![2009-2010 Bill 3418: Voting identification - South … · Web view[3418]1 South Carolina General Assembly 118th Session, 2009-2010 H. 3418 STATUS INFORMATION General Bill Sponsors:](https://static.fdocuments.us/doc/165x107/5ae5700d7f8b9aee078b8eb4/2009-2010-bill-3418-voting-identification-south-view34181-south-carolina.jpg)
#3418. Cross Comparison of Targeted Gene Expression ......Colorectal cancer (CRC) is one of the...
1
#3418. Cross Comparison of Targeted Gene Expression Technologies for Patient Stratification Raghavee Venkatramanan 1 , Spencer Chee 1 , Ryan Fleming 1 , Tuuli Saloranta 1 , Inah Golez 1 , Elliot Swanson 1 , Kimberly Kruse 1 , Vickie Satele 1 , Saman Tahir 1 , Sally Dow 1 , Evan Anderson 1 , Bri Hudson 1 , Kerry Deutsch 1 , Steve Anderson 2 , Fang Yin Lo 1 , Corey Braastad 1 and Anup Madan 1 ; 1 Covance Inc., Seattle, WA; 2 Laboratory Corporation of America ® Holdings, Research Triangle Park, NC Presented at AACR 2018 Introduction Colorectal cancer (CRC) is one of the major causes of global cancer mortality. Until recently, KRAS has been the only predictive biomarker for anti-EGFR therapy for metastatic CRC, and yet predicting prognosis in clinical practice is still poor. Therefore, a more accurate method for prognosis of CRC patients is needed. Gene expression profiling has shown great promise in predicting prognosis of individual patients in diverse cancers. The development of RNA-sequencing has greatly facilitated identification of biomarkers that can be used to stratify patients for targeted therapies. Despite the decrease in cost of sequencing in last few years, the time and the resources needed for analysis limits its use in clinical trials for patient selection. Targeted gene expression technologies like qPCR and NanoString enable highly customizable assays that can be conveniently performed for patient recruitment. The aim of this study was to investigate potential alternatives for gene profiling using a novel NanoString PlexSet technology. The PlexSet system comes with pre-packaged and custom code sets in identifying genetic markers. Up to 8 samples can be pooled to each nCounter cartridge lane, enabling a total of 96 samples per run, thus lowering cost. For this study, gene expression signature was developed using RNA-Seq data where we have profiled 74 CRC samples, 20 of which have matching normal samples. A RAS signature score based on expression profile was calculated for each sample. In order to look for potential gene signatures, differential gene expression analysis was performed between the following groups: (a) samples with high versus those with low RAS signature scores in the 54 CRC (b) KRAS mutant versus wildtype samples and (c) tumor versus normal samples in the clinical study. We hypothesized that our genes of interest is most likely significantly differentially expressed in one of these groups. The count of significantly expressed gene for the groups (a-c) are 1560, 34 and 24, respectively. Therefore, significantly differentially expressed genes between groups were selected and ranked based on frequency of occurrence. These genes of interest are being analyzed using NanoString PlexSet to evaluate the potential of using NanoString PlexSet system for targeted gene expression profiling. Results of these analyses are presented here. Conclusion This study shows that the newly developed NanoString PlexSet technology can be a good choice as a potential alternative for gene profiling. Leveraging on previous data, we selected genes that were expressed significantly to build a panel and test for reproducibility. The standard RNA Seq data was further compared to Nanostring, HTG EdgeSeq Oncology Biomarker panel data and were seen to have high correlation scores across platforms. NanoString PlexSet data and RNA Sequencing data for 32 samples had high correlation for the genes compared. We are in process of designing TLDA card to compare other low throughput technologies. References 1. Smith, Gillian, et al. “Mutations in APC, Kirsten-ras, and p53—alternative genetic pathways to colorectal cancer.” Proceedings of the National Academy of Sciences 99.14 (2002): 9433-9438. 2. Omolo, Bernard, et al. “Adaptation of a RAS pathway activation signature from FF to FFPE tissues in colorectal cancer.” BMC Medical Genomics 9.1 (2016): 65. 3. NanoString Product Bulletin – PlexSet™ Figure 1. Model of the genetic changes required for progression of adenoma to carcinoma in the development of colorectal cancer. 1 Based on recent studies, RAS pathway has been identified to predict response to targeted therapies. 