Introduction

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is an effective, yet high-risk, therapy for hematologic and lymphoid malignancies, disorders of the hematopoietic or immune systems and metabolism uncurable by other procedures. The overall survival has improved in both malignant and non-malignant disorders due less toxic treatment protocols, better patient and donor selection and improved supportive care1,2,3. Yet, treatment-related mortality (TRM) and morbidity remain a challenge4,5,6,7. Our knowledge on HSCT-associated thrombotic microangiopathy (HSCT-TMA) has expanded over the past decade, and it is now recognized as a severe complication in HSCT8,9,10,11. An early elevation of the soluble terminal complement complex (SC5b-9) in blood has been reported to have a high sensitivity, but a suboptimal specificity to detect the development of TMA in pediatric HSCT12,13. In adults, complement activation during the first 12 weeks post-transplant has been shown to be associated with an increased non-relapse mortality and poorer overall survival14.

In this prospective cohort study, we assessed activation of the complement system peri- and post-transplant by measuring C3a and SC5b-9 in blood. C3a is a sensitive marker of complement activation generated as a response to plasma exposure to external complement activators (like microbes) or tissue injury. C3a generation may also reflect an imbalance in complement regulation, especially in the alternative complement pathway. This may be the consequence of genetic or acquired factors. For the current study, the potentially relevant reason for complement activation could be the changes in endothelial or blood cell surfaces related to the HSCT procedure possibly altering the ability of the surfaces to protect themselves against complement activation. Our aim was to see, if early complement activation occurs, and whether it is related to the clinical course of individual patients. We also studied the incidence of acute adverse events and key vascular complications during the pediatric HSCT process, i.e. capillary leak syndrome (CLS), HSCT-related thrombotic microangiopathy (HSCT-TMA) and veno-occlusive disease/sinusoidal obstruction syndrome (VOD/SOS).

Materials and methods

Patients

A total of 135 patients underwent an allo-HSCT between 1/2014 and 12/2020 at the Helsinki University Children´s Hospital, Finland. The prospective cohort consisted of 42 patients, who underwent allo-HSCT within the given timeframe. This unselected cohort of 42 pediatric patients was formed on the basis of receiving written, informed consents from the guardians. Out of the 42 allo-HSCT patients EDTA (0.01 M) plasma samples were obtained from 23, and, in addition, from 3 patients with an autologous graft. The remaining 19 allo-HSCT patients in the cohort were followed up meticulously to evaluate the prevalence of acute adverse events and key vascular complications. The inclusion criterion was simple, first allogeneic HSCT, and with multiple blood sampling the exclusion criterion patient´s weight less than 15 kg. As reference for the vascular complications we used our previously described, and retrospectively analyzed, patient group15 consisting of 122 pediatric patients with an allo-HSCT between 1/2001 and 12/2013 at the Helsinki University Children´s Hospital, Finland. The clinical follow-up for both cohorts was 1 year after HSCT. The key demographic and clinical data of both groups are given in Table 1.

Table 1 The demographics of the prospective study cohort transplanted during 2013–2020 at the Helsinki university children´s hospital and the previously published retrospective cohort transplanted at the Helsinki university children´s hospital during 2001–2013.
Full size table

The study was approved by the Research Ethics Committee of the Helsinki University Hospital (79/13/03/03/2016, update 3082/2018). All research was performed in accordance with the Declaration of Helsinki. Written, informed consents were obtained from the guardians and patients over 6 years of age. No organs or tissues were procured from prisoners.

Outcome measures

The aim of this study was to assess the activation of the complement system in pediatric HSCT and its 6-month follow-up. The complement activation was correlated with the acute toxicity and key vascular complications, i.e., capillary leak syndrome, TMA, and VOD. The prevalence of the complications was compared between our two cohorts from different era.

