Article Text

Download PDFPDF
Cerebral oxygenation and echocardiographic parameters in preterm neonates with a patent ductus arteriosus: an observational study
  1. Laura Dix1,
  2. Mirella Molenschot2,
  3. Johannes Breur2,
  4. Willem de Vries1,
  5. Daniel Vijlbrief1,
  6. Floris Groenendaal1,
  7. Frank van Bel1,
  8. Petra Lemmers1
  1. 1Department of Neonatology, Wilhelmina Children's Hospital, University Medical Center Utrecht, The Netherlands
  2. 2Department of Paediatric Cardiology, Wilhelmina Children's Hospital, University Medical Center Utrecht, The Netherlands
  1. Correspondence to Laura Dix, Department of Neonatology, Wilhelmina Children's Hospital/University Medical Center, Room KE04.123.1, PO Box 85090, Utrecht 3584AE, The Netherlands; l.m.l.dix-2{at}


Background A haemodynamically significant patent ductus arteriosus (hsPDA) is clinically suspected and confirmed by echocardiographic examination. A hsPDA decreases cerebral blood flow and oxygen saturation by the ductal steal phenomenon.

Aim To determine the relationship between echocardiographic parameters, cerebral oxygenation and a hsPDA in preterm infants.

Methods 380 preterm infants (<32 weeks gestational age) born between 2008 and 2010 were included. Blinded echocardiographic examination was performed on the second, fourth and sixth day after birth. Examinations were deblinded when hsPDA was clinically suspected. Regional cerebral oxygen saturation (rScO2) was continuously monitored by near-infrared spectroscopy during 72 h after birth, and afterwards for at least 1 h before echocardiography. Echocardiographic parameters included ductal diameter, end-diastolic flow in the left pulmonary artery, left atrium/aorta ratio and ductal flow pattern.

Results rScO2 was significantly related only to ductal diameter over time. Mixed modelling analysed the course of rScO2 over time, where infants were divided into four groups: a closed duct, an open haemodynamically insignificant duct (non-sPDA), a hsPDA, which was successfully closed during study period (SC hsPDA) or a hsPDA, which was unsuccessfully closed during study period (UC hsPDA). SC hsPDA infants showed the highest rScO2 on day 6, while UC hsPDA infants had the lowest rScO2 values.

Conclusions Ductal diameter is the only echocardiographic parameter significantly related to cerebral oxygenation over time. Cerebral oxygenation takes a different course over time depending on the status of the duct. Low cerebral oxygenation may be suggestive of a hsPDA.

  • preterm infants
  • cerebral oxygenation
  • patent ductus arteriosus
  • echocardiography
  • near infrared spectroscopy

Statistics from

What is already known on this topic?

  • A haemodynamically significant patent ductus arteriosus (hsPDA) is a common problem among preterm newborns. Ductal shunting can cause pulmonary overflow and systemic hypoperfusion, also affecting cerebral blood flow. The effect of ductal steal on cerebral oxygenation can be assessed with near-infrared spectroscopy.

What this study adds?

  • Cerebral oxygenation is related to the ductal diameter, as measured by echocardiographic examination. Cerebral oxygenation takes a different course over time, depending on the status of the duct. Even an echocardiographic haemodynamically non-significant open duct seems to affect cerebral oxygenation. Cerebral oxygenation can help the clinician in diagnosing a clinically relevant haemodynamically significant patent ductus arteriosus.


Currently, clinical suspicion of a haemodynamically significant patent ductus arteriosus (hsPDA) arises when symptoms occur such as respiratory instability or cardiac murmurs. The diagnosis is subsequently confirmed by echocardiographic examination.1 ,2 Echocardiographic parameters that point to a hsPDA include a larger ductal diameter, higher end-diastolic flow in the left pulmonary artery (LPAed), an increased left atrium/aorta ratio (LA/Ao ratio) as well as the ductal flow pattern of the growing or pulsatile type.3 ,4 These flow patterns show a left to right shunt through the duct on Doppler examination, where the pulsatile type has a higher peak flow velocity of approximately 1.5 m/s compared with the growing type.3

