[Establishment of Risk Evaluation Model of Peritoneal Metastasis in Gastric Cancer and Its Predictive Value].
Junjie Zhao,Rongjian Zhou,Qi Zhang,Ping Shu,Haojie Li,Xuefei Wang,Zhenbin Shen,Fenglin Liu,Weidong Chen,Jing Qin,Yihong Sun
DOI: https://doi.org/10.3760/cma.j.issn.1671-0274.2017.01.010
2017-01-01
Abstract:OBJECTIVE:To establish an evaluation model of peritoneal metastasis in gastric cancer, and to assess its clinical significance.METHODS:Clinical and pathologic data of the consecutive cases of gastric cancer admitted between April 2015 and December 2015 in Department of General Surgery, Zhongshan Hospital of Fudan University were analyzed retrospectively. A total of 710 patients were enrolled in the study after 18 patients with other distant metastasis were excluded. The correlations between peritoneal metastasis and different factors were studied through univariate (Pearson's test or Fisher's exact test) and multivariate analyses (Binary Logistic regression). Independent predictable factors for peritoneal metastasis were combined to establish a risk evaluation model (nomogram). The nomogram was created with R software using the 'rms' package. In the nomogram, each factor had different scores, and every patient could have a total score by adding all the scores of each factor. A higher total score represented higher risk of peritoneal metastasis. Receiver operating characteristic (ROC) curve analysis was used to compare the sensitivity and specificity of the established nomogram. Delong. Delong. Clarke-Pearson test was used to compare the difference of the area under the curve (AUC). The cut-off value was determined by the AUC, when the ROC curve had the biggest AUC, the model had the best sensitivity and specificity.RESULTS:Among 710 patients, 47 patients had peritoneal metastasis (6.6%), including 30 male (30/506, 5.9%) and 17 female (17/204, 8.3%); 31 were ≥ 60 years old (31/429, 7.2%); 38 had tumor ≥ 3 cm(38/461, 8.2%). Lauren classification indicated that 2 patients were intestinal type(2/245, 0.8%), 8 patients were mixed type(8/208, 3.8%), 11 patients were diffuse type(11/142, 7.7%), and others had no associated data. CA19-9 of 13 patients was ≥ 37 kU/L(13/61, 21.3%); CA125 of 11 patients was ≥ 35 kU/L(11/36, 30.6%); CA72-4 of 11 patients was ≥ 10 kU/L(11/39, 28.2%). Neutrophil/lymphocyte ratio (NLR) of 26 patients was ≥ 2.37(26/231, 11.3%). Multivariate analysis showed that Lauren classification (HR=8.95, 95%CI:1.32-60.59, P=0.025), CA125(HR=17.45, 95%CI:5.54-54.89, P=0.001), CA72-4(HR=20.06, 95%CI:5.05-79.68, P=0.001), and NLR (HR=4.16, 95%CI:1.17-14.75, P=0.032) were independent risk factors of peritoneal metastasis in gastric cancer. In the nomogram, the highest score was 241, including diffuse or mixed Lauren classification (54 score), CA125 ≥ 35 kU/L (66 score), CA72-4 ≥ 10 kU/L (100 score), and NLR ≥ 2.37 (21 score), which represented a highest risk of peritoneal metastasis (more than 90%). The AUC of nomogram was 0.912, which was superior than any single variable (AUC of Lauren classification: 0.678; AUC of CA125: 0.720; AUC of CA72-4: 0.792; AUC of NLR: 0.613, all P=0.000). The total score of nomogram increased according to the TNM stage, and was highest in the peritoneal metastasis group (F=49.1, P=0.000). When the cut-off value calculated by ROC analysis was set at 140, the model could best balanced the sensitivity (0.79) and the specificity (0.87). Only 5% of patients had peritoneal metastasis when their nomogram scores were lower than 140, while 58% of patients had peritoneal metastasis when their scores were ≥ 140(χ2=69.1, P=0.000).CONCLUSION:The risk evaluation model established with Lauren classification, CA125, CA72-4 and NLR can effectively predict the risk of peritoneal metastasis in gastric cancer, and provide the reference to preoperative staging and choice of therapeutic strategy.