Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is a well-established therapy for hematologic and lymphoid malignancies, but also for disorders of the immune or hematopoietic systems and metabolism incurable by other treatments. Despite the improvement in prognosis, toxicity and treatment-related mortality (TRM) still remain challenges (1, 2). In our recent study (3) nearly half (56/122, 45.9%) of the pediatric patients, who underwent allo-HSCT, had at least one severe adverse event (grade 3 or 4 by the Common Terminology Criteria for Adverse Events, CTCAE, 4.03 classification). Endothelial cell activation and dysfunction have recently been indicated in many severe complications of allo-HSCT (4). Thrombotic microangiopathy (TMA) is one of the best known endotheliopathy syndromes and a relatively common complication after HSCT affecting 10-20% of the patients (5–8). Previously, we identified a total of 19/122 (15.6%) patients with vascular complications, fulfilling the criteria of capillary leak syndrome (CLS), veno-occlusive disease/sinusoidal obstruction syndrome (VOD/SOS) or TMA. The patients with endotheliopathy had a poorer 5-year overall survival than those without it (77% versus 26%, p<0.001) (3).

Endothelial cells can be damaged and activated during the HSCT process by several mechanisms, including previous chemotherapy, conditioning (e.g. irradiation), drugs used in the transplantation process (such as G-CSF or calcineurin inhibitors), cytokines produced by activated immune cells or the injured tissues, infections, engraftment and alloreactivity (9–11). Injury and inflammation involve neutrophil activation leading to formation of neutrophil extracellular traps (NETs). NETs and other released intracellular materials are thought to activate the complement system (12, 13). On the other hand, overactive or activated complement system has been shown to disturb the normal function of the endothelium in many disease states (14–18). Previously, endotheliopathy in HSCT was considered to follow the “two-hit” –hypothesis. First, many risk factors in the conventional chemo or conditioning phase make the endothelium more vulnerable and procoagulant (Hit 1). This is followed by one or more factors (medications, infections, possible antibodies, or alloreactivity) in the post-HSCT recovery phase (Hit 2) promoting endothelial injury and related problems (19). Although not emphasized earlier, the second hit could, in addition to procoagulant activity, also involve “procomplement” activity with changes in the ability of endothelial cells to protect themselves against complement attack (5). Currently a “three hit” –hypothesis, where the first hit is the patients´ genetic predisposition to endothelial complications, is considered most relevant to the disease pathogenesis (5, 11, 20, 21).

Problems characterized by endothelial dysfunction vary in their clinical phenotype. Still, all of these share a common denominator: endothelial injury accompanied by proinflammatory, prothrombotic and proapoptotic processes likely related to complement activation (4, 8, 10). Here we focused on the following three entities: capillary leak syndrome (CLS), HSCT-related thrombotic microangiopathy (HSCT-TMA) and veno-occlusive disease/sinusoidal obstruction syndrome (VOD/SOS). Other endotheliopathy syndromes, such as engraftment syndrome (ES), idiopathic pneumonia syndrome (IPS), posterior reversible encephalopathy syndrome (PRES), peri-engraftment respiratory distress syndrome (PERDS) and refractory acute GVHD will not be addressed here.

In this retrospective study, because of the postulated role of complement, we studied genetic complement system variants in patients with documented endotheliopathy. Thereby, our aim was to lay basis for understanding potential mechanisms behind the acute complications seen post-HSCT.

The patient data and clinical findings of our patient cohort have been published before (3). In brief, the original patient cohort consisted of 122 patients with an allogeneic HSCT between 2001 and 2013 at the Helsinki University Children´s Hospital, Finland. We evaluated the acute adverse events emerging ≤ 100 days post-transplant and prevalence of key vascular complications. We identified a total of 19/122 (15.6%) patients with vascular complications, fulfilling the criteria of capillary leak syndrome (CLS), veno-occlusive disease/sinusoidal obstruction syndrome (VOD/SOS) or thrombotic microangiopathy (TMA). These patients had a poorer 5-year overall survival than those without vascular complications (26% versus 77%, p<0.001). These endotheliopathy-related adverse events appeared early on post-HSCT, varied in their clinical phenotype and predicted a poor outcome.

