Re‐appraisal of the universal definition of tumor rupture among patients with high‐risk gastrointestinal stromal tumors

Abstract Aim Tumor rupture has been indicated as a risk factor for recurrence of gastrointestinal stromal tumors (GISTs). The universal definition of tumor rupture was proposed. This study evaluated whether the universal definition was more accurate in identification of GISTs with high recurrent risk than subjective judgment. Methods The study included 507 patients with high‐risk GISTs who underwent complete resection between December 2012 and December 2015. We conducted a questionnaire survey in participating institutes to re‐diagnose tumor rupture based on the universal definition according to their surgical and pathological findings. We compared the clinical outcomes of tumor rupture based on the definition to those based on the surgeon's judgment and clarified the clinical importance of the rupture. Results Sixty‐four patients were initially registered to have tumor rupture by surgeon's judgment, and it became 90 patients who had tumor rupture after reevaluation. Although there were significant differences in recurrence‐free survival (RFS) between no rupture and rupture for both initial registration and reevaluation (p = 0.002, <0.001, respectively), a significant difference in overall survival was only observed after reevaluation (p = 0.011). Tumor rupture was significantly associated with large tumor size, mixed cell type in histology, R1 resection, frequent adjuvant therapy and recurrence, but not with location, mitosis, and genotype. Adjuvant therapy more than 3 years improved RFS of patients with tumor rupture. Conclusion This study suggested that tumor rupture based on the universal definition more accurately identified GISTs with poor prognostic outcomes than the subjective judgment.


| INTRODUC TI ON
Gastrointestinal stromal tumor (GIST) is the most frequent sarcoma in the gastrointestinal tract.Complete resection is the gold standard in the treatment of primary GISTs.2][3] Tumor size, mitosis count, tumor site, and tumor rupture are considered to be risk factors for recurrence. 1,4-9Among the four factors, tumor size, mitosis count, and tumor site are documented objectively even in retrospective previous studies, and are included in the most risk stratification and nomograms.The prognosis of patients with ruptured GISTs was reported as poor, 5,8,[10][11][12][13] however, as the diagnosis of tumor rupture is clinical judgment and is infrequent, prognostic outcomes of patients with ruptured GISTs were inconsistent.Clinical significance of tumor rupture is still controversial due to the lack of a commonly shared definition and low statistical power.Therefore, tumor rupture is not always included in the risk stratification or nomograms.
It has been shown that surgeons have varied criteria in their judgment of tumor rupture. 14Therefore, the universal definition of tumor rupture in GIST was proposed in 2019. 15This observational substudy on patients from the prospective registry study (the STAR ReGISTry study 16 ) evaluated whether a clinical decision based on the universal definition more efficiently identifies GISTs with high recurrent risk compared with individual judgment.We also examined prognostic outcomes of ruptured GISTs and evaluated the clinical significance of tumor rupture based on the universal definition.

| Patients
A total of 541 patients with primary high-risk GISTs were enrolled to the prospective observational registry study (the STAR ReGISTry study) after complete resection (R0 or R1) between December 2012 and December 2015. 17The modified NIH consensus criteria were adopted for the risk stratification. 1 Other inclusion criteria include KIT-positive or mutation-positive in KIT, PDGFRA, and/or other genes, no metastatic disease (neither metastasis nor peritoneal disease), older than 20 years of age, and written informed consent.
Patients with other cancers, cardiovascular, brain, hepatic, or renal diseases and recurrent diseases were excluded.Surgical specimens were confirmed by the central pathological diagnosis.
Among 541 patients, 26 patients were excluded for various reasons including unmet inclusion criteria and non-GISTs by the central pathological diagnosis and, finally, 515 patients were evaluated in the STAR ReGISTry study. 16This study also evaluated the same 515 patients.The clinicopathological data of registered patients were collected from participating institutions.The tumor rupture was judged by a physician or a surgeon of each institution in the initial registry.However, the definition of tumor rupture was proposed in 2019. 15Thus, all registered patients were reevaluated according to the classification of the universal definition of tumor rupture.The data of eight patients were not available in terms of the rupture classification and were excluded from the study.Finally, 507 GIST patients with or without tumor rupture were retrospectively analyzed.
The treatment for GIST patients was not prescribed in the protocol, however, the adoption of the Clinical Practice Guidelines 2012 18 was recommended among participating institutions.

