The Role of 2-[ F-18 ] Fluoro 2-Deoxy D-Glucose Positron Emission Tomography in the Preoperative Staging of Gastric Cancer

Objective: FDG-PET can contribute significantly to the preoperative period of patients with planned curative resections or with isolated liver or lung metastasis. In this study, we analyze the importance and diagnostic value of FDG-PET in the preoperative evaluation of gastric cancers by correlating its value with computerized tomography (CT). Materials and Methods:A prospective study was conducted. Abdominal and pelvic CTs were obtained for preoperative staging in each patient. Separate stagings were performed according to the findings of conventional methods and PET scans, and the results were compared with the operative and histological findings. Results: When the patients were evaluated according to the pathological stage, eight patients were Stage 1 (26.7%), five were Stage 2 (16.7%), eight were Stage 3 (26.7%), and nine were Stage 4 (30%). FDG-PET and CT down-staged the disease in eight (26.7%) and twelve (40%) patients, respectively. FDG-PET and CT up-staged the disease in two (6.7%) and five patients (16.7%), respectively. FDG-PET accurately staged the disease in eighteen patients (60%) (p= 0.182). Conclusion: To develop highly sensitive radioactive tests that show invasive local lymph node and peritoneal metastases, more studies in the preoperative staging of gastric cancer are needed. FDG-PET indicates a high specificity in the staging of gastric cancer. In patients who are planned for curative surgery, FDG-PET provides better staging when used with CT.


Introduction
The treatment of gastric cancer starts with a good staging.The thickness of gastric tumors on the gastric wall, lymph node invasion, and metastases can be identified by modern endoscopy, endoscopic ultrasound, conventional imaging methods, and laparoscopic methods.
More effective imaging studies are preferred to decrease morbidity by preventing unnecessary surgical intervention and by determining patients who would benefit the most from curative surgery [1][2][3][4].Positron Emission Tomography (PET) is a nuclear medicine imaging method.It functions on the basis of imaging with the help of detectors indicating the distribution of fluoro-deoxy glucose (FDG), generated via fluorescent marking of glucose.The distribution in the body is greater in tumor tissue compared to normal tissue.This method has been widely used in the field of oncology since the 1990s [4,5].Although there are many publications describing its use in the staging of lung, colon, esophagus, and head and neck tumors, there is little published literature on its role in gastric cancer.Its value in gastric cancer has not been highlighted yet.FDG-PET can contribute significantly to the preoperative period of patients with planned curative resections or with isolated liver or lung metastasis [5,6] In this study, we analyze the importance and diagnostic value of FDG-PET in the preoperative evaluation of gastric cancers by correlating its value with computerized tomography (CT).

Materials and Methods
Thirty patients [22 males (73.3%) and 8 females (26.7%); median age: 61 years (min-max: 35-80 years)] referred to the GATA General Surgery Department from January 2005 to July 2007 for advanced examinations and treatment with the diagnosis of "gastric adenocarcinoma" according to endoscopic biopsies were included in the study.A prospective study was conducted.Abdominal and pelvic CTs were obtained for preoperative staging in each patient, and FDG-PET was performed in addition to other preoperative laboratory tests and radiological imaging.At most, seven days were allowed between CT and PET.
Written consent was obtained from all patients after supplying information about the aim of the study and the tests that would be performed.Approval from the local ethics committee was received for the study.Pregnancy, lactation, possibility of pregnancy, patients with a history of hypersensitivity to FDG-PET, and patients with uncontrolled advanced diabetes were excluded from the study.
All patients drank an oral solution of 750-1000 mL of water and 50 mL of Urografin approximately one hour prior to the examination.Following the intravenous introduction of a minimum of 90 ml and a maximum of 150 ml of contrast medium (Omnipaque 350) according to the weight of the individual, images in 5 mm sections were obtained in the prone position by Multislace BT (Philips Medical Systems MX 8000 IDT) equipment.Radiologists blind to the FDG-PET findings interpreted the images.Computerized tomography images were evaluated and interpreted in terms of the width of the tumor, invasion of neighboring structures, liver metastasis, lymph node invasion, and extrahepatic metastasis."Wall thickness and mass image" were accepted as significant for primary tumors.
The FDG-PET images were obtained by an ECAT EXACT trademark PET scanner (Siemens/ CTI, Knoxville, TN, USA).The fasting blood glucose levels were measured prior to the examination.After eight hours of fasting and an intravenous injection of 10 mCi (370 MBq-555 MBq) F-18 FDG, consequent transmission and emission images were obtained at the 60th minute in the supine position.The images were evaluated by whole body projections and in three orthogonal plane sections.The metabolic activity regions with an SUVmax value higher than 2.3 were accepted as positive for malignancy.
The FDG-PET images were correlated with the abdominopelvic CT images.Separate stagings were performed according to the findings of conventional methods and PET scans, and the results were compared with the operative and histological findings in patients who had curative resections and complete dissections and with clinical follow-up findings in patients who did not have curative resections and complete dissections.
The computerized tomography and PET stagings were performed according to the 2002 6 th edition of TNM classification.The important factors for staging, such as the depth of tumor invasion on the wall (T), lymph node involvement (N), and presence of metastasis (M), were identified by both the CT and PET scans.The sensitivity, specificity, positive predictive values, and negative predictive values of both examinations in all stages were calculated and were compared according to the pathological stage.
Total or subtotal gastrectomy and D1 or D2 lymph node dissections, according to the lymph node classification of the Japanese Research Society for the Study of Gastric Cancer, were performed on patients.If present, any other organs with tumor involvement were also resected.
The SPSS 15.0 statistical package program was used for statistics.The frequency, percentages and median (min-max) were used for descriptive statistics.The validity was evaluated by the indicators specificity, sensitivity, positive predictive values and negative predictive values.The non-parametric Mann-Whitney U and McNemar tests were used for comparisons between the groups.Correlations between the parameters were evaluated by the Spearman correlation coefficient.A p value <0.05 was accepted as statistically significant.

