Related Experiment Video
Updated: Jan 7, 2026
Enhancing Tumor Content through Tumor Macrodissection
Published on: February 12, 2022
Clinical characteristics and prognostic analysis of three hundred and nineteen cases of primary gastrointestinal
Jia-Jia Ma1, Huan Zhang1, Cui-Cui Wang1
1Pathology Center, Xinjiang Medical University Affiliated Tumor Hospital, Urumqi 830000, Xinjiang Uygur Autonomous Region, China.
Background:
Primary gastrointestinal diffuse large B-cell lymphoma (PGI-DLBCL), the most prevalent extranodal non-Hodgkin lymphoma, poses significant diagnostic and therapeutic challenges due to its non-specific symptoms and poor prognosis.
Aim:
To develop and validate a risk model for the early identification of PGI-DLBCL using Least Absolute Shrinkage and Selection-Cox regression, with the aim of guiding clinical decision-making.
Methods:
The clinical data of patients diagnosed with PGI-DLBCL at the Tumor Hospital Affiliated to Xinjiang Medical University were analyzed retrospectively from January 2010 to April 2022.
Results:
A total of 319 patients with PGI-DLBCL were included and divided into training (n = 223) and validation (n = 96) cohorts. The median age was 55 years, with 48.9% male and 51.1% female patients. Key clinical features included Eastern Cooperative Oncology Group performance status ≥ 2 (40.8%), advanced-stage disease (stage IV: 27.6%), extranodal involvement ≥ 2 sites (47%), tumor > 5 cm (46.1%), elevated beta-2 microglobulin (50.5%), elevated lactate dehydrogenase (27%), high International Prognostic Index (3-5: 69.9%), non-germinal center B-cell-like subtype (59.9%), and B symptoms (55.8%). Immunohistochemical analysis showed frequent expression of CD10 (51.1%), B-cell lymphoma 6 (53.3%), multiple myeloma oncogene 1 (40.1%), B-cell lymphoma 2 (49.2%), myelocytomatosis viral oncogene homolog (48.3%), Ki-67 (67.1%), and CD5 (42.6%); Epstein-Barr virus-encoded RNA was positive in 3.1%. Based on Least Absolute Shrinkage and Selection regression and subsequent univariate and multivariate Cox regression analyses, extranodal sites ≥ 2, B symptoms, mixed lesion type, and negative multiple myeloma oncogene 1 expression were identified as independent risk factors for PGI-DLBCL. The risk model stratified patients into high- and low-risk groups with significantly different overall survival (P < 0.05). Area under the curve values for 1-, 3-, and 5-year overall survival were 0.625, 0.663, and 0.723 in the training cohort, with consistent performance in the validation cohort. Decision curve analysis indicated favorable clinical utility.
Conclusion:
PGI-DLBCL in our cohort showed distinctive clinical features and a predominance of the non-germinal center B-cell-like subtype. Decision curve analysis confirmed the clinical applicability of our prognostic model. Although molecular biomarkers will be needed to improve predictive precision, our model offers a practical tool for early risk identification and individualized management in clinical practice.
