Cerebral desaturation during neonatal congenital heart surgery is associated with perioperative brain structure alterations but not with neurodevelopmental outcome at 1 year

Abstract

Open in new tabDownload slide

OBJECTIVES

The significance of intraoperative cerebral desaturation (CD) measured by near-infrared spectroscopy (NIRS) to predict neurological outcome after congenital heart surgery is uncertain. The goal of this study was to compare brain structure changes and neurodevelopmental outcome in patients with severe congenital heart disease with and without intraoperative CD.

METHODS

Neonates requiring congenital heart surgery were enrolled in a cohort study. NIRS data from their first cardiac operation were collected. Pre- and postoperative brain magnetic resonance imaging results and Bayley-III scores at 1 year were compared between patients with and without CD, defined by 2 NIRS thresholds: regional cerebral oxygen saturation (rSO₂) of 45% (45%rSO₂) and rSO₂ below 20% of baseline value (20%BLrSO₂).

RESULTS

Thirty-two patients (72% male) with d-transposition of the great arteries (n = 24, 75%) and other complex types of congenital heart diseases (n = 8, 25%) were analysed. Perioperative relative lateral ventricle volume change was increased in patients with versus without intraoperative CD (P = 0.003 for 45%rSO₂, P = 0.008 for 20%BLrSO₂). For 45%rSO₂, the effect of CD remained significant after adjusting for age at postoperative scan, time between scans and cardiac diagnosis (P = 0.019). New intracranial lesions occurred predominantly in CD groups (6/6 patients for 45%rSO₂, 5/6 patients for 20%BLrSO₂). Neurodevelopmental outcome at 1 year was not associated with intraoperative CD.

CONCLUSIONS

This study demonstrates the clinical relevance of NIRS monitoring during congenital heart surgery. The occurrence of intraoperative CD is associated with perioperative lateral ventricle volume change and new intracranial lesions.

INTRODUCTION

Survival of patients with severe congenital heart diseases (CHD) undergoing neonatal cardiopulmonary bypass (CPB) has increased due to advances in surgical and perioperative management. The well-known risk of long-term neurodevelopmental (ND) deficits requires a better understanding of predictive factors. Patients with CHD are at risk for brain dysmaturity at birth [1]. In addition, perinatal brain growth and development continue at a slower pace compared to healthy controls [2].

As a major invasive event, neonatal CPB may lead to intraoperative cerebral desaturation (CD) [3]. Near-infrared spectroscopy (NIRS) is used for intraoperative, non-invasive neuromonitoring [4] by measuring regional cerebral oxygenation saturation (rSO₂) using near-infrared light-emitting optodes placed on the patient’s forehead [5].

Intraoperative CD detected by NIRS has been associated with new postoperative brain injuries [6, 7]. As an additional short-term ND variable, perioperative brain growth may provide further information on the effects of intraoperative CD. Whereas brain volumes increase by more than 100% within the first year of life, the brain growth rate of patients with CHD may be negatively impacted by perioperative CD [8]. Little knowledge exists on the effect of CD during congenital heart surgery on perioperative brain growth.

The goal of this study was to assess the impact of intraoperative NIRS (i) on perioperative brain growth, (ii) on changes of brain morphology, both assessed by magnetic resonance imaging (MRI) and (iii) on the ND outcome at 1 year in patients with CHD.

PATIENTS AND METHODS

Ethics statement

This study was approved by the ethical committee of Kanton Zürich, Switzerland (Kantonale Ethikkommission Zürich, StV-23/619/04). Parental written informed consent was obtained.

Study design and patient population

This study is part of a prospective cohort study of the Research Group Heart and Brain [9]. Between December 2009 and March 2020, neonates with severe types of CHD requiring neonatal cardiac surgery within the first 6 weeks of life were consecutively included in the cohort study. Neonates <36 weeks of gestational age at birth and those with clinical suspicion or evidence of genetic malformation syndromes associated with ND disabilities were excluded. For this secondary, exploratory analysis, we included neonates undergoing CPB for CHD at our institution with systematic intraoperative NIRS data and pre- and postoperative brain MRI.

