HIV/AIDS • CID 2010:51 (15 November) • 1209
H I V / A I D S M A J O R A R T I C L E
Modeling the Probability of Sustained Virological
Response to Therapy with Pegylated Interferon
plus Ribavirin in Patients Coinfected
with Hepatitis C Virus and HIV
Jose Medrano,1Karin Neukam,3Norma Rallo ´n,1Antonio Rivero,4Salvador Resino,2Susanna Naggie,6Antonio Caruz,5
Aida Calvino,2Juan Macı ´as,3Jose Miguel Benito,1Carlos Sa ´nchez-Piedra,1Eugenia Vispo,1Pablo Barreiro,1
John McHutchison,6Juan Antonio Pineda,3and Vincent Soriano1
1Department of Infectious Diseases, Hospital Carlos III, and
of Microbiology, Instituto de Salud Carlos III, Majadahonda, Madrid,3Infectious Diseases Unit, Hospital Universitario de Valme, Grupo Guadalquivir,
Universidad de Jaen, Jaen, Spain; and
2Laboratory of Molecular Epidemiology of Infectious Diseases, National Centre
4Infectious Diseases Unit, Hospital Universitario Reina Sofia, Grupo Guadalquivir, Cordoba, and
6Duke Clinical Research Institute, Durham, North Carolina
5Immunogenetics Unit, Faculty of Sciences,
sustained virological response to pegylated interferon plus ribavirin (pegIFN-RBV) treatment for chronic hepatitis
C virus (HCV) infection. Given that therapy is poorly tolerated and rates of response arelowerinpatientscoinfected
with HCV and human immunodeficiency virus (HIV), the recognition of predictors of response is a high priority
in this population.
A baseline noninvasive index was derived on the basis of the probability of achieving sustained
virological response in a group of 159 HIV-HCV–coinfected patients treated at one clinic in Spain. The index was
then validated using data from a separate cohort of 86 coinfected individuals. Only individuals who had completed
a course of pegIFN-RBV therapy and had validated outcomes were considered.
The final score included 4 variables: 2 host-related variables (IL28B SNP rs12979860 and liverstiffness)
and 2 HCV-related variables (genotype and viral load). The area under the receiver operating characteristic curve
was 0.89 in the derivation group and 0.85 in the validation group.
The probability of achieving sustained virological response with pegIFN-RBV therapy in HIV-
HCV–coinfected patients can be reliably estimated prior to initiation of therapy using an index that includes 4
A single-nucleotide polymorphism (SNP) near the IL28B gene (rs12979860) strongly predicts
Hepatitis C virus (HCV) infects 1175 million people
worldwide  and is the leading cause of end-stage
liver disease, as well as the primary indication for liver
transplantation in Western countries . Both HCV
and human immunodeficiencyvirus(HIV)shareroutes
of transmission and establish chronic infections; there-
fore, coinfection is common. Overall, 20% of HIV-in-
Received 1 May 2010; accepted 28 July 2010; electronically published 19
Reprints or correspondence: Dr Vincent Soriano, Dept of Infectious Diseases,
Hospital Carlos III, Calle Sinesio Delgado 10, Madrid 28029, Spain (vsoriano@
Clinical Infectious Diseases2010;51(10):1209–1216
? 2010 by the Infectious Diseases Society of America. All rights reserved.
fected individuals worldwide are chronically infected
with HCV, with coinfection rates of up to 90% in per-
sons with a history of injection drug use . Thecourse
of HCV-associated liver disease is accelerated in coin-
fected individuals [3, 4]; thus, HCV has emerged as an
important cause of mortality in this population ,
especially since successful antiretroviral therapy has
dramatically reduced opportunistic illnesses.
