Long-term follow-up results of the multicenter phase II trial of regorafenib in patients with metastatic and/or unresectable GI stromal tumor after failure of standard tyrosine kinase inhibitor therapy

Our study provides the longest follow-up of advanced GIST patients treated with regorafenib in this phase II trial. Particular benefit among exon 11 KIT mutations and SDH-deficient GIST were observed. Liberal use of dose reductions and treatment breaks were required to optimize long-term drug exposure.

Abstract

Background

This investigator-initiated trial provided the justification for the phase III GRID study resulting in worldwide regulatory approval of regorafenib as a third-line therapy for patients with metastatic gastrointestinal stromal tumors (GIST). We report the genotype analyses, long-term safety, and activity results from this initial trial of regorafenib in GIST.

Patients and methods

The trial was conducted between February 2010 and January 2014, among adult patients with metastatic GIST, after failure of at least imatinib and sunitinib. Patients received regorafenib orally, 160 mg once daily, days 1–21 of a 28-day cycle. Clinical benefit rate (CBR), defined as complete or partial response (PR), or stable disease lasting ≥16 weeks per RECIST 1.1, progression-free survival (PFS), overall survival (OS), long-term safety data, and metabolic response by functional imaging were assessed.

Results

Thirty-three patients received at least one dose of regorafenib. The median follow-up was 41 months. CBR was documented in 25 of 33 patients [76%; 95% confidence interval (CI) 58% to 89%], including six PRs. The median PFS was 13.2 months (95% CI 9.2–18.3 months) including four patients who remained progression-free at study closure, each achieving clinical benefit for more than 3 years (range 36.8–43.5 months). The median OS was 25 months (95% CI 13.2–39.1 months). Patients whose tumors harbored a KIT exon 11 mutation demonstrated the longest median PFS (13.4 months), whereas patients with KIT/PDGFRA wild-type, non-SDH-deficient tumors experienced a median 1.6 months PFS (P < 0.0001). Long-term safety profile is consistent with previous reports; hand–foot skin reaction and hypertension were the most common reasons for dose reduction. Notably, regorafenib induced objective responses and durable benefit in SDH-deficient GIST.

Conclusions

Long-term follow-up of patients with metastatic GIST treated with regorafenib suggests particular benefit among patients with primary KIT exon 11 mutations and those with SDH-deficient GIST. Dose modifications are frequently required to manage treatment-related toxicities.

Clinical trial number

NCT01068769.

introduction

The seminal discovery of activating kinase mutations (in KIT and later PDGFRA receptor tyrosine kinase) identified the pivotal role oncoproteins play in aberrant signal transduction in gastrointestinal stromal tumors (GIST) [1, 2]. Given the remarkable impact of imatinib on improvements in progression-free survival (PFS) and overall survival (OS) of patients with advanced GIST, continued efforts have been made over the past decade to further improve survival outcomes [3–6]. In 2006, sunitinib malate was approved by the Food and Drug Administration (FDA) as a second-line therapy for metastatic GIST, after clinical trials demonstrated significant superiority over placebo in terms of PFS among patients with disease progression or intolerance to imatinib [7].

Preclinical academic studies at Dana Farber Cancer Institute identified regorafenib, an orally available multi-kinase inhibitor targeting KIT, PDGFRA as well as angiogenic pathways such as VEGF receptors 1–3, TIE2, and fibroblast growth factor receptors, as a promising agent in human GIST model systems [8]. On the basis of this, an investigator-initiated academic clinical trial was designed and conducted. In 2012, our group published the early results of this phase II trial assessing the efficacy and safety of regorafenib in GIST patients, after failure of standard tyrosine kinase inhibitors, including at least imatinib and sunitinib. This phase II trial served as the basis for design of the phase III international GRID trial of regorafenib versus placebo, which ultimately led to worldwide regulatory approval of regorafenib for patients with advanced GIST after imatinib and sunitinib failure [9].

