WO2019122415A1

WO2019122415A1 - A gene signature for prognosticating gleason 7 prostate cancer

Info

Publication number: WO2019122415A1
Application number: PCT/EP2018/086766
Authority: WO
Inventors: Manuel Salto-Tellez; Chee Wee ONG; Ian Mills; David Waugh
Original assignee: The Queen's University Of Belfast
Priority date: 2017-12-22
Filing date: 2018-12-21
Publication date: 2019-06-27
Also published as: GB201721746D0

Abstract

Provided are methods, kits and biomarker panels for prognosticating Gleason 7 prostate cancer in a subject or for stratifying a subject's Gleason 7 prostate cancer into high risk and low risk categories of post-operative death, comprising use of a gene signature comprising, or consisting of, one or more genes selected from ACTA2, ACTG2, MYH11, TPM2, ROCK2, and COX4I1.

Description

A gene signature for prognosticating Gleason 7 prostate cancer

Technical Field

This invention relates to a gene signature for the stratification and prognostication of Gleason 7 prostate cancer.

Background Art

Prostate cancer remains the second most common cancer-related death for men in developed nations. In the United Kingdom, prostate cancer has become the most common form of cancer in males, surpassing incidence rates of lung and bowel cancers since 1998. These facts could be attributed to the overall aging population of the region and the prevalent use of prostate-specific antigen as a screening biomarker. Despite advances in the understanding of prostate cancer biology, the current standard of care is still dependent on the Gleason grading of the patient’s tumour. To prevent overtreatment, there has been an emerging routine of assigning non-treatment or active surveillance approach for patients whose primary prostate biopsies were graded a lower than Gleason 7 score. Such practise poses a dilemma whereby the intent of avoiding treatment could be counteracted by the opportunity for early treatment intervention of aggressive tumours. Thus, molecular characterizations by integrated genomic analysis could be useful in further stratifying the low Gleason scoring tumours that could not be distinguished by current practices.

The molecular characterizations of cancer subtypes have been more successful in other cancer types such as breast and colorectal cancers. In the past decade, several major studies have been carried out using next-generation high-throughput strategies to characterize the molecular landscape of prostate cancers. Together, these studies have shown that the most common genomic events in prostate cancers involved androgen receptor amplification or mutation, the loss of PTEN and the high frequency of gene arrangements surrounding the oncogenic transcription factor ERG.

Particularly, the loss of the tumour suppressor PTEN was widely associated with advanced staging of the disease and poor prognosis. Although most of these studies have allowed researchers to gain a better understanding of the major genomic alterations leading to prostate cancers, they were unable to provide further insights relating to clinical decision surrounding low-risk prostate cancers. Therefore, we postulated that an in-depth molecular characterization of low-risk prostate cancers may aid in the further stratification of patients who might benefit from early treatment intervention.

Thus, as can be seen, Gleason 7 tumours represent a particular conundrum with respect to treatment as there is currently no consensus or molecular test that has the power to differentiate the indolent Gleason 7 tumours from those life-threatening tumours. A molecular test that informs a clinician whether a Gleason 7 tumour is indolent or has characteristics that are life-threatening will greatly improve clinical decision-making with respect to sparing patients from non-essential treatment or alternatively, introducing early treatment intervention of aggressive tumours. Currently, because of this prognostic uncertainty, many patients with Gleason 7 prostate cancer are overtreated. In this study, we examined a group of untreated Gleason 7 tumours through immunohistochemistry, whole genome gene expression and methylation analysis. This integrated analytical approach enabled us to identify subtypes of patients associated with PTEN status and a gene expression signature that predicts survival outcome. To test the significance of the signature, we then further validated the effect of the gene expression signature in three independently published prostate cancer datasets. In addition, through in silico analysis of a time-based gene expression database of genes regulated by androgen receptor (AR) signalling, our gene signature enables us to describe the underlying biological processes that linked prostate cancer progression with PTEN/AKT and AR signalling pathway.

Therefore, the present invention has been developed with a view to providing an improved stratification and prognostication of Gleason 7 prostate cancers.

Summary of the Invention

Accordingly, the invention provides a method of prognosticating Gleason 7 prostate cancer in a subject, the method comprising:

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of one or more genes selected from a panel of genes comprising, or consisting of: ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1 ,

(ii) predicting a risk of post-operative death for the subject based on the expression level of said one or more genes.

The present inventors have surprisingly found that the hazard risk ratio associated with expression levels of ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1 , relative to a corresponding reference gene expression level, allowed the prediction of survival outcome for Gleason 7 prostate cancer patients. Gleason 7 prostate cancer discussed herein can correspond to prostate cancer determined to have a Gleason score of 3 + 4. As is understood in the field, prostate cancer having a Gleason score of 3 + 4 is particularly difficult to prognosticate, as is stratification of patients into high and low-risk categories and assignment of a suitable treatment option. As will be noted from Table 5 below, the above genes were among 14 genes having a p value of <0.05 which indicated a significant differential expression of the 14 genes between patient groups with good and poor outcomes. The gene ranking was based on the statistical differences in terms of gene expression between the“low” risk and“high” risk groups. Further, as noted from Table 6 below, ACTA2,

ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I 1 showed a hazard risk ratio of 2.33 to 4.96 and a p- value of < 0.01 indicating that each of these genes are individually predictive of survival outcomes in Gleason 7 prostate cancer patients. Optionally, the risk of post-operative death is the risk of post-operative death within 5 years. Further optionally, the risk of post-operative death within 5 years is determined to be high or low, wherein high risk is a risk that is > 2-fold greater, optionally > 3-fold greater, optionally > 3.54-fold greater, optionally > 4-fold greater, optionally > 5-fold greater, than low risk.

Optionally, an increased expression relative to a corresponding reference gene expression level for ACTA2 and/or TPM2 corresponds to a high risk of post-operative death, optionally post-operative death within 5 years. Optionally, an unchanged or decreased expression relative to a corresponding reference gene expression level for ACTA2 and/or TPM2 corresponds to a low risk of post-operative death, optionally post-operative death within 5 years.

Optionally, a decreased expression relative to a corresponding reference gene expression level for COX4I1 , MYH11 , ACTG2, and/or ROCK2 corresponds to a high risk of post-operative death, optionally post-operative death within 5 years. Optionally, an unchanged or increased expression relative to a corresponding reference gene expression level for COX4I1 , MYH11 , ACTG2, and/or ROCK2 corresponds to a low risk of post-operative death, optionally post-operative death within 5 years.

Optionally, the Gleason 7 prostate cancer in the subject described herein has normal or increased expression levels of phosphatase and tensin homolog (PTEN; accession number NM_000314). Optionally, the Gleason 7 prostate cancer in the subject has normal or increased expression of PTEN and low gene methylation levels. In other words, the Gleason 7 prostate cancer in the subject is designated as having high PTEN expression (PTEN-H), i.e. having normal or increased expression of the PTEN gene and low gene methylation levels. Optionally, Gleason 7 prostate cancer in a subject may be designated as having moderate PTEN expression (PTEN-M), having high PTEN gene expression and high gene methylation levels, i.e. having normal or increased expression of the PTEN gene and high gene methylation levels. Optionally, Gleason 7 prostate cancer in a subject may be designated as having low PTEN expression (PTEN-L) having low PTEN gene expression and high gene methylation levels, i.e. having decreased expression of the PTEN gene and high gene methylation levels. Optionally, Gleason 7 prostate cancer in a subject may be designated as having high PTEN gene methylation levels corresponding to methylation value of greater or equal to 0.5 when assessed using the lllumina® Infinium HD assay. Optionally, Gleason 7 prostate cancer in a subject may be designated as having low PTEN gene methylation levels corresponding to methylation value of lower than 0.5 when assessed using the lllumina® Infinium HD assay. In a clinical diagnostic setting, PTEN methylation can also by assessed by, for example, digestion-based assay followed by real-time PCR. Additionally or alternatively, PTEN methylation can be detected by pyrosequencing after bisulphite conversion.

Optionally, the expression level of PTEN is designated to be increased, decreased or substantially the same (i.e.“normal”) relative to a corresponding PTEN reference gene expression level. It will be understood that the PTEN reference gene expression level is the gene expression level of PTEN in non-tumour regions of a subject’s prostate. In other words, the reference gene expression level is the expression level of PTEN in normal, non-tumour prostate tissue. Optionally, the gene expression level may be determined by immunohistochemistry. Such immunohistochemistry techniques are known in the art and may employ an anti-PTEN antibody, such as described in the examples herein. Optionally, or alternatively, the gene expression level may be determined, for example, by macro- dissecting non-tumour regions of the subject’s prostate, using the tissue to generate pooled“normallike” epithelium from the tissue specimens and then determining expression level of the gene(s) using any of various techniques well known in the art such as real-time PCR, quantitative real-time PCR, nucleic acid microarrays, and lllumina®’s Human Whole-Genome DASL HT Assay, an array- based method for expression profiling of partially degraded RNA samples. Thus, up-regulation or down-regulation of PTEN expression relative to the corresponding reference gene expression level means the increased or decreased expression, respectively, of PTEN relative to the level of expression of PTEN in non-tumour regions of the subject’s prostate.

Optionally, the invention provides a method of prognosticating Gleason 7 prostate cancer in a subject, the method comprising:

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of at least one gene, optionally at least two genes, optionally at least three genes, optionally at least four genes, optionally at least five genes, optionally at least six genes, selected from a panel of genes comprising, or consisting of: ACTA2, ACTG2, MYH11 , TPM2, ROCK2, and COX4I1 ,

(ii) predicting a risk of post-operative death for the subject based on the expression level of said genes.

