研究目的：造血及免疫重建是异基因造血干细胞移植成功的关键。尽管已有研究在单细胞水平探讨了移植后造血干细胞和祖细胞的重建动力学，并探讨了对急性移植物抗宿主病的预测意义，但对巨核细胞重建动态及异质性的系统性解析仍不足。血小板植入水平是评价巨核谱系重建的重要指标，既往研究表明其与急性移植物抗宿主病相关，但现有研究多局限于静态血小板计数，缺乏对血小板轨迹动态变化与急性移植物抗宿主病发生关系的深入探索。基于此，本研究旨在从分子和数量层面系统解析移植后巨核细胞-血小板谱系的重建动态，阐明巨核细胞重建的分子特征以及血小板重建的数量轨迹特征，并评估其临床预后价值。

研究方法：（1）通过单细胞转录组测序技术解析后造血干细胞移植后不同时间点巨核细胞的分子重建异质性。（2）基于多中心患者队列，构建基于多元高斯分布的Two-Stage模型对移植后早期血小板进行动态轨迹聚类。（3）构建基于早期血小板轨迹的重度aGVHD多因素风险预测模型。

研究结果：

（1）首次在单细胞层面揭示移植后正常巨核细胞重建的动态异质性：本研究通过整合移植后3个关键时间点的单细胞转录组测序数据，首次在单细胞层面阐明异基因造血干细胞移植后不同时间点正常状态下巨核细胞异质性重建的动态演变规律，填补了正常重建巨核细胞时序性功能重塑的研究空白，为区分生理性与病理性重建提供了动态基准。

（2）明确巨核细胞分化的两条基因调控路径：本研究发现，移植后巨核细胞分化存在两条路径：早期免疫特征路径（“CXCR4”、“IL4R”高表达）和晚期产板特征路径（“BCL2L1”、“GATA2”高表达）。这种时序性功能转换提示巨核细胞可能在移植后生理重建过程中从免疫应激状态向稳态血小板生成的动态转变。

（3）前瞻性队列研究发现高剂量rhTPO可以促进移植后产板型巨核细胞的生成和血小板植入：本研究基于前瞻性队列研究，发现提高重组人血小板生成素（rhTPO）给药剂量可显著促进移植后巨核细胞向产板型巨核细胞分化，并加快血小板植入时间、提高血小板植入水平。同时，在血小板植入更快的患者组中，发现Ⅱ-Ⅳ°aGVHD的发生率显著降低。这提示血小板的植入时间与水平可能与aGVHD发生密切相关。

（4）移植后血小板重建轨迹异质性显著，并与aGVHD存在联系：本研究通过对1192例患者移植后外周血细胞数量进行轨迹建模分析，发现外周血细胞重建过程存在异质性。其中，血小板重建过程呈现显著异质性，其动态轨迹与中性粒细胞、淋巴细胞呈同步趋势。进一步分析发现血小板在aGVHD出现前已表现出轨迹组间的水平差异，而中性粒细胞及淋巴细胞计数变化滞后于aGVHD发生。

（5）开发血小板动态轨迹模型及在线工具：针对传统静态指标的局限性，本研究基于多中心患者队列，创新性开发Two-Stage血小板轨迹模型，利用移植后0-10天血小板连续数量将患者分为低轨迹、中轨迹、高轨迹3种血小板轨迹组。与中、高血小板轨迹组患者相比，我们发现，低轨迹组患者重度aGVHD发生率显著升高。同时，为实现临床转化应用，进一步开发了在线血小板轨迹计算器（https://fengliang.shinyapps.io/PTPC/）。

（6）构建重度aGVHD多因素综合预测模型：本研究通过整合血小板轨迹与多种临床风险因素（年龄、移植类型等），构建了移植后早期预测重度aGVHD的Cox多因素模型。并进一步将患者进行风险分组，其中高风险组患者重度aGVHD累积发生率显著高于低风险组（训练集46.4% vs. 9.5%，验证集54.5% vs. 15.9%，均p<0.001）。对该模型进行评估，其AUC为0.741，较单一指标提升14.3%，为临床早期识别重度aGVHD患者提供了更为简便可靠的工具。

研究结论：本研究通过整合单细胞转录组学、前瞻性临床队列以及动态轨迹建模，系统阐明了造血干细胞移植后巨核细胞及血小板重建的动态规律及其临床转化价值：首次从单细胞层面揭示了巨核细胞从免疫调控向产板功能的时序性切换的现象，发现高剂量rhTPO通过促进该切换加速血小板植入并降低aGVHD发生；并基于多中心队列，创新性构建基于早期血小板动态轨迹的重度aGVHD风险预测模型（AUC=0.741），为重度aGVHD早期预警提供精准工具。

Objectives: Hematopoietic and immune reconstitution is crucial for the success of allogeneic hematopoietic stem cell transplantation (allo-HSCT). Although previous studies have explored the reconstitution dynamics of hematopoietic stem and progenitor cells at the single-cell level and their predictive value for acute graft-versus-host disease (aGVHD), a systematic analysis of megakaryocyte reconstitution dynamics and heterogeneity remains insufficient. Platelet engraftment is an important indicator of megakaryocytic lineage reconstitution, and previous studies have suggested its association with aGVHD. However, existing research has mostly focused on static platelet counts, lacking an in-depth investigation of the dynamic trajectory of platelet changes and its relationship with aGVHD development. Based on this, our study aims to systematically analyze the reconstitution dynamics of the megakaryocyte-platelet lineage at both molecular and quantitative levels, elucidate the molecular characteristics of megakaryocyte reconstitution, and characterize the quantitative trajectory of platelet reconstitution while assessing its clinical prognostic value.

