In brief summary:
|increased albumin = decreased ionised calcium|
|increased pH = decreased ionised calcium|
|increased lactate = decreased ionised calcium|
|increased phosphate = decreased ionised calcium|
|increased bicarbonate = decreased ionised calcium|
|increased citrate = decreased ionised calcium|
|Presence of heparin in the sample = decreased ionised calcium|
Free fatty acids
|Increase in free fatty acids = decreased ionised calcium|
One can find an excellent overview of calcium in the Electrolyte Quintet series from the Lancet. It is an article you migth have to pay for. The freegan can instead avail themselves of the excellent calcium chapter from the free-to-view Chapter 143 from the 3rd edition of Clinical Methods: The History, Physical, and Laboratory Examinations.
The distribution of calcium in the body fluid compartments is also briefly discussed in the electrolyte section of this site.
Physiology of calcium ionisation
First of all, let us address the misnomer of "ionised" calcium. It's all ionised, people. Unless it is in a covalent bond with something, it is present as an ion. To be sure, it might be clinging to the side of an albumin molecule, or it might be complexed with a chelator like citrate, but it is still an ion, and not a member of a molecule per se. However, the trend of calling unbound calcium "ionised calcium" persists, and is prevalent at all levels of academic medicine. It is only this fraction which actually does anything useful, and is also the fraction which enjoys tight regulation by PTH and vitamin D; whereas the bound fraction is physiologically useless. Respecting tradition and eschewing pedantry, hereafter "ionised calcium" will be used to in reference to the free fraction.
With that unpleasantness now behind us, let us consider the serum transport of calcium.
Generally speaking, when one reads a textbook on this topic, one typically encounters a figure of 45-55% for the amount of calcium which is present in the serum in its free unbound state. The Lancet article reports that each 10 g/L of albumin bind about 0.2 mmol/L of calcium. This means the totally deconditioned ICU patient with 10g/L of albumin will have an ionised calcium which is about 0.6mmol/L higher than the patient with a normal albumin, if the total serum calcium level is the same for both. Not all protein-bound calcium is bound purely to albumin - some (about 10%) is complexed with globulins.
Binding of calcium to albumin
Albumin binds calcium with twelve available cation-binding sites. Of these, usually only 10-15% are occupied (i.e. only one or two sites). This means there are plenty of other sites to bind other cations (eg. magnesium) and the divalent cation species rarely enter into binding site competition with one another.
However, the elusive hydrogen (or hydronium) cation is a constant source of competition, and as its concentration increases in acidosis, calcium binding sites become less available, and albumin-calcium binding decreases, increasing the free fraction. The influence of pH on ionised calcium is discussed in slightly greater detail elsewhere.
"Diffusable" calcium complexes
Of the total body calcium, the albumin-bound fraction is confined to the plasma, whereas all the small-anion complexes (eg. calcium and sulfate) are little enough to pass through various biological membranes, and can be described as "diffusable" or ultrafitrable" calcium. Among these complexes, binding to phosphate increases with alkalosis (i.e. the hydrogen ion competes with calcium) whereas the sulfate and citrate binding seems to be fairly pH-independent, at least within the survivable range. In the day-to-day bedside management of ICU patients, the presence of these species plays little role, except when vast amounts of citrate are administered (eg. in massive transfusion). There are also more exotic circumstances of largely historical interest. For instance, embarrassed cardiologists from the 1970s report discovering that EDTA and citrate in their radiocontrast media sucked the calcium right out of people's coronary arteries, with occasionally lethal results.
Within the context of routine practice, the most relevant interaction of calcium is with lactate. Dissociated lactate decreases the pH, which influences calcium-albumin binding (increasing the free fraction of calcium) - but that is a boring pH-related effect. The more exciting effect is the chelation of calcium by lactate. Lactate has a higher affinity for calcium than albumin, and can strip calcium ions from its imidazoline bindings sites. However, its ability to chelate calcium is about two orders of magnitude weaker than than of "real" chelators, such as citrate (this ability is discussed in terms of the association constant, which is 1300L/mol for citrate and only 14L/mol for lactate). Thus, much of the time in lactic acidosis one sees an increase in ionised calcium, rather than a decrease - the acidosis forces more calcium off albumin than the lactate has capacity to chelate.