Current therapy for chronic hepatitis C is based on
a combination of pegylated interferon-a (pegIFN) and
ribavirin (RBV), generally administered for 6–18
months . Unfortunately, the regimen is poorly tol-
erated, and overall only 60% of patients are cured. Of
note, this figure is lower among HIV-HCV–coinfected
patients [2, 7]. Thus, the identification of predictors of
treatment success is particularly desirable in this pop-
1210 • CID 2010:51 (15 November) • HIV/AIDS
nodeficiency Virus (HIV) and Hepatitis C Virus (HCV)
Main Characteristics of the Study Population of Patients Coinfected with Human Immu-
Characteristic Derivation cohort Validation cohortPa
No. of patients
Age, median years (range)
Prior intravenous drug use
BMI, median (range)
Liver fibrosis stage (Metavir score)b
Liver stiffness, median kPa (range)
Baseline CD4 cell count, cells/mL (range)
Median plasma HIV-RNA level, log copies/mL (range)
Serum HCV RNA level 1850,000 IU/mL
kilograms divided by the square of height in meters); HAART, highly active antiretroviral therapy; kPa, kilopascals; RNA,
ribonucleic acid; SNP , single nucleotide polymorphism; SVR, sustained virological response.
aP values in bold type indicate statistical significance.
bF0–F1, !7.2 kPa; F2, 7.2–9.5 kPa; F3, 9.6–14.5 kPa; F4, ?14.5 kPa.
cThis definition includes null, partial response, breakthrough, and relapse.
Data are no. (%) of patients, unless otherwise indicated. BMI, body mass index (calculated as weight in
ulation, to ensure adequate selection of the best candidates,
who largely remain untreated at most medical centers [8–10].
The best baseline predictors of response to current HCV
therapy are infection with HCV genotypes 2 or 3, low baseline
serum HCV RNA level, and null-minimal liver fibrosis .
Once treatment has been initiated, the best predictor of cure
is the achievement of undetectable serum HCV RNA level at
week 4 [2, 6]. Recently, 3 independent genomewide association
studies have identified severalsingle-nucleotidepolymorphisms
(SNPs) around the IL28B gene as strongly associated with treat-
ment outcomes in HCV-monoinfected individuals [12–16].
The SNP with the strongest association, rs12979860, is located
on chromosome 19. In patients infected with HCV genotype
1, the CC genotype is associated with a 12-fold greater rate of
sustained virological response (SVR) than the CT and TT ge-
notypes. Similar findings have recently been reproduced in
HIV-HCV–coinfected patients . Modeling the impact of
baseline predictors of treatment outcome that incorporate this
new genetic variant may provide clinicians the opportunity to
make more adequate therapeutic decisions in HIV-HCV–coin-
fected patients [18–21].
To construct a clinically useful predictive index of SVR, a pop-
ulation of HIV-HCV–coinfected patients regularly followed up
at Hospital Carlos III in Madrid was used as the derivation
cohort. The main characteristics of the whole cohort and HCV
HIV/AIDS • CID 2010:51 (15 November) • 1211
apy with Pegylated Interferon and Ribavirin in PatientsCoinfected
with Human Immunodeficiency Virus and Hepatitis C Virus (HCV)
Predictors of Sustained Virological Response to Ther-
PredictorOR (95% CI)P
HCV genotypes 1 and 4
Serum HCV RNA level (in log IU/mL)
Liver stiffness (in kilopascals)
SNP rs12979860 (CT or TT)
gression analysis. CI, confidence interval; OR, odds ratio; SNP , single-nucle-
Predictors were determined by stepwise multivariate logistic re-
virological response in the derivation and validation groups. AUROC, area
under the receiver operating characteristic curve; CI, confidence interval.
Diagnostic performance of the predictive index of sustained
therapy uptake have been described elsewhere . For the
external validation of the predictive index, an independent co-
hort of HIV-HCV–coinfected patients treated during the same
period at another 2 clinics was evaluated (the validation co-
hort). To participate in the study, written informed consent
for genetic testing was obtained from all individuals, and the
study protocol was evaluated and approved by the hospital
From an initial population of 672 HIV-
HCV–coinfected patients receiving regular follow-up at Hos-
pital Carlos III, a total of 159 consecutive individuals who
initiated therapy with pegIFN-RBV, had validated outcomes,
and were tested for the rs12979860 SNP were selected .
