Method of treating a patient having renal cancer
11015190 · 2021-05-25
Assignee
Inventors
- Min-Han Tan (Singapore, SG)
- Yukti Choudhury (Singapore, SG)
- Puay Hoon Tan (Singapore, SG)
- Bin Tean Teh (Singapore, SG)
Cpc classification
C12Q2600/106
CHEMISTRY; METALLURGY
C12Q2600/166
CHEMISTRY; METALLURGY
C12N15/1072
CHEMISTRY; METALLURGY
International classification
Abstract
A method of making a prognosis as to whether a patient having renal cancer is likely to survive in a tumour tissue sample obtained from the patient is provided. The method comprising determining the level of expression for each marker of a panel of markers comprising at least one housekeeping gene selected from the group consisting of ACTB, RPL13A, RPL9, and RPS29 and any combinations thereof and at least one prognostic gene selected from the group consisting of CXCL5, EFNA5, EMCN, G6PC, GFPT2, HIST2H3C, IGFBP1, LAMB3, MMP9, MOCOS, PLG, PRAME, RARRES1, SDPR, SLC6A19, TK1, KDELR3 and TSPAN7 and any combinations thereof, comparing the level of expression of each marker with a predetermined reference level associated with each marker, and determining the differential expression of each marker in the tumour tissue sample based on the expression parameter for each marker to provide a prognosis for renal cancer.
Claims
1. A method of treating a patient having renal cancer comprising the steps of: a) obtaining a tumor tissue sample from the patient, b) detecting the level of expression for each marker of a panel of markers, wherein the panel comprises at least one housekeeping gene selected from the group consisting of ACTB, RPL13A, RPL9, and RPS29, or any combinations thereof; and prognostic genes CXCL5, EFNA5, EMCN, LAMB3, PLG, PRAME, RARRES1, and SLC6A19; c) comparing the level of expression of each marker with a predetermined reference level associated with each marker to obtain an expression parameter, wherein the predetermined reference level is based on the expression of the at least one housekeeping gene, and wherein said comparison comprises normalizing the expression level of each marker with the predetermined reference level; and d) determining the differential expression of each marker in the tumor tissue sample based on the expression parameter for each marker, e) administering an effective amount of Tyrosine Kinase Inhibitor (TKI) therapy if the patient exhibits at least a doubling of expression level for EMCN, PLG and SLC6A19 markers, or at least a halving of expression level for CXCL5, EFNA5, LAMB3, PRAME and RARRES1 markers, or administering a different anti-renal cancer treatment if the patient does not exhibit at least a doubling of expression level for EMCN, PLG and SLC6A19 markers, or does not exhibit at least a halving of expression level for CXCL5, EFNA5, LAMB3, PRAME and RARRES1 markers, wherein the renal cancer is clear cell renal cell carcinoma (ccRCC) or a mixture of renal tumors comprising ccRCC, and the patient suffers from ccRCC, or the patient suffers from ccRCC and undergoes anti-cancer treatment.
2. The method according to claim 1, wherein the tumor tissue sample is obtained from tissue selected from the group consisting of frozen tissue, tissue biopsies, circulating tumor cells, and bodily fluids selected from the group consisting of ascites, effusions, cerebrospinal and urine.
3. The method according to claim 1, wherein the patient having renal cancer is a patient that has previously received and/or is currently undergoing-anti-cancer treatment; wherein the anti-cancer treatment is selected from the group consisting of a chemotherapeutic treatment, a surgical treatment, a treatment with radiation therapy, immunotherapy, targeted therapy, a small molecule therapeutic or any combination thereof; wherein if the anti-cancer treatment comprises a chemotherapeutic treatment, then the chemotherapeutic treatment comprises treatment with a protein kinase inhibitor, receptor tyrosine kinase inhibitor, antimetabolite, platinum complex, spindle poison, DNA crosslinking drug and alkylating agent, bleomycin, antibiotic, and topoisomerase inhibitor or combinations thereof; wherein if the chemotherapeutic treatment comprises treatment with a receptor tyrosine kinase inhibitor, then the receptor tyrosine kinase inhibitor is selected from the group consisting of sunitinib, pazopanib, axitinib, sorafenib or combinations thereof; wherein if the chemotherapeutic treatment comprises treatment with a protein kinase inhibitor, then the protein kinase inhibitor is temsirolimus or everolimus; wherein if the anti-cancer treatment comprises a targeted therapy, then the targeted therapy comprises anti-tumor antibodies such as bevacizumab, interferon and combinations thereof; wherein if the anti-cancer treatment comprises immunotherapy, then the immunotherapy is interferon, high-dose-interleukin 2 or combinations thereof; and wherein if the anti-cancer treatment comprises a small molecule therapeutic, then the small molecule therapeutic comprises tivozantinib.
4. The method of claim 1, wherein the at least one housekeeping gene comprises ACTB, RPL13A, RPL9, and RPS29 and the predetermined reference level is the average expression level of the four housekeeping genes.
5. The method of claim 1, the predetermined reference level is the geometric mean C.sub.T value of the at least one housekeeping gene.
6. The method according to claim 1, wherein detecting the level of expression for each marker comprises: a. contacting a nucleic acid sequence obtained or derived from said sample with at least one primer and/or at least one probe for amplification of a sequence of the marker; b. amplifying said nucleic acid sequence using said at least one primer or probe with a polymerase enzyme; and c. detecting the level of expression of the marker in said sample.
7. The method of claim 1, wherein the TKI therapy is selected from the group consisting of sunitinib, pazopanib, axitinib, sorafenib, or combinations thereof.
Description
BRIEF DESCRIPTION OF THE DRAWINGS
(1) The invention will be better understood with reference to the detailed description when considered in conjunction with the non-limiting examples and the accompanying drawings, in which:
(2)
(3)
(4)
(5)
(6)
(7)
DETAILED DESCRIPTION OF THE PRESENT INVENTION
(8) In a first aspect the present invention refers to a method of making a prognosis as to whether a patient having renal cancer is likely to survive in a tumour tissue sample obtained from the patient. The method may comprise: determining the level of expression for each marker of a panel of markers, wherein the panel comprises at least one housekeeping gene selected from the group consisting of ACTB, RPL13A, RPL9, and RPS29 and any combinations thereof; and at least one prognostic gene selected from the group consisting of CXCL5, EFNA5, EMCN, G6PC, GFPT2, HIST2H3C, IGFBP1, LAMB3, MMP9, MOCOS, PLG, PRAME, RARRES1, SDPR, SLC6A19, TK1, KDELR3 and TSPAN7 and any combinations thereof; determining whether an expression parameter for each marker in the tumour tissue sample is achieved by comparing the level of expression of each marker with a predetermined reference level associated with each marker; determining the differential expression of each marker in the tumour tissue sample based on the expression parameter for each marker to provide a prognosis for renal cancer.
(9) The prognostic gene may be selected from the group consisting of CXCL5, LAMB3, MMP9, PRAME, RARRES1, PLG, SLC6A19, EFNA5, IGFBP1 and EMCN and any combinations thereof.
(10) In one embodiment, the renal cancer may be clear cell renal cell carcinoma (ccRCC) or a mixture of renal tumours comprising ccRCC.
(11) In one embodiment, the patient may suffer from ccRCC or wherein the patient suffers from ccRCC and undergoes anti-cancer treatment. The likelihood of survival may correlate with a decrease in the likelihood of metastasis, disease recurrence or early death.
(12) The tumour tissue sample may be obtained from tissue selected from the group consisting of frozen tissue, tissue biopsies, circulating tumor cells, bodily fluids or other biological sample.
(13) In one embodiment, the bodily fluids are selected from the group consisting of ascites, effusions, cerebrospinal and urine.
(14) The patient may be undergoing anti-cancer treatment. The anti-cancer treatment may be selected from the group consisting of a chemotherapeutic treatment, a surgical treatment, a treatment with radiation therapy, immunotherapy, targeted therapy, small molecule therapeutics or any combination thereof. In some embodiments then chemotherapeutic treatment may comprise treatment with a protein kinase inhibitor, receptor tyrosine kinase inhibitor, antimetabolite, platinum complex, spindle poison, DNA crosslinking drug and alkylating agent, bleomycin, antibiotic, and topoisomerase inhibitor or combinations thereof.
(15) The receptor tyrosine kinase inhibitor may be selected from the group consisting of sunitinib, pazopanib, axitinib, sorafenib or combinations thereof.
(16) In some embodiments the protein kinase inhibitor may be temsirolimus or everolimus.
(17) The targeted therapy may comprise anti-tumour antibodies such as bevacizumab, interferon and combinations thereof.
