Intermittent Hypoxemia in Preterm Infants

SYNOPSIS

Intermittent hypoxemia events are common during early postnatal life, particularly in preterm infants. These events have been associated with multiple morbidities including retinopathy of prematurity, sleep disordered breathing, neurodevelopmental impairment, and mortality. The relationship between IH and poor outcomes may be dependent on the patterns (frequency, duration, and timing) of the IH events. Current treatment modalities used in the clinical setting have been only partially successful in reducing the incidence of apnea and accompanying IH but the risks and benefits of more aggressive interventions should include knowledge of the relationship between IH and morbidity.

INTRODUCTION

Although maintaining adequate oxygenation is a fundamental aspect of newborn care, we are only beginning to appreciate how even subtle alterations in oxygen levels can affect both short and long term outcomes. Before the implementation of non-invasive technologies, oxygen assessment was limited to intermittent arterial sampling which only gave a glimpse of the true instability of oxygen levels that can occur during early postnatal life. Current continuous recordings of oxygen saturation reveal a much higher frequency of intermittent hypoxemia (IH) events that were previously un-documented in medical charts and provide insight to high risk patterns associated with both short and long term morbidity. This chapter summarizes what is currently known about the technology used to assess oxygen levels, patterns of IH during early postnatal life, underlying mechanisms associated with IH, and high risk IH patterns that may possibly induce a pathological cascade.

HISTORICAL PROSPECTIVE

Prior to the mid-1970s intermittent arterial sampling formed the basis for assessing and managing supplemental oxygen administration. There was no clear evidence that arterial oxygen tension [PaO₂] or oxygen saturation [SaO₂] exhibited frequent fluctuations in preterm infants even though apnea of prematurity was known to be a problem. This changed with the advent of transcutaneous PO₂ [TcPO₂] monitors which were widely employed during the 1980s.

It was remarkable to observe how positional changes and associated procedures, such as spinal taps, could cause TcPO₂ to fall [Gleason ‘83]¹. This led to the widespread acceptance of “gentle” care for fragile preterm infants.

Unfortunately TcPO₂ electrodes [later combined with TcPCO₂ electrodes] were cumbersome, required frequent recalibration, and resulted in site erythema from heating to 43°−44°C. There was also the realization in the 1980s that TcPO₂ increasingly underestimated PaO₂ with advancing postnatal age; this was especially a problem with the increasing incidence of bronchopulmonary dysplasia [BPD] [Rome ‘84²; Solimano ‘86³; Ramanathan ‘87⁴]. This was a new population of extremely low birth weight infants who no longer had arterial access. A solution was found in pulse oximetry and this technology has dominated Neonatology since approximately 1990. As a result, the available literature on IH episodes in preterm infants is almost exclusively based on pulse oximetry.

PULSE OXIMETRY

As stabilization of oxygenation is one of the primary challenges in the NICU, pulse oximetry plays an important role in patient care. Bedside discussions often include oximetry based histogram data to note percent time in any given oxygen saturation range and/or nursing notation of IH events. However, treatment decisions based on medical chart documentation may be problematic as they significantly underestimate the true incidence of even prolonged events [Poets⁵]. Adding to the confusion is that there is currently no standard definition for a clinically relevant IH event and, therefore, corresponding pulse oximeter alarm settings. Most research trials have defined an intermittent hypoxemia event as a fall <85% or 80% but in the clinical setting a low threshold alarm will be determined by the individual NICU oxygen saturation target. There is also wide variation in practice pertaining to the clinical significance of the duration of an IH event. For example, health care workers in some NICUs use a long pulse oximeter alarm delay as a tool to minimize nuisance alarms due to short self-resolving IH while others consider even a short event one that requires intervention. Correspondingly, alarm delay criteria can vary widely between NICUs depending on the manufacturer (ranging from 0 to 15 sec) and staff perception of the duration needed for a clinically relevant desaturation event.