Optionally, the invention provides a method of prognosticating Gleason 7 prostate cancer in a subject, the method comprising: (i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of at least one gene, optionally at least two genes, optionally at least three genes, optionally at least four genes, selected from a panel of genes comprising, or consisting of: ACTA2, ACTG2, MYH11 , and TPM2,

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of a panel of genes comprising, or consisting of: ACTA2, ACTG2, MYH11 , and TPM2,

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of a panel of genes comprising, or consisting of: ACTA2, ACTG2, MYH11 , TPM2, ROCK2, and COX4I1 ,

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of two or more genes selected from a panel of genes comprising: at least one of ACTA2, ACTG2, MYH11 , TPM2, ROCK2, and COX4I1 , and at least one of TPT1 , TRPM4, UBA52 RNY1 , SUMF2, CLUAP1 , LILRB3, RNU1-3,

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of two or more genes selected from a panel of genes comprising: at least two, optionally at least three, optionally at least four, optionally at least five, optionally six, of ACTA2, ACTG2, MYH11 , TPM2, ROCK2, and COX4I1 , and at least one of TPT1 , TRPM4, UBA52, RNY1 , SUMF2, CLUAP1 , LILRB3, and RNU1-3,

(ii) predicting a risk of post-operative death for the subject based on the expression level of said genes. Optionally, the invention provides a method of prognosticating Gleason 7 prostate cancer in a subject, the method comprising:

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of two or more genes selected from a panel of genes comprising: at least one of ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I 1 , and at least two, optionally at least three, optionally at least four, optionally at least five, optionally at least six, optionally at least seven, optionally eight, of TPT1 , TRPM4, UBA52, RNY1 , SUMF2, CLUAP1 , LILRB3, and RNU1-3,

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of two or more genes selected from a panel of genes comprising: at least one of ACTA2, ACTG2, MYH 1 1 , TPM2, ROCK2, and COX4I 1 , and at least one of TPT1 , TRPM4, UBA52 RNY1 , SUMF2, CLUAP1 , LILRB3, RNU1-3, CDC37, SLC44A4, DOPEY2, YIF1A, RPL9, RPS10, SNORA61 , KLK3, PHGDH, SERPINB6, LRRC26, AKT1 , CCR6, HGS, FASN, RPS19, ATP9A,

CLTB, TOMM7, RPS29, and VIPR1 ,

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of two or more genes selected from a panel of genes comprising: at least two, optionally at least three, optionally at least four, optionally at least five, optionally six, of ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I 1 , and at least one of TPT1 , TRPM4, UBA52, RNY1 , SUMF2, CLUAP1 , LILRB3, RNU1-3, CDC37, SLC44A4, DOPEY2, YIF1A, RPL9, RPS10,

SNORA61 , KLK3, PHGDH, SERPINB6, LRRC26, AKT1 , CCR6, HGS, FASN, RPS19, ATP9A,

CLTB, TOMM7, RPS29, and VIPR1 ,

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of two or more genes selected from a panel of genes comprising: at least one of ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1 , and at least 2, optionally at least 3, optionally at least 4, optionally at least 5, optionally at least 6, optionally at least 7, optionally at least 8, optionally at least 9, optionally at least 10, optionally at least 1 1 , optionally at least 12, optionally at least 13, optionally at least 14, optionally at least 15, optionally at least 16, optionally at least 17, optionally at least 18, optionally at least 19, optionally at least 20, optionally at least 21 , optionally at least 22, optionally at least 23, optionally at least 24, optionally at least 25, optionally at least 26, optionally at least 27, optionally at least 28, optionally 29, of TPT1 , TRPM4, UBA52, RNY1 , SUMF2, CLUAP1 , LILRB3, RNU1-3, CDC37, SLC44A4, DOPEY2, YIF1A, RPL9, RPS10, SNORA61 , KLK3, PHGDH, SERPINB6, LRRC26, AKT1 , CCR6, HGS, FASN, RPS19, ATP9A, CLTB, TOMM7, RPS29, and VIPR1 ,

Optionally, an increased expression relative to a corresponding reference gene expression level for ACTA2, TPM2, UBA52, LILRB3, TPT1 , CDC37, YIF1A, RPS10, SNORA61 , KLK3, PHGDH, LRRC26, CCR6, FASN, ATP9A, CLTB, TOMM7, and/or VIPR1 corresponds to a high risk of postoperative death. Optionally, an unchanged or decreased expression relative to a corresponding reference gene expression level for ACTA2, TPM2, UBA52, LILRB3, TPT1 , CDC37, YIF1A, RPS10, SNORA61 , KLK3, PHGDH, LRRC26, CCR6, FASN, ATP9A, CLTB, TOMM7, and/or VIPR1 corresponds to a low risk of post-operative death.

Optionally, a decreased expression relative to a corresponding reference gene expression level for COX4I1 , MYH1 1 , ACTG2, RNY1 , TRPM4, ROCK2, RNU1-3, CLUAP1 , SUMP2, SLC44A4,

DOPEY2, RPL9, SERPINB6, AKT1 , HGS, RPS19, and/or RPS29 corresponds to a high risk of postoperative death, optionally post-operative death within 5 years. Optionally, an unchanged or increased expression relative to a corresponding reference gene expression level for COX4I1 ,

MYH1 1 , COX4I1 , MYH1 1 , ACTG2, RNY1 , TRPM4, ROCK2, RNU1-3, CLUAP1 , SUMP2, SLC44A4, DOPEY2, RPL9, SERPINB6, AKT1 , HGS, RPS19, and/or RPS29 corresponds to a low risk of postoperative death, optionally post-operative death within 5 years.

In another aspect, the invention provides a method of prognosticating Gleason 7 prostate cancer in a subject, the method comprising:

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of three or more genes selected from a panel of genes comprising: ACTA2, TPM2, RNY1 , COX4I 1 , LRRC26, TOMM7, RPS10, UBA52, YIF1A, ACTG2, ATP9A, CLUAP1 , VIPR1 , CDC37, MYH1 1 , SUMF2, TRPM4, PHGDH, RPS29, SERPINB6, CLTB, SLC44A4, RNU1-3,

LILRB3, HGS, DOPEY2, KLK3, TPT 1 , AKT 1 , FASN, RPS19, SNORA61 , ROCK2, CCR6, and RPL9,

(ii) predicting a risk of post-operative death for the subject based on the expression level of said three or more genes.

“Prognositicating”, as used herein, means that an outcome for a subject’s Gleason 7 is predicted based on the determined expression of one or more genes selected from a panel of genes comprising: ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1 , and/or the determined expression of three or more genes selected from a panel of genes comprising: ACTA2, TPM2, RNY1 , COX4I 1 , LRRC26, TOMM7, RPS10, UBA52, YIF1A, ACTG2, ATP9A, CLUAP1 , VIPR1 , CDC37, MYH1 1 , SUMF2, TRPM4, PHGDH, RPS29, SERPINB6, CLTB, SLC44A4, RNU1-3,

LILRB3, HGS, DOPEY2, KLK3, TPT 1 , AKT 1 , FASN, RPS19, SNORA61 , ROCK2, CCR6, and RPL9, which prediction can determine the risk of post-operative death, such as post-operative death within 5 years of a prostatectomy. The risk may be high or low, that is, a high risk of post-operative death within 5 years relative to a low risk of post-operative death within 5 years. By“high risk”, it is meant that the risk of post-operative death within 5 years is > 2-fold greater, optionally > 3-fold greater, optionally > 3.54-fold greater, optionally > 4-fold greater, optionally > 5-fold greater, than“low risk”. Conversely, by“low risk”, it is meant that the risk of post-operative death within 5 years is > 2-fold lower, optionally > 3-fold lower, optionally > 3.54-fold lower, optionally > 4-fold lower, optionally > 5- fold lower, than“high risk”.

Optionally, the expression level of at least 3, optionally at least 4, optionally at least 5, optionally at least 6, optionally at least 7, optionally at least 8, optionally at least 9, optionally at least 10, optionally at least 1 1 , optionally at least 12, optionally at least 13, optionally at least 14, optionally at least 15, optionally at least 16, optionally at least 17, optionally at least 18, optionally at least 19, optionally at least 20, optionally at least 21 , optionally at least 22, optionally at least 23, optionally at least 24, optionally at least 25, optionally at least 26, optionally at least 27, optionally at least 28, optionally at least 29, optionally at least 30, optionally at least 31 , optionally at least 32, optionally at least 33, optionally at least 34, or optionally at least all 35, of said genes is determined.

Optionally, the expression level of 3, optionally 4, optionally 5, optionally 6, optionally 7, optionally 8, optionally 9, optionally 10, optionally 1 1 , optionally 12, optionally 13, optionally 14, optionally 15, optionally 16, optionally 17, optionally 18, optionally 19, optionally 20, optionally 21 , optionally 22, optionally 23, optionally 24, optionally 25, optionally 26, optionally 27, optionally 28, optionally 29, optionally 30, optionally 31 , optionally 32, optionally 33, optionally 34, or optionally 35, of said genes is determined.

Optionally, the gene panel comprises ACTA2, TPM2, RNY1 , COX4I1 , LRRC26, TOMM7, RPS10, UBA52, YIF1A, ACTG2, ATP9A, CLUAP1 , VIPR1 , CDC37, MYH1 1 , SUMF2, TRPM4, PHGDH, RPS29, SERPINB6, CLTB, SLC44A4, RNU1-3, LILRB3, HGS, DOPEY2, KLK3, TPT 1 , AKT 1 , FASN, RPS19, SNORA61 , ROCK2, CCR6, and RPL9. Optionally, the gene panel consists of ACTA2, TPM2, RNY1 , COX4I 1 , LRRC26, TOMM7, RPS10, UBA52, YIF1A, ACTG2, ATP9A, CLUAP1 , VIPR1 , CDC37, MYH1 1 , SUMF2, TRPM4, PHGDH, RPS29, SERPINB6, CLTB, SLC44A4, RNU1-3, LILRB3, HGS, DOPEY2, KLK3, TPT 1 , AKT 1 , FASN, RPS19, SNORA61 , ROCK2, CCR6, and RPL9

Optionally, the gene panel comprises COX4I1 , MYH1 1 , ACTG2, ACTA2, TPM2, RNY1 , TRPM4, UBA52, ROCK2, LILRB3, RNU1-3, TPT1 , CLUAP1 , and SUMF2. Optionally, the gene panel consists of COX4I 1 , MYH1 1 , ACTG2, ACTA2, TPM2, RNY1 , TRPM4, UBA52, ROCK2, LILRB3, RNU1-3, TPT1 , CLUAP1 , and SUMF2. Optionally, the three or more genes are selected from COX4I1 , MYH1 1 , ACTG2, ACTA2, TPM2, RNY1 , TRPM4, UBA52, ROCK2, LILRB3, RNU1-3, TPT1 , CLUAP1 , and SUMF2. Optionally, the three or more genes comprise ACTA2, optionally ACTA2 and ACTG2, optionally ACTA2, ACTG2 and MYH 1 1 , further optionally ACTA2, ACTG2, MYH 1 1 and TPM2. Optionally, the three or more genes consist of ACTA2, ACTG2 and MYH1 1 , optionally ACTA2, ACTG2, MYH1 1 and TPM2.

Optionally, the expression level of each of said genes is based on the up-regulation or down- regulation of gene expression relative to the corresponding reference gene expression level. It will be understood that the reference gene expression level is the gene expression level of the gene in non-tumour regions of a subject’s prostate. In other words, the reference gene expression level is the expression level of the gene in normal, non-tumour prostate tissue. The gene expression level may be determined, for example, by macro-dissecting non-tumour regions of the subject’s prostate, using the tissue to generate pooled“normal-like” epithelium from the tissue specimens and then determining expression level of the gene(s) using any of various techniques well known in the art such as real-time PCR, quantitative real-time PCR, nucleic acid microarrays, and lllumina®’s Human Whole-Genome DASL HT Assay, an array-based method for expression profiling of partially degraded RNA samples. Thus, up-regulation or down-regulation of gene expression relative to the corresponding reference gene expression level means the increased or decreased expression, respectively, of a gene from the gene panel relative to the level of expression of said gene in nontumour regions of the subject’s prostate. Thus, the expression level of a gene described herein may be understood to be the relative expression level of said gene, that is, the relative expression level of the gene compared to the expression level of the corresponding reference gene.

Optionally, an increased relative expression of ACTA2, TPM2, UBA52, LILRB3, TPT1 , CDC37, YIF1A, RPS10, SNORA61 , KLK3, PHGDH, LRRC26, CCR6, FASN, ATP9A, CLTB, TOMM7, and VIPR1 , corresponds to a high risk of post-operative death, optionally post-operative death within 5 years.

