Characterization and visualization of compound combination responses in a high throughout setting
1 like898 views
This document summarizes a study characterizing drug combination responses in a high-throughput screening setting. It describes the workflow used, which involves running single agent dose responses, then 6x6 matrices to identify potential synergies followed by 10x10 matrices for confirmation. Methods for visualizing and reporting the combination results are discussed, including heatmaps and clustering response surfaces. The primary focus of the study is investigating combinations of Ibrutinib, a Btk inhibitor, for treating diffuse large B-cell lymphoma.
Characterization and visualization of compound combination responses in a high throughout setting
- 1. Characteriza*on and visualiza*on of compound combina*on responses in a high throughout se8ng Rajarshi Guha, Lesley Mathews, John Keller, Paul Shinn, Craig Thomas, Anton Simeonov, Marc Ferrar NIH-‐NCATS April 7, 2013, New Orleans
- 2. Outline Why combine? Physical infrastructure & workflow Summarizing and exploring the data hRp://origin.arstechnica.com/news.media/pills-‐4.jpg
- 3. Screening for Novel Drug Combina*ons • Drug combina*ons offer advantages for both efficacy and poten*al reduc*on of target related toxici*es • Combina*on studies also offer insight into systems level interac*ons
- 4. How to Test Combina*ons • Many procedures described in the literature – Fixed dose ra*o (aka ray) – Ray contour C5,D5 C5 – Checkerboard C4,D4 C4 – Gene*c algorithm C3,D3 C3 C2,D2 C2 C1,D5 C1,D4 C1,D3 C1,D2 C1,D1 C1 D5 D4 D3 D2 D1 0
- 5. Scaling Response Surface Screening 5e+07 Combination type • Response surfaces All pairs Fixed library Dose matrix size 4e+07 imply a DxD matrix 4 Number of combinations 6 10 for each combina*on 3e+07 • All pairs screening is 2e+07 imprac*cal for more 1e+07 than tens of 0e+00 compounds 250 500 750 Number of compounds 1000 • Instead we consider N compounds versus a fixed size library
- 6. Mechanism Interroga*on PlateE Top 10 Panther gene classes Top 10 Panther gene classes 200 kinase nucleic acid binding Number of compounds 150 receptor signaling molecule transferase 100 50 Top 10 enriched GeneGo pathway maps Development EGFR signaling pathway 0 Some pathways of EMT in cancer cells &D I II III ed al t Development VEGF signaling via VEGFR2 - generic cascades d en e ue e ic as e ov R as lim lin as in Ph pr Ph ec nt Ph pp Ap Apoptosis and survival Anti-apoptotic action of Gastrin co Pr Su is D Cell adhesion Chemokines and adhesion Cytoskeleton remodeling TGF, WNT and cytoskeletal remodeling Regulation of lipid metabolism RXR-dependent regulation of lipid metabolism via PPAR, RAR and VDR Transcription PPAR Pathway Translation Non-genomic (rapid) action of Androgen Receptor Development VEGF signaling and activation 0 5 10 15 -log10(pValue)
- 7. Combina*on Screening Workflow Run single agent dose responses 6x6 matrices for poten5al synergies 10x10 for confirma5on + self-‐cross Acoustic dispense, 15 min for 1260 wells, 14 min for 1200 wells"
- 10. Repor*ng Combina*on Results • These web pages and matrix layouts are a useful first step • Does not scale as we grow MIPE • S*ll need to do a beRer job of ranking and aggrega*ng combina*on responses taking into account – Response matrix – Compounds, targets and pathways
- 11. A Simpler Visual Summary • Convert mul*ple individual 1 7 13 19 25 31 heatmaps, to a single heatmap 2 3 8 9 14 15 20 21 26 27 32 33 by unrolling response matrices 4 10 16 22 28 34 • Examine effects of A at fixed 5 6 11 12 17 18 23 24 29 30 35 36 concentra*ons, on dose response of B {1, 2, 3, 4, …, 34, 35, 36} • Zoom in on combina*ons that show extensive ac*vity throughout the dose matrix
- 12. A Simpler Visual Summary 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 Concentration Combination
- 13. When are Combina*ons Similar? • Differences and their aggregates such as RMSD can lead to degeneracy 0.06 • Instead we’re interested in 0.04 0.010 the shape of the surface 0.005 0.02 0.00 0.000 0 25 50 75 100 0 25 50 75 100 • How to characterize shape? 0.15 – Parametrized fits 0.10 0.05 – Distribu*on of responses 0.00 0 50 100 D, p value
- 14. Similarity via the KS Test • Quan*fy distance between response distribu*ons via KS test – If p-‐value > 0.05, we assume 9 distance is 0 • But ignores the spa5al density 6 distribu*on of the responses 3 on the concentra*on grid 0 0.00 0.25 0.50 0.75 1.00 D
- 15. Similarity via the Syrjala Test • Syrjala test used to compare 10.0 popula*on distribu*ons over a spa*al grid 7.5 – Invariant to grid orienta*on density 5.0 – Provides an empirical p-‐value 2.5 • Less degenerate than just considering 1D distribu*ons 0.0 0.00 0.25 0.50 0.75 D Syrjala, S.E., “A Sta*s*cal Test for a Difference between the Spa*al Distribu*ons of Two Popula*ons”, Ecology, 1996, 77(1), 75-‐80
- 16. Datasets • Primary focus is on inves*ga*ng combina*ons with Ibru*nib for treatment of DLBCL – Btk inhibitor – In Phase II trials – Experiments run in the TMD8 cell line, tes*ng for cell viability
- 17. 0.8 Clustering Response Surfaces C1 (24) 0.6 0.4 C3(35) C2(47) 0.2 C4(24) 0.0
- 18. Cluster C3 0.30 0.25 0.20 0.15 0.10 0.05 0.00 302 281 128 174 285 153 177 210 144 35 60 457 180 39 111 272 288 166 231 104 106 417 319 44 218 279 219 121 119 34 102 286 230 178 179 macromolecule catabolic process regulation of interferon-gamma-mediated signaling pathway • Vargatef, vorinostat, ubiquitin-dependent protein catabolic process cellular process involved in reproduction flavopiridol, … negative regulation of cell cycle peptidyl-amino acid modification • Not par*cularly interphase specific given the cell cycle checkpoint range of primary peptidyl-tyrosine phosphorylation response to stress targets 0 1 2 3 -log10(Pvalue)
- 19. 0.08 0.06 0.04 0.02 0.00 361 Cluster C4 254 215 164 143 82 125 327 241 194 145 116 139 371 163 165 384 339 322 217 184 150 52 136 cellular carbohydrate biosynthetic process regulation of polysaccharide biosynthetic process cellular macromolecule localization • Focus on sugar peptidyl-serine phosphorylation metabolism regulation of generation of precursor metabolites and energy • Ruboxistaurin, cellular polysaccharide metabolic process cycloheximide, 2-‐ glucan metabolic process methoxyestradiol, … glucan biosynthetic process regulation of glycogen biosynthetic process • PI3K/Akt/mTOR glycogen metabolic process signalling pathways 0 1 2 3 -log10(Pvalue)
- 20. Combina*ons across Cell Lines • Cellular background affects responses • Can we group cell lines based on combina*on response?