The brain of preterm infants is vulnerable to disturbances in perfusion and oxygenation. The immature brain develops very fast in the third trimester of pregnancy and uses a significant amount of oxygen.5 ,6 Also, cerebral autoregulation is limited in preterm infants.7 Preterm infants with a hsPDA are especially susceptible to brain damage for several reasons. First of all, the redistribution of blood volume through the open duct from the systemic circulation into the pulmonary circulation (the ducal steal phenomenon) results in decreased cerebral perfusion.8 ,9 Previous studies reported a decrease in cerebral blood flow (CBF) and oxygenation in presence of a hsPDA, as shown by Doppler examination and near-infrared spectroscopy (NIRS).8 ,10 However, not all studies show a statistically significant effect of a duct on cerebral oxygenation.11 ,12 Second, a hsPDA has been associated with decreased blood pressure, even despite a compensatory increase in left ventricular output.13 Finally, a hsPDA has been related to periventricular or intraventricular haemorrhages (PIVH).14 ,15 Cerebral perfusion may be affected to various degrees by a hsPDA, depending on the magnitude of ductal steal, autoregulatory ability of the cerebral vasculature and haemodynamic factors such as cerebral perfusion pressure, myocardial function and left ventricular output. The aim of this study was to investigate the association between the ductus arteriosus, echocardiographic parameters and cerebral oxygenation.

Materials and methods


In this prospective observational study, all infants born <32 weeks of gestational age (GA) admitted to the Neonatal Intensive Care Unit of the Wilhelmina Children's Hospital Utrecht between 2008 and 2010 were included. hsPDA treatment was initiated if clinical suspicion arose with symptoms such as ventilator problems, cardiac murmur or feeding intolerance, and echocardiographic confirmation of the diagnosis. The study was approved by the medical ethical committee of the University Medical Center Utrecht.

Obstetric, intrapartum and neonatal data were obtained from hospital records. Physiological parameters that were monitored included: systemic oxygen saturation (SaO2) using a pulse-oximetry probe (Covidien, Mansfield, Massachusetts, USA), arterial blood pressure by means of an indwelling catheter and heart rate using gel electrodes. Haemoglobin was determined on a regular base. Cranial ultrasound was performed daily and PIVH was graded according to the classification of Papile et al.16 Need for blood pressure support was divided into three categories: none: no support or only fluid administration; mild: dopamine ≤5 µg/kg or severe: dopamine >5 µg/kg or the addition of dobutamine, adrenaline or corticosteroids. The use of inotropes was restricted to blood pressure support management. The diagnosis of moderate-to-severe infant respiratory distress syndrome was based on clinical signs and the need for surfactant therapy. Bronchopulmonary dysplasia was diagnosed, if there was a need for additional oxygen at the corrected age of 36 weeks GA. All treatment decisions were made by attending neonatologists. Infants with haemodynamically significant congenital heart defects or chromosomal abnormalities were excluded.

Infants were divided into four different groups according to the status of their duct during the study period. They either had a closed duct, an open but haemodynamically non-significant PDA (non-sPDA, ie, did not receive treatment during the entire study period) or a haemodynamically significant PDA (hsPDA, ie, received treatment during study period). Infants with a hsPDA were subdivided with regard to their response to treatment: either the duct was successfully closed during the study period (SC hsPDA) or the duct was unsuccessfully closed at the end of the study period (UC hsPDA).

Echocardiographic assessment of the ductus arteriosus

Echocardiographic examinations were performed by attending paediatric cardiologists according to study protocol. All cardiologists were well trained and practiced in neonatal echocardiography. Three serial echocardiographic examinations were performed, on the second, fourth and sixth day after birth. Echocardiographic parameters to assess the duct included internal ductal diameter, LPAed, LA/Ao ratio and presence of a ductal flow type associated with a hsPDA, that is, growing or pulsatile flow pattern.4 Internal ductal diameter was measured on the most constricted point (anatomical diameter based on two-dimensional (2D) echocardiographic examination). The most constricted point was located either by delineating the duct or by visualisation of ductal flow acceleration. In a duct without visible constriction but with a hsPDA-associated flow pattern, the diameter was measured at the pulmonary end of the duct. hsPDA confirmation on echocardiographic examination was based on ductal diameter >1.4 mm, presence of growing or pulsatile ductal flow pattern and/or LPAed >0.2 m/s and/or a LA/Ao ratio >1.4.4 ,17

Echocardiographic examinations were blinded for neonatologists. Indication to deblind the results of the examinations included respiratory discomfort (supplementary oxygen need, unable to wean from respiratory support) or haemodynamic complications (ie, blood pressure instability, cardiac murmur or wide pulse pressure).

