Main

Sodium bicarbonate (NaHCO3) is a commonly used medication to treat metabolic acidemia from a variety of causes. Intravenous NaHCO3 acts by neutralizing excess acid in the blood to yield carbonic acid, which then dissociates into carbon dioxide and water, restoring physiologic pH. The efficacy of NaHCO3 treatment for mild-to-moderate acidemia, however, is widely debated, and controversy exists over whether any true benefit results from the therapy (1,2,3,4). In fact, some data suggest that NaHCO3 may be harmful in certain populations. In preterm infants, for example, the use of NaHCO3 has been linked to intraventricular hemorrhage, hypernatremia, and death (2,5,6). Nevertheless, treatment of metabolic acidemia with NaHCO3 remains a common practice in many pediatric intensive care units and operating rooms.

Further understanding of the cerebral hemodynamic effects of rapid administration of NaHCO3 may illuminate the link between NaHCO3 and brain injury. It is known that administration of NaHCO3 causes an immediate and transient increase in the production of nonmetabolic CO2 (7,8,9,10), as well as a slight increase in plasma pH (7,11) and serum osmolality (12,13). This increase in serum osmolality leads to a flow of intracellular water into the extracellular space to restore osmotic equilibrium and to an increase in arterial hemoglobin concentration and a decrease in hematocrit (9,10,12,13). However, little definitive and quantitative information exists regarding the effects of NaHCO3 on cerebral hemodynamics.

Several publications on the effects of NaHCO3 on cerebral blood flow (CBF) report conflicting observations (7,11,14,15,16,17,18,19). Lou et al. (17) observed substantial decreases in CBF measured by the Xe-133 clearance technique 5 min after NaHCO3 administration in seven newborn infants with respiratory distress. By contrast, Nakashima et al. (9) reported significant increases in CBF in five healthy adult volunteers after NaHCO3 administration. Finally, in a study of six neonatal dogs, Young et al. (19) observed no changes in CBF (measured with radioactive tracers) 30 min after NaHCO3 injection. These conflicting results may reflect the wide variety of experimental subjects studied (both humans and animals), the severity and cause of the acidemia, the dosage and rapidity of injection of NaHCO3, the use of mechanical ventilation, the anesthetic state, the method of CBF measurement, and the time frame for assessing the cerebral hemodynamic effects following drug administration.

The current observational pilot investigation aimed to quantify the immediate cerebral hemodynamic effects of a rapid (10–20 s) bolus administration of NaHCO3. Pilot data were taken 1–15 min after the bolus injection and were obtained from a subset of preoperative patients with hypoplastic left heart syndrome (HLHS) who were treated for mild acidemia during part of a larger presurgical brain imaging study. Noninvasive diffuse optical spectroscopies, namely, diffuse optical spectroscopy (DOS) and diffuse correlation spectroscopy (DCS), were used for 15 min postinjection to monitor regional changes in cerebral oxy- and deoxyhemoglobin concentrations (ΔHbO2 and ΔHb, respectively), changes in total hemoglobin concentration (ΔTHC), and changes in CBF relative to baseline (ΔrCBF) before rapid NaHCO3 administration.

Results

As seen in Table 1 , 305 patients were approached for this investigation; parental consent was obtained in 133, and 91 were studied with DOS/DCS. Of the 91 patients with HLHS monitored with DOS/DCS, 22 received NaHCO3 treatment for a mild or moderate base deficit: n = 8 pre-Norwood, n = 8 pre-Glenn, and n = 6 pre-Fontan. Furthermore, we selected 22 age- and gender-matched control patients from the remaining 69 patients. These patients received no interventions but were monitored with DOS/DCS as part of the presurgical brain magnetic resonance imaging study. Patient characteristics for the treated and control groups are summarized in Table 2 . NaHCO3-treated patients were mostly of male gender (64%) and ranged in age from 1 d to 4 y.

