FormalPara Key Summary Points
Low serum chloride levels are associated with adverse prognosis in patients with acute or chronic heart failure (HF) regardless of left ventricular ejection fraction and independently of other prognostic markers such as N-terminal pro-B-type natriuretic peptide levels.
It is not clear how hypochloraemia develops in patients with HF but it may be linked to neurohormonal activation, high-dose loop diuretic usage, and metabolic alkalosis.
It is not known whether hypochloraemia is a marker or mediator of adverse outcome in patients with heart failure, although there are several putative mechanisms that might suggest the latter. For example, hypochloraemia might be linked to increased neurohormonal activation, diuretic resistance, and increased risk of sudden cardiac death.
Acetazolamide may increase natriuresis and diuresis while also increasing chloride reabsorption and bicarbonate excretion and thus might be a useful treatment for patients with HF, hypochloraemia, metabolic alkalosis, and diuretic resistance.

Introduction

A link between low serum chloride concentrations, loop diuretics, and risk of death in patients following a heart attack was first reported in 1979 [1], and the first reported association between low serum chloride concentrations and increased risk of death amongst patients with heart failure (HF) was in 2007 [2]. The authors of neither paper made even a passing reference to the chloride findings in the discussion [1, 2], perhaps owing to a lack of understanding regarding the importance of serum chloride: the potential prognostic significance of low chloride has, until recently, not been appreciated.

Hypochloraemia is a common electrolyte disturbance and marker of adverse outcome amongst patients with HF independent of other prognostic markers, including hyponatremia (Table 1) [3,4,5,6,7,8,9,10,11,12]. The mechanisms are poorly understood. In this review, we will discuss the etiology of hypochloraemia in patients with HF, explore the possible mechanisms behind its association with adverse outcome, and consider what, if anything, might be done about it. The present article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

Table 1 Summary of reports of hypochloraemia in patients with heart failure

Hypochloraemia and Heart Failure

Chloride is the main anion in the plasma and extracellular fluid [13], and is freely filtered in the glomerulus of the kidney into the urinary space (tubular lumen). Renal tubular cells are asymmetric with an apical surface facing the urinary space and a basolateral membrane facing the renal interstitium (peritubular capillaries). The majority of chloride reabsorption occurs in the proximal convoluted tubule (PCT), paracellularly in the intercellular space passively along an electrochemical gradient as the permeability to chloride anions exceeds that of other anions such as bicarbonate [14, 15]. Active, trans-cellular, reabsorption occurs via Cl/anion counter transports (antiporters or exchangers) in particular formate amongst others (sulphate, iodide, oxalate, hydroxyl, and bicarbonate) on the apical membrane, and by a sodium-driven Cl/HCO3 antiporter and K+/Cl symporter on the basolateral membrane [16]. In the loop of Henle (LoH), further chloride reabsorption takes place via Na+/K+/2Cl co-transporters (NKCC2) on the apical membrane (the site of action of loop diuretics) and voltage-gated chloride channels on the basolateral membrane [17]. In the distal convoluted tubule (DCT) and collecting duct (CD) (responsible for ~ 5% of chloride reabsorption) chloride is reabsorbed by thiazide-sensitive Na+/Cl co-transporter and Cl/HCO3 antiporter and returns to the bloodstream via voltage-gated chloride channels on the basolateral membrane (Fig. 1) [18].

Fig. 1
figure 1

Chloride reabsorption along the nephron. The majority of renal chloride reabsorption occurs in the proximal convoluted tubule, paracellularly along an electrochemical gradient although transcellular Cl/anion transport also plays a role

At first sight, the origin of hypochloraemia seems likely to be similar to the putative etiology of hyponatremia in patients with HF: low chloride results from either haemodilution or depletion due to loop diuretics [19]. However, patients with hypochloraemia appear to fall into two phenotypes; those with concurrent hyponatremia and those with normal sodium concentrations [3]. The group with normal sodium has higher bicarbonate, and lower potassium concentrations (and a higher rate of clinically significant hypokalaemia (defined as a serum K+ < 3.5 mmol/l) [3].

An Association with Metabolic Alkalosis?

In other disease states, such as severe vomiting or mineralocorticoid excess, hypochloraemia is associated with metabolic alkalosis (HCO3 > 30 mmol/l) [20, 21]. Metabolic alkalosis is the most common acid-base abnormality in patients with HF, affecting up to half of patients admitted to hospital [22]. While activation of the renin–angiotensin–aldosterone system (RAAS) is usually linked to sodium homeostasis [23], data from in vitro and in vivo animal studies suggest that neurohormonal activation might play a significant role in the development and maintenance of a metabolic alkalosis in patients with HF (Fig. 2).