Results
Early gastric cancer (EGC) was identified in 9 patients, and advanced gastric cancer (AGC) was identified in 21 of 22 male patients and in 8 female patients.Among the Early Stage Gastric cancers, 7 were diagnosed by FDG-PET (77.7%).The other 21 patients had advanced stage gastric cancer, and 20 of them were diagnosed by FDG-PET (95%), with a median SUVmax value of 5.2 (min-max; 2.3-24.9).The tumors of patients who could not be diagnosed by FDG-PET contained mucinous and stromal tumor components, each in one patient, and one patient had a signet ring cell adenocarcinoma.Nine of the tumors (30%) were T1, four were T2 (13.3%), and eight were T4.Three patients (10%) had a signet ring cell component, three had (10%) undifferentiated tumors, seven (23%) had intermediate tumors, and seventeen (56.7%) had well-differentiated adenocarcinoma.Two tumor foci were found in seven patients, while perineural invasion was identified in three patients.The CT was not helpful in identifying the T-level in two patients (6.7%).
Total gastrectomy and subtotal gastrectomy were performed in 25 and 4 patients, respectively.D2 and modified D1 lymph node dissection was performed in 24 and 5 patients, respectively.Diagnostic laparoscopy was performed in one patient.Splenectomies were performed in seven additional patients.
An intermediate level (r=0.48) and positive statistically significant difference was found between the tumor dimensions and the SUV (p=0.007).The median SUV in tumors greater than 3 cm was 6.1 (min-max: 2.6-24.9).The median SUV in tumors smaller than 3 cm was 4.6 (min-max: 2.3-5.4).This difference was statistically significant (p=0.011).When the tumor invasion and tumor depth increased, the SUV was observed to increase.Statistically, this increase was interpreted as a positive intermediate level difference (r=0.39;p=0.032).
Lymph node evaluations could not be carried out by FDG-PET and CT in three (10%) and five (16.7%) patients, respectively.No significant differences were found between the SUV in patients with no node involvement, with a median SUV of 4.75 (min-max: 2.3-8.5), and in patients with node involvement, with a median SUV of 5.85 (min-max: 2.6-24.9)(p=0.785).
Lymph node involvement was identified as N1 in ten patients (30%), N2 in four patients (13.3%), and N3/4 in three (10%) patients.In patients with N3/4 lymph node involvement, the PPV (positive predictive value) and sensitivity of CT were '0' .This result may be due to the small number of cases.In general, FDG-PET was considered a more specific test compared to CT (Tables 1 and 2).
Both CT and FDG-PET demonstrated liver metastases in two patients and a preoperative right adrenal metastasis in one patient.In one patient with a 3x5 cm metastasis in the left lobe of the liver, a left lateral segmentectomy was added to the surgical treatment because the metastasis was limited to the left lobe of the liver.In the other patient with liver metastasis, the metastases were multiple and nodular, and a segment 4 resection was performed to include the biggest nodule with a 3 cm diameter.
Preoperative CT and FDG-PET demonstrated multiple peritoneal implants and ascites in one of the two patients with this condition.