Data collection

Blood samples were obtained according to a predefined sampling plan in the two sets. In the first set (n = 6) the samples were obtained pre-transplant, and thereafter consecutively at 10 min, 30 min, 2, 4, and 24 h after the start of the graft infusion. The purpose was to detect possible, procedure-related complement activation. In the second set (n = 26) the purpose was to relate the complement activation to possible complications. Blood samples were collected pre-transplant and at 48 h, 2 weeks, 3 weeks, 1, 3 and 6 months post-transplant. The 5 ml blood samples were collected into EDTA-tubes, processed without delay, and the separated plasmas divided into three aliquots. The plasma samples were quickly frozen at -20° C and transferred to storage at -70° C. The predefined clinical data was collected at 0, 7, 14, 28, 100 days and 1-year post-transplant from the medical records. The following data was collected: the diagnosis, conditioning, transplant (including ABO-match, HLA-match, cells infused, patient/donor viral status), engraftment, laboratory values (hemoglobin, leukocytes, neutrophils, thrombocytes, lactate dehydrogenase, haptoglobin, Coombs, albumin, bilirubin, alanine aminotransferase, creatinine, partial thromboplastin time) infections, medications, acute adverse events pre-transplant, possible endotheliopathy symptoms, supportive care, and GVHD status. The collection and grading of the acute adverse events was as previously reported15.

Treatment characteristics

The treatment of the retrospective cohort was as described previously15. The first line treatment of ALL and AML in the prospective cohort was according to the pertinent, Nordic (NOPHO) treatment protocols16,17,18. The patients received myeloablative conditioning with cyclophosphamide (Cy, 100–200 mg/kg) or busulfan (12-16 mg/kg), or etoposide (60 mg/kg), or cytarabine (36mg/m2), or combination of fludarabine up to 150 mg/m², treosulfan (42 mg/m²), and thiotepa (10 mg/m²) with or without fTBI. A patient with severe aplastic anemia (SAA) received a conditioning with Cy (120 mg/kg) and fludarabine (150 mg/m²) without total body irradiation (TBI). Patients with NHL and CML were treated according to pertinent international guidelines, and for conditioning they, as well as those with a non-malignant hematopoietic disease or immunodeficiency, received myeloablative conditioning with Cy (100–200 mg/kg) or busulfan (12-16 mg/kg) or combination of fludarabine up to 150 mg/m², treosulfan (42 mg/m²), and thiotepa (10 mg/m²) and etoposide (60 mg/kg) without fTBI. The conditioning was classified into three groups (myeloablative with or without fTBI or reduced intensity) as shown in Table 1.

Prophylactic defibrotide (6.25 mg/kg q. 6 h) was used for a high-risk of VOD (n = 15; 35.7%) for the first three weeks post-transplant. Acute GVHD (aGVHD) was graded according to the Glucksberg classification19. Anti-thymocyte globulin (ATG) was used in 29 (69%) and alemtuzumab in 4 (9.5%) patients as serotherapy for the prophylaxis of GVHD also consisting of daily cyclosporine in 71.4% (n = 30) and mycophenolate mofetil in 26.2% (n = 11) of patients. One patient (2.4%) did not have any GVHD-prophylaxis. 33.3% (N = 14) of the patients needed therapy with prednisolone (2 mg/kg/day) prior to day + 100 post-HSCT for grade 2 (gut or liver) or 3 (skin) aGVHD.

Vascular complications

The definitions of early complications of vascular origin and the diagnostic criteria used for capillary leak20, TMA and VOD21 were as described previously15, and the same for both cohorts. TMA was diagnosed using the criteria proposed by Cho et al.22 or having clinical evidence supporting the diagnosis when, in the absence of specific laboratory aberrations, the criteria (cf. Table 2) were only partially met. The TMA Harmonization Panel Consensus Recommended Diagnostic Criteria have just recently been published requiring ≥ 4 of the 7 criteria within 14 days at 2 consecutive time points to be met23. Our TMA-patients also met these criteria. 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 were not addressed.

Table 2 Diagnostic criteria for the vascular complications.
Full size table

Laboratory analyses to detect complement activation

We analyzed the complement activation products C3a and SC5b-9 to monitor for possible complement activation during HSCT. A cohort of 6 patients (5 allo-HSCT and one autologous) were sampled peri-transplant at 0–10 min, 30 min, 2, 4 and 24 h post-transplant. For a longer-term follow-up, a cohort of 26 patients (23 allo-HSCT and 3 autologous) were sampled at seven timepoints from pre-transplantation to 6 months post-transplant as described above. The quantitation of the C3a activation fragment of the complement protein C3 was performed with the MicroVue C3a Plus Enzyme Immunoassay (Quidel Corporation A031, San Diego, CA, USA). For the quantitation of the soluble terminal SC5b-9 complex we used the MicroVue SC5b-9 Plus Enzyme Immunoassay (Quidel Corporation A020, San Diego, CA, USA). The samples were initially diluted 1:200 for C3a and 1:10 for SC5b-9 tests. We followed the test protocols of the manufacturer, except for the incubation time with the 3,3´, 5,5´-tetramethylbenzidine (TMB) substrate. The reaction was stopped after 8 min, when the color intensity reached the observation window. The absorbance (OD) was measured in addition to 450 nm also at 540 nm to correct for non-specific absorbance. The Four Parameter Logistic (4PL) curve fit by MyCurveFit (version 1.0.1102.823 (230120), MyAssays Ltd, Brighton, Sussex, UK, was used for generating the standard curve.

