Cell-free hemoglobin mediated oxidative stress is associated with acute kidney injury and renal replacement therapy in severe falciparum malaria: an observational study

Background Intravascular hemolysis is an intrinsic feature of severe malaria pathophysiology but the pathogenic role of cell-free hemoglobin-mediated oxidative stress in severe malaria associated acute kidney injury (AKI) is unknown. Methods As part of a prospective observational study, enrolment plasma cell-free hemoglobin (CFH), lipid peroxidation markers (F2-isoprostanes (F2-IsoPs) and isofurans (IsoFs)), red cell deformability, and serum creatinine were quantified in Bangladeshi patients with severe falciparum malaria (n = 107), uncomplicated malaria (n = 80) and sepsis (n = 28). The relationships between these indices and kidney function and clinical outcomes were examined. Results AKI was diagnosed at enrolment in 58% (62/107) of consecutive patients with severe malaria, defined by an increase in creatinine ≥1.5 times expected baseline. Severe malaria patients with AKI had significantly higher plasma cell-free hemoglobin (geometric mean CFH: 8.8 μM; 95% CI, 6.2–12.3 μM), F2-isoprostane (56.7 pg/ml; 95% CI, 45.3–71.0 pg/ml) and isofuran (109.2 pg/ml; 95% CI, 85.1–140.1 pg/ml) concentrations on enrolment compared to those without AKI (CFH: 5.1 μM; 95% CI, 4.0–6.6 μM; P = 0.018; F2-IsoPs: 27.8 pg/ml; 95% CI, 23.7–32.7 pg/ml; P < 0.001; IsoFs: 41.7 pg/ml; 95% CI, 30.2–57.6 pg/ml; P < 0.001). Cell-free hemoglobin correlated with markers of hemolysis, parasite burden (P. falciparum histidine rich protein 2 (PfHRP2)), and F2-IsoPs. Plasma F2-IsoPs and IsoFs inversely correlated with pH, positively correlated with creatinine, PfHRP2 and fractional excretion of sodium, and were higher in patients later requiring hemodialysis. Plasma F2-IsoP concentrations also inversely correlated with red cell deformability and were higher in fatal cases. Mixed effects modeling including an interaction term for CFH and time showed that F2-IsoPs, IsoFs, PfHRP2, CFH, and red cell rigidity were independently associated with increasing creatinine over 72 h. Multivariable logistic regression showed that admission F2-IsoPs, IsoFs and red cell deformability were associated with the need for subsequent hemodialysis. Conclusions Cell-free hemoglobin and lipid peroxidation are associated with acute kidney injury and disease severity in falciparum malaria, suggesting a pathophysiological role in renal tubular injury. Evaluation of adjunctive therapies targeting cell-free hemoglobin-mediated oxidative stress is warranted. Electronic supplementary material The online version of this article (doi:10.1186/s12879-017-2373-1) contains supplementary material, which is available to authorized users.


