Hypomagnesemia is a risk factor for nonrecovery of renal function and mortality in AIDS patients with acute kidney injury

The objective of the present study was to determine the prevalence of electrolyte disturbances in AIDS patients developing acute kidney injury in the hospital setting, as well as to determine whether such disturbances constitute a risk factor for nephrotoxic and ischemic injury. A prospective, observational cohort study was carried out. Hospitalized AIDS patients were evaluated for age; gender; coinfection with hepatitis; diabetes mellitus; hypertension; time since HIV seroconversion; CD4 count; HIV viral load; proteinuria; serum levels of creatinine, urea, sodium, potassium and magnesium; antiretroviral use; nephrotoxic drug use; sepsis; intensive care unit (ICU) admission, and the need for dialysis. Each of these characteristics was correlated with the development of acute kidney injury, with recovery of renal function and with survival. Fifty-four patients developed acute kidney injury: 72% were males, 59% had been HIV-infected for >5 years, 72% had CD4 counts <200 cells/mm3, 87% developed electrolyte disturbances, 33% recovered renal function, and 56% survived. ICU admission, dialysis, sepsis and hypomagnesemia were all significantly associated with nonrecovery of renal function and with mortality. Nonrecovery of renal function was significantly associated with hypomagnesemia, as was mortality in the multivariate analysis. The risks for nonrecovery of renal function and for death were 6.94 and 6.92 times greater, respectively, for patients with hypomagnesemia. In hospitalized AIDS patients, hypomagnesemia is a risk factor for nonrecovery of renal function and for in-hospital mortality. To determine whether hypomagnesemia is a determinant or simply a marker of critical illness, further studies involving magnesium supplementation in AIDS patients are warranted.


Introduction
It is recognized that renal complications have a substantive impact on the course of AIDS and alter the prognosis of the disease (1,2).In patients with AIDS, kidney disease can occur as a primary manifestation of HIV infection, as a secondary complication of an underlying illness or due to the side effects of antiretroviral therapy (3,4).Kidney disease has emerged as a leading cause of death among AIDS patients in the era of highly active antiretroviral therapy (HAART) (5).
Acute kidney injury (AKI) is one of the most common causes of renal failure in AIDS patients (6), as well as a strong predictor of in-hospital morbidity and mortality in the general population (7).Patients with AIDS frequently require hospitalization and are often treated with drug combinations, factors that contribute to the development of AKI (1), which has been reported to occur in up to 20% of hospitalized AIDS patients (8).Acute tubular necrosis (ATN) is the most common cause of AKI in hospitalized patients.The risk of AKI related to volume depletion, hemodynamic stress, nephrotoxic

O N L I N E P R O V I S I O N A L
medications, or the administration of contrast medium is greater among patients hospitalized with complications of AIDS (4).Sepsis is considered to be the leading cause of ischemic AKI in AIDS patients.In such patients, impaired renal handling of water and sodium also causes volume depletion (9).In hospitalized AIDS patients, AKI resulting from druginduced ATN is common.Antiretroviral drug use can cause renal failure through a variety of mechanisms: direct renal tubular toxicity (ATN, Fanconi-like syndrome, distal tubular acidosis, etc.), obstructions (crystal deposition in the kidney), and glomerular lesions (10,11).Other potential causes of AKI in AIDS patients include thrombotic microangiopathy and opportunistic infections of the renal parenchyma (12).Patients with AIDS who develop AKI can also present underlying diseases such as HIV-associated nephropathy and other glomerular diseases (13), as well as non-HIV-related kidney diseases such as diabetes and hypertension, as well as hepatitis C-or B-related nephropathy (14).Electrolyte disturbances are common in AIDS patients (15).Marked changes in electrolyte balance can result from the many and varied drugs used to control AIDS infection or can be secondary to opportunistic infections.
It has been shown that hypokalemia constitutes a risk factor for the development of AKI (16,17).Hypokalemia also causes a marked increase in gentamicin nephrotoxicity (18).In addition, hypokalemia is a common complication of treatment with amphotericin B and potentiates amphotericin B nephrotoxicity (19,20).The antiretroviral drugs zidovudine (AZT) and didanosine (ddI) are widely used in AIDS patients, and data suggest that their use in patients with hypokalemia can have adverse effects on renal function (21).It has been shown that both drugs decrease glomerular filtration rate and renal blood flow in hypokalemic rats, but that only AZT reduces glomerular filtration rate and renal blood flow in rats with hypomagnesemia.Therefore, chronic AZT and ddI administration can produce AKI in AIDS patients with hypokalemia or hypomagnesemia (21).Seguro et al. (17) showed that hypokalemia also enhances the tubular damage resulting from ischemic injury.In addition, magnesium supplementation has been shown to protect against postischemic AKI (22).
Although hypomagnesemia is also a common disturbance in AIDS patients, its effects in such patients have not been studied thoroughly.
The aim of the present study was to determine the prevalence of electrolyte disturbances in AIDS patients who develop AKI in the hospital setting, as well as whether such disturbances constitute a risk factor for nephrotoxic and ischemic injury.We also evaluated the recovery of renal function and mortality rates in this population of patients.

