Download Hepatitis B or Hepatitis C Virus Infection Is a Risk Factor for Severe

Hepatitis B or Hepatitis C Virus Infection Is a Risk Factor for Severe
Hepatic Cytolysis after Initiation of a Protease Inhibitor-Containing
Antiretroviral Regimen in Human Immunodeficiency Virus-Infected
Patients
In a cohort of 1,047 human immunodeficiency virus type 1-infected patients started on protease
inhibitors (PIs), the incidence of severe hepatic cytolysis (alanine aminotransferase concentration
five times or more above the upper limit of the normal level ≥ 5N) was 5% patient-years after a
mean follow-up of 5 months. Only positivity for hepatitis C virus antibodies (hazard ratio [HR],
7.95; P < 10−3) or hepatitis B virus surface antigen (HR, 6.67; P < 10−3) was associated with
severe cytolysis. Before starting patients on PIs, assessment of liver enzyme levels and viral
coinfections is necessary.
Several cases of acute hepatitis have been reported after exposure to protease inhibitor (PI)containing regimens (2, 10, 11, 13, 19). At least two mechanisms may be involved in drugrelated hepatitis: either a toxic effect of the PIs or other antiretroviral drugs or an enhanced
inflammatory response against hepatitis B virus (HBV) or hepatitis C virus (HCV) induced by an
immune reconstitution (4, 18). Our report aims at estimating the incidence of severe hepatic
cytolysis among patients exposed to PIs in a multicenter cohort study of human
immunodeficiency virus (HIV) type 1 (HIV-1)-infected patients started on PIs, the anti-proteases
cohort, named APROCO (ANRS EP11), and assessing the determinants of the occurrence of
severe cytolysis.
APROCO was set up to study the clinical and immunovirological evolution in HIV-1-infected
patients started on PI-containing regimens. Patients were enrolled at the initiation of PI therapy
from May 1997 to July 1998 and were monitored at month 1 (M1), M4, and then every 4 months
in 47 French AIDS centers. Patients eligible for this analysis were those who had a serum alanine
aminotransferase (ALT) concentration under fivefold the upper limit of the normal value (<5N)
at the baseline. The HBV and HCV infection statuses at the time of inclusion in the study were
retrospectively recorded. Clinicians were asked to report the most recent results. For HCV and
HBV surface (HBs) antigen, this was part of routine care, but this might have been performed
more often if the patient had a potential risk of contamination.
Severe adverse events (i.e., events graded 3 or 4 according to the grading scheme of the AIDS
Clinical Trials Group [6]), had to be reported to the sponsor within 48 h after recognition. In this
classification, a case of severe cytolysis was defined as an increase in the ALT level to ≥5N. A
validation committee reviewed the cases and classified them as “not related” or “related” to PIs
(12). Cox regression models were used for the analysis of potential determinants of severe
hepatic cytolysis.
Among the initial cohort of 1,080 patients, 1,047 (96.9 %) had a baseline ALT of <5N (median
age, 35 years; proportion of men, 77%). The main HIV transmission route categories were
homosexuality (39%), heterosexuality (34%), and intravenous drug use (17%). The serological
status for hepatitis viruses was known for 613 patients: 26% (n = 159) were HCV seropositive
and 4% (n = 45) had HBs antigen. After a mean follow-up of 5 months, severe cytolysis
developed in 23 patients, yielding an incidence of 5 per 100 patient-years (95% confidence
interval, 3.2 to 7.6). The median time from cohort entry to an ALT of ≥5N was 95 days
(interquartile range, 34 to 121 days). Median (minimum to maximum) ALT and aspartate
aminotransferase (AST) concentration (fold N) were 1.1 (0.5 to 4.8) and 1.1 (0.4 to 5.1) at the
initiation of PI, 3.4 (0.6 to 85.6) and 1.8 (0.6 to 55.7) at M1, 5.3 (0.4 to 17.8) and 3.8 (0.7 to 7.7)
at M4, 1.3 (0.4 to 7.8) and 1.2 (0.5 to 5.5) at M8, 2.0 (0.4 to 5.8) and 2.0 (0.5 to 4.9) at M12, 2.9
(0.7 to 7.7) and 2.6 (0.7 to 4.7) at M16, 1.2 (0.4 to 3.9) and 1.2 (0.7 to 4.1) at M20, and 1.6 (0.3
to 7.4) and 1.1 (0.6 to 3.3) at M24. It was associated with at least one clinical manifestation,
mainly jaundice (n = 6) and abdominal pain (n = 5), in 11 patients (48%).