(Continued from previous page)
Conclusions: Cell-free hemoglobin and lipid peroxidation are associated with acute kidney injury and disease severity in falciparum malaria, suggesting a pathophysiological role in renal tubular injury. Evaluation of adjunctive therapies targeting cell-free hemoglobin-mediated oxidative stress is warranted.
Keywords: Acute kidney injury, Pathophysiology, Falciparum malaria, Cell-free hemoglobin, Oxidative stress Background Severe falciparum malaria is characterized by intravascular hemolysis, where cell-free hemoglobin (CFH) increases with disease severity [1]. Sources of CFH include rupture of parasitized red blood cells (RBC) at schizogony, and destruction of uninfected erythrocytes, most prominently in patients with blackwater fever (hemoglobinuria) [2]. In 400 BC, Hippocrates first associated blackwater fever with anuria and mortality; findings that consistently resurfaced after Firth's report in 1886 [3,4]. More recently, this condition of fulminant hemolysis has been associated with kidney dysfunction in up to 64% of patients [5], but the underlying mechanisms have not been fully characterized.
When the degree of intravascular hemolysis exceeds the scavenging capacity of plasma haptoglobin for hemoglobin, CFH dimers are filtered by the glomeruli and reabsorbed by the proximal tubule. Once the reabsorptive capacity is exceeded, hemoglobin appears in the urine [6]. CFH is independently associated with AKI in patients post-cardiopulmonary bypass, and with mortality in bacterial sepsis [7][8][9]. Hemoproteins, hemoglobin and myoglobin, are pathogenic as prooxidants when released heme is not scavenged by hemopexin. Heme redox cycling between ferric and ferryl states then generates globin radicals inducing lipid peroxidation [10]. In vivo studies on oxidative injury have been hampered by the paucity of stable and specific markers of oxidative stress.
CFH-mediated non-enzymatic lipid peroxidation of arachidonic acid generates isomers of prostaglandins, F 2 -isoprostanes (F 2 -IsoPs) and isofurans (IsoFs) [11,12]. F 2 -IsoPs are generated at low oxygen tension whereas IsoFs are generated at higher oxygen tension and together are considered robust in vivo measures of oxidative stress [11,12]. In the current study, the hypothesis was that CFH-mediated oxidative stress could cause renal damage either through a direct effect on renal tubules, through a reduction in red cell deformability (RCD), or through the vasoconstrictive properties of F 2 -IsoPs.
Arachidonic acids, such as red cell membrane phospholipids, are particularly vulnerable to free radicalmediated lipid peroxidation. Oxidative stress-induced reduction of RCD has been proposed to play a role in renal insufficiency [13]. In malaria, the high arachidonic acid content of infected RBC membranes reduces with intracellular parasite maturation, suggesting membrane peroxidation occurs during parasite development [14]. F 2 -IsoPs are considered not just bystanders of oxidative injury but are bioactive renal vasoconstrictors [12]. Both F 2 -IsoPs and IsoFs have been associated with AKI in patients with rhabdomyolysis and hemolysis postcardiopulmonary bypass [15][16][17]. Other plasma and urinary markers of oxidative stress have been shown to be significantly elevated in severe malaria compared to uncomplicated malaria [18][19][20][21].
The role of plasma CFH-mediated lipid peroxidation in the pathophysiology of severe malaria and AKI has not been described. Gaining a better understanding of the pathophysiology in malaria will help towards the development of targeted therapies. This study aimed to examine the generation of CFH-mediated lipid peroxidation and its role in AKI and malaria severity by analyzing the associations between CFH, F 2 -IsoPs, IsoFs, red cell deformability, and creatinine in patients with falciparum malaria.

Study aim, design and setting
The aim of this study was to assess CFH-mediated lipid peroxidation and its role in AKI and disease severity in falciparum malaria. This prospective observational study was conducted at Chittagong Medical College Hospital, Bangladesh from 2011 to 2014. This tertiary hospital receives referrals from malaria hypoendemic areas, and has basic facilities for intensive care and hemodialysis.

Patient characteristics
Patients admitted with slide confirmed severe or uncomplicated P. falciparum malaria were recruited upon diagnosis. Positive microscopy of peripheral blood required the presence of asexual stages of P. falciparum. Uncomplicated malaria was defined as asexual P. falciparum slide positivity without severity criteria. Criteria for severe malaria were: coma (Glasgow Coma Score < 11), shock (systolic blood pressure (SBP) < 80 mmHg with cool extremities), anemia, (hematocrit <20% plus parasitemia >100,000/μl), jaundice (total bilirubin >51.3 μmol/L plus parasitemia >100,000/μl), hyperparasitemia (asexual parasitemia >10%), acidosis (bicarbonate <15 mmol/L), hyperlactatemia (lactate >4 mmol/L), hypoglycemia (glucose <2.22 mmol/L), convulsions (≥ 2 in 24 h), pulmonary edema, and/or AKI (serum creatinine >3 mg/dl). Patients were treated with parenteral artesunate (Guilin No.2 Pharmaceuticals, Guangxi, China) followed by artemether/lumefantrine (Novartis, Basel, Switzerland) and managed according to WHO treatment guidelines [22]. Hemodialysis was initiated according to local nephrologists. Indications for dialysis in this setting include: (1) anuria for more than 24 h, (2) severe electrolyte and acid-base disturbance, (3) serum creatinine >3 mg/dl with urine output <0.5 ml/kg/h for 12 h, or (4) gradual rise in creatinine despite normal urine output. A control group of sepsis patients hospitalized with suspected bacterial infection and at least 2/4 systemic inflammatory response criteria (n = 28) was also recruited [23]. Data from non-pregnant patients aged ≥10 years is presented. Patients were followed until discharge or death; those with AKI were followed in hospital until renal recovery, if possible, as gauged by local nephrologists. Follow up after hospital discharge was challenging due to the distances patients travel to the teriary care center.

