Red blood cell transfusions post diagnosis of necrotizing enterocolitis and the deterioration of necrotizing enterocolitis in full-term and near-term infants: a propensity score adjustment retrospective cohort study

Necrotizing enterocolitis (NEC) remains one of the common and serious gastrointestinal inflammatory diseases in newborn infants, with an estimated incidence of 1–5 per 1000 live births [1], and mortality rates ranging from 20 to 30% among neonates requiring surgery [2]. The spectrum of NEC ranges from a slowly evolving and benign form to a more serious or advanced form. According to the staging system proposed by Bell in the 1970s originally, modified by Walsh and Kliegman subsequently [3], NEC was classified into 3 stages. Stage III NEC cases had a higher rate of mortality, higher cost of hospitalization, and more significant sequelae than those less severe NEC cases [4]. For infants with stage III NEC, some of them have stage III documented within the first 24 to 48 h after the onset of NEC, but most of them were progressed from less severe conditions, such as stage I or II [3]. Therefore, if it were somehow possible to prevent stage I or II NEC from progressing to stage III, the prognosis of NEC would probably be improved. However, little information is available to explain why some less severe NEC episodes are managed successfully with medical intensive care, while certain cases deteriorate to stage III. Previous studies had found that broad-spectrum antibiotics plus anaerobic antimicrobial therapy [5, 6] and probiotics [7] could not prevent the deterioration of NEC. Thus, it is a priority to explore the risk factors for the deterioration of NEC.

Red blood cell transfusion (RBCT), a common intervention in neonatal intensive care units, plays a controversial and doubtful role in the treatment of NEC. Several observational studies demonstrated that a positive association was observed between RBCT and the subsequent development of NEC [8, 9]. However, no association [10, 11], even a negative association [12, 13] between RBCT and the subsequent development of NEC were observed in other studies. In clinical practice, RBCT was an alternative strategy for treating evolving NEC infants with anemia. Beena G. et al. found that infants were more likely to receive RBCT after diagnosis of NEC than before, and approximately 90% premature infants with NEC received RBCT after diagnosis of NEC [12]. What a effect does RBCT have on the deterioration of NEC is unclear. The main literature about the associations between NEC and RBCT focused on premature infants [10, 12], and premature infants presented with different patterns of disease and have different outcomes compared with full term infant [14]. It suggests NEC in preterm infants may be a different clinical entity from the full term infants. In addition, the previous literature [10, 12] about the associations between NEC and RBCT focused on association between RBCT and the onset of NEC instead of the deterioration of NEC. Considering most infants with stage I NEC are not really NEC, but much more probability of feeding intolerance, to minimize ascertainment bias from feeding intolerance, we focused on the near-term and full term infants (gestational age ≥ 34 weeks and ≤ 42 weeks) with stage II NEC to explore the association between RBCT and the deterioration of NEC in the present study.

Of the 273 eligible near-term and full-term newborns with NEC, 8 infants with intestinal perforation or intestinal malformation (aproctia, intestinal atresia, Hirschsprung’s disease), 15 infants with Bell’s stage III NEC at the onset of the disease and 8 infants with incomplete information were excluded. Finally, 242 infants with Bell’s stage II NEC met the inclusion criterion (seen in Fig. 1).

Overall, of the 242 infants enrolled, 60 (24.7%) infants had a history of RBCT post NEC diagnosis, and 40 (16.5%) infants deteriorated from Bell’s stage II to stage III. The demographic features of all infants were shown in Table 1. The infants with RBCT post NEC had a higher rate of caesarean delivery (78.3% vs 62.1%, P < 0.05), an elder of age at diagnosis of NEC (10.5 (3.25–26.75) vs 8 (2–16), P < 0.05), and a lower birth weight (2763 g ± 614 g vs 2989 g ± 580 g, P < 0.05).

As shown in Table 2, the infants with RBCT post NEC had a higher risk in sepsis (55.0% vs 21.4%, P = 0.000), and no significant difference was observed in other complications.

