Introduction

A time-honored contraindication for administering sodium bicarbonate is the presence of hypercapnic respiratory failure in a patient breathing spontaneously [5]. Such view appears well justified considering the potential ill effects of this therapy, including an increased carbon dioxide load caused by decomposition of the infused alkali in the process of buffering and a depressive effect on ventilation owing to alkalinization of body fluids [5, 6, 39]. In agreement with this tenet, a recent perspective concluded that sodium bicarbonate therapy is not recommended for the management of respiratory acidosis as a simple acid-base disorder; contrariwise, prescription of this alkalinizing agent is appropriate for treating severe acidemia caused by mixed respiratory and metabolic acidosis or permissive hypercapnia, a condition wherein ventilatory assistance is provided to the patient [7]. A most recent review questioned whether even under those circumstances administration of sodium bicarbonate yields benefit [16].

The conventional viewpoint notwithstanding, it is notable that limited credit has been given to several older reports describing salutary effects of sodium bicarbonate administration in certain patients with respiratory acidosis, including a reduction in arterial carbon dioxide tension (PaCO2) and improvement in cardiopulmonary distress [10, 12, 15, 19, 26, 28, 32, 33, 35, 36, 41, 48]. Such reports dealt with patients with acute severe asthma or exacerbation of chronic obstructive pulmonary disease. The determinants of such beneficial effects of sodium bicarbonate therapy in those patients remain uncertain. Proposed mechanisms include improvement of pulmonary ventilation that results from lessening the fatigue of respiratory muscles and ameliorating bronchoconstriction [19, 28, 36]. Of interest, sodium bicarbonate has been administered to improve performance of racing horses, a practice that was declared illegal [27]. Additionally, several studies in healthy volunteers that examined the impact of sodium bicarbonate administration on the ventilatory efficiency during exercise reported discordant conclusions ranging from no effects to salutary actions [13, 22, 29, 38].

In a previous study, wherein an equivalent amount of 1-N NaHCO3 solution was infused to unanesthetized dogs with normal acid-base status or one of the four cardinal acid-base disorders, we demonstrated a striking disparity in the postinfusion PaCO2 among the study groups: acute alkali administration increased PaCO2 in normal animals and those with metabolic acidosis, metabolic alkalosis, or respiratory alkalosis, but it failed to raise PaCO2 (actually it tended to decrease PaCO2) in respiratory acidosis [3]. Such disparity defies the conventional view considering that sodium bicarbonate administration produces a transient increase in carbon dioxide load and a sustained decrease in alveolar ventilation [6, 45]. An attempt to investigate and interpret the aberrant PaCO2 response to sodium bicarbonate administration in respiratory acidosis was not made in the original paper [3]. In view of the recent finding of improvement in ventilatory efficiency by sodium bicarbonate administration in healthy volunteers during exercise [13], we carried out a new analysis of the previously conducted study to examine the aberrant response of PaCO2 to sodium bicarbonate infusion in respiratory acidosis.

Materials and methods

The impact of an acute sodium bicarbonate load on the time-course of PaCO2 was investigated in normal dogs and dogs with each of the four cardinal acid-base disorders. Previous reports from these alkali-infusion studies examined the apparent space of distribution of sodium and bicarbonate, the osmotic and nonosmotic sodium storage, and the determinants of the resulting hypokalemia [1,2,3, 8].

Animal preparation

Fifty-two acute studies of hypertonic NaHCO3 infusion were carried out on 40 female mongrel dogs ranging in weight between 10.2 and 18.2 kg. The animals were fed 30 g/kg per day of a synthetic diet until the day of infusion. The diet contained <1.0 mmol sodium/100 g, <0.1 mmol potassium/100 g, and <0.5 mmol chloride/100 g [3]. The daily diet was supplemented with 2.5 mmol/kg body weight of potassium as neutral phosphate and 2.5 mmol/kg body weight of sodium chloride, except as noted below. The diet was homogenized with twice its weight of distilled water before feeding. Blood samples were obtained by percutaneous arterial puncture and rectal temperature was measured at the time of blood sampling.