2 RNA Sequencing experiments were conducted previously to calculate and stratify patient samples based on the RAS scores and this study has the advantage of leveraging the RNASeq data from those previous studies. Figure 5. NanoString PlexSet technology. A) NanoString PlexSet technology, showing the steps involved in generating data. 3 B) The PlexSet technology probe design includes both target-specific and tag-specific sequences. These can be modified by reassigning reporter tags to new targets. In the above diagram. Probes A and B can be replaced with new oligos enabling multiplexing. At present, 96 custom probes across 96 samples can be multiplexed to get 99216 data points per run. C) Initial titration tests were performed to select suitable input volume. 221ng input was selected for a binding density of 1 and 300ng of input was selected to get a binding density of 1.6, which was preferred and run in the final experiment. Figure 3. Result correlation plots for sequencing and standard NanoString technology. Sequencing technology has been the global standard for analyzing aberration in genomic expression and mutation data. Preliminary analysis was performed between RNA Seq, HTG panel and NanoString data on 32 samples across 461 common genes. RNASeq data was de-multiplexed, trimmed and aligned to the reference genome followed by calculating gene counts. NanoString data was quality controlled, normalized for technical variations and Housekeeping genes using nSolver 3.0 suite to get the final counts data. HTG counts data was obtained from the manufacturer. Scripts using R v3.4 software was used to preform Pearson correlation analysis. Correlation coefficients were plot in a heat map for visualization. Both such comparison shows high correlation amongst samples. Figure 7. Result correlation plots for NanoString technologies against RNASeq data. Raw counts of the NanoString PlexSet data was de-multiplexed, quality controlled and normalized for technical variations only using nSolver 4.0 suite. The samples were subjected to Background subtraction and positive control normalization. These counts were correlated using Pearson correlation against raw counts from RNASeq data compiled using feature counts software, to plot the correlation coefficients. It is noted that all but 5 samples are highly correlated with Pearson average R 2 value of 0.814 (excluding the 5 samples that did show a positive trend). With a diverse set of genes, these expression value tend to differ. Both NanoString platforms show high correlation with RNA Seq data. Figure 2. Exploring potential biomarker genes. Describes the steps involved in selecting the genes that were used to design the NanoString PlexSet panel. A mix of 155 CRC tumor and normal samples were selected and sequenced to examine three separate comparisons. RAS scores were calculated and samples were stratified into three different groups as Low, Medium and High. Along with this, tumor vs normal pairs and RAS mutant vs wildtype pairs were identified to perform differential gene expression analysis. Differentially expressed genes were identified after each analysis and were ranked to calculate the frequency. Frequently occurred gene from these 5 comparisons were selected along with other interesting genes identified form pathway analysis to create a mixed bag of assays to be designed in the panel. Selected Genes as Poten�al Markers for Pa�ent Stra�fica�on Using PlexSet Panel AGBL2 CDH9 LOC100507053 OBP2B RTN2 ZNF257 ALK CELF3 LOC285441 OLFM3 SLC14A1 AOAH DCBLD2 MANBA PLCZ1 SMTNL2 ABCF1 ATP6V1C2 FAM149A MANSC1 PPBP SNORD115-44 ALAS1 C1orf173 IL7R MBTPS2 REN SPACA7 G6PD C6orf10 KRT6B MED23 REST TTLL4 GUSB CALML3 LCN2 MEPE RGPD3 VTA1 POLR2A CAPN6 LEMD1 MYO18B RPS16 XIRP2 SDHA Figure 4. Endogenous and housekeeping genes selected. The table gives the list of endogenous genes and housekeepers (highlighted in red) that were selected to design the target and reporter of the PlexSet Assay. A total of 41 endogenous genes with 6 housekeeping genes were used to design the custom PlexSet panel. Figure 6. Comparing universal human reference (UHR). Correlation plot of UHR samples that were run at different time points to see the reproducibility of the assay and the technology used. All four runs had high correlation with average R 2 value of 0.99. Figure 8. Result Correlation plots for PlexSet data and NanoString data for UHR (Universal human reference). Normalized NanoString PlexSet data and normalized NanoString data from regular NanoString IO360 panels that had some overlapping genes were analyzed. Cell line samples had a very high correlation with an average R 2 of 0.88 across two different NanoString platforms as opposed to FFPE samples.
