See Article by Bartoli et al

This is your last chance. After this, there is no turning back. You take the blue pill—the story ends, you wake up in your bed and believe whatever you want to believe. You take the red pill—you stay in Wonderland and I show you how deep the rabbit-hole goes.

—Morpheus in The Matrix¹

The advent of continuous flow left ventricular assist devices (CF LVADs) for end-stage heart failure has led to marked improvements in survival, device miniaturization and durability, yet serious adverse hematologic events limit this therapy. Although changes in CF LVAD design have led to a reduction in device thrombosis, gastrointestinal bleeding (GIB) remains a major source of morbidity occurring in ≈30% of patients.^2,3

The exact pathophysiology of GIB during CF LVAD remains unclear, but it has become apparent that 2 major factors contribute toward excessive bleeding: (1) a hematologic disturbance, specifically an acquired von Willebrand factor (VWF) syndrome (avWs) and (2) angiodysplasia formation in the gastrointestinal tract. VWF is a glycoprotein synthesized in endothelial cells (ECs) and megakaryocytes that undergoes post-translational modification to form ultralarge or high-molecular-weight multimers (HMWMs).⁴ These HMWMs are stored in the Weibel Palade bodies of ECs and platelets and released in response to hemorrhage. On release, HMWMs serve as bridges crosslinking exposed collagen and platelets to form a platelet plug at the site of injury within a blood vessel.⁴ HMWMs are cleaved by ADAMTS13 (a disintegrin and metalloproteinase with a thrombospondin type 1 motif, member 13) into smaller multimers at its A2 domain. Although VWF multimers of all sizes contribute to coagulation, the larger the multimer, the greater the hemostatic potential. Increased shear stress, as produced by CF LVADs, elongates the multimers and causes changes in the conformation of the A2 domain making it more susceptible to cleavage by ADAMTS13.⁵ Thus, shear forces generated during passage of blood elements through a CF LVAD accelerate degradation of HMWMs via increased proteolysis and to create avWs.

Although various subtypes of von Willebrand disease were known to be associated with angiodysplasia, novel in vitro studies by Starke et al⁶ further demonstrated VWF as a negative regulator of angiogenesis. In their investigation, inhibition of VWF expression in human umbilical vein ECs by siRNA led to an increase in VEGF (vascular endothelial growth factor)-dependent cell proliferation, migration, and tubule formation.⁶ It was suggested that VWF inhibits angiogenesis through the following 2 spatially separate mechanisms: (1) extracellular binding of VWF to integrin αvβ3 reduced VEGF-2–dependent proliferation of ECs and (2) defective Weibel Palade body formation from loss of VWF led to ineffective intracellular storage and unregulated release of Ang-2 (angiopoietin-2), thereby creating angiodysplasia.^6,7

Further investigation into dysregulated angiogenesis has shed some light toward a possible pathogenesis of angiodysplasia during CF LVAD support. Tabit et al reported higher levels of thrombin induced Ang-2 in patients with a CF LVAD. Moreover, antibody blockade of Ang-2 resulted in a reduction of tubule formation by human umbilical vein ECs after exposure to serum from CF LVAD patients.⁸ Additionally, retrospective studies have associated angiotensin-converting enzyme inhibitors/angiotensin receptor blockers⁹ and digoxin¹⁰—medications that can reduce Ang-2 levels—with reduced angiodysplasia-related GIB during CF LVAD support.

In this issue of Circulation: Heart Failure, Bartoli et al¹¹ further relate avWs to clinical bleeding and propose a potentially unifying mechanism that associates VWF multimer loss and fragment formation to the development of angiodysplasia during CF LVAD support. The authors confirm previous findings¹² and show a reduction in HMWMs with an increase in low-molecular-weight multimers, irrespective of clinical bleeding events. More interestingly, they note greater levels of VWF 450-kDA fragments in 7 (20%) patients who experienced an angiodysplasia-related GIB in comparison with nonbleeders (P=0.02).¹¹ To further investigate the relation between VWF multimers and angiodysplasia, they assess tubule length and migration of human umbilical vein ECs plated on matrigels with (1) exposure of fragment-rich whole blood from CF LVAD patients and (2) addition of 450-kDA VWF fragments isolated from a shear stress mock loop. In both experiments, there is reduction in tubule length and migration as compared with pre-CF LVAD controls. Overall, the authors conclude that CF LVAD-related shear stress creates VWF fragments, which in turn are associated with angiodysplasia and ultimately lead to the clinical manifestation of mucosal bleeding.

The investigators are to be commended for exploring a multifaceted and prevalent device-related adverse event in a prospective method on both clinical and mechanistic levels. An enigmatic clinical inquiry into device-related GIB has been why bleeding does not occur in all patients exposed to a CF LVAD since all patients develop avWs. In this study, Bartoli et al offer some granularity by proposing for the first time specific phenotypes of CF LVAD-related avWs, defined by a ratio of VWF multimers to fragments. With precise phenotypic identification, there could be potential for identifying at-risk individuals and developing targeted therapy, which would be a major step toward addressing the current Achilles heel of CF LVADs; yet, caution must be exercised, and findings must be framed within the scope of the present methodology aimed only at associating VWF profile with angiodysplasia.

