More Related Content
What's hot
What's hot (20)
Viewers also liked
Viewers also liked (7)
Similar to Presentatie maastricht
Similar to Presentatie maastricht (20)
Presentatie maastricht
- 1. Discovering novel biomarkers in breast cancer Rianne Fijten
- 2. Breast cancer subtypes • ER status • PR status • HER2 status • EGFR status
- 4. Copy number alterations • Alterations in allele Normal Deleted Amplified number • Can affect gene expression • In cancer: selection for regions containing oncogenes or tumour suppressors
- 5. Public data: in vitro • NCI60: 59 cell lines for 9 tumour types • CCLE: 947 cell lines for 36 tumour types mRNA Mutation Drug CN Proteome Metabolome expression status sensitivity
- 6. Public data: in vivo • Mostly transcriptome data – Microarray – SNP array – RNA sequencing • Very dependent on availability of datasets and quality of data
- 7. Research question Can we use copy number data as a starting point for exploratory biomarker discovery in breast cancer?
- 8. Identification of amplified genes • CCLE: Cancer Cell Line Encyclopedia – 56 breast cancer cell lines • Amplification CN > 3
- 9. Copy numbers in breast cancer Amplification CN > 3 Deletion CN < 1
- 10. Correlation with mRNA expression • Significant and valid correlation with mRNA values
- 11. Result: 30 genes – 4 amplified regions
- 12. In vivo validation – Array Express Gene Average CN Gene Average CN • 900+ early stage breast ERBB2 TRMT12 4.401254 3.039269 UTP23 DSCC1 2.510294 2.45963 cancer tumour samples NSMCE2 2.833896 VAPB 2.451618 RAD21 2.805559 AURKA 2.408208 • Copy number data KIAA0196 2.75552 RAB22A 2.399712 SQLE 2.70128 STX16 2.371037 RNF139 2.667317 RAE1 2.281577 TAF2 2.650698 CSTF1 2.258596 EIF3H 2.64878 GRB7 - WDR67 2.625884 MRPL13 - DERL1 2.625829 PGAP3 - ATAD2 2.608554 RBM38 - NDUFB9 2.587144 STARD3 - C20orf43 2.550113 TCAP - TMEM65 2.52641 WDYHV1 -
- 13. In vivo validation – The Cancer Genome Atlas (TCGA) • 900+ samples of patient invasive breast carcinomas • Copy number + mRNA data
- 14. In vivo validation – The Cancer Genome Atlas (TCGA) Gene Ratio CN Ratio mRNA Correlation Gene Ratio CN Ratio mRNA Correlation ATAD2 0.43 0.162 0.690 * RAD21 0.419 0.157 0.850 * AURKA 0.215 0.168 0.580 * RAE1 0.215 0.092 0.920 * C20ORF43 0.237 0.124 0.840 * RBM38 0.215 0.157 0.520 * CSTF1 0.215 0.108 0.900 * RNF139 0.441 0.135 0.760 * DERL1 0.419 0.162 0.840 * SQLE 0.43 0.157 0.740 * DSCC1 0.419 0.173 0.750 * STARD3 0.194 0.119 0.940 * EIF3H 0.43 0.135 0.600 * STX16 0.226 0.119 0.690 * ERBB2 0.194 0.135 0.840 * TAF2 0.419 0.168 0.830 * GRB7 0.183 0.135 0.880 * TCAP 0.194 0.141 0.720 * KIAA0196 0.441 0.173 0.720 * TMEM65 - - - MRPL13 0.419 0.168 0.870 * TRMT12 0.441 0.157 0.780 * NDUFB9 0.43 0.157 0.870 * UTP23 0.419 0.157 0.790 * NSMCE2 - - - VAPB 0.226 0.119 0.850 * PGAP3 0.194 0.135 0.870 * WDR67 0.419 0.189 0.660 * RAB22A 0.226 0.097 0.860 * WDYHV1 0.43 0.135 0.730 *
- 15. Survival analysis - KMPlot • 3000 breast cancer patient samples incl. survival data • Compare survival between patients with high and low gene expression
- 16. Survival analysis - KMPlot • Significant differences in 11 of 26 genes RNF139 DERL1 STARD3
- 17. Conclusions Results obtained in vitro were validated using in vivo datasets Some show differences in breast cancer patient survival
- 18. Of interest: SQLE Steroid degradation Cholesterol Steroid hormone biosynthesis
- 19. SQLE • CCLE – Allele copies: 3.084107 – Significant mRNA correlation • TCGA – CN affected: 43% – mRNA affected: 18.5% – Significant correlation • AE – Allele copies: 2.70128 • Survival
- 20. SQLE Extended survival analysis ER positive Luminal A (ER+ and low grade) Other receptor/molecular subtypes showed no significant difference
- 21. Cholesterol and cancer • Patients with cancer have abnormal levels of HDL– and LDL-cholesterol • Transformed cells and tumors exhibit abnormal regulation of LDL-R and HMG-CoA Reductase. • Transformed cells may require or utilize more cholesterol than normal cells, and this may be associated with their increased rate of proliferation. Llaverias, G.; Danilo, C.; Mercier, I.; Daumer, K.; Capozza, F.; Williams, T. M.; Sotgia, F.; Lisanti, M. P.; Frank, P. G. Role of cholesterol in the development and progression of breast cancer. The American journal of pathology 2011, 178, 402–12.
