Preclinical model-based evaluation of Imatinib resistance induced by KIT mutations and its overcoming strategies in gastrointestinal stromal tumor (GIST)

Abstract

Background: The potential correlation between KIT secondary mutations and Imatinib-resistance in gastrointestinal stromal tumor (GIST) has been hinted, yet their specific linkage and underlying mechanisms remained unelucidated, also the development of substitute strategies dealing with this resistance was urgently needed. Methods: In this study, we explored the distribution of the most prevalent forms of KIT mutation in Chinese GIST patients, after that, we established cell lines that was overexpressed with mutant KIT, and by performing RNA sequencing, immunoblotting and cell viability, we analyzed their functional and mechanistic relevance with Imatinib-resistance in GIST cell lines. Additionally, we evaluated the tumor inhibition efficacy of four regimens in Imatinib-resistant GIST cell lines and patient-derived xenograft (PDX) models. Results: We found that KIT exon 13-V654A and exon 17-N822K were the most common secondary mutations in GIST with primary exon 11 mutations. These two secondary mutations induced Imatinib resistance by activating PI3K-Akt signaling pathway, while PI3K-Akt inhibition rescued the resistance. By assessing the feasibility of other four tyrosine kinase inhibitor (TKIs, Sunitinib/Regorafenib/Avapritinib/Ripretinib) against Imatinib-resistant GIST, we found that Sunitinib was more suitable for KIT exon 13 secondary mutations, the rest were more effective for KIT exon 17 secondary mutations, while all four TKIs displayed efficacy for KIT exon 9 mutations, emphasizing their clinical applications against Imatinib resistance. Conclusions: We demonstrated the mechanism by which KIT secondary mutations on exon 13/17 cause Imatinib resistance to GIST, and validated that several novel TKIs were valuable therapeutic options against Imatinib-resistance for both secondary- and primary-KIT mutations.

Introduction

Gastrointestinal stromal tumor (GIST), the most common alimentary canal interstitial tumor, was considered to be originated from the mesenchymal cells of Cajal (the pacemaker cells of the gastrointestinal tract) or related stem cells [1,2]. The prevalence of GIST continues to increase, with approximately 5,000 new cases each year. While GISTs can arise anywhere from esophagus to rectum, the majority originate from the stomach (60-70%) and small bowel (20-30%) [3]. Traditionally, surgery was the only approach of radical cure for GIST with a 5-year survival rate of 48-54% [4], while patients with irresectable or metastatic disease survived only for a median of 18-24 months after diagnosis with a 5-year survival rate of 5-10% [5,6].

GISTs are driven and characterized by C-Kit (KIT) or platelet-derived growth factor receptor alpha (PDGFRA) activating mutations, which approximately account for 80% or 10% of GIST patients, respectively [7]. With the development of targeted therapies, Imatinib mesylate, a selective inhibitor against mutant forms of type II tyrosine kinases, such as KIT, PDGFRA and ABL, has been used as a paragon of first-line treatment for patients with advanced GIST, which dramatically improved the 5-year survival and recurrence rate [8-11]. However, most patients with initial clinical benefit from Imatinib eventually progress, typically within 29 months [12,13]. Therefore, it is necessary to uncover the underlying molecular mechanisms of acquired resistance to Imatinib so as to further improve the survival of GIST patients.

Mainly observed on the ATP-binding pocket encoded by exon 13 and 14 and the activation loop encoded by exon 17 and 18, secondary mutations of KIT gene were found after Imatinib failure in up to 90% of GIST patients. Thus, tumor subclones carrying heterogeneous secondary KIT mutations were considered to re-activate KIT downstream signaling and continuously drive GIST proliferation and survival [14-17]. However, the molecular correlation between KIT mutations and Imatinib resistance, as well as the underlying mechanisms were still uncertain. In order to develop substitute regimens after Imatinib resistance, the applicable drug spectrum for GIST with different KIT mutations also remained to be established.

In our study, we analyzed 2273 Chinese GIST patients to identify the mutational spectrum of KIT mutations. By applying cell lines and PDX models, we evaluated the impact and underlying mechanisms of the most representative types of secondary KIT mutations on Imatinib sensitivity, and compared the anti-GIST efficacy of four candidate TKIs (Sunitinib, Regorafenib, Avapritinib and Ripretinib) to replace Imatinib for specific KIT mutations.

