Mark genes for prostate cancer classification

专利摘要:
43 Abstract The present invention relates to a method for classifying a prostate cancer in asubject, the method comprising the steps of a) determining a gene expression|eve| or gene expression pattern of the genes F3 and |GFBP3 in a sample fromthe subject and b) classifying the tumor by comparing the gene expression |eve|determined in a) with a reference gene expression of the same genes inreference patients known to have a high risk or low risk tumor respectively. lnaddition the invention relates to a method for determining prognosis of a subjectdiagnosed with prostate cancer, a method for making a treatment decision for asubject diagnosed with prostate cancer and a solid support or a kit for classifyinga tumor in a subject diagnosed with prostate cancer.
公开号:SE1150982A1
申请号:SE1150982
申请日:2011-10-24
公开日:2013-04-25
发明作者:Chunde Li；Zhuochun Peng；Lambert Skoog
申请人:Chundsell Medicals Ab；
IPC主号:

专利说明:

castration-resistant metastases within three years or die from cardiovascular and other side effects of castration treatment. At present, there are no methods that can predict the type of patients who would benefit from castration treatment.
The majority of prostate cancers progress so slowly that they can never reach a life-threatening condition, mainly due to old age and other competing diseases. However, a small proportion of prostate cancer progresses very quickly and can kill patients in less than five years. At diagnosis, with conventional parameters including age, tumor grade, Gleason score, clinical grade, and comorbidity, the prediction of cancer-specific and overall survival can reach an accuracy of up to 60-70%. Even patients with the same clinical prognostic parameters can show large differences in survival as well as in response to treatment. Consequently, prostate cancer is a pathological (morphological) diagnosis that can include several different biological subgroups or subtypes.
There is a need for a method that can distinguish these biological subgroups or subtypes of prostate cancer patients. There is also a need for a method that can classify these subtypes into aggressive or high-risk tumors and less aggressive or low-risk tumors, as well as a method that can predict survival for patients with the respective tumor subtype.
Furthermore, there is a need for a method that can be used to make treatment decisions for patients who have a tumor of the respective subtype, possibly also with regard to clinical parameters.
Prior Art Patent document WO2008 / 013492 A1 describes an approach for identifying genes related to embryonic stem cells, termed ES tumor predictor genes (ESTP genes), which may be important for the function of cancer stem cells. 641 ESTP genes were identified and found useful for the classification of prostate cancer tumors. Patent document WO09021338 A1 describes a method for prognosis of a cancerous disease, e.g. prostate cancer, in a patient by detecting a signature of splicing events. F3 is mentioned as one of many genes that can be used.
Patent document WO0171355 describes simultaneous analysis of PSA, IGF-1 and IGFBP-3 in blood plasma to predict the risk of a man getting prostate cancer.
US2003054419 A1 describes a method for determining the risk of progression in a prostate cancer patient after therapy, in which the levels of TGF-β1, IGFBP-2 or IGFBP-3 in plasma are measured.
Patent documents WO10006048 A and US2009298082 A describe the respective methods for predicting the survival of a patient diagnosed with prostate cancer and whether a patient with PSA recurrence will later develop systemic disease state. In both documents, IGF BP-3 is mentioned as one of many genes that, together with other molecular markers, can be used.
Documents WO00105154 and WO06028867A describe a method for determining a prognosis for an individual having cancer and a method for diagnosing multiple myeloma. c-MAF is mentioned as one of many genes that can be used.
WO1010188A describes a method for interfering with the activity of CTGF, wherein the activity of CTGT is associated with metastases of prostate cancer.
Object of the Invention It is an object of the present invention to provide molecular markers which are useful for classification, for predicting prognosis and for guiding decisions regarding the treatment of prostate cancer in a patient. Another object of the present invention is to provide new methods for classifying prostate cancer in a patient, as well as for using the classification for prognosis in the patient and for making decisions regarding the treatment of the patient.
A further object of the present invention is to provide a method of treating a patient having prostate cancer, which method is based on the patient's tumor subtype. Yet another object of the present invention is to provide tools for classifying prostate cancer or tumors in a patient.
Description of the Invention Identification of genes and gene signatures that are significantly correlated with survival in prostate cancer patients To support the concept of biological subtype, previous studies using cDNA microarrays based on whole by classified breast cancer as well as prostate cancer with molecular clinical subtypes. The present description further expands the concept and meaning. Instead of using only statistical analysis, the selection of gene marker candidates in the present study was driven by the cancer stem cell (CSC) / embryonic stem cell (ESC) hypothesis, with the aim of identifying only a few ESC / CSC gene markers of paramount importance. This approach has been shown to be effective because the most significant predictive gene markers identified in the present study were from the list of identified embryonic stem cell gene predictors (ESCGPs).
The inventors' hypothesis was that the biological aggressiveness of prostate cancer and the ability to respond to castration treatment is largely determined by main gene expression patterns in prostate cancer stem cells (CSCs) (Visvader, Nature 2011, 4692314-22; Ratajczak et al, Differentiation 2011, 10 15 20 25 30 81 : 153-161; Lang et al., J Pathol 2009, 2171299-306). Another hypothesis was that genes that have an important function in embryonic stem cells (ESCs) may also be important in prostate CSCs. Thus, direct measurement of expression patterns of genes related to embryonic stem cell genes in prostate cancer cells would reflect the biological aggressiveness of the cancer and allow prediction of the effect of castration treatment as well as prediction of patient survival.
Based on this hypothesis, the inventors have previously identified genes, ie. embryonic stem cell gene predictors (ESCGPs) that have a consistently high or consistently low level of expression in ECS lines (WO 2008/013492 A1).
Briefly, the ESCGPs were identified by analysis of previously published datasets of microarray data from whole genome cDNA derived from 5 human ESC lines and 115 human normal tissues from different organs using a simple single-class SAM (Significance Analysis of Microarrays), wherein the genes were ranked according to their degree of consistency in expression levels in the ESCs.
This was based on the concept that genes with either consistently high or consistently low expression levels in all ESC lines can have significant functions in maintaining ESC status / condition and that changes in their expression in different patterns can lead to differentiation in different directions. These ESC genes may also have functions in maintaining different status of CSCs and thus different expression patterns of ESC genes in CSCs may classify tumors into different subtypes with different aggressiveness and sensitivity to different types of treatment. Based on this list of ESCGPs, the present study identified some important prognostic and predictive gene markers for prostate cancer.
From a list of 641 ESCGPs identified in WO 2008/013492 A1, a subset of 33 ESCGPs were selected in the present study as candidates that could enable the classification of prostate cancer using fewer markers.
The candidates were selected according to three criteria as described in Example 2A (see also Figure 1), ie. according to their ranking position in the ESCGP list and according to their ranking positions in gene lists from a previous study (Lapointe et al., 10 15 20 25 30 Proc Natl Acad Sci USA 2004, 101: 81 1-816), which identified genes that could potentially be used for classification of prostate cancer subtypes and genes that could differentiate prostate cancer from normal tissue.
Furthermore, 5 genes that were not from the ESCGP list were selected according to a fourth criterion; they were described and known to be significant in prostate cancer. These described genes were used as a control to evaluate the importance of ESCGP genes relative to non-ESCGP genes in the classification of prostate cancer. Furthermore, they could optionally be included in a molecular marker signature for use in classifying prostate cancer.
Expression of the 33 selected ESCGPs and the 5 described genes were examined in three different prostate cancer cell lines by RT-PCR (see Example 2B). Of these 33 genes, 24 genes (19 ESCGPs and 5 described genes) were identified that had different expression patterns in the less aggressive cell line LNCap compared to the aggressive cell lines DU145 and PC3 (see Figure 2). These 24 genes were considered more likely to be useful for tumor classification to distinguish less aggressive from more aggressive cancer. Thus, the 24 genes (25 gene markers) were selected for optimization of multiplex quantitative PCR and evaluation of the ability to classify prostate cancer in fine needle aspiration (FNA) samples from 189 patients with prostate cancer with known clinical outcome (see Example 3A). Genes whose expression profile correlated with survival were first identified using analysis of a training set, ie. a subset of the entire group of 189 patients. The ability of the identified genes to classify tumors was then confirmed by analysis of the entire patient population.
All patients in the present group had clinically significant prostate cancer and the majority (80%) of the patients had not been treated with radical prostatectomy or full dose radiation therapy but only with castration treatment when the disease was advanced. Thus, the survival data were not affected by the curative effects of radical treatments, which, when used at an early stage in some patients with biologically aggressive cancer, can eliminate the cancer and thus the threat to life. In the present group, the follow-up period was 7-20 years and the majority (94.5%) of the patients died, which enabled a complete analysis of the actual overall survival time with minimally censored data. These properties ensured the discovery of new biomarkers for prediction of survival and were unique compared to most previous studies, where PSA recurrence and progression-free survival have been used as surrogates for general and cancer-specific survival. In the present study, both general and cancer-specific survival were used to evaluate the clinical value of prognostic biomarkers. Cancer-specific survival is mainly determined by the biological aggressiveness of cancer cells.
However, the accuracy and significance of the correlation between prognostic as well as predictive parameters (such as clinical parameters and / or biomarker expression) and cancer-specific survival may be affected by how cancer-specific survival is defined and how much data is censored due to other competing causes of death. On the other hand, overall survival is survival data without any censorship of causes of death and includes all causes of death.
Therefore, overall survival reflects not only the biological aggressiveness of cancer cells but also many other factors such as competing diseases or comorbidities, complications as well as side effects of treatment, age and life expectancy. Overall survival may be more important for prostate cancer patients than cancer-specific survival because most patients are diagnosed at an advanced age and usually have other competing diseases such as cardiovascular disease, diabetes mellitus or other malignancies (Daskivich et al, Cancer 2011, Apr 8. doi: 10.1002 / cncr.26104. [Epub before publication]).
Ten molecular marker genes showed significant correlation with overall and / or cancer-specific survival in univariate analysis (see Table 1), and can be used for classification of prostate tumors, for prognosis and also to make treatment decisions for patients, depending on the classification. of the patient's tumor. These markers were F3 (coagulation factor III; coagulation factor III), WNT5B (wingless-type MMTV integration site family, member 5B; wingles-type MMTV integration site family, member 5B), VGLL3 (rudimentary like 3 (Drosophila); vestigial like 3 (Drosophila)) , CTGF (connective tissue growth factor), IGFBP3 (insulin-like growth factor binding protein 3; insulin-like growth factor binding protein 3), c-MAF-α (long form of v-maf musculoaponeurotic fibrosarcoma oncogen homolog (avian); long form of v-maf musculoaponeurotic fibrosarcoma oncogene homolog (avian), c-MAF-b (short form of v-maf musculoaponeurotic fibrosarcoma oncogene homolog (avian); short form of v-maf musculoaponeurotic fibrosarcoma oncogene homolog (avian )- AMACR- methylacyl-CoA racemas; alpha-methylacyl-CoA racemase), MUC1 (cell surface-associated mucin 1; mucin 1, cell surface associated) and EZH2 (enhancer of zeste homolog 2 (Drosophila); enhancer of zeste homolog 2 (Drosophila)). Five of these ten genes (F3, WNT5B, CTGF, VGLL3 and IFGBP3) were ESCGPs identified from the list of genes with consistently high or low expression levels in embryonic stem cells. Two of these genes (c-MAF-a and c-MAF-b) were previously described genes known to have important functions in myeloma.