Study procedures
After enrolment, a medical history and physical examination were performed. Patients were seen 6-hourly until discharge or death. Enrolment venous blood samples were analyzed for electrolytes, glucose, pH and bicarbonate using a bedside analyzer (iSTAT, Abbott). Parasitemia was assessed 6-hourly from thick and thin smears until parasite clearance. Blood and urine for creatinine measurement were collected every 24 h for three days. Serum, heparinized plasma and urine were frozen in liquid nitrogen within one hour of collection. Serum creatinine was measured using an Olympus analyzer (Beckman Coulter Inc.).

Assays
Plasma and urine F 2 -isoprostane and isofuran concentrations were determined by gas chromatography-mass spectrometry at Vanderbilt University, as described [11,24]. The 24-h urinary excretion rate of urine F 2isoprostane and isofuran concentrations were calculated as: concentration x volume × 24 h/time of collection [25]. Plasma CFH concentrations were measured by ELISA (Bethyl Laboratories), as described [1]. Plasma Plasmodium falciparum histidine rich protein 2 (PfHRP2), a biomarker of total parasite burden, was quantified using commercial sandwich ELISA (Celisa, Cellabs), as described [26].
Red cell deformability (RCD) was measured at enrolment using a laser-assisted optical rotational cell analyzer (LORCA, Mechatronic) immediately after blood collection [27,28]. The deformability was measured by ellipticity of RBCs and described by the elongation index; (long minus short axes lengths) divided by (long plus short axes lengths). RCD was assessed at shear stresses ranging from 0.3 to 30 Pa. In capillaries, shear stresses of 1.7 Pa and above are encountered [29].

Acute kidney injury
Patients were classified according to AKI status at enrolment as defined by an increase in serum creatinine ≥1.5 times expected baseline, known or presumed to have occurred within the prior seven days (as per the Kidney Disease Improving Global Outcomes (KDIGO) classification system) [30]. Since urine output was not available for all patients, and the duration of illness at presentation was always greater than 48 h, these criteria were not incorporated for enrolment AKI diagnosis. As preadmission creatinine values were not available, expected baseline creatinine values were calculated as recommended using the Modification of Diet in Renal Disease formula assuming a glomerular filtration rate (GFR) of 75 ml/min/1.73m 2 for participants 19 years and older [30] and using the Bedside Schwartz formula assuming a GFR of 100 ml/min/1.73m 2 for those 18 years and younger [31][32][33]. The highest KDIGO AKI staging was assessed both on enrolment and during admission in order to accurately present the heterogeneous AKI status at the time of enrolment and subsequent kidney function during admission in those that survived. Stage 2 AKI was defined as an increase to ≥ 2.0-2.9 times expected baseline; Stage 3 as either an increase to ≥ 3 times expected baseline, an increase in creatinine to ≥4 mg/dl, initiation of RRT, or in patients <18 years a decrease in GFR to <35 ml/min/1.73 m 2 [30]. Creatinine used for enrolment AKI stratification were performed on samples drawn prior to hemodialysis.