Study design and population
This was a prospective cohort study conducted at the Emílio Ribas Institute of Infectology in São Paulo, Brazil.All hospitalized AIDS patients over 18 years of age were eligible for inclusion.Those developing AKI between May and August of 2006 were selected for study.Patients presenting serum creatinine levels <1.5 mg/dL were excluded, as were pregnant patients and those with pre-existing chronic renal disease.Clinical, biochemical and demographic data were collected at enrollment and prospectively.The study design was approved by the Ethics in Research Committee of the Emílio Ribas Institute of Infectology and written informed consent was obtained from all patients, who participated in the study.

Data collection
Data related to the following variables were collected: gender, age, coinfection with hepatitis B, coinfection with hepatitis C, diabetes mellitus, hypertension, interval between HIV seroconversion and the occurrence of AKI, use of antiretroviral therapy, nephrotoxic drugs administered in the hospital, sepsis, and dehydration.Laboratory testing included the following parameters: CD4 count, HIV viral load, serum levels of urea, creatinine, sodium, potassium and magnesium, urinary protein excretion, and urinalysis.

Outcome measures
Each patient was evaluated prospectively throughout the entire hospital stay.The primary outcome measures were survival, duration of AKI, need for renal replacement therapy, and admission to the intensive care unit (ICU).

Comorbidities and laboratory data
Underlying clinical conditions and laboratory test results at the time of AKI development were reviewed in order to determine the potential cause of AKI.
A diagnosis of AKI was made if serum creatinine levels remained ≥1.5 mg/dL for more than 3 consecutive days during the hospital stay.Patients presenting serum creatinine levels ≥1.5 mg/dL for less than 3 days were considered to have prerenal azotemia and were excluded from the analysis.The criteria for not to consider prerenal azotemia after 3 days were based on the clinical analysis of the nephrologist (all patients were evaluated by the same nephrologist).The clinical signs of salt depletion were analyzed: postural hypotension, reduction in central venous pressure and skin turgor.

O N L I N E P R O V I S I O N A L
Also, we did not consider prerenal azotemia if saline infusion was given for the extracellular volume expansion and the serum levels of creatinine remained the same or even increased despite the treatment.Recovery of renal function was defined as a return of serum creatinine levels to <1.5 mg/dL.Coinfection with hepatitis B was defined as serum positivity for hepatitis B surface antigen or hepatitis Be antigen.Coinfection with hepatitis C was defined as serum positivity for hepatitis C antibodies or detectable hepatitis C virus RNA.
Diabetes mellitus was defined as glucose intolerance requiring dietary or pharmacologic management.Patients under treatment with antihypertensive medications at the time of admission were classified as hypertensive.Hyponatremia was defined as serum sodium levels <135 mEq/L, whereas serum sodium levels >145 mEq/L were considered to be indicative of hypernatremia.Hyperkalemia was defined as serum potassium levels >5 mEq/L, and hypokalemia was defined as serum potassium levels <3.5 mEq/L.Hypomagnesemia was defined as serum magnesium levels <1.7 mg/dL, and hypermagnesemia was defined as serum magnesium levels >2.1 mg/dL.
HIV RNA levels were quantified by the reverse transcriptase polymerase chain reaction using the standard and ultrasensitive versions of the Amplicor HIV Monitor Assay (Roche Diagnostics, USA).

