Clinicopathologic Features and Molecular Characteristics of Glucose Metabolism Contributing to ¹⁸F-fluorodeoxyglucose Uptake in Gastrointestinal Stromal Tumors

Fluorine-18 fluorodeoxyglucose (18F-FDG) positron emission tomography–computed tomography (PET/CT) is useful in the preoperative diagnosis of gastrointestinal stromal tumors (GISTs); however, the molecular characteristics of glucose metabolism of GIST are unknown. We evaluated 18F-FDG uptake on preoperative PET/CT of 40 patients and analyzed the expression of glycolytic enzymes in resected GIST tissues by qRT-PCR, western blotting, and immunohistochemistry. Results of receiver operating characteristic curve analysis showed that the maximum standardized uptake value (SUVmax) cut-off value of 4.99 had a sensitivity of 89.5%, specificity was 76.2%, and accuracy of 82.5% for identifying tumors with a high risk of malignancy. We found that 18F-FDG uptake correlated positively with tumor size, risk grade, and expression levels of glucose transporter 1 (GLUT1), hexokinase 1 (HK1), and lactate dehydrogenase A (LDHA). Elevated HK and LDH activity was found in high-risk tumors. Among the isoforms of GLUT and HK, GLUT1 and HK1 expression increased with higher tumor risk grade. In addition, overexpression of glycolytic enzymes M2 isoform of pyruvate kinase (PKM2) and LDHA was observed in GISTs, especially in high-risk tumors. These results suggest that upregulation of GLUT1, HK1, PKM2, and LDHA may play an important role in GIST tumorigenesis and may be useful in the preoperative prediction of malignant potential.


Introduction
Gastrointestinal stromal tumors (GISTs), the most common non-epithelial neoplasms of the gastrointestinal (GI) tract, are defined as "KIT or PDGFRA mutation-driven mesenchymal tumors that can occur anywhere in the GI tract" [1]. Preoperative diagnosis of GISTs and assessment of their malignant potential are difficult because most GISTs are located in the submucosa. Tumor grading is therefore based primarily on mitotic index and tumor diameter.
The clinical usefulness of 18 F-fluorodeoxyglucose (FDG) positron emission tomographycomputed tomography (PET/CT) has been demonstrated in tumor staging, treatment response assessment, and prognosis prediction in various tumors [2][3][4][5][6]. This technique is also used in tumor staging and evaluation of targeted therapy response in GISTs [7][8][9][10][11][12]. Many tumors depend on aerobic glycolysis for rapid growth beyond that supported by the vasculature. 18 F-FDG is an analogue of glucose that allows noninvasive evaluation of the tumor's glucose metabolism, which can predict treatment response and patient prognosis. 18 F-FDG enters the cell through glucose transporters (GLUT), is phosphorylated to 18 F-FDG-6-PO 4 by hexokinase (HK), and is then "trapped" in the cell as it is not further metabolized, allowing PET/CT acquisition 1 hour after 18 F-FDG injection. Upregulation of GLUT and HK expression is associated with increased glucose metabolism and 18 F-FDG uptake in tumors. For example, previous studies have shown that increased GLUT1 expression correlates with higher 18 F-FDG uptake in lung and breast carcinomas [13][14][15].
Although 18 F-FDG PET/CT is useful for the preoperative diagnosis of GISTs, the detailed molecular mechanisms underlying glucose metabolism in these tumors and specific characteristics associated with tumor risk grade are not well understood. Tumor cells consume large amounts of glucose and produce large amounts of lactate compared to normal cells, even in the presence of oxygen. This metabolic switch from oxidative phosphorylation to increased glycolysis (i.e., the Warburg effect) is a common characteristic of malignant tumors [16,17] and regulated by transcription factors, such as hypoxia inducible factor-1α (HIF-1α), v-myc avian myelocytomatosis viral oncogene homolog (c-Myc), and tumor suppressor p53 (p53) [18][19][20]. In gastric cancer, 18 F-FDG accumulation represents tissue hypoxia, rather than GLUT expression [21]. However, analysis of the enzymes involved in glycolysis has not been performed in GISTs, and it is unclear whether the Warburg effect occurs in GISTs. Qualitative and quantitative analysis of glycolytic enzyme expression and their relationship with GIST tumor risk grade may clarify whether the Warburg effect occurs in GISTs. In this study, we aimed to identify 1) the relationship between maximum standardized uptake value (SUVmax) on preoperative 18 F-FDG PET/CT with GLUT and HK expression in GISTs, 2) the specific isoforms of GLUT and HK that are upregulated according to GIST tumor risk grade, and 3) alterations in the expression of various glycolytic enzymes according to tumor risk grade. By performing this study, we expect to identify the molecular biomarkers predictive of malignant GISTs that can be used in preoperative biopsy or cytology specimens and the molecular mechanisms of GIST detection by 18 F-FDG PET/CT.

