Clonal Relationships of Adjacent Gleason Pattern 3 and Gleason Pattern 5 Lesions in Gleason Scores 3+5=8 and 5+3=8

Abstract

Genomic studies have demonstrated a high level of intra-tumoral heterogeneity in prostate cancer. There is strong evidence suggesting that individual tumor foci can arise as genetically distinct, clonally independent lesions. However, recent studies have also demonstrated that adjacent Gleason pattern (GP) 3 and GP4 lesions can originate from the same clone but follow divergent genetic and morphologic evolution. The clonal relationship of adjacent GP3 and GP5 lesions has thus far not been investigated. Here we analyzed a cohort of 14 cases—11 biopsy and 3 radical prostatectomy specimens—with a Gleason score of 3+5=8 or 5+3=8 present in the same biopsy or in a single dominant tumor nodule at radical prostatectomy. Clonal and subclonal relationships between GP3 and GP5 lesions were assessed using genetically validated immunohistochemical assays for ERG, PTEN, and P53. 9/14 (64%) cases showed ERG reactivity in both GP3 and GP5 lesions. Only 1/14 (7%) cases showed a discordant pattern with ERG staining present only in GP3. PTEN expression was lost in 2/14 (14%) cases with perfect concordance between GP5 and GP3. P53 nuclear reactivity was present in 1/14 (7%) case in both GP5 and GP3. This study provides first evidence that the majority of adjacent GP3 and GP5 lesions share driver alterations and are clonally related. In addition, we observed a lower-than-expected rate of PTEN loss in GP5 in the context of Gleason score 3+5=8 or 5+3=8 tumors.

1. INTRODUCTION

The Gleason grading system is a valuable tool for clinical decision making [1,2]. Although the grouping in the current Grade Group (GG) system is based on robust prognostic information, some controversies exist that result from heterogeneity of morphologies combined in a given GG [3–5]. In particular, tumors with Gleason score (GS) 3+5=8 or 5+3=8 have been challenging to fit into the prognostic framework of the GG system [6–10]. While the outcomes for GS 4+4=8 (GG4) disease have been well documented, it is less certain if lesions with GS 3+5=8 or 5+3=8 should be considered GG4 or GG5. This uncertainty is due to several factors. First, 3+5=8 and 5+3=8 disease is rare with contemporary grading and accounts for only a small fraction (<1%) of cases [2,11]. Second, only a very small number of studies have investigated the outcomes of GS 3+5=8 and 5+3=8 cancers, and large-scale studies with centralized expert pathology review are missing [7,10]. Third, very little is known about the biology and clonal relationships in tumors containing both Gleason pattern (GP) 3 and GP5.

A particular challenge is the classification of GS 3+5=8 and GS 5+3=8 cancer on biopsy, since it is unclear if such lesions represent one tumor mass or if they are representative of two independent tumor foci with distinct morphologies and GG. More broadly, much of the complexity of primary prostate cancer grading is rooted in the multifocal nature of the disease [12]. Around 80% of primary prostate cancers harbor multiple distinct tumor foci, which can differ substantially in grade [13–16]. Molecular studies have revealed that in 60–70% of cases such separate tumor foci are genomically distinct [17]. These findings demonstrate that, in contrast to most other tumor types, prostate cancers are often composed of tumors arising from independent clones [18].

Over the past decade, several recurrent genomic alterations were catalogued in prostate cancer [19,20]. The most common and clinically relevant alterations include translocations involving the V-Ets Avian Erythroblastosis Virus E26 Oncogene Related gene ERG, copy-number loss affecting the tumor suppressor gene phosphatase and tensin homolog PTEN, and mutations and copy-number alterations in TP53. Importantly, these alterations are the basis for molecular subclassifications of prostate cancers and have been shown (e.g., for PTEN and TP53) to provide relevant prognostic information [21,22]. Although these genomic changes were identified using DNA sequencing approaches, validated immunohistochemical tools were developed that allow for the assessment of ERG, PTEN, and TP53 alterations in situ [23–28]. Given the tight association between these immunohistochemical markers and the underlying genomic alterations, these markers have been extensively used to delineate clonal relationships between different tumor cell populations [18].

To gain insights into the molecular features and clonal composition of GP3 and GP5 lesions, we applied genetically validated immunohistochemical assays to determine ERG, PTEN, and P53 status in GS 3+5 or 5+3 tumors. We demonstrate that, GP5 tumors show an unexpectedly low rate of PTEN alterations. Furthermore, we observed that the vast majority of adjacent GP3 and GP5 lesions had concordant ERG, PTEN, and P53 alterations. Collectively, these findings suggest that, in GS 3+5 and 5+3 tumors, GP5 and GP3 components are clonally related and GP5 lesions show a lower-than-expected frequency of alterations associated with aggressive disease.