Transcript of #3418. Cross Comparison of Targeted Gene Expression ......Colorectal cancer (CRC) is one of the...
#3418. Cross Comparison of Targeted Gene Expression Technologies for Patient StratificationRaghavee Venkatramanan1, Spencer Chee1, Ryan Fleming1, Tuuli Saloranta1, Inah Golez1, Elliot Swanson1, Kimberly Kruse1, Vickie Satele1, Saman Tahir1, Sally Dow1, Evan Anderson1, Bri Hudson1, Kerry Deutsch1, Steve Anderson2, Fang Yin Lo1, Corey Braastad1 and Anup Madan1; 1Covance Inc., Seattle, WA; 2Laboratory Corporation of America® Holdings, Research Triangle Park, NC
Presented at AACR 2018
IntroductionColorectal cancer (CRC) is one of the major causes of global cancer mortality. Until recently, KRAS has been the only predictive biomarker for anti-EGFR therapy for metastatic CRC, and yet predicting prognosis in clinical practice is still poor. Therefore, a more accurate method for prognosis of CRC patients is needed.
Gene expression profiling has shown great promise in predicting prognosis of individual patients in diverse cancers. The development of RNA-sequencing has greatly facilitated identification of biomarkers that can be used to stratify patients for targeted therapies. Despite the decrease in cost of sequencing in last few years, the time and the resources needed for analysis limits its use in clinical trials for patient selection. Targeted gene expression technologies like qPCR and NanoString enable highly customizable assays that can be conveniently performed for patient recruitment.
The aim of this study was to investigate potential alternatives for gene profiling using a novel NanoString PlexSet technology. The PlexSet system comes with pre-packaged and custom code sets in identifying genetic markers. Up to 8 samples can be pooled to each nCounter cartridge lane, enabling a total of 96 samples per run, thus lowering cost.
For this study, gene expression signature was developed using RNA-Seq data where we have profiled 74 CRC samples, 20 of which have matching normal samples. A RAS signature score based on expression profile was calculated for each sample. In order to look for potential gene signatures, differential gene expression analysis was performed between the following groups: (a) samples with high versus those with low RAS signature scores in the 54 CRC (b) KRAS mutant versus wildtype samples and (c) tumor versus normal samples in the clinical study.
We hypothesized that our genes of interest is most likely significantly differentially expressed in one of these groups. The count of significantly expressed gene for the groups (a-c) are 1560, 34 and 24, respectively. Therefore, significantly differentially expressed genes between groups were selected and ranked based on frequency of occurrence. These genes of interest are being analyzed using NanoString PlexSet to evaluate the potential of using NanoString PlexSet system for targeted gene expression profiling. Results of these analyses are presented here.
ConclusionThis study shows that the newly developed NanoString PlexSet technology can be a good choice as a potential alternative for gene profiling. Leveraging on previous data, we selected genes that were expressed significantly to build a panel and test for reproducibility. The standard RNA Seq data was further compared to Nanostring, HTG EdgeSeq Oncology Biomarker panel data and were seen to have high correlation scores across platforms.
NanoString PlexSet data and RNA Sequencing data for 32 samples had high correlation for the genes compared. We are in process of designing TLDA card to compare other low throughput technologies.
References1. Smith, Gillian, et al. “Mutations in APC, Kirsten-ras, and p53—alternative genetic pathways to
colorectal cancer.” Proceedings of the National Academy of Sciences 99.14 (2002): 9433-9438.2. Omolo, Bernard, et al. “Adaptation of a RAS pathway activation signature from FF to FFPE tissues
in colorectal cancer.” BMC Medical Genomics 9.1 (2016): 65.3. NanoString Product Bulletin – PlexSet™
Figure 1. Model of the genetic changes required for progression of adenoma to carcinoma in the development of colorectal cancer.1 Based on recent studies, RAS pathway has been identified to predict response to targeted therapies.2 RNA Sequencing experiments were conducted previously to calculate and stratify patient samples based on the RAS scores and this study has the advantage of leveraging the RNASeq data from those previous studies.