Table 1 Summary of patient recruitment
Full size table
Table 2 Patient characteristics
Full size table

Arterial blood gas data obtained before the administration of NaHCO3 are summarized in Table 3 for patients in the treated and control groups. Patients received a median (interquartile range) dose of 1.1 (0.9, 1.8) mEq/kg NaHCO3 to treat a median (interquartile range) base deficit of −4 (−6, −3) mEq/l. Of note, the majority of patients were normocapnic but mildly hypoxemic with median (interquartile range) arterial oxygen tension of 6.3 (8.0, 6.7) kPa. The below-normal partial pressures of oxygen were expected due to the presence of intracardiac shunting, a consequence of single-ventricle physiology. Furthermore, arterial blood samples were not drawn after NaHCO3 administration; thus, changes in the parameters listed in Table 3 due to NaHCO3 are not reported. Baseline heart rate (HR), mean arterial pressure (MAP), and transcutaneous oxygen saturation (SpO2) are also reported in Table 3 for both treated and control groups. No differences in these baseline parameters between treated and age-matched controls were observed.

Table 3 Baseline systemic hemodynamics
Full size table

Figure 1 provides box plots of ΔHb, ΔHbO2, ΔTHC, and ΔrCBF over time for the control and treated groups following the injection of intravenous NaHCO3. As compared with age-matched controls, patients showed significant increases in ΔrCBF at 1 min after NaHCO3 injection (P = 0.0084). No significant changes in DOS measures of ΔHb, ΔHbO2, or ΔTHC were observed, nor were any significant differences in these parameters between the treatment group and the control group observed at any time following the injection. In addition, MAP, HR, and SpO2 did not change following the administration of NaHCO3 in the treated patients as compared with controls ( Figure 2 ).

Figure 1
figure 1

Box plots of changes from baseline in (a) oxy-, (b) deoxy-, and (c) total hemoglobin concentrations (ΔHb, ΔHbO2, and ΔTHC, respectively) and (d) relative cerebral blood flow (ΔrCBF) at times 1, 5, 10, and 15 min following NaHCO3 administration (gray). The control group, which received no intervention, is shown in white. The dotted gray lines indicate no change from baseline levels. *P < 0.05. NaHCO3, sodium bicarbonate.

Full size image
Figure 2
figure 2

Box plots of changes from baseline in (a) heart rate, (b) mean arterial pressure, and (c) transcutaneous oxygen saturation at times 1, 5, 10, and 15 min following NaHCO3 administration (gray). The control group, which received no intervention, is shown in white. The dotted gray lines indicate no change from baseline levels. NaHCO3, sodium bicarbonate.

Full size image

In the NaHCO3-treated group, at 1 min postinjection, the increase in ΔrCBF was highly correlated with NaHCO3 dosage (R2 = 0.71; P = 2.1 × 106; slope (95% confidence interval) = 45.7 (32.5, 58.9)%/mEq/kg; Figure 3 ). No relationship between change in ΔrCBF at 1 min postinjection and baseline pH, Pco2, or Po2 was observed (all P > 0.1). The relationship between ΔrCBF and NaHCO3 dosage weakened slightly by 5 min postinjection (R2 = 0.51; P = 6.0e-4) and was no longer highly significant by 10 min after NaHCO3 administration (at 10 min, R2 = 0.12; P = 0.085, and at 15 min, R2 = 0.23; P = 0.042). Furthermore, no relationship was observed between cardiac physiology, age, weight, or arterial hemoglobin concentration and the change in ΔHb, ΔHbO2, ΔTHC, or ΔrCBF at any time point.

Figure 3
figure 3

Relationship between the dose of sodium bicarbonate administered (in mEq/kg) and the associated change in cerebral blood flow (%) 1 min after injection. The black line indicates the best linear fit to the data, and the 95% confidence interval to the fit is shown in the gray shaded region. Open circles denote pre-Fontan patients, black circles denote pre-Glenn patients, and gray circles denote pre-Norwood patients.