Materials and methods

Origination of patients and clinical information

Patients’ specimens for KIT mutations assessment and xenograft construction were collected with the assistance of the gastrointestinal surgery by the Department of Pathology and Department of Gastrointestinal Oncology, Peking University Cancer Hospital & Institute. The diagnosis of GIST and KIT mutational spectrum were confirmed by histological analysis or Sanger sequencing for all cases. The experimental applications of patient samples and information (including basic parameters, therapeutic responses and medical images) have been approved by the Institutional Ethics Committee, Peking University Cancer Hospital & Institute and performed in accordance with the Declaration of Helsinki Principles. Written informed consents were obtained from all providers.

Clinical information of all patients was recorded and followed up by Department of Gastrointestinal Oncology, Peking University Cancer Hospital & Institute. Responses of patients were defined based on the Response Evaluation Criteria in Solid Tumors (RECIST 1.1). The progression-free survival (PFS) was calculated respectively, which started from the treatment of Imatinib and ended with progressive disease or death for any cause.

Cell culture, virus transfection and Imatinib-resistance induction

GIST-T1 and GIST-882 are human GIST cell lines with a KIT exon 11 heterozygous 560_578 deletion mutation and a primary KIT exon 13 homozygous K642E missense mutation, respectively [18,19]. Both cell lines were gifts from Dr. Wang (Changzheng Hospital, Shanghai, China), and has been maintained and used in our laboratory. Under 37°C, 5% CO₂ and appropriate humidity, GIST cells were cultured with Iscove, Modified Dulbecco Medium (Gibco, USA) supplemented with 20% fetal bovine serum (Gibco, USA) and 1% penicillin-streptomycin (HyClone, USA).

KIT cDNA clones carrying exon 13-V654A or exon 17-N822K were constructed with pLenti-EF1a-EGFP-F2A-Puro-CMV (Clontech), and were integrated into lentiviral particles by using a Transfect lentiviral packaging reagent (OBIO, Shanghai, China). GIST-T1 and GIST-882 were transfected with lentiviral particles containing vector, KIT/V654A or KIT/N822K plasmids, respectively, then administrated with puromycin (0.5-1 μg/mL, Gibco, USA) to select cell clones expressing KIT.

GIST-T1R and GIST-882R sub-cell lines were cultured by continuously culturing the progenitor cell lines with low-dose (16 nM) to high-dose (400 nM) Imatinib for 6 months.

Cell viability assay

GIST cells in the logarithmic growth phase were evenly planted in 96-well plates with 1×10⁴ cells per well. After 24 hours, the culture medium was replaced with fresh medium containing drugs (Imatinib/BEZ235/Sunitinib/Regorafenib/Avapritinib/Ripretinib) and cultured at 37°C for another 72 hours. Imatinib mesylate, Sunitinib, Regorafenib, Avapritinib, and Ripretinib were purchased from MedChemExpress. BEZ235 was purchased from Selleck.

The cell culture medium was then discarded and replaced by fresh medium containing 10% cell counting kit-8 (CCK-8, DOJINDO, Japan), and cultured under 37°C for 2-3 hours. The absorbance at 450 nm was measured with a microplate reader (Infinite F50, TECAN, Switzerland) and used to calculated cell viability. All the above experiments were repeated for at least three times.

Establishment of patient-derived xenografts

Tumor specimens from patients were first placed in fresh RPMI 1640 medium, cut into a 3-mm square-fragments and washed 3 times with RPMI 1640. The tumor samples (P0) were then stuffed into 18-gauge needles and planted subcutaneously into 6-8 weeks NOD/SCID mice (Huafukang Biotechnology Co., Ltd., Beijing, China). Tumor growth and body weight were measured twice a week. When the volume of xenograft tumor reached 750-1000 mm³, the mice were sacrificed and the subcutaneous tumors (P1) were stripped, and then re-implanted into NOD/SCID mice according to the previous steps, sequentially establishing P2 and P3 passages of PDX. All procedures were performed in Peking University Cancer Hospital animal rooms under sterile conditions in accordance with the Guidelines for the Care and Use of Laboratory Animals of National Institutes of Health. This experiment was approved by the Ethics Committee of Peking University Cancer Hospital.

In vivo drug sensitivity assay

For established GIST PDX models, when xenograft tumors reached 150-250 mm³, mice were randomly divided into control and experimental groups (5 mice for each group). The control group was given saline and the treatment group was given different drugs by gavage. The drugs and doses in the treatment groups were as follows: Imatinib 50 mg/kg twice a day, Sunitinib 40 mg/kg daily, Regorafenib 30 mg/kg daily, Avapritinib 30 mg/kg daily, Ripretinib 30 mg/kg daily. Mice were treated for 3-4 weeks and measured for tumor size and body weight every two days. Tumor volume was calculated using the following formula: volume = (length × width²)/2. Tumor growth inhibition (TGI) rate was calculated using the following formula: TGI = 1-ΔT/ΔC×100% (ΔT = tumor volume changes of the drug treated group, ΔC = tumor volume changes of the control group on the final day of the study).