Five groups of animals were studied:

Group 1. Normal (4 dogs, 4 studies). Plasma bicarbonate concentration ([HCO3-]p) ranged between 21.3 and 22.8 mmol/L.

Group 2. Respiratory acidosis (7 dogs, 9 studies). Graded degrees of the disorder were produced by exposing the animals to CO2 within a large environmental chamber [44]. The animals were maintained at a given level of inspired CO2 (8 or 11%) for at least 7 days, a period known to be adequate for the development of a chronic steady state [4, 43]. A normal atmospheric O2 concentration of 21% was maintained in all studies. [HCO3-]p ranged between 31.5 and 40.7 mmol/L.

Group 3. Respiratory alkalosis (7 dogs, 7 studies). Sustained hyperventilation was produced by exposing the animals to an hypoxic atmosphere within a large environmental chamber [44]. The ambient O2 concentration within the chamber was lowered from 21% to 9% over a period of 2-3 days. The chamber atmosphere was maintained at a level of 9% O2 for at least 7 days, a period known to be adequate for the development of a chronic steady state [18]. [HCO3-]p ranged between 13.4 and 17.3 mmol/L.

Group 4. Metabolic acidosis (9 dogs, 15 studies). Graded severity of the disorder was induced by adding HCl or L-lysine-monohydrochloride (Sigma Chemical Co., St. Louis, MO) at a dose ranging between 2 and 8 mmol/kg body weight to the daily diet for at least 7 days [17]. [HCO3-]p ranged between 9.5 and 18.6 mmol/L.

Group 5. Metabolic alkalosis (13 dogs, 17 studies). Graded severity of the disorder was induced by administering ethacrynic acid, 50 mg orally, for 1-5 days. In this group, sodium neutral-phosphate was substituted for the NaCl supplement administered to the other groups [30]. Five days were allowed to lapse after the last dose of ethacrynic acid before evidence for a chronic steady state of acid-base equilibrium was sought. [HCO3-]p ranged between 25.9 and 47.7 mmol/L.

Acute experimental protocol

On the day of the acute experiment, the diet was withheld. Two arterial blood samples were obtained 30 min apart for baseline measurements of electrolyte and acid-base composition. Animals were accepted for study only if (a) the mean of the two control values for [HCO3-]p differed by no >2 mmol/L from the value obtained on the previous day and (b) the mean of the two control observations for PaCO2 differed by no >3 mm Hg or 10% (whichever was greater) from the value obtained on the previous day.

Once the presence of a steady-state of acid-base equilibrium was established, sodium bicarbonate was administered intravenously as a 1-N solution at a dose of 5 mmol/kg body weight (5 mL/kg body weight) over a 10-min period. Arterial blood samples were drawn anaerobically at 30, 60, and 90 min from the midpoint of the infusion. The acute experimental protocol was carried out in unanesthetized animals. Importantly, the composition of the breathing atmosphere remained unchanged before and after sodium bicarbonate infusion in all study groups, including the chamber atmosphere of the respiratory alkalosis and respiratory acidosis groups.

Analytical methods

Acid-base status was determined by direct measurement of arterial blood pH and total CO2; PaCO2 and [HCO3-]p were calculated from the Henderson-Hasselbalch equation. pH and pK, and the solubility coefficient of CO2 were corrected for temperature; pK was also corrected for pH. Methods used for determining sodium, potassium, and chloride have been reported previously [3].

Statistical analysis

Continuous variables are reported as mean + SD. To compare study groups at time points of observation, we used paired t-test or Wilcoxon rank-sum test depending on the normality of data distribution and equality of variances. We used simple linear regression to model the relationship between PaCO2 and blood pH at various time points of observation across the 5 study groups. There were no statistically significant deviations from linearity for these continuous variables. The terms “significant” or “significantly different” are used to describe differences which have a p value of <0.05. Analyses were performed using Stata/IC 17.0 software (Stata Corp. 2021. Stata Statistical Software: Release 17. College Station, TX: StataCorp LLC).

Results

Table 1 depicts the baseline blood acid-base status and plasma electrolyte composition of the 5 groups of dogs. These values were typical of normal dogs [31] and dogs with HCl-induced metabolic acidosis [17], diuretic-induced metabolic alkalosis [30], respiratory acidosis [4, 43], or respiratory alkalosis [18].

