Platelet and red cell transfusions for neonates: lifesavers or Trojan horses?

KEYWORDS:

• Anemia
• thrombocytopenia
• transfusion
• neonate
• preterms

1. Introduction

Red blood cell (RBC) and platelet transfusions are frequently administered to neonates. Up to 90% of extremely low birth weight infants and 58% of preterm infants (<32 weeks gestational age (GA)) receive one or more RBC transfusions [1]. Severe thrombocytopenia (platelet count <50x10⁹/L) occurs in approximately 5–10% of preterm neonates, of which over 75% are treated with platelet transfusions [2]. Unfortunately, due to lack of high-quality evidence, neonatal transfusion guidelines are largely based on expert opinion. In recent years, neonatal transfusion research has gained momentum, challenging preconceptions regarding transfusion efficacy by showing that the patient benefits are oftentimes limited. In line with data from studies in adults and contrary to commonly held beliefs, recent neonatal trials suggest that transfusions may indeed have inadvertent harmful effects, such as increased risk of bleeding or mortality [3,4]. Additionally, adult studies have further confirmed that transfusion thresholds for both RBC and platelet transfusions can be lowered without increasing risk of adverse outcomes. Whilst the latter results will urgently need replicating in neonatal patients, findings like these underline the urgent need for high-quality studies to support neonatal transfusion guidelines. Recently, two large-scale neonatal transfusion trials have been completed which we will discuss in the context of current evidence and guidelines. We will further describe research priority areas and discuss recommendations for clinical practice.

2. Current evidence

2.1. Red blood cell transfusions

Neonatal anemia is a universal challenge, encountered on almost every neonatal intensive care unit (NICU) around the world. Commonly, iatrogenic blood loss is the main culprit: in the first 6 weeks of life, iatrogenic blood loss has been reported to range from 11 to 22 ml/kg per week, equivalent to 15–30% of the total circulating blood volume per week [5]. Additionally, the preterm infant’s reduced hematopoietic capacity and comorbidities such as sepsis, or medication, result in even lower hemoglobin levels. RBC transfusions are generally given based on predefined hemoglobin or hematocrit thresholds, which also vary depending on comorbidity, age and ventilatory status of the neonate. Evidence supporting neonatal RBC transfusion guidelines is limited. A 2011 Cochrane systematic review included four randomized controlled trials comparing a liberal and a restrictive RBC transfusion threshold. The authors concluded that a liberal transfusion threshold was not superior over a restrictive threshold with regards to the primary outcomes of death or major morbidities. However, neurodevelopmental follow up data were not available, thus limiting the interpretation from these studies to the duration of hospitalization only [6]. Recently, with completion of the ETTNO trial (‘Effects of Transfusion Thresholds on Neurocognitive Outcome of Extremely Low Birth-Weight Infants) more knowledge on neurodevelopmental outcomes following RBC transfusions has become available. In this study, Franz et al. studied 920 neonates with birthweights between 400 and 999 g, randomized to either a liberal and restrictive hematocrit-based RBC transfusion arm. The study results have been presented at the EAPS congress in Paris in November 2018 but are not yet published, and therefore preliminary. The primary outcome was a composite of death or neurodevelopmental impairment at 2 years corrected age. In the liberal arm, 200 out of 450 (44%) neonates developed the primary outcome, versus 205/478 (43%) in the restrictive arm (odds ratio (OR) 0.95 (95% confidence interval (CI) 0.72–1.25)). Secondary outcomes such as necrotizing enterocolitis (NEC), bronchopulmonary dysplasia (BPD) and retinopathy of prematurity (ROP) did not differ between the two groups. Another large randomized controlled trial (TOP: Transfusion Of Prematures, NCT01702805) with a similar design is currently ongoing and will provide further information on the effects of liberal versus restrictive RBC thresholds in preterm neonates. Results are expected in the course of the coming years. While awaiting these results, current evidence continues to suggest that restrictive RBC thresholds are appropriate in extremely low birth weight neonates [7].

