A total of 19 adult patients who underwent allo-HSCT between January 2019 and January 2020 in Son Espases Hospital (HUSE) were recruited for this study. The study was conducted according to the ethical guidelines of the 1975 Declaration of Helsinki and approved by CEIC (Balearic Islands Clinical Research Ethics Committee). Written informed consent to have their clinical information published in medical or scientific journals was obtained from all subjects.

Different baseline diagnoses and indications for allo-HSCT were enrolled in this real-world study. All the diagnoses were evaluated according to the World Health Organization 2016 criteria (Table 1).

Peripheral blood from patients was collected in heparin (for plasma extraction) and BD vacutainer (serum extraction tubes) and immediately processed, upon the arrival of the patients to the center before the start of the conditioning regimen (D-1), on infusion (D0) and every 5 days afterward (D5, D10, D15, D20 and D30), if available. A total of 107 serum and plasma samples were collected and stored at -70°C until further use. Demographic and clinical data were recorded, as well as, routine laboratory tests including coagulation and inflammatory markers like ferritin or C-reactive protein (CRP). These variables were used to calculate EASIX and m-EASIX prognostic scores.

Neutrophils were isolated from the peripheral blood of healthy donors using dextran sedimentation. Briefly, peripheral blood collected in heparin tubes was incubated with 6% dextran for 45 min at 37°C in a 5% CO2 atmosphere. After sedimentation, neutrophil-rich supernatant at the upper layer was collected in a sterile tube, centrifuged and washed with RPMI-1640 medium. Red blood cell (RBC) lysis was performed using an RBC lysis buffer (1 mL of physiological saline solution plus 3 mL of distilled water) and vortex for 30 seconds. Afterward, 1 mL of 3.5% saline solution was added and centrifuged for 5 minutes. This process was repeated three times.

Isolated neutrophils were suspended in RPMI-1640 medium supplemented with 10% heat-inactivated fetal calf serum, glutamine (2 mM) and antibiotics (penicillin and streptomycin). Samples were filtered to remove cell accumulations using 50 µm pore size filters. Then, cells were cultured (2×10⁶ cells/mL) in Millicell EZ SLIDE glass (Millipore, Merck KGaA) and each well was incubated with different patients’ plasma samples. Negative and positive controls, stimulated with phorbol 12-myristate 13-acetate (PMA; 75 ng/mL) and ionomycin (1.5 ug/mL) (both from Sigma-Aldrich), were also prepared in each experiment (Figure 1A). Cultures were maintained for 4 hours at 37°C in a 5% CO2 atmosphere.

A staining protocol was used to evaluate NETosis, in vitro, by using fluorescent microscopy (Nikon Eclipse E400). Briefly, after incubation, neutrophils were fixated with 2% of paraformaldehyde and stained with Iodide solution (1.0 µg/ml Merck Life Science S.L.) and anti-myeloperoxidase (MPO)-FITC (Milteny Biotec, REAfinity™, USA) for extracellular DNA and MPO identification. Immunofluorescence images were independently interpreted by two observers, who were blinded to the sample identity and any discrepancy was solved by consensus. For each sample, 10 fields were counted and the number of visible NETs in each one was registered. The median count from both observers was used for the analyses. The fluorescent microscopic evaluation and quantification of positive NETs were carried out at 20x magnification. Inconclusive results were solved by 110x magnification with immersion oil. Neutrophils were considered positive, for NETs formation, according to single-cell morphology (nuclear loss of lobules and ejection of DNA from the cell membrane) together with positive double immunolabeling of NET structures using anti-MPO and propidium iodide solution (co-localization of DNA and MPO) (Figure 1B).

The Human Inflammatory Cytokine kit BD Cytometric Bead Array assay (BD Biosciences, U.S.A) was used to quantify serum concentrations of IL‐6, IL‐8, IL‐10, TNF-α and IL-12p70 by flow cytometry (FACS Verse BD Biosciences, U.S.A), according to the manufacturer´s instructions.

The serum concentration of Citrullinated histone 3 (H3Cit) was measured using a sandwich enzyme-linked immunosorbent assay (ELISA) developed by Cayman Chemicals (Ann Arbor, MI, USA) according to the manufacturer´s instructions. This assay employed a monoclonal antibody specific for histone H3 citrullinated at R2, R8 and R17 (clone 11D3). The lower limit detection of the assay was 0.1 ng/ml.

