Patients: Participants with chronic HCV infection were recruited at two centres: Newcastle-upon-Tyne and Imperial College London. All participants gave written, informed consent and the study had ethical approval (Northumberland REC 07/H0902/45 and Fife and Forth Valley REC 07/S0501/21). The research was performed in accordance with the relevant guidelines/regulations set out by the Northumberland and Fife and Forth Valley research ethics committees, and was performed in accordance with the Declaration of Helsinki of 1975.
All participants were age ≥ 18 years, HCV-RNA positive for >6 months, and not on a lipid modulating agent for 3-months prior to the study. Patients with hepatitis B, hepatitis delta, or HIV co-infection, or alcohol dependency were excluded. All participants attended following a >8 hour overnight fast for sample collection. The fasted cohort consisted of 112 fasting sera (39 G3, 73 G1); 25% had compensated cirrhosis evidenced by Fibroscan >12.5 KPa (Echosens, Paris, France). Baseline clinical and demographic data are shown in Table 1.
Table 1
Clinical and laboratory characteristics of fasting cohort.
| Fasting samples cohort N =112 |
Post prandial status | Fasting >8 hours | P value |
HCV genotype | 1 | 3 | |
HCV Viraemic | yes | |
N | 73 | 39 | |
Age years | 48.3 ± 9.9 | 48.1 ± 10.6 | 0.918 |
Male (%) / female | 50 (68%) / 23 (32%) | 30 (77%) / 9 (23%) | |
BMI kg/M2 | 25.4 ± 4.0 | 25.3 ± 3.0 | 0.902 |
Fibroscan LSM KPa Median (Q1-Q3) | 7.35 (5.3 – 16.1) | 8.8 (6.5 – 16.4) | 0.143 |
% Cirrhosis (LSM ≥ 12.5 KPa) | 25% | 25% | NS |
ALT IU/L | 96.8 ± 80.5 | 117.2 ± 68.4 | 0.030 |
AST IU/L | 76.7 ± 63.4 | 91.2 ± 46.0 | 0.022 |
Total cholesterol mmol/L | 4.62 ± 0.95 | 3.74 ± 0.91 | <0.001 |
HDL cholesterol mmol/L | 1.26 ± 0.36 | 1.26 ± 0.45 | 0.953 |
Non-HDL cholesterol mmol/L | 3.36 ± 0.95 | 2.43 ± 0.82 | <0.001 |
Triglycerides mmol/L | 1.31 ± 0.68 | 1.01 ± 0.72 | 0.035 |
apoB g/L | 0.88 ± 0.26 | 0.64 ± 0.20 | <0.001 |
apoA1 g/L | 1.47 ± 0.29 | 1.41 ± 0.32 | 0.328 |
Fasting glucose mmol/L | 5.0 ± 0.69 | 5.44 ± 1.22 | 0.095 |
Fasting insulin µIU/mL | 8.07 ± 5.68 | 7.37 ± 4.11 | 0.783 |
HOMA-IR | 1.77 ± 1.52 | 1.88 ± 1.32 | 0.463 |
NEFA mM | 0.50 ± 0.04 | 0.54 ± 0.06 | 0.596 |
NEFA = non-esterified fatty acids; HOMA-IR = Homeostatic Model Assessment for Insulin Resistance |
In addition, a second cohort of non-fasted serum samples were obtained from the HCV Research UK Clinical Database and Biobank (Glasgow, UK) and comprised 150 treatment naïve chronic HCV patients (75 HCV-G1, 75 HCV-G3), matched to the fasted cohort for age, sex, body-mass index (BMI) and the presence of cirrhosis. A further 100 samples (50 HCV-G1, 50 HCV-G3) were obtained from the HCV Research UK Clinical Database and Biobank from individuals following a sustained virological response (SVR) after successful antiviral treatment (the SVR cohort).
Liver function tests and serum glucose measurements: Standard serum liver function test and serum glucose measurements were performed on the serum samples from all participants. Aspartate aminotransferase (AST) and alanine aminotransferase (ALT) and serum glucose were measured by standard biochemical methodologies using British National Health Service (NHS) laboratory protocols (https://www.england.nhs.uk/wp-content/uploads/2021/09/B0960-optimising-blood-testing-secondary-care.pdf).
Fasting lipid profiling. Fasting serum lipids were measured using standard enzymatic methods. Where appropriate, LDL cholesterol was calculated using the Friedewald equation. Apolipoprotein B concentrations were measured by automated rate nephelometric methods (BNII, Dade Behring Ltd, Milton Keynes, Buckinghamshire, UK). Insulin was measured by ELISA (Linco Research Inc, St Charles, Missouri, USA). Lathosterol, desmosterol, cholestanol and sitosterol were measured by gas-chromatography mass spectrometry, (GC-MS), as described previously by Kelley (18).
