Heterogeneity of TMPRSS2 Gene Rearrangements in Multifocal Prostate Adenocarcinoma: Molecular Evidence for an Independent Group of Diseases

Introduction

Prostate adenocarcinoma (PCA) is often a multifocal disease, generally consisting of a dominant (index) tumor and one or multiple separate smaller tumors ( 1– 4). Multifocal PCA frequently shows histologic heterogeneity among different tumor foci exhibiting different Gleason grades ( 1, 3). Arora et al. showed that of the 115 PCA prostatectomy specimens, 87% (100) contained two or more widely separated tumors and only 9% of multifocal PCAs had all tumor foci with primary and secondary Gleason grades that were the same as the corresponding overall Gleason grades assigned to the whole specimen ( 1). Due to this anatomic and histologic heterogeneity, needle biopsies are often not representative of the entire tumor.

The histologic and biological heterogeneity of multifocal PCA suggests that they arise independently within the same gland; however, the molecular basis for this hypothesis is poorly understood. Recently, our group identified gene rearrangements in a majority of PCAs, with the 5′ untranslated region of the androgen-regulated gene TMPRSS2 (21q22.3) fused to the 3′ region of the oncogenic ETS transcription factor family members ERG (21q22.2), ETV1 (7q21.2), or ETV4 (17q21; refs. 5, 6). We showed that most PCAs harboring gene rearrangements can be identified using a 5′ and 3′ break-apart fluorescence in situ hybridization (FISH) strategy, flanking the TMPRSS2 locus ( 7). Using this approach, 65% of clinically localized PCAs showed rearrangement of TMPRSS2, with the majority (55%) being fused with the ETS partner ERG ( 7). In addition, as TMPRSS2 and ERG are located ∼3 Mb apart on chromosome 21, the rearrangement between these two partners occurs through either a translocation between chromosome 21's or intrachromosomal deletion ( 8).

It is currently unknown whether different tumor foci in multifocal PCAs harbor these gene rearrangements uniformly. As this molecular aberration is specific to prostate carcinogenesis, analyzing gene fusion status of multifocal PCA should provide insight into tumor origin and the molecular basis of multifocal PCA heterogeneity. Hence, we sought to investigate TMPRSS2 gene rearrangement in multifocal PCAs.

Materials and Methods

Study population, clinical data, and prostate sample collection. The study consisted of 43 patients, who underwent radical prostatectomy with or without bilateral pelvic lymphadenectomy between 2004 and 2006 at the University of Michigan. None of the patients received preoperative radiation or androgen deprivation therapy. This study was approved by the Institutional Review Board at the University of Michigan hospital. Fresh prostates removed after surgery were weighed, measured, inked, and fixed in 10% neutral formalin. Seminal vesicles, apex, and base were amputated, and the remaining prostate was serially sectioned at 3-mm to 5-mm intervals perpendicular to the long axis of the gland from the base to apex. Tumor maps were generated by tracking each section and reconstructing them as a whole-mount section ( Figs. 1A and 2A ). A cancer was considered spatially separate or multifocal if it was 3 mm or more from the closest cancer in any single section or if it was 4 mm or more from the closest cancer on the adjacent section above or below, as described previously and as outlined in Fig. 1A ( 1, 4). The largest tumor focus was designated as the index tumor (Index T), whereas additional tumor foci were labeled as secondary tumor 1 (Sec T1, for the second tumor focus from a case) and secondary tumor 2 (Sec T2, when three different foci could be identified). Each focus was assigned a primary and secondary grade ( 9). A tissue microarray representing a total of 93 tumor foci, which included an index tumor and at least one separate secondary tumor, was constructed from these prostatectomy specimens. Three cores were taken from each representative tumor focus to construct the tissue microarray as described ( 7).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

A, schematic representation of an approach to the analysis of multifocal PCA. Diagrams were constructed by outlining each section and then reconstructing them as a whole-mount section, and tumor maps subsequently were generated. A PCA was considered spatially separate or multifocal if it was 3 mm or more from the closest cancer in any single section or if it was 4 mm or more from the nearest cancer on the adjacent section, above or below. B, assay approach to detect TMPRSS2 gene rearrangement in multifocal PCA. Interphase FISH on formalin-fixed, paraffin-embedded tissues to detect TMPRSS2 gene rearrangement in PCA. For this assay, the chromosomal location of the gene is indicated (box), with the direction of transcription indicated by the arrow. 5′ and 3′ BAC are indicated (ovals), with the number identifying the BAC as described below and the color indicating the probe color in the accompanying images. B1, two colocalized signals (yellow arrows) in a case lacking TMPRSS2 rearrangement. B2, TMPRSS2 rearrangement positive case (split positive), as indicated by one pair of split of 5′ and 3′ signals. B3, a TMPRSS2 rearrangement positive case (deletion positive); PCA case showing loss of one green-labeled probe 3′ to TMPRSS2. BACs: 1, RP11-35C4; 2, RP11-120C17.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