RNA/DNA extraction and sequencing

Total RNA was collected from specimens using Trizol method. Quality control was performed to ensure RNA integrity using an Agilent 2100 Bioanalyzer (Agilent Technologies, USA). RNA sequencing was performed by Novogene (Beijing, China) using an Illumina HiSeq instrument (Illumina, San Diego, CA, USA). Genomic DNA was collected by using QIAamp DNA Mini Kit (Qiagen, Hilden, Germany), and was testified by Next generation sequencing (NGS) with an Illumina HiSeq 2000 system (Illumina, San Diego, CA, USA). Expressional and genomic alterations were enriched by referring to the Kyoto Encyclopedia of Genes and Genomes (KEGG, https://www.kegg.jp/) and the Database for Annotation, Visualization, and Integrated Discovery Bioinformatics Resources (DAVID, http://david.abcc.ncifcrf.gov). These procedures were similar to our previous reports [20,21].

Immunoblotting

Total proteins from cells or tissue were extracted with CytoBuster protein extraction reagent (EMD Millipore, USA) supplemented with protease inhibitors and phosphatase inhibitors (Roche, Germany). BCA method (Applygen, Beijing, China) was used to measure protein concentration. 30 μg proteins of each sample were electrophoresed by 10% SDS-PAGE and transferred to the Immobilon^®-PSQ Membrane (Merck Millipore, Germany). The membrane was blocked with 5% albumin bovine V (Biotopped, Beijing, China) solution and incubated with primary and secondary antibodies under proper concentrations. Primary antibodies for phospho-KIT Y719 (#3319), KIT (#3074), phospho-AKT S473 (#4060), AKT (#4691), phosphor-S6 S240/S244 (#4858) and S6 (#2217), anti-rabbit IgG, HRP-linked antibody (#7074), anti-mouse IgG, and HRP-linked antibody (#7076) were from Cell Signaling Technology (Danvers, MA, USA), while antibody for β-actin was purchased from Sigma-Aldrich (St. Louis, Missouri, USA). The protein blots were illuminated with HRP substrate luminol reagent (Millipore, USA) and visualized using an Amersham Imager 600 (GE Healthcare, UK).

Immunohistochemistry and hematoxylin-eosin staining

PDX tumor tissues were fixed in formalin, embedded in paraffin, and then sliced into tissue sections. The tissue sections were subjected to deparaffinization, hydration, endogenous peroxidase treatment, and antigen retrieval steps, and then incubated with primary antibodies, IgG-HRP polymer (ZSJQB, Beijing, China) and diaminobenzidine substrate. To evaluate the histological characteristics of GIST, the tissue sections were stained with hematoxylin and eosin. Two pathologists from the Peking University Cancer Hospital’s pathology department who were blinded to this study interpreted the sections.

Statistical analysis

Unpaired t test was used to analyze differences in cell viability between different groups. ANOVA was used to compare tumor growth among different groups with P<0.05 as statistically significant. The connection between KIT mutations and Imatinib treatment response were analyzed using Fisher’s exact test. Log-rank test and Kaplan-Meier curve were used to analyze the correlation between the KIT mutation and PFS. Statistical analysis was performed with SPSS 20.0 software, and formatted with Graphpad Prism 6 software. For all analysis, P<0.05 was considered statistically significant.

Results

Specific mutations of KIT contributed to GIST’s acquired resistance to Imatinib

We recruited 2273 Chinese GIST patients and assessed the clinicopathological characteristics of KIT mutations for validation (Supplementary Table 1). Among these patients, 3.0% (69) had wild-type KIT and wild-type PDGFRA, 4.8% (109) had mutated PDGFRA, while KIT and PDGFRA dual mutations were not observed. 92.2% (2095) patients harbored primary KIT mutations, in which exon 11-mutation (77.8%, 1769) was the dominant type, while mutations on exon 9 (11.5%, 263), exon 13 (1.5%, 33), and exon 17 (1.3%, 30) were also detected (Figure 1A). In these GIST patients, 130 received Imatinib therapy and developed resistance to Imatinib. Secondary mutations were detected in 38.5% (50) of them, including exon 13 (14.6%, 19), exon 17 (17.7%, 23), exon 14 (2.3%, 3) and exon 18 (3.9%, 5) (Figure 1B).

Figure 1.