Statistical analysis
Statistical analysis was based on the frequency of nonrecovery of renal function and on the mortality rate, as well as on the odds ratios estimated using univariate and multivariate logistic regression.In the univariate analysis, factors associated with nonrecovery and mortality were analyzed individually.The initial multivariate model included all variables that presented an association of P ≤ 0.15 in the univariate analysis.In the final model, only variables that maintained an association of P ≤ 0.10 were retained, and we employed a Wald statistic backward stepwise selection.The statistical analysis was performed using the Statistical Package for the Social Sciences, version 14.0 for Windows (SPSS Inc., USA).

Clinical, biochemical and demographic characteristics
Of the 54 patients selected for inclusion, 15 were females and 39 were males.Most of the patients (59%) had been HIV-infected for at least 5 years, and 72% presented CD4 counts <200 cells/mm 3 .Table 1 shows the general characteristics of the cohort.

Outcomes
Unlike what is seen in the general population, only 18 patients (33%) recovered renal function.There were 21 patients (39%) who required dialysis.It is of note that only 2 (10%) of these 21 patients recovered renal function, whereas 13 (62%) died in the hospital.Of the 33 patients (61%) who did not require dialysis, 16 (49%) recovered renal function and 11 (33%) died in the hospital.In terms of mortality, the difference between AIDS patients who recovered renal function and those who did not was significant (P < 0.001).
Of the cohort as a whole, 28 patients (52%) were admitted to the ICU, and 18 (64%) of them died in the hospital.Of the 26 patients (48%) who were not admitted to the ICU, only 6 (23%) died in the hospital.The overall in-hospital mortality rate was 44%.

O N L I N E P R O V I S I O N A L
analysis (Table 2), we found that the nonrecovery of renal function correlated significantly with ICU admission (P = 0.004), the need for dialysis (P = 0.008), sepsis (P = 0.010), dehydration (P = 0.010), and hypomagnesemia (P = 0.020).
In the multivariate logistic regression analysis (Table 3), the independent risk factors for nonrecovery of renal function were hypomagnesemia (P = 0.03) and ICU admission (P = 0.082).The risk for nonrecovery of renal function was 3.8 times greater in patients admitted to the ICU than in those who did not require intensive care (P = 0.02).Interestingly, hypomagnesemia was also a risk factor for the nonrecovery of renal function.Of the 23 patients who presented serum magnesium levels ≥1.7 mg/dL, 11 (48%) recovered renal function.However, of the 18 patients who presented serum magnesium levels <1.7 mg/dL, only 2 (11%) recovered renal function (P = 0.030).The risk for the nonrecovery of renal function was 6.9 times greater in patients with hypomagnesemia than in those without.
Mortality was quite high (64%) among the patients admitted to the ICU.
In multiple logistic regression analysis (Table 5), the independent risk factors for mortality were hypomagnesemia (P = 0.033) and sepsis (P = 0.003).
The most interesting finding in the present study was the correlation between hypomagnesemia and mortality (P = 0.033).Of the 18 patients who presented hypomagnesemia, 11 (62%) died in the hospital, compared with only 5 (22%) of the 23 patients presenting normal serum levels of magnesium.We found that the risk of death for patients presenting serum magnesium levels <1.7 mg/dL was 6.92 times greater than that for those presenting serum magnesium levels ≥1.7 mg/dL.