Pulse oximeters have the advantage of obtaining longitudinal documentation of oxygen levels in a noninvasive and rapidly responding manner but there are limitations that may affect measurement accuracy. The most widely acknowledged disadvantage of pulse oximetry is in the inability to detect hyperoxemia at levels of SpO₂ exceeding approximately 97% [Poets ‘93]⁶.

Additional factors including probe position, motion and ambient light interference, low perfusion, skin pigmentation, and variations in hemoglobin may result in delayed waveform recognition and/or underestimation of oxygen saturation levels [Poets, Southall 1994⁷ and Trivedi 1997⁸]. Most manufacturers report an overall error of ±2–3% of full scale but even under ideal conditions, accuracy may diminish with decreasing SpO2 [Rosychuck⁹]. In a study of 1664 preterm infants, overall mean differences between SaO2 and SpO2 (Masimo) were −1.8±2.9% but less than 40% of infants were within 3% of the corresponding SaO2 when SpO2 fell below 88%. During very low oxygen saturation levels of <70% that are impractical or unethical to target in infants, newborn lamb models have shown even larger differences ranging from 13–17% [Dawson]¹⁰. Therefore, although pulse oximetry is often used to identify periods of IH, treatment decisions based on the absolute SpO2 value should be made with caution especially at low levels of oxygen saturation as can occur in cardiac patients.

INCIDENCE OF IH EVENTS AND UNDERLYING MECHANISMS

There is considerable current interest in the postnatal time course of IH events in preterm infants. This is precipitated by the fact that persistent IH events frequently delay hospital discharge and may be perceived as increasing the vulnerability of these infants. Barrington documented over 20 years ago that apneic events of >12 seconds are common in very low birth weight preterm infants prior to discharge, many of which would, presumably, have been associated with IH [Barrington ‘96¹¹]. Data in infants of 24–28 weeks’ gestation show a marked change in IH events over time with relatively few hypoxemic episodes occurring during the first week of postnatal life, a progressive increase in weeks 2–3, a plateau around weeks 4–6, and finally a decrease in weeks 6–8 [Di Fiore 2010¹²] [FIGURE 1]. It has been proposed that this postnatal increase in IH events may be related to a documented postnatal increase in periodic breathing which is frequently associated with episodic desaturation [Patel ‘16]¹³. In a subsequent study, the incidence of IH events was significantly increased in a low [85–89%] versus high [91–95%] baseline SpO₂ target [Di Fiore ‘12]¹⁴ This finding is consistent with earlier data in preterm infants with BPD [McEvoy ‘93]¹⁵. More recently IH events [comprising a >10% fall in baseline SaO₂] were reported in a majority of preterm infants during home recordings of SpO₂, but declined between 36 and 44 weeks’ postmenstrual age [Hunt ‘11¹⁶].

Figure 1.

The progression of intermittent hypoxemic events (IH) during early postnatal life. Preterm infants exhibit relatively few IH during the first week of life followed by an increase during weeks 2–4 and a decrease thereafter.

From Di Fiore JM, Bloom JN, Orge F, et al. A higher incidence of intermittent hypoxemic episodes is associated with severe retinopathy of prematurity. The Journal of pediatrics. Jul 2010;157(1):69–73