Specific mutations of KIT in GIST were associated with acquired Imatinib resistance. (A) Distribution of KIT/PDGFRA mutations in GIST population. (B) Distribution of secondary mutations in KIT exon 11 mutant-GIST patients developed resistance to Imatinib. (C) ORR (overall response rate) and (D) PFS (progression-free survival) for different KIT gene types. (E) Establishment of KIT expressing cell lines with lentivirus transfection. (F) Viability curve of GIST cells overexpressing wild type-, E13V654A-, or E17N882K-KIT after 72 hours exposure to Imatinib. All experiments were repeated three times independently. *P<0.05.

Patients harboring primary KIT exon 11 mutation displayed the best, while patients harboring primary KIT exon 9 mutation displayed the worst ORR (overall response rate) and PFS (progression-free survival) to Imatinib, thus they were considered as Imatinib-sensitive and Imatinib-insensitive. After the emerging of KIT secondary mutations, the ORR and PFS of exon 11-mutated patients to Imatinib was impaired (ORR: 36.7% vs. 73.2% vs. 38%; median PFS: 9 vs. 58 vs. 25 months) (Figure 1C, 1D), indicating the acquisition of Imatinib resistance. Therefore, we concluded three representative categories of GIST patients according to their KIT status and responses to Imatinib, i.e., Imatinib primary sensitive group (exon 11 mutations), Imatinib primary insensitive group (exon 9 mutations), and Imatinib secondary resistant group (exon 11 mutations plus secondary mutations). Since acquired resistance has been a major reason that impairs the responses to Imatinib, we first investigated those individuals, in which the dominant forms of secondary KIT mutations for exon 13 and 17 were V654A and N822K, which was in accordance with our previous findings in a 50-case GIST patient cohort [22].

GIST-T1 and GIST-882 were two cell lines harboring primary KIT mutations. Since GIST-T1 carried a KIT exon 11-560_578del heterozygous mutation, it has a higher sensitivity to Imatinib than GIST-882 carrying an exon 13-K642E homozygous mutation. To decipher whether the secondary mutations we observed in patient cohort were involved in the acquired resistance to Imatinib, we performed lentiviruses-directed transfection of KIT exon 13-V654A or exon 17-N822K into GIST-T1/GIST-882 cells (Figure 1E). According to cell viability assay, V654A/N882K transfection significantly impaired sensitivity to Imatinib, while wild-type (WT) KIT expressing groups displayed comparable sensitivity to vector groups (Figure 1F). The IC₅₀ values for GIST cells to Imatinib were shown in Table 1. These data suggested that WT-KIT had a minimum impact on the sensitivity to Imatinib, while the acquisition of specific KIT mutations (i.e., exon 13-V654A and exon 17-N822K) contributed to GIST’s acquired resistance to Imatinib. We also verified that Imatinib failed to increase phenotypic characteristics of the GIST-T1 and GIST-882 cells expressing E13V654A and E17N822K forms of KIT, including cell apoptosis, cycle, migration, invasion, plate clone formation, and wound-healing assay experiments which were presented in Supplementary Figures 1, 2, 3.

Table 1.

IC₅₀ values for Imatinib in GIST cell groups

GIST cell line	IC50 (nM)
	vector	E13V654A-KIT	E17N822K-KIT	WT-KIT
GIST-T1	16.83	28.62	30.70	16.02
GIST-882	33.49	130.19	604.8	44.89

Abbreviations: E13V654A, exon 13-V654A; E17N822K, exon 17-N822K; WT, wild-type.

KIT exon 13-V654A and exon 17-N822K mutations contributed to Imatinib resistance by inducing PI3K-Akt activation in GIST cells

In order to explore the effects of KIT secondary mutations on the molecular characteristics of GIST, we performed RNA-sequencing for the groups stably overexpressing exon 13-V654A- or exon 17-N822K-mutated KIT, and investigated their expressional diversities. For GIST-T1, 2706 genes (1681 upregulated and 1025 downregulated) were differentially expressed between V654A and vector groups, while 1664 genes (971 upregulated and 693 downregulated) were differentially expressed between N822K and vector groups. For GIST-882, 3532 genes (1864 upregulated and 1668 downregulated) were differentially expressed between V654A and vector groups, while 1411 genes (972 upregulated and 439 downregulated) were differentially expressed between N822K and vector groups (Figure 2A). By referring to KEGG database, we performed pathway enrichment analysis for V654A- and N822K-transfected groups. Differentially expressed genes were mainly distributed in pathways stimulating carcinogenic progression and aggressiveness, among which PI3K-Akt signaling was consistently enriched for all four pairs of comparisons (Figure 2B), providing potential clues that these KIT secondary mutations contributed to Imatinib-resistance through stimulating PI3K-Akt signaling.

