Relationship Between Viremia and Specific Organ Damage in Ebola Patients: A Cohort Study

Abstract

Background

Pathogenesis of Ebola virus disease remains poorly understood. We used concomitant determination of routine laboratory biomarkers and Ebola viremia to explore the potential role of viral replication in specific organ damage.

Methods

We recruited patients with detectable Ebola viremia admitted to the EMERGENCY Organizzazione Non Governativa Organizzazione Non Lucrativa di Utilità Sociale (ONG ONLUS) Ebola Treatment Center in Sierra Leone. Repeated measure of Ebola viremia, alanine aminotransferase (ALT), aspartate aminotransferase (AST), bilirubin, creatine phosphokinase (CPK), lactate dehydrogenase (LDH), activated prothrombin time (aPTT), international normalized ratio (INR), creatinine, and blood urea nitrogen (BUN) were recorded. Patients were followed up from admission until death or discharge.

Results

One hundred patients (49 survivors and 51 nonsurvivors) were included in the analysis. Unadjusted analysis to compare survivors and nonsurvivors provided evidence that all biomarkers were significantly above the normal range and that the extent of these abnormalities was generally higher in nonsurvivors than in survivors. Multivariable mixed-effects models provided strong evidence for a biological gradient (suggestive of a direct role in organ damage) between the viremia levels and either ALT, AST, CPK LDH, aPTT, and INR. In contrast, no direct linear association was found between viremia and either creatinine, BUN, or bilirubin.

Conclusions

This study provides evidence to support that Ebola virus may have a direct role in muscular damage and imbalance of the coagulation system. We did not find strong evidence suggestive of a direct role of Ebola virus in kidney damage. The role of the virus in liver damage remains unclear, but our evidence suggests that acute severe liver injury is not a typical feature of Ebola virus disease.

On 9 June 2016, the World Health Organization (WHO) officially declared the end of the Ebola virus disease (EVD) epidemic, which caused 28616 confirmed cases with 11310 deaths [1]. During this outbreak, logistical difficulties, lack of political will, and weak coordination between partners prevented the rapid implementation of scientifically sound prospective clinical and pathogenesis studies [2, 3]. Indeed, good-quality trials to assess efficacy of novel treatments [4] and vaccines [5] for EVD came late when the epidemic had already claimed most of its victims.

Understanding clinical aspects of the pathogenesis of EVD may serve to better define targets for developing new therapeutic, preventive, and disease-monitoring strategies [6]. The only defined biomarker unequivocally associated with patients’ outcome is the level of Ebola virus (EBOV) RNA in the blood (viremia) [7–9]. However, marked biochemical abnormalities, reflecting the potential capability of the virus to produce damage in different body compartments, have been observed [10, 11]. Nevertheless, studies exploring the relationship between EBOV viremia, biochemical abnormalities, and specific clinical disorders are scanty. Using prospectively collected repeated measures, we conducted an analysis to explore the potential association between the EBOV viremia and biomarkers reflecting skeletal muscle damage, liver damage, renal impairment, and coagulation abnormalities.

METHODS

Study Design and Aims

The study was based on a cohort of patients with confirmed EVD. The analyses were designed to assess the presence of a biological gradient between EBOV viremia and biomarkers of skeletal muscle damage, liver damage, renal impairment, and coagulation abnormalities.

Clinical Setting and Patient Care

The study was set in the EMERGENCY Organizzazione Non Governativa Organizzazione Non Lucrativa di Utilità Sociale (ONG ONLUS) Ebola Treatment Center (EMERGENCY-ETC), which was operational between December 2014 and May 2015, in Goderich (Sierra Leone). This center offered advanced supportive care to patients and laboratory monitoring through an internal laboratory, managed by the Istituto Nazionale Malattie Infettive (INMI) Lazzaro Spallanzani, where highly infectious biological samples were processed for molecular diagnosis of EVD and standard biochemical assays. Patients were treated with supportive care using a standardized protocol, described elsewhere [7]. In general, blood collection (for EBOV molecular testing and standard biochemical assays) was performed within 6 hours since patient admission and eventually once a day during the acute phase of the disease. Patients with confirmed EVD who recovered were discharged after 2 consecutive undetectable EBOV RNA in blood samples taken at least 2 days apart.