Immature respiratory control resulting in apnea and respiratory pauses as well as ineffective and/or obstructed inspiratory efforts are the major precipitants of IH events [DiFiore¹⁷] but several physiologic parameters likely contribute to the resultant desaturation. The most important of these is probably pulmonary oxygen stores which reflect lung volume. Preterm infants are at risk for a low basal functional residual capacity because of both atelectasis and high chest wall compliance. Therapy with continuous positive airway pressure (CPAP) clearly benefits these infants by splinting the upper airway, stabilizing lung volume, and increasing baseline SpO₂, thereby minimizing the risk of desaturation. Other physiologic parameters including blood oxygen capacity comprising blood volume and hemoglobin content [Abu Jawdeh ‘14¹⁸; Sands ‘10¹⁹] may also be implicated in IH. Current neonatal practice attempts to minimize the use of mechanical ventilation via endotracheal intubation and favors the widespread use of CPAP or other non-invasive ventilation techniques. Synchronization methods during non-invasive ventilation to decrease IH events continue to be a challenge [DeWaal]²⁰. However, IH episodes are also common even in intubated, ventilated infants as a consequence of ineffective ventilator support and loss of the infant’s lung volume [Dimaguila ‘97²¹; Esquer ‘07²²]. Interestingly, intrauterine growth status also appears to be a factor. Small-for-gestation [SGA] [versus appropriately grown] preterm infants are more vulnerable at a low baseline SpO₂ target as manifest by higher mortality and increased incidence of IH events [Di Fiore ‘17²³]. The mechanism of this increased vulnerability to IH is unclear but pulmonary hypertension in SGA infants and resultant hypoxia-induced pulmonary vasoconstriction may contribute.

INTERMITTENT HYPOXEMIA AND OUTCOMES

Stabilizing the extreme fluctuations in oxygenation in preterm infants often requires a balance between the risks and benefits of oxygen supplementation and constant adjustments of FiO2 which can be labor intensive. Therefore, with the current NICU focus on aggressive weaning protocols, knowledge of the association between IH and morbidity would be of great benefit in guiding clinical practice.

The evidence that suggests that IH may initiate a pathological cascade is derived from animal models which shows that intermittent hypoxia during adulthood increases extracellular superoxide concentration [Fabian 2004²⁴], induces HIF-1α expression [Yuan 2008²⁵], degrades HIF-2α expression, and downregulates superoxide dismutase [Nanduri 2009²⁶] leading to overall pro-oxidant signaling. In addition, exposure to IH during early postnatal life disrupts expression patterns of proteins involved with dopamine signaling [Decker²⁷] and causes a pro-inflammatory response including elevation of TNF-α and IL-1β [Del Rio²⁸]. Thus, in infants intermittent hypoxemia may induce a pathological cascade via a pro-oxidant, pro-inflammatory, or neurotransmitter imbalance pathway [Figure 2]

Figure 2.

The relationship between intermittent hypoxemia and outcomes may be dependent on the pattern of the hypoxemic events. IH have been associated with morbidity in preterm infants but not all IH have been shown to have deleterious effects. Therefore, these studies may represent “severe” IH in terms of frequency, severity, duration and timing that induce inflammation, neurotransmitter imbalance and reactive oxygen species with a high risk of poor outcomes. In contrast, the effect of “mild” IH on outcomes is unclear.

Intermittent hypoxemia during early postnatal life has been associated with multiple poor outcomes including retinopathy of prematurity (ROP), growth restriction, sleep disordered breathing, neurodevelopmental impairment and mortality. Although it is well known that early postnatal hyperoxic exposure is the major risk factor for severe ROP, there is evidence that later IH causes rebound overexpression of growth factors (ie VEGF, EPO) associated with hypoxia-inducible factors (HIF) that may play a role in neovascularization (Chow 2003²⁹). Multiple animal models with various intermittent hypoxic paradigms have induced retinopathy [Penn³⁰] and have shown that the level of neovascularization may be dependent on the pattern of intermittent hypoxic exposure [Coleman³¹]. For example, rats exposed to cycles of IH in an equally dispersed (2 hrs apart) versus clustered (10 min apart) pattern have more severe oxygen-induced retinopathy including vascular tufts, leaky vessels, retinal hemorrhage and vascular overgrowth in the clustered paradigm. Studies in preterm infants have shown similar findings with a higher number of IH during early postnatal life in infants with severe ROP requiring laser therapy [DiFiore 2010¹²]. Closer examination of IH patterns revealed an association between ROP and IH of longer duration [DiFiore 2012³², Poets 2016³³], less severity, and a specific time interval between events of 1–20 minutes [DiFiore 2012³²]. The time between IH events may play an important role in initiating a pathophysiologic response as rodent studies have revealed transient alterations in pro-oxidant signaling occurring during the resolution of the hypoxemic event [Fabian 2004²⁴, Yuan²⁵].

