prostate cancer suggests a significant role in prostate cancer development (1, 15, 16). The elucidation of the oncogenic activation of members of the ETS-family of transcription factors was envisioned to have significant clinical impact in the treatment of prostate cancer and provided rational therapeutic targets (2). However, although recurrent gene fusions involving ALK, ROS, MET, and ABL have been tremendously successful therapeutic targets in other cancers, fusions involving ETS-family of transcription factors have proven notoriously difficult to target (2, 17). ETS members are associated with the regulation of cell growth, proliferation, differentiation, and apoptosis, through activation or repression of target genes (18). However, in vivo studies recapitulating ERG or ETV1 expression in mice fell short of generating carcinoma, resulting only in the development of prostatic intraepithelial neoplasia (PIN; refs. 12, 19, 20). Thus although early acquisition of the fusion suggests a role in prostate cancer initiation, ETS fusions could potentially be primers to tumorigenesis, with other subsequent driver mutations, such as PTEN loss (21), dictating cancer progression. With no consistent association of TMPRSS2-ERG gene fusion with clinicopathologic features nor cancer outcome, the prognostic role of TMPRSS2-ERG gene fusions in prostate cancer remains limited (22).

Although significant biological differences have been observed among different ETS family fusions, the clinical impact is still emerging (3). The structures of TMPRSS2-ERG fusions have been demonstrated to impart differing biological functions in vitro, which could potentially influence clinical responses (23, 24). The mechanism of the gene fusion may additionally influence prognosis (25, 26). Two mechanisms of TMPRSS2-ERG fusion have been described that while resulting in structurally identical fusion products, proceed through either an interstitial deletion or an insertional recombination, with the key difference lying in the presence or absence of genes contained in the approximately 3 Mb interstitial region (27). A recent report by Linn and colleagues investigated the potential role of these interstitial genes using mouse models (28). Seventeen known protein encoding genes lie between TMPRSS2 and ERG, with BACE2, ETS2, HMGN1, and MX1 having potential function in tumorigenesis, and their in vitro studies in cell lines indicated a potential role for ETS2 as a tumor-suppressor gene in prostate cancer (28). In addition, loss of just a single copy of ETS2 resulted in the development of larger high-grade PIN lesions, compared with controls, and Lin and colleagues suggested that loss of these interstitial genes could result in more aggressive prostate cancer. Of note, the limitation of mouse models expressing TMPRSS2-ERG gene fusions to progress beyond PIN necessitated a PTEN-null model to drive tumor progression (28).

In order to better stratify prostate cancer patients, we need to increase our understanding of the mechanisms underlying tumor progression. The advent of high-throughput next-generation sequencing (NGS) has dramatically propelled our knowledge of somatic variation present in tumors. Here, we utilize mate pair NGS of 133 prostate cancers, ranging from very low- to high-risk disease, to characterize the presence, diversity, and structure of all ETS-family rearrangements. Specifically, we tested the hypothesis that the retention of interstitial sequence between TMPRSS2 and ERG is more commonly found in low-risk prostate cancer.

Microarray expression data

Expression analyses used an Affymetrix (U133PLUS2) microarray dataset from a prior study (29) that included expression profiles of laser capture microdissected-derived collections of prostate tumor cells from Mayo patients including 28 cases of Gleason score 6 (low-risk) and 36 cases of Gleason score 7 and higher prostate cancer (intermediate- and high-risk) as well as lymph node metastases. Normalized expression values were computed by the “gcrma” R/Bioconductor package (https://www.bioconductor.org/). The ERG status of tumors was based on the log2 intensity values of “213541_s_at” probeset above or below a threshold of 6. Associations of fusion and BCR status with gene expression levels were calculated by group t tests and corrected for multiple hypotheses testing using the “qvalue” package in R. Reported values are P values after multiple hypothesis correction (q values).

Tissue collection and processing

Consecutive cases selected from the Mayo Clinic prostate cancer–frozen repository from radical prostatectomy specimens were initially grouped according to Gleason grade group (30), with 78 cases of grade group 1 (Gleason 3+3), 15 cases of grade group 2 (Gleason 3+4), 13 cases of grade group 3 (Gleason 4+3), 8 cases of grade group 4 (Gleason 4+4), and 17 cases of grade group 5 (Gleason scores 9 and 10). Grade group 1 was additionally subgrouped according to tumor volume as very low-risk (clinically insignificant confined grade group 1 Gleason score 6 cancer with tumor volume less than 0.7 cm3; INS GS6; 53 cases) and low-risk larger volume grade group 1 Gleason score 6 (tumor volume greater than 0.7 cm3; LV GS6; 26 cases). Grade groups 2 and 3 were combined to an intermediate-risk group Gleason score 7 (GS7; 29 cases) and grade groups 4 and 5 similarly grouped together as high-risk Gleason score 8 and higher cancers (GS8+; 25 cases). Processing of radical prostatectomy specimens occurred as previously described (31). All specimens were handled in the fresh state and analyzed initially with a frozen microtome technique. The prostate was inked, serially sectioned in the horizontal plane, and right apex, left apex, and bladder base margins (perpendicular) were examined microscopically followed by standard sections from the inferior, mid, and superior posterior prostate and right and left vesicles. In cases of insignificant Gleason score 6 cancers, the entire posterior prostate was examined with additional sections of the right and left anterior prostate. Clinicopathologic data from this patient cohort have been previously described (21) and are provided in Supplementary Table S1. The study was approved by an Institutional Review Board.

Isolation of DNA and mate pair sequencing

Tumor was collected by laser capture microdissection (Arcturus instrument) from 10-μm unstained fresh-frozen sections and DNA amplified directly from cells, as previously described (32, 33). In Gleason score 7, the Gleason patterns 3 and 4 were collected and analyzed separately. Indexed libraries were prepared using the Illumina Mate-Pair (MP) Kit following the manufacturer's instructions and sequenced as two libraries per lane on the Illumina HiSeq2000 platform (34). Data were processed using previously described binary indexing mapping algorithms developed in our group (35). The read-to-reference-genome-mapping algorithm was modified to map both mate pair reads across the whole genome. Discordant mapping mate pair reads covered by at least five associates were identified for further analysis. Concordant mapping mate pair reads were used to determine frequency coverage levels across the genome (35). All raw data mate pair reads mapping to chromosome 21 for each case studied are deposited in the NCBI Sequence Read Archive (SRA) database, which can be accessed at the following link: https://trace.ncbi.nlm.nih.gov/Traces/sra/sra.cgi?study=SRP119412.

Statistical analyses

A χ2 test was used to identify associations of fusion status and Gleason score. Altogether, there were a total of 99 samples in three fusion categories including tumors with no fusion (NEG, n = 46), tumors with ERG fusion but retaining interstitial regions (RET, n = 20), and tumors with ERG fusion and deleted interstitial regions (DEL, n = 33). Gleason scores were categorized as either GS6 (including “INS GS6” and “LV GS6”) or GS7 and higher. Using “pwr.chisq.test” function in the “pwr” package, we determined there was 90.5% power to detect a medium effect size of 0.35 or higher at 0.05 significant levels. Reported P value was based on the χ2 analysis.

The three groups of samples were also analyzed for associations with BCR. The powerCT function in the “powerSurvEpi” R package indicated that given the number of samples in RET and DEL, there was a 73% power to detect a relative risk of 3.74 or higher with a type I error of 0.05. Using RET group as the baseline, the associations of DEL and NEG groups with the BRC were computed by using the coxph function in the R “survival” package. Reported results are the P values by the likelihood ratio tests.