Eligibility Criteria

All patients with confirmed EVD (WHO definition [12]) admitted to the EMERGENCY-ETC between 13 December 2014 and 30 May 2015 who had at least 1 available quantifiable EBOV viremia were enrolled in this study.

Variables Studied

For each patient, we collected information about age, sex, dates of renal replacement therapy, clinical outcome (either nonsurvivors or survivors), and repeated measurements of EBOV viremia, liver function tests, renal function tests, coagulation parameters, and muscle enzymes. In detail, we collected repeated measure of 10 biomarkers including:

• EBOV viremia in log₁₀ copies/mL (continuous variable);

• total bilirubin in mg/dL (continuous variable);

• international normalized ratio (INR) (continuous variable);

• activated prothrombin time (aPTT) in seconds (continuous variable);

• creatinine in mg/dL (continuous variable);

• blood urea nitrogen (BUN) in mg/dL (continuous variable);

• alanine aminotransferase (ALT) in U/L (right-censored variable with upper limit at 1000 U/L);

• aspartate aminotransferase (AST) in U/L (right-censored variable with upper limit at 1000 U/L);

• creatine phosphokinase (CPK) in U/L (right-censored variable with upper limit at 2000 U/L);

• lactate dehydrogenase (LDH) (right-censored variable with upper limit at 4000 U/L).

ALT, AST, CPK, and LDH are considered as right-censored variables because, to minimize risk of accidental exposure of laboratory workers, samples with values exceeding methods upper limit of quantification were not always diluted to obtain a quantitative measure. Thus the measures of these biomarkers were recorded either as a punctual point measure (ie, samples below the upper limit of quantification and those exceeding the limit but reanalyzed after dilution) or as right-censored measure (ie, undiluted sample above the upper limit of quantification).

Laboratory Methods

Blood samples were collected by trained doctors and nurses according to recommended safety and infection control precautions [13]. Samples were not inactivated before being tested.

AST, ALT, LDH, total bilirubin, CPK, BUN, and creatinine were measured by SpotChem EZ clinical chemistry analyzer (Woodley Equipment Company Ltd). INR and aPTT were measured by a Hemochron JR Signature plus machine (Whitmire Medical Ltd). EBOV RNA testing was performed using a real-time reverse-transcription polymerase chain reaction (RT-PCR) assay (RealStar Filovirus Screen RT-PCR 1.0 kit, Altona Diagnostics), with a limit of detection of 3.11 log₁₀ copies/mL of EBOV RNA. Viral RNA quantification was based on a standard reference curve provided by the kit producers, spanning up to 9 log₁₀ copies/mL of EBOV RNA.

Statistics and Modeling

Unadjusted analysis to compare survivors and nonsurvivors was carried out to describe the study sample and to show the distribution of baseline patient characteristics between survivors and nonsurvivors. Distribution free statistics including Kruskal-Wallis (for continuous variables) and Pearson χ² (categorical variable) were used to assess significant differences between survivors and nonsurvivors. Right-censored variables were transformed into 3-level categorical variables (level 1, within normal range; level 2, above normal but below censoring cutoff; level 3, above the cutoff of censoring).

Association between biomarkers and EBOV viremia was assessed in 9 separate multilevel mixed-effects (MME) regression models using concomitant determinations. We considered a biomarker’s determination to be concomitant to and EBOV viremia determination when the 2 determinations were made no more than 1 day apart from each other. All MME regression models were set for allowing for random intercept at patient level, random slope at EBOV viremia level, and unstructured variance-covariance matrix. Each biomarker served as the dependent variable for 1 model only. The same set of independent variables (EBOV viremia; clinical outcome; dialysis; sex and age) was used in all the MME regression models. Continuous dependent variables (bilirubin, BUN, creatinine, INR, and aPTT) were assessed by MME linear regression models [7, 14]. Right-censored dependent variables [15, 16] (ALT, AST, LDH, and CPK) were assessed by using MME interval regression models. MME interval regression is a generalization of other censored regression estimators [17], such as Tobit [18], and can be used when the dependent variable is measured as point data, interval-censored data, left-censored data, or right-censored data [17].