Among the 23 patients with severe cytolysis, intravenous drug use was the most frequent HIV
transmission route category (52%); 16 (70%) were positive for HCV antibodies and 5 (22%)
were positive for HBs antigen (Table 1). The median change between M0 and M1 was 72 ×
106/liter for the CD4+ cell count and −1.84 log10 copies/ml for the HIV RNA level. At the onset
of severe cytolysis, two patients were receiving saquinavir (SQV), five patients were receiving
ritonavir (RTV), seven patients were receiving indinavir (IDV), five patients were receiving
nelfinavir (NFV), one patient was receiving SQV and RTV, one patient was receiving IDV and
NFV, and one patient was receiving RTV and NFV. NFV had been discontinued 17 days before
severe cytolysis in one patient, and no other PI was used at the onset of severe cytolysis. The
initially prescribed PIs were stopped after the occurrence of severe cytolysis in 17 other patients,
among whom 6 were switched to another PI and 1 was switched to nevirapine. Among these 17
patients, the ALT concentration decreased to <5N in the 13 patients for whom follow-up data
were available. Death occurred in one patient with HCV-related cirrhosis, despite withdrawal of
PI, 150 days after the initiation of PI therapy and 3 days after the onset of severe cytolysis.
Among the five patients who continued to take PIs, the ALT concentration decreased to <5N in
two patients and the ALT concentration remained at ≥5N in three patients. Complete data for
variables entered in the multivariate analysis were available for 570 patients (Table 2). The
patients not included were comparable for HIV transmission route category (P = 0.99) and
baseline level of ALT (P = 0.58). Positivity for HCV antibodies (hazard ratio [HR], 7.95; P <
10−3) and positivity for HBs antigen (HR, 6.67; P < 10−3) were identified as the only risk factors
for severe cytolysis. After adjustment for hepatitis virus status, we found no association between
severe cytolysis and the type of PI or the response at M1 in terms of CD4+ cell count and plasma
HIV RNA level. The interaction between HCV and HBV status was not significant.
Among the present cohort of HIV-infected patients treated with PI-containing regimens, a higher
incidence of severe hepatic cytolysis was detected in patients positive for HCV antibodies or
HBs antigen. In large phase III trials assessing the efficacies of PI-containing regimens (3, 5, 7–
9, 15), when reported, the incidences of severe hepatic cytolysis were 2 and 3% in the groups
receiving IDV (7, 15) and 9% in the group receiving RTV (3). The prevalence of HBV or HCV
coinfection was never mentioned, but abnormal liver enzyme levels were a common exclusion
criterion. Therefore, it is likely that coinfected patients have most frequently been excluded, and
this may have precluded the detection of this adverse event and its risk factors in some of these
trials. Monitoring of severe adverse events after expanded use of PIs therefore seems mandatory
to confirm or assess their incidence and identify risk factors in unselected population samples.
Although an increase in the ALT concentration greater than 5N may not be considered severe
from a hepatologist's point of view, it seems to be a reasonable threshold considering that it
provides an assessment of hepatocellular necrosis and seems to be a reasonable threshold in the
context of adverse events surveillance from international standardized toxicity tables (6).
Severe cytolysis in HIV-infected patients may be related to several causes such as drug treatment
regimens that include PIs and other antiretroviral drugs, concomitant infections or neoplasms,
and immune restoration, which may interact with each other. Drug-related hepatitis may be
suggested by a decrease in ALT levels after withdrawal of PI treatment, a positive response upon
rechallenge, and the presence of an hepatic eosinophilic infiltrate (1, 2, 10). In the present study,
ALT levels decreased to normal after the withdrawal of PI treatment in some patients; however,
the observation of a return of ALT levels to normal levels in other patients who continued PI
treatment suggests that the drug may not have been the only cause of cytolysis.
The first cases of severe cytolysis were described in patients receiving IDV only (2, 10, 13, 19).
We did not confirm these results in our study, and we suggest that a selection bias might have
occurred in previous studies, related to clinical habits for prescription. Through its wide
composition, APROCO may be considered fairly representative of the French population of
HIV-infected patients routinely started on a PI-containing regimen and is a particular contributor
to the surveillance for these adverse events. However, all patients in APROCO were started on a
PI, and it was therefore not possible to compare them to patients not receiving a PI. Moreover, as
patients were not randomized to receive each PI, confounding by indication may exist, so
comparisons were adjusted for other potential risk factors to avoid confounding as much as
possible.