The infants with RBCT post NEC diagnosis had the a higher rate in receiving transfusion of fresh frozen plasma (25% vs 2.2%, P = 0.000), platelet (11.7% vs 1.6%, P = 0.003), cryoprecipitate (16.7% vs 4.4%, P = 0.004) and mechanical ventilation (6.7% vs 0, P = 0.001) post NEC diagnosis. Furthermore, they were also in the a higher risk of blood transfusion prior to NEC diagnosis (16.7% vs 2.7%, P = 0.000), intravenous albumin (68.3% vs 29.7%, P = 0.000) or immunoglobulin (15.0% vs 3.8%, P = 0.007), anaerobic antimicrobial therapy (90.0% vs 62.6%, P = 0.000), vasoactive agents support (41.7% vs 17.0%, P = 0.000), and had a longer duration for broad spectrum antibiotics usage (P = 0.000), cessation of enteral feeding (P = 0.000), and gastrointestinal decompression (P = 0.000). Infants with CRP > 8 mg/L within 24 h of NEC diagnosis was significantly more in the group of infants with RBCT post NEC diagnosis than that in those without RBCT post NEC diagnosis and the lowest Hct post NEC diagnosis and prior NEC deterioration was significantly lower in the group of infants with RBCT post NEC diagnosis than that in those without RBCT post NEC diagnosis (Table 3).

RBCT is a common treatment to increase circulatory haemoglobin, improve tissue oxygenation, and maintain the stable of hemodynamics and the desired oxygenation [20]. Due to the iatrogenic blood loss, the small blood volume and immature hematopoietic system [21], newborn infants with severe complications received blood transfusion more frequently. It was reported that a transfusion rate of 13.4% in a population of patients, which included term neonates; 27.8% of the transfused patients were with a gestational age more than 34 weeks [22], 50 to 94% in infants with birth weight less than 1500 g and up to 95% in infants with birth weight less than 1000 g [23, 24]. Since 1980s, more and more evidences showed that RBCT played a controversial and doubtful role in the treatment of NEC [8‐13, 25]. Some observational studies had reported that RBCT was associated with subsequent onset of NEC [8, 9]. However, no negative association between RBCT and the onset of NEC was proved in other studies [10, 11], even a protective effect was observed [12, 13]. Given that RBCT played a controversial and doubtful role in the onset of NEC, the neonatologist are facing an important dilemma in clinical practice. Firstly focusing on the relationship between RBCT post NEC diagnosis and deterioration of NEC on near-term and full term infants, the findings of the present study seem to indicate that RBCT post NEC diagnosis was associated with NEC deterioration from stage II to stage III.

Beena G and colleagues have found that infants were more likely to receive RBCT after diagnosed NEC than before, and approximately 90% NEC infants, with gestational age 23–32 weeks, received RBCT after diagnosed NEC [12]. Consistent with the previous study [12], infants were more likely to receive RBCT after diagnosis of NEC than before (24.8% versus 6.2%, p = 0.000) in the present study. However, the rates for RBCT after and before diagnosis of NEC in the present study are much lower than that in previous study (69% versus 90% after diagnosis and 6.2% versus 24.8% before diagnosis), and it can be explained by the higher gestational age and the larger birth weight of the near-term and full term infants in the present study.

Although a positive association between RBCT post NEC diagnosis and the deterioration of NEC was observed in this retrospective study, the underlying pathogenic mechanisms of this association were unknown. To some extent, the mechanisms which have been used to explain the transfusion associated NEC (TANEC) might be suitable for explaining the deterioration of NEC after transfusion. In most cases, when hemoglobin or hematocrit value fell below a certain threshold, a transfusion was invariably given to increase circulatory haemoglobin and improve tissue oxygenation. It was quite possible that infants with a history of RBCT post diagnosis of NEC were with a parallel diagnosis of severe anemia post diagnosis of NEC. And in our study, the lowest Hct value in the group of infants with RBCT post NEC diagnosis was significantly lower than that in those without RBCT post NEC diagnosis (Table 3). Could anemia per se be the reason for deterioration of NEC? Anemia can lead to poor perfusion and oxygen delivery to the intestine, resulting in a relative intestinal hypoxia and subsequent mucosal injury [20]. Animal anemia model had already shown that anemia could impair gut blood flow [26]. In anemia infants, intestinal injury was also detected prior to the RBCT [27], which established that infants with anemia already had harbored intestinal mucosal injury. What’s more, anemia could impair the normal maturation of vascular auto-regulation in the intestine, predisposing to intestinal ischemic injury [28, 29]. In murine model of anemia, inflammatory state in the intestinal mucosa with macrophage infiltration was found [30]. All these may possibly constitute an injury relevant to the deterioration of NEC, and anemia might be one possible mechanism for deterioration of NEC after RBCT.