Full size image

Discussion

In this pilot observational study, we quantified the cerebral hemodynamic effects of NaHCO3 administered rapidly to treat metabolic acidemia in paralyzed, mechanically ventilated children with single-ventricle physiology. DCS demonstrated significant increases in CBF immediately (within 2 min) following bolus administration of NaHCO3. These increases in CBF were strongly associated with the dosage of the NaHCO3, increasing in a linear fashion. This relationship between CBF and NaHCO3 dose was observed within all stages of HLHS cardiac physiology studied herein. The current investigation is the first to describe a dose-dependent response of CBF to NaHCO3. Population-averaged changes in oxy- or deoxyhemoglobin concentrations were not significantly different between the control and the treated groups. Quantification of the changes in cerebral hemodynamics that occur as a consequence of bolus injection of NaHCO3 may have a significant and beneficial impact on the treatment of metabolic acidemia in patients with congenital heart disease, especially in patients with bidirectional Glenn, in which pulmonary blood flow is dependent on CBF.

Although these data were obtained on mechanically ventilated patients with single-ventricle physiology, these results may be generalized to a larger pediatric population. Understanding the cerebral hemodynamic effects of NaHCO3 administration may be especially important in vulnerable populations such as premature infants, patients with impaired autoregulation from hypoxic–ischemic injury, or patients with focal or global cerebral edema for which the risk of rapid fluctuations in CBF does not outweigh the benefit of treating metabolic acidemia.

The mechanisms that govern the cerebral hemodynamic responses to a rapid NaHCO3 infusion are complex and not fully understood. Potentially, the observed increase in CBF was caused by an increase in the concentration of CO2, produced as a by-product of the reaction of NaHCO3 with acid, leading to an intracellular acidosis (8,9). Although we did not obtain post-NaHCO3 administration arterial blood gas measurements, previous work suggests that NaHCO3 causes significant increases in the partial pressure of arterial CO2 in mechanically ventilated patients (20). CO2 is a potent vasodilator that induces increases in CBF through local effects on cerebral vasculature. Because our population was paralyzed under general anesthesia, the normal mechanism of responding to elevated arterial CO2 tension by increasing minute ventilation was eliminated. Therefore, it is possible that a more potent effect from NaHCO3 may have been observed in our population as compared with patients who are awake and spontaneously breathing.

Relatedly, the increase in CBF may reflect the increase in plasma osmolality following infusion. Siegel et al. (13) demonstrated an increase in osmolality and a decrease in hematocrit in critically ill neonates following the treatment of metabolic acidemia with NaHCO3. Both increased osmolality and decreased hematocrit have been linked to an increase in CBF via vasodilation and decreased viscosity, respectively (21). Therefore, in addition to the vasodilatory effects of CO2, hyperosmolality and/or a drop in hematocrit could be responsible for our observed increase in CBF.

Of note, we did not observe significant population-averaged changes in oxy-, deoxy-, or total hemoglobin concentrations. Vasodilation caused by CO2 and/or hyperosmolality following NaHCO3 administration might be expected to lead to increases in oxy- and total hemoglobin concentrations, as well as a slight decrease in deoxyhemoglobin concentration. However, a decrease in hematocrit after NaHCO3 (as shown in ref. 13) would likely be accompanied by a drop in oxy- and total hemoglobin concentrations, as well as an increase in deoxyhemoglobin concentration (22). Possibly, these two phenomena (i.e., vasodilation and a concomitatant drop in hematocrit) have opposite effects on tissue hemoglobin concentrations, leading to a population-averaged effect of no net change (i.e., within the error bars of our measured concentration changes).