(18) The immunotherapy may be interferon, high-dose-interleukin 2 or combinations thereof.
(19) The small molecule therapeutic may comprise tivozantinib.
(20) The patient may be a mammal or a human. In some embodiments, the human may be an ethnic Asian, or an ethnic Caucasian, or an ethnic African.
(21) The predetermined reference level may be determined based on the expression of the at least one housekeeping gene. In some embodiments, the determining of whether or not the biomarkers are differentially expressed in the sample further comprises normalizing the levels of expression of the markers to the at least one housekeeping gene selected from the group consisting of ACTB, RPL13A, RPL9, and RPS29 and any combinations thereof.
(22) In some embodiments, the levels of expression of the markers may be normalized against the geometric mean C.sub.T value of the at least one housekeeping gene.
(23) In some embodiments, an increase in the expression parameter of prognostic genes selected from the group consisting of EMCN, G6PC, PLG, SDPR, SLC6A19 and TSPAN7 may indicate an increased chance of survival. In some embodiments, a decrease or absence in the expression parameter of prognostic genes selected from the group consisting of G6PC, PLG, SDPR, SLC6A19 and TSPAN7 may indicate a decreased chance of survival.
(24) An increase in the expression parameter of prognostic genes selected from the group consisting of CXCL5, EFNA5, GFPT2, HIST2H3C, IGFBP1, LAMB3, MMP9, MOCOS, PRAME, RARRES1 and TK1 may indicate a decreased chance of survival.
(25) A decrease or absence in the expression parameter of prognostic genes selected from the group consisting of GFPT2 and LAMB3 may indicate an increased chance of survival. In some embodiments, the prognosis may be determined by a multivariate algorithm. The multivariate algorithm may be multivariate logistic correlation analysis or linear discriminant analysis (LDA).
(26) The multivariate algorithm may use a class prediction approach.
(27) In a second aspect the present invention refers to a method for predicting responsiveness to an anti-renal cancer treatment in a patient having or at risk of developing renal cancer in a tumour tissue sample obtained from the patient comprising: determining the level of expression for each marker of a panel of markers, wherein the panel comprises at least one housekeeping gene selected from the group consisting of ACTB, RPL13A, RPL9, and RPS29 and any combinations thereof; and at least one prognostic gene selected from the group consisting of CXCL5, EFNA5, EMCN, G6PC, GFPT2, HIST2H3C, IGFBP1, LAMB3, MMP9, MOCOS, PLG, PRAME, RARRES1, SDPR, SLC6A19, TK1, KDELR3 and TSPAN7 and any combinations thereof; determining whether an expression parameter for each marker in the tumour tissue sample is achieved by comparing the level of expression of each marker with a predetermined reference level associated with each marker; wherein differential expression of each marker in the tumour tissue sample based on the expression parameter for each marker is indicative of the responsiveness of the patient to the anti-renal cancer treatment.
(28) The prognostic gene in may be selected from the group consisting of CXCL5, LAMB3, MMP9, PRAME, RARRES1, PLG, SLC6A19, EFNA5, IGFBP1 and EMCN and any combinations thereof.
(29) In some embodiments, an increase in the expression parameter of prognostic genes selected from the group consisting of EMCN, G6PC, PLG, SDPR, SLC6A19 and TSPAN7 may indicate an increased responsiveness to the treatment and an increased chance of survival.
(30) In some embodiments, a decrease or absence in the expression parameter of prognostic genes selected from the group consisting of EMCN, G6PC, PLG, SDPR, SLC6A19 and TSPAN7 may indicate a decreased responsiveness to the treatment and a decreased chance of survival.
(31) In one embodiment, an increase in the expression parameter of prognostic genes selected from the group consisting of CXCL5, EFNA5, GFPT2, HIST2H3C, IGFBP1, LAMB3, MMP9, MOCOS, PRAME, RARRES1 and TK1 may indicate a decreased responsiveness to the treatment and a decreased chance of survival.
(32) In one embodiment a, decrease or absence in the expression parameter of prognostic genes selected from the group consisting of CXCL5, EFNA5, GFPT2, HIST2H3C, IGFBP1, LAMB3, MMP9, MOCOS, PRAME, RARRES1 and TK1 may indicate an increased responsiveness to the treatment and an increased chance of survival.
(33) In some embodiments, the renal cancer is a clear cell renal cell carcinoma (ccRCC).
(34) In one embodiment, the step of determining the level of expression of each marker may comprise: contacting a nucleic acid sequence obtained or derived from said sample with at least one primer and/or at least one probe for amplification of a sequence of the marker; amplifying said nucleic acid sequence using said at least one primer or probe with a polymerase enzyme; detecting the level of expression of the marker in said sample.
(35) In one embodiment, the amplification step may be performed by polymerase chain reaction (PCR).
(36) In a second aspect the present invention refers to a kit comprising: at least one reagent to determine the level of expression for each marker of a panel of markers in a tissue tumour sample, wherein the panel comprises at least one housekeeping gene selected from the group consisting of ACTB, RPL13A, RPL9, and RPS29 and any combinations thereof; and at least one prognostic gene selected from the group consisting of CXCL5, EFNA5, EMCN, G6PC, GFPT2, HIST2H3C, IGFBP1, LAMB3, MMP9, MOCOS, PLG, PRAME, RARRES1, SDPR, SLC6A19, TK1, KDELR3 and TSPAN7 and any combinations thereof.
(37) In one embodiment, the prognostic gene may be selected from the group consisting of CXCL5, LAMB3, MMP9, PRAME, RARRES1, PLG, SLC6A19, EFNA5, IGFBP1 and EMCN and any combinations thereof.
(38) In one embodiment, the at least one reagent may comprise at least one primer and/or at least one probe for amplification of a sequence comprising the marker.
EXAMPLES
Example 1
Development of a Prognostic Assay for Clear Cell Renal Cell Carcinoma (ccRCC)
(39) Methods
(40) Study Population
(41) The study was done in a retrospective manner with patient cohort that included 279 ccRCC patients who underwent resection for clear-cell renal cell carcinoma at National Cancer Centre, Singapore between 1999 and 2012. All patients had histologically confirmed ccRCCs for which formalin-fixed paraffin-embedded (FFPE) primary or metastatic tumour blocks were available. Fifty-five FFPE samples were used for Illumina's Whole Genome DASL analysis. For qPCR analysis 214 FFPE samples were used. This group included 48 patients with metastatic ccRCCs treated with anti-VEGF targeted treatment (sunitinib or pazopanib) as first-, second or third-line therapy. The patient database included patients' baseline pathological features and disease outcome data including date of death or last follow-up. Disease-specific survival time was calculated from date of first diagnosis. All patients provided written informed consent, and the study was approved by SingHealth Institutional Review Board.
(42) RNA Extraction, cDNA Synthesis and Quantitative PCR
(43) FFPE blocks were sectioned into 4-μm sections, stained with hematoxylin-eosin for confirmation of histological diagnosis and tumour tissue content (>70%). Following deparaffinization of 3-6 FFPE sections per sample, macrodissection was performed with a sterile single-use scalpel to remove nontumour elements and RNA was extracted using Qiagen's RNeasy FFPE Kit. Integrity of RNA samples was assessed by verifying amplification of housekeeping gene RPL13A using Power SYBR Green RNA-to-CT, 1-Step kit (Applied Biosystems) with 100 ng of RNA template. Samples that had cycle threshold (Ct) value for RPL13A below 29, were used for Illumina Whole genome-DASL expression analysis or for subsequent conversion to complementary DNA. All DASL (cDNA-mediated Annealing, Selection, extension, and Ligation) assays were performed in Biopolis Shared Facilities, A*STAR, Singapore.
(44) For selected genes, primers were designed for amplicons of size 55-90 bp, with at least primer spanning exon-exon boundaries. For qPCR, 1 μg extracted. RNA was reverse transcribed with random hexamer primers using High Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Relative expression of each target gene was measured by real-time qPCR with Power SYBR Green Master Mix (Applied Biosystems) on a Bio Rad CFX96 machine. cDNA was diluted four-fold and 1 μl of the diluted cDNA was used as template in a 10 μl reaction with primers at a final concentration of 200 nM. Products were checked for specificity of amplification with melt curve and positive and negative controls were run for each plate.
(45) Identification of Prognostic Genes
(46) Quantile-normalized gene expression data for 55 ccRCCs analyzed by Illumina's DASL Assay using HumanHT-12 v4 BeadChip (24,526 features) was obtained using Genomestudio. Data was further analyzed with R 2.13.1 using packages Genefilter, Survival and siggenes. Data was first filtered to include probes that were present in at least one sample defined by detection p-value <0.05 (19,303 features). Further, genes with expression values greater than 3000 in at least one sample and with covariance greater than 0.75 were selected (3740 features) as reliably expressed features with reasonably high variation in expression among 55 samples. Unsupervised hierarchical clustering (Ward clustering) of samples by the selected features was done and resulting clusters were cut at the highest level to generate two main expression subgroups. Significance analysis of microarrays (SAM) was applied to select features that were significantly differentially expressed between the two subgroups (p<0.05).
