La revue Pediatrics – organe officiel de la American Academy of Pediatrics – a publié le 14 mars sur son site l’étude “Increased Odds of 
Necrotizing Enterocolitis After Transfusion of Red Blood Cells in Premature Infants » (Pediatrics 2011;127:635-641). Traduction : Risque accru d’entérocolite nécrosante après transfusion de globules rouges chez des nourrissons prématurés. (Image tirée de ce site français: article décrivant l’entérocolite nécrosante en général).
David A. Paul et son équipe de Newark et de Philadelphia ont mis en place une étude rétrospective de cohorte incluant 2311 nourrissons prématurés ayant un poids de naissance inférieur à 1.500 grammes et pris en charge dans un seul centre de néonatalogie. Les investigateurs n’ont retenu que les cas d’entérocolite nécrosante (ECN) intervenus dans les 48 heures après le début d’une transfusion de globules rouges stockées dans des poches de perfusion (PRBC : packed red blood cells).
Des extraits du texte complet – réservé aux abonnés –, publiés en fin d’article, complètent l’annonce des résultats bruts par des éléments très intéressants et un rappel bienvenu de la littérature médicale à ce sujet. Des études passées, certes à une échelle moindre, ont déjà montré une association entre transfusion et ECN.
Résultats de l’étude de David Paul et al : 122 nourrissons (5,3%) ont développé une entérocolite nécrosante, et sur les 122 cas, 33 (27%) ont eu cette complication dans les 48 heures après la transfusion. Au total, l’ECN a été constatée dans 33 (1,4%) de toutes les transfusions (2315).
Après ajustement, les nourrissons ayant reçu une transfusion avaient un risque plus élevé ( ) de développer une entérocolite nécrosante par rapport aux nourrissons non transfusés. Les globules rouges transfusées venaient pour la plupart (83%) de donneurs mâles et avaient été prélevées depuis 5 jours en moyenne. Impossible pour les auteurs de dire s’il y a un lien avec les donneurs, même s’ils tentent de donner des explications (voir extraits du texte complet à la fin).
Conclusions: “In our study sample, PRBC transfusion was associated with increased odds of NEC. The rate of NEC after transfusion was 1.4%. From our data we could not determine if PRBC transfusions were part of the causal pathway for NEC or were indicative of other factors that may be causal for NEC.”
Autrement dit, les auteurs ne comprennent pas si les transfusions ont été des co-facteurs dans la chaîne causale menant à l’entérocolite nécrosante ou si les résultats pointent vers d’autres facteurs de causalité pour l’ECN.
Dans le texte complet, réservé aux abonnés, on apprend d’autres éléments qui, s’ils n’entrent pas dans les critères retenus, n’en sont pas moins révélateurs : ainsi, si 27% des ECN sont apparues dans les 48 heures suivant (le début de) la transfusion, 49% supplémentaires d’ECN sont apparues plus de 48 heures après…En outre, de tous les nourrissons ayant fait une entérocolite nécrosante, 80% avaient été transfusés à un moment ou à un autre avant le diagnostic.
Les auteurs s’interrogent sur le mécanisme qui pourrait lier transfusion et ECN : serait-ce la transfusion qui affecte la quantité d’oxygène arrivant aux intestins ? La cause de l’ECN serait-elle un afflux trop faible de sang et d’oxygène ? Mais ils n’excluent pas que l’entérocolite nécrosante puisse être liée aux transfusions par un autre mécanisme.
L’investigateur principal reconnaît que cette étude n’apporte pas de solutions quant à prévenir ce risque accru d’ECN. Une possibilité serait de stopper l’alimentation des bébés pendant la transfusion. Il n’y a pas de consensus en néonatalogie quant à la conduite à tenir dans cette situation.
Le texte donne d’autres pistes d’explication, et c’est pourquoi j’ai choisi de publier quelques extraits, juste après les deux tableaux.
Résultats : figures 3 et 4
Discussion (fragments du texte complet) :
« To date, our study is among the largest conducted to investigate the potential association of NEC and PRBC transfusion.
Our data support the results of the study of Mally and colleagues, who previously described an association of NEC and PRBC transfusions in a smaller, single-center cohort of premature infants.1 Furthermore, in our sample both the rate of NEC and the proportion of NEC after transfusion, 1.4% and 27%, respectively, were similar to the proportions, 0.8% and 35%, described by Mally et al.1 A recent retrospective study of infants with extremely low birth weight demonstrated an increase in neonatal morbidities, including NEC, in infants who received transfusions.2 Furthermore, Christensen et al showed that 35% of NEC that required surgical treatment was preceded by PRBC transfusion,3 and Josephson and colleagues reported that 38% of infants given transfusions received red blood cells within 48 hours of NEC.5 Josephson et al also reported increased odds of NEC after transfusion in infants older than 4 weeks.5 In previous studies in which premature infants were randomly assigned to higher or lower PRBC transfusion thresholds there were no differences in the occurrence of NEC.13,14 Similarly, results of other studies of infants treated with erythropoietin have shown a decrease in PRBC transfusion but no decrease in NEC compared with infants treated with placebo.10,15 None of these studies investigated NEC as the primary outcome.