Figure 2.

Dysregulated downstream pathways induced by two crucial KIT mutations. (A) Differentially expressed genes and (B) significantly enriched pathways comparing E13V654A/E17N822K-KIT with vector groups were demonstrated for GIST-T1/GIST-882 cells.

For verification, we verified the activation of PI3K-Akt signaling under different KIT mutational status. In accordance with total KIT level, the expressional intensity of phosphorylated KIT was higher in WT, V654A and N822K groups. While the total Akt and S6 expressions remained unaffected, p-Akt and p-S6 were evidently elevated in WT, V654A and N822K groups. More importantly, the administration of Imatinib (16 nM in GIST-T1, 64 nM in GIST-882, 72 hours) failed to quench the phosphorylation of KIT/Akt/S6 in V654A and N822K groups as succeed in vector and WT groups (Figure 3A). Furthermore, we tested the combination of Imatinib with the PI3K-Akt pathway inhibitor BEZ235 in V654A/N882K transfected groups. The overactivation of PI3K-Akt signaling induced by KIT secondary mutations was rescued by BEZ235 (Figure 3B), while BEZ235 augmented the growth-inhibiting efficacy of Imatinib (Figure 3C), emphasizing that exon 13-V654A and exon 17-N822K mutations contributed to Imatinib resistance by stimulating PI3K-Akt signaling.

Figure 3.

The activation of PI3K-Akt signaling under different KIT status. Western blot-quantified expression of p-KIT, p-Akt and p-S6 in (A) Imatinib-treated GIST cells transfected with E13V654A-KIT (exon 13-V654A) or E17N822K-KIT (exon 17-N822K), or (B) Imatinib and/or BEZ235-treated GIST cells transfected with E13V654A/E17N822K-KIT. (C) Viability curve of GIST cells transfected with E13V654A/E17N822K-KIT after 72 hours exposure to Imatinib with or without BEZ235.

Additionally, we constructed Imatinib-resistant GIST-T1R and GIST-882R sub-cells by continuously culturing the progenitor cell lines with low-dose (16 nM) to high-dose (400 nM) Imatinib for 6 months. A resistance to Imatinib was induced in both cell lines (Supplementary Figure 4A), while western blot assay showed that Imatinib with the same concentration as in progenitor cells (16 nM in GIST-T1, 64 nM in GIST-882, 72 hours) failed to suppress PI3K-Akt signaling in these resistant cells (Supplementary Figure 4B), supporting the notion that activated PI3K-Akt signaling contributed to Imatinib resistance in GIST.

Evaluation of substitutive inhibitors in GIST cells harboring Imatinib-resistant KIT mutations

With the rapid development of precision medicine, a series of new agents (such as Sunitinib, Regorafenib, Avapritinib and Ripretinib) targeting oncogenic tyrosine kinases have been developed for the treatment of KIT-related cancers, and were currently under evaluation for potential clinical applications. Consequently, we evaluated the inhibition efficiency of the above four newly developed TKIs (tyrosine kinase inhibitors) against Imatinib-sensitive or -resistant GIST cellular groups.

In GIST-T1, when compared with vector (exon 11 mutation) group, IC₅₀ values of V654A (exon 11+13 mutation) group were apparently higher for Regorafenib/Avapritinib/Ripretinib and lower for Sunitinib, while IC₅₀ values of N822K (exon 11+17 mutation) group were higher for Sunitinib and lower for Regorafenib/Avapritinib/Ripretinib. In contrast, in GIST-882, when compared with vector (exon 13 mutation) group, IC₅₀ values of V654A (exon 13+13 mutation) group were higher for Sunitinib/Avapritinib/Ripretinib, while IC₅₀ values of N822K (exon 13+17 mutation) group were higher for Sunitinib and lower for Regorafenib/Avapritinib/Ripretinib (Table 2). These inconsistencies between two cell lines might be due to the diverse sensitivity endowed by their original KIT status (GIST-T1, KIT exon 11 mutation (560_578 deletion), Imatinib sensitive; GIST-882, KIT exon 13 mutation (K642E), Imatinib insensitive). Concomitantly, these data suggested that after acquiring KIT secondary mutation-associated Imatinib resistance, Sunitinib was a better option for KIT secondary exon 13 mutations, while Regorafenib/Avapritinib/Ripretinib had better performance for KIT secondary exon 17 mutations (Figure 4A).

Table 2.

