Cardiovascular Findings in Severe Malaria: A Review

Background: Severe malaria remains a leading cause of death worldwide. A greater understanding of its impact on multiple organ systems is essential in reducing the burden of disease. In this review we will summarize previously reported cardiovascular parameters of both adults and children with severe malaria. Method: For this systematic review we searched MEDLINE and PUBMED for all papers published on cardiac function in severe malaria from January 1, 1990 until September 1, 2019. Severe malaria was defined as per World Health Organization. Publications were included if there was data from echocardiography, Pulse Contour Cardiac Output (PiCCO), or Pulmonary Arterial catheters (PAC) reported. Studies were excluded if related to medication induced cardiac dysfunction, malaria in pregnancy, or included subjects with known pre-existing heart disease. Results: Twenty-four studies met inclusion criteria, the majority of which were studies of adult patients or a mixed cohort. Six solely involved pediatric patients. Significant heterogeneity existed in the cardiac parameters measured and results reported. One pediatric and one adult study suggested a reduced preload state during severe malaria. Cardiac systolic function was reported primarily within, or above, normative numeric ranges established in uninfected pediatric patients without anemia. Extensive variability existed in adult studies with reports of an elevated cardiac index in two studies, normal cardiac function in two studies, and descriptions of decreased function in two studies. Two reports suggest afterload in pediatric severe malaria is reduced. Reports of changes in the systemic vascular resistance of adults with severe malaria are inconsistent, with two trials demonstrating an increase and two suggesting a decrease. Studies demonstrated a mild rise in pulmonary pressure in both pediatric and adult patients that normalized by discharge. Conclusion: Based on limited data, the cardiovascular effects of severe malaria appear to be heterogeneous and vary depending on age. Further detailed studies are required to explore and understand the overall hemodynamic effects of this high burden disease.


Introduction
Despite the burden of malaria falling worldwide, progress in outcomes has stalled with over 400,000 individuals dying annually. A majority of these deaths occur in young children under the age of five [1]. Severe Malaria (SM) affects multiple organ systems, with frequent respiratory, cardiac, renal and neurological manifestations. Cardiovascular abnormalities in SM have long been appreciated [2] and take a multitude of forms [3][4][5][6][7]. Systemic inflammatory effects due to cytokine release, as well as extensive sequestration of the parasitized red blood cells causing microvascular obstruction in the coronary vessels, are likely contributors to the reported hemodynamic abnormalities [8,9].
In the 'Mortality after fluid bolus in African children with severe infection' (FEAST) trial, significantly increased mortality was noted in children who received 20-40 ml/kg of either 5% albumin or normal saline boluses [10]. Those with WHO defined shock at the time of randomization had a substantially increased absolute mortality risk (28%) with fluid bolus therapy. On later analysis, the excess mortality due to fluid bolus was reported to be caused by cardiogenic shock in all subgroups, including those with severe malaria [11]. This conclusion was based on the presence of clinical signs of shock at the point of demise. No objective measurements of cardiovascular function were performed in the FEAST trial. It remains unclear if previous studies that measured cardiovascular parameters directly support the theory that cardiovascular failure is a major contributor to mortality in SM either before or after fluid bolus administration.
We therefore undertook this systematic review of previously published manuscripts describing cardiovascular parameters measured directly during SM infection in pediatric and adult patients. Our aim was to summarize the results in an attempt to improve the overall understanding of the effects of SM on the cardiovascular system.

Methods
This review was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [12].

Data sources and Search Strategy
A search was undertaken of Medline and Embase for studies published up until Sept 2019. Search terms included a combination of the following: 'malaria,' 'cardiac,' 'cardiovascular,' 'hemodynamic,' 'echocardiography,' 'ultrasound,' 'heart' and 'myocardial.' Relevant references cited in eligible studies were also sought.

Study Selection
Studies involving either pediatric or adult patients with severe malaria and documented echocardiography or other hemodynamic monitoring parameters published in English were considered. SM was defined as per World Health Organization (WHO) definition [13]; presence of plasmodium falciparum asexual parasitemia and no other confirmed cause for the patient's symptoms or signs, accompanied by one or more clinical or laboratory features. Papers reviewing the pharmacodynamics of or cardiotoxicity from anti-malarial drugs or vaccines were excluded. Studies related to malaria in pregnancy or those involving patients with known cardiac disease were also not included. Studies whose primary focus was on myocardial infarction/heart failure in adults with SM were excluded given the high probability that the results were secondary to cardiovascular complications in the setting of coronary disease exacerbated by fever and anemia rather than a direct effect of malaria on the cardiovascular system. Those studies describing the effect of malaria on the cardiac rhythm or electrocardiogram were outside the scope of this review.

