异基因造血干细胞移植（allo-HSCT）是当前治疗晚期慢性髓性恶性肿瘤（如骨髓纤维化、慢性髓性白血病等）的关键手段，亦是目前唯一有潜力实现长期生存甚至治愈的治疗选择。然而，移植相关的毒性风险和个体间预后差异显著，使得精准评估患者移植后结局成为临床决策中的重要挑战。在近期召开的第51届欧洲血液与骨髓移植学会年会（EBMT 2025）上，来自西班牙圣地亚哥-德孔波斯特拉大学医院的Adrian Mosquera教授，分享了《利用机器学习预测慢性髓性恶性肿瘤患者异基因造血干细胞移植后结局》的精彩内容，为解决上述问题提供了重要方向。《肿瘤瞭望-血液时讯》现场特邀Adrian Mosquera-Orgeira教授，就该话题做进一步分享。
《肿瘤瞭望-血液时讯》新药时代，您如何看待移植在慢性髓性恶性肿瘤患者中的应用和地位？
Adrian Mosquera-Orgeira教授：长期以来，对于进入加速期或更晚期的慢性髓性恶性肿瘤患者，造血干细胞移植一直是唯一的治疗选择。特别是在疾病进入终末期时，异基因造血干细胞移植几乎成为不可替代的治疗手段。目前，我们正积极探索新药在这类患者中的应用前景，例如部分患者可使用的靶向药物——莫洛替尼（Momelotinib）。问题在于：是否应该让某些特定患者接受异基因造血干细胞移植？以及何时是进行移植的最佳时机？
在这方面，我们仍有许多工作要做，尤其是在优化患者风险分层方面。这样才能为临床医生提供更科学的决策依据，帮助其判断哪些患者适合接受移植、何时进行移植更为合适。例如，新药能否改善患者状况，从而降低其接受移植的风险，并尽可能提高生存率，这些都是我们亟需深入研究的问题。
Oncology Frontier-Hematology Frontier：In the era of novel drugs, what are your views on the application and role of transplantation for patients with chronic myeloid malignancies?
Pro. Adrian Mosquera-Orgeira：For a long time the only alternative for patients with chronic myeloid malignancies who progressed to accelerated phases or more.End-stage disease was obviously allogeneic stem cell transplantation.It's obvious that there is an interest now to explore and now better the niche for new drugs for new patients as we have new targeted drugs like momelotinib for some patients.The issue is whether you decide a particular patient should be a candidate for allogeneic stem cell transplantation or not and at which point.And I think in that regard there's a lot of effort that needs to be done in order to better risk stratify patients and enable doctors making more informed decisions.About who and when should be a candidate for allogeneic stem cell transplantation.For example, how can these new drugs be used to improve and downgrade the risk of a transplant for a particular patient and therefore maximizing the likelihood of survival.
《肿瘤瞭望-血液时讯》在本次慢性髓性恶性肿瘤工作组会议（CMWP）专场，您分享了“利用机器学习预测慢性髓性恶性肿瘤患者异基因造血干细胞移植后结局”的精彩内容，请您简要介绍下研究的主要发现和重要意义？
Adrian Mosquera-Orgeira教授：这与我刚才的观点高度一致。关键在于，随着新药物的出现，我们有了更多将其个体化应用于新发患者的机会。然而，目前尚无法确定所有患者在疾病进展后是否都应接受异基因造血干细胞移植。原因在于，已有明确证据显示，某些特定患者群体的死亡率较高。因此，我们的目标是更精准地识别那些预后不佳的患者群体——无论是总生存率还是无复发生存率较低的患者。这类患者在接受异基因造血干细胞移植后，往往疗效有限，预后较差。
为实现这一目标，我们正在借助人工智能技术，确切地说是机器学习工具。我们利用EBMT注册数据库进行模型训练，该数据库汇集了来自多个欧洲国家的数千例接受移植治疗的患者数据。基于此，我们开发了一种用于识别移植后生存率极低的高风险患者的工具。目前，该工具可识别出大约25%至30%的高风险患者群体，他们在接受移植治疗后的1至2年内，总生存率和无复发生存率均明显偏低。这一研究结果具有较强的说服力，未来有望为临床医生在决定是否为患者实施异基因造血干细胞移植，或优先采用现有新药治疗时提供重要参考。
Oncology Frontier-Hematology Frontier：At this CMWP session, you presented on using machine learning to predict outcomes post-allogeneic HCT for chronic myeloid malignancies. Could you briefly share the key findings and their significance?