Another limitation was that only those patients with a known serologic status could be included
in the multivariate analysis and a potential selection bias could tend to overestimate the excess
risk associated with hepatitis status. In fact, the proportion of patients with known serologic
HCV status on entry into the study was the same (both 61%) for the group of patients with AST
or ALT levels of at least >1N (n = 305) and the group with normal liver enzyme levels (n = 742).
It is therefore unlikely that only patients with abnormal liver enzyme levels underwent HCV
antibody testing. Moreover, we performed a robustness analysis in which patients with unknown
serologic status were considered HCV positive: the HR for HCV antibody-positive patients
versus HCV antibody-negative patients was 2.1, which could be considered the minimal HR for
severe cytolysis associated with the presence of HCV. Thus, our results clearly confirm the
relationship between severe cytolysis in patients started on a PI-containing regimen and HBV or
HCV infection (2, 4, 11, 18).
It was an hypothesis that restoration of immune status in patients coinfected with HIV and HCV
or HIV and HBV may lead to an enhanced inflammatory response against hepatitis viruses
mediated by CD8+ and CD4+ cells (4, 14, 18). Our data do not support the hypothesis that in
patients positive for HBs antigen or HCV antibodies the likelihood of severe hepatic cytolysis is
related to the intensity of the immunological response to highly active antiretroviral therapy
(HAART). Nevertheless, our study had several limitations. First, a 1-month delay may be too
short to assess the immunological response, and markers other than the peripheral CD4+ cell
count might be more relevant for study of the HCV-specific immune response (20). Second,
HCV or HBV status was assessed only by serology and did not take the plasma HCV RNA or
HBV DNA concentration into consideration, the latter of which is not determined as part of
routine care. Nevertheless, several studies have shown that HAART does not significantly
modify HCV replication (16, 21). Third, without assessment of histopathological lesions by liver
biopsy, the discussion of the underlying pathologic mechanism remains speculative.
In conclusion, because nearly 90% of HCV-infected patients did not develop severe hepatic
cytolysis, our data do not question the recommendation that patients coinfected with HIV and
HCV be treated with HAART, according to recent guidelines on the management of HIVinfected patients (17). However, before starting these patients on a PI, assessment for liver
enzyme levels and viral coinfections is necessary. In patients with HIV and HBV or HCV
coinfection, careful monitoring of liver enzyme levels is suggested during the first months after
the initation of such treatments. It may be hypothesized that other markers, such as HCV RNA or
HBV DNA levels, might be useful to provide a better understanding of the pathophysiology of
acute cytolysis in antiretroviral-treated patients coinfected with HIV and HBV or HIV and HCV.
We thank all patients and investigators at the clinical sites. We are grateful to Nicholas Moore
and Hervé Zylberberg for valuable discussions during the preparation of the manuscript, to
Sylvie Lawson-Ayayi for special contribution to data collection, and to Valérie Journot for
contribution to the statistical analysis.
This study was supported by grants from the Agence Nationale de Recherches sur le Sida
(ANRS) through the Action Coordonnée no. 7 (cohorts), which was the sponsor, and received
additional grants from the following pharmaceutical companies: Abbott, Boerhinger-Ingelheim,
Roche, Bristol-Myers Squibb, Merck Dohm Chibret, Glaxo-Wellcome.
Members of the APROCO Study Group are as follows:
Scientific Committee: Steering Committee, principal investigators C. Leport and F. Raffi;
methodology, G. Chêne and R. Salamon; social sciences, J.-P. Moatti and J. Pierret; virology, F.
Brun-Vézinet and H. Fleury; pharmacy, G. Peytavin; Other members of the Scientific
Committee, D. Costagliola, P. Dellamonica, C. Katlama, L. Meyer, M. Morin, D. Sicard, A.
Sobel, and F. Vincent-Ballereau; Events Validation Committee of the Scientific Committee, M.
Dupon, V. Le Moing, B. Marchou, T. May, P. Morlat, A. Waldner-Combernoux; observers to
the Scientific Committee, F. Agid, F. Bourdillon, J.-F. Delfraissy, J. Dormont, J.-Y. Lacut, Y.
Souteyrand, and J.-L. Vildé. Monitoring and statistical analysis were conducted by V. Cailleton,
D. Carricaburu, C. Deveaud, G. Dupouy, S. Dutoit, J.-L. Ecobichon, C. Egouy, C. Jadand, P.
Joly, V. Journot, S. Lawson-Ayayi, C. Lewden, B. Masquelier, W. Nouioua, G. Palmer, M.