Statistical analysis
Groups were compared using Wilcoxon rank-sum test or Student's t-tests depending on the distribution of the data. Correlations were assessed using Spearman's correlation coefficient. As the hypothesis was that F 2 -IsoPs, IsoFs, and CFH contribute to AKI, these pre-specified variables were assessed in a mixed effects model using creatinine as the dependent variable, and in a logistic regression model using hemodialysis as the dependant variable. As the F 2 -IsoPs and IsoFs were highly collinear (r p = 0.67; p < 0.001), their association with creatinine was assessed in separate multivariable models. The multivariable models included adjustment for age, SBP, PfHRP2 and RCD as these are considered physiologically relevant in contributing to AKI. In the mixed effect model, age, SBP, and time were modeled as fixed effects while the rest were treated as random effects. All prespecified and known AKI risk factor variables that were associated with creatinine were included in a multivariable mixed effects model. The interaction between enrolment CFH and time was also assessed. To account for hemodialysis confounding the decline in creatinine concentrations, creatinine values were adjusted at each time point following dialysis until time of death by using a creatinine rise of 1.5 mg/dl per day as proposed for anephric states [34]. Known risk factors for hemodialysis, including additional markers of disease severity (number of severity criteria, GCS, and lactate), were assessed using backward stepwise logistic regression including variables (Table 5) with a p-value of <0.10 on univariable analysis. Selection of the final model was based on the Akaike information criteria (AIC). Software used were STATA14.0 (Stata), and Prism 6 (Graphpad Software).

Red cell deformability, renal function and oxidative stress
RCD at enrolment was lower at shear stresses between 1.69 and 30.00 Pa in patients with AKI on enrolment compared to those without AKI (Figure 4). Decreased RCD correlated with higher creatinine at all these shear stresses and most strongly at 9.49 Pa (r s = − 0.31, P = 0.019). RCD was also lower at shear stresses between 1.69 and 9.49 Pa in patients who subsequently required hemodialysis (all P < 0.05). RCD at low shear stress was inversely associated with plasma   Fig. 2 Cell-free hemoglobin, and oxidative stress measures at enrolment in patients with severe malaria by AKI status. Plasma cell-free hemoglobin (n = 105), F 2 -isoprostanes (n = 64) and isofurans (n = 64) were significantly more elevated on enrolment in those with acute kidney injury. Geometric mean and 95% CI are shown. Abbreviations: AKI, acute kidney injury; CFH, cell-free hemoglobin; pF 2 -IsoP, plasma F 2 -isoprostanes; pIsoF, plasma isofurans

Predictors of renal function
Since plasma F 2 -IsoPs and IsoFs were collinear, two different multivariable mixed effects models were considered to assess the associations of each of them with change in serum creatinine over time (   associated with the subsequent need for hemodialysis, but RCD and CFH were not (overall model fit R 2 = 0.41; Table 5).