IC₅₀ values for the four KIT-inhibiting TKIs in GIST cell groups

	IC50 (nM)
	GIST-T1-vector	GIST-T1-V654A	GIST-T1-N822K	GIST-882-vector	GIST-882-V654A	GIST-882-N822K
Sunitinib	15.91	13.02	61.05	4.17	29.34	>1000
Regorafenib	88.97	134.50	13.29	116.20	80.27	18.07
Avapritinib	139.70	213.10	30.50	161.70	221.40	119.50
Ripretinib	53.21	>1000	19.53	470.20	1972	345.30

Figure 4.

Inhibition efficiency of Sunitinib, Regorafenib, Avapritinib and Ripretinib against Imatinib-sensitive or -resistant GIST cell groups. A. Viability curve of GIST-T1/GIST-882 cells after 72 hours exposure to Sunitinib, Regorafenib, Avapritinib and Ripretinib. B. Western blot quantification of protein expressional intensity of p-KIT, p-Akt and p-S6 in vector, E13V654A and E17N822K for GIST-T1/GIST-882 cells after exposure to Imatinib, Sunitinib, Regorafenib, Avapritinib and Ripretinib. All experiments were repeated three times independently.

On the other hand, these TKIs’ pathway-inhibitive efficiency in parental and Imatinib-resistant GIST cells was also assessed by western blot. In line with the IC₅₀ spectrum, secondary mutation-induced KIT phosphorylation and PI3K-Akt activation were more efficiently inhibited by Sunitinib for secondary exon 13 mutations or by Regorafenib/Avapritinib/Ripretinib for exon 17 mutations (Figure 4B).

Anti-tumor effects of TKIs in Imatinib-resistant GIST PDXs

Although we have previously proved the efficacy of multiple inhibitors in GIST cells harboring KIT secondary mutations, more powerful in vivo evidences remained to be achieved. For further validation, we established two PDX models (namely, GIST-1 and GIST-2) by applying lesions from Imatinib-resistant GIST patients. By referring to a literature concerning in vivo drug concentrations [23], we tested the anticancer efficacy of above-mentioned agents (Imatinib, Sunitinib, Regorafenib, Avapritinib and Ripretinib). For PDX-GIST-1 which carries KIT exon 11 (546_554 deletion) and KIT exon 17 (Y823D) mutations (primary Imatinib sensitive + acquired resistance), the efficacy of Imatinib was compromised, while Avapritinib and Regorafenib showed significantly higher tumor growth inhibition than other agents (Figure 5A). For PDX-GIST-2 that carries KIT exon 9 502-503 duplication mutation (primary Imatinib insensitive), a resistance to Imatinib was also observed. The best inhibitory effects were achieved by Avapritinib/Sunitinib and followed by Regorafenib, while Ripretinib had only weak antitumor effects (Figure 5B). All these regimens displayed comparable adverse effect on body weight as Imatinib, indicating acceptable toxicity (Supplementary Figure 5A). In addition to the reduction in tumor size, microscopic evaluation showed that Avapritinib/Regorafenib/Sunitinib treatment for PDX-GIST-1, as well as Sunitinib/Avapritinib/Regorafenib treatment for PDX-GIST-2, exhibited reductions in nuclear density (Supplementary Figure 5B). This reduction was due to mucus-like degeneration and found in GIST and other mesenchymal tumors, and the main feature was that connective tissue mucosa replaced tumor cells [23].

Figure 5.

Anti-tumor effects of TKIs in Imatinib-resistant GIST PDX. Antitumor activity of TKIs in human GIST PDX models with (A) KIT exon 11 (del546-554) plus exon 17 (Y823D) or (B) KIT exon 9 (dup502-503) mutations. Lysates were extracted from GIST-1 (C) and GIST-2 (D) xenografts after 21 days of treatment with the corresponding inhibitors and analyzed by western blotting to explore the downstream signaling responses. (E) A schematic diagram of the molecular mechanisms of acquired resistance revealed in this study.

We then compared the molecular changes induced by treatment of these TKIs. KIT phosphorylation and PI3K-Akt activation (represented by p-Akt and p-S6) were strongly repressed by Sunitinib/Regorafenib/Avapritinib yet mildly or seldom affected by other agents including Imatinib (Figure 5C, 5D). These data were in agreement with in vitro studies, validating that these TKIs exerted similar inhibitive effect on PI3K-Akt signaling as BEZ235, and blocked tumor growth induced by KIT mutation and subsequent PI3K-Akt in Imatinib-tolerant GIST. The resistance mechanism and spectrum for proper regimens were systematically summarized in Figure 5E.

