Key points

Introduction

The first 15 years of active therapy of gastrointestinal stromal tumor (GIST) set the stage for the development of treatment of solid tumors, thanks to small molecule oral kinase inhibitors such as imatinib. The discovery of KIT mutations in GIST led to the understanding of KIT as a driver of GIST growth and has led to the understanding of the clonal evolution of a solid tumor over time. Imatinib as a KIT inhibitor in GIST in vitro and in people with metastatic disease sparked a wave of activity leading to approval of 3 agents in many countries for metastatic GIST, specifically imatinib, sunitinib, and regorafenib. In parallel with the developments in metastatic disease, a series of clinical trials has been conducted that led to the present standard of care after resection of higher-risk GIST, specifically 3 years of imatinib.

Despite an apparently complete and coherent set of data outlining therapy for primary and metastatic disease, patients still progress despite all lines of therapy, indicating the need for further research on adapting treatment as metastatic disease evolves. In addition, several issues remain regarding the use of 3 years of imatinib in the adjuvant setting. It is this latter series of questions that is addressed in this review.

Adjuvant imatinib therapy has been observed applied blindly to patients with larger tumors with higher mitotic rates. However, the genetics of GISTs, which has impact on the treatment of metastatic disease, also contributes to the decision making of who should receive adjuvant therapy. These data are still evolving and remain one of the most interesting issues of the “first-order” problems in GIST management. The treatment of large primary tumors in which neoadjuvant imatinib is used preoperatively is also an important area of clinical interest, in which there is some practical experience. Finally, how many scans are necessary to help most efficiently identify recurrence in patients who have either not received or have received adjuvant therapy? The kinetics of recurrence has important implications for patient care as well.

It is hoped that in this review that some of these evolving “knowns” and “known unknowns” in the adjuvant therapy of GIST are addressed, so that clinicians can contribute to the next series of studies that will better define the role of adjuvant therapy for GIST.

Introduction

The first 15 years of active therapy of gastrointestinal stromal tumor (GIST) set the stage for the development of treatment of solid tumors, thanks to small molecule oral kinase inhibitors such as imatinib. The discovery of KIT mutations in GIST led to the understanding of KIT as a driver of GIST growth and has led to the understanding of the clonal evolution of a solid tumor over time. Imatinib as a KIT inhibitor in GIST in vitro and in people with metastatic disease sparked a wave of activity leading to approval of 3 agents in many countries for metastatic GIST, specifically imatinib, sunitinib, and regorafenib. In parallel with the developments in metastatic disease, a series of clinical trials has been conducted that led to the present standard of care after resection of higher-risk GIST, specifically 3 years of imatinib.

Despite an apparently complete and coherent set of data outlining therapy for primary and metastatic disease, patients still progress despite all lines of therapy, indicating the need for further research on adapting treatment as metastatic disease evolves. In addition, several issues remain regarding the use of 3 years of imatinib in the adjuvant setting. It is this latter series of questions that is addressed in this review.

Adjuvant imatinib therapy has been observed applied blindly to patients with larger tumors with higher mitotic rates. However, the genetics of GISTs, which has impact on the treatment of metastatic disease, also contributes to the decision making of who should receive adjuvant therapy. These data are still evolving and remain one of the most interesting issues of the “first-order” problems in GIST management. The treatment of large primary tumors in which neoadjuvant imatinib is used preoperatively is also an important area of clinical interest, in which there is some practical experience. Finally, how many scans are necessary to help most efficiently identify recurrence in patients who have either not received or have received adjuvant therapy? The kinetics of recurrence has important implications for patient care as well.

It is hoped that in this review that some of these evolving “knowns” and “known unknowns” in the adjuvant therapy of GIST are addressed, so that clinicians can contribute to the next series of studies that will better define the role of adjuvant therapy for GIST.

Case series from the era before imatinib therapy

It was clear from older series of GIST patients that there was a significant recurrence and death rate from recurrent disease. In the oldest large series from MD Anderson, Ng and colleagues showed that what was then called gastrointestinal leiomyosarcoma had a very high recurrence and mortality risk. A series of 200 GIST patients from Memorial Sloan Kettering Cancer Center showed a high recurrence and disease-specific death rate. Notably, both of these series involved patients with larger tumors than is seen in more contemporary series of primary disease. GIST proved unyielding to standard cytotoxic chemotherapy, for reasons that remain unknown. As a result, the finding of patients with metastatic disease responding to imatinib quickly led to the implementation of adjuvant trials to attempt to improve the cure rate.