Although it is interesting that patients with elevated VWF fragments shortly after LVAD implantation were noted to develop angiodysplasia-related bleeding, and in an in vitro experiment, isolated VWF fragments reduced tubule formation, the authors rightly mention that such results may not implicate in vivo causality. In the ex vivo experiment, with exposure of fragment-rich whole blood from CF LVAD patients to human umbilical vein ECs, there is reduced tubule formation, but it is uncertain whether there are other antiangiogenic mediators that may impede normal microvessel development. More pressing is the uncertainty in reconciling the matrix, that is, the role of decreased in vitro tubule formation related to VWF fragments and increased tubule formation with loss of VWF/elevated Ang-2, as shown in previous studies,^6,8 toward angiodysplasia formation. In other words, whether VWF fragment associated reduction in tubule formation and migration is indeed equivalent to in vivo CF LVAD-related angiodysplasia bleeding remains to be further established.

As with any intriguing findings, there tend to be more questions than answers. A few queries at the forefront are as follows: (1) Why do only a subset of CF LVAD patients form VWF fragments? (2) How do VWF fragments prevent normal microvessel formation? (3) What is the role of pulsatility in determining the VWF profile, as recent animal model studies suggest preservation of HMWMs under pulsatile conditions?¹³ (4) Is there device-specific VWF fragment formation? and (5) How can this VWF fragment-related hypothesis be integrated with the recent observation that small bowel mucosal angiodysplasia may precede CF LVAD implantation?¹⁴ Moreover, if VWF profile is related to GIB, since differing VWF breakdown patterns are reported among various CF LVADs,¹⁵ there should be differences in the prevalence of GIB between such devices; but this was not reported in the 2-year outcomes of the MOMENTUM trial (The Multicenter Study of MagLev Technology in Patients Undergoing Mechanical Circulatory Support Therapy With HeartMate 3).² Though likely not the sole culprit toward angiodysplasia-related GIB, perhaps knowledge of VWF fragment formation may eventually inform therapeutic use of downstream antiangiogenic mediators and tailored anticoagulation therapies to modify disease burden. Bartoli et al¹¹ have made another important contribution to exploration of the von Willebrand matrix, but the mechanistic underpinnings of GIB in CF LVAD are yet to be fully elucidated. For now, we shall take the red pill with a few video capsules and remain in wonderland.

Dr Jorde is a nonpaid consultant to Abbott. The other authors report no conflicts.

https://www.ahajournals.org/journal/circheartfailure

The opinions expressed in this article are not necessarily those of the editors or of the American Heart Association.

参见Bartoli等人的文章

这是您最后的机会。此后，就没有回头路了。您服用蓝色药丸-故事结束了，您在床上醒来，相信了您想相信的一切。您服用红色药丸-您留在仙境中，我向您展示了兔子洞有多深。

— 《黑客帝国》中的莫非斯¹

用于末期心力衰竭的连续血流左心室辅助设备（CF LVADs）的出现已显着改善了生存率，设备小型化和耐用性，但严重的血液学不良事件限制了该治疗方法。尽管CF LVAD设计的改变已导致装置血栓形成的减少，但胃肠道出血（GIB）仍是约30％的患者发病的主要来源。^2,3

尚不清楚CF LVAD期间GIB的确切病理生理，但已经发现有2个主要因素导致出血过多：（1）血液学异常，特别是获得性血管性血友病因子（VWF）综合征（avWs）和（2）血管增生在胃肠道形成。VWF是在内皮细胞（EC）和巨核细胞中合成的糖蛋白，经过翻译后修饰后形成超大或高分子量多聚体（HMWM）。⁴这些HMWM储存在EC和血小板的Weibel Palade体中，并因出血而释放。释放后，HMWM充当桥梁，使暴露的胶原蛋白和血小板交联，从而在血管损伤部位形成血小板栓塞。⁴HMWM被ADAMTS13（具有血小板反应蛋白1型基序的整合素和金属蛋白酶，成员13）切割成较小的多聚体，位于其A2结构域。尽管各种尺寸的VWF多聚体均有助于凝血，但多聚体越大，止血潜力越大。CF LVAD产生的剪切应力增加会延长多聚体，并导致A2结构域构象发生变化，使其更易于被ADAMTS13切割。⁵因此，在血液元素通过CF LVAD的过程中产生的剪切力通过增加的蛋白水解作用加速了HMWMs的降解并产生了avWs。

尽管已知血管性血友病的各种亚型与血管增生有关，但Starke等[ ^6]进行的新的体外研究进一步证明，VWF是血管新生的负调节剂。在他们的研究中，siRNA抑制人脐静脉EC中VWF的表达导致依赖VEGF（血管内皮生长因子）的细胞增殖，迁移和肾小管形成增加。⁶提示VWF通过以下2种在空间上独立的机制抑制血管生成：（1）VWF与整联蛋白αvβ3的细胞外结合减少了VEGF-2依赖性EC的增殖，以及（2）由于VWF丧失导致的有缺陷的Weibel Palade体形成导致无效细胞内储存和Ang-2（angiopoietin-2）的不受控制的释放，从而导致血管增生。^6,7