Three of the significant genes (EZH2, AMACR and MUC1) are genes previously described in connection with prostate cancer. Several previous studies have identified biomarkers such as AMACR, EXH2, MUC1 and AZGP1 and a stem cell-like signature that are correlated with relapse-free survival after radical prostatectomy (Varambally et al, Nature 2002, 419: 624-9; Rubin et al, JAMA 2002, 28711662- 70; Oon et al, Nat Rev Urol 2011, 81131-8; Lapointe et al, Cancer Res 2007, 6728504-10; Rubin et al, Cancer Epidemiol Biomarkers Prev 2005, 1421424-32; Strawbridge et al, Biomark lnsights 2008, 32303 -15; Glinsky et al, J Clin Oncol 2008, 2846-53; Glinsky et al, J Clin lnvest 2005, 115: 1503-21). The present results show that the expression level of MUC1, AMACR and EXH2 in FNA samples from prostate cancer actually correlates with either cancer-specific or overall survival. However, of the previously described gene markers, only the c-MAF-α correlation was as strong as Table 1. Cox monovalent proportional hazard analysis of ESCGPs and clinical parameters.
Variable PSA> 50 vs. PSASSO (ng / ml) WHO Tumor Degree Low vs. Medium / High Clinical Stage 1 'Progress vs. Limited Age 1 PSA § F3 ll WNT5B | VGLL3 | c-MAF-a | CTGF | | GFBP3 | c-MAF-b | EzH2 | AMAcR | Muc1 | WNT11 | BAsP1 | AZGP1 | co | _12A1 | EGR1 | LRRN1 | ERBB3 | cYR61 | FBP1 | PTN | LRP4 "THBS1 | GREM1 | METTL7A | 001-11 | Sample number * 161 181 175 185 161 92 89 152 174 100 169 69 144 148 143 177 177 148 176 175 182 89 79 79 27 27 27 35 35 35 General Survival Risk ratio (95% C |) 2.34 (165-331) 1.59 (1.16-2.18) 1.70 (1.23-2_35) 1.04 (102-106) 1.00 (1.00-1.00) 1.11 (1_04-1.17) 1.14 (104-125) 1.08 (1.03-1_13) 1.09 (1.03-1_17) 1.13 (1.03-1.23) 1.04 (0_98-1.10) 1.13 (0.96-1.33) 0.94 (0_84-1.05) 1.08 (1.02-1.16) 1.07 (1.00-1.13) 1.02 (0.97 -mos) 1.99 (093-106) 0.99 (0.94-1.05) 1.02 (0_9s-1.07) 1.03 (095-111) 1.02 (098-107) 1.04 (098-110) 1.12 (099-127) 1_11 (1.00-1_23 ) 0.99 (079-124) 1.06 (090-124) 1.03 (091-117) 1.04 (0.94-1.16) 1.08 (091-128) 1.12 (0.96-1.30) P-value <0.001 0.004 0.001 <0.001 0.005 0.001 0.004 0.002 0.007 0.008 0.16 0.13 0.26 0.01 0.04 0.38 0.87 0.81 0.34 0.47 0.26 0.23 0.07 0.06 0.93 0.47 0.66 0.40 0.37 0.14 Cancer Survival Risk Ratio (95% Cl) 2.61 (1.68-4.05) 1.94 (1 _28-2.94) 2.20 (1 .44-388) 1.03 (100-105) 1.00 (1 .00-1 _00) 1.14 (1_06-1.22) 1.26 (1 11-142) 1.07 (1 .01-1 .14) 1.09 (1_01-1.19) 1.15 (1.02-1.29) 1.09 (1_01-1.17) 1.28 (1 .04-1 _57) 0.85 (074-098) 1.08 (1.00-1.17) 1.06 (0_98-1.14) 1.02 (096-109) 0.97 (089-1 _05) 1.02 (096-108) 0.97 (091-103) 1.07 (0.97-1.18) 1.03 (098-1 _09) 1.04 (0.96-1 _13) 1.08 (0.91-1 _28) 1.08 (093-1 _25) 1.02 (073-1 _42) 1.11 (086-139) 1.02 (088-125) 1.12 (095-133) 0.96 (0_72-1_25 ) 0.94 (072-1 _24) P-value <0.001 0.002 <0.001 0.04 0.004 <0.001 <0.001 0.02 0.03 0.02 0.02 0.02 0.02 0.02 0.06 0.13 0.55 0.45 0.55 0.36 0.17 0.29 0.29 0.36 0.36 0.32 0.91 0.37 0.88 0.17 0.69 0.67 10 15 20 25 * Each ESCGP has its own sample number because not all ESCGPs have been profiled across all samples, 1 'Groups of clinical stages were classified using the Tumor Node Metastasis (TNM) system and PSA value.
Advanced clinical stage de ier was initiated as any TNM stage where T23, N1, M1 or PSA> 100.0 ng / ml.
Limited clinical stage they were initiated as T1-2NOM0 and PSAs100.0 ng / ml. 1: Age was modeled as a continuous variable. The risk ratio is stated for each 1.0-year increase in age.
§ The PSA value was modeled as a continuous variable. The risk ratio is stated for each increase of 1.0 ng / ml PSA in serum.
| The centered deita Ct value of the gene was modeled as a continuous variable. It corresponds inversely to the gene expression level. The risk ratio is given for each increase of 1.0 unit of the gene's centered delta Ct value. the correlation with the ESCGPs F3, | GFBP3 and VGLL3, identified in the present study.
The expression levels (inversely correlated with the centered delta-Ct value) of all these significant genes except EZH2 showed a positive correlation with survival time (Table 1, Risk ratio> 1). Only the expression level of EZH2 in the FNA samples was inversely correlated with patient survival. This negative correlation of EZH2 was consistent with its documented role as an oncogene. The present results for the EZH2, AMACR, | GFBP3 and c-MAF-α genes are in line with relevant results from previous studies (Varambally et al, Nature 2002, 419: 624-9; Rubin et al, Cancer Epidemiol Biomarkers Prev 2005, 14: 1424-32; Mehta et al., Cancer Res 2011, 7115154-63; Li et al., Genes Chromosomes Cancer 1999, 242175-82). | GFBP3 has a well-proven function in suppressing the metastatic process in prostate cancer (Mehta et al, Cancer Res 2011, 71: 5154-63).
The positive correlation between MUC1 and F3 and survival was unexpected.
F3 and MUC1 have documented functions in promoting the development of cancer (Strawbridge et al, Biomark lnsights 2008, 31303-15; Kasthuri et al, J Clin Oncol 2009, 27: 4834-8). The positive correlation with survival may indicate that prostate cancer cells with high expression levels of F3 and MUC1 are highly androgen dependent and sensitive to castration treatment (Strawbridge et al, Biomark lnsights 2008, 32303-15; Kasthuri et al, J Clin Oncol 2009, 27: 4834- 8; Mitchell et al., Neoplasia 2002, 4: 9-18; Brodin et al., Semin Thromb Hemost 2001, 37: 87-94). There are some prognostic and predictive markers with similar dual aspects in other cancers, such as HER-2 / neu / ERBB2 amplification in breast cancer, where breast cancer with amplification of HER- 10 15 20 25 30 35 11 2 / neu / ERBB2 has aggressive biological as well as clinical features but which respond to Tratsuzumab (Herceptin) treatment with concomitant prolonged survival.
The function of VGLL3 in prostate cancer is still unknown.
Furthermore, multivariate analysis was performed to identify genes that, independent of all clinical parameters, show correlation with survival (see Example 3A).
Four genes (F3, IGFBPS, CTGF and AMACR) showed correlation with both overall and cancer-specific survival independent of all clinical parameters (Figure 4A-K). All four genes except AMACR came from the list of ESCGPs. Two genes (VVNT5B and EZH2) showed, independent of clinical parameters, correlation with cancer-specific survival and one gene (VGLL3) showed, independent of clinical parameters, correlation with overall survival.
To study the possible additive or synergistic effects of multiple genes with respect to prediction of survival, different combinations of the ten significant genes were tested in a series of unsupervised hierarchical cluster analyzes (see Example 3B and Figure 6-7). It is important that two signatures were identified that could similarly classify tumors into three subgroups or subtypes with a significant difference in overall and cancer-specific survival. The first ESCGP signature (Signature 1) includes the genes VGLL3, IgFBP3 and F3. The second ESCGP signature (Signature 2) includes the genes c-MAF-α, IgFBP3 and F3. The classification had a strong correlation with and can be used for the prediction of a patient's overall and cancer-specific survival (see Figure 6-7 and Table 2-3). This prognostic and predictive expression signature was independent of age, PSA level, tumor grade, and clinical stage.
The major marker genes found in this study showed correlation with both overall and cancer-specific survival. This was partly due to the possibility that prostate cancer or side effects of the treatments also contributed to deaths directly caused by other diseases. This may also be due in part to the fact that the ESCGP signature can be shared by both cancer stem cells and certain types of normal stem cells in the body. Therefore, the ESCFP signature may be important for the development of both cancer and other diseases. To 12 Table 2. “,. risk analysis according to Cox of EåCüPsignature 1 and clinical parameters (Univariate and Multivarial analysis). variable Overall survival Cancer survival Univariate analysis Mullivanate Assay Univariate assay Mumvaria! Analysis Number of samples' Risk quotas P value Rlsk quota P value Qrskkvol P value Rvsk quota P value 195% Cl) (95%) Cl) i95% Cll l95% Cl) ESCGP signature lf sfuppi vs Group: fn sasiz 9141 var < 0 uoi 417 rz zr-io m; <0 om rem 04-19 se) <0 om r iziz 56-19 ss) fo oi »Gmppzvs srupea av 345m 7943561 <0 om 25i Psßvso vs PsAsso ing / mn a 2 93 n Visa sei <9 001 2 09 rim-s 94) o oz z 33: i ns -iii <0 om i ris: o 11-4 os) u ia wHo rumfgiau Làgivs meaeinøgi / Hegi av issii 03-2661 om iir-p 69-2110) oss 193 ri 04-151) u: ii izo ic 61-2 39) oso Kliniskt Stadium 1 Frarrlâkndell VS. Limited 87 2.13; 1 32-3 45) O 002 l 68 (0 91-3 0G) 010 3 B (l 94-7 70) <0 G01 3 52; 155-8 45) O 003 Ål0er§ 87 'lfJišl103-109l <0 00) 5.031,1 00-106) 005 l 06 11 02-1 10) 0003 1030994 08 ) 011 "The number of samples for the cluster analysis was 95. 87 samples had all clinical information including age at diagnosis, PSA value, tumor grade according to WHO and clinical stage. Univariate and multivariate analysis was performed on these 87 samples. 1- ESCGP signature 1 included expression signature for VGLL3, IGFBP3 and F3 The classified specimens of subtypes of tumors: Group 1, Group 2 and Group 3 by cluster analysis (Figure 6, panel A) 1: Groups of clinical stages were classified using the Tumor Node Metastasis system (TNM) and PSA value Advanced clinical stage degenerated as any TNM stage where TaB N1, M1 or PSA> 100.0 ng / ml Limited clinical stage degenerated as T1-2NOM0 and PSAs100.0 ng / ml § Age was modeled as a continuous variable.The risk ratio is stated for each 1.0-year increase in age.
Table 3. Cox Proportional Risk Analysis of ESCGP Signature 2 and Clinical Parameters (Univariate and Multivarial Analysis). variable General Survival Cancer Survival Univariate analysis Multivarlate Analysis Univariate analysis Mulbvariai Analysis Number of samples' Ratio ratio P-varrle Ratio ratio P-value Grlløølvs Grupp 87 3 lñjt 71-5 81.
GIUDD 2 vs 'SrUDD 3 37 2 02! L.Û9-3.77) 6.03 138: O 67-2 83i U 38 219m 95-5 05) i.)' J7 S12: Cl 644.543 0.29 PSA> 50 VS PSAS5Û (Mimi) S7 2 93 il 76-4 362 <0 001 339 ll 04-3113) Q 94 3 33i173-6 41) <3 001 11110 79-3 71) 51.75 'WHO Tufmrgrad Lågt VS Medelhógl / Hogl 87 155 il 03-2561- 004 138 'O 93-2 28) 0.21 193 il 04-3 57) G04 14210 74-273) 0.29 Clinical Stage I Progressed) 45 Limited 87 2 13 ll 32-3 45) Ü M2 179HO2-314) 3 04 3 87 ll * lå-Y 75)) <0 (101 3 6411 67- Elli 0 001 Ålder§ 37 lü fi i * 03-109) <0Û0l lßlßlt 01-1 08) 002 lfJ6 "O2-110 | 0003 l 06101994 08) 0.09 * The number of samples for the cluster analysis was 95. 87 samples had all clinical information including age at diagnosis, PSA value, tumor grade according to WHO and clinical stage, single and multivariate analysis was performed on these 87 samples fESCGP signal urz included expression signature for c-MAF-a , IGFBP3 and F3 The classified sample of subtypes of tumors: Group 1, Group 2 and Group 3 by cluster analysis (Figure 7). stages were classified using the Tumor Node Metastasis (TNM) system and PSA value. Advanced clinical stage de ier was initiated as any TNM stage where T23, N1, M1 or PSA> 100.0 ng / ml. Limited clinical stage de ier nier as T1-2NOM0 and PSAs100.0 ng / ml § Age was modeled as a continuous variable. The risk ratio is stated for each 1.0-year increase in age. For example, GFBP3 has been identified to have important suppressive functions in both cancer and diabetes (Yeap et al, Eur J Endocrinol 2011, 1641715-23; Mehta et al, Cancer Res 2011, 71: 5451-63). .
Embodiments of the present invention According to a first aspect, the present invention provides a method for classifying prostate cancer in a patient, comprising a) determining a gene expression level for genes F3 and | GFBP3 in a sample from the patient, in other words determining the gene expression pattern of said gene gene b) by comparing the level of gene expression, ie. the gene expression pattern, which was determined in a) with a reference gene expression for the same genes in reference patients known to have a high-risk and low-risk tumor, respectively; and c) concluding that if the ia) determined gene expression level / gene expression pattern matches the reference gene expression of the reference patients with a high-risk tumor, the tumor of the patient is a high-risk tumor, and if the ia) determined gene expression level corresponds to the reference gene expression of the reference patient the patient a low-risk tumor.
In a preferred embodiment, the expression level of the genes F3 and | GFBP3 and any of VGLL3 and c-MAF in a) is determined, and is thus used to classify the tumor. Preferably, the expression level is determined by F3, | GFBP3 and VGLL3.
These gene signatures have been shown to be particularly useful for the classification of prostate cancer tumors (Figure 6-7) and the resulting classification has been shown to be significantly correlated with survival in prostate cancer patients (Figures 6 and 9-12, Tables 2 and 3).
Thus, according to one embodiment, step a) further comprises determining a gene expression level for one or more of the genes VGLL3 and c-MAF, preferably VGLL3.
