Introduction

Refeeding syndrome (RS) in preterm infants is a condition of metabolic derangements attributable to impaired transplacental nutrient transfer followed by a relative abundance of nutrient availability, usually provided through parenteral nutrition (PN). This commonly occurs in pregnancies complicated by fetal growth restriction (FGR), also called intrauterine growth restriction (IUGR), and it is important to note that infants born with weights considered appropriate for gestational age may still have experienced FGR. Fetal hypoxemia and malnutrition due to placental insufficiency contribute to approximately 70% of newborns with FGR; however, FGR can also occur secondary to genetic disorders or congenital infection. The etiology of placental insufficiency is either maternal disease or abnormal placentation. FGR newborns, whether preterm or full term, are born with weights considered lower than expected based on growth potential and may have obvious absence or decreases of fat and muscle mass. The nutrient dyscrasias most described in full-term newborns with FGR include hypoglycemia and hypocalcemia, although hyperglycemia has also been observed [1]. In preterm infants with FGR receiving PN, a constellation of hypophosphatemia, hypokalemia, hypomagnesemia, and hyperglycemia is described [1, 2]. This pattern of electrolyte and glucose abnormalities is strikingly similar to the adolescent and adult diagnosis of refeeding syndrome (RS) [3]. This Perspective focuses on the pathophysiology and expected biochemical abnormalities of neonatal RS, with a focus on preterm infants, and summarizes recommendations for lab monitoring and nutrition management for these infants.

Physiology of refeeding syndrome

Transport of nutrients including glucose, amino acids, and fatty acids is altered in situations of placental dysfunction associated with FGR [4, 5]. The accumulated alterations in nutrient transport likely trigger the physiologic responses in RS: 1) dependence on alternate mechanisms for nutrient availability such as gluconeogenesis and proteolysis, and 2) minimization of cellular metabolic activity. When this cellular activity is minimized, the intracellular need for constituents of metabolic pathways is decreased. Inactive cells have minimal need for glucose, potassium (K), phosphorus (P), or magnesium (Mg), so these nutrients remain in extracellular compartments including the circulation. Therefore, circulating K, P, and Mg measured in starved patients will be normal or even elevated in blood despite total body levels being low due to poor nutrient delivery [3]. When nutrition is introduced to the starved patient, dormant metabolism is activated with a sudden high intracellular need for P, K, and Mg with a resulting large influx from the extracellular circulation. This rapid extracellular to intracellular influx leads to measurable mineral deficiencies that may be severe.

Biochemical indicators of refeeding syndrome in FGR infants

Both SGA status and FGR are amongst the most consistently identified risk factors for RS in preterm infants and suggest infants who may develop biochemical abnormalities consistent with RS. The development of RS has not consistently differed based on infant sex, although a recent cohort analysis identified a trend toward increased occurrence in males [6]. These are common biochemical findings in RS:

  1. 1.

    Hypophosphatemia, generally defined as serum P < 4 mg/dL, is the hallmark electrolyte derangement of RS and alone may be considered sufficient for diagnosis [7]. Reported incidence of RS ranges between 20 and 90% as primarily studied in very low birthweight (VLBW) populations but also infants with birthweights up to 2500 g [8,9,10,11,12]. This range is a result of variation in defining hypophosphatemia and timing of blood sampling. Studies utilize a range of cutoff values, though, including <3.5 mg/dL and even define severe hypophosphatemia as <2.5 mg/dL [8]. Hypophosphatemia is seen as early as 24 hours after birth and may persist for many days unless corrective measures are taken.

  2. 2.

    Hypokalemia, generally defined as serum K < 3.5 mEq/L, may accompany hypophosphatemia [8, 12, 13]. As with other lab findings, a wide range in incidence is reported, ranging from less than 10% to approximately two-thirds of infants studied [12].

  3. 3.