Joensuu and colleagues have compiled the largest retrospective analysis of GIST involving patients from the era before imatinib, which in addition to data from Z9000 and Z9001 discussed later, form the basis of discussion of the risk of recurrence of GIST on the specific mutation driving the tumor. The Z9000 and Z9001 genetic data also allow at least some insight into the impact of imatinib on these relatively small subsets of patients.

From a population of more than 3000 GIST patients, mutation analysis was conducted in 1505 of the patients. Relapse-free survival (RFS) was the principal endpoint for the analysis. A total of 301 unique KIT mutations and 33 PDGFRA mutations were observed. The most common mutations overall were deletion of WK557-558 ( KIT exon 11), substitution of D842V ( PDGFRA exon 18), and duplication of AY502-503 ( KIT exon 9). These mutations were associated with a similar RFS to other GISTs.

Although mitotic rate still was more important than genomic status of the GIST in terms of prognosis, specific genetic subtypes were associated with better outcomes than others. In particular, patients with PDGFRA mutations had superior RFS compared with GISTs with KIT mutations (hazard ratio [HR], 0.34; P = .004).

Looking specifically at KIT -mutated GIST, patients with KIT exon 9 mutated GIST showed a numerically inferior PFS compared with KIT exon 11 mutated GIST ( P = .07). Approximately 80% of KIT exon 9 mutations were seen in small bowel GIST. In terms of rarer KIT mutant GISTs, both primary exon 13 and 17 primary GISTs were nongastric and did not portend a different prognosis than other KIT mutant GIST.

Notably, some subtypes of KIT exon 11 mutations were substantially lower-risk tumors than the most common mutation subtypes. For example, only 1 in 35 GISTs with KIT exon 11 duplication mutations recurred. Patients with deletions of only one codon of KIT exon 11 had better RFS than those with another deletion type, and specific KIT exon 11 substitution mutations (W557R, V559A, and L576P) were also associated with better RFS. Patients with no KIT or PDGFRA mutation also had a lower risk of recurrence compared with all patients (HR 0.52; P <.001). Eighty-eight percent of PDGFRA mutations were found in gastric GISTs.

These data form the foundation of the expected genomic subtypes by location, which recur at different rates after surgery, impacting who recurs after therapy, be it surgery, imatinib, or both.

Selected trials of adjuvant imatinib

The key trial of imatinib in the adjuvant setting that has led to the 2016 standard of care of 3-year imatinib for higher-risk GIST is discussed after setting the stage discussing other important trials.

Z9000 (1-year imatinib)

Two trials led by DeMatteo with the American College of Surgeons Oncology Group, trials Z9000 (phase II) and Z9001 (phase III), led to the approval of imatinib in the adjuvant setting.

The Z9000 study of imatinib 400 mg oral daily for 1 year after resection of primary high-risk GIST was heartening, whereby high risk was defined as tumor diameter of 10 cm or more, intraperitoneal tumor rupture, or up to 4 peritoneal implants. A total of 106 people were treated. With a median follow-up of 7.7 years, RFS rates were 96%, 60%, and 40% at 1, 3, and 5 years, respectively. Overall survival (OS) rates were 99%, 97%, and 83%, much better than historical 5-year OS rates of 35%.

The KIT and PDGFRA mutation data from Z9000 confirmed the higher-risk nature of KIT exon 11 deletion mutations compared with other mutation classes. KIT exon 11 mutant GIST patients had clear benefit from imatinib overall. Patients with KIT exon 9 mutant GIST had approximately the same relapse risk with or without imatinib, calling into question the use of imatinib in this group of GIST. Conversely, in the European Organisation for Research and Treatment of Cancer metastatic disease study discussed later, higher doses of imatinib appear to be more effective in the metastatic setting, suggesting higher-dose therapy for KIT exon 9 mutant GIST if it is to be used. Patients without KIT or PDGFRA mutations appear to have no difference in RFS regardless of whether they received imatinib or not, implying that such patients are not good candidates for adjuvant imatinib. Finally, patients with PDGFRA mutation other than D842V appeared to benefit; the very low risk of recurrence of and the resistance of such GIST to imatinib are 2 reasons to not use adjuvant imatinib in PDGFRA D842V patients (see later discussion).

