Intrafocal heterogeneity of ERG protein expression and gene fusion pattern in prostate cancer

Upgrading of Grade Group 1 Prostate Cancer at Prostatectomy: Germline Risk Factors in a Prospective Cohort

BACKGROUND

Localized prostate tumors show significant spatial heterogeneity, with regions of high-grade disease adjacent to lower grade disease. Consequently, prostate cancer biopsies are prone to sampling bias, potentially leading to underestimation of tumor grade. To study the clinical, epidemiologic, and molecular hallmarks of this phenomenon, we conducted a prospective study of grade upgrading: differences in detected prostate cancer grade between biopsy and surgery.

METHODS

We established a prospective, multi-institutional cohort of men with grade group 1 (GG1) prostate cancer on biopsy who underwent radical prostatectomy. Upgrading was defined as detection of GG2+ in the resected tumor. Germline DNA from 192 subjects was subjected to whole-genome sequencing to quantify ancestry, pathogenic variants in DNA damage response genes, and polygenic risk.

RESULTS

Of 285 men, 67% upgraded at surgery. PSA density and percent of cancer in pre-prostatectomy positive biopsy cores were significantly associated with upgrading. No assessed genetic risk factor was predictive of upgrading, including polygenic risk scores for prostate cancer diagnosis.

CONCLUSIONS

In a cohort of patients with low-grade prostate cancer, a majority upgraded at radical prostatectomy. PSA density and percent of cancer in pre-prostatectomy positive biopsy cores portended the presence of higher-grade disease, while germline genetics was not informative in this setting. Patients with low-risk prostate cancer, but elevated PSA density or percent cancer in positive biopsy cores, may benefit from repeat biopsy, additional imaging or other approaches to complement active surveillance.

IMPACT

Further risk stratification of patients with low-risk prostate cancer may provide useful context for active surveillance decision-making.

Prostate cancer: Molecular aspects, consequences, and opportunities of the multifocal nature

Prostate cancer is unique compared to other major cancers due to the presence of multiple primary malignant foci in the majority of patients at the time of diagnosis. Each malignant focus has distinct somatic mutations and gene expression patterns, which represents a challenge for the development of prognostic tests for localized prostate cancer. Additionally, the molecular heterogeneity of advanced prostate cancer has important implications for management, particularly for patients with metastatic and locally recurrent cancer. Studies have shown that prostate cancers with mutations in DNA damage response genes are more sensitive to drugs inhibiting the poly ADP-ribose polymerase (PARP) enzyme. However, testing for such mutations should consider both spatial and temporal heterogeneity. Here, we summarize studies where multiregional genomics and transcriptomics analyses have been performed for primary prostate cancer. We further discuss the vast interfocal heterogeneity and how prognostic biomarkers and a molecular definition of the index tumor should be developed. The concept of focal treatments in prostate cancer has been evolving as a demand from patients and clinicians and is one example where there is a need for defining an index tumor. Here, biomarkers must have proven value for individual malignant foci. The potential discovery and implementation of biomarkers that are agnostic to heterogeneity are also explored as an alternative to multisample testing. Thus, deciding upon whole-organ treatment, such as radical prostatectomy, should depend on information from biomarkers which are informative for the whole organ.

High Homogeneity of Mesothelin Expression in Primary and Metastatic Ovarian Cancer

To study the extent of heterogeneity of mesothelin overexpression in primary ovarian cancers and their peritoneal and lymph node metastases, a tissue microarray (TMA) was constructed from multiple sites of 220 ovarian cancers and analyzed by immunohistochemistry. One tissue core each was taken from up to 18 different tumor blocks per cancer, resulting in a total of 2460 tissue spots from 423 tumor sites (188 primary cancers, 162 peritoneal carcinosis, and 73 lymph node metastases). Positive mesothelin expression was found in 2041 of the 2342 (87%) arrayed tissue spots and in 372 of the 392 (95%) tumor sites that were interpretable for mesothelin immunohistochemistry. Intratumoral heterogeneity was found in 23% of 168 primary cancer sites interpretable for mesothelin and decreased to 12% in 154 peritoneal carcinosis and to 6% in 71 lymph node metastases ( P <0.0001). Heterogeneity between the primary tumor and matched peritoneal carcinosis was found in 16% of 102 cancers with interpretable mesothelin results. In these cancers, the mesothelin status switched from positive in the primary tumor to negative in the peritoneal carcinosis (3 cancers) in or vice versa (2 cancers), or a mixture of positive and negative peritoneal carcinoses was found (11 cancers). No such switch was seen between the mesothelin-interpretable primary tumors and their nodal metastases of 59 cancers, and only 1 mesothelin-positive tumor had a mixture of positive and negative lymph node metastases. In conclusion, mesothelin expression is frequent and highly homogeneous in ovarian cancer.