Patient Selection
Our patient selection criteria specified the inclusion of patients diagnosed with GIST who underwent surgery after preoperative 18 F-FDG PET/CT, and a total of 40 GIST patients were included in our study. The cases were identified prospectively and consecutively between 2003 and 2013 at Severance Hospital, Yonsei University College of Medicine and from the Liver Cancer Specimen Bank, National Research Resource Bank Program of the Korea Science and Engineering Foundation of the Ministry of Science and Technology. Written informed consent for use of GIST tissues was obtained from all patients, and use of these tissues for research purposes was approved by the Institutional Review Board of Yonsei University of College of Medicine (IRB approval No.4-2015-0227). Risk of malignancy was categorized according to the system described by Fletcher et al [22].

Mutational analysis
DNA extraction was performed by using the QIAamp DNA FFPE Tissue Kit (Qiagen, GmbH, Hilden, Germany) according to the manufacturer's instructions. Primer sequences used to amplify v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog (KIT) exons are shown in S1 Table. PCR was carried out using a Veriti thermal cycler (Life Technologies, USA) with the following amplification conditions: 35 cycles of denaturation at 94°C for 30 s, annealing at 60°C for 30 s, and extension at 72°C for 1 min. The amplified products were purified using Agencourt AMPure XP (Beckman Coulter Genomics, Danvers, MA), and direct sequencing was performed using the BigDye Terminator Ready Reaction Cycle Sequencing kit and an ABI Prism 3130 genetic analyzer (Life Technologies).

PET/CT protocol and quantification
All patients underwent routine 18 F-FDG PET/CT scans with either the DSTe PET/CT scanner (GE Healthcare, Milwaukee, WI) or the Biograph TruePoint 40 PET/CT scanner (Siemens Medical Systems, CTI, Knoxv/ille, TN). Before 18 F-FDG injections, all patients fasted for at least 6 hours, and peripheral blood glucose levels were confirmed to be !140 mg/dL. The 18 F-FDG dose of approximately 5.5 MBq/kg body weight was administered intravenously 1 hour before image acquisition. After the initial low-dose CT (DSTe: 30mA, 130 kVp, Biograph TruePoint: 36 mA, 120 kVp), standard PET imaging was performed from neck to the proximal thighs (acquisition time, 3 min/bed) in three-dimensional mode. Images were then reconstructed using ordered-subset expectation maximization (2 iterations, 20 subsets). Written informed consent was obtained from all patients undergoing 18 F-FDG PET/CT scans.
Images were reviewed by an experienced nuclear medicine specialist on a GE AW 4.0 workstation. On PET scans, a volume of interest (VOI) was drawn on the primary lesion; SUVmax of the GIST was recorded.