The relationship between IH events and growth restriction is currently limited to neonatal rodent models. For example, repetitive exposure to IH during the first week of life significantly reduced body weight by the 3^rd day with subsequent growth restriction with every day of exposure [Pozo³⁴]. After 21 days of recovery in room air, the rat pups exhibited catch-up growth suggesting that IH events commonly seen in preterm infants during early postnatal life may play a role in weight gain, a common criteria for hospital discharge. However, to our knowledge, there are currently no available data looking at the potential effect of IH on growth restriction in human neonates.

IH events can have both short and long term effects on respiratory stability and sleep disordered breathing. In rodents, early postnatal exposure to intermittent hypoxia enhanced the acute hypoxic chemoreflex immediately following intermittent hypoxia exposure [Julien³⁵] in males with no effect in females. In contrast, early postnatal exposure to intermittent hypoxia followed by recovery to young adulthood reduced the acute ventilatory response to hypoxia [Reeves³⁶] suggesting that intermittent hypoxemia events during early postnatal life may have divergent short versus long term effects on respiratory control. The early postnatal increase in peripheral chemoreceptor activity may be one explanation why periodic breathing often appears within 2–4wks of life in preterm infants [Patel¹³] while the long term reduction in peripheral chemoreceptor activity may be one component contributing to preterm birth being a risk factor for sleep disordered breathing at 8–11 years of age [Rosen³⁷].

Human trials in neonates have revealed associations between both delayed resolution [Janvier 2004³⁸, Pillekamp³⁹] and increased frequency and/or severity of cardiorespiratory events [Taylor⁴⁰, Greene 2014⁴¹] and neurodevelopmental impairment. The findings of these single center studies have been confirmed by a recent retrospective analysis in the multicenter Canadian Oxygen Trial (COT)³³. Using continuous recordings of oxygen saturation in a large cohort of >1000 infants, the authors found a correlation between time spent with hypoxemia during the first few months of life and adverse 18-month outcomes including late death or disability, cognitive language delay, and motor impairment. Additional clinical trials have shown a relationship between bronchopulmonary dysplasia and poor cognitive outcomes [Short 2003⁴², Gray 2004⁴³, Anderson 2006⁴⁴, Twilharr 2018]⁴⁵ which begs the question is it the oxygen exposure or the increased IH events that can occur with chronic lung disease that may put these infants at risk for neurodevelopmental impairment? Data in rodents have shown that neonatal exposure to hyperoxia (65% O2) alone had no effect on long term working memory while hyperoxia+IH resulted in neurofunctional handicap [Ratner 2007⁴⁶]. Taken together these studies suggest that exposure to IH events during a critical period of brain development may have long term consequences on brain dysfunction.

Lastly, patterns of oxygenation during the first few days of life have been associated with mortality. The surfactant positive pressure and oxygenation trial (SUPPORT) randomized infants to a lower (85–89%) versus higher (90–95%) oxygen saturation target to reduce the incidence of severe ROP [SUPPORT⁴⁷]. One unexpected finding of the trial was a higher mortality in the lower target group that was limited to infants with intrauterine growth restriction [Walsh⁴⁸]. Closer examination of actual achieved levels of oxygenation (as opposed to the randomized target) revealed an association between lower oxygen saturation (≤92%) during the first three days of life and decreased 90 day survival in both appropriate (AGA) and small (SGA) for gestational age infants. However, SGA infants also had an enhanced mortality associated with an increased frequency of IH events that was not seen in the AGA infant cohort [DiFiore 2017²³].