(47) Quantitative PCR Data Analysis
(48) Expression data for 18 genes for all FFPE RNA samples was collected as cycle threshold (Ct) values. Expression was normalized by subtracting Ct values from the geometric average of Ct values for four housekeeping genes. When Ct exceeded 36, expression was not subject to data transformation and deemed to be of value zero. Following normalization, a value of +14.72 was added to the delta Ct values to scale negative values to start from 0 and to fall on a positive scale, where a 1-unit increase reflects a doubling of RNA. These transformed delta Ct values were used for further analysis.
(49) Statistical Analysis
(50) To determine correlation of expression by DASL and qPCR, transformed qPCR data was converted to a linear scale by the function 2.sup.ΔΔCt. Linear correlation between the linear qPCR expression data and DASL expression values in 55 samples was determined by Pearson's product-moment correlation. The ten most significantly correlated genes representing a successful translation for measuring expression on the qPCR platform were selected to generate a model for a prognostic score with which to correlate clinical outcomes.
(51) Expression values from qPCR for 10 genes were used to generate a model using multivariate logistic regression analysis for the dichotomous outcome of good or poor prognosis classification from the original prognosis groups generated by unsupervised hierarchical clustering of DASL expression data. The intercept and coefficients for each of 10 genes were generated using this multivariate analysis. The cut-off score to classify prognosis groups was determined as the score at which misclassification of 55 ccRCC samples using the prognostic score, with respect to their original class assignment was minimized (9%).
(52) Prognostic scores for 214 FFPE RNA samples were determined from expression values of 10 genes according to the derived algorithm. Based on the pre-determined cut-off, samples were stratified into low- and high-scoring groups corresponding to poor and good prognosis categories respectively. The distribution of disease-specific survival was estimated by the Kaplan-Meier method with log-rank test to assess the association of the prognosis category with primary end-point of disease-specific mortality. Other clinical covariates including age, tumour stage and tumour grade were compared to outcome using univariate and multivariate Cox proportional hazards modeling. Wald and likelihood ratio tests were done for multivariate modeling to assess statistical significance. Cox model was also used to estimate hazard ratios and 95% CI for prognosis category. A subset of 34 samples corresponding to patients that underwent TKI therapy with evaluable clinical response data, were stratified into prognosis groups. Objective response was classified as partial response, stable disease, progressive disease or mixed response. The partial response and stable disease categories were classified as responders and progressive disease and mixed response categories were classified as non-responders. The significance of association between prognosis groups and response outcome was measured using Fisher's exact method. A logistic regression analysis was used to assess the association between prognostic score as a continuous score and response to TKI as a categorical variable.
(53) Results
(54) Material from 55 patients was analyzed in the screening cohort initially for whole-genome expression profiling and subsequent selection and confirmation of prognostic genes, and material from 214 patients was analyzed to validate the performance of prognostic genes. Baseline characteristics of patients in the initial screening group are described in Table 1.
(55) TABLE-US-00001 TABLE 1 Characteristics of patients in initial screening group used for DASL study Number of samples 55 Age Range 32-81 Median 59 Gender-n (%) Male 43 (78%) Female 12 (22%) TNM stage I 33 (60%) II 2 (3.6%) III 14 (25.5%) IV 6 (10.9%) Tumour T stage 1 33 (60%) 2 3 (5.5%) 3 18 (32.7%) 4 1 (1.8%) Tumour M stage 0 41 (74.5%) 1 6 (10.9%) X 8 (14.5%) Tumour Grade 1 9 (16.3%) 2 29 (52.7%) 3 9 (16.4%) 4 8 (14.5%) Primary tumour Range 1.9-16 size Median 5 <4 16 (29%) 4 to 7 26 (47%) >7 13 (24%) Unknown ECOS PS 0 37 (67.2%) 1 13 (23.6%) 2 0 (0%) Unknown 5 (9%) Follow-up Range 0.036-6.42 duration Mean 4.19 (yr) Patient Status Deaths (cancer- 7 (12.7%) related) Deaths (other 2 (3.6%) causes) Alive with disease 2 (3.6%) No evidence 44 (80%) of disease Post-nephrectomy Sunitinib 4 (7.2%) treatment Pazopanib 1 (1.8%) Interferon 1 (1.8%) 5FU + PTK/ZK 1 (1.8%) 2 patients without nephrectomy-1 definite metastasis.
(56) All patients in the screening group underwent nephrectomy for primary tumour, except two patients who were operated on metastatic tumours in the lungs. Following retrieval of paraffin blocks for primary or metastatic tissue, RNA was extracted from FFPE sections and integrity of RNA validated before whole-genome expression profiling was done for 55 ccRCC. Characteristics of patients from the validation group are described in Table 2. Eleven patients did not undergo nephrectomy and tissue was obtained from biopsies or metastatic tissue.
(57) TABLE-US-00002 TABLE 2 Characteristics of patients in confirmation group Number of samples 224 Age Range 29-91 Median 56.5 Gender Male 152 (68%) Female 72 (32%) TNM stage I 106 (47.3%) II 26 (11.6%) III 55 (24.6%) IV 36 (16.1%) Unknown 1 (0.5%) Tumour T stage 1 109 (50.9%) 2 33 (15.4%) 3 67 (31.3%) 4 1 (0.5%) X 3 (1.4%) NA 1 (0.5%) Tumour M stage 0 181 (84.5%) 1 32 (15%) Unknown 1 (0.5%) Tumour Grade 1 22 (10.3%) 2 113 (52%) 3 63 (27%) 4 25 (10.3%) Primary tumour Range 1.5-25 size Median 5.5 ECOS PS 0 126 (56.2%) 1 47 (21%) 2 3 (1.4%) Unknown 48 (21.4%) Follow-up Range 0.049-12.19 duration (yr) Mean 5.54 Patient Status Deaths (cancer- 49 (21.9%%) related) Deaths (other 14 (6.3%) causes) Alive with 27 (12%) disease No evidence 134 (59.8%) of disease 11 patients without nephrectomy.