Patients with poor drug compliance and/or who had discon-
tinued therapy because of side effects were excluded, as were
patients infected with hepatitis B virus. Treatmentwasprovided
from November 2004 through December 2008. Inclusion cri-
teria included interferon naiveness, liver fibrosis assessment us-
ing elastometry within 6 months prior to initiation of therapy,
serum HCV RNA level, and genotyping.
Among a total of 154 HIV-HCV–coin-
fected patients who similarly had completed a course of HCV
therapy during the same period at 2 different HIV clinics, 86
patients with the same inclusion criteria were eligible for the
Hepatitis C therapy.
In both cohorts, treatment regimens
included pegIFN alpha 2a or 2b at standard doses (180 mg/
week or 1.5 mg/kg/week, respectively) plus weight-adjusted
RBV dosing (1000 mg/day for patients weighing !75 kg and
1200 mg/day for patients weighing ?75 kg). Following inter-
national guidelines , patients with HCV genotypes 1 or 4
received either 48 or 72 weeks of treatment, and patients with
HCV genotype 3 were treated for 24 or 48 weeks, according
to virological response at week 4. No patients in this series were
infected with HCV genotype 2. Early stopping rules were ap-
plied for subjectswithsuboptimalvirologicalresponsesatweeks
12 and 24 .
SVR was the primary outcome and
was defined as undetectable serum HCV RNA level at 24 weeks
after completion of treatment. For the purpose of this study,
patients with relapse were considered along with nonrespond-
ers, who were patients who experienced suboptimal virological
response during the treatment period and for this reason did
not complete the planned duration of therapy. As previously
mentioned, patients with poor drug compliance and/or who
discontinued therapy prematurely because of side effects were
excluded from the study.
HCV viremia and genotyping.
serum HCV RNA level was measured using a real-time poly-
merase chain reaction (PCR) assay (COBAS TaqMan; Roche)
with a lower limit of detection of 10 IU/mL. HCV genotyping
was performed using a commercial reverse-transcriptase PCR
hybridization assay (Versant HCV Genotype v2.0 LiPA; Sie-
mens), which maximally reduces the chances of HCV genotype
Liver fibrosis stage.
The extent of hepatic fibrosis was mea-
sured in both cohorts using transientelastographybyFibroScan
(Echosens). Details about this noninvasive method, the ex-
amination procedure, and correlation of hepatic fibrosis esti-
mates with liver biopsy findings have been reported elsewhere
[23–25]. Liver stiffness valuesare expressedinkilopascals(kPa).
For clinical purposes, stages of liver fibrosis were defined ac-
cording to evaluations performed both in HCV-monoinfected
and HIV-HCV–coinfected patients [26, 27], using the Metavir
score as follows: F0-F1, !7.2 kPa; F2, 7.2–9.5 kPa; F3, 9.6–14.5
kPa; and F4, 114.5 kPa.
rs12979860 SNP genotyping.
cohort, IL28B genotyping was performed at the Duke Institute
for Genome Sciences and Policy. Genotyping was conducted
In both cohorts of patients,
For patients in the derivation
Diagnostic Accuracy and Predictors of Sustained Virological Response in the Study Population
Group and cutoff
Estimation group (
94.2 (87.1–97.5) 56.2 (44.8–67)
71.7 (62.8–79.2) 89.1 (77–95.3)
0.10 (0.04–0.25) 20.76 (7.5–57.3)
81.4 (71.9–88.2) 80.8 (70.3–88.2) 83.3 (73.9–89.8) 78.7 (68.1–86.4) 4.24 (2.62–6.87)
0.23 (0.15–0.36) 18.44 (8.3–40.9)
60.5 (49.9–70.1) 91.8 (83.2–96.2) 89.7 (79.2–95.2) 66.3 (56.7–74.8) 7.36 (3.35–16.14) 0.43 (0.33–0.57) 17.08 (6.7–43.7)
Validation group (
41 (27.1–56.6) 65.2 (53.1–75.5)
80 (58.4–91.9) 1.55 (1.18–2.04)
85.1 (72.3–92.6) 69.2 (53.6–81.4) 76.9 (63.9–86.3) 79.4 (63.2–89.7) 2.77 (1.7–4.5)
0.22 (0.11–0.44) 12.86 (4.5–36.8)
66 (51.7–77.8) 92.3 (79.7–97.3) 91.2 (77–97)
69.2 (55.7–80.1) 8.57 (2.84–25.93) 0.37 (0.24–0.56) 23.25 (6.2–87.3)
DOR, diagnostic odds ratio; LR, likelihood ratio; NPV, negative predictive value; PPV, positive predictive value.