Optionally, an unchanged or decreased relative expression of ACTA2, TPM2, UBA52, LILRB3,

TPT1 , CDC37, YIF1A, RPS10, SNORA61 , KLK3, PHGDH, LRRC26, CCR6, FASN, ATP9A, CLTB, TOMM7, and VIPR1 , corresponds to a low risk of post-operative death, optionally post-operative death within 5 years.

Optionally, a decreased relative expression of one or more genes selected from COX4I1 , MYH1 1 , ACTG2, RNY1 , TRPM4, ROCK2, RNU1-3, CLUAP1 , SUMP2, SLC44A4, DOPEY2, RPL9,

SERPINB6, AKT1 , HGS, RPS19, and RPS29, corresponds to a high risk of post-operative death, optionally post-operative death within 5 years.

Optionally, an unchanged or increased relative expression of one or more genes selected from COX4I1 , MYH1 1 , ACTG2, RNY1 , TRPM4, ROCK2, RNU1-3, CLUAP1 , SUMP2, SLC44A4,

DOPEY2, RPL9, SERPINB6, AKT1 , HGS, RPS19, and RPS29, corresponds to a low risk of postoperative death, optionally post-operative death within 5 years. It will be understood that“relative expression”, as used herein, is gene expression relative to the corresponding reference gene expression level. By“unchanged”,“substantially the same” or similar, as used herein, it is meant that the expression of the gene relative to the corresponding reference gene expression level is substantially similar, optionally ±15%, optionally ±10%, optionally ±5%, optionally ±2%, optionally ±1 % of the expression level of the corresponding reference gene. By “increased” relative gene expression, it is meant that the expression of the gene relative to the corresponding reference gene expression level is increased by at least about 10%, optionally at least about 15%, optionally at least about 20%, optionally at least about 25%, optionally at least about 30%, optionally at least about 40%, optionally at least about 50%, optionally at least about 60%, optionally at least about 70%, optionally at least about 80%, optionally at least about 90%, optionally at least about 100%, optionally greater than 100%, of the expression level of the corresponding reference gene. By“decreased” relative gene expression, it is meant that the expression of the gene relative to the corresponding reference gene expression level is decreased by at least about 10%, optionally at least about 15%, optionally at least about 20%, optionally at least about 25%, optionally at least about 30%, optionally at least about 40%, optionally at least about 50%, optionally at least about 60%, optionally at least about 70%, optionally at least about 80%, optionally at least about 90%, optionally about 100%, optionally 100%, of the expression level of the corresponding reference gene. Further, it will be understood that the increased, decreased or unchanged relative expression of the genes is determined, which relative expression levels correspond to a high or low risk of postoperative death, such as post-operative death within 5 years, and wherein said relative expression of the genes from the gene panels described herein allows a risk of post-operative death for the subject to be predicted according to the methods of the invention.

In a further aspect, the present invention provides a method for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death, the method comprising:

(ii) stratifying the Gleason 7 prostate cancer into high risk and low risk categories of postoperative death based on the expression level of said one or more genes.

Optionally, the risk of post-operative death is the risk of post-operative death within 5 years. Further optionally, the risk of post-operative death within 5 years is determined to be high or low, wherein high risk is a risk that is > 2-fold greater, optionally > 3-fold greater, optionally > 3.54-fold greater, optionally > 4-fold greater, optionally > 5-fold greater, as compared to low risk.

Optionally, an increased expression relative to a corresponding reference gene expression level for ACTA2 and/or TPM2 corresponds to a high risk of post-operative death, optionally post-operative death within 5 years, and thus stratifies the subject’s Gleason 7 prostate cancer into the low risk category. Optionally, an unchanged or decreased expression relative to a corresponding reference gene expression level for ACTA2 and/or TPM2 corresponds to a low risk of post-operative death, optionally post-operative death within 5 years, and thus stratifies the subject’s Gleason 7 prostate cancer into the high risk category.

Optionally, a decreased expression relative to a corresponding reference gene expression level for COX4I1 , MYH11 , ACTG2, and/or ROCK2 corresponds to a high risk of post-operative death, optionally post-operative death within 5 years, and thus stratifies the subject’s Gleason 7 prostate cancer into the high risk category. Optionally, an unchanged or increased expression relative to a corresponding reference gene expression level for COX4I1 , MYH11 , ACTG2, and/or ROCK2 corresponds to a low risk of post-operative death, optionally post-operative death within 5 years, and thus stratifies the subject’s Gleason 7 prostate cancer into the low risk category.

Optionally, as described above, the Gleason 7 prostate cancer in the subject described herein has an increased expression of PTEN. Optionally, the Gleason 7 prostate cancer in the subject has an increased expression of PTEN and low methylation. In other words, the Gleason 7 prostate cancer in the subject is designated as having high PTEN expression (PTEN-H; high PTEN gene expression and low methylation).

Optionally, the invention provides a method for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death, the method comprising:

(ii) stratifying the Gleason 7 prostate cancer into high risk and low risk categories of postoperative death based on the expression level of said genes.

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of at least one gene, optionally at least two genes, optionally at least three genes, optionally at least four genes, selected from a panel of genes comprising, or consisting of: ACTA2, ACTG2, MYH11 , and TPM2,

Optionally, the invention provides a method for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death, the method comprising: (i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of two or more genes selected from a panel of genes comprising: at least two, optionally at least three, optionally at least four, optionally at least five, optionally six, of ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1 , and at least one of TPT1 , TRPM4, UBA52, RNY1 , SUMF2, CLUAP1 , LILRB3, and RNU1-3,

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of two or more genes selected from a panel of genes comprising: at least one of ACTA2, ACTG2, MYH 1 1 , TPM2, ROCK2, and COX4I 1 , and at least one of TPT1 , TRPM4, UBA52 RNY1 , SUMF2, CLUAP1 , LILRB3, RNU1-3, CDC37, SLC44A4, DOPEY2, YIF1A, RPL9, RPS10, SNORA61 , KLK3, PHGDH, SERPINB6, LRRC26, AKT1 , CCR6, HGS, FASN, RPS19, ATP9A, CLTB, TOMM7, RPS29, and VIPR1 ,

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of two or more genes selected from a panel of genes comprising: at least two, optionally at least three, optionally at least four, optionally at least five, optionally six, of ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1 , and at least one of TPT1 , TRPM4, UBA52, RNY1 , SUMF2, CLUAP1 , LILRB3, RNU1-3, CDC37, SLC44A4, DOPEY2, YIF1A, RPL9, RPS10, SNORA61 , KLK3, PHGDH, SERPINB6, LRRC26, AKT1 , CCR6, HGS, FASN, RPS19, ATP9A, CLTB, TOMM7, RPS29, and VIPR1 ,

(ii) stratifying the Gleason 7 prostate cancer into high risk and low risk categories of postoperative death based on the expression level of said genes. Optionally, the invention provides a method for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death, the method comprising:

Optionally, an increased expression relative to a corresponding reference gene expression level for ACTA2, TPM2, UBA52, LILRB3, TPT1 , CDC37, YIF1A, RPS10, SNORA61 , KLK3, PHGDH,

LRRC26, CCR6, FASN, ATP9A, CLTB, TOMM7, and/or VIPR1 corresponds to a high risk of postoperative death, and thus stratifies the subject’s Gleason 7 prostate cancer into the low risk category. Optionally, an unchanged or decreased expression relative to a corresponding reference gene expression level for ACTA2, TPM2, UBA52, LILRB3, TPT1 , CDC37, YIF1A, RPS10,

SNORA61 , KLK3, PHGDH, LRRC26, CCR6, FASN, ATP9A, CLTB, TOMM7, and/or VIPR1 corresponds to a low risk of post-operative death, and thus stratifies the subject’s Gleason 7 prostate cancer into the high risk category.

DOPEY2, RPL9, SERPINB6, AKT1 , HGS, RPS19, and/or RPS29 corresponds to a high risk of postoperative death, optionally post-operative death within 5 years, and thus stratifies the subject’s Gleason 7 prostate cancer into the high risk category. Optionally, an unchanged or increased expression relative to a corresponding reference gene expression level for COX4I 1 , MYH1 1 ,

COX4I1 , MYH1 1 , ACTG2, RNY1 , TRPM4, ROCK2, RNU1-3, CLUAP1 , SUMP2, SLC44A4,

DOPEY2, RPL9, SERPINB6, AKT1 , HGS, RPS19, and/or RPS29 corresponds to a low risk of postoperative death, optionally post-operative death within 5 years, and thus stratifies the subject’s Gleason 7 prostate cancer into the low risk category.

In a further aspect, the present invention provides a method for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death, the method comprising: (i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of three or more of genes selected of from a gene panel comprising ACTA2, TPM2, RNY1 , COX4I 1 , LRRC26, TOMM7, RPS10, UBA52, YIF1A, ACTG2, ATP9A, CLUAP1 , VIPR1 , CDC37, MYH1 1 , SUMF2, TRPM4, PHGDH, RPS29, SERPINB6, CLTB, SLC44A4, RNU1-3,

LILRB3, HGS, DOPEY2, KLK3, TPT 1 , AKT 1 , FASN, RPS19, SNORA61 , ROCK2, CCR6, and RPL9;

(ii) stratifying the Gleason 7 prostate cancer into high risk and low risk categories of postoperative death based on the expression level of said three or more genes.

By“stratifying”, as used herein, it is meant that a Gleason 7 prostate cancer can be categorised as high risk or low risk of post-operative death, such as post-operative death within 5 years of a prostatectomy, based on the determined expression of one or more genes selected from a panel of genes comprising: ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1 , and/or the determined expression of three or more genes selected from a panel of genes comprising: ACTA2, TPM2,

RNY1 , COX4I 1 , LRRC26, TOMM7, RPS10, UBA52, YIF1A, ACTG2, ATP9A, CLUAP1 , VIPR1 , CDC37, MYH1 1 , SUMF2, TRPM4, PHGDH, RPS29, SERPINB6, CLTB, SLC44A4, RNU1-3,

LILRB3, HGS, DOPEY2, KLK3, TPT 1 , AKT 1 , FASN, RPS19, SNORA61 , ROCK2, CCR6, and RPL9. The risk may be high or low, that is, a high risk of post-operative death within 5 years relative to a low risk of post-operative death within 5 years. By“high risk”, it is meant that the risk of postoperative death within 5 years is > 2-fold greater, optionally > 3-fold greater, optionally > 3.54-fold greater, optionally > 4-fold greater, optionally > 5-fold greater, than“low risk”. Conversely, by“low risk”, it is meant that the risk of post-operative death within 5 years is > 2-fold lower, optionally > 3- fold lower, optionally > 3.54-fold lower, optionally > 4-fold lower, optionally > 5-fold lower, than“high risk”.

