“I don’t get no respect”: the role of chloride in acute kidney injury

Abstract

Acute kidney injury (AKI) is a major public health problem that complicates 10–40% of hospital admissions. Importantly, AKI is independently associated with increased risk of progression to chronic kidney disease, end-stage renal disease, cardiovascular events, and increased risk of in-hospital and long-term mortality. The chloride content of intravenous fluid has garnered much attention over the last decade, as well as its association with excess use and adverse outcomes, including AKI. Numerous studies show that changes in serum chloride concentration, independent of serum sodium and bicarbonate, are associated with increased risk of AKI, morbidity, and mortality. This comprehensive review details the complex renal physiology regarding the role of chloride in regulating renal blood flow, glomerular filtration rate, tubuloglomerular feedback, and tubular injury, as well as the findings of clinical research related to the chloride content of intravenous fluids, changes in serum chloride concentration, and AKI. Chloride is underappreciated in both physiology and pathophysiology. Although the exact mechanism is debated, avoidance of excessive chloride administration is a reasonable treatment option for all patients and especially in those at risk for AKI. Therefore, high-risk patients and those with “incipient” AKI should receive balanced solutions rather than normal saline to minimize the risk of AKI.

INTRODUCTION

Acute kidney injury (AKI) is a major public health problem that complicates 10–40% of hospital admissions. Importantly, AKI is independently associated with increased risk of progression to chronic kidney disease (CKD), end-stage renal disease, cardiovascular events, and increased risk of in-hospital and long-term mortality (33, 34). It is the hope of the scientific community that better understanding of the mechanisms and contributors to AKI can lead toward interventions to reduce the risk of AKI and the adverse sequelae after AKI.

Fluids are frequently prescribed as prophylaxis against or for the treatment of AKI. 0.9% sodium chloride solution, also known as “normal saline,” is the most frequently used intravenous resuscitation fluid worldwide (133). Some have even proposed that it should be “the central feature of a rite of medical school passage” (55). Unfortunately, medical trainees lack knowledge regarding the daily salt and water needs of patients and the electrolyte contents of intravenous fluids (114). The chloride content of intravenous fluid has garnered much attention over the last decade, as well as its association with excess use and adverse outcomes, including AKI.

This comprehensive review first details the findings from clinical research in adults related to the chloride content of intravenous fluids, changes in serum chloride concentration, and AKI. The second half focuses on the complex renal physiology regarding the role of chloride in regulating renal blood flow (RBF), glomerular filtration rate (GFR), and tubuloglomerular feedback (TGF) and, lastly, proposes potential mechanisms of tubular injury.

CHLORIDE BACKGROUND

The chloride ion has major physiological significance, which includes the regulation of extracellular and intracellular volume and acid-base homeostasis. Chloride is the most abundant extracellular anion and accounts for about one-third of extracellular fluid tonicity (17). Intracellular chloride ranges from 20 to 40% of total body chloride (37) and is most concentrated within erythrocytes (70–80 mEq/l) (18). About 50% of total body chloride is in plasma, interstitium, and lymph, and the remaining percentage is in connective tissues, cartilage, and bone (42). Actually, it was observed over a century ago that one-third of the body’s chloride resides in the skin (226). The kidney is a major regulator of chloride homeostasis, and tubular reabsorption of chloride is critical for maintaining extracellular fluid volume (57, 69, 229). Unfortunately, comprehensive tubular physiology will not be discussed in this study and has been reviewed elsewhere (38, 44, 45, 119, 137, 204).

INTRAVENOUS CRYSTALLOID FLUID BACKGROUND

The first reports of using intravenous solutions for volume resuscitation were described in Great Britain in the early 1830s by Dr. William B. O’Shaughnessy in dogs (142) and Dr. Thomas Latta in humans (108) during the second cholera pandemic. The composition of intravenous fluids has changed significantly since then, including modifications by Dr. Sydney Ringer (177) and Dr. Alexis Hartmann (68). Dr. Hartog Jakob Hamburger, who is also known for describing the “chloride shift” phenomenon in erythrocytes, is credited with first suggesting that 0.92% saline solution is “normal” for mammalian blood (109). Further detail into this history can be found in these enlightening reviews (8, 10).

Intravenous fluids are the most commonly used “treatment” or “pharmacological agent” to treat hospitalized patients. Although little thought may enter the mind of some clinicians on the choice of fluid composition, it certainly stands to reason that, if certain fluid composition were to have even small effects that are more or less favorable on outcomes than other fluid composition, in theory, then this could impact millions upon millions of patients over time because of the frequency and widespread use. Thus a deep analysis on the association and effect of the composition of various intravenous fluid preparations is necessary and an important public health issue.

The choice of fluid depends on the optimal electrolyte concentration that would improve the underlying disorder (Table 1). Although normal saline is an effective intravenous fluid for a wide range of conditions (71), a growing concern regarding this solution is the abnormal chloride concentration at 154 mEq/l, which is 40–50% higher than plasma. Recommendations advising against excessive normal saline administration were first published over 100 yr ago (48) and reiterated in the 1970s (227) but were largely ignored until recently. Excessive use leads to sustained increases in serum chloride levels, which can be toxic to cells, imposes unnecessary demands on cellular energy metabolism (225), and contributes to excess morbidity and mortality (113). On the other hand, balanced solutions have lower chloride concentrations that are more reflective of normal physiological levels. The anion buffer to maintain electroneutrality in these fluids is commonly bicarbonate precursors, which prevent the development of hyperchloremic metabolic acidosis from large-volume resuscitation (184).

Table 1.

Composition of adult plasma and commonly used intravenous crystalloid fluids

	Plasma	Normal Saline	Lactated Ringer’s	Hartmann’s Solution	Plasma-Lyte
Na+, mEq/l	135–145	154	130	131	140
Cl−, mEq/l	96–108	154	109	112	98
K+, mEq/l	3.5–5.2	0	4	5	5
Ca2+, mg/dl	8.5–10.5	0	6	8	0
Mg2+, mg/dl	1.5–2.5	0	0	0	3.6
Buffer, mEq/l	Bicarbonate 22–30	0	Lactate 28	Lactate 28	Acetate and gluconate
Osmolarity, mosmol/l	285–295	308	273	279	295
Osmolality, mosmol/kgH2O	282–292	286	254	255	N/A

Balanced solutions should be avoided in patients with increased intracranial pressure or the syndrome of inappropriate antidiuretic hormone secretion because these fluids are generally hypotonic. Additionally, these solutions should not be used in patients with metabolic alkalosis or lactic acidosis with impaired lactate metabolism from liver failure or impaired lactate clearance from renal failure. Ultimately, the decision to choose an intravenous fluid and its rate of administration should be tailored to each patient’s needs.

Changes in Plasma Chloride Concentration from Intravenous Fluids

Chloride is largely restricted to the extracellular space (3, 80, 206, 243), which is 40% of total body water (TBW), unlike sodium that distributes over TBW (206). Intravascular volume accounts for 8% of TBW, and extravascular volume accounts for 32%. One liter of normal saline contains 154 mEq chloride and distributes 25% to the intravascular space and 75% to the interstitium. The average normal plasma chloride concentration is 104 mEq/l (28). If 25% of chloride and water from normal saline enters the intravascular space, 1 liter of normal saline would increase plasma chloride in a 70-kg man from 104.0 to 107.5 mEq/l and 104.0 to 109.1 in a 55-kg woman. Therefore, the predicted change in plasma chloride from an intravenous infusion of isotonic crystalloid fluid can be calculated from the equation

$$\begin{array}{l}\text{Predicted\hspace{0.17em}}\Delta \text{\hspace{0.17em}plasma\hspace{0.17em}}\left[{\text{Cl}}^{-}\right]=\\ \frac{\left(\text{initial\hspace{0.17em}plasma\hspace{0.17em}}\left[{\text{Cl}}^{-}\right]\times \text{plasma\hspace{0.17em}volume}\right)+0.25{\text{\hspace{0.17em}infusate\hspace{0.17em}Cl}}^{-}\text{\hspace{0.17em}content}}{\text{plasma\hspace{0.17em}volume}+0.25\text{\hspace{0.17em}infusate\hspace{0.17em}volume}}\end{array}$$

where

$$\text{plasma\hspace{0.17em}volume}=0.08\left(\text{\%water}\times \text{weight}\right)$$

In men, % water is 60% and in women, 50%; add 10% for lean individuals and subtract 10% for obese individuals.

