Background

Gastrointestinal stromal tumors (GISTs) are neoplasms arising from mesenchymal tissue of the GI tract. The most common sites of GISTs are the stomach and small intestine. Most GISTs can be managed with curative surgery followed by adjuvant imatinib treatment or not. However, approximately 40% of patients may develop metastasis [1]. Imatinib is a tyrosine kinase inhibitor and has been shown to inhibit KIT phosphorylation and cell proliferation of c-KIT mutated HMC-1 cells by Heinrich in 2000 [2]. In 2001, Joensuu et al. have reported that an advanced GIST patient with c-KIT exon 11 mutation, who progressed after multiple treatment, including surgery, chemotherapy, thalidomide and interferon alfa, had a good response to imatinib treatment [3]. After that, imatinib has been shown to induce an overall response rate of approximately 45–68% and a median progression-free survival (PFS) and overall survival (OS) of 18–26 and 51–57 months, respectively, according to randomized phase II and III clinical trials [4,5,6]. Imatinib has become the standard first-line therapy for advanced GISTs. Many prognostic factors, such as sex, tumor size, mitotic count, Eastern Cooperative Oncology Group performance status (ECOG PS), neutrophil count, albumin level, and genetic alterations, have been reported to be associated with the outcome of imatinib treatment for advanced GISTs [6,7,8,9].

GIST patients may develop resistance to imatinib; however, sunitinib has been approved as a second-line treatment and is associated with a PFS of 6.8 months for advanced GIST patients who are intolerant to or in whom imatinib failed [10]. Moreover, regorafenib was approved in 2013 as a third-line treatment for advanced GIST patients who are intolerant to imatinib or sunitinib or in whom these treatments failed [11]. Both sunitnib and regorafenib are multi-targeted tyrosine kinase inhibitors. They competitively inhibit ATP-binding sites of several receptor tyrosine kinases, including KIT [12]. There are also prognostic factors reported for sunitinib treatment in advanced GIST patients, such as the neutrophil to lymphocyte ratio (NLR), neutrophil and platelet count and genotype [13,14,15,16]. However, the results of these studies are controversial. Similarly, performance status, biological factors (NLR, platelet to lymphocyte ratio), and genotype have also been reported to be associated with the outcome of regorafenib treatment in advanced GIST patients [17, 18]. Regarding genotype of GIST, around 75% of GISTs harbor c-KIT mutations. Around 10% to 20% of GISTs harbor PDGFRA and around 5% to 10% of GISTs do not harbor c-KIT or PDGFRA mutation. The genotype of GISTs is associated with clinical features and sensitivities to tyrosine kinase inhibitors [8, 19,20,21,22]. For example, PDGFRA mutation is more associated gastric location whereas c-KIT exon 9 mutation is more associated with intestinal location [19]. GIST patients with c-KIT exon 9 mutation is associated with lower response rate to imatinib than those with c-KIT exon 11 mutation and higher dose of imatinib is suggested for GIST patients harboring c-KIT exon 9 mutation [19, 22]. Imatinib, sunitinib and regorafenib have been approved and reimbursed by the health bureaus in Taiwan since 2002, 2010, and 2016, respectively. Thus, we conducted a registry study of GIST patients to understand the demographics, genotypes, treatment patterns, and treatment outcomes of these patients in Taiwan. In this registry study, we analyzed the baseline characteristics, treatment outcomes and prognostic factors for recurrent or metastatic GISTs in patients who received tyrosine kinase inhibitor (TKI) treatment.

