https://orcid.org/0000-0003-3387-9093
Abstract
We hypothesized that oxidative stress in Ugandan children with severe malaria is associated with mortality.
We evaluated biomarkers of oxidative stress in children with cerebral malaria (CM, n = 77) or severe malarial anemia (SMA, n = 79), who were enrolled in a randomized clinical trial of immediate vs delayed iron therapy, compared with community children (CC, n = 83). Associations between admission biomarkers and risk of death during hospitalization or risk of readmission within 6 months were analyzed.
Nine children with CM and none with SMA died during hospitalization. Children with CM or SMA had higher levels of heme oxygenase-1 (HO-1) (P < .001) and lower superoxide dismutase (SOD) activity than CC (P < .02). Children with CM had a higher risk of death with increasing HO-1 concentration (odds ratio [OR], 6.07 [95% confidence interval {CI}, 1.17–31.31]; P = .03) but a lower risk of death with increasing SOD activity (OR, 0.02 [95% CI, .001–.70]; P = .03). There were no associations between oxidative stress biomarkers on admission and risk of readmission within 6 months of enrollment.
Children with CM or SMA develop oxidative stress in response to severe malaria. Oxidative stress is associated with higher mortality in children with CM but not with SMA.
NCT01093989.
Plasmodium falciparum malaria remains a leading cause of morbidity and mortality in children. African children <5 years of age constitute 80% of the mortality in the region due to malaria, which approximates 480 000 deaths annually [1]. The pathophysiological mechanisms leading to severe disease and mortality in malaria-infected children have not been fully elucidated. Clinical outcomes are determined by complex interactions between host and parasitic factors, including microvascular obstruction by sequestered parasitized erythrocytes causing hypoxic injury, dysregulated host immunity and inflammatory responses, and parasite-specific variant surface antigen–mediated pathogenicity [2–4]. There is a need to better define mechanisms of disease with the goal of identifying biomarkers to predict adverse outcomes or discovering therapeutic targets.
Physiological levels of reactive oxygen species (ROS) under homeostatic conditions are a normal part of human cell functioning and are critical for antimicrobial defense [5]. But an excess of free radicals and reactive nonradicals, such as hydrogen peroxide, produced by parasitized erythrocytes, macrophages, neutrophils, and endothelial cells during malaria infections, can overwhelm the antioxidant defense system [6–8]. The resulting imbalance triggers oxidative stress and contributes to malaria pathobiology. Although prospective studies of African children with malaria have demonstrated significant perturbations of finely regulated pro-oxidant/antioxidant balance, the association between molecular markers of oxidative stress and mortality in children remains unclear [9–11]. The purpose of this study was to investigate the relationship between measures of oxidative stress or compensatory responses and mortality in Ugandan children with severe malaria.
Intraerythrocytic parasites release redox-active free heme and large quantities of iron-containing hemozoin from the food vacuole upon rupture of the erythrocyte. Hemozoin, an insoluble product of hemoglobin metabolism, can lead to oxidative stress in erythrocytes and host tissues [12]. In addition, some antimalarial agents, such as chloroquine and artemisinin, exert their activities in part by inducing production of free radicals and oxidative stress in infected erythrocytes, which can cause deleterious off target effects in their host [13]. Free radicals modify intracellular targets such as DNA, proteins, and lipids, and mediate damage by causing oxidation of proteins and peroxidation of membrane phospholipids [14]. Because ROS have short lifespans and extremely low concentrations, they cannot easily be quantified directly [15], but their activity can be measured indirectly by estimating their toxic byproducts, such as malondialdehyde (MDA); proteins involved in the hosts’ defense against oxidant injury, such as superoxide dismutase (SOD) and heme oxygenase-1 (HO-1) [5]; and other oxidative stress measures, such as F2-isoprostanes [16, 17] and biopterins [18, 19].
The World Health Organization estimates that Uganda carries the third-highest burden of P. falciparum malaria cases globally, with an annual incidence of >250 cases per 1000 population [1]. Ugandan children are at particularly high risk of morbidity and mortality from malaria. In this prospective study of Ugandan children hospitalized for treatment of severe malaria, we tested our hypothesis that increased oxidative stress is associated with increased mortality.
METHODS
Research Setting and Study Design
The study was conducted at the Paediatric Acute Care Unit of the Mulago National Referral Hospital, Kampala, Uganda, a 1790-bed regional hospital and major referral center for children. We report on oxidative stress biomarkers in children enrolled in a randomized clinical trial of immediate vs delayed iron therapy in children with severe malaria compared with community children (CC) without severe malaria to investigate how timing of iron supplementation impacted long-term iron status and incidence of subsequent malaria episodes [20, 21]. Enrollment of children was conducted from June 2010 to December 2013. Hospitalized children with a diagnosis of malaria were eligible for the study if they had cerebral malaria (CM) or severe malarial anemia (SMA). CM was diagnosed in children with coma (Blantyre Coma Score ≤2 or Glasgow Coma Score ≤8) and detection of P. falciparum on a blood smear, without another explanation for coma. SMA was defined as a hemoglobin concentration ≤5 g/dL in children with P. falciparum observed on thick or thin blood smear by microscopy. Healthy CC were recruited from the same households or neighborhoods as the children with CM or SMA; all groups had similar numbers.
Inclusion and Exclusion Criteria
Children were eligible if they were 18 months to 5 years of age and had CM or SMA. Community controls were children who were healthy during the previous 4 weeks and were within the same age range as the CM and SMA cases. Children with known chronic illnesses were excluded from this study.
