Temporal effects of caffeine on intrapulmonary shunt in preterm ventilated infants

Introduction

Premature infants commonly develop respiratory distress which can lead to chronic pulmonary insufficiency of prematurity with long term sequelae persisting into adolescence and young adulthood [1].

Caffeine is a well-established aspect of preterm newborn respiratory care [2]. Its administration can reduce the frequency of apnoea of prematurity, the duration of mechanical ventilation and the development of bronchopulmonary dysplasia (BPD) [3].

Caffeine increases respiratory drive and alveolar ventilation [4], increases diaphragmatic muscle strength and improves lung compliance [5], increases diaphragmatic electrical activity [6] and exerts a diuretic action [7]. It is plausible that the diuretic action might offload interstitial fluid thereby decreasing intrapulmonary shunting and improving ventilation, but this phenomenon and the duration of any effect has not been studied.

Precise measurements of ventilation to perfusion ratios require complex nuclear medicine investigations which cannot be performed in sick prematurely born, ventilated infants. We have, however, described a non-invasive method to assess ventilation to perfusion relationships and right to left intrapulmonary shunting using the oxyhemoglobin dissociation curve (ODC) and reported results in extremely preterm infants with BPD [8] or pulmonary interstitial emphysema [9], as well as in healthy term infants [10].

We aimed to evaluate the temporal effects of caffeine administration on respiratory parameters in premature, ventilated infants using the non-invasive ODC method.

Materials and methods

A retrospective observational cohort was studied on a Neonatal Intensive Care Unit between September 2021 and April 2023.

Infants born less than 31 weeks of completed gestational age (GA) who were prescribed caffeine citrate by the clinical team were eligible for recruitment into the study if they were receiving invasive mechanical ventilation on day three after birth. All infants had been entered into a randomized study of closed-loop automated oxygen control [11]. This is a prospective randomised controlled trial that has been approved by the Yorkshire and the Humber-Sheffield Research Ethics Committee. Informed written consent was given by the parents/legal guardians of all study participants. Infants with chromosomal or major congenital anomalies and those extubated before day three were excluded from the study.

Baseline demographic data were collected from the medical notes including a full course of antenatal steroids (yes/no), gender, gestational age (GA) (weeks), birth weight (grams), corrected gestational age (CGA) and weight at study, mode of delivery (spontaneous vaginal delivery (SVD) or caesarean section), administration of surfactant (yes/no), duration of mechanical ventilation (hours) at study entry, dose of caffeine prescribed (milligram/kilogram), method of oxygen saturation monitoring (automated/manual), total duration of mechanical ventilation (MV) (days), treatment for patient ductus arteriosus (PDA) (yes/no), intraventricular hemorrhage (IVH) grade 3 or 4 or periventricular leukomalacia (PVL) (yes/no), survival to discharge (yes/no), post menstrual age (PMA) at discharge (weeks), weight at discharge (grams), diagnosis of BPD at 36 weeks (yes/no) and home oxygen at discharge (yes/no).

All infants were supported by patient-triggered ventilation with a targeted tidal volume using the SLE6000 neonatal ventilator (SLE, South Croydon, United Kingdom). Targeted tidal volumes were set between 5 and 6 mL/kg. All ventilated infants received surfactant in the delivery suite. The clinical team routinely prescribed a loading dose of caffeine citrate (20 mg/kg) to all infants less than 34 weeks of gestation who required either invasive or non-invasive respiratory support, followed by a maintenance dose of 5 mg/kg/day as per trust guidelines. The infants had hourly manual recordings of ventilatory measures and inspiratory tidal volumes (Vt) on the nursing charts and second by second recordings of paired transcutaneous oxygen saturation (SpO₂) and the fraction of inspired oxygen concentration (FiO₂) which were downloaded from the ventilators. Blood gas analysis was performed every four to 6 h as per unit protocol.

Infants randomised to closed-loop automated oxygen control were also connected to the Oxygenie closed-loop oxygen saturation monitoring software (SLE) which has been described in a previous publication [11]. Manual adjustments to the inspired oxygen concentration were allowed at any point during the study including in the automated oxygen control period if deemed appropriate by the clinical team. Ventilation settings were manually adjusted by the clinical team as per the unit’s protocol.