According to a further embodiment, step a) also comprises determining a gene expression level for one or more of the genes WNT5B and CTGF, EZH2, AMACR and MUC1. According to a second aspect, the present invention provides a method for classifying prostate cancer in a patient, comprising the steps of: a) determining a gene expression level, for at least one of genes selected from F3, IgGBP3, VGLL3, c-MAF, WNT5B and / or CTGF in a sample from the patient; b) classifying the tumor by comparing the gene expression level determined in a) with a reference gene expression for the same gene (s) in reference patients who are known to have a high-risk and a low-risk tumor, respectively; and c) concluding that if the ia) determined gene expression level matches the reference gene expression of the reference patients with a high risk tumor, the tumor of the patient is a high risk tumor, and if the ia) determined gene expression level is consistent with the reference gene expression of the reference patients with a low risk tumor, .
The second aspect of the present invention is based on the fact recognized herein that the expression of any of F3, IgFBP3, VGLL3, c-MAF, WNT5B and CTGF in samples from patients having prostate cancer may serve as an indicator of disease state in said patient. The inventors have found that there is a positive correlation between the gene expression levels of any of said genes and survival. More specifically, the inventors of the present invention have found a correlation between a high level of expression of any of F3, IGFBP3, VGLL3, c-MAF, WNT5B and CTGF and longer survival, thus low risk tumors. On the other hand, a low expression level of any of said genes is correlated with shorter survival and thus high-risk tumors.
According to an embodiment of this second aspect, the expression level of at least two, such as two, three or four of the genes F3, IgGBP3, VGLL3, c-MAF, WNT5B and CTGF is determined in step a) of the method of the present invention and is thus used to classify the tumor. .
In a further embodiment, the expression level of all genes F3, IgFBP3, VGLL3, c-MAF, WNT5B and CTGF is determined in step a) of the method of the present invention and is thus used to classify the tumor. 10 15 20 25 30 35 15 According to another embodiment, the level of expression is also determined, i.e. in addition to any of the above combinations, for at least one of the genes EZH2, AMACR and MUC1 and is thus used for the classification.
Whether the expression level of one of said genes in a prostate cancer patient is high or low can be determined by comparing the gene expression level in a sample from the patient with a gene expression reference value for the same gene (s) of a reference patient, or group of reference patients known to have a high risk. - respectively low-risk tumor. If the expression level of the selected gene (s) in the patient sample is equal to or higher than the expression level of the same gene in a reference patient known to have a low-risk tumor, the patient's tumor may be classified as low risk. If the expression level of the selected gene (s) in the patient sample is equal to or lower than the expression level of the same gene in a reference patient known to have a high-risk tumor, the patient's tumor may be classified as high risk. When a group of reference patients is used for comparison, the mean or median expression level of the selected gene (s) in the group can be used as the gene expression reference value.
By matching the gene expression level of the selected gene with the reference gene expression of a reference patient is meant, when the gene expression level is determined for a gene, that when the expression level of the selected gene is equal to or higher than the reference gene expression of a reference patient known to have a low risk tumor. that reference gene expression. Similarly, when the expression level of the selected gene is equal to or lower than the reference gene expression of a reference patient known to have a high-risk tumor, the gene expression level matches that of the reference gene expression.
By matching the gene expression level of the selected gene with the reference expression of a reference patient is meant, when the gene expression level is determined for two or more genes, that the overall gene expression pattern of the two or more selected genes must match the overall reference gene expression pattern of the two or more reference patients. . Thus, the expression of both or all of the selected genes, as evaluated individually, need not completely match the reference gene expression of the selected genes individually.
Rather, a very high level of gene expression for one of the genes may compensate for a lower level of gene expression for the other gene (s), and the expression pattern would still be considered to match. By gene expression pattern is meant the gene expression level of the genes in a selection of two or more genes.
Matching of gene expression profiles obtained from the patient and the reference patient can be done, for example, by means of hierarchical clustering of gene expression data from both the patient and reference samples by methods known in the art (e.g. Eisen et al. Proc Natl Acad Sci USA 1998, 95214863-). 8). Clustering methods are suitable for evaluating trends in large amounts of data. Unsupervised cluster analysis such as hierarchical cluster analysis is used to advantage to detect groups or classes in data sets that would not be easily recognized by simply looking through the data. If the patient, whose tumor is to be classified, clustered or grouped together with reference patients who are known to have a low-risk tumor, the patient's tumor is also classified as a low-risk tumor. If the patient, whose tumor is to be classified, clustered or grouped together with reference patients who are known to have a high-risk tumor, the patient's tumor is also classified as a high-risk tumor.
By a high-risk tumor is meant that the tumor subtype, as determined by using a group of patients with known tumor subtype and known survival, is associated with a shorter overall and / or cancer-specific survival time than a low-risk tumor. For example, the subtype can be defined as a tumor subtype with certain clinical parameters or with some expression of certain genes. When determining whether there is a significant difference in survival time between patients with known subtypes and known survival times, one can use the calculation of risk ratio (Hazard ratio), which is well known in the art (Cox DR, J Royal Statist Soc B 1972, 342187-220) . One risk in a group is the rate at which events, such as deaths, occur. The risk in one group is assumed to be a constant proportion of the risk in the other group. This share is the risk ratio. Thus, if the risk ratio is, significantly, higher or lower than one, there is a higher risk in one group compared to the other.
The classification of the tumor may also include more classes than high risk and low risk, such as one or more intermediate risk group (s). The sample from the patient may be a tumor sample, such as a tumor sample obtained by fine needle aspiration (FNA), needle biopsy or by surgery. Alternatively, the sample may be a blood sample, plasma, serum, cerebrospinal fluid, urine, semen, exudate or a stool sample obtained from the patient. In particular, the gene expression level of IgFBP3 and F3 can be advantageously determined by analysis of a blood sample.
In one embodiment, the level of gene expression for the selected genes is determined by quantifying the amount of RNA or mRNA expressed from the genes.
The amount of RNA or mRNA can be determined, for example, using a method selected from microarray technology, Northern blotting and quantitative PCR (qPCR), such as quantitative real-time PCR (qrt-PCR), selectable multiplex PCR, or any other method of measurement known in the art. of gene expression.
For example, in the present study, the inventors have developed a simple multiplex quantitative PCR (qPCR) method to measure the expression level of a plurality of selected marker genes in fine needle aspiration (FNA) samples from the prostate. The developed method can also be used to measure expression levels in any tumor or blood sample taken from the patient.
An important technical advantage of this approach is that, although the marker genes of the present invention are identified by a stem cell approach and are thought to be important for a cancer stem cell function, there is no need to directly isolate the CSCs from the tumor samples. The simple and robust multiplex qPCR method established in the present study can be applied directly to fresh samples from routine needle biopsies or aspiration cytology to predict survival and effect of castration treatment at the time of diagnosis. All samples analyzed in the present study were fresh-frozen cytological cell smears that would ensure isolation of high quality pure cancer cell RNA molecules for qPCR analysis. In some cases, however, the isolation of RNA was not successful due to too few cells on the glass slides with FNA cytology smears. In future clinical applications, this problem can be easily solved by immediately using fresh cell suspensions from FNA or microdissected tumor samples from needle biopsies to isolate RNA. Since the marker genes of the present invention (F3, IGFBP3, VGLL3, c-MAF, WNT5B and CTGF) encode proteins, it is also possible to use immunohistochemistry and other protein analytical methods to measure their expression of protein as an estimate or a function of their gene expression.
Thus, according to an embodiment of the present invention, the level of gene expression can be indirectly determined by measuring the amount of protein encoded by said genes. The amount of protein can be determined, for example, by the use of methods such as EL1SA, R1A and mass spectrometry, as well as other methods of protein detection known in the art.
Those skilled in the art will appreciate that the utility of the present invention is not limited to the quantification of gene expression of any particular variant of the marker genes of the present invention. As non-limiting examples, the marker genes may have coding sequences and amino acid sequences as specified in Table 4.
Table 4 Gene / Encoding Protein Sequence Sequence Gene Full Name SEQ ID NO: SEQ ID NO: IGFBP3 insulin-like growth factor binding protein 3; 1 11 insulin-like growth factor binding protein 3 F3 coagulation factor III; coagulation factor lll 2 12 VGLL3 Rudimentary 3 (Drosophila); vestigial like 3 3 13 (Drosophila) c-MAF-a long form of v-maf musculoaponeurotic 4 14 fi brosarcoma oncogenic homolog (avian); long form of v-maf musculoaponeurotic fi brosarcoma oncogene homolog (avian) c-MAF-b short form of v-maf musculoaponeurotic 15 brosarcoma oncogenic homolog (avian); short form of v-maf musculoaponeurotic fi brosarcoma oncogene homolog (avian) WNT5B wingless-type MMTV integration site family, 6 16 member 5B; wingless-type MMTV integration site family, member 5B CTGF connective tissue growth factor; connective tissue growth 7 17 factor EZH2 enhancer by zestehomolog 2 (Drosophila); 8 18 enhancer of sixth homolog 2 (Drosophila) AMACR alpha-methylacyl-CoA racemas; alpha-methylacyl-9 19 CoA racemase MUC1 cell surface-associated mucin 1; mucin 1, cell surface 10 associated In some embodiments, those cDNA sequences or amino acid sequences that are at least 85% identical to or similar to the listed sequences, such as at least 90%, 91%, 92%, 93 %, 94%, 95%, 96%, 97%, 98% or at least 99% identical to or similar to the sequences listed in Table 4.
The term "% identity", as used in this specification, is calculated as follows. The search sequence is aligned with the target sequence by the CLUSTAL W algorithm (Thompson, J.D., Higgins, D.G. and Gibson, T.J., Nucleic Acid Research, 22: 4673-4680 (1994)). A comparison is made over a window that corresponds to the shortest of the fitted sequences. The shortest of the fitted sequences may in some cases be the target sequence. In other cases, the search sequence may be the shortest of the matched sequences.
The amino acid residues are compared in each position and the percentage of positions in the search sequence that have identical equivalents in the target sequence is indicated as% identity.
The term "% similarity" as used in this specification is calculated as follows: The alignment of the sequences and the sequence comparison are performed in the same manner as described for the calculation of% identity. However, "similarity" should be interpreted as follows. amino acid residues are considered similar if they belong to the same group of amino acid residues Non-limiting examples of amino acid residue groups are the hydrophobic group comprising the amino acid residues Ala, Val, Phe, Pro, Leu, Ile, Trp, Met and Cys; the basic group comprising the amino acid residues Lys , Arg and His; the acidic group comprising the amino acid residues Glu and Asp; the hydrophilic group comprising the uncharged amino acids Gln, Asn, Ser, Thr and Tyr; and the neutral group comprising the amino acid residue Gly. Thus, amino acid residues in each position and the percentage of positions in the search sequence that has similar equivalents in the target sequence is specified as% - similarity.
The method of the present invention for classifying a tumor in a patient having prostate cancer can have many benefits. For example, according to an embodiment of the invention, it may be used to predict the survival of said patient. For a patient with a tumor that is classified as a low-risk tumor, it is indicated that the patient has a good prognosis, while for a patient with a tumor that is classified as a high-risk tumor, it is indicated that the patient has a bad prognosis. 10 15 20 25 30 35 20 A poor prognosis for a patient may mean that the patient has a reduced probability of survival or a shorter survival time compared to a patient who has been predicted to have a good prognosis. A poor prognosis can also mean that the patient has an increased risk of recurrence or metastasis compared to a patient with a good prognosis. For example, the five-year survival of a patient with a low-risk tumor may be 90% or less, such as 85%, 80%, 75%, 70%, 60% or less, while the probability of five-year survival of a patient with a high-risk tumor may be 50% or lower, such as 45%, 40%, 30%, 20%, 10% or lower. The median length of survival in patients with low-risk tumors may also be 6 years or longer, such as 7 years, 8 years, 9 years, 10 years or longer, while the median length of survival in patients with high-risk tumors may be 5 years or less, such as 4 years, 3 years, 2 years, 1 year or shorter.