Outcome Measures
For the present substudy, the primary outcome was death during the initial hospitalization. The secondary outcome was readmission to a healthcare facility for any cause within 6 months of enrollment. A repeat malaria infection was diagnosed if a child had an axillary temperature ≥37.5°C and P. falciparum on a blood smear.
Measure of Iron Deficiency
Iron deficiency was defined as zinc protoporphyrin (ZPP) ≥80 μmol/mol heme, which was measured using a hematofluorometer (Aviv Biomedical, Lakewood, New Jersey) [20]. ZPP, a hemoglobin precursor, is a sensitive biomarker of iron deficiency that can be measured rapidly near the point of care and may detect iron deficiency before anemia is diagnosed.
Measures of Oxidative Stress and Compensatory Responses
MDA is a marker of oxidative stress and its concentration (µmol/L) was measured with the Malondialdehyde high-performance liquid chromatography (HPLC) Assay (Eagle Biosciences, Amherst, New Hampshire). HO-1 concentration (ng/mL) was assayed using an HO-1 (human) enzyme-linked immunosorbent assay kit (Enzo Life Sciences, Farmingdale, New York). This enzyme is a surrogate for oxidative stress as it is expressed in response to endothelial damage and has an antioxidant function [22, 23]. SOD activity was measured as a marker of antioxidant activity using the OxiSelect Superoxide Dismutase Activity Assay (Cell Biolabs, San Diego, California). This enzyme mediates neuroprotective effects and protects endothelial integrity [24, 25]. Positive control proteins were provided by the same manufacturers: MDA HPLC used a control set, HO-1 concentrations and SOD activity were interpolated on standard curves generated with control proteins, and interplate sample replicates were tested to ensure reproducibility of results. All 3 tests were measured at the time of admission before study randomization and 28 days later. Heparinized vacutainers were used for collection of venous blood samples. Plasma was separated within 3 hours of collection and stored at −80°C at the study site until testing was performed. Samples were processed and stored in the same manner at both timepoints.
Therapeutic Interventions
All children with CM or SMA were initially treated with either intravenous (IV) quinine (June 2010–December 2012) or IV artesunate (January 2013–December 2013) according to Ugandan National Clinical Guidelines. As soon as oral medication was tolerated, artemether/lumefantrine was the prescribed agent for the full duration of the study. Children with SMA also received 1–5 blood transfusions to treat anemia. Community children with positive malarial blood smears but no clinical evidence of malaria were eligible for antimalarial treatment according to national guidelines and were retained in the study.
Iron Therapy
As previously described [20, 21], all enrolled children, except CC without evidence of iron deficiency, were randomized to receive a 3-month course of ferrous sulfate suspension (2 mg/kg/day) that started either during the initial hospitalization when oral medication could be tolerated (immediate arm) or on day 28 (delayed arm). Although a placebo was not provided to the delayed arm, treatment and follow-up care were identical in all other respects. The treatment allocation was determined by simple random assignment that was stratified by study group (CM, SMA, and CC). Adherence to therapy was monitored using caregiver-administered treatment diaries and biweekly verification by study monitors.
Statistical Analyses
The original trial was powered to find significant differences in hemoglobin concentration and number of malaria episodes between study arms as previously described [21]. For descriptive statistics, median values with interquartile range (IQR) were calculated for skewed continuous data, and proportions were used for categorical data. Statistical comparisons among groups were performed using the Kruskal–Wallis rank test for continuous data, Dunn test for post hoc pairwise comparisons between groups using the Benjamini–Hochberg approach, and Pearson χ2 test for categorical data. Binary logistic regression analysis was performed to model the relationship between measures of oxidative stress (log10-transformed) and death during hospitalization, adjusted for age. Cox proportional hazards regression analysis was used to model the temporal relationship between measures of oxidative stress (log10-transformed) and risk of postdischarge readmission, adjusted for age and iron treatment study arm. To determine if the effect of oxidative stress on readmission was modified by the assignment to a particular iron treatment study arm, we used interaction terms for iron treatment and oxidative stress biomarkers in the age-adjusted Cox proportional hazards regression models. Wilcoxon matched-pairs signed-rank tests were used to compare oxidative stress biomarkers at baseline to 1-month follow-up values. A P value < .05 was considered statistically significant. Analyses were performed using Stata version 16.0 (StataCorp), SPSS version 27 (IBM), and Prism version 9 (GraphPad).
Ethics Statement
Providers obtained written informed consent from parents or guardians of subjects prior to their enrollment in the study. The institutional review boards for human studies at Makerere University School of Medicine, the University of Minnesota, and Rhode Island Hospital/Lifespan Inc approved the study. The Uganda National Council for Science and Technology provided regulatory approval, and the Uganda National Drug Authority provided ethical approval. The original trial was registered at ClinicalTrials.gov (identifier NCT01093989).
RESULTS
Patient Enrollment and Clinical Characteristics
A total of 239 children were enrolled in this study as previously described (Figure 1) [21]. All children with SMA (n = 77) and CM (n = 79), as well as 35 of 83 (42%) children in the CC group, had iron deficiency. Children with ZPP ≥80 µmol/mol heme were randomized to begin a 3-month iron therapy regimen (2 mg/kg/day as liquid oral ferrous sulfate) concurrently with antimalarial treatment on day 0 or as soon as oral medication could be tolerated (immediate group) or 28 days later (delayed group). More than 90% of children initially discharged from the hospital were monitored for the 6-month duration of this study. As reported previously, 9 children (11.3%) with CM and no children with SMA died during the hospitalization. Of the 9 children with CM who died during hospitalization, 1 was in the immediate iron arm and 8 were in the delayed iron arm, but none had initiated iron therapy before death. Two children with SMA, both of whom received iron immediately, died after discharge. There were no observed differences in demographic, clinical, or conventional baseline laboratory findings between children who died before discharge and those who died subsequently. Two children withdrew from the study and 1 child with SMA, who received immediate iron therapy, was lost to follow-up (Figure 1) [21].
