[Gastric signet ring cell adenocarcinoma: A distinct entity]

Predicting Overall Survival in Patients with Nonmetastatic Gastric Signet Ring Cell Carcinoma: A Machine Learning Approach

Background and Aims

Accurate prediction is essential for the survival of patients with nonmetastatic gastric signet ring cell carcinoma (GSRC) and medical decision-making. Current models rely on prespecified variables, limiting their performance and not being suitable for individual patients. Our study is aimed at developing a more precise model for predicting 1-, 3-, and 5-year overall survival (OS) in patients with nonmetastatic GSRC based on a machine learning approach.

Methods

We selected 2127 GSRC patients diagnosed from 2004 to 2014 from the Surveillance, Epidemiology, and End Results (SEER) database and then randomly partitioned them into a training and validation cohort. We compared the performance of several machine learning-based models and finally chose the eXtreme gradient boosting (XGBoost) model as the optimal method to predict the OS in patients with nonmetastatic GSRC. The model was assessed using the receiver operating characteristic curve (ROC).

Results

In the training cohort, for predicting OS rates at 1-, 3-, and 5-year, the AUCs of the XGBoost model were 0.842, 0.831, and 0.838, respectively, while in the testing cohort, the AUCs of 1-, 3-, and 5-year OS rates were 0.749, 0.823, and 0.829, respectively. Besides, the XGBoost model also performed better when compared with the American Joint Committee on Cancer (AJCC) stage. The performance for this model was stably maintained when stratified by age and ethnicity.

Conclusion

The XGBoost-based model accurately predicts the 1-, 3-, and 5-year OS in patients with nonmetastatic GSRC. Machine learning is a promising way to predict the survival outcomes of tumor patients.

Comparison between signet-ring cell carcinoma and non-signet-ring cell carcinoma of the stomach: clinicopathological parameters, epidemiological data, outcome, and prognosis-a cohort study of 123 patients from a non-endemic country

Background

Signet-ring cell carcinoma of the stomach (SRCC) is a particular gastric cancer entity. Its incidence is increasing. Its diagnosis is pathological; it corresponds to adenocarcinoma with a majority of signet-ring cells component (> 50%).

These histological features give it its aggressiveness characteristics. This has repercussions on the prognostic level and implications for the alternatives of therapy, especially since some authors suggest a potential chemoresistance.

This survey aimed to identify the epidemiological, pathological, therapeutic, and prognostic characteristics of SRCC as a separate disease entity.

Methods

This was a retrospective study of 123 patients admitted for gastric adenocarcinoma to Habib Thameur Hospital in Tunis over 11 years from January 2006 to December 2016. A comparative study was performed between 2 groups: the SRCC group with 62 patients and the non-SRCC (non-signet-ring cell carcinoma of the stomach) with 61 patients.

Results

The prevalence of SRCC in our series was 50%. SRCC affected significantly younger patients (55 vs 62 years; p = 0.004). The infiltrative character was more common in SRCC tumors (30.6 vs 14.8%; p = 0.060), whereas the budding character was more often noted in non-SRCC tumors (78.7 vs 58.1%; p = 0.039). There was no significant difference in tumor localization between both groups. Linitis plastica was noted in 14 patients with SRCC against a single patient with non-SRCC (p = 0.001). The tumor size was more important in the non-SRCC group (6.84 vs 6.39 cm; p = 0.551). Peritoneal carcinomatosis was noted in 4.3% of cases in the SRCC group versus 2.2% of cases in the NSRCC group (p = 0.570). Total gastrectomy was more often performed in the SRCC group (87 vs 56%; p = 0.001). Resection was more often curative in the non-SRCC group (84.4 vs 78.3%; p = 0.063). Postoperative chemotherapy was more commonly indicated in the SRCC group (67.4 vs 53.3%; p = 0.339). Tumor recurrence was more common in the non-SRCC group (35.7 vs 32%; p = 0.776). The most common type of recurrence was peritoneal carcinomatosis in the SRCC group (62.5%) and hepatic metastasis in the non-SRCC group (60%; p = 0.096). The overall 5-year survival in the SRCC group was lower than in the non-SRCC group, with no statistically significant difference (47.1 vs 51.5%; p = 0.715). The overall survival was more important for SRCC in early cancer (100 vs 80%; p = 0.408), whereas it was higher for non-SRCC in advanced cancer (48.1 vs 41.9%; p = 0.635).