Methods

Patients, study design and data collection

This was a longitudinal multicenter registry study (Taiwan Cooperative Oncology Group T1218). Eleven hospitals located from the northern to southern Taiwan, including Taiwan University Hospital, Taipei Veterans General Hospital, Mackay Memorial Hospital, Tri-Service General Hospital, Linkou Chang Gung Memorial Hospital, China Medical University Hospital, Changhua Christian Hospital, National Cheng Kung University Hospital, Kaohsiung Medical University Hospital, Kaohsiung Veterans General Hospital, and Kaohsiung Chang Gung Memorial Hospital, participated in this study. Pathologically proven GIST patients diagnosed between January 1, 2010, and December 31, 2020, were included. The data for enrolled patients were collected via chart review. The collected data included baseline characteristics, treatment strategies (surgery, TKI treatment, and any other therapy), genetic profiles, best treatment responses, PFS, OS and adverse events associated with TKI therapy. We retrospectively collected the data prior to enrollment and prospectively collected the data after enrollment. The study protocol was reviewed and approved by the Institutional Review Board of each participating institution. All patients except those who had died signed informed consent. In the current study, we analyzed the clinical outcomes of patients who received TKI treatment. Patients who received TKIs (imatinib, sunitinib and regorafenib) for less than 30 days were excluded from the analysis.

Genetic analysis of GIST patients

Some patients had c-KIT and PDGFRA aberrations detected prior to enrollment. The data were recorded in our database system. Some patients did not undergo a check for c-KIT and/or PDGFRA before enrollment. Their tumor samples were sent to our central laboratory for c-KIT and PDGFRA assessment by Sanger sequencing. The detailed methods of DNA extraction, PCR, and Sanger sequencing, and the primer sequences for c-KIT and PDGFRA were described in Supplementary Method. Wild-type c-KIT and PDGFRA were defined as no aberration detected for c-KIT exons 9, 11, 13, 14 and 17 and PDGFRA exons 12, 14, and 18 and was called “wild-type c-KIT/PDGFRA” in this study. Ninety-five percent of tumor samples used for assessment of c-KIT and PDGFRA were free of TKI treatment.

Statistical analysis

All statistical analyses were performed using SAS statistical software (Version 9.4, SAS Institute Inc., Cary, NC, U.S.A.). Descriptive analyses were performed to examine the baseline characteristics and genetic alterations of patients. Data were summarized using descriptive statistics (number of patients, mean, standard deviation, median, minimum, and maximum) for continuous variables and using frequency and percentage for categorical variables. Fisher’s exact test was used to compare the differences in baseline characteristics for categorical variables, while t-test was performed for continuous variables. If the normality assumption for a continuous endpoint was violated, the nonparametric Wilcoxon–Mann–Whitney test was applied. PFS was defined as the time from the date of TKI use (imatinib, sunitinib, or regorafenib) to the date of disease progression. OS was defined as the time from the date of TKI use to the date of patient death due to any cause or to the last date that the patient was known to be alive. Probabilities of PFS and OS were estimated by the method of Kaplan–Meier. Univariate Cox proportional hazards model was used to evaluate the risk associated with baseline characteristics for the PFS and OS. Baseline variables with significant difference from the univariate analyses were selected in multivariate analysis. Hazard ratios and 95% confidence intervals were also calculated. All tests were two-tailed. A P value < 0.05 was considered significant.

Results

Demographics of advanced/metastatic or recurrent GIST patients who received imatinib treatment

There were 224 patients who received imatinib treatment due to a diagnosis of advanced/metastatic disease or recurrent disease after prior surgery with or without adjuvant imatinib therapy. They were classified into 4 groups by the timing of imatinib use for controlling advanced/metastatic or recurrent disease but not for adjuvant purpose. Eighty-nine patients were treated with primary front-line imatinib. Seventy-six patients were treated with imatinib due to recurrent disease after surgery with adjuvant imatinib. Twenty-seven patients were treated with imatinib due to recurrent disease after prior surgery without adjuvant imatinib. Thirty-two patients were treated with imatinib after palliative surgery. The male to female ratio was 1.43. The median age of all 224 patients was 57.8 years (19.6 to 85.1). The ECOG PS scores among the 121 patients with known ECOG PS were 0 for 55 patients and 1 for 55 patients. Eighty-five (37.9%) patients had GISTs of the stomach. Ninety-seven (43.3%) patients had GISTs of the small intestine, which included duodenum, jejunum, ileum, and other small intestine with 20, 43, 21, and 13 patients, respectively. The other primary sites of GISTs included the esophagus, colon, rectum, peritoneum, vagina, prostate and one undetermined site. The baseline characteristics, including stage, of these 224 patients by primary site of the stomach and nonstomach are shown in Table 1. The known genetic data among the patients are also listed in Table 1. One hundred thirty-six (76.8%) patients had c-KIT exon 11 mutations with mutational type of deletion, missense mutation, deletion + missense mutation, deletion and insertion, and duplication. Eighteen (10.2%) patients had c-KIT exon 9 mutations. Seventeen of the 18 patients had 502–503 duplication mutation and the other one had insertion mutation. Three (1.7%) patients had PDGFRA mutations. Two of the 3 patients had D842V mutation and the other one had V561D mutation. Seventeen (9.6%) patients had wild-type c-KIT/PDGFRA. The percentages of the baseline albumin level and NLR in the patients are also listed in Table 1.