| ME THODS
This study is an observational study as the STAR ReGISTry study. 16ter identifying cases of tumor rupture registered in the STAR ReGISTry study, a questionnaire was sent to all participating institutions to reevaluate enrolled patients based on surgical and pathological findings based on the universal definition.The subtypes of tumor rupture were also requested to classify according to the definition in this questionnaire.We compared the clinical outcomes of tumor rupture based on the universal definition with those based on the surgeon's initial judgment at registry.Additionally, we evaluated prognostic effects of adjuvant therapy according to treatment duration: less than 1 year, between 1 year and 3 years, and more than

| The classification of tumor rupture in GIST
The classification of tumor rupture in GIST was proposed by Nishida et al in 2019. 15The spectrum of tumor rupture in GIST

| Statistical analysis
Recurrence-free survival (RFS) and overall survival (OS) of this study were defined as the same as in the STAR ReGISTry study.RFS was defined as the time from the date of surgery to the first documented recurrence or death from any cause.OS was defined as the time from the date of surgery to the death from any cause.Categorical variables were tabulated using n (%) and continuous variables using median and interquartile range (IQR).For comparisons of categorical variables, a chi-squared test was used, and for comparisons of continuous variables, a Mann-Whitney U test was used.The cumulative survival rates for RFS and OS were estimated by the Kaplan-Meier method.The hazard ratio (HR) and its 95% confidence interval (95%CI) were calculated using the Cox's proportional hazards model.The p values were two-sided, and p values <0.05 were considered significant.All statistical analysis were performed using SPSS PASW version 18.0 (SPSS Inc.).

| Changes in tumor rupture after definition
Among 507 patients, 64 patients were registered to have tumor rupture in the initial case report of the STAR ReGISTry study, and 90 patients were reported to have tumor rupture by the questionnaire.
Changes in subtypes of tumor rupture are shown in Table 1.The total number of subtypes of tumor ruptured patients was 74 (some patients had multi-subtypes) in the initial report.Eleven patients with blood-stained ascites (type 2), 14 with adjacent organ infiltration (type 4), one with piecemeal resection/intralesional dissection (type 5), and three with incisional biopsy (type 6) were added as ruptured GIST patients by the reevaluation (some overlapping).Of 90 patients, 17 patients had multiple factors.One patient was type 1A and 2 and 3 with three types of overlap, and the remaining 16 had two factors of overlap: seven with types 1A and 2, one with types 1B and 2, one with types 1B and 3, two with types 2 and 4, two with types 2 and 5, two with types 3 and 4, and one with types 4 and 5.
Comparisons between each group were difficult to analyze due to the small number of cases.

| Clinicopathological characteristics of patients with tumor rupture
Among 507 registered patients, we compared clinicopathological features of patients without tumor rupture (NTR group; N = 417) and those with the rupture (TR group; N = 90) according to the universal definition.Between NTR and TR groups, there were no significant differences in age, gender, performance status (PS), tumor location (stomach, small intestine and rectum, or others), mitosis, and genotype (Table 2).Microscopically residual tumor (R1) was significantly higher in the TR group compared with the NTR group (TR and NTR: 14.4% vs. 1.0%, p < 0.001).Tumor size was significantly larger in the TR group (median size of TR and NTR: 10 and 7 cm, p < 0.001), and, thus, open surgery was more frequent in the TR group (TR and NTR: 84.4 vs. 69.3%,p = 0.002).
Although the number of patients who underwent neoadjuvant therapy was similar between the two groups, patients with adjuvant therapy was significantly higher in the TR group (TR and NTR: 87.8 vs. 78.9%,p = 0.033).The duration of adjuvant therapy was significantly shorter in the TR group (median duration of TR and NTR: 2.5 vs. 3.0 years, p = 0.023).In the histological cell type, mixed (spindle and epithelioid) features are more predominant in the TR group (16.7%) compared with the NTR group (7.7%).

TA B L E 1
The distribution of tumor rupture patients based on the defined classification.