Optionally, the gene panel comprises ACTA2, TPM2, RNY1 , COX4I 1 , LRRC26, TOMM7, RPS10, UBA52, YIF1A, ACTG2, ATP9A, CLUAP1 , VIPR1 , CDC37, MYH1 1 , SUMF2, TRPM4, PHGDH, RPS29, SERPINB6, CLTB, SLC44A4, RNU1-3, LILRB3, HGS, DOPEY2, KLK3, TPT 1 , AKT 1 , FASN, RPS19, SNORA61 , ROCK2, CCR6, and RPL9. Optionally, the gene panel consists of ACTA2, TPM2, RNY1 , COX4I 1 , LRRC26, TOMM7, RPS10, UBA52, YIF1A, ACTG2, ATP9A, CLUAP1 , VIPR1 , CDC37, MYH1 1 , SUMF2, TRPM4, PHGDH, RPS29, SERPINB6, CLTB, SLC44A4, RNU1-3, LILRB3, HGS, DOPEY2, KLK3, TPT 1 , AKT 1 , FASN, RPS19, SNORA61 , ROCK2, CCR6, and RPL9

Optionally, the expression level of each of said genes is based on the up-regulation or down- regulation of gene expression relative to the corresponding reference gene expression level. As described above, it will be understood that the reference gene expression level is the gene expression level of the gene in non-tumour regions of a subject’s prostate.

Optionally, an increased relative expression of ACTA2, TPM2, UBA52, LILRB3, TPT1 , CDC37, YIF1A, RPS10, SNORA61 , KLK3, PHGDH, LRRC26, CCR6, FASN, ATP9A, CLTB, TOMM7, and VIPR1 , stratifies the subject’s Gleason 7 prostate cancer into the high risk category.

TPT1 , CDC37, YIF1A, RPS10, SNORA61 , KLK3, PHGDH, LRRC26, CCR6, FASN, ATP9A, CLTB, TOMM7, and VIPR1 , stratifies the subject’s Gleason 7 prostate cancer into the low risk category.

SERPINB6, AKT1 , HGS, RPS19, and RPS29, stratifies the subject’s Gleason 7 prostate cancer into the high risk category. Optionally, an unchanged or increased relative expression of one or more genes selected from COX4I1 , MYH1 1 , ACTG2, RNY1 , TRPM4, ROCK2, RNU1-3, CLUAP1 , SUMP2, SLC44A4,

DOPEY2, RPL9, SERPINB6, AKT1 , HGS, RPS19, and RPS29, stratifies the subject’s Gleason 7 prostate cancer into the low risk category.

It will be understood that the increased, decreased or unchanged relative expression of three or more genes is determined, which relative expression levels correspond to a high or low risk of postoperative death, such as post-operative death within 5 years, and which allow a Gleason 7 prostate cancer to be stratified into high risk or low risk categories of post-operative death according to the method of the invention.

In a further aspect, the present invention provides a method for treating prostate cancer in a subject, the method comprising:

(a) prognosticating Gleason 7 prostate cancer in the subject according to methods for prognosticating Gleason 7 prostate cancer described herein, or stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death according to methods for stratifying a subject’s Gleason 7 prostate cancer described herein, and

(b) administering a prostate cancer treatment to the patient when the predicted risk of postoperative death is high.

In a further aspect, the present invention provides a prostate cancer treatment for use in treating prostate cancer in a subject, wherein said use comprises administering the prostate cancer treatment to the subject when the risk of post-operative death is predicted to be high according to any of the methods described herein for prognosticating Gleason 7 prostate cancer, and/or when the subject’s Gleason 7 prostate cancer is stratified into the high risk category according to any of the methods described herein for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death.

In a further aspect, the present invention provides a prostate cancer treatment for use in treating prostate cancer in a subject, wherein said subject has a Gleason 7 prostate cancer exhibiting an altered relative expression of one or more genes selected from: increased relative expression of ACTA2 and TPM2, and decreased relative expression of ACTG2, MYH1 1 , ROCK2, and COX4I 1 ; or three or more genes selected from: increased relative expression of ACTA2, TPM2, UBA52, LILRB3, TPT1 , CDC37, YIF1A, RPS10, SNORA61 , KLK3, PHGDH, LRRC26, CCR6, FASN, ATP9A, CLTB, TOMM7, and VIPR1 , and decreased relative expression of COX4I1 , MYH1 1 , ACTG2, RNY1 , TRPM4, ROCK2, RNU1-3, CLUAP1 , SUMP2, SLC44A4, DOPEY2, RPL9, SERPINB6, AKT1 , HGS, RPS19, and RPS29.

Optionally, present invention provides a prostate cancer treatment for use in treating prostate cancer in a subject, wherein said subject has a Gleason 7 prostate cancer exhibiting an altered relative expression of one or more genes selected from: increased relative expression of ACTA2 and TPM2, and decreased relative expression of ACTG2 and MYH1 1.

Optionally, present invention provides a prostate cancer treatment for use in treating prostate cancer in a subject, wherein said subject has a Gleason 7 prostate cancer exhibiting an altered relative expression of three or more genes selected from: increased relative expression of ACTA2, TPM2, UBA52, LILRB3, and TPT1 , and decreased relative expression of COX4I1 , MYH1 1 , ACTG2, RNY1 , TRPM4, ROCK2, RNU1-3, CLUAP1 , and SUMP2. Thus, it will be understood that when the risk of post-operative death, such as post-operative death within 5-years, is high, a subject will be treated accordingly. As such, the present invention helps prevent overtreatment of a subject having Gleason 7 prostate cancer wherein the risk of postoperative death is low as prognosticated or stratified by the methods of the present invention. Such subjects can more confidently be assigned a non-treatment or active surveillance approach, whereas subjects with a high risk of post-operative death can be actively treated with an appropriate prostate cancer treatment.

Optionally, in the method for treating prostate cancer in a subject described herein, or the prostate cancer treatment for use in treating prostate cancer in a subject as described herein, the prostate cancer treatment is selected from radiotherapy, androgen deprivation therapy, or a combination of radiotherapy and androgen deprivation therapy. Further optionally, the prostate cancer treatment is a treatment selected from one or more of the treatments known in the art, such as those disclosed in Table 1 below.

Table 1. Treatment options for different stages of prostate cancer and their survival.

5 year relative

survival rates

Stage T reatment

(unadjusted for age

and race)

Localised i. Surveillance strategy 100%

(T1 ,T2) ii. Surgery (Radical prostatectomy)

iii. Radiation therapy (Brachytherapy

or external beam radiation)

iv. Cryotherapy _

Locally i. Surgery ~ 90%

advanced ii. Radiation therapy

iii. Combination of surgery, radiation

_ or hormonal therapy _

Metastatic i. Surgical castration ~ 30%

ii. Medical castration with Luteinizing

hormone-releasing hormone

_ (LHRH) agonist _

Hormone- i. Continued LHRH agonist < 20%

resistant ii. Second-line hormone therapy

iii. Clinical trial If deemed to be“high risk”, then the treatment approach would typically be an intensive treatment which would be a combination of radiotherapy and androgen deprivation therapy - i.e. a combination of the locally advanced and metastatic treatment strategies listed in Table 1. As will be understood, by employing the present invention, a skilled physician can decide the whether to actively treat a Gleason 7 prostate cancer, or decide on a non-treatment or active surveillance approach, based on the prognostication of the Gleason 7 prostate cancer in a subject, or the stratification of the subject’s Gleason 7 prostate cancer into the high or low risk category, according to the methods described herein.

In a further aspect, the present invention provides a kit for prognosticating Gleason 7 prostate cancer in a subject, or for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death, by determining the expression level of one or more genes from a panel of genes comprising, or consisting of: ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1 , the kit comprising, or consisting of,

(1 ) means for detecting and quantifying RNA and/or single-stranded cDNA corresponding to the one or more genes from the gene panel, the detecting and quantifying means comprising oligonucleotides having the complementary sequence of the RNA and/or single-stranded cDNA corresponding to the one or more genes from the gene panel, and optionally,

(2) means for extracting RNA from a biological sample containing prostate cancer cells; and/or

(3) means for reverse transcribing the RNA to form single-stranded cDNA.

Optionally, the gene panel comprises ACTA2, ACTG2, MYH1 1 , and TPM2. Optionally, the gene panel consists of ACTA2, ACTG2, MYH1 1 , and TPM2.

In a further aspect, the present invention provides a kit for prognosticating Gleason 7 prostate cancer in a subject, or for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death, by determining the expression level of three or more genes from a panel of genes comprising, or consisting of: ACTA2, TPM2, RNY1 , COX4I 1 , LRRC26, TOMM7, RPS10, UBA52, YIF1A, ACTG2, ATP9A, CLUAP1 , VIPR1 , CDC37, MYH1 1 , SUMF2, TRPM4, PHGDH, RPS29, SERPINB6, CLTB, SLC44A4, RNU1-3, LILRB3, HGS, DOPEY2, KLK3, TPT1 , AKT1 , FASN, RPS19, SNORA61 , ROCK2, CCR6, and RPL9, the kit comprising, or consisting of,

(1 ) means for detecting and quantifying RNA and/or single-stranded cDNA corresponding to three or more genes from the gene panel, the detecting and quantifying means comprising oligonucleotides having the complementary sequence of the RNA and/or single-stranded cDNA corresponding to the three or more genes from the gene panel, and optionally,

(3) means for reverse transcribing the RNA to form single-stranded cDNA.

Optionally, the gene panel comprises ACTA2, TPM2, RNY1 , COX4I 1 , LRRC26, TOMM7, RPS10, UBA52, YIF1A, ACTG2, ATP9A, CLUAP1 , VIPR1 , CDC37, MYH1 1 , SUMF2, TRPM4, PHGDH, RPS29, SERPINB6, CLTB, SLC44A4, RNU1-3, LILRB3, HGS, DOPEY2, KLK3, TPT 1 , AKT 1 , FASN, RPS19, SNORA61 , ROCK2, CCR6, and RPL9. Optionally, the gene panel consists of ACTA2, TPM2, RNY1 , COX4I 1 , LRRC26, TOMM7, RPS10, UBA52, YIF1A, ACTG2, ATP9A, CLUAP1 , VIPR1 , CDC37, MYH1 1 , SUMF2, TRPM4, PHGDH, RPS29, SERPINB6, CLTB, SLC44A4, RNU1-3, LILRB3, HGS, DOPEY2, KLK3, TPT 1 , AKT 1 , FASN, RPS19, SNORA61 , ROCK2, CCR6, and RPL9.

Optionally, the gene panel comprises COX4I 1 , MYH 1 1 , ACTG2, ACTA2, TPM2, RNY1 , TRPM4, UBA52, ROCK2, LILRB3, RNU1-3, TPT1 , CLUAP1 , and SUMF2. Optionally, the gene panel consists of COX4I1 , MYH 1 1 , ACTG2, ACTA2, TPM2, RNY1 , TRPM4, UBA52, ROCK2, LILRB3, RNU1-3, TPT1 , CLUAP1 , and SUMF2.

Optionally, the kit described herein further comprises instructions for using the kit. Optionally, the kit further comprises instructions for using the kit according to the methods for prognosticating Gleason 7 prostate cancer in a subject, or for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death, as described herein.