Discussion
In this study, plasma cell-free hemoglobin and oxidative stress markers (F 2 -IsoPs and IsoFs) were strongly associated with the presence of AKI on admission, and oxidative stress markers predicted subsequent creatinine elevation and hemodialysis requirement during admission in adult patients with severe falciparum malaria. Earlier studies have shown that urine F 2 -IsoPs and other urine and plasma oxidative stress markers are elevated in severe malaria compared to uncomplicated malaria [18][19][20][21]. CFH-mediated oxidative stress has been shown to contribute to AKI in diseases and medical procedures inducing hemolysis [15-17, 35, 36]. In malaria, intravascular hemolysis involves both parasitized and non-parasitized red blood cells [2]. In the current study, plasma CFH was associated with higher parasite burden, as measured by PfHRP2, consistent with a previous study [1]. This suggests that the hemoglobin released at schizont rupture, the end of the 48-h intra-erythrocytic lifecycle, contributes to plasma CFH concentration.
Redox cycling of hemoglobin forms a radical species that can initiate lipid peroxidation of arachidonic acid to generate F 2 -IsoPs and IsoFs [11,12,37]. In this study, CFH concentrations correlated with increased levels of plasma F 2 -IsoPs, a marker of oxidative stress. Several factors influence the generation of lipid peroxidation markers. The oxidative capacity of plasma CFH is largely dependent on its redox state and the fate of the heme moiety. Heme is water-insoluble; it binds to hemopexin, albumin, lipoproteins, and cell membranes [38], and its oxidative capacity differs considerably between these fractions. In addition, other sources of free radicals, such as those created during phagocytic oxidative burst, may initiate lipid peroxidation to generate plasma F 2 -IsoPs and IsoFs. Indeed, renal histopathology of AKI in severe malaria shows accumulation of host monocytes (in addition to parasitized red blood cells) in the renal microvasculature [39]. Heme-containing myoglobin concentrations are also increased in severe malaria, although to a much lesser extent than CFH [1]. The oxidative capacity of myoglobin increases at low pH because of increased pseudoperoxidase activity [10]. This could also apply to CFH, given that both plasma F 2 -IsoPs and IsoFs were associated with reduced venous blood pH in the current study.
F 2 -IsoPs are potent renal vasoconstrictors, which reduce renal blood flow and GFR [12]. In this study, concentrations of plasma CFH, F 2 -IsoPs, and IsoFs at enrolment were all higher in patients with AKI, whereas F 2 -IsoPs and IsoFs correlated with the fractional excretion of sodium (an indirect measure of renal tubular injury) and were independently associated with increased creatinine over 72 h and the subsequent requirement for hemodialysis. A common problem of AKI biomarkers is that their concentrations can be increased as a result of renal dysfunction, rather than be a cause of it. Indeed, the lower 24-h urine F 2 -IsoP excretion concentrations in patients with established AKI suggest that hemoglobin dimers and/or F 2 -IsoPs are not filtered well at low creatinine clearances. However, the group of patients without AKI on enrolment that subsequently developed AKI also showed elevated enrolment plasma IsoF concentrations. This was then followed by concomitant increases in 24 h plasma F 2 -IsoPs and IsoF with a subsequent peak in serum creatinine at 48 h. This time course suggests a role of plasma F 2 -IsoPs and IsoFs in the pathogenesis of AKI. The delayed effect of CFHmediated oxidative stress on renal function is also suggested by the multivariable mixed effects model, in which time modified the effect of enrolment CFH on creatinine, where the slope of change in creatinine increased with time. Plasma F 2 -IsoPs were also associated with reduced RBC deformability. Red cell rigidity has been well-described in severe malaria [27]. It is thought to be caused by oxidative damage to RBC membranes, and contribute to microvascular flow obstruction [27,28]. RBC membranes are rich in arachidonic acid. Heme-mediated lipid peroxidation of RBC membranes could be an important cause of reduced RCD, and a significant source of plasma F 2 -IsoPs and IsoFs in severe malaria. RCD at low shear stresses is mainly determined by membrane properties, whereas at high shear stresses surface-to-volume relationships become more important [40,41]. In this study, the correlation between plasma F 2 -IsoPs and decreased RCD was strongest at low shear stress, suggesting RBC membrane damage. In the current study, RCD was lower in patients with AKI, inversely correlated with serum creatinine, and independently associated with an increase in serum creatinine over 72 h. Reduced RCD could further compromise renal medullary perfusion concomitant with F 2 -IsoP-and IsoF-induced vasoconstriction and obstructing microvascular sequestered parasitized red blood cells, all contributing to renal tubular damage [39].
This study has some limitations. The sample size of this detailed study was relatively small. Additional biomarkers of AKI were not assessed. However, both changes in serum creatinine and hemodialysis are well established indicators of kidney dysfunction and injury. Living or post-mortem renal biopsy as a standard to evaluate the underlying kidney pathology was not ethical or feasible in this study. Quantification of plasma and urine hemoglobin oxidation states was not performed. However, studies have shown that methemoglobin is the predominant form in patients with malaria and blackwater fever [42]. Myoglobin as an alternative source of heme-mediated oxidative damage was not assessed, but other studies have shown a 100-fold less increase in plasma myoglobin compared to CFH in patients with severe malaria [1].