This equation does not include additional fluid intakes and losses, nor does it account for Gibbs-Donnan equilibrium between the plasma and interstitium. Tissue buffering by erythrocytes, the kidney, and muscle will slightly decrease, whereas the acute hypoalbuminemia would slightly increase the chloride estimate. Additional calculations for men and women, balanced solutions, and a 2-l infusion are found in Table 2.

Table 2.

Calculated change in plasma chloride from 1- and 2-liter infusions of various crystalloid fluids for a 70-kg man and 55-kg woman

Infusion Volume	Calculated Change in Plasma Cl− in an Average Man, mEq/l	Calculated Change in Plasma Cl− in an Average Woman, mEq/l
Normal saline, 154 mEq/l
1 liter	104.0 to 107.5	104.0 to 109.1
2 liters	104.0 to 110.5	104.0 to 113.3
Hartmann’s solution, 112 mEq/l
1 liter	104.0 to 104.6	104.0 to 104.8
2 liters	104.0 to 105.0	104.0 to 105.5
Lactated Ringer’s, 109 mEq/l
1 liter	104.0 to 104.3	104.0 to 104.5
2 liters	104.0 to 104.7	104.0 to 104.9
Plasma-Lyte, 98 mEq/l
1 liter	104.0 to 103.6	104.0 to 103.4
2 liters	104.0 to 103.2	104.0 to 102.9

The rise in serum chloride after a 1- or 2-l infusion of various fluids over 1 h has been measured in several clinical studies of healthy volunteers and is summarized along with results from the SALT-ED trial in Table 3. Notably, in one of these studies, a 2-l normal saline bolus over 1 h decreased renal artery flow velocity and renal cortical tissue perfusion compared with Plasma-Lyte (31). Of note, there was no difference in urinary neutrophil gelatinase-associated lipocalcin, a biomarker of distal tubular injury, between groups.

Table 3.

Measured change in serum chloride from 1- and 2-liter infusions of various crystalloid fluids

Infusion Volume and Duration	Mean Change in Serum [Cl−], mEq/l	Comments, [Cl−] in mEq/l
Normal saline
1 liter over 1 h (115)	104 to 107.5 in the 1st hour followed by a gradual decrease of 1 over the next 6 h	• Gradually decreased by 1 over the next 6 h
2 liters over 1 h (172)	103.0 to ~107.5 in the 1st hour	• Gradually decreased to 106 over the next 5 h
2 liters over 1 h (116)	103 to 108 in the 1st hour	• Gradually decreased to 106 over the next 5 h
2 liters over 1 h (31)	103 to 108.5 in the 1st hour	• Gradually decreased to 106 over the next 3 h
2 liters over 25 min (40)	Peaked at 108.3 within the 1st 2.5 h postinfusion compared with 104.1 in control subjects, who did not receive any infusion	• Serum chloride remained elevated around 105 for up to 22 h postinfusion
		• Serum renin and aldosterone concentrations were significantly reduced for 2 days postinfusion
Median fluid volume administration of 1 liter (195)	103.1 to ~105.5 during slightly less than the 1st 24 h from ED arrival	• 6,639 SALT-ED trial participants
Hartmann’s solution
2 liters over 1 h (172)	102.5 to 104.5	• Cl remained around 104.5 for the next 3 h then decreased to around 103.5 for the next 2 h
		• Hartmann’s solution induced 1.7 times greater natriuresis compared with normal saline despite the lower sodium concentration of the fluid
		• After 6 h, 56% of the infused saline was retained compared with 30% of the Hartmann’s solution
Balanced solution
1 liter (195)	102.8 to a peak of ~104.5 at 24 h from ED arrival, slightly longer than the normal saline group	• Median fluid volume administration of 1 liter in the ED
		• 6,708 SALT-ED participants received balanced solutions
Plasma-Lyte
2 liters over 1 h (31)	103.5 to 104.5 over the 1st hour and remained at that level for the next 2 h	• Serum chloride increased despite the calculated decrease in Table 2

ED, emergency department.

RELEVANCE OF CHLORIDE ACROSS DIFFERENT CLINICAL SETTINGS

Hyperchloremia-associated AKI is thought to be related to renal vasoconstriction mediated by TGF and possibly other mechanisms, which will be thoroughly reviewed later in this article. Numerous studies show that changes in serum chloride concentration, independent of serum sodium and bicarbonate, are associated with increased risk of AKI, morbidity, and mortality. When evaluating the results, one should consider several confounding factors that could influence the association between chloride and AKI. Serum creatinine can be diluted from large-volume fluid administration, which could delay the diagnosis of AKI and lead to underestimation of AKI severity (120); positive fluid balance during AKI is independently associated with excess mortality (21, 56, 156, 191); present research does not explain whether the risk factor for AKI is the serum chloride concentration itself, the change in serum chloride concentration over time, the total quantity of chloride administered, or other factors; arterial chloride would have a greater influence on RBF and GFR than venous chloride. Most clinical trials analyze venous blood, which has a slightly lower chloride concentration than arterial blood (102 vs. 104 mEq/l) and may underrepresent the incidence of arterial hyperchloremia and lower its association with AKI. Intravenous fluids used for maintenance, replacement, or as a medication diluent contribute significantly to fluid volume intake. Normal saline is commonly used for this purpose and can be a large unrecognized source of chloride administration in the ICU. In fact, a large-scale retrospective study from a single-center ICU revealed that maintenance and replacement fluids contributed greater to daily chloride intake than resuscitation fluids (223). Interestingly, concentrated electrolytes, such as potassium chloride repletion, accounted for more than half of the daily chloride contribution provided by resuscitation fluids. Another single-center study showed that 3.5 liters of medication diluent were given during the first 7 days of ICU admission, which accounted for 63% of the total intravenous fluid volume given during that time (121). Patients who received >3 liters of normal saline as medication diluent had the greatest risk for developing hyperchloremia compared with 5% dextrose in water (D5W). Unadjusted analysis revealed the D5W group was associated with less AKI but was not statistically significant after adjusting for baseline covariates.

Overall, intravenous fluids are an important chloride exposure across numerous clinical settings, and care should be taken to minimize overexposure.

Large-Scale Clinical Trials

Several large clinical trials have been performed in critically ill patients. The first major clinical investigation was a pre-post clinical trial performed by Yunos and colleagues (250) in over 750 participants to investigate the effect of intravenous chloride concentration on renal outcomes. In this single-center study, after a 6-mo period of observation, a 6-mo phase-out period, and 6-mo intervention period, during which intravenous fluids were changed over to balanced solutions, the increase in serum creatinine level during ICU stay was correlated with log chloride intake in the 200 patients in which detailed data were obtained. The chloride-restrictive intravenous strategy intervention period was associated with a 50% decrease in the incidence of RIFLE-defined AKI and a decrease in renal replacement therapy (RRT) use (10% during the control period vs. 6.3% during the intervention period). These findings were confirmed in extended follow-up although sensitivity analyses revealed an increase in AKI incidence over time, suggesting that unidentified confounders may have contributed to the changes in the incidence of AKI (249). The initially impressive results were not achieved via a true randomized trial, as the prepost study design may have led to changes in other processes of care that resulted in the apparent improvement in renal function over time. Moreover, there were no differences in long-term RRT outcomes or nonrenal outcomes after the initial hospitalization period. Lastly, the trial included high-chloride (4% succinylated gelatin and 4% albumin) and low-chloride (20% albumin) colloids, which limited generalizability. Yunos and colleagues (251) also performed a before and after study of over 5,000 consecutive emergency department (ED)-treated hospital admissions that revealed that chloride restriction in the ED was associated with significantly lower rates of kidney disease, improving global outcomes (KDIGO) defined stage 3 AKI during the hospital admission but with no difference in rates of need for RRT.

