Introduction
Gastrointestinal stromal tumors (GISTs) represent the most common type of gastrointestinal mesenchymal tumors and are detected most often in the stomach, followed by the small intestine and other sites in the colon, esophagus, and peritoneal cavity [
1‐
5]. Surgical resection with a negative margin remains the only therapeutic modality for cure.
Adjuvant therapy has recently been established for patients with a high-risk of recurrence GISTs [
6,
7]. Three randomized phase III trials evaluating the efficacy of adjuvant therapy with imatinib mesylate (IM) demonstrated the efficacy of adjuvant therapy with IM for high-risk GIST recurrence after resection and revealed that recurrence-free survival (RFS) was significantly prolonged in patients treated with IM for a duration of 1–3 years, as compared with that in controls [
7‐
10]. Nonetheless, the eligibility criteria for randomization of patients in each trial were different. The first trial (i.e., the American College of Surgeons Oncology Group Z9001 study) targeted patients with GIST measuring ≥ 3 cm in size and reported that adjuvant IM treatment for 1 year improved the RFS rate, as compared with that in controls. Following this study, patients in the high-risk group according to the National Institutes of Health consensus criteria (NIHC) and patients with tumor rupture were randomized in the Scandinavian Sarcoma Group XVIII/Arbeitsgemeinschaft Internistische Onkologie study [
9]. The European Organisation for Research and Treatment of Cancer 62,024 study was the largest phase III trial that targeted patients with primary GIST who had high and intermediate risk based on the NIHC [
8]. For these reasons, there is no consensus on the indications for adjuvant therapy. Therefore, predicting the patient-specific risk of GIST recurrence plays a crucial role in determining indications for adjuvant therapy.
Several risk-stratification systems for analyzing patients after radical resection of GISTs have been proposed. The NIHC was proposed in 2002 and have been widely used globally since then [
11]. This classification is based on tumor size and the number of mitotic counts per 50 high-power fields (HPFs). However, further study showed that the location of GISTs was also one of the important and independent prognostic factors for recurrence in patients after radical resection. Miettinen et al. proposed the Armed Forces Institute of Pathology Criteria; this classification involved the location of GISTs in addition to the tumor size and number of mitotic counts [
3]. The presence of tumor rupture, which is a strong adverse prognostic factor for GISTs and potentially important in determining patient prognosis [
12‐
14], is absent in these classification systems. Therefore, Joensuu et al. proposed the modified NIHC (m-NIHC) [
15], which considers ruptured GISTs as high-risk tumors irrespective of other features since tumor rupture is a clinically malignant factor for recurrence. Previously, we compared these three risk classifications and reported that m-NIHC exhibited the highest sensitivity for predicting recurrence; hence, we proposed it for the identification of candidates for adjuvant therapy [
14]. Despite tumor size and mitotic count showing a nonlinear association with the risk of recurrence, these were estimated by linear modelling. Therefore, current methods of estimating individuals’ survival outcomes were faulty.
Joensuu et al. proposed contour maps, which enable risk classification of recurrence in individual patients. Contour maps consider tumor size, location, mitotic count, and rupture and consider the tumor size and mitotic count as continuous nonlinear variables [
16]. They reported that contour maps were more detailed than other criteria with respect to the 10-year risk of GIST recurrence and were appropriate for estimating the individualized outcomes. However, although contour maps have been referred to in Japanese guidelines [
17], their validity has not been confirmed in Asian patients. Therefore, the present study aimed to clarify the validity of contour maps in Japanese patients with GIST and explore the new strategy for adjuvant therapy.
Discussion
The Kinki GIST registry comprises the largest Japanese population in retrospective and prospective studies [
14,
18‐
22]. Because this cohort study included consecutive patients in each hospital, this patient population reflects the characteristics of real-world Japanese patients with GIST. The proportion of patients with gastric GISTs was higher than that reported by previous international studies [
12,
16]. In addition, there were fewer high-risk GISTs and more low-risk GISTs based on the m-NIHC, as compared with those in international reports [
15,
16,
23]. This might be attributable to the well-established screening system for gastric cancer in Japan, which can detect several asymptomatic GISTs [
14,
24]. Finding asymptomatic tumors using established screening systems improves prognosis. In fact, the overall prognosis for patients who did not receive adjuvant therapy in this study was better than that previously reported for both 10-year OS and RFS rates (80.1% and 89.6%, respectively) [
16,
23,
25]. The 10-year RFS rate of the high-risk group was significantly lower than that of the other risk groups. This result suggests that the m-NIHC had a high sensitivity (76.3%) for predicting recurrence, and this criterion is clinically important in the selection of adjuvant therapy candidates, as previously described [
26].