Intraoperative near-infrared spectroscopy

Cerebral rSO₂ was routinely measured during cardiac surgery using INVOS 5100C Cerebral/Somatic oximeter (Medtronic, Minneapolis, MN, USA). Unilateral NIRS monitoring of the right cerebral hemisphere was used in every patient, whereas bilateral NIRS monitoring was used in patients requiring Norwood stage I palliation or any type of surgery involving the aortic arch. NIRS baseline values were set after intubation, in haemodynamically stable patients with appropriate inspired oxygen concentration before starting surgery. The INVOS 5100C was configured at an rSO₂ 20% below baseline value, and cumulative CD over time was produced as the area under the curve (AUC). We used the INVOS Analytics Tool (https://www.medtronic.com/covidien/en-us/products/cerebral-somatic-oximetry/invos-analytics-tool.html; accessed 18 March 2021). All NIRS data files were reviewed for data accuracy (B.K.). AUC values using a threshold of 45% absolute rSO₂ (45%rSO₂) and 20% below baseline rSO₂ (20%BLrSO₂) were calculated as frequently done by others [3, 10, 11]. Because AUC values were non-normally distributed, patients were binarized for statistical analysis into a CD group (AUC > 0, in 1 or both cerebral hemispheres) and a no-CD group (AUC = 0) for both NIRS thresholds. The groups were subsequently labelled as 45%rSO₂ or 20%BLrSO₂, CD and no-CD group.

Cardiopulmonary bypass

For the arterial switch operation (ASO), the arterial CPB cannula was placed in the distal ascending aorta. For Norwood I and isolated aortic arch reconstruction, the ductus arteriosus was cannulated for perfusion of the descending aortic territory and the brachiocephalic trunk, for antegrade cerebral perfusion (ACP). During CPB a mean arterial pressure of 40–50 mmHg was targeted using a blood flow rate of 100–150 ml/kg/min. For ACP a pump rate of 30–50 ml/kg/min was used with monitoring of the mean arterial pressure via the right radial artery. ASO was performed under mild-to-moderate hypothermia (28–32°C), and Norwood stage I palliation, under moderate-to-deep hypothermia (18–25°C). Hemofiltration was performed routinely during CPB.

Brain magnetic resonance imaging and image post processing

Brain MRI was performed during natural sleep on a 3.0 T GE MR750 scanner (GE Healthcare, Milwaukee, WI, USA) using an eight-channel head coil (GE Signa MR750). No contrast agent was used. The scanning protocol has been described previously [12]. Structural 2-dimensional T2-weighted fast spin-echo images acquired in all 3 planes were used for three-dimensional (3D) super-resolution images (resolution of 0.5x0.5x0.5 mm). Each image was bias field corrected using the “N4ITK” filter [13] from the 3DSlicer software [14] and denoised with the program “btkNLMDenoising” of the Baby Brain Toolkit (BTK) [15]. The 3D super-resolution image was created with the slice-to-volume reconstruction toolkit software [16].

Brain volumetry

The automated developing Human Connectome Project structural pipeline [17] was used to segment the reconstructed 3D image data. The intracranial volumes analysed were (i) total brain volume (TBV), (ii) bilateral lateral ventricle volume (LVV) and (iii) cerebrospinal fluid volume excluding lateral ventricles (CSFV) (Fig. 1). TBV analysis was done using absolute volumes in milliliters. Volumetric analysis of LVV and CSF was conducted using absolute as well as relative volumes as a percentage of TBV to adjust for interindividual differences (relative LVV and CSFV). A volumetric change rate per week, quantifying differences between the pre- and postoperative scans, was subsequently calculated for TBV, LVV and CSFV.

Brain lesion scoring

Brain lesions were scored by a neuroradiologist (R.K.) and neonatologist experienced in brain imaging (C.H.). Postoperative intracranial lesions were defined as new if they had not been present on the preoperative scan or had increased in size or number.