A, representative reconstructed map of the prostate sections in case 17. Index tumor from this case is represented as A, and two secondary tumors in the same prostate as B (secondary tumor 1) and C (secondary tumor 2); relative location of the three tumor foci. A summary of Gleason score, tumor size, and TMPRSS2 rearrangement status for each of these foci is presented in the boxes. B, comparison of TMPRSS2 rearrangement status of different tumor foci in 93 multifocal tumors from 43 PCA cases. The TMPRSS2 gene aberration status of each tumor focus is indicated individually in the form of a heat map using a color schema. B1, gene rearrangement status of cases with two tumor foci; index tumor (Index T) with one additional tumor focus called secondary tumor 1 (Sec T1). B2, the cases where an index tumor was accompanied by two additional secondary tumors, secondary tumor 1 and secondary tumor 2 (Sec T2). C, flow chart demonstrating summary of TMPRSS2 rearrangement status in 93 multifocal tumors from 43 PCA cases. N, negative for TMPRSS2 rearrangement; S, split of 5′ and 3′ ends of TMPRSS2; D, deletion of 3′ end of TMPRSS2; N/N, all tumor foci negative for TMPRSS2 rearrangement; D/D, all foci deleted; S/S, all foci split for TMPRSS2; S/N, one focus split and other remaining negative; S/D, one focus split and other deleted for TMPRSS2; D/N, one focus deleted and other focus negative for TMPRSS2 rearrangement.

TMPRSS2 gene rearrangement assessment by FISH. FISH was done as described previously ( 5– 7). Slides were examined using ImagingZ1 microscope (Carl Zeiss) and imaged with a CCD camera (Metafer). FISH signals were scored manually (100× oil immersion) in morphologically intact, nonoverlapping nuclei. A minimum of 50 cancer cells from each tumor focus was recorded. Cases without 50 evaluable cancer cells were reported as insufficient. Cores with very weak or no signals were recorded as insufficiently hybridized. All bacterial artificial chromosomes (BAC) were obtained from BACPAC Resource Center (Oakland, CA). For TMPRSS2 rearrangement detection, we used RP11-35C4 (5′ to TMPRSS2) and RP11-120C17 (3′ to TMPRSS2).

A case was considered discordant when one tumor focus showed TMPRSS2 rearrangement (split or deletion positive; Fig. 1B), whereas a second or third tumor focus did not. If a case showed different mechanisms of gene rearrangement (one focus through translocation and another through deletion), the case was considered discordant for gene rearrangement status. Conversely, a case was considered concordant when all the tumor foci in a single case showed a similar mechanism of TMPRSS2 gene rearrangement, exclusively through a translocation (split) or a deletion. Cases with all the tumor foci negative for TMPRSS2 gene rearrangement were also labeled as concordant.

Statistical analysis. The relationship between overall TMPRSS2 rearrangement, pathologic stage, and Gleason score was evaluated by the Pearson χ² test of association.

Results and Discussion

As TMPRSS2 is the most common 5′ partner of the ETS family genes ( 7), we monitored TMPRSS2 gene rearrangement as a tool to study the origin and molecular basis of multifocal PCA heterogeneity. A normal signal pattern for TMPRSS2 in 4′,6-diamidino-2-phenylindole–stained nuclei is two pairs of colocalized green and red signals ( Fig. 1B). A rearrangement was indicated by the break apart of one of the two colocalized signals (without deletion; Fig. 1B) or by one pair of colocalized signals and loss of one 3′ probe (green, with deletion; Fig. 1B).

A total of 93 tumor foci from 43 radical prostatectomy specimens were analyzed. In 36 specimens, two tumor foci and, in the remaining seven cases, three tumor foci could be evaluated for TMPRSS2 rearrangement. Overall, 30 of 43 cases (70%) showed TMPRSS2 rearrangement within at least one tumor focus, whereas the remaining 13 of 43 (30%) tumors lacked TMPRSS2 gene rearrangement in all tumor foci. We observed TMPRSS2 rearrangement in 27% of the cases through translocation (split of 5′ and 3′ TMPRSS2 signals; Fig. 1B), 63% through deletion (loss of one 3′ TMPRSS2 signal; Fig. 1B), and 10% through both mechanisms in separate tumor foci. In the 30 rearranged cases, uniform gene rearrangement mechanism was seen in 9 of 30 cases (30%). Rearrangement in the index tumor was seen in 25 of 30 cases (83%), whereas in 5 of 30 cases (17%) the rearrangement was seen only in secondary tumors. Overall, 21 of 43 cases (49%) were concordant for TMPRSS2 gene rearrangement mechanism or concordant by lack of TMPRSS2 rearrangement in all foci. Although heterogeneity for TMPRSS2 rearrangement between individual tumor foci was common, individual tumor cells within each tumor focus were homogeneous for TMPRSS2 rearrangements, suggesting that individual tumor foci develop through clonal expansion. Table 1 summarizes the TMPRSS2 gene rearrangement status and its assessment in 93 tumor foci from 43 cases with their relevant clinical and pathologic characteristics. Overall TMPRSS2 rearrangement status was not significantly associated with pathologic stage (P = 0.98), Gleason grade (P = 0.93), or tumor size (P = 0.13). Figure 2B represents a heat map of TMPRSS2 rearrangement status between different tumor foci across 43 cases in this cohort.