Discussion
There are few publications regarding FDG-PET and its use in gastric cancers [7,8].Its use in gastric cancers is new and uncommon compared to its use in other malignancies [8,9].Thus, its indications in gastric cancers are not yet sufficiently evident [10].
In the studies previously performed, FDG-PET was reported to be successful in identifying the primary tumor in EGC in 25.9-63% of the cases and in AGC in 83-98% of the cases [4,[8][9][10].In our study, EGC and AGC were identified in 77% and 95% of the cases, respectively.Three cases consisting of one of each of the signet ring cell type, mucinous adenocancer, and stromal variant were not demonstrated with FDG-PET.The low diagnostic rate in these cases is similar to those in other publications [11][12][13][14].The reason for this low rate is interpreted in two ways.The first theory is metabolically inert mucus decreases the concentration of FDG in the membrane, thus decreasing the uptake by the tumor [15].Another theory is that a decreased expression of GLUT-1, an enzyme functioning as a glucose transporter in the membranes of signet ring cells and mucinous cells, results in a decreased uptake of FDG [15][16][17][18].In some studies, the median SUV value was 2.8±1.6 for EGC and 79± 8 for AGC, and this difference was found to be statistically significant [18][19][20][21].No difference was found in our study.
There are reports suggesting that the mean SUV increases with increasing tumor depth [10,22].An intermediate level positive correlation was found in this study.
The specificity of FDG-PET and CT in patients in the N1 group was found to be 100% and 95%, respectively.The specificity of FDG-PET in the N1 group in the literature is reported to be 90-97% [11,12,14].The specificity of CT is reported to be 45-89% [12,13].In our study, the specificity of FDG-PET in the N2 group was 88.4%.In the literature, the specificity of FDG-PET is reported to be 90-96%.We found the specificity of CT to be 84% in the same group, while it is reported to be 86-90% in the literature [12][13][14].In the N1 and N2 groups, the specificity of FDG-PET and CT were similar.
The sensitivity of FDG-PET in the N1 group is reported to be 17.6-56% in the literature [11,12,14].In our study, the FDG-PET sensitivity was 40% in this group.The CT sensitivity in the same group is reported to be 58-89.3% in the literature.The sensitivity of CT in this study was 40%, which is a relatively low rate.The FDG-PET sensitivity in the N2 group in the literature is reported to be 34-42.1% [12][13][14], while it was 50% in our study.The sensitivity of CT in the N2 group was observed to be 40-63.1% in the literature [12][13][14].The sensitivity of CT was 25% in our study, which is a rather low rate.The reason for this low rate is that there were only four patients in this group, and CT was diagnostic in only one case.
The N3/4 groups are evaluated together because of the low number of cases.N3 group lymph node involvement was identified in four patients by FDG-PET.Two of them were hepatoduodenal, one was mesenteric, and one was paraaortic lymph node involvement.In these patients, only that group of lymph nodes was dissected in addition to the standard procedure.After histopathologic examination, one hepatoduodenal and one mesenteric lymph node involvement were correctly interpreted by FDG-PET.D3 dissections were performed in all other cases.We attempted to show that the rest of the patients were not group N3 by observations during surgery, conventional methods, and FDG-PET.Statistically significant results were not obtained in the N3/4 groups because the number of cases was low.In two of four patients, FDG-PET changed the stage and the surgical procedure and thus increased survival.
A significant difference was not found between the mean SUV in tumors with nodal involvement (N1, N2, N3/4) and those with no nodal involvement (N0).However, the SUV was reported to increase with increased nodal involvement in some studies [11,13].The primary involvement of perigastric nodes cannot be differentiated by FDG-PET due to the low longitudinal resolution.However the specificity of FDG-PET in N1 node involvement is rather high [14].The more important node group is the N2 group because lymph node involvement in this group changes the dissection from D1 to D2, which changes the treatment plan.
The sensitivity of both examinations was demonstrated to be low in the N2 group.This result indicates that the low diagnostic yield of FDG-PET and CT, which do not identify the lymph nodes preoperatively in the N2 group, mandates additional examinations and imaging methods.Both imaging modalities have high rates of specificity.This result supports the opinion that when an N2 node is identified with FDG-PET, a D2 dissection is recommended.The same opinions are valid for N3 node involvement.
FDG-PET plays an important role in determining the magnitude of lymphadenectomies (D1-D3) or in decreasing the rate of unnecessary laparotomies.Neither modality is sufficient in the evaluation of N3/4 metastasis because of the low prevalence of N3/4 involvement.
N1 node involvement is a cornerstone in changing the surgical approach.Endoscopic mucosal resection or laparoscopic or open gastrectomy, in addition to limited lymph node dissection, is an adequate treatment method in EGC without N1 node involvement.The high specificity in the N1 group decreases the rate of unnecessary surgical resection [21][22][23].
In a patient in whom distant metastasis and ascites were not identified by preoperative CT and FDG-PET, multiple peritoneal implants and ascites were diagnosed with laparoscopy, and tumor cells were found in the cytologic samples that were taken.The case was accepted as inoperable, and chemotherapy and radiotherapy were initiated.FDG-PET identified paraaortic lymph node involvement in the same case; however, no paraaortic involvement was found in surgery.
In a case with multiple millimetric metastases in the liver and multiple intra-abdominal peritoneal implants and ascites identified preoperatively by CT, FDG-PET was not successful in identifying ascites.
Metastases in the left lobe of the liver and right adrenal gland were demonstrated by CT in another patient, and uptakes in the same metastatic regions were observed on the FDG-PET scans.
An exploratory laparotomy is invasive and expensive, although it is the gold standard method for identifying peritoneal metastases [22,23].The sensitivity and specificity of CT in identifying peritoneal metastasis and implants have been reported as 76.5% and 91.6%, respectively, while the specificity and sensitivity of FDG-PET have been reported as 98.9% and 35.3% [24,25].
Chen et al. reported that FDG-PET could not detect liver metastasis and that a CT could not detect splenic metastasis.In this report, two hematological metastases were not detected, and false negative results were obtained [11].Our results demonstrate that CT should be the first option with FDG-PET for preoperative assessment of gastric cancer among non-invasive methods to identify metastasis and peritoneal involvement.