Table 1 baseline characteristics of the patients who received imatinib treatment
Full size table

Genetic mutation profiles among advanced/metastatic or recurrent GIST patients who received TKI treatment

We analyzed the genetic alterations in patients to understand whether these alterations were different according to the baseline patient characteristics (Table 2). Higher percentage of male patients had c-KIT exon 9 mutations (men vs women, 72.2% vs 27.8%), PDGFRA mutations (men vs women, 100.0% vs 0%), and wild-type c-KIT/PDGFRA (men vs women, 82.4% vs 17.6%) than female patients (P = 0.0215). Higher percentage of nonstomach GIST patients had c-KIT exon 9 mutations (nonstomach vs stomach, 94.4% vs 5.6%) and wild-type c-KIT/PDGFRA (nonstomach vs stomach, 76.5% vs 23.5%) than stomach GIST patients (P = 0.01). Higher percentage of patients aged less than 60 years had c-KIT exon 9 mutations (< 60 years vs ≥ 60 years, 83.3% vs 16.7%) and wild-type c-KIT/PDGFRA (< 60 years vs ≥ 60 years, 76.5% vs 23.5%) than those aged more than 60 years (P = 0.0205). The genetic alterations were not significantly different among patients with different ECOG PS scores.

Table 2 Genetic alterations in advanced/metastatic or recurrent GIST patients who received TKI treatment
Full size table

PFS and OS of advanced/metastatic or recurrent GIST patients who received imatinib treatment

We analyzed the PFS of 224 advanced/metastatic or recurrent GIST patients who received imatinib treatment in our registry study. The median PFS was not reached, and the 48-month PFS rate was 50.5% (Fig. 1A). We investigated whether the PFS or OS was different among patients with initial advanced/metastatic disease with or without palliative surgery or recurrent disease after prior surgery with or without adjuvant imatinib treatment. The PFS among these 4 groups was not significantly different, as shown in Supplementary Fig. 1. The PFS analysis in patients was also not significantly different by sex, ECOG PS score, baseline albumin level, and baseline NLR, as shown in Supplementary Fig. 2. The median PFS of all patients is shown in Fig. 1A. The median PFS was significantly different in patients who received imatinib by primary site and genetic profiles. The median PFS of patients with a primary site of the stomach was not reached, and the median PFS of patients with a nonstomach primary site was 41.45 (95% confidence interval (CI), 27.68–55.22) months (Fig. 1B) (P = 0.026). Patients with c-KIT exon 11 mutations (not reached) and wild-type c-KIT/PDGFRA (not reached) had a longer PFS than patients with c-KIT exon 9 (median 12.5 (95% CI, 7.12–17.89) months) and PDGFRA mutations (median 2.93 (95% CI, 0.00–6.14) months) (P < 0.001) (Fig. 1C).

