Scholarlink

Objective <p>To characterize the longitudinal trajectories of multi-category biomarkers and evaluate their association with 21-day all-cause mortality in critically ill burn patients with sepsis.</p> Methods <p>In this retrospective single-center cohort study, we analyzed 943 adult burn patients with sepsis, defined per Sepsis-3.0 criteria. Serial measurements of 15 biomarkers across nutritional, immunoglobulin, lymphocyte subset, inflammatory, and other categories were collected over 21&#xa0;days. We employed linear mixed-effects models (LME) to compare trajectories between survivors and non-survivors, Cox regression to assess associations with mortality, time-dependent ROC to evaluate predictive performance, and k-means clustering to identify patient phenotypes based on integrated ALB, IL-6, and IgG trajectories.</p> Results <p>The 21-day mortality was 17.92%. LME revealed significantly different trajectories for 11 biomarkers between survivors and non-survivors (<i>P</i> &lt; 0.05). Univariate Cox analysis identified multiple significant biomarkers, with transferrin (HR = 0.985, <i>P</i> = 6.84 × 10⁻<sup>11</sup>) and IgM (HR = 0.284, <i>P</i> = 1.24 × 10⁻<sup>5</sup>) as strong protective factors, and mitochondrial DNA (HR = 1.002, <i>P</i> = 1.89 × 10⁻⁹) as a risk factor. In multivariate analysis, only the Burn Index remained an independent risk factor (HR = 1.066, <i>P</i> &lt; 0.001). Time-dependent ROC showed peak predictive accuracy at Day 7 (albumin AUC = 0.729). Clustering identified three distinct phenotypes—“Rapid Recovery” (mortality 5.2%), “Persistent Inflammatory &amp; Catabolic” (mortality 38.0%), and “Intermediate” (mortality 18.7%; <i>P</i> &lt; 0.001)—with starkly different biomarker trends and clinical profiles.</p> Conclusions <p>The dynamic patterns of multi-category biomarkers are strongly associated with short-term survival in burn sepsis. While burn severity is a dominant baseline risk factor, longitudinal trajectory analysis captures the essence of the host’s recovery or failure, effectively stratifying patients into prognostically distinct subgroups. This trajectory-based phenotyping highlights the potential of monitoring the host response over time to improve risk assessment and guide personalized management.</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

Longitudinal biomarker trajectories and their prognostic utility for 21-day mortality in burn patients with sepsis: a retrospective cohort study

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月21日

查看全文
  • Jiaqi Lou,
  • Hong Kong,
  • Jiliang Li,
  • Ziyi Xiang,
  • Xiaoyu Zhu,
  • Shengyong Cui,
  • Neng Huang,
  • Sida Xu,
  • Guoying Jin,
  • Xin Le,
  • Youfen Fan

摘要

Objective

To characterize the longitudinal trajectories of multi-category biomarkers and evaluate their association with 21-day all-cause mortality in critically ill burn patients with sepsis.

Methods

In this retrospective single-center cohort study, we analyzed 943 adult burn patients with sepsis, defined per Sepsis-3.0 criteria. Serial measurements of 15 biomarkers across nutritional, immunoglobulin, lymphocyte subset, inflammatory, and other categories were collected over 21 days. We employed linear mixed-effects models (LME) to compare trajectories between survivors and non-survivors, Cox regression to assess associations with mortality, time-dependent ROC to evaluate predictive performance, and k-means clustering to identify patient phenotypes based on integrated ALB, IL-6, and IgG trajectories.

Results

The 21-day mortality was 17.92%. LME revealed significantly different trajectories for 11 biomarkers between survivors and non-survivors (P < 0.05). Univariate Cox analysis identified multiple significant biomarkers, with transferrin (HR = 0.985, P = 6.84 × 10⁻11) and IgM (HR = 0.284, P = 1.24 × 10⁻5) as strong protective factors, and mitochondrial DNA (HR = 1.002, P = 1.89 × 10⁻⁹) as a risk factor. In multivariate analysis, only the Burn Index remained an independent risk factor (HR = 1.066, P < 0.001). Time-dependent ROC showed peak predictive accuracy at Day 7 (albumin AUC = 0.729). Clustering identified three distinct phenotypes—“Rapid Recovery” (mortality 5.2%), “Persistent Inflammatory & Catabolic” (mortality 38.0%), and “Intermediate” (mortality 18.7%; P < 0.001)—with starkly different biomarker trends and clinical profiles.

Conclusions

The dynamic patterns of multi-category biomarkers are strongly associated with short-term survival in burn sepsis. While burn severity is a dominant baseline risk factor, longitudinal trajectory analysis captures the essence of the host’s recovery or failure, effectively stratifying patients into prognostically distinct subgroups. This trajectory-based phenotyping highlights the potential of monitoring the host response over time to improve risk assessment and guide personalized management.