AI & the Hunt for the Holy Grail of Cancer Targets
The human mutant K-Ras protein, long considered the ‘holy grail’ of cancer targets, is implicated in about 30% of all human cancers, acting much like a central on-off switch. Despite extensive research, it earned the label of being “undruggable” due to daunting challenges related to its structure and biological functions.
Targeting the enzyme’s active site, which binds GTP, proved to be one such obstacle. Many similar enzymes are involved in essential cellular processes, and the enzyme binds its natural substrate, GTP, very tightly so it is challenging to compete it out with inhibitors. Additionally, it is crucial for any effective cancer treatment to selectively inhibit the mutated, overactive form of K-Ras without affecting the non-mutated form, which plays vital roles in healthy tissues.
A significant breakthrough occurred in 2013 when Kevan Shokat’s team at UCSF developed a small molecule probe that specifically and covalently inhibits the K-Ras G12C mutant protein. This discovery provided the starting point drug hunters had been waiting for, eventually leading to the FDA approval of the first K-Ras G12C targeted drug, sotorasib (Amgen), for treating non-small cell lung cancer (NSCLC) eight years later.
The development of the first K-Ras G12C inhibitors was a significant milestone in cancer research, igniting intense interest among researchers and pharmaceutical companies to find the most effective treatments for K-Ras-driven cancers. Clinical trials of the current inhibitors have shown an increase in progression-free survival (PFS). However, they also quickly revealed the challenge of drug resistance, a common issue in cancer therapy.
An even greater limitation is that the G12C mutation targeted by these treatments is found in just 13–14% of non-small cell lung cancer (NSCLC) patients and in under 3% of those with colorectal cancer (CRC) or pancreatic ductal adenocarcinoma (PDAC). The G12D and G12V mutations are the most prevalent in CRC (12.5%, 8.5%) and PDAC (37.0%, 28.2%) and are notably more challenging due to the decreased reactivity of these amino acid substitutions. This has prompted the exploration of alternative targeting strategies, including Pan-Ras inhibitors and inhibitors targeting auxiliary proteins such as SOS1 or SHP2. However, these strategies do not offer the same level of precision as direct covalent K-Ras G12C inhibitors, leading to an increased risk of off-target effects on healthy tissues. Active clinical trials will establish the efficacy of these approaches as standalone treatments.
Given that no single K-Ras inhibitor is expected to be a universal solution for cancer therapy, researchers are increasingly focusing on drug combinations. Amgen is currently conducting a large clinical trial, “CodeBreak 101” (NCT04185883), which is evaluating 13 different combination treatment regimens using their drug, sotorasib. This trial, which includes 14 different treatment arms (13 combinations plus sotorasib as a monotherapy), represents a substantial investment and underscores our limited understanding of how to rationally design treatment combinations.
Among these, an early promising combination is the pairing of sotorasib, a K-Ras G12C inhibitor, with an SHP2 inhibitor (RMC-4630) for non-small cell lung cancer (NSCLC). Targeting two points within the same pathway appears to produce higher response rates and reduced resistance. However, these initial results are from a small Phase 1b trial, are limited to K-Ras G12C mutants, and specific to NSCLC. Developing more effective treatments for a broader patient population requires the innovation of drugs that extend their targeting beyond K-Ras G12C mutant cancers.
These efforts to disrupt the K-Ras signaling pathway have centered on inhibitors, including direct K-Ras inhibitors, aimed at upstream and downstream targets. Complementary to these advancements, AI methods are now driving extensive analyses of human clinical cancer data in search of new vulnerabilities in K-Ras G12V driven cancers. Similarly, the Synfini team has leveraged proprietary AI-based approaches and twenty years of human clinical tumor data to discover several promising drug targets.
These targets are proteins required for tumor growth in cells harboring the K-Ras G12V mutation, but not in cells with the non-mutant form of the protein, a biological phenomenon known as synthetic lethality. While the idea of exploiting synthetic lethality in mutant K-Ras isn’t new, the availability of extensive clinical genomic data combined with AI’s capabilities has enabled Synfini to identify new targets directly from patient data. These targets have undergone rigorous validation in the lab using human cell lines and CRISPR gene editing, and their efficacy has been confirmed in mouse tumor models, where they were shown to selectively inhibit tumor growth in K-Ras G12V mutant CRC models.
The next step is the development of drug candidates targeting these proteins for clinical testing. Synfini’s drug discovery platform, which leverages AI-driven chemistry and rapid design-make-test-analyze (DMTA) cycles, is set to streamline the creation of these first-in-class drug candidates. Utilizing this technology for the newly identified K-Ras cancer targets could lead to the development of new, more targeted cancer treatments.
The success of these drugs in clinical settings, similar to previous K-Ras therapies, will depend on carefully choosing patients for treatment. Since these targets were discovered through analysis of human tumor data, it enables more precise identification of patients who are most likely to benefit, embodying the principles of precision medicine. This targeted approach is crucial for addressing the needs of the vast number of cancer patients who have yet to find effective treatment options through existing K-Ras pathway-targeted therapies.