The normality of the distribution was assessed using the Shapiro test. Normally distributed variables were expressed as mean (± standard deviation) and non-normally distributed data were presented as median (min-max). T-test and ANOVA tests or their non-parametric counterparts, the Mann-Whitney and Kruskal-Wallis tests, were used to compare differences between two groups or more than two groups, respectively. Correlations were analyzed using the Spearman correlation test. A p-value lower than 0.05 was considered significant.

ROC analysis was used to determine the power of variables to differentiate groups, and the area under the curve was calculated; significant cut-off levels were calculated using a Youden index. Logistic regression analysis was used to compare independent variables. ROC curves were performed using 35 negative and positive controls. Two cut-off values obtained by ROC analysis were used in the study. The cut-off value for NETS positivity was ≥ 0.85 NETs counts per field (sensitivity 100% specificity 96.77%). The second cut-off value used to identify strong NETS positivity was ≥ 2.27 NETs counts per field (sensitivity 100% specificity 100%). The statistical analysis was performed using GraphPad Prism software version 5.0. and R version 4.2.

Table 1 summarizes the main patient and allo-HSCT information. Briefly outlined, the median age of the study cohort at the time of the allo-HSCT was 52.8 years (range 18-69) and 12 (63%) were male and 7 (37%) female. The median number of previous lines of therapy before transplant was 1 (0 to 3). Disease status before allo-HSCT was as follows: complete response (CR) in 8 patients (42%), partial response (PR) in 1 (5%) patient and stable disease (SD) in 10 (53%). Overall, 58% of patients received reduced-intensity conditioning (RIC) allo-HSCT, and most patients received grafts from related and unrelated HLA-matched donors (73%), followed by 9/10 mismatched unrelated donors (16%) and haploidentical donors (10%). A proportion of 18 grafts were obtained from peripheral blood and only 1 from bone marrow.

Table 2 displays the complications presented by patients, divided into two periods: (i) before day 15 after transplant, referred to as early complications (58%), and (ii) during admission (75%). We considered as relevant complications: (i) aGVHD, (ii) infection, (iii) sepsis, (iv) veno-occlusive disease (VOD), hemorrhage (v) and (vi) transplant-associated thrombotic microangiopathy (TR-TMA). We did not consider mucositis as a severe complication, but it was included in the table (Table 2). Table 2 also depicts patients who developed aGVHD (47%) and chronic (c)GVHD (10% of patients).

The most frequent early complication was infection (42%) and aGVHD (10%). During admission, the most frequent associated complications were also infections (68%), of which 1 developed sepsis, followed by aGVHD (16%) (Table 2). Overall, during follow-up and before day 100, 47% of patients presented aGVHD; The grading for these patients was as follows: 2 (grade I), 4 (grade II) and 3 (grade III). After day 100, 1 patient had a moderate overlap GVHD and 1 patient developed a moderate cGVHD (Table 2).

We were interested in evaluating whether plasma samples from allo-HSCT patients were able to induce NETs formation, in vitro, in human neutrophils from healthy controls. We studied the NETs induction capacity (NETsIC) in 107 plasma samples from 19 allo-HSCT patients. Plasma samples from allo-HSCT patients were stronger NETs inducers than plasma from healthy controls (14.4 counts vs 3.5 counts; p<0.001) (Figure 1C).

Then, we evaluated how the NETsIC evolved throughout the transplantation process. To do so, we compared the NETsIC of plasma samples from allo-HSCT patients between the different time points assessed (D-1, D0, D5, D10, D15 and D20) with those taken from healthy controls. Accordingly, the NETsIC of plasma samples from allo-HSCT patients remained significantly elevated compared to controls at all time points studied (D-1, D0, D5, D10, D15 and D20) (Figure 1D). However, regarding allo-HSCT patient samples, similar levels were found between the different time points of transplantation tested (Figure 1D).

