The Bernard-Cannon concept
The term “milieu interieur” was coined by Claud Bernard in 1865 to mean the composition of body fluids constituting the body internal environment. The term “homeostasis” was coined by WB Cannon in 1929 to mean constancy of the internal environment after its distortion by external stresses. (1) The Bernard-Cannon concept can be achieved through a variety of biochemical, endocrinal and metabolic regulatory mechanisms.  Many of these mechanisms operate on the principle of negative feedback.  Deviations from a given normal set point are detected by a sensor, and signals from the sensor trigger compensatory changes that continue until the set point is reached.
Maintaining homeostasis  is a function of many organ systems.   However, those systems do not play independent roles but rather they play a collective effect, or an inter-related effect.
Advantages of homeostasis mechanisms include increased chance of survival, allowing free and independent life, under different stressful circumstances with proper functioning of the brain and higher faculties for decision-making during human life.

Homeostasis Mechanisms
The buffering properties of body fluids, and the renal and respiratory adjustments to the presence of excess acid or alkali are important acid-base homeostasis mechanisms.
I. The Henderson-Hasselbalch model  
Acid-base balance is generally explained in terms of carbonic acid/ bicarbonate system, as the most important buffer system.  This system describes the relationship between the respiratory and metabolic components of acid-base balance; the partial pressure of carbon dioxide (PaCO2), and bicarbonate, respectively.  This relationship is defined by the Henderson-Hasselbalch equation.  It incorporates a carbon dioxide solubility coefficient of 0.0307, and an apparent overall dissociation constant for carbonic acid of 6.105.  The reliability of these constants has been questioned, specially when applied to plasma of critically ill patients(2).  
As bicarbonate is not an independent variable, it varies with changing PaCO2.  Thus, the changes of HCO3- per se cannot be used as an indicator of the metabolic contribution of any disorder unless this is taken into account.  This means that empirically derived correction formulae have to be used to adjust the bicarbonate for both acute and chronic changes in PaCO2.The interaction between the respiratory and metabolic components is that of an association , rather than that of a cause.
A number of methods of assessing the metabolic component have been devised including “the buffer base”, “the standard bicarbonate: SHCO3-”, “the base excess: BE”, and “the Siggard-Andersen nomogram”.
The term “buffer base” calculates the sum of all the plasma buffer anions i.e. HCO3-  plus the non volatile weak acid buffers. The later results from the metabolism of methionine and cystine in dietary proteins and the incomplete metabolism of carbohydrates and fats.
The term “standard bicarbonate ,SHCO3” is defined as “the plasma HCO3- in blood, fully saturated with oxygen, and equilibrated with PaCO2 of 40 mmHg at 37°C”.  It is used as a reference value to assess the metabolic change in HCO3-. (3)
“The base excess ,BE” is a more accurate measure of metabolic disturbances. Actual BE is “the concentration of strong acid or base required to return the PH of invitro whole  blood sample to normal, while maintaining PaCO2 at 40mmHg by equilibration at 37°C” (3).  The standard base excess (SBE) uses the blood diluted threefold with its own plasma giving an effective haemoglobin concentration of about 5g/dl (4).  With a negative SBE (base deficit), PH is lower than normal, and an alkali must be added, while with a positive SBE (base exess), PH is higher than normal, and an acid must be added to titrate the PH back to normal. However, two equal opposite effects can exist producing a normal BE.  Also, two metabolic acid-base abnormalities can exist while BE is normal. For example, hyperchloremic metabolic acidosis may develop by renal absorption of Cl-, with compensation by the alkalinizing effect of low albumin concentration, often developing in critically ill patients.  Hypoalbuminemia has been shown to be a common cause of metabolic alkalosis in the critically ill(4,5).
An original acid-base nomogram was developed by Siggaard-Andersen in the 1960s.  It shows reference areas for the different classes of acid-base disturbances, with PH on the abscissa and PaCO2 on a logarithmic ordinate (6).  It illustrates metabolic and respiratory patient conditions, and differentiates between acute and chronic cases. An important value on the monogram is the standard base excess (SBE), which allows estimation of base excess or deficit of the ECF.  