Savès, and M. Souville. Promotion was done by the Agence Nationale de Recherches sur le Sida
(ANRS, Action Coordonnée no. 7). Other support was provided by the Association des
Professeurs de Pathologie Infectieuse et Tropicale and associated pharmaceutical companies: J.
P. Chauvin (Abbott), D. Delavelle (Boerhinger-Ingelheim), E. Dohin (Roche), B. Gallet (BristolMyers Squib), M.-C. Gervais (Merck Dohm Chibret), and D. Lapierre (Glaxo-Wellcome).
Clinical centers (coordinators) were as follows: Amiens (J. L. Schmit); Angers (J.-M.
Chennebault), Belfort (J.-P. Faller), Besançon (J.-M. Estavoyer, R. Laurent, and D. Vuitton),
Bordeaux (J. Beylot, J.-Y. Lacut, M. Le Bras, J.-M. Ragnaud), Bourg-en-Bresse (P. Granier),
Brest (M. Garré), Caen (C. Bazin), Compiègne (P. Veyssier), Corbeil Essonnes (A. Devidas),
Créteil (A. Sobel), Dijon (H. Portier), Garches (C. Perronne), Lagny (P. Lagarde), Libourne (J.
Ceccaldi), Lyon (D. Peyramond), Meaux (C. Allard), Montpellier (J. Reynes), Nancy (P.
Canton), Nantes (F. Raffi), Nice (J.-P. Cassuto and P. Dellamonica), Orléans (P. Arsac), Paris (F.
Bricaire, C. Caulin, J. Frottier, S. Herson, J.-C. Imbert, J.-E. Malkin, W. Rozenbaum, D. Sicard,
F. Vachon, and J.-L. Vildé), Poitiers (B. Becq-Giraudon), Reims (G. Rémy), Rennes (F. Cartier),
Saint-Etienne (F. Lucht), Saint-Mandé (R. Roué), Strasbourg (J.-M. Lang), Toulon (D. Jaubert),
Toulouse (P. Massip), and Tours (P. Choutet).
Nucleotide and Amino Acid Complexity of Hepatitis C Virus
Quasispecies in Serum and Liver
The quasispecies nature of the hepatitis C virus (HCV) is thought to play a central role in
maintaining and modulating viral replication. Several studies have tried to unravel, through the
parameters that characterize HCV circulating quasispecies, prognostic markers of the disease. In
a previous work we demonstrated that the parameters of circulating viral quasispecies do not
always reflect those of the intrahepatic virus. Here, we have analyzed paired serum and liver
quasispecies from 39 genotype 1b-infected patients with different degrees of liver damage,
ranging from minimal changes to cirrhosis. Viral level was quantified by real-time reverse
transcription-PCR, and viral heterogeneity was characterized through the cloning and sequencing
of 540 HCV variants of a genomic fragment encompassing the E2-NS2 junction. Although in
95% of patients, serum and liver consensus HCV amino acid sequences were identical,
quasispecies complexity varied considerably between the viruses isolated from each
compartment. Patients with HCV quasispecies in serum more complex (26%) than, less complex
(28%) than, or similarly complex (41%) to those in liver were found. Among the last, a
significant correlation between fibrosis and all the parameters that measure the viral amino acid
complexity was found. Correlation between fibrosis and serum viral load was found as well (R =
0.7). With regard to the origin of the differences in quasispecies complexity between serum and
liver populations, sequence analysis argued against extrahepatic replication as a quantitatively
important contributing factor and supported the idea of a differential effect or different selective
forces on the virus depending on whether it is circulating in serum or replicating in the liver.
Hepatitis C virus (HCV) is an enveloped virus classified in the family Flaviviridae (7, 29, 45).
Its genome consists of a single-stranded RNA, with plus polarity, of 9,600 nucleotides, which
does not integrate in the host genome, yet persistence is the rule. The damage caused during
infection ranges from minimal changes to cirrhosis of the liver and hepatocarcinoma, but little is
known about the mechanism of hepatocyte injury due to chronic infection. It seems very likely
that the pathogenesis of HCV infection is directly related to a strong interplay between the host
defense mechanisms and the virus's ability to evade them efficiently. Moreover, in order to
persist, HCV must regulate its lytic potential and avoid elimination by the host immune system.
Due to the quasispecies structure of the HCV viral population infecting single patients (39), the
virus may use a variety of strategies to fulfill both requirements (11, 17).