Flow diagram of patient enrollment, treatment randomization, and mortality. Abbreviations: D, delayed iron therapy; I, immediate iron therapy; LTFU, lost to follow-up; ZPP, zinc protoporphyrin.
There were no significant differences in age or sex distributions among the study groups (Table 1). Baseline parasitemia levels were similar between the CM and SMA groups (Table 1). Overall, 9 (9.3%) children in the immediate iron group and 6 (7.2%) children in the delayed iron group had a repeat malaria infection in the first 28 days after enrollment, but this difference was not significant (P = .62) [20].
Demographic and Parasitological Data for Hospitalized Children With Cerebral Malaria or Severe Malarial Anemia, and Community Children
| CM | SMA | CC | ||
|---|---|---|---|---|
| Characteristic | (n = 79)a | (n = 77) | (n = 83) | P Value |
| Age, y, median (IQR) | 3.27 (2.17–3.92) | 2.63 (2.00–3.53) | 2.93 (2.39–3.78) | .07 |
| Female sex, No. (%) | 37 (46.8) | 31 (40.3) | 48 (57.8) | .41 |
| Peripheral parasite density, parasites/µL, median (IQR) | 47 400 (9760–181 080) | 35 740 (10 040–85 980) | 0 (0–0) | <.001b |
| CM | SMA | CC | ||
|---|---|---|---|---|
| Characteristic | (n = 79)a | (n = 77) | (n = 83) | P Value |
| Age, y, median (IQR) | 3.27 (2.17–3.92) | 2.63 (2.00–3.53) | 2.93 (2.39–3.78) | .07 |
| Female sex, No. (%) | 37 (46.8) | 31 (40.3) | 48 (57.8) | .41 |
| Peripheral parasite density, parasites/µL, median (IQR) | 47 400 (9760–181 080) | 35 740 (10 040–85 980) | 0 (0–0) | <.001b |
Bold font indicates statistical significance.
Abbreviations: CC, community children; CM, cerebral malaria; IQR, interquartile range; SMA, severe malarial anemia.
Children with available parasite level results: n = 74.
Post hoc analysis revealed that CC differed significantly from CM and from SMA. Statistical tests included the Kruskal–Wallis rank test for comparing continuous variables, Dunn test for pairwise comparisons between groups using the Benjamini–Hochberg approach, and Pearson χ2 test for categorical variables.
Demographic and Parasitological Data for Hospitalized Children With Cerebral Malaria or Severe Malarial Anemia, and Community Children
| CM | SMA | CC | ||
|---|---|---|---|---|
| Characteristic | (n = 79)a | (n = 77) | (n = 83) | P Value |
| Age, y, median (IQR) | 3.27 (2.17–3.92) | 2.63 (2.00–3.53) | 2.93 (2.39–3.78) | .07 |
| Female sex, No. (%) | 37 (46.8) | 31 (40.3) | 48 (57.8) | .41 |
| Peripheral parasite density, parasites/µL, median (IQR) | 47 400 (9760–181 080) | 35 740 (10 040–85 980) | 0 (0–0) | <.001b |
| CM | SMA | CC | ||
|---|---|---|---|---|
| Characteristic | (n = 79)a | (n = 77) | (n = 83) | P Value |
| Age, y, median (IQR) | 3.27 (2.17–3.92) | 2.63 (2.00–3.53) | 2.93 (2.39–3.78) | .07 |
| Female sex, No. (%) | 37 (46.8) | 31 (40.3) | 48 (57.8) | .41 |
| Peripheral parasite density, parasites/µL, median (IQR) | 47 400 (9760–181 080) | 35 740 (10 040–85 980) | 0 (0–0) | <.001b |
Bold font indicates statistical significance.
Abbreviations: CC, community children; CM, cerebral malaria; IQR, interquartile range; SMA, severe malarial anemia.
Children with available parasite level results: n = 74.
Post hoc analysis revealed that CC differed significantly from CM and from SMA. Statistical tests included the Kruskal–Wallis rank test for comparing continuous variables, Dunn test for pairwise comparisons between groups using the Benjamini–Hochberg approach, and Pearson χ2 test for categorical variables.
Baseline Measurements of Select Plasma Oxidative Stress Biomarkers in Children With CM or SMA
Concentrations of MDA and HO-1 and activity of SOD in available plasma were compared across patient groups (Figure 2). MDA levels (µmol/L, median [IQR]) were similar across all groups (CM: 1.0 [0.62–1.3]; SMA: 1.2 [0.90–1.5]; CC: 0.98 [0.76–1.3]; Figure 2A). Children with CM or SMA had significantly higher levels of HO-1 (ng/mL, median [IQR]) compared with CC (CM: 16.2 [8.5–27.4]; SMA: 14.9 [9.6–32.5]; CC: 3.2 [1.9–9.1]; P < .001, each analysis; Figure 2B). On the other hand, SOD activity (mUnits [mU]/µL, median [IQR]) was significantly lower in children with CM (219.2 [65.3–451.6]) compared with SMA (407.5 [149.8–1051.4]; P < .02), as well as CC (884.7 [155.4–1338.5]; P < .001); children with SMA also had significantly lower SOD activity than CC (P < .02; Figure 2C).
