The Tulane University Institutional Review Board and the Sierra Leone Ethics and Scientific Research Committee approved this project. Patients were referred to the Kenema Government Hospital (KGH) Lassa Ward from the hospital’s general ward or from regional health centers on the basis of suspicion of Lassa fever. Patients who met the case definition of Lassa fever as defined by Khan et al.  were admitted and cared for by the ward’s trained staff. After the initial cases of Ebola were detected, patients were referred if they presented with an illness that met the World Health Organization case definition for Ebola. We obtained samples using the collection and processing protocols at Kenema Government Hospital under the emergency-response guidelines established by the Sierra Leone Ministry of Health and Sanitation. Diagnostic tests for the presence of Ebola virus (EBOV) were performed on site by means of quantitative reverse-transcriptase–polymerase chain reaction assays with the use of the SuperScript III One-Step RT-PCR System with Platinum Taq DNA Polymerase (Life Technologies).
Serum processing for metabolomics analyses
Small blood volumes (approximately 5 mL) for serum separation were collected from patients presenting to KGH with febrile illnesses that met preclinical criteria of suspected Lassa fever or Ebola. Patient samples received a coded designation and were collected in serum vacutainer tubes. Blood samples were allowed to coagulate for 20 min at room temperature. Serum was separated from coagulated blood by centrifugation (200×g, 20 min at room temperature). For subjects for which there was excess serum not needed for clinical evaluations, aliquots of the serum fraction were stored in cryovials at − 20 °C prior to processing for metabolite analysis.
Serum metabolite analysis was performed as previously described . Briefly, serum samples were depleted of protein by addition to one part sera (100 μL) of 4 parts ice-cold methanol (400 μL), the mixture was vortexed vigorously for 10 s, and incubated 1 h at − 20 °C followed by centrifugation at 14,000×g, 15 min, 4 °C. The supernatant was collected and transferred to a new, sterile vial and dried under vacuum. The resultant smal l-molecule containing pellets were stored in desiccated, sealed containers and shipped to Tulane University where they were gamma-irradiated. Small molecule containing pellets were dissolved in a solution of 95:5 water:acetonitrile transferred to autosampler vials, and held at − 20 °C or 4 °C immediately prior to analysis. All reagents utilized were high-pressure liquid chromatography (HPLC) grade.
Liquid Chromatography Mass Spectrometry
The Liquid Chromatography Mass Spectrometry (LCMS) method was performed as previously described with minor changes . Briefly, detection of metabolites was performed via HPLC separation with ESI–MS (electrospray mass spectrometry) detection. HLPC was performed with an aqueous norma l-phase, hydrophilic interaction chromatography HPLC column: a Cogent Diamond Hydride Type-C column with 4 μm particles and dimensions of 150 mm length and 2.1 mm diameter or an Agilent Zorbax 300-SB-C18 column with 3.0 μm particles and dimensions of 150 mm length and 0.3 mm diameter was used with an Agilent 1290 HPLC system (Agilent Technologies, Santa Clara, CA). The column were maintained at 60 °C with a flow rate of 900 μL/min. Chromatography was as follows: solvent consisted of H2O with 0.1% (v/v) formic acid for channel “A” and acetonitrile with 0.1% formic acid for channel “B”. Following column equilibration at 98% B, the sample was injected via autosampler, and the column was flushed for 2.0 min to waste. From 2.0 min to 14.5 min, the gradient was linearly ramped from 98 to 65% B. From 14.5 min to 16.0 min, the gradient was ramped from 65 to 25% B. From 14.5 to 18.0 min the column was held at 25% B, and from 18.0 to 18.2 min the gradient was ramped from 25 to 98% B. From 18.2 to 20.0 min the column was re-equilibrated with 98% B. An Agilent 6538 Quad-Time of Flight with dua l-electron spray ion source mass spectrometer was used for all analyses. Resolution was approximately 20,000 and accuracy was 1 ppm. Source parameters: drying gas 12 L/min, nebulizer 60 psi, capillary voltage 3500 V, capillary exit 100 V. Spectra were collected in positive mode from 50 to 1700 m/z at a rate of 1 Hz.
Authenticated standards of synthetic platelet-activating factor (PAF) C-16 (#878110) and lysoPAF C-16 (#878119) at a concentration of 5 mg/mL in chloroform were purchased from Avanti Polar Lipids (Birmingham, AL). l-threonine (T8625) was purchased from Sigma-Aldrich. The molecules were diluted in 95:5 water:acetonitrile solution and analyzed with the identical method for metabolite detection.
Data analysis and visualization
Raw spectral data in.d format where uploaded to XCMS Online (Versions 2.3.0 or 2.2.3) and processed as pairwise comparisons using parameters optimized for data acquired with UPLC on an Agilent 6538 MS.
Statistics and machine learning
Statistical analyses were carried out using the R statistical software package or Graphpad Prism. Multiple comparisons were performed by Analysis of Variance (ANOVA); p < 0.5 was considered significant. Raw mass spectral intensity values and a unique identifier for specific spectral features were extrapolated from XCMS output and compiled into.csv file types.