Z9001 (0 vs 1-year adjuvant imatinib)

The Z9001 was the first trial to compare imatinib and placebo in the adjuvant setting, testing 1 year of treatment. A total of 713 patients with resected primary localized GIST at least 3 cm in greatest dimension were treated. In the initial analysis with median follow-up of ∼20 months, RFS was superior with imatinib versus placebo (98% vs 83%; P <.001). OS was not different (99.2% vs 99.7%; P = .47). The benefit of imatinib could be stratified by tumor size, with RFS in favor of treatment only for larger tumors (for 6–10 cm size tumors, RFS was 98% for imatinib vs 76% for placebo; P = .05); the difference in RFS was greater for primary GISTs >10 cm in size (77% versus 41%; P <.0001).

The most recent follow-up of the Z9001 focused on genetic data and their impact on outcomes in the trial. With a median follow-up of 74 months, RFS remained superior for 1 year of imatinib (HR 0.6; 95% CI 0.43–0.75). Examining the control and imatinib arms separately, in each case by multivariate analysis, the tumor genotype was not significantly associated with RFS in comparison with the known significant factors of larger tumor, small bowel location, and high mitotic rate. OS was not impacted by 1 year of imatinib; patients with relapse did well for extended periods of time on imatinib after relapse.

Comparing the 2 arms of the trial, imatinib was associated with superior RFS in patients with KIT exon 11 deletion, but not KIT exon 11 insertion or point mutation, KIT exon 9 mutations, PDGFRA mutation, or for tumors without PDGFRA or KIT mutation.

European Organisation for Research and Treatment of Cancer intergroup trial (0 vs 2-year adjuvant imatinib)

A large international study examined 2 versus 0 years of imatinib in the adjuvant setting for intermediate- to high-risk resected GIST, using the novel endpoint of failure of first tyrosine kinase inhibitor as a primary endpoint, meaning that patients had to fail imatinib in the adjuvant and recurrent setting to be declared a failure. Eight hundred thirty-five eligible patients were accrued internationally. With 4.7-year median follow-up, 5-year imatinib failure-free survival was not different (97% for imatinib vs 84% for placebo), and OS was essentially identical (100% for imatinib vs 99% for placebo), but RFS was significantly different at 3 years (84% vs 66%) and at 5 years (69% vs 63%) ( P <.001). The data across these studies are consistent with the relapse of patients most commonly 6 to 18 months after completion of adjuvant therapy. These data were supported by a smaller phase II trial of 2 years of imatinib for KIT exon 11 mutant GIST from Korea, with consistent data.

Scandinavian Sarcoma Group XVIII/Arbeitsgemeinschaft Internistische Onkologie (1- vs 3-year adjuvant imatinib)

The study that defines present day adjuvant therapy for GIST is the Scandinavian Sarcoma Group (SSG) XVIII/Arbeitsgemeinschaft Internistische Onkologie (AIO) trial of 3 years of imatinib versus 1 year of imatinib, in which both RFS and OS were superior for 3 years of imatinib versus 1 year. Two hundred evaluable patients were treated on each arm, and all had higher-risk disease as defined by having at least 1 of the following features: (1) longest tumor diameter greater than 10 cm, (2) mitotic rate greater than 10 per 50 high-power fields (HPF) of the microscope, (3) tumor diameter greater than 5.0 cm and mitotic count greater than 5/50 HPF, or (4) tumor rupture before surgery or at surgery. With a median follow-up of 54 months, RFS was superior for 3-year imatinib, 66% versus 48% ( P <.0001); 5-year OS was 92% versus 82% in favor of 3-year imatinib ( P = .019). In a 2016 follow-up analysis with median follow up of 90 months, 5-year RFS was 71% for imatinib versus 52% for placebo, and 5-year OS was 92% versus 85%. Mutation analysis of this group of patients, not published as of the 2016 update, will provide a unique resource as to who should or should not receive adjuvant imatinib on the basis of their mutation status and other known risk factors.