2.2. Platelet transfusions

Thrombocytopenia occurs in 5–10% of all preterm neonates in the NICU. Whilst early onset thrombocytopenia (within 3 days after birth) is often associated with intrauterine growth restriction and often resolves within days, late onset thrombocytopenia is most often seen in the context of sepsis or inflammatory processes, such as NEC. Interestingly, widespread international and inter-hospital variation exists with regards to prophylactic platelet transfusion thresholds, resulting in thresholds ranging from 20x10⁹/L to 100x10⁹/L [8,9]. Fustolo-Gunnink et al. have recently performed a systematic review to assess whether platelet count was causally associated with bleeding risk and whether prophylactic platelet transfusions reduce bleeding risk in preterm neonates [10]. Summarizing the evidence from six studies (of which only one randomized trial) they concluded that there is insufficient evidence to prove a causal association between platelet count and risk of bleeding. But, more importantly, none of the included studies showed reduction of bleeding risk associated with prophylactic platelet transfusions. In fact, the results suggest the opposite, that platelet transfusions might be associated with increased risk of bleeding. This disturbing finding was confirmed by the results of the to date largest randomized controlled trial in this area, the Platelets for Neonatal Transfusion (PlaNet-2) trial [4]. In this trial, 660 neonates with gestational age <34 weeks and severe thrombocytopenia were randomized between a 25x10⁹/L and a 50x10⁹/L prophylactic platelet transfusion threshold. The primary outcome was the incidence of major bleeding and/or mortality within 28 days after randomization. This primary outcome occurred in 19% of neonates in the restrictive arm versus 26% in the liberal arm (OR 1.57 (95% CI 1.06–2.32)). These results indicate for one that platelet transfusions appear to be harmful, and secondly, they support the use of a 25x10⁹/L platelet transfusion threshold in preterm neonates.

3. Paradigm shift

The results of the aforementioned studies, as well as recent developments in adult hematology, challenge the prevailing dogmas on the benefits of transfusion and indicate a paradigm shift. Evidently, we should always focus on ‘giving the right blood products to the rights patients’ and there is certainly room for improvement there. In addition, the previously widely accepted adage of ‘better safe than sorry’ may soon be replaced by a less proactive ‘less is more’ prerogative. However, at present, there remains a lot of controversy and disagreement with regards to this change. Why should clinicians accept the paradigm shift and transfuse less? We propose the following arguments:

1. There are several plausible hypotheses that could explain why RBC and platelet transfusions might not be effective or could cause harm. For example:

   1. Platelets possess previously unknown inflammatory and immunomodulatory properties [11].

   2. Red blood cell transfusions in neonates are associated with production of inflammatory cytokines and immunoactivation of the endothelium, indicating possible transfusion-related immunomodulation (TRIM). TRIM is associated with several major comorbidities in adults, but studies in neonates are lacking [12].

   3. Neonatal transfusions are large volume transfusions when compared to adult transfusions: a 15 ml/kg transfusion equals 17% of the total circulating volume of a one-kilogram preterm neonate. This is important, as significant changes in cerebral blood flow and cerebral artery pressure are associated with increased risk of intracranial bleeding in neonates [13].

   4. There is a high likelihood of a developmental mismatch between transfused adult platelets and the immature neonatal hemostatic system [14].

(B) The two recently completed trials provide the best available evidence, despite inevitable limitations. For example, the PlaNeT-2 trial has been criticized for including few neonates with early onset thrombocytopenia. Therefore, some clinicians hesitate to use the 25x10⁹/L threshold for early onset thrombocytopenia. However, there is undisputed proof of harm in the overall PlaNeT-2 population and no solid indication that this effect is different in subgroups in this or other studies. Therefore, we advocate for a 25x10⁹/L threshold even in early onset thrombocytopenia, perhaps while awaiting the results of additional trials to confirm these results in early onset thrombocytopenia. As for RBC transfusions, we are currently awaiting the full publication of the ETTNO trial results, so we cannot presently discuss any limitations of this trial or make recommendations on RBC transfusions based on the preliminary results alone.

(C) In adults, trials have similarly related platelet transfusions to recipient harm, thus suggesting restrictive RBC and platelet transfusion thresholds [3,15].

(D) Transfusion related adverse events are likely to be underreported because they are not easily recognized as such. And, even while underreported, they appear to be more common in neonates than in children or adults [16].

(E) Several studies have suggested associations between RBC or platelet transfusions and other adverse outcomes. In general, transfusions are associated with a potential risk of infectious disease transmission (especially hepatitis), transfusion-related lung injury (TRALI) and transfusion-associated circulatory overload (TACO). In preterm neonates, RBC transfusions have been associated with increased rates of intraventricular hemorrhage (IVH), NEC, ROP and iron overload [7]. In addition, the PlaNeT-2 trial showed increased rates of BPD in the high threshold platelet transfusion arm. Though confirmation of these results in additional studies is needed, in the context of the recent trials these associations do warrant caution when transfusing preterm neonates.