Our study did not enable us to determine if PRBC transfusion is causal for NEC. There are, however, a number of biologically plausible etiologies for NEC caused by PRBC transfusion. NEC is known to be associated with epithelial injury and an intense intestinal immune response.7 PRBCs have been postulated to lead to an exaggerated intestinal immune response similar to transfusion-associated lung injury (TRALI) observed in adults. Our previous research has shown no systemic cytokine response after PRBC transfusion.
16 However, because intestinal ischemia is part of the pathophysiology of NEC, any inflammatory response after transfusion may be initially limited to the intestine. TRALI can be immune mediated and non–immune mediated.17 Immune-mediated TRALI seems unlikely to be associated with NEC, because more than 80% of PRBC transfusions that were administered before the occurrence of NEC were from male donors and immunemediated TRALI is usually observed after exposure to plasma from female donors.17 Additional study is needed to determine the significance of the male preponderance of blood donors in the PRBC transfusion preceding NEC in our study sample.
PRBC transfusion has been associated with increased morbidity after adult cardiac surgery.18 In addition, donated PRBCs are known to have diminished capacity to donate nitric oxide to the endothelium, which allows microvascular dilation.6 Thus, PRBC transfusion may alter intestinal oxygen delivery, an effect that may lead to NEC. Transfusion of PRBCs ha
s been shown to alter mesenteric artery blood flow.19 In our study population, infants who developed NEC within 48 hours of transfusion and those who developed NEC _48 hours after transfusion had similar hematocrits, and a similar percentage were not receiving oral nutrition at the time of transfusion. These similarities suggest that the cases of NEC that occurred within 48 hours of transfusion were unlikely to have been related to severe anemia or different feeding practices during transfusion. It is also feasible that the early stages of NEC may have preceded transfusion. Subtle early symptoms of NEC, such as apnea and bradycardia, may have influenced the clinical decision to administer transfusion to infants before more fulminant clinical signs of NEC developed.
Our study had a number of important limitations. The retrospective nature of our investigation did not allow us to determine causality of PRBC transfusion and NEC. Because we limited NEC diagnosis to Bell’s stage 2 or greater, it is possible, although unlikely, that some mild cases of NEC were missed, which may have led to an underestimation of the occurrence of NEC after transfusion. We assessed the temporal association of NEC and PRBCs on the basis of the established criteria of Mally et al.1 It is possible that infants may have developed some subtle signs of NEC before 48 hours after PRBC transfusion but did not show overt clinical manifestations until after the 48 hours after transfusion. Such situations also may have led to an underestimation of the rate of NEC. Conversely, the pathophysiologic process of NEC may have been subtly initiated before transfusion, but the signs of this process may have been interpreted as symptomatic anemia and prompted the administration of a PRBC transfusion that was considered to have occurred before the onset of NEC and, thus, led to overestimation of the rate of NEC associated with PRBC transfusion.
Conclusions:
Our analyses of data from 1 of the largest study samples to date show an association between PRBC transfusion and NEC in infants with VLBW. Transfusion of PRBCs is frequently performed in the NICU, and NEC in preterm infants has been associated with mortality20 as well as neurodevelopmental disabilities. 21 Our study results add to a growing body of data that confirm the association of PRBC transfusion and NEC; 27% to 38% of NEC cases occur within
48 hours after transfusion.1,3,5 Although we could not determine the mechanism of NEC onset after transfusion, our data do not suggest that causality is based on extremely low hematocrit or transfusion of older blood.
Our data that show a preponderance of male donors are intriguing because they suggest a potential cause, but this finding requires confirmation. In our study sample, infants who developed
NEC after transfusion developed NEC at a later postnatal age but a similar postmenstrual age compared with infants who developed NEC but never received a PRBC transfusion. This finding supports the results of the study by Josephson et al, which showed increased odds of late-onset NEC (in infants older than 4 weeks) after transfusion.5 These data suggest a developmental period during which preterm infants are vulnerable to intestinal injury after transfusion. Because donated PRBCs have been shown to have decreased capacity to supply nitric oxide,6 and mesenteric blood flow has been shown to be altered in newborns after transfusion,19 we speculate that PRBC transfusion may lead to alterations in oxygen delivery to the gastrointestinal microcirculation during this vulnerable period. (…)”
Elena Pasca
Copyright Pharmacritique pour les commentaires
Pharmacritique 