Abstract: Objective:To explore the computed tomography (CT) features of gastric cancer invasion to the pancreas and significance in the assessment of resectability of primary lesions.
Methods:The retrospective cohort study was conducted. The clinical data of 31 gastric cancer patients who were admitted to the Peking University Cancer Hospital between February 2011 and August 2016 were collected. Of 31 patients receiving CT examinations, 〖HQK〗11 who were diagnosed with suspected pancreas invasion by preoperative CT examinations but operation confirmed no invasion were allocated into the pancreas negative (PN) group, 11 who were confirmed as pancreas invasion and underwent radical gastrectomy of gastric cancer combined with pancreas resection were allocated into the pancreas invasion (PI) group, and 9 who were confirmed as pancreas invasion and had unresectable primary lesions were allocated into the pancreas invasion nonresected (PINR) group. Observation indicators: (1) morphologic type of contact surface between gastric cancer and pancreas; (2) comparison of CT findings among the 3 groups: primary lesion location, tumor thickness, Borrmann type, serosa pattern of gastric cancer, judging obvious region invaded by gastric cancer, contact or invasion site with pancreas, contact length between gastric cancer and pancreas, pattern, clarity and CT values of contact surface or peripancreas invaded and normal peripancreas; (3) treatment or followup situations. All the patients underwent radical resection and palliative resection for gastric cancer or nonoperation according to results of exploration. Telephone interview was performed to detect the survival of patients up to February 2017. Measurement data with skewed distribution were described as M (Qn), and comparisons among groups were done by the KruskalWallis test. Comparison of count data were done by the Fisher exact probability.
Results:(1) Morphologic type of contact surface between gastric cancer and pancreas: there were 4 types according to results of CT examination. Type Ⅰ: pancreas contacted with gastric cancer and there was no change in the morphology and radian of contact surface. TypeⅡ: pancreas contacted with gastric cancer and radian of contact surface became flattened or shallow depression. Type Ⅲ: contact surface showed a inserted sign or obvious depression. Type Ⅳ: pancreas didn′t contact with gastric cancer and there was increased density in fat space between pancreas and gastric cancer, with a smudge sign or strip and sheetlike opacity. Of 31 patients, type Ⅰ, Ⅱ, Ⅲ and Ⅳ were detected in 5, 10, 4 and 12 patients, respectively. (2) Comparison of CT findings among the 3 groups: nodular protrusion, spiculation and strip shape, clounding patch opacity of serosa pattern of gastric cancer were detected in 1, 6, 4 patients in the PN group and 5, 4, 2 patients in the PI group and 0, 2, 7 patients in the PINR group, respectively, with a statistically significant difference (x²=10.054, P<0.05). Two, 8 and 8 patients in the PN, PI and PINR groups had obvious tumor invasion located at a adjacent region between stomach and pancreas, with a statistically significant difference (x²=11.259, P<0.05). Contact or invasion site with pancreas located at head, body and tail of pancreas was detected in 6, 5, 0 patients in the PN group and 1, 7, 3 patients in the PI group and 5, 4, 0 patients in the PINR group, respectively, with a statistically significant difference (x²=8.390, P<0.05). Type Ⅰ, Ⅱ, Ⅲ and Ⅳof contact surface between gastric cancer and pancreas were detected in 5, 6, 0, 0 patients in the PN group and 0, 4, 4, 3 patients in the PI group and 0, 0, 0, 9 patients in the PINR group, respectively, with a statistically significant difference (x²=29.291, P<0.05). Number of patients with clear and ambiguous contact surface was 10, 1 patients in the PN group and 0, 11 patients in the PI group and 0, 9 patients in the PINR group, respectively, with a statistically significant difference (x²=26.227, P<0.05). CT values of contact surface or peripancreas invaded were -46 HU (-57 HU, -20 HU) in the PN group and -34 HU (-41 HU, -25 HU) in the PI group and -10 HU (-15 HU, -10 HU) in the PINR group, respectively, with a statistically significant difference (Z=15.306, P<0.05). CT values of normal peripancreas were -87 HU (-96 HU, -76 HU) in the PN group and -88 HU (-70 HU, -1 HU) in the PI group and -83 HU (-98 HU, -74 HU) in the PINR group, respectively, with statistically significant differences in CT values between contact surface or peripancreas invaded and normal peripancreas among the 3 groups (Z=12.581, 13.780, 7.793, P<0.05). (3) Treatment or followup situations: of 31 patients, 22 underwent radical gastrectomy and 9 underwent simplex exploration or short surgery. All the 31 patients were followed up for 6.0-71.0 months, with a median time of 13.5 months. Postoperative 1 and 2year survival rates were 82.6% and 77.1%.
Conclusions:There are significant differences in pancreatic invasion and resectability between CT features of contact surface of gastric cancer and pancreas and tumor classification. CT features include that pancreas contacts with gastric cancer in the PN group, radian of contact surface becomes flattened and with a inserted sign in the PI group, and there are increased density in fat space between pancreas and gastric cancer and a smudge sign or strip and sheetlike opacity in the PI-NR group.