HIV/AIDS • CID 2010:51 (15 November) • 1213
in a blinded fashion on DNA specimens collected from each
individual, using the 5?nuclease assay with allele-specific
TaqMan probes, as previously described . For patients from
the validation cohort, similar primers and conditions wereused
at the Immunogenetics Unit of the Jaen University.
Overall, results are presented as me-
dians (lower and upper quartiles) for continuous variables and
as frequencies and percentages for categorical data. Analysis of
normality was performed using the Kolmogorov-Smirnov test.
Categorical data and proportions were analyzed using the x2
test or the Fisher’s exact test, as required. Student’s t test was
used to compare the means of the 2 groups with normal dis-
tributions, and the Mann-Whitney U test was used to compare
variables with nonnormal distributions.
The study population was represented by all consecutive
HIV-HCV–coinfected individuals who had completed a course
of pegIFN-RBV therapy and for whom baseline liver stiffness
and IL28B genotyping results were available. Similar criteria
were used to select individuals to be enrolled in the validation
cohort among patients who attended 2 distinct HIV clinics. In
this way, the whole study population was split as 64.9% and
35.1%, respectively, which represents nearly optimal propor-
tions for ensuring the reliability of independent external vali-
dation of indexes.
Multiple association tests were conducted using univariate
logistic regression to identify independent variables associated
with SVR, the primary outcome. In the last analysis, we in-
cluded all variables with P values !.05 in the univariate anal-
ysis. Then, a forward stepwise logistic regression analysis was
conducted with P values for entry and exit of .05 and .10,
respectively. Thereafter, to express the likelihood of achieving
SVR as a probability ranging from 0 to 1, an index to predict
SVR through a logistic probability function was built .
The accuracy of this index was obtained and compared by
calculating area under the receiver operating characteristic
(AUROC) curves for both the derivation and validation co-
horts. Criteria to qualify for accuracy were as follows: 0.90–
1 p excellent, 0.80–0.90 p good, 0.70–0.80 p fair, and 0.60–
0.70 p poor .
The ability of the predictive index to discriminate between
SVR and non-SVR was assessed using several cutoffs. For ob-
taining the highest sensitivity and negative predictive value
(NPV), the lowest cutoff was established at quartile 0.25.
Conversely, for obtaining the highest specificity and positive
predictive value (PPV), the highest cutoff was established at
quartile 0.75. Finally, an intermediate cutoff fitting both the
highest sensitivity and the highest specificity was established
at quartile 0.5.
The diagnostic odds ratio (DOR), which expresses the
strength of the association between test results and disease,
was also determined. Briefly, DOR is the ratio of the odds of
a positive result in a given person with the target condition,
compared with a person without the condition . A DOR
of 1 suggests that the test provides no diagnostic evidence.
Finally, the likelihood ratios (LRs), which describe how many
times a person with the target condition is more likely to have
a particular test result, compared with a person without the
condition, were also calculated. LRs contribute to change, after
the test has been made, the probability that a target condition
is present. Binary tests have 2 LRs, positive and negative. A LR
of 1 indicates no diagnostic value.