Neurodevelopmental outcome

ND outcome was assessed at 1 year, using the Bayley Scales of Infant and Toddler Development, Third Edition (Bayley-III) [18]. The Bayley-III assessment provides 3 main composite scores: the cognitive, the language and the motor composite score, with a mean score of 100 and a standard deviation (SD) of 15 for healthy infants. Bayley-III scoring was administered by developmental paediatricians [12] blinded to NIRS and MRI data.

Clinical variables

Patient- and procedure-related factors collected for analysis are specified in Table 1. Perioperative factors for analysis included length of hospital stay (LOHS), length of paediatric intensive care unit stay and the number of cardiac invasive interventions (surgical and catheter-based procedures including main neonatal heart surgery). Cardiac diagnoses included d-transposition of the great arteries (d-TGA) and other complex CHD types (labelled as “others”). The socioeconomic status (SES) was calculated using a 6-point scale of both paternal occupation and maternal education and ranged from 2 (lowest) to 12 (highest) [19].

Statistical analyses

Statistical analyses were conducted using R (version 3.6.3, The R Software Foundation for Statistical Computing, Vienna, Austria). Descriptive statistics were expressed as mean and SD, median and interquartile range for continuous and discrete variables, and number and percentage for categorical variables. Normality of variables was assessed by visual inspection.

Patient characteristics, perioperative volumetric change rates, Bayley composite scores and perioperative factors were compared between patients with and without intraoperative CD using an unpaired t-test or Fisher’s exact test. Perioperative brain changes of the entire study cohort were analysed using the paired t-test. Analysis of covariance models were applied to evaluate the impact of CD on volumetric changes, while adjusting for age at postoperative MRI, time between pre- and postoperative MRI and cardiac diagnosis (d-TGA vs others). Covariates were included for theoretical criteria. Linear models were used to evaluate CD group association with Bayley-III outcome adjusting for SES and cardiac diagnosis (d-TGA vs others).

A P-value of <0.025 (Bonferroni correction for using 2 NIRS thresholds) was considered significant for all tests.

RESULTS

Patient characteristics

Intraoperative NIRS data of 75 patients were available. Thirty-two patients had pre- and postoperative brain MRI scans and were included for analysis. Of these, 31 (97%) patients had completed ND outcome testing at 1 year (Fig. 2).

Cardiac surgery was performed at a mean postmenstrual age of 41.6 (SD: 1.4) weeks or at 14.4 (SD: 5.8) days of life, respectively. Most patients (n = 24, 75%) underwent ASO. In 17 (71%) patients with d-TGA, a preoperative balloon atrial septostomy was performed. Patient- and procedure-related factors are summarized in Table 1. In total, 8 patients (25%) were diagnosed prenatally.

Intraoperative near-infrared spectroscopy

Seventeen (53%) patients had bilateral cerebral NIRS monitoring and 15 (47%) patients had unilateral NIRS monitoring. Mean baseline rSO₂ of the entire patient cohort was 50.9 (SD: 10.3). The difference between right- and left-sided baseline values in patients with bilateral NIRS monitoring was a mean of 5.2 (SD: 4.2). Applying the 45%rSO₂ threshold, 8 (25%) patients were in the no-CD group and 24 patients (75%) were in the CD group. Median (interquartile range) 45%rSO₂ AUC values for the right and left cerebral hemispheres were 0 (0–16.0) and 47.0 (0–337.0), respectively. Applying the 20%BLrSO₂ threshold, 17 (53%) patients were in the no-CD group, and 15 (47%) patients were in the CD group. The 20%BLrSO₂ AUC values for the right and left cerebral hemispheres were 0 (0–102.5) and 0 (0–103.0), respectively. Baseline rSO₂ was lower in patients in the CD group than in the no-CD group of the 45%rSO₂ threshold.