Fig. 1
figure 1

PFS of advanced or recurrent GIST patients who received imatinib treatment. A PFS of all GIST patients. B PFS of GIST patients by primary site. C PFS of GIST patients by genetic alterations

Full size image

The median OS was not reached, and the 48-month OS rate was 79.5% (Fig. 2A). The OS among the 4 groups of patients who received imatinib was not significantly different. The OS was also not significantly different in patients according to sex, ECOG PS score, primary site, and genetic alterations, as shown in Supplementary Fig. 3. Regarding genetic alterations in patients, the OS among all groups with different genetic alterations was not significantly different. Patients with PDGFRA mutations had a significantly shorter OS (median OS 18.00 (95% CI, 4.10–31.89) months) than patients with other genetic alterations. The OS of patients with c-KIT exon 9 mutations was not significantly different from that of patients with c-KIT exon 11 mutations. The OS of all patients is shown in Fig. 2A. The OS was significantly different in patients according to age (Fig. 2B), baseline albumin level (Fig. 2C) and baseline NLR (Fig. 2D). The OS of patients aged ≥ 60 years was worse than that of patients aged < 60 years (P = 0.011). The OS of patients with a baseline albumin level < 3.2 g/dl was worse than that of patients with a baseline albumin level ≥ 3.2 g/dl (P = 0.014). The OS of patients with a baseline NLR ≥ 3.0 was worse than that of patients with a baseline NLR < 3.0 (P = 0.041).

Fig. 2
figure 2

OS of advanced or recurrent GIST patients who received imatinib treatment. A OS of all GIST patients. B OS of GIST patients by age. C OS of GIST patients by baseline albumin level. D OS of GIST patients by baseline NLR

Full size image

Univariate and multivariable Cox regression analysis of PFS and OS in advanced/metastatic or recurrent GIST patients who received imatinib treatment

We performed univariate and multivariable Cox regression analyses to evaluate the risk associated with baseline characteristics for the PFS of these patients. Supplementary Table 1 shows the univariate Cox regression analysis for the PFS of these patients by sex, age, ECOG PS score, primary site, baseline albumin level, baseline NLR and genetic alterations. In univariate analysis, patients with mutations in c-KIT exon 9 (hazard ratio (HR) = 3.057, 95% CI, 1.694–5.519, P = 0.0002) or PDGFRA (HR = 13.178, 95% CI, 3.015–57.597, P = 0.0006) had a higher HR than patients with mutations in c-KIT exon 11. Patients with a primary site in the stomach had a lower risk (HR = 0.617, 95% CI, 0.401–0.949, P = 0.0280) than those with a nonstomach primary site. In multivariable analysis (Table 3), only c-KIT exon 9 (HR = 2.997, 95% CI, 1.620–5.544, P = 0.0005) and PDGFRA mutations (HR = 13.609, 95% CI, 3.029–61.138, P = 0.0007) were still statistically significant, with a higher HR for PFS in these patients.

Table 3 Multivariate Cox regression analysis for PFS of recurrent/metastatic GIST patients treated with imatinib treatment
Full size table

The univariate Cox regression analysis for the OS of these patients by sex, age, ECOG PS score, primary site, baseline albumin level, baseline NLR and genetic alterations is shown in Supplementary Table 2. PDGFRA mutation (HR = 4.815, 95% CI, 1.140–20.345, P = 0.0325), age ≥ 60 years (HR = 2.074, 95% CI, 1.164–3.694, P = 0.0133), and baseline NLR ≥ 3.0 (HR = 1.987, 95% CI, 1.014–3.895, P = 0.0454) were risk factors for poor OS in patients. Baseline albumin level ≥ 3.2 g/dl (HR = 0.365, 95% CI, 0.158–0.841, P = 0.0180) was a favorable factor for OS in these patients. In multivariate analysis, only PDGFRA mutation (HR = 98.670, 95% CI, 5.200–1872.32, P = 0.0022) was a risk factor for poor OS in these patients (Supplementary Table 3).

c-KIT exon 11 mutation and the impact of exon 11 mutational type on the PFS and OS of advanced/metastatic or recurrent GIST patients who received imatinib treatment