Monitoring of cerebral oxygenation

NIRS-monitored regional cerebral oxygen saturation (rScO2) was used as a reliable parameter of cerebral oxygenation as well as a surrogate of CBF.18 It is common practice in our neonatal intensive care for all preterm infants with GA <32 weeks to monitor rScO2 during at least 72 h after birth. We used the INVOS 4100/5100C near-infrared spectrometer (Covidien) with a transducer (SomaSensor SAFB-SM, Covidien) containing a light-emitting diode with two near-infrared wavelengths (ie, 730 and 810 nm) and two distant sensors. The INVOS oximeter is a continuous wave spatially resolved spectrometer.19 It is important to realise when using NIRS to assess cerebral oxygenation that rScO2 can be merely used as a trend monitoring parameter and not as a quantitative measurement. Spatially resolved spectroscopy calculates rScO2 from differences in absorption of near-infrared light.20 ,21 The NIRS (small) adult sensor was attached to the frontoparietal side of the infants' head.22 A stable and reliable 1 h period was selected prior to each echocardiographic examination. A 1 h baseline period was selected between 6 and 12 h after birth. Fractional tissue oxygen extraction (FTOE) was calculated as (SaO2 – rScO2)/SaO2. Analysis of rScO2 was performed using Signal Base software (University Medical Center Utrecht, the Netherlands), designed for rScO2 signal analysis. Measurements were considered as non-representative when rScO2 was below 30% or when abundant artefacts were present and arterial oxygen saturation was below 85%.

Statistical analysis

Data are summarised as mean±SD, count and percentage or as median and range, where appropriate. Student's t test, analysis of variance with Tukey honest significant difference correction, Mann–Whitney U test, Pearson's χ2 test or Kruskal–Wallis test was used to compare patient characteristics and differences in dependent variables between the groups. Two main associations were assessed over time. First is the association between echocardiographic parameters and cerebral oxygenation on the three consecutive time points: (1) second day of life, when the first echocardiographic examination was performed, (2) fourth day of life, during the second echocardiographic examination and (3) sixth day of life, during the third echocardiographic examination. The second association is between the status of the duct and cerebral oxygenation, over the same three time points, with an additional baseline moment: (0) baseline, between 6 and 12 h after birth. In both analyses, the factors GA and being born small for gestational age (SGA) were taken into account as they both can influence cerebral oxygenation. Cerebral oxygenation increases with increasing GA, and being SGA can lead to relative cerebral luxury perfusion.23 Mixed model analysis was used for statistical analysis. This approach was chosen because it is appropriate in an unbalanced data structure with missing values. Both main effects and interactions were examined. The final model was selected based on the best −2 log-likelihood. Statistical significance was set at p<0.05. All statistical analysis was performed with SPSS (IBM Statistics SPSS V.20) or in R for Windows 64-bit, V.3.1.1 (R Core Team, R Foundation for Statistical Computing, Vienna, Austria) with the non-linear mixed-effects package.


A total of 391 infants were included between 2008 and 2010. Eleven infants were excluded for either chromosomal or genetic abnormalities (n=4), metabolic disorders (n=1), congenital anomalies (n=4) or because we were unable to collect data (n=2). This resulted in a study population of 380 infants. Clinical characteristics are shown in table 1. During the study period, 76 infants developed a hsPDA and were treated with indomethacin (n=61) or surgical ligation (n=15). All infants with a hsPDA were treated with indomethacin as soon as possible after diagnosis was confirmed. When indomethacin failed to close the duct or a contraindication was present, surgical ligation was performed. Subanalysis showed no substantial differences in clinical characteristics between infants treated with indomethacin and infants with surgical ligation. During the first echocardiographic examination, 180 infants had a closed duct and 124 a non-sPDA.

