Unmeasured ions have long captured the imagination of intensivists. Potential candidates include L-lactate, β-hydroxybutyrate, D-lactate, salicylate, formate and oxalate in toxicological situations, pyroglutamate, semisynthetic penicillins, sulphate and hippurate in renal failure, and occasionally urate and amino acids with catabolic states and total parenteral nutrition. Reports of increased tricyclic acid (TCA) cycle anions in shock are now emerging 12.

Their presence is often inferred from the anion gap (AG), calculated as [Na⁺] + [K⁺] - ([Cl^-] + [HCO₃^-]). When its reference range is exceeded, a search for unmeasured anions should commence, irrespective of the overall metabolic acid-base status, because a competing metabolic alkalosis can mask their presence. Likely culprits vary with the clinical scenario, but the search usually starts with L-lactate and β-hydroxybutyrate. During this process, stoichiometry is tracked between ΔAG (measured AG – normal AG) and the summed concentrations of suspect anions (always in mEq/l, because we are dealing in electrical neutrality). If ΔAG – [suspect anions] exceeds 2 to 3 mEq/l, then the involvement of other unmeasured anions is likely.

Unfortunately, the AG is a flawed instrument. Both sensitivity and specificity are reduced by perturbations of albumin (remembering that albumin negative charge forms the bulk of the normal AG), pH, [Ca²⁺], [Mg²⁺] and [phosphate] 3. The most promising alternative is the strong ion gap (SIG) 45 Like the AG, the SIG quantifies unmeasured anions minus unmeasured cations, but unlike its predecessor it is insulated from variations in [albumin], [phosphate], pH, [L-lactate], [Ca²⁺] and [Mg²⁺] 6.

In the previous issue of Critical Care, Bruegger and colleagues 1 combine SIG calculations with capillary electrophoresis, and report that anions associated with the TCA cycle, specifically citrate and acetate, contribute to the metabolic acidosis of canine haemorrhagic shock. Their data originate from an earlier experiment designed to investigate the benefits of a perflurocarbon-based oxygen carrier during resuscitation from a predefined oxygen debt 7. Capillary electrophoresis on specimens before shock, during shock and on resuscitation revealed maximal citrate elevations of 1.9 mEq/l, whereas the peak acetate increase was 3.4 mEq/l after shock. Together, these accounted for around 60% of corresponding SIG increases. Minor increases were noted in fumarate, sulphate and α-ketoglutarate. L-lactate reached 5.6 mmol/l.

Although these findings fuel ongoing speculation concerning TCA anions in shock, several potential confounders are worthy of comment. During preparation, the animals acquired major metabolic perturbations, with severe baseline hypoalbuminaemia (1.5 g/dl) and impressive hyperchloraemia (130 mmol/l), but (from the parent study) only mild anaemia (11 g/dl) 7. This suggests the administration of large fluid volumes during the surgical preparation phase. Most surprising in this context was a massive baseline plasma acetate (2.4 mEq/l), which is 40 times the level reported from a previous study in dogs (0.06 mmol/l) 8. The postshock acetate peaked at 5.8 mEq/l, over 30 times that in the previous report (0.19 mmol/l).

To our knowledge such prodigious acetate levels are unprecedented outside the setting of exogenous administration 9. In the parent study 7, Ringer's solution 15 ml/kg per hour was documented as infused during all but the shock phase. If this was Ringer's acetate, and if the animals had received both saline (as stated by Bruegger and colleagues 1) and Ringer's acetate, then this would explain much. Of relevance is a report that exogenous acetate can elevate hepatic citrate 10. Although the authors acknowledge that they re-infused blood containing citrate phosphate dextrose solution during the shock phase, thus introducing exogenous citrate, they clearly stated that no acetate-containing solutions were administered. Hence, apart from possible assay problems, these acetate concentrations are unexplained. A final caveat is that charge and dissociation indices for human albumin used in this study differ from those for canine albumin 1112, although the effect on SIG calculations is probably small.

Until now, talk of unmeasured ions in critical illness has largely been speculative, based on discrepancies in AG or SIG. Nonetheless, since the late 1960s reports have emerged of accumulating TCA cycle intermediates in shock and dysoxic states 1314. The pattern reported by Forni and colleagues 2 in human metabolic acidosis differed substantially from the findings reported by Bruegger and coworkers 1, with relatively small increases in isocitrate, α-ketoglutarate, malate and D-lactate, and in some cases citrate and succinate. Only on aggregate were these sufficient to inflate the AG. They did not measure acetate.

It is insufficient to invoke 'tissue stress' to explain such increases in TCA anions. Elevations must be considered within the context of anaplerosis and cataplerosis 15, which combine to maintain adequate concentrations of TCA intermediates. It is unclear how dysoxic states disturb this delicate balance.

So, do unmeasured anions appear in shock? It is highly likely. Do at least some have a mitochondrial source? This is very probable. Can there be massive acetate and citrate concentrations? We need more information.

Abbreviations

Competing interests