The effect of caffeine on respiratory parameters was studied on day three after birth before and after administration of the second maintenance caffeine dose, at four time points: 1 h before the administration of caffeine (pre-caffeine) and at 1, 4 and 20 h after its administration. These time points were selected to avoid the changes in lung mechanics that occur in the first 48 h after birth following the administration of surfactant [12]. For these time points, data on peak inflation pressure (PIP), positive end-expiratory pressure (PEEP), mean airway pressure (MAP), tidal volume (Vt), inspiratory time, ventilatory rate, SpO₂ and FiO₂ were collected.

For the SpO₂ and FiO₂, the scatter plots of the recorded paired SpO₂/FiO₂ values 15 min before and 15 min after each time point were reviewed and the most frequent three to five pairs before and after the administration of caffeine were used in our analyses.

Using the paired values of SpO₂ and FiO₂, an oxyhemoglobin dissociation curve was constructed for each time point for each infant and compared with an ideal reference neonatal oxyhemoglobin dissociation curve. Using the paired values of SpO₂ and FiO₂, the ventilation perfusion ratio (V_A/Q), the right shift of the oxyhemoglobin dissociation curve and the percentage of right-to-left shunt were calculated for the four time points. V_A/Q, shift and right-to-left shunting were derived using software based on the Lockwood algorithm, which derives results for each dataset from a two-compartment model: shunt, shift and V_A/Q for a single homogeneous ventilated compartment [13]. The hemoglobin level at the time of the assessment was used in the calculations.

Results

The 21 infants included in the study had a median gestational age of 27 (23.7–30.7) weeks and median birth weight of 780 (range: 475–1,640) grams and were studied at a median corrected post menstrual age of 27.3 (range: 24–31) weeks. 10 infants were male. 10 infants had an SVD and the remaining 11 infants were delivered by caesarean section (in only one case was the mother in labour). All infants were ventilated within 48 h of birth for respiratory distress (Table 1).

There were no significant differences in the median mean airway pressures, tidal volumes and partial arterial pressures of carbon dioxide pre and post caffeine administration (Table 2).

There was a significant reduction in the median shunt percentage from baseline (1 h before caffeine administration) compared to 1 h post caffeine administration (8 (range: 7–9)% vs. 4 (range: 0–6)%, p=0.042) (Table 3, Figure 1), and 4 h post caffeine administration (8 (range: 7–9)% vs. 0 (range: 0–3)%, p=0.042), but the (V_A/Q) (Figure 1) and the right shift of the oxyhemoglobin dissociation curve did not differ significantly between these time points. At 20 h, there were no significant differences between the shunt, (V_A/Q) and the right shift of the oxyhemoglobin dissociation curve compared to baseline values.

Figure 1:

Boxplot of percentage change in right to left shunt over time. The ends of each box represent the upper and lower quartiles with the median being marked by a horizontal line inside the box. The whiskers are the two lines outside the box that extend to the highest and lowest observations.

Subgroup analysis in infants receiving closed loop automated oxygen control (n=8), did not reveal any significant differences in the median shunt percentage, the (V_A/Q) and right shift of the oxyhemoglobin dissociation curve from baseline compared to all timepoints after the administration of caffeine. Similarly, when only infants on manual oxygen control where examined (n=13), the administration of caffeine had no significant effect on shunt, (V_A/Q) and right shift of the curve.

The differences in the median shunt percentage at 1 h after the administration of caffeine did not differ significantly from baseline between infants delivered via SVD or born through caesarean section (median (range): 0 (0–9) %, [SVD] vs. 0 (0–8) %, [caesarean section], p=0.131). In addition, 4 h after the administration of caffeine, the differences in the median shunt percentage were comparable between the two groups (median (range): 0 (0–9) %, [SVD] vs. 0 (0–8)%, [caesarean section], p=0.131).Subgroup analysis based on the mode of delivery did not reveal any significant differences in the median shunt percentage, the (V_A/Q) and right shift of the oxyhemoglobin dissociation curve from baseline compared to all timepoints after the administration of caffeine.

Discussion

We have demonstrated that caffeine administration was associated with a transient decrease in the right to left shunt at 1 and 4 h after administration. This action was short-lived as it was not present by 20 h of age.