对血管生成失调的进一步研究为CF LVAD支持期间血管增生的可能发病机理提供了一些启示。Tabit等报道CF LVAD患者中凝血酶诱导的Ang-2水平更高。此外，在暴露于CF LVAD患者的血清后，Ang-2抗体的阻断导致人脐静脉ECs的小管形成减少。⁸此外，回顾性研究还发现在CF LVAD支持期间，血管紧张素转换酶抑制剂/血管紧张素受体阻滞剂⁹和地高辛^10（可降低Ang-2水平的药物）与血管增生相关的GIB降低有关。

在《循环：心力衰竭》一期中，Bartoli等人¹¹进一步将avWs与临床出血联系起来，并提出了潜在的统一机制，该机制将CFWF LVAD支持期间VWF多聚体的丢失和片段形成与血管增生的发展相关联。作者证实了先前的发现^12，并显示出HMWM的减少与低分子量多聚体的增加，而与临床出血事件无关。更有趣的是，他们注意到与非出血者相比，在7名（20％）经历了血管增生相关的GIB的患者中，VWF 450-kDA片段的水平更高（P = 0.02）。¹¹为了进一步研究VWF多聚体与血管增生之间的关系，他们评估了（1）CF LVAD患者富含片段的全血和（2）添加了450-kDA VWF的方法，将小管长度和人脐静脉ECs转移到了基质胶上。从剪切应力模拟循环中分离出的碎片。在两个实验中，与CF-LVAD前对照相比，肾小管长度和迁移均减少。总体而言，作者得出结论，CF LVAD相关的切应力产生VWF碎片，而VWF碎片又与血管增生相关，并最终导致粘膜出血的临床表现。

研究人员应以前瞻性的方法在临床和机制水平上探索多方面且普遍存在的与器械相关的不良事件，这是值得赞扬的。对与设备相关的GIB的一种神秘的临床研究一直是为什么在所有暴露于CF LVAD的患者中都没有出血的原因，因为所有患者都出现了avW。在这项研究中，Bartoli等人首次提出了CF LVAD相关avW的特定表型，该表型由VWF多聚体与片段的比例定义，从而提供了一些粒度。通过精确的表型鉴定，可能存在鉴定高危人群和开发靶向治疗的潜力，这将是解决当前CF LVAD致命弱点的重要一步；但是，必须谨慎行事，

尽管有趣的是，在LVAD植入后不久发现VWF片段升高的患者会发生血管增生相关的出血，并且在体外实验中，分离的VWF片段减少了肾小管的形成，但作者正确地提到，这种结果可能并不暗示体内因果关系。在离体实验中，来自CF LVAD患者的富含片段的全血暴露于人脐静脉EC，从而减少了肾小管的形成，但尚不确定是否存在其他可能抑制正常微血管发育的抗血管生成介质。迫在眉睫的是调和基质的不确定性，即，与VWF片段相关的体外肾小管形成减少，以及VWF / Ang-2升高引起的肾小管形成增加的作用，如先前研究[ ^6,8]所示。向血管增生形成。换句话说，VWF片段相关的肾小管形成和迁移减少是否确实等同于体内CF LVAD相关的血管增生性出血尚待进一步证实。

与任何有趣的发现一样，问题多于答案。最前沿的几个问题如下：（1）为什么只有CF LVAD患者的一部分会形成VWF片段？（2）VWF片段如何阻止正常的微血管形成？（3）脉搏性在确定VWF谱中的作用是什么，因为最近的动物模型研究表明，在脉动条件下可以保存HMWMs？¹³（4）是否存在特定于设备的VWF片段形成？（5）如何将这种VWF片段相关假说与最近的观察结合起来，即小肠粘膜血管增生可能在CF LVAD植入之前？¹⁴此外，如果VWF配置文件与GIB相关，则由于在各种CF LVAD之间报告了不同的VWF击穿模式，因此¹⁵这些设备之间的GIB患病率应有所不同；但这并未在MOMENTUM试验的2年结果中报告（MagLev技术在接受HeartMate 3机械循环支持疗法的患者中的多中心研究）。²尽管不是血管增生相关性GIB的唯一罪魁祸首，也许对VWF片段形成的了解可能最终会指导下游抗血管生成介质的治疗应用以及量身定制的抗凝疗法以改变疾病负担。Bartoli等人¹¹von Willebrand矩阵的探索做出了另一重要贡献，但CF LVAD中GIB的机制基础尚待充分阐明。现在，我们将把红色药丸和一些视频胶囊一起服用，并留在仙境中。

乔德博士是雅培的无偿顾问。其他作者报告没有冲突。

https://www.ahajournals.org/journal/circheartfailure

本文表达的观点不一定是编辑者或美国心脏协会的观点。