According to one embodiment of the present invention, the classification of a tumor can be used to improve the prediction of survival using clinical parameters. For example, the inventors have shown (Example 3C) that when subtype classification using Signature 1 (VGLL3, IgFBP3 and F3) is added to conventional prediction models which use only clinical parameters, the accuracy of the prediction significantly improves.
According to one aspect of the present invention, there is provided a method of making decisions regarding future treatment of the patient, the decision being dependent on the classification of the present invention. Patients who have a tumor that has been classified as a high-risk tumor need more radical or curative treatments than patients with low-risk tumors, and in addition at an earlier stage. Radical or curative treatment includes treatment regimens selected from prostatectomy, radiation, chemotherapy, castration or a combination thereof. Patients with tumors that have been classified as low-risk tumors need less or no radical or curative treatment at all, but may be prescribed attentive waiting / "wait-and-see" or active monitoring.
According to certain embodiments of the present invention, patients with limited tumor risk of high or intermediate risk tumor, need radical or curative treatments without delay, while patients with limited cancer of low risk tumor type can be safely assigned attentive waiting with minimal concern as castration treatment can still be a guarantee of long-term survival in case the disease progresses. For patients with cancer 10 15 20 25 30 35 21 who are advanced at diagnosis, those with a low-risk subtype may benefit most from castration treatment or anti-androgen therapy, while patients with a high-risk subtype or intermediate risk subtype may need to be treated early with chemotherapy or other new therapies.
In one aspect, the invention further provides a method of treating the patient who has been diagnosed with prostate cancer, and whose tumor has been classified according to the invention, in accordance with the treatment decision made as above.
In one aspect, the invention provides the use of any of the genes | GFBP3, F3, VGLL3, c-MAF, WNT5B and / or CTGF or of the proteins encoded thereby as a prognostic marker (s) for prostate cancer. In various embodiments of this aspect, the invention provides the use of a combination of two, three or more of the genes IGF BP3, F3, VGLL3, c-MAF, WNT5B and / or CTGF or the proteins encoded thereby as prognostic markers of prostate cancer. A particularly useful embodiment provides the use of a combination of the genes | GRBP3 and F3 and, optionally, any of VGLL3 and c-MAF, or the proteins encoded thereby as prognostic markers for prostate cancer.
In one aspect, the invention provides a solid support or kit for classifying a tumor in a patient diagnosed with prostate cancer, comprising nucleic acid probes (probes) or antibodies useful for determining gene expression and specific for a combination of at least two of the genes | GFBP3 , F3, VGLL3, c-MAF, WNT5B and CTGF. According to one embodiment thereof, said solid support or kit comprises nucleic acid probes or antibodies specific for | GFBP3 and F3. In another embodiment, the solid support or kit comprises nucleic acid probes or antibodies specific for | GFBP3 and F3 and either or both of VGLL3 and c-MAF. In a further embodiment, the solid support or kit further comprises nucleic acid probes or antibodies specific for EXH2, AMACR and MUC1.
The solid support may be an array, such as a cDNA microarray, a polynucleotide array or a protein array. For example, the nucleic acid probes for any of the kit embodiments may be selected only from the sequences described in Table 6. Such a kit is particularly useful for determining gene expression levels using multiplex PCR, e.g. multiplex quantitative PCR. The kit may also include additional reagents necessary for the measurement of gene expression level, such as secondary labeled probes or affinity ligands for the detection and / or quantification of bound or amplified nucleic acids or antibodies, depending on the method chosen. Such labels may also be directly bound to or linked to nucleic acid probes or antibodies.
The kit may further comprise various excipients to enable the kit to be used easily and efficiently, for example solvents, wash buffers and so on.
Furthermore, the kit may preferably also include reference samples or information on reference gene expression levels obtained from patients with known high-risk and low-risk tumors using the same method.
Table 6 SEQ SEQ SEQ ID NO Probe Sequence ID ID ID ID symbol (5'-3 ') NO: Sensing primer sequence (S'-3') NO: Antisense primer sequence (5'-3 ') NR: AMACR CTGCTGGAGCCCWCCGCCGC 21C 24 GGAAATGCTGCGAGGAGTGG 25 CGTGTCTTCCAGTCGGTAAGC 26 EZH 2 ACACGCTTCCGCCAACAAACTGGTCC 27 GCGGGACGAAGAATAATCATGG 28 TGTCTCAGTCGCATGTACTCTG 29 F3 ACAACAGACACAGAGTGTGACCTCACCGA 30 AGTCAGGAGATFGGAAAAGCAAATG 31 CCGTGCCAAGTACGTCTGC 32 IGFBPB ACCCAGAACTTCTCCTCCGAGTCCAAGC 33 GACTACGAGTCTCAGAGCACAG 34 CTCTACGGCAGGGACCATATTC 35 IGFBP3 ACAGATACCCAGAACTTCTCCTCCGAGTCCA 36 TACAAAGTTGACTACGAGTCTCAGAG 37 AGTGTGTCTTCCATTTCTCTACGG 38 c »MAF TFITCATAACTGAGCCCACTCGCAAGTTGG 39 AGCGACAACCCGTCCTCTC 40 GGCGTATCCCACTGATG GC 41 c ~ MAF CAATCCATGAGCCAGACACCCATFCCCT 42 TCGAGITFGTGGTGGTGGTG 43 CTAGCAAGTTATGGAGAATTTCAGATTG 44 c-MAF 'ITTTCATAACTGAGCCCACTCGCAAGTFGG 45 AGCGACAACCCGTCCTCTC 46 GGCGTATCCCACTGATGGC 47 c-MAF TTFTCATAACTGAGCCCACACCCGCGCGCGCGCGCGG TCCCACTGATGGC S0 MUC1 CCCCTCCCCACCCATTTCACCACCA 51 CGCCTGCCTGAATCTGTTCTG 52 CTGTAAGCACTGTGAGGAGCAG S3 VGLL3 AGACAGCTCAGCTCTCTCAAGCCAGC 54 AAAGCAAGATGGGGCTAACCC 55 TCCAAAAGGAAGTTGGGAAACTATTC 56 VGLL3 TGCTGTAGACCTGTATCGAATCCCACGC 57 TGGAGCCTITCATGGAACAGTAG 58 TACCACGGTGATTCCTTACTCTTG S9 VG LL3 CTGAATACCGCTAACTTCTTCTGCTGGCC 60 CCCCACAGCCTACTATCAGC 61 GACTTCCAGAGAGTCCTGCATC 62 VGLL3 AGACAGCTCAGCTCTCTCAAGCCAGC 63 AAAGCAAGATGGGGCTAACCC 64 GGTCCAAAAGGAAGTTGGGAAAC 65 WNTSB AGCCCTGCGACCGGCCTCGT 66 GGTGCTCATGAACCTGCAAAAC 67 AGGCTACGTCTGCCATCTTATAC 68 Beskrivninq of fiqurerna Figure 1 illustrates the approach to identifying key candidate ESCGPs in prostate cancer. A. Gradual identification of candidate ESCGPs for predicting prostate cancer prognosis. B. 19 highly ranked ESCGPs and control genes were selected according to 4 criteria as described in Example 2A. C.
The expression of these 24 genes in prostate cancer cell lines was verified by qPCR.
The gene expression pattern was visualized using Treeview software with gene median-centered delta-Ct values. The level of gene expression increased from light gray to black while the delta Ct values decreased from light gray to black. White represents missing data.
Figure 2 illustrates the expression of ESCGPs by RT-PCR in prostate cancer cell lines as described in Example 2B. The expression pattern of 34 ESCGPs and 5 control genes (c-MAF, AZGP1, AMACR, MUC1 and EZH2) was verified by RT-PCR in the three prostate cancer lines (LNCaP, DU145 and PC3) with 50ng cDNA as a template molecule for each reaction. G | yceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the internal charge control gene.
Figure 3 illustrates verification of the accuracy of 4-plex qPCR by comparison with single (single) qPCR. A series of cDNA dilution assays (the standard cDNA curve method) compared the results of single qPCR and 4-plex qPCR.
The optimized conditions for 4-plex qPCR were defined as those that gave the result that was most similar to the result of simple qPCR.
Figure 4A-k shows result tables from multivariate analysis performed to identify marker genes showing correlation with survival independent of all clinical parameters (see Example 3A).
Figure 5 illustrates tumor subtype classification for a training set of patients by ESCGP signature 1 and ESCGP signature 2. In the training set, 28 of 36 FNA samples had expression data for the four significant genes (F3, | GFBP3, VGLL3 and c-MAF-a). A series of cluster analyzes of different gene combinations showed that two gene combinations or signatures could similarly classify samples into three subtypes with high correlation with survival. The first (ESCGP signature 1) included F3, | GFBP3 and VGLL3 and the second (ESCGP signature 2) included F3, | GFBP3 and c-MAF-α. The gene expression level increases with decreasing ACt value.
Figure 6 illustrates differences in survival between tumor subtypes classified using ESCGP signature 1 (F3, | GFBP3 and VGLL3). A. FNA samples from 95 patients were classified into three tumor subtypes or groups (Group 1, Group 10, and Group 3) using ESCGP signature 1 (F3, IGFBP3 and VGLL3) as described in Example 3B. Each patient's clinical parameters are represented as shown by different boxes. Empty boxes represent longer survival, lower PSA level, limited clinical stage and high / medium differentiated tumor degree. Boxes with different fillings represent shorter survival, higher PSA level, advanced clinical stage, low differentiated tumor degree. The level of gene expression increases with decreasing ACt value.
B. Analysis of overall and cancer-specific survival in three subgroups is shown by Kaplan-Meier curves. C. Kaplan-Meier survival curves for patients with PSAs50ng / ml at diagnosis. D. Kaplan-Meier Survival Curves for Patients Aged S73 at Diagnosis. E and F are statistical box diagrams that show the difference in survival between the three subtypes or groups. The ends of the boxes are 25th and 75th quartiles and the line through the middle of the box shows the median value with 95% confidence interval (Kl). The p-values were calculated using t-tests and the p-values marked with an asterisk were of statistical significance.
Figure 7 illustrates the classification of tumor subtype in the entire patient group using ESGCP signature 2. The same 95 FNA samples were classified into three main tumor subtypes or groups (Group 1, Group 2 and Group 3) using signature 2 (F3, IGFBP 3 and c-MAF -a). The gene expression level increases with decreasing ACt value.
Figure 8 illustrates Kaplan-Meier patient-group survival curves defined by PSA, age, clinical stage, and tumor grade. A. 87 of the 95 patients in Figure 6 had serum PSA data at diagnosis and survival. The patients were divided into two groups, one with PSA> 50ng / ml and the other with PSAs50ng / ml. B. 92 of the 95 patients in Figure 6 had data regarding age at diagnosis and survival. The patients were divided into two groups, one with an age of 73 years and the other with an age> 73 years. C. 89 of the 95 patients in Figure 6 had clinical stage and survival data. Patients were divided into two groups based on clinical stage, one with limited stage (TsT2 and NO and MO and PSAs100ng / ml) and the other with advanced stage (T> T2 or N1 or M1 or PSA> 100 ng / ml). D. 92 of the 95 patients in Figure 6 had tumor grade and survival data. The patients were divided into two groups, one with low differentiated cancer and the other with high or medium differentiated cancer information. All p-values were calculated using the Log-rank test method.
Figure 9 illustrates Kaplan-Meier survival curves for the three tumor subtypes classified using ESCGP signature 1, in patients within the same group defined using clinical parameters. Of the 95 patients in Figure 6, 48 of the 95 patients showed PSAs5O ng / ml, 39 showed PSA> 50 ng / ml (B), 40 were of an age S73 (C), 52 were of an age> 73 (D) ), 38 had limited stage (E), 51 had advanced stage (F), 39 had high or moderate differentiated cancer (G) and 53 had low differentiated cancer (H). Patients in the group with the same clinical parameters could still be classified with ESCGP signature 1 (F3, IgFBP3 and VGLL3) in high-risk subtype (Group 1), intermediate-risk group (Group 2) and low-risk subtype (Group 3) with clearly different survival. The upper and lower part of each panel show general and cancer-specific survival, respectively. The log rank test was used to calculate the significance or p-value for the difference in survival between the subtypes or groups.
Figure 10 illustrates Kaplan-Meier survival curves for the three tumor subtypes classified using ESCGP signature 1 in patients primarily treated with castration treatment only. Of the 95 patients in Figure 6, 65 received castration treatment as the primary treatment. A clear difference in survival between the three tumor subtypes classified according to ESCGP signature 1 could be observed.
Figure 11 illustrates Kaplan-Meier survival curves for the three tumor subtypes classified using ESCGP signature 1 in patients primarily treated with castration therapy only and within the same group as defined by clinical parameters. Of the 95 patients in Figure 6, 65 received castration therapy as the primary treatment. Of these 65 patients, 29 showed PSAs50 ng / ml (A), 37 showed PSAs> 50 ng / ml (B), 24 were of an age S73 (C), 41 were of an age> 73 (D), 22 showed limited stage (E), 44 showed advanced stage (F), 26 showed high or medium differentiated cancer (G) and 39 showed low differentiated cancer (H). A clear difference in survival could still be observed between high-risk (Group 1) and low-risk subtype (Group 3) in patients within the same group of clinical parameters.