Statistical analysis

The statistical analyses were performed using the IBM SPSS Statistics, version 26.0, Armonk, NY, USA. Cross tabulation with the Fisher´s exact, χ2 -test for categorical variables and the independent samples T-test test for continuous variables were employed. Out of the five patients with two allo-HSCTs during our time frame only the first was included. P-value < 0.05 was employed to indicate statistical significance between the observed differences.

Results

Patient characteristics

Key patient characteristics are presented in Table 1. When compared to the retrospective cases, the patients in the prospective cohort were older at the time of HSCT (median 11.0 years [range 1.7–18.1] compared to 7.7 years [range 0.5–17.2], p < 0.01), and had more matched unrelated donors (73.8% vs. 50.8%, p = 0.045). In the prospective group 61.9% (n = 26) had acute GVHD (aGVHD) of any grade before day 100 post-transplant. Of these, 26.9% (n = 7) had a severe, grade 3 or 4 aGVHD in at least one organ (skin, gut or liver). The overall survival was better in the prospective (76.2%) than in the retrospective group (68.9%)(n.s). TBI was used significantly less in the prospective group (42.9%, 18/42 vs. 80.3%, 98/122; p < 0.01).

Acute adverse effects

Out of our total of 42 patients, 39 (92.9%) had at least one adverse event (grade 2–4) and 23 (54.8%) at least one severe, grade 3 or 4 adverse event (cf. Table 3). Four patients had 50% (29/58) of all the severe adverse events encountered. Three of these were diagnosed with TMA.

Table 3 Distribution and severity of adverse events in the pediatric allo-HSCT cohort of 42 patients ≤ 100 days post-transplant graded according to CTCAE 4.03.
Full size table

Vascular complications

There was no statistically significant difference in the endotheliopathy incidence between the prospective and retrospective cohorts (9.5% vs. 15.6% respectively). Four patients (3 male and 1 female) were identified with vascular complications fulfilling the criteria of a capillary leak syndrome (CLS), veno-occlusive disease/sinusoidal obstruction syndrome (VOD/SOS) or thrombotic microangiopathy (TMA). Of these, two had TMA, one CLS followed by TMA and one CLS followed by pulmonary VOD. Patients with the vascular complications had more sibling (n = 3) than matched unrelated donors (MUD) (n = 1) when compared to those without endotheliopathy (p = 0.049). The endotheliopathy patients had more severe, grade 3–4 aGVHD in at least one organ (skin, gut or liver) (75% vs. 10.5%, p = 0.011), and two of the four (50%) died due to toxicity.

Complement activation

In the technical controls, the acceptable C3a and SC5b-9 concentration ranges for the high- and low-level controls of the certificate of analysis were met.

The C3a and SC5b-9 alterations during HSCT are shown in Fig. 1. The C3a levels among both the allo- and auto-HSCT patients were quite similar and peaked at two timepoints, at 30 min and 24 h (Fig. 1A). The SC5b-9 peak at 30 min in the patients with an autologous graft followed the same pattern as C3a (Fig. 1A,B). The three allo-HSCT patients (allo 1P, 3P and 5P) had a clear elevation in SC5b-9, two at 30 min (Fig. 1C). The patient 3P developed CLS during the second week and pulmonary VOD 2 months post-transplant. 4 out of these 5 allo-HSCT patients had a transplant with both an ABO major and minor match. One patient (Fig. 1C, Patient Allo 5P) had a major ABO-mismatch and > 50% increase in SC5b-9 but not in C3a. The patient had no endotheliopathy. Also, the auto-HSCT patient had a 1.65-times higher level of SC5b-9 at 30 min. Altogether, these results showed that the HSCT was associated with a procedure-related complement activation.