We analyzed the mutational type patterns of the 136 recurrent or advanced/metastatic GIST patients with c-KIT exon 11 mutation who received imatinib treatment (Supplementary Table 4). The number of patients with c-KIT exon 11 deletion, missense mutation, deletion + missense mutation, deletion and insertion, and duplication was 69, 24, 32, 8, and 2, respectively. One patient’s mutation type was unknown. There was no difference in the percentage of mutational type between GISTs of the stomach or nonstomach (P = 0.2263). The PFS of patients with the 3 major types (deletion, missense mutation, deletion + missense mutation) of c-KIT exon 11 mutations is shown in Supplementary Fig. 4A. The PFS of patients with deletions was better than that of patients with deletions + missense mutations and missense mutations in c-KIT exon 11 (P = 0.023). The median PFS of patients with deletion and deletion + missense mutations in c-KIT exon 11 was not reached. The median PFS of patients with c-KIT exon 11 missense mutations was 33.13 (95% CI, 14.41–51.84) months. Supplementary Fig. 4B shows the OS of patients with the 3 major types of mutations in c-KIT exon 11. The OS of patients with deletions was better than that of patients with deletions + missense mutations and missense mutations in c-KIT exon 11 (P = 0.004).

PFS and OS of advanced/metastatic or recurrent GIST patients who received sunitinib treatment

Most of the patients who were intolerant to or in whom imatinib treatment failed were treated with sunitinib. We analyzed the PFS and OS of 93 patients who received sunitinib treatment after imatinib failure. The baseline characteristics of these patients are listed in Supplementary Table 5. Sixty-five patients had nonstomach GISTs, and the other 28 patients had stomach GISTs. There was no difference in the distribution of sex and ECOG PS score between patients with stomach and nonstomach GISTs. However, only one stomach GIST patient but 10 nonstomach GIST patients had c-KIT exon 9 mutations. The median PFS of all patients treated with sunitinib was 12.76 (95% CI, 11.01–14.52) months, as shown in Fig. 3A. There was no difference in PFS in patients by sex, ECOG PS score, and primary site (Supplementary Fig. 5). However, the PFS was different among patients with different genetic alterations. Patients with c-KIT exon 9 mutations had a longer PFS than those with c-KIT exon 11 mutation, PDGFRA mutation and wild-type c-KIT/PDGFRA (P = 0.003), as shown in Fig. 3B. The median PFS of the patient with c-KIT exon 9 mutation was 25.26 months, whereas the median PFS values for patients with c-KIT exon 11 mutation, PDGFRA mutation, and wild-type c-KIT/PDGFRA were 11.74, 2.17, and 4.01 months, respectively. The median OS of the 93 patients was not reached, with 36-month and 60-month survival rates of 53.8% and 45.2%, respectively (Fig. 3C). The OS was not significantly different in patients by sex, ECOG PS score, and primary site (Supplementary Fig. 5). The OS was different in patients by genetic alterations, with a longer OS in patients with c-KIT exon 9 mutation than in those with PDGFRA mutation and c-KIT exon 11 mutation (P < 0.001). However, the OS was not significantly different between patients with c-KIT exon 9 and exon 11 mutations (P = 0.135) (Fig. 3D).

Fig. 3
figure 3

PFS and OS of advanced or recurrent GIST patients who received sunitinib treatment. A Median PFS of all patients. B PFS of GIST patients by genetic alterations. C Median OS of all patients. D OS of GIST patients by genetic alterations

Full size image

PFS and OS of advanced/metastatic or recurrent GIST patients who received regorafenib treatment

The patients who were intolerant to or in whom imatinib and sunitinib failed received regorafenib treatment. We analyzed the PFS and OS of the 42 patients who received regorafenib treatment. The median PFS of patients treated with regorafenib was 7.14 (95% CI, 3.39–10.89) months (Fig. 4A). There was no difference in PFS by primary site, sex or ECOG PS score (Supplementary Fig. 6). The median OS was not reached, with 12-month and 24-month survival rates of 68.1% and 51.8%, respectively (Fig. 4B). There was no significant difference in OS by primary site, sex, or ECOG PS score (Supplementary Fig. 6). Because only one case had a PDGFRA mutation and one case had wild-type c-KIT/PDGFRA, these two cases were not included in the analysis for the effect of genetic alterations on PFS and OS. The difference in PFS and OS by c-KIT exon 9 and c-KIT exon 11 mutations was not significant in patients who received regorafenib treatment (Supplementary Fig. 7).