The dynamic component of the quasispecies structure is responsible for the rapid virus evolution
(12). It works through a complex mixture of genomic sequences (quasispecies) which behaves as
a single unit when facing changes in the environment. The genetic interaction within the viral
population allows the system to distinguish the best possibility at any given time and, therefore,
to avoid replicative efforts in unrewarding directions (12–16). It has been experimentally proven
for RNA phage (14) and viruses (27) that the use of this formula is intended to produce
successful adaptation to environmental changes (12). This mechanism has been invoked in the
HCV virus model to explain both the high frequency of viral persistence and the wide range of
disease (3, 4, 10, 18, 21, 39). Phylogenetic methods may be used for understanding virus
evolution.
The static component of the quasispecies structure of the viral population can be analyzed
through the total number of viral particles (viral load), the proportion of different viral genomes
present in that total (normalized Shannon entropy [55, 59]), and the degree of polymorphism (Pn)
among the variants (21, 40). These parameters can be measured at single time points and are of
analytical interest because they may fluctuate over a wide range of values and may be used to
categorize quasispecies in relation to the clinical state of the patient. Many previous studies have
examined the significance of both parameters independently. The wide range of
clinicopathological correlations between serum and intrahepatic RNA levels (20, 23, 24, 31, 32,
35, 38, 42, 47, 50, 51), together with discrepancies in the literature among authors who find
correlation between quasispecies complexity and liver damage (25, 28, 33, 61) and those who do
not (22, 36, 48, 56), suggests that the relation between viral load and liver injury is more
complex than expected.
Recently, we have proven that, within an infected patient, the composition of the circulating viral
population does not necessarily reflect the composition of the hepatic population (5), although
the causes for this difference remain obscure. Heterogeneous quasispecies in peripheral blood
mononuclear cells (PBMC) of humans and in chimpanzees have been described (34, 37, 49, 54,
57), and it has been proposed that replication in this tissue might contribute to HCV serum
quasispecies complexity. In the present study we have evaluated the implications of serum and
liver quasispecies complexity in the natural course of the disease. To do this, we have performed
an analysis of the viral population parameters of a genomic region encompassing the envelope 2nonstructural region 2 (E2-NS2) junction in paired serum and liver samples from 39 patients
with chronic hepatitis C.
Viral isolates. HCV was isolated from paired serum and liver samples from 39 HCV-infected
patients. The degree of liver damage was semiquantified according to the scoring system of
Ishak et al. (30). This showed that 7 patients had mild chronic hepatitis, 17 had moderate
hepatitis, 11 had severe hepatitis, and 4 had established cirrhosis (Table 1). A parenteral risk
factor with a known date of infection was present for 24 patients. The estimated mean duration
of infection of these patients was 28 ± 14 years. Demographic variables are summarized in Table
1. All patients were infected with genotype 1b and had detectable HCV RNA, but none was
positive for other hepatitis viruses or human immunodeficiency virus (HIV).
Six patients had received a 6- to 12-month course of interferon treatment 2 to 4 years before the
samples were obtained. Written informed consent was obtained from all patients before they
underwent liver biopsy. In all cases blood was drawn in Vacutainer tubes and centrifuged within
2 h and the serum was stored at −80°C. Three-millimeter-long fragments of liver biopsies were
frozen in liquid nitrogen.
RNA extraction, reverse transcription-PCR, cloning, and sequencing. Virus RNA was
extracted from both serum (140-μl) and liver (0.05-g) samples with QIAamp viral RNA binding
columns (Qiagen). Isolated HCV RNA was reverse transcribed into cDNA, with genotype 1bspecific primers from a region encompassing the E2(p7)-NS2 region (MJJ3, 5′CTCGAGCGTTGAGGGGGG-3′, positions 2602 to 2585). Nested PCR was performed with
specific oligonucleotides to amplify a 212-bp fragment (outer set, MJJ3 and MJJ4 [5′TGTGCCTGCTTGTGGATG-3′; positions 2194 to 2211]; inner set, MJJ5 [5′CTAGAATTCAAAAATATTGTAACCACCA-3′; positions 2547 to 2530] and MJJ6 [5′ACAGGATCCAGTCCTTCCTTGTGTTCTTCT-3′; positions 2299 to 2317]). Amplified
products were purified with QIAquick PCR purification kit (Qiagen) and cloned in Escherichia
coli DH5α. Individual clones were sequenced by the dideoxy chain terminator method with the
DNA sequencing kit (Perkin-Elmer) and the Abi Prism 310 genetic analyzer (Perkin-Elmer).
Viral RNA quantitation. Quantitation of HCV RNA in both serum and liver samples was
performed by using the Taqman technology (Roche Molecular Diagnostics Systems) and the Abi
Prism 7700 real-time sequence detection system (Perkin-Elmer), as already described (41).