Table 1 Baseline blood acid-base status and plasma electrolyte composition in the five groups of dogs
Full size table

As expected, sodium bicarbonate infusion elicited significant increases in [HCO3-]p and blood pH in all groups at 30 min, which gradually lessened thereafter. Table 2 shows the changes (Δ) in blood acid-base composition from baseline in the 5 groups of dogs at each time point. The mean Δ[HCO3-]p at 30 min ranged from 8.3 mmol/L to 9.5 mmol/L and was not different among study groups. Contrariwise, mean ΔpH from baseline at 30 min attained a wide range, being largest in metabolic acidosis and respiratory alkalosis (0.17 units and 0.15 units, respectively), intermediate in normal and respiratory acidosis (0.10 units and 0.09 units, respectively), and smallest in metabolic alkalosis (0.06 units). The same pattern persisted at the subsequent time points of 60 min and 90 min, with mean ΔpH values from baseline decreasing progressively in association with the gradual dissipation of Δ[HCO3-]p. Such wide differences in mean ΔpH are predicted by the Henderson-Hasselbalch equation, wherein similar Δ[HCO3-]p would yield larger changes in ΔpH in states of low baseline [HCO3-]p and PaCO2 compared with those of high baseline [HCO3-]p and PaCO2.

Table 2 Changes in blood acid-base status from baseline following infusion of sodium bicarbonate in all study groups
Full size table

Table 2 depicts the impact of acute sodium bicarbonate infusion on PaCO2 at 30 min, a timeframe in which distribution and equilibration of the administered alkali in body fluids should be complete; during this early phase, interaction of bicarbonate with the body’s non bicarbonate buffers should have occurred resulting in release of carbon dioxide thereby increasing its body stores. As expected, sodium bicarbonate infusion elicited significant increases in PaCO2 from baseline, mean ΔPaCO2 being 5.5 mm Hg in normals, 6.2 mm Hg in metabolic acidosis, 5.4 mm Hg in metabolic alkalosis, and 2.9 mm Hg in respiratory alkalosis. Strikingly, no significant increase in PaCO2 occurred in respiratory acidosis, mean ΔPaCO2 being 0.3 mm Hg. Mean ΔPaCO2 from baseline remained significantly increased and essentially unchanged at 60 min and 90 min in all study groups, whereas values in respiratory acidosis showed a tendency toward reduction (–1.0 mm Hg and – 0.5 mm Hg, respectively) that did not reach statistical significance. As shown in Table 2, at 30 min, 60 min, and 90 min, mean ΔPaCO2 from baseline of each of the four study groups were significantly different than respiratory acidosis (only exception being respiratory alkalosis at 30 min).

Significant and positive linear regressions of the PaCO2 vs blood pH relationship were obtained between control and 30 min in normal (slope 45.2 mm Hg/pH unit), respiratory alkalosis (13.5 mm Hg/pH unit), metabolic acidosis (30.3 mm Hg/pH unit), and metabolic alkalosis (89.6 mm Hg/pH unit) (Figs. 1 and 2). This relationship remained significant and positive between control and 60 min and control and 90 min in the four study groups (with the exception of normal between control and 90 min). In general, the ΔPaCO2/ΔpH slope increased progressively, values between control and 90 min being higher than those between control and 30 min (59.3 mm Hg/pH unit vs 45.2 mm Hg/pH unit in normal, respectively; 23.7 mm Hg/pH unit vs 13.5 mm Hg/pH unit in respiratory alkalosis, respectively; 31.3 mm Hg/pH unit vs 30.3 mm Hg/pH unit in metabolic acidosis, respectively; and 119.6 mm Hg/pH unit vs 89.6 mm Hg/pH unit in metabolic alkalosis, respectively). In sharp contrast, the ΔPaCO2/ΔpH slope was negative in respiratory acidosis (– 43.2 mm Hg/pH unit between control and 30 min; –76.1 mm Hg/pH unit between control and 60 min; and –108.0 mm Hg/pH unit between control and 90 min). Only the relationship between control and 90 min reached statistical significance (Fig. 3).