Little work has been published on the cerebral effects of NaHCO3 to treat metabolic acidemia in human pediatric populations, and to our knowledge, only one publication has investigated the effects in patients with single-ventricle physiology (14). Overall, our results are consistent with several reports of the cerebral hemodynamic effects of NaHCO3 used to correct metabolic acidemia, although we did observe some disparities with other reports. van Alfen-van der Velden et al. (11) used continuous-wave near-infrared spectroscopy and transcranial Doppler ultrasound to study 15 preterm infants with metabolic acidosis treated with bolus administration of NaHCO3. Their cohort presented with more severe acidemia than our cohort, i.e., a base deficit less than −6 mmol/l and pH < 7.3, and moreover, their cohort received half the dose (mEq/kg) of NaHCO3 than our population. As with our results, van Alfen-van der Velden et al. did not observe substantial changes in total hemoglobin concentration at 5 and 15 min post-NaHCO3 (they report changes in cerebral blood volume). By contrast to our findings, however, they also did not observe a significant change in CBF as measured by blood flow velocity in the internal carotid artery. This discrepancy may be due to the fact that Doppler ultrasound measures macrovascular changes in arterial flow velocity, whereas DCS measures microvascular flow directly in cortical tissue, and these two quantities may be disparate. Alternatively, the discrepancy may arise from the differences in age and physiology between the populations, or from the fact that 9 of 15 patients in their study (11) were spontaneously ventilating, thus permitting the patient to increase their minute ventilation to exhale the extra CO2 produced by NaHCO3.

Lou et al. (17) used the Xe-133 clearance technique to measure CBF changes 5 min after NaHCO3 injection in seven asphyxiated neonates with respiratory distress and acidosis (base deficit less than −8 mEq/l). Of note, they found profound decreases in global CBF in these infants, contrary to our findings, despite the fact that they administered the same base deficit–dependent dose per kilogram of NaHCO3. It is not clear why these results are contradictory; however, a possible explanation could be the difference in patient population. Unlike our otherwise healthy population with palliated congenital heart disease, their cohort was younger, i.e., preterm neonates, and their cohort had suffered asphyxia and potential damage to the blood–brain barrier. Therefore, bicarbonate ions, which are normally nonpermeable ions, may have been able to penetrate from the plasma to the extracellular fluid, leading to cerebrovasoconstriction and hence decreased CBF.

Bradley et al. (14) monitored 14 patients following bidrectional superior cavopulmonary connections with transcranial Doppler ultrasound of the middle or anterior cerebral artery. Unlike our study, these patients were not acidemic at baseline, i.e., population-averaged baseline pH = 7.39. However, the authors also observed a significant increase in CBF velocity for up to 15 min after a 4 mEq/kg NaHCO3 bolus administration, similar to the findings observed in our patients. In addition, they observed an increase in systemic arterial saturations following bicarbonate administration, contrary to the findings in our bidirectional Glenn population.

In summary, a handful of publications that investigated the cerebral effects of NaHCO3 reported findings consistent with the ones presented herein. The discrepancies that do arise may reflect the severity and cause of the acidemia, the dosage and injection rate of NaHCO3, the use of mechanical ventilation, the differences in patient population, the anesthetic state, the method of CBF measurement, and the time frame for assessing the cerebral hemodynamic effects following drug administration.

Study Limitations

The results presented herein have several limitations. First, we did not draw a post-NaHCO3 arterial blood gas measurement because this was merely an observational pilot study. The current clinical practice at The Children’s Hospital of Philadelphia following NaHCO3 administration is to not draw another arterial blood sample to confirm increases in pH, CO2 tension, and bicarbonate ion concentration. Therefore, although we suggest that arterial CO2 levels increased following NaHCO3 administration due to the abundance of literature suggesting this effect (9,10,15,20) and due to the observed dose-dependent increases in CBF, we cannot definitively confirm that CO2 increased in our cohort. Furthermore, we did not measure baseline albumin concentration, an important nonbicarbonate buffer that may also influence CO2 release following NaHCO3 injection and thus may affect subsequent cerebral hemodynamic changes (20,23).

