Scholarlink

Background <p>This study aimed to longitudinally investigate the evolution and prognosis of radiologic progression patterns (RPPs) in glioblastoma (GBM) using multi-state model (MSM).</p> Methods <p>A retrospective analysis of 119 GBM patients with confirmed progression identified four RPPs: complete (cT1, T2-circumscribed, T2-diffuse), classic-T1, non-responder, and non-local. About 11 genes were analyzed on 69 patients. Prognostic and molecular differences among RPPs were compared. A unidirectional MSM with 6 states (“surgery”, “complete”, “classic-T1”, “non-responder”, “non-local”, “death”) and 14 transitions was constructed to systematically analyze the trajectories and outcomes of various progression patterns in GBM.</p> Results <p>Significant differences in first progression-free survival and overall survival were observed among RPPs (<i>P</i> &lt; 0.001), while key gene alterations showed no significant differences. In the 6&#xa0;months post-surgery, cumulative risk for non-responder increased, with state probabilities peaking at 5&#xa0;months (20%). Between 6 and 24&#xa0;months, risks for classic-T1 and complete rose, with peak state probabilities at 10–12&#xa0;months (12.7%) for classic-T1 and 16&#xa0;months (8.7%) for complete. Non-local progression had a lower risk, with state probabilities peaking at 9–10&#xa0;months (4.7%). The cumulative mortality risk rose rapidly after progression, sequentially associated with non-local, non-responder, classic-T1, and complete. Classic-T1 progression linked to non-rim enhancement, non-response to age and subtotal resection or absence of radiotherapy, and non-local to male and multiple lesions at diagnosis.</p> Conclusion <p>RPPs could stratified treatment efficacy and prognosis in GBM, with each RPP demonstrating specific temporal risk profiles and outcomes, offering valuable insights into 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 analysis of radiologic progression patterns in glioblastoma: investigating prognosis using a multi-state model

  • 期刊论文

  • 发表日期：

    2025年3月31日

查看全文
  • Hong Guo,
  • Xiaoguang Li,
  • Haipeng Tong,
  • Jingqin Fang,
  • Xuesong Du,
  • Chao Cong,
  • Yu Guo,
  • Tian Xie,
  • Qisheng Ran,
  • Hang Shi,
  • Suihan Liu,
  • Xiao Chen,
  • Weiguo Zhang

摘要

Background

This study aimed to longitudinally investigate the evolution and prognosis of radiologic progression patterns (RPPs) in glioblastoma (GBM) using multi-state model (MSM).

Methods

A retrospective analysis of 119 GBM patients with confirmed progression identified four RPPs: complete (cT1, T2-circumscribed, T2-diffuse), classic-T1, non-responder, and non-local. About 11 genes were analyzed on 69 patients. Prognostic and molecular differences among RPPs were compared. A unidirectional MSM with 6 states (“surgery”, “complete”, “classic-T1”, “non-responder”, “non-local”, “death”) and 14 transitions was constructed to systematically analyze the trajectories and outcomes of various progression patterns in GBM.

Results

Significant differences in first progression-free survival and overall survival were observed among RPPs (P < 0.001), while key gene alterations showed no significant differences. In the 6 months post-surgery, cumulative risk for non-responder increased, with state probabilities peaking at 5 months (20%). Between 6 and 24 months, risks for classic-T1 and complete rose, with peak state probabilities at 10–12 months (12.7%) for classic-T1 and 16 months (8.7%) for complete. Non-local progression had a lower risk, with state probabilities peaking at 9–10 months (4.7%). The cumulative mortality risk rose rapidly after progression, sequentially associated with non-local, non-responder, classic-T1, and complete. Classic-T1 progression linked to non-rim enhancement, non-response to age and subtotal resection or absence of radiotherapy, and non-local to male and multiple lesions at diagnosis.

Conclusion

RPPs could stratified treatment efficacy and prognosis in GBM, with each RPP demonstrating specific temporal risk profiles and outcomes, offering valuable insights into personalized management.