Immunohistochemistry
Formalin-fixed and paraffin-embedded GIST tissue specimens were cut into 4-μm thick sections, and immunohistochemistry (IHC) analysis was performed using the Ventana Discovery

Quantitative reverse transcription-polymerase chain reaction
Primer sequences are listed in S2 Table. Quantitative reverse transcription-polymerase chain reaction (qRT-PCR) analysis was carried out in a final reaction volume of 20 μl with Premix Ex Taq II (Takara Bio, Otsu, Japan) according to the manufacturer's protocol. All reactions were run in triplicate on the StepOnePlus Real-Time PCR System (Applied Biosystems, Foster City, CA).

Hexokinase and lactate dehydrogenase activity assay
Hexokinase (HK) and lactate dehydrogenase (LDH) activities in GIST tissues were determined using a colorimetric (450 nm) kit (Sigma-Aldrich, St Louis, MO) according to the manufacturer's recommended protocol. Each sample was diluted as necessary to fall in the linear range of the standard curve and assayed in duplicate. Briefly, 10 μl samples were mixed with assay buffer in HK activity assay and 5 μl samples were mixed with assay buffer in LDH activity assay. Absorbance was measured at room temperature every 5 minutes (HK activity) and at 37°C every 2-3 minutes (LDH activity). Both enzyme activities were assessed using the following equation: where B is the amount of NADH generated between T initial and T final , and (T final -T initial ) indicates the reaction time. V is the sample volume (mL) added to the reaction well. All data were expressed as milliunit per mg of total protein. One unit of HK and LDH is the amount of enzyme that generate 1 μmole of NADH per minute.

Statistical analysis
Relationships between tumor risk grade and other parameters were evaluated using either chisquare test or one-way analysis of variance. Correlation analysis was performed with Pearson's or Spearman's correlation test. Student's t-test was used to compare two groups of continuous variables. The area under the receiver operating characteristic (ROC) curve was used to determine the SUV cut-off level able to predict tumor risk grade with the highest sensitivity. Data are expressed as mean ± standard deviation (SD); P < .05 was considered significant. Statistical analysis was performed using SPSS for Windows (version 21.0; SPSS Inc., USA).

Clinicopathologic characteristics and 18 F-FDG uptake of 40 GISTs
Of the 40 patients included in the study, 22 were women and 18 were men (S3 Table), and mean patient age was 59 years (range 20-83). Of the 40 GIST lesions, 18 were located in the stomach, 22 were located in the small or large intestine. Five lesions were not well visualized on PET (SUVmax<2.5), 12 showed moderate FDG uptake (2.5<SUVmax<5.0), and 23 showed intense FDG uptake (SUVmax!5.0). High-risk tumors were more common in men (Table 1). Histologic features of representative cases are shown in Fig 1A. FDG uptake did not differ significantly between gastric GISTs and non-gastric GISTs (data not shown); however, SUVmax strongly correlated with tumor size (r = 0.716, P < .001). In  Correlation between 18 F-FDG uptake and tumor risk grade Tumor risk grade [22] correlated significantly with 18 F-FDG uptake (P < .001) ( Table 1 and Fig 1B). SUVmax was lower for low-risk (4.3±3.2) and intermediate-risk tumors (5.4±2.7) than for high-risk tumors (10.9±5.2; P < .001).
To evaluate the usefulness of 18 F-FDG PET/CT in predicting the malignant potential of GISTs, we compared high-risk tumors with intermediate-and low-risk tumors. ROC curve analysis showed that a SUVmax cut-off of 4.99 was the most sensitive for predicting malignancy, and area under curve was 0.875 (P < .001). Using this SUVmax cut-off value to differentiate high-risk tumors from low-and intermediate-risk tumors, sensitivity was 89.5% (16/ 18), specificity was 76.2% (17/22), and accuracy was 82.5% (32/40).