We now report the final analysis of the initial phase II trial of regorafenib in patients with advanced GIST, thus providing the longest follow-up on regorafenib safety and efficacy in patients with GIST. The analysis provides an update on PFS and OS, as well as correlations with ¹⁸fluoro-deoxy-glucose-positron emission tomography (FDG-PET) response and mutation categorization not included in the original report.

methods

patients and study design

This investigator-initiated multicenter open-label phase II trial with regorafenib was conducted between February 2010 and January 2014, in adult patients with histologically confirmed metastatic and/or unresectable GIST, with prior failure of imatinib due to disease progression or clinically significant adverse tolerability and failure of sunitinib due solely to disease progression. As previously described, patients received regorafenib orally at this initial dose of 160 mg once daily, on days 1–21 of a 28-day cycle [8].

efficacy and safety assessments

Primary efficacy objectives for this trial were clinical benefit rate (CBR), defined as the proportion of patients with a confirmed complete response (CR) or partial response (PR), or stable disease (SD) lasting ≥16 weeks per RECIST 1.1. Secondary efficacy objectives included PFS (defined as time from date of registration to date of objective disease progression or death resulting from any cause; patients without disease progression were censored at the date of last disease assessment); OS (defined as time from date of registration to date of death or date last known alive) [10]; comparison of changes in FDG-PET imaging per EORTC criteria [11] from baseline to end of first week of cycle 1 regorafenib dosing, with comparisons to computed tomography (CT) tumor evaluation response, PFS, and OS; and relationships of primary GIST genotype with FDG-PET response, CT tumor response, CBR, and PFS. Toxicity was assessed per Common Terminology Criteria for Adverse Events 4.0.

statistical analysis

All enrolled patients who received at least one dose of study drug were eligible for efficacy and toxicity analyses. Ninety-five percent two-sided exact binomial confidence intervals were computed for CBR. PFS and OS were estimated using the Kaplan–Meier method. Reverse censoring using the Kaplan–Meier method was used to calculate the median follow-up. Pairwise comparisons of the differences in CBR, PFS, and best overall response by genotype and FDG-PET response were assessed by Fisher's exact and log-rank tests, as appropriate. A two-sided P < 0.05 defined statistical significance. STATA v 13 (StataCorp, College Station, TX, USA) was used for all analyses.

results

patients

Thirty-four patients were enrolled at three centers in the United States. One participant was judged ineligible for study drug dosing after enrollment; therefore, 33 patients were eligible and received at least one dose of regorafenib. As of the close of study, 14 January 2014, the median follow-up of living patients was 41 months (range 3.2–44.9 months). Twenty-nine patients discontinued protocol treatment before study closure due to progressive disease (60%), physician decision (12%), death while on study drug (3%), intratumoral bleeding and ileus possibly drug-related (3%), intercurrent illness (3%), alternative treatment (3%), and patient choice (3%). The primary GIST genotype was determined in tumors from 30 of 33 patients (Table 1).

Table 1.

Clinical benefit rate and progression-free survival by genotype

Genotype	Total	Patients with clinical benefit (n)	CBR %	95% CI CBR	Median PFS (months)
Exon 11	19 (59%)	15	79	54% to 94%	13.4
Exon 9	3 (9%)	2	67	9% to 99%	5.7
SDH deficient	6 (18%)	6	100	54% to 100%	10
BRAF exon 15 mutation	1 (3%)	0	0	—	NAa
SDH unknown, BRAF intact	1 (3%)	0	0	—	NAb
Unknown genotype	3 (9%)	2	67	9% to 99%	NAc
Total	33	25	76	58% to 89%	13.2

^aMedian PFS not calculated for single patients. Patient with BRAF exon 15 mutation genotype had a progression free period of 1.6 months.

^bMedian PFS not calculated for single patients. Patient with BRAF intact, SDH unknown genotype had a progression-free period of 3.6 months.

^cMedian PFS calculated for known genotypes only. Unknown genotypes included in total median progression-free survival analysis.

efficacy

clinical benefit rate

Clinical benefit was confirmed in 25 of 33 patients [76%; 95% confidence interval (CI) 58% to 89%], including 6 PRs and 19 instances of SD lasting at least 16 weeks. Five patients experienced SD for <16 weeks, two patients demonstrated disease progression at first tumor evaluation, and one patient was not evaluable for response due to study withdrawal before the first disease assessment. Among the six patients who experienced PR, two patients had tumors that were SDH-deficient—a subtype of disease without activating KIT/PDGR mutations, two harbored an exon 11 mutation, one an exon 9 mutation, and one was of unknown mutation status. The CBR varied among genotypes (Table 1).

progression-free survival

The overall median PFS was 13.2 months (95% CI 9.2–18.3 months; Figure 1). Twenty patients had documented progressive disease following RECIST criteria, four patients died without documented disease progression, and two patients had unconfirmed disease progression and died during follow-up.