Brain magnetic resonance imaging findings

Pre- and postoperative brain MRI scans were performed within a time interval of 19.3 (SD: 7.1) days. The absolute TBV and LVV increased perioperatively, whereas relative LVV and absolute and relative CSFV showed no evidence of a significant change (Table 2). Mean volume change per week was 13.08 (SD: 6.70) ml for absolute TBV, -0.01 (SD: 0.15) %TBV for relative LVV, and –0.35 (SD: 1.90) %TBV for relative CSFV. No pathological liquor space enlargement was found.

New postoperative intracranial lesions were found in 6 of 32 (19%) patients. New extra-axial haemorrhages occurred most frequently (n = 4), followed by new punctate white matter lesions (n = 3). No new cerebral stroke or parenchymal haemorrhage was found postoperatively in any patient (Supplementary Table 1).

A comparison of brain MRI findings of patients with and without CD is presented in Table 3. For both NIRS thresholds, absolute and relative LVV (%TBV) change per week differed between the no-CD and the CD group. The relative LVV change per week predominantly decreased in the no-CD groups and predominantly increased in the CD groups (for 45%rSO₂ -0.14%TBV in the no-CD group and 0.03%TBV in the CD group, P = 0.003; for 20%BLrSO₂ -0.08%TBV in the no-CD group and 0.06 in the CD group, P = 0.008) (Fig. 3). When adjusting the analysis for CHD diagnosis (d-TGA vs others) in an analysis of covariance model or when analysing the subpopulation of d-TGA patients only (24 patients, 75%), the increase of relative LVV change per week remained higher in the CD group compared to the no-CD group of the 45%rSO₂ NIRS threshold (adjusted P = 0.019, Supplementary Table 2; P = 0008, Supplementary Table 3). A supplementary brain deformation analysis supports these results by visualization of LVV shrinkage in the 45%rSO₂ no-CD group and LVV growth in the 45%rSO₂ CD group (Supplementary Analysis 1).

For the 45%rSO₂ threshold, new brain lesions occurred solely in the CD group. For the 20%BLrSO₂ threshold, 5 of 6 patients with new lesions were in the CD group.

Neurodevelopmental outcome

The ND outcome determined by Bayley-III at 1 year of age for the entire cohort was 107.6 (SD: 15.6) for cognitive, 93.9 (SD: 14.4) for language and 93.2 (SD: 4.6) for motor composite score.

A comparison of neurodevelopmental outcomes of patients with and without CD is presented in Table 3. ND outcomes at 1 year of age did not differ between the CD and the no-CD groups even after adjustment for SES and CHD diagnosis (d-TGA vs others). In a supplementary analysis of the study cohort including all available patients with NIRS and Bayley-III data (n = 71) irrespective of MRI examinations, ND outcomes remained comparable between patients with and without CD.

Perioperative course

For the entire population, LOHS was 34.3 (SD: 11.9) days and the ICU stay, 15.4 (SD: 12.6) days. A comparison of the perioperative course of patients with and without CD is presented in Table 3. LOHS was prolonged in the 20%BLrSO₂ CD group and tended to be prolonged in the 45%rSO₂ CD group, compared with the respective no-CD groups (for 45%rSO₂ 26.6 days vs 36.9 days, P = 0.032; for 20%BLrSO₂ 28.4 days vs 41.1 days, P = 0.001) (Fig. 3).

A comparison of patient- and procedure-related characteristics of patients with and without CD is presented in Table 1. The 20%BLrSO₂ groups differed in cardiac diagnosis (d-TGA vs other; P = 0.013) and type of surgery (ASO, Norwood I, other; P = 0.010). A total of 94% of patients in the no-CD group and 53% patients in the CD group had d-TGA diagnosis and ASO surgery.

DISCUSSION

Around 50% of neonates with CHD in our analysis had intraoperative CD according to the predefined NIRS thresholds (45%rSO₂ and 20%BLrSO₂). These patients showed an increased volume change in absolute and relative brain LVV compared to patients without CD. The differences in volume change were most prominent for LVV and were a trend for CSFV, but the change rate of TBV did not differ between the groups. Patients with CD had a higher incidence of new intracranial lesions, with extra-axial haemorrhages and white matter lesions. Of note, at 1 year of age, no difference in ND outcome was detected in our patient population.