(58) In order to identify biologically relevant subtypes of ccRCC based on gene expression profiling, whole-genome expression data for the screening group of 55 ccRCC was processed to derive a gene set that is reliably detectable (expressed) with elements of this set retaining reasonably high variation of expression among the samples. As RNA from FFPE material is of a highly degraded nature, preprocessing of data was done to only include features that had significant expression (p<0.05) as defined by. Illumina detection platform. A set of 3740 features which had significant expression in at least one sample (expression values more than 3000) and a coefficient of variation greater than 0.75, was used for performing unsupervised hierarchical clustering of 55 ccRCCs. Samples were partitioned into two main groups based on differential expression of these 3740 features (
(59) Having delineated two main biologically-defined prognostic subtypes, genes that were most significantly differentially expressed between these two subtypes, and expression levels of which could potentially be used to accurately classify ccRCCs were identified. To achieve this, significance analysis of microarrays (SAM) analysis was applied to the highest level of the resulting clustering dendrogram (
(60) TABLE-US-00003 TABLE 3 List of genes significantly differentially expressed between Prognosis subtypes 1 and 2. Prognosis Prognosis Prognosis Prognosis Subtype Subtype Subtype Subtype Gene q. value R. fold 1 Mean 1 SD 2 Mean 2 SD NME1 0.00 1.86 4115.43 1416.31 7308.75 1247.38 TSPAN7 0.00 0.15 3354.05 1756.40 559.59 412.96 PLG 0.00 0.08 4141.75 6073.62 65.42 4.21 F12 0.01 3.68 1712.15 1060.05 5794.40 2836.39 TUBB3 0.01 5.46 1661.67 1721.87 5592.52 2669.45 KDELR3 0.01 4.24 1009.97 1294.46 2802.80 1147.52 TUBA1C 0.01 1.38 29840.58 4981.53 41013.67 5110.15 CTHRC1 0.01 4.27 2791.22 1957.13 10061.91 5098.79 PPAP2A 0.01 0.65 15673.18 2879.62 10147.16 1896.57 VIPR1 0.01 0.12 27517.56 16607.77 3697.29 2695.32 RAG1AP1 0.01 1.23 7458.21 961.35 9093.07 599.98 SEC61G 0.01 1.53 13658.87 3362.98 20565.09 3389.91 C15orf58 0.01 2.01 2693.30 1167.46 4974.76 1099.97 FKBP10 0.01 2.07 4606.42 2337.31 8518.46 1914.65 ANKRD56 0.01 0.33 386.91 456.28 71.59 7.79 SLC7A5 0.01 2.85 12127.31 7564.27 26289.58 6651.07 TK1 0.01 5.71 2332.11 2041.67 7540.38 3417.20 FATE1 0.01 0.34 470.08 502.89 102.66 27.37 CCT3 0.01 1.53 4860.83 1587.98 7230.99 1248.00 RGS7BP 0.01 0.09 2723.24 2375.56 379.83 877.70 MOCOS 0.01 11.35 334.56 846.20 2690.46 2250.93 MAZ 0.01 1.47 3572.83 960.98 5133.94 857.64 PYGB 0.01 3.16 1047.23 754.87 2636.35 1121.14 EFNA5 0.01 8.03 5266.97 4969.20 19699.94 7750.11 ILF2 0.01 1.44 13618.75 3364.73 19160.67 2735.72 PTHLH 0.01 6.69 6441.70 7953.95 19633.73 9861.15 TRIB3 0.01 4.75 2091.17 2214.14 7387.63 5363.56 MSI1 0.01 0.13 2616.70 1663.41 421.20 475.86 IGFBP1 0.01 6.59 931.63 2215.31 3952.64 4619.23 TMED9 0.01 1.30 11224.71 2442.54 14282.52 1309.85 IL6 0.01 12.63 1869.99 2796.27 11071.49 7797.10 CD36 0.01 0.45 13308.98 4037.75 6142.54 3025.02 CDCP1 0.01 3.13 2578.14 1793.62 6624.40 2654.82 B3GNT4 0.01 4.00 2393.20 2418.73 6054.66 3356.73 G0S2 0.01 1.97 11673.35 5799.14 21565.14 7144.86 TMCC3 0.01 0.44 274.45 318.11 78.09 9.91 RARRES1 0.01 4.40 877.72 958.54 3050.94 2010.85 PPP1R13B 0.01 0.67 42295.88 7515.15 28350.28 5818.83 LAMB3 0.01 5.19 1295.49 2012.96 4883.82 3909.33 UBE2C 0.01 2.85 1059.59 765.06 2522.13 1160.76 RPL26L1 0.01 1.63 7649.03 2706.42 12071.41 2848.66 SLC6A18 0.01 0.15 4033.38 6391.15 100.80 18.15 HSD3B7 0.01 2.07 11014.58 6432.93 20571.02 7174.79 SDPR 0.01 0.22 1522.06 940.14 341.22 314.94 ABCA12 0.01 5.20 480.44 415.90 2301.93 1504.30 RARB 0.01 0.38 2687.22 1102.62 1032.38 490.48 GXYLT2 0.01 4.01 5156.94 4453.50 12491.42 4449.21 EMCN 0.01 0.14 7130.05 2572.85 1838.98 2335.68 F2RL3 0.02 0.15 8055.35 5449.34 1685.99 2041.84 ATP1B2 0.02 0.42 344.71 386.24 93.67 24.18 PNPLA7 0.02 0.51 22112.68 4675.99 11732.24 4373.74 SNORD78 0.02 1.72 6037.43 2431.38 9792.73 2374.50 SLC6A19 0.02 0.20 1804.62 2186.33 150.45 105.00 TIMELESS 0.02 1.43 12956.99 3295.02 18022.91 2939.25 GPRC5A 0.02 4.63 4826.77 4537.78 12772.09 5006.02 GRAMD1C 0.02 0.35 370.07 390.16 90.48 47.34 SRPX2 0.02 3.16 4013.34 3254.21 11279.39 6145.79 TSTA3 0.02 2.64 3982.77 2454.89 8440.76 3228.31 NUDT1 0.02 2.14 1552.81 875.10 3179.63 1321.63 SHISA4 0.02 1.77 967.83 569.27 1544.70 398.83 HSPC159 0.02 0.58 4829.10 1467.58 2805.99 934.44 ADAM12 0.02 3.08 510.35 365.30 1409.79 605.81 NOMO1 0.02 1.41 6920.06 1924.52 9613.80 1698.10 TIPRL 0.02 1.71 990.44 358.25 1653.15 488.74 MRPS12 0.02 1.21 7714.22 802.63 9329.89 973.51 MMP9 0.02 5.69 4988.74 4208.08 21201.30 15556.89 HIST2H3C 0.02 5.02 2303.06 2130.49 6786.64 4112.68 RNF128 0.02 1.80 3769.33 1199.91 6758.63 2503.35 LGSN 0.02 0.28 572.34 602.79 109.70 99.75 SLC9A3 0.02 0.17 3341.76 4618.18 349.19 416.19 SOX8 0.02 0.12 4564.53 3560.46 929.41 1464.94 AUTS2 0.02 0.57 19373.23 3306.57 11543.49 4626.93 SLC39A1 0.02 1.24 15305.51 2179.30 18910.64 2179.79 MIOX 0.02 0.08 13111.28 7515.13 1802.21 1971.93 RGS5 0.02 0.66 40485.22 5700.72 27324.05 6703.61 PECAM1 0.02 0.62 18350.76 3858.62 11604.99 3634.17 TRPM8 0.02 5.74 318.32 378.30 1697.47 1305.48 GFPT2 0.02 13.91 391.18 720.30 4088.86 4168.47 RANBP1 0.02 1.28 14423.19 2714.14 18210.15 2250.26 C3orf71 0.02 0.42 384.37 604.97 74.61 11.05 RHBDL2 0.02 2.29 506.95 559.64 810.42 298.25 KCNE4 0.02 0.18 1221.87 1424.44 185.22 321.98 C1R 0.02 1.92 14156.01 7227.32 23310.54 5139.30 PLOD2 0.02 1.26 24600.38 4623.15 30622.17 3447.21 IVNS1ABP 0.02 0.77 16167.92 2196.01 12503.06 2084.96 SND1 0.02 1.39 7357.16 2047.46 9949.13 1553.75 HIST2H3A 0.02 2.44 