Fig. 1
figure 1

PaCO2 vs blood pH relationship between control and 30 min, control and 60 min, and control and 90 min following sodium bicarbonate infusion in normal dogs (left panel) and dogs with respiratory alkalosis (right panel)

Full size image
Fig. 2
figure 2

PaCO2 vs blood pH relationship between control and 30 min, control and 60 min, and control and 90 min following sodium bicarbonate infusion in dogs with metabolic acidosis (left panel) and dogs with metabolic alkalosis (right panel)

Full size image
Fig. 3
figure 3

PaCO2 vs blood pH relationship between control and 30 min (left panel), control and 60 min (middle panel), and control and 90 min (right panel) following sodium bicarbonate infusion in dogs with respiratory acidosis

Full size image

Discussion

The present study reveals important insights into the disparate response of PaCO2 to an acute sodium bicarbonate infusion of animals with respiratory acidosis compared with normal animals and those with the remaining cardinal acid-base disorders. First, an increase in PaCO2 was evident at 30 min, the first point of observation, in all groups save for respiratory acidosis. Second, the observed early effect of sodium bicarbonate on PaCO2 in all study groups remained essentially unchanged in the later stages of the postinfusion period (60 min and 90 min). Third, the PaCO2 vs blood pH relationship indicates that the decreased acidity following sodium bicarbonate infusion was accompanied by an elevation in PaCO2 at all time points in all study groups except for respiratory acidosis, in which there was a tendency toward reduced values. In the latter group, the ΔPaCO2/ΔpH slope was negative contrasting with the positive slope of the four other groups. What factors might be responsible for the disparate response of PaCO2 to an alkalinization of body fluids in experimental respiratory acidosis?

Could differences in carbon dioxide released from buffering-induced decomposition of administered bicarbonate account for the disparate response in respiratory acidosis? Classic observations in normal volunteers revealed a ~16% transient increase in carbon dioxide output in the early phase of a rapid infusion of sodium bicarbonate that was accompanied by a rise in PaCO2 [45]. Titration by hemoglobin and plasma proteins is largely responsible for such decomposition of bicarbonate, which in our studies averaged to ~9% of retained bicarbonate at 30 min and was very similar among groups [3]. Importantly, decomposition of bicarbonate is completed within minutes from its infusion and its effects on PaCO2 should have vanished by 30 min, our first timepoint of observation [45]. The stability of PaCO2 values throughout the 90 min of observation in all study groups excludes differences in bicarbonate decomposition from playing a role in the disparate response of PaCO2 to bicarbonate infusion. Rather, it indicates that the aberrant response of PaCO2 does not reflect changes in carbon dioxide production but differences in carbon dioxide excretion.

Could differences in baseline blood pH or baseline PaCO2 be responsible for the discordant response of PaCO2 to bicarbonate infusion in respiratory acidosis? Baseline blood pH was decreased (metabolic acidosis), normal (normal group), or increased (metabolic alkalosis and respiratory alkalosis) in the four groups and bicarbonate infusion increased PaCO2 in all of them. Consequently, differences in baseline blood pH could not explain our observations. Further, respiratory acidosis, an acidemic disorder, sharply deviated from the postinfusion PaCO2 response in metabolic acidosis, the other acidemic disorder. Turning to baseline PaCO2, it was diminished (metabolic acidosis and respiratory alkalosis), normal (normal group), or increased (metabolic alkalosis) in these groups and bicarbonate infusion increased PaCO2 in all of them. Thus, differences in baseline PaCO2 could also not account for our results. Furthermore, respiratory acidosis is characterized by an increased baseline PaCO2 like metabolic alkalosis, and yet it did not yield a rise in PaCO2 following bicarbonate infusion but showed a trend toward decreased levels.

Might differences in baseline pulmonary ventilation explain the disparate response of PaCO2 to bicarbonate infusion in respiratory acidosis? Baseline pulmonary ventilation was diminished (metabolic alkalosis), normal (normal group), or increased (metabolic acidosis and respiratory alkalosis) and bicarbonate infusion increased PaCO2 in all of them. Consequently, differences in baseline pulmonary ventilation could not account for our findings. Because a CO2-rich atmosphere was responsible for the experimental respiratory acidosis in our study, pulmonary ventilation was increased at baseline in this disorder like in metabolic acidosis and respiratory alkalosis [24]. Thus, differences in baseline pulmonary ventilation could not explain the aberrant response of PaCO2 to bicarbonate infusion in respiratory acidosis.