The present study found a higher-than-expected frequency of rare complement gene variants possibly related to vascular complications following HSCT in children. Particularly interesting associations were found to the terminal pathway complement components C6 and C9 as well as to the main phagocytic complement receptor type 3 (CR3, CD11b/CD18). While pathway-burden analysis indicated that rare missense variants in genes coding for classical pathway are enriched in patients in general, from the point of view of endotheliopathy, main links were found to the terminal pathway complement components.

We found five patients with mutations in the ITGAM gene encoding for CD11b (integrin alpha-M), a central part of the key complement receptor CR3. CD11b alpha chain forms the CR3 receptor together with the integrin beta 2 chain (CD18 encoded by ITGB2). CR3 binds structures coated with iC3b, which is an inactivated form of C3b. CR3 regulates the adherence of neutrophils and monocytes to stimulated endothelium. Importantly, CR3 mediates the phagocytosis of complement iC3b-coated microbes and other particles. CR3 functions also as a receptor for the coagulation system proteins fibrinogen and factor X (FX). Mutated ITGAM may thus result in diminished neutrophil adhesion and migration, reduced phagocytosis, increased cytokine production and elevated anti-dsDNA antibody formation, which may further trigger the complement cascade (33). Genetic variants in the ITGAM gene are strongly associated with systemic lupus erythematosus (SLE) and its complications (lupus nephritis, etc) (33), as well as with pre-eclampsia (34). CR3 and CR4 are both β2-integrins that specifically recognize iC3b. It has been previously found that variant rs2230424 in CR4, located in the first blade of the β-propeller domain, results in a decreased binding capacity of CR4 to iC3b (34). We speculate, that ITGAM variant (rs775141495), also in the first blade of the β-propeller domain, may also interfere with the binding of CR3 to iC3b.

Acute respiratory distress syndrome (ARDS) is characterized by an increased permeability of endothelium and epithelium leading to an inflammatory reaction with accumulation of neutrophils and a protein-rich edema of the lungs. Later, also macrophages, strongly expressing CR3, promote the inflammatory response. Biomarkers of the endothelial damage have been linked to increased morbidity and mortality (35–37). One of our patients with the ITGAM variant (rs775141495; Arg292His) had an endotheliopathy-like clinical phenotype with ARDS. The patient´s ARDS symptoms evolved from day 30 post-transplant, thus matching the timeline of the post-engraftment acute phase (30-100 days post-transplant) pulmonary complications. An impaired cellular immunity plays a central role in this phase in both the infectious and non-infectious complications with the latter including many entities related to endothelial damage (38, 39). Our patient with the ITGAM variant Arg292His with CLS followed by TMA/VOD was treated with defibrotide (6.25mg/kg iv q. 6 hours) and survived. The other one with ARDS was not treated with defibrotide and died. Defibrotide is found to diminish the endothelial activation and enhance the local antithrombotic effects, while having also anti-inflammatory effects by interacting with the endothelium (40). In the clinical trial by Frame et al. (41) defibrotide was found safe, and even with the study not being designed to address the efficacy of defibrotide, the survival rates of the patients treated with defibrotide were promising, when compared to other patients with SARS-CoV-2 ARDS.

Mannan-binding lectin-associated serine protease -1 (MASP-1) is an initial part of the lectin pathway and can promote activation of both complement and coagulation. It is a procoagulant by inducing fibrin formation with thrombin, but it also directly activates prothrombin, FXIII and thrombin-activatable fibrinolysis inhibitor (TAFI) (42, 43). Thrombin, on the other hand, has been reported to directly activate the complement cascade by cleaving C3 and C5 (44). The lectin pathway was suggested to mediate complement activation in SARS-CoV-2 infection, and the MASP2-inhibitor narsoplimab had a promising effect in critically ill COVID-19 patients (45). The lectin pathway may also play a role in HSCT-related TMA. Adult patients with HSCT-TMA treated in a phase II study with narsoplimab showed significant improvement in TMA-related laboratory values, and the treatment resulted in a good clinical response with improved overall survival (46). Our patient with the MASP-1 variant Ala544Thr had severe and persistent diffuse gastrointestinal bleeding leading to ICU-admission twice and ending up being lethal. Massive haemorrhage is thought to be a possible trigger of complement activation (47, 48) and activated complement could contribute to tissue damage seen at the site of the haemorrhage (49, 50). We speculate that the MASP-1 variant rs1712360640 with severe GI-bleeding could lead to a more vigorous activation of the complement and coagulation cascade, leading to endothelium-related complications.