3017.32 1640.84 6997.48 3282.49 SPAG4 0.02 4.19 944.27 905.11 2703.90 1724.03 PTTG1 0.03 2.55 3413.64 2018.54 6923.91 2785.22 PHF1 0.03 0.71 15466.02 2806.57 11095.66 2229.15 C5orf46 0.03 5.84 6436.88 7059.01 16967.96 9955.13 SLC22A12 0.03 0.15 5367.45 4545.15 944.03 1032.47 AGTR1 0.03 0.21 3255.00 3016.64 628.76 585.23 PSMD14 0.03 1.39 7875.00 2200.16 10668.21 1715.25 CXCL5 0.03 9.19 195.44 372.29 2151.50 1844.54 IDH1 0.03 1.75 2973.17 955.92 5258.68 1988.51 PLAT 0.03 0.55 11980.85 3772.90 6618.52 2408.05 NEFL 0.03 2.69 16495.66 13376.70 32780.54 12463.63 CPM 0.03 0.55 4182.38 1303.46 2327.10 846.03 PADI4 0.03 0.29 1110.93 1312.93 171.64 106.03 KDR 0.03 0.36 7657.95 2956.14 3128.36 1821.88 IL1R2 0.03 3.09 1176.97 1243.79 3178.45 2300.68 HOTAIR 0.03 0.27 861.92 922.39 146.96 143.07 G6PC 0.03 0.09 6369.61 5723.08 1047.48 1599.24 MAGED1 0.03 1.39 10911.15 2612.86 14978.58 2875.12 GMPPA 0.03 1.56 3386.76 1102.11 5152.45 1375.77 CDR2 0.03 1.95 813.52 533.19 1374.27 507.74 TFIP11 0.03 1.41 7714.60 1503.71 10949.21 2743.95 STEAP3 0.03 1.83 3492.48 797.29 6733.09 2748.95 TMED3 0.03 1.30 24669.33 3174.73 32213.75 5136.95 WT1 0.03 6.95 718.66 1168.99 2785.68 2041.81 KLHL4 0.03 8.99 293.22 509.57 2244.72 1688.17 DSN1 0.03 1.52 8991.83 2590.53 13481.01 3544.08 MOGS 0.03 1.49 9123.15 3058.61 12947.63 2643.80 PNKD 0.03 1.79 5442.82 2108.12 9636.16 3638.06 PSMG3 0.03 3.52 401.33 350.68 1139.42 563.20 CYP1B1 0.03 2.97 3589.24 2886.74 7466.17 3150.56 C1orf117 0.03 5.31 1033.97 1467.58 2811.54 1911.20 B3GAT3 0.03 2.74 2710.42 1694.24 5293.41 2097.10 TUBA1A 0.03 1.36 23827.72 5695.64 31936.94 5864.45 CD82 0.03 1.73 2995.34 987.65 5238.59 1954.52 GOLSYN 0.03 6.94 1107.34 1425.21 4308.70 3186.70 PDGFRL 0.03 3.46 3218.74 2619.40 8990.74 4508.20 NEB 0.03 5.37 768.54 1709.39 3361.68 3710.18 TMEM45A 0.03 2.54 3039.47 2519.30 5447.44 1844.59 OBFC2B 0.03 2.67 4391.14 2492.56 8814.26 3534.06 LOC653604 0.03 2.52 6280.04 2741.96 13158.67 4437.94 SERPINA5 0.04 2.58 331.45 301.81 797.15 547.28 SLC19A3 0.04 1.77 15002.93 7349.70 24208.80 6971.49 SERPINA3 0.04 15.12 600.16 1313.92 7220.24 7349.07 ANAPC11 0.04 1.57 3566.78 1141.81 5548.05 1850.44 RCN1 0.04 1.19 15840.27 2089.52 18846.90 1946.14 TNIP2 0.04 1.86 2319.03 1180.42 4177.59 1921.20 SNORD83A 0.04 1.29 26374.70 4606.54 33810.72 5363.56 KCNG1 0.04 3.51 247.90 187.88 960.40 879.42 USP1 0.04 2.30 720.81 443.87 1348.10 595.69 TBX18 0.04 2.23 218.05 142.79 472.71 240.90 SLITRK2 0.04 4.32 3694.49 3754.85 8343.69 6095.61 KDELR2 0.04 1.32 11824.82 2520.56 15525.01 2816.41 ARMET 0.04 1.34 19602.19 4371.38 26039.10 5453.29 POP7 0.04 1.46 14696.18 4082.27 21225.29 4591.96 RASGEF1C 0.04 3.32 420.55 289.63 1543.87 1180.95 LGALS1 0.04 1.26 28437.27 4116.78 35722.06 5439.03 MGC5139 0.04 2.42 2947.56 1963.00 5594.60 2648.39 SPOCK1 0.04 3.22 4504.99 4992.34 11127.36 7300.38 NUP62 0.04 1.17 17428.86 1801.70 20302.79 2002.33 ARL4C 0.04 2.09 549.72 300.12 1126.80 596.83 LOC730347 0.04 2.17 2811.49 1765.40 4949.16 1691.70 FABP6 0.04 3.31 9199.16 6736.40 23030.53 14403.94 KIAA1244 0.04 2.87 747.48 588.06 1806.86 872.23 CLEC16A 0.04 1.25 16304.25 3061.72 20247.88 2565.45 SLC2A1 0.04 1.28 13905.90 2717.29 17594.25 2386.18 HK2 0.04 2.04 1144.94 660.80 2137.48 962.07 PDIA6 0.04 1.33 10335.70 2265.40 13602.50 2462.30 ANLN 0.04 2.00 2042.31 1039.76 3547.06 1238.50 LPAR2 0.04 1.73 1339.13 726.76 2025.40 586.99 MDM2 0.04 1.36 7460.56 1639.34 10101.95 1903.06 PPIB 0.04 1.30 14737.26 3450.28 18853.32 2921.23 NCAM1 0.04 6.25 1025.39 1072.53 7226.46 5917.43 PRAME 0.04 9.84 431.97 722.63 4797.21 5023.85 ARPC1B 0.04 1.26 16618.74 2912.39 20889.46 3187.11 NUDT5 0.04 2.33 3148.63 1991.61 6246.28 2918.45 C3 0.04 1.46 20173.35 7404.26 27838.40 5422.49 SNORD88C 0.04 1.31 12948.89 3757.04 16413.60 2424.29 COL22A1 0.04 8.22 321.92 559.54 2753.71 3864.98 PYGO2 0.04 2.06 2853.56 1614.03 4811.90 1602.52 FN1 0.04 1.22 31291.54 4665.65 38000.05 4653.10 PSMC4 0.04 1.46 4979.45 1680.57 6995.79 1551.24 HM13 0.04 1.27 8337.88 1109.64 10683.99 2139.23 BUD13 0.04 2.09 1315.15 752.01 2621.05 1657.03 LRWD1 0.04 1.88 1960.54 1007.66 3072.47 933.63 CXCL1 0.04 1.66 3105.68 1698.17 4489.29 1029.72 QSOX1 0.04 1.71 2118.22 663.00 3767.46 1771.45 SNORA71C 0.04 1.47 9011.31 3219.43 12675.94 2903.71 POMT2 0.04 2.49 2862.90 2098.33 5999.21 3637.75 CBX8 0.04 1.99 2970.56 1684.57 5320.43 2315.35 FBF1 0.04 2.47 270.72 289.88 555.44 297.07 GPATCH4 0.04 1.79 1642.15 956.83 2518.50 750.32 F8A3 0.04 1.85 999.66 758.57 1507.96 435.39 VSIG4 0.04 2.25 341.93 254.50 746.76 431.58 GCNT3 0.04 7.66 1312.74 1687.81 5389.18 4069.88 PRKDC 0.04 1.28 10418.59 2036.98 13212.66 2059.89 MED8 0.04 1.55 2065.10 734.24 3135.14 922.28 LGALS3BP 0.04 1.60 7812.87 3036.80 12232.48 3953.69 C1orf35 0.04 2.59 1137.04 766.63 2533.97 1391.58 C11orf48 0.04 1.43 5177.48 2207.45 7013.08 1430.97 MIR210 0.04 3.02 2837.84 1931.78 5805.25 3066.74 SLC7A5P1 0.04 1.96 6131.99 2648.39 10379.97 2987.96 SPP1 0.04 1.20 40041.43 5945.15 47925.17 5558.27 PSMB4 0.04 1.37 7695.26 1949.18 10370.90 2144.71 RPTOR 0.04 2.46 1988.63 1173.89 3981.19 2126.65 LOC100128731 0.04 1.19 28519.44 3852.65 33884.81 3774.57 SEMA4F 0.04 3.31 1045.97 930.08 2906.89 2207.10 PATE2 0.04 1.60 278.78 118.54 437.78 141.35 C12orf57 0.04 1.80 3088.65 1777.46 4837.80 1412.47 TAGLN2 0.04 1.21 23283.51 3205.06 28266.89 3700.43 PSMA5 0.04 1.40 15420.83 4581.14 21223.75 4486.60 SNORD37 0.04 1.32 15577.80 2364.61 20698.82 4146.51 PSMB3 0.04 1.31 17435.08 3601.29 22745.00 4145.14 CCDC97 0.04 1.38 7777.20 2637.50 10209.00 1726.19 ARPC1A 0.04 1.30 15340.75 3646.75 19694.30 3457.62 LONP1 0.04 1.35 14402.92 3878.06 19191.42 3808.25 CSNK1D 0.05 1.17 19638.21 3069.24 22891.03 2213.97 CDCA3 0.05 2.46 4403.47 3203.44 9312.46 4525.62 SFT2D1 0.05 2.13 3094.38 2231.55 4803.86 1427.04 PDPN 0.05 3.04 908.04 760.73 2600.69 2195.71 NME2 0.05 1.29 4250.92 862.28 5448.03 946.46 NUF2 0.05 2.47 593.66 360.37 1405.52 759.75 TFAP2A 0.05 2.56 198.66 189.96 479.09 326.65 C1S 0.05 1.78 10576.96 5285.45 17097.44 4818.86 KIAA1715 0.05 1.76 3323.82 1406.42 5550.09 2102.36 C17orf58 0.05 1.89 2285.68 893.56 3916.49 1415.70 DHCR24 0.05 2.23 639.67 374.97 1344.86 693.28 TREM1 0.05 2.47 282.49 161.14 756.12 512.59 C9orf140 0.05 2.29 1120.27 817.81 2130.00 994.48 IL10RB 0.05 1.55 1919.54 895.28 2750.80 697.57 SCARNA20 0.05 1.55 16418.28 5117.96 25271.34 7798.24 SEC23B 0.05 1.20 10747.90 2345.62 12624.34 1010.84 HIST2H4B 0.05 1.20 8614.78 1475.84 10265.09 1148.24 SD: standard deviation