Taken together, we concluded that Sunitinib/Regorafenib/Avapritinib were powerful regimens against Imatinib-tolerant GIST. The high tumor growth inhibition for both cell lines and PDXs indicated that under tolerable dosages, a complete therapeutic coverage by selecting these regimens was achievable for Imatinib-tolerant populations, including both KIT primary and secondary mutation-related resistance.

Discussion

The average annual incidence of gastrointestinal stromal tumor (GIST) was about 19 per million people [24]. As a pioneer of targeted therapy for solid tumors, Imatinib has been used in patients with recurrent metastatic GIST, with an overall response rate of up to 50% [25]. However, although most targeted drugs respond initially, the inevitable emergence of resistance has been a major problem that prominently hinders therapeutic responses. After receiving Imatinib treatment, approximately 9-13% of GIST patients developed primary resistance [26], which is currently considered to be closely related to the genotype of wild-type, KIT gene exon 9 mutations, and PDGFRA gene exon 18 D842V mutation [15,27]. Furthermore, most patients responded sensitively to Imatinib developed the acquired resistance during treatment, which severely impaired the rate of complete remission in advanced GIST populations. Secondary gene mutations, predominantly happened on KIT exon 13/17, were detected in 46.7-83% of GIST patients that acquired resistance [15-17,27-29], suggesting that secondary mutations are an important cause of acquired resistance. In this study, we applied multiple models to evaluate the impact of KIT secondary mutations on acquired Imatinib resistance and their potential therapeutic regimens. V654A and N822K were the most common type of KIT exon 13 and exon 17 mutations, whose transfections in GIST cells mimicked the acquisition of Imatinib resistance. Our data confirmed that the sensitivity of Imatinib in GIST cells was decreased after acquiring these secondary mutations, which was in accordance with previous research [30].

We validated that the acquired resistance to Imatinib induced by specific KIT mutations was most likely to be originated from an overactivation of PI3K-Akt signaling, as indicated by previous studies [31]. Notably, the overactivation of PI3K-Akt signaling was also observed in GIST cell lines with induced Imatinib resistance. Although we did not assess the genomic changes of KIT in these cells, the administration of the PI3K-Akt pathway inhibitor BEZ235, in certain degrees, augmented the anti-GIST efficacy of Imatinib against V654A and N822K for both cell lines, validating that PI3K-Akt signaling contributed to this resistance, and the combination with PI3K-Akt inhibitors might be an option in rescuing secondary mutation-associated Imatinib resistance.

However, the development of PI3K-Akt inhibitors in clinical practice was lagged behind due to failures in multiple trials [32], while it has also been reported that KIT protein was still the main dependence for Imatinib-resistant GIST cell lines to activate downstream signals and maintain growth [15]. As inspired by this KIT-dependency, seeking other tyrosine kinase inhibitors targeting a broader spectrum of KIT secondary mutations could be considered as a better strategy than PI3K-Akt inhibitors in order to overcome Imatinib resistance at current stage [30]. Increasing evidences suggested that several TKIs, including Sunitinib and Regorafenib, displayed promising prospects in multiple types of cancer [15]. These efforts led to the approval of Sunitinib and Regorafenib as second-line and third-line therapies respectively, for patients with advanced GIST [33-36]. GIST patients with Imatinib-resistant tumors are treated with Sunitinib, which potently inhibits KIT ATP-binding pocket mutations [37]. However, Sunitinib is ineffective against A-loop mutants, which account for 50% of Imatinib-resistant mutations [38]. This may explain why overall response rates (ORR) are low (7%) and median progression-free survival (PFS) is short (6.2 months) [39]. Regorafenib was approved as third-line therapy, but showed only moderate activity, with ORR of 4.5% and median PFS of 4.8 months [38].

A series of other multi-kinase inhibitors are currently assessed by phase I to phase III clinical studies [40,41], among which two TKIs are found with promising perspectives. The first is Avapritinib (BLU-285, Blueprint Medicines), an oral, highly-selective and potent investigational inhibitor of KIT and PDGFRA activation loop mutation [42]. In vitro, Avapritinib disrupts KIT signaling as assessed by inhibition of both KIT phosphorylation and activation of downstream proteins such as Akt in human mast cell and leukemia cell lines. In vivo, Avapritinib achieves dose-dependent tumor growth inhibition in a mouse model of systemic mastocytosis (SM). Moreover, Avapritinib also inhibits PDGFRA D842V [38], the mutation responsible for one out of five primary gastric GIST, for which there is no effective treatment available [26]. The other one is Ripretinib, a switch control type II inhibitor of KIT, which arrests KIT in an inactive state regardless of its activating mutations, such as KIT D816V [43]. Several additional oncogenic kinases, including FLT-3, PDGFRA, PDGFRB, KDR, TIE2 and FMS, are also recognized by Ripretinib [44-46].