Optionally, means for detecting and quantifying RNA and/or single-stranded cDNA include means known in the art such as probes, e.g. oligonucleotides, having the complementary sequence of the RNA and/or single-stranded cDNA corresponding to the one or more genes selected from a panel of genes comprising: ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1 , and/or the determined expression of three or more genes selected from a panel of genes comprising: ACTA2, TPM2,

LILRB3, HGS, DOPEY2, KLK3, TPT 1 , AKT 1 , FASN, RPS19, SNORA61 , ROCK2, CCR6, and RPL9. Said probes may be bound to a solid support such as a chip, membrane or a plurality of beads. By “complementary sequence” it is meant a nucleotide sequence that is sufficiently complementary to allow specific hybridisation of the probe to a corresponding nucleotide sequence in the RNA and/or single-stranded cDNA. Optionally, the nucleotide sequence has at least about 50%, optionally at least about 60%, optionally at least about 70%, optionally at least about 80%, optionally at least about 90%, optionally at least about 95% sequence complementary to the corresponding nucleotide sequence in the RNA and/or single-stranded cDNA. The means may further included fluorophore-, silver-, or chemiluminescence- labels to label the target RNA and/or single-stranded cDNA. Means for extracting RNA from a biological sample containing prostate cancer cells include means known in the art such as phenol, chloroform, ethanol, Trizol®, and/or filter columns for RNA isolation. Means for reverse transcribing the RNA to form single-stranded cDNA includes include means known in the art such as reverse transcriptase, DNA polymerase, deoxynucleotides (dNTPs), primers such as a random primer mix, nuclease-free water, and/or magnesium chloride.

In a further aspect, the present invention provides for the use of the kit described herein for prognosticating Gleason 7 prostate cancer in a subject or for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death. In a further aspect, the present invention provides a panel of biomarkers useful for prognosticating Gleason 7 prostate cancer in a subject or for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death, wherein the panel comprises, or consists of: ACTA2, ACTG2, MYH 1 1 , TPM2, ROCK2, and COX4I 1 , optionally the panel comprises, or consists of: ACTA2, ACTG2, MYH1 1 , and TPM2.

Optionally, the panel of biomarkers further comprises one or more of TPT 1 , TRPM4, UBA52

RNY1 , SUMF2, CLUAP1 , LILRB3, and RNU1-3. Optionally, the panel of biomarkers further comprises one or more of TPT1 , TRPM4, UBA52, RNY1 , SUMF2, CLUAP1 , LILRB3, RNU1-3, CDC37, SLC44A4, DOPEY2, YIF1A, RPL9, RPS10, SNORA61 , KLK3, PHGDH, SERPINB6, LRRC26, AKT1 , CCR6, HGS, FASN, RPS19, ATP9A, CLTB, TOMM7, RPS29, and VIPR1.

In a further aspect, the present invention provides a panel of biomarkers useful for prognosticating Gleason 7 prostate cancer in a subject or for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death, wherein the panel comprises, or consists of, three or more genes selected from ACTA2, TPM2, RNY1 , COX4I 1 , LRRC26, TOMM7, RPS10, UBA52, YIF1A, ACTG2, ATP9A, CLUAP1 , VIPR1 , CDC37, MYH1 1 , SUMF2, TRPM4, PHGDH, RPS29, SERPINB6, CLTB, SLC44A4, RNU1-3, LILRB3, HGS, DOPEY2, KLK3, TPT 1 , AKT 1 , FASN, RPS19, SNORA61 , ROCK2, CCR6, and RPL9.

In a further aspect, the present invention provides a panel of biomarkers useful for prognosticating Gleason 7 prostate cancer in a subject or for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death, wherein the panel comprises, or consists of, three or more genes selected from COX4I1 , MYH1 1 , ACTG2, ACTA2, TPM2, RNY1 , TRPM4, UBA52, ROCK2, LILRB3, RNU1-3, TPT1 , CLUAP1 , and SUMF2.

Optionally,“Gleason 7 prostate cancer” corresponds to prostate cancer having a Gleason score of 3 + 4.

“Prostate cancer” as used herein includes carcinomas, including, carcinoma in situ, invasive carcinoma, metastatic carcinoma and pre-malignant conditions.

By“about”, as used herein, it is meant that the recited value may be precisely the recited value, optionally ± 10% of the recited value, further optionally ± 20% of the recited value.

Brief Description of the Drawings

The embodiments of the present invention will now be described, by way of example, with reference to the accompanying drawings, in which:

Figure 1 depicts representative immunohistochemical staining of biomarkers commonly associated with prostate cancer development. The pictures shown were captured at 20x magnification. Figure 2 depicts the expression percentages of markers in the Northern Ireland cohort (n=49) assessed by immunohistochemistry.

Figure 3 depicts a scatterplot showing the correlation between RT-qPCR-derived and whole genome DASL-derived gene expression values for PTEN expression (normalized to values of 0 to 1 ). The R value shown is Spearman rho coefficient.

Figure 4 depicts the number of DNA alterations in Northern Ireland cohort as identified by Ion Torrent® Am pi iseq Cancer Hotspot and the corresponding PTEN subtypes and clinical

characteristics.

Figure 5 depicts integrated analysis of untreated Gleason 7 prostate tumours revealing distinct subgroups associated with PTEN status. (A) shows the bivariate correlation analysis of whole genome gene expression and methylation data showing the stratification of three subgroups of cases associated with their PTEN status: PTEN-H (high PTEN gene expression and low methylation), PTEN-M (high PTEN gene expression and methylation), and PTEN-L (low PTEN gene expression and high methylation). (B) shows unsupervised hierarchical clustering of whole genome gene expression data showing clustering of patient subgroups associated with PTEN subtypes identified by bivariate analysis, and the panel of 35 genes that are differentially expressed and distinguish the PTEN-H cases and the other two subtypes (PTEN-M and PTEN-L).

Figure 6 depicts the panel of 35 genes which discriminate those cases with PTEN-H status from the others (PTEN-M and PTEN-L) identified from the unsupervised cluster analysis (Figure 5B). (A) shows the differential gene expression fold changes (log2) of each of the 35 genes. Using risk index estimation, the gene signature was subsequently assessed for its prognostic value (Figure 9) taking into consideration the gene expression status of each member (i.e. weighted by whether they are up- regulated or down-regulated as depicted). (B) shows the functional significance of the genes in relation to their biological roles.

Figure 7 shows that the prognostic value of genes of the 35-gene signature was conserved in three independently published datasets from Taylor et al. [Taylor BS, Schultz N, Hieronymus H, Gopalan A, Xiao Y, Carver BS, et al. Integrative genomic profiling of human prostate cancer. Cancer Cell 2010; 18: 1 1-22], Sboner et al. [Sboner A, Demichelis F, Calza S, Pawitan Y, Setlur SR, Hoshida Y, et al. Molecular sampling of prostate cancer: a dilemma for predicting disease progression. BMC Med Genomics 2010; 3: 8] and Gulzar et al. [Gulzar ZG, McKenney JK, Brooks JD. Increased expression of NuSAP in recurrent prostate cancer is mediated by E2F1. Oncogene 2013; 32: 70-7]. The depicted survival plots were derived from Kaplan-Meier analysis (the p-values summarized the differences by the log-rank tests). Figure 8 depicts a line chart showing the median Area-Under-the-Curve (AUC) values with 95% confidence interval (i.e. lower endpoint to upper endpoint) calculated for each different number- combination of genes.

Figure 9 depicts a schematic diagram summarizing the biological processes underlying the association of the 35-gene expression signature with PTEN/AKT and AR signalling pathways. The interplay between these two oncogenic pathways was modulated by PTEN and was differentially regulated with tumour progression. TPT1 has been previously reported to modulate the expression of PKR. Apart from TPT 1 , actins and troponin-associated factors from the 35-gene expression signature (TPM2, ACTA2 and ACTG2) were also reported to induce the activation of PKR. Notably, PKR has been reported to be important for the tumour suppressive actively of PTEN, independent of the PI3K/AKT signalling pathway. Among the four genes from the 35-gene expression signature that were highly regulated by AR (FASN, KLK3, TRPM4 and VIPR1 ), TRPM4 has been reported to be associated with tumour proliferation through the b-catenin signalling pathway. The dotted lines in the diagram signify additional biological associations that have yet to be reported and warrant further investigation.

Detailed Description

METHODS

Patient datasets

For this study, the main clinical cohort is referred to as the Northern Ireland dataset (n=49). Of the 49 cases, 28 cases were selected for whole genome gene expression and methylation analysis. These 28 cases were selected as they represent potentially treatable cases associated with a single biomarker aberration phenotype determined by immunohistochemistry, for instance, cases with just loss of PTEN. For subsequent validation of the gene expression signature, the datasets used are referred as the Taylor’s dataset (n=66), the Sboner’s dataset (n=1 17) and the Gulzar’s dataset (n=64). To address the possible confounding effect based on combined Gleason score of 4 + 3, all the selected cases were diagnosed with the combined Gleason score of 3 + 4.

Patient specimens and nucleic acids extraction

Formalin-fixed and paraffin-embedded (FFPE) radical prostatectomy tissue specimens from the Northern Ireland dataset were obtained from the Northern Ireland Biobank. Sequential 4pm section of each case were sectioned and placed on glass slides (a total of 20pm). The diagnostic histological sections were reviewed by a tissue pathologist, after clinical reporting by a prostate cancer specialist pathologist, and the Gleason score was confirmed. The tumour regions were then macro-dissected into sterile 1.5ml Eppendorf® tubes. Similarly, non-tumour regions were macro- dissected and used to generate pooled“normal-like” epithelium from the tissue specimens. Dissections were carried out according to standard techniques for transcript profiling studies, which techniques involve an experienced histopathologist marking up tissue sections based on tissue morphology for subsequent macrodissection using a scalpel blade. The RNA was extracted using the Qiagen® RNAeasy kit. Genomic DNA was extracted using the Promega® Genomic DNA extraction kit. All standard procedures were taken according to the manufacturer’s protocol.

Extracted DNA and RNA was quantified and assessed for their quality using the Agilent®

Bioanalyser chips.

Tissue microarray construction and immunohistochemistry

For the construction of tissue microarrays, representative formalin-fixed paraffin-embedded tissues from selected resection materials were cored (0.6 mm) and arrayed into donor recipient block using a tissue microarrayer (Beecher Instruments, Sun Prairie, Wl, USA). Consecutive TMA paraffin sections of 4pm thickness were cut and placed onto silanated slides for immunohistochemical detection. Immunohistochemistry were performed for PTEN. Standard processing steps for each antibody were performed according to manufacturer’s instructions. Briefly, heat-induced antigen retrieval with epitope retrieval ER1 solution (Leica Biosystems) was performed for 20 min prior to incubation with primary antibody. Slides were incubated with primary antibody at optimised concentration. After incubation, slides were washed with Bond® washing buffer (Leica Biosystems) and incubated with secondary antibody (Bond® Polymer Refine kit, Leica Biosystems).

Subsequently chromogenic detection was achieved by incubation with 3,30-diaminobenzidine (DAB) followed by Bond® DAB enhancer (Leica Biosystems). All slides were counterstained with haematoxylin and dehydrated through ascending ethanol to xylene before mounting. Further description of each commercially available antibodies and dilution used are described in Table 2.

Table 2. Immunohistochemistry conditions and antibodies used in this study.

Protein Gene description Clone Manufacturer Conditions

PTEN Phosphatase and tensin homolog 6H2.1 Dako 1 :1600

Establishing loss of PTEN by immunohistochemistry

Using visual scoring methods of the staining intensity, the evaluation of PTEN expression by immunohistochemistry was carried out by two observers. Briefly, the staining of the cytoplasm of the tumour cores was scored from 0 to 3. Where present, expression in morphologically non-malignant cells was used as an internal control. Any staining of intensity greater than 2 in the tumour nucleus or cytoplasm was then classified as retention of PTEN expression.