Phenotyping of body fat distribution. A subgroup of 13 consecutively-attending participants from the fasted cohort (6 HCV-G1, 7 HCV-G3) at Imperial College London had additional detailed clinical phenotyping performed by determination of whole body fat distribution using in vivo magnetic resonance spectroscopy (MRS) to quantify intra-hepatocellular lipid (IHCL), intra-myocellular lipids in tibialis (T IMCL) and soleus muscles (S IMCL), and distribution of adipose tissue fat (% visceral and non-visceral fat) using magnetic resonance imaging, as previously described in detail by Thomas and colleagues (19).
Ultra Performance Liquid Chromatography Mass Spectroscopy (UPLC-MS) lipidomics. All samples were thawed at 4⁰C and prepared for UPLC-MS analysis by isopropanol protein precipitation by addition of 150µL of cold isopropanol to each 50 µL serum sample (ratio 3:1), as previously described by Sarafian and colleagues in 2014 (20). Quality control (QC) samples were prepared by pooling equal volumes of all samples and injecting into the mass spectrometry system at regular intervals throughout the analytical runs, in order to define the system suitability, analytical stability, and sample repeatability. Serum lipid UPLC-MS profiling was performed using an ACQUITY UPLC system (Waters Ltd., Elstree, UK), coupled to a Q-ToF Premier mass spectrometer (Waters MS Technologies Ltd, Manchester, UK) using an electrospray (ESI) ion source operated in both positive and negative electrospray ionisation modes (ESI+ and ESI-).
Liquid chromatography (LC) conditions have been previously described by Eliasson and colleagues in 2012 (21). Separation was done in a Waters Acquity UPLC HSS CSH column (1.7 µm, 2.1 × 100 mm) maintained at 55⁰C. Mobile phases consisted of acetonitrile (ACN)/H2O (60:40) (A) and iso-propyl alcohol (IPA)/ACN (90:10) (B), both containing 10 mM ammonium formate and 0.1% (v/v) formic acid. The flow rate was set at 0.4 mL/min. Injection volume was 5 µL and 15 µL for positive (ESI +ve) and negative (ESI –ve) modes, respectively.
ESI conditions were as follows: capillary voltage for ESI- 2500V, for ESI +ve 3000 V, cone voltage 25 V for ESI -ve and 30V for ESI +ve, source temperature 120⁰C, desolvation temperature 400⁰C, cone gas flow 25L/h, desolvation gas 800L/h. Data were collected in centroid mode. For mass accuracy, leucine enkephalin (555.2692 Da calculated monoisotopic molecular weight) was used as a lock mass. Lock mass scans were collected every 30s and averaged over 3 scans to perform mass correction. Instrument calibration was performed using sodium formate prior to each ESI mode.
To equilibrate the system, ten conditioning QC samples were performed at the start of acquisition. QC samples were run periodically after 10 sample injections to monitor instrument performance. Data-dependent acquisition (DDA) and MSE analysis of the QC sample was performed to obtain MS/MS information for metabolite annotation. Candidate metabolites were annotated using accurate m/z values, fragmentation patterns, retention times, and the METLIN database (https://metlin.scripps.edu/).
Ms Data Pre-processing
The UPLC-MS raw data were acquired using MassLynx software version 4.1 (Waters, Manchester, UK) and converted in NetCDF files using Databridge; a module within MassLynx software 4.1. The CDF files were pre-processed using XCMS package within the R statistical software version (Rx64 3.2.5), and in-house developed scripts.
Statistical analysis.
Where continuous data were normally distributed, two-sample t-tests were used to compare means between control groups. The Kruskal-Wallis test was used for comparison of non-parametric data. Pearson’s r correlation coefficient was used to determine relationships between continuous variables and Spearman’s rank analysis for correlation between non-parametric variables. P < 0.05 was taken to indicate statistical significance. All statistical analyses were carried using Minitab version 16 (Minitab, State College, PA, USA).
Multivariate Statistical Analysis
The supervised and unsupervised multivariate models were generated using SIMCA (version 14.1, Umetrics, Umeå, Sweden). Principal component analysis (PCA) and orthogonal projections to latent structures discriminant analysis (OPLS-DA) were performed on all spectral data after pareto-scaling and log transformation for detection of patterns, trends and outliers; and construction of discriminant models were generated for classification and the discovery of potential biomarkers respectively.