Interestingly, not all patterns of IH may be deleterious and the effects of “mild” hypoxemia are not clear. For example, “mild” IH has been associated with later impaired sensorimotor performance in mice [Juliano 2015]⁴⁹. In contrast, various paradigms in both human and animal models have suggested that “mild” hypoxia/hypoxemia may be benign or even beneficial. In rodents, exposure to a single brief (5 min) cycle of hypoxia during the first 24 hrs after birth [Martin 2010⁵⁰], or longer (4hrs) and milder cycles of hypoxia during the first 3–4 wks of life [Zhang⁵¹] enhanced long term spatial learning [Zhang⁵¹], memory [Martin⁵⁰] and structural changes in both the hippocampus [Zhang⁵¹, Martin⁵⁰] and frontal cortex [Martin⁵⁰]. In neonates, short [Poets³³] or tightly clustered (<1min apart) [Difiore 2012¹²] IH events were not associated with morbidity but the risk associated with such patterns may be confounded by other factors including intrauterine growth restriction [Di Fiore 2017²³]. In adults, various therapeutic “mild” IH paradigms are currently being examined for sleep apnea, systemic hypertension, depression and neural inflammation [Navarrete⁵²]. These combined findings suggest that the compensatory versus maladaptive effects of IH may be dependent on the “dose” of IH. As premature infants exhibit a wide array of IH patterns with increasing postnatal age [Di Fiore 2012³²] it is, therefore, possible that early postnatal IH configurations contain both pathological and beneficial components [Figure 2].

THERAPEUTIC OPTIONS

Management strategies for IH events focus primarily on their prevention. The aggressiveness of these approaches clearly depends on the likelihood of adverse effects of IH and, unfortunately, unanswered questions remain. Desaturation events are a consequence of immature respiratory control superimposed upon an immature respiratory system and, in many cases, there is probably vulnerability to pulmonary hypertension. Two mainstays of management are, therefore, optimizing baseline oxygenation and enhancing respiratory control [figure 3].

Figure 3.

Treatment strategies for intermittent hypoxemia include a multi-pronged approach; xanthines to enhance respiratory control, supplemental oxygen and pressure support to optimize baseline level of oxygenation and red blood cell transfusion to improve oxygen stores.

1. Optimizing Baseline Oxygenation. There appears to be widespread consensus that baseline SpO₂ should be maintained in the 90–95% range, and that risks associated with restricting the upper limit to 89% outweigh the benefits [Manja ‘17⁵³,Stenson ‘16⁵⁴]. As already noted, low baseline SpO₂ significantly increases the risk of desaturation, especially in SGA infants [Di Fiore ‘12⁵⁵, ‘17²³]. Over several decades CPAP has proven an important means to stabilize oxygenation by supporting functional residual capacity [FRC] and splinting the upper airway to prevent its closure during apnea.

   The role of red blood cell transfusions in decreasing apnea [and resultant desaturation] has been controversial. Most recent data show improvement in apnea and IH after red blood cell transfusion. Age-dependent improvement in frequency and severity of IH after transfusion have been documented beyond the first week of life when IH events begin to increase [Abu Jawdeh¹⁸]. Another study documented a decrease in apnea associated with desaturation and bradycardia in the 12 hours after red cell transfusion [Zagol ‘12⁵⁶]. It would appear that the benefits of increasing red cell mass on IH outweigh the risk of transfusion in selected infants.

   There is great interest in the rapid emergence of automated [vs manual] control of supplemental oxygen delivery in high risk neonates as addressed elsewhere in this volume. Initial automated systems focused on ventilated infants and had great success in reducing hyperoxic, but not hypoxic, episodes [Claure ‘11⁵⁷]. More recent studies by several European groups have had success in reducing IH events with automated versus manual FiO₂ control and further documented this benefit in infants on noninvasive as well as invasive ventilator support [Waitz ‘15⁵⁸, van Kaam ‘15⁵⁹]. This technology is being employed clinically in Europe and undergoing further investigative trials in the U.S. While the available data are very encouraging, it should be noted that the percent time that SpO₂ was in range [91–95%] only increased from 58 to 62% [van Kaam ‘15⁵⁹] emphasizing the major challenge in keeping preterm infants in any given target even with rapid responding computer controlled feedback.

