Scholarlink

Background <p>Evidence for adjuvant chemotherapy (ACT) in very old gastric cancer patients is limited, and inadequate lymph node (LN) examination may affect staging and treatment selection. We assessed whether LN examination modifies the association between ACT and survival in older adults.</p> Methods <p>From SEER (2010–2015), we identified patients aged ≥ 75 years who underwent resection for gastric cancer. LN examination was grouped as &lt; 15 vs. ≥ 15. Overall survival (OS) was evaluated using Cox models. Cancer-specific death was analyzed with cumulative incidence functions and Fine–Gray regression. To mitigate treatment-selection bias, we used multivariable adjustment, propensity score matching (PSM), and stabilized inverse probability of treatment weighting (IPTW). Effect modification was tested using ACT×LN interaction terms; LN count was also explored continuously with restricted cubic splines.</p> Results <p>The cohort included 825 patients; 386 were retained after PSM. ACT–outcome estimates differed by LN examination. ACT was more favorably associated with outcomes among patients with ≥ 15 nodes examined, while estimates were weaker and less consistent when &lt; 15 nodes were assessed. Interaction signals appeared in crude and some adjusted models but were attenuated, while remaining directionally consistent, in IPTW sensitivity analyses, suggesting susceptibility to residual confounding and modeling assumptions. Competing-risk analyses highlighted substantial non–cancer mortality. In a low-risk–exclusion cohort (<i>N</i> = 306), ACT remained favorably associated with outcomes, with less evidence of interaction.</p> Conclusions <p>Among resected gastric cancer patients aged ≥ 75 years, the observed association between ACT and survival differed by LN examination extent, with more favorable estimates generally seen when ≥ 15 nodes are examined.</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

Adjuvant chemotherapy and lymph node examination interact to shape survival in older adults (≥ 75 years) with resected gastric cancer: a SEER-based propensity score–matched and competing-risk analysis

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月11日

查看全文
  • Jiajun Luo,
  • Ziwei Wang

摘要

Background

Evidence for adjuvant chemotherapy (ACT) in very old gastric cancer patients is limited, and inadequate lymph node (LN) examination may affect staging and treatment selection. We assessed whether LN examination modifies the association between ACT and survival in older adults.

Methods

From SEER (2010–2015), we identified patients aged ≥ 75 years who underwent resection for gastric cancer. LN examination was grouped as < 15 vs. ≥ 15. Overall survival (OS) was evaluated using Cox models. Cancer-specific death was analyzed with cumulative incidence functions and Fine–Gray regression. To mitigate treatment-selection bias, we used multivariable adjustment, propensity score matching (PSM), and stabilized inverse probability of treatment weighting (IPTW). Effect modification was tested using ACT×LN interaction terms; LN count was also explored continuously with restricted cubic splines.

Results

The cohort included 825 patients; 386 were retained after PSM. ACT–outcome estimates differed by LN examination. ACT was more favorably associated with outcomes among patients with ≥ 15 nodes examined, while estimates were weaker and less consistent when < 15 nodes were assessed. Interaction signals appeared in crude and some adjusted models but were attenuated, while remaining directionally consistent, in IPTW sensitivity analyses, suggesting susceptibility to residual confounding and modeling assumptions. Competing-risk analyses highlighted substantial non–cancer mortality. In a low-risk–exclusion cohort (N = 306), ACT remained favorably associated with outcomes, with less evidence of interaction.

Conclusions

Among resected gastric cancer patients aged ≥ 75 years, the observed association between ACT and survival differed by LN examination extent, with more favorable estimates generally seen when ≥ 15 nodes are examined.