Briefly, the method uses a dual-label fluorogenic hybridization probe that specifically anneals
the template between PCR primers. When the probe is intact, the quencher at the 3′ end
suppresses the emission of the reporter at the 5′ end. The nuclease degradation of the probe,
during the PCR, releases the reporter, and the sequence detector (Abi Prism 7700) measures the
amplified product in direct proportion to the increase of fluorescence emission. The total RNA
concentration from liver biopsies was estimated by determining the absorbance at 260 nm (2).
The serum HCV RNA concentration was expressed as the number of viral genome copies per
milliliter. The liver HCV RNA quantity was expressed as the number of genomes per microgram
of total RNA.
HCV population parameters. The term complexity, as it refers to genetic information, was
adopted to describe, in quantitative terms, the genetic information contained in viral quasispecies
(39, 44). The quasispecies complexity can be divided into two different parameters: Pn and
frequency of different sequences (Shannon entropy). Pn was calculated as the number of
polymorphic sites divided by the number of nucleotides (or amino acids) sequenced; variability
of the quasispecies increased as Pn increased. Shannon entropy has been defined in terms of the
probabilities of the different sequences than can appear at a given time point (55, 59). This
measure was calculated as S = −Σi(pi ln pi) where pi is the frequency of each sequence in the
viral quasispecies. The resulting number was normalized as a function of the number of clones
analyzed, thus allowing comparisons of complexity among different isolates. The normalized
entropy, Sn, was calculated as Sn = S/ln N, where N is the total number of sequences analyzed. Sn
varied from 0 (no diversity) to 1 (maximum diversity) (55). This is a measure of the specific
heterogeneity of a given specimen without considering the number of particles present in that
specimen. Total heterogeneity of a specimen, per milliliter of serum or per microgram of total
RNA, was calculated as the product of normalized Shannon entropy and the natural logarithm of
the number of particles.
The proportion of synonymous substitutions per potential synonymous site (ds) and the
proportion nonsynonymous substitutions per potential nonsynonymous site (dn) were calculated
with SNAP.pl (synonymous, nonsynonymous analysis program) from the HIV sequence
database (http://hiv-web.lanl.gov).
In order to evaluate the relationships among the viruses isolated from the 39 patients, distances
between all possible pairs of sequences (intrapatient and interpatient distances among serum,
liver, and serum/liver sequences) were calculated by the Kimura two-parameter modification
method and unrooted phylogenetic trees were constructed from the Windows Easy Tree software
package (http://www.tdi.es).
We have estimated the degree of polymorphism, distance, and specific and total heterogeneity at
both the nucleotide and amino acid levels and the ratio of synonymous to nonsynonymous
substitutions (ds/dn) for 39 chronic hepatitis C patients.
Statistical analysis. Data were expressed as means ± standard deviations. Pearson's or
Superman's correlation was carried out by using the following variables: age, fibrosis,
necroinflammation, alanine transaminase level, (ALT) HCV RNA level, degree of
polymorphism, Shannon entropy, and total heterogeneity in serum and liver viral samples (in
nucleotide and amino acid sequences). Differences in viral complexity between distinct
histological groups were analyzed by t test.
Nucleotide sequence accession numbers. The EMBL database accession numbers for the
sequences presented in this article are AJ247658 through AJ248197.
Table 1 summarizes viral population parameters of complexity in serum-liver pairs in the 39
patients studied. Viral levels ranged from 2.7 × 104 to 2.9 × 107 copies/ml of serum (mean, 2.6 ×
106 ± 4.8 × 106 copies/ml) and from 0.021 to 1.9 × 102 copies/μg of total liver RNA (mean, 24 ±
42 copies/μg of total liver RNA). The characteristic quasispecies structure was found in all
samples, which were subsequently analyzed by using population parameters. On average, 7 E2NS2 sequences (4 to 12) were obtained from each sample (Table 1). Overall, as shown in Table
2, mean polymorphism values and proportions of variants present in both serum and liver
samples were high, although they varied widely from patient to patient. This variability indicates
that in this region both parameters were of analytical interest and might be used to characterize
HCV quasispecies. Serum and liver viral sequences appeared to be strongly selected for
synonymous replacements (mean percentage of synonymous mutations in serum and liver, 70%
± 23% and 71% ± 25%, respectively; mean ratio of synonymous to nonsynonymous [ds/dn]
substitutions in serum and liver, 1.6 ± 1.2 and 1.6 ± 0.9, respectively) (Table 1). Samples from
four patients contained sequences differing in more than 10 residues (5% of the total fragment
length) from the other sequences from the same compartment. In these cases, patients are said to
have double populations.