Fig. 1
figure 1

Complement activation during and after the HSCT procedure. Allogeneic and autologous patients are shown separately, number of patients is given. (A) The mean C3a concentration changes during the peri-transplantation period showing peaks at 30 min and at 24 h. (B) The mean SC5b-9 concentration changes during the peri-transplantation period with a peak at 30 min. (C) The C3a and SC5b-9 variations of each patient (allo n = 5, auto n = 1) during the peri-transplantation period.

Full size image

We also monitored complement activation long-term (up to 6 months). When analyzed as a group, 16/23 (69.6%) of the allo-HSCT patients had an increase in the SC5b-9 at some point during the first month post-transplant, when compared to their pre-transplantation level. We categorized the patients into two groups based on the alteration in their SC5b-9. Figure 2 shows the C3a- and C5b-9 elevation trends for each of the patients with the one with TMA shown separately (Fig. 3) during the 6-month follow-up. Figure 2A shows the patients with at least 50% elevations in SC5b-9 at some point, and Fig. 2B those with no or less than a 50% elevation. Patients who died are marked. After the SC5b-9 peak during the HSCT (cf. above) the endotheliopathy patient with CLS followed by pulmonary VOD (Patient 3P on Fig. 1C) did not have further elevations in C3a or SC5b-9 (Fig. 2B, here Patient 10).

One patient (Patient 3) showed a massive elevation in SC5b-9 peaking at 1-month post-transplant, at nearly 40 times the pre-transplant level (Fig. 3), but the C3a was not elevated at any point. The patient had a severe TMA starting with an increased platelet consumption at D18 followed by red cell fragmentation and increased consumption of blood products, severe hypoventilation, gastrointestinal bleeding, epidermal necrolysis, acute kidney injury and severe aGVHD. The patient died of toxicity on D74 post-transplant.

Among the patients with an allogeneic graft no peak in C3a could be seen, while the autologous had a C3a peak at 48 hours (Fig. 4A). The mean SC5b-9 levels did not show a general pattern (Fig. 4B). The one allo-patient (Patient 3, c.f. Fig. 3) with extremely high SC5b-9 levels separating strongly from all the other patients was excluded from the analysis.

Fig. 2
figure 2

The C3a and SC5b-9 levels of each of the patients with an allogeneic HSCT. (A) with at least a 50% elevation in SC5b-9 at some point, and (B) without elevation or less than 50% increase in SC5b-9. Also given the severe (gr 3–4) adverse events and clinical findings that include in the diagnostic criteria of endotheliopathies. Patients who died are marked with †. AKI acute kidney injury, CLS capillary leak syndrome, GI gastrointestinal, FUO fever of unknown origin, VOD veno-occlusive disease.

Full size image
Fig. 3
figure 3

The C3a and SC5b-9 concentration changes of a patient with severe TMA. Also shown the timepoints, when each problem was first noticed. GI gastrointestinal, TEN toxic epidermal necrolysis, LD lactate dehydrogenase, AKI acute kidney injury.

Full size image
Fig. 4
figure 4

The mean concentration changes of the complement activation markers during the first 6 months of the transplantation period. Allogeneic and autologous patients shown separately, number of patients is given. (A) The mean concentration of C3a. (B) The mean concentration of SC5b-9. The one allo-patient (Patient 3, cf. Fig.  3) with extremely high SC5b-9 levels separating strongly from all the other patients was excluded from the analysis.

Full size image

Discussion

Treatment-related mortality and morbidity remain significant problems in pediatric allo-HSCT24. Our nationally centralized, single-center study shows that 93% of the patients had at least one, grade 2–4 adverse event. One third of the events were severe (grade 3 or 4), and patients with endothelial damage burdened with a multitude of major, clinical problems. One half of the adverse events seen in our cohort were accumulated into a group of four patients of which three had TMA. TMA has been established as a key contributor to toxicity post-HSCT. Yet, the criteria for its definition and diagnostics are only emerging. An international expert group published recommendations for transplantation related TMA (TA-TMA) in 2022 23. The recommendation gives an opportunity to diagnose TA-TMA using clinical and laboratory parameters, without tissue biopsies. The clinical phenotype of the TA-TMA varies, so the recommendation suggests ≥ 4 of the 7 key features to be fulfilled twice within 14 days to set the diagnosis. It is important that all these features are easily followed-up in every patient and the laboratory tests widely available. Jodele et al.25 showed that patients with a moderate risk for TA-TMA may benefit from routine TMA-screening, early TMA-diagnosis, and targeted therapies in a timely manner.