Although the Siggaard-Andersen monogram may be used to establish whether an acid-base abnormality is present or not, it does not identify its underlying cause (6).  It only allows the conclusion that the PH and PaCO2 values fall within an area of specific acid-base abnormality.  It does not show how the diagnostic point has arrived at this particular area.  What is currently needed clinically for acid-base estimation, is an arterial blood sample collected in a heparinized blood-gas syringe, and immediately analyzed in the blood-gas analyzer, measured at 37°C. 
Classification of acid-base disorders:
There are four primary acid-base disturbances. Respiratory acidosis or alkalosis refers to an increase or decrease in PaCO2 respectively, while metabolic acidosis or alkalosis refers to a decrease or increase in extracellular base excess in the form of plasma HCO3- respectively.  Academia or alkalemia refers to a decrease or increase in serum PH (<7.42 and > 7.38 ) ,respectively.  
For each primary acid-base disturbance, there is a secondary physiologic compensation, the quality and quantity of which can be predicted (7). Qualitatively the compensatory variable moves in the same direction as the primary disturbance. 
A primary decrease in serum HCO3- (metabolic acidosis) stimulates hyperventilation and a secondary decrease in PaCO2 (compensatory respiratory alkalosis). Compensation tends to return the PH towards normal. To quantitate this physiologic compensation, expected PaCO2 corresponding to the serum HCO3- value equals (1.5 x HCO3-  + 8 ±2).  If the patient PaCO2 value is greater than that predicted, it means less adequate or partial compensation. If the patient PaCO2 value is less than that predicted it refers to over-compensation by a primary respiratory alkalosis, a condition of type II mixed disorder.
A primary increase in serum HCO3- (metabolic alkalosis) causes hypoventilation and a physiologic increase in PaCO2 (compensatory respiratory acidosis).  To quantatiate this physiologic compensation, a formula is used with expected PaCO2 value which equals (HCO3- x 0.7 + 20 ± 1.5) to deduce  partial compensation or  primary respiratory acidosis denoting a mixed type II disorder.
The causes of metabolic alkalosis are usually classified by their response or resistance to chloride (or saline) administration. Not surprisingly,  patients who are saline responsive are volume depleted  with low urinary chloride of < 20 mEq/L.  This is due to vomiting, gastric suction, diuretics or post-hypercapnia. So, they improve with extracellular volume repletion. Patients with metabolic alkalosis who are resistant to chloride (or saline) administration have high urinary chloride of > 20 mEq/L.  For example, those with excess mineralocorticoids as in Cushing’s syndrome with hyperaldosteronism are volume expanded and improve with diuretic therapy. Hypokalemia contributes to metabolic alkalosis and carries significant therapeutic implications.
The compensation of primary respiratory acidosis or alkalosis involves renal reabsorption or excretion of HCO3- , respectively, a process that is not immediate. So, the acute (24 hours) response is less than that achieved in more chronic conditions. Complete HCO3- compensation by the kidney takes five days approximately. In respiratory acidosis, each 10mmHg increase in PaCO2 is compensated by  increases of 1 and 3 mmoL in serum HCO3- level for the acute and chronic respiratory disorder, respectively. In respiratory alkalosis, each 10 mmHg decrease in PaCO2 is compensated by corresponding HCO3- decreases of 2 and 5 mmol, respectively. Again, any deviation of the compensatory reaction from that predicted, indicates partial compensation or the presence of another primary acid-base disturbance referring to a mixed type II disorder.
The co-existence of two or more primary acid-base disturbances is termed mixed disorder. Clues to mixed disorders include a change in PaCO2  and serum HCO3- in opposite directions (type I) or over compensation by PaCO2 or HCO3- in the same direction (type II).  It is not unusual to find multiple causes of acid-base disturbances in the same critically-ill patient.
The Henderson-Hasselbalch approach works well whenever total proteins, albumin, and phosphate concentrations in the serum are approximately normal. However, when they are markedly abnormal, this approach frequently provides erroneous conclusions regarding the cause of acid-base disturbance. In this way, this approach is more descriptive than mechanistic. So, new tools are needed to identify the etiology of simple or complex acid-base abnormalities.These tools include the  anion gap (AG) concept and the physico- chemical Stewart approach.