Baseline measurements of select plasma oxidative stress biomarkers in children with cerebral malaria (CM) or severe malarial anemia (SMA) compared with community children (CC). Shown are plasma malondialdehyde concentration (A), heme oxygenase-1 concentration (B), and superoxide dismutase activity (C) in children with CM or SMA compared with CC. Circles represent plasma biomarker values in individual children. Boxes show median and interquartile range; bars show minimum and maximum values. Statistical comparisons were performed with the Kruskal–Wallis rank test, and post hoc analysis used Dunn test for pairwise comparisons using the Benjamini–Hochberg approach. *P < .02; **P < .001. Abbreviations: CC, community children; CM, cerebral malaria; HO-1, heme oxygenase-1; MDA, malondialdehyde; SMA, severe malarial anemia; SOD, superoxide dismutase.
Associations Between Baseline Oxidative Stress Biomarker Values and Death During Admission in Children With CM
Children with CM had a 6-fold increase in risk of death during hospitalization for every 1 ng/mL increase in HO-1 concentration at baseline (odds ratio [OR], 6.07 [95% confidence interval {CI}, 1.17–31.31]; P = .03, Table 2). Conversely, children with CM had a 50-fold decrease in risk of death for every 1 mU/μL increase in SOD activity (OR, 0.02 [95% CI, .001–.70]; P = .03; Table 2). No significant associations between MDA and death were observed. Associations between biomarkers of oxidative stress and mortality among children with SMA were not assessed because there were no deaths in children with SMA during their hospitalization.
Associations Between Select Baseline Plasma Oxidative Stress Biomarkers and Death Among Hospitalized Children With Cerebral Malaria
| Death During Hospitalizationa | Risk of Deathb | ||||
|---|---|---|---|---|---|
| Died | Survived | ||||
| Biomarker | Median (IQR) (no./No., %) | Median (IQR) (no./No., %) | P Value | Adjusted OR (95% CI) | P Value |
| MDA, μmol/L | 0.90 (0.25–1.79) (3/42, 7) | 1.05 (0.62–1.34) (39/42, 93) | .75 | 0.04 (.001–18.6) | .31 |
| HO-1, ng/mL | 28.32 (26.13–37.42) (9/79, 11) | 15.23 (7.68–26.60) (70/79, 89) | .02 | 6.07 (1.17–31.31) | .03 |
| SOD activity, mU/uL | 78.96 (70.02–79.43) (5/57, 9) | 244.68 (65.30–507.87) (52/57, 91) | .10 | 0.02 (.001–.70) | .03 |
| Death During Hospitalizationa | Risk of Deathb | ||||
|---|---|---|---|---|---|
| Died | Survived | ||||
| Biomarker | Median (IQR) (no./No., %) | Median (IQR) (no./No., %) | P Value | Adjusted OR (95% CI) | P Value |
| MDA, μmol/L | 0.90 (0.25–1.79) (3/42, 7) | 1.05 (0.62–1.34) (39/42, 93) | .75 | 0.04 (.001–18.6) | .31 |
| HO-1, ng/mL | 28.32 (26.13–37.42) (9/79, 11) | 15.23 (7.68–26.60) (70/79, 89) | .02 | 6.07 (1.17–31.31) | .03 |
| SOD activity, mU/uL | 78.96 (70.02–79.43) (5/57, 9) | 244.68 (65.30–507.87) (52/57, 91) | .10 | 0.02 (.001–.70) | .03 |
Bold font indicates statistical significance. no./No. indicates number with specified outcome/number tested.
Abbreviations: CI, confidence interval; HO-1, heme oxygenase-1; IQR, interquartile range; MDA, malondialdehyde; OR, odds ratio; SOD, superoxide dismutase.
Wilcoxon rank-sum tests were used for univariate comparisons of biomarker values.
Risk of death, adjusted for age, was calculated using binary logistic regression models with log-transformed biomarker values.
Associations Between Select Baseline Plasma Oxidative Stress Biomarkers and Death Among Hospitalized Children With Cerebral Malaria
| Death During Hospitalizationa | Risk of Deathb | ||||
|---|---|---|---|---|---|
| Died | Survived | ||||
| Biomarker | Median (IQR) (no./No., %) | Median (IQR) (no./No., %) | P Value | Adjusted OR (95% CI) | P Value |
| MDA, μmol/L | 0.90 (0.25–1.79) (3/42, 7) | 1.05 (0.62–1.34) (39/42, 93) | .75 | 0.04 (.001–18.6) | .31 |
| HO-1, ng/mL | 28.32 (26.13–37.42) (9/79, 11) | 15.23 (7.68–26.60) (70/79, 89) | .02 | 6.07 (1.17–31.31) | .03 |
| SOD activity, mU/uL | 78.96 (70.02–79.43) (5/57, 9) | 244.68 (65.30–507.87) (52/57, 91) | .10 | 0.02 (.001–.70) | .03 |
| Death During Hospitalizationa | Risk of Deathb | ||||
|---|---|---|---|---|---|
| Died | Survived | ||||
| Biomarker | Median (IQR) (no./No., %) | Median (IQR) (no./No., %) | P Value | Adjusted OR (95% CI) | P Value |
| MDA, μmol/L | 0.90 (0.25–1.79) (3/42, 7) | 1.05 (0.62–1.34) (39/42, 93) | .75 | 0.04 (.001–18.6) | .31 |
| HO-1, ng/mL | 28.32 (26.13–37.42) (9/79, 11) | 15.23 (7.68–26.60) (70/79, 89) | .02 | 6.07 (1.17–31.31) | .03 |
| SOD activity, mU/uL | 78.96 (70.02–79.43) (5/57, 9) | 244.68 (65.30–507.87) (52/57, 91) | .10 | 0.02 (.001–.70) | .03 |
Bold font indicates statistical significance. no./No. indicates number with specified outcome/number tested.