Thus, Avapritinib and Ripretinib were currently under clinical investigation (NCT02508532, NCT03465722, NCT03862885 for Avapritinib, NCT04148092, NCT03353753, NCT036735-01, NCT02571036 for Ripretinib) in solid tumor patients (including GIST). Although Sunitinib, Regorafenib, Avapritinib and Ripretinib have been recognized as promising agents against GIST, their antitumor mechanisms and feasible types of KIT status still remained unexplored. Thus, we explored their therapeutic potential against Imatinib-resistant GIST, and proposed they were suitable options for KIT exon 13 and exon 17 secondary mutations. Notably, an inconsistency between GIST-T1- and GIST-882-related groups was observed regarding the responses to Imatinib and/or BEZ235, as well as to the other four KIT-targeted inhibitors. Since GIST-T1 carried a KIT exon 11-560_578 deletion while GIST-882 carried an exon 13-K642E mutation, it is of note that GIST-T1 resembled the primarily sensitive type while GIST-882 resembled the primarily insensitive type of GIST, which explained their diversity to drug treatments [18,19].

For KIT exon 13-mutation related Imatinib resistance, a previous report pointed out that Ripretinib was most effective to GIST-430 cells carrying KIT exon 11 and 13-V654A mutations [43], yet our research exhibited that Ripretinib was less effective to KIT exon 13 mutations. For KIT exon 17 mutation related Imatinib-resistance, although Ripretinib showed strong antitumor effects in vitro in Imatinib-resistant GIST cells with KIT exon 17-N822K mutation, its inhibition (represented by tumor growth inhibition) on the in vivo growth of PDX-GIST-1 carrying KIT exon 11 and exon 17-Y823D mutations was less efficient than Avapritinib, Regorafenib or even Sunitinib. Therefore, higher dosages of Ripretinib may be required to achieve the same effect against Imatinib-resistant GIST as Regorafenib and Avapritinb, which complied with a report that 50 mg/Bid was needed in the body to achieve an eminent antitumor effect [43]. Despite these diversities, investigation for both cells consistently indicated that Sunitinib was an option for KIT secondary exon 13 mutations, while Regorafenib/Avapritinib/Ripretinib were better candidates for secondary exon 17 mutations (with acquired Imatinib resistance). Moreover, our findings suggested that Sunitinib was also effective for PDX-GIST-1 carrying KIT exon 17 mutations, which finding was in consistent with a previous research [47]. However, due to the diversity between cellular and PDX data, it remained to be answered whether Suntinib could be considered applicable for both KIT exon 13 and exon 17 mutations.

Apart from exon 13 and exon 17 mutations, we also verified the anti-tumor effects of the four TKIs for GIST (PDX-GIST-2) that carried KIT exon 9 mutations. As previously described, GIST KIT exon 9 mutation was considered to be a primary insensitive phenotype. Both primary insensitivity (represented by KIT exon 9) and acquired resistance (represented by KIT exon 11+exon 13, or exon 11+exon 17 mutations) to Imatinib could be effectively overcame by combining these TKIs, suggesting that although with heterogenic efficacies, these TKIs were valuable therapeutic complements or even potential substitutions for Imatinib. Thus, it is worthy to keep promoting the development of relevant clinical trials.

In conclusion, we proved that the overactivation of PI3K-Akt signaling induced by multiple KIT mutations mediated Imatinib-resistance in GIST. The inhibition of PI3K-Akt signaling restored the efficacy of Imatinib, while other newly developed TKIs, especially Sunitinib, Regorafenib and Avapritinib, were realistic choice for clinical applications. By shedding light upon the new strategies for major problem of current targeted therapy, our work expanded the armory against Imatinib-resistant GIST, which might benefit a large number of patients.

Conclusion

Through applying multiple GIST models, we unveiled the involvement of specific KIT mutations in activating PI3K-Akt signaling and inducing Imatinib resistance. By evaluating the preclinical efficacy of PI3K-Akt inhibitor BEZ235 as well as several newly developed TKIs, we verified that these agents were potential options to overcome Imatinib resistance for GIST patients in the future.

Disclosure of conflict of interest

None.

Abbreviations

GIST

gastrointestinal stromal tumor

TKI

tyrosine kinase inhibitor

PDGFRA

platelet-derived growth factor receptor alpha

PDX

patient-derived xenograft

ORR

overall response rate

PFS

progression-free survival

WT

wild-type

KEGG

Kyoto Encyclopedia of Genes and Genomes

PD

progressive disease

PR

partial response

CR

complete response

SD

stable disease