Figure 12 illustrates prediction of survival time using a parametric model. Prediction of survival time was modeled using the parametric model with the assumption of Weibull's distribution. A. General (left part) and cancer-specific (right part) survival was predicted using clinical parameters including PSA (> 50 ng / ml vs. S50 ng / ml), clinical stage (advanced vs. limited), tumor grade (low vs. high + medium differentiated) and age at diagnosis. B. General (left) and cancer-specific (right) survival were predicted using clinical parameters along with tumor subtypes or groups classified using ESCGP signature 1. The Y-axis represents the actual survival time while the X-axis represents the predicted survival time. Survival for 5 years and 8 years are marked in respective graphs for simplified interpretation. C. The table presents estimated improvement in survival prediction by adding parameters from the tumor subtype classification using ESCGP signature 1. D. The table represents contributions from ESCGP signature 1 and from clinical parameters for the prediction of general and cancer-specific survival, respectively.
Examples General methods Bioinformatics analysis Bioinformatics analysis for identification of embryonic stem cell gene predictors (ESCGPs) has been described previously (WO 2008/013492 A1). Briefly, the gene expression database was retrieved from previously published cDNA microarray from the Stanford Microarray Database (SMD, http://smd.stanford.edu/). The criteria used to collect data were as follows: Gene / spot selection: all genes or clones on arrays were selected, control points and blank points were not included.
Data collapse and retrieval: line data were retrieved and averaged with SUID; The UID column contains NAME. 10 15 20 25 30 27 Data collected: Log (base2) of R / G Normalized Ratio (Mean).
Selected data filters: Point is not flagged by the experimenter.
Data filter for the GENEPIX result set: Channel 1 Medium Intensity / Median Background Intensity> 1.5 AND normalized Channel 2 (Medium Intensityl Median Background Intensity)> 1.5.
To perform unattended hierarchical cluster analysis of mean linkage, the Cluster program (version 3.0) was used and the TreeView program was used to visualize the cluster result (Eisen et al, Proc Natl Acad Sci U S A 1998, 95214863-8). SAM (significant analysis of microarrays) was performed as previously described (Tusher et al, Proc Natl Acad Sci U S A 2001, 9825116-21).
Data Centering of Acquired CDNA Microarray Dataset: cDNA microarray data from 5 human ESC lines (Sperger et al, Proc Natl Acad Sci USA 2003, 100: 13350-5) and 115 human normal tissues from various organs (Shyamsundar et al, Genome Biol 2005, 6: R22) were obtained from SMD according to the parameters described above. The data set was divided into subgroups according to different array rounds. The genes were centered within each array round using the gene centering function of the Cluster program.
The subgroups were recombined and arrays were centered using the array centering function of the Cluster program. After centering, the amount of data was saved and converted to Excel format.
Prostate cancer cells / lines Three prostate cancer cell lines LNCaP, DU145 and PC3 were purchased from the American Type Culture Collection (ATCC). Cell culture was performed using medium and methods as instructed by the ATCC. LNCaP, DU145 and PC3 cells are maintained with lscove's Modified Dulbeccds Medium (IMDM, Cat. No. 21980-032, Invitrogen) supplemented with 10% Fetal Calf Serum (Cat. No. 10082-147, Invitrogen) and 50 units / ml and and 50 ug / ml Penicillin / Streptomycin (Cat No.15140-163, Invitrogen). 10 15 20 25 30 28 FNA samples FNA samples (fine needle aspiration) from the prostate were taken according to a routine approach for cytology diagnosis at the Department of Clinical Cytology and Pathology, Karolinska Hospital, Stockholm, Sweden. FNA samples were obtained from 241 patients at the time of diagnosis before any treatment. At least one fresh cytology smear from each patient was Giemsa stained for clinical cytology diagnosis.
Remaining duplicates of fresh smears were transferred to the freezer and kept fresh frozen at -80 ° C until the isolation of RNA samples. Most FNA cytology smears with prostate cancer diagnosis were estimated to contain more than 80% tumor cells due to the well-known selection effect that the aspiration sampling method can enrich cancer cells due to their reduced cell adhesion. Of the 241 patients, good quality RNA isolation from 193 patients was successful. Of these, 189 were diagnosed with prostate cancer while 4 did not have prostate cancer.
Clinical features of the group In total, fresh-frozen FNA samples from 189 prostate cancer patients were analyzed in the present study. These 189 patients were diagnosed between 1986 and 2001. All 189 patients had clinical symptoms that led to the diagnosis of prostate cancer. Under the supervision of an oncologist, a trainee physician collected relevant clinical data such as age at diagnosis, date of diagnosis, cytology and biopsy diagnosis, serum PSA at diagnosis, clinical stage, primary treatment, and so on. Table 5 shows details regarding clinical features of these 189 patients.
Data on the date of diagnosis, date of death and cause of death for all patients were first obtained from regional and national registers and then verified using original medical records. The date of censorship of data was 31 December 2008. At that time, 22 of the 189 patients were still alive, 163 had died and 4 had no data in the registers. Prostate cancer-specific death was defined as the primary or secondary cause of death being prostate cancer or metastases. Deaths from other causes were defined as the primary and secondary causes of death were not prostate cancer or 29 TABLE 5 Table 5. Clinical features of patients, features Exercise set Validation set 1 Validation set 2 Complete set Final aspiration fl FNA) ~ samples Reproduced FNA 65, 88 (Number (189) Median survival (Nlln-ltâax) is 7 65 (0 07-17 80) 4 00 (0 21-15 67) 4 32 (0 19-15 08) 4 32 (0 07-17 80) Prostate specific death, Number (9 / 0) '13 (361) 40 (61 6) 45 (51 1) 98 (51 8) Other death, Number (Wu) 19 (52 8) 21 (32 3) 25 (28 4) 65 (34 4) Living Number (%) 3 (B 3) 3 (4 ö) 16 (18 2) 221116) Missing, Number (%) 'l (2 8) 1 (1 5) 2 (2 3) 4 (2 1) Age ar 'll fl edelálder àr 70 4: 7 8 72 1:87 73 828.9 72 6: 8 7 Missing 1' l 2 4 PSA level (ng / lnl), Number (%) 1> 500 10135 7) 23143 4) 35143 8 ) 68 (42 2) S500 18 (64 3) 3066 6) 45156 3) 93 (57 8) Missing 8 12 8 28 Clinical Stage Number (170): Frarnskridet 13 (40 6) 321542) 53 (60 7) 96 ( 549) Limited 19 (59 4) 27 (45 8) 31 (39 3) 79 (45 1) Missing 4 ö 4 14 WHO Tilmorgrad, Antal (fl / n) § Lågt 14 (389) 31 (50 O) 54 (62 1) 99 (53 5) lvledelhögt I Högt 22 (61 l) 31 (500) 33 (37 9 ) 86 (46 5) Missing 0 3 1 4 Treatment, Number 1%) | Radical prostatectomy 1 (3 2) 31.5 0) 4 (4 9) 8 (4 7) Radiation 51161) 2103) 11 (130) | 8 (10 5) Hormone 'Ablatio testis 19 (61 .3) 531883) 62 (76 5) 134 (77 9) Never treated 6 (_19 4) 2 (3 3) 4 (4 9) 12 (7 0) Missing 5 5 7 17 metastases. These cases also included patients who died of diseases or conditions that could be exacerbated by prostate cancer or be related to side effects and complications of treatments.
All 189 patients showed clinical symptoms that led to digital examination of the rectum, PSA tests and subsequent FNA of the prostate. In advanced disease, castration treatment was the only primary treatment for most patients (77.9%). Isolation of RNA AllPrep DNA / RNA Mini Kit (Cat No. 80204, QIAGEN) was used to isolate total RNA from prostate cancer cell lines. RNAqueous® Micro Kit (Cat No. 1931, Ambion) for isolating total RNA less than 100 ng was used to isolate total RNA from fresh-frozen FNA samples from prostate cancer patients. Quantity and quality of RNA were checked using the Agilent RNA 6000 Nano Kit (Cat No. 5067-1511, Agilent) on a 2100 RNA Bioanalyzer (Agilent).
RNA samples with an RNA integrity number (RIN) greater than 7 were considered eligible. In the present study, qualified total RNA was isolated from 193 of the 241 FNA samples for further cDNA synthesis and quantitative PCR experiments (qPCR). RT-PCR For reverse transcription (RT) reactions, PCR synthesis was performed for PCR ( polymerase chain reaction) using the Cloned AMV First-Strand cDNA Synthesis Kit (Cat. No. 12328-032, Invitrogen) according to the manufacturer's instructions.
For reverse transcription (RT), a maximum of 2 pg of total RNA was used in a reaction volume of 20 μl. The expression pattern of 33 ESCGPs and 5 control genes was validated by RT-PCR in prostate cancer cell lines with gene-specific primer pairs (Figure 2). 50 ng of cDNA was used for each reaction and the experiment was repeated three times. Conventional methods were used to design primers and conditions for PCR cycling. 4-plex quantitative real-time PCR Synthesis of the first strand of quantitative PCR (qPCR) cDNA was performed using a QuantiTect® Reverse Transcription Kit (Cat. No. 205311, QIAGEN). For each qPCR, up to 1 μg of total RNA was used in a 20 μl reaction volume. The reaction was run on an ABI 7500 real time cycler which in real time could simultaneously monitor the density of four different fluorescent dyes (4-plex). No passive reference was chosen for this combination of four dyes. The conditions for 4-plex qPCR were 1 cycle at 50 ° C for 2 minutes; 1 cycle at 94 ° C for 10 minutes; 40 cycles at 94 ° C for 1 minute and 1 cycle at 60 ° C for 1.5 minutes. Fixed baseline start value 31 and final value were chosen for analysis of Ct values (Schmittgen and Livak, Nat Protoc 2008, 3: 1101-8; Wittwer et al, Methods 2001, 251430-42).
Optimization of 4-plex quantitative rea / time PCR A 4-plex qPCR contains four pairs of gene-specific primers and four gene-specific Taqman probes, each of which is double-labeled with a fluorophore at the 5 'end and a quencher at the 3' end. the end. In our study, Cy5, FAM, Texas Red and VlC were used to label the 5 'end while BHQ-3, BHQ-1, BHQ-2 and TAMRA were used as 3' quenchers. The four different combinations of fluorophore-quenching pairs enabled specific detection of PCR products from the 4 different genes. A total of 45 predicted 4-plex probes and 24 primer pairs were designed with Beacon Designer 7.0 software (Primer Biosoft) for the 19 ESCGPs and 5 control genes. Sequence information regarding probes and primers for the genes of the present invention is shown in Table 6.
To validate whether 4-plex qPCR has the same specificity and efficiency as single-probe qPCR, the standard cDNA curve method was used. Total RNA cDNA purified from LNCap, DU145 and PC3 cells was diluted to a concentration range of 10 pg, 100 pg, 1000 pg, 10000 pg, 100,000 pg and used as a template for both single probe qPCR and 4- plex-qPCR.
Standard curves are made based on the Ct value of each probe and the amount of cDNA.
The slope and r values of the standard cDNA curves derived from single-probe qPCR and 4-plex qPCR for the same gene were compared. Optimization of concentrations of probes and primer pairs was performed until there was no significant difference in these values between single-probe and 4-plex qPCR.
These results show that 0.2 μM probes and 0.2 μM primer pairs were the best concentrations for 4-plex qPCR. Validation of the results for 4-plex-qPCR is presented in Figure 3.
Normalization and centering of Ct value from results of qPCR Ct (cycle threshold) is a measure of the number of PCR cycles (in real-time PCR) required to obtain a fluorescent signal or sufficient PCR product. In the present study, the Ct value of a gene in a sample was generated after real-time PCR using 7500 software (version 2.0.5, ABI). To normalize the Ct values for each gene, the delta Ct value was calculated according to an equation ACt = CtgenX-CtGApDH where Ctgenx was the Ct value of the gene being analyzed and CtGApDH was the Ct value of the base gene GAPDH (glyceraldehyde-3-phosphate dehydrogenase) (Schmittgen and Livak, Nat Protoc 2008, 3: 1101-8; Wittwer et al, Methods 2001, 251430-42). Thus, the expression level of each gene in the sample was normalized with the expression level of GAPDH. The ACt value was inversely correlated with the gene expression level. Each panel of 4-plex-qPCT contains a specific GAPDH probe, respectively. Samples with weak signals were excluded from analysis (Ct value in GAPDH> 28). The Ct value was set to 40 (set as the maximum Ct value) for samples with weak signals for genes to be analyzed. The delta Ct value for genes in all samples was centered using the gene median centering function of a Cluster program (version 3.0) (Eisen et al, Proc Natl Acad Sci U S A 1998, 95114863-8). The centered deIta-Ct value was used for statistical analyzes.