Data extraction and synthesis
Data extraction was performed using a standardized form by one review author (GW). The following data was retrieved: author details, year and journal of publication, patient demographics, clinical cardiovascular and echocardiography findings, and results of any invasive cardiac output monitoring including Pulse Contour Cardiac Output (PiCCO) or Pulmonary Arterial catheters (PAC). The findings were then placed into broad categories, including pediatric or adult (some of which included a small number of children) studies and separated by parameters studied.
The methodological quality of the studies included was not assessed.

Study Findings: Pediatric
A total of seven studies included in the review evaluated pediatric patients only. All took place in Sub-Saharan Africa. The majority of cardiovascular parameters evaluated are available in Table 2.

Significant findings from included pediatric studies related to preload
Multiple studies in pediatric patients demonstrated indices indicative of decreased preload in infected patients: • Left ventricular end-diastolic diameter index (LV-EDDI), which can be a marker of preload, was slightly lower upon presentation (53.15 mm/m 2 ) than at follow up (53.20 mm/m 2 , p = 0.028) [28]. • Left ventricular end diastolic diameter (LVEDD) tended to be lower on admission (3.17 cm ± 0.4) and improved with time (3.27 cm ± 0.33 on discharge, p = 0.47) [21].
º Normative values for LVEDD for healthy children are 3.5 cm-5.6 cm. • The inferior vena cava collapsibility index (IVCCI), an indirect marker of preload, was higher for all children on admission (43.8 ± 19.5), than at the time of discharge (21.7 ± 9.8) and was significantly worse in children with acidosis (52.1 ± 21.9) [21].

Significant findings from included pediatric studies related to cardiac function
Markers of cardiac function were variable in different studies, using heterogenous timepoints, markers and indices of function: • Left ventricular ejection fraction (EF) was preserved in all participants [20,25].
º There was no difference between acidotic and non-acidotic patients on admission or discharge º Those with hypotension upon presentation were not studied [21]. • 100/104 had normal EF upon admission [34] • Left ventricular end systolic volume index (LV-ESVI), reflecting global left ventricular function, was comparable on day 0 (21.7 ml/m 2 ± 7.1) to the follow up LV-ESVI on day 42 (22.0 ml/m 2 ± 7.4) [28]. • Cardiac index (CI), stroke index (SI) or velocity time integral (VTI), all surrogates for stroke volume, were all markedly increased during the disease process and decreased after recovery [28,34].   [34]. � CI 6.4 l/min/m 2 with SMA vs 5.4 l/min/m 2 without SMA, p ≤ 0.001. LVEDDI 57 mm/m 2 in SMA vs 51.5 mm/m 2 without SMA, p ≤ 0.001 [28]. � Authors concluded their findings suggested an increased cardiac output occurred from a raised stroke volume, with subsequent normalization following transfusion and antimalarial treatment. • No evidence of septal flattening in systole or diastole [25], indicating normal pressures of both the right and left ventricle.

Significant findings from included pediatric studies related to afterload
While no study directly measured SVR or other parameters of afterload, mean arterial pressure may have been lower in more ill children or those earlier in the time course: • Mean arterial pressure (MAP) was lower in children with SMA than in those with SM [34].

Significant findings from included pediatric studies related to pulmonary artery pressures
Pediatric patient pulmonary arterial pressures were higher than control patients: • Janka reported an increase in pulmonary arterial pressures (mean PAP = 31 mmHg) in comparison to controls (mean PAP = 21 mmHg) [20].
º No potentially causative left ventricular dysfunction was seen. • Murphy stated that no right ventricular enlargement was seen in her patients, and only three of 26 had trace tricuspid regurgitation [25].

Significant findings from included pediatric studies related to structural alterations
No studies mentioned the presence of pericardial effusions or other structural abnormalities.

Significant findings from included pediatric studies related to cardiac biomarkers
Troponin subtypes varied from normal to elevated in some studies, while BNP and NT-proBNP levels were higher in the two pediatric studies that followed it: • Troponin T and CK-MB levels were the same as in control patients [20].
• Forty-eight percent of all children (n = 50) had elevated levels of Troponin I (cTnI).
º NT-proBNP improved prior to discharge. º Normative values of BNP in healthy children are < 100 pg/mL.
• Nineteen percent of all children (n = 20) had mildly elevated levels of BNP at Time 0 [34].