4. How can we improve neonatal transfusion guidelines?

We urgently need new, large-randomized controlled trials to answer the current plethora of questions regarding transfusion thresholds, patient harm, adequate preventive measures, to name but a few.

In addition, acknowledging the financial and professional impact that setting up new trials has in times of limited health-care resources, our focus should be on large, high-quality observational studies. When these are well performed and of substantial size, one may draw careful causal conclusions, or at least use these studies to guide the design and direction of new-randomized trials [17]. Furthermore, we need to set up so-called ‘vein-to-vein’ databases, with which donor and transfusion product data can be directly linked to recipient data [18].

One of the main issues with neonatal transfusion research is the lack of well-validated efficacy parameters. RBC transfusions are given with the aim to improve tissue oxygenation, but we lack a good parameter to assess this outcome. The same applies to platelet transfusions, which are given under the assumption that a platelet transfusion will restore inadequate primary hemostasis due to thrombocytopenia. We do not have evidence to support this assumption, as we have no accurate way of measuring primary hemostasis in preterm neonates. Therefore, we need studies in which these parameters are identified and validated.

Ultimately, the decision to transfuse should be individualized. Neonates should not be transfused based on hemoglobin or hematocrit levels or platelet counts alone, but other risk factors for adverse outcomes of anemia or thrombocytopenia should be taken into account as well. For example, a recently published dynamic prediction model for major bleeding in preterm neonates with thrombocytopenia, once externally validated, could be used to define individualized platelet transfusion thresholds in a subsequent-randomized controlled trial [19].

5. Prevention is always better than cure, but can we prevent anemia and thrombocytopenia?

Yes, we can. As a preventative measure to counteract anemia, the WHO recommends delayed cord clamping by at least 30–60 s and a recently published systematic review supports this approach in preterm neonates [20,21]. Providers of neonatal care have a responsibility to reduce iatrogenic blood loss by avoiding or minimizing often unnecessary laboratory testing [22]. Several recent studies have shown that this can result in significant reduction of the number of RBC transfusions needed in neonates [23]. We recently performed an observational cohort study in 20 extreme preterm infants (<28 weeks) admitted to our NICU and recorded the amount of blood drawn for laboratory testing during the first 28 days of life, as well as the need for RBC transfusions (unpublished data, manuscript submitted). The median volume of phlebotomy loss was 24.2 ml/kg and the median volume of red blood cell transfusions was 30 ml/kg during the same time period, which was almost equal to the amount of iatrogenic blood loss (Figure 1). Decreasing the frequency and amount of phlebotomy loss may thus save substantial transfusions and complications. Unfortunately, methods to reduce phlebotomy losses using small volume laboratory measurements and point-of-care blood sample analysis have not yet been widely implemented in clinical practice [24]. More studies are needed to improve the technique and stimulate implementation. Looking ahead, prophylactic erythropoietin treatment is a promising strategy but additional studies are needed before clinical implementation can be advised [25]. Lastly, development of and adherence to transfusion guidelines will result in consistent and comparable transfusion practices among NICUs, and may reduce the number of (unnecessary or out-of-protocol) transfusions [26].

Figure 1. Median cumulative iatrogenic blood loss in ml/kg birth weight during the first 28 days of life per week of gestational age.

6. Recommendations for clinical practice

As a first step, neonatologists and hematologists should be made aware of the evidence behind the recent paradigm shift in neonatal hematology, resulting in a ‘Less is more’ attitude. Recognizing that high-quality care can be provided without frequent high-volume blood tests, and that transfusions are associated with harm, will alert care providers to reconsider current practice. Based on the currently best available evidence, we recommend restrictive RBC transfusion guidelines and a prophylactic platelet transfusion threshold of 25x10⁹/L. In addition, several prophylactic measures should be implemented into routine care: in agreement with existing guidelines, delayed cord clamping for at least 60 s should be performed in stable newborn infants. Furthermore, the number of phlebotomies should be limited wherever possible. New, randomized controlled multicenter trials are essential to test other measures for improving the care for neonates suffering from anemia or thrombocytopenia. Lastly, it may also be worth mentioning that common transfusion service policies of ignoring ABO for platelet transfusion may be contributing to the increased bleeding seen in these fragile patients. In adults, ABO mismatched platelet transfusions are associated with a myriad of adverse outcomes [27]. Thus, this is the time for international research initiatives to join forces and expertise to improve our knowledge and inform evidence-based neonatal transfusion guidelines.

Declaration of interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

This paper was not funded.