II. The Anion Gap
An electrolyte abnormality is often the first laboratory sign of an acid-base disorder.  The principle of electroneutrality in body fluids necessitates that the number of positive charges contributed by cations equals the number of negative charges contributed by anions:
(Na+) + (K+) + (Ca++) + (Mg++) = (HCO3-) + (Cl-)+(HPO4)+(SO4)+ (proteins) + (organic acids).
For simplification, K+, Ca++, and Mg++ are considered unmeasured cations (UC), while HPO-4, SO-4, proteins, and organic acids are considered unmeasured anions (UA).
Na+ + UC = HCO3- + Cl- + UA
A total of 23 mEq/L UA (Proteins 15, organic acids 5, HPO4- 2, SO4-  1) exceed a total of 11 mEq/L UC (Ca++   5, K++ 4.5, Mg++  1.5). Those UA and UC are not routinely measured and formulate the AG, with a normal value of about 12 mEq/L (8).
The AG can be calculated through one of two formulas.
AG = UA – UC  
AG = Na+ - (HCO3- +Cl-)
The term AG is actually a misnomer   because in any body fluid, the sum of cations must equal the sum of anions, and in reality there is no true AG.

An increased AG may be due to an increase in the concentration of UA, a decrease in the concentration of UC, or both (9).
The AG may be high when organic acids such  lactate or ketone bodies (generated by incomplete combustion of carbohydrates or fatty acids, respectively) accumulate in the blood, when concentrations of phosphates and sulfates (derived from tissue metabolism) increase, when concentration of plasma proteins increase, and when K+, Ca++ or Mg++ concentrations decrease.
The AG concept is used to differentiate normal from high AG metabolic acidosis.
Normal AG metabolic acidosis is due to loss of HCO3-, either from the gastrointestinal tract or from the renal system, with a compensatory increase in Cl- plasma concentration. (10)  Again, normal AG acidosis may develop from ingestion of NH4Cl or carbonic anhydrase inhibitors; or from parenteral administration of large volumes of saline (0.9% NaCl).  In such normal AG or hyperchloremic metabolic acidosis, NaHCO3 therapy may be indicated.  Acidosis is mostly corrected by HCO3- to increase the HCO3- deficit while Na+ ions correct the strong ion difference (SID),  which equals Na+ - Cl- towards normal.
In high AG metabolic acidosis (usually due to excess  endogenous or exogenous organic acids), excess anions are buffered by HCO3-, and the physician should search for the underlying etiology(11). Treatment of the underlying cause, rather than HCO3- therapy, is mandatory, specially when AG values exceed 25 mEq/L.
The calculated AG is often unreliable in detecting increased concentrations of the gap anions (12).  It should be corrected or adjusted before categorizing  AG metabolic acidosis. For example, in hypoalbuminemia, the correction factor for albumin was defined as 0.25 to quantify the effect of the change of albumin concentration on the AG.(13)
Adjusted AG = calculated AG + 0.25  (normal albumin – measured albumin concentration) 
It has been reported that for each gram decline in serum albumin level, 2.5-3 mEq/L decrease of AG will take place (14). 
As another example, in an occult neoplasm like multiple myeloma, the positively charged gamma globulin may need adjustment for the AG to be valuable (15).
Mixed acid-base disorders cannot be diagnosed by simple AG calculation. In this situation, the change in AG (delta AG), should be compared to the change in HCO3- concentration (delta HCO3-) . This ratio shows the “deviation from 1: 1 correlation” (16)
                                 
 
When                 is zero, it diagnoses normal AG metabolic acidosis. When the ratio is unity, it diagnoses high AG metabolic acidosis because the ∆ AG increase equals the ∆ HCO3 decrease.  Deviations  from unity signal mixed acid-base disorders.  