Interstitial gene expression and BCR

The expression level of interstitial genes between TMPRSS2 and ERG with BCR, defined as PSA ≥ 0.4 post–radical prostatectomy, was evaluated in microarray data on a set of 64 radical prostatectomy specimens. Cases were initially stratified for ERG expression and the expression levels of interstitial genes plotted (Fig. 1A). Although no details of the mechanism of ERG fusion were available in this case study, a drop in expression was evident for the majority of genes in the ERG-positive group. Most significant drops in expression levels were associated with IGSF5 (q value < 10−9) and FAM3B (q value < 10−5). However, mean ETS2 levels were not decreased in the ERG-positive group, indicative of potential additional mechanistic regulatory factors influencing gene expression through promoter-specific effects. In addition, no significant association of interstitial gene expression loss with BCR was observed in the ERG-positive cases (Fig. 1B). Thus, to evaluate further associations of clinical outcome, the precise mechanism of TMPRSS2-ERG gene fusions was pursued in an independent prostate cancer cohort.

Figure 1.

Interstitial gene expressions from microarray data. Box plots represent log2 expression (y axis) of interstitial genes between ERG and TMPRSS2 (x axis) in ERG fusion–negative (gray) and positive (white) prostate cancer cases (A) and in ERG fusion–positive cases without (gray) or with (white) BCR (B). Thick black bars represent median expression levels. Genes with significant differential expression *, q < 0.05.

Figure 1.

Interstitial gene expressions from microarray data. Box plots represent log2 expression (y axis) of interstitial genes between ERG and TMPRSS2 (x axis) in ERG fusion–negative (gray) and positive (white) prostate cancer cases (A) and in ERG fusion–positive cases without (gray) or with (white) BCR (B). Thick black bars represent median expression levels. Genes with significant differential expression *, q < 0.05.

Close modal

Spectrum of TMPRSS2-ERG fusions

The mechanism of TMPRSS2-ERG gene fusion was investigated in an independent dataset of 133 prostate cancers from radical prostatectomy specimens of varying Gleason grade and tumor volumes. Cases were split into four groups: very low-risk (clinical insignificant Gleason score 6; INS GS6) 53 (40%) cases, low-risk large volume Gleason score 6 (LV GS6); 26 (20%) cases, intermediate-risk Gleason score 7 (GS7); 29 (22%) cases and high-risk Gleason score 8 and higher (GS8+); 25 (19%) cases, as described in Materials and Methods and in Supplementary Table S1. The tiling of the genome with larger fragments (∼3 kb) through mate pair sequencing (MPseq) enabled efficient genome-wide profiling of chromosomal rearrangements. The profile of all ETS family rearrangements, focusing specifically on TMPRSS2-ERG fusion and interstitial deletions between the two genes, was investigated in all cases.

As expected, TMPRSS2-ERG fusions were observed at high frequency in our population, present in 45% (60 of 133 cases) of the study cohort. In the very low-risk, low-risk, and intermediate-risk groups, fusions were observed in 43%, 59%, and 52% of cases, respectively (Fig. 2A). In the intermediate group, identical TMPRSS2-ERG fusions were observed in adjacent Gleason patterns 3 and 4 of the same tumor, which were isolated individually. The high-risk group had a much-reduced ERG fusion–positive population at 24%. This high-risk group was previously reported to have high frequency in PTEN deletion events, many in the absence of TMPRSS2-ERG fusions (21). Figure 2B illustrates the spectrum of TMPRSS2-ERG gene fusion junctions present in our study. Fusions of TMPRSS2 exon 1 or 2 to ERG exons 3 and 4 were predominant and observed similarly between the different risk groups. Ninety percent of the fusions involved exon 3 or 4 of ERG. Similarly, 85% of fusions originated from exon 1 or 2 from TMPRSS2. The high selection of fusions at the foremost-5′ exons of ERG retained the oncogenic functional domains. Similarly, fusions with the foremost-5′ exons of TMPRSS2 enable gain of promoter functions with minimal interference of TMPRSS2 structural domains. Although the variation in fusion structure is significant, the low occurrence of fusions at later exons of both TMPRSS2 and ERG potentially indicates lesser oncogenic potential, fitting the early presentation of these alterations with key functions in tumor initiation. Interestingly, the majority of fusions varying from ERG exon 3/4 fusions were from the very low-risk (insignificant) prostate cancer (5 of 6 cases).

Figure 2.

Spectrum of TMPRSS2-ERG gene fusions in prostate cancer. A, Percentage of cases in very low-risk (GS6 INS; gray), low-risk (GS6 LV; checkered), intermediate-risk (GS7; black), high-risk (GS8+; hashed), or from all prostate cancer cases (black and white) with TMPRSS2-ERG gene fusions. B, Break down of cases from each group according to exon fusion position. T# and E# refer to specific exons in TMPRSS2 and ERG genes, respectively. C, Percentage of cases of Gleason score 6 (GS6; gray) or Gleason score 7 and above (GS7+; black) that are ERG positive through deletion or no deletion of interstitial region between TMPRSS2 and ERG genes. D, Distribution of ERG-positive cases with or without interstitial deletion according to ERG exon fusion position.

Figure 2.

Spectrum of TMPRSS2-ERG gene fusions in prostate cancer. A, Percentage of cases in very low-risk (GS6 INS; gray), low-risk (GS6 LV; checkered), intermediate-risk (GS7; black), high-risk (GS8+; hashed), or from all prostate cancer cases (black and white) with TMPRSS2-ERG gene fusions. B, Break down of cases from each group according to exon fusion position. T# and E# refer to specific exons in TMPRSS2 and ERG genes, respectively. C, Percentage of cases of Gleason score 6 (GS6; gray) or Gleason score 7 and above (GS7+; black) that are ERG positive through deletion or no deletion of interstitial region between TMPRSS2 and ERG genes. D, Distribution of ERG-positive cases with or without interstitial deletion according to ERG exon fusion position.

Close modal

TMPRSS2-ERG fusion and interstitial deletion

In addition to using discordant mapping reads to define chromosomal breakpoints, MPseq utilizes concordant reads to define copy-number variations across the genome. Frequency coverage across chromosome 21 for each case is presented in Supplementary Fig. S1 and Supplementary Table S2. We utilized the copy-number data to investigate interstitial chromosomal sequence loss between TMPRSS2 and ERG on chromosome 21 in different Gleason score groupings and the presence or absence of ERG fusions (Fig. 2C). No difference in the distribution of TMPRSS2-ERG fusion with interstitial deletion was seen between Gleason score 6 cancer and Gleason score 7 and higher (P = 0.30). However, TMPRSS2-ERG–positive cases in the absence of a deletion were significantly enriched in the very low-risk and low-risk Gleason score 6 groups (P = 0.0006; Fig. 2C). The pattern of exon fusions did not appear to change significantly between the groups; however, the size of samples in different groups was too small for rigorous statistical considerations (Fig. 2D).