2. MATERIALS AND METHODS

2.1. Patient cohort

This study was approved by the Institutional Research Board at Johns Hopkins. We interrogated the database of the Johns Hopkins Hospital Department of Pathology to identify 22 cases with a diagnosis of GS 3+5 or 5+3 between 2010 and 2019. Upon re-review and determination of available tissue for immunohistochemistry, we included 14 cases, 11 biopsy and 3 radical prostatectomy specimens in this study.

2.2. Immunohistochemistry

Immunostaining for ERG, PTEN, and P53 was performed on the Ventana Discovery Ultra Immunohistochemistry/In Situ Hybridization system (Roche Diagnostics) as described previously [27,29,30]. All antibodies were extensively validated using gold standard assays (FISH, NGS) [26–28]. ERG immunohistochemical analysis was performed on the Ventana Benchmark autostaining system using a rabbit monoclonal antibody (EPR 3864) after antigen retrieval in CC1 buffer followed by detection using the Optiview horseradish peroxidase system (Roche/Ventana Medical Systems, Tucson, AZ). PTEN staining was performed on a Ventana automated staining platform (Ventana Discovery Ultra; Ventana Medical Systems, Tucson, AZ) using a rabbit antihuman PTEN antibody (Clone D4.3 XP; Cell Signaling Technologies, Danvers, MA). P53 immunohistochemistry analysis was performed on the Ventana Benchmark autostaining system using a mouse monoclonal antibody (BP53-11) after antigen retrieval in CC1 buffer followed by detection using the iView system (Roche/Ventana Medical Systems, Tucson, AZ). All bright field slides were scanned using a Ventana DP200 system (Roche Diagnostics, Indianapolis, IN). Images were scored blindly by an urologic pathologist (M.C.H.). A lesion was considered to be ERG positive if any tumor nuclei showed ERG positivity. Endothelial cells served as an internal positive control. A tumor biopsy was considered to have PTEN protein loss if the intensity of cytoplasmic and nuclear staining for PTEN was markedly decreased or entirely negative across >10% of tumor cells. Adjacent benign glands and stroma provided an internal positive control. For P53, nuclear staining in >10% of tumor nuclei was considered positive [26].

2.3. Statistical analyses

Observed and expected concordances between adjacent GP3 and GP5 were calculated separately for each immunohistochemical marker (ERG, PTEN, and P53). Observed concordance was calculated as the proportion of adjacent GP3 and GP5 lesion pairs for which there was agreement of the marker status (i.e., both lesions were positive or both lesions were negative). Expected concordance was calculated assuming the probabilities of positivity or negativity in adjacent GP3 and GP5 lesions were independent and therefore equal to product of the marginal proportions. Observed concordance was evaluated relative to the expected concordance using 2-sided exact binomial tests. P<0.05 was considered statistically significant.

3. RESULTS

We identified 11 biopsy and 3 radical prostatectomy specimens with GS 3+5 or GS 5+3 that had sufficient material for immunohistochemical studies. We observed ERG positivity in 10/14 (71%) GP3 and 9/14 (64%) GP5 lesions (Figure 1, 2). Only 1/14 (7%) case showed a discordant ERG staining pattern between GP3 and GP5; in this case (55712), the GP3 was positive while the GP5 was negative (Figure 1, 2). PTEN loss was concordantly present in GP3 and GP5 in 2/14 (14%) cases. P53 nuclear staining was observed in 1/14 (7%) case and was concordant between GP3 and GP5 (Figure 1, 2).

Figure 1.

A. Representative micrographs showing ERG-negative GP5 and ERG-positive GP3 lesions in case 1. Arrowheads indicate ERG positive endothelial cells. B. Representative micrographs showing nuclear P53 staining in GP3 and GP5 components of case 11. Arrows indicate benign glands. Scale bars denote 20 μm.

Figure 2.

Overview of immunohistochemical findings for GP3 and GP5 tumors. Altered (yellow) denotes positivity for nuclear staining for ERG and P53 or loss of reactivity for PTEN. Non-altered (blue) denoted wild type staining pattern.

To contextualize these findings, we compared the observed concordance with expected concordance derived from reported frequencies in the literature. ERG alterations are present in 47% of GP3 and 54% of GP5 localized prostate cancers [28,31,32]. Assuming adjacent GP3 and GP5 lesions arose from two independent clones, the expected concordance would be 50%. In contrast, the observed concordance was 93% (95% CI 66%−100%; P<0.001). Similarly, the rate of PTEN loss was shown to be 14% in GP3 and 41% in GP5 lesions [33]. The expected concordance of PTEN in clonally independent lesions would be 56.7%, but the observed concordance was 100% (95% CI 0.77%−100%; P<0.001). Finally, the frequency of P53 alterations was shown to be 0% in GP3 and 10% in GP5 [22]; the expected concordance would be 90%, but the observed concordance was 100% (95% CI 77%−100%; P=0.4). These results indicate a higher-than-expected concordance of ERG and PTEN between adjacent GP3 and GP5 lesions suggesting that in the context of GS 3+5 or GS5+3 adjacent GP3 and GP5 lesions are clonally related.