Figure 1.

(A) Progression-free survival and (B) overall survival.

Patients with a KIT exon 11 mutant tumor demonstrated the longest median PFS (13.4 months, n = 15), followed by SDH-deficient GIST (10 months, n = 6), exon 9 mutant tumors (5.7 months, n = 2) and wild type, non-SDH-deficient tumors (1.6 months, n = 2) (P < 0.0001). Of note, four patients (12%) remained evaluable for response and were without disease progression at the time of study closure, each achieved this clinical benefit for more than 3 years (range 36.8–43.5 months). The mutation categories among these patients without disease progression and still on regorafenib treatment at the time of study closure were two SDH-deficient mutant tumors (38.8 and 40.9 months), one exon 11 mutant tumor (43.5 months), and one exon 9 mutant tumor (36.8 months). All four patients required dose modifications during the study; dose modification occurred early in the course of the study (first two cycles) for three of the four patients, but were required as late as cycles 13 and 17 as well. Detailed background medical history of these patients is provided in supplementary material, available at Annals of Oncology online.

overall survival

A total of nine patients (27%) were alive at the end of study; the median OS was 25 months (95% CI 13.2–39.1 months; Figure 1) with a 1 year survival rate of 79% (95% CI 61% to 89%). The median OS was not statistically different across genotypes (P = 0.77).

correlative FDG-PET studies

FDG-PET and clinical benefit

Thirty-one of 33 patients underwent FDG-PET imaging at baseline and at days 18–21 of cycle 1. Complete metabolic response was observed in five patients (four exon 11 mutant tumors and one SDH-deficient mutant tumor), whereas partial and stable metabolic responses were observed in 16 and 6 patients, respectively. There was no statistically significant difference in predefined clinical benefit (RECIST CR, PR, or SD lasting at least 16 weeks) achieved when comparing those who achieved FDG-PET response in cycle 1 and those who did not, although there was a trend in favor of those achieving metabolic response. Approximately 85% of patients with a metabolic response in cycle 1 achieved clinical benefit, whereas 60% of patients who did not achieve a metabolic response achieved clinical benefit (P = 0.18).

There was a trend toward longer RECIST PFS among patients who achieved metabolic response at days 18–21 (Figure 2). The median RECIST PFS for metabolic responders was 15.8 months (95% CI 10.0–21.4) compared with 7.3 months (95% CI 1.6–14.9) in metabolic non-responders, although this did not reach statistical significance (P = 0.4).

Figure 2.

Progression-free survival by ¹⁸fluoro-deoxy-glucose-positron emission tomography response cycle 1, on-drug.

FDG-PET and mutation category

Approximately 83% of patients with an exon 11 mutation experienced a metabolic response at cycle 1. In addition, one patient with SDH-deficient GIST and FDG avid disease at baseline experienced a complete metabolic response at cycle 1.

safety

As of close of study, 593 cycles of regorafenib were administered with a median of 15 cycles administered per patient (range 1–45 cycles). All 33 patients experienced a grade 2 or higher treatment-related toxicity, and 31 patients experienced at least one grade 3 or 4 toxicity from the study drug. Toxicities at least possibly related to the study drug, occurring in at least 25% of patients, are shown in Figure 3. The most common toxicities were hand–foot skin reaction, fatigue, diarrhea, and hypertension occurring in 91%, 85%, 79%, and 76% of patients, respectively. Grade 3 toxicities occurring in >5% of patients and all grade 4 toxicities are provided in supplementary Table S1, available at Annals of Oncology online. The highest grade toxicities occurred early in the course of treatment and the median time to initial highest grade toxicity was during cycle 1. Twenty-two patients (67%) experienced at least one more grade 3 or 4 toxicity while on study drug.