Number of reevaluated report of tumor rupture based on defined classification
Tumor fracture (type 1A) 21  21 Tumor spillage (type 1B) 5 5 Blood-stained ascites (type 2) Gastrointestinal perforation on tumor (type 3) Adjacent organ infiltration (type 4) Piecemeal resection or intralesional dissection (type 5) Incisional biopsy (type 6) 0 3 Total 74 103 Recurrence was significantly frequent in the TR group (TR and NTR: 52.2 vs. 29.7%,p < 0.001).The liver was the most frequent site in the NTR group (55.3%) as the first site of recurrence, the peritoneum was the most frequent site in the TR group (63.8%,Table 2).

| Comparison of RFS and OS between NTR group and TR group
We evaluated prognostic outcomes of patients with ruptured GISTs before and after reevaluation.
OS was not different between NTR and TR groups in the initial report (p = 0.210).The 3-year/5-year OS rates of NTR and TR groups were 97.9/93.1% (NTR, 3y/5y) and 96.7/85.4% (TR, 3y/5y), TA B L E 2 Patients' clinicopathological characteristics according to the status of tumor rupture.respectively.In contrast, OS was significantly different between the two groups in the reevaluated report (p = 0.011, Figure 1C, D).
The median RFSs (95%CI) of the TR group in the initial and reevaluated reports were 6.9 (3.0-10.7)and 5.1 (3.0-7.3)years, respectively.Median OSs could not be estimated in both situations.

| Effects of various duration of adjuvant therapy on ruptured GISTs
RFS and OS were evaluated according to adjuvant duration among 90 patients with tumor rupture based on the universal definition.
Clinicopathological comparison of these patients according to various adjuvant therapy durations was shown in Table 3.No significant differences were found among the three groups.Median duration of adjuvant imatinib therapy for less than 1 year and that for between 1 and 3 years were 0.41 (IQR: 0.085-0.72)years and 2.7 (IQR: 1.9-2.9)years, respectively, and that of adjuvant therapy more than 3 years was estimated to be 3.3 (IQR: 3.0-5.0)years and five patients were still on the therapy at data cut-off.
In RFS, there was no difference between patients with adjuvant therapy less than 1 year and those with adjuvant for 1 to 3 years (p = 0.356); however, there were significant differences between patients with adjuvant for 1 to 3 years and those with adjuvant for more than 3 years (p = 0.014) and between less than 1 year and more than 3 years (p = 0.013, Figure 2A).The median RFS (95%CI) and the OS was similar among three groups of adjuvant duration (less than 1 year, 1 to 3 years, and more than 3 years, Figure 2B).The 3-year/5-year OS rates of patients with adjuvant therapy of less than 1 year, 1 to 3 years, and more than 3 years were 88.2/81.5%,96.8/82.4%, and 100.0/100.0%,respectively.The median OS was not estimated in all groups.
There were 16 patients who relapsed during adjuvant therapy.
RFS and OS were evaluated excluding these patients (Figure S1), the results of RFS and OS were similar to Figure 2A, B.

| Prognostic factors of OS and RFS
Prognostic analysis of Cox proportional hazards model for RFS and OS after resection of GISTs is shown in Tables 4 and 5

| DISCUSS ION
The universal definition of tumor rupture in GIST was proposed in 2019. 15Using data from a prospective registry of high-risk GISTs, we and from 0 to 3, respectively, suggesting that these three types may not be commonly recognized by surgeons as tumor rupture.
Blood-stained ascites was defined as preoperative tumor rupture in the SSGXVIII/AIO tria. 19In our study, among 30 patients with blood-stained ascites, one had gastrointestinal perforation at the tumor site, and eight had tumor fracture or spillage at laparotomy.Thus, blood-stained ascites is considered to be tumor rupture.
Incisional biopsy is technically similar to a piecemeal resection and is not recommended in oncology, rather, endoscopic or percutaneous needle biopsy is recommended as a biopsy for histological diagnosis.
Histologic invasion of adjacent organs is very infrequent in GISTs.
Tumor cell infiltration into adjacent organs implicates a disruption of the peritoneal lining and fibrous capsule between GISTs and neighboring organs and potential exposure of tumor cells to the peritoneal cavity.Retrospective studies have indicated that GISTs with histological invasion have worse prognostic outcomes than GISTs without it.In our study, patients with these three types of tumor rupture appear to have high recurrences similar to those with type 1A/B, type 3, and/or type 5 rupture (Figure 3).Furthermore, although the liver was the most frequent locus as the first recurrence site in the NTR group, the peritoneum was the most frequent in the TR group as indicated by previous report. 20,21These results indicate that GISTs conducted in accordance with the World Medical Association Declaration of Helsinki, Ethical Principles for Medical Research Involving Human Subjects (Amended in Seoul in October 2008), and the Ethics Guidelines for Clinical Research (Ministry of Health, Labour and Welfare Notice No. 415, 2008).Ethical approval was initially obtained from the institutional review board (IRB) of the National Cancer Center and then from the IRBs of the other participating hospitals (No. 2016-250).Written informed consent recurrence, but not with location, mitosis, and genotype.Adjuvant therapy more than 3 years improved RFS of patients with tumor rupture.Conclusion: This study suggested that tumor rupture based on the universal definition more accurately identified GISTs with poor prognostic outcomes than the subjective judgment.K E Y W O R D S gastrointestinal stromal tumor, high-risk GIST, tumor rupture, universal definition was obtained from all participating patients.The STAR ReGISTry study was registered with the UMIN Clinical Trials Registry, number UMIN000009531.
was represented in six different clinical scenarios.The classification was the following six definitions for "tumor rupture": (type 1) tumor fracture or spillage; (type 2) blood-stained ascites; (type 3) gastrointestinal perforation at the tumor site; (type 4) microscopic infiltration of an adjacent organ; (type 5) piecemeal resection or intralesional dissection; or (type 6) incisional biopsy.Furthermore, tumor fracture is classified as type 1A and tumor spillage is classified as type 1B.

