Precision Oncology in Gastrointestinal Stromal Tumors

GIST (gastrointestinal stromal tumors) represent 20% of sarcomatous tumors and 1–2% of primary gastrointestinal cancers. They have an excellent prognosis when localized and resectable, though their prognosis is poor in the metastatic setting, with limited options after the second line until recently. Four lines are now standard in KIT-mutated GIST and one in PDGFRA-mutated GIST. An exponential growth of new treatments is expected in this era of molecular diagnostic techniques and systematic sequencing. Currently, the main challenge remains the emergence of resistance linked to secondary mutations caused by selective pressure induced by TKIs. Repeating biopsies to tailor treatments might be a step in the right direction, and liquid biopsies at progression may offer a non-invasive alternative. New molecules with wider KIT inhibition are under investigation and could change the catalog and the sequence of existing treatments. Combination therapies may also be an approach to overcome current resistance mechanisms. Here, we review the current epidemiology and biology of GIST and discuss future management options, with an emphasis on genome-oriented therapies.


Epidemiology
GIST (gastrointestinal stromal tumors) represent 20% of sarcomatous tumors and 1-2% of primary gastrointestinal cancers. They typically present in older individuals and are most common in the stomach (60-70%), followed by the small intestine (20-25%), colon and rectum (5%), and esophagus (<5%) [1]. GIST can also occur in children and young adults, especially when a genetic predisposition is involved [2]. They are equally common in male and female patients [3].

Management of Localized GIST
Surgery is the treatment of choice for resectable GIST if no major functional losses are to be expected. The tumor should be resected completely, and rupture needs to be avoided [5]. Patients at high risk of recurrence need post-operative imatinib for a duration of a minimum of 3 years if well tolerated [15].
Different scores are used to determine the risk: The majority of these gain-of-function mutations are found in the KIT gene, with two-thirds and the most frequent within exon 11. Exon 9 mutations are often found in small or large bowel tumors and represent approximately 10% of GIST. Deletions, deletioninsertion, point mutations, duplications, insertions, and inversion have been identified in KIT. Deletions in exon 11, especially involving codons 557 and 558, are associated with a poorer natural prognosis compared to exon 11 point mutations [11]. Mutations in exons 13,17, and 18 are rare [12].
Furthermore, 5 to 10% of mutations are related to the PDGFRA oncogene, more specifically to exons 12, 14, and 18. PDGFRA mutations are more commonly found in the stomach. Exon 18 of PDGFRA is the most frequently mutated region, with exon 18 D842V mutations accounting for 70% of PDGFRA-mutant cases, while exons 12 or 14 are rarely mutated. [12] KIT and PDGFRA mutations are considered mutually exclusive [13].
Despite extensive sequencing, some GIST remain "wild-type" GIST, but a number have been identified nowadays with the modern techniques to carry low-frequency KIT or PDGFRA mutations, while others were associated with fusions of NTRK (NTRK3-ETV6) and FGFR1 (FGFR1-HOOK3, FGFR1-TACC1), alterations in the RAS-MAPK pathway with BRAF mutations, NF1 mutations, or SDHA deficiency caused by a germline mutation in the suppressor genes encoding the SDH complex or by SDHC promotor methylation [13,14].

Management of Localized GIST
Surgery is the treatment of choice for resectable GIST if no major functional losses are to be expected. The tumor should be resected completely, and rupture needs to be avoided [5]. Patients at high risk of recurrence need post-operative imatinib for a duration of a minimum of 3 years if well tolerated [15].
Different scores are used to determine the risk: The "Miettinen classification", also known as the Armed Forces Institute of Pathology (AFIP) classification, includes the mitotic index, the size of the tumor, and the location.
Joensuu's classification, also known as modified NIH classification, uses the parameters above and integrates the pejorative nature of a perforation. It aims to better split the GIST between intermediate and high risk [16].
All classifications stratify patients into very low, low, intermediate, and high-risk categories of recurrence. The indication for adjuvant treatment depends on the risk score, but also on the mutational status [17]. As an example, patients with the PDGFRA D842V mutation do not receive adjuvant treatment as this is a resistance mutation to imatinib [18].
Molecular prognostic factors such as the level of tumor genome rearrangement have been investigated and are currently evaluated in intermediate-risk GIST in the GI-GIST trial (NCT02576080).