Estimates for intercept (baseline) and independent variables coefficients were optimized by using the natural-log transformed dependent variables. Fixed-effects measures of the variation of dependent variables (biomarkers) according to the level of exposure to independent variables were reported in back-transformed form and interpreted as proportional variation from the baseline (according to standard interpretation for log-transformed measures) [19] for providing readers with biomarker measures conventionally used in medical practice. Full model parameters (in natural log form) including fixed- and random-effects coefficients were reported in the Supplementary Data.

A biological gradient suggestive of a potential direct role of EBOV viremia in alteration of individual biomarker was considered present if the P value for log-linear association was <.050.

A further analysis was carried out to assess the potential association between CPK levels and creatinine levels (as proxy of renal damage). In this analysis, CPK was added as a 5-level independent categorical variable (according to Common Terminology Criteria for Adverse Events [CTCAE], version 4.03) [20] to the MME linear model for association between creatinine and EBOV viremia levels.

All analyses and plots were implemented by Stata version 13.1 statistical package.

RESULTS

Descriptive and Unadjusted Analysis to Compare Survivors and Nonsurvivors

One hundred six patients with confirmed EVD diagnosis were admitted to the Center. Of these, 6 were excluded from the study for the following reasons: 2 were referred to the Center without an available EBOV viremia result and died soon after arrival (with no additional testing), and 4 had already unquantifiable level of EBOV viremia at arrival. Overall, 100 patients were included in the analysis.

Median time between admission and outcome, either EBOV viremia clearance or death, was 8 and 4 days, respectively. Unadjusted analysis to assess the distribution of patients’ characteristics according to clinical outcome is shown in Table 1. There was no difference between survivors and nonsurvivors regarding sex and age. The analysis provided good statistical evidence that peak levels of EBOV viremia and those of all biomarkers assessed, apart from LDH, were significantly lower in survivors than in nonsurvivors (P < .050).

Biological Gradient Between Ebola Virus Viremia and Biomarker Levels

MME regression models made on concomitant determinations of EBOV viremia and biomarker levels provided strong statistical evidence for the presence of a biological gradient between EBOV viremia and the levels of AST, ALT, CPK, LDH, INR, and aPTT, suggesting that EBOV can directly affect skeletal muscle tissue, coagulation system, and the liver (Table 2 and Figure 1). Proportional increases per 1 log₁₀ EBOV viremia were: AST, 1.67 (95% confidence interval [CI], 1.46–1.90; P < .001); ALT, 1.24 (95% CI, 1.12–1.37; P < .001); CPK, 1.21 (95% CI, 1.08–1.37; P = .001); LDH, 1.27 (95% CI, 1.09–1.47; P = .002); INR, 1.06 (95% CI, 1.03–1.09; P < .001); and aPTT, 1.12 (95% CI, 1.08–1.17, P < .001). The analysis did not provide any evidence for a direct linear relationship (suggestive of biological gradient) between EBOV viremia levels and either creatinine (P = .973), BUN (P = .205), or bilirubin (P = .741) values.

In addition to the above reported findings, other significant associations were observed (Table 2). In particular: (1) patient outcome was associated with the levels of all biomarkers (P < .050), apart from CPK and LDH (P = .104 and P = .218, respectively); (2) renal dialysis was associated with the level of CPK (P < .001), LDH (P = .043), bilirubin (P = .006), INR (P = .001), and aPTT (P = .001); (3) sex was associated with creatinine (P = .037) and bilirubin (P = .020) levels; and (4) age was associated with INR (P = .011) and creatinine (P = .025) levels.

Association Between Creatinine and Creatine Phosphokinase Levels

To explore the potential role of CPK levels on renal damage, we set a further model by including CPK as 5-level categorical variable in the MME linear regression model already used to assess the association between creatinine and EBOV viremia level.

Table 3 shows results of the MME linear regression model to assess the proportional variation of creatinine according to CPK levels. This analysis highlighted a significant positive association between CPK and creatinine levels (P < .001) independent from EBOV viremia, age, sex, renal dialysis, and clinical outcome. Due to the laboratory methodology for CPK determination with cutoff at 2000 U/L, we could not use CPK values as a continuous independent variable; therefore, the grading according to CTCAE version 4.03 was adopted. Nevertheless, the variation of creatinine according to the grade of CPK strongly suggests a linear relationship between these 2 biomarkers (Figure 2), which is consistent with a biological gradient.