Fig. 2
figure 2

Possible association between hypochloraemia, metabolic alkalosis, and neurohormonal activation in patients with heart failure. Our proposed theoretical link between hypochloraemia, metabolic alkalosis, and neurohormonal activation in patients with heart failure is based on various in vivo and in vitro animal experiments. The dotted greyed lines denote that loop diuretics are only a contributing factor in this proposed model, rather than the driving force

In vitro and in vivo studies suggest that when noradrenaline [24], and angiotensin II [25] levels increase, bicarbonate reabsorption in the first segment of the PCT increases. Additionally, in vivo studies in rats show that aldosterone increases the activity of the H+-ATPase pump in the CD which increases H+ secretion into the urine [26]. The increased acidification of the urine might result in a net gain of bicarbonate by the body.

Loop diuretics might also contribute to a metabolic alkalosis: a so-called “contraction-alkalosis” due to decreased extracellular fluid volume resulting in increased bicarbonate concentration [27] is well recognized in the literature, but may be an over-simplification. In vivo, increased sodium delivery to the CD (due to apical NKCC2 co-transporter inhibition) increases the activity of the H+-ATPase pump, increasing H+ secretion into the urine [28]. In vitro studies in rats have found that hypokalaemia (a potential complication of loop diuretic use) promotes bicarbonate reabsorption in the PCT [29, 30], and hypokalaemia increases RAAS activation in humans with HF [31, 32], which might further drive bicarbonate reabsorption (Fig. 2).

In vitro and in vivo studies in both rabbits and rats suggest that increased bicarbonate reabsorption is accompanied by increased chloride excretion [33,34,35]. The same process may occur in humans [36]. One small study (N = 51) found that patients with HF and hypochloraemia had higher serum bicarbonate, and greater fractional chloride excretion than those with normal chloride levels while having similar fractional sodium excretion (Table 1) [8]. “Chloride wasting nephropathy”—persistent urinary chloride excretion—is seen in patients with hyperaldosteronism [37] and/or severe potassium depletion [38], and similar metabolic states have been reported in patients with HF [39].

An additional factor contributing to the maintenance of an alkalosis is that as serum concentrations of chloride fall (either due to increased excretion in response to increased bicarbonate reabsorption, or diuretic use, or both), there is less and less chloride filtered into in the urinary space. A threshold of low serum chloride may be reached beyond which bicarbonate excretion is inhibited as there is less chloride in the urine to exchange with bicarbonate [36, 37].

Patients can thus be trapped in a cycle of hypochloraemia and alkalosis, which is only partly due to loop diuretic usage (Fig. 2): for example, among patients admitted with HF, those with serum bicarbonate concentrations above the median (≥ 28 mmol/l) had more severe disease (lower left ventricular ejection fraction, worse renal function, and higher natriuretic peptide levels) but were on lower doses of loop diuretic than patients with serum bicarbonate below the median [25].

Chloride and Outcome

Whether a low chloride concentration is a marker or a mediator of adverse outcome is unknown although there are possible pathophysiological mechanisms, which might suggest the latter (Fig. 3).

Fig. 3
figure 3

Confirmed and possible associations between hypochloraemia and adverse outcome in patients with heart failure. The dotted lines denote possible links demonstrated in animal studies and the thick lines denote confirmed links in patients with heart failure

Diuretic Resistance

The with-no-lysine (WNK) kinases (WNK1, WNK3, and WNK4) are the first step in an enzymatic cascade which increases activity of the Na+/K+/2Cl and Na+/Cl co-transporters [40,41,42,43]. Chloride binds to the catalytic site of the kinases, thus inactivating them [44, 45]. In vitro and in vivo studies suggest that the activity of WNK1 and WNK4 is reduced at high chloride concentrations [44, 46], but increased at lower concentrations [47, 48]. Thus, hypochloraemia may increase the activity of both Na+/K+/2Cl and Na+/Cl co-transporters, meaning greater doses of loop diuretic are required to induce a diuresis. In addition, chronic use of loop diuretic leads to an increase in sodium delivery to the distal tubule with consequent hypertrophy of cells in the distal nephron. The hypertrophied cells reabsorb sodium more avidly, an effect that can be mitigated by increasing the dose of loop diuretic and/or the additional use of a thiazide diuretic [49].

Consistent with this idea, patients with hypochloraemia take higher doses of loop diuretics than those with normal chloride levels [3,4,5,6,7,8,9,10,11,12], but whether a high-dose diuretic is the cause of hypochloraemia or becomes necessary because of hypochloraemia-induced diuretic resistance is uncertain.

Effect on the RAAS

Renin secretion is controlled by the macula densa. These specialized cells are sensitive to sodium chloride, low concentrations of which in the urinary space leads to renin secretion from the juxtaglomerular cells of the afferent and efferent arterioles. Increased chloride (but not increased sodium) delivery to the macula densa suppresses renin release from the granular cells in the afferent arteriole and a subsequent fall in angiotensin II levels [50, 51]. Chloride and renin are inversely related in patients with HF [8]. This is the rate-limiting event in the RAAS.