Although the multifocality and morphologic heterogeneity of PCA is well recognized ( 1, 3, 4), until the findings of this study it was unclear whether this represented a single disease or multiple independent diseases of distinct clonal origin. The molecular basis of multifocal PCA has been explored using comparative analyses of an allelotype, the marker of genetic linkage on specific chromosomal regions, based on multiple microsatellite loci in different areas of a given tumor ( 2, 10– 13). This procedure identifies regions or polymorphic markers likely to harbor genes altered in prostate carcinogenesis ( 2, 10, 13). However, the interpretation of patterns of such allelic imbalance between different foci of PCA remains complex and depends on the tissue purity and analysis of multiple loci, which are usually considered critical for this type of analysis. Based on this approach, some studies have provided evidence suggestive of a genetic relationship ( 13), whereas other studies do not suggest common clonality of multifocal PCAs ( 2, 10, 12).

Recently, our group identified recurrent gene fusions between the androgen-regulated gene TMPRSS2 and the oncogenic ETS family transcription factors as a crucial molecular event in the majority of PCAs ( 5, 6). As this molecular abnormality is specific to prostate carcinogenesis, in this study we monitored TMPRSS2 gene rearrangement as a tool to study the origin and molecular basis of multifocal PCA heterogeneity ( 7).

As previously shown by our group and others, this study confirms that a majority of PCAs harbor TMPRSS2-ETS family gene rearrangements ( 6– 8, 14), suggesting an important role in prostate cancer development and progression. The 70% frequency of TMPRSS2 gene rearrangement seen in these cases is similar to the 65% frequency we recently reported in a cohort of clinically localized PCAs in American men treated with radical prostatectomy ( 7). Interestingly, when different tumor foci of individual cases were compared for the TMPRSS2 gene rearrangement status, only 9 of 30 cases (30%) showed similar gene fusions across foci, demonstrating remarkable heterogeneity between multiple tumor foci. Of 21 discordant multifocal TMPRSS2 rearranged cases (70%), the mechanism of heterogeneity between different tumor foci was through either the presence or absence of TMPRSS2 rearrangement, or the presence of TMPRSS2 rearrangement through translocation in one tumor focus and deletion in another focus. The mechanism of TMPRSS2 gene rearrangement through translocation or deletion has been shown previously to be potentially significant ( 8). Perner et al. reported a significant association between tumors with TMPRSS2-ERG rearrangement through deletions and higher tumor stage and the presence of positive pelvic lymph nodes when compared with cancers without TMPRSS2-ERG fusions ( 8). This molecular heterogeneity is likely to occur in the process of tumor origin and/or progression, and therefore, our results suggest that this genetic heterogeneity may be an underlying molecular mechanism for diverse clinical and morphologic manifestations of PCAs. In multifocal PCA, the index (or the largest) tumor focus is usually considered to be biologically most relevant or significant ( 1, 4, 9). Our results indicate that an index tumor is usually (83%) representative of this molecular abnormality; however, in 17% of the cases, TMPRSS2 rearrangement was seen in only secondary tumors. Therefore, this result also suggests that, in a small proportion of cases, the secondary tumors can also have significant alterations that may be biologically significant. Therefore, the size of a PCA and its degree of histologic differentiation may not always reflect the extent of its genetic alterations. Although TMPRSS2 rearrangement status monitors all reported ETS rearrangements, other molecular mechanisms may play an important role in foci lacking TMPRSS2 rearrangement. We hypothesize that uncharacterized 5′ partners to ETS family genes or additional new 5′ and 3′ partners may be important in tumor development. Future work will focus on identifying other novel rearrangement that may have biological significance.

With the widespread use of serum prostate-specific antigen screening, over 90% of PCAs diagnosed in American men are clinically localized ( 15). These clinically localized PCAs are biologically diverse, ranging from clinically indolent tumors to a subset of aggressive tumors with potential for recurrence and metastasis ( 16, 17). Biomarkers that can accurately identify and stratify these prognostic subsets of clinically localized PCAs currently remain the field of major investigation. Interestingly, it is noted that no single biomarker or group of biomarkers that can successfully diagnose or predict disease recurrence 100% of the time has yet been identified. Although many previous studies, including recent ones, have suggested the distinct molecular subtypes of gene fusions as potential prognostic biomarkers, these studies, by definition, have ignored the heterogeneity of multifocal PCA ( 7, 8, 18, 19). Therefore, as shown earlier by our group in androgen-independent prostate cancers, the current study further emphasizes that understanding this heterogeneity is the key to the development of such future diagnostic or prognostic molecular biomarkers, specifically in small needle biopsy samples ( 20).

In summary, this report uses TMPRSS2 rearrangement status to provide compelling evidence that suggests diverse molecular origins of multifocal PCA. Our findings show that a significant proportion of multifocal PCAs is a heterogeneous group of diseases that arise from multiple, independent clonal expansions. Understanding this heterogeneity is critical to the future development of diagnostic or prognostic biomarkers.