Atypical EGFR Mutations Show Heterogeneity

Oncogenic EGFR mutations—which occur in approximately 15% of non–small cell lung cancers (NSCLC) in the West and 30% to 40% in East Asia—have long been categorized into two types. The best known are classical mutations, such as L858R substitutions and exon 19 deletions, which account for two thirds of oncogenic mutations; most targeted therapy development is focused on these alterations. However, more than 70 types of so-called atypical EGFR mutations represent the remaining third, and which treatment patients should receive if their tumors have these atypical genetic aberrations isn't always clear.

“One of the challenges we face is that it's not practical to do clinical trials for each of those 70 mutations,” says senior author John Heymach, MD, PhD, of The University of Texas MD Anderson Cancer Center in Houston. “Up to this point, we haven't had an evidence-based way to group them.”

Atypical mutations have commonly been classified based on the exon in which they occur. But because proteins, including EGFR, take on three-dimensional folded structures, mutations near each other in the linear gene sequence won't necessarily have comparable effects on protein function or drug binding.

To address this, Heymach's team used sequencing data from 16,715 patients with EGFR-mutant NSCLC, then turned to a computational technique called molecular dynamics simulation to determine how the types of mutations in patients' tumors might affect EGFR structure and binding to first-generation (reversible, noncovalent), second-generation (irreversible, covalent), or third-generation (irreversible, covalent, T790M-targeting, and wild-type sparing) EGFR inhibitors (Nature 2021;597:732–7).

To validate the EGFR mutation types identified in silico, the group performed in vitro experiments testing the efficacy of EGFR inhibitors—including some approved agents and some under investigation—against cells harboring mutations of interest. They then compared these results with patient data, specifically retrospective progression-free survival reported in clinical trials and information about time to treatment failure pulled from real-world datasets.

The team's findings revealed a more rational way to group atypical EGFR mutations than simply relying on the affected exon. Specifically, they uncovered four classes of mutations:

• Classical-like mutations, which had little effect on EGFR structure, produced drug-binding characteristics similar to those of typical EGFR mutations, being sensitive to and selective for all EGFR inhibitors, particularly third-generation inhibitors.

• T790M-like mutations, which affected the protein's hydrophobic drug-binding cleft, blocked activity of first- and second-generation inhibitors, and were commonly found in NSCLCs with acquired resistance to these drugs.

• Exon 20 loop insertions, characterized by the presence of additional amino acid residues following the C-terminal end of the EGFR function–modulating αC-helix, exhibited heterogeneity in drug activity but generally responded only to second-generation inhibitors.

• P-loop αC-helix compression mutations, which alter the orientation of the kinase activation– and catalysis-mediating P-loop or αC-helix, were most sensitive to second-generation inhibitors.

However, the research must be prospectively validated in patients before it can be applied broadly. “I think we really have to do a big meta-analysis, collect all the information, and try to sort out whether the picture they describe is really applicable in the clinic,” says Tony Mok, MD, of the Chinese University of Hong Kong.

Still, the study provides a framework for future analyses. “There are other people who have looked at this, but not in such a comprehensive manner,” Mok adds. “I have admiration for this team for sorting out all these differences.” –Nicole Haloupek