Study recruitment started on December 1, 2021, and the first patient was enrolled on December 17, 2021. Patients were recruited in NICUs of 4 tertiary Italian hospitals and CB-RBC units were collected by 6 public CB banks. By August 31, 2023, 58 patients were enrolled and completed their follow-up: 14 died without reaching the primary endpoint (40 weeks of PMA or discharge, whichever occurred first), whereas 44 reached the primary endpoint and were evaluated for ROP (Fig. 1A). Figure 1B shows the patients’ distribution according to the types of blood products received and the two sets of analysis. Table 1 illustrates clinical and laboratory characteristics of the 58 patients: 28 were assigned to the arm A (A-RBCs) and 30 to the arm B (CB-RBCs). The arms were well balanced for prenatal and neonatal characteristics, hematological parameters at birth, and follow-up duration. The overall mortality was 24.1% (95% CI 15.0-36.6) and the mortality rate was similar in the two arms (Table 1). Moreover. the mortality rate was comparable among groups regardless the type of transfusions received [31.3% (95% CI 14.2–55.6) in non-transfused patients, 29.2% (95% CI 14.9–49.2) in neonates receiving only A-RBCs, 25.0% (95% CI 4.4–59.1) in those receiving only CB-RBCs, and 0 (95% CI 0-27.8) in those receiving both A-RBCs and CB-RBCs, p = 0.235)].
Adverse events and imputability of previous transfusions.
During the study period, 336 adverse events were recorded: 165 occurred in the arm A and 171 in the arm B (Table 2). Individual patients could have multiple different types of events or multiple episodes of the same events. There was no difference between arms regarding the incidence of any of the events listed (Table 2). The rate of adverse events was also assessed in the “per protocol” setting (eTable 1), comparing patients enrolled according to the type of blood products received before the PMA of 32 weeks (Fig. 1B). Patients receiving only A-RBCs or both types of RBCs experienced more adverse events than non-transfused neonates or those transfused with only CB-RBCs (p 0.014). A slightly higher incidence of surgery was observed in patients receiving both RBC types (p = 0.052), suggesting a particular complexity in the clinical course of these neonates. The severity of various adverse events was comparable between arms (Table 2). Nevertheless, when evaluated in the “per-protocol” setting, neonates receiving only A-RBCs or both types of RBCs tended to exhibit more severe forms of bradycardia (p = 0.002), pulmonary hypertension (p = 0.018), and hsPDA (p = 0.040) than non-transfused neonates or neonates receiving only CB-RBCs. In total, 25 adverse events were reported as fatal, with more than one event resulting in death: sepsis (6), bradycardia (6), acute renal failure (3), pulmonary hemorrhage (3), apnea (2), pneumonia (1), pneumothorax (1), NEC (1), coagulopathy (1), and tachyarrhythmia (1). There was no difference between arms A and B regarding the incidence of fatal adverse events.
The imputability of adverse events to previous transfusions was scored in 285 out of the 336 episodes. No adverse events were definitely attributable to previous transfusions, whereas 8 cases were reported as possible and five as likely connected to transfusions (eFigure 1). The patient charts were individually reviewed to determine which types of RBC products were involved. The transfusions preceding the adverse events always consisted of A-RBCs, including an IVH in a patient in the arm B who received also A-RBCs. Taken together, these results confirm that CB-RBC transfusions in ELGANs are at least as safe as standard RBC transfusions and do not increase the incidence or severity of clinical complications arising during the clinical course of these patients.
Transfusion needs, efficacy, and post-transfusion biochemical parameters.
Forty-two of 58 patients (72.4%) required transfusions: 20 (71.4%) in the control arm and 22 (73.3%) in the experimental arm (p = 0.990). A total of 153 RBC transfusions were administered, 72 in arm A and 81 in arm B, with similar transfusion needs in the two groups (3.0, IQR 1.0-5.7 and 3.0, IQR 1.0–7.0 RBC units in arm A and B, and respectively, p = 0.867). Among 22 transfused patients in arm B, 8 received exclusively CB-RBCs, 10 both RBC types, and 4 exclusively A-RBCs (Fig. 1B and eTable 2). PMA at first transfusion, indications to transfuse, and intervals between consecutive transfusions were also similar (eTable 2). In terms of storage duration, RBC content, transfused dose, and effects on hematological and biochemical parameters, CB-RBC units were significantly older than A-RBC units (3 days, IQR 2–5 and 8 days, IQR 6–12, in A-RBCs and CB-RBCs, respectively, p < 0.001), had a lower Hct (61.6%, IQR 5.7–72.2 and 57.1%, IQR 53.0-62.3, in A-RBCs and CB-RBCs, respectively, p = 0.009), and elicited a lower hematocrit increment (Δ Hct, 12.0, IQR 9.1–16.5, and 10.4 IQR 7.5–14.0, in A-RBCs and CB-RBCs, respectively, p = 0.038). However, we found no correlation between ΔHct and storage duration (either with CB-RBCs or A-RBCs), so that the lower Hct increment after CB-RBCs should likely to be ascribed to the lower RBC content of these units. Nonetheless, the time interval to the next transfusion event after CB-RBC transfusions was similar to that after A-RBC transfusions. Moreover, serum potassium and lactate levels, as well as pH values, recorded within 24 hours after CB-RBC or A-RBC transfusions were comparable (eTable 3). In total, 230 CB units were processed into CB-RBC concentrates during the study period, but only 21.3% were transfused. eTable 4 provides characteristics of CB units at collection, after processing, and the hemolysis rate at the end of the storage.