Besides anemia, RBCT itself may also be the involved mechanism for deterioration of NEC. RBCT can alter blood flow to the bowel and augment intestinal injury theoretically [20]. During RBCT, there existed a sudden increasement of viscosity in the circulating blood, which had been proven to associate with the intestinal injury [20, 31]. And the affinity of RBCs for oxygen increases during storage, which shifts the recipient’s oxygen dissociation curve to the left, predisposing recipients to ischemia [32, 33]. What’s more, erythrocyte nitric oxide bioactivity losses rapidly during storage, which impairs the ability of RBCs to effect hypoxic vasodilation. Thus RBCs transfused to patients would act as a nitric oxide sink, resulting in vasoconstriction and ischemic insult to the intestine [34]. In principle, RBCTs are given to improve oxygen delivery to tissues, but, paradoxically, RBCTs fail to do it [32, 35], and may even make it worse [33, 36]. Another mechanism of TANEC, which centers on immunological injury, is also the plausible mechanism for intestinal injury post RBCT. Transfusion-related immunomodulation (TRIM) is speculated to be that RBCT trigger immune cell activation and related immunomodulation, whose likely manifestation is increasement of proinflamamatory cytokine production and endothelial activation [37]. In murine model of transfusion-associated NEC, MohanKumar et al. found that RBCT would activate intestinal mucosa cells, which were infiltrated by the macrophage, via a TLR4-mediated mechanism to cause bowel injury [30]. Keir et al. evaluated changes in circulating proinflammatory cytokine concentration and related downstream pathway in response to RBCT, assessed the potential proinflammatory effects for RBCs to be biologically active, and supported the existence of TRIM [38]. TRIM encompasses not only adverse proinflammatory and immunosuppressive responses but also the whole spectrum of posttransfusion effects on organs and tissues [37, 38], which may partly explain the association between RBCT and the pathogenesis of several high neonatal morbidities affecting the brain [39‐41], eyes [40, 41], lung [41, 42], and gut [40, 41]. Some other studies had also confirmed that proinflammatory cytokine concentration increased in infants who underwent a RBCT [21, 43]. Cytokines have been confirmed to play an important role in mediating intestinal inflammation and injury in the pathogenesis of NEC [44, 45]. TRIM is the mechanism to explain how RBCT can induce an increase in cytokine serum level and to partly explain the association between RBCT and the deterioration of NEC. Although we regarded the foregoing as different possible pathogenic mechanisms, they might not be mutually exclusive; conceivably, elements of these mechanisms could operate at the same time.

There were some limitations existing in the present study, such as the inherent errors and bias of retrospective studies. This retrospective cohort study was also limited by its cohort design, failure to take some unreported clinical risk factors for NEC and/or exposures to RBCTs into consideration, and there still existed some unmeasured confounders which might affect the propensity score. In order to minimize ascertainment bias, we identified only stage II NEC cases in our hospital databases, and excluded all suspected NEC cases, resulting in a relatively small number of NEC cases included in this study. Because of the small sample, we did not further divide the included cases into the subgroups of Bell’s stage. We were unable to define the precise time of onset and diagnosis of NEC. The time of onset of NEC, defined as the day of the first presentation of non-specific signs and symptoms, and the time of diagnosis of NEC, defined as the day of the abdominal X-ray or the ultrasound results fit the diagnostic criteria for stage II NEC, may be made several hours after NEC occured and confirmed. Absence of a standardized feeding protocol during the study period may have influenced the results of our study [46, 47]. As no breast milk bank in our hospital, the formula was used for all infants during hospitalization. And lack of early colostrums feeding would contribute to the deterioration of NEC [4], we did not know whether formula feeding could have an influence on the deterioration of NEC in our study or not. Most importantly, in this study, the association between RBCT post NEC diagnosis and the deterioration of NEC, even with a statistically significant difference, does not prove a cause and effect relationship. Thus, the antecedent RBCT might have no pathogenic role on the deterioration of confirmed NEC. Rather, the RBCT might be an epiphenomenologic marker of deterioration of NEC. In order to determine whether RBCT cause the deterioration of NEC will require a different experimental approach.

In conclusion, NEC infants who required RBCT post NEC diagnosis were more likely to associate with the deterioration from stage II to III in full-term and near-term infants.