(61) As the aim was to identify prognostic subtypes- and implement, an expression-based method to classify ccRCCs into these prognostic subtypes, qPCR assays were designed for a set of 37 genes that were previously identified as prognostic genes on DASL platform. Expression levels measured by DASL method was reproduced by qPCR for these 37 genes in 6 randomly selected tumour samples that were also assessed by DASL assay. The list of selected genes and the primer sequences used to produce amplicons amenable to PCR amplification from FFPE RNA are described in Table 4. Separately, four genes (β-actin, RPL9, RPL13A, RPS29) whose expression was the least variable (lowest coefficients of variance) among the 55 ccRCCs were also identified to serve as normalization genes for qPCR (
(62) TABLE-US-00004 TABLE 4 List of prognostic and normalization genes and primers used to measure their expression levels using qPCR. Gene Forward Primer Reverse Primer AGTR1_v1 GTC GGC ACC AGG CCA TCT TCA GTA TGT ATT T GAA GAG TTG (SEQ ID NO. 1) (SEQ ID NO. 2) AGTR1_v3 GCT CAG AGG AGG CCA TCT TCA GTA TGT ATT TGA GAA GAG TTG (SEQ ID NO. 3) (SEQ ID NO. 4) CTHRC1 GTG GTG GAC CTG GAA TGC CAT TGG TAT AAT GGA AT CCC CAG (SEQ ID NO. 5) (SEQ ID NO. 6) CXCL5 CGC AAG GAG TTC CAG GGA GGC TAC ATC CCA CAC TT (SEQ ID NO. 7) (SEQ ID NO. 8) EFNA5 CAG CAG ATG ACA CAA AAG GCG GCT CCG TAC A GGG TAT C (SEQ ID NO. 9) (SEQ ID NO. 10) EMCN GAT CAA CCT CAG GTG CAG AGT GCT TCT GAT AAA GAG CAC CAG AC (SEQ ID NO. 11) (SEQ ID NO. 12) F2RL3 AGG TGG TGA TGA CCA GGG TGT CAC CAG CAC TGT CAT T (SEQ ID NO. 13) (SEQ ID NO. 14) G6PC CTG TCA GGC ATT CTT GAG GCT GGC GCT GTT G ATT ATA GAT (SEQ ID NO. 15) (SEQ ID NO. 16) GFPT2 GAG GAT ATG ACG CAT TCC ACA GTT TTG ACT TCC ACA GAC TTG GC (SEQ ID NO. 17) (SEQ ID NO. 18) HIST2H3C CAG AAG TCC ACG CAG GTC CGT CTT GAG CT AAA GTC CT (SEQ ID NO. 19) (SEQ ID NO. 20) IGFBP1 CAG ACA GTG TGA CCT CTT CCC ATT GAC ATC CA CCA AGG GT (SEQ ID NO. 21) (SEQ ID NO. 22) IL6 GAT TCA ATG AGG CTC TCA AAT CTG AGA CTT GCC TTC TGG AGG T (SEQ ID NO. 23) (SEQ ID NO. 24) KDELR3_v1 CTT GTA TGT GAC CGT AGA CTG TCT CAA AGT CCT CTG AAG GT (SEQ ID NO. 25) (SEQ ID NO. 26) KDELR3_v2 GCT GGA GAT CCT GAT CAG CAA GAC CTG GAC TT TGG AGA G (SEQ ID NO. 27) (SEQ ID NO. 28) KDR CCA GAT GAC AAC CTG GGC ACC ATT CAG ACG GA CCA CCA A (SEQ ID NO. 29) (SEQ ID NO. 30) KLHL4 CTG TGT GGA ACG GAG GAA CAC TCA GTA TGA TC GAG GTG CCA (SEQ ID NO. 31) (SEQ ID NO. 32) LAMB3 CAG AGG CAG AGG CCA ACT CCA TGT AGC TGT T CTT TCA TCC (SEQ ID NO. 33) (SEQ ID NO. 34) MIOX GTG CGG GAG TTC GTC AAT GAG CCC AAC AAG TTC CTG GT (SEQ ID NO. 35) (SEQ ID NO. 36) MMP9 CAG TAC CGA GAG CCA CCT GGT TCA AAA GCC TA ACT CAC TC (SEQ ID NO. 37) (SEQ ID NO. 38) MOCOS GGT GAA TGA GGC CTT TCC ATT CTC ACA GTA TC ATC ACT GGT G (SEQ ID NO. 39) (SEQ ID NO. 40) MSI1 TCG AGG GAC AGG TGG GAG TCG AAC CTC TCA CTG GA (SEQ ID NO. 41) (SEQ ID NO. 42) NUDT5 CCA GGG GAT GGA GTT CTT CAG CTA GAG TTT G CCA GAG CA (SEQ ID NO. 43) (SEQ ID NO. 44) PLG ATG GCT GAA AAC CCT CCA TAA TCA AGG AAG TC TTA GGA TGA GAG (SEQ ID NO. 45) (SEQ ID NO. 46) PRAME GGA TCA GTT GCT CAT CAC ATC CCC CAG GCA C TTC CGA A (SEQ ID NO. 47) (SEQ ID NO. 48) PTHLH CGC CTC AAA AGA GTG AAG GAA GAA GCT GTG TCG TCG CC (SEQ ID NO. 49) (SEQ ID NO. 50) RARRES1 GGC AGT GGA AAA CAG GGA ATT ATT CTA ATG ATG A TCC TGT GTT G (SEQ ID NO. 51) (SEQ ID NO. 52) RGS7BP TCA AGA TGA CAG GAA CCT TCT CTT CAG CCT TCT CCG TCT TC (SEQ ID NO. 53) (SEQ ID NO. 54) SDPR GCT CAT CTT CCA CTC CTC CTT CCC GGA GGA AA TTC CAC (SEQ ID NO. 55) (SEQ ID NO. 56) SERPINA3 GTC TCC CAG GTG GGA GGG TGA TTT GTC CAT AA TGA CTG C (SEQ ID NO. 57) (SEQ ID NO. 58) SLC6A19 GAC CCT GGC TAC CCA GTT CGG GTA GAG GAA T GGA GAT C (SEQ ID NO. 59) (SEQ ID NO. 60) SLC9A3 CCA TCA AGG AGA CTA GCC AGG AAC AAG ACT TGG A TCG ATC C (SEQ ID NO. 61) (SEQ ID NO. 62) TK1 AGA AGG AGG TCG GCC TGA GGC CTT AGG TGA TT CTT GAA G (SEQ ID NO. 63) (SEQ ID NO. 64) TRPM8 CAG AGG AAA TGA CCT TGA GAT CAT GGC ATC G TAA GCT TTG TAT (SEQ ID NO. 65) CC (SEQ ID NO. 66) TSPAN7 GGA ATC GCA TTC CGT CAT TCT TGA TCC CAG T AAG ACT TCT CC (SEQ ID NO. 67) (SEQ ID NO. 68) TUBB3 CAT CTT TGG TCA CAC ATC CAG GAC GAG TGG GG CGA ATC C (SEQ ID NO. 69) (SEQ ID NO. 70) VIPR1 GTA CAC TAC ATC CCA CAA AAC CCT ATG TTC GCC GGA AAG AC (SEQ ID NO. 71) (SEQ ID NO. 72) WT1 GGA AGC ACA CTG CCT TCG TTC ACA GTG AGA A GTC CTT GA (SEQ ID NO. 73) (SEQ ID NO. 74) ACTB* CAA GAT CAT TGC CCA CAT CTG CTG TCC TCC TG GAA GGT G (SEQ ID NO. 75) (SEQ ID NO. 76) RPL13A* CAC TTG GGG ACA GTA ACC CCT TGG GCA TGA G TTG TGC AT (SEQ ID NO. 77) (SEQ ID NO. 78) RPL9* CGG ATG AGA CCA CAA GCT CAA TGT GGT GTT G CAT TTC CTT C (SEQ ID NO. 79) (SEQ ID NO. 80) RPS29* GCT CTT GTC GTG CGT ACT GAC GGA TCT GTT C AAC ACT G (SEQ ID NO: 81) (SEQ ID NO. 82) *Normalization genes v1 and v2 refer to different transcript variants of the same gene for which unique primers were designed.