RBC transfusions and ROP.
The efficacy of CB-RBC transfusions in preventing severe ROP was preliminarily assessed in 44 evaluable patients (Fig. 1B). Twenty-five patients (56.8%, 95%CI 42.2–70.3) had any stage of ROP, 10 (22.7%, 95% CI 12.8–36.9) had severe ROP, and 9 (20.4% 95%CI 11.1–34.5) required treatment. Therapy consisted of one or more administrations of anti-vascular endothelial growth factor (VEGF), combined with laser therapy (six cases) or vitrectomy (one case). All but one patient with severe ROP had received transfusions. The clinical course of the study population was typical for critically ill preterm neonates: 17 (38.6%) had IVH, 10 (22.7%) developed BPD, 4 (9.1%) surgical NEC, and 19 (43.2%) hsPDA. All but two patients experienced sepsis, 22 (50%) required inotropic agents, and all received oxygen therapy. Thirty-six patients (81.8%) needed invasive ventilation and 38 (86.3%) non-invasive ventilation. Erythropoietin was administered to 7 patients (15.9%) and 34 (77.3%) received RBC transfusions before discharge.
According to the arm allocation (intention-to-treat set), 14 (70.0%) out of 20 patients in the arm A and 19 (79.2%) out of 24 in the arm B received transfusions, with similar transfusion needs. Due to CB-RBC unavailability, ten neonates (52.6% of transfused patients) in arm B received both A-RBC and CB-RBC units and three patients (15.7%) exclusively received A-RBC units. The proportions of patients developing ROP (of any stage), severe ROP, or ROP requiring treatment in arm A and arm B were comparable (eTable 5). The two arms also did not differ for any other reported outcomes. Similar results were obtained excluding from the analysis non-transfused patients (data not shown).
Considering the high rate of protocol deviations in arm B, we further analyzed the data according to the per-protocol set. We categorized patients according to the types of RBC units received before the PMA of 32 weeks (treatment period). Thirteen neonates did not receive transfusions, 15 received only A-RBCs, 6 only CB-RBCs, and 10 both types of RBCs (Fig. 1B). Severe ROP occurred in one out of 13 non-transfused patients (7.7%), in 5 of 15 receiving exclusively A-RBCs (30.0%), and in 4 of 10 transfused with both RBC types (40.0%). Severe ROP was not observed among infants receiving only CB-RBCs (p = 0.107).
The AUC of HbF at various weeks of PMA was compared for patients with and without severe ROP. The median AUC was 649 (IQR 462–778) for patients without severe ROP and 498 (IQR 259–597) for those with severe ROP (p = 0.035) (Fig. 2A). The AUC profiles of patients who did, or did not, develop severe ROP differed mainly in the earliest weeks of life. Therefore, we focused on to the type of RBC units received before the PMA of 30 weeks. In total, 16 patients were not transfused, 8 received only CB-RBCs, 15 received only A-RBCs, and 5 received both RBC types. Figure 2B illustrates the AUC of HbF in these groups. Among infants transfused exclusively with CB-RBCs, none developed severe ROP, whereas 2 non-transfused patients (12.5%), 4 patients in the A-RBC group (26.7%), and 4 (80%) receiving both A-RBCs and CB-RBCs did develop this complication (p = 0.005). Considering only transfused patients, 8 of 20 (40.0%) receiving A-RBCs before the PMA of 30 weeks developed severe ROP, with a relative risk of 1.66 (95% CI 1.06–2.20) as compared to those receiving only CB-RBCs (p = 0.040). Table 3 shows that neonates receiving only CB-RBCs and those exposed to A-RBCs exhibited comparable characteristics at birth (GA, birth weight, Apgar, and CRIB scores), transfusion burden, rates of IVH, sepsis, NEC, and BPD. Apart from a slightly shorter duration of oxygen therapy, they experienced similar periods of invasive or non-invasive ventilation and more frequently required inotropic medications (Table 3). Collectively these data suggest that transfusing ELGANs with HbF-enriched RBCs in the first weeks of life may protect them from severe ROP.
To further corroborate these findings, logistic regression analysis was used to explore if A-RBC or CB-RBC transfusions exerted a diverging effect on severe ROP development. The model included severe ROP as the dependent variable and the number of A-RBC and CB-RBC transfusions before 30 week-PMA as independent variables. After adjusting for covariates with recognized effects on ROP (i.e., GA, BW, number of days on oxygen therapy), we found that the amount of A-RBC, but not CB-RBC transfusions significantly predicted severe ROP, with an OR of 1.90 for each unit of A-RBCs transfused (95% CI 1.13–3.17, p = 0.014).