Cancer gene targeted next-generation DNA sequencing The Ion Ampliseq® (Life Technologies®, Carlsbad, CA, USA) assay simultaneously amplified 50 oncogenes and tumour suppressor genes covering 2,800 COSMIC mutations in a single-tube reaction. A minimum of 50ng of FFPE DNA was used for molecular profiling according to the manufacturer’s instructions with the Ion PGM system. Briefly, the pooled DNA was paired and amplified with Ion Torrent® adapters to produce a DNA template library. The resulting library then underwent sample emulsion PCR in which copies of the DNA template were allowed to amplify in the Ion Sphere Particles (ISP). Subsequently, the ISPs were recovered and barcoded. Next, barcoded samples were sequenced on the Ion Torrent® PGM for 65 cycles, as per the recommended protocol. Finally, the resulting data were analysed for single nucleotide polymorphism (SNP) by the proprietary Variant Caller Plugin within the Ion Torrent® software suite (Life Technologies®, Carlsbad, CA,

USA).

Whole genome gene expression

Whole genome gene expression analysis was performed using the lllumina® (San Diego, CA, USA) WG-DASL assay according to manufacturer’s protocol. Briefly, 100ng of FFPE RNA was converted to cDNA by the WG-DASL assay using biotinylated-tagged random nonamer and oligo (dT) primers. The biontinylated cDNA was then mounted onto a streptavidin-coated support and further extended and ligated by gene-specific oligonucleotides (DAP). Subsequently, PCR amplification was performed. The resulting PCR products were eluted and hybridized to the lllumina® Human-Ref v3.0 Beadchip and scanned with the lllumina® iScan Reader. The image intensity values from the microarray images generated were then analysed by the GenomeStudio Gene Expression Module (lllumina®, San Diego, CA, USA) software. The processed gene expression values were subsequently used for further analysis in this study.

Whole genome methylation

Whole genome methylation analysis was performed using the lllumina® Infinium HD (San Diego,

CA, USA) assay according to manufacturer’s protocol. Briefly, 1000ng of genomic DNA extracted from the FFPE samples was firstly treated with sodium bisulphite to convert unmethylated cytosines to uracils. The bisulphite treated DNA was denatured isothermally amplified overnight. After amplification, the post-amplified DNA was fragmented using a proprietary enzymatic process and precipitated using isopropanol. The precipitated DNA was then collected by centrifugation and resuspended in a hybridization buffer. The hybridized product was then hybridized onto the Infinium 450K Beadchip. The loaded chip underwent further extension and staining steps. Subsequently, the lllumina® iScan reader was used to derive image intensity values off the stained chip from the high- resolution scans of the chip. The image intensity values was processed and normalized by the GenomeStudio Gene Expression Module (lllumina®, San Diego, CA, USA). The processed methylation values were subsequently used for further analysis in this study.

Statistical analysis Gene expression data and associated clinical parameters deposited at Gene Expression Omnibus were downloaded for the Taylor’s, Sboner’s and Gulzar’s datasets. For the Northern Ireland’s dataset, gene expression and methylation data were exported from GenomeStudio software (lllumina®, San Diego, CA, USA).

The NMF package for the R statistical software was used for the cluster analysis for the whole genome gene expression and methylation values. The NMF method allowed identification of clusters in an unsupervised manner based on the Euclidean distance and average linkage.

Survival analyses were conducted using the Kaplan-Meier component from survival R packages.

For risk estimation, the cases were stratified into low- and high-risk categories using the

SurvExpress R package available from the SurvExpress biomarker validation tool. The stratification into low and high-risk categories is based on the calculated prognostic index (PI) obtained through the linear component of the Cox proportional hazard models, PI = b-iC-i + b₂C₂ + ... + b_rC_r where X, is the expression value and b, is obtained from the Cox fitting. For each individual cohort, the median PI value is used to split the groups into“low” or“high” risk, wherein“risk” is the risk of death within five years of prostatectomy surgery. The PI value is calculated by (i) classical Cox regression analysis for a particular gene, or (ii) assigning the gene a weightage corresponding to the gene expression value. These values are then fitted in the R statistical package (http://cran.r-project.org) using the“survival package” (see e.g. Aguirre-Gamboa, Raul, et al.,“SurvExpress: An Online Biomarker Validation Tool and Database for Cancer Gene Expression Data Using Survival Analysis”, PloS One, 2013). Cox regression analysis is known in the art and is frequently used as a statistical model for generating survival statistics and understanding prognosis of a particular treatment/marker - see e.g. Collett, D.,“Modelling Survival Data in Medical Research”, Chapman and Hall / CRC,

1993 (page 368). Sampling was performed over 1000 iterations to obtain an average PI and its standard deviation.

RESULTS

Distinct gene expression profile associated with PTEN expression

The loss of the tumour suppressor PTEN gene has previously been observed to be lost in 50 to 70% of all prostate cancer cases. Similarly, the loss of PTEN at the protein level was observed with high frequency (65.3%) in our Northern Ireland cohort when assessed by immunohistochemistry (Figures 1 and 2). This was further validated at the mRNA level with the whole genome gene expression assay as well as single marker real-time PCR analysis with high concordance (Figure 3). However, we did not observe any identifiable PTEN genomic alterations through high throughput gene-targeted sequencing (see METHODS:“Cancer gene targeted next-generation DNA sequencing”, and Figure 4). Concurrently, the high levels of PTEN methylation status suggested that the high occurrence of PTEN loss in the cohort could be due to gene methylation status. These observations also led us to hypothesize that there could be different subgroups of patients with differing PTEN status based on their gene expression and methylation status. To confirm this, we performed a bivariate correlation analysis between the gene expression and the methylation values based on the beta-values (ranging from 0 to 1 ). The bivariate correlation analysis showed distinct stratification of the cases into high PTEN expressing (PTEN-H; high PTEN gene expression and low methylation), moderate PTEN expression (PTEN-M; high PTEN gene expression and methylation)) and low PTEN expressing (PTEN-L; low PTEN gene expression and high methylation) subgroups (P<0.001 ) (Figure 5A). We then sought to investigate if these subtypes could be recapitulated by cluster analysis. By performing unsupervised hierarchical clustering analysis, the separation of PTEN-H cases from the PTEN-M and PTEN-L cases was reiterated. Through cluster analysis, we also observed a distinct expression profile that distinguished the PTEN-H cases from the other two subtypes (Figure 5B).

Identification of a 35-gene expression signature associated with prognostic outcome in prostate cancers

From the cluster analysis of the gene expression data, we were able to identify a group of 35 genes that was differentially expressed in the PTEN-H compared to PTEN-M and PTEN-L patients in the Northern Ireland’s gene expression dataset (Figure 5B). The 35 genes, their accession numbers, the corresponding lllumina® Array probe ID, and Qualified Gene Ontology (GO) term are provided in Table 3. Next, we sought to investigate the prognostic potential of genes of this signature in three published prostate cancer gene expression datasets: Taylor et al., Sboner et al. and Gulzar et al. - see Table 4. All three datasets which contain extensive clinical-pathological features, as well as time to recurrence [Taylor and Gulzar] and time to death [Sboner]. We further limited the survival analysis to Gleason 7 (3 + 4) cases in each dataset to resemble clinical features similar to the Northern Ireland’s dataset. Next, we used a risk estimation index based on the Cox proportional hazard modelling to determine the prognostic effect of the 35-gene expression signature. The risk index allows us to take into consideration the degree of expression of each individual gene within the signature by attributing a weighting value to each gene based on their expression level (whether up- or down-regulated relative to corresponding gene expression level in a pooled“normal-like” sample). We then dichotomized the cases into high and low risk categories based on the beta-values derived from the risk index estimation. We observed strong survival discrimination (i.e. survival for 5 years following surgery) in the Taylor’s (HR=6.92, 95% Cl=2.73-17.54, p<0.0001 ; Figure 7) and Gulzar’s datasets (HR=6.40, 95% Cl=2.28-17.96, p=0.0009; Figure 7). To a lesser extent, a similar effect was also observed in the Sboner’s dataset (HR=1.77, 95% Cl=1.29-2.41 , p=0.0004; Figure 7).

Table 3. Genes of 35-gene signature, and their accession numbers, lllumina® Array probe ID, and Qualified Gene Ontology (GO) term.

Gene Accession lllumina® Qualified Gene Ontology (GO) term

Number Array Probe ID

ACTA2 NM_001613.1 ILMN_1671703 Epithelial cell development (G0:0002064) ACTG2 NM_001615.3 ILMN_1795325 ATP binding (G0:0005524)

AKT1 NM_005163.2 ILMN_2388507 Cellular response to cytokine stimulus

(G0:0071345)

ATP9A NM_0006045.1 ILMN_2089073 Integral component of membrane

(G0:0016021 )

CCR6 NM_031409.3 ILMN_2387696 Cellular response to cytokine stimulus

(G0:0071345)

CDC37 NM_007065.3 ILMN_1668369 Protein targeting (G0:0006605)

CLTB NM_007097.2 ILMN_1674609 Protein binding (G0:0005515)

CLUAP1 NM_024793.1 ILMN_1750596 Cellular component assembly involved in morphogenesis (G0:0010927)

COX4I1 NM_001861 .2 ILMN_1652207 Response to extracellular stimulus

(G0:0009991 )

DOPEY2 NM_015018.2 ILMN_1739156 Post-Golgi vesicle-mediated transport

(G0:0006892)

FASN NM_004104.4 ILMN_1784871 Cellular response to cytokine stimulus

(G0:0071345)

HGS NM_004712.3 ILMN_1715994 Protein targeting (G0:0006605)

KLK3 NM_ ILMN_1655426 Regulation of vasculature development

001030050.1 (GO:1901342)

LILRB3 NM_006864.2 ILMN_1784884 Regulation of myeloid leukocyte

differentiation (G0:0002761 )

LRRC26 NM_001013653 ILMN_1680757 Protein binding (G0:0005515)

MYH1 1 NM_002474.2 ILMN_1660086 Cellular component assembly involved in morphogenesis (G0:0010927)

PHGDH NM_006623.2 ILMN_1704537 Cellular amino acid metabolic process

(G0:0006520)

RNU1-3 NR 004408.1 ILMN 3246273 Small nuclear RNAs

RNY1 NR_004391 ILMN_3237623 Small cytoplasmic ribonucleoprotein

ROCK2 NM_004850.3 ILMN_1659099 Regulation of cell cycle process

(G0:0010564)

RPL9 NM_001024921 .2 ILMN_1729033 Protein targeting (G0:0006605)

RPS10 NM_001014.3 ILMN_1686954 Protein targeting (G0:0006605)

RPS19 NM_001022.3 ILMN_1784717 Regulation of innate immune response

(G0:0045088)

RPS29 NM_001032.3 ILMN_1694742 Protein targeting (G0:0006605)

SERPINB6 NM 004568.4 ILMN_1712400 Protease binding (G0:0002020 protease) SLC44A4 NM_025257.2 ILMN_1730977 Positive regulation of cell growth

(G0:0030307)

SNORA61 NR_002987.1 ILMN_3245458 Small nucleolar RNA

SUMF2 NM_001042470.1 ILMN_1685371 Protein binding (G0:0005515)

TOMM7 NM_019059.2 ILMN_1674069 Protein targeting (G0:0006605)

TPM2 NM_213674.1 ILMN_1789196 Regulation of ATPase activity

(G0:0043462)

TPT1 NM_003295.1 ILMN_1789614 Stem cell maintenance (G0:0019827)

TRPM4 NM_017636.2 ILMN_1679401 Leukocyte migration (G0:0050900)

UBA52 NM_001033930.1 ILMN_1782977 Regulation of cell cycle process

(G0:0010564)

VIPR1 NM_004624.2 ILMN_1707959 G-protein coupled receptor signalling

pathway, coupled to cyclic nucleotide second messenger (G0:0007187)

YIF1A NM_020470.1 ILMN_1712975 ER-nucleus signalling pathway

(G0:0006984)

Table 4. Demographics of prostate cancer datasets used in this study.