In clinical daily use, low rSO₂ values in NIRS neuromonitoring trigger prompt evaluation of modifiable factors by surgeons, anaesthetists and perfusionists to preclude CD [4], which among others includes repositioning the aortic cannula, increasing the CPB flow and blood pressure, elevating the blood haemoglobin level and administering drugs with attributed neuroprotective properties [10]. A clinical impact of the 2 NIRS thresholds, 20%BLrSO₂ and 45%rSO_2, has been reported by their associations with cerebral ischaemia, higher frequency of neurological complications, prolonged LOHS [10] and new or worsened ischaemic brain lesions [7].

Despite undergoing congenital heart surgery and the need for paediatric intensive care, brain growth and brain development in neonates with CHD continues, although at a slower pace compared to healthy controls [2]. Our findings with a measured TBV growth with a mean of 13.1 ml per week are comparable with the findings of others [2].

The perioperative increase of cerebrospinal fluid in patients with CD, particularly in the lateral ventricles, may be caused by injury to the brain parenchyma due to intraoperative hypoxia with subsequent passive ventricular dilatation. Nevertheless, we did not find a difference in TBV change between CD and no-CD groups that would implicate parenchymal loss (TBV decrease) or even parenchymal oedema (TBV increase). The lack of significance when looking at the difference in relative CSFV change is possibly biased by the anatomical difficulties in brain segmentation. Whereas the anatomical margins of lateral ventricles are rather smooth, the extracerebral fluid space margins are more challenging to segment due to the complex, convoluted shape of the cortical surface.

A further mechanism potentially leading to a change in liquor hydrodynamics is the perioperative alteration in venous pressure due to perioperative fluid management. Liquor space enlargement has been described in patients with CHD at 2 years of age undergoing the Fontan procedure [20], where it was explained by a severely altered venous backflow due to the existing Glenn physiology with a chronic increase of central venous blood pressure. In this age group, CSF enlargement was also linked to ND outcome [20]. We did not measure intracranial pressure but we estimate the central venous pressure to have been elevated within the first postoperative days. The use of diuretics and their subsequent effect on brain water content, intracranial pressure and plasma osmolality may be contributing factors [21, 22]. Liquor volume has been described to be enlarged in adults with CHD [23], but for our study population it is unclear whether the observed changes in altered intracranial liquor spaces will persist.

New postoperative intracranial lesions predominantly occurred in the CD groups of both NIRS thresholds: in 3 patients with d-TGA, 2 patients with HLHS and 1 patient with l-TGA. New postoperative lesions have been associated with intraoperative [6] and postoperative [7] rSO₂.

ND outcome at 1 year of age was comparable for patients with and without intraoperative CD. Lower intraoperative rSO₂ [24], lower rSO₂ during a postoperative period following CPB [11] and rSO₂ nadir in combination with LOHS [25] were found to predict ND outcome at 1 or 2 years of age. Nevertheless, other studies [4] revealed no correlation between intraoperative NIRS and ND outcome [3, 26, 27]. Still, a critical rSO₂ score as a cut-off value remains difficult to define, and furthermore, this value may vary according to the metabolic requirements of the neonatal brain during different perioperative periods [28].

We found cardiac diagnosis and corresponding cardiac surgery to be associated with CD. Underlying reasons for this finding may lie in the delicate haemodynamic pathophysiology associated with CHD and the technique required for surgical repair or palliation. In patients with HLHS or for aortic repair, ACP and moderate-to-deep hypothermic circulatory arrest are required for surgery, whereas they are usually not required for ASO or intracardiac surgery. In this context, we interpret the finding of a lower baseline rSO₂ in patients with CD in the 45%rSO₂ group. An a priori lower baseline rSO₂ in patients may be considered to reflect the severity of the underlying CHD associated with a varying degree of venous blood shunting and low cardiac output. Our findings underscore that special care must be taken to prevent or expeditiously ameliorate cerebral desaturation particularly in patients with a complex CHD and/or low baseline rSO₂. Other procedure-related characteristics (e.g. duration of surgery, bypass time) were not associated with intraoperative CD.