Mechanisms of TMPRSS2-ERG fusions

TMPRSS2-ERG fusion events proceed by two distinct mechanisms: direct fusions or complex rearrangement events. In our series, direct fusions encompassed the most common mechanism involving 36 (63%) of the ERG-positive cases. Figure 3A illustrates a direct deletion-fusion in a representative case, together with the characteristic copy loss of interstitial genes between TMPRSS2 and ERG on chromosome 21. The remaining 21 cases (37%) had complex rearrangements involving TMPRSS2-ERG potentially occurring via concurrent fusion and rearrangement. Figure 3B exemplifies a complex rearrangement with loss of the interstitial gene region for a very low-risk Gleason score 6 case in which a 50 kb region within the DSCAM gene is retained between the TMPRSS2-ERG fusion and the rest of the interstitial sequence lost. Figure 3C exemplifies a complex rearrangement where the interstitial gene region is retained for a very low-risk Gleason score 6 case in which the interstitial regions translocate to within UMODL1, an adjacent gene on chromosome 21. The complex TMPRSS2-ERG fusion cases were divided into three classes (Table 1). The first 19 cases presented with complex events in the presence of standard direct (5′)TMPRSS2-ERG(3′) fusion events. Three of these events resulted in loss of interstitial regions between TMPRSS2 and ERG, but contained links from the 3′ region of TMPRSS2 and the 5′ region of ERG at the breakpoints to other sites. The remaining 16 cases presented with complex balanced rearrangements that retained the interstitial region sequences (in one case, there was a copy-number gain in this region). Several of these cases were characterized by complex translocations involving up to three different chromosomes (Supplementary Fig. S2). A very low-risk Gleason score 6 case, and a Gleason score 7 case, involved complex rearrangements involving four and three different chromosomes, respectively, but ultimately retained the interstitial regions between TMPRSS2 and ERG (Supplementary Fig. S2). The next three cases in Table 1, PR007, PR013, and PR042, involved direct TMPRSS2-ERG fusions with additional independent nonrelated events hitting either the TMPRSS2 or ERG genes, each of these three fusions occurred via a standard interstitial deletion mechanism.

Figure 3.

TMPRSS2-ERG fusion mechanisms. A, Direct interstitial deletion TMPRSS2-ERG fusion in very low-risk insignificant Gleason score 6 case, PR057. B, Complex interstitial deletion TMPRSS2-ERG fusion in very low-risk insignificant Gleason score 6 case, PR058. C, Complex no interstitial deletion TMPRSS2-ERG fusion in very low-grade insignificant Gleason score 6 case, PR014. Normal genome structure represented above rearranged genome region for each case, with specific genome positions and exons indicated. Coverage across chromosome 21 for each case is illustrated in lower left corner of each example with the presence or absence of interstitial gene regions between TMPRSS2 and ERG. D,SLC45A3- and NDRG1-driven ERG fusions. Positions of fusion breakpoints (vertical gray arrows) with case name and ERG exon fusion position (E#) indicated in the SLC45A3/ELK4 1q32.1 (top) and NDRG1 8q24.22 (bottom) gene regions, illustrated as horizontal dark arrows, with exon positions indicated as vertical lines.

Figure 3.

TMPRSS2-ERG fusion mechanisms. A, Direct interstitial deletion TMPRSS2-ERG fusion in very low-risk insignificant Gleason score 6 case, PR057. B, Complex interstitial deletion TMPRSS2-ERG fusion in very low-risk insignificant Gleason score 6 case, PR058. C, Complex no interstitial deletion TMPRSS2-ERG fusion in very low-grade insignificant Gleason score 6 case, PR014. Normal genome structure represented above rearranged genome region for each case, with specific genome positions and exons indicated. Coverage across chromosome 21 for each case is illustrated in lower left corner of each example with the presence or absence of interstitial gene regions between TMPRSS2 and ERG. D,SLC45A3- and NDRG1-driven ERG fusions. Positions of fusion breakpoints (vertical gray arrows) with case name and ERG exon fusion position (E#) indicated in the SLC45A3/ELK4 1q32.1 (top) and NDRG1 8q24.22 (bottom) gene regions, illustrated as horizontal dark arrows, with exon positions indicated as vertical lines.

Close modal
Table 1.