C6 is an essential part of the membrane attack complex (MAC). The variant (rs142896559; Gly452Glu) seen in our patient with CLS in located in the MAC perforin (MACPF) domain of the C6 protein (51). The variants seen in C6 and C9 are both located in the MACPF domain. The functional consequences of this variant in C6 are currently not known. However, the essential role of C6 in the generation of the complement membrane attack complex (MAC) and location of the variant in one of the key domains suggests that it could be involved in abnormal assembly of MAC.

While the functional consequences of the C9 variant (rs183125896; Thr279Ala) observed in four patients are also unclear, interestingly, reports on other C9 variants in connection to TMA have recently been published (52, 53). It is therefore possible that, in the HSCT patients, this mutation, normally tolerable, could contribute to triggering TMA.

Many variants were overrepresented in the whole HSCT population ( Table 2 ). Interestingly, also transplant patients without vascular problems showed enrichment of variants in many complement genes. This could be related to the underlying reasons for the HSCT (usually malignancy, but also hematopoietic syndromes, immunological or metabolic disorders; Table 1 ). In this scenario, complement may have functioned insufficiently to prevent the disorder. Furthermore, the variants observed in the endotheliopathy patients were enriched in the patient population, although the small sample size of endotheliopathy patients did not warrant direct statistical comparison. A reason for this could be an incompatibility in the complement system between the recipient and the donor. Components from the donor and the recipient (e.g. between an activating and a regulating factor) may have a mismatch with each other and lead to events that harm the vascular endothelium. This hypothesis is supported by the fact that patients with endotheliopathy had more matched unrelated donors than those without endotheliopathy (p=0.031).

The limitation of our study is the size of our cohort. Gene burden analysis may be influenced by the relatively small N and the findings should be evaluated using a larger validation cohort. The strength of our study is the uniform Finnish ethnicity of the patients. Because of the centralization of more than 95% of the paediatric HSCTs in Finland to one, single centre, the cohort covers the whole Finland and represents the national population well. Thus, the frequency of variants in the patient set and gnomAD non-cancer Finns was comparable. Comparison to gnomAD Europeans ( Supplementary Table S2 ) showed differences in complement variant distribution between Finns and the rest of Europeans. This is understandable, because it is well known that Finns are a genetically relatively uniform but distinct population in Europe (54).

Treatment-related toxicity is still a significant clinical problem in paediatric allo-HSCT. Injury to the vascular endothelium is relatively common after HSCT with varying phenotypic appearances suggesting hitherto unidentified underlying mechanisms. Rare and ultrarare variants in complement genes are overexpressed in patients with injury to the vascular endothelium. These nonsynonymous variants in complement genes may be related to endotheliopathy and poor prognosis seen in these patients. Our hypothesis is that in addition to these variants, also genetic incompatibility between the donor and the recipient predisposes to endothelial damage and problems seen post-HSCT. Studies to address this hypothesis are underway.

The datasets presented in this article are not readily available because of ethical and privacy restrictions. Requests to access the datasets should be directed to the corresponding author/s.

The studies involving humans were approved by The Research Ethics Committee of the Helsinki University Hospital. The studies were conducted in accordance with the local legislation and institutional requirements. The ethics committee/institutional review board waived the requirement of written informed consent for participation from the participants or the participants’ legal guardians/next of kin because all samples and data were obtained following authorization by the ethics committee and the national supervisory authority Valvira. Written, informed consents were not attainable due to the retrospective nature of the study also including deceased patients.

LL performed research and collected data; JK analyzed and interpreted data; KV and OK supervised the students and interpreted data; IL and SM designed research, supervised the students, and interpreted data. LL, JK, KV, IL, and SM wrote the manuscript. All authors contributed to the article and approved the submitted version.