    Hypercalcemia, which can be defined as an ionized calcium (Ca) >1.5 mmol/L or a total > 2.8 mmol/L, is also common and can be worsened with the practice of adding only Ca to early PN solutions or administering formulations with a higher protein content [14]. Hypercalcemia occurs in response to hypophosphatemia. To normalize the low circulating P in RS, P is removed from bone. Since Ca and P complex together to form the mineral portion of bone, if low circulating levels leads to removal of one of these two minerals, the other also is removed. Consequently, hypophosphatemia leads to elevated circulating Ca.

  4. 4.

    Hypomagnesemia, defined as serum Mg < 1.5 mg/dL, is far less common than hypophosphatemia and hypokalemia but has not been studied as carefully [8, 15]. Maternal preeclampsia is often associated with fetal FGR or infants born SGA; therefore, if the mother received magnesium for the treatment of preeclampsia or other indications, the neonate’s serum magnesium may be in normal range or elevated [8].

  5. 5.

    Hyperglycemia also occurs in RS. The glucose values defining hyperglycemia vary and include levels above 150 or 180 mg/dL [8, 16]. Independent and time-related associations suggest that hypophosphatemia precedes hyperglycemia [16]. The low P status may impair insulin production, yet infants born after FGR also have hyperinsulinism as a result of insulin insensitivity [17, 18].

The incidence of simultaneously having three or more biochemical abnormalities is less than 10% [8], suggesting clinicians may consider categorizing infants as developing RS based on the occurrence of hypophosphatemia alone or with just two lab abnormalities. The interdependent metabolism of amino acids, electrolytes, and minerals underly the development of these lab abnormalities. Without sufficient postnatal electrolyte and mineral provisions, cellular uptake and utilization of amino acids, P and K may lead to lower blood concentrations. Increased electrolyte and mineral provisions have generally not accompanied earlier and higher parenteral amino acid infusions [14], creating scenarios of increased P and K utilization without sufficient exogenous sources in early PN. As Ca and P would be expected to be released from bone in response to the low P, renal responses to these chemical disturbances appear appropriate and include resorption of P [19] and coinciding calciuria in infants with higher serum Ca [14].

Potential morbidities related to refeeding syndrome

Preterm infant morbidities that occur in association with RS can be considered serious and even life-threatening. Described associations derive from various study methodologies including case reports, observational cohorts as well as secondary analyses of clinical trials. Clinical risk factors for RS include FGR or small for gestational age (SGA) size of the newborn [8], preeclampsia [8], and increased resistive indices in the umbilical artery [9]. Still, RS can occur in infants whose birth weights are considered appropriate. This may partially reflect the distinctions of FGR being an entity of slower fetal growth across time as compared to SGA status reflecting a cross-sectional and dichotomous (yes/no) measure at birth. Most concerning, the development of RS may be associated with an increased risk of mortality although this is not confirmed in all cohorts [6, 20]. The development of RS has no reported favorable outcomes.

Respiratory outcomes may be impacted by RS, more specifically by the hypophosphatemic state, and may be attributed to respiratory muscle weakness. In a single-center cohort, VLBW infants with hypophosphatemia (<4 mg/dL) had increased risk of mechanical ventilation for ≥3 days compared to infants without hypophosphatemia [8]. Any hypophosphatemia during the first 7 days remained independently associated with an increased risk of bronchopulmonary dysplasia [8], yet an association with this chronic lung disease was not detected in other cohorts which used cutoff values including <3.5 mg/dL [9, 11, 20].

Immunologic and hematologic impairments also occur in RS. In a clinical trial of VLBW infants receiving higher versus lower parenteral amino acid doses, infants with lower serum P concentrations within the first week showed an increased risk of sepsis [21]. However, not all studies substantiate an increased risk of sepsis in infants with hypophosphatemia or RS. A separate cohort study by Cormack et al. identified only a statistical trend towards more sepsis amongst extremely low birth weight (ELBW) infants with RS, specifically defined as having both hypophosphatemia (<1.4 mmol/L) and hypercalcemia (>2.8 mmol/L) [6]. And a third cohort of hypophosphatemic VLBW infants (<4 mg/dL) by Ross et al. found no increased risk of sepsis [8]. ELBW infants with P levels in the lowest quintile carried a significantly increased risk of severe intraventricular hemorrhage;[6] causal mechanisms are yet to be identified. A case report suggested hemolytic jaundice resulted from RS, although this was not substantiated through comprehensive laboratory results [19].

