Scholarlink

Background <p>The Lauren classification (intestinal versus diffuse type) is pivotal for prognosing gastric cancer and guiding treatment selection. This study developed an integrated preoperative prediction model that merges preoperatively available clinicopathological characteristics with spectral CT parameters to distinguish Lauren subtypes and predict disease-free survival (DFS).</p> Methods <p>This single-centre, retrospective diagnostic-accuracy study included 224 chemotherapy-treated gastric cancer patients with histopathologically confirmed intestinal or diffuse-type Lauren classification and two-year disease-free survival outcomes. Portal-venous phase spectral computed tomography (CT) parameters (CT₇₀ₖₑ<sub>v</sub>, iodine concentration, and normalized iodine concentration) and clinicopathological variables (tumor differentiation and Borrmann type) were assessed. Independent predictors were identified using multivariate logistic regression, and three models were constructed: clinicopathological, spectral CT, and combined. Diagnostic performance was evaluated using receiver operating characteristic (ROC) analysis, calibration curves, decision curve analysis, and 1000-bootstrap internal validation. Disease-free survival was further analysed using Kaplan–Meier curves, the log-rank test, and Cox proportional hazards regression.</p> Results <p>Multivariate analysis identified poor differentiation (OR = 9.80, P &lt; &lt; 0.001) and Borrmann type III/IV (OR = 2.65, <i>P</i> = 0.004) as clinicopathological predictors, while portal-venous phase CT at 70&#xa0;keV (OR = 1.34, P &lt; &lt; 0.001), iodine concentration (OR = 2.08, <i>P</i> = 0.001), and normalized iodine concentration (OR = 2.09, P &lt; &lt; 0.001) were significant spectral CT predictors of diffuse-type histology. The combined model demonstrated an area under the curve (AUC) of 0.901 (95% CI: 0.859–0.944) for Lauren classification, outperforming both the pathological model (AUC = 0.810) and the spectral CT model (AUC = 0.861). The optimism-corrected AUC was 0.901 (95% CI: 0.857–0.943) based on 1000-bootstrap validation. For disease-free survival (DFS) prediction, the model achieved an AUC of 0.896 (95% CI: 0.851–0.941) at a cut-off value of 0.646, with an optimism-corrected AUC of 0.895 (95% CI: 0.850–0.939). Calibration curves and decision curve analysis indicated satisfactory agreement and a superior net benefit. Kaplan-Meier analysis revealed significant separation between high-risk and low-risk groups (2-year DFS rates: 7.2% versus 84.4%; log-rank chi2 = 158.88, P &lt; &lt; 0.001), and Cox regression analysis confirmed a substantially increased recurrence risk in the high-risk group (HR = 10.97, 95% CI: 6.68–17.99, P &lt; &lt; 0.001).</p> Conclusion <p>The integrated model provides accurate preoperative Lauren classification and robust DFS prediction, serving as a multimodal tool for gastric cancer risk stratification and treatment planning.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Lauren classification-based combined model integrating clinicopathological and spectral CT features for disease-free survival prediction in gastric cancer

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月3日

查看全文
  • Haibo You,
  • Tiezhu Ren,
  • Min Xu,
  • Qianqian Chen,
  • Long Ma,
  • Yue Peng,
  • Chenyang Zhang,
  • Xinyu Liu,
  • Wenjuan Zhang

摘要

Background

The Lauren classification (intestinal versus diffuse type) is pivotal for prognosing gastric cancer and guiding treatment selection. This study developed an integrated preoperative prediction model that merges preoperatively available clinicopathological characteristics with spectral CT parameters to distinguish Lauren subtypes and predict disease-free survival (DFS).

Methods

This single-centre, retrospective diagnostic-accuracy study included 224 chemotherapy-treated gastric cancer patients with histopathologically confirmed intestinal or diffuse-type Lauren classification and two-year disease-free survival outcomes. Portal-venous phase spectral computed tomography (CT) parameters (CT₇₀ₖₑv, iodine concentration, and normalized iodine concentration) and clinicopathological variables (tumor differentiation and Borrmann type) were assessed. Independent predictors were identified using multivariate logistic regression, and three models were constructed: clinicopathological, spectral CT, and combined. Diagnostic performance was evaluated using receiver operating characteristic (ROC) analysis, calibration curves, decision curve analysis, and 1000-bootstrap internal validation. Disease-free survival was further analysed using Kaplan–Meier curves, the log-rank test, and Cox proportional hazards regression.

Results

Multivariate analysis identified poor differentiation (OR = 9.80, P < < 0.001) and Borrmann type III/IV (OR = 2.65, P = 0.004) as clinicopathological predictors, while portal-venous phase CT at 70 keV (OR = 1.34, P < < 0.001), iodine concentration (OR = 2.08, P = 0.001), and normalized iodine concentration (OR = 2.09, P < < 0.001) were significant spectral CT predictors of diffuse-type histology. The combined model demonstrated an area under the curve (AUC) of 0.901 (95% CI: 0.859–0.944) for Lauren classification, outperforming both the pathological model (AUC = 0.810) and the spectral CT model (AUC = 0.861). The optimism-corrected AUC was 0.901 (95% CI: 0.857–0.943) based on 1000-bootstrap validation. For disease-free survival (DFS) prediction, the model achieved an AUC of 0.896 (95% CI: 0.851–0.941) at a cut-off value of 0.646, with an optimism-corrected AUC of 0.895 (95% CI: 0.850–0.939). Calibration curves and decision curve analysis indicated satisfactory agreement and a superior net benefit. Kaplan-Meier analysis revealed significant separation between high-risk and low-risk groups (2-year DFS rates: 7.2% versus 84.4%; log-rank chi2 = 158.88, P < < 0.001), and Cox regression analysis confirmed a substantially increased recurrence risk in the high-risk group (HR = 10.97, 95% CI: 6.68–17.99, P < < 0.001).

Conclusion

The integrated model provides accurate preoperative Lauren classification and robust DFS prediction, serving as a multimodal tool for gastric cancer risk stratification and treatment planning.