First-Line Treatment in the Metastatic Setting: Imatinib
Imatinib is an inhibitor of KIT, PDGFRA, and BCR-ABL tyrosine kinase and is used to treat inoperable or metastatic GIST. Standard dosage is 400 mg for all sensitive mutations. Indeed, sensitivity to imatinib for exon 9 and exon 11 mutated GIST was demonstrated in the lead trial evaluating imatinib in GIST [19], while KIT exon 9 mutations are treated with 800 mg/d as this provides longer progression-free survival (PFS) [20][21][22].
Currently, imatinib is continued until progression or intolerance in the metastatic setting. The exon 13 and 17 activation loop mutations are essentially secondary mutations that occur on imatinib therapy. Primary exon 13 mutations exist and are often sensitive to imatinib in vitro. Given the rarity of this mutation, in vivo sensitivity is not clear as little evidence exists to date [23].
Currently, imatinib is continued until progression or intolerance in the metastatic setting. Primary PDGFRA mutations occur mainly in exon 18 and exon 12, which, respectively, encode the activation loop in the juxta-membrane domain, but more rarely in exon 14, which encodes the ATP-binding domain [24]. While D842V, the most common exon 18 mutation, confers a primary mutation to imatinib, other types of PDGFRA mutations are sensitive to imatinib [25].
3.2. Primary Resistance to Treatment 3.2.1. Pseudo-Resistance: Imatinib Plasma-Levels and Pharmacokinetics Demetri et al. showed in a small group of patients that imatinib plasma levels above 1100 ng/mL were associated with clinical benefit and a longer time to disease progression [26]. Pointing in the same direction, in chronic myeloid leukemia, Gotta et al. proved in a prospective randomized controlled trial that imatinib dose monitoring helps in achieving efficient plasma concentrations [27]. In another trial, durable effective imatinib concentrations were reached only by 33.3% [28], raising the question if imatinib should be dosed individually.
Several treatments such as proton pump inhibitors can have interactions with oral oncology treatments. Indeed, the concomitant use of TKIs and proton pump inhibitors can reduce TKI absorption, thus potentially reducing the effectiveness of TKIs [29].
Genetic polymorphisms of cytochrome P450 can participate in interpatient variability in imatinib blood levels [30]. At progression, the use of high-dose imatinib (800 mg daily) has shown benefit to patients with advanced or metastatic GIST that progressed on the standard dose [31].
Compliance issues are also a major factor to consider when suspecting pseudoprogression. Indeed, in a study made on imatinib-treated patients who had a diagnosis of CML or GIST, compliance with imatinib was about 75%, with 30% of patients interrupting therapy for at least 30 consecutive days in the first year [32].
Avapritinib has been specifically developed to target this mutation and the NAVIGA-TOR study showed high efficacy of this drug with above 90% overall response rate (ORR) and a duration of response of 70% at 1 year [34][35][36][37].
To note, among 167 patients starting on 300 mg of avapritinib, 37.0% of all patients and 52.0% of patients older than 65 years showed cognitive dysfunction. This toxicity decreased faster to a lower grade with dose modification (1.3-3.1 weeks) than without (4.9-7.6 weeks). Median PFS was 11.4 months with dose modification and 7.2 months without. As a result, early recognition of neuro-toxicity and adapted dose modification can help maintain patients on this treatment. It can be recommended to assess cognitive function at baseline and monitored [38].

Rare Non-KIT/PDGFRA Molecular Subtypes
In the past, GIST have been classified as KIT or PDGFRA mutated according to testing that included the frequent mutations. GIST missing these mutations have been classified as "wild type" (WT). Later, the concept of quadruple negative WT-GIST was coined, and additional subgroups were identified. Recently, new techniques have detected low-frequency mutations [39]. The latter probably explains why some GIST considered to be WT respond to TKI [40,41].
Within the WT-GIST category, the following "new" subtypes have been identified.