Comparison of serum and liver quasispecies. The same consensus nucleotide sequence was
found in 46% of patients. In 33% there were one to four ambiguities (when the consensus residue
at a given position is not defined by 60% or more of the sequences, the consensus is not assigned
to any residue [10]) in serum or liver consensus sequences. Twenty-one percent of patients had
different nucleotide consensus sequences. At the amino acid level, 82% of the patients had
identical sequences; in 13% there were one or two ambiguities, and 5% had different sequences
in each compartment. Except for patients 9 and 36, phylogenetic analysis (Fig. 1) showed that
sequences obtained from each individual did not segregate according to their tissues of origin.
Mean nucleotide and amino acid intrapatient distances for serum, liver, and serum/liver
sequences were similar, while such distances were found to be seven or eight times higher in the
interpatient analysis (Table 3). Nevertheless, as previously reported, HCV quasispecies structure
(in terms of complexity) in serum did not always reflect the quasispecies structure in the liver
(Table 4) at both the nucleotide and amino acid levels.
Patients were initially classified into three groups according to the similarity between HCV
quasispecies complexities of serum and liver for each parameter, although the percentage of
patients that were included in each group differed according to the parameter chosen (Table 3).
Patients were considered to have the same level of complexity when the ratio between liver and
serum values for a given parameter was between 0.5 and 2 (group A), to have less-complex
serum HCV quasispecies when the ratio was 2 or higher (group B), and to have more-complex
serum HCV quasispecies when the ratio was 0.5 or less (group C).
Since a phenotypic criterion, i.e., liver damage, had been used to select the study patients, the
phenotypic parameters (those referring to amino acid sequence) were used to further study the
correlations between quasispecies complexity for each compartment and clinicopathological
variables. In those cases in which there was disagreement between the two amino acid
parameters, final classification was based on nucleotide parameters. Accordingly, 16 patients
(41%) were classified into group A, 11 (28%) were classified into group B, and 10 (26%) were
classified into group C. Two patients (5%) were not included in the classification because they
had different serum and liver HCV consensus amino acid sequences.
Correlation between the values of quasispecies parameters in serum and in liver. Statistical
analysis of the results showed that there was a correlation between serum and liver viral RNA
concentrations (R = 0.4; P = 0.02). There was a strong correlation among the six population
parameters within each compartment (0.4 < R < 0.96; P < 0.02) but not between compartments,
with the exception of the degree of nucleotide polymorphism and the intrapatient nucleotide
distance (R = 0.6; P < 0.01). Total heterogeneity, which integrates viral load and normalized
Shannon entropy, was correlated at both nucleotide and amino acid levels within and between
compartments (0.4 < R < 0.9; P < 0.02).
Correlation of viral quasispecies parameters and liver damage. Viral population parameters
for the 39 patients were classified according to the degree of liver damage. Significant
correlation between serum viral load and fibrosis, necroinflammation, and ALT level was found
(R = 0.7, P < 0.01; R = 0.5, P < 0.01; and R = 0.4, P = 0.02, respectively). Fibrosis correlated
with total nucleotide heterogeneity of liver quasispecies (R = 0.4, P = 0.03) and total amino acid
heterogeneity of both circulating and hepatic viral populations (R = 0.4, P = 0.01; and R = 0.3, P
= 0.038, respectively). Necroinflammation correlated with total nucleotide heterogeneity of
hepatic virus (R = 0.4, P = 0.03) and total amino acid heterogeneity of circulating virus (R = 0.4,
P = 0.03).
Correlations between clinical and viral parameters according to liver/serum complexity
ratio. Within group A patients, viral load and amino acid complexities of serum and liver
quasispecies correlated with the degree of fibrosis (Table 5). In contrast, quasispecies complexity
at the nucleotide level did not correlate with fibrosis. Among these patients, viral HCV RNA and
amino acid complexities of circulating and hepatic viruses were higher in those with severe liver
damage than in those with mild or moderate disease (data not shown). In contrast, among
patients from groups B and C no correlation between viral parameters and the degree of fibrosis
was found.