The first high-quality randomized controlled trial (RCT), 0.9% Saline vs. Plasma-Lyte 148 for Intensive Care Unit Fluid Therapy (SPLIT), was a large multicenter double-blind cluster randomized, double-crossover clinical trial of 2,262 participants comparing normal saline to balanced solutions. Counter to the previous findings by Yunos and colleagues (251), SPLIT did not demonstrate reduced risk of AKI, rates of RRT, or in-hospital mortality with balanced solutions compared with normal saline. However, the study was limited by relatively modest fluid volume administration (median 2 l), the cohort was composed of mostly surgical patients (40% were cardiac surgery), participants were not severely ill (mean Acute Physiology and Chronic Health Evaluation II illness severity score of 14), and serum chloride levels were not reported (248). Additionally, over 90% of trial participants were exposed to intravenous fluids before study enrollment, which were largely balanced crystalloid. The overall incidence of AKI in the study was low (9%), the confidence intervals were wide, and thus these findings did not necessarily exclude a beneficial role for balanced fluids in higher-risk groups.

Most recently were the two largest randomized trials, the Isotonic Solutions and Major Adverse Renal Events Trial (SMART) (196) and the Saline Against Lactated Ringer’s or Plasma-Lyte in the Emergency Department (SALT-ED) trial that compared normal saline to balanced solutions (195). Both studies were conducted at the same institution and had low-volume administration (medians slightly over 1 l). SMART was a single-center, cluster-randomized, multiple-crossover trial of 15,802 critically ill adults from 5 ICUs. The participants who received balanced solutions had ~10% lower odds of the primary outcome, major adverse kidney events within 30 days (MAKE-30) (composite of death, new RRT, or persistent renal dysfunction). Although both study groups had similar hospital-free survival, subgroup analysis revealed the greatest benefit with balanced solutions was among those who received larger fluid volumes and those with sepsis.

SALT-ED had a similar study design and enrolled 13,347 participants who were treated in the ED and subsequently hospitalized outside an ICU. Although the primary outcome of hospital-free days was not significant between groups, balanced solutions resulted in 18% lower odds of MAKE-30 compared with normal saline. Additionally, those who presented to the ED with renal dysfunction or hyperchloremia received the largest benefit from balanced solutions for avoiding AKI and MAKE-30. Furthermore, among participants with KDIGO stage 2 AKI or higher at presentation, the use of balanced solutions was associated with greater AKI resolution during the hospitalization.

Although the effect sizes of SMART and SALT-ED trials were relatively modest (~10–20% reductions in AKI-related end points), because these were prevention trials, and the use of intravenous fluids is ubiquitous in the ED and ICU, the potential for the public health magnitude of a widespread culture shift to balanced chloride solutions would result in millions of fewer AKI episodes.

The Plasma-Lyte 148 versUS Saline Study (PLUS) is an ongoing multicenter blinded RCT across 50 ICUs studying the effect of Plasma-Lyte and normal saline on 90-day all-cause mortality, mean and peak creatinine concentration, and incidence of RRT (62). Additionally, ongoing is the pragmatic multicenter RCT Balanced Solution versus Saline in Intensive Care Study (BaSICS) comparing rapid infusion (999 ml/h) to slow infusion (333 ml/h) with either Plasma-Lyte 148 or normal saline on 90-day all-cause mortality, AKI incidence, and other important clinical renal outcomes (252). The results of these studies will likely have large clinical implications.

Surgery

The surgical literature contains the most clinical evidence comparing normal saline and balanced solutions on renal outcomes. A Cochrane review of perioperative normal saline vs. balanced solutions showed less hyperchloremia and metabolic acidosis with balanced solutions but no difference in rates of AKI and RRT (24). An update to this Cochrane review still identified insufficient evidence on the effects of perioperative fluid therapy and risk of postoperative renal failure, calling for larger and more robust trials (9). However, a meta-analysis of perioperative normal saline vs. balanced solutions demonstrated a 64% increased risk of AKI and longer intubation time with normal saline but no association with mortality (100).

Three randomized trials failed to show an effect of perioperative chloride administration on postoperative renal function. A small single-center single-blind randomized trial that compared intraoperative normal saline or a balanced solution during cardiac surgery revealed no difference in the kidney injury biomarker panel of urinary tissue inhibitor of metalloproteinase 2 (TIMP-2) and insulin-like growth factor-binding protein 7 (IGFBP7) between groups (110). A double-blind RCT of normal saline vs. lactated Ringer’s for perioperative fluid management during abdominal aortic aneurysm repair showed no difference in postoperative serum creatinine values. Of note, patients with baseline serum creatinine >1.4 mg/dl were excluded (233). Additionally, a clinical trial in cardiac surgery of perioperative administration of chloride-limited or chloride-rich fluids did not demonstrate any difference in rates of AKI (128)· However, this study had significant limitations because of an unforeseen institutional change in vancomycin exposure, which was lower in the chloride-rich group and may have confounded a beneficial effect in the chloride-limited group. Also, participants in the chloride-rich group still received balanced solutions and had a mean normal saline administration of 1.5 liters compared with 3 liters in the chloride-limited group.

Retrospective analyses of surgical patients demonstrate an association with normal saline or hyperchloremia and AKI and are summarized in Tables 4 and 5.

Table 4.

Retrospective studies comparing normal saline to balanced solutions across different clinical settings

Study Description	Intravenous Fluid Comparison	Key Results	Comments	Association With AKI
Surgery
Nationwide hospital database of patients undergoing either elective or emergency open abdominal surgery (197)	Normal saline (30,994 patients) vs. Plasma-Lyte (926 patients)	Normal saline associated with more postoperative infections, acute kidney infection (AKI) requiring renal replacement therapy (RRT), blood transfusions, and increased use of hospital resources compared with Plasma-Lyte	Patients who received normal saline were more likely to have undergone emergency rather than elective surgery, but rates of AKI remained significant even after adjustment	Yes
Organ transplantation
Single-center study of 158 patients undergoing liver transplantation (136)	Normal saline vs. lactated Ringer’s	>3.2 liters of normal saline administration within the 1st postoperative day was associated with AKI	These patients may have been more hypotensive and at increased risk for AKI, as blood pressure was not included in the analysis	Yes
Sepsis
Propensity matched cohort study using a multihospital electronic health record database of adults with systemic inflammatory response syndrome (SIRS) rather than overt sepsis (198)	Normal saline (1,558 patients) vs. balanced solution (1,558 patients)	Normal saline was not associated with greater rates in AKI	Normal saline was associated with greater in-hospital morbidity and more cardiac, infectious, and coagulopathy complications	No
Single-center cohort using a publicly available database of 8,085 intensive care unit (ICU) patients (253)	Normal saline vs. lactated Ringer’s	Lactated Ringer’s administration was associated with less AKI and reduced mortality during days 3 to 7 after ICU admission	The protective effect of lactated Ringer’s on AKI was more pronounced, as total fluid volume increased and was lost in subgroup analysis when <7 liters of fluid was infused	Yes
Propensity matched study of a nationwide cohort of septic ICU patients (170)	Normal saline (3,365 patients) vs. balanced solutions (3,365 patients)	No difference in AKI or need for RRT	Patients that received balanced solutions had a lower absolute mortality rate regardless of the total fluid volume received	No
Single-center cohort study of critically ill medical patients with severe sepsis or septic shock (72)	Normal saline (209 patients) vs. balanced solutions (201 patients)	Increased odds of AKI in the normal saline group	Both groups received large fluid volumes during the acute resuscitation period (median volume >6.5 liters)	Yes
			Median ICU length of stay was 2 days longer in the normal saline group but no difference in in-hospital length of stay or all cause in-hospital mortality

Table 5.