All tests were 2-tailed, with P values !.05 considered signif-
icant. All statisticalanalyseswereperformedusingSPSS,version
16.0 (SPSS) and Stata, version 9.1 (Stata Corp).
prised 245 patients, who were split into 2 groups according to
the clinics at which they were treated: (1) the derivation cohort,
based at Hospital Carlos III, Madrid (
patients), which was used to develop the predictive index, and
(2) the validation cohort, including patients from 2 distinct
clinics ( ;35.1%), whichwasusedtovalidatetheaccuracyn p 86
of the index to predictSVR inanindependentgroupofpatients.
Table 1 displays the main characteristics of the study popula-
tion. Groups were well matched in terms of most baseline
characteristics except for liver stiffness and serum HCV RNA
level, which displayed higher values in the validation cohort.
All but 2 individuals (99%) were white, with a median body
weight of 23.2 kg (range, 20.1–25.3 kg). pegIFN alpha 2a was
the most commonly used modality, as only 12% of patients in
this series had been treated with pegIFN alpha 2b. There were
no significant differences in treatment outcome when com-
paring subjects treated with distinct pegIFNmodalities.Overall,
sustained HCV clearance was achieved by ∼55% of patients
included in the derivation and validation cohorts.
Predictive index of SVR.
In the derivation group, several
baseline variables associated with SVR by forward stepwise lo-
gistic regression analysis were identified (Table 2). Both serum
HCV RNA level (in log IU/mL) and liver stiffness (in kPa)
were evaluated as continuous variables,usinglogarithmictrans-
formations. HCV genotype was categorized as genotypes 1 and
4 versus genotypes 2 and 3, and the rs12979860 polymorphism
was categorized as CT or TT versus CC. With these definitions,
all variables were significantly inversely associated with SVR.
The strength of this negative association was higher for the SNP
CT/TT versus CC (odds ratio [OR], 0.170; 95% confidence
interval [CI], 0.065–0.442), followed by HCV genotypes 1 and
4 versus genotypes 2 and 3 (OR, 0.186; 95% CI, 0.091–0.381),
high serum HCV RNA level (OR, 0.212; 95% CI, 0.078–0.577),
and elevated liver stiffness (OR, 0.920; 95% CI, 0.865–0.977).
All these variables were used to build the predictive index,
; 64.9% of alln p 159
1214 • CID 2010:51 (15 November) • HIV/AIDS
named Prometheus, which can be downloaded for free at http://
ability of achieving SVR estimated by the logistic model was
obtained with the following transformation:
x p [13.940+ ?1.549?HCV genotype 1&4
+ ?1.682?log HCV VL+ ?0.084?Stiffness
+(?1.772?rs12979860 CT or TT)] .
AUROC values for the predictive index were similar in the
derivation cohort (0.892; 95% CI, 0.843–0.942) and validation
cohort (0.848; 95% CI, 0.763–0.933), ensuring the reliability of
the predictive index of SVR in the external cohort (Figure 1).
Diagnostic accuracy to predict SVR.
off (0.25), 41 patients in the derivation group were correctly
identified as nonresponders (true-negative findings without
SVR), with 5 misclassified patients (false-negative findings with
SVR) (Table 3). Thus, lack of SVR was predicted with 89%
certainty (NPV). The negative LR was 0.10, and the DOR was
∼20. For the validation cohort, values for sensitivity, NPV, neg-
ative LR, and DOR were similar to those obtained for the der-
Using the highest cutoff (0.75), 52 patients in the derivation
cohort were correctly identified as responders (true-positive
findings with SVR), with 6 patients misclassified (false-positive
findings without SVR) (Table 3). Thus, SVR was predicted with
89.7% certainty (PPV). The positive LR was 7.37, and the DOR
was 17.08. For the validation cohort, values for specificity, NPV,
negative LR, and DOR were similar to those obtained for the
When the intermediate cutoff (0.50) was used in the deri-
vation cohort, 129 patients were correctly identified (70 true-
positive findings and 59 true-negative findings), leaving 30 pa-
tients misclassified (14 false-positive findings and 16 false-neg-
ative findings) (Table 3). Thus, certainty for achieving SVR or
not achieving SVR was 83.3% and 78.7%, respectively. The
DOR was 18.44. For the validation cohort, values forsensitivity,
specificity, PPV, NPV, positive LR, negative LR, and DOR were
similar to those obtained for the derivation cohort.