Cardiac myocyte function is susceptible to electrolyte derangements known to occur in RS including hypophosphatemia and hypokalemia. Bradycardia and complete heart block have been reported in preterm infants born SGA. Fortunately, these electrophysiologic disturbances are reversible with electrolyte and mineral repletion [19, 22] and do not appear to cause long-term electrophysiologic abnormalities or cardiac dysfunction [22].

Potential strategies to avoid, diagnose, and treat refeeding syndrome in FGR infants

While the ideal scenario is complete avoidance of RS, investigation of precisely how to accomplish that using clinical trials to determine the optimal balance of parenteral macro- and micronutrients is needed. Until such studies are implemented, it appears warranted that clinicians develop a consistent approach to RS in the neonatal intensive care unit (NICU) that involves a multidisciplinary group to address: 1) creating a process to identify at-risk preterm infants, 2) a PN prescribing strategy for infants identified as being at risk for RS, 3) a standardized feeding guideline to enhance early introduction of enteral nutrition, and 4) a lab monitoring strategy. Given the consistently identified risk factors, we suggest that infants showing evidence of growth restriction be identified for monitoring. While definitions of growth restriction have varied, recent consensus expert opinion provides definitions of weight, length, head circumference as plotted on sex-specific growth charts as well as additional considerations that can be utilized for deciding which infants warrant monitoring [23]. It is important to note that not all infants that develop RS are categorized as having experienced FGR or SGA. Whether every extremely preterm infant should have extensive monitoring for RS should be a decision for each individual NICU.

Parenteral nutrition strategies

The general approach to PN in preterm infants is a daily advancement of macronutrients to pre-specified goals. However, in infants at risk of or showing evidence of RS, it may be safer to maintain lower doses of macronutrients without advancement throughout that higher risk period of the first 5 days after birth, possibly longer if electrolyte abnormalities develop [6]. There is no recommended specific dose of parenteral amino acids, intravenous lipids, or dextrose proven to prevent RS through clinical trials. Yet lower amino acid doses, even as early as the first postnatal day, are associated with lower risk of hypophosphatemia in some but not all cohort studies [6, 10, 11].

The availability of “stock” or “vanilla” PN solutions, available at any hour from the pharmacy, supports early amino acid administration to minimize early nitrogen losses. However, these solutions often do not automatically contain electrolytes and minerals. The administration of sodium (Na), P, and Ca in the first days lower the risk of RS and/or reduce the severity of lab abnormalities [6, 8]. In fact, it may be most valuable to provide electrolytes and minerals with the PN administered upon admission to the NICU. While concern exists about fluid retention and balance with early Na administration, providing Na glycerophosphate with early amino acids reduced the incidence of hypophosphatemia without causing hypernatremia [11]. Moreover, since hypokalemia is frequently identified in preterm infants with or at risk for RS, adding K phosphate to PN early on is feasible.

Developing and planning a process for early administration of micronutrients should include pharmacists to ensure PN admixture stability and safety [24]. The process may differ between NICUs based on whether stock PN solutions are compounded by a centralized, external compounding agency versus on-site. One report of a quality improvement initiative to provide PN P prior to 24 h versus 72 h post-birth reduced the incidence of hypercalcemia from 50 to 21% in the first postnatal week [25]. When providing early administration of Ca and P, an appropriate ratio should be utilized. Suggested molar ratios of Ca:P are 0.8 to 1.1:1 for VLBW infants in the first week [13]. When Mg is added to PN varies among NICUs and commonly relates to whether a mother received antenatal Mg and/or approaches to monitoring infant’s blood concentrations.

