KIT ATP-Binding Pocket/Activation Loop Mutations in GI Stromal Tumor: Emerging Mechanisms of Kinase Inhibitor Escape

PURPOSE

Imatinib resistance in GI stromal tumors (GISTs) is primarily caused by secondary KIT mutations, and clonal heterogeneity of these secondary mutations represents a major treatment obstacle. KIT inhibitors used after imatinib have clinical activity, albeit with limited benefit. Ripretinib is a potent inhibitor of secondary KIT mutations in the activation loop (AL). However, clinical benefit in fourth line remains limited and the molecular mechanisms of ripretinib resistance are largely unknown.

PATIENTS AND METHODS

Progressing lesions of 25 patients with GISTs refractory to ripretinib were sequenced for KIT resistance mutations. Resistant genotypes were validated and characterized using novel cell line models and in silico modeling.

RESULTS

GISTs progressing on ripretinib were enriched for secondary mutations in the ATP-binding pocket (AP), which frequently occur in cis with preexisting AL mutations, resulting in highly resistant AP/AL genotypes. AP/AL mutations were rarely observed in a cohort of progressing GIST samples from the preripretinib era but represented 50% of secondary KIT mutations in patients with tumors resistant to ripretinib. In GIST cell lines harboring secondary KIT AL mutations, the sole genomic escape mechanisms during ripretinib drug selection were AP/AL mutations. Ripretinib and sunitinib synergize against mixed clones with secondary AP or AL mutants but do not suppress clones with AP/AL genotypes.

CONCLUSION

Our findings underscore that KIT remains the central oncogenic driver even in late lines of GIST therapy. KIT-inhibitor combinations may suppress resistance because of secondary KIT mutations. However, the emergence of KIT AP/AL mutations after ripretinib treatment calls for new strategies in the development of next-generation KIT inhibitors.

Abstract 3887: ATP-binding pocket substitutions as secondary or tertiary in-cis mutations are major on-target ripretinib resistance mechanisms in gastrointestinal stromal tumor

Introduction: Ripretinib (Rip) is a kinase inhibitor with broad preclinical activity against mutant KIT. Based on the INVICTUS trial, Rip was approved for patients with Gastrointestinal Stromal Tumors (GIST) after treatment with 3 or more kinase inhibitors. Most patients in this ≥4th-line setting progress on Rip within one year. Here, we characterized Rip-progressing GIST samples to identify resistance mechanisms.

Methods: Progressing lesions in 25 patients were analyzed by NGS after Rip failure. KIT mutations (muts) were recapitulated by gene editing. Activities of Rip, sunitinib (SU), and novel TKIs were characterized by cell viability assays and immunoblot. Mutagenesis experiments were performed using ENU. SU- and Rip-resistant cell lines were pooled and treated with various regimens, with clonal composition deconvoluted by cDNA-based amplicon sequencing.

Results: 19/25 Rip-progressing GISTs displayed muts in the ATP-binding pocket (ATP-BP; e13/14). Four of these had pre-existing muts in the activation loop (AL; e17/18), which were confirmed to be in-cis with the primary and ATP-BP muts (triple in cis; TIC). Mutagenesis screens using a double KIT-mutant GIST line (e11 + AL) as a starting point confirmed that TIC-muts are a predominant escape mechanism when treated with Rip. A GIST subline (T1-triple) with TIC-muts in e11, e18 (A829P), and e13 (V654A) was highly resistant to Rip (GR50 > 2µM). Structural analyses of ATP-BP muts suggest steric interference and loss of van der Waals interactions that impede Rip binding and thereby confer Rip resistance. Another 3/25 Rip-progressing GISTs harbored pathogenic KIT muts in e9 only. A novel GIST cell line with primary KIT e9 mut (T1-e9) was 14-fold less sensitive to Rip than isogenic e11-driven cells (GR50 = 115 vs 8nM). Notably, adding a typical AL mut to T1-e9 (T1-e9-N822K) sensitized the cells to Rip (GR50 = 20nM). Immunoblots showed >95% inhibition of phospho-KIT in AL-mutant cell lines at Rip 100nM, whereas inhibition was weaker in T1-e9 (77%) and T1-V654A (46%), and absent in T1-triple. A compound screen of FDA-approved kinase inhibitors identified Nintedanib (NIN) as the most active compound against T1-triple. Clonal outgrowth assays of mixed cultures revealed SU (93% inhibition), and a weekly switch between Rip and either SU (96%) or NIN (79%) as the most effective inhibitors of pooled cell growth.

Conclusions: KIT e9 primary muts and e13/14 (ATP-BP) secondary muts are enriched in post-progression biopsies following Rip treatment. KIT TIC-muts are novel frequent events driving GIST clinical progression and confer a high degree of resistance. Strategies to overcome resistance may include combinations or sequences of approved drugs, and novel drugs that more efficiently inhibit TIC-mutant KIT.

Citation Format: Thomas Mühlenberg, Johanna Falkenhorst, Tom Schulz, Benjamin S. Fletcher, Alina Teuber, Dawid Krzeciesa, Isabella Klooster, Jonas Lategahn, Wen-Bin Ou, Meijun Lundberg, Margaret von Mehren, Susanne Grunewald, Alicia I. Tüns, Mehdi Brahmi, Michael C. Heinrich, Cesar Serrano, Hans-Ulrich Schildhaus, Sonja Sievers, Jürgen Treckmann, Lydia Wilson, Chandrajit P. Raut, Adrian Marino-Enriquez, Suzanne George, Daniel Rauh, Jonathan A. Fletcher, Sebastian Bauer. ATP-binding pocket substitutions as secondary or tertiary in-cis mutations are major on-target ripretinib resistance mechanisms in gastrointestinal stromal tumor. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 3887.

Plasma Sequencing for Patients with GIST-Limitations and Opportunities in an Academic Setting

Circulating tumor DNA (ctDNA) from circulating free DNA (cfDNA) in GIST is of interest for the detection of heterogeneous resistance mutations and treatment monitoring. However, methodologies for use in a local setting are not standardized and are error-prone and difficult to interpret. We established a workflow to evaluate routine tumor tissue NGS (Illumina-based next generation sequencing) panels and pipelines for ctDNA sequencing in an academic setting. Regular blood collection (Sarstedt) EDTA tubes were sufficient for direct processing whereas specialized tubes (STRECK) were better for transportation. Mutation detection rate was higher in automatically extracted (AE) than manually extracted (ME) samples. Sensitivity and specificity for specific mutation detection was higher using digital droplet (dd)PCR compared to NGS. In a retrospective analysis of NGS and clinical data (133 samples from 38 patients), cfDNA concentration correlated with tumor load and mutation detection. A clinical routine pipeline and a novel research pipeline yielded different results, but known and resistance-mediating mutations were detected by both and correlated with the resistance spectrum of TKIs used. In conclusion, NGS routine panel analysis was not sensitive and specific enough to replace solid biopsies in GIST. However, more precise methods (hybridization capture NGS, ddPCR) may comprise important research tools to investigate resistance. Future clinical trials need to compare methodology and protocols.