Chelation of calcium by heparin in the sampling syringe
Heparin in the sample can also bind calcium, as it is a molecule with vast amounst of negative charge. Generally, it is recommended that you keep the heparin concentration of your sample below 15 units/ml. This is usually only an issue of the heparin content in the syringe, and systemic heparinisation will never interfere with your systemic calcium ionisation. To arrive at a systemic serum concentration of 15u/ml, one would have to administer 75,000 units of heparin to the patient as a bolus, which is a dose not usually seen in routine practice.
Enhancement of calcium-albumin binding by free fatty acids
The presence of free fatty acids, particularly olic and palmitic acids, has the effect of enhancing albumin-calcium binding; thus the higher the free fatty acid levels, the lower the serum calcium. The mechanism for this is not completely understood. It was discovered by accident, when albumin solutions were observed to lose some of their capacity to bind calcium after being washed with ether, suggesting that some sort of ether-soluble "calcium binding enhancer" was present.
Cis-unsaturated fatty acids (such as the TPN favourites, oleic and linoleic acids) seem to have a far greater effect on calcium binding than unsaturated free fatty acids. The effect seems to be due to a conformational change in the albumin molecule, an idea which is supported by changes in the wavelength of maximum fluorescence emission in albumin (i.e. as the calcium binding changes with increasing fatty acid concentration, so does the maximum fluorescence emission of albumin).
The levels of cis-unsaturated FFAs required to exert a clinically significant effect are actually achievable in vivo, provided the patient is receiving some sort of nice fatty lipid infusion. And indeed, this effect has been confirmed in vivo, among patients being infused with TPN.
Discrepancies between total and ionised calcium measurements
Observe the following colourful cylinder diagrams. The total calcium remains the same, but the ionised fraction changes significantly, with the appropriate symptoms developing.
Now, when the serum calcium is reported by the automated analyser from the central laboratory, one typically receives both a calcium value (oh, say, 2.2mmol/L) as well as a "corrected calcium" value. The measurement of this "formal" value is arrived at by colorimetry. Its an automated technique, where an the sample is diluted, an indicator is added which turns some vivid hue, and an absorption spectrophotometer is used to determine the indicator-calcium complex concentration. The indicator is usually some high-affinity molecule which is able to outcompete all the other serum calcium chelators, and thus is expected to bind all of the calcium present in the sample. This technique has superceded the old-school method of determining serum calcium by adding oxalate to the blood sample and measuring the volume of the resulting insoluble precipitate.
The profound uselessness of the corrected calcium level in the ICU
So, if the formal technique measures the total calcium, then why must it be corrected for albumin? Surely, the albumin should play no role in the determination of calcium by colorimetry?
The "corrected calcium" level is actually a mathematical construct.
It is determined by the following formula, first described by Payne in 1973:
So, say the hypoalbuminaemic patient's measured total calcium is reported as 2.2 mmol/L. Their albumin is 10g/L, and plugged into the formula one arrives at a corrected value of 2.8 mmol/L, making that patient hypercalcaemic.
Does this mean that this patients serum actually contains this much calcium?
Of course not.
The calcium correction formulae were developed in the seventies because, to quote Payne himself, "Measurement of ionized or diffusible calcium is at present technically complex... a screening procedure to detect patients with abnormal protein concentrations and to allow for the effect of changes in protein-bound calcium on the measurement of total calcium is desirable". In short, this method reports for you the calcium level your patient would have if their albumin were normal, so you can decide whether their ionised calcium might be high. A nominally normocalcemic patient might be clinically hypercalcemic because their ionised calcium fraction is raised due to their low albumin.
Thus, the correction of measured calcium is totally redundant in the technology-infested Borg empire of the modern ICU environment, where ionised calcium measurements are immediately accessible. Furthermore, Payne's formula breaks down in the presence of raised albumin, and may be wildly misleading when something other than albumin is doing all the calcium-binding (eg. in citrate toxicity).