TMPRSS2 and ERG gene fusion events

CaseGroupERG fusionTMPRSS2 eventsERG eventsETS expressionFusion mechanism
Complex TMPRSS2-ERG fusion events 11 GS7 T3-E3 Duplication event  ERG Retention 
 14 GS6 INS T1-E4 T1-UMODL1(21q) E4/UMODL1(21q) ERG Retention 
 20 GS6 INS T1-E4 T5-ZMIZ1(10q), C10orf11(10q)/T2 E4/ng(10q) ERG Retention 
 21 GS7 T1-E4 T1-PXDN(2p)Bal, FOXP1(3p)/T4 FOXP1Pr(3p)/E5 ERG Retention 
 28 GS7 T1-E4  E5-GRIN3A(9q)(GP4) ERG Gain 
 29 GS6 INS T1-E4 T2-E4C, DSCAM(21q)/T5  ERG Deletion 
 44 GS6 LV T1-E3 T1-ZNF589(3p)  ERG Retention 
 47 GS6 INS T2-E3 T4/ng(2p) ng(2p)/E4 ERG Retention 
 68 GS6 INS T2-E5 Complex rearrangement between TMPRSS2 and ERG ERG Retention 
 80 GS6 LV T1-E4 E3-T2 HMGN1(21q)-E4, E4/RTN1(14q) ERG Retention 
 84 GS6 INS T1-E4 C6orf106(6p)-T6,T6/HSP90AA1(14q) E4/ANPEP(4q) ERG Retention 
 87 GS6 INS T1-E1  EP/PARM1(4q) ERG Retention 
 92 GS6 INS T1-E5 C8orf37AS1(8q)-T1,T(Pr)/TTC3(21q) IGSF5pA/E5 ERG Retention 
 96 GS6 INS T2-E4 Associated internal E5/T3 rearrangement event ERG Retention 
 99 GS6 INS T3-E4 T4/ng(3q) PTPRM(18p)/E4 ERG Retention 
 108 GS8+ T1-E3 T11/CAMTA1(1p) DSCAM(21q)-E4, E3/RERE(1p) ERG Deletion 
 130 GS8+ T1-E3 T1-BACE2(21q) E3-BACE2(21q) ERG Retention 
 132 GS6 INS T1-E3 T1-BDP1(5q), T4/ng(5q)  ERG Deletion 
Nonrelated GS7 T1-E4 T7-KLK2 (19q13.33)  ERG Deletion 
 13 GS7 T2-E4  COMPLEX Inversion around ERG ERG Deletion 
 42 GS6 INS T1-E3  E4/ng(16q12.2) Bal ERG Deletion 
Complex nondirect 35 GS7 T6-E3 C Complex fusion additional events predicted ERG Retention 
 48 GS6 LV T1-E3 C PRDM15-T2, T1-ng(21q22.2) ng(21q22.2)/E4 ERG Retention 
 58 GS6 INS T1-E4 C T1-DSCAM DSCAM-E5 ERG Deletion 
 72 GS6 INS T1-E C CHODL(21q)/T1  ERG Retention 
 97 GS6 INS T1-E1 C T2-E1 E1/TRABD2B(1p33) ERG Deletion 
 74 GS6 LV T4-E4 C T4/SLC1A3(5p) E4-CAPSL(5p) nd Retention 
 95 GS6 IND T5-E3 C Complex inverted fusion no additional events detected nd Retention 
SLC45A3 events 19 GS6 INS S1-E3  SLC45A3(1q)-E3 ERG Retention 
 43 GS6 LV S1-E6 T4-PDE4D(5q) SLC45A3(1q)-E6, E5/ELK4(1q) ERG Retention 
 66 GS6 INS S1-E3  SLC45A3(1q)-E3 ERG Retention 
 70 GS6 INS S5-E4 T2-ELK4(1q),ELK4/SLC45A3-T12 SLC45A3(1q)-E4, E12/ng(9p) ERG/ELK4 Retention 
 89 GS6 INS S1-E4  SLC45A3(1q)-E4 ERG Retention 
NDRG1 events 17 GS6 INS N3-E4  NDRG1(8q)-E4 ERG Retention 
 79 GS6 LV N5-E5  NDRG1(8q)-E5 ERG Retention 
 133 GS6 INS N1-E4  NDRG1(8q)-E4, CLDN14(21q)/E4 ERG Retention 
ETV events 62 GS7 T2-V1 T2-ETV5(3q), T4-DGKG(3q)  ETV5 Retention 
 69 GS6 INS T1-V2 T1-ETV1(7p), ICA1(7p)-T1  ETV1 Retention 
 91 GS6 INS T1-V1 T1-ETV4(17q), T4/ng(9q)  ETV4 Retention 
 93 GS6 INS T2-V3 T2-ETV1(7p), T4-ANKRD10(13q)  ETV1 Retention 
CaseGroupERG fusionTMPRSS2 eventsERG eventsETS expressionFusion mechanism
Complex TMPRSS2-ERG fusion events 11 GS7 T3-E3 Duplication event  ERG Retention 
 14 GS6 INS T1-E4 T1-UMODL1(21q) E4/UMODL1(21q) ERG Retention 
 20 GS6 INS T1-E4 T5-ZMIZ1(10q), C10orf11(10q)/T2 E4/ng(10q) ERG Retention 
 21 GS7 T1-E4 T1-PXDN(2p)Bal, FOXP1(3p)/T4 FOXP1Pr(3p)/E5 ERG Retention 
 28 GS7 T1-E4  E5-GRIN3A(9q)(GP4) ERG Gain 
 29 GS6 INS T1-E4 T2-E4C, DSCAM(21q)/T5  ERG Deletion 
 44 GS6 LV T1-E3 T1-ZNF589(3p)  ERG Retention 
 47 GS6 INS T2-E3 T4/ng(2p) ng(2p)/E4 ERG Retention 
 68 GS6 INS T2-E5 Complex rearrangement between TMPRSS2 and ERG ERG Retention 
 80 GS6 LV T1-E4 E3-T2 HMGN1(21q)-E4, E4/RTN1(14q) ERG Retention 
 84 GS6 INS T1-E4 C6orf106(6p)-T6,T6/HSP90AA1(14q) E4/ANPEP(4q) ERG Retention 
 87 GS6 INS T1-E1  EP/PARM1(4q) ERG Retention 
 92 GS6 INS T1-E5 C8orf37AS1(8q)-T1,T(Pr)/TTC3(21q) IGSF5pA/E5 ERG Retention 
 96 GS6 INS T2-E4 Associated internal E5/T3 rearrangement event ERG Retention 
 99 GS6 INS T3-E4 T4/ng(3q) PTPRM(18p)/E4 ERG Retention 
 108 GS8+ T1-E3 T11/CAMTA1(1p) DSCAM(21q)-E4, E3/RERE(1p) ERG Deletion 
 130 GS8+ T1-E3 T1-BACE2(21q) E3-BACE2(21q) ERG Retention 
 132 GS6 INS T1-E3 T1-BDP1(5q), T4/ng(5q)  ERG Deletion 
Nonrelated GS7 T1-E4 T7-KLK2 (19q13.33)  ERG Deletion 
 13 GS7 T2-E4  COMPLEX Inversion around ERG ERG Deletion 
 42 GS6 INS T1-E3  E4/ng(16q12.2) Bal ERG Deletion 
Complex nondirect 35 GS7 T6-E3 C Complex fusion additional events predicted ERG Retention 
 48 GS6 LV T1-E3 C PRDM15-T2, T1-ng(21q22.2) ng(21q22.2)/E4 ERG Retention 
 58 GS6 INS T1-E4 C T1-DSCAM DSCAM-E5 ERG Deletion 
 72 GS6 INS T1-E C CHODL(21q)/T1  ERG Retention 
 97 GS6 INS T1-E1 C T2-E1 E1/TRABD2B(1p33) ERG Deletion 
 74 GS6 LV T4-E4 C T4/SLC1A3(5p) E4-CAPSL(5p) nd Retention 
 95 GS6 IND T5-E3 C Complex inverted fusion no additional events detected nd Retention 
SLC45A3 events 19 GS6 INS S1-E3  SLC45A3(1q)-E3 ERG Retention 
 43 GS6 LV S1-E6 T4-PDE4D(5q) SLC45A3(1q)-E6, E5/ELK4(1q) ERG Retention 
 66 GS6 INS S1-E3  SLC45A3(1q)-E3 ERG Retention 
 70 GS6 INS S5-E4 T2-ELK4(1q),ELK4/SLC45A3-T12 SLC45A3(1q)-E4, E12/ng(9p) ERG/ELK4 Retention 
 89 GS6 INS S1-E4  SLC45A3(1q)-E4 ERG Retention 
NDRG1 events 17 GS6 INS N3-E4  NDRG1(8q)-E4 ERG Retention 
 79 GS6 LV N5-E5  NDRG1(8q)-E5 ERG Retention 
 133 GS6 INS N1-E4  NDRG1(8q)-E4, CLDN14(21q)/E4 ERG Retention 
ETV events 62 GS7 T2-V1 T2-ETV5(3q), T4-DGKG(3q)  ETV5 Retention 
 69 GS6 INS T1-V2 T1-ETV1(7p), ICA1(7p)-T1  ETV1 Retention 
 91 GS6 INS T1-V1 T1-ETV4(17q), T4/ng(9q)  ETV4 Retention 
 93 GS6 INS T2-V3 T2-ETV1(7p), T4-ANKRD10(13q)  ETV1 Retention 

NOTE: Case groups, very low-risk Gleason score 6, GS6 INS; low-risk Gleason score 6, LV GS6; intermediate-risk Gleason score 7, GS7; and high-risk Gleason score 8 and higher, GS8+. Fusion genes: T, TMPRSS2; E, ERG; S, SLC45A3; N, NDRG1; V, ETV1, -4, or -5; C, complex event. Number following gene shorthand indicates fusing exon. Productive and nonproductive aligned fusions separated by “-” or “/,” respectively. Chromosomal cytobands are indicated in parenthesis after the gene names in the events columns.

The next seven cases listed in Table 1 involved complex nondirect (5′)TMPRSS2-ERG(3′) fusions (Table 1). For these cases, no direct paired sequencing reads were detected connecting TMPRSS2 and ERG, but combinations of rearrangements were discovered for each case bringing the two genes together for productive fusions (Fig. 3B). Two of these seven cases resulted in loss of the interstitial sequence.

The majority of the complex TMPRSS2-ERG fusion events (15 of 25, 60%) occurred in the very low-risk group. Overall, 76% (19 of 25) occurred in the very low- and low-risk Gleason score 6 cancers compared with 16% (4 of 25) in Gleason score 7 and 8% (2 of 25) in Gleason score 8 and higher.

TMPRSS2-independent ERG rearrangements

In addition to the cases with TMPRSS2-ERG fusion, eight cases contained ERG fusions with other androgen-driven genes, SLC45A3 in five cases, and NDRG1 in three cases (Table 1, bottom section). All of these cases were Gleason score 6 cancers, with six very low- and two low-risk tumors. Similar to the TMPRSS2-ERG fusions, the majority of these cases (6 of 8) fused with exon 3 or 4 of ERG, with two additional fusions at exons 5 and 6 (Fig. 3D). Six of the fusions also occurred at foremost 5′-exons of SLC45A3 or NDRG1.

Four additional cases presented with TMPRSS2 fusions to other ETS family genes, two to ETV1 and one each to ETV4 and ETV5. The ETV1 and ETV4 fusions were in very low-risk Gleason score 6 cancers and the ETV5 in a Gleason score 7 cancer (Table 1). Thus, although the number of cases is relatively small, like the complex TMPRSS2-ERG fusions, these alternative TMPRSS2 and ERG fusions were more common in the low-risk prostate cancers.