Pro. Adrian Mosquera-Orgeira：Oh, this is very in line with my previous answer.So the point is that we have new available drugs, we have a better opportunity to tailor these drugs to new patients and we are not so sure about the fact that whether all patients should be delivered to an allogeneic stem cell transplantation when this is progresses.Because we have strong evidence that mortality is high in particular subsets of patients.So the idea was to better identify those patients who develop poor or who have poor survival, either overall survival or relapse-free survival.When they are treated with an allogeneic stem cell transplantation.And the idea here was to use artificial intelligence, actually machine learning tools.Trained with the EBMT registry which encompasses thousands of patients treated across many countries mostly from Europe and then deliver a tool which can be used to better identify high-risk patients that is those patients who have very poor survival after allogeneic stem cell transplantation.And our tool actually identifies 25-30 percent of patients who are high risk they have a poor survival in terms of one two year.OS and relapse free survival and the results are quite compelling and probably will help clinicians deciding whether a patient should go for allogeneic stem cell transplantation or should be considered for medical, novel medical therapies that we have at this moment.
《肿瘤瞭望-血液时讯》在CMWP专场上，多位专家围绕异基因造血干细胞移植在不同慢性髓性恶性肿瘤的应用和优化进行了探讨，您如何看待异基因造血干细胞移植在骨髓纤维化中的优化策略？
Adrian Mosquera-Orgeira教授：我们最近在《Blood》杂志发表了一篇关于骨髓纤维化患者异基因造血干细胞移植的研究论文。当前，这是唯一可能实现治愈的治疗手段。除了异基因造血干细胞移植，目前尚无其他方法可以根治慢性骨髓增殖性疾病。但问题在于，移植相关毒性仍然较大。大约30%的骨髓纤维化患者在移植后预后较差。因此，我们亟需优化风险分层策略，使医生团队和患者本人能够在充分知情的基础上做出更合理的治疗决策。
这也意味着我们必须更加智慧地应用人工智能技术，尤其是在精准识别高风险患者方面，为这一临床需求提供解决思路。然而，目前这项工作仍处于起步阶段，远未完成。一旦能够更准确地识别出高风险人群，下一步就应聚焦于如何在移植前改善其身体状况。比如，是否可以通过新药降低肿瘤负荷、提升体能状态，使部分高风险患者也有机会接受移植治疗？这无疑是值得探索的重要方向。不过，这一过程不会简单。慢性骨髓增殖性疾病通常伴有复杂的克隆进化过程和分子生物学特征，因此未来的治疗路径也将更加复杂，亟需更强大的工具和更深入的研究支持。
Oncology Frontier-Hematology Frontier：At the CMWP session, experts discussed the use and optimization of allogeneic HCT in various chronic myeloid malignancies. What's your view on optimizing allogeneic HCT for myelofibrosis?
Pro. Adrian Mosquera-Orgeira：Allogeneic stem cell transplantation for myelofibrosis, well that's the point of the paper that we've just published in Blood.I mean, it's the only curative therapy that we have at this moment.There is no other way to cure chronic myeloproliferative diseases without an allogeneic stem cell transplantation at this moment.The problem is toxicity.Roughly 30% of patients with myelofibrosis who get the transplantation do perform very poorly.So we need to better risk stratify that to take more informed decisions for the doctor, for the team of doctors treating these patients and also for the patient himself.And that means using AI in a smart way in order to try to shed light on a medical need.That said, it doesn't mean All the work is done.Now that we can better identify high-risk patients, now it's time to study how can we improve their condition before allogeneic stem cell transplantation.For example, one question would be, can these new drugs deliver an improvement in overall tumor burden and performance status that can make that risk lower, and then the transplantation for some of those patients could still be an option?Well, that's something that could be considered.Address in the future but we needed these tools before and that's why the future is probably going to become more complex because at the very end we are dealing with complex.Disorders with lots of clonal evolution and complex molecular biology underneath so it is not going to be straightforward.