The clinical usefulness of the AG, in  patient with metabolic acidosis, is that it shows whether the acidosis is associated with an increase in UA (e.g. lactate) if the AG is raised, or, if normal, a hyperchloremic acidosis (3).  However, the AG has few limitations:
- Limits of normal serum Na+, HCO3-, and Cl- levels are set within 95% of normal population. This may affect the fidelity of measured ion levels used in AG calculation (17).
- About two thirds of the AG are formed by proteins .  A decrease in serum albumin, commonly encountered in critically ill patients, can affect the calculated AG (18), and should be adjusted.  So, hypoalbuminemia, can mask an increased concentration of the AG (14).  Significant hyperlactatemia can be also missed in septic or hepatic patients with hypoalbuminemia. (19)

1. Acidosis due to lactic acid accumulation with volume depletion and hemoconcentration may cause hyperproteinemia and hyperphosphatemia leading to high AG metabolic acidosis.
2. Variations in UC concentrations as Ca++ and Mg++ can affect the AG calculation .  As an example, an increase in serum Ca++ due to hyperparathyroidism, would lower the AG (11).  On the other hand, an increase in serum Ca++ due to neoplasms like multiple myeloma, would show a low AG due to the positively charged gamma globulins(15).
3. Mixed acid-base disorders cannot be diagnosed by simple calculation of the AG, but through the ratio of ΔAG compared to Δ HCO3- concentrations (16). 

Limitations in Henderson-Hasselbalch model and AG concept       urged      the need to a further physico-chemical approach for acid-base     evaluation.
III. The Physico-chemical approach
An alternative concept in the understanding of acid-base regulation is the “physico-chemical approach” based on the work of Stewart, almost two decades ago (20).
In the mathematically based Stewart approach, there are dependent and independent variables. The hydrogen ion concentration (H+) and bicarbonate (HCO3-) are dependent variables.The strong ion difference (SID), the partial pressure of carbon dioxide (PaCO2), and the total concentration of non-volatile weak acids (ATOT) represented mainly by albumin and inorganic phosphate, are independent PH regulating variables. 
The dependent and independent variables should be decided as fundamental of this approach.  
Carbon dioxide diffuses freely across all membranes in the body, hence, it cannot be used to regulate PH. Given a constant PaCO2, metabolic acid-base effects can be calculated, (20).  
Strong electrolytes dissociate (ionize) almost completely when in solution into strong ions, and the non-ionized fraction remains negligible across the physiological PH range.  
The total concentration of weak acids (ATOT) includes ionized (A-) and non-ionized weak acids (HA).  ATOT can be roughly estimated by multiplying total plasma proteins (g/L) by a Van Slyke factor (21) of 0.243, and coined as Prot-.
If an independent variable is altered (SID, ATOT) ,then the dependent variable will change.  If a change in a dependent variable is observed, then there must have been a change in one independent variable at least. So, we must know all the independent variables in order to calculate the dependent ones, through mathematical equations.
Stewart designed a computer program to solve the resulting equations for the dependent variable for differing values of independent variables (20).
The SID, being the most important variable, is calculated as the difference between the positively charged strong base cation Na+, and the negatively charged strong acid anion Cl-.  To satisfy electrical neutrality, the SID should be balanced by changes in H+ or OH-. The normal value of SID is 39 ± 1 mmol/L (22). A decrease in SID of <38 mmol/L is associated with acidosis where the H+ is greater than the OH- ion concentration.  An increase in SID of >40 mmol/L is associated with alkalosis where the OH- ion concentration is greater than the H+ ion concentration.  In both conditions the difference between calculated SID and normal SID estimates the effect of a change in SID (23).  The kidneys are the most important regulators of SID for acid-base purposes. The concentration of strong ions in plasma can be altered by adjusting their absorption or secretion. Plasma Na+ is tightly controlled for preservation of intravascular volume by tonicity and plasma K+ is also closely controlled to ensure appropriate cardiac and neuromuscular functions.  Hence, we are left with plasma cl- as the strong ion that the kidney uses to regulate the acid-base status without interfering with other important homeostatic mechanisms.  This is the major factor that changes the SID relative to a constant Na+ concentration (24).  In the compensation of acidosis, the removal of cl- ions in the urine will increase the value of the SID in plasma, and thus help to return plasma PH towards normal. In the compensation of alkalosis, reabsorption of additional cl- ions by renal tubular cells will reduce the plasma SID and therefore lower the plasma PH. This is based on the fact that in order to maintain electrical neutrality within all compartments, movement of HCO3- ion out of the cell is balanced by movement of Cl- ions from plasma into the cells (the chloride shift). This Cl- movement between compartments helps  to keep a normal plasma PH.

