Deep Sequencing of Hepatitis C Virus Reveals Genetic Compartmentalization in Cerebrospinal Fluid From Cognitively Impaired Patients

Damien C. Tully; Simon Hjerrild; Peter D. Leutscher; Signe G. Renvillard; Colin B. Ogilvie; David J. Bean; Poul Videbech; Todd M. Allen; Jane A. McKeating; Nicola F. Fletcher


Liver International. 2016;36(10):1418-1424. 

In This Article


Hepatitis C virus RNA was detected in all plasma samples and in CSF from four of six patients (Patients 2, 4, 5 and 6) and subjected to deep sequencing. Despite failing to detect HCV RNA in the CSF from patients 1 and 3 low levels of virus may still be present but beneath the detection limit of standard assays. In the neurocognitive group the plasma viral burden was generally higher (median of 8.9 × 106 IU/ml) compared to the normal cognitive group (median of 1.93 × 106 IU/ml) although this was not statistically significant. Viral sequences recovered from the CSF were compared with their respective counterparts from plasma. The number of reads generated from plasma for all subjects was a median of 2807 (IQR: 2494–5620) while a median of 1690 (IQR: 1003–2228) reads were generated from CSF samples. Both sets of samples had similar sequencing depth for plasma (394x ± 165) and CSF (317x ± 101). Moreover, the mean reads lengths were 369 (±28) bp for plasma and 379 (±20) bp for CSF. The hypervariable region (HVR) of E2 is the most rapidly evolving region of the viral genome and provides a simple indicator of genetic variability. Significant genetic compartmentalization was detected between CSF and plasma-derived HVR populations in patient 4 (P < 0.001) where a genetically distinct CSF-derived virus population was observed (Fig. 1A). The CSF-derived viruses contained amino acid substitutions that were less charged and more hydrophobic than the major variant circulating in the periphery. Furthermore, additional genetic changes in E1 were found within the CSF that were absent in the plasma. Examination of viral sequences from patient 6 revealed that the blood contained five HVR variants of which two were detected in the CSF. Upon closer inspection a unique viral mutation was found in the C-terminus of E1 resulting in this variant being sequestered within the CNS (P < 0.001) (Fig. 1B). These data support the presence of an autonomous replicating viral population harboured within the CSF. In the normal neurocognitive group no independent viral evolution was detected in the CSF of one sequenced patient (patient 5) despite a sequencing depth to detect minority variants (Fig. 1C). In this instance the monotypic variant found within CSF comprised the major variant circulating in the blood quasispecies. Only in a single subject (patient 2) was a well-equilibrated viral population maintained between the periphery and CSF with a similar distribution of variants observed at both sites (Fig. 1D).

Figure 1.

Evidence for hepatitis C virus (HCV) genetic compartmentalization in plasma and cerebrospinal fluid (CSF) from subjects with neurocognitive impairment. Phylogenetic analyses of the HCV E2 hypervariable region (HVR) are shown unless otherwise stated. Sequences from the CSF are labelled with solid red circles and sequences derived from plasma are labelled with solid blue squares. The genetic distance (0.01) represents the number of substitutions per site and is indicated by the horizontal line below each phylogeny. (A) One subject (Patient 4) demonstrates the presence of a statistically significant compartmentalized population within the CSF relative to the plasma as assessed using the Slatkin–Maddison test and supported by the test of panmixia. (B) Patient 6 shows evidence of compartmentalization where plasma and CSF populations are not uniformly equilibrated and a minor CSF subpopulation exists from the analysis of partial E1 sequences. (C) Patient 5 demonstrates a monotypic variant present within the CSF while multiple plasma variants are detected. (D) Comparison of HCV amino acid sequence diversity derived from a 3.2-kb fragment of blood and CSF for patient 2 illustrating a well-equilibrated population between blood plasma and CSF. Plotted is the percentage of amino acid diversity at each position spanning the HCV proteins Core to NS2 with respect to the dominant amino acid found within each sample.

Quantification of 27 cytokines in plasma highlighted the levels of inflammatory markers in plasma which were elevated in cognitively impaired HCV-infected patients compared to non-cognitively impaired HCV-infected patients (Fig. S1). Using an unsupervised data reduction tool, principal component analysis (PCA), we attempted to explore the inflammatory profile that was associated with neurocognitive impairment. Principal component (PC1) described the vast majority of the variation in the data set with positive loadings (indicated by the right side of the quadrant) corresponding to HCV-infected patients, whereas negative loadings (indicated by the left side of the quadrant) correspond to healthy subjects (Fig. 2A). Strikingly, the patients demonstrating normal neurocognition cluster together while those patients having neurocognitive impairment cluster independently. Unsupervised hierarchical clustering of all 12 patients revealed two major clusters linked to infected and normal status (Fig. 2B). While examination of the HCV-infected cluster revealed subclusters depicting subjects with normal neurocognition behaviour (patients 1 and 5), a significant similarity within neurocognitive impairment group was observed (Fig. 2B). Cytokine expression profiles in CSF were also measured with IP-10, MCP-1 and GM-CSF detected in CSF from HCV-positive patients (data not shown). However, we failed to detect increased cytokine expression in the CSF from HCV-infected patients with neurocognitive impairment compared to those with normal neurocognition (data not shown). While we did not measure cytokine levels in CSF from uninfected individuals we noted that the cytokine levels measured in the CSF from our HCV-infected cohort are comparable with previous reports of normal CSF, suggesting minimal perturbation.[15] Although the expression levels of MCP-1 in CSF were significantly higher than in plasma (P < 0.002, Mann–Whitney test) the concentrations did not show any significant difference between the two neurocognitive groups.

Figure 2.

Immune activation markers are elevated upon hepatitis C virus (HCV) infection and tend to associate with a state of impaired neurocognition. (A) Unsupervised analysis of cytokine expression by Principal Component Analysis (PCA) was used to reduce dimensionality and extract principal components composed of combinations of 27 different analytes. Principal component 1 (PC1) and 2 (PC2) are depicted in a two-dimensional scatter plot for each individual. Open red dots represent HCV-infected subjects with neurocognitive impairment; blue diamonds are HCV-infected subjects with normal neurocognition; green cross depicts healthy controls. (B) Unsupervised hierarchical clustering of HCV-infected samples analysed using Luminex. Cytokine expression is depicted with a colour scale; red denoting high expression level and blue denoting low expression level. Patient IDs are listed as HCV-infected (1–6) and healthy uninfected controls (7–12) with symbols as described in panel A.