Retrospective studies evaluating the association between serum chloride concentration and AKI across different clinical settings

Study Description	Key Results	Comments	Association With AKI
Surgery
Single-center study in 726 patients who underwent craniotomy for a brain tumor resection (147)	Perioperative hyperchloremia was associated with postoperative AKI	39 patients with AKI	Yes
Organ transplantation
Single-center study of 213 donors and kidney allograft recipients (221)	No association between terminal serum chloride concentration in the deceased donor and kidney allograft function in the recipient up to 1 mo posttransplant	127 donors had hyperchloremia	No
		Dose of norepinephrine was higher in donors with hyperchloremia
Sepsis
Single-center cohort study of 240 patients with severe sepsis or septic shock (209)	Change in serum chloride concentration ≥5 mEq/l was associated with 5.7× higher odds of AKI after multivariable analysis	Excluded patients with chronic kidney disease (CKD)	Yes
	The average fluid volume administered per patient was 4.8 liters, and a greater increase in chloride was associated with more severe AKI	A trend was observed for the association between hyperchloremia and the need for RRT
	Patients without hyperchloremia but who had an increase in chloride ≥5 mEq/l showed 8.3× higher odds of developing AKI
Single-center study of 1,045 ICU patients with severe sepsis or septic shock (246)	No association of admission hyperchloremia, hypochloremia, or change in serum chloride with AKI within the 1st 72 h of admission	Median cumulative fluid balance in the 1st 72 h of ICU stay was slightly over 2 liters, and mean APACHE II score was ~13	No
Single-center study of baseline serum chloride levels in 834 septic patients (144)	More AKI in hypochloremic and hyperchloremic groups compared with normochloremic group	~1.5 liters mean daily infused normal saline volume in all groups	Yes
Single-center cohort study of 1,221 critically ill patients (258)	AKI was associated with mean and maximum serum chloride concentrations	AKI in 357 patients	Yes
		The most severe AKI was associated with a maximum serum chloride of 116 mEq/l
Contrast nephropathy
Single-center study of 401 patients admitted with ST-segment myocardial infarction undergoing percutaneous coronary intervention (154)	Baseline hyperchloremia was not associated with increased risk for postprocedure AKI	Did not report on IV fluid administration	No
Single-center analysis of 13,088 patients, mostly hospitalized, who underwent a contrast-enhanced abdominal computed tomography scan (143)	Increased risk of contrast-induced AKI in patients with baseline hypochloremia and normal serum creatinine (<1.2 mg/dl)	Low incidence of hyperchloremia in only 248 (1.8%) patients	Yes
	In those with hypochloremia, a positive change in serum chloride within 72 h after contrast exposure was associated with decreased AKI incidence
Congestive heart failure
Analysis of 2 cohorts with 1,318 and 876 patients admitted for acute decompensated congestive heart failure (CHF) (58)	Admission serum chloride levels <99 compared with >103 mEq/l were associated with slightly higher admission creatinine (1.3 vs. 1.2 mg/dl) in 1 of the 2 cohorts	Serum chloride tertiles were also inversely associated with mortality, systolic blood pressure, and blood urea nitrogen	Yes
Post hoc analysis of a double-blind randomized controlled trial (RCT) of 2,699 chronic stable CHF patients (216)	Hypochloremia was associated with higher baseline serum creatinine, lower glomerular filtration rate (GFR) and a greater percentage of patients with eGFR <60 ml·min−1·1.73 m−2	Interestingly, no association was seen between hyponatremia and renal function	Yes
Post hoc analysis of an RCT of 1,960 patients hospitalized with acute decompensated CHF (212)	No association between baseline chloride and subsequent worsening renal function during decongestion	279 patients had hypochloremia Lower serum chloride concentration was associated with poorer diuretic response	No
Single-center analysis of 277 patients diagnosed with pulmonary arterial hypertension (134)	Serum chloride ≤100 mEq/l 6 mo after diagnosis was associated with higher mortality and worse renal function than those with a serum chloride >100 mEq/l	The lower chloride group was older, with decreased functional capacity, took more diuretics, had higher pulmonary artery wedge pressure but lower mean pulmonary artery pressure, transpulmonary gradient, and pulmonary vascular resistance	Yes

Kidney Transplantation

Renal vasoconstriction is of clinical relevance in kidney transplantation, especially among deceased donor transplant recipients who are at higher risk for delayed graft function (DGF). In this setting, avoidance of hyperchloremia may help maintain RBF and GFR. A Cochrane review of adult kidney transplant recipients comparing perioperative normal saline to balanced solutions found less hyperchloremic metabolic acidosis but no difference in rates of hyperkalemia or DGF with balanced solutions. However, data lack regarding important transplant outcomes, such as graft survival and acute rejection (231). In an RCT assessing perioperative fluid administration in living donor kidney transplant recipients, the normal saline group had higher postoperative serum chloride, lower blood pH and serum bicarbonate, but no difference in serum creatinine, urine output, or graft failure compared with Plasma-Lyte (89). Additionally, a prospective randomized double-blind trial of mostly living donor kidney transplant recipients did not demonstrate a difference in serum creatinine concentration on postoperative day 3 when comparing normal saline to lactated Ringer’s solution. This study was terminated early for safety reasons, as 5 of 26 patients who received normal saline developed hyperkalemia (K >6 mEq/l), whereas hyperkalemia did not occur in the lactated Ringer’s group despite the 4 mEq/l K concentration of the fluid. Additionally, eight patients in the normal saline group developed metabolic acidosis requiring treatment compared with none in the lactated Ringer’s group (140). These observations suggest that the risk of hyperkalemia from potassium-containing balanced solutions is low during uncomplicated kidney transplant surgery and that hyperkalemia is likely secondary to metabolic acidosis. A prospective randomized trial comparing normal saline to an acetate-buffered crystalloid in deceased donor kidney transplant recipients did not demonstrate a difference in postoperative creatinine or urine output over 1 wk (167). Of note, significantly more patients in the normal saline group needed intraoperative vasopressors.

Additional research is needed, and the Better Evidence for Selecting Transplant Fluids (BEST-Fluids) is a multicenter RCT comparing the impact of normal saline to Plasma-Lyte use before, during, and after deceased donor kidney transplant on key kidney transplant outcomes (7).

Sepsis and the ICU

Hyperchloremia is common among patients with sepsis (49), and, despite limited evidence, the most recent Surviving Sepsis Campaign guideline recommends the avoidance of hyperchloremia during treatment (176). Septic AKI may be due in part to tubular dysfunction leading to increased distal chloride delivery to the macula densa, which would stimulate TGF to reduce RBF and GFR. Mouse models of AKI demonstrate downregulation of the apical chloride cotransporter NKCC2 and the basolateral chloride channels CLC-K1 and K2 (in addition to nonchloride channels and transporters), which increases chloride delivery to the macula densa and also contributes to sepsis-mediated salt wasting (130). It is presently unknown whether NKCC2 in the macula densa is also downregulated during sepsis, which could alter the TGF response.