Study findings: Adults
Sixteen studies included in the review evaluated a majority of adult patients. Studies took place worldwide. Table 3 reports studies describing echocardiography findings in patients with myocarditis, pericarditis, myocardial ischemia and cardiac dysfunction thought to be primarily related to SM. Table 4 represents selected clinical and measured cardiovascular findings in adult studies reporting invasive measurements of cardiac output.

Significant findings from included primarily adult studies related to preload
Adult studies of preload found varying markers: • Fluid loading resulted in a rise in cardiac index and oxygen delivery [26] with a mean change of 0.75 l/min/m 2 in CI (-.41 to 1.1) and 26 ml/min/m 2 (-2 to 54) in DO 2 .
º Normative values are >680 ml/m 2 . º There was no correlation betwe e n the baseline CVP and the likelihood of the CI being volume responsive, nor between the change in CVP and change in CI with fluid loading [22].

Significant findings from included primarily adult studies related to cardiac function
A majority of adult studies found some degree of cardiac dysfunction or low end of normal, although some studies found higher cardiac function: • EF was marginally reduced in those with circulatory failure and pulmonary edema (56.34 ± 1.04 all patients) compared to those without (59.11 ± 1.12 p = 0.01) [30].
º Nine of 28 SFM cases had cardiac involvement and was found to be more common than in P. Vivax (p < 0.001). • Hanson reported the median CI for the entire cohort was 3.1 (2.27-5.24 l/min/m 2 ) and was thus considered normal although 10 patients had a CI less than 3 [22]. • When using an expired gas collection model, CI was increased in SM, 4167 ml/min/m 2 median (3564 to 4876 ml/min/m 2 ) [36].

Significant findings from included primarily adult studies related to afterload
Studies had conflicting results with regards to measures of afterload in adult patients: • Mortality was higher in the 50% of patients in the Hanson's series who developed severe generalized edema. The GEDVI fell in the first 24 hours in all patients who died despite continuing fluid administration [29].
º There was no significant difference between the CVP of the patient who did and did not have pulmonary edema on admission. There was no relationship between the baseline CVP and the volume of fluid that was required to resuscitate the patient. • Significantly elevated systemic vascular resistance index (SVRI) was found in SM patients [23].
Significant findings from included primarily adult studies related to afterload pulmonary artery pressure (PAP) • Mild TR and mild pulmonary arterial hypertension (PAH) identified [33].
• PVR was found to be low and persistent throughout the study [14].

Significant findings from included primarily adult studies related to structural alterations
Pericardial effusion was a rare finding in adult studies: No patient had any evidence of pericardial effusion or regional hypokinesia [30].
• One patient of 27 in one study [33] and two case reports [32] of SM induced pericarditis had evidence of a pericardial effusion.

Significant findings from included primarily adult studies related to cardiac biomarkers
Cardiac biomarkers also varied between studies: • Herr demonstrated that Troponin T was not raised in malarial cases compared to healthy controls [23]. • H-FABP and myoglobin were almost twice those of controls in patients with SM (H-FABP: 1.9 ng/ml ± 1.1, and in controls 1.1 ng/ml ± 1.6, p < 0.001. Myoglobin 43.6 mg/l ± 12.5 and in controls 27.8 ±15.0, p =< 0.001) [24].
º NT-proBNP and CK-MB were not significantly elevated [23]. • Cardiac markers, both Troponin-I and CK-MB were increased in 14% cases and were found normal in 3 out of 17 patients who presented with cardiovascular involvement [30].

