Historic overview

The study of acid–base equilibrium and its relationship to the diet and disease has been a subject of considerable speculation for at least several centuries. But, before the 19th century, little was known about the concepts of acids and bases, and no means were available to quantify the acid or alkaline load of foods, or of the pH of physiological processes. Not until the second half of the 19th century did nutritional science began to develop, and the chemical components of food start to be analysed. Henry Clapp Sherman's book in 1919, [The] Chemistry of Food and Nutrition, describes the history of nutritional studies performed up to that date (Sherman was the first to quantify nutritional acid load in 1912).

During the next half century, a number of clinical practitioners outside of academia began to observe health improvements in their patients from consuming unprocessed raw fruits and vegetables, instead of the processed food which were increasingly becoming the standard fare, and developed the concept that life is an equilibrium between acid and base. However, it was not until the 1960 s that input–output acid–base balance studies were performed for the first time in healthy adults and patients with chronic renal acidosis⁽Reference Lennon, Lemann and Litzow^1, Reference Goodman, Lemann and Lennon²⁾. There is still a great deal of debate on how to define acids and bases. In the 1980s, Peter Stewart challenged traditional thought and mathematically determined that hydrogen ion and bicarbonate concentrations were dependent variables. He considered the independent variables, which by definition would determine the dependent variables, to be the strong ion difference (the difference in the net charge of cations and anions fully dissociated in solution), the partially dissociated weak acids (albumin, phosphate), and the partial pressure of carbon dioxide (P_CO2) of the solution. While a full discussion of the implications of the differences between theories is beyond the scope of the present paper, of significance is the lack of consensus in the fundamental understanding of the mechanisms, as well as the practical applications of acid–base chemistry in physiological systems. A lengthy review published in January 2008 emphasises this point⁽Reference Kurtz, Kraut and Ornekian³⁾, and excellent reviews of acid–base history are available⁽Reference Manz^4, Reference Story⁵⁾.

Definitions: acidosis, acidaemia and diet-induced acidosis

For the purposes of the present review, some definitions are in order, though it should be clear that most definitions are not universally accepted. The term acidosis is often used interchangeably with the term acidaemia, with the latter referring to a blood pH of less than 7·35. Correctly used, the term acidosis refers to a process, or a trend toward acidaemia, without necessarily reaching a pH of less than 7·35, or actual acidaemia. The concentration of H⁺ in blood plasma and various other body solutions is among the most tightly regulated variables in human physiology. Acidosis only becomes acidaemia when compensatory measures to correct it fail. To illustrate the difference between acidosis and acidaemia, take the following example: two processes occurring simultaneously in the same individual, such as a respiratory acidosis combined with a metabolic alkalosis. In this case, if the respiratory trend toward acidosis is greater than the metabolic trend, a pH of less than 7·35 may be reached, and would be considered acidaemia, despite the presence of a metabolic alkalosis. The intensity of each ‘process’ will determine the pH, but the terms themselves (acidosis, alkalosis) do not indicate a certain pH.

What should also be defined, at least broadly, is the net acid load of the diet, the primary topic of the present paper. Diet net acid load can be estimated from measurements of urinary excretion of ammonium, titratable acids and bicarbonate (called net acid excretion; NAE), or can be calculated from dietary constituents (called net endogenous acid production; NEAP). The techniques for measuring or quantifying the acid load of the diet will be discussed in a later section, but, in general, food contributes a net acid or base effect due to the balance between the acid-forming constituents, such as sulfuric acid produced from the catabolism of methionine and cystine in dietary proteins, and the base-forming constituents, for example, bicarbonate, produced from the metabolism of the K salts of organic anions in plant foods. As predicted by the Stewart hypothesis, sodium chloride appears to affect systemic acid–base status independently of the net acid load of the diet, perhaps by affecting renal excretion of Cl⁻ /, or by a strong ion effect⁽Reference Frassetto, Morris and Sebastian⁶⁾. The effects of sodium chloride are especially relevant, given the high salt content of the typical diet in industrialised countries.