Abbreviations: CI, confidence interval; HO-1, heme oxygenase-1; IQR, interquartile range; MDA, malondialdehyde; OR, odds ratio; SOD, superoxide dismutase.
Wilcoxon rank-sum tests were used for univariate comparisons of biomarker values.
Risk of death, adjusted for age, was calculated using binary logistic regression models with log-transformed biomarker values.
Impact of Iron Therapy on Biomarkers of Oxidative Stress
Initial high HO-1 concentrations in children with CM or SMA tended to decrease, regardless of whether they had received iron supplementation or not, and approximated normal values after 1 month as indicated by HO-1 levels in CC (Table 3). Initial low SOD activity in children with CM or SMA rebounded, with or without iron treatment, and reached an approximately normal range of values after 1 month as indicated by SOD activity in CC (Table 4).
Comparisons of Heme Oxygenase-1 Concentrations at Baseline and After 1 Month
| Study Group by Timing of Iron Therapy | Baseline | 1 Month | |
|---|---|---|---|
| Median (IQR) | Median (IQR) | P Valuea | |
| Cerebral malariab | |||
| Immediate (n = 37) | 13.33 (7.36–24.57) | 8.64 (5.68–17.95) | .01 |
| Delayed (n = 31) | 15.74 (10.68–36.60) | 9.60 (5.81–12.98) | <.001 |
| Severe malarial anemiab | |||
| Immediate (n = 35) | 18.88 (9.54–41.25) | 9.68 (5.75–15.30) | .005 |
| Delayed (n = 35) | 13.55 (8.56–30.40) | 11.19 (8.32–20.95) | .30 |
| Community childrenb | |||
| Immediate (n = 19) | 3.43 (2.17–10.67) | 7.88 (6.06–13.41) | .23 |
| Delayed (n = 16) | 3.17 (2.27–3.57) | 6.78 (5.67–10.99) | <.001 |
| No iron (n = 45) | 2.82 (1.89–11.27) | 6.98 (5.44–12.21) | .01 |
| Study Group by Timing of Iron Therapy | Baseline | 1 Month | |
|---|---|---|---|
| Median (IQR) | Median (IQR) | P Valuea | |
| Cerebral malariab | |||
| Immediate (n = 37) | 13.33 (7.36–24.57) | 8.64 (5.68–17.95) | .01 |
| Delayed (n = 31) | 15.74 (10.68–36.60) | 9.60 (5.81–12.98) | <.001 |
| Severe malarial anemiab | |||
| Immediate (n = 35) | 18.88 (9.54–41.25) | 9.68 (5.75–15.30) | .005 |
| Delayed (n = 35) | 13.55 (8.56–30.40) | 11.19 (8.32–20.95) | .30 |
| Community childrenb | |||
| Immediate (n = 19) | 3.43 (2.17–10.67) | 7.88 (6.06–13.41) | .23 |
| Delayed (n = 16) | 3.17 (2.27–3.57) | 6.78 (5.67–10.99) | <.001 |
| No iron (n = 45) | 2.82 (1.89–11.27) | 6.98 (5.44–12.21) | .01 |
Bold font indicates statistical significance.
Abbreviation: IQR, interquartile range.
Wilcoxon matched-pairs signed-rank test.
Only subjects with available data were included.
Comparisons of Heme Oxygenase-1 Concentrations at Baseline and After 1 Month
| Study Group by Timing of Iron Therapy | Baseline | 1 Month | |
|---|---|---|---|
| Median (IQR) | Median (IQR) | P Valuea | |
| Cerebral malariab | |||
| Immediate (n = 37) | 13.33 (7.36–24.57) | 8.64 (5.68–17.95) | .01 |
| Delayed (n = 31) | 15.74 (10.68–36.60) | 9.60 (5.81–12.98) | <.001 |
| Severe malarial anemiab | |||
| Immediate (n = 35) | 18.88 (9.54–41.25) | 9.68 (5.75–15.30) | .005 |
| Delayed (n = 35) | 13.55 (8.56–30.40) | 11.19 (8.32–20.95) | .30 |
| Community childrenb | |||
| Immediate (n = 19) | 3.43 (2.17–10.67) | 7.88 (6.06–13.41) | .23 |
| Delayed (n = 16) | 3.17 (2.27–3.57) | 6.78 (5.67–10.99) | <.001 |
| No iron (n = 45) | 2.82 (1.89–11.27) | 6.98 (5.44–12.21) | .01 |
| Study Group by Timing of Iron Therapy | Baseline | 1 Month | |
|---|---|---|---|
| Median (IQR) | Median (IQR) | P Valuea | |
| Cerebral malariab | |||
| Immediate (n = 37) | 13.33 (7.36–24.57) | 8.64 (5.68–17.95) | .01 |
| Delayed (n = 31) | 15.74 (10.68–36.60) | 9.60 (5.81–12.98) | <.001 |
| Severe malarial anemiab | |||
| Immediate (n = 35) | 18.88 (9.54–41.25) | 9.68 (5.75–15.30) | .005 |
| Delayed (n = 35) | 13.55 (8.56–30.40) | 11.19 (8.32–20.95) | .30 |
| Community childrenb | |||
| Immediate (n = 19) | 3.43 (2.17–10.67) | 7.88 (6.06–13.41) | .23 |
| Delayed (n = 16) | 3.17 (2.27–3.57) | 6.78 (5.67–10.99) | <.001 |
| No iron (n = 45) | 2.82 (1.89–11.27) | 6.98 (5.44–12.21) | .01 |
Bold font indicates statistical significance.
Abbreviation: IQR, interquartile range.
Wilcoxon matched-pairs signed-rank test.
Only subjects with available data were included.