Valuable information about the impact of baseline acid-base status and baseline pulmonary function on the response of PaCO2 to sodium bicarbonate infusion was obtained by analyzing the linear regressions of the PaCO2 vs blood pH relationship. Such linear regressions, an index of the ventilatory response to sodium bicarbonate infusion, were positive at all times in all acid-base disorders, the sole exception being respiratory acidosis, in which the linear regressions were negative. The ΔPaCO2/ΔpH slope after bicarbonate infusion demonstrated a larger effect at 90 min compared with 30 min in all study groups, reflecting relative stability of PaCO2 levels in association with decreasing blood pH values. The largest ΔPaCO2/ΔpH slopes after sodium bicarbonate infusion were found in metabolic alkalosis (89.6 mm Hg/pH unit and 119.6 mm Hg/pH unit at 30 min and 90 min, respectively) indicating that this group exhibited the greatest depression of pulmonary ventilation following bicarbonate administration; notably this group of animals had baseline alkalemia and baseline hypercapnia, the latter signifying depressed pulmonary ventilation prior to bicarbonate infusion.

Contrariwise, respiratory alkalosis featured the smallest ventilatory depression postinfusion (ΔPaCO2/ΔpH slopes being 13.5 mm Hg/pH unit and 23.7 mm Hg/pH unit at 30 min and 90 min, respectively), a study group exhibiting at baseline mild alkalemia and hypocapnia, which were secondary to hypoxemia-induced hyperventilation; importantly, postinfusion hypercapnia developed despite maintaining the same level of hypoxemia. Intermediate values for ΔPaCO2/ΔpH slopes were found in metabolic acidosis (ΔPaCO2/ΔpH slopes being 30.3 mm Hg/pH unit and 31.3 mm Hg/pH unit at 30 min and 90 min, respectively), in which baseline hyperventilation and hypocapnia were driven by acidemia that was partially corrected by sodium bicarbonate infusion; and in normal animals (ΔPaCO2/ΔpH slopes being 45.2 mm Hg/pH unit and 59.3 mm Hg/pH unit at 30 min and 90 min, respectively), in which no abnormalities in acid-base status and pulmonary ventilation at baseline were present.

In sharp contrast with the above results, the ΔPaCO2/ΔpH slopes in respiratory acidosis were consistently negative at all points of observation (being – 43.2 mm Hg/pH unit, –76.1 mm Hg/pH unit, and –108.0 mm Hg/pH unit at 30 min, 60 min, and 90 min, respectively). Like respiratory alkalosis, these animals had a stimulus to pulmonary ventilation both before and after bicarbonate infusion; yet, the PaCO2 vs blood pH relationship postinfusion was opposite in direction between the two disorders. Taken together, another explanation is required for the aberrant response of PaCO2 to sodium bicarbonate infusion in respiratory acidosis.

The effects of an acute bicarbonate infusion on PaCO2 and pulmonary function have long been the subject of investigation. In normal volunteers, PaCO2 increased immediately after sodium bicarbonate infusion, as did ventilation and carbon dioxide excretion [45]. A sudden increase in end-tidal CO2 tension, assessed by capnography, is known to occur following a bolus administration of sodium bicarbonate [9]. This excess carbon dioxide excretion caused by partial decomposition of bicarbonate is completed within 10 min postinfusion [34]. The early stimulation of ventilation results from the rapid entry of the released carbon dioxide postinfusion into the cerebrospinal fluid decreasing its pH (at the time that blood pH increases) and stimulating central chemoreceptors in the brain medulla [39]. However, this initial response is soon followed by persistent hypoventilation effected by the delayed entry of bicarbonate into the cerebrospinal fluid resulting in its alkalinization [23, 37, 39, 45]. The sustained, depressive effect of bicarbonate on pulmonary ventilation is apparent in the four groups of animals in our study that exhibited a rise in PaCO2 throughout the period of observation.