Other ETS family and androgen-driven fusions

Additional cases contained breakpoints affecting other ETS family members, occurring independently to the common TMPRSS2 and ERG fusions (Table 2). ETV1 was affected in five additional cases, whereas only two were in the correct orientation to produce a fusion product. In a very low-risk Gleason score 6 case, the previously reported androgen-driven gene HNRNPA2B1 (13, 14) was fused to the promoter of ETV1 (Table 2, PR135). In another very low-risk cancer case (Table 2, PR60), ETV1 was joined with a novel partner PCBP2 on chromosome 12. Other novel 5′ drivers were observed for ETV family members, ETV3, ETV4, ETV6, SPDEF, and FLJ1, involving EMC1, FASN, VMP1, PACSIN1, and ZZZ3, respectively. All but PACSIN1 were significantly expressed in prostate tissue from expression data (data not shown). Previously reported androgen-driven fusions HMGN2P46-ETV4 and SLC45A3-ELK4 (13, 14) were each seen in a single case. SLC45A3 was also observed fused to two previously unreported non-ETS family genes: ZBTB7B and AADACL2. Nonproductive fusions with ETS family members were also seen in a number of cases with predicted incorrect orientation for expression (Table 2). Evaluation of productive fusions with a large panel of androgen-driven/prostate-specific genes yielded just six additional events, with one of the most functionally significant involving a KLK3-TP53 fusion in a case in the Gleason score 8 and higher category, which might have affected PSA levels for that patient. Finally, as previously reported (13, 14), the overlap of additional ETV fusion events with TMPRSS2-ERG fusions was minimal.

Table 2.

Additional ETS gene family and androgen-driven gene fusions

CaseGroupProductive fusionsNonproductive fusionsOther androgen drivenPredicted expressionERG fusion status
GS7  ETV1/ETV1   nd 
GS7   KLKP1-SRPK2 SRPK2 T1-E4 
13 GS7 EMC1-ETV3Pr  ETV3 T2-E4 
14 GS6 INS  ELK3/ATF71P   T1-E4 
15 GS6 INS FASN-ETV4   ETV4 nd 
17 GS6 INS  NDRG1/TMEM132B ZBTB16-MBOAT2 MBOAT2 N3-E4 
21 GS7  NDRG1/ng   T1-E4 
23 GS7   APPBP2-CUEDC1 CUEDC1 T1-E3 
31 GS7 VMP1-ETV6 C   ETV6 T1-E4 
32 GS7 HMGN2P46-LGSN   LGSN nd 
34 GS7 SLC45A3-ZBTB7B ETV6/ng  ZBTB7B nd 
35 GS7  ETV6/YBX3 C   T6-E3 
45 GS6 LV ZZZ3-FLI1   FLJ1 T1-E3 
52 GS7  FLJ1/ng   T1-E1 
54 GS6 LV  ETV4/GOSR2   T1-E4 
60 GS6 INS PCBP2-ETV1 MAP3K12/ETV1  ETV1 nd 
63 GS7  ELF5/ng   nd 
65 GS6 INS SLC45A3-ELK4 ETV1/ng  ELK4 nd 
69 GS6 INS TMPRSS2-ETV1 ETV1/ICA1pA  ETV1 nd 
73 GS6 INS SLC45A3-AADACL2 SLC45A3/LIPH  AADACL2 nd 
77 GS6 LV   TSC22D1-CLYBL CLYBL T2-E3 
88 GS6 INS  ETV6/ABCC9   nd 
89 GS6 INS  ELK4Pr/FGGY   nd 
97 GS6 INS ETV6-CD163   ETV6 T1-E1 C 
105 GS8+  ETV1/FOXA1 KLK3-TP53 TP53 nd 
116 GS8+  GABPA/TXLNG2P   T1-E3 
118 GS8+ HMGN2P46-ETV4, PACSIN1-SPDEF   ETV4 SPDEF nd 
122 GS8+  ETV3L/CCT3 ELF2/ng   nd 
128 GS8+   DHCR24-GLIS1 GLIS1 nd 
129 GS7  ETV3/ng   nd 
133 GS6 INS SLC30A4pA-NDRG1 NDRG1/DCTN6  NDRG1 N1-E4 
135 GS6 INS HNRNPA2B1-ETV1   ETV1 nd 
CaseGroupProductive fusionsNonproductive fusionsOther androgen drivenPredicted expressionERG fusion status
GS7  ETV1/ETV1   nd 
GS7   KLKP1-SRPK2 SRPK2 T1-E4 
13 GS7 EMC1-ETV3Pr  ETV3 T2-E4 
14 GS6 INS  ELK3/ATF71P   T1-E4 
15 GS6 INS FASN-ETV4   ETV4 nd 
17 GS6 INS  NDRG1/TMEM132B ZBTB16-MBOAT2 MBOAT2 N3-E4 
21 GS7  NDRG1/ng   T1-E4 
23 GS7   APPBP2-CUEDC1 CUEDC1 T1-E3 
31 GS7 VMP1-ETV6 C   ETV6 T1-E4 
32 GS7 HMGN2P46-LGSN   LGSN nd 
34 GS7 SLC45A3-ZBTB7B ETV6/ng  ZBTB7B nd 
35 GS7  ETV6/YBX3 C   T6-E3 
45 GS6 LV ZZZ3-FLI1   FLJ1 T1-E3 
52 GS7  FLJ1/ng   T1-E1 
54 GS6 LV  ETV4/GOSR2   T1-E4 
60 GS6 INS PCBP2-ETV1 MAP3K12/ETV1  ETV1 nd 
63 GS7  ELF5/ng   nd 
65 GS6 INS SLC45A3-ELK4 ETV1/ng  ELK4 nd 
69 GS6 INS TMPRSS2-ETV1 ETV1/ICA1pA  ETV1 nd 
73 GS6 INS SLC45A3-AADACL2 SLC45A3/LIPH  AADACL2 nd 
77 GS6 LV   TSC22D1-CLYBL CLYBL T2-E3 
88 GS6 INS  ETV6/ABCC9   nd 
89 GS6 INS  ELK4Pr/FGGY   nd 
97 GS6 INS ETV6-CD163   ETV6 T1-E1 C 
105 GS8+  ETV1/FOXA1 KLK3-TP53 TP53 nd 
116 GS8+  GABPA/TXLNG2P   T1-E3 
118 GS8+ HMGN2P46-ETV4, PACSIN1-SPDEF   ETV4 SPDEF nd 
122 GS8+  ETV3L/CCT3 ELF2/ng   nd 
128 GS8+   DHCR24-GLIS1 GLIS1 nd 
129 GS7  ETV3/ng   nd 
133 GS6 INS SLC30A4pA-NDRG1 NDRG1/DCTN6  NDRG1 N1-E4 
135 GS6 INS HNRNPA2B1-ETV1   ETV1 nd 

NOTE: Case groups, very low-risk Gleason score 6, GS6 INS; low-risk Gleason score 6, LV GS6; intermediate-risk Gleason score 7, GS7; and high-risk Gleason score 8 and higher, GS8+. Productive and nonproductive aligned fusions separated by “-” or “/,” respectively. ETS family genes indicated in bold text. C, complex fusion events; nd, no fusion detected. TMPRSS2 (T) and ERG (E) gene fusions indicated with exon position.

Associations of BCR and interstitial gene deletion

ERG translocation with no loss of the interstitial sequence between TMPRSS2 and ERG, including translocation to alternative driver genes to TMPRSS2, was more frequent in the Gleason score 6 cancers, with an increased association with very low- and low-risk disease (Fig. 2C). The clinicopathologic features of the MPseq cases were available for 118 (89%; Supplementary Table S1). Of these cases, 49 represented the control ERG-negative group with no ETS fusions, 34 and 22 contained TMPRSS2-ERG fusions with interstitial region deletion or retention, respectively, with 12 cases excluded due to no clinical follow-up on BCR, the presence of alternative ETS fusions to ERG, or the presence of interstitial region gains/homozygous losses (Supplementary Table S1). There was no significant difference in the BCR rate between ERG fusion–negative and –positive cases (Fig. 4A, P = 0.15). However, in univariate analysis cancers with ERG fusion without interstitial deletion were associated with a significantly lower BCR incidence than cancers with ERG fusion and interstitial deletion (P value = 0.028; Fig. 4B) and also compared with the ERG-negative tumors (P value = 0.014). Cancers with an ERG fusion with deletion had risk profiles similar to the cancers without the fusion (Fig. 4C).