Comparisons of Superoxide Dismutase Activity at Baseline and After 1 Month
| Baseline | 1 Month | ||
|---|---|---|---|
| Study Group by Timing of Iron Therapy | Median (IQR) | Median (IQR) | P Valuea |
| Cerebral malariab | |||
| Immediate (n = 29) | 257.02 (65.30–561.16) | 577.48 (332.89–708.33) | .06 |
| Delayed (n = 22) | 188.25 (66.30–377.14) | 464.99 (365.18–602.41) | .009 |
| Severe malarial anemiab | |||
| Immediate (n = 25) | 444.32 (150.13–1152.81) | 579.52 (384.17–755.92) | .99 |
| Delayed (n = 22) | 400.07 (149.80–767.80) | 529.78 (390.74–812.69) | .17 |
| Community childrenb | |||
| Immediate (n = 14) | 819.11 (97.62–1404.56) | 449.91 (192.59–844.37) | .22 |
| Delayed (n = 9) | 716.64 (88.87–910.33) | 590.50 (377.27–759.39) | .86 |
| No iron (n = 24) | 955.85 (415.88–1364.39) | 614.21 (493.20–850.26) | .35 |
| Baseline | 1 Month | ||
|---|---|---|---|
| Study Group by Timing of Iron Therapy | Median (IQR) | Median (IQR) | P Valuea |
| Cerebral malariab | |||
| Immediate (n = 29) | 257.02 (65.30–561.16) | 577.48 (332.89–708.33) | .06 |
| Delayed (n = 22) | 188.25 (66.30–377.14) | 464.99 (365.18–602.41) | .009 |
| Severe malarial anemiab | |||
| Immediate (n = 25) | 444.32 (150.13–1152.81) | 579.52 (384.17–755.92) | .99 |
| Delayed (n = 22) | 400.07 (149.80–767.80) | 529.78 (390.74–812.69) | .17 |
| Community childrenb | |||
| Immediate (n = 14) | 819.11 (97.62–1404.56) | 449.91 (192.59–844.37) | .22 |
| Delayed (n = 9) | 716.64 (88.87–910.33) | 590.50 (377.27–759.39) | .86 |
| No iron (n = 24) | 955.85 (415.88–1364.39) | 614.21 (493.20–850.26) | .35 |
Bold font indicates statistical significance.
Abbreviation: IQR, interquartile range.
Wilcoxon matched-pairs signed-ranks test.
Only subjects with available data were included.
Comparisons of Superoxide Dismutase Activity at Baseline and After 1 Month
| Baseline | 1 Month | ||
|---|---|---|---|
| Study Group by Timing of Iron Therapy | Median (IQR) | Median (IQR) | P Valuea |
| Cerebral malariab | |||
| Immediate (n = 29) | 257.02 (65.30–561.16) | 577.48 (332.89–708.33) | .06 |
| Delayed (n = 22) | 188.25 (66.30–377.14) | 464.99 (365.18–602.41) | .009 |
| Severe malarial anemiab | |||
| Immediate (n = 25) | 444.32 (150.13–1152.81) | 579.52 (384.17–755.92) | .99 |
| Delayed (n = 22) | 400.07 (149.80–767.80) | 529.78 (390.74–812.69) | .17 |
| Community childrenb | |||
| Immediate (n = 14) | 819.11 (97.62–1404.56) | 449.91 (192.59–844.37) | .22 |
| Delayed (n = 9) | 716.64 (88.87–910.33) | 590.50 (377.27–759.39) | .86 |
| No iron (n = 24) | 955.85 (415.88–1364.39) | 614.21 (493.20–850.26) | .35 |
| Baseline | 1 Month | ||
|---|---|---|---|
| Study Group by Timing of Iron Therapy | Median (IQR) | Median (IQR) | P Valuea |
| Cerebral malariab | |||
| Immediate (n = 29) | 257.02 (65.30–561.16) | 577.48 (332.89–708.33) | .06 |
| Delayed (n = 22) | 188.25 (66.30–377.14) | 464.99 (365.18–602.41) | .009 |
| Severe malarial anemiab | |||
| Immediate (n = 25) | 444.32 (150.13–1152.81) | 579.52 (384.17–755.92) | .99 |
| Delayed (n = 22) | 400.07 (149.80–767.80) | 529.78 (390.74–812.69) | .17 |
| Community childrenb | |||
| Immediate (n = 14) | 819.11 (97.62–1404.56) | 449.91 (192.59–844.37) | .22 |
| Delayed (n = 9) | 716.64 (88.87–910.33) | 590.50 (377.27–759.39) | .86 |
| No iron (n = 24) | 955.85 (415.88–1364.39) | 614.21 (493.20–850.26) | .35 |
Bold font indicates statistical significance.
Abbreviation: IQR, interquartile range.
Wilcoxon matched-pairs signed-ranks test.
Only subjects with available data were included.
Association Between Oxidative Stress and Readmission to Hospital
There were no associations between admission markers of oxidative stress and risk of readmission for any cause within 6 months of enrollment, after adjustment for age and iron treatment category (Table 5). There was no interaction between admission oxidative stress markers and iron treatment on risk of readmission.