With an increasing understanding of acid–base chemistry has come recognition of the significant differences between contemporary diets and diets more typical of Homo sapiens ancestors. Although of course we do not know exactly what our hominid ancestors ate, studies in hunter–gatherer tribes suggest a relatively high intake of plant foods compared with modern-day humans⁽Reference Cordain, Miller and Eaton⁷⁾. In a recent study estimating the net acid load (NEAP) of 159 hypothetical pre-agricultural diets, 87 % were found to be base producing, with an estimated mean NEAP of negative 88 mEq/d. In comparison, calculations from the US Third National Health and Nutrition Examination Survey (NHANES III) found the average American diet to be acid producing, with an NEAP of positive 48 mEq/d⁽Reference Sebastian, Frassetto and Sellmeyer⁸⁾. This represents a switch from the net base-producing diet we ate for the majority of our evolutionary history to the net acid-producing diet we have been eating for only several thousand years. Much of the current research highlights this change and the potential long-term physiological consequences of a chronic low-grade metabolic acidosis among those eating the typical Western diet, and the effects of reducing or eliminating this diet acid load by altering the diet or giving base supplements.

Is acidosis a real physiological phenomenon?

The human body tends to maintain a tightly controlled pH of about 7·40 in the extracellular fluid by respiratory excretion of carbon dioxide and renal excretion of non-carbonic (non-volatile) acid or base⁽Reference Paulev and Zubieta-Calleja⁹⁾. Everyday metabolism produces acid as non-volatile sulfate from amino acid catabolism, non-metabolised organic acids and phosphoric and other acids. The kidney reabsorbs all of the filtered bicarbonate () and generates new bicarbonate in the collecting duct. Under normal steady-state conditions, the net quantity of acid secreted and the consequent renal generation of new bicarbonate equals the rate of metabolic proton generation, preserving pH balance. In metabolic acidosis, either non-volatile acid accumulates, or is lost (for example, in diarrhoea) and this can be happening even when the plasma is within the range considered to be normal (24–28 mmol/l)⁽Reference Gluck¹⁰⁾.

While acute acid loading may only temporarily disrupt acid–base equilibrium, a chronic perturbation occurs when metabolism of the diet repeatedly releases non-carbonic acids into the systemic circulation in amounts that exceed the amount of base released concomitantly (for example, bicarbonate from combustion of organic acid salts of K in vegetable foods)⁽Reference Frassetto, Todd and Morris¹¹⁾. The size of the discrepancy between acid and base production determines the NEAP rate. To maintain equilibrium when there is a net retention of acid (H⁺), at least three compensatory physiological responses are activated: buffering, increased ventilation, and increased renal reabsorption and generation of . The major reservoir of base is the skeleton (in the form of alkaline salts of Ca) which provides the buffer needed to maintain blood pH and plasma bicarbonate concentrations. To some degree, skeletal muscle also acts as a buffer⁽Reference Bailey, England and Long¹²⁾. Respiratory ventilation increases within minutes if the acidosis is great enough, and the kidneys compensate by increasing reabsorption, H⁺ secretion, and production of the urinary buffer ammonia, all in response to an acidic load.

Thus the loss of bone mass due to acidosis has generally been considered a passive process, a physico-chemical dissolution of the matrix. However, bone dissolution is more than just a passive process, but rather is one of active resorption by osteoclasts, with extracellular H⁺ being a key inducer of osteoclastic activity. Indeed, extracellular H⁺ has been suggested to be the ‘long-sought osteoclast activation factor’, partly because in vitro experiments show osteoclasts to be almost inactive at a pH above 7·4, with pH reductions of ≤  0·1 sufficient to cause a doubling of resorption pit formation, removing both mineral and organic components of bone⁽Reference Arnett¹³⁾. A number of in vitro studies have shown that even very small changes in pH have an effect especially on osteoclastic resorptive activity and on osteoblastic activity, with acidosis exerting a selective, inhibitory action on matrix mineralisation⁽Reference Brandao-Burch, Utting and Orriss^14–Reference Bushinsky¹⁶⁾. Most recently, Frick et al. have demonstrated that ovarian cancer G protein-coupled receptor-1 (OGR1) is the proton-sensing receptor on the osteoblast that leads to osteoclast activation⁽Reference Frick, Krieger and Nehrke¹⁷⁾.