If we focused particularly on each time point evaluated, we observed that total NETs counts were highly variable between patients (Figure 1D). Therefore, using ROC analysis we established two cut-off values. The first one (≥8.5 total NETs counts) was used to discriminate between patients whose plasma was able to trigger NETosis (NETs+IC group) and those who did not (NETs-IC group) (Figure 2A). The second cut-off value (≥ 27.2 total NETs counts) was established to identify allo-HSCT samples with a strong NETsIC (Figure 2C). As depicted in Figure 2B, in all time points evaluated over half of the plasma samples were capable of inducing NETosis. Remarkably, at D5 after graft infusion, 73% of the patient´s plasma was able to induce NETosis. These results suggested that most of the allo-HSCT patient’s plasma could induce NETosis throughout the transplantation process. Interestingly, when we examined which samples had a strong NETsIC we found that the maximum was reached at D10 after transplant with 38.8% of positive samples (Figure 2D).

To rule out that treatments previous to the conditioning regimen may have influenced the ability of plasma samples to induce NETosis, the NETsIC was compared between treated and the 4 untreated patients (Table 1), and no statistically significant differences were found (Supplementary Table S1).

Multiple NETs inducers including cytokines such as IL-6 or IL-8 have been described (14, 15), Thus, we quantified serum levels of IL-8, IL-6, IL-10, TNF-α and IL-12p70 in the 107 patient’s serum samples before, during and after allo-HSCT. It is noteworthy, that IL-8 followed by IL-6 experienced the largest increases in allo-HSCT patients during the evaluated period (Figure 3A). Indeed, these two cytokines presented a similar profile: peak levels were reached at D10 after stem cell infusion and returned to pretransplant values, approximately, at D20 after infusion. Specifically, when we performed a longitudinal analysis, the highest differences were found between D-1 and D10 for IL-6 (8.17 pg/ml vs 119.4 pg/ml p<0.001) and IL-8 (16.06 pg/ml vs 168.1 pg/ml p<0.001).

Serum levels of IL-10 increased slightly (p=0.07) and remained slightly elevated until D20. TNF-α and IL-12p70 levels did not increase throughout the studied period (Figure 3A).

Then, we evaluated the potential relationship between NETosis and cytokines levels in allo-HSCT settings (Figures 3B–F). We compared the cytokines levels between patients with positive and negative NETsIC. We only found significantly higher levels of IL-10 before transplant (2.94 pg/ml vs 1.03 pg/ml; p<0.05) and IL-6 (82.5 pg/ml vs 20.2 pg/ml p<0.05) at D5 after graft infusion in NETs+IC group compared to NETs–IC group (Table 3). No differences in IL-8 levels were found between patients with positive and negative NETsIC. Interestingly, as depicted in Figure 3C, IL-8 levels seemed to be higher in the NETs+IC patients group until D5 after transplant but this trend reversed on D10. However, no statistically significant difference was found in IL-8 levels between patients with positive or negative NETsIC. Furthermore, we did not find any correlation between cytokines levels and the NETsIC of the plasma samples from allo-HSCT patients (Figures 4A–C).

In our cohort, cytokine levels were not affected by previous treatments to the conditioning regimen (Supplementary Table S1).

Citrullinated histone 3 (H3Cit) is one of the main components of NETs structures and the evaluation of H3Cit levels is an indirect way of measuring NETosis. We determined H3Cit levels in serum samples from allo-HSCT patients previous to the conditioning regimen (D-1) and at day 5 (D5) post-transplant and correlated them with NETsIC. Higher H3Cit levels were found in the group of patients with a positive NETsIC (NETs+IC group) compared to patients with a negative NETsIC (NETs-IC group) (1.6 ng/mL vs 1.05 ng/mL; p<0.05) (Figure 4D). However, we did not find a correlation between H3Cit levels and NETsIC (p<0.05; r2 0.17; Sy.x 0.57) (Figure 4F). No differences were found when H3Cit levels were evaluated before and after allo-HSCT (Figure 4E).