(63) These 18 genes were quantified in all 55 ccRCCs by qPCR and the correlation of expressions between qPCR and DASL platforms was calculated to determine those genes with the best performance on the qPCR platform, thus embodying the ideal nature of a prognostic gene determined on a distinct platform. 10 genes which were the most significantly correlated (Pearson's correlation p<0.05) in expression between the two platforms were selected (Table 5). Ten genes with the highest correlation were selected as the final prognostic gene set. These included CXCL5, EFNA5, EMCN, IGFBP1, LAMB3, MMP9, PLG, PRAME, RARRES1 and SLC6A19. Univariate logistic correlation analysis for qPCR gene expression with DASL prognosis subtype classification showed that their direction of association was as expected and significant for most prognostic genes (Table 6 and
(64) TABLE-US-00005 TABLE 5 Correlation of expression levels determined by DASL assay and qPCR for 18 prognostic genes. Correlation Gene coefficient t p-value MMP9 0.789638 9.3691 7.81E−13 PRAME 0.708578 7.3105 1.43E−09 CXCL5 0.581672 5.2059 3.2E−06 LAMB3 0.57174 5.0733 5.13E−06 RARRES1 0.518783 4.4178 4.96E−05 PLG 0.472899 3.9073 0.000267 SLC6A19 0.467393 3.849 0.000321 EFNA5 0.373134 2.9279 0.005019 IGFBP1 0.322731 2.4823 0.01625 EMCN 0.267176 2.0184 0.04862 HIST2H3C 0.205662 1.5299 0.132 GFPT2 0.169278 1.2504 0.2166 G6PC 0.104169 0.7625 0.4491 TSPAN7 0.097224 0.7112 0.4801 TK1 0.062664 0.4571 0.6495 MOCOS 0.062524 0.4561 0.6502 KDELR3 0.06103 0.4451 0.658 SDPR −0.08401 −0.6138 0.542
(65) TABLE-US-00006 TABLE 6 Univariate logistic regression analysis for 10 prognostic genes and prognostic subtype assignment. Gene Estimate Std. Error t value Pr (>|t|) SLC6A19 −0.04188 0.00783 −5.348 1.93E−06 EMCN −0.03883 0.009622 −4.036 0.000176 PLG −0.02696 0.008596 −3.137 0.00279 PRAME 0.04576 0.01543 2.966 0.00451 RARRES1 0.05143 0.01989 2.586 0.0125 LAMB3 0.02465 0.01042 2.366 0.0217 CXCL5 0.04105 0.02068 1.985 0.0524 EFNA5 0.0215 0.01154 1.863 0.068 IGFBP1 0.01722 0.01087 1.585 0.119 MMP9 0.01845 0.01234 1.496 0.141
(66) The requirement to develop a prognostic model based on qPCR assays for a limited gene set that can reproduce classification patterns achieved by the DASL expression analysis, dictated that the ten selected prognostic genes are able to reconstruct the sample, distribution of good and poor prognosis subtypes originally produced. Therefore, these ten genes were used for developing a prognostic algorithm using a multivariate logistic correlation analysis for dichotomized outcome corresponding to prognostic subtype assignment (good or poor prognosis) determined by DASL expression analysis. The algorithm based on normalized and scaled Ct values for each gene is summarized in Table 7 with coefficients. The scoring formula is described as the linear combination below:
−0.00534×CXCL5−0.0111×LAMB3+0.001578×MMP9−0.02952×PRAME−0.01001×RARRES1+0.012535×PLG+0.021832×SLC6A19−0.01494×EFNA5+0.014457×IGFBP1+0.020579×EMCN
(67) TABLE-US-00007 TABLE 7 Summary of algorithm coefficients values for qPCR expression of 10 genes derived from the 55 ccRCC screening cohort. Value Gene Coefficients CXCL5 −0.00534 LAMB3 −0.0111 MMP9 0.001578 PRAME −0.02952 RARRES1 −0.01001 PLG 0.012535 SLC6A19 0.021832 EFNA5 −0.01494 IGFBP1 0.014457 EMCN 0.020579 Intercept 0.578108
(68) The scoring algorithm was used to derive a prognostic score for each of the 55 ccRCCs in the screening cohort and a cut-off value for assignment to good or poor subtypes was selected such that misclassification of samples was minimized.
(69) To validate the performance of the ten-gene prognostic algorithm in a separate confirmation cohort of 214 ccRCCs, prognostic scores were calculated based on the qPCR expression values of these ten genes with weighted coefficients determined by multivariate logistic correlation. QPCR expression data was processed in the same manner as it was for the 55 ccRCC screening cohort and Ct values (only those less than 36 considered) were normalized to housekeeping genes and scaled to lie on a positive scale by the addition of a positive constant (14.72). Based on the cutoff determined from the screening cohort, samples were assigned to high- and low-scoring categories corresponding to good and poor prognosis subtypes respectively. Kaplan-Meier analysis showed a significant difference in disease-specific survival between high- and low-scoring groups (
(70) TABLE-US-00008 TABLE 8 Multivariate analysis of 10-gene prediction score and clinicopathologic features Adjusted hazard 95% Adjusted ratio of confidence p-value variable predictor intervals LR test Unadjusted 9.578 3.61-25.4 0.000003527 Age 7.8576 2.89-21.3 7.138E−07 Stage Group 5.857 1.7-20.0 0 Grade 6.181 2.1-17.9 2.513E−07
(71) For a subset of patients of 38 patients from the patient cohort that received TKI treatment in first-, second- or third-line setting, stratification was done based on the prognostic score into high- and low-scoring groups. For two patients, no evaluable clinical response was available. There was a statistically significant difference between number of clinical responders in the two prognostic groups (Table 9) (response rates 51% vs. 100%, p=0.01361 by Fisher's exact test). Further, univariate logistic regression analysis with dichotomized objective response and prognostic score suggested that the score tends to significance (p=0.0535).
(72) TABLE-US-00009 TABLE 9 Clinical response count data for 36 TKI-receiving patients classified into prognosis categories based on 10-gene prognostic score. Prognosis Prognosis Subtype 1 Subtype 2 Non-responders 13 0 Responders 14 9 Fisher's exact test p-value = 0.01361
(73) Determination of Prognosis Based on Expression of Eight Genes—Two-Class Assignment Model
(74) Eighteen selected genes were quantified in all 55 ccRCCs by qPCR and the correlation of expressions between qPCR and DASL platforms was calculated to determine those genes with the best performance on the qPCR platform, thus embodying the ideal nature of a prognostic gene determined on a distinct platform. Univariate logistic correlation analysis for qPCR gene expression with DASL prognosis subtype classification showed that their direction of association was as expected and significant for most prognostic genes (Table 6 and
(75) The linear discriminant analysis (LDA) method was employed to develop a model for assigning prognostic subtype, the categorical dependent variable, based on the linear combination of qPCR expression values of eight genes. The model was trained on qPCR expression data for eight genes in the initial screening cohort of 55 samples for which prognostic class assignment was derived from hierarchical clustering of the DASL dataset. The coefficients of linear discriminants for the eight prognostic genes are in Table 10. The projection of samples onto linear discriminant coordinate is achieved simply by summing up the values of gene expression multiplied by the relevant LDA coefficient as per Table 10. The pre-determined centroids of the two prognostic classes with LDA projection values are 0.781 for good prognosis and 0.228 for poor prognosis. It follows that a sample is assigned to either prognosis group based on the nearness of its LDA projection to either centroid value.
(76) The following description summarizes the method of prognostic class assignment based on expression of eight genes:
(77) ld=[“CXCL5”]*0.01822631+[“EFNA5”]*0.03641317+[“EMCN”]*−0.07433829+[“LAMB3”]*0.05900340+[“PLG”]*−0.05058604+[“PRAME”]*0.13194489+[“RARRES1”]*0.05541879+[“SLC6A 19”]*−0.10545562
(78) If for a given sample, the distance of calculated ld from centroid1 is less than its distance from centroid2 the sample belongs to good prognosis group i.e. if (abs(ld−centroid1)<abs(ld−centroid2)).
(79) If for a given sample distance of calculated ld from centroid2 is less than its distance from centroid1 the sample belongs to poor prognosis group i.e. if (abs(ld−centroid1)>abs(ld−centroid2)).
(80) TABLE-US-00010 TABLE 10 Coefficients of linear discriminants for eight genes in LDA model for ccRCC prognostic subtype assignment Gene Coefficients CXCL5 −0.0182 EFNA5 −0.0364 EMCN 0.0743 LAMB3 −0.0590 PLG 0.0506 PRAME −0.132 RARRES1 −0.055 SLC6A19 0.105
(81) The output from an LDA prediction based on the expression of eight genes can be converted to a continuous score, rather than a discrete two-class assignment. This involves normalization of the output ld, described above, to a scale that lies from 1 to 100. The two extremes (minv and maxv) of the LDA projection are first estimated based on the centroid1=0.781 and centroid2=0.226 as follows:
distance=abs(centroid1−centroid2)
min v=centroid2−(distance/2.0);
max v=centroid1−(distance/2.0);
range=max v−min v
(82) The two ends of the LDA projection are scaled to lie on a linear scale of 1-100 and a particular sample's calculated ld score is converted to a score between 1 and 100 as follows:
score=(int)(((ld−min v)/range)*99.0)+1;
score=min(rk, 100); if score exceeds 100, it is forced to be 100
score=max(rk, 1); if score is less than 1, it is forced to be 1
(83) This is a simple mathematical manipulation of the LDA prediction output that provides a continuous score rather than discrete class assignment.