Figure 4.

ERG fusion and BCR. Survival curves of prostate cancer cases free of BCR with years from surgery (radical prostatectomy) illustrated for ERG-positive (black line) and -negative (gray line) cases (A) and with ERG-positive cases split into those with interstitial deletion (ERG+/IntDel; solid line) or without interstitial deletion (ERG+/IntRet; hatched line; B). Number of cases in each group indicated (n). C, Distribution of very low-risk insignificant (GS6 IND; solid gray shading), low-risk large volume (GS6 LV; checkered gray shading), Gleason score 7 (GS7; solid black shading), and 8+ (GS8+; hatched gray shading) cases with ERG fusion status and loss or no loss of interstitial genes.

Figure 4.

ERG fusion and BCR. Survival curves of prostate cancer cases free of BCR with years from surgery (radical prostatectomy) illustrated for ERG-positive (black line) and -negative (gray line) cases (A) and with ERG-positive cases split into those with interstitial deletion (ERG+/IntDel; solid line) or without interstitial deletion (ERG+/IntRet; hatched line; B). Number of cases in each group indicated (n). C, Distribution of very low-risk insignificant (GS6 IND; solid gray shading), low-risk large volume (GS6 LV; checkered gray shading), Gleason score 7 (GS7; solid black shading), and 8+ (GS8+; hatched gray shading) cases with ERG fusion status and loss or no loss of interstitial genes.

Close modal

The mechanism of TMPRSS2-ERG gene fusion on chromosome 21q22, through either deletion of interstitial regions or insertional chromosomal rearrangements, has been hypothesized to influence the clinical outcome of prostate cancer tumors (24, 27). PTEN-null murine models encompassing TMPRSS2-ERG fusion through interstitial loss were recently reported to develop poorly differentiated carcinomas that did not occur in mice with fusions retaining the interstitial genes, implicating ETS2 as a tumor-suppressor gene (28). Subsequent in vitro studies supported a role for ETS2 as a tumor-suppressor gene in prostate cancer cell lines. Growth inhibition and apoptosis following downregulation of ETS2 expression in human prostate cancer cells has also been reported (36). Although ERG and ETS2 are both members of the ETS family of transcription factors, ERG contains an additional activation domain at its C-terminus (12). It is therefore compelling to hypothesize that somatically induced expression of ERG, or other ETS factors such as ETV1, may interfere with the functions of ETS2, which is expressed in normal prostate tissue (28). Our analysis of interstitial gene levels in RNA expression data from 64 cases of prostate cancer failed to demonstrate any association between ERG and ETS2 expression levels (Fig. 1A). As the majority of the ERG-positive cases would be expected to have occurred through interstitial deletion, it is clear that the gene expression is a poor surrogate for allelic loss. Pathway regulation of these interstitial genes in prostate cancer would be expected to additionally affect expression levels even from a single retained allele. Significantly reduced expression of a number of interstitial genes, specifically IGSF5 and FAM3B, was present in the ERG fusion group (Fig. 1A); however, the role of interstitial genes in prostate cancer remains to be determined. As no copy number data were available for these tumors, no subcategorization according to interstitial gene loss was possible.

Mate pair sequencing afforded us the ability to precisely map all rearrangements affecting the TMPRSS2 and ERG gene loci and other ETS family genes. We used these data to analyze association of different TMPRSS2-ERG fusion mechanisms with BCR in an independent set of 133 prostate cancers, ranging from very low-risk, insignificant Gleason score 6 to high-risk Gleason score 9 prostate cancers. TMPRSS2-ERG gene fusions were observed in 44% of the cases, in line with a recent study on 1,577 prostate cancers from 8 independent cohorts, which reported 46% occurrence (range, 38%–64%; ref. 5). Eight cases (6%) presented with breakpoints in the ETV1 gene, but only half were predicted to be productive fusions. In total, 14 cases (11%) presented with predicted productive fusions on other ETS genes including ETV3 (1 case), ETV4 (3 cases), ETV5 (1 case), ETV6 (2 cases), ELK4 (2 cases), and FLJ1 (1 case) also consistent with other larger cohort studies (3–5, 37–39). The number of ERG rearrangements retaining the intervening sequences was 38% (26/68), with TMPRSS2 as the 5′-partner in 73% (19/26), also consistent with a study by Esgueva and colleagues on 540 prostate cancers (38%, with 71% TMPRSS2; ref. 37). The frequency of SLC45A3- and NDGR1-ERG gene fusions (6%) was also consistent with the study by Esgueva. However, we believe this is the first study to identify these cases selectively presenting in very low-risk cancers.

ERG and other ETS family member fusion sites had bias toward the foremost 5′-exons, as did the TMPRSS2, SLC45A3, and NRGD1 driver's fusion sites. Interestingly, the majority of fusions differing from this norm were more commonly present in very low-risk Gleason score 6 cancer. Thus, retention or loss of functional domains of ETS 3′-genes and the degree of inclusion of domains from the 5′-driver gene would be expected to affect functionality of fusion proteins. Recently, the retention of specific domains with differentially fused TMPRSS2-ERG fusion proteins, specifically SPOP binding sites within exon 3 of ERG, was demonstrated to alter the half-life of the expressed fusion proteins (24). The frequency of breakpoints occurring within 21q22.2 (ERG) and 21q22.13 (TMPRSS2) compared with the rest of chromosome 21 (Supplementary Fig. S3) supports a driver mechanism and selection for these fusions in prostate cancer. The frequent occurrence in very small volume Gleason score 6 cancers also supports the predicted early occurrence of these fusions (1, 15, 16).

Our dataset predicted a better outcome for prostate tumors with TMPRSS2-ERG fusions that retained the interstitial genes. Despite the limited number of cases, there was a statistically significant difference in a univariate analysis of time to BCR (P value = 0.028). Only four of 22 ERG-positive patients (18%) with retained sequences developed BCR compared with 35% (12 of 34) with deletion. There was no significant difference in recurrence between cancers with TMPRSS2-ERG fusion with interstitial deletion and cancers without ETS fusions (P value = 0.90). This result exemplifies the complex heterogeneity in prostate cancer, with multiple additional factors contributing to prognosis both in fusion-negative and -positive groups. Although a meta-analysis by Petterssen and colleagues (4) did not demonstrate a worse outcome for men whose prostate cancer had an ERG gene fusions occurring by deletion, they additionally acknowledged caution from this observation from limited analysis with small numbers of cases. Esgueva and colleagues similarly reported on the mechanism of ERG fusions; however, statistical analysis revealed no significant association between assessed gene rearrangements and clinical features such as Gleason grade, stage, or BCR (37). However, higher numbers of ERG rearrangements through interstitial deletion (28%) presented with BCR compared with nondeletion (17%). Data on the association of interstitial loss and BCR available from a very recent copy-number analysis dataset from 284 clinically significant prostate cancers (39) revealed slightly better outcome in ERG-expressing tumors with interstitial retention (Supplementary Fig. S4). Although these studies failed to yield significant separation of the groups, in contrast to our dataset, these studies focused on nonindolent cancers. Our study greatly benefited from our unique access to a significant number of fresh-frozen very-low volume prostate cancers and our laser capture microdissection expertise. The fact that ERG fusion is considered an early event in prostate cancer progression and that additional driver somatic mutations, such PTEN loss, are necessary to develop poorly differentiated carcinomas in ERG fusion–positive prostate cancer murine models indicates these early ERG rearranged tumors as boiling pots, primed for these additional mutations. Hence, the retention of a potential tumor-suppressor gene in the interstitial region could be considered a damper on progression, which could retain many prostate tumors in the low-risk state and thus be enriched in this population. As such, interstitial retention alone may not be a marker for clinical significant prostate cancer, but combined with the detected absence of other alterations, such as PTEN loss, could better predict prognosis.