Table 1

Important clinical characteristics

Infants with an open duct were significantly younger and had lower birth weights than infants with a closed duct. There was no significant difference in number of SGA infants between groups. Infants with a hsPDA were generally sicker, for example, a larger proportion was ventilated, suffered from sepsis or needed more blood pressure support. Also, the incidence of retinopathy of prematurity was higher in infants with a hsPDA. Lower GA increased the risk of hsPDA. A hsPDA was present in 41 of the 88 infants (47%) with GA <28 weeks, 22 of 118 infants (19%) with GA between 28 and 30 weeks and 13 of 174 infants (7%) with GA between 30 and 32 weeks GA developed a hsPDA.

In 188 infants, the echocardiographic examination was deblinded: 75 showed a hsPDA, 60 showed a non-sPDA and 53 showed a closed duct. For 158 infants, deblinding occurred during the study period. Indications were circulatory (n=110), pulmonary (n=27), infection (n=3), neurological (n=12), cardiologist request (n=3) or due to other reasons (eg, suspicion of a congenital malformation, n=5). Thirty infants were deblinded after the study period due to minor cardiac malformations, which needed to be re-evaluated at the outpatient clinic.

Cerebral oxygen saturation and echocardiographic parameters

Figure 1 shows NIRS-monitored rScO2 values and the echocardiographic parameters LA/Ao ratio and ductal diameter in the four groups (closed duct, non-sPDA, SC hsPDA and UC hsPDA) on three time points (day 2, 4 and 6 of life). Ductal diameter can only be measured in infants with an open duct. Echocardiographic parameters differed statistically significant between the four patient groups, as shown in table 2. For the parameter ductal flow pattern 69% of the data were missing, and for LPAed 67% of the data were missing. Therefore, these parameters were not included in the analysis. Mixed model analysis assessed the association between echocardiographic parameters and rScO2, independent of group membership (ie, status of the duct). The effect of time was taken into account, with a random intercept per patient and a fixed effect for rScO2, time, GA, SGA and the echocardiographic parameter. A statistically significant effect was only obtained for the echocardiographic parameter ductal diameter. The best model was obtained with the association between rScO2 and ductal diameter, GA, being SGA and time. The resulting model was: rScO2=36.29–2.38×time−0.95×ductal diameter+3.10×SGA+1.15×GA (all coefficients p<0.05). The LA/Ao ratio did not show a statistically significant relation with rScO2. The same analysis for FTOE did not show a significant relationship with ductal parameters diameter or LA/Ao ratio. Subanalysis with only hsPDA infants did not improve the association between rScO2 and echocardiographic parameters.

Table 2

Cerebral oxygenation and echocardiographic parameters

Figure 1

Boxplots of cerebral oxygenation and echocardiographic parameters on the second, fourth and sixth day of life. Infants either have a closed duct, a haemodynamically non-significant patent ductus arteriosus (non-sPDA), a successfully closed haemodynamically significant patent ductus arteriosus (SC hsPDA) or an unsuccessfully closed hsPDA (UC hsPDA). Black chequered: rScO2 (%); Grey dots: LA/Ao ratio; Grey solid: ductal diameter (mm). rScO2, regional cerebral oxygen saturation; LA/Ao, left atrium/aorta ratio.

Cerebral oxygen saturation and the ductus arteriosus

Over time, mean rScO2 values tended to be lower in infants with an open duct. However, these differences were not statistically significant. Mixed model analysis assessed the course of rScO2 during the first 6 days after birth depending on the status of the duct, with a random intercept per patient and a fixed effect for: time (as categorical factor), GA, SGA, group (by ductal status) and the interaction time×group. Infants were divided into four groups according to the status of their duct during the study period. They either had a closed duct (n=179), an open non-sPDA (n=125), a hsPDA (n=76), which was either effectively treated (SC hsPDA, n=32) or remained open at the end of the study period (UC hsPDA, n=44). The different slopes and corresponding p values are shown in table 3. Mean rScO2 is significantly determined by time, GA, group (ductal status) and the interaction between time×group. Mean blood pressure had no significant effect.

