NCI-H1373 | The Challenging Development Journey of Quinazoline-Containing KRAS[G12C] Covalent Inhibitors

This article is from an article published in JMC on April 1, 2026, Optimization of Covalent 6-Cyanoquinazoline KRASG12C Inhibitors for the Treatment of Solid Tumors.

What we can learn from this article:

1. How to reasonably rely on X-ray eutectic results in the SAR process of compounds? We can learn about the "rationality" of SBDD design, including the construction of hydrogen bond interactions and the occupation of residue hydrophobic pockets.
2. KRASG12C is a validated target, so the article focuses more on the comprehensive indicators of the developed molecule, including PD, PK, and safety. The compound in this article ultimately faild in terms of safety.

3. Under physiological conditions, the charge of alkaline groups can reduce membrane absorption. By introducing substituents and inducing changes in pKa, the proportion of charge can be altered.

Optimization of Covalent 6-Cyanoquinazoline KRASG12C Inhibitors for the Treatment of Solid Tumors

The overall landscape of the RAS signaling pathway is shown below, where the RAS family plays an important role in signal transduction and cell proliferation. The RAS family includes three proteins: KRAS, HRAS, and NRAS, with KRAS having the most frequent mutations.

The left figure a represents the mutation frequency of three proteins in the RAS family, while the right figure b represents the mutation types of KRAS in different locations.

The main mutation site of KRAS is position 12, which causes KRAS to mainly bind to GTP, thereby continuously activating downstream pathways. Due to the pM level affinity between KRAS and GDP or GTP, and given that there are almost no obvious pockets on the protein surface, there have been bottlenecks in the development of KRAS inhibitors for a long time in the past. The development of KRAS [G12C] inhibitors that this article focuses on originated from the concept validation achieved by Shokat et al. in 2013, which showed that this mutant form could be covalently targeted. After the development, the world's first KRAS inhibitor sotorasib was launched, as well as the subsequent new generation drug adagrasib.

The recommended dosage and frequency of administration for the two existing drugs by the FDA are sotorasib (960 mg QD) and adagrasib (600 mg BID), respectively. Researchers at Johnson&Johnson have developed a new drug design targeting this mutant form. Their research started with a compound JNJ-74699157, whose structure has not been disclosed but has undergone Phase I clinical trials. However, at a dosage of 100mg, dose limiting skeletal muscle toxicity was found in patients with advanced solid tumors (manifested as elevated creatine kinase in the blood, muscle weakness, and muscle spasms, which can recover after discontinuation), and there was also insufficient efficacy, so clinical research was discontinued. More importantly, the originally planned effective dosage for human administration of the drug was 1.5 g QD. 

Goal: Based on the above scientific facts, Johnson&Johnson researchers hope to develop drugs to ① reduce the dosage of medication; ② We also hope to maintain 100% binding to the KRAS [G12C] mutant after 24 hours; ③ Reduce skeletal muscle toxicity. 

In this development route, researchers conducted SAR studies on ARS-1620 (a covalent molecule based on quinazolines discovered by M. Janes et al. in 2018 and published in the Cell journal). The three SAR patterns discovered are shown on the right side of the following figure:

Among the series of compounds in this stage, 13a has better overall properties. In the NCI-H1373 human lung adenocarcinoma xenograft mouse model, target binding experiments showed that at a dosing frequency of 30mpk and QDx3, 100% target binding could be achieved within 24 hours.

VITRO BIOTECH provides a series of NCI-H1373 models: NCI-H1373 cell line | NCI-H1373-Luc Cell line | NCI-H1373-Luc-GFP

And this concentration can achieve long-term and almost complete inhibition of tumors. (Oral administration, solvent containing 20% HP beta CD).

The difference between the two doses indicates that achieving 100% binding to the target is crucial for later pharmacological effects. However, skeletal muscle toxicity was observed in the mouse model (7-day toxicity test) for 13a. Researchers are committed to improving the above symptoms, thus further exploring the drug structure.

Later, they entered the design phase based on crystal structure:

The main difference between 13ab and 13ba is that - Cl substitution becomes - CN substitution. The introduction of lipophilic groups reduces activity, membrane permeability, LogD, and covalent reaction rate. This rational transformation is based on the crystal structure, and they hope to build an interaction with R68.

The original explanation for the lack of improvement in activity even after 13ba constructed hydrogen bonds with Arg68 is ligand desorption penalty. Analyzing this type of molecule from multiple dimensions can provide more comprehensive insights into understanding the substitution effects of compounds. Researchers analyzed a series of molecules containing - CN substitution from the perspective of in vitro safety and compared them with - Cl substituted molecules to see if they avoided skeletal muscle toxicity. The two dimensions of evaluation are: human primary skeletal muscle cell confluence experiment (left figure) and drug-induced phospholipid disease experiment (right figure).

To improve the safety profile, a critical change was made: replacing a -Cl substituent with a -CN group (compound 13ba). This modification aimed to introduce a hydrogen bond interaction with Arg68, based on co-crystal structure analysis. Although this change reduced lipophilicity (LogD), membrane permeability, and covalent reaction rate, it was associated with improved in vitro safety markers (reduced drug-induced phospholipidosis and less impact on primary human skeletal muscle cell confluence).

Further Optimization for Activity and ADME:

Subsequent SAR focused on the N-methylpyrrolidine moiety to optimize electrostatic interactions with Glu62. Introducing a pyrrolizidine group (e.g., 13de) improved the covalent reaction rate (kobs/[I]) but decreased permeability, likely due to increased pKa and higher charge at physiological pH. Modifications to mitigate this, such as adding a methoxy group to the pyrrolizidine ring to lower the pKa via an inductive effect, were explored.

Promising Candidate 13de:

Compound 13de emerged with a favorable balance. It showed high in vivo target engagement at low doses (5 and 15 mg/kg) and achieved significant, dose-dependent tumor growth inhibition in mouse models. Its reaction rate was attributed to non-covalent binding, not intrinsic reactivity, as its GSH reactivity was lower than that of afatinib.

Termination of the Series:

Despite these advancements, the development of the entire 6-cyanoquinazoline series was ultimately halted. The decision was primarily based on the persistent observation of skeletal muscle toxicity in in vivo models, a recurring liability that could not be resolved through the SAR strategies explored, coupled with generally poor and variable oral bioavailability across species.

Conclusion and Lessons:

This study highlights the complex challenges in developing covalent KRAS[G12C] inhibitors. Despite achieving potent target engagement and efficacy, the project was abandoned due to the insurmountable issue of skeletal muscle toxicity and suboptimal pharmacokinetic properties. The work underscores that for such inhibitors, a holistic optimization of potency, ADME, and safety is critical, and that toxicity issues observed early in lead compounds can be difficult to overcome, even with rational, structure-based design.