Reference Institution Sample size Time Median Database period follow-up

(months)

Taylor et al., Memorial 218 2000-2006 57.4 GSE21032 Cancer Cell Sloan-Kettering (66 Gleason7)

2010 Cancer Center

Sboner et al., Weill Cornell 281 1977- 102.9 GSE16560 BMC Med Medical (1 17 Gleason 7) 1999

Genomics College

2010

Gulzar et al., Stanford 98 1998-2007 37.3 GSE40272

Oncogene University (64 Gleason 7)

2013

Ong et al., Northern 62 2009-2013 N.A. N.A.

(unpublished) Ireland Biobank (49 Gleason 7)

Identification of genes from the 35-gene signature associated with prognostic outcome in prostate cancers whose expression distinguishes between high and low risk cases

We generated a gene ranking based on each gene’s statistical power to define expression between high risk and low risk cases and the results depicted in Table 5. The p value of 0.05 is based on significant differential expression of the indicated genes between patient groups with good and poor outcomes. The gene ranking was based on the statistical differences in terms of gene expression between the“low” risk and“high” risk groups. For instance, COX4I1 has the greatest differential expression between the risk categories. As can be seen, 14 genes achieve a p value of 0.05 or less (italicised and underlined). Table 5. Gene ranking based on each gene’s statistical power to define expression between high risk and low risk cases. Gene ranking was based on a Chi-squared statistical test of the expression difference for each gene between the’’low” and’’high” risk groups - the p value generated by this test has been used to rank the genes from the most to least significant.

The 14 genes noted above were analysed further based on the p value for the hazard ratio with a 95% confidence interval. Hazard ratio is an actionable risk prediction measure and is used to indicate which of the 14 genes having a p value of 0.05 or less might be most informative for clinical risk prediction. Thus, the prognostic effect of these 14 genes was determined by univariate Cox proportional hazard analysis. The results are summarised in Table 6, in which the 14 genes are ranked in the order of their prognostic value as a single biomarker. As depicted in Table 6, six genes achieved a hazard ratio with a p value of <0.1 (italicised), and four genes achieved a hazard ratio with a p value of <0.05 (italicised and underlined), thus indicating that these genes (ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2 and COX4I1 ) are most informative for clinical risk prediction.

Table 6. The hazard risk ratio for each gene in predicting overall survival outcome, ranked in the order of individual prognostic value.

The Prognostic Index may be calculated as follows:

Prognostic Index = $_Gene1 _XGene1 + Gene2 _XGene2 + + b(O_qG_q i) / ' (Gere i)

Where X_Gene is the gene expression value derived from quantile-normalised and log2 transformed values from high-throughput gene expression assays and the

is the weightage value associated for the specific gene.

For example, in the case of the 14 genes indicated in Table 6, PI = 0.994 XACTA2 - 1.936 XACTG2 - 0.594X_MYHII + 0.086XTPM2 - 0.692 XROCK2 - 1.17X_Gox4n + 8.223 ctrti - 1.31 5XTRPM4 + 4.70QXUBAS2 - 0.201 XRNYI - 2.012 XSUMF2 - 0.941 XQLUAPI + 1.1 15XULRB3 - 0.178 XRNUIG3

The threshold cut-off for risk stratification (for the corresponding number of genes) can be obtained for Table 6A. For example, using the Taylor’s dataset as a training cohort, high risk cases (using 14 genes) can be defined as any cases with prognostic index value > -83.151. Table 6A. Cut-off threshold value for risk stratification calculated using prognostic index

The utility of genes, and gene combinations, from the 35-gene signature in risk prediction was also examined in an Area-Under-the-Curve (AUC) assessment. AUC assessment reflects the specificity and sensitivity of a biomarker as a predictor of survival. A value of higher than about 0.6 can be understood to indicate that the gene(s) has clinical utility as a predictor of risk. The line chart in Figure 8 shows the median AUC values with 95% confidence interval (i.e. lower endpoint to upper endpoint) calculated for each different number-combination of genes. As can be seen, gene combinations comprising one and three or more genes exhibited an AUC greater than 0.6. Two gene combinations exhibited an AUC close to 0.6. Thus, for instance, any random combination of 18 genes will give a median AUC value of 0.72 (ranging from 0.70 to 0.74), indicating that combinations of 18 genes from the present gene signature have clinical utility as a predictor of disease outcome. Thus, gene signatures as described herein can outperform PSA testing, where reported AUC values range from 0.523 to 0.626 (Etzioni et al., 2007,“Is prostate-specific antigen velocity useful in early detection of prostate cancer? A critical appraisal of the evidence”, Journal of the National Cancer Institute, 99(20), 1510-1515; Loeb and Catalona, 2014,“The Prostate Health Index: a new test for the detection of prostate cancer”, Therapeutic Advances in Urology, 6(2): 74-77).

Next, we took the genes in Table 6 and, starting from the most significant gene, we added an additional gene (in the order of their significance) and reported the AUC in Table 7.

Table 7. The AUC value of ACTA2 and combinations of genes with ACTA2 following addition of gene(s) based on their order of significance as reported in Table 6. The cut-off PI values for high- risk cases are obtained from Table 6A.

Next, we took the genes in Table 6 and, starting from the most significant gene, we added an additional gene (in the order of their significance) and reported the hazard risk ratio in Table 8.

Table 8. Hazard risk ratios of ACTA2 and combinations of genes (thereafter) following addition of gene(s) based on their order of significance as reported in Table 6 in Taylor et al. (GSE21032) prostate cancer dataset. The cut-off PI values for high-risk cases are obtained from Table 6A.

DISCUSSION

The loss of genomic PTEN was widely recognised as a common genomic aberration in prostate cancer. Generally, PTEN loss is suggested to be more frequently occurring in metastatic prostate cancers than in primary tumours. The frequency of loss of PTEN protein expression examined by immunohistochemistry in our study was consistent with previously reported findings.

Through cluster analysis, we have also observed a distinct gene expression profile for subgroup of PTEN expressing cases (Figure 5B). These PTEN-H cases were marked by the differential expression of a panel of 35 genes that discriminate them from the other two subtypes, PTEN-M and PTEN-L (Figure 5B).

Within our PTEN-H gene signature, there are also several components that suggest that possible role of PKR activation in mediating PTEN expression, particularly through TPT1. The TPT1 gene has been previously shown to be associated with disease progression in prostate and colorectal cancers. Functionally, TPT1 was reported to be involved in several biological processes, including rapamycin signalling as well as mitosis and nuclear reprogramming. Furthermore, gene silencing and knockdown of TPT1 was reported to inhibit cell proliferation and invasion. Notably, it is postulated that the activation of TPT1 modulates the activity serine-threonine kinase PKR. A recent study has previously established the role of PKR in the tumour suppressive activity of PTEN as an alternative link that is independent of PI3K signalling pathway. In that study, it was reported that the activation of PKR resulted in the phosphorylation of EIF2 leading to subsequent inhibition of cell proliferation. Apart from TPT1 , actin and tropomycin has also been described to induce the in vitro activation of PKR. Interestingly, within our PTEN-H signature, there are several representative factors present, namely TPM2, ACTA2 and ACTG2. Taken together, our PTEN-H signature uncovered an alternative network of tumour suppressive features, independent of PI3K signalling, contributed by PKR activity and its interaction with other proteins (Figure 9).

Our finding shows a gene signature associated with survival outcome in untreated Gleason 7 prostate cancers that can identify high risks patients that would benefit from early treatment intervention. Currently, biomarkers evaluated in most studies correlate with grade and high-risk staging, hence possibly confounded by tumour proliferation. In our study we have essentially excluded that bias and proliferative component by limiting the cases analysed to Gleason 7 (3 + 4) cases. Our signature therefore allows us to further postulate that there is limited prospect of progression for the low risk patients despite pathological presentation of neoplastic phenotypes and overexpression of previously reported prognostic markers such as TPT1. We hypothesised that this is due to the numerous tumour suppressive checkpoints, such as the role of PKR activity as well as functional androgen receptor signalling (Figure 9). Furthermore this also implies that some of these genes may become“oncogenic” or contributes to aggressive disease once cell cycle checkpoints are compromised. Without such changes, they are but rather regulators of normal energy balance or moderators of stress. In summary, our study is the first that evaluates markers in a single Gleason score setting. More importantly, our gene expression signature allows for better stratification and prognostication of Gleason 7 prostate cancers.

The invention is not limited to the embodiments described herein but can be amended or modified without departing from the scope of the present invention.

Claims

1. A method of prognosticating Gleason 7 prostate cancer in a subject, the method comprising:

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of one or more genes selected from a panel of genes comprising: ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1 ,

2. The method of Claim 1 , wherein an increased expression relative to a corresponding reference gene expression level for ACTA2 and/or TPM2 corresponds to a high risk of post-operative death.

3. The method of Claim 1 , wherein unchanged or decreased expression relative to a corresponding reference gene expression level for ACTA2 and/or TPM2 corresponds to a low risk of post-operative death.

4. The method of Claim 1 , wherein a decreased expression relative to a corresponding reference gene expression level for COX4I1 , MYH1 1 , ACTG2, and/or ROCK2 corresponds to a high risk of post-operative death.

5. The method of Claim 1 , wherein an unchanged or increased expression relative to a corresponding reference gene expression level for COX4I 1 , MYH 1 1 , ACTG2, and/or ROCK2 corresponds to a low risk of post-operative death.

6. The method of any one of the preceding claims, wherein the Gleason 7 prostate cancer in the subject has normal or increased expression levels of phosphatase and tensin homolog (PTEN; accession number NM_000314).

7. The method of Claim 6, wherein the Gleason 7 prostate cancer in the subject has normal or increased expression of PTEN and low gene methylation levels.