Using the lowest cut-
The accurate prediction of the likelihood of response to
pegIFN-RBV before initiation of therapy is of much interest
for the identification of potentially curable HIV-HCV–coin-
fected patients. For this purpose, baselineindependentvariables
predicting treatment outcomes must be identified. Ideally, they
should be easy to obtain, and the predictive index must be
validated in an independent cohort, showing high accuracy and
cost-effectiveness . It should be noted that our predictive
model was built on the basis of a population that excluded
nonadherent patients and patients who discontinued therapy
because of side effects and only evaluated the subjects who had
completed a course of therapy. Thus, selection of candidates,
good adherence to drug therapy, and adequate management of
side effects will maximize the predictive value of our index [19,
30]. In clinical practice, baseline comorbidities, such as alco-
hol abuse, neuropsychiatric disorders, active drug abuse, devel-
opment of drug-related adverse events, and prior failure of in-
terferon-based therapies, could act as confounding variables,
reducing the power of prediction using this index. Moreover,
it must be highlighted that both the original and the valida-
tion groups in our study were mainly represented by white
individuals, most of whom had normal body mass indexes.
Thus, the performance of our model in different ethnicities
or in obese (or slim) persons might be lower.
Given that our model was developed in HIV-HCV–coin-
fected populations, in whom other conditions interfering with
treatment outcomes (eg, insulin resistance, drug interactions,
and immunodeficiency) are generally more frequent than in
HCV-monoinfected persons [3, 7, 31, 32], it is reasonable to
assume that our predictive model might perform even better
in HCV-monoinfected patients. Studies are ongoing to test this
hypothesis. If true, the HIV status should be explored as an
independent variable capable of improving the accuracy of this
score, to generalize its use for chronically HCV-infected pa-
tients, either coinfected with HIV or not . It is noteworthy
that HIV variables, at least considering a broad range of CD4
cell counts and viral load values, did not seem to influence the
results of using our model. However, since we did not have a
control group of HCV-monoinfected individualswhohadcom-
pleted a course of pegIFN-RBV therapy, we could not exclude
that HIV infection by itself might have an impact on the per-
formance of the model.
The 4 variables included in our final model have previously
been shown to predict achievement of SVR [12, 33, 34], al-
though prior attempts to model SVR by combining baseline
predictors of response are scarce . To our knowledge, this
is the first attempt to evaluate the predictive value of the newest
treatment outcome predictors in conjunction in a single score
based on noninvasive tools prior to initiation of therapy. Of
note, genetic IL28B testing and liver fibrosis assessment using
elastometry are incorporated for the first time. The use of elas-
tometry instead of histological analysis for assessment of liver
fibrosis stage has several advantages in addition to its nonin-
vasive nature. Liver stiffness has shown to accurately predict
HIV/AIDS • CID 2010:51 (15 November) • 1215
chronic hepatitis C . Moreover, the wide linear range of
values using elastometry (0–74) allowed more accurate predic-
tion of SVR than did the few nonlinear histologic fibrosis stages
(F0–F4, using the Metavir score) . Altogether, we think that
elastography may have fitted better in our model while ren-
dering it more easily available and cost-effective.