A study of metabolic acidosis in dogs with experimental endotoxemic sepsis demonstrated that volume resuscitation with normal saline accounted for 42% of the acid load, whereas 52% of the acid load was postulated to be from sepsis-induced hyperchloremia mediated by disproportionate shifts in sodium and chloride, and the remaining 6% came from lactate (86). In a rat model of abdominal sepsis, hyperchloremic acidosis induced by hydrochloric acid infusion increased proinflammatory cytokines IL-6, IL-10, and TNF-α (87). Sheep with abdominal sepsis given normal saline were the quickest to develop oliguria, had more cardiac and microcirculation dysfunction, and had greater hemodynamic instability compared with animals that received either lactated Ringer’s or Plasma-Lyte (153). A short-term study in septic rats revealed more metabolic acidosis and lower urine output with normal saline infusion despite similar renal hemodynamics and creatinine clearance compared with Plasma-Lyte (152). Lastly, a rat model of endotoxemic shock comparing normal saline to a novel balanced solution showed attenuation of oxidative and nitrosative stress markers but did not demonstrate differences in plasma creatinine levels and urine output or improve cortical and medullary microvascular renal oxygenation (46).

A recent systematic review and meta-analysis of critically ill patients from six RCTs concluded that balanced solutions and normal saline had no significant difference on AKI, new RRT, in-hospital mortality, and overall ICU mortality (255). Retrospective studies in sepsis have mixed results and are summarized in Tables 4 and 5.

Sepsis is a common condition that often warrants large-volume resuscitation. Therefore, further prospective randomized trials should be performed to evaluate a role of chloride in mediating septic AKI. The Fluids in Sepsis and Septic Shock (FISSH) trial is an on-going multicenter pilot RCT investigating the feasibility of a large-scale trial comparing lactated Ringer’s with normal saline in patients with septic shock (179). Secondary outcomes include mortality, receipt of RRT, ICU and hospital lengths of stay and surrogate outcomes of incidence of acidosis, hyperkalemia, and AKI.

Diabetic Ketoacidosis

AKI secondary to profound volume depletion is commonly encountered with diabetic ketoacidosis, often necessitating large-volume resuscitation. Urinary losses of ketone anions (potential bicarbonate) contribute to impaired normalization of serum bicarbonate upon resolution of diabetic ketoacidosis and lead to a persistent hyperchloremic metabolic acidosis (146). A prospective randomized double-blind study showed lower postresuscitation serum chloride and higher serum bicarbonate levels with Plasma-Lyte A compared with normal saline (122). A retrospective analysis of 23 critically ill patients revealed more hyperchloremia, delayed recovery of acidosis, and lower initial urine output from large-volume resuscitation with normal saline (median 4.4 liters administered at 12 h from baseline) compared with Plasma-Lyte (median 3.1 liters administered at 12 h from baseline) but no difference in serum creatinine decline (32). Further prospective research is required in this patient population to study the effects of different fluids on renal and metabolic outcomes.

Contrast Nephropathy

Intravenous fluids are frequently administered to prevent contrast-induced AKI (CIAKI). Iodinated intravenous contrast media causes kidney injury through multiple mechanisms, including afferent vasoconstriction mediated by TGF. In an isolated perfused rat juxtaglomerular apparatus, iodixanol caused direct tubular cell damage and increased oxidative stress, which enhanced the TGF with greater afferent vasoconstriction and a decrease in medullary perfusion pressure compared with control perfusate (112).

Clinical evidence does not support the role of chloride in CIAKI, as isotonic solution is better than half-isotonic solution in preventing contrast-induced AKI after percutaneous coronary intervention (132). These findings may be related to fluid tonicity rather than the chloride concentration itself. Literature on the prevention of CIAKI with normal saline or sodium bicarbonate is controversial without a clear benefit from the avoidance of chloride (208). A recent randomized trial comparing sodium chloride to sodium bicarbonate had equal effect on preventing CIAKI, persistent decline in renal function at 90 days, need for RRT, or death (234). Two retrospective studies have been conducted regarding contrast nephropathy and chloride and are summarized in Table 5.

Evidence comparing normal saline to balanced solutions for the prevention of contrast nephropathy is lacking. An ongoing randomized multicenter trial, Balanced Salt Solution Vs. 0.9% Saline Infusion for Prevention of Contrast-Induced Acute Kidney Injury (BASIC Trial), seeks to study the rates of contrast-induced AKI and other renal outcomes with normal saline compared with a balanced solution when administered at 3 ml/kg for 1 h before contrast then 1.5 ml·kg⁻¹·h⁻¹ for 4 h after contrast in patients at high risk for CIAKI (76).

Congestive Heart Failure

The conditions reviewed to this point have largely focused on hyperchloremia. On the other hand, congestive heart failure (CHF) is commonly associated with hypochloremia. Dilutional hypochloremia from neurohormonal activation and nonosmotic release of vasopressin and chloride depletion from loop diuretics both contribute. In rats, hypochloremia causes TGF-mediated reductions in GFR independent of extracellular volume (52). Hypochloremia activates the renin-angiotensin-aldosterone system (RAAS) despite the significant volume expansion with CHF. In fact, hypochloremia and lower serum chloride levels, even without overt hypochloremia, are associated with increased mortality in the general population (36) and among patients with hypertension (126), CKD (123), and CHF (58, 60, 216). In those with CHF, hypochloremia is also associated with adverse renal outcomes. It has been hypothesized that changes in the serum chloride concentration primarily influence changes in the plasma volume by regulating vasopressin and RAAS activity during worsening CHF (83).

Retrospective studies, including acute and chronic CHF and pulmonary hypertension, are summarized in Table 5. One notable study was a post hoc analysis from the Renal Optimization Strategies Evaluation in Acute Heart Failure (ROSE-AHF) trial of patients hospitalized for acute CHF with renal dysfunction, which demonstrated that a change in chloride during decongestive therapy was inversely related to serum cystatin C, whereas baseline serum chloride was not associated with change in cystatin C during decongestive therapy at 72 h (59). Additionally, lower chloride levels were associated with decreased diuretic responsiveness.

Similar findings were found in a prospective observational cohort study and an interventional pilot study of patients with chronic, stable CHF (63). Hypochloremia was associated with higher plasma renin activity and greater loop diuretic resistance than those without hypochloremia. Chloride repletion with lysine chloride (21 g/day for 3 days) along with low-sodium diet were associated with a mean increase in serum chloride levels of 2.2 mEq/l, and the majority of participants experienced findings such as hemoconcentration, weight loss, reduction in NT-pro B-type natriuretic peptide, increased plasma renin activity, and increased blood urea nitrogen:creatinine ratio. These data reflect the notion that hypochloremia may, not only be associated with poor outcomes in CHF, but also be causal and provide a modifiable treatment target in patients with CHF. An ongoing trial seeks to elicit the therapeutic potential of sodium-free chloride supplementation in patients with acute decompensated CHF (www.clinicaltrials.gov, identifier NCT03446651).

THE REGULATION OF RBF AND GFR BY CHLORIDE

Extracellular chloride has been shown to be essential for the constriction of isolated rodent afferent arterioles and cultured mesangial cells and juxtaglomerular cells. Studies have demonstrated vasoconstrictor responses that are highly sensitive to changes in extracellular chloride within the physiological range (65). Sustained vessel or cell contraction has been shown to be dependent on the influx of extracellular calcium driven by a G protein-coupled receptor-initiated increase in intracellular calcium and activation of calcium-activated chloride channels. Chloride efflux-mediated cell depolarization activates voltage-dependent calcium channels, leading to additional calcium influx and robust cell contraction (27) (Fig. 1). Furthermore, changes in extracellular chloride concentration influence the response to vasoconstrictors.

Fig. 1.