Overexpression of GLUT1 and HK1 in GISTs according to tumor risk grade
Because isoforms of GLUT and HK are overexpressed in tumors and associated with 18 F-FDG uptake, we evaluated the expression of four isoforms of GLUT (GLUT 1, 2, 3, and 4) and two isoforms of HK (HK1 and HK2) in the 40 GISTs. First, we analyzed total HK activity. We found that HK activity was significantly upregulated in high-risk GISTs, compared to low-risk and intermediate-risk GISTs (P < .05) (Fig 2A). Results of qRT-PCR analysis showed gradual increases in GLUT1 and HK1 expression with higher tumor risk grade (Fig 3A). In contrast, We next analyzed GLUT1 and HK1 protein expression by IHC and western blot analysis. GLUT1 and HK1 protein were detected in most GISTs; however, GLUT1 was not detected in four GISTs, and HK1 was not detected in five GISTs ( Table 2). Most of the tumors lacking expression of these two proteins were low-risk GISTs (Table 2 and Fig 4A). Results of IHC staining showed that SUVmax correlated with both GLUT1 expression (r s = 0.465, P = 0.002) and HK1 expression (r s = 0.446, P = 0.004) (Fig 4B). These findings suggest that GLUT1 and HK1 expression are related to PET signals. Results of western blotting analysis showed expression patterns of GLUT1 and HK1 that were similar to qRT-PCR and IHC results (Fig 4C).

Enhanced expression of PKM2 and LDHA in high-risk GISTs
In order to elucidate the presence of the Warburg effect in GISTs, we evaluated mRNA and protein expression levels for PKM2 and LDHA. Results of qRT-PCR showed a gradual increase in PKM2 and LDHA expression with higher tumor risk grade (Fig 3B). PKM2 expression was significantly higher in high-risk GISTs (3.39±3.51) than in low-risk (0.69±0.67) and intermediate-risk GISTs (0.85±0.36; P < .05). Similarly, LDHA expression was significantly higher in high-risk GISTs (1.37±1.09) than in low-risk (0.49±0.63) and intermediate-risk GISTs (0.89 ±0.92; P < .05).
Next, we evaluated protein levels of PKM2 and LDHA in the 40 GISTs by IHC and western blotting analysis. These two glycolytic enzymes were detected in most GISTs; however, PKM2 was not expressed in eight GISTs ( Table 2). Most of the tumors lacking protein expression of these enzymes were low-risk GISTs. Protein levels of PKM2 and LDHA were increased in intermediate-and high-risk GISTs (Table 2 and Fig 5A). We also found a positive correlation between SUVmax and LDHA protein expression. SUVmax was significantly higher in tumors with strong LDHA expression than in tumors with weak LDHA expression (r s = 0.466, P = 0.002) (Fig 5B). Results of western blotting analysis showed that PKM2 and LDHA expression patterns were similar to IHC results (Fig 5C). Additionally, we measured LDH activity in our GIST tissues to evaluate metabolism status. LDH activity was significantly upregulated in high-risk GISTs, compared to low-risk and intermediate-risk GISTs (P < .01) (Fig 2B). These findings suggest that upregulation of LDHA is involved in the malignant potential and PET signal of GISTs.
We further evaluated the relationship between the glycolysis-modulating transcription factors and glycolytic markers that we had selected (GLUT1, HK1, PKM2, and LDHA). First, we analyzed HIF-1α, c-Myc, and p53 expression using qRT-PCR and IHC. We found that mRNA expression levels of HIF-1α were higher in high-risk GISTs

Discussion
In this study, we evaluated clinicopathologic factors contributing to 18 F-FDG uptake in the GISTs of 40 consecutive patients. Positive correlations were found between 18 F-FDG uptake Table 2. Relationships between GLUT1, HK1, PKM2, and LDHA immunohistochemical expression and GIST tumor risk grade.