Other predictors of SVR have used early viral kinetics while
receiving treatment (eg, at week 4) [2, 6]. In HIV-HCV–coin-
fected patients, PPV values of ∼75% for HCV genotypes 1 and
4, and even greater values for HCV genotypes 2 and 3, have
been reported for subjects achieving rapid virological response
while receiving therapy . These figures are similar to those
obtained using our model, which is based on baseline parame-
ters, before initiation of treatment. Thus, it might provide infor-
mation on the probability of response without the need to be-
The values obtained using our proposed predictive index ex-
press a probability that ranges from 0 to 1, giving a tailored
estimate at baseline of the chances of clearance of HCV if a
course of therapy is conducted according to current recom-
mendations. As for many other prognostic tools used in clinical
practice, stratifying the outcomes and the risk of nonresponse
may guide therapeutic decisions . The quartiles we used
categorized the chances of clearance of HCV as high, inter-
mediate, or low when probabilities predicted by the modelwere
10.75, 0.75–0.50, and !0.50, respectively. Initiation of therapy
prior to the establishment of end-organ disease is a preemptive
approach that could be encouraged by prognostic scores. In
this regard, the cost-effectiveness of early interventions, when
the chances for HCV eradication are the greatest, should be
explored through longitudinal studies in patients in whom the
risk of liver fibrosis progression is well established .
In summary, the predictive SVR index proposed here that is
based on 4 baseline noninvasive parameters may be of great
value for making adequatetherapeuticdecisionsforHIV-HCV–
coinfected patients. Use of current hepatitis C therapy could
be encouraged for subjects without advancedliverfibrosiswhen
the chances of response are the greatest. Conversely, advice to
wait for new direct-acting antivirals against HCV may be more
appropriate for subjects with mild liver fibrosis who have min-
imal chances of response to pegIFN-RBV treatment. As with
any other diagnostic tools used in clinical practice, misclassi-
fication may occur, and understanding the limits of estimated
outcomes with predictive indexes is important before its wide-
de Inmunodeficiencia Adquirida (SIDA), Network of EuropeanAIDSTrials
(European Union 6th Framework Programme; LSHP-CT-2006–037570),
Fundacio ´n para la Investigacio ´n y la Prevencio ´n del SIDA en Espan ˜a
(360799/09), Red de Investigacion en SIDA (RIS, FIS-RD06/0006), Agencia
Fundacion Investigacion y Educacion en Sindrome
Lain Entralgo, Fundacio ´n Progreso y Salud of Consejerı ´a de Salud de la
Junta de Andalucia (AI-0021) and Instituto de Salud Carlos III (Rı ´o Hor-
tega; CM009, PI07/90201, UIPY-1467/07 and PI08/0738).
Potential conflicts of interest.
All authors: no conflicts.
1. Lauer G, Walker B. Hepatitis C virus infection. N Engl J Med 2001;
2. Soriano V, Puoti M, Sulkowski M, et al. Care of patients coinfected
with HIV and hepatitis C virus: 2007 updated recommendations from
the HCV-HIV International Panel. AIDS 2007;21:1073–1089.
3. Graham C, Baden L, Yu E, et al. Influence of HIV infection on the
course of hepatitis C virus infection: a meta-analysis. Clin Infect Dis
4. Martin-Carbonero L, Benhamou Y, Puoti M, et al. Incidence and pre-
dictors of severe liver fibrosis in HIV-infected patients with chronic
hepatitis C: a European collaborative study. Clin Infect Dis 2004;38:
5. Weber R, Sabin C, Friis-Moller N, et al. Liver-related deaths in persons
infected with the human immunodefiency virus: the D:A:D study. Arch
Intern Med 2006;166:1632–1641.
6. Ghany M, Strader D, Thomas D, Seeff L, American Association for
the Study of Liver Diseases. Diagnosis, management, and treatment of
hepatitis C: an update. Hepatology 2009;49:1335–1374.
7. Medrano J, Barreiro P, Resino S, et al. Rate and timing of hepatitis C
virus relapse after a successful course of pegylated interferon plus ri-
bavirin in HIV-infected and HIV-uninfected patients. Clin Infect Dis
8. Mehta S, Genberg B, Astemborski J, et al. Limited uptake of hepatitis
C treatment among injection drug users. J Community Health 2008;
9. Grebely J, Raffa J, Lai C, et al. Low uptake of treatment for hepatitis
C virus infection in a large community-based study of inner city res-
idents. J Viral Hepat 2009;16:352–358.
