The neonate undergoes profound changes in pulmonary and systemic circulations over the first week of life, with overlying developmental influences on circulatory controls of cerebral and renal-mesenteric systemic beds (1–4). While both local autoregulatory mechanisms and sympathetic nervous system modulation of regional resistance are present throughout the systemic circulation, the differential physiologic controls over cerebral and somatic regions result from profoundly different magnitudes of these mechanisms. In addition to vulnerabilities related to the transitional circulation and the neonatal myocardium, neonates with univentricular heart disease have vulnerabilities related to decreased ventricular mass, obligate mixing with arterial desaturation, and the pulmonary-systemic tradeoff inherent in parallel circulation (5–8). Such circulatory limitations increase the risk for hypoxic-ischemic injury in both cerebral and renal/somatic circulations (3, 9).

In neonates with univentricular parallel circulatory anatomy, the various systemic regional circulations compete with each other and with the pulmonary circulation. Monitoring cerebral and renal/somatic circulations with near-infrared spectroscopy (NIRS) in neonates before, during, and after surgical palliation of hypoplastic left heart syndrome (HLHS) with the Norwood procedure, has helped uncover circulatory vulnerabilities, describe relationships between monitored parameters and outcomes, and determine targets for intervention (3, 8, 10–14). Both cerebral and renal-somatic organ beds are at risk for hypoxic injury following Norwood palliation of HLHS (9, 11, 15–21). Hypothermic cardiopulmonary bypass produces significant postoperative changes in cerebrovascular resistance (3, 22–27). Resulting early postoperative cerebral desaturation has been linked to later neurodevelopmental delays (11, 28–30). Perioperative renal/somatic desaturation has been associated with acute kidney injury as well as necrotizing enterocolitis (31–33). The combination of arterial blood pressure and renal oxygenation by NIRS can define a low cardiac output condition associated with increased mortality (13).

Modulation of systemic and pulmonary vascular resistance with induced changes in arterial carbon dioxide tension has been utilized to affect systemic-pulmonary balance, with particularly prominent effects noted on the cerebral circulation (34–37). In univentricular parallel circulation, deliberate modulation of regional resistances may be helpful to avoid end-organ injury, since total oxygen delivery may be critically limited (5, 38, 39). In this report, we describe the complex effects of carbon dioxide on systemic cerebral and renal regional circulations in neonates following surgical palliation of HLHS with a Norwood procedure.

The patient population included neonates with HLHS undergoing a Norwood procedure for arch reconstruction and systemic to pulmonary shunt placement in the first month of life. Patients underwent preoperative stabilization with prostaglandin infusion, and treatment with sedatives, inotropic-vasoactive infusions, and mechanical ventilation as necessary. Monitoring included continuous assessment of arterial oxyhemoglobin saturation (SaO2) by pulse oximetry, mean arterial pressure (MABP) transduced from an umbilical or radial artery catheter, central venous or atrial pressure from an umbilical venous or atrial catheter, and regional tissue oxyhemoglobin saturation by NIRS with probes over the midline-right forehead and right T12 to L2 flank regions for cerebral (rSO2C) and renal-somatic (rSO2R) fields, respectively. Surgical procedures were performed with deep hypothermic cardiopulmonary bypass (CPB) at 18–20 degrees Celsius with pH stat blood gas management, alpha-adrenergic blockade (AAB), and with selective antegrade cerebral perfusion (ACP) to limit duration of deep hypothermic circulatory arrest (DHCA), under high dose opioid based volatile supplemented anesthesia (3, 13, 40–42). Infusions of milrinone, epinephrine, and norepinephrine were titrated prior to weaning from CPB to achieve a calculated systemic vascular resistance index of about 12 Wood units. Physiologic targets following cardiopulmonary bypass included SaO2 75–85%, MABP > 50 mmHg, rSO2C > 50%, rSO2R > 60%. Patients were recovered in the cardiac ICU with continuation of opioid and vasoactive infusions, red blood cell transfusion, elective mechanical ventilation, controlled normothermia, and planned delayed sternal closure to achieve these physiologic goals.