(84) The performance of the eight-gene prognostic algorithm was validated in a separate cohort of 224 ccRCCs, SGH-224 (
(85) TABLE-US-00011 TABLE 11 Multivariate survival analysis of eight-gene prognostic subtype classifier and clinicopathologic features 95% Adjusted confidence p-value Adjusted variable HR intervals LR test Unadjusted eight- 4.44 2.53-7.81 5.72 × 10.sup.-7 gene prognostic classifier Age 3.96 2.24-6.99 1.12 × 10.sup.-7 Stage 1.89 1.03-3.46 0 Grade 2.86 1.51-5.28 8.85 × 10.sup.-8 HR = hazard ratio; LR = likelihood ratio.
(86) In order to demonstrate its utility and validate it in a multiple-centre, multiple-platform setting, the eight-gene prognostic algorithm was applied to the TCGA-419 dataset. Kaplan-Meier analysis confirmed that survival was significantly different between the prognostic, subtypes (median survival times 91.7 vs. 44.6 months, HR 2.26 CI 1.59-3.21, p=3.04×10.sup.−6 by log rank test) (
(87) For a subset of 48 metastatic RCC patients from SGH-224 cohort receiving tyrosine kinase inhibitor treatment in first-, second- or third-line setting, prognostic classification using eight genes algorithm was done. Characteristics of TKI-receiving patients are available in Table 12. There was a statistically significant difference between number of clinical responders in the two prognostic groups (Table 13) (response rates 67% vs. 100%, p=6.24×10.sup.−4 by Fisher's exact test). Further, univariate logistic regression analysis with clinical benefit as a categorical variable and prognostic class assignment showed a significant correlation between the two (OR 0.429, p=5.Math.96×10.sup.−4).
(88) TABLE-US-00012 TABLE 12 Characteristics of patients receiving TKI treatment from SGH-224 FFPE validation group Number of samples 48 Age Range 43-86 Median 57.5 Gender-n (%) Male 39 (81.3%) Female 9 (18.7%) Stage-n (%) I 14 (8.3%) II 8 (16.6%) III 13 (27.1%) IV 23 (47.9%) Tumour T stage-n (%) 1 5 (10.4%) 2 15 (31.2%) 3 24 (50%) 4 1 (2.1%) X 3 (6.3%) Tumour M stage-n (%) 0 24 (50%) 1 23 (47.9%) X 1 (2.1%) Tumour Grade-n (%) 1 1 (2.1%) 2 15 (31.3%) 3 18 (37.5%) 4 13 (27.1%) Unknown 1 (2.1%) Primary tumour size Range 4-25 Median 9 ECOG Performance 0 20 (41.7%) Status-n (%) 1 13 (27.1%) 1+ 1 (2.1%) 2 1 (2.1%) Unknown 13 (27.1%) Follow-up duration (yr) Range 0.19-12.3 Mean 3.71 Patient Status-n (%) Deaths (cancer- 28 (58.3%) related) Death (other 1 (2.1%) causes) Alive with disease 18 (37.5%) No evidence of disease 1 (2.1%) Post-nephrectomy Sunitinib First line 32 (66.7%) treatment-n (%) Second line 5 (10.4%) Third line 5 (10.4%) Pazopanib First line 5 (10.4%) Second line 1 (2.1%) Previous nephrectomy 36 (75%) MSKCC good 11 (22.9%) risk factors intermediate 21 (43.8%) poor 13 (27.1%) unknown 3 (6.3%)
(89) TABLE-US-00013 TABLE 13 Clinical response count data for 48 TKI- receiving patients classified based on eight-gene prognostic algorithm Good prognosis subtype Poor prognosis subtype Patients with 19 16 clinical benefit Patients with 0 12 progressive disease.sup.† Fisher's exact test p-value = 6.237 × 10.sup.-4 .sup.†Patients with progressive disease were those for whom disease progressed as evaluated radiologically, after initiating TKI treatment. Patients with clinical benefit were those that showed partial response to TKI therapy or had stable disease-after initiating TKI therapy, as evaluated radiologically. One patient had non-evaluable response data.
(90) A practical molecular assay has been developed that is capable of stratifying ccRCC patients into prognostic groups that are essentially manifestations of the underlying biological heterogeneity of ccRCC. The prognostic assay also functions well as a predictive assay for clinical response to TKI therapy, suggesting a fundamental overlap between patients with good prognosis and those likely to respond to therapy upon relapse. This assay therefore can delineate an aggressive subtype of the disease, which patients prognosed with, face worse survival outcomes and are also unlikely to benefit from TKI therapy.
(91) Previous efforts to develop a multigene assay for prognostic classification of ccRCC have identified 16 genes, expression levels of which were associated with recurrence-free interval. Of note, in this previous study, prognostic genes were selected from a predetermined set of 732 genes, based on their function and individual association with recurrence-free interval. In contrast, in this study the choice of prognostic genes was driven by the identification of gene expression-based subtypes of ccRCC, without the introduction of previous knowledge of gene function or clinical outcome. With this truly unbiased approach to gene selection unraveling the latent differences among ccRCC, the selected prognostic genes are more likely to continue to embody these differences even in smaller numbers as in a multigene assay.
(92) The choice of FFPE material to develop and confirm the prognostic assay allows greater scope for validation on a large-scale given the relative abundance of such material. Several studies in the past have molecularly profiled ccRCCs to identify subtypes with differing outcomes. However, all have made use of fresh-frozen material limiting the validation of identified prognostic gene signatures.
(93) The platform used for initial screening of potential prognostic genes was the whole-genome high-throughput Illumina's DASL analysis but the final platform qPCR was chosen to confirm and validate the prognostic assay. Although the success rate of translation to the qPCR platform from the DASL platform was relatively low (10 out of 37 genes tested=27%), the ability of the final selected gene set to discriminate prognostic groups in the validation group using the qPCR platform reinforces the platform-independence of the assay. The qPCR platform is preferable for a prognostic assay as it is practical, inexpensive, flexible and readily transferable to other laboratories.
(94) The 10 genes finally selected to be included in the prognostic assay, CXCL5, EFNA5, EMCN, IGFBP1, LAMB3, MMP9, PLG, PRAME, RARRES1 and SLC6A19, represent genes from the chemokine signaling, migration and invasion, angiogenesis, growth factor signaling, ECM-interacting, retinoic acid signaling and transporter families. Given the non-predeterministic manner in which they were selected the wide variety of cellular functions encompassed by the prognostic genes is not surprising. Notwithstanding, the biological subtypes of ccRCC likely differ in the pathway functions described here.
(95) The prognostic algorithm developed based on 10 genes was initially applied to the 55 ccRCCs in the screening cohort that also served as the confirmation cohort. The purpose was to identify a suitable cutoff for the score such that the best discrimination of prognostic groups was achieved. The cutoff chosen represents the score at which reasonably high level of concordance of sample assignment to class was achieved by both DASL and qPCR platforms. This cutoff score may be subject to refinement as more samples from independent study centres are analyzed with this assay for prognostic classification.
(96) An additional utility of the 10-gene prognostic assay was the ability to predict benefit from TKI therapy, measured as objective response assessed by tumour shrinkage. A prognostic assay can serve as a predictive assay, if the prognostic subtypes represent biological subtypes of the disease with differential disease progression and accompanying susceptibility to therapeutic intervention. Previous attempts to identify predictive biomarkers for TKI therapy for ccRCC, which is the most common therapy administered presently have looked at baseline levels of cytokine and angiogenic factors in plasma and serum, levels of factors involved in VEGF signaling pathways in plasma, immunohistochemical analyses of HIF-1α, CAIX, PTEN and p21 in primary tumours. Association of genetic polymorphisms with response to sunitinib has been extensively analyzed for targets of sunitinib (VEGFRs, PDGFRs) and drug metabolism genes and with pazopanib for angiogenesis- and exposure-related genes. Most of these studies looked at the progression-free survival times after drug administration, but did not address the objective response criteria, which is more relevant in the clinical setting for decision-making. Progression-free survival time can indicate association of drug-response with specific factors but may ultimately arise from different underlying ccRCC biology, which is what was sought to be identified in this study.
(97) This simple 10-gene based assay that can run on FFPE material of primary and metastatic tumours can thus define subsets of biologically distinct ccRCCs. Using a prognostic risk scoring system, patients can be stratified to an aggressive subtype which are likely to face poor clinical outcome in the localized setting and also less likely to benefit from TKI therapy in the metastatic setting.