Suppr超能文献

基于机器学习共识聚类识别的肾移植受者差异。

Differences between kidney retransplant recipients as identified by machine learning consensus clustering.

机构信息

Division of Nephrology and Hypertension, Department of Medicine, Mayo Clinic, Rochester, Minnesota, USA.

Division of Nephrology, University of Mississippi Medical Center, Jackson, Mississippi, USA.

出版信息

Clin Transplant. 2023 May;37(5):e14943. doi: 10.1111/ctr.14943. Epub 2023 Feb 27.

Abstract

BACKGROUND

Our study aimed to characterize kidney retransplant recipients using an unsupervised machine-learning approach.

METHODS

We performed consensus cluster analysis based on the recipient-, donor-, and transplant-related characteristics in 17 443 kidney retransplant recipients in the OPTN/UNOS database from 2010 to 2019. We identified each cluster's key characteristics using the standardized mean difference of >.3. We compared the posttransplant outcomes, including death-censored graft failure and patient death among the assigned clusters RESULTS: Consensus cluster analysis identified three distinct clusters of kidney retransplant recipients. Cluster 1 recipients were predominantly white and were less sensitized. They were most likely to receive a living donor kidney transplant and more likely to be preemptive (30%) or need ≤1 year of dialysis (32%). In contrast, cluster 2 recipients were the most sensitized (median PRA 95%). They were more likely to have been on dialysis >1 year, and receive a nationally allocated, low HLA mismatch, standard KDPI deceased donor kidney. Recipients in cluster 3 were more likely to be minorities (37% Black; 15% Hispanic). They were moderately sensitized with a median PRA of 87% and were also most likely to have been on dialysis >1 year. They received locally allocated high HLA mismatch kidneys from standard KDPI deceased donors. Thymoglobulin was the most commonly used induction agent for all three clusters. Cluster 1 had the most favorable patient and graft survival, while cluster 3 had the worst patient and graft survival.

CONCLUSION

The use of an unsupervised machine learning approach characterized kidney retransplant recipients into three clinically distinct clusters with differing posttransplant outcomes. Recipients with moderate allosensitization, such as those represented in cluster 3, are perhaps more disadvantaged in the kidney retransplantation process. Potential opportunities for improvement specific to these re-transplant recipients include working to improve opportunities to improve access to living donor kidney transplantation, living donor paired exchange and identifying strategies for better HLA matching.

摘要

背景

本研究旨在采用无监督机器学习方法对肾再移植受者进行特征描述。

方法

我们对 2010 年至 2019 年 OPTN/UNOS 数据库中 17443 例肾再移植受者的受者、供者和移植相关特征进行了共识聚类分析。我们使用>.3 的标准化均数差来确定每个聚类的关键特征。我们比较了分配给各聚类的受者的移植后结局,包括死亡风险调整移植物失功和患者死亡。

结果

共识聚类分析确定了肾再移植受者的三个不同聚类。聚类 1 受者主要为白人,致敏程度较低。他们最有可能接受活体供者肾移植,更有可能是预先指定(30%)或需要<1 年透析(32%)。相比之下,聚类 2 受者致敏程度最高(中位 PRA95%)。他们更有可能已透析>1 年,并接受国家分配的低 HLA 错配、标准 KDPI 死亡供者肾。聚类 3 受者更有可能为少数民族(37%黑人;15%西班牙裔)。他们致敏程度中等,中位 PRA 为 87%,也最有可能已透析>1 年。他们接受来自标准 KDPI 死亡供者的局部分配的 HLA 错配高的肾。对于所有三个聚类,他克莫司都是最常用的诱导剂。聚类 1 的患者和移植物存活率最高,而聚类 3 的患者和移植物存活率最差。

结论

采用无监督机器学习方法可将肾再移植受者分为具有不同移植后结局的三个临床显著聚类。具有中度同种异体致敏的受者,如聚类 3 中的受者,在肾再移植过程中可能处于更为不利的地位。针对这些再移植受者,可能有机会改善其获得活体供者肾移植、活体供者配对交换的机会,并寻找更好的 HLA 匹配策略。

文献检索

告别复杂PubMed语法,用中文像聊天一样搜索,搜遍4000万医学文献。AI智能推荐,让科研检索更轻松。

立即免费搜索

文件翻译

保留排版,准确专业,支持PDF/Word/PPT等文件格式,支持 12+语言互译。

免费翻译文档

深度研究

AI帮你快速写综述,25分钟生成高质量综述,智能提取关键信息,辅助科研写作。

立即免费体验