In vivo studies have generally supported the in vitro findings that acid-promoting diets are associated with both increased Ca and increased bone matrix protein excretion (used as a marker for estimating bone loss), and that neutralising the acid intake with diet or bicarbonate supplements decreases urine Ca and bone matrix protein excretion⁽Reference Buclin, Cosma and Appenzeller^18–Reference Sebastian, Harris and Ottaway²¹⁾. In a trial of 170 postmenopausal women, for example, potassium bicarbonate supplementation reduced daily urinary Ca excretion, and one could predict which women would benefit most – those with the greatest urinary Ca loss⁽Reference Frassetto, Morris and Sebastian²²⁾.

Additionally, the same mechanism may be involved in Ca nephrolithiasis⁽Reference Osther^23–Reference Osther, Bollerslev and Hansen²⁵⁾. A common risk factor for Ca stone formation appears to be hypocitraturia, which has been associated with a low urinary K level and a more acidic urinary pH, both of which can be predicted by dietary intake⁽Reference Domrongkitchaiporn, Stitchantrakul and Kochakarn^26, Reference Borghi, Schianchi and Meschi²⁷⁾. However, it appears that dietary acid load is a better predictor of urinary citrate than the intake of most individual nutrients, including dietary K. In a trial of 187 patients with renal Ca stone disease, an inverse correlation ( − 0·18) was found between daily potential renal acid load (PRAL; defined below) and daily urinary citrate (P < 0·01). PRAL was also inversely related to urinary pH and K, and although dietary protein and K were correlated with citrate excretion, these relationships did not reach statistical significance⁽Reference Trinchieri, Lizzano and Marchesotti²⁸⁾. A low-salt, low-animal protein diet has also been shown to reduce risk of stone recurrence (relative risk 0·49) with men followed over 5 years⁽Reference Borghi, Schianchi and Meschi²⁷⁾. While there are several possible explanations for this effect, one is that the reduced salt and/or animal protein diet is likely to result in a lower acid load.

Finally, metabolic acidaemia has been associated with increased protein breakdown in rats⁽Reference Mitch, Medina and Grieber²⁹⁾ and possibly subjects with end-stage renal disease⁽Reference Papadoyannakis, Stefanidis and McGeown³⁰⁾. Mitch et al. demonstrated that increased activation of a ubiquitin–proteasome pathway (via an insulin- or insulin-like growth factor-1-mediated transporter⁽Reference Du and Mitch³¹⁾) increases the production of amino acids which are then made into glutamate in the liver, transported to the proximal renal tubular cells, and then excreted as ammonia (NH₃) and complexed to H⁺⁽Reference Mitch, Medina and Grieber²⁹⁾. This mechanism allows the kidneys to dramatically increase the amount of acid excreted daily.

What are the causes of acidosis?

The causes of metabolic acidosis include increased consumption or generation of organic acids, as well as either insufficient production of bicarbonate, or renal and/or gastrointestinal loss of bicarbonate, such as that seen in renal disease, diarrhoea, pancreatic drainage and biliary fistula. Causes of renal tubular acidosis include Sjögren's syndrome, systemic lupus erythematosus, urinary tract obstruction, fever, aldosterone deficiency and glucocorticoid administration⁽Reference Gluck¹⁰⁾. Many of these medical conditions are typically associated with overt laboratory abnormalities; either a frank acidaemia, a decrease in plasma bicarbonate, or an increase in the anion gap.

