Prognostic impact and reasons for variability by tumor location in gastric cancer

Abstract

Background: Gastric cancer (GC) is a highly prevalent gastrointestinal tract tumor. Several trials have demonstrated that the location of GC can affect patient prognosis. However, the factors determining tumor location remain unclear.

Aim: To investigate the tumor location of patients, we went on to study the influencing factors that lead to changes in the location of GC.

Methods: A retrospective evaluation was carried out on 3287 patients who underwent gastrectomy for GC in Zhejiang Cancer Hospital. The patients were followed up post-diagnosis and post-gastrectomy. The clinicopathological variables and overall survival of the patients were recorded. By analyzing the location of GC, the tumor location was divided into four categories: "Upper", "middle", "lower", and "total". Statistical software was utilized to analyze the relationship of each variable with the location of GC.

Results: A total of 3287 patients were included in this study. The clinicopathological indices of gender, age, serum levels of carcinoembryonic antigen (CEA), carbohydrate antigen (CA19-9) and CA72-4 levels, were significantly associated with tumor location in patients with GC. In addition, there was a strong correlation between GC location and the prognosis of postoperative patients. Specifically, patients with "lower" and "middle" GC demonstrated a better prognosis than those with tumors in other categories.

Conclusion: The five clinicopathological indices of gender, age, CEA, CA19-9 and CA72-4 levels exhibit varying degrees of influence on the tumor location. The tumor location correlates with patient prognosis following surgery.

Keywords: Clinicopathologic characteristics; Gastric cancer; Overall survival; Tumor location; Tumor marker.

Figure 1

Flow chart of the data selection and processing steps. AFP: Alpha-fetoprotein; CEA: Carcinoembryonic antigen; CA: Carbohydrate antigen; BMI: Body mass index; VIMP: Variable importance.

Figure 2

Forest plot of factors influencing survival in patients with gastric cancer. The forest map displays hazard ratios (HRs) and 95% confidence intervals (95%CIs) for various factors that influence tumor prognosis according to multiple Cox regression analyses. Each dot represents the HR for a specific factor, the horizontal line represents the 95%CI, and the vertical blank line represents the reference line (HR = 1.0), indicating no impact. Factors with good HR and CI on the right side of the line were significant predictors of poor prognosis. Values in parentheses are the highest normal levels for that tumor marker. CEA: Carcinoembryonic antigen; AFP: Alpha-fetoprotein; CA: Carbohydrate antigen; BMI: Body mass index; 95%CI: 95% confidence interval; HR: Hazard ratio; pTNM: Pathological tumor-node-metastasis.

Figure 3

Kaplan-Meier survival curves for different locations of gastric cancer, where different colors represent different locations. Red: Lower 1/3; green: Middle 1/3; blue: Total; purple: Upper 1/3). The solid line represents the Kaplan-Meier estimates for each group. The shaded area around the survival curve represents the 95% confidence interval, representing the range of fluctuations in the true survival estimate. A P value less than 0.05 was considered to indicate statistical significance.

Figure 4

Variable importance of the predictors for the predictive models for gastric cancer location. A: The minimum depth method was used to calculate the influencing factors related to the location of gastric cancer (GC). The vertical dashed line represents the minimum depth threshold; B: The variable importance method was used to calculate the influencing factors related to the location of GC. The green dots indicate strong correlations among impact factors, and the orange dots indicate weak correlations among impact factors. CEA: Carcinoembryonic antigen; AFP: Alpha-fetoprotein; CA: Carbohydrate antigen; BMI: Body mass index; pTNM: Pathological tumor-node-metastasis.

Figure 5

Random forest plot of influencing factors related to gastric cancer location. According to the random forest model, the importance of each parameter to the degree of influence on the location of gastric cancer was calculated, and the histogram was drawn in order of importance. CEA: Carcinoembryonic antigen; CA: Carbohydrate antigen; BMI: Body mass index; pTNM: Pathological tumor-node-metastasis.