SDH-Deficient (5 to 8%) [42]
About 20-40% of WT-GIST show defects in succinate dehydrogenase (SDH) complexes. Some mutations can be germline [43] and involve genetic testing. Few data are available in SDH-deficient GIST due to their rarity. The early, large trials did not classify WT-GIST further [44].
A retrospective analysis of 87 patients with SDH-deficient GIST found better response rates with sunitinib than with imatinib, which had very little activity in this cohort. Nilotinib and vandetanib have also been used, but limited data do not support the use of these drugs [45].
The dysfunction of the SDH complex in these tumors leads to a pseudo-hypoxic phenotype suggesting a potential benefit of TKIs with anti-angiogenic activity. In the phase II REGISTRI trial, an ORR of nearly 20% was achieved with sunitinib, and regorafenib was associated with an 87% disease control rate [46]. Other TKIs such as regorafenib and pazopanib showed limited efficacy.
A novel third-generation TKI, olverembatinib, has shown antitumor activity in patients with TKI-resistant SDH-deficient GIST in a phase Ib/II study with 2 partial responses in 6 evaluable patients and 1 with stable disease for 36 cycles, calling for its further investigation [47].
As SDH-loss causes succinate accumulation and activation of pseudohypoxia signaling [48] via overexpression of HIF-proteins, specific HIFa inhibitors such as belzutifan are under development in different types of tumors associated with overexpression of HIF (NCT04895748, NCT04924075) and could be good candidates for SDH-deficient GIST in the future.
Furthermore, temozolomide has potential interest in this subgroup with promising results in five patients (100% disease control rate, 40% ORR), motivating an ongoing phase II study (NCT03556384). Overexpression of FGF/FGFR signaling pathways has also been reported in SDH-deficient GIST, and a new pan-FGFR inhibitor, rogaratinib, is being evaluated in a phase II trial (NCT04595747).
NF1 mutations (0.1 to 2.4%) Neurofibromatosis 1 (NF1) is an autosomal dominant disorder caused by germline mutations in the NF1 tumor suppressor gene causing its inactivation. There is an estimated 7% of individuals with NF1 who will develop GIST during their life [42].
To date, there is no standard treatment for NF1-mutated GIST, which do not respond to TKIs, and surgery remains the main option for these patients. However, these tumors appear to be more indolent [55].
MEK inhibitors might have clinical efficacy in other NF1-associated tumors, given the role of the RAF/MEK/ERK pathway in neurofibromas. A phase II trial with mirdametinib, a MEK inhibitor, for adolescents and young adults with NF1-associated plexiform neurofibromas showed a 42% partial response and a significant decrease in pain ratings [56].

KIT and PDGFRA Secondary Mutations
In the second line, 67% of patients have one or more secondary mutations involving KIT exon 17, exon 13, and exon 14, causing resistance to imatinib [57].
Point mutations associated with imatinib resistance are usually located in the drug/ ATP-binding pocket of the receptor (encoded by exons 13 and 14) or in the activation loop (encoded by exon 17) [58].
For avapritinib, mechanisms of secondary resistance in PDGFRA-mutant GIST involve compound mutations of exons 13, 14, and 15 of PDGFRA with codon 658 and 680 mutations representing a recurrent cause of resistance [59].

Second-Line Sunitinib
Sunitinib was approved after a phase III trial in patients with GIST failing or intolerant to imatinib and showed a longer time to progression (27.3 versus 6.4 weeks, p < 0.0001) in patients with sunitinib than placebo.
Of interest, higher response rates were observed among GIST with a primary KIT exon 9 mutation [60].
Antiangiogenic effects of sunitinib treatment may contribute to its effectiveness. Indeed, sunitinib selectively inhibits PDGFRB and VEGFR in addition to KIT and PDGFRA, whereas imatinib inhibits PDGFRB but not VEGFR. Sunitinib has, however, little activity against secondary mutations involving the KIT activation loop (exons 17 and 18) [61].

Third-Line Regorafenib
Regorafenib is a multikinase inhibitor and also inhibits VEGFR and showed a significant improvement in median PFS (4.8 vs. 0.9 months, HR 0.27, p < 0·0001) over placebo in patients already treated with imatinib and sunitinib, leading to its approval in the third line [62]. It has become the treatment of choice in patients with an exon 17 mutation, as these do not respond to sunitinib [63].

Third-Line Pazopanib
Pazopanib has been tested in the third line, following imatinib and sunitinib failure, and showed modest benefit with a median PFS of 3.4 months (95% CI 2.4-5.6) vs. 2.3 months in best supportive care only [64].
In this setting, ripretinib demonstrated a PFS benefit regardless of the primary mutation [67]. Ripretinib did not improve PFS when tested in the second-line INTRIGUE trial against sunitinib, but improved response rates (23.9 vs. 14.6%). OS data are immature. Sunitinib showed better PFS in the exon 9 subgroup [68].
The failure of ripretinib in the second line compared to sunitinib might be explained by the emergence of secondary exon 13 mutations, which decrease ripretinib efficacy compared to sunitinib [68].
Interestingly, a Chinese phase II study, testing ripretinib in the second line, showed a benefit in PFS for ripretinib [69].

Third-Line Avapritinib
VOYAGER, a randomized, phase III trial, tested avapritinib versus regorafenib in the third line. In molecularly unselected patients, the primary end point was not met. There was no significant difference in median PFS between avapritinib and regorafenib, but in the selected subgroup of patients with the PDGFRA exon 18 D842V mutation, which is resistant to the other TKIs, avapritinib showed high response rates [70] (Table 1).