- 21. Working in Combina*on Space • Each cell line is represented as a vector of response matrices L 1 L2 • “Distance” between two , = d1 cell lines is a func*on of the distance between component , = d2 response matrices , = d3 D ( L1, L2 ) = F({d1, d2 ,…, dn }) , = d4 • F can be min, max, mean, … , = d5
- 22. 0.00 0.05 0.10 0.15 0.20 0.25 0 1 2 3 4 INA-6 KMS-34 MM-MM1 INA-6 min sum 8226 L363 XG-1 OPM-1 U266 XG-2 ANBL-6 FR4 SKMM-1 AMO-1 EJM XG-6 OPM-1 MOLP-8 XG-2 ANBL-6 OCI-MY1 KMS-20 KMS-20 XG-7 L363 OCI-MY1 KMS-11LB XG-1 AMO-1 8226 XG-6 EJM FR4 U266 KMS-34 KMS-11LB MOLP-8 SKMM-1 XG-7 MM-MM1 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.1 0.2 0.3 0.4 0.5 0.6 L363 L363 OPM-1 OPM-1 euc max XG-2 XG-2 KMS-34 KMS-20 INA-6 XG-1 KMS-11LB XG-7 SKMM-1 ANBL-6 EJM OCI-MY1 U266 U266 MM-MM1 XG-6 FR4 INA-6 AMO-1 MOLP-8 XG-6 AMO-1 Many Choices to Make 8226 KMS-34 MOLP-8 KMS-11LB ANBL-6 SKMM-1 OCI-MY1 MM-MM1 XG-1 EJM KMS-20 FR4 XG-7 8226
- 23. Exploi*ng Polypharmacology • Vargatef exhibited anomalous matrix response compared to other VEGFR inhibitors Linifanib Axitinib Sorafenib Vatalanib Motesanib Tivozanib Brivanib Telatinib Cabozantinib Cediranib BMS-794833 Lenvatinib OSI-632 Foretinib Regorafenib Vargatef
- 24. Exploi*ng Polypharmacology Vargatef DCC-2036 PD-166285 GDC-0941 • PD-‐166285 is a SRC & FGFR inhibitor PI-103 GDC-0980 Bardoxolone methyl AT-7519 AT7519 • Lestaurnib has ac*vity against FLT3 SNS-032 NCGC00188382-01 Lestaurtinib CNF-2024 Src Lyn Lck ISOX Belinostat PF-477736 AZD-7762 Flt-3 PDGFRb PDGFRa FGFR-4 FGFR-3 FGFR-2 Chk1 IC50 = 105 nM FGFR-1 VEGFR-3 VEGFR-2 VEGFR-1 0 200 400 600 Potency (nM) Hilberg, F. et al, Cancer Res., 2008, 68, 4774-‐4782
- 25. Predic*ng Synergies • Related to response surface methodologies • LiRle work on predic*ng drug response surfaces – Peng et al, PLoS One, 2011 – Jin et al, Bioinforma5cs, 2011 – Boik & Newman, BMC Pharmacology, 2008 – Lehar et al, Mol Syst Bio, 2007 • But synergy is not always objec*ve and doesn’t really correlate with structure
- 26. Structural Similarity vs Synergy beta gamma 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 Similarity 0.85 0.90 0.95 1.00 1.05 1.10 1.15 0.75 0.85 0.95 1.05 ssnum Win 3x3 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0 5 10 15 20 25 -40 -30 -20 -10 0 Synergy measure
- 27. Predic*on Strategy • Don’t directly predict synergy • Use single agent data to generate a model surface • Predict combina*on responses • Characterize synergy of predicted response with respect to model surface • Reduced to a mixture predic*on problem • Will likely be beRer addressed by (also) considering target connec*vity
- 28. Conclusions • Use response surfaces as first class descriptors of drug combina*ons – Surrogate for underlying target network connec*vity (?) • Response surface similarity based on distribu*ons is (fundamentally) non-‐parametric • Going from single -‐ chemical space to combina*on space opens up interes*ng possibili*es • Manual inspec*on is s*ll a vital step
- 29. Acknowledgements • Lou Staudt • Beverly Mock, John Simmons