Discussion
The results of the present study emphasize the variety of clinical causes of AKI in hospitalized AIDS patients, the use of multiple nephrotoxic drugs for these patients and the impact that electrolyte disturbances have on the course of the disease.
Our study population was predominantly male.The majority of the patients evaluated had been HIV-infected for at least 5 years, presented low CD4 counts and were not receiving regular antiretroviral therapy.Noncompliance with treatment continues to be a challenge in the treatment of these patients.
Our findings are compatible with those of many other studies in which renal dysfunction has been shown to occur predominantly in the more advanced stages of HIV infection (14,15).Sepsis is considered to be a major cause of AKI in AIDS patients (5,6,9).In a sample of patients with advanced HIV infection, approximately half of those presenting AKI also had sepsis, which causes prerenal azotemia by promoting systemic vasodilation and arterial hypotension (6).Sepsis was the leading cause of renal failure in our study population as well.
Sixty-five percent of the patients presenting AKI were receiving potentially nephrotoxic drugs, contrast medium being the most common.Contrast-induced AKI continues to be a leading cause of hospital-acquired renal failure (23).Various studies have demonstrated that diabetic nephropathy, volume depletion, dehydration, hypercholesterolemia, and preexisting renal insufficiency are the most important risk factors for the development of contrast-induced AKI (23).
There are no data on whether AIDS can predispose to contrast-induced AKI.However, AIDS patients are typically under treatment with other potentially nephrotoxic drugs, as well as having sepsis, presenting a high risk for volume depletion (due to dehydration and diarrhea) and being frequently submitted to imaging studies involving the administration of contrast medium.
In view of these facts, it is not surprising that contrast medium was the most common potentially nephrotoxic drug employed in our cohort, and that vancomycin was the second most common.It should be borne in mind that vancomycin is used only in patients with severe infections, which might explain the fact that a significant correlation between vancomycin use and mortality, as well as between vancomycin use and nonrecovery of renal function, was found only in the univariate analysis.
In the present study, hyponatremia was the most common electrolyte disturbance.Hyponatremia occurs in 36-56% of hospitalized AIDS patients and typically indicates inappropriate arginine vasopressin secretion or volume depletion secondary to gastrointestinal fluid loss (15,24).