We report median V_A/Q values of 0.49 (range: 0.19–0.76). This included some severely abnormal values compared with the values of V_A/Q in healthy term infants (range 0.8–1.1) [10]. This is not surprising as our cohort consisted of very preterm infants with a median gestational age of 27 weeks and median birth weight of 780 g who had respiratory failure. Our results are comparable to the V_A/Q values reported for very preterm infants with BPD diagnosed at 28 days of life [8] or evolving BPD and receiving treatment with postnatal steroids [15], as well as for infants with pulmonary interstitial emphysema [9]. In a cohort of 24 infants with a median gestational age of 25 weeks who were studied at a median postmenstrual age of 33 weeks, there was median shunt values of 8 % (IQR: 0.3–16.5 %) and V_A/Q of 0.40 (IQR: 0.30–0.48 %). Low gestational age at birth, low birth weight and weight at the time of study were significantly related to shunt, shift and V_A/Q [8].

At birth, the airways must be cleared of liquid to allow the entry of air and the onset of pulmonary gas exchange [16]^. In healthy term infants, most of the pulmonary liquid is cleared within 2 h of commencing spontaneous breathing [17]. Preterm infants are often unable to generate the intrathoracic pressure needed to achieve this [18]. Following clearance from the airways, the liquid enters the pulmonary tissue and increases pulmonary interstitial tissue pressure which can cause the fluid to flow back to the airways [19]. Failure of the lungs to fully aerate after birth is a common problem in very preterm infants and adversely affects transition to postnatal life. It is plausible that caffeine, through its diuretic action, enhances the rate of pulmonary fluid clearance and improves pulmonary function [20]. This mechanism of action could explain the reduction in shunt that we observed 1 and 4 h after the administration of caffeine. This effect, though, was transient and there were no significant differences between the percentages of shunt observed at baseline and 20 h following caffeine administration. We did not find any significant differences in the effect of caffeine administration between infants born following SVD or elective caesarian section, but this may be because we studied infants on day three after birth.

The peak plasma concentration of caffeine with both oral and intravenous routes is reached within 30 min to 2 h of administration [4]. Caffeine’s half-life in preterm infants is estimated to be between 60 and 140 h [21] and therefore we might have expected maintenance of caffeine plasma concentrations for the total duration of our study and sustained improvement in right to left shunt at all subsequent timepoints. Animal data, however, suggest that the maximum diuretic effect of caffeine is observed at 30–60 min after beginning a caffeine infusion and subsequently decreases [20]. This finding would explain the transient improvement in shunt we report up to 4 h following the administration of caffeine.

Our study has strengths and some limitations. We described the acute effects of caffeine on respiratory parameters at multiple time points and we used indices that describe both the oxygenation and the ventilation to perfusion relationships. In calculating the relative position of the FiO₂ vs. SpO₂ curve in our cohort, we used three to five paired points of SpO₂ and FiO₂ measurements, as this described more of the plateau of their relationship and enable the algorithm to calculate the shunt, V_A/Q and shift values, whereas in previous neonatal studies only two measurements were used [9, 10]. Moreover, despite the retrospective nature of our study, the paired SpO₂ and FiO₂ measurements were derived from hundreds of electronic recordings for each timepoint and not a single manual entry from the patient’s chart, increasing the accuracy of our measurements. As this was a novel study and there was no previous study reporting specifically on the effect of caffeine on V_A/Q we could not base our sample size calculation on this parameter, but instead used data from two large cohorts with outcomes that were as relevant as possible. The participants were receiving either automatic or manual control of their inspired oxygen concentration. Infants on automatic oxygen control had significantly higher percentages of right to left shunt at baseline and, thus, more severe respiratory disease, but the median differences in shunt at 1 and 4 h after the administration of caffeine were comparable between the two groups. Therefore, the different modes of oxygen control did not seem to influence our results. To our knowledge, this is the first neonatal study to date to assess the temporal effects of caffeine on intrapulmonary shunting and ventilation to perfusion relationships in preterm neonates.

In conclusion, we have demonstrated that a maintenance dose of caffeine citrate in ventilated preterm infants on day three after birth led to a transient decrease in intrapulmonary shunt from one to 4 h after administration. We speculate that this phenomenon is attributed to its diuretic action.