8. The method of any one of the preceding claims, the method comprising:

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of at least one gene, optionally at least two genes, optionally at least three genes, optionally at least four genes, optionally at least five genes, optionally at least six genes, selected from a panel of genes comprising: ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1 ,

9. The method of any one of the preceding claims, the method comprising: (i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of at least one gene, optionally at least two genes, optionally at least three genes, optionally at least four genes, selected from a panel of genes comprising: ACTA2, ACTG2, MYH1 1 , and TPM2,

10. The method of any one of the Claims 1 to 8, the method comprising:

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of two or more genes selected from a panel of genes comprising: at least one of ACTA2, ACTG2, MYH 1 1 , TPM2, ROCK2, and COX4I 1 , and at least one of TPT1 , TRPM4, UBA52 RNY1 , SUMF2, CLUAP1 , LILRB3, RNU1-3,

1 1. The method of Claim 10, the method comprising:

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of two or more genes selected from a panel of genes comprising: at least one, optionally at least two, optionally at least three, optionally at least four, optionally at least five, optionally six, of ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1 , and at least one, optionally at least two, optionally at least three, optionally at least four, optionally at least five, optionally at least six, optionally at least seven, optionally eight, of TPT1 , TRPM4, UBA52, RNY1 , SUMF2, CLUAP1 , LILRB3, and RNU1-3,

12. The method of any one of the Claims 1 to 8, the method comprising:

CLTB, TOMM7, RPS29, and VIPR1 ,

13. The method of Claim 12, the method comprising:

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of two or more genes selected from a panel of genes comprising: at least one, optionally at least two, optionally at least three, optionally at least four, optionally at least five, optionally six, of ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1 , and at least one, optionally at least 2, optionally at least 3, optionally at least 4, optionally at least 5, optionally at least 6, optionally at least 7, optionally at least 8, optionally at least 9, optionally at least 10, optionally at least 1 1 , optionally at least 12, optionally at least 13, optionally at least 14, optionally at least 15, optionally at least 16, optionally at least 17, optionally at least 18, optionally at least 19, optionally at least 20, optionally at least 21 , optionally at least 22, optionally at least 23, optionally at least 24, optionally at least 25, optionally at least 26, optionally at least 27, optionally at least 28, optionally 29 of TPT1 , TRPM4, UBA52, RNY1 , SUMF2, CLUAP1 , LILRB3, RNU1-3, CDC37, SLC44A4, DOPEY2, YIF1A, RPL9, RPS10, SNORA61 , KLK3, PHGDH, SERPINB6, LRRC26, AKT1 , CCR6, HGS, FASN, RPS19, ATP9A, CLTB, TOMM7, RPS29, and VIPR1 ,

14. The method of any one of Claims 10-13, wherein decreased expression relative to a corresponding reference gene expression level for ACTA2, TPM2, UBA52, LILRB3, TPT1 , CDC37, YIF1A, RPS10, SNORA61 , KLK3, PHGDH, LRRC26, CCR6, FASN, ATP9A, CLTB, TOMM7, and/or VIPR1 corresponds to a high risk of post-operative death.

15. The method of any one of Claims 10-13, wherein an unchanged or decreased expression relative to a corresponding reference gene expression level for ACTA2, TPM2, UBA52, LILRB3, TPT1 , CDC37, YIF1A, RPS10, SNORA61 , KLK3, PHGDH, LRRC26, CCR6, FASN, ATP9A, CLTB, TOMM7, and/or VIPR1 corresponds to a low risk of post-operative death.

16. The method of any one of Claims 10-13, wherein a decreased expression relative to a corresponding reference gene expression level for COX4I 1 , MYH 1 1 , ACTG2, RNY1 , TRPM4, ROCK2, RNU1-3, CLUAP1 , SUMP2, SLC44A4, DOPEY2, RPL9, SERPINB6, AKT1 , HGS, RPS19, and/or RPS29 corresponds to a high risk of post-operative death.

17. The method of any one of Claims 10-13, wherein an unchanged or increased expression relative to a corresponding reference gene expression level for COX4I 1 , MYH1 1 , COX4I1 , MYH1 1 , ACTG2, RNY1 , TRPM4, ROCK2, RNU1-3, CLUAP1 , SUMP2, SLC44A4, DOPEY2, RPL9,

SERPINB6, AKT1 , HGS, RPS19, and/or RPS29 corresponds to a low risk of post-operative death.

18. A method for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death, the method comprising:

(i) deter determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of one or more genes selected from a panel of genes comprising, or consisting of: ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1 ,

19. The method of Claim 18, wherein an increased expression relative to a corresponding reference gene expression level for ACTA2 and/or TPM2 corresponds to a high risk of post-operative death.

20. The method of Claim 18, wherein unchanged or decreased expression relative to a corresponding reference gene expression level for ACTA2 and/or TPM2 corresponds to a low risk of post-operative death.

21. The method of Claim 18, wherein a decreased expression relative to a corresponding reference gene expression level for COX4I1 , MYH1 1 , ACTG2, and/or ROCK2 corresponds to a high risk of post-operative death.

22. The method of Claim 18, wherein an unchanged or increased expression relative to a corresponding reference gene expression level for COX4I 1 , MYH 1 1 , ACTG2, and/or ROCK2 corresponds to a low risk of post-operative death.

23. The method of any one of Claims 18-22, wherein the Gleason 7 prostate cancer in the subject has normal or increased expression levels of phosphatase and tensin homolog (PTEN; accession number NM_000314).

24. The method of Claim 23, wherein the Gleason 7 prostate cancer in the subject has normal or increased expression of PTEN and low gene methylation levels.

25. The method of any one of Claims 18-24, the method comprising:

26. The method of any one of Claims 18-25, the method comprising:

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of at least one gene, optionally at least two genes, optionally at least three genes, optionally at least four genes, selected from a panel of genes comprising: ACTA2, ACTG2, MYH1 1 , and TPM2,

27. The method of any one of Claims 18-25, the method comprising: (i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of two or more genes selected from a panel of genes comprising: at least one of ACTA2, ACTG2, MYH 1 1 , TPM2, ROCK2, and COX4I 1 , and at least one of TPT1 , TRPM4, UBA52 RNY1 , SUMF2, CLUAP1 , LILRB3, RNU1-3,

28. The method of Claim 27, the method comprising:

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of two or more genes selected from a panel of genes comprising: at least one, optionally at least two, optionally at least three, optionally at least four, optionally at least five, optionally six, of ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I 1 , and at least one, optionally at least two, optionally at least three, optionally at least four, optionally at least five, optionally at least six, optionally at least seven, optionally eight, of TPT1 , TRPM4, UBA52, RNY1 , SUMF2, CLUAP1 , LILRB3, and RNU1-3,

29. The method of any one of Claims 18-25, the method comprising:

(i) determining, in a Gleason 7 prostate cancer sample taken from the subject, the expression level of two or more genes selected from a panel of genes comprising: at least one of ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I 1 , and at least one of TPT1 , TRPM4, UBA52 RNY1 , SUMF2, CLUAP1 , LILRB3, RNU1-3, CDC37, SLC44A4, DOPEY2, YIF1A, RPL9, RPS10, SNORA61 , KLK3, PHGDH, SERPINB6, LRRC26, AKT1 , CCR6, HGS, FASN, RPS19, ATP9A,

CLTB, TOMM7, RPS29, and VIPR1 ,

30. The method of Claim 29, the method comprising:

31. The method of any one of Claims 27-30, wherein decreased expression relative to a corresponding reference gene expression level for ACT A2, TPM2, UBA52, LILRB3, TPT1 , CDC37, YIF1A, RPS10, SNORA61 , KLK3, PHGDH, LRRC26, CCR6, FASN, ATP9A, CLTB, TOMM7, and/or VIPR1 corresponds to a high risk of post-operative death.

32. The method of any one of Claims 27-30, wherein an unchanged or decreased expression relative to a corresponding reference gene expression level for ACTA2, TPM2, UBA52, LILRB3, TPT1 , CDC37, YIF1A, RPS10, SNORA61 , KLK3, PHGDH, LRRC26, CCR6, FASN, ATP9A, CLTB, TOMM7, and/or VIPR1 corresponds to a low risk of post-operative death.

33. The method of any one of Claims 27-30, wherein a decreased expression relative to a corresponding reference gene expression level for COX4I 1 , MYH 1 1 , ACTG2, RNY1 , TRPM4, ROCK2, RNU1-3, CLUAP1 , SUMP2, SLC44A4, DOPEY2, RPL9, SERPINB6, AKT1 , HGS, RPS19, and/or RPS29 corresponds to a high risk of post-operative death.

34. The method of any one of Claims 27-30, wherein an unchanged or increased expression relative to a corresponding reference gene expression level for COX4I 1 , MYH1 1 , COX4I1 , MYH1 1 , ACTG2, RNY1 , TRPM4, ROCK2, RNU1-3, CLUAP1 , SUMP2, SLC44A4, DOPEY2, RPL9,

35. The method of Claim 1 , wherein the risk of post-operative death is the risk of post-operative death within 5 years.

36. The method of Claim 2, wherein the risk of post-operative death within 5 years is determined to be high or low, wherein high risk is a risk that is >2-fold greater, optionally >3-fold greater, optionally >3.54-fold greater, optionally >4-fold greater, optionally >5-fold greater, than low risk.

37. A method for treating prostate cancer in a subject, the method comprising:

(a) prognosticating Gleason 7 prostate cancer in the subject according to the method of any one of the Claims 1-17 or stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death according to the method of any one of the Claims 18-34, and

38. A prostate cancer treatment for use in treating prostate cancer in a subject, wherein said use comprises administering the a prostate cancer treatment to the subject when the risk of post- operative death is predicted to be high according to the method of any one of Claims 1-17 or when a subject’s Gleason 7 prostate cancer is stratified into the high risk category according to the method of any one of Claims 18-34.

39. A prostate cancer treatment for use in treating prostate cancer in a subject, wherein said subject has Gleason 7 prostate cancer exhibiting an altered relative expression of one or more genes selected from: increased relative expression of ACTA2 and TPM2, and decreased relative expression of COX4I1 , MYH 1 1 , ACTG2, and ROCK2.

40. A prostate cancer treatment for use in treating prostate cancer in a subject, wherein said subject has Gleason 7 prostate cancer exhibiting an altered relative expression of one or more genes selected from: increased relative expression of ACTA2 and TPM2, and decreased relative expression of MYH1 1 and ACTG2.

41. The method for treating prostate cancer in a subject according to Claim 37, or the prostate cancer treatment for use according to any one of Claims 38-40, wherein the prostate cancer treatment is radiotherapy, androgen deprivation therapy, or a combination of radiotherapy and androgen deprivation therapy.

42. A kit for prognosticating Gleason 7 prostate cancer in a subject, or for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of post-operative death, by determining the expression level of one or more genes from a panel of genes comprising: ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1 , the kit comprising

(3) means for reverse transcribing the RNA to form single-stranded cDNA.

43. The kit of Claim 42, wherein the gene panel comprises ACTA2, ACTG2, MYH 1 1 , TPM2.

44. Use of the kit of Claim 42 or 43 for prognosticating Gleason 7 prostate cancer in a subject or for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of postoperative death.

45. A panel of biomarkers useful for prognosticating Gleason 7 prostate cancer in a subject or for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of postoperative death, wherein the panel comprises, or consists of, one or more genes selected from ACTA2, ACTG2, MYH1 1 , TPM2, ROCK2, and COX4I1.

46. A panel of biomarkers useful for prognosticating Gleason 7 prostate cancer in a subject or for stratifying a subject’s Gleason 7 prostate cancer into high risk and low risk categories of postoperative death, wherein the panel comprises, or consists of, one or more genes selected from ACTA2, ACTG2, MYH1 1 , and TPM2.