Methods: 1. Deciphering the molecular reconstruction heterogeneity of megakaryocytes at different time points after hematopoietic stem cell transplantation using single-cell transcriptome sequencing. 2. Based on a multi-center patient cohort, a Two-Stage model based on multivariate Gaussian distribution was constructed to perform dynamic trajectory clustering of early post-transplant platelets. 3. Constructing a multifactorial risk prediction model for severe aGVHD based on early platelet trajectories.

Results

(1) First Revealing the Dynamic Heterogeneity of Normal Megakaryocyte Reconstitution at the Single-Cell Level After Transplantation：This study, by integrating single-cell transcriptomic sequencing data from three key post-transplantation time points, for the first time elucidates the dynamic evolution of normal megakaryocyte heterogeneity reconstitution at different stages after allogeneic hematopoietic stem cell transplantation (allo-HSCT) at the single-cell level. It fills the research gap in the temporal functional remodeling of normally reconstituted megakaryocytes and provides a dynamic reference for distinguishing physiological from pathological reconstitution.

(2) Identification of Two Genetic Regulatory Pathways of Megakaryocyte Differentiation:
This study reveals that megakaryocyte differentiation after transplantation follows two distinct pathways: an early immune-featured pathway (characterized by high expression of CXCR4 and IL4R) and a late platelet-producing pathway (marked by high expression of BCL2L1 and GATA2). This sequential functional transition suggests that megakaryocytes may dynamically shift from an immune-stress state to steady-state platelet production during post-transplant physiological reconstruction.

(3) A prospective cohort study found that high-dose rhTPO promotes the generation of platelet-producing megakaryocytes and platelet engraftment after transplantation:

Based on a prospective cohort study, we found that increasing the dosage of recombinant human thrombopoietin (rhTPO) significantly promotes the differentiation of megakaryocytes into platelet-producing megakaryocytes after transplantation, accelerates platelet engraftment, and enhances platelet engraftment levels. Additionally, in the group of patients with faster platelet engraftment, the incidence of grade II-IV aGVHD was significantly lower. This suggests that platelet engraftment time and levels may be closely related to the occurrence of aGVHD.

(4) Significant Heterogeneity in Post-Transplant Platelet Reconstitution Trajectories and Their Association with aGVHD: This study conducted trajectory modeling analysis on post-transplant peripheral blood cell counts in 1,192 patients and identified heterogeneity in the reconstitution process. Notably, platelet reconstitution exhibited significant heterogeneity, with its dynamic trajectory synchronizing with those of neutrophils and lymphocytes. Further analysis revealed that platelet levels differed between trajectory groups before the onset of aGVHD, whereas changes in neutrophil and lymphocyte counts lagged behind the occurrence of aGVHD.

(5) Development of a Platelet Dynamic Trajectory Model and Online Tool: To address the limitations of traditional static indicators, this study innovatively developed a Two-Stage Platelet Trajectory Model based on a multi-center patient cohort, classifying patients into three platelet trajectory groups—low, medium, and high—based on continuous platelet counts from days 0 to 10 post-transplant. Compared to patients in the medium and high platelet trajectory groups, we found that patients in the low trajectory group had a significantly higher incidence of severe aGVHD. To facilitate clinical translation, we further developed an online platelet trajectory calculator (https://fengliang.shinyapps.io/PTPC/).

(6) Construction of a Multivariate Predictive Model for Severe aGVHD: By integrating platelet trajectories with multiple clinical risk factors (such as age and transplant type), we constructed a Cox multivariate model for the early post-transplant prediction of severe aGVHD. Patients were further stratified into risk groups, where those in the high-risk group exhibited a significantly higher cumulative incidence of severe aGVHD compared to the low-risk group (training cohort: 46.4% vs. 9.5%; validation cohort: 54.5% vs. 15.9%, both p<0.001). Model evaluation showed an AUC of 0.741, representing a 14.3% improvement over single-indicator models, providing a more practical and reliable tool for the early clinical identification of severe aGVHD.

Conclusions: By integrating single-cell transcriptomics, a prospective clinical cohort, and dynamic trajectory modeling, this study systematically elucidates the dynamic reconstruction of megakaryocytes and platelets after hematopoietic stem cell transplantation (HSCT) and its clinical implications. For the first time, we revealed at the single-cell level the temporal transition of megakaryocytes from immune regulation to platelet production, and found that high-dose rhTPO accelerates platelet engraftment and reduces aGVHD incidence by promoting this transition. Additionally, using a multi-center patient cohort, we innovatively developed an early severe aGVHD risk prediction model based on platelet dynamic trajectories (AUC = 0.741), providing a precise tool for early severe aGVHD warning.