Demographic surgical physiologic and hemodynamic data were collected at hourly intervals for the first post-operative 48 h following the Norwood procedure and were stored in a clinical registry with IRB approval. The primary physiologic outcome measures were cerebral (rSO2C) and renal (rSO2R) field saturations, regional venous saturations (SvO2C, SvO2R), and arterio-venous differences (ΔSavO2C, ΔSavO2R) calculated from simultaneous measures using a 25%/75% arterial/venous field saturation model. Arterial blood samples were obtained at standard intervals with linear interpolation for values synchronous with hourly recordings of physiologic monitoring data. Details of measures and derivations are shown in Table 1. Physiologic data were excluded from analysis during periods of extracorporeal circulatory support. Data were summarized as mean plus or minus standard deviation, median and interquartile range and 5–95% confidence intervals.

Univariable fixed-effects models were used to test association of the primary variables regional field saturations and arteriovenous differences with linear and non-linear components for PaCO2 and postoperative time. Multivariable mixed effect models were then used to define associations of primary variables with interactions of PaCO2 and time, using other demographic, treatment, and physiologic measures as covariates. The combined effects of PaCO2 and time on the primary physiologic outcome measures were derived from these models with defined values for covariates and displayed in graphical form. Finally, the adjusted marginal effects of changes in PaCO2 on outcome measures were calculated for each postoperative hour. All models used cross-sectional panel regression methods to allow within-patient and between-patient variance specifications robust to non-independent clustering and non-normality. Model coefficients were expressed with point estimate ± standard error, with significant effects identified at p < 0.05 (^*), p < 0.01 (^**), and p < 0.001 (^***) levels. Analyses were performed using Stata Software (v17, RRID:SCR_012763) and Python Programming Language (v3.8, RRID:SCR_008394).

This study used NIRS to measure cerebral and renal/somatic oxygenation and blood flow following the Norwood procedure in nearly 200 neonates. Several novel observations were evident in this study. The major finding was the differential effect of changes in PaCO2 on cerebral and renal/somatic oxygenation and blood flow. These findings are consistent with known effect of increasing PaCO2 on cerebral vasodilation (34, 43, 44) and pulmonary vasoconstriction (45, 46), both of which might increase cerebral blood flow (47–50). However, the reduction in renal/somatic blood flow with increased PaCO2 suggests that the effects of PaCO2 on the pulmonary circulation were not predominant, and an alternative mechanism might be at work. If total systemic blood flow is relatively constant, then a reduction in cerebrovascular resistance with increased PaCO2 will divert blood flow from renal/somatic to the cerebral bed, consistent with the effects we observed. Our findings suggest that manipulation of PaCO2 is an intervention to alter the distribution of systemic blood flow by exploiting the differential sensitivity of cerebral and renal circulations to PaCO2, particularly when total systemic flow is limited.

The effects of CO2 exhibited significant time-dependence, with increased CO2 sensitivity of the cerebral circulation in the early postoperative period. This finding might be related to the pH-stat CPB management strategy that utilizes deliberate induction of extreme hypercapnia (PaCO2 80–100 torr) to counteract the leftward shift in oxyhemoglobin affinity during deep hypothermia. Regardless of duration of whole-body CPB, ACP, or DHCA, these patients were treated using pH-stat blood gas management strategies to maintain tissue oxygen tension during all phases of deep hypothermia. Hysteresis in the CO2-cerebrovascular resistance relationship (45) could therefore explain the observed shift over the first postoperative day. These observations suggest that targeted modulation of PaCO2 with deliberate hypercapnia in the early postoperative period is a physiologically rational strategy to normalize cerebral hemodynamics following significant exposure to pH-stat hypothermic CPB.

The observed relationships of Hb and SaO2 on regional oxygenation and blood flow were largely consistent with known effects and each contributed significantly to oxygenation in both cerebral and renal fields. The regression coefficient for SaO2 on both rSO2C and rSO2R was very close to the theoretical 25% arterial contribution to field saturation from the NIRS device calibration supporting the generalizability of the validation model (51–57). This empiric validation of the model also implies corresponding accuracy for derivation of regional venous saturation, arteriovenous difference, and fractional extraction by applying this model to synchronous SaO2 and rSO2 measures.