In a previous work (5), we found that the structure of replicating HCV quasispecies in the liver
does not always reflect that of circulating HCV virions. We observed that two of four patients
had a twofold-more-complex HCV quasispecies in liver than in serum. Subsequently, others
have reported finding more-complex circulating quasispecies (49). The present study, involving
a large number of genotype 1b-infected patients with a wide range of liver lesions, confirms and
expands these observations. Most patients (95%) had HCV quasispecies with the same consensus
amino acid sequence in serum and liver at the E2(p7)-NS2 junction; sequences from the majority
of patients (95%) (Fig. 1) did not cluster separately between the two compartments, and in the
same line, the intrapatient nucleotide and amino acid distances were seven and eight times lower
than the interpatient distances. However, the amino acid complexities of the quasispecies in this
region showed a twofold or higher difference between the two compartments in 54% of the
patients. Accordingly, patients could be classified into three groups as a function of the degree of
similarity in the complexities of viral quasispecies in both compartments. The origin and the
clinical implications of this finding are unknown. In our previous work, we tried to explain the
higher complexity in the liver by suggesting the existence of distinct functional compartments
with different replication kinetics (5). Alternatively, since the final fate of sequences found in the
replicating pool is unknown, the finding of highest complexity in the liver quasispecies might be
explained by an excess contribution of sequences that will not be incorporated into mature
virions and released to the circulation. The opposite finding, that is, higher complexity in the
circulating pool, does not have a readily obvious explanation. The possible contribution of
variants replicating in extrahepatic sites has been proposed. In fact, several studies have reported
the presence of distinct viral quasispecies in PBMC of infected patients (34, 37, 49, 54, 57). Any
extrahepatic contribution to the circulating pool should lead to the presence of readily obvious
mixed populations in the serum. This should be more apparent in long-standing infections (6, 26,
46), in which virus in the two replicating compartments (i.e., PBMC and liver) would evolve
separately from a common ancestor. However, in our study a double virus population in the
serum was found in only 1 of the 14 patients with long-standing infection. In addition, in two of
the four patients who had a double population of sequences in the serum, the corresponding
sequences were also present in the liver. All these data, the phylogenetic clustering of serum and
liver sequences (in 37 of 39 patients; Fig. 1) and the finding that virus level was correlated
between both compartments, argue against a significant contribution (in quantitative terms) of
extrahepatic HCV replication to the serum (9, 43).
Alternatively, differences in the clearance rates of some variants might be responsible for the
observed differences (19). Rapid elimination of a major variant by circulating antibodies could
lead to an overrepresentation of the mutant repertoire. In that situation, the observed differences
between the circulating and the hepatic virus would be more apparent than real.
Several studies have tried to correlate the complexity of the circulating quasispecies and degree
of liver damage (22, 25, 28, 33, 36, 48, 56, 61). In the present study we found no correlation
between quasispecies complexity at the nucleotide level and liver damage. In contrast, a
significant correlation between quasispecies complexity at the amino acid level, in both serum
and liver, and the extent of liver fibrosis was observed, albeit in only those patients with similar
levels of complexity in both compartments. Hence, techniques that can only provide estimates of
nucleotide diversity would not have predictive value with regard to liver damage. The finding
that only amino acid complexity correlates with liver damage might have pathogenic relevance
since the interaction between virus and the host immune system occurs at the phenotypic level.
This observation would fit theoretical models of HIV diversification, in which antigenic variants
are not completely replaced by emerging ones, so that the continuous accumulation of variants
could reflect the history of immune evasion and cell destruction (52, 53). In these patients (group
A), the good correlation between serum viral load and degree of fibrosis would allow the
monitoring of disease progression in individual patients by HCV RNA quantitation (1, 8, 41, 58,
60). Nevertheless, the potential use of amino acid parameters of complexity and/or viral load as
an indirect measure of ongoing liver damage is limited for two reasons. First, correlation is
restricted to those patients with similar levels of quasispecies complexity in both compartments,
and these cannot be differentiated from the other two groups by any clinical or readily accessible
parameter. Second, it is currently unknown whether the liver/serum complexity ratio is a stable
parameter or fluctuates over time. Longitudinal studies of sequential serum and liver pairs would
be required to clarify this issue. It is possible that coincident quasispecies complexities represent
a steady-state level, in which more complexity implies more damage. Such an equilibrium may
transiently be lost when viral or immune factors influence the complexity of the circulating or
replicating pool. Further investigation of the dynamic behavior of viral quasispecies in both
compartments would increase our understanding of the influence of quasispecies complexity in
liver damage.
This investigation was supported in part by grant 94/1682 from the Fondo de Investigaciones
Sanitarias (FIS), by grant 97-0148 from the Comisión Interministerial de Ciencia y Tecnología
(CICYT), and by the Fundació per a la Recerca Biomèdica i la Docència de l'Hospital Vall
d'Hebron.