This study is the first to validate contour maps in the Japanese population. Recurrent cases are plotted in the contour maps, which estimate the gradient recurrence risk for each patient. First, we evaluated the patients’ recurrence risk according to the contour maps, and their prognoses were evaluated. The predicted recurrence in each patient aligned well with their prognoses, with rates of recurrence within 10 years increasing with the predicted risk of recurrence. Furthermore, the actual recurrence rate was lower than the estimated recurrence rate, suggesting that the recurrence rate in Japanese patients might have more optimistic prognoses than those in other populations.
Undergoing adjuvant IM for 3 years has been established as the standard treatment for patients with a high risk of relapse, based on randomized trials [
9,
17,
27]. However, the benefit for RFS seemed to decrease after the end of adjuvant therapy, with an increasing risk of relapse during this period [
7‐
9]. Thus, adjuvant therapy cannot eradicate micro-residual GISTs but may merely delay the time of recurrence. Nishida et al. reported that patients with potential micrometastases are recommended to take adjuvant therapy for more than 3 years [
28]. Furthermore, some clinical trials are ongoing to determine whether extending adjuvant therapy with IM over 3 years can further reduce the risk of relapse and improve OS in patients with high-risk GISTs (NCT02413736, NCT02260505 and NCT01742299). However, it remains unclear for which high-risk patients are eligible for over 3 years. According to our results, the prognosis was different in the high-risk group according to m-NIHC. And, it may be useful for determination of the candidate for more intensive therapy. Thus, patients and surgeons will be able to obtain more detailed information from the contour maps with high sensitivity to recurrence (Online Resource 1) to make personalized decisions for postoperative treatment.
Tumor rupture is a serious risk factor for recurrence. Most ruptured GISTs are associated with recurrence at follow-up, and patients with ruptured GISTs have significantly shorter RFS than that those without rupture [
22,
29,
30]. In the present study, the rate of tumor rupture was 1.8% in this population. The 10 year-RFS rate of the patients with tumor rupture was quite worse than that of high risk patients without tumor rupture (Online Resource 2). Therefore, it is considered reasonable to separate ruptured GIST patients and non-ruptured patients in the contour maps.
Neoadjuvant therapy with IM has been shown by both prospective and retrospective studies to effectively decrease tumor size, thereby facilitating ease of surgery and resulting in less morbid, organ-preserving operations [
31‐
33]. For these patients, conventional risk classification is not applicable because pathological features are affected by IM. In particular, the mitotic counts of the specimen after neoadjuvant therapy have been reported to decrease remarkably, and we cannot fit patients into the m-NIHC or contour maps [
34].
18F-Fluorodeoxyglucose positron emission tomography/CT (
18F-FDG PET/CT) is noninvasive and has been reported to be an effective imaging technique for assessment of malignancy and monitoring responses to IM therapy in GIST [
34‐
39]. Proper selection of patients who require neoadjuvant therapy and risk evaluation after neoadjuvant therapy is important. However, prognostic factors such as mitotic count, exact tumor size, and tumor rupture during surgery can only be assessed postoperatively. In this regard,
18F-FDG PET/CT may aid in predicting the prognosis of neoadjuvant therapy patients. Further studies are needed to evaluate the possible role of
18F-FDG PET/CT in pathological assessment.
This study has some limitations. First, as this study was a registry study that included patients whose data were collected retrospectively, the follow-up period and way it was conducted were not standardized. Second, the very low-risk GISTs classification as defined by the m-NIHC included many incidental conditions of other serious diseases. In fact, while there were few cases of recurrence, 38 patients died during the observation period, and the 10-year OS rate was 69.4%, similar to that in the high-risk group (Online Resource 3). Of the patients who died in the very low-risk group, 92.1% were due to other diseases that were mainly the triggers for finding those small GISTs. Therefore, OS of the very low-risk GIST group does not reflect its own prognosis. Third, mutational analysis is known to be important in deciding whether to administer IM [
7,
40‐
42]. However, most patients in this study did not undergo mutational analysis. Therefore, we could not evaluate the response to adjuvant therapy by genetic mutations. Additionally, since this study was excluded the high-risk patients with perioperative therapy, it might have some bias. However, this study was mainly based on patients admitted before 2012, when the standard treatment of 3-year adjuvant therapy was established, the ratio of patients with perioperative therapy was relatively low. Nevertheless, since the data were based on consecutive patients in each participating hospital, the study reflects real-world data on Japanese patients.
In conclusion, contour maps are effective in predicting GIST recurrence after primary surgery in the Japanese population. Contour maps may be useful in combination with the m-NIHC for the consideration of more individual indications for adjuvant therapy.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.