Second, we tracked changes in cerebral and systemic hemodynamics for only 15 min following NaHCO3 administration. Despite this limited monitoring time period, by 15 min postinjection, CBF was no longer significantly elevated. Therefore, we believe a 15-min window was sufficient to capture the rapid and transient effects of NaHCO3. In addition, we were limited to studying the effects of rapid infusion of NaHCO3. Future work will investigate variation of the infusion time in order to compare the potential beneficial effects of rapid vs. slow infusions.

Third, diffuse optical spectroscopies probe tissues located at shallow depths in the frontal cortex in the region under the optical probe. Although we presume that our frontal cortex measurements are indicative of whole-brain response to NaHCO3, absolute quantification of cerebral hemodynamics in other regions of the brain is beyond the scope of this work.

Conclusions

NaHCO3 is a commonly used medication administered for rapid correction of metabolic acidemia in adult, pediatric, and neonatal intensive care units. In pediatric patients with HLHS, we observed substantial increases in CBF following bolus intravenous NaHCO3 administration. These changes in CBF were linearly related to the dose of NaHCO3. On average, cerebral oxy- and deoxyhemoglobin concentrations did not change with NaHCO3 administration. Future work will benefit from the investigation of the effects of infusion rate on the CBF response to NaHCO3.

Methods

Study Protocol

Infants and children with HLHS at various stages of palliation were enrolled, and parental consent was obtained for a presurgical brain magnetic resonance imaging and hypercapnia study (described in refs. 24,25) approved by the institutional review board at The Children’s Hospital of Philadelphia. After induction of general anesthesia with paralysis in the operating room, patients were tracheally intubated. The anesthetic consisted of sevoflurane in room air for patients older than 3 mo and fentanyl (5 µg/kg) for neonates. Supplemental oxygen was not utilized. Patients were mechanically ventilated using a tidal volume to achieve an arterial CO2 of 4.93–5.60 kPa. An arterial catheter was placed in the umbilical artery in pre-Norwood patients and in an ulnar or radial artery in pre-Glenn and pre-Fontan patients. Patients were then transferred to the magnetic resonance imaging table, and a noninvasive optical probe (see description below) was placed on the forehead for continuous (0.2 Hz) optical monitoring of cerebral hemodynamics. HR via electrocardiogram and peripheral hemoglobin–oxygen saturation (SpO2) via pulse oximetry were monitored and continuously recorded at a rate of 0.5 Hz throughout the duration of the study. Noninvasive (cuff) MAP was measured every 3 min.

After the patient was stabilized, an arterial blood gas measurement was drawn. Blood gas analysis was performed using an i-STAT handheld blood analyzer (Abbott Laboratories, Princeton, NJ) to derive blood pH, partial pressure of carbon dioxide and oxygen, base excess or base deficit, bicarbonate ion concentration, hemoglobin concentration, and hematocrit. As part of routine clinical care during the study, any calculated base deficit less than −2 mEq/l was treated at the attending anesthesiologist’s discretion (not protocolized) with bolus intravenous administration of 8.4% NaHCO3 over a 10–30 s period. All patients in this study were hemodynamically stable, including those patients with a base deficit between −2 and −3, for whom the decision to treat was solely based on the practitioner’s preference. The acuity of the patient’s illness did not play a role in the decision to treat. The following formula was used to calculate the NaHCO3 dosage: patient weight (kg) × 1/3 (l/kg) × base deficit (mEq/l). If NaHCO3 was given, a second blood gas measurement was not obtained following NaHCO3 administration.