Statistical analysis of survival correlation Overall survival and prostate cancer-specific survival were used as respective endpoints in survival analysis for the correlation with molecular and clinical parameters. Survival time was defined as the time from the date of diagnosis to the date of death and was used as a continuous variable. For simplified interpretation, long, intermediate and short survival times were defined as> 8, 5-8 and <5 years, respectively. For patients primarily treated with castration treatment only, the lead time before treatment was defined as the time from the date of diagnosis to the date the castration treatment was started and used as a continuous variable. The centered delta Ct value for each gene, age at diagnosis and serum level of PSA at diagnosis were used as continuous variables. With unsupervised hierarchical cluster analysis, samples were classified into three groups or subtypes and the grouping was used as a non-continuous variable. PSA was also analyzed as a non-continuous variable with two categories S50 ng / ml or> 50 ng / ml. The WHO tumor grade was integrated into two categories: high-medium highly differentiated or low differentiated. The clinical stage was integrated into two categories: advanced (any of T2T3 or N1 or M1 or PSA2100 ng / ml) or limited (T multivariate analysis of proportional hazard ratio according to Cox and Cox regression was performed with Stata statistical software ( Version 10.1, StataCorp LP) Kaplan-Meyer analysis as well as statistical block diagrams were performed with JMP® statistical software (version 8.0.1, SAS Institute lnc.).
Study design The study was performed in three steps: 1) identification of an embryonic stem cell gene predictor signature (ESCGP signature) with 641 genes. 2) selection of a subset of important candidate genes from the ESCGP signature for classification of prostate cancer subtypes and optimization of multiplex qPCR in prostate cancer cell lines. 3) verification of the clinical significance by measuring expression levels of these selected genes in FNA samples from prostate cancer patients with 7-20 year survival data.
This resulted in the identification of a subset of gene markers that show a significant correlation with either overall or cancer-specific survival.
Example 1. Identification of an ESCGP signature An ESCGP signature for the classification of different types of cancer was identified as described in patent document WO 2008/013492 A1. Briefly, previously published datasets of microarray data for whole genome cDNA, including from 5 human ESC lines and 115 human normal tissues from various organs, were obtained from the Stanford Microarray Database (SMD) according to the parameters described above. Data centering of the acquired data sets was also performed as described above. Data sets from normal tissue were used to facilitate data centering. After centering, the sub-dataset for ESC lines was isolated from the entire dataset. A single-class SAM was performed with only this ESC line data set, through which all genes were ranked according to how consistent their expression levels were across the 5 ESC lines. Using a q-value of 50.05 as the limit value, the assay identified 328 genes with consistently high and 313 with consistently low expression levels in the ESCs. The 641 genes were named embryonic stem cell gene predictors (ESCGPs). Example 2A: Selection of important ESCGP candidates in prostate cancer.
From the list of 614 ESCGPs, a subset of 33 ESCGPs as well as 5 control genes were selected as candidates that could enable the classification of prostate cancer using fewer ESCGPs. The candidates were selected according to four criteria (see Figure 1B); i) position in the ranking list of 641 ESCGP (referred to as the "ESCGP list" in Figure S1 B); ii) position in the ranking list identified by Lapointe et al (Proc Natl Acad Sci U S A 2004, 101: 81 1-816) comprising significant genes for classification of subtypes of prostate cancer (designated “PCa vs. PCa” in Figure 1B); iii) position in the ranking in the gene list identified by Lapointe et al (Proc Natl Acad Sci U S A 2004, 1012811-816) including significant genes that differentiate between prostate cancer and normal tissue (designated “Normal vs. PCa” in Figure 1B); and iv) genes from previous important publications (Lapointe et al, Proc Natl Acad Sci U S A 2004, 101181 1-816; Varambally et al, Nature 2002, 4191624-629; Rubin et al, JAMA 2002, 28721662-70). In Figure 1B, the genes were marked with “1” if present and “0” if not present in the respective gene lists. Thus, some genes met all four criteria, while other genes met 1-3 of the four criteria. AZGP1, c-MAF, AMACR, MUC1 and EZH were not identified in the list of ESCGPs but were included as important control genes as they were found to be important in prostate cancer in previous studies. A few genes such as c-MAF have different RNA transcripts ((http: //www.ncbi.n | m.nih.gov/gene/4094).
Primers and probes were designed to target these respective different RNA transcripts. Example 2B: verification of expression of the selected genes in prostate cancer cell lines.
Expression of the 33 selected ESCGPs and 5 control genes in three different prostate cancer cell lines was validated by RT-PCR using gene-specific primer pairs (see Figure 2). The cell lines used for analysis were LNCap, the bay is derived from a less aggressive cancer, and DU145 and PC3, both derived from aggressive cancers. Of the 38 genes analyzed, 14 had similar expression in all three cell lines and were considered less likely to be valuable for tumor classification. The remaining 24 genes had different expression patterns in the less aggressive cell line LNCap and the aggressive cell lines DU145 and PC3, and were thus considered more likely to be useful for tumor classification to distinguish between less aggressive and more aggressive cancers. Thus, a total of 24 genes (25 gene markers) were selected for optimization of multiplex qPCR and evaluation of ability to classify prostate cancer.
Example 3A: Focused gene expression profiling of FNA samples from prostate cancer and identification of significant ESCGPs that correlate with survival.
Expression of the 24 genes (25 gene markers) was analyzed in fine needle aspiration (FNA) samples from 189 prostate cancer patients using multiplex qPCR and then analyzed for correlation with survival data. Clinical features of the patient group as well as the statistical analysis are described above.
Not all candidate genes could be assayed in each FNA sample due to insufficient total RNA in most FNA samples. To compromise this limitation, the group of 189 patients was divided into three parts according to the chronology of the experiment. The three sections contained samples from 36, 65 and 88 patients, respectively (Table 5). Only genes that showed significant correlation with survival in the first subset were included, along with new candidate genes, in subsequent subsets. Survival analysis was performed in each and every one of the three subsets as well as in the final complete group (Table 1, Figure 5-7). This compromised screening process ensured the detection of the most significant gene markers but may miss a few gene markers with more modest significance.
Analysis of correlation with survival was performed both for known clinical parameters in the patients and for gene expression of the selected candidate genes. In univariate analysis, all clinical parameters showed significant correlation with both overall and cancer-specific survival (Table 1). Ten of the 25 gene markers, F3 (coagulation factor lll; coagulation factor lll), WNT5B (wingless-type MMTV integration site family, member 5B; wingles-type MMTV integration site family, member 5B), VGLL3 (rudimentary like 3 (Drosophila); vestigial like 3 ( Drosophila)), CTGF (connective tissue growth factor), IGF BP3 (insulin-like growth factor binding protein 3; insulin-like growth factor binding protein 3), c-MAF-a (long form of v-maf musculoaponeurotic fibrosarcoma oncogen homolog (avian) ; long form of v-maf musculoaponeurotic fibrosarcoma oncogene homolog (avian)), c-MAF-b (short form of v-maf musculoaponeurotic fibrosarcoma oncogene homolog (avian); short form of v-maf musculoaponeurotic fibrosarcoma oncogene homolog, avian) AMACR (alpha-methylacyl-CoA racemas; alpha-methylacyl-CoA racemase), MUC1 (cell surface-associated mucin 1; mucin 1, cell surface associated) and EZH2 (enhancer of zeste homolog 2 (Drosophila); enhancer of zeste homolog 2 (Drosophila)) up showed significant correlation with either overall and / or cancer-specific survival (Table 1). A p-value <0.05 is considered significant throughout the study. The expression levels (inversely correlated to the delta Ct value) for all significant genes except EZH2 showed a positive correlation with survival time (value <1 in Table 1).
Each of the 10 gene markers with significant correlation with survival according to univariate analysis was analyzed together with clinical parameters that include age at diagnosis, two-category PSA, tumor grade and clinical stage with multivariate analysis (Figure 4A-K). Multivariate analysis indicates how much the significance of the gene variable is affected by clinical parameters. The number of patients included in the multivariate assay was smaller than that in the univariate assay due to lack of data for various parameters.
In summary, 4 genes (F3, IgFBP3, CTGF and AMACR) showed correlation with both overall and cancer-specific survival independent of all clinical parameters. All four genes except AMACR were from the list of ESCGPs. Two genes (VVNT5B and EZH2) showed independent correlation with cancer-specific survival and one gene (VGLL3) showed independent correlation with overall survival.
Example 3B: Identification of significant ESCGP signatures that correlate with survival.
To study possible additive effects or synergy effects of several genes in the prediction of survival, the inventors tested different combinations of the ten significant genes in a series of unsupervised hierarchical cluster analyzes, using data from patients in the first set (training set). Two signatures could similarly classify tumors into three subgroups or subtypes with significant difference in overall and cancer-specific survival (Figure 5). The first ESCGP signature (Signature 1) includes the marker genes VGLL3, IgFBP3 and F3. The second ESCGP signature (Signature 2) includes the marker genes c-MAF-α, IgFBP3 and F3. The classification of tumor subtype using the respective signatures was confirmed using data from patients in the whole group (Figures 6 and 7).
ESCGP Signature 1 (VGLL3, IgFBP3 and F3) showed better results than ESCGP Signature 2 (c-MAF-a, IgGBP3 and F3) (Tables 2 and 3). Of the 189 patients, 87 had data for both all clinical parameters and for subtype classification according to Signature 1. Multivariate analysis for general and cancer-specific survival showed that subtype classification with Signature 1 was the most significant parameter and independent of age, PSA level, tumor grade and clinical stage (Table 2).
The median overall survival was 2.60 years for the high-risk subtype, 3.85 years for the intermediate risk subtype and 7.98 years for the low-risk subtype (Figure 6E), corresponding to a risk ratio of 5.86 (95% at 2.91-11.78 , P <0.001) for the high-risk and 10 15 20 25 30 38 3.45 (95% Kl 1.79-6.66, P <0.001) for the intermediate subtype over the low-risk subtype (Table 3). The difference in overall survival was attributed to both cancer-specific and non-cancer-specific survival (Figure 6E).
Interestingly, the median survival time for non-specific deaths was 3.54 years for high-risk, 3.70 years for intermediate risk and 7.98 years for the low-risk subtype. Within five years of diagnosis, the proportion of deaths not directly caused by prostate cancer was only 4/31 (12.9%) in the low-risk subtype compared with 9/31 (29%) in the high-risk subtype and 9/32 (28%) in the intermediate risk subtype, respectively. . Of the three cases with the shortest survival time in the low-risk subtype (symbolized dots), PC39 and PC140 were never treated after prostate cancer diagnosis and died of other diseases, and PC234 was diagnosed at age 81, treated only with castration treatment, and died of prostate cancer.
Furthermore, Kaplan-Meier curves showed clear survival difference between the three subtypes classified with the tumor ESCGP signature 1. Overall survival of high-risk subtype (Group 1), intermediate risk subtype (Group 2) and low-risk subtype (Group 3) was 20%, 40% and 80, respectively. % at 5 years and 10.3%, 25.0% and 64.4% at 8 years, respectively. The survival difference between the high-risk and low-risk subtypes was much more impressive than the results with any clinical parameters and was observed within each patient group or became even clearer within the same patient group defined by PSA, clinical stage, tumor grade or age (Figure 6C-D).
For example, 48 of 92 patients had serum levels of PSA s 50 ng / ml at diagnosis. Of these 48 patients, overall survival at 8 years was 21.4% for the high-risk subtype, 47.1% for the intermediate risk subtype and 76.5% for the low-risk subtype. Most impressive were 40 of 92 patients at the age of S73. Of these 40 young patients, the overall survival at 8 years was 7.1% for the high-risk subtype, 44.4% for the intermediate risk subtype and 88.2% for the low-risk subtype. In addition, the difference in survival between the classified groups of patient groups treated only with castration treatment was also seen (Figures 6-11). Example 3C: Improved survival prediction by adding the ESCGP Signature to clinical parameters.
Parametric model was used for survival prediction to estimate how much subtype classification with the signature with VGLL3, IGFBP3 and F3 (Signature 1) could improve prediction using all clinical parameters (Figure 12). Compared to the prediction model that used only clinical parameters, the addition of subtype classification with Signature 1 improves the accuracy of the prediction of overall survival from 70.1% up to 78.2% and for cancer-specific survival from 65.5% to 71.3% at 5 years (Figure 12C).
Based on the Cox regression analysis, the nested likelihood ratio test showed that the subtype classification with Signature 1 significantly contributes to the improvement of the regression rate in a multivariate model together with clinical parameters (Figure 12D).