Figure 5. NanoString PlexSet technology. A) NanoString PlexSet technology, showing the steps involved in generating data.3B) The PlexSet technology probe design includes both target-specific and tag-specific sequences. These can be modified by reassigning reporter tags to new targets. In the above diagram. Probes A and B can be replaced with new oligos enabling multiplexing. At present, 96 custom probes across 96 samples can be multiplexed to get 99216 data points per run. C) Initial titration tests were performed to select suitable input volume. 221ng input was selected for a binding density of 1 and 300ng of input was selected to get a binding density of 1.6, which was preferred and run in the final experiment.
Figure 3. Result correlation plots for sequencing and standard NanoString technology. Sequencing technology has been the global standard for analyzing aberration in genomic expression and mutation data. Preliminary analysis was performed between RNA Seq, HTG panel and NanoString data on 32 samples across 461 common genes. RNASeq data was de-multiplexed, trimmed and aligned to the reference genome followed by calculating gene counts. NanoString data was quality controlled, normalized for technical variations and Housekeeping genes using nSolver 3.0 suite to get the final counts data. HTG counts data was obtained from the manufacturer. Scripts using R v3.4 software was used to preform Pearson correlation analysis. Correlation coefficients were plot in a heat map for visualization. Both such comparison shows high correlation amongst samples.
Figure 7. Result correlation plots for NanoString technologies against RNASeq data. Raw counts of the NanoString PlexSet data was de-multiplexed, quality controlled and normalized for technical variations only using nSolver 4.0 suite. The samples were subjected to Background subtraction and positive control normalization. These counts were correlated using Pearson correlation against raw counts from RNASeq data compiled using feature counts software, to plot the correlation coefficients. It is noted that all but 5 samples are highly correlated with Pearson average R2 value of 0.814 (excluding the 5 samples that did show a positive trend). With a diverse set of genes, these expression value tend to differ. Both NanoString platforms show high correlation with RNA Seq data.
Figure 2. Exploring potential biomarker genes. Describes the steps involved in selecting the genes that were used to design the NanoString PlexSet panel. A mix of 155 CRC tumor and normal samples were selected and sequenced to examine three separate comparisons. RAS scores were calculated and samples were stratified into three different groups as Low, Medium and High. Along with this, tumor vs normal pairs and RAS mutant vs wildtype pairs were identified to perform differential gene expression analysis. Differentially expressed genes were identified after each analysis and were ranked to calculate the frequency. Frequently occurred gene from these 5 comparisons were selected along with other interesting genes identified form pathway analysis to create a mixed bag of assays to be designed in the panel.
Selected Genes as Poten�al Markers for Pa�ent Stra�fica�on Using PlexSet Panel
AGBL2 CDH9 LOC100507053 OBP2B RTN2 ZNF257ALK CELF3 LOC285441 OLFM3 SLC14A1AOAH DCBLD2 MANBA PLCZ1 SMTNL2 ABCF1ATP6V1C2 FAM149A MANSC1 PPBP SNORD115-44 ALAS1C1orf173 IL7R MBTPS2 REN SPACA7 G6PDC6orf10 KRT6B MED23 REST TTLL4 GUSBCALML3 LCN2 MEPE RGPD3 VTA1 POLR2ACAPN6 LEMD1 MYO18B RPS16 XIRP2 SDHA
Figure 4. Endogenous and housekeeping genes selected. The table gives the list of endogenous genes and housekeepers (highlighted in red) that were selected to design the target and reporter of the PlexSet Assay. A total of 41 endogenous genes with 6 housekeeping genes were used to design the custom PlexSet panel.
Figure 6. Comparing universal human reference (UHR). Correlation plot of UHR samples that were run at different time points to see the reproducibility of the assay and the technology used. All four runs had high correlation with average R2 value of 0.99.
Figure 8. Result Correlation plots for PlexSet data and NanoString data for UHR (Universal human reference). Normalized NanoString PlexSet data and normalized NanoString data from regular NanoString IO360 panels that had some overlapping genes were analyzed. Cell line samples had a very high correlation with an average R2 of 0.88 across two different NanoString platforms as opposed to FFPE samples.
