The Physicochemical Approach to Acid–Base Balance

Introduction

The analysis of acid–base disorders uses several approaches. Most clinicians follow the traditional approach¹ using pH, PCO₂, and HCO₃⁻. Some prefer the base excess approach.² A third approach is the physicochemical approach, known as strong ion difference (SID) approach, developed by Peter Stewart in 1978³ and touted by his advocates as the most comprehensive and quantitative way to understand and manage acid–base disorders, with substantial popularity among some clinicians. The physicochemical approach's key concepts are (1) changes in H⁺ depend on three independent variables: SID (difference between strong cations and strong anions), A_TOT, defined as total undissociated weak acids and their anions, excluding HCO₃⁻, and PCO₂; (2) these three independent variables explain acid–base disorders better than the Henderson–Hasselbalch equation; and (3) HCO₃⁻, a mere dependent variable, plays no important role in acid–base regulation. In this Perspective, we shall critically appraise the physicochemical approach through more thorough quantitative analysis than previous reviews^4–6 to shed light on some overlooked misconceptions and flaws.

Dichotomous Classification of All Ions into Strong and Weak Ions

Stewart classified all ions as either strong or weak electrolytes. The former dissociate nearly completely at physiologic pH, with a pK of 4.0 separating strong from weak ions. This classification contains many contradictions.

• Lactate (pK of 3.7) is considered a strong anion, but not beta-hydroxybutyrate (pK of 4.7) or urate (pK of 5.4). Yet, all three mostly dissociate at pH 7.4.

• Na⁺, a strong cation, precipitates as monosodium urate crystals at urate concentration of mere 0.5 mmol/L (8 mg/dl) at physiologic pH.

• Plasma Ca⁺⁺ and Mg⁺⁺, considered strong cations, are protein bound at 50% and 30%, respectively.

New Classification of Acid–Base Disorders According to PCO₂, SID, and A_TOT

The physicochemical approach classifies acid–base disorders into six types with three independent variables: high and low values of PCO₂, SID, and A_TOT: (1) respiratory acidosis, (2) respiratory alkalosis, (3) high SID alkalosis, (4) low SID acidosis, (5) high A_TOT acidosis, and (6) low A_TOT alkalosis. Disorders due to abnormal HCO₃⁻ are conspicuously missing in this classification because HCO₃⁻ is a mere dependent variable.

Stewart's advocates claim that high SID and low A_TOT increase pH, whereas low SID and high A_TOT decrease pH, but no formula consistently predicts pH from these variables. In fact, many invalidating counter-examples exist: intravenous administration of tromethamine produces HCO₃⁻ and alkalosis, yet SID decreases because tromethamine cation replaces Na⁺ with unchanged Cl⁻, and administration of Na₂HPO₄/NaH₂PO_4, at 4/1 ratio, would keep pH unchanged, but SID increases because Na⁺ increases with unchanged Cl⁻. By contrast, changes in HCO₃⁻, without primary changes in PCO₂, predictably, change pH.

Two new disorders, hypoproteinemic alkalosis and hyperproteinemic acidosis, representing low and high A_TOT states, respectively, are similarly based on incorrect arguments and an imprecise experimental design. Proponents of these disorders base their proposition on eight patients who presented with an inverse association between serum albumin and HCO₃⁻, ignoring the dictum: correlation does not prove causation.⁷ Furthermore, in vitro experiments do not duplicate in vivo situations because major body buffers, residing outside of plasma, mitigate plasma HCO₃⁻ changes, and renal compensation corrects residual abnormalities, unless conditions increasing renal HCO₃⁻ threshold coexist; hypoalbuminemia is not one of them.

Furthermore, our analysis of an in vitro study,⁸ widely cited as experimental proof of validity of the new entities, strongly suggested that HCO₃⁻ changes were not caused by changes in albumin concentration, but most likely by changes in test solution volumes.⁸ HCO₃⁻ concentration can change only by (1) removal or addition, (2) generation or consumption by a buffer, and (3) changes in volume. Study protocol clearly precluded the first mechanism. HCO₃⁻ generation/consumption requires a buffer and CO₂ as a HCO₃⁻ source; neither was present according to the study protocol. The only plausible explanation is volume change.⁸ We speculate that a high HCO₃ concentration already present in the test solutions was maintained when small volumes of albumin-containing solution were added and reduced when larger volumes added.

The idea for an inverse relationship between HCO₃⁻ and albumin probably originated from the belief that loss of albumin, a major weak anion of plasma, must be compensated by another major weak anion, HCO₃⁻, not realizing that albumin loss requires accompanying cations, mainly Na⁺. Subsequent restoration of electrical charge balance occurs mostly by reduced Na⁺ and increased Cl⁻ and slightly by reduced HCO₃⁻. This exemplifies that changes in weak ions can change SIDs to maintain electrical neutrality. A major concept of SID has been the contention that only changes in SID can change weak ions, not the other way around.