We propose that such ventilatory depression also occurred in respiratory acidosis but carbon dioxide retention did not develop because of a concurrent process that facilitated its excretion. This process must reflect improvement in ventilatory efficiency following bicarbonate infusion in our animals with experimental respiratory acidosis. Such an action of acute sodium bicarbonate administration has previously been demonstrated in the response to exercise of healthy volunteers [13].

Ventilatory efficiency, also known as ventilatory equivalent for carbon dioxide, is assessed by the ratio of total pulmonary ventilation to carbon dioxide excretion (VE/VCO2); notably, a decreased VE/VCO2 signifies greater ventilatory efficiency and vice versa, an increased VE/VCO2 indicates less ventilatory efficiency [50, 51]. Insight into the determinants of ventilatory efficiency can be obtained by utilizing the modified alveolar ventilation equation (VE/VCO2=863/[PaCO2×(1–VD/VT)]) that establishes the relationship among PaCO2, VCO2, VE, and the ratio of dead space to tidal volume (VD/VT) [13, 50]. The equation predicts that improved ventilatory efficiency (i.e., decreased VE/VCO2) in the context of constant/decreasing PaCO2 and constant VCO2, conditions present in our respiratory acidosis studies, must result from diminished ratio of dead space ventilation to total ventilation, a ratio that is equivalent to the VD/VT term. We propose that sodium bicarbonate administration in respiratory acidosis improves ventilatory efficiency, i.e., alveolar ventilation-pulmonary perfusion (VA/Q) matching because of salutary effects on VD/VT.

Respiratory acidosis in healthy, unanesthetized dogs breathing spontaneously is expected to be associated with bronchial constriction of larger airways that is mediated by enhanced vagus tone [21, 28]. Additionally, respiratory acidosis has been shown to induce pulmonary vasoconstriction leading to reduced pulmonary blood flow [20, 25]. Therefore, increased VD/VT and consequent ventilation/perfusion inequality in our dogs with experimental respiratory acidosis could emanate from a ventilation defect, a perfusion defect, or a combination of the two. We propose that this defect(s) is then ameliorated by the alkalinizing effect of sodium bicarbonate. Contrasting with our proposal, several studies examining the effects of respiratory acidosis on ventilation-perfusion matching reported salutary actions [14, 46, 47, 49]. Such studies, however, were performed under vastly different experimental conditions, including general anesthesia, muscle relaxants, mechanical ventilation, and even background septic shock, rendering their conclusions non applicable to our observations.

An abnormally high VE/VCO2 ratio is found in individuals with exertional breathlessness or exercise intolerance, the elderly, and patients with heart failure or a variety of chronic pulmonary disorders [11, 13, 40, 42, 51]. The common basis for this abnormality appears to be an increased VD/VT largely owing to expansion of the “physiologic” dead space [51]. The precise mechanism responsible for the prevailing ventilation/perfusion inequality varies among conditions and can involve either term of the ratio or both terms. In chronic lung disease, the main mechanism relates to abnormal ventilation, whereas in heart failure the key defect is reduced pulmonary blood flow.

The results of our investigation might help explain the underlying mechanisms of the beneficial effects of sodium bicarbonate administration in patients with respiratory acidosis secondary to acute severe asthma or acute exacerbation of chronic obstructive pulmonary disease. The substantially larger impairment in the VE/VCO2 ratio present in those patients resulting from their increased VD/VT in comparison with our experimental animals might explain the more dramatic beneficial effects of sodium bicarbonate administration, including the sizable reduction of PaCO2, reported in the clinical studies [10, 12, 15, 19, 26, 28, 32, 33, 35, 36, 41, 48]. Indeed, in our studies of experimental respiratory acidosis, there was a tendency for a reduction of PaCO2 and negative ΔPaCO2/ΔpH slopes following sodium bicarbonate infusion (Table 2 and Fig. 3).

Absence of direct measurements of ventilation and VD/VT in the current investigation constitutes an important experimental shortcoming. Therefore, our proposal that sodium bicarbonate administration in respiratory acidosis improves ventilatory efficiency must remain conjectural.