Abstract 2425
Background
Patients (pts) in 80405 w/ mCRC originating from left-sided (L) 1° tumor had superior OS and PFS vs. pts w/ right-sided (R) 1°, but the underlying biological explanation is mostly unknown. We applied causal machine learning methods to the study to discover molecular pathways linking 1° side to OS and PFS.
Methods
430 pts including 128 RAS mutant (mut) had clinical data, mut status in 10 genes, MSI status, gene expression from 816 genes by Nanostring, and consensus molecular subtypes (CMS). For this analysis, transverse included w/ R-sided tumors. Using a Bayesian causal machine learning platform, we built an ensemble of network models for OS and PFS.
Results
Bayesian networks revealed a molecular pathway downstream of 1° side and driving OS and PFS: MSI-H status, BRAF mut, and extracellular matrix remodeling and angiogenesis-related genes including EGF7A, ALOX5, VIM, MAP4K1, COL10A1, COL6A3, and COL5A2. BRAF mut and angiogenesis signature are prognostic for OS and PFS and downstream of 1° side (Table); angiogenesis signature is downstream of BRAF (OR 13.1, p < 0.001). CMS4 and RAS mut are prognostic for OS and PFS but not causally downstream of 1° side. In angiogenesis-high pts (n = 90), 1° side is no longer prognostic for OS or PFS. In angiogenesis-low pts, 1° side remains prognostic only for OS. There is a significant angiogenesis by biologic interaction for OS (p = 0.008) favoring bevacizumab in angiogenesis-high pts.Table: 458PD
| R vs L | OS | PFS | |||
|---|---|---|---|---|---|
| OR (p-value) | HR (95% CI) | p-value | HR (95% CI) | p-value | |
| RAS | 1.7 (0.01) | 1.42 (1.14-1.77) | 0.002 | 1.24 (1.00-1.54) | 0.048 |
| CMS4 | 1.2 (0.5) | 1.68 (1.33-2.13) | <0.001 | 1.60 (1.28-2.01) | <0.001 |
| angio sig. | 3.9 (<0.001) | 1.71 (1.32-2.21) | <0.001 | 1.71 (1.34-2.19) | <0.001 |
| BRAF | 7.7 (<0.001) | 1.76 (1.28-2.40) | <0.001 | 1.57 (1.16-2.12) | 0.003 |
| R vs L (all) | - | 1.50 (1.22-1.86) | <0.001 | 1.24 (1.02-1.51) | 0.03 |
| R vs l (angio +) | - | 1.26 (0.76-2.08) | 0.36 | 1.15 (0.72-1.82) | 0.56 |
| R vs L (angio -) | - | 1.47 (1.15-1.88) | 0.002 | 1.15 (0.91-1.44) | 0.23 |
Conclusions
Impact of 1° side on OS and PFS is associated with BRAF and angiogenesis-related pathways, consistent with the relative clinical benefit of anti-VEGF therapy in R-sided and BRAF mut pts.
Clinical trial identification
NCT00265850.
Legal entity responsible for the study
Alliance for Clinical Trials in Oncology.
Funding
U10CA180821, U10CA180882, U10CA180830, U10CA180888; Eli Lilly and Company, Genentech, Pfizer.
Editorial Acknowledgement
Disclosure
R.K. Das, L. Furchtgott, B. Hayete, B. Harms, D. Cunha, J. Latourelle, D. Wuest, I. Khalil, C. Washburn: Employee: GNS Healthcare. K. Rich: Consulting: GNS Healthcare. A. Nixon: Research funding: Seattle Genetics, Tracon Pharma, Amgen, Novartis, Genentech/Roche, Acceleron, MedPacto, Leadiant Biosciences; Consultant: Pfizer, Eli Lilly. H-J. Lenz: Advisory board, lectures: BMS, Bayer, Roche, Merck Serono, Genentech. A. Venook: Research funding: Genentech, Roche, Merck. All other authors have declared no conflicts of interest.