4. DISCUSSION

Prostate cancer is known to be a multifocal and multiclonal disease. The ingenuity of the Gleason scoring system lies in the integration of this heterogeneity. Although Gleason grading has been shown to provide robust risk stratification, controversies remain in cases where opposing extremes of the scoring scale are observed to coexist (i.e., GP3 and GP5). Although GS 3+5 and GS 5+3 tumors are rare, they do occur in practice and often present a diagnostic and clinical challenge [11].

Several prior reports demonstrated differences in prostate cancer specific outcomes between GS 3+5 and GS 4+4 [6,7], suggesting that the presence of any GP5 is associated with worse prognosis [34]. However, the majority of these studies included retrospective analyses from multiple institutions without centralized pathology re-review. Complicating these analyses is the relatively poor interobserver reproducibility of GP5 diagnoses even among urologic pathologists [35]. In addition, choices of treatment modality (radical prostatectomy vs external beam radiation) might have introduced bias in these cohorts. Interestingly, more recent studies suggest that the prognosis of GS 3+5 tumors are similar to GS 4+4 [8–10].

To better understand the molecular features of GP3 and GP5 and determine their clonal relationship in this context, we studied ERG, PTEN, and P53 expression in a retrospective cohort of cases harboring GS 3+5 and GS 5+3 tumors. Unexpectedly, we observed that the frequency of PTEN and TP53 alterations observed in GP5 in this study was lower than reported previously for GP5 in the context of GP4 [22,33]. This potentially suggests that GP5 lesions when adjacent to GP3 lesions could show different molecular features. More broadly, this finding is reminiscent of a prior study from our group suggesting a context dependent difference in the frequency of PTEN alterations [36]. Given the established role of PTEN loss as an indicator for more aggressive disease, it remains to be seen if the context in which GP5 arises determines the molecular alteration pattern and biological behavior. These findings also highlight the potential importance of ancillary molecular testing in conjunction with histomorphologic grading to predict the biological behavior and prognosis.

Multifocal tumor growth is common in prostate cancer. However, individual tumors often lack well-defined invasion fronts and show infiltrative growth resulting in colliding tumor foci that form tumor masses that are challenging to distinguish [18,37]. Distinct tumor foci often show different histological grades, and more recent genomic analyses demonstrate that up to 70% of topographically distinct tumor foci show non-overlapping mutational profiles, suggesting a clonally independent origin [Arora:2004hm; 13,18,38–41].

Given the striking morphological differences between GP3 and GP5 lesions, we initially hypothesized that these GP, when co-occurring in one tumor, would most likely represent clonally distinct tumor foci. However, our analyses revealed that the majority of adjacent GP3 and GP5 showed a shared pattern of ERG and PTEN immunoreactivity [18]. Although, we could not statistically conclude that the observed concordance of P53 in our study differed from what would be expected from lesions arising from independent clones, which was at least in part due to the low frequency of this alternation, it is worth noting that the only P53 altered case showed nuclear staining in both, the GP3 and GP5 component. Nonetheless, our findings suggest a tight clonal relationship of GP3 and GP5 in the context of GS 3+5 or GS 5+3 tumors.

This finding is reminiscent of earlier clonality analyses demonstrating that at least a subset of adjacent GP3 and GP4 lesions are clonally related and share initiating driver alterations (e.g., ERG rearrangements), but also show evidence of branching subclonal evolution (exemplified by subclonal PTEN loss) [42–45]. These data provide support for the notion that lesions with distinct GP can arise from a common ancestry clone but evolve by accumulating additional molecular and histomorphological alterations. Collectively our analyses suggest that GP3 and GP5 likely arise from a common ancestral clone and, although they show divergent morphologic features, they share key driver alterations. Parenthetically, the observation that GP3 and GP5 are clonally related is also of relevance to the ongoing discussion regarding the reclassification of GS 3+3 as noncancer [46–48]. Given the potential for clonal and morphologic evolution, defining a GS 3+3 as nonmalignant is not aligned with our current molecular understanding of prostate cancer and would cause major challenges in patient care [46].

This study has several limitations. First, given the relatively rare occurrence of GS 3+5 or GS 5+3 tumors, the number of cases included in this retrospective single institution study is small. Although the low case number did not preclude the assessment of molecular associations for adjacent GP3 and GP5, we were not able to make conclusions regarding the prognosis of GS 3+5 or GS 5+3 relative to GS 4+4. Assembling a large multi-institutional cohort with long-term follow-up and centralized slide re-review should be the goal of future studies. Second, for all clonality studies, cases that harbor a particular genomic alteration are considered most informative. Despite ERG not being universally positive, we have observed that 9/10 ERG positive cases show concordant ERG staining in GP3 and GP5. Therefore, even in this subset analysis, our findings support a clonal relationship. Lastly, due to limited available tissue, we focused in in situ methods to determine molecular alterations in GP3 and GP5. In subsequent studies it would be informative to use unbiased (RNA-seq, whole genome sequencing) methods to further characterize the molecular makeup of GS 3+5 or GS 5+3 tumors.