- 22. Cholesterol and cancer • Inhibition of squalene synthase decrease proliferation in prostate cell line 1 • inhibition of most enzymes involved in cholesterol biosynthesis from lanosterol results in cell proliferation inhibition 2 1. Fukuma, Y.; Matsui, H.; Koike, H.; Sekine, Y.; Shechter, I.; Ohtake, N.; Nakata, S.; Ito, K.; Suzuki, K. Role of squalene synthase in prostate cancer risk and the biological aggressiveness of human prostate cancer. Prostate cancer and prostatic diseases 2012, 15, 339–45. 2. Lasunción, M. A.; Martín-Sánchez, C.; Canfrán-Duque, A.; Busto, R. Post-lanosterol biosynthesis of cholesterol and cancer. Current opinion in pharmacology 2012, 12, 717–23.
- 23. Conclusion SQLE or the entire cholesterol pathway may play a crucial role in ER+ breast cancer
- 24. Systems Biology approach • Survival analysis for cholesterol pathway genes (alone and groups) • Create biomarker profile containing SQLE and other genes • Survival analysis for all genes in amplified regions • Amplified miRNAs
Editor's Notes
- To just make clear what it is about, first some background. Copy number alterations are alterations in the number of alleles of a gene. Having more or less than 2 alleles can affect gene expression of that specific gene. In cancer, this can play a big role, especially when amplifying oncogenes or deleting tumor suppressor genes. Deletion: <2 alleles (either 1 or 0). Amplification: >2, so can be 3 or even 10.
- A lot of data is available online, and does not need to be generated independently in the lab. Especially a lot of info is available for cell lines. The NCI60 and CCLE are examples of this. Both have various datasets available including copy number data, mRNA expression data and drug sensitivity. This allows scientists to easily look at different aspects of cancer processes in these cell lines. NCI60 has all of the data displayed in the figure. CCLE has less data, but has not been around for very long, and will probably expand in the future.
- In vitro results do not always ensure that the same thing is happening in vivo. Therefore, it would be good to be able to use public in vivo data too. Some databases I’ve used are displayed on the bottom. Most databases only contain transcriptomic data, but other datasets are gradually becoming available more. However, as these datasets were not generated in a uniform format and by the same people, the quality is not always as good as you would like.
- The research question for my project
- Because amplifications are more easily treated than deletions, I focussed on amplifications in this project. In order to identify amplified genes, I used the copy number data of the CCLE database. As said, it contains 947 cell lines for 36 tumour types. For breast cancer, it has 56 different cell lines available. Based on literature, the cut-off for amplifications was set on a copy number > 3.
- This is an overview of the average copy number values for all breast cancer cell lines in the CCLE database. They are sorted by chromosomal location and each color represents a different chromosome. Some regions on chromosome 8, 17 and 20 go into the red area, indicating amplified regions. In addition, some genes on other chromosomes show amplifications, but are not part of regions.
- Copy number alterations do not always result in similar alterations in mRNA expression of a gene. I therefore integrated the mRNA expression with the copy number data and only took the genes that had significant and valid correlations between the two. A valid correlation meant that the regression line in the correlation figure actually showed the right trend. See figure.
- The result of the amplification cut-off and correlation with mRNA expression was a list of 30 genes. These can from 4 amplified regions on chromosome 8, 17 and 20.
- As in vitro and in vivo can differ a lot, in vivo validation of these amplified regions is required. Therefore, I downloaded a dataset from Array Express containing 900+ early stage breast cancer tumour samples for which the copy numbers were profiled. I looked at the average copy numbers for each of the genes in the list. As seen in the table, only two of the genes have CN values above 3, but the others do show a gain of alleles, if not an amplification (all have CN> 2.25). This indicates that these amplification regions are also (to some extent) present in vivo.
- Another dataset used to validate the in vitro findings. This is a database again containing 900+ breast cancer tumour samples, now of invasive carcinomas. Two types of data were available: CN and mRNA expression. This gave me the opportunity to replicate what I did with the CCLE data (find amplifications and correlate with mRNA.
- These are the results. The cut-off criteria I used for this analysis were CN>3 and mRNA expression fold change >2 compared to controls. The ratio values indicate the percentage of patients that met the criteria. So 0.43 means that 43% of the patients had a CN > 3 for that specific gene. The correlation coefficient is also given and an asterisk indicates a significant p-value (<0.05). This data shows that a substantial percentage of patients show amplification and overexpression of these genes. This thus validates the in vitro findings.
- Some of the genes may potentially modulate survival outcomes in patients. Therefore, I used a database called KMPlot which contains 3000+ breast cancer patients that have been genotyped and have survival data available. It will enable the generation of Kaplan meier plots, where outcomes of patients with high gene expression vs. low gene expression can be visualized.
- Out of the 30 genes, 26 were in the database. Of those 26, 11 showed significant differences between patients with high expression vs. low expression. Some examples are shown in the figures.
- Conclusions: in vitro findings seem to be present in vivo too. Some of the genes show significant differences in breast cancer patient survival outcomes too.
- A gene of particular interest: SQLE, squaleneepoxidase/squalenemonooxygenase. It converts squalene to 2,3-oxidosqualene, a step in the cholesterol biosynthesis pathway
- A summary of the results obtained from the previously explained steps
- I wanted to know whether there was a breast cancer subtype that might be the driver behind the poor survival outcome of patients with high SQLE expression. I found that this driver is very likely the ER+ status of some tumours. The figure shows the two subtypes that were significant. The rest of the subtypes was not (ER-, PR+/-, HER2+, basal, Luminal B)
- Some literature research indicating the importance of cholesterol
- Inhibition of other genes in the pathway reduced cell proliferation. Indication that inhibition of SQLE may be a potential breast cancer treatment
- As I don’t have enough time, I will stop here. I have done many other things in addition to this, but have mainly been focussing on a more Systems Biology approach of the entire cholesterol pathway. Currently I am testing a biomarker profile I set up containing four genes (including SQLE) that all influence survival outcome in patients with ER+ tumours. In addition, I have done survival analysis for the entire amplified regions, instead of just one gene.