Fig. 4
figure 4

PFS and OS of advanced or recurrent GIST patients who received regorafenib treatment. A PFS of all patients. B OS of all patients

Full size image

Discussion

This study shows a longer PFS and OS in advanced and recurrent GIST patients who received TKI therapies diagnosed between 2010 and 2020 in Taiwan. Genetic aberrations are prognostic factors for PFS and OS in patients who received imatinib and sunitinib treatment.

Imatinib was approved by the Food and Drug Administration (FDA) for the treatment of advanced GISTs in 2001. The median PFS and OS of advanced GIST patients who received imatinib treatment were 18–26 and 51–57 months, respectively, according to registered clinical trials [4,5,6]. The survival of our patients was longer, with 48-month PFS and OS rates of patients treated with imatinib of 50.5% and 79.5%, respectively. The data from the Netherlands Cancer Registry show that the 5-year OS rate was 48.2% for patients with primary metastatic GIST diagnosed between 2001 and 2012 [23]. Data from the GOLD ReGISTry, a global prospective, observational registry study between 2007 and 2011, showed that the estimated 30-month PFS and OS rates were 59.8% and 82.7%, respectively, for 1095 advanced GIST patients [24]. Data from the Dutch GIST Registry show that the median PFS and OS of 420 advanced GIST patients diagnosed between 2009 and 2021 who were treated with imatinib were 33.0 and 68.0 months, respectively [25]. These real-world data all showed longer survival for advanced GIST patients. Less extensive disease and earlier treatment with imatinib were possible reasons for the longer survival in these data since the patients were diagnosed after the approval of imatinib. In addition, most physicians are familiar with the management of adverse events of imatinib, which affects patient compliance with the drug. On the other hand, sunitinib and regorafenib were approved by the FDA for the treatment of GISTs in 2006 and 2013, respectively. The use of imatinib, sunitinib and regorafenib has been reimbursed in Taiwan since 2002, 2010, and 2016, respectively. Sunitinib and regorafenib treatment were available as further treatment for the patients in whom imatinib and sunitinib failed after the approval date, which explains the longer OS of the patients than the previous registration trials.

In the current study, we showed the incidence of c-KIT and PDGFRA aberrations in advanced or recurrent GIST patients (Table 2). The percentages of genetic aberrations in c-KIT and PDGFRA and wild-type c-KIT/PDGFRA were 88.7%, 1.7% and 9.6%, respectively. The percentage of PDGFRA aberration reported in GISTs, including localized and advanced stage, was 10–16.3% [26,27,28]. Our data showed a much lower percentage of PDGFRA in advanced or recurrent GISTs, which is consistent with < 4% of PDGFRA mutations in advanced GISTs from the S0033 trial [9, 29]. Similar to other studies, most c-KIT exon 9 mutations were detected in nonstomach GISTs [26, 27]. In the current study, we found that men had more c-KIT exon 9 mutations than women, and the 3 patients with PDGFRA mutations were all men. Because the sample size was small, more data are needed for further confirmation.