Figure 3.

Toxicities of any grade, at least possibly study drug-related, occurring in at least 25% of patients.

Schedule or dose modifications for toxicities were common. Sixty-four percent of patients had a treatment delay within a cycle of a median of 6 days at some point in their treatment course, and cycle delay during more than one cycle was required for 45% of patients (median: 3, range 1–9). Approximately 88% of patients had a dose withheld during at least one cycle, of which 56% had a dose withheld during more than one cycle (median: 2 cycles, range 1–6 cycles).

Dose reduction was required in 27 patients (82%) due to protocol-defined toxicity, of which 66% of patients (18) were re-escalated while on study drug. In total, 13 patients (40%) tolerated a final dose of regorafenib of 160 mg, 9 patients (27%) tolerated a final dose of 120 mg, 7 patients (21%) tolerated a final dose of 80 mg, and 4 patients (12%) tolerated a final dose of 40 mg. The most common reasons for dose reduction were hand–foot skin reaction and grade 3 or higher non-hematological toxicities.

discussion

Regorafenib remains the only drug approved for metastatic GIST since 2006. The phase III trial results demonstrated significant improvement in PFS among GIST patients treated with regorafenib compared with placebo.

Our initial report of this phase II trial of regorafenib among metastatic GIST patients was at 11 month median follow-up. In this updated analysis, the median follow-up is 41 months, which enables us to provide comprehensive safety follow-up and include subgroup analyses based on these long-term results.

Efficacy analysis demonstrate longer median PFS (13.2 months) than we previously estimated (10.0 months), and notably longer PFS than previously reported in phase II trials of TKIs in patients with advanced GIST after imatinib and sunitinib: sorafenib (4.9 months), pazopanib (1.9 months), or the phase III trial with regorafenib (4.8 months) [5, 6, 9]. A retrospective analysis of sorafenib as third- or fourth-line treatment for GIST has also shown a PFS of 6.4 months [4]. Differences in PFS between our trial and previously reported trials with sorafenib or pazopanib may be a result of variation in drug efficacy or reflect the nature of non-randomized phase II trials and heterogeneity in patient selection and outcome. However, there are increasing data that regorafenib has effectiveness against KIT exon 17 mutations which are known to be more prominent following exposure to multiple TKIs [12]. The KIT inhibitory profile of pazopanib against known TKI resistance mutations is not well defined against. The shorter PFS observed in the phase III randomized control trial with regorafenib in GIST is probably due to inherent differences in patient selection with a limited number of centers in the phase II trial compared with the broader, multinational phase III setting. Nevertheless, this long-term follow-up enabled us to provide the first actuarial median OS estimate of 25 months for regorafenib as third-line or later treatment for GIST patients. As this treatment was given after failure or intolerance to imatinib and sunitinib, and there is no approved therapy for GIST beyond this line of treatment, the median OS of 2 years potentially reflects a true benefit of regorafenib.

In contrast to treatment with imatinib as a first-line therapy, where rapid and complete metabolic response predicts excellent disease control, in this more heavily pretreated population, the correlation between metabolic response of FDG-PET at days 18–21 of the first cycle and RECIST response appears less strong and was not predictive of outcome in a statistically significant manner [13]. This is likely due to heterogeneity of secondary resistance mutations which emerge after extensive tyrosine kinase inhibitors exposure and are known to have variable sensitivities to available TKIs [12].

Acquired secondary KIT mutations (exons 13, 14, 17, and 18) are commonly identified in pretreated GIST patients [14]. In our cohort, pre-enrollment tumor for secondary mutations was available on nine study participants—one exon 13 secondary mutation, seven exon 17, and one exon 18 secondary mutations. Interestingly, the median PFS for patients with tumor that harbored a known secondary mutation in exon 17 was 22 months (95% CI 6–NR), suggesting effective inhibition of this common KIT activation loop resistance mutation by regorafenib.

