Scholarlink

<p>The prevailing fixed-hierarchy approach to donor selection for allogeneic hematopoietic cell transplantation (HCT) fails to capture critical interactions between donor age and type. We addressed this by analyzing 1713 adult recipients of matched related (MRD), matched unrelated (MUD), or haploidentical donors, receiving post-transplantation cyclophosphamide, using both regularized-Cox and XGBoost machine learning models. Our analysis revealed that the overall, independent association of donor age with survival was modest; however, its importance became pivotal within specific, context-dependent trade-offs. For example, while age-matched MRD and MUD were equivalent, a younger MUD was superior to an older MRD for elderly recipients. The survival advantage of MUD over haploidentical donors was consistent and magnified with increasing recipient age. We identified a non-linear relationship between baseline risk and the benefit of a matched versus a haploidentical donor; with the gain being maximal for intermediate-risk patients but attenuated in the highest-risk stratum. Our framework quantifies the gap that persists even between HLA-favorable haploidentical donors and matched donors. These findings provide a quantitative framework –the principles of which we have made explorable via an interactive web application –for a personalized, risk-adapted approach to donor selection. The clinical utility of this model must be confirmed in future validation studies.</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

Modeling donor age and type trade-offs for personalized selection in allogeneic hematopoietic cell transplantation with post-transplant cyclophosphamide prophylaxis

  • 期刊论文

  • 发表日期：

    2026年1月12日

查看全文
  • Rohtesh S. Mehta,
  • Yosra M. Aljawai,
  • Partow Kebriaei,
  • Gabriela Rondon,
  • Warren Fingrut,
  • Portia Smallbone,
  • Betul Oran,
  • Amanda Olson,
  • Richard E. Champlin,
  • Elizabeth J. Shpall

摘要

The prevailing fixed-hierarchy approach to donor selection for allogeneic hematopoietic cell transplantation (HCT) fails to capture critical interactions between donor age and type. We addressed this by analyzing 1713 adult recipients of matched related (MRD), matched unrelated (MUD), or haploidentical donors, receiving post-transplantation cyclophosphamide, using both regularized-Cox and XGBoost machine learning models. Our analysis revealed that the overall, independent association of donor age with survival was modest; however, its importance became pivotal within specific, context-dependent trade-offs. For example, while age-matched MRD and MUD were equivalent, a younger MUD was superior to an older MRD for elderly recipients. The survival advantage of MUD over haploidentical donors was consistent and magnified with increasing recipient age. We identified a non-linear relationship between baseline risk and the benefit of a matched versus a haploidentical donor; with the gain being maximal for intermediate-risk patients but attenuated in the highest-risk stratum. Our framework quantifies the gap that persists even between HLA-favorable haploidentical donors and matched donors. These findings provide a quantitative framework –the principles of which we have made explorable via an interactive web application –for a personalized, risk-adapted approach to donor selection. The clinical utility of this model must be confirmed in future validation studies.