DISCUSSION

To our knowledge, this is the first study to explore the association between the level of EBOV viremia and biomarkers of specific organ damage. This association could be hypothesized from previous evidence showing that EBOV can be replicated in several different body compartments [21].

Previous studies have shown that viremia level is the strongest predictor of clinical outcome in patients with EVD [7–9, 22], suggesting a direct role of EBOV in the host tissue damage. However, the pathogenesis of renal and liver function impairment, coagulation disorders, and muscle damage remains poorly understood. In this study, we shed light on EBOV pathogenesis using data from real clinical practice collected during the 2014–2015 EBOV epidemic in Western Africa. There are several important findings from our study.

First, unadjusted analyses to assess distribution of peak values of biomarkers between survivors and nonsurvivors confirmed that the highest EBOV viremia levels are strongly associated with unfavorable clinical outcome and that EVD is characterized by a systemic syndrome with a variable degree of clinical severity [23, 24]. In fact, we provide evidence that all the assessed biomarkers were above the normal range but also that the observed abnormalities were much more pronounced in nonsurvivors than in survivors. Similar metabolic alterations either in survivors or nonsurvivors were reported in a clinical study that extensively assessed biomarkers in a patient infected with Sudan Ebola virus [25, 26].

Second, the models set using repeated biomarkers measures, provided strong evidence that a biological gradient exists between EBOV viremia and the levels of ALT, AST, LDH, CPK, aPTT, and INR, suggesting that the virus may be directly involved in the tissue damage leading to biochemical alterations [27]. The notion that Ebola infection may have a major impact on the coagulation system is old [28]. Though significant hemorrhage was infrequently reported in the recent West Africa EBOV epidemic [29], clinical studies including extensive monitoring of coagulation system provided evidence that coagulation imbalance was frequent and associated with unfavorable clinical outcome even in absence of evident hemorrhage [30]. In line with these findings, our models predicted that INR and aPTT were mildly to moderately elevated in patients with EVD. Moreover, the levels of INR and aPTT were significantly associated with patient outcome and showed a direct biological gradient with EBOV viremia level. These observations suggest that EBOV may directly affect the balance of coagulation pathways and that coagulation parameters may have a value as a potential prognostic indicator of disease severity in patients with EVD. The hypothesis that EBOV may have a direct role in skeletal muscle damage has been already suggested in previous studies and is consistent with the frequently reported symptoms of myalgia at disease onset and persistent muscular weakness in survivors. Hunt et al [11] found that median value of CPK was 1949 U/L (ie, about 5 times above the normal) value with AST:ALT ratio >2 in a cohort of 118 patients with confirmed EVD in Sierra Leone. Similar results were reported by Cournac et al [10], who analyzed a smaller cohort of 22 patients in Guinea. Both studies also suggested an association between CPK and either patient outcome or the degree of the renal function impairment, but neither of them implemented multivariable analyses to explore potential co-factors. Our study confirms previous results and provides new evidence to support the hypothesis that EBOV is directly involved in the muscular tissue damage. In fact, our predictions emphasized that CPK levels were always above normal levels in patients with detectable EBOV viremia and that a strong biological gradient exists between EBOV viremia and CPK levels, suggesting that viral replication may have a direct role in muscular damage. The hypothesis that EBOV can cause rhabdomyolysis is also supported by the evidence that a biological gradient exists between EBOV viremia and the level of other biomarkers possibly associated with muscular tissue damage, such as LDH and AST.

Third, our models found no evidence for a biological gradient between the levels of biomarkers of kidney damage (creatinine and BUN) and EBOV viremia. However, similar to other recent prospective studies, [31] we still found that creatinine and BUN were strongly associated with clinical outcome. These findings suggest that kidney damage may be multifactorial in patients with EVD and that the virus-triggered mechanisms may be complex. One hypothesis is that kidney damage may be the consequence of rhabdomyolysis. In support of this hypothesis, we have here shown a strong direct association between the degree of muscular damage and the level of creatinine which is also independent from the EBOV viremia. This finding is in line with postmortem analyses carried out in patients with EVD and other filovirus infections, showing evidence of acute tubular necrosis with no significant inflammation, suggesting that damage through myoglobin due to rhabdomyolysis is a possible mechanism [24, 32]. Nevertheless, the values of CPK we estimated, even at the highest level of EBOV viremia, were not consistent with those expected for acute renal failure due to rhabdomyolysis, suggesting that coexisting conditions such as dehydration and acidosis, nearly always present in African EVD patients, may play a role as co-factors of renal dysfunction [33]. Moreover, the hypothesis that EBOV does not have a direct pathogenetic effect on kidney tissues is indirectly supported by the observation that viable EBOV can be found in convalescent (asymptomatic) patients’ urine [34, 35]. Specific pathogenesis of EBOV on different body compartments could have been investigated, at best, through postmortem analysis [24]. However, performing autopsy during outbreaks is very challenging, due to the hazardous nature of EVD corpses. This emphasizes the need of including also pathologists and high biocontainment mobile cabinets in networks for preparedness and response to future outbreaks [24].