In the current study, we analyzed the potential prognostic factors for PFS and OS in advanced or recurrent GIST patients treated with TKI therapy. For the patients treated with imatinib, genetic aberrations and primary site were prognostic factors for PFS in univariate Cox regression analysis. However, only c-KIT exon 9 and PDGFRA mutations were associated with a poor PFS compared with c-KIT exon 11 mutations in multivariate analysis. Genetic aberrations, age, baseline albumin level and baseline NLR were prognostic factors for OS in univariate analysis. However, only PDGFRA mutation was associated with a poor OS in multivariate analysis. Gold et al. reported that the mutational status of c-KIT and PDGFRA was not associated with the outcome of metastatic GIST patients before the use of TKIs [30]. c-KIT exon 11 mutation has been reported to be associated with a better OS than c-KIT exon 9 or wild-type advanced GIST treated with imatinib [8, 20]. In our current analysis, sunitinib was available for all patients in whom imatinib treatment failed. We also observed that patients with c-KIT exon 9 mutations had a longer PFS and OS under sunitinib treatment. Therefore, it is reasonable that the OS of patients with c-KIT exon 9 and exon 11 mutations was not significantly different in our current study. However, patients with PDGFRA mutations had the worst survival since this mutation is generally not responsive to imatinib, sunitinib or regorafenib [21]. In the current study, two patients had PDGFRA D842V mutation and the PFS of these two patients to front-line imatinib were 2.9 and 0.9 month, respectively. The first patient then received sunitinib and regorafenib with the PFS of 2.2 and 1.2 months, respectively. The second patient developed enlargement of huge tumor and massive ascites after approximately 1 month’s imatinib and then received surgical removal of tumor with suspected seeding tumor in liver. This patient continued imatinib treatment after surgery with stable disease for 73 months and then received sunitinib treatment with a PFS of 2.7 months. This patient then received avapretinib treatment and is still kept stable disease. Consistent with previous studies, our data showed shorter PFS for patients with PDGFRA D842V mutation treated with imatinib, sunitinib, and regorafenib. For the patients with c-KIT exon 11 mutations, Incorvaia et al. have reported that 60 metastatic GIST patients with their tumor harboring deletion or insertion/deletion in codons 557 and/or 558 (D-557/8) had shorter PFS to first-line imatinib than the patients with their tumors harboring mutations other than D-557/8 [31]. In our current study, because the data of deletion site were not available in some patients, the analysis for D-557/8 was not performed. We showed that the PFS of the patients with c-KIT exon 11 deletion was longer than those with c-KIT exon 11 deletion + missense mutation or missense mutation when they received first-line imatinib treatment. Our result is not consistent with Incorvaia et al.’s result. Although several study groups have reported the poor prognostic role of D-557/8 or deletion in c-KIT exon 11 on recurrence free survival in resected GIST patients, the impact of mutational type of c-KIT exon 11 on survival of advanced GIST patients needs more data for further confirmation [32,33,34]. The genetic study for c-KIT and PDGFRA is not routinely performed for GISTs because it is not reimbursed in Taiwan and the application of sunitinib or regorafenib is feasible without genetic data. Our result suggests that genetic test is strongly indicated for the patients experiencing resistance to imatinib, particularly early resistance due to PDGFRA exon 18 mutations which may benefit from novel TKI, avapretinib, therapy [35]. Regarding the other risk factors associated with the survival of advanced GIST patients treated with TKIs, such as ECOG PS score, age, sex, baseline neutrophil count, and baseline albumin level [5, 13, 20, 36], we could not identify their prognostic role after multivariate analysis.

There are some limitations of this study. The lack of patient data, particularly ECOG PS score and baseline albumin level, is a limitation of this study. Another limitation of this study is the patient selection bias that we enrolled the patients treated in the 11 medical centers but not in regional hospitals. The resources, availability of medications, and practical principles in regional hospitals may differ from that in medical centers and affect the survival of the cancer patients [37]. To overcome these problems, a prospective registry study with predefined baseline characteristics and biomarkers been evaluated and checked in GIST patients from medical centers and regional hospitals may provide more comprehensive information. However, previous studies were analyzed earlier, and imatinib was the major treatment for these patients [5, 13, 20, 36]. Our patient population had more treatment options, namely, sunitinib and regorafenib, after imatinib failure. Recently, novel TKIs, such as ripretinib and avapritinib, have been evaluated and approved for refractory advanced GIST or PDGFRA exon 18-mutated GIST patients by FDA in May 2020 and January 2020, respectively, based on the results of randomized phase III trials [35, 38]. The efficacy of other novel agents, such as the heat shock protein 90 inhibitor TAS-116, has also been evaluated in clinical trials [39, 40]. Therefore, we expect that the survival of advanced GIST patients will be longer and that the effect of mutational status will probably become less significant after the availability of effective novel agents for GIST treatment in the near future.

Conclusions

Our current study demonstrates real-world evidence of a longer survival of advanced or recurrent GIST patients in the era of TKIs and identifies mutational status as a prognostic factor for survival of these patients. Other novel agents are under investigation and are expected to prolong the survival of advanced GIST patients in the near future.