   The application of near-infrared spectroscopy as a tool for assessing the relationship between SpO2 and cerebral tissue oxygen saturation is still the subject of ongoing research. Although interpretation of absolute values of cerebral tissue oxygenation have yet to be determined recent data have shown a greater adverse impact from IH than bradycardic events [Schmid ‘15⁶⁰]. It remains to be seen whether this noninvasive technique can be used as a prognostic marker [Korček 2017⁶¹].

   Finally, other novel new approaches are being explored. Two investigative groups are exploring afferent stimulation at the body surface to decrease apnea and resultant IH. Smith and colleagues are employing stochastic resonance stimulation via gentle mattress vibrations while Kesavan et al are employing extremity vibration devices to enhance limb proprioceptive afferents [Smith ‘15⁶², Kesavan ‘16⁶³]. While both approaches show promise in decreasing IH events, it is unclear if stabilization of oxygenation is a contributing mechanism.

2. Role of Caffeine Therapy. Xanthine therapy, notably caffeine, has become a mainstay of neonatal care and improves not only respiratory, but also neurodevelopmentl outcomes [Schmidt ‘17⁶⁴]. While its mechanisms of benefit may be multifactorial, enhanced neonatal respiratory control with caffeine is well accepted [Abu-Shaweesh ‘17⁶⁵]. Although earlier data questioned whether caffeine improved hypoxemic episodes in preterm infants [Bucher ‘88⁶⁶], recent data clearly demonstrate a reduction in IH [Rhein⁶⁷]. This is not surprising, given that caffeine reduces the incidence of apnea of prematurity.

   There are still interesting issues regarding optimal caffeine administration for preterm infants. The first issue is when this therapy should begin. Available data support early [even prophylactic] dosing in high risk neonates, however, these data are largely based on retrospective reviews and associations rather than randomized trials [Lodha ‘15⁶⁸]. The second issue is when this therapy should stop. Rhein and colleagues have demonstrated a reduction in IH events with prolonged therapy until approximately 36 weeks’ postmenstrual age and even beyond [Rhein⁶⁷]. This is under further study, but runs the potential risk of either prolonging hospitalization or increasing home monitor use. The third and most challenging issue is whether doses higher than those traditionally used should be employed in either a loading or maintenance mode. Available data comparing high versus standard dosing are limited by low quality outcome measures, small sample sizes, and diverse caffeine dosing regimens [Vliegaerthart 2018]⁶⁹. Benefit versus safety of such an approach must be carefully studied as adenosine receptor subtype inhibition [the presumed main mechanism of action for xanthines] is variable and dose dependent, raising potential safety concerns [Chavez-Valdez⁷⁰,McPherson⁷¹]. The challenge is to weigh the potential consequences of IH episodes against the pros and cons of any therapeutic intervention.

SUMMARY

In summary, intermittent hypoxemic events are ubiquitous in preterm infants and are a major challenge in clinical care. Early arterial blood gas sampling limited our ability to monitor oxygen levels over small windows of time while implementation of pulse oximetry has increased our knowledge of the transient progression of IH during early postnatal life. Current treatment modalities used in the clinical setting have only been partially successful in reducing the incidence of apnea and accompanying IH but the risks and benefits of more aggressive interventions must include knowledge of the relationship between IH and morbidity. Future clinical challenges that may assist in mitigating IH and possible sequelae include identification of optimum oxygen saturation targets, recognition of high risk (and possibly beneficial) IH patterns, and implementation of FiO2 automated controllers and other novel therapies to avoid periods of both hyper- and hypoxemia.

KEY POINTS.

1. Intermittent hypoxemia events are common in preterm infants during early postnatal life.

2. In neonates, intermittent hypoxemia events have been associated with multiple morbidities including retinopathy of prematurity, sleep disordered breathing, neurodevelopmental impairment and death.

3. The relationship between IH and morbidity may be dependent on the pattern of the IH events, although this needs further investigation.