Sudden Death

Chloride channels play a role in ventricular repolarization [52, 53], and in regulating the positive chronotropic effect of cardiac pacemaker activity [54]. Myocyte volume and pH are regulated, in part, by chloride-dependent co-transporters [55, 56]. Abnormalities of the chloride channels and co-transporters may be arrhythmogenic [57, 58] and can impair contractility [59]. Consistent with these observations, a large study of outpatients with HF found that patients with hypochloraemia had an increased risk of sudden death (Table 1) [3].

A Therapeutic Target?

Hypertonic saline (HS) increases diuresis and may improve outcome when given alongside intravenous furosemide in patients admitted with HF [60, 61]. However, data on changes in chloride levels are absent from almost all reports of HS and whether any observed benefit is due a change in chloride levels is pure speculation. A proof-of-concept study of oral chloride supplementation in patients with HF (N = 10) found that lysine chloride increased chloride levels but required enormous doses to affect only small changes in serum chloride (Table 1) [8]. Further work is ongoing (NCT03446651) [62].

Acetazolamide

Acetazolamide (ACZ) is a carbonic anhydrase (CA) inhibitor. CA catalyses the interconversion between carbon dioxide and water on the one hand, and hydrogen (H+) and bicarbonate ions on the other (Fig. 4). CA on the apical membrane of the PCT cell converts free H+ and bicarbonate to water and carbon dioxide in the urinary space; the water then diffuses back into the cell via aquaporin 1 channels, carbon dioxide freely diffuses across the apical membrane [63, 64]. There, the water and carbon dioxide are converted back to H+ and bicarbonate ions by intracellular CA.

Fig. 4
figure 4

Renal carbonic anhydrase and acetazolamide. Inhibition of renal carbonic anhydrase with acetazolamide might increase luminal bicarbonate concentrations, reduce intracellular hydrogen ion concentrations thus reducing sodium reabsorption via the Na+/H+ antiporter, and reduce movement of chloride out of the peritubular capillaries. ACZ acetazolamide

The newly formed H+ ions in the cell are excreted in exchange for urinary sodium via Na+/H+ co-transporters on the apical membrane [65, 66], and bicarbonate returns to the circulation via Na+/HCO3 and Cl/HCO3 antiporters on the basolateral membrane (Fig. 4) [67]. Inhibition of intracellular CA reduces production of intracellular H+, thus reducing sodium reabsorption via the Na+/H+ antiporters on the apical membrane, and inhibition of luminal CA reduces production of water and carbon dioxide, thus increasing urinary bicarbonate levels (Fig. 4).

ACZ increases bicarbonate excretion and chloride reabsorption in vivo [32, 68], and increases serum chloride levels in humans [69, 70]. The reasons behind this are not clear but may result from two potential mechanisms: firstly, increased HCO3 in the urinary space increases the negative charge thus increasing the electrochemical gradient along which chloride is reabsorbed in the PCT. Secondly, in vivo studies suggest that ACZ, separately from CA inhibition, also inhibits the basolateral Cl-/HCO3 antiporter in the PCT thus reducing movement of chloride out of the blood and into the cell (Fig. 4).

There are thus three ways in which ACZ might be beneficial for patients with HF: (1) increasing sodium excretion and increasing diuresis [71, 72]; (2) increasing bicarbonate excretion, which may reduce metabolic alkalosis [73, 74]; and (3) increasing renal chloride reabsorption, which may reverse hypochloraemia [69, 70].

The ADVOR study of ACZ in patients admitted with HF is aiming to recruit ~ 500 patients, the largest study of ACZ in patients with HF to date. The primary endpoint is treatment success (i.e., clinical decongestion defined as the absence of pleural effusion, ascites, and significant peripheral edema) after 3 days of treatment. Secondary endpoints include mortality and morbidity alongside changes in natriuresis, body weight, and natriuretic peptide levels [75]. There is no planned analysis of either chloride or bicarbonate changes but the data will give an insight into the usefulness of ACZ as a treatment for patients with HF.

Future Perspective: Is Prevention Better Than Cure?

Amongst patients admitted with HF, those with hypochloraemia that resolves by the time of discharge have a similar post-discharge prognosis to those with normal chloride concentrations throughout admission [5]. Conversely, incident hypochloraemia during admission is associated with an increased risk of adverse outcome post-discharge [5]. If hypochloraemia results from the inevitable combination of severe HF and high-dose loop diuretics, it may be that prevention of hypochloraemia, rather than the correction of an existing abnormality, may have the greater effect on outcome. Whether acetazolamide might be best employed as a preventative measure is unknown, but should be the focus of future research.

Conclusions

Hypochloraemia is a common electrolyte abnormality in patients with HF and is an important marker of poor prognosis. There are many unknowns as to how hypochloraemia develops and whether it has a pathophysiological effect in patients with HF. If the latter is true, it may be a therapeutic target. As ever, more work is needed.