Determining the Impact of Spatial Heterogeneity on Genomic Prognostic Biomarkers for Localized Prostate Cancer

Localized prostate tumors show remarkably diverse clinical courses, with some being cured by local therapy alone, while others rapidly relapse and have a lethal course despite precision surgery or radiotherapy. Many genomic biomarkers have been developed to predict this clinical behavior, but these are confounded by the extreme spatial heterogeneity of prostate tumors: most are multifocal and harbor multiple subclonal populations. To quantify the influence of spatial heterogeneity on genomic prognostic biomarkers, we developed a case-control high-risk cohort (n = 42) using a prospective registry, risk matched by clinicopathologic prognostic indices. Half of the cohort had early biochemical recurrence (BCR; ie, ≤18 mo), while half remained without evidence of disease for at least 48 mo after radical prostatectomy. We then genomically profiled multiple tumor foci per patient, analyzing 119 total specimens. These data allowed us to validate three published genomic prognostic biomarkers and quantify their sensitivity to tumor spatial heterogeneity. Remarkably, all three biomarkers robustly predicted early BCR, and all three were robust to spatiogenomic variability. These data suggest that DNA-based genomic biomarkers can overcome intratumoral heterogeneity: single biopsies may be sufficient to estimate the risk of early BCR after radical treatment in patients with high-risk disease. PATIENT SUMMARY: We investigated whether heterogeneity between tumor regions within the prostate affects the accuracy of DNA-based biomarkers predicting early relapse after prostatectomy. We observed persistent accuracy in predicting disease relapse, suggesting that spatial heterogeneity may not hinder biomarker performance.

The expression profile and heterogeneity analysis of ERG in 633 consecutive prostate cancers from a single center

BACKGROUND

Overexpression of ERG protein resulting from TMPRSS2:ERG rearrangement is highly specific for prostate cancer (PCa). However, the biological function of this fusion protein and its relationship with clinicopathological features still remain controversial.

METHOD

In this study, we evaluated ERG protein expression/gene rearrangement and heterogeneity by immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH) in a cohort of 633 consecutive PCa initially diagnosed by core-needle biopsy in the West China Hospital.

RESULT

Overall, ERG protein expression was detected in 16.7% (106 of 633) cases, and frequently observed in PCa patients less than 60 years of age (31.9% vs 15.5%, P = 0.004) and in PCa with Gleason score less than 8 (20.0% vs 13.4%, P = 0.027), but infrequently observed in cases with intraductal carcinoma of the prostate (IDC-P) (10.0% vs 18.6%, P = 0.012). Follow-up analysis found that patients who progressed to castration-resistant prostate cancer (CRPC) have a lower frequency of ERG protein expression at initial biopsies compared to androgen deprivation therapy (ADT)-sensitive cases (14.1% vs 23.5%, P = 0.042), but Kaplan-Meier curve showed that ERG protein expression was not an independent prognostic marker. Of all the 106 ERG-positive cases, eight cases (7.5%) exhibited heterogeneous expression of ERG protein, in which ERG was only positive in tumors with Gleason pattern 3, but negative in Gleason pattern 4. The FISH analysis was consistent with IHC in six of these cases. In the other two cases, ERG rearrangement was detected in tumors with both Gleason pattern 3 and 4 by FISH, despite the negative protein expression in Gleason pattern 4. In case 1, a repeated biopsy was performed when the disease progressed to CRPC, and no ERG-positive cells were identified neither by IHC nor FISH.

CONCLUSION

This was by far the largest series of ERG expression and heterogeneity analysis in Chinese PCa. The ERG rearrangement seemed to be frequently expressed in patients with relatively younger age and lower Gleason score and infrequently expressed in PCa with the IDC-P. PCa with positive ERG were less frequently to progress to CRPC, but there was no prognostic significance of ERG expression. In heterogeneous cases, ERG protein was detectable only in tumors with Gleason pattern 3 but not in pattern 4. Tumor cells with positive ERG expression/rearrangement seemed easily response to ADT.

Exploring the transcriptome of hormone-naive multifocal prostate cancer and matched lymph node metastases

Background

The current inability to predict whether a primary prostate cancer (PC) will progress to metastatic disease leads to overtreatment of indolent PCs as well as undertreatment of aggressive PCs. Here, we explored the transcriptional changes associated with metastatic progression of multifocal hormone-naive PC.

Methods

Using total RNA-sequencing, we analysed laser micro-dissected primary PC foci (n = 23), adjacent normal prostate tissue samples (n = 23) and lymph node metastases (n = 9) from ten hormone-naive PC patients. Genes important for PC progression were identified using differential gene expression and clustering analysis. From these, two multi-gene-based expression signatures (models) were developed, and their prognostic potential was evaluated using Cox-regression and Kaplan–Meier analyses in three independent radical prostatectomy (RP) cohorts (>650 patients).

Results

We identified several novel PC-associated transcripts deregulated during PC progression, and these transcripts were used to develop two novel gene-expression-based prognostic models. The models showed independent prognostic potential in three RP cohorts (n = 405, n = 107 and n = 91), using biochemical recurrence after RP as the primary clinical endpoint.

Conclusions

We identified several transcripts deregulated during PC progression and developed two new prognostic models for PC risk stratification, each of which showed independent prognostic value beyond routine clinicopathological factors in three independent RP cohorts.