(P 177) PREDICTING THE IMATINIB RESPONSE OF PDGFRA EXON 18-MUTANT GASTROINTESTINAL STROMAL TUMOR USING IN VITRO MODELS

Objective: Gastrointestinal stromal tumor (GIST) is the most common sarcoma, with 4,000-6,000 cases diagnosed annually in the United States. The two most common GIST subtypes harbor oncogenic mutations in the homologous receptor tyrosine kinases KIT or platelet derived growth factor receptor alpha (PDGFRA) and are treated with tyrosine kinase inhibitors (TKIs). Imatinib, a type II TKl, is effective as a front-line (1L) therapy for advanced KIT-mutant GIST but is considered to be generally ineffective for PDGFRA-mutant GIST. Indeed, the most common PDGFRA mutation, exon 18 D842V, is notoriously resistant to all FDA-approved GIST type II TKIs. While avapritinib (AVA), a type I TKI, is FDA-approved for the treatment of all PDGFRA exon 18 mutant GIST, in some countries AVA is only available for D842V-mutant patients. Therefore, we investigated whether IM could be effective for GIST with non-D842V exon 18 mutations, thus reducing the number of patients who require 1L AVA.

We curated mutation data from a large patient cohort of PDGFRA-exon 18 mutant GISTs (n=967) and found ~67% (n=652/967) were D842V, with the other ~33% either having non-D842V point mutations (n=54/967) or complex in/dels around the 842 residue (n=261/967). We hypothesized that IM sensitivity could be predicted based on the nature of the amino acid in the 842-codon position.

Methods: To test our hypothesis, we modeled every amino acid in the 842-codon position as both a point mutation and in a 4-residue deletion context around the 842 position. These were expressed in Ba/F3s and we performed transformation assays to determine the oncogenic potential of each mutation. Mutations that drove IL3-independent growth were assessed for IM sensitivity (IC50) using immunoblotting for phosphorylated-PDGFRA and total PDGFRA.

Results: 13 out of 20 amino acids confer IM sensitivity when in the 842 position, contrary to the theory that all PDGFRA exon 18 mutations are IM resistant. Except for alanine, all other hydrophobic amino acids (7 out of 8), conferred IM resistance (IC50>150nM). In contrast, 842 position variants with amino acids from all other classes (polar, negatively charged, positively charged, or special case) were IM sensitive (IC50< 150nM). Therefore, we predict that ~86% of non-D842V mutations (n=270/315) in our 967-patient cohort would be IM sensitive. When we apply these predictions to real world clinical outcome data, the median PFS for IM therapy of predicted IM-resistant mutations was 3 months (n=55) compared to 38 months for patients with predicted IM-sensitive PDGFRA mutations (n=20, HR 3.798, CI 2.14—6.75, p< 0.0001). To further understand the structural implications of these 842 mutations on the kinase, we are currently performing molecular simulations with drug docking and will report this in our presentation.

Conclusion: Our predictive model can identify which PDGFRA exon 18 mutant GIST patients would benefit from 1L IM therapy and classify those that require AVA. Even though AVA is FDA-approved for 1L treatment of all PDGFRA exon 18 mutant GIST in the United States, other regulatory authorities have different use criteria. For instance, AVA is only approved for D842V patients in Europe. We suggest that patients with predicted IM-sensitive exon 18 mutations should receive 1L IM. Notably, IM has been in clinical use for over two decades, is more widely available, more cost-effective, and has a superior safety and tolerability profile compared to AVA. Additionally, if a patient progresses on IM, resistance profiles are known and there are additional FDA-approved treatments, whereas there are no known effective agents for patients that progress on AVA. Our work highlights a paradigm shifting approach in which clinical guidelines for the TKI treatment for PDGFRA-mutant GIST should be based on specific patient mutations, allowing for the optimization of patient care.