After adjusting our brain growth analysis for diagnosis (and corresponding surgery), we were able to validate the association of intraoperative CD and altered LVV change. Despite the fact that in the recent literature, type of CHD was found to be most predictive for brain growth trajectory [29] and occurrence of brain lesions [6], we revealed a significant effect of intraoperative CD on short-term intracranial liquor space development even after a statistical adjustment for CHD diagnosis. Furthermore, we found CD to be associated with longer LOHS. CD may therefore be a marker for a vulnerable intraoperative and complicated clinical course, independent of the type of CHD diagnosis.

Limitations

Several limitations apply to this secondary exploratory study. NIRS monitoring was temporally limited only to the intraoperative period and spatially limited to the frontal brain cortex because the optodes were placed on the patients’ foreheads. Individual NIRS baseline values must be set by the user.

A further limitation of CHD studies is the small and heterogeneous study cohort. In particular, operations involving selective cerebral perfusion and single ventricle hearts differ significantly from normal whole body perfusion and biventricular congenital hearts. Nevertheless, we found CD to be associated with alterations in LVV change, independent of CHD diagnosis.

For safety reasons, study MRI scans were only performed in patients whose haemodynamic conditions were stable. Alterations in brain volume measurements and incidence of brain lesions may therefore be under-reported, and ND outcome at 1 year of age may be overestimated.

The ND outcome measurement is limited to the age of 1 year.

CONCLUSIONS

The association between intraoperative CD, perioperative change of LVV and new postoperative lesions in neonates undergoing surgery for CHD highlights the clinical relevance of intraoperative NIRS monitoring. However, the occurrence of CD is not associated with adverse ND outcome at 1 year of age. The long-term effect of the observed differences, as well as their impact on ND outcome beyond 1 year of age, need to be assessed in further studies.

SUPPLEMENTARY MATERIAL

Supplementary material is available at EJCTS online.

ACKNOWLEDGEMENTS

The authors thank Ulrike Held, Epidemiology, Biostatistics and Prevention Institute, Department of Biostatistics, University of Zurich; Zurich, Switzerland.

Funding

The work was supported by the Swiss National Science Foundation (grant number 320030_184932), Vontobel Foundation, OPO Foundation, the Prof. Dr. Max Cloetta Foundation, the Anna Müller Grocholski Foundation, the Foundation for Research in Science and the Humanities at the University of Zurich, the EMDO Foundation, the Hasler Foundation, the Zurich Neuroscience Center PhD Grant, the Forschungszentrum für das Kind (FZK) Grant, the University Research Priority Program Adaptive Brain Circuits in Development and Learning (AdaBD) and the Swiss National Science Foundation SPARK Grant (grant number CRSK-3_190638).

Conflict of interest: none declared.

Data Availability Statement

The data underlying this article are available in the article and in its online supplementary material. Further data cannot be shared publicly for ethical reasons.

REFERENCES

Abbreviations

 
• ACP

  Antegrade cerebral perfusion

 
• ASO

  Arterial switch operation

 
• AUC

  Area under the curve

 
• CD

  Cerebral desaturation

 
• CHD

  Congenital heart disease

 
• CPB

  Cardiopulmonary bypass

 
• CSFV

  Cerebrospinal fluid volume excluding lateral ventricles

 
• d-TGA

  d-transposition of the great arteries

 
• LOHS

  Length of hospital stay

 
• LVV

  Lateral ventricle volume

 
• MRI

  Magnetic resonance imaging

 
• NIRS

  Near-infrared spectroscopy

 
• ND

  Neurodevelopmental

 
• rSO₂

  Regional Oxygen saturation

 
• SD

  Standard deviation

 
• SES

  Socioeconomic status

 
• TBV

  Total brain volume

 
• 20%BLrSO₂

  20% below baseline rSO₂

 
• 45%rSO₂

  45% of absolute rSO₂

Author notes