F 2 ( 6 . 2 - 8 . 3 )
I G U R E 1 RFS and OS after surgery according to the status of tumor rupture.A and B show the RFS curves.C and D show the OS curves.A and C include 64 patients of tumor rupture group at the initial report.B and D include 90 patients of the reevaluated report of tumor rupture based on the universal definition.The green line indicates the non-tumor rupture group, and the blue line indicates the tumor rupture group./5-year RFS rates of patients with adjuvant therapy of less than 1 year, 1 to 3 years, and more than 3 years were 1.9 (1.1-2.7) years and 40.3/40.3%,4.3 (3.6-4.9) years and 71.0/39.8% and 7.years and 100.0/81.6%,respectively.
compared individual surgeon judgment of tumor rupture with those based on the universal definition.The results showed that the universal definition identified GISTs with poorer prognostic outcome.Adjuvant therapy for more than 3 years improved RFS of patients with tumor rupture.When a prospective registry study started, there was no definition of tumor rupture, hence, diagnosis of tumor rupture was left to individual investigators in the initial case reports.After the definition, tumor rupture was reevaluated based on the universal definition.Comparing two judgments, patients with tumor rupture were almost the same for tumor fracture/spillage (type 1A/B), gastrointestinal perforation on the tumor (type 3), and piecemeal resection/ intralesional dissection (type 5), suggesting that type 1A/B, type 3, and type 5 are commonly shared as tumor rupture among participating investigators.On the contrary, the number of patients for blood-stained ascites (type 2), adjacent organ infiltration (type 4), and incisional biopsy (type 6) increased from 19 to 30, from 7 to 21,

3
with any one of six rupture types may harbor potential dissemination of tumor cells in the abdominal cavity.Collectively, inclusion of blood-stained ascites, incisional biopsy, and histologic invasion may facilitate selection of patients with worse prognostic outcomes than those with high-risk GISTs.The 5-year RFS rates of patients with tumor rupture according to individual surgeons' judgment and of those based on the universal definition were 55.2% and 50.5%, respectively.Furthermore, tumor rupture based on the universal definition, but not rupture according to individual judgment, was significantly related with poorer OS than those without rupture.These results indicate that diagnosis of F I G U R E 2 RFS and OS after surgery according to various adjuvant therapy duration among 90 patients of reevaluated report of tumor rupture based on the universal definition.A shows the RFS curves and B shows the OS curves.The red line indicates the more than 3-year adjuvant group, the green line indicates the 1-year to 3-year adjuvant group and the blue line indicates the less than 1-year adjuvant group.RFS and OS after surgery according to 90 patients of reevaluated report of tumor rupture based on the universal definition including non-tumor rupture patients.A shows the RFS curves and B shows the OS curves.The blue line indicates type 1A in the universal definition, the green line indicates type 1B, the orange line indicates type 2, the purple line indicates type 3, the yellow line indicates type 4, the red line indicates type 5, the light blue line indicates type 6, and the black line indicates the non-tumor rupture group.

year group (N = 37) 1 to 3 years group (N = 31) More than 3 years group (N = 22) p value
tion of adjuvant therapy were significantly associated with RFS.Age, performance status, tumor size, tumor rupture, mitosis, genotyping, histological type, and duration of adjuvant therapy were significantly associated with OS.Multivariate Cox proportional hazards model analysis revealed that mitosis, histological type, and duration of adjuvant therapy were significantly associated with both RFS and OS.TA B L E 3 Patients' clinicopathological characteristics according to various adjuvant therapy duration.Less than 1