Conclusions
In summary, the findings of this study suggest that CFH and systemic F 2 -IsoPs and IsoFs contribute to the pathogenesis of AKI in severe malaria and these markers of oxidative stress are associated with need for hemodialysis and in-hospital mortality. CFH can have a direct oxidative damaging effect on renal tubules. In addition, CFH induced lipid peroxidation may have a dual effect in the mechanism of AKI in malaria. Firstly, the lipid peroxidation metabolites, plasma F 2 -IsoPs, may cause direct renal vasoconstriction. Secondly, lipid peroxidation of RBC membranes may result in reduced RCD, which in turn could exacerbate ischemia in the renal medulla.
Therapies targeted at reducing hemoprotein-mediated oxidative stress have been shown to improve renal function and mortality [9,17,35]. Treatments that reduce CFH levels or CFH-mediated oxidative stress, such as haptoglobin or paracetamol respectively, may have potential as an adjunctive therapy to improve kidney function and survival in severe malaria.

Additional file
Additional file 1: Figure S1. Supplement to Plewes K, Kingston HWF, Ghose A, et al. Cell-free hemoglobin mediated oxidative stress is associated with acute kidney injury and renal replacement therapy in severe falciparum malaria: an observational study. Cell-free hemoglobin, and oxidative stress measures at enrolment stratified by malaria severity. Plasma cell-free hemoglobin (n = 185), F 2 -isoprostanes (n = 82) and isofurans (n = 82) were significantly more elevated on enrolment in those with severe malaria compared to uncomplicated malaria. Geometric mean and 95% CIs shown. Abbreviations: AKI, acute kidney injury; CFH, cell-free hemoglobin; pF 2 -IsoP, plasma F 2 -isoprostanes; pIsoF, plasma isofurans. (DOCX 1707 kb) Abbreviations AKI: acute kidney injury; CFH: cell-free hemoglobin; F 2 -IsoPs: F 2 -isoprostanes; GFR: glomerular filtration rate; IsoF: Isofurans; KDIGO: Kidney Diseases Improving Global Outcomes; PfHRP2: Plasmodium falciparum histidine rich protein 2; RBC: red blood cell; RCD: red cell deformability; RRT: renal replacement therapy National Foundation for Infectious Diseases, Young Investigator recipient of the Merle A. Sande/Pfizer Fellowship in International Infectious Diseases; and the Clinician Investigator Program at the University of British Columbia, Canada.

Availability of data and materials
The datasets generated and analyzed during the current study are not publicly available. The Mahidol Oxford Tropical Medicine Research Unit has a Data Access Committee that reviews all data requests on a case by case basis. MORU is committed to ensuring that data sharing is planned for at the inception of a study: including during negotiations with funders and collaborating sites, during evaluation of compliance with local and international ethics and regulatory requirements, and during the design and conduct of consent processes. Queries and applications for datasets should be directed to the corresponding author who will discuss with the MORU Data Access committee. For further information please refer to the MORU Data Sharing Policy (http://www.tropmedres.ac/data-sharing-policy).
Authors' contributions LJR and AMD conceived of the study. KP contributed to study design, patient enrollment, acquisition of patient samples, performed the statistical analysis and wrote the manuscript. HWK, RJM, MTH, SJL, HI, TWY, PC, and KS contributed to the study design, patient enrolment, sample collection and manuscript revision. AG, MMUH, MSH, SA and MAH contributed to study design, clinical care of the patients and manuscript revision. KAP performed the cell-free hemoglobin measurements, and provided interpretation of the data and revision of the manuscript. SJL, MM and AMD advised and assisted with statistical analysis, and critically revised the manuscript for intellectual content. LJR provided the F 2 -IsoP and IsoF quantification. MAF, GDHT, TWY, NMA, NPJD, NJW, LJR, and AMD provided interpretation of the data and critical manuscript revisions. All authors read and approved of the final manuscript, and agreed to be accountable for all aspects of the work.

Competing interests
All authors declare that they have no competing interests.

Consent for publication
Informed written consent for publication was obtained from each patient or legally acceptable representative.
Ethics approval and consent to participate Informed written consent to participate was obtained from each patient or legally acceptable representative. All procedures were in accordance with the ethical standards of Declaration of Helsinki 2008. Ethical approval was obtained from Chittagong Medical College Ethical Review Committee, and Oxford Tropical Research Ethics Committee (reference 21-11).

Publisher's Note
Springer nature remains neutral with regard jurisdictional claims in published maps and institutional affiliations.
Author details