权利要求:
Claims (25)
[1] 1. ) A method of classifying a prostate cancer in a subject, comprising: a) determining a gene expression level of the genes F3 and |GFBP3 in asample from the subject; b) classifying the tumor by comparing the gene expression level determinedin a) with a reference gene expression of the same genes in referencepatients known to have a high risk or low risk tumor respectively; and c) concluding that if the gene expression level determined in a) matches thereference gene expression of the reference patients with a high risktumor, the tumor in the subject is a high risk tumor, and that if the geneexpression level determined in a) matches the reference gene expressionof the reference patients with a low risk tumor, the tumor in the subject isa low risk tumor.
[2] 2. ) The method of claim 1, wherein a) further comprises determining a geneexpression level of one or more of the genes VGLL3 and c-MAF, preferablyVGLL3.
[3] 3. ) The method of claim 1-2, wherein a) further comprises determining a geneexpression level for one or more of the genes WNT5B and CTGF, EZH2,AMACR and MUC1.
[4] 4. ) The method of any of claims 1-3, wherein the gene expression level isdetermined by quantifying the amount of RNA or mRNA expressed from saidgenes.
[5] 5. ) The method of claim 4, wherein said amount of RNA or mRNA is determinedby use of a method selected from microarray technology, Northern blottingand quantitative PCR (qPCR), such as real time quantitative PCR (qrt-PCR), optionally multiplex PCR.
[6] 6. ) The method of any of claims 1-3, wherein the gene expression level isdetermined by quantifying the amount of protein encoded by said genes.
[7] 7. ) The method of claim 6, wherein said amount of protein is determined by useof a method selected from immunohistochemistry, Western blotting, ELISA,RIA and mass spectrometry.
[8] 8. ) The method of any of claims 1-7, wherein the sample is a tumor sampleobtained from the subject.
[9] 9. ) The method of any of claims 1-7, wherein the sample is a blood sampleobtained from the subject.
[10] 10. )A method for determining prognosis of a subject diagnosed with prostatecancer and having a tumor, said method comprising a) classifying the tumor using the method according to any of claims 1-9;and b) concluding that a low risk tumor indicates that the subject has a goodprognosis and a high risk tumor indicates that the subject has a poorprognosis.
[11] 11. )The method of claim 10, wherein the poor prognosis is a decrease in thelikelihood of survival compared to the good prognosis.
[12] 12. )The method of claim 10, wherein the poor prognosis is a decrease in time ofoverall survival compared to the good prognosis. 41
[13] 13. )The method of claim 10, wherein the poor prognosis implicates a decrease intime of cancer specific survival compared to the good prognosis.
[14] 14. )The method of claim 10, wherein the poor prognosis implicates an increasedrisk of mortality compared to the good prognosis.
[15] 15. )A method of making a treatment decision for a subject diagnosed withprostate cancer and having a tumor, said method comprising a) c|assifying the tumor using the method according to any of c|aims 1-9;and b) making a treatment decision for said subject dependent on theclassification obtained in a).
[16] 16. )The method of claim 15, wherein said treatment decision is to give radica| orcurative treatment to a patient with |oca|ized cancer and a high risk tumor.
[17] 17. )The method of claim 15, wherein said treatment decision is to awaittreatment or to give hormone therapy, such as castration therapy or anti-androgen therapy, to a patient with |oca|ized cancer and a low risk tumor.
[18] 18. )The method of claim 15, wherein said treatment decision is to givechemotherapy to a patient with advanced cancer and a high risk tumor.
[19] 19. )The method of claim 15, wherein said treatment decision is to give castrationtherapy to a patient with advanced cancer and a low risk tumor.
[20] 20. )A method of treating a subject diagnosed with prostate cancer and having a tumor, said method comprising 42 a) making a treatment decision using the method according to any of claims15-19; and b) treating said subject in accordance with said treatment decision.
[21] 21. )Use of the genes F3 and |GFBP3 and, optionally, either of VGLL3 and c-MAF, or the proteins encoded therefrom as prognostic markers for prostatecancen
[22] 22. )A solid support or a kit for classifying a tumor in a subject diagnosed withprostate cancer, comprising nucleic acid probes or antibodies that are usefulfor determining gene expression, such as RNA or protein expression, of thegenes F3 and |GFBP3 and, optionally, either of VGLL3 and c-MAF.
[23] 23. )The solid support or kit of claim 22, further comprising nucleic acid probes orantibodies that are useful for determining gene expression of one or more ofthe genes WNT5B, CTGF, EZH2, AMACR and MUC1.
[24] 24. )The solid support of any of claims 22-23, wherein the solid support is anarray such as a microarray or protein array.
[25] 25. )The kit of any of claims 22-24, wherein the nucleic acid probes are selectedfrom SEQ ID NOs: 21-68.