Table 3

Mixed-effects model of cerebral oxygenation

Figure 2 shows an example of the course of cerebral oxygen saturation over time in infants of 29 weeks GA. Based on the model shown in figure 2, infants who developed a hsPDA started with the lowest rScO2. After 6 days, infants with a SC hsPDA showed the highest rScO2, infants with a closed duct had slightly lower rScO2 values followed by infants with a non-sPDA and infants with a UC hsPDA showed the lowest rScO2. The inverse association existed for FTOE. Infants with a SC hsPDA showed the highest rScO2 with the lowest FTOE, while infants with an UC hsPDA showed the lowest rScO2 with the highest FTOE after 6 days. Subanalysis with hsPDA infants showed that rScO2 was lower in infants who underwent surgery than infants treated with indomethacin (mean 55.3%±9.0 vs mean 63.4%±8.4 SD; p<0.05).

Figure 2

Regional cerebral oxygen saturation (rScO2) for infants of 29 weeks gestational age during the first 6 days of life depending on the status of the duct: closed, haemodynamically non-significant patent ductus arteriosus (non-sPDA), successfully closed haemodynamically significant patent ductus arteriosus (SC hsPDA closed) or unsuccessfully closed hsPDA (UC hsPDA). Time points: (0) baseline within 12 h after birth; (1) day 2; (2) day 4; (3) day 6.


This present study shows that cerebral oxygenation is significantly related only to the echocardiographic parameter ductal diameter, together with GA and being SGA over time. Although not significant, a trend of lower cerebral oxygenation and higher LA/Ao ratio is seen in infants with a hsPDA. Depending on the status of the duct, cerebral oxygenation takes a different course over time. Consistent with previous findings, infants who were surgically ligated showed lower cerebral oxygen saturation than infants who received pharmacological treatment.24 These findings support our hypothesis that a hsPDA has a systemic effect through the ductal steal phenomenon, leading to a reduction of cerebral perfusion and oxygenation. However, a venous component cannot be excluded. Increased pulmonary circulation increases preload of the right heart. An increased venous compartment could lead to venous congestion in the brain and thereby decreasing cerebral oxygen saturation.

A hsPDA has been associated with low cerebral oxygen saturation,8 and PIVH.3 ,15 ,25 The use of cerebral oxygenation as a screening tool to detect a hsPDA has been suggested previously.12 Although it should be noted that NIRS determined rScO2 can be merely used as a trend monitoring parameter to assess changes in cerebral oxygenation and not as a quantitative measurement. It estimates changes in oxygenation with an assumption of path length of the near-infrared light bundle.26 ,27 Assessment of cerebral oxygenation over time can alert the clinician to a (developing) hsPDA. As there is still a lot of debate concerning hsPDA diagnosis and treatment and solid criteria are lacking, cerebral oxygenation could be of additional value in the decision-making process.

It is important to realise while interpreting these results that the haemodynamic impact of ductal steal may be quite variable.3 Ductal flow depends on changes in pulmonary pressure and systemic vascular pressure, both of which are very dynamic and may cause temporary changes in the magnitude and even direction of ductal shunting.3 The consequences of a hsPDA may differ between infants, and hsPDA scoring systems have already been suggested.28–30 To date, there is no real gold standard for the diagnosis of a hsPDA, and more precisely the magnitude of ductal flow and steal. After echocardiographic examination, the attending neonatologist draws the final conclusion. It may be unclear what the exact differences between a hsPDA and a non-sPDA are, as diagnosis is partly subjective. The fluctuating magnitude of the shunt and ductal steal over time prohibits an objective distinction. Also, echocardiographic parameters may be difficult to interpret early in the course of a (developing) hsPDA. Ductal diameter is measured by 2D echocardiography on the most constricted point, which may be difficult to identify. The LA/Ao ratio is derived of left atrium enlargement after volume overload. It signifies increased workload of the left heart after extensive ductal shunting, it does not directly assess haemodynamic severity of the duct.31 It takes time for left atrium enlargement to develop and our centre starts treatment early, so it might not be sufficiently developed yet during the study period. LPAed and ductal flow pattern both seemed promising as reflections of hsPDA severity, as they are directly related to ductal steal.32 Unfortunately, our study contained too many missing values to make valid statements concerning the course of these parameters over time. However, a previous study has shown that ductal diameter is significantly associated with ductal flow pattern.33 Although intrarater and interrater variability of echocardiographic examinations were not included in the study design, all examinations were performed by well-trained and experienced paediatric cardiologists. Echocardiographic examination in itself can challenge the clinical stability of sick preterm infants, sometimes requiring adaptation of ventilator settings. Therefore, non-invasive monitoring with NIRS may sometimes be a reasonable alternative to assess hsPDA severity by evaluating the effect of ductal steal on cerebral oxygenation. Without a gold standard for hsPDA diagnosis, assessment of cerebral saturation can aid the clinician together with echocardiographic examination. Our group previously showed that cerebral oxygenation is lower in infants with a hsPDA. Prolonged low cerebral oxygenation (rScO2 <40%) has been associated with brain damage.34 ,35 RScO2 recovers to normal values after ductal closure.8