In comparison, diet-induced ‘low-grade’ metabolic acidosis has only very small decreases in blood pH and plasma bicarbonate within the range considered to be normal. Within that range, this means that the system equilibrates nearer the lower end of normal rather than the higher end of normal (see Fig. 1). But, if the duration of the acidosis is prolonged or chronically present, even a low degree of acidosis becomes significant. This less severe but more chronic ‘low-grade’ acidosis is thought to be brought about primarily by two factors: advancing age with a consequent decline in renal function, and diet, which may promote acidosis by both its net acid load, as well as its sodium chloride content. With age, the severity of diet-dependent acidosis increases independently of the diet, most likely due to a decline in kidney functional capacity with age⁽Reference Frassetto, Morris and Sellmeyer^20, Reference Du and Mitch^32, Reference Frassetto, Morris and Sebastian³³⁾. Renal insufficiency contributes to a metabolic acidosis by reducing conservation of filtered bicarbonate and excretion of acid.

Fig. 1 (a) Blood pH at constant net acid excretion (NAE) rate (r 0·34; P < 0·005). (b) Blood pH at constant urinary chloride excretion rate (renal NAE (RNAE) = 56 mEq/d) (r 0·22; P < 0·05). (c) Plasma bicarbonate at constant NAE rate (r − 0·36; P < 0·005). (d) Plasma bicarbonate at constant urinary chloride excretion rate (RNAE = 56 mEq/d) (r − 0·45; P < 0·001). Adapted from Frassetto et al. ⁽Reference Frassetto, Morris and Sebastian⁶⁾.

Diet's contribution to an acidotic state is now well documented. In the European Prospective Investigation into Cancer and Nutrition (EPIC) study of over 22 000 men and women, after adjusting for age, BMI, physical activity and smoking, a more alkaline diet (as calculated by PRAL (a diet-only-based estimate of the production of endogenous acid) was significantly associated with a more alkaline urine, both before and after adjustment for age, physical activity and smoking, and after exclusion for urinary protein, glucose and ketones, and for those diagnosed with high blood pressure and/or diuretic medication use (all factors known to influence urine pH)⁽Reference Frassetto, Morris and Sebastian³⁴⁾.

How is acidosis diagnosed?

Several approaches have been used to estimate the acid–base equilibrium of the body, as well as the net acid contribution from the diet. As discussed previously, this is an area without general consensus, and one that lacks a definitive marker of acid–base status. Because plasma pH is so tightly regulated, without acidaemia an acidosis may not be detected, despite the stress on the body's buffering capacity. Additionally, it has been shown that plasma bicarbonate level may be normal even when acidosis is present, and so neither plasma bicarbonate nor the serum anion gap is a sensitive indicator of acidosis⁽Reference Gluck^10, Reference Frassetto and Sebastian³⁵⁾. This situation is analogous to the impaired glucose tolerance that precedes frank blood sugar elevation. Waiting for acidaemia before recognising acidosis is not a sound clinical strategy.

NEAP represents the amount of net acid (acid minus base) produced by the metabolic system every day (i.e. a combination of cellular metabolism and exogenous acid and base loads from the diet). NEAP can either be measured by quantifying the inorganic constituents of diet, urine and stool, and of the total organic anion content of the urine (calculated as the sum of urinary inorganic sulfate and organic acid salts minus dietary organic anions less faecal organic anions)⁽Reference Lennon, Lemann and Litzow¹⁾, or from estimations of acid or base production from the constituents of the diet. Several methodologies have been proposed to estimate NEAP from dietary components⁽Reference Frassetto, Todd and Morris^11, Reference Welch, Mulligan and Bingham36,37⁾. To avoid confusion, when the NEAP is estimated (which it usually is), it is identified as ‘estimated NEAP’, and the algorithm used for the estimation clearly specified (see Table 1)⁽Reference Remer and Manz³⁸⁾.

Table 1 Algorithms to determine net endogenous acid production (NEAP)

PRAL, potential renal acid load; OA_est, organic acids estimated; P_i, inorganic P; TA, titratable.