In Stewart’s terminology, weak plasma acids as albumin are incompletely dissociated substances in the plasma. As albumin is the major contributor of weak acid anions, hypoalbuminemia has an alkalinizing effect while hyperalbuminemia has an acidifying effect (25). 
Plasma HCO3-, according to Stewart approach, can be calculated from the difference between SID and weak plasma proteins, and usually proves identical with actual HCO3- calculated by the Henderson-Hasselbalch equation (26).
HCO3-  = SID – Prot -
This modified the Henderson-Hasselbalch equation  in the following way(27):
PH = 6.1 + log 
For a particular value of PaCO2, PH is determined by a difference between SID and ATOT.  This equation enables to understand the scope of causes of PH deviation from normal, and to correct the abnormalities by adjusting the values of the three independent factors; SID, ATOT, and PaCo2.  The SID alone cannot determine the final PH value because a value of ATOT can either pronounce or suppress the effect of SID.  When ATOT is decreased, it has an alkalinizing effect. Recently, it has been proved that this approach enables to define the influence of strong ions, and weak acids on PH of body fluids (24,28).  In this way the Stewart approach is mechanistic rather than descriptive.It simply shows a cause-response relationship.

Classification of acid-base disorders:
Acid-base classification due to Stewart approach is based on derangements of the independent variable(s). 
Respiratory acidosis and alkalosis are those in which the first independent variable affected is the PaCO2.  A change in the plasma SID may then occur as a compensatory response.
Metabolic acidosis arises from conditions that cause either a reduction in the plasma SID or increase in ATOT. Stewart approach explains the commonly occurring metabolic acidosis following the administration of large volumes of normal saline. The concentration of Cl- in plasma increases to a greater extent than that of Na+ when normal saline (unlike plasma) contains Na+ and Cl- in equal amounts. This leads to a reduction in the value of plasma SID and a consequent decrease in PH. Metabolic alkalosis arises from conditions that cause either an increase in the plasma SID or a decrease in ATOT.It follows that the changes in electrolytes would change the acid-base status. 
Considering changes in  SID and plasma weak acids with consideration of the BE, the physico-chemical Stewart approach can be a useful tool to formulate water, electrolyte and acid-base homeostasis, on the basis of electroneutrality and conservation of mass. This approach should be encouraged to accompany the every-day clinical use of the traditional Henderson-Hasselbalch approach for more accurate acid-base manipulation and improved patient care.
Manipulation of acid-base  in clinical practice will remain an art that combines intelligent synthesis of the clinical history, physical examination, and laboratory data, in the context of the individual patient and the nature and course of his / her illness.
References

1-    Ganong WF.  The general and cellular basis of medical physiology. In: Review of Medical physiology, 17th ed, John Dolan and Chris Langan (editors).  Appleton and Lang, California 1995: 1-42.
2-    Rosan RC, Enalnder D, Ellis J. Unpredictable error in calculated bicarbonate homeostasis during pediatric intensive care: The delusion of fixed pk. Clin Chem 1983; 29: 69-73.   
3-    Androgue HJ, Madias NE. Arterial blood gas monitoring: Acid-base assessment.  In: Tobin MJ (Ed).  Principles and Practice of Intensive Care Monitoring, New York; McGraw-Hill Inc.  1998: 217-41.
4-    Fencl V, Jabor A, Kazda A, et al.Diagnosis of metabolic acid-base disturbances in critically ill patients.  Am J Resp Crit Care Med 2000; 162: 2246-51.
5-    Wilkes P. Hypoproteinemia, strong-ion difference, and acid-base status in critically ill patients.  J Appl Physiol 1998; 85: 1740-8.