Many risk factors for the development of TA-TMA have been suggested including transplant- (myeloablative conditioning with TBI, the use of HLA-mismatched donors, second HSCT), treatment- (the use of calcineurin inhibitors) or patient-related (age, race, genetics) factors26,27,28,29. The use of TBI has somewhat decreased and that of matched, unrelated donors increased. Yet, the incidence of endotheliopathies (9.5% vs. 15.6%, p = 0.443) appears not to have changed as shown by our two patient cohorts.

The elevation of soluble terminal complement complex (SC5b-9) levels in blood, as an indicator of terminal pathway complement activation, can be seen in pediatric HSCT-TMAs12,30,31,32. Still, only some patients with TMA have elevated SC5b-9 or C3a33,34. Although not needed for diagnosis, they act as indicators of a high-risk of TMA with an ensuing, and increased, risk for multiple organ dysfunction syndrome (MODS) and high non-relapse mortality9,11,23,34. We report one patient with markedly elevated SC5b-9, severe TMA, and death due to toxicity. The other two patients diagnosed with TMA did not have plasma samples for complement activation studies and the one with CLS and VOD (Patient 3P in Fig. 1 / Patient 10 in Fig. 2) had a clear elevation in SC5b-9 at 30 min but not later on. A recent review article sets an SC5b-9 level of ≥ 252 ng/ml on day 28 to have a sensitivity of 100% and specificity of 53 − 51% for later TA-TMA35. Our patient with TMA had his SC5b-9 at 9900 ng/ml on day 32 while the average SC5b-9 level peaked at 714 ng/ml [range 172–9960 ng/ml, SD +- 2120] among the allo- and at 259 ng/ml [range 165–887 ng/ml, SD +- 84] among the auto-HSCT patients. Thus, the very high SC5b-9 level was clearly predictive for a poor prognosis.

High C3a levels reflect generalized complement activation. C3a is very sensitive biomarker for complement activation, but also the least stable of the components. With its short half-life, it reflects ongoing activation at the time of sampling36. The patient with TMA and massive elevation in SC5b-9 did not show an elevated C3a at any point. This may have been due to the difference in kinetics of C3a and SC5b-9. C3a disappears rapidly from circulation, whereas SC5b-9 integrates activation of the whole complement cascade and remains elevated for a longer time than C3a. An elevated C3 has also been postulated as predictive and prognostic for pediatric TA-TMA37. (37). In our study, we did not measure C3 levels, but had several patients (1, 2, 5, 6, 7, 9, 10, 11, 12, 13, 17, 18, 20, and 21) with manifold elevations in C3a with no apparent correlation to the clinical, adverse events during the first 100 days post-transplant.

SC5b-9 is a more stable activation product than C3a and is indicative of the activation of the whole complement cascade. Yet not all TMA leads to systemic complement activation. SC5b-9 in blood and complement activation on cells and endothelium do not always go parallel38,39. The plasma samples in our study were obtained according to our predefined sampling plan and not when adverse events were observed. It is possible that some episodes of complement activation may have been missed, but we describe a reliable trend in the complement activation markers throughout the HSCT. To our knowledge, this is the first publication to describe complement activation peri-transplant.

The limitation to our study is the size and heterogeneity of our cohort. Nevertheless, the cohort well represents the “real-life” setting in a mid-size, European, pediatric SCT unit. Furthermore, due to our predefined sampling times, we may have missed some short-term episodes of complement activation. The strength of our study is the prospective design and standardized handling of the blood samples rendering the results on complement activation reliable. The uniform treatment of the primary malignant disease using the Nordic treatment protocols and the centralization of more than 95% of the pediatric HSCTs in Finland to one, single center, also makes the patient cohort homogenous. Thus, our study cohort is representative of a national, pediatric allo-HSCT population within the given timeframe.

The purpose of the current study was to explore the incidence of acute adverse events and vascular complications as well as perturbations in the complement system during a pediatric HSCT process. Terminal complement activation seems to be only a part of the TA-TMA story. We reported ten patients with at least 50% elevation in SC5b-9 during the first 6 months post-transplant. We also reported C3a level elevations during the HSCT process, but neither clearly correlated with the clinical adverse events. This is a comforting sign from the point of view of potential inflammatory reactions but could also indicate a role for other mechanisms of systemic reactions or local tissue injury as causes for the adverse effects related to HSCT.