The net amount of acid produced daily (NEAP) is closely tied to the net amount of acid excreted (NAE) by the kidneys daily. NAE can also be measured by quantifying the amount of acids in the form of ammonium and titratable acid in the urine and subtracting the amount of base (bicarbonate)⁽Reference Frassetto, Morris and Sebastian³⁴⁾. When measured directly, nearly 90 % of the variance in renal NAE among subjects is accounted for by differences in NEAP, and because of this closeness in value, they are often considered equivalent⁽Reference Frassetto, Morris and Sellmeyer²⁰⁾.

Another test often used to estimate the NEAP is the 24 h urine pH. Urinary pH represents an index of the diet-dependent NAE (correlation with calculated NAE, r 0·83; P < 0·001)⁽Reference Remer, Dimitriou and Manz³⁹⁾, as well as the PRAL⁽Reference Frassetto, Morris and Sebastian^34, Reference Frassetto, Lanham-New and Macdonald⁴⁰⁾. Additionally, urine pH can be adjusted to a target pH based on PRAL calculations for dietary intake. However, pH strip measurement of the first voided urine was not found to be predictive of the NEAP reflected by the 24 h urine NAE⁽Reference Remer and Manz⁴¹⁾.

Lastly, while urinary Ca is not considered a specific marker for acid–base equilibrium, it has been used as a marker of bone turnover, and elevated levels are a risk factor for kidney stone formation⁽Reference Michaud, Troiano and Subar^42–Reference Pietschmann, Breslau and Pak⁴⁵⁾. As discussed below, it does not necessarily indicate increased bone Ca loss, however. It could, for example, represent a change in intestinal absorption.

In summary, analysed NAE appears to be the closest approximation of ‘true’ NEAP, and is more easily determined. For simplicity, however, various computational models for estimating NEAP are often employed, each with its own advantages and disadvantages (see Table 1). For a more complete discussion of the pros and cons for each methodology, the reader is referred to a recent review⁽Reference Stitchantrakul, Kochakarn and Ruangraksa⁴⁶⁾.

How is acidosis normalised?

The normalisation of a low-grade chronic metabolic acidosis has been accomplished by two methods: change in dietary patterns and alkaline supplementation. Dietary factors that affect net acid production include the quantity and type of protein intake, fruits and vegetables and table salt (sodium chloride). Alkali supplementation is generally in the form of potassium or sodium bicarbonate or citrate.

Increased fruit and vegetable consumption, as well as K and Mg alkali intake, is consistently associated with a base-producing diet⁽Reference Damasio, Massarino and Durand⁴⁷⁾. It has been shown that a vegetarian diet has a considerably lower NEAP than both a high and moderate omnivorous protein intake⁽Reference Welch, Mulligan and Bingham³⁶⁾.

Reducing protein consumption down to the US dietary recommended intake in a trial of thirty-nine healthy premenopausal women has also been shown to reduce Ca excretion and raise urinary pH, as well as reduce markers of bone resorption⁽Reference Frassetto, Morris and Sebastian⁴⁸⁾. It should be emphasised that this trial did not evaluate a low-protein diet, but rather lowered what could be considered a high protein intake to a level of 0·8 g/kg for this population. Because renal formation is dependent upon adequate protein intake, an extremely protein-deficient diet may also increase acidosis⁽Reference Tucker, Hannan and Chen⁴⁹⁾. In fact, in a recent study of 161 postmenopausal women, protein intake had a positive association with lumbar bone mineral density, but only after adjusting for the negative effect of the sulfur content of the protein (sulfate), perhaps ‘reconciling reports of positive impacts of dietary protein on bone health with reports of a negative impact of the acid load from sulfur-containing amino acids’⁽Reference Ince, Anderson and Neer⁵⁰⁾. In children, a greater protein intake has been associated with greater bone strength, though this effect is negated if alkalinising nutrients are lacking. It should be noted, however, that clearly bone may be influenced by these minerals in ways unrelated to acid–base chemistry⁽Reference Remer⁵¹⁾.