O N L I N E P R O V I S I O N A L
The second most common electrolyte disturbance detected in our study population was hypokalemia, which, since it can be caused by drugs such as gentamicin, amphotericin B, acyclovir, and tenofovir, is frequently observed in AIDS patients (18,20,25,26).In addition, the clinical symptoms of AIDS (diarrhea, malnutrition and anorexia) predispose to hypokalemia.It is also known that hypokalemia enhances the tubular damage resulting from ischemic injury (16).In contrast to what would be expected, hypokalemia did not correlate with the nonrecovery of renal function in the present study.
Hypomagnesemia is a common entity, occurring in up to 12% of hospitalized patients (27).The incidence can be as high as 65% in patients in intensive care settings, in which nutrition, diuretics, hypoalbuminemia, and aminoglycosides can play important roles (27,28).Gastrointestinal losses, renal losses (due to volume expansion, as well as to the use of pentamidine, amphotericin and diuretic drugs), redistribution and hypokalemia are all potential causes of hypomagnesemia (27,28).
Hypomagnesemia is considered to be a common electrolyte disturbance in patients with AIDS (15).Since AIDS patients commonly present diarrhea and malnutrition and are treated with drugs that can cause magnesium loss, they are at high risk for developing hypomagnesemia, which reportedly occurs in approximately 29% of such patients (15).Magnesium plays a basic role in various organ functions, and the kidney is the main regulator of magnesium (27,28).In mammals, the magnesium ion plays an important role, since the intracellular level of free magnesium regulates intermediary metabolism, DNA/RNA structure, DNA/RNA synthesis, cell growth, cell reproduction, membrane structure, potassium transport, calcium ion transport, signal transduction modulation, fat synthesis, and protein synthesis (27)(28)(29).Since the adenosine triphosphate-magnesium complex is bound and then hydrolyzed by enzymes, magnesium is a cofactor for most adenosine triphosphates (28,29).
It has been shown that hypomagnesemia is a risk factor for AKI (22).de Araujo et al. (21) demonstrated that AZT induces a decrease in glomerular filtration rate and in renal blood flow in rats with hypomagnesemia.Hypomagnesemia also predisposes to AZT nephrotoxicity (21).Therefore, chronic administration of AZT can produce AKI in AIDS patients with hypomagnesemia.de Araujo et al. ( 22) also showed that magnesium supplementation protects against postischemic AKI.These investigators found that renal blood flow in ischemic rats receiving magnesium supplementation is comparable to that observed in nonischemic rats.It is known that hypomagnesemia leads to hypertension, atherogenesis and stroke, although the mechanisms involved remain ill-defined (30).
It has also been shown that patients with hypertension, ischemic heart disease or stroke exhibit significant magnesium depletion (31).Recent in vitro studies of aortic and cerebrovascular smooth muscle cells from humans, rats and dogs have demonstrated that reductions in serum magnesium levels to below those considered normal induce the expression of at least two proto-oncogenes (c-fos and c-jun), as well as of the nuclear factor-kappa B (NF-κB) transcription factor (30).Since it is known that mRNA expression of c-fos, c-jun and NF-κB increases after renal ischemia/reperfusion, these factors might also be involved in the pro-inflammatory and oxidative renal processes (32).Renal ischemia-induced hypomagnesemia might be associated with up-regulation of proto-oncogenes and of NF-κB.In various cells, including peripheral blood mononuclear cells, magnesium is involved in myriad reactions and functions: enzymatic reactions, operation of channel, receptor and intracellular signaling molecules, as well as nucleic acid and protein conformation (27,33).Here, we have demonstrated for the first time that hypomagnesemia is a risk factor for mortality and for nonrecovery of renal function in AIDS patients.
The relationship between hypomagnesemia and mortality has been well described.Rubeiz et al. (34) reported that mortality rates were nearly twice as high in hypomagnesemic patients as in patients presenting normal serum levels of magnesium.Soliman et al. (35) demonstrated that developing ionized hypomagnesemia during an ICU stay correlated with a poor prognosis.However, there are no available reports showing that magnesium supplementation can improve mortality in hypomagnesemic patients in the ICU setting.In animal models, magnesium deficiency has been shown to increase production of inflammatory cytokines (36).
As previously stated, sepsis was the leading cause of AKI in our study population.Salem et al. (37) showed that progressive magnesium deficiency and hypomagnesemia are strongly associated with increased mortality in experimental sepsis.In patients with sepsis or septic shock, the prevalence of hypomagnesemia is high, and sepsis has been identified as an independent risk factor for developing hypomagnesemia during an ICU stay (35).Moreover, studies have suggested that low magnesium concentration may contribute to the pathogenesis of coronary atherosclerosis or acute thrombosis (38).
In order to classify the severity of AKI in hospitalized patients, we used the definition of AKI established by the Adult AIDS Clinical Trial Groups (39).This might constitute a limitation of the study, since it is possible that, by using this definition, we missed some cases of AKI in which there were smaller increases in serum creatinine.
In conclusion, hypomagnesemia in hospitalized AIDS patients is a risk factor for nonrecovery of renal function and for in-hospital mortality.Studies involving magnesium supplementation in AIDS patients are warranted in order to ascertain whether hypomagnesemia is a determinant or simply a marker of critical illness.For each variable: OR = 1.000 indicates the reference category.In addition to these variables, the initial multivariate model included time since HIV seroconversion, highly active antiretroviral therapy, nephrotic syndrome, need for dialysis, electrolyte disturbances, dehydration, and sepsis.

Table 3 .
Final results of the multivariate analysis of sociodemographic and clinical characteristics associated with nonrecovery of renal function. .= degrees of freedom; OR = odds ratio estimated by multivariate logistic regression; ICU = intensive care unit.For each variable: OR = 1.000 indicates the reference category.In addition to these variables, the initial multivariate model included age, time since HIV seroconversion, need for dialysis, electrolyte disturbances, dehydration, and sepsis.

Table 5 .
Final results of the multivariate analysis of sociodemographic and clinical characteristics associated with mortality. .= degrees of freedom; OR = odds ratio estimated by multivariate logistic regression.