Discussion
Improved understanding of the cardiovascular effects of SM is an imperative step in the identification of appropriate adjunctive therapies that may reduce the morbidity and mortality of this global health menace. The results of this systematic review highlight the heterogeneity of reported effects of severe malaria on the cardiovascular system. Numerous reasons may account for the variability in the reported results including the relatively small number of patients evaluated as well as differences in approaches and parameters measured. Suggestion of a reduced preload state was present in two pediatric studies and one adult study. This would fit the common clinical presentation and suspicion of hypovolemia, especially in children who are commonly unwell for a few days prior to admission with vomiting, fever and a reduced intake. Through the use of tracer dilution methods and bioelectrical impedance, Planche et al. demonstrated that mild dehydration was often present in children with SM (mean (SD) depletion of TBW of 37(±33) ml/kg) [37]. Maitland et al. recruited patients with evidence of compensated shock (tachycardia, prolonged capillary refill and low central venous pressure (CVP) < 5cmH 2 O) and reported an improvement in haemodynamics and acidosis with administration of fluid bolus, concurrently with an increase in CVP [38]. However, other studies do not demonstrate significant preload loss and with the minimal literature available, it is difficult to make definitive conclusions, except that it may be present in some SM cases.
Cardiac systolic function in both pediatric and adult studies was demonstrated to be 'normal' or 'supranormal' in a number of studies. However, the finding that there are 'normal' cardiac functional parameters does not mean that there is an adequate cardiac output in the face of severe anemia and cytokine induced capillary leak. In two pediatric studies there appeared to be an increase in cardiac index, as expected in those with severe anemia, predominately produced by a rise in stroke volume. A significant proportion of the remaining studies performed echocardiography after the initial resuscitation period. It is therefore difficult to determine what overall effect fluid loading had on cardiac function and if it was normal or reduced prior to the intervention in these studies.
Presumably due to sequestration in the myocardial vessels with secondary ischemic changes, myocarditis with diminished cardiac function is not uncommonly reported in adults with SM [19,39,40]. Similar findings have not clearly been reported in pediatric patients except one included case report [27]. Cardiac biomarkers are largely normal or at the most, modestly elevated, with return to normal following resolution of the acute infection in the presented studies. Myocardial oxygen demand is high during acute SM due to tachycardia, and thus elevations in these values may occur secondary to myocardial stress rather than any significant myocardial injury.
The effect of SM on afterload is unclear as the results evaluating this in the available literature are mixed. In a limited number of pediatric patients, blood pressure tended to increase from admission to follow up in SM patients, perhaps indicating initial reduced afterload. Other adult trials also suggest low to normal SVR. This is in comparison to evidence for increased systemic vascular resistance in two different adult studies. The interaction of the plasmodium infected erythrocytes with the endothelium has been extensively studied. Intuitively, pro-inflammatory cytokine release known to occur in SM should result in reduced systemic vascular resistance [13,14,28,41]. Alternatively, extensive microvascular and widespread obstruction from parasitized and sequestered red blood cells may result in increased SVR [22]. At an individual level, variations in the overall severity and impact of each of these pathophysiological processes may occur in patients with SM. If this is the case, this may explain the differing results of available studies. It is important to note that anti-malarials such as intravenous quinine, are commonly considered cardiodepressants, and their use may be at least partially responsible for some of the cardiovascular effects reported in the literature before artesunate became standard of care.
Approximately 10-15% of children and 10% of adults with severe malaria present with shock [41][42][43][44][45], and those that do have a very high mortality [46]. These patients have traditionally been managed with fluid resuscitation and vasoactive medications to support failing haemodynamics. Additionally, adults have a propensity towards ill-defined and frequently fatal pulmonary edema during the management of SM [29,30,47,48]. Such pulmonary edema associated with a mortality rate of 80% in resource limited settings [48]. One would expect these apparently preload deficient and hypoperfused patients, with mostly normal cardiac function and variable afterload, to respond well to fluid therapy. Indeed, the cardiac function of these shocked patients may not be considered normal in the face of SM and they could actually have myocarditis. It is clear, however, that fluid loading increases mortality in children despite improving perfusion at one hour [10,11] and recent re-analysis of the data discovered that the detrimental effect on mortality risk persisted for up to 4 days post randomization [49]. Fluid loading also increases the risk of pulmonary edema in adults and is associated with worse outcomes [50]. The overall results of the available studies presented here, that directly measure myocardial function, do not support clear mechanisms by which a poor response to fluid resuscitation would occur.
Future research needs to be done in larger numbers of patients meeting strict and uniform inclusion criteria with the same instruments of measurements and parameters recorded. In order to conclusively understand the effect of SM on the cardiovascular system, cardiac function needs to be closely and quantitatively tracked over the disease duration.

Conclusion
Available studies of children and adults with SM report variable changes in preload, myocardial contractility, and systemic vascular resistance. It remains unclear if there is legitimate heterogeneity in these measurements across a cohort of patients with SM or if findings are limited by patient numbers and disparate approaches to and timing of measurements. Larger, more detailed studies are required to explore and understand the cardiovascular abnormalities in this multi-systemic high burden disease, including if there is a subset of SMA patients with under-recognized cardiac dysfunction despite normal indices.

Additional File
The additional file for this article can be found as follows: • Supplementary Material. PRISMA-P checklist. DOI: https://doi.org/10.5334/gh.789.s1

Funding Information
This research received no specific grant from any funding agency in the public, commercial or not for profit sectors.

Author Contributions
NO was responsible for study conception. GW undertook the systematic review, extracted data and completed the first draft of the manuscript. GW, ND, YC and NO were involved in drafting and editing the final manuscript.