Succinate dehydrogenase is a member of a complex chain of enzymatic reactions responsible for the oxidation of succinate to fumarate. This complex chain is an important regulator of hypoxia-inducible factor-1α (HIF-1α), a subunit of HIF1, which is regulated by the Von Hippel Lindau protein and serves as a transcriptional activator of IGF2 and VEGF [15]. In SDH-deficient GIST, succinate dehydrogenase inactivation through mutation or methylation leads to accumulation of HIF-1α and formation of a transcription factor inducing expression of genes involved in angiogenesis (including VEGF) and glycolysis. Indeed, consistent with these findings, SDH-deficient mutant tumors, including SDH-deficient GIST tumors, have high expression of VEGF [16].

In our study, all six patients with SDH-deficient GIST (KIT/PDGFRA wild-type GIST) experienced clinical benefit from regorafenib with tumor response or stabilization for 16 weeks or more. Two SDH-deficient patients also experienced a PR per RECIST 1.1, one of whom also achieved a complete metabolic response by FDG-PET. Objective tumor response in this specific GIST subtype is quite uncommon, although clinical benefit, as well as tumor response, has been observed in this population with other potent VEGFR inhibitors, such as sorafenib (one PR lasting more than 16 months) and pazopanib (one patient with 20% tumor reduction and ongoing response after 17 cycles), suggesting that perhaps, in addition to the rather indolent nature of this subtype, potent VEGFR inhibition may be important mechanism of disease control in SDH-deficient GIST [6, 17].

The safety profile in our long-term follow-up is consistent with previous large-scale clinical trial reports [9]. Hand–foot skin reaction, fatigue, diarrhea, and hypertension were the most frequently observed drug-related adverse events. The rates of hypertension and hand–foot skin reaction (39% and 36%, respectively) observed in our study are considerably higher than previous reports most probably reflecting the relative high median of cycles administered per patient (15 cycles) and the long-term drug exposure and toxicity. These adverse events, especially hand–foot skin reaction and hypertension, were the most common reasons for dose reduction, and were manageable among most patients by dose modifications, schedule interruptions, and proper supportive care. With this approach, ultimately, two-thirds of the patients in this trial tolerated a final dose of 160 or 120 mg of regorafenib after dose reduction and re-escalation. Our experience suggests that close follow-up and liberal use of dose reductions and treatment breaks, particularly in the first cycle of therapy, may be required to optimize long-term exposure of regorafenib in patients with advanced GIST.

Our cohort of advanced GIST patients previously experienced treatments failure with at least imatinib and sunitinib; however, regorafenib provided long-term benefit to a subset of patients, particularly patients with SDH-deficient and primary exon 11 mutant tumors. Although dose modifications due to toxicities were very common, drug discontinuation was rare, and a small subset of patients remained on regorafenib treatment for more than 3 years. Increased activity of regorafenib in specific refractory GIST molecular subtypes warrants further investigation.

funding

This work was supported in part by Dana Farber Cancer Institute via an unrestricted grant for this academic investigator-initiated trial from Bayer Healthcare Pharmaceuticals and by the NCI cancer center support grant (biostatistics CORE) P30 CA006516-50. Supported in part VA Merit Review Grants to MCH (1I01BX000338-01, 2I01BX000338-05).

disclosure

MCH: Consultant: Novartis, Ariad, Blueprint Medicines, Pfizer, MolecularMD. Speakers honorarium: Novartis, Pfizer. Research funding: Deciphera, Ariad, Inhibikase, Blueprint Medicines. Equity interest: MolecularMD. Expert testimony: Novartis. GDD consultant: Bayer, Novartis, Pfizer, Lilly, EMD-Serono, Sanofi Oncology, Janssen Oncology, PharmaMar, Daiichi-Sankyo, Ariad, WIRB Copernicus Group, Ziopharm, Polaris Pharmaceuticals, Kymab, Genocea, Adaptimmune, Blueprint Medicines, Kolltan Pharmaceuticals, G1 Therapeutics, Caris Life Science, Champions Oncology, Bessor Pharmaceuticals. Research funding: Bayer, Novartis, Pfizer, Janssen Oncology. Equity interest: Blueprint Medicines, Kolltan Pharmaceuticals, G1 therapeutics, Caris Life Sceince, Champions Oncology, Bessor Pharmaceuticals. SG research funding: Pfizer, Bayer, ARIAD, Blueprint Medicines, Deciphera. All remaining authors have declared no conflicts of interest.