Immunotherapy
The activity of anti-PD-1/PD-L1 and CTLA-4 drugs and combinations have been investigated in GIST. In a randomized phase II trial with 40 patients, nivolumab with or without ipilimumab showed only modest response rates. Other clinical trials based on the same combination are ongoing [71]. No or little clinical efficacy was seen with ipilimumab plus dasatinib [72], epacadostat with pembrolizumab [73], and in the subgroup of 31 patients with GIST of the PEMBROSARC trial treated with pembrolizumab and metronomic cyclophosphamide, which showed a low 2.3% ORR [74]. Ongoing trials are investigating avelumab with axitinib (NCT04258956, AXAGIST) or with regorafenib (NCT0347595, REGOMUNE), spartalizumab with imatinib (NCT03609424), or with TNO155 or ribociclib (NCT04000529).
The retrospective analysis of the Sarc028 study showed the presence of tertiary lymphoid structures (TLSs), found to be associated with better response and longer PFS [75]. Today, immunotherapy in GIST requires further evaluation in prospective trials, possibly guided by new biomarkers.
Bezuclastinib (CGT9486) showed good clinical benefit and a median PFS of 12 months in an early phase trial and is currently tested with or without sunitinib in a phase 3 clinical trial (NCT05208047).
THE 630, a pan KIT inhibitor is in testing in a phase I/II study and showed good preclinical results (NCT05160168).

Pimitespib
Heat shock protein 90 is necessary for the stabilization of KIT and PDGFRA. Several HSP90 inhibitors had preclinical activity in GIST.
In a randomized placebo-controlled phase III trial (CHAPTERGIST-301), pimitespib, also known as TAS 116, a novel HSP90 inhibitor tested in the fourth line, showed an improved PFS and OS compared with placebo in patients with previously treated advanced GIST. Exploratory pharmacogenomic analysis showed a benefit irrespective of KIT mutation status. As a result, this therapy has been approved in the fourth line in Japan for the treatment of metastatic GIST. Visual impairment was reported in patients receiving pimitespib, with 13% of night blindness, and two cases of retinal vein occlusion and visual impairment, which resolved with discontinuation [76].
A phase I study evaluating pimitespib in combination with imatinib (NCT05245968, CHAPTERGIST-101) is ongoing.

Intratumoral Vaccination
Ilixadencel is a cell-based immune primer injected intratumorally that has been clinically investigated in metastatic renal cell carcinoma and hepatocellular carcinoma. It has been evaluated in a phase I study and presented an acceptable safety profile and radiological tumor responses in 33% of treated patients. Further investigation is needed [77].

Discussion
Despite the rarity of GIST, multicentric and multinational trials have led to the approval of four lines of treatment, and an exponential growth of new treatments is expected in this era of molecular diagnostic techniques and systematic sequencing.
Currently, the main challenge remains the emergence of resistance linked to secondary mutations caused by selection pressure induced by TKIs [78]. Future studies might select patients according to the secondary mutations, either on the basis of a (repeated) solid or liquid biopsy. Indeed, the failure of ripretinib in the second line compared to sunitinib might be explained by the emergence of secondary exon 13 mutations, which decrease ripretinib efficacy compared to sunitinib [62]. Furthermore, liquid biopsies may overcome the challenge of tumor heterogeneity and aid in detecting relevant resistance clones. An exploratory study of the NAVIGATOR trial for PDGFRA-mutant GIST showed that mutant ctDNA was detected in 63% of patients, and in this same population, the median sum of target lesions was 18.2 cm, suggesting that ctDNA detection might be potentially limited by the tumor burden [59]. Further research is needed to validate the optimal approach [79][80][81][82].
Imatinib has been the first-line standard since the discovery of its efficacy in GIST, and current studies focus mainly on later lines given the excellent tolerance profile of the molecule. New studies might focus on earlier lines to delay the emergence of resistance. Nonetheless, imatinib remains the gold standard, and the use of an experimental first-line TKI does not seem to reduce imatinib effectiveness in the second line [83,84]. Sequencing TKIs is an important challenge in the management of GIST, and a tailored approach based on an identified resistance mechanism will be an important part of future therapies.

Conclusions
GIST have a rich molecular landscape that is being unraveled thanks to modern genomic analyses. Four lines of therapy are currently standard, but the prognosis after the second line is poor and optimal management of secondary mutations remains a challenge.
Repeating biopsies to tailor treatments might be a step in the right direction, and liquid biopsies at progression may offer a non-invasive alternative.
New molecules with wider KIT inhibition are being tested and could change the catalog and the sequence of existing treatments. Combination therapies may also be an approach to overcome current resistance mechanisms [85], either by targeting these directly or avoiding their development.

Conflicts of Interest:
The authors declare no conflict of interest.