6-    Siggaard-Andersen O. An acid-base chart for arterial blood with normal and pathophysiological reference area. Scand J Clin Lab Invest 1971; 27: 239-45.
7-    Muther R.S.  Recognition and management of life-threatening acid-base disturbances. The second international conference on critical care medicine. Conference syllabus, pp 89-96, 14-18 October 2002, Cairo, Egypt.
8-    Sendak MJ. Monitoring and management of perioperative electrolyte abnormalities, acid-base disorders, and fluid replacement.Chapter 47 In:Principles and Practice of Anesthesiology, second ed, Longnesker DE ,Tinker JH and Morgan GE (editors) Mosby,1998:942-1010.
9-    Cogan MG.  Fluid and electrolytes: Physiology and pathophysiology.  In: Cogan MG (Ed), 1st edition. Lang Medical book, 1991: 313-9.
10-  Caravaca F, Arrobas M, Pizzare JL, et al.  Metabolic acidosis in advanced renal failure : difference between diabetic and non-diabetic patients.  Am J Kid Dis 1999; 33: 892-8.
11-  Oster JR, Gutierrex P, Schlessinger FB.  Effect of hypercalcemia on the anion gap.  Nephron 1990; 55: 164-9.
12-  Salem MS, Mujais SC.  Gaps in the anion gap.  Arch Intern Med 1992; 152: 1625-9.
13-  Story DA, Morimatsu H, Bellomo R.  Strong ions, weak acids and base excess: a simplified Fencl-Stewart approach to clinical acid-base disorders.  Br J Anaesth 2004; 92: 54-60.
14-  Gabow PA.  Disorders associated with an altered anion gap.  Kidney Int 1985; 27: 472-83.
15-  Fang JT, Kuo CT.  IgA myeloma associated with decreased anion gap and renal failure.  Renal Failure 1997; 19: 481-3.
16-  Martin L. Diagnosing acid-base disorders from serum electrolytes: the anion gap and the bicarbonate gap.  In: All you really need to know to interpret arterial blood gases 2nd ed, Lippincott Williams & Wilkins Feb. 1999; 31: 7-10.
17-  Nanji AA.  Misleading biochemical laboratory test results.  Can Med Assoc J 1984; 130: 1435-41.
18-  Figge J, Jabor A, Kazda A, et al.  Anion gap and hypoalbuminemia.  Crit Care Med 1998; 26: 1807-1
19-  Levrant J, Bounatirou T, Tchai C, et al.  Reliability of anion gap as an indicator of blood lactate in critically ill patients.  Intensive Care Med 1997; 23: 417-22.
20-  Stewart PA.  Modern quantitative acid-base chemistry.  Can J Physiol Pharmacol 1983 ; 61 : 1444-61.
21-  Van Slyke DD, Hastings AB, Hiller A.  Studies of gas and electrolyte equilibria in blood.  XIV: Amount of alkali bound by serum albumin and globulin. J Biol Chem 1928; 79: 769-80.  Quoted from reference 29.
22-  Kazda A, Jabon A. Mixed acid-base balance disturbances and their diagnosis. Klin Biochem Metab 2000; 8: 161-71.
23- Boyle M, Lawrence J.  An easy method of mentally estimating   the metabolic component of acid/base balance using the Fencl Stewart approach.  Anaesth Intensive Care 2003; 31: 538-47. 
24-  Otto S.; Karel M. Relation between PH and the strong ion difference in body fluids.Biomed2005;149:64-73. 
 25- Figge J, Mydosh T, Fencl V. Serum proteins and acid-base equilibria: a follow-up. J Lab Clin Med 1992; 120: 713-19.
26-  Scheingraber S, Boehme J, Scharbert G, et al.  Moniroting of acid-base and regulating variables during abdominal lavage.  Anaesth Intensive care 2004; 32: 637-43.
27-  Matousovic K, Marlinek K, Kvapil M.  Acid-base balance of body fluids, and its quantitative physico-chemical evaluation. Aktual Nefrol 2002; 8: 150-6.
28-  Corey HE. Stewart and beyond new models of acid-base balance. Kidney Int 2003; 64: 777-87