NETosis has been extensively related to thrombotic mechanisms (16–18). For that reason, we analyzed the coagulation parameters, prothrombin time (PT) and fibrinogen, after ruling out prior anticoagulation treatments in the patients. We did not find differences when PT and fibrinogen levels were compared between NETs+IC and NETs-IC patient groups (Table 4). However, when we evaluated the PT and fibrinogen levels in all the allo-HSCT samples with a strong NETsIC we found higher fibrinogen levels (744 mg/dL vs 598 mg/dL; p<0.05) and a trend toward lower PT (73.2% vs 80.2%; p=0.1) compared to samples with a non-strong NETsIC. Moreover, prior to consolidation treatment (D-1), significantly lower PT was found in patients with strong NETsIC (Table 5).

In our cohort, NETosis was not associated with sex, age, donor type, stem cell source or HLA identity (P > 0.05). We also assessed hematological and serological parameters and correlated them with NETosis. Before conditioning settings (D-1) we found a positive correlation between the EASIX score, a surrogate marker of inflammation and morbimortality, and the NETsIC of allo-HSCT patients samples (p<0.014. r=0.5) (Figure 4G). In fact, the EASIX score was significantly higher in patients with a strong NETsIC (p<0.05) (Table 5). Moreover, the subgroup of patients with a strong NETsIC exhibited a more compromised hematological profile characterized by significantly diminished hemoglobin levels (p<0.05) and reduced platelet counts (p<0.05) (Table 5).

It is well known that conditioning regimens damage host tissues and cause the release of inflammatory mediators. Thus, we also evaluated hematological and serological parameters at D5 after graft infusion. Conversely to D-1, the EASIX score was not significantly correlated with NETsIC. However, the M-EASIX score, which replaces creatinine with CRP, was significantly higher in the NETs+IC group compared to the NETs-IC group (Table 4). In addition, the inflammatory markers CRP and IL-6, as we mentioned above, were higher in the NETs+IC group compared to the NETs-IC group, suggesting a pro-inflammatory state in patients with a positive NETsIC (Tables 3, 4). Furthermore, at D10 after transplant, allo-HSCT patients with a strong NETsIC continued having low platelets counts (p<0.05) as well as higher mean platelet volume (p<0.05) and neutrophil-to-lymphocyte ratio (p<0.05) (Table 5).

Surprisingly, patients who underwent an RIC regimen presented a higher capacity to induce NETosis than patients on other conditioning regimens. Specifically, 78% of plasma samples from patients who underwent non-myeloablative conditioning presented a positive NETsIC (p<0.05) (Table 6).

A substantial proportion of patients showed severe complications after allo-HSCT (Table 2). For that reason, we evaluated hematological, biochemical parameters and NETsIC in patients according to the presence of severe complications (Table 2).

Interestingly, patients who developed early severe complications had, at D5 after transplant, significantly higher IL-6 and IL-8 levels compared to patients without complications in that period (95.91 pg/ml vs 25.02 pg/ml, p<0.01; 205.8 pg/ml vs 36.23 pg/ml; p<0.05; respectively) (Figures 5A, B).

Note that, as depicted in Table 2, only 16% of patients did not show complications during admission.

Despite the limited sample size, we found before (D-1), during (D0) and after transplant (D10 and D15) a higher NETsIC in patients who presented complications during admission compared to patients without them (Table 7 and Figure 5F). Patients with complications during admission also had higher levels of IL-8 and IL-10 (Table 7 and Figures 5D, E). Moreover, previous to the conditioning regimen (D-1) there is a trend toward a higher percentage of patients who developed GVHD either during or after admission in the NETs+IC group (p=0.1). Specifically, 66.7% of the patients who developed GVHD during or after admission (during follow-up in ambulatory care) had a positive NETsIC (Table 6).

Although there were no differences in IL-6 levels (Figure 5C), 2 patients who died during the study, after D5 (septic shock and cerebral bleeding), presented increased levels of IL-6 at D5 after transplant thus, in this small cohort, IL-6 levels correlated with overall mortality (p<0.05) (data not shown).

Regarding to EASIX score, previous to the conditioning regimen (D-1) there is a trend toward higher EASIX score in patients who developed GVHD either during or after admission (p=0.095) (data not shown). Furthermore, when we take into account complications during the patient’s whole admission the EASIX score at D-1 and D0 could predict overall clinically significant complications (Table 7). Interestingly the M-EASIX score, at D0 was able to predict early complications before day 15 (12.3 vs 50.8; p=0.026) but lacked prediction capability for overall complications during admission in all time points (Table 7).