Enteral nutrition strategies

While details of PN formulations are important, it is imperative that clinical strategies also emphasize early introduction and advancement of enteral nutrition. Although no studies have specifically focused on enteral nutrition strategies to prevent or avoid RS, delayed initiation of enteral feedings in preterm infants at risk for RS, especially those with significant abnormalities of umbilical blood flow, often occurs out of fear of necrotizing enterocolitis (NEC) [26,27,28,29]. Therefore, studies of how preterm infant RS relates to enteral nutrition are needed and should be based on the existing evidence supporting enteral nutrition in this population. Several randomized trials have compared early versus delayed initiation of enteral feedings among preterm infants with FGR and/or with evidence of abnormal antenatal Doppler studies [30, 31]. They found no differences in the occurrence of feeding intolerance or NEC. Moreover, full enteral feedings were reached sooner with early feedings and the duration of PN and the rate of cholestatic jaundice were lower in the group fed earlier. In these earlier trials, about half of infants were exclusively formula fed. More recently, in a trial of 2804 infants born <32 weeks’ or weighing <1500 g and randomized to feeding advancement of 18 ml/kg/d versus 30 ml/kg/d, planned subgroup analyses showed no increased risk of NEC for infants with birth weight <10th percentile or with abnormalities of umbilical blood flow [32]. Over 90% of these infants were fed partially or completely with human milk.

Thus, for infants with FGR with or without abnormal Doppler studies or those at risk of developing RS, there is no advantage to delaying the initiation of feeds. Moreover, all or most of the feedings should be in the form of human milk, which has been shown to reduce the risk of NEC [33]. Feedings can be advanced at rates similar to those used in their normally grown counterparts.

Electrolyte monitoring and response to dyscrasias

For infants who received adequate fetal nutrition, in general, serum P concentrations are higher in the first 72 h postnatal and then decline [16, 34]. However, among preterm infants with FGR low serum P and K are commonly seen as early as postnatal day 1–3 [34, 35]. Therefore, preterm infants, especially those of VLBW or with FGR, should have close monitoring of serum electrolytes including P, K, and Mg starting at 24 h postnatal age. The frequency of follow up evaluation depends on the presence of abnormalities and the utilization of exacerbating interventions, i.e., high doses of parenteral amino acids [13, 14]. At a minimum, daily monitoring of infants at risk of RS through 5 days after birth may be considered as the risks of hypophosphatemia, hypokalemia and hypomagnesemia are higher during the first week [6, 8].

Clinical center-specific guidelines should consider that, while not all laboratory abnormalities may be avoided, mitigating severe abnormalities should be a priority. Guidelines should define these abnormalities and can even distinguish cutoff values to indicate abnormal versus severely abnormal. As an example, considering the range of cutoffs used to define hypophosphatemia, a center may choose to utilize a cutoff value of 3.5 or 4 mg/dL to define a low value and consider <2.5 mg/dL as severe and worthy of immediate correction. Setting these thresholds can guide providers as to the seriousness of the abnormal laboratory value and the time sensitivity for responding (i.e., distinguishing whether action is needed immediately once the lab result is available, versus adjusting at a later, pre-defined time such as during clinical rounds). Interventions should address correcting the single or numerous laboratory abnormalities and consider any modifications to provision of other nutrients (e.g., responding to hypophosphatemia requires P repletion and review of Ca administration and Ca:P ratio). Whether to alter the Ca:P ratio in response to cases of refractory hypophosphatemia is often context-dependent and may be best considered in discussion with a neonatal pharmacist. When considering such changes, the discussion should also include a determination of endpoints for such alterations in the ratio.

Closing perspectives

RS risk has been consistently identified amongst preterm infants with evidence of impaired growth in utero. In addition, consistent laboratory abnormalities occur in RS. This suggests that standardized monitoring and adapting PN formulations can mitigate, if not prevent, these abnormalities and perhaps reduce the risk of life-threatening morbidities in these infants. Future considerations for research should include clinical trials for varying macro- and micronutrient doses, perhaps utilizing enrollment strategies that stratify for both gestational age and FGR status. Until such data are available, we suggest that individual NICUs account for site-specific clinical practices to utilize these consistent findings to develop a screening and management approach for at-risk infants.