类似技术:

---

公开号 | 公开日 | 专利标题

---

JP2019013255A|2019-01-31|Gene expression profile algorithm and test for determining prognosis of prostate cancer

---

US20210017606A1|2021-01-21|Marker Genes for Prostate Cancer Classification

---

KR20140024907A|2014-03-03|Biomarkers for lung cancer

---

JP2008521412A|2008-06-26|Lung cancer prognosis judging means

---

JP2007049991A|2007-03-01|Prediction of recurrence of breast cancer in bone

---

TW200526958A|2005-08-16|Materials and methods relating to breast cancer classification

---

WO2012158780A2|2012-11-22|Lung cancer signature

---

CN102333887B|2014-04-02|Method for detecting metastasis of gi cancer

---

WO2015154715A1|2015-10-15|A method of diagnosis, prognosis or treatment of a cancer

---

KR20070115891A|2007-12-06|Pharmacogenomic markers for prognosis of solid tumors

---

Kuang et al.2019|tRNA‐based prognostic score in predicting survival outcomes of lung adenocarcinomas

---

Guo et al.2011|Screening and identification of specific markers for bladder transitional cell carcinoma from urine urothelial cells with suppressive subtractive hybridization and cDNA microarray

---

US20150024956A1|2015-01-22|Methods for diagnosis and/or prognosis of gynecological cancer

---

US8911940B2|2014-12-16|Methods of assessing a risk of cancer progression

---

US9195796B2|2015-11-24|Malignancy-risk signature from histologically normal breast tissue

---

WO2012125411A1|2012-09-20|Methods of predicting prognosis in cancer

---

US20080119367A1|2008-05-22|Prognosis of Renal Cell Carcinoma

---

US20150011411A1|2015-01-08|Biomarkers of cancer

---

CA2844132A1|2013-02-07|Hypoxia-related gene signatures for cancer classification

---

US20200123619A1|2020-04-23|Mirna signature expression in cancer

---

WO2021047992A1|2021-03-18|Superior biomarker signature to predict the response of a breast cancer patient to chemotherapy

---

CN113817834A|2021-12-21|Genetic factor-based lung cancer prognosis staging detection reagent, system and application

---

CN113930511A|2022-01-14|Application of GOLT1B in prognosis of breast cancer

---

EP3752837A1|2020-12-23|Patient classification and prognostic method

同族专利:

---

公开号 | 公开日

---

CN104024436A|2014-09-03|

---

JP2015501151A|2015-01-15|

---

ES2689547T3|2018-11-14|

---

US9790555B2|2017-10-17|

---

CN104024436B|2016-10-19|

---

US20180080088A1|2018-03-22|

---

SE536352C2|2013-09-03|

---

CA2852020C|2021-06-08|

---

EP2771481A1|2014-09-03|

---

EP2771481B1|2018-07-04|

---

JP6049739B2|2016-12-21|

---

US20140243433A1|2014-08-28|

---

DK2771481T3|2018-10-15|

---

CA2852020A1|2013-05-02|

---

US20210017606A1|2021-01-21|

---

WO2013060739A1|2013-05-02|

引用文献:

---

公开号 | 申请日 | 公开日 | 申请人 | 专利标题

---

---

US6582908B2|1990-12-06|2003-06-24|Affymetrix, Inc.|Oligonucleotides|

---

US7001735B2|2000-03-01|2006-02-21|Albert Einstein College Of Medicine Of Yeshiva University|Glucose transporter/sensor protein and uses thereof|

---

AU5005101A|2000-03-23|2001-10-03|Beth Israel Hospital|Use of igf-i, igfbp-3 and psa to predict the risk of prostate cancer|

---

US20030054419A1|2001-02-07|2003-03-20|Slawin Kevin M.|Method to determine prognosis after therapy for prostate cancer|

---

US7935679B2|2001-11-07|2011-05-03|Board Of Trustees Of The University Of Arkansas|Gene expression profiling based identification of CKS1B as a potential therapeutic target in multiple myeloma|

---

US20070099209A1|2005-06-13|2007-05-03|The Regents Of The University Of Michigan|Compositions and methods for treating and diagnosing cancer|

---

WO2008013492A1|2006-07-28|2008-01-31|Chundsell Medicals Ab|Embryonic stem cell markers for cancer diagnosis and prognosis|

---

US20100056445A1|2006-11-07|2010-03-04|Thomas Jefferson University|Adiponectin for the treatment and diagnosis of albuminuria|

---

US9353415B2|2006-12-19|2016-05-31|Thomson Reuters Llc|Methods for functional analysis of high-throughput experimental data and gene groups identified therefrom|

---

AU2007350331A1|2007-03-30|2008-10-09|Source Precision Medicine, Inc. D/B/A Source Mdx|Gene expression profiling for identification, monitoring, and treatment of prostate cancer|

---

WO2009021338A1|2007-08-15|2009-02-19|UNIVERSITé DE SHERBROOKE|Alternative splicing gene variants in cancer detection|

---

WO2009105154A2|2008-02-19|2009-08-27|The Jackson Laboratory|Diagnostic and prognostic methods for cancer|

---

CN104031984A|2008-02-28|2014-09-10|俄亥俄州立大学研究基金会|Microrna-based methods and compositions for the diagnosis, prognosis and treatment of prostate related disorders|

---

US20110136683A1|2008-05-28|2011-06-09|Genomedx Biosciences, Inc.|Systems and Methods for Expression-Based Discrimination of Distinct Clinical Disease States in Prostate Cancer|

---

US20090298082A1|2008-05-30|2009-12-03|Klee George G|Biomarker panels for predicting prostate cancer outcomes|

---

US20120009581A1|2008-07-08|2012-01-12|Bankaitis-Davis Danute M|Gene Expression Profiling for Predicting the Survivability of Prostate Cancer Subjects|

---

US20120156216A1|2009-03-02|2012-06-21|Youngman Oh|Methods and Compositions for Treatment of Tumor Metastasis|

---

US20110166030A1|2009-09-30|2011-07-07|Yixin Wang|Prediction of response to docetaxel therapy based on the presence of TMPRSSG2:ERG fusion in circulating tumor cells|CN110527728A|2013-08-08|2019-12-03|纽约州州立大学研究基金会|The keratin of biomarker as cervix cancer and survival period|

---

WO2018222217A1|2017-05-29|2018-12-06|Epelbaum Michael|Lifespan in regression-based estimation of phenomenon-specific probabilities of mortality or survivorship|

---

CN106796236A|2013-12-30|2017-05-31|促进军事医学的亨利·M·杰克逊基金会公司|Prostate cancer gene profile and the method using the gene profile|

---

EP3037545A1|2014-12-23|2016-06-29|The Provost, Fellows, Foundation Scholars, & the other members of Board, of the College of the Holy & Undiv. Trinity of Queen Elizabeth near Dublin|A DNA-methylation test for prostate cancer|

---

EP3359692A4|2015-10-05|2019-05-01|Cedars-Sinai Medical Center|Method of classifying and diagnosing cancer|

---

CN107362353A|2016-05-11|2017-11-21|上海长海医院|A kind of related drug target of prostate cancer of castration resistance and its application|

---

CN106399509A|2016-09-23|2017-02-15|北京致成生物医学科技有限公司|Product for diagnosing prostate cancer and use thereof|

---

CN107460238A|2017-07-07|2017-12-12|沈阳宁沪科技有限公司|A kind of noninvasive high flux methylates prostate cancer diagnosis, research and treatment method|

---

CN108647493B|2018-05-09|2022-01-18|中国科学院昆明动物研究所|Individualized prognosis evaluation method for renal clear cell carcinoma|

---

EP3730941A1|2019-04-23|2020-10-28|Institut Jean Paoli & Irène Calmettes|Method for determining a reference tumor aggressiveness molecular gradient for a pancreatic ductal adenocarcinoma|

法律状态:

优先权:

---

申请号 | 申请日 | 专利标题

---

SE1150982A|SE536352C2|2011-10-24|2011-10-24|Cursor genes for classification of prostate cancer|SE1150982A| SE536352C2|2011-10-24|2011-10-24|Cursor genes for classification of prostate cancer|

---

ES12775526.2T| ES2689547T3|2011-10-24|2012-10-24|Marker genes for prostate cancer classification|

---

CN201280064273.1A| CN104024436B|2011-10-24|2012-10-24|Marker gene for carcinoma of prostate classification|

---

CA2852020A| CA2852020C|2011-10-24|2012-10-24|Marker genes for prostate cancer classification|

---

EP12775526.2A| EP2771481B1|2011-10-24|2012-10-24|Marker genes for prostate cancer classification|

---

US14/352,107| US9790555B2|2011-10-24|2012-10-24|Marker genes for prostate cancer classification|

---

DK12775526.2T| DK2771481T3|2011-10-24|2012-10-24|MARKET GENERATIONS FOR CLASSIFICATION OF PROSTATACANCES|

---

JP2014536297A| JP6049739B2|2011-10-24|2012-10-24|Marker genes for classification of prostate cancer|

---

PCT/EP2012/071077| WO2013060739A1|2011-10-24|2012-10-24|Marker genes for prostate cancer classification|

---

US15/725,678| US20180080088A1|2011-10-24|2017-10-05|Marker genes for prostate cancer classification|

---

US16/904,789| US20210017606A1|2011-10-24|2020-06-18|Marker Genes for Prostate Cancer Classification|