Associations Between Initial Oxidative Stress Biomarkers and Risk of All-Cause Readmission Within 6 Months of Enrollment
| CM | SMA | |||||
|---|---|---|---|---|---|---|
| Biomarker on Admissiona | No. | Adjusted HR (95% CI)b | P Value | No. | Adjusted HR (95% CI)b | P Value |
| MDA | 39 | 0.92 (.09–9.30) | .94 | 33 | 0.31 (.002–39.8) | .64 |
| SOD activity | 52 | 0.71 (.26–1.95) | .51 | 53 | 1.50 (.60–3.71) | .39 |
| HO-1 | 70 | 1.16 (.35–3.77) | .81 | 75 | 1.12 (.39–3.21) | .83 |
| CM | SMA | |||||
|---|---|---|---|---|---|---|
| Biomarker on Admissiona | No. | Adjusted HR (95% CI)b | P Value | No. | Adjusted HR (95% CI)b | P Value |
| MDA | 39 | 0.92 (.09–9.30) | .94 | 33 | 0.31 (.002–39.8) | .64 |
| SOD activity | 52 | 0.71 (.26–1.95) | .51 | 53 | 1.50 (.60–3.71) | .39 |
| HO-1 | 70 | 1.16 (.35–3.77) | .81 | 75 | 1.12 (.39–3.21) | .83 |
Abbreviations: CI, confidence interval; CM, cerebral malaria, HO-1, heme oxygenase-1; HR, hazard ratio; MDA, malondialdehyde; SMA, severe malarial anemia; SOD, superoxide dismutase.
Only subjects with available data were included.
Cox proportional hazards model adjusted for age and iron treatment. There were no significant interaction effects between biomarkers of oxidative stress and iron treatment.
Associations Between Initial Oxidative Stress Biomarkers and Risk of All-Cause Readmission Within 6 Months of Enrollment
| CM | SMA | |||||
|---|---|---|---|---|---|---|
| Biomarker on Admissiona | No. | Adjusted HR (95% CI)b | P Value | No. | Adjusted HR (95% CI)b | P Value |
| MDA | 39 | 0.92 (.09–9.30) | .94 | 33 | 0.31 (.002–39.8) | .64 |
| SOD activity | 52 | 0.71 (.26–1.95) | .51 | 53 | 1.50 (.60–3.71) | .39 |
| HO-1 | 70 | 1.16 (.35–3.77) | .81 | 75 | 1.12 (.39–3.21) | .83 |
| CM | SMA | |||||
|---|---|---|---|---|---|---|
| Biomarker on Admissiona | No. | Adjusted HR (95% CI)b | P Value | No. | Adjusted HR (95% CI)b | P Value |
| MDA | 39 | 0.92 (.09–9.30) | .94 | 33 | 0.31 (.002–39.8) | .64 |
| SOD activity | 52 | 0.71 (.26–1.95) | .51 | 53 | 1.50 (.60–3.71) | .39 |
| HO-1 | 70 | 1.16 (.35–3.77) | .81 | 75 | 1.12 (.39–3.21) | .83 |
Abbreviations: CI, confidence interval; CM, cerebral malaria, HO-1, heme oxygenase-1; HR, hazard ratio; MDA, malondialdehyde; SMA, severe malarial anemia; SOD, superoxide dismutase.
Only subjects with available data were included.
Cox proportional hazards model adjusted for age and iron treatment. There were no significant interaction effects between biomarkers of oxidative stress and iron treatment.
DISCUSSION
In this substudy within a randomized clinical trial of iron therapy in Ugandan children with severe malaria compared with healthy CC [20, 21], we analyzed associations between select biomarkers of oxidative stress measured at initial enrollment and in-hospital mortality and secondarily, subsequent all-cause readmissions. Levels of HO-1, a surrogate marker of oxidative stress, were significantly higher in the CM and SMA groups compared with CC, and conversely, the activity of SOD, an antioxidant enzyme, was significantly lower in children with CM and SMA than CC, and lowest in children with CM. These perturbations were not impacted by iron therapy and normalized within 1 month of testing. Notably, the odds of death in children with CM, adjusted for age, were 6-fold higher for every 1 ng/mL increase in HO-1 concentration measured at the time of admission. Furthermore, the age-adjusted odds of death in children with CM were 50-fold higher for every 1 mU/μL decrease in SOD activity. On the other hand, no association between death and MDA, another biomarker of oxidative stress, was found in children with CM. Together, these novel findings show that oxidative stress is present in children with CM or SMA, and associated with mortality in children with CM.
Cerebral malaria develops in <1% of children infected with P. falciparum, mostly in children aged <5 years [26]. However, cumulatively, CM represents a massive global burden of disease because it can cause permanent neurological sequelae or mortality in up to 30% of affected children [26]. Precise neuropathological mechanisms leading to CM and its complications are not known but are thought to be related to complex interactions among erythrocytes parasitized with certain PfEMP1-expressing strains of P. falciparum, endothelial cells and their receptors; host innate and adaptive immune responses; and neuroactive mediators such as nitric oxide [2, 4, 27–29]. Detailed knowledge about the potential role of dysregulated oxidation reduction in the explanatory model could lead to novel therapeutic or preventive strategies for at-risk children. Our observations align with 3 lines of evidence showing the important protective role of antioxidants in CM in humans [30, 31], animal models [23, 25], and in vitro models of endothelial damage [24].
We previously reported that key biomarkers of inflammation (hepcidin, ferritin, and C-reactive protein) were substantially elevated in the current cohort of children, who presented to hospital with severe malaria [21]. However, hepcidin was significantly higher in the CM group compared with the SMA group, suggesting that inflammation is even greater in that condition [20]. That observation aligns with our new finding that biomarkers of oxidative stress were associated with significantly higher mortality in children with CM. Oxidative stress is thought to contribute to vascular endothelial dysfunction and increased permeability, a hallmark of cerebral edema, which is a prominent feature of CM [29]. Compensatory antioxidant responses include SOD activity, among others, that provides early protection against oxidative stress by reducing superoxide radicals and hydrogen peroxide, which is further reduced to water by both catalase and glutathione peroxidase. Reduction in SOD activity is explained partly by scavenging of nitric oxide by superoxide (·O2−) that leads to the formation of peroxynitrite (ONOO−), and subsequently to SOD inhibition [32]. SOD also reacts with hydrogen peroxide in a peroxidase reaction that inhibits its antioxidant activity [33, 34]. The association between increased SOD activity in the present study and decreased mortality supports the importance of antioxidant activity in protection against severe malaria. It is possible the increased hemolysis that occurs in SMA as compared to CM contributed in part to the increased SOD activity seen in children with SMA, since SOD is found in erythrocytes [35].