Finally, increasing sodium chloride intake dose-dependently decreases blood pH and plasma bicarbonate levels⁽Reference Thorpe, Mojtahedi and Chapman-Novakofski⁵²⁾, independent of the partial pressure of carbon dioxide (P_CO2), creatinine clearance and dietary acid load⁽Reference Frassetto, Morris and Sebastian⁶⁾. This effect may be due to a decrease in the strong ion difference, as total chloride concentration increases relative to total Na concentration, an effect that may increase H⁺ concentration⁽Reference Alexy, Remer and Manz⁵³⁾. Subjects who are particularly sensitive to salt, generally defined as an increase of 3 to 5 mmHg for a given salt load, have more of a metabolic acidosis than those subjects who are salt resistant⁽Reference Frings-Meuthen, Baecker and Heer⁵⁴⁾. So, while everyone's net acid load would improve by lowering their dietary salt intake, some individuals should benefit more than others from this dietary intervention.

A number of supplemental interventions have also been used. Salts of carbonic acid are available in a variety of formats. These include sodium or potassium bicarbonate and calcium carbonate. Alkali salts are also available as citrate, acetate or hydroxides. As suggested above, giving Na salts may be partly counterproductive, given their other effects, and so most studies use K or Ca alkali salts. These salts dose-dependently decrease NAE⁽Reference Kellum^55, Reference Sharma, Kribben and Schattenfroh⁵⁶⁾.

Caution using alkali therapy without careful consideration and expertise in subjects with heart, lung or kidney disease is needed. In congestive heart failure, sodium bicarbonate impairs arterial oxygenation and reduces systemic and myocardial oxygen consumption in these patients, which may lead to transient myocardial ischaemia⁽Reference Lemann, Gray and Pleuss⁵⁷⁾. Additionally there may be several simultaneous processes affecting acid–base status among patients with congestive heart failure⁽Reference Frassetto, Morris, Sebastian, Burckhardt and Dawson-Hughes⁵⁸⁾. Similarly, bicarbonate loading may worsen exercise response in chronic obstructive pulmonary disease patients⁽Reference Bersin, Chatterjee and Arieff⁵⁹⁾. Finally, subjects with kidney failure may develop elevated blood K levels and potentially fatal cardiac arrhythmias if given K alkali salts, or volume overload and breathing problems if given Na alkali salts.

Is acidosis clinically relevant?

Finally, from a clinical standpoint, the most important question is whether or not acidosis has pathophysiological effects that could be ameliorated or abolished by reversing the acidosis. These detrimental effects are associated with long-term acidosis (years to decades), and offer significant potential for prevention-based interventions. Areas of recent interest on the effects of chronic low-grade metabolic acidosis include effects on bone, muscle and kidney stone formation. While the data are generally consistent for the effects of alkali salts on urine Ca, bone biomarker or urinary N excretion (that is, shorter-term effects), the data are less consistent for the longer-term effects of alkali salts on bone mineral density, muscle function or fracture incidence. Positive correlations have been demonstrated using epidemiological or cohort studies, but results of prospective, randomised, placebo-controlled studies on long-term outcomes are mixed.

Conclusion

The lack of consensus for both qualitative and quantitative aspects of acid–base chemistry in physiological systems as well as measurement has caused considerable confusion for both researchers and clinicians. This confusion has also complicated the search for cause and effect and made clinical application difficult and controversial. Nonetheless, the available research makes a compelling case that diet-induced acidosis is a real phenomenon, has significant clinical relevance, may largely be prevented through dietary changes, and should be recognised and treated.

Both dietary interventions (lowering protein and increasing fruit and vegetable consumption) and nutritional supplementation (with K and Mg salts) have been shown to normalise acidosis, but with discordant results on whether this is then associated with clinical improvement in bone, muscle or other physiological or pathophysiological conditions. A positive NEAP diet results in increased urine Ca, N and bone marker excretion, and predisposes to kidney stones. Whether or not, over the longer term, this translates to lower bone density, increased bone and muscle loss with ageing is unclear and requires further investigation.