Determining prognostic factors for gastric cancer using the regression tree method

Area of Science:

• Oncology
• Biostatistics
• Surgical Oncology

Background:

• The regression tree method, a statistical technique, has been underutilized in prognostic analysis.
• Gastric cancer prognosis requires robust analytical methods for effective treatment planning.

Purpose of the Study:

• To investigate prognostic factors for gastric cancer using the regression tree method.
• To identify key predictors of patient outcomes in serosa-negative gastric cancer.
• To explore the utility of regression trees in visualizing complex prognostic data.

Main Methods:

• A cohort of 555 patients with serosa-negative gastric cancer, post-curative resection, was analyzed.
• The regression tree method was applied to data from a randomized controlled trial (JCOG 8801 study).
• Prognostic factors were identified and their impact on survival was assessed.

Main Results:

• Lymph node metastasis, particularly the extent of lymphatic spread, was the primary prognostic factor.
• Age, tumor size, depth of invasion, and dose intensity significantly influenced prognosis.
• The regression tree analysis yielded nine terminal nodes and four distinct prognostic clusters, with a 5-year survival rate of 0.986 in the best-performing cluster.

Conclusions:

• Lymph node metastasis is the most critical prognostic factor in serosa-negative gastric cancer.
• The regression tree method offers a visual and interpretable approach to prognostic factor analysis.
• This method facilitates the identification of patient subgroups with varying prognoses, supporting individualized treatment strategies.