Conclusion

Apart from its epidemiological and pathological features, SRCC seems to have a worse prognosis. Indeed, it is diagnosed at a more advanced stage and has a worse prognosis in advanced cancer than non-SRCC. It is therefore to be considered as a particular entity of gastric adenocarcinoma requiring a specific therapeutic protocol where the place of chemotherapy remains to be more investigated.

Development of prognostic predictive model with neutrophil-lymphocyte ratio (NLR) in patients with gastric signet ring carcinoma

The risk factors have not been well-defined for prognosis in gastric signet ring cell carcinoma (GSRC) patients. This study is designed to prognosticate survival in GSRC patients by establishing and verifying a predictive model with neutrophil-lymphocyte ratio (NLR).A total of 147 GSRC patients from Department of Surgical Oncology, Neimenggu Baogang Hospital, Inner Mongolia Medical University were retrospectively reviewed. A predictive model was established using Cox proportional hazards. The performance of the model was evaluated by ROC curves.In present study, we found that overall survival (OS) (P < .001, Fig. 1A) and tumor recurrence rate (P = .036, Fig. 1B) in the NLR ≤ 2.8 group were significantly better than those in the NLR > 2.8 group. These results showed that NLR ≤ 2.8 was significant prognostic factor related with both OS and tumor recurrence in patients with GSRC. After adjusting for competing risk factors, NLR ≤ 2.8 (hazard ratio [HR]: 2.625, 95% confidence interval [CI]: 1.505-5.3166, P = .003), tumor size (HR: 3.024, 95% CI: 1.521-4.186, P = .005), and tumor metastasis (HR: 3.303, 95% CI: 1.25-4.525, P = .012) remained independent predictors of tumor recurrence rate and OS. Our results showed that comparing with the model without NLR (area under ROC curve: 0.798), the model with NLR (area under ROC curve: 0.826) had significant better predictive power than the model without NLR, which further confirmed the value of NLR in predicting prognosis of patients with GSRC.In conclusion, a high NLR value independently predicts poor survival in patients with GSRC after surgery. The NLR may help oncologists evaluate outcomes of patients received surgical resection and chemotherapy in order to choose alternative therapies for patients with high NLR value.

Screening of differential microRNA expression in gastric signet ring cell carcinoma and gastric adenocarcinoma and target gene prediction

Gastric signet ring cell carcinoma (GSRCC) is a unique pathological type of gastric carcinoma that is extremely invasive and has a poor prognosis after diagnosis. The expression of microRNAs has been closely linked to the carcinogenesis of gastric cancer and has been considered as a powerful prognostic marker. Distinctive expression of miRNAs in GSRCC was investigated in the present study. Samples of GSRCC were compared to that of intestinal gastric adenocarcinoma using Agilent microarray technique, and two differentially expressed miRNAs were identified, hsa-miR-665 and hsa-miR‑95. qRT-PCR verification showed downregulation of both miRNAs in signet ring cell carcinoma and upregulation in gastric adenocarcinoma, which was not consistent with the results obtained by the microarray. Target gene prediction using online databases conferred two strong candidate genes, GLI2 and PLCG1. GO/KO analysis of these two genes showed close correlations with carcinogenesis and chemoresistance. It was concluded that hsa-miR-665 and hsa-miR-95 were downregulated in GSRCC but upregulated in intestinal gastric adenocarcinoma, and the relatively differential expression of the miRNAs negatively controlling their target genes could be closely related to the high invasive metastasis and chemoresistance of GSRCC.