Finally, our analyses showed that most patients had raised concentrations of biomarkers suggestive of liver damage including ALT, AST, and bilirubin and that increased level of these biomarkers were significantly associated with clinical outcome. However, the mild elevation of bilirubin and the observation that AST levels were steadily much higher than those of ALT suggest that severe acute viral hepatitis is not a typical manifestation of EVD subsequent to EBOV [33, 36]. Similar evidence has been reported in clinical studies in patients infected either with Sudan Ebola or [29] Taï Forest Ebola [37] and in animal models investigating pathogenesis of Bundibugyo virus [38].

In conclusion, our study provides new evidence to support the hypothesis that EBOV may play a direct role in muscular damage and imbalance of coagulation system. The study did not provide unequivocal evidence about the pathogenesis of kidney damage. In fact, though it is possible that rhabdomyolysis may contribute to the kidney damage, it is also likely that other predisposing conditions should coexist. Finally, our study also demonstrates that merging classical epidemiological study design (ie, historical cohort) with state-of-the-art statistical techniques made it possible to tame the complexity of the data structure (heteroskedastic observations with censored measures) and, thus, to best exploit the information contained in the clinical datasets.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. Goderich ETC has been built by the Royal Engineers as proxy for the Department for International Development, UK Government (DFID), in cooperation with EMERGENCY’s Technical Division. The ETC’s operations have been cofunded by DFID and EMERGENCY’s private donations. The upgrading of the virology laboratory has been supported by a grant from the Italian Ministry of Foreign Affairs–Direzione General per la Cooperazione allo Sviluppo (DGCS). INMI’s research activities have been performed thanks to grants from the Italian Ministry of Health (Ricerca Corrente IRCCS and Ricerca Finalizzata). Deployment of laboratory personnel was possible thanks to a grant from the Italian Ministry of Foreign Affairs–DGCS. We thank the EMLab European consortium (IFS/2011/272–372; www.emlab.eu), which has been where we have developed the knowledge and experience that have allowed us to establish the virology laboratory at Goderich ETC. We thank the people who worked at the ETC in Goderich, Sierra Leone, and all members of the INMI-EMERGENCY EBOV Sierra Leone Study group.

Financial support. This work was supported by the Italian Ministry of Health, Ricerca Corrente IRCCS; Italian Ministry of Foreign Affairs; EMERGENCY’s private donations; and Royal Engineers for the UK DFID.

Potential conflicts of interest. All authors: No reported conflicts of interest. 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.

INMI-EMERGENCY EBOV Sierra Leone Study Group

Affiliated to National Institute for Infectious Diseases INMI “Lazzaro Spallanzani,” IRCCS, Rome, Italy: Mirella Biava, Angela Cannas, Roberta Chiappini, Sabrina Coen, Francesca Colavita, Germana Grassi, Daniele Lapa, Antonio Mazzarelli, Silvia Meschi, Claudia Minosse, Serena Quartu, Maria Beatrice Valli, Carolina Venditti Antonella Vulcano, Paola Zaccaro.

Affiliated to EMERGENCY, Milan, Italy: Umar Ahmad, Elisabetta Checcarelli, Michela Delli Guanti, Elena Giovanella, Davide Gottardello, Maurizio Guastalegname, Milos Jocic, Giorgio Monti, Clare Parsons, Nicola Rossi, Giampiero Salvati, Giovanna Scaccabarozzi, Erminio Sisillo, Paola Tagliabue, Marta Turella, Caterina Valdatta.

Giuseppe Ippolito (INMI) and Gino Strada (EMERGENCY) have authorized authorship on behalf of the Study Group.

References

Author notes