One limitation of this study, despite being the largest study of its kind to date, was the relatively small sample size for statistical analysis. The reported association with the risk of BCR in ERG fusion–positive tumors with different interstitial deletions status was significant in a univariate but not in a multivariate analysis. This may be the result of the close association of retention of interstitial genes and the Gleason score, which was the best predictor of BCR in our study. Nevertheless, our finding is important not only because it adds to our understanding of the initial events that influence the course of prostate cancer progression, but also because of its translational potential. A number of studies from our laboratory and also from other investigators have reported that the genomic DNA rearrangements are often shared between disparate Gleason patterns in the same tumor (33, 40–42). Therefore, knowledge of the exact fusion mechanism in any Gleason pattern in a tumor can be telling about the probability of BCR. One scenario where this information would be valuable is in estimating the PSA progression risk in patients with GS6 biopsies given the estimated 30% rate of missing high-grade areas by the biopsy needle during the biopsy procedures in these patients.

In conclusion, MPseq provided in depth whole-genome assessment of rearrangements within each tumor and detailed the complexity often associated with TMPRSS2-ERG fusions. Specifically, the technique enabled precise localization of the interstitial sequences through insertion in adjacent areas of the genome. Our data supported the notion that the retention of interstitial genes is more frequently associated with very low- and low-risk prostate cancers and that determining the status of the interstitial genes may be useful in the risk stratification of patients with newly diagnosed prostate cancer.

G. Vasmatzis is CEO at WholeGenome LLC. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S.J. Murphy, F. Kosari, R.J. Karnes, G. Vasmatzis, J.C. Cheville

Development of methodology: S.J. Murphy, G. Vasmatzis, J.C. Cheville

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.J. Murphy, F. Kosari, R.J. Karnes, A. Nasir, L.J. Rangel, G. Vasmatzis, J.C. Cheville

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.J. Murphy, F. Kosari, R.J. Karnes, S.H. Johnson, A.G. Gaitatzes, J.B. Smadbeck, L.J. Rangel, G. Vasmatzis, J.C. Cheville

Writing, review, and/or revision of the manuscript: S.J. Murphy, F. Kosari, R.J. Karnes, S.H. Johnson, L.J. Rangel, G. Vasmatzis, J.C. Cheville

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R.J. Karnes, A. Nasir, S.H. Johnson, J.C. Cheville

Study supervision: J.C. Cheville

We would like to thank Bruce Eckloff and Bob Sikkink from the Mayo Genomics Sequencing Core for the Mate Pair sequencing.

This work was supported by James and Dorothy Nelson Benefactor Funds and Mayo Clinic Center for Individualized Medicine (CIM).

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.