A hsPDA is a common complication in all preterm infants, but occurs most often in the youngest infants. In this study, the prevalence of hsPDA was also highest in infants born before 28 weeks of GA. However, in our centre all infants born before 32 weeks of gestation with a hsPDA are usually treated early and proactively. There is an ongoing debate among neonatologist when and for which patients treatment is most suitable. To increase generalisability of our results to centres who only treat infants with GA <28 weeks, we have added GA as an independent factor in our analyses.

rScO2 takes a different course over time depending on the status of the duct. Overall, a general decline is seen over time. This finding is in accordance with a previous study by our group, providing reference values of cerebral oxygenation during the first 3 days of life.23 This decline is presumably caused by physiological changes in haemoglobin, as well as relative low haemoglobin levels due to frequent blood testing on a neonatal intensive care unit. Roche-Labarbe et al36 reported a similar declining pattern. As stated previously, in our centre hsPDA treatment is initiated early. This is presumably why extremely low values of cerebral oxygenation were not observed during the first 6 days after birth. The reasoning behind early treatment is underlined by a recent study by Rozé et al, who reported that early screening and treatment for a hsPDA was associated with lower in-hospital mortality and likelihood of pulmonary haemorrhage.37 In our study, infants with a non-sPDA show slightly lower cerebral oxygenation after 6 days and a sharper decline over time compared with infants with a closed duct. This could mean a (small) haemodynamic effect of the duct after all, and the term haemodynamically insignificant might not be entirely appropriate. Cerebral oxygenation is influenced by perfusion as well as oxygen transport capacity. Haemoglobin and SaO2 were kept within normal range. The influence of other factors affecting tissue oxygenation other than hsPDA, such as circulatory complications, was negligible in our study population. Future research should focus on the contributing effect of NIRS-monitored cerebral oxygenation in hsPDA diagnosis, especially when echocardiographic examination is inconclusive or in contradiction with clinical symptoms.


Cerebral oxygenation is related to ductal diameter during the first 6 days of life in infants born <32 weeks GA, together with GA and being SGA. A larger diameter is associated with a lower cerebral oxygen saturation. There appears to be a trend, although not significant, towards lower cerebral oxygenation and a higher LA/Ao ratio. Cerebral oxygenation takes a different course over time, depending on the status of the duct. Infants with an effectively closed hsPDA recover to normal cerebral oxygenation, whereas infants with a hsPDA that remains open show the lowest values of cerebral oxygenation at day 6. These infants could represent a separate patient group who do not respond well to treatment, or are recognised or develop a hsPDA at an older stage. Infants with an open but haemodynamically insignificant ductus arteriosus show lower cerebral oxygenation and a sharper decline over time compared with infants with a closed duct. This could be due to the ductal steal phenomenon. Monitoring cerebral oxygenation can help the clinician in diagnosing and evaluating treatment of a hsPDA.


View Abstract


  • Twitter Follow Daniel C Vijlbrief @vijlbrief01, @WillemBdeVries, @HansBreur, Willem B de Vries, @WillemBdeVries and Johannes Breur, @HansBreur

  • Contributors LD: collected data, performed analysis and wrote both the main as the revised manuscript. MM: conceptualising study concept, collected data, critically revised content of manuscript. HB: conceptualising study concept, collected data, critically revised content of manuscript. WdV: conceptualising study concept, critically revised content of manuscript. DV: conceptualising study concept, critically revised content of manuscript. FG: statistics FVB: conceptualising study concept, critically revised content of manuscript. PL: conceptualising study concept, data collection, critically revised content of manuscript. Final responsibility for manuscript content.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval Medical Ethical Committee of the UMC Utrecht.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.