The differential relationship of MABP to cerebral and renal oxygenation deserves comments. Univentricular parallel circulation is characterized by an accentuated flow-resistance tradeoff in the systemic circulation, particularly in mesenteric and renal beds. Activation of the sympathetic nervous system with mesenteric/renal vasoconstriction is common following major operative intervention, contributing to systemic hypoperfusion, cellular hypoxia and shock even with preservation of blood pressure (40, 58, 59). We have previously described a potential flow-pressure tradeoff (60) and low cardiac output state utilizing renal NIRS (13) following the Norwood procedure with a more complex relationship on renal oxygenation described herein. Pressure autoregulation for cerebral circulation undergoes important developmental changes, and can be impaired by hypothermia and hypercapnia (37, 61–64). The positive association between MABP and rSO2C could be explained by MABP below a theoretical lower limit of autoregulation or with impairment of autoregulation by the above mechanisms. Cerebral pressure autoregulation is likely more continuously variable than absolute (26, 62, 65, 66).

Postnatal age showed a small but significant negative relationship with cerebral rSO2C, even though age at operation was relatively tightly constrained. This postnatal decrease in rSO2C has been previously described in healthy neonates, ascribed to an increase in cerebral metabolic rate following birth. These findings suggest potential advantage of operation at younger age. Although CPB time had no apparent effect, this may be related to potential protective effects of antegrade cerebral perfusion and alpha-adrenergic blockade. These effects were measurable in the renal circulation with associated increased rSO2R and lower ΔSavO2.

Although the RV-PA (Sano) shunt type has been associated with better early survival (67, 68) possibly related to improved myocardial oxygen supply-demand balance, the global hemodynamic effects of the RV-PA shunt are not clearly different from the MBT shunt, with very similar mean arterial pressure and saturations (69–71). Accordingly, we found no influence of shunt type on determinants of regional saturation or blood flow in multivariable models. Similarly, we did not a find a difference related to survival status, suggesting that regional circulator controls may not be important determinants of survival. However, this study cannot directly answer whether specific manipulation of PaCO2 to alter oxygenation and blood flow in regional circulations could also alter outcomes.

Like all patients with complete mixing of pulmonary and systemic venous blood, neonates with HLHS following a Norwood procedure will have significant hemoglobin desaturation in arterial blood, adding to risk of impaired systemic oxygen delivery, but limiting exposure of tissues to hyperoxic conditions, especially compared to normal term and preterm neonates (72–76). Although changes in both PaO2 and PaCO2 during re-oxygenation following birth and resuscitation have the potential to cause tissue injury through free radical pathways (77–80), this concern may have more relevance during cardiopulmonary bypass, ECMO support, or beyond the extremes of PaCO2 and PaO2 observed in this report (Table 3). Given the 5–95% intervals of PaCO2 (36–63 torr), PaO2 (38–56 torr), and SaO2 (75–91%), we would not expect significant effect of CO2 manipulation on tissue redox states or free radical generation. Both hypocapnia and hypercapnia have been associated with risk of cerebral injury in premature neonates, especially when rapid or extreme, with alteration in cerebral vascular resistance as a potential mechanism. This report describes the changes in both cerebral and renal-somatic blood flow and oxygenation associated with changes in PaCO2, and these regional NIRS measures would be thus rational targets for goal-directed ventilation strategies including dynamic manipulation of PaCO2.

The observational nature of this study is its major weakness. However, the population size and longitudinal data collection methods make the sample large enough for application of complex multivariable methods. Although multivariable multilevel and within-subject techniques were used to reduce bias and confounding, only limited causal association can be inferred. The population was restricted to patients with HLHS, undergoing Norwood repair utilizing deep hypothermic CPB, with known circulatory limitations. Although the findings are consistent with known physiologic mechanisms, caution should be observed when generalizing the observed effects to patients with other conditions and treatment strategies.

In summary, we found that arterial PaCO2 had differential effects on cerebral and renal circulation in neonates with HLHS following the Norwood procedure. The effects are measurable with a monitoring strategy that utilizes cerebral and somatic NIRS, thus facilitating individualized goal-directed interventions (81, 82).

The data analyzed in this study is subject to the following licenses/restrictions: a subset of the dataset will be available by contacting the corresponding author. Requests to access these datasets should be directed to George M. Hoffman, ghoffman@mcw.edu.

The studies involving human participants were reviewed and approved by Children's Hospital of Wisconsin Institutional Review Board. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

GH was responsible for conception and design of the study, database maintenance, statistical analysis, and wrote the first draft of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

This study was supported by institutional resources at Children's Hospital and Medical College of Wisconsin.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.