HO-1 (previously named heat-shock protein 32), an early indicator of endothelial damage, is an inducible enzyme that catalyzes heme into its constituent components including biliverdin and subsequently bilirubin, which are potent antioxidants, as well as ferrous iron, which catalyzes the Fenton reaction producing highly reactive oxygen species [14]. HO-1 induction has been shown to prevent blood-brain barrier disruption, brain microvasculature congestion, and neuro-inflammation in the murine model [23]. Therefore, expression of cerebral HO-1 is considered a critical adaptive response to various causes of stress, such as infection, ischemia, trauma, and neurotoxins, and has been shown to be cyto- and neuroprotective [31]. At the molecular level, HO-1 is upregulated by ROS, certain cytokines, and possibly hemozoin [31] and, in turn, it downregulates expression of various adhesion molecules thereby moderating the severity of cellular and tissue damage caused by free heme released from parasitized erythrocytes [14]. In the present study, higher plasma levels of HO-1, which may have provided greater antioxidant activity, were associated with increased risk of death, unlike the correlation with SOD activity, in which lower values, consistent with less antioxidant activity, were associated with death. In the case of HO-1, plasma levels may reflect oxidative stress more than antioxidant activity [36]. It is possible that other measures (eg, HO-1 gene expression, HMOX1 gene polymorphisms) may more accurately reflect the activity of HO-1 than plasma levels [37]. Alternatively, HO-1 levels may have a narrow therapeutic range, outside of which the increased HO-1 can be harmful. Walther et al also demonstrated elevated plasma HO-1 levels in children with severe malaria [38], and reported in vitro studies showing that moderate induction of HO-1 is associated with protection against heme-mediated damage [39], while at higher activity, HO-1 can mediate oxidative stress and cellular injury [40]. The consistent association of elevated HO-1 levels with malaria disease severity in the present study and the study by Walther et al, and the association of HO-1 levels in the present study with increased mortality, both suggest that it is a marker of oxidative stress and, along with the in vitro studies, supports a potential mechanistic role for HO-1 in mediating oxidative stress.
Based on our hypothesis that oxidative stress is important in the pathogenesis of CM leading to death, it was surprising to find that the levels of plasma MDA, another biomarker of oxidative stress, were not higher in children with severe malaria than CC, particularly since elevated MDA levels were seen in children with severe malaria in a previous study [9]. Possible reasons for this discrepancy are that (1) plasma levels of MDA do not consistently reflect localized tissue production, for example in cerebral vasculature; (2) SOD and HO-1 are more sensitive indicators of disease severity in this setting; (3) relatively larger compensatory responses in SOD activity occurred during the single timeframe that we captured; or (4) the assay was insufficiently sensitive.
Study limitations include the evaluation of children at a single study site. Multicenter studies are needed to verify the generalizability of these findings. In addition, we tested only systemic measures of oxidative stress. This may not reflect oxidative stress in the central nervous system compartment, which may have important prognostic and mechanistic implications, as described by Rubach et al [19] who measured biopterins in cerebrospinal fluid. Evaluation of long-term neurocognitive outcomes may shed light on other potential damaging effects of oxidative stress caused by severe malaria. In this observational cohort study, we could not determine whether HO-1 and SOD directly mediate the lethal effects of CM or are products of severe disease that serve as markers for risk of death in children with CM. To determine the potential value of antioxidants as therapeutics, a randomized controlled trial will be needed.
In conclusion, we found that oxidative stress, as evaluated by plasma HO-1 levels and SOD activity, is associated with mortality in Ugandan children with CM. If future studies confirm these findings, benefits could include point-of-care tests to predict mortality and guide medical interventions, as well as new therapeutics that supplement [41] or enhance expression of certain antioxidants.
Notes
Author contributions. D. B. B., B. H., I. C. M., and C. C. J. conceptualized and designed the study. R. O. O., S. E. C., and C. C. J. were responsible for the conduct of the study and data collection. B. H. performed the bioassays. D. B. B., C. B., and I. C. M. analyzed and interpreted the data. D. B. B. and I. C. M. prepared the initial draft of the manuscript. All authors read and approved the final manuscript and agree to be accountable for all aspects of the work.
Acknowledgments. We thank the children and their parents who participated in this study, and the study team for their dedicated effort in treating the children and collecting the data.
Disclaimer. The funding agencies had no role in the design of the study; collection, analysis, and interpretation of data; or in writing of the manuscript.
Financial support. This work was supported by the National Institute of Neurological Disorders and Stroke (grant number R01 NS055349 to C. C. J.); the Eunice Kennedy Shriver National Institute of Child Health and Human Development (grant number U01 HD064698 to C. C. J.) and the National Institute of Allergy and Infectious Diseases (grant number K08 AI100997 to I. C. M.), all at the National Institutes of Health.
Potential conflicts of interest. All authors: No reported conflicts.
All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
References
Author notes
aD. B. B. and B. H. contributed equally to this work.
bI. C. M. and C. C. J. contributed equally to this work.
Presented in part: 69th Annual Meeting of the American Society of Tropical Medicine and Hygiene, Virtual, 15–19 November 2020.