Trivialization of HCO₃⁻ and Exaltation of A_TOT as Determinants of pH

Stewart trivializes buffering role of HCO₃⁻ while exalting that of albumin, the most prominent component of A_TOT. Facts are opposite. HCO₃⁻ is the most important alkaline buffer, whereas total absence of albumin has little effect on acid–base balance.

Although the human body contains more histidine than HCO₃⁻ (1580 versus 680 mmol), HCO₃⁻ is far more important in metabolic acidosis. For example, blood pH reduction from 7.4 to 7.1 in severe metabolic acidosis decreases HCO₃⁻ from 24 to 3 mEq/L buffering about 600 mmol of H⁺; the same pH change would change histidine⁰/histidine⁺ ratio from 4/1 to 2/1 (histidine pK of 6.8) buffering mere 205 mmol of H⁺.

Furthermore, importance of PCO₂ as buffer depends on HCO₃⁻. At very low HCO₃⁻ concentration, PCO₂ is an unimportant buffer. For example, at urine pH of 5.1 and PCO₂ of 40 mm Hg, HCO₃⁻ concentration would be 0.1 mmol/L. As PCO₂ of 40 mm Hg at voiding decreases to 0.3 mm Hg following air exposure and vigorous agitation, urine pH barely changes⁹ because HCO₃⁻ also decreases similarly as PCO₂ by reacting with other dominant urinary buffers, but consuming negligible amounts of them because of its very low concentration and therefore hardly changing ratios of the other buffer pairs.

Albumin has many important functions: principal determinant of plasma oncotic pressure, carrier of fatty acids, and binding protein of Ca⁺⁺ and Mg⁺⁺ and numerous drugs. However, buffering function is not one of them. Main relevant acid–base role of albumin comes from its abundant negative charges, which must be counted in calculating serum anion gap and strong ion gap.

Negligible importance of albumin as buffer can be further appreciated by comparing its histidine content, main buffer, with that of hemoglobin. Each albumin contains 16 histidine, and thus plasma in 2.75 L, at 42 g/L of albumin (0.63 mmol/L), contains 28 mmol of histidine. By contrast, 5 L of whole blood at 150 g/L hemoglobin, with 38 histidines in each tetramer, contains 440 mmol of histidine, 15.7 times more than plasma albumin. Reducing hemoglobin from 15 to 14 g/dl would reduce histidine content by 29 mmol, whereas complete loss of plasma albumin would result in loss of 28 mmol.

H⁺ buffering by histidine reduces net negative charge of plasma proteins by increasing cationic charges, not through decreased anionic charges. The impression that albumin is a weak anion probably derives from the coincidence that net charge difference of amino acids (n=15) nearly equals number of histidine molecules (n=16) in albumin. This unique coincidence is not duplicated in hemoglobin, which has a neutral charge, despite having 15 times more histidine than albumin. If a genetic mutation exchanged all anionic amino acids for cationic amino acids in albumin, its buffering capacity would now be only slightly greater than the native albumin because of downward shift of pK of some lysines, but net charge would now be strongly positive.

Serum Strong Ion Gap Is Better than Serum Anion Gap as a Clinical Bedside Tool

Strong ion gap is calculated as apparent SID, measured as (Na+K+Ca+Mg)−(Cl+lactate+sulfate) minus effective SID, measured as weak anions+HCO₃⁻+protein anions.¹⁰ Despite the use of complex terms implying some deep connection between SID and strong ion gap, the latter is merely another form of serum anion gap when all known electrolytes are included in its calculation. Its use does not require understanding the concept of SID. Yet, SID advocates often claim the (unproven) greater accuracy of strong ion gap over simple serum anion gap^10,11 as evidence of superiority of the concept of SID. These claims are of course as meaningless as claiming serum anion gap including K+ is more accurate than without it.

SID advocates pay utmost care to accurate measurement of phosphate and albumin,^10,11 while ignoring contribution of other ions such as plasma globulins and organic anions. Routine laboratory errors in measurements of serum Na⁺ and Cl⁻ often far exceed errors from omitting to correct pH effects on serum phosphate charges. Moreover, the test choice in clinical medicine is not always accuracy, but sometimes convenience; accordingly, the simplest formula for anion gap excluding K⁺ is most widely used.

Conclusions

We have critically and quantitatively reviewed and analyzed Stewart's approach and disagree with almost all of its claims. We find no merits in it and conclude that it introduces confusion and unnecessary complexity.

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/JSN/E929.

Funding

None.

Author Contributions

Conceptualization: Man S. Oh, Jaime Uribarri.

Writing – original draft: Man S. Oh, Jaime Uribarri.

Writing – review & editing: Man S. Oh, Jaime Uribarri.