Left: with reduced perfusion pressure, a low macula densa chloride concentration stimulates the release of renin and local prostaglandins, which serve to maintain glomerular filtration rate (GFR) by afferent vasodilatation and efferent vasoconstriction. Right: excess chloride from normal saline and/or hyperchloremia increases distal chloride delivery to the macula densa, signaling a decrease in GFR from afferent vasoconstriction, mediated by thromboxane and adenosine, and attenuation of efferent vasoconstriction from reduced renin and angiotensin II levels. In the macula densa, influx of chloride via NKCC2 increases intracellular chloride and subsequent basolateral chloride exit with ATP. In, juxtaglomerular (JG) cells, angiotensin II increases intracellular calcium, which triggers chloride efflux through calcium-activated chloride channels, leading to cell depolarization and inhibition of renin release. Calcium-activated chloride channels are also present in mesangial cells and afferent arteriole smooth muscle cells (AASMC), where they are under the influence of multiple vasoconstrictors acting through G protein-coupled receptors. However, unlike in JG cells, chloride efflux-mediated cell depolarization triggers calcium influx through voltage-dependent calcium channels, leading to robust cell contraction. COX2, cyclooxygenase 2; PGE₂, prostaglandin E₂. [Printed with permission from Mount Sinai Health System.]

Afferent Arteriole

Angiotensin II (ANG II) (27, 51, 73, 74, 211) and endothelin (210) have been shown to increase chloride conductance and vasoconstriction of the afferent arteriole. Additionally, the extracellular to intracellular chloride gradient was shown to be crucial in determining the afferent arteriolar responsiveness. However, some studies demonstrated conflicting results, possibly dependent on the experimental model (65, 169, 211). ATP released from the macula densa with chloride is metabolized to adenosine, which activates adenosine A1 receptors, leading to afferent arteriole and mesangial cell contraction, leading to a reduction in GFR. Intriguingly, adenosine similarly affects afferent arteriole vasoconstriction through an increase in chloride and calcium conductance (64). It is tempting to speculate that chloride may influence the vasoconstrictive response to adenosine from TGF and may be involved in hyperchloremia-associated AKI.

Mesangial Cells

Mesangial cells are important in regulating RBF and GFR (186), and in vitro evidence suggests that these cells can be manipulated by chloride. Mesangial cell chloride permeability is enhanced by activation of calcium-activated chloride channels causing membrane depolarization and cell contraction in response to ATP (155), ANG II, and vasopressin (102, 103, 149–151). On the contrary, mesangial cells have been shown to produce prostaglandin E₂ (PGE₂) in response to low extracellular chloride, which attenuated ANG II and vasopressin-mediated cell contractility (150). Additionally, low extracellular chloride increased basal intracellular calcium in cultured rat mesangial cells, which decreased the response to vasopressor stimulation with either endothelin-1 or vasopressin (207).

Changes in plasma or mesangial interstitial chloride concentrations modulate the chloride gradient across the cell membrane, influencing the activity of chloride channels in mesangial cells and afferent arteriole smooth muscle cells and thus cell contraction. Accordingly, the interstitial chloride concentration of the glomerulus, as determined by basolateral exit from the macula densa, could directly determine vascular tone reactivity. On the other hand, changes in blood chloride concentration may be influential. Conceivably, hyperchloremia may raise the overall renal interstitial chloride concentration, including the glomerular mesangium, leading to afferent vasoconstriction independent of chloride-mediated changes in TGF.

Juxtaglomerular Cells

Juxtaglomerular cells are the primary cells responsible for secreting renin. Calcium-activated chloride channels have also been shown to mediate chloride efflux in isolated juxtaglomerular cells to inhibit renin release (75, 104) in response to ANG II (135) and endothelins (178).

Macula Densa and TGF

TGF regulates renal hemodynamics based on the tubular fluid composition at the macula densa. Although controversial, evidence supports a role for chloride in the macula densa-sensing mechanism, renin release, and regulation of GFR (Fig. 2). Schnermann and colleagues identified a role for chloride in regulating TGF (189) and renin secretion (118) independent of sodium. Although this effect was further studied and replicated by themselves and others (52, 53, 93–95, 180), some investigations dispute these findings (13, 15, 16). NKCC2 is present on the apical membrane of the macula densa (106, 185), and its activity is determined by luminal chloride (107). Because chloride is rate limiting for NKCC2, luminal chloride serves as the sensing mechanism of TGF (43, 97). Increases in luminal chloride at the macula densa increase intracellular chloride via NKCC2 activity (181, 182). Basolateral chloride efflux from the macula densa into the mesangial interstitium occurs during TGF signaling along with ATP via maxi anion channels (12), ultimately leading to mesangial cell and afferent arteriole contraction. In microperfused Amphiuma tubules, the interstitial chloride concentration of the juxtaglomerular apparatus (JGA), as measured by microelectrode, was 159 mM without tubular perfusion but increased to 645 mM during perfusion (159). This hypertonic state was unique to the JGA, as the interstitial chloride outside the JGA was similar to plasma. However, it is unknown whether chloride serves to regulate TGF in humans in this fashion.

Fig. 2.

Representation of the effect of macula densa chloride concentration on tubuloglomerular feedback (TGF) and single-nephron glomerular filtration rate (SNGFR). A leftward shift from the gray to black curve occurs during an increase in TGF responsiveness, such as during hypovolemia. SNGFR would be further depressed for the same macula densa chloride concentration (22, 187, 218).

Animal studies have demonstrated that changes in extracellular chloride can regulate RBF and GFR. This effect was demonstrated by Wilcox and coworkers (238) through several experiments during the 1980s with autotransplanted kidneys in greyhound dogs. Vasoconstriction was induced by intrarenal infusion of hyperosmotic sodium chloride or ammonium chloride, whereas vasodilation occurred with hyperosmotic dextrose, sodium bicarbonate, and sodium acetate infusions. Salt deprivation further exacerbated chloride-induced renal vasoconstriction, suggesting an influential role for baseline RAAS tone. Subsequently, RBF was found to be regulated by a balance between chloride-mediated synthesis of renal vasodilating PGE₂ and vasoconstricting thromboxane (241). However, another study did not show a significant role for thromboxane (6). Further studies demonstrated decreased renin release and reduced lymphatic ANG II levels during hyperchloremia (240). Additionally, efferent vasoconstriction was attenuated during hyperchloremia and intrarenal renin, and ANG II levels were lower than during normochloremia (239). The renal vasoconstriction response and decrease in GFR induced by hyperosmotic chloride infusion in rats were significantly blunted with inhibition of either cyclooxygenase (COX) or thromboxane synthetase (23). Hyperchloremic dogs given indomethacin and captopril demonstrated chloride-induced ANG II release into lymph and a chloride-induced reduction in GFR without a change in renal vascular tone (239). Although tempting to apply these findings to clinical practice, the use of hyperosmotic solutions in these studies limits translation to normal in vivo conditions.

TGF ultimately involves a complex interplay of additional factors, such as luminal osmolality (139), cell volume (91), cytosolic calcium (14, 161, 162, 175), tubular flow (203), ATP (12, 92, 174), adenosine (88, 173), nitric oxide (29, 98, 111), ANG II (99, 163, 232, 256), atrial natriuretic peptide (70), dopamine (257), and other autocrine and paracrine factors. Greater detail of other regulatory pathways in TGF are discussed in these reviews (11, 26, 50, 138, 164, 188, 205, 217).