1.
Tomlins
SA
,
Rhodes
DR
,
Perner
S
,
Dhanasekaran
SM
,
Mehra
R
,
Sun
XW
et al 
Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer
.
Science
2005
;
310
:
644
8
.
2.
Mertens
F
,
Johansson
B
,
Fioretos
T
,
Mitelman
F
. 
The emerging complexity of gene fusions in cancer
.
Nat Rev Cancer
2015
;
15
:
371
81
.
3.
Minner
S
,
Enodien
M
,
Sirma
H
,
Luebke
AM
,
Krohn
A
,
Mayer
PS
, et al
ERG status is unrelated to PSA recurrence in radically operated prostate cancer in the absence of antihormonal therapy
.
Clin Cancer Res
2011
;
17
:
5878
88
.
4.
Pettersson
A
,
Graff
RE
,
Bauer
SR
,
Pitt
MJ
,
Lis
RT
,
Stack
EC
, et al
The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis
.
Cancer Epidemiol Biomarkers Prev
2012
;
21
:
1497
509
.
5.
Tomlins
SA
,
Alshalalfa
M
,
Davicioni
E
,
Erho
N
,
Yousefi
K
,
Zhao
S
, et al
Characterization of 1577 primary prostate cancers reveals novel biological and clinicopathologic insights into molecular subtypes
.
Eur Urol
2015
;
68
:
555
67
.
6.
Hoogland
AM
,
Jenster
G
,
van Weerden
WM
,
Trapman
J
,
van der Kwast
T
,
Roobol
MJ
, et al
ERG immunohistochemistry is not predictive for PSA recurrence, local recurrence or overall survival after radical prostatectomy for prostate cancer
.
Mod Pathol
2012
;
25
:
471
9
.
7.
Haffner
MC
,
Aryee
MJ
,
Toubaji
A
,
Esopi
DM
,
Albadine
R
,
Gurel
B
, et al
Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements
.
Nat Genet
2010
;
42
:
668
75
.
8.
Dal Pra
A
,
Lalonde
E
,
Sykes
J
,
Warde
F
,
Ishkanian
A
,
Meng
A
, et al
TMPRSS2-ERG status is not prognostic following prostate cancer radiotherapy: implications for fusion status and DSB repair
.
Clin Cancer Res
2013
;
19
:
5202
9
9.
Karnes
RJ
,
Cheville
JC
,
Ida
CM
,
Sebo
TJ
,
Nair
AA
,
Tang
H
, et al
The ability of biomarkers to predict systemic progression in men with high-risk prostate cancer treated surgically is dependent on ERG status
.
Cancer Res
2010
;
70
:
8994
9002
10.
Berg
KD
,
Vainer
B
,
Thomsen
FB
,
Røder
MA
,
Gerds
TA
,
Toft
BG
, et al
ERG protein expression in diagnostic specimens is associated with increased risk of progression during active surveillance for prostate cancer
.
Eur Urol
2014
;
66
:
851
60
.
11.
Hollenhorst
PC
,
McIntosh
LP
,
Graves
BJ
. 
Genomic and biochemical insights into the specificity of ETS transcription factors
.
Annu Rev Biochem
2011
;
80
:
437
71
.
12.
Gutierrez-Hartmann
A
,
Duval
DL
,
Bradford
AP
. 
ETS transcription factors in endocrine systems
.
Trends Endocrinol Metab
2007
;
18
:
150
8
.
13.
Tomlins
SA
,
Laxman
B
,
Dhanasekaran
SM
,
Helgeson
BE
,
Cao
X
,
Morris
DS
. 
Distinct classes of chromosomal rearrangements create oncogenic ETS gene fusions in prostate cancer
.
Nature
2007
;
448
;
595
9
.
14.
Kumar-Sinha
C
,
Tomlins
SA
,
Chinnaiyan
AM
. 
Recurrent gene fusions in prostate cancer
.
Nat Rev Cancer
2008
;
8
:
497
511
.
15.
Rostad
K
,
Mannelqvist
M
,
Halvorsen
OJ
,
Oyan
AM
,
TH
,
Stordrange
L
et al 
ERG upregulation and related ETS transcription factors in prostate cancer
.
Int J Oncol
2007
;
30
:
19
32
.
16.
Clark
J
,
Attard
G
,
Jhavar
S
,
Flohr
P
,
Reid
A
,
De-Bono
J
, et al
Complex patterns of ETS gene alteration arise during cancer development in the human prostate
.
Oncogene
2008
;
27
:
1993
2003
.
17.
Darnell
JE
. 
Transcription factors as targets for cancer therapy
.
Nat Rev Cancer
2002
;
2
:
740
9
.
18.
Oikawa
T
,
Yamada
T
. 
Molecular biology of the Ets family of transcription factors
.
Gene
2003
;
303
:
11
34
.
19.
Tomlins
SA
,
Laxman
B
,
Varambally
S
,
Cao
X
,
Yu
J
,
Helgeson
BE
, et al
Role of the TMPRSS2-ERG gene fusion in prostate cancer
.
Neoplasia
2008
;
10
:
177
88
.
20.
Klezovitch
O
,
Risk
M
,
Coleman
I
,
Lucas
JM
,
Null
M
,
True
L
, et al
A causal role for ERG in neoplastic transformation of prostate epithelium
.
Proc Natl Acad Sci U S A
2008
;
105
:
2105
10
.
21.
Murphy
SJ
,
Karnes
RJ
,
Kosari
F
,
Castellar
BE
,
Kipp
BR
,
Johnson
SH
, et al
Integrated analysis of the genomic instability of PTEN in clinically insignificant and significant prostate cancer
.
Mod Pathol
2016
;
29
:
143
56
.
22.
Dubovenko
A
,
Serebryiskaya
T
,
Nikolsky
Y
,
Nikolskaya
T
,
Perlina
A
,
JeBailey
L
, et al
Reconstitution of the ERG gene expression network reveals new biomarkers and therapeutic targets in ERG positive prostate tumors
.
J Cancer
2015
;
6
:
490
501
.
23.
St John
J
,
Powell
K
,
Conley-LaComb
MK
,
Chinni
SR
. 
TMPRSS2-ERG fusion gene expression in prostate tumor cells and its clinical and biological significance in prostate cancer progression
.
J Cancer Sci Ther
2012
:
4
;
94
101
.
24.
An
J
,
Ren
S
,
Murphy
SJ
,
Dalangood
S
,
Chang
C
,
Pang
X
et al 
Truncated ERG oncoproteins from TMPRSS2-ERG fusions are resistant to SPOP-mediated proteasome degradation
.
Mol Cell
2015
;
59
:
904
1016
.
25.
Perner
S
,
Demichelis
F
,
Beroukhim
R
,
Schmidt
FH
,
Mosquera
JM
,
Setlur
S
, et al
TMPRSS2: ERG fusion-associated deletions provide insight into the heterogeneity of prostate cancer
.
Cancer Res
2006
;
66
:
8337
41
.
26.
Hermans
KG
,
van Marion
R
,
van Dekken
H
,
Jenster
G
,
van Weerden
WM
,
Trapman
J
. 
TMPRSS2:ERG fusion by translocation or interstitial deletion is highly relevant in androgen-dependent prostate cancer, but is bypassed in late-stage androgen receptor-negative prostate cancer
.
Cancer Res
2006
;
66
:
10658
63
.
27.
Teixeira
MR
. 
Chromosome mechanisms giving rise to the TMPRSS2–ERG fusion oncogene in prostate cancer and HGPIN lesions
.
Am J Surg Pathol
2008
;
32
:
642
4
.
28.
Linn
DE
,
Penney
KL
,
Bronson
RT
,
Mucci
LA
,
Li
Z
. 
Deletion of interstitial genes between TMPRSS2 and ERG promotes prostate cancer progression
.
Cancer Res
2016
;
76
:
1869
81
.
29.
Kosari
F
,
Munz
JM
,
Savci-Heijink
CD
,
Spiro
C
,
Klee
EW
,
Kube
DM
, et al
Identification of prognostic biomarkers for prostate cancer
.
Clin Cancer Res
2008
;
14
:
1734
43
.
30.
Epstein
JI
,
Zelefsky
MJ
,
Sjoberg
DD
,
Nelson
JB
,
Egevad
L
,
Magi-Galluzzi
C
, et al
A contemporary prostate cancer grading system: a validated alternative to the gleason score
.
Eur Urol
2015
;
69
:
428
35
.
31.
Sebo
TJ
,
Cheville
JC
,
Riehle
DL
,
Lohse
DL
,
Pankratz
VS
,
Myers
RP
, et al
Perineural invasion and MIB-1 positivity in addition to Gleason score are significant preoperative predictors of progression after radical retropubic prostatectomy for prostate cancer
.
Am J Surg Pathol
2002
;
26
:
7
14
.
32.
Murphy
SJ
,
Cheville
JC
,
Zarei
S
,
Johnson
SH
,
Sikkink
RA
,
Kosari
F
, et al
Mate pair sequencing of whole-genome-amplified DNA following laser capture microdissection of prostate cancer
.
DNA Res
2012
;
19
:
395
406
.
33.
Kovtun
IV
,
Cheville
JC
,
Murphy
SJ
,
Johnson
SH
,
Zarei
S
,
Kosari
F
, et al
Lineage relationship of Gleason patterns in Gleason score 7 prostate cancer
.
Cancer Res
2013
;
73
:
3275
84
.
34.
Vasmatzis
G
,
Johnson
SH
,
Knudson
RA
,
Ketterling
RP
,
Braggio
E
,
Fonseca
R
, et al
Genome-wide analysis reveals recurrent structural abnormalities of TP63 and other p53-related genes in peripheral T-cell lymphomas
.
Blood
2012
;
120
:
2280
9
.
35.
Drucker
TM
,
Johnson
SH
,
Murphy
SJ
,
Cradic
KW
,
Therneau
TM
,
Vasmatzis
G
. 
BIMA V3: an aligner customized for mate pair library sequencing
.
Bioinformatics
2014
;
30
:
1627
9
.
36.
Carbone
GM
,
Napoli
S
,
Valentini
A
,
Cavalli
F
,
Watson
DK
,
Catapano
CV
. 
Triplex DNA-mediated downregulation of Ets2 expression results in growth inhibition and apoptosis in human prostate cancer cells
.
Nucleic Acids Res
2004
;
32
:
4358
67
.
37.
Esgueva
R
,
Perner
S
,
J LaFargue
C
,
Scheble
V
,
Stephan
C
,
Lein
M
, et al
Prevalence of TMPRSS2-ERG and SLC45A3-ERG gene fusions in a large prostatectomy cohort
.
Mod Pathol
2010
;
23
:
539
46
.
38.
The Cancer Genome Atlas Research Network (TCGA)
,
Abeshouse
A
,
Ahn
J
,
Akbani
R
,
Ally
A
,
Amin
S
, et al
The molecular taxonomy of primary prostate cancer
.
Cell
2015
;
163
:
1011
25
.
39.
Fraser
M
,
Sabelnykova
VY
,
Yamaguchi
TN
,
Heisler
LE
,
Livingstone
J
,
Huang
V
, et al
Genomic hallmarks of localized, non-indolent prostate cancer
.
Nature
2017
;
541
:
359
64
.
40.
Boutros
PC
,
Fraser
M
,
Harding
NJ
,
de Borja
R
,
Trudel
D
,
Lalonde
E
, et al
Spatial genomic heterogeneity within localized, multifocal prostate cancer
.
Nat Genet
2015
;
47
:
736
45
.
41.
Cooper
CS
,
Eeles
R
,
Wedge
DC
,
Van Loo
P
,
Gundem
G
,
Alexandrov
LB
, et al
Analysis of the genetic phylogeny of multifocal prostate cancer identifies multiple independent clonal expansions in neoplastic and morphologically normal prostate tissue
.
Nat Genet
2015
;
47
:
367
72
.
42.
Gundem
G
,
Van Loo
P
,
Kremeyer
B
,
Alexandrov
LB
,
Tubio
JMC
,
Papaemmanuli
E
, et al
The evolutionary history of lethal metastatic prostate cancer
.
Nature
2015
;
520
:
353
7
.