Category
Tumor and several clinicopathologic features and metabolic factors. SUVmax correlated with tumor size; expression of GLUT1, HK1, and LDHA; and NIH risk group. 18 F-FDG uptake was also significantly higher in GISTs with mitotic count >10/50 HPF compared to GISTs with lower mitotic count. The optimal SUVmax cut-off value for identifying tumors with a high risk of malignancy (NIH risk classification) was 4.99 (sensitivity, 89.5%; specificity, 76.2%). These findings suggest that PET/CT may be useful for preoperative assessment of malignant potential.
We demonstrated significant overexpression of GLUT1 and HK1 protein in high-risk GISTs by IHC and western blot analysis. We also demonstrated that HK activity in upregulated in high-risk GISTs. GLUT and HK are important for glucose uptake, and the GLUT family of transporters has been implicated in 18 F-FDG uptake, with GLUT1 and GLUT3 in particular playing important roles in 18 F-FDG accumulation [13,23]. The role of HK in 18 F-FDG uptake has been studied in various tumors [15,24,25]. In our study, the isoforms GLUT1 and HK1 were specifically overexpressed in high-risk GISTs, however, the expression levels of GLUT2, 3, 4, and HK2 showed no relationship with GIST risk grade. Overexpression of GLUT1 in the cell membrane had been reported in many tumors, and overexpression of GLUT2 and GLUT3 has been reported in hepatocellular carcinoma and malignant lymphoma, respectively [26,27]. Although GLUT1, as well as GLUT3 and GLUT4, expression was found in one GIST cell line (GIST-T1) [28], only the expression of GLUT1 was related to GIST risk grade. The paired overexpression of GLUT1 and HK1 in GISTs, and the correlation between increased overexpression of these two proteins and tumor risk grade provide evidence for increased glucose uptake and abnormal glucose metabolism in GISTs, which may be useful in preoperative diagnosis and the development of novel therapeutic targets.
In this study, we demonstrated that the Warburg effect exists in GISTs and observed a correlation between 18 F-FDG uptake and tumor risk grade. Most GISTs evaluated in this study showed significant overexpression of the glycolytic enzymes PKM2 and LDHA, with the degree of overexpression increasing with higher tumor risk grade. We tried to identify which transcription factors induce glycolytic markers in GISTs and found moderate correlation between HIF-1α expression and GLUT1, HK1, and LDHA expression. This metabolic switch from oxidative phosphorylation to increased glycolysis is one of the principle biochemical characteristics of malignant cells [16]. In addition, PKM2 possesses protein tyrosine kinase activity and plays a role in modulating gene expression, thereby contributing to tumorigenesis [29]. For example, enhanced PKM2 expression correlates with aggressive tumor behavior (in vivo tumor growth, tumor cell proliferation, migration) in colon cancer [30]. LDHA expression is also elevated in many types of cancers and is linked to tumor growth, maintenance, and invasion [31][32][33][34]. Therefore, LDHA inhibition may restrict the energy supply in cancer cells, thereby decreasing their tumorigenicity [24,35]. This enzyme may also be useful as a diagnostic marker or predictive biomarker for many types of cancer, as well as a therapeutic target for new anti-cancer treatments. In addition, the correlation between 18 F-FDG accumulation and LDHA expression and the possible modulation of 18 F-FDG uptake through LDHA-AKT-GLUT1 signaling has been reported in lung adenocarcinoma [24]. Based on these reports, our findings showing overexpression of PKM2 and LDHA and the correlation between degree of expression and tumor risk grade in GISTs indicate that overexpression of PKM2 and LDHA may play important roles in GIST tumorigenesis and suggest their usefulness as potential therapeutic targets.
In conclusion, we evaluated the usefulness of preoperative 18 F-FDG PET/CT for the prediction of malignant potential in GISTs and the relationship between tumor risk grade and the expression of proteins involved in glucose metabolism. Our results showed that 18 F-FDG uptake correlates with tumor size, tumor risk grade, and expression levels of GLUT1, HK1, and LDHA. The increased expression of GLUT1, HK1, PKM2, and LDHA with higher tumor risk grades indicates important roles for these proteins in GIST tumorigenesis and suggests their usefulness in the preoperative prediction of malignant potential.