COX-2

COX-2 metabolites of arachidonic acid are essential mediators of RBF and renal salt handing (66, 67). At the macula densa, luminal chloride regulates COX-2, such that low extracellular chloride increases COX-2 expression (30) and PGE₂ release (165, 245) to activate juxtaglomerular cells and stimulate renin secretion. Interestingly, indomethacin was shown to have opposing effects in isolated perfused rat kidneys, which was dependent on the chloride concentration of the perfusate. During COX blockade, GFR decreased with 87 mEq/l chloride solution, whereas GFR increased with 117 mEq/l chloride solution (247). 20-Hydroxyeicosatetraenoic acid (20-HETE), a cytochrome p450 metabolite of arachidonic acid, was shown to cause afferent vasoconstriction by inhibition of calcium/stretch-activated BK (maxi K) channels in the vascular smooth muscle (260) and to modulate the myogenic response of the afferent arteriole to TGF (54). Another study suggested that the renal response to hyperchloremia may in fact be mediated by a COX-2-dependent 20-HETE metabolite (6). Overall, chloride influences COX-2 and is likely involved in hyperchloremia-associated AKI.

HYPERCHLOREMIA-ASSOCIATED AKI

The TGF mechanism is a compromise between the ability to increase GFR at the expense of intravascular volume and the ability to conserve volume at the expense of GFR (219). Inappropriate increases in GFR could lead to excessive loss of extracellular fluid volume and hypotension if TGF were not intact. Reduced renal artery pressure increases the sensitivity of the TGF mechanism and contributes to prolonged reductions in GFR (84, 194). Changes in extracellular volume alter the TGF response curve such that the TGF response is enhanced during volume depletion and blunted during volume expansion (129). Injury to the proximal tubule (222) or loop of Henle (124, 160) during AKI impairs solute reabsorption and increases distal solute delivery. TGF-mediated afferent arteriole vasoconstriction is a common hemodynamic insult in prerenal, ischemic, nephrotoxic, and obstructive AKI models (202). TGF may be intact (125) or impaired (2) during experimental AKI, and, in some cases, tubular injury and GFR can dissociate from normal autoregulation, contributing to excess urinary losses or worse tubular injury (19). During AKI, volume loss is minimized by a TGF-mediated decrease in GFR and oliguria in a phenomenon referred to as “acute renal success” (220). In the setting of increased TGF responsiveness, such as in hypotension, a decline in single-nephron glomerular filtration rate (SNGFR) can occur (22). A full TGF response occurs at a tubular lumen concentration of 60 mEq/l sodium chloride. Between 25 and 45 mEq/l sodium chloride, the range of normal sodium chloride concentrations at the macula densa, a change of 1 mEq/l leads to a change in SNGFR of ~0.5 nl/min (187). A normal adult kidney has about 1,000,000 nephrons, so an increase of 1 mEq/l in 2,000,000 nephrons will decrease GFR by 1 ml/min. An increase in distal chloride from 30 to 60 mEq/l, which has been previously reported under experimental conditions after a normal saline infusion (235), could theoretically decrease GFR by 30 ml/min.

Prolonged afferent arteriole vasoconstriction can lead to ischemia and additional tubular injury (148), hindering the reabsorption of solutes, including chloride. Despite renal autoregulation, the kidney is at risk for ischemic AKI if there is an inadequate supply of oxygen and metabolism to meet the demand of salt reabsorption and oxygen consumption (20). Ischemic AKI is typically associated with low systemic perfusion without overt hypotension, includes prerenal azotemia and acute tubular necrosis, and accounts for more than half of renal failure cases among hospitalized patients (1). Renal autoregulation serves to maintain normal blood flow during this decrease in perfusion pressure. However, as renal perfusion pressure decreases below autoregulatory range, vasoconstrictors increase afferent arteriole resistance, which decreases postglomerular capillary blood flow to the tubules. Increased luminal chloride sensed by the macula densa would further vasoconstrict the afferent arteriole. Elderly patients or those with atherosclerosis, hypertension, or CKD with arteriolar narrowing contributing to impaired autoregulation are particularly susceptible to renal hypoperfusion and ischemic AKI (1). Additionally, afferent arteriole vasodilation leading to glomerular hyperfiltration in patients with CKD from diabetes or obesity would limit the autoregulatory ability to further vasodilate the afferent arteriole, making the kidney susceptible to ischemic AKI during renal hypoperfusion (1).

Although not studied clinically, hyperchloremia-associated AKI may be most pronounced in states of high RAAS activity. Volume expansion with intravenous fluids has been shown to inhibit proximal tubule sodium chloride reabsorption (35, 171), leading to increased distal chloride delivery (41, 235). Additionally, acute volume expansion in rats with an isotonic chloride solution resulted in greater TGF inhibition compared with several nonchloride solutions (235). Chloride in intravenous solutions has been shown to influence renin release. In fact, several studies have demonstrated that infusion of chloride salt solutions suppressed renin in salt-restricted rats (90, 117) and humans (77), whereas sodium bicarbonate had no effect. Additional details regarding this mechanism have been explained by Kotchen and associates (96). Volume resuscitation with normal saline clearly raises the serum chloride level to a greater extent than balanced solutions, which would increase the tubular chloride concentration at the macula densa, inhibit renin secretion, signal afferent arteriole vasoconstriction, attenuate efferent arteriole vasoconstriction, and decrease glomerular hydrostatic pressure, translating to decrease in GFR (Fig. 1). In other words, normal saline may inappropriately inhibit TGF in the setting of renal hypoperfusion. Hyperchloremia may also influence both the afferent arteriole vasoreactivity to agonists and the balance between vasodilating PGE₂ and vasoconstrictive thromboxane synthesis. Therefore, concomitant postresuscitation hyperchloremia and AKI would maximize distal chloride delivery. TGF-mediated afferent vasoconstriction in addition to the direct effect of the hyperchloremia on contraction of afferent arteriole smooth muscle cells and mesangial cells would decrease RBF and further contribute to AKI. Additionally, simultaneous afferent arteriole vasoconstriction from the administration of excess chloride and venous congestion from excess volume administration may contribute to further decreases in RBF and thus AKI (39).

Clinically, patients with AKI may have an impaired ability to concentrate (5) and acidify urine (242). Both of these can increase serum chloride and predispose to hyperchloremia, which may plausibly exacerbate the existing AKI or impede its recovery. In critically ill patients with metabolic acidosis, decreased renal function was determined to be related to inadequate urinary excretion of chloride (131). Accordingly, chloride loading by large-volume resuscitation may contribute to pronounced acidosis in these patients.

The effect of hyperchloremia on GFR is analogous to giving a COX antagonist, angiotensin-converting enzyme inhibitor, or ANG II receptor blocker during hypovolemia and renin-dependent renal perfusion. The effects of chloride on RBF and TGF may also share similarities to the disease pathology of salt-sensitive hypertension. With regard to AKI, it is possible that salt sensitivity predisposes to the development of hyperchloremia-associated AKI, as these individuals may have an augmented response to excess distal chloride delivery, causing exaggerated afferent vasoconstriction. Experimental data suggest that the renal dysfunction of spontaneously hypertensive rats may be due to impaired PGE₂ production by mesangial cells in response to low extracellular chloride (149). Interestingly, skin chloride accumulation was associated with the development of immune cell-mediated salt-sensitive hypertension in mice (237). The significance of chloride in blood pressure regulation is increasingly recognized and can be found in these thought-provoking reviews (127, 228, 230).

Overall, this section provides rationale for a possible mechanism explaining the association seen in clinical studies between chloride and AKI.

HOW CHLORIDE CAN HELP DIAGNOSE AKI

The KDIGO AKI Workgroup definition and staging system are the preferred definition for AKI, which is defined as an increase in serum creatinine by ≥0.3 mg/dl within 48 h, or increase in serum creatinine to ≥1.5 times baseline, which is known or presumed to have occurred within the prior 7 days, or urine volume <0.5 ml·kg⁻¹·h⁻¹ for 6 h (85). AKI is then subdivided into stage 1, stage 2, and stage 3 based on severity.