Cerebral Monitoring

A hybrid diffuse optical instrument combining DOS and DCS was used to noninvasively monitor cerebral hemodynamics. This instrument has been described previously (24), and the techniques and theoretical analysis have been described at length (26). Briefly, DOS uses three near-infrared wavelengths, 688, 787, and 826 nm, and uses a modified Beer–Lambert law (22) to quantify changes in tissue oxy- and deoxyhemoglobin concentrations, ΔHbO2 and ΔHb, respectively, in the region of brain approximately 1–1.5 cm under the optical probe, i.e., in the cortex surface region. The sum of these changes gives variation of total hemoglobin concentration (ΔTHC = ΔHb + ΔHbO2), a quantity that is generally assumed to be proportional to the change in cerebral blood volume. DCS monitors temporal fluctuations of the reflected light intensity, specifically, the temporal intensity autocorrelation function of detected near-infrared light is computed using a semi-infinite homogeneous approximation in order to derive a blood flow index (26). Previous studies have shown that changes of blood flow index in various model systems agree with ΔrCBF measured by other techniques (24,25,26,27,28). The sources and detectors for both DOS and DCS were separated by 2.5 cm and held in place by a black rubber probe.

Data Analysis

To quantify the effects of NaHCO3 on hemodynamics, a 1-min mean of each of the following parameters was obtained immediately before and at 1, 5, 10, and 15 min after the injection of NaHCO3: ΔHbO2, ΔHb, ΔTHC, ΔrCBF, ΔHR, ΔMAP, and ΔSpO2. These time intervals were chosen because the effects of NaHCO3 were expected to be clearly evident due to the rapid onset and transient action of NaHCO3. Changes in each DOS parameter, i.e., Hb, HbO2, and THC, and vital sign parameter, i.e., HR, MAP, and SpO2, were quantified as the difference between a 1-min average taken 1, 5, 10, and 15 min after NaHCO3 injection and a 1-min average immediately before the injection. Relative changes in DCS-measured CBF were quantified using the following formula:

Here, the angle brackets indicate the mean taken over a 1-min time period, and the subscripts pre and post denote means taken before and after NaHCO3 injection, respectively. Note that the four post averages were quantified at the time points specified above.

A subset of patients who did not receive NaHCO3 treatment were used as controls. These control patients were individually matched with each NaHCO3-treated patient for both age (within 4 mo) and cardiac physiology. Vital sign and DOS/DCS monitoring were acquired continuously for these patients, although they received no intervention. These control patients were intended to elucidate the normal physiologic variations that occur during the monitoring period. The baseline period for these patients was the first minute of DOS/DCS data acquisition, and changes in DOS, DCS, and vital sign parameters were computed in the same manner as described above at 1, 5, 10, and 15 min after the baseline.

Statistical Analyses

A Wilcoxon signed-rank test was carried out to test whether the NaHCO3-treated group showed significantly different changes in vital signs and cerebral hemodynamics as compared with age- and physiology-matched controls. Furthermore, to quantify the relationship between the dosage of NaHCO3 and the subsequent change in CBF measured with DCS, we fit a simple linear regression model; using this model, we estimated the Pearson’s correlation coefficient. The Pearson’s correlation coefficient, R, varies from 0 to 1.0 and measures the extent to which a linear model explains variability in the data. Analyses were performed using R 2.11 statistical software (R Foundation for Statistical Computing, Vienna, Austria). Hypotheses tests and associated P values were two sided. A Hochberg correction was used to adjust the P values for multiple comparisons. Statistical significance was declared for P values < 0.05.

Statement of Financial Support

This work was supported by the National Institutes of Health at the University of Pennsylvania (NS-060653, PI: A.G.Y.) and at The Children’s Hospital of Philadelphia (grant HL090615, PI: M.A.F.; grant NS072338, PI: D.J.L.; grant NS-052380, PI: D.J.L.; grant T32AL07915, PI: D.A.G.; grant T32NS007413, ­PI: E.M.B.); by the National Center for Research Resources and the ­National ­Institute of Biomedical Imaging and Bioengineering of the National ­Institutes of Health through grant P41-EB015893; by the Dana Foundation; and by the Steve and Judy Wolfson Family Trust.

Disclosure

The authors declared no conflict of interest.