The AKI staging does not distinguish between various etiologies of AKI, which is often important in caring for patients, as patients with prerenal forms of AKI often need volume repletion, and those with intrinsic AKI will not benefit as much from volume administration. The two classic clinical assessments that nephrologists use to distinguish between prerenal vs. intrinsic AKI are examination of the urine sediment for casts and cell types and assessment of urine electrolytes. Renal salt output is directly determined by effective vascular volume (145, 193), and the quick, easy, and inexpensive measurement of urinary electrolytes can assist in the diagnosis of AKI. The most common parameters used for assessment to distinguish between prerenal and intrinsic AKI are urine sodium (U_Na) and its fractional excretion (FE_Na) (47). However, false-positives can occur when the FE_Na is <1% despite true kidney injury, which can be seen with urinary obstruction, glomerular disease, or other sodium-avid states (254), or false-negatives when the FE_Na is high with concurrent metabolic alkalosis with bicarbonaturia (259). One of the major reasons for reduced FE_Na utility relates to the use of this test in disease states not included in the initial studies. Such diseases disturb the typical tubular response to sodium reabsorption and make Fe_Na poorly reflective of the actual cause of AKI (158).

In addition to U_Na, urinary chloride (U_Cl) excretion is a sensitive marker of volume depletion (224). Traditionally, U_Cl has been used to distinguish between saline-responsive and saline-resistant forms of metabolic alkalosis (192). A low spot U_Cl can also assist in diagnosing prerenal AKI (79, 190, 201). However, conditions that cause a high U_Cl concentration and would decrease the specificity for detecting prerenal AKI include mineralocorticoid excess, severe potassium depletion, metabolic acidosis not attributable to distal renal tubular acidosis, diuretic use, and advanced CKD.

Few studies have examined the test characteristics of U_Na vs. U_Cl. One study revealed spot U_Cl concentration to be more sensitive to prerenal azotemia than spot U_Na [20 of 21 patients (95%) vs. 13 of 21 patients (62%)]. Specificity was high for both chloride (87%) and sodium (91%) (4). Interestingly, 7 of 21 (33%) patients with prerenal azotemia had a U_Cl <20 mEq/l and U_Na >20 mEq/l. These patients did not have acid/base disturbances and had a lower urine:plasma creatinine ratio (43 vs. 88), higher FE_Na (0.72% vs. 0.28%), and blood urea nitrogen (74 mg/dl vs. 53 mg/dl), suggestive of more severe AKI. Additionally, U_Na was elevated (mean 65 mEq/l) among all eight patients with concomitant gastric alkalosis and contributed to decreased recognition of prerenal AKI.

Another study demonstrated similar sensitivity of U_Na and U_Cl during various states of renal salt retention (U_Na and/or U_Cl <10 mEq/l), including hypovolemia (200). U_Na exceeded U_Cl by at least 15 mEq/l in 16 of 110 episodes (15%). This disparity was associated with obligate excretion of poorly reabsorbed anions, a decreasing serum bicarbonate level (obligate sodium excretion with bicarbonaturia), or severe renal insufficiency. In 14 of 110 episodes, U_Cl exceeded U_Na by at least 15 mEq/l (13%). These patients were more often oliguric and had a higher mean serum chloride than patients without a discrepancy in urinary electrolytes.

U_Cl alone is at least as accurate as U_Na for identifying prerenal AKI. However, measurement of U_Na or U_Cl alone will fail to detect renal salt retention in a substantial number of cases because of urinary excretion of excess nonchloride anions or nonsodium cations. Despite the limitations of urinary electrolytes, to maximize diagnostic accuracy, both U_Na and U_Cl should be analyzed together, and a low concentration of either is indicative of a salt-retaining state.

ADDITIONAL CONTRIBUTION OF ANEMIA TO HYPERCHLOREMIA-ASSOCIATED AKI

As mentioned earlier, erythrocytes contain the highest intracellular chloride concentration in the body (70–80 mEq/l) (18). This high concentration maximizes the bicarbonate-carrying capacity of the venous circulation to balance pH changes from carbon dioxide transport in a phenomenon called the Hamburger phenomenon or “chloride shift” (236). Carbon dioxide enters the erythrocyte and is converted to bicarbonate, which leaves the cell through the anion exchanger AE1 (band 3) in exchange for chloride.

Metabolic acidosis decreases oxygen affinity to hemoglobin, decreases cardiac contractility and cardiac output, and increases resistance to vasopressor agents (101). The kidney is particularly susceptible to hypoxic injury (81), and excess chloride administration in the setting of anemia may therefore exacerbate renal injury. Hyperchloremia induced by normal saline in young healthy cows showed decreased oxygen binding to hemoglobin (25). Additionally, among healthy volunteers, chloride was shown to compete with oxygen for binding to hemoglobin at physiologically relevant changes in Po_2. Chloride also induced a conformational change in hemoglobin and decreased its affinity for oxygen (168). Because acidemia decreases hemoglobin affinity for oxygen and hyperchloremia often occurs simultaneously with metabolic acidosis, these findings suggest that hyperchloremia may be a second hit during states of low tissue oxygen delivery.

Clinical evidence supports an association with anemia and increased risk of AKI (61) and need for RRT (82). Additionally, excess saline administration can also cause a dilutional anemia, and these patients may also be on vasopressors, which can worsen renal medullary perfusion and oxygenation independent of systemic blood pressure (105). Anemia may act as a second hit, and future investigation in various patient populations should evaluate the risk of AKI in patients with concomitant hyperchloremia and anemia.

WNK KINASES

The with-no-lysine (WNK) kinases are master regulators of ion transport and cell volume throughout the body, including the kidney, are implicated in Mendelian forms of hypertension [familial hyperkalemia and hypertension (FHHt)], and have attracted great attention since their discovery in 2001 (199). Historically, FHHt has been described as a “chloride shunt” phenomenon because of the increase in distal chloride reabsorption (183). Intracellular chloride competitively blocks autophosphorylation and activation of the WNK kinases (166, 214, 215), including control over NCC (244), NKCC2 (78, 213), and other cation-chloride cotransporters and chloride channels. The WNK1 and WNK4 isoforms are present within the mouse macula densa (141) although their role here is presently unknown. Because macula densa cells experience wide variations in intracellular chloride, it is plausible that multiple WNK isoforms may be involved in macula densa cell signaling. Interestingly, WNK4 is sensitive to changes in intracellular chloride at levels found during basal conditions in the macula densa and distal convoluted tubule (0–40 mEq/l) (214). During activation of TGF, macula densa intracellular chloride can rise above the threshold for WNK1 inhibition (60 mEq/l) (214). The role of WNK kinases in either TGF or AKI is largely unknown, and future investigation of these chloride-sensitive kinases is of great interest.

CONCLUSIONS

Chloride is underappreciated in both physiology and pathophysiology. Animal studies and recent clinical trials and observations suggest that chloride may influence AKI through adverse effects on RBF and GFR. Although the exact mechanism is debated, avoidance of excessive chloride administration is a reasonable treatment option for all patients and especially in those at risk for AKI. Therefore, high-risk patients and those with “incipient” AKI should receive balanced solutions rather than normal saline to minimize the risk of AKI (157). Lastly, future research should focus on chloride in the study of renal physiology and pathophysiology across a variety of clinical settings.

GRANTS

This work was supported in part by National Institute of Diabetes and Digestive and Kidney Diseases Grants R01DK096549, R01HL085757, U01DK082185, and U01DK106962 (S. G. Coca) and T32DK007757 (J. L. Rein).

DISCLOSURES

S. G. Coca serves as a consultant to Janssen Pharmaceuticals, Quark Biopharma, and CHF Solutions and serves on the Scientific Advisory Boards of pulseData, LLC and RenalytixAI, LLC. J. L. Rein has no conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

J.L.R. and S.G.C. prepared figures; J.L.R. and S.G.C. drafted manuscript; J.L.R. and S.G.C. edited and revised manuscript; J.L.R. and S.G.C. approved final version of manuscript.