10. Rauch A, Egger M, Reichen J, Furrer H, Swiss HIV Cohort Study.
Chronic hepatitis C in HIV-infected patients: low eligibility and ap-
plicability of therapy with pegylated interferon-alpha plus ribavirin. J
Acquir Immune Defic Syndr 2005;38:238–240.
11. Kau A, Vermehren J, Sarrazin C. Treatment predictors of a sustained
virologic response in hepatitis B and C. J Hepatol 2008;49:634–651.
12. Ge D, Fellay J, Thompson A, et al. Genetic variation in IL28B predicts
hepatitis C treatment-induced viral clearance. Nature 2009;461:399–
13. Suppiah V, Moldovan M, Ahlenstiel G, et al. IL28B is associated with
response to chronic hepatitis C interferon-alpha and ribavirin therapy.
Nat Genet 2009;41:1100–1104.
14. Tanaka Y, Nishida N, Sugiyama M, et al. Genome-wide association of
IL28B with response to pegylated interferon-alpha and ribavirin ther-
apy for chronic hepatitis C. Nat Genet 2009;41:1105–1109.
15. Rauch A, Kutalik Z, Descombes P, et al. Genetic variation in IL28B is
associated with chronic hepatitis C and treatment failure: a genome-
wide association study. Gastroenterology 2010;138:1338–1345.
16. Thomas D, Thio C, Martin M, et al. Genetic variation in IL28B and
spontaneous clearance of hepatitis C virus. Nature 2009;461:798–801.
17. Rallon N, Naggie S, Benito JM, et al. Association of a single nucleotide
polymorphism near the interleukin-28B gene with response to hepati-
tis C therapy in HIV/hepatitis C virus-coinfected patients. AIDS 2010;
18. Moons K, Altman D, Vergouwe Y, Royston P. Prognosis and prognostic
research: application and impact of prognostic models in clinical prac-
tice. BMJ 2009;338:b606.
19. Royston P, Moons K, Altman D, Vergouwe Y. Prognosis andprognostic
research: developing a prognostic model. BMJ 2009;338:b604.
20. Moons K, Royston P, Vergouwe Y, Grobbee D, Altman D. Prognosis
and prognostic research: what, why, and how? BMJ 2009;338:b375.
21. Medrano J, Resino S, Vispo E, et al. Hepatitis C virus (HCV) treatment
1216 • CID 2010:51 (15 November) • HIV/AIDS Download full-text
uptake and changes in the prevalence of HCV genotypes in HIV-HCV
coinfected patients. J Viral Hepat 2010; published online 27 April.
22. Chevaliez S, Bouvier-Alias M, Brillet R, Pawlotsky JM. Hepatitis C
virus (HCV) genotype 1 subtype identification in new HCV drug de-
velopment and future clinical practice. PLoS One 2009;4:e8209.
23. Ziol M, Handra-Luca A, Kettaneh A, et al. Noninvasive assessment of
liver fibrosis by measurement of stiffness in patients with chronic hep-
atitis C. Hepatology 2005;41:48–54.
24. Castera L, Foucher J, Bernard P, et al. Pitfalls of liver stiffness mea-
surement: a 5-year prospective study of 13,369 examinations. Hepa-
25. Castera L, Vergniol J, Foucher J, et al. Prospective comparison of tran-
sient elastography, Fibrotest, APRI, and liver biopsy for the assessment
of fibrosis in chronic hepatitis C. Gastroenterology 2005;128:343–350.
26. de Ledinghen V, Douvin C, Kettaneh A, et al. Diagnosis of hepatic fi-
brosis and cirrhosis by transient elastography in HIV/hepatitis C vi-
rus-coinfected patients. J Acquir Immune Defic Syndr 2006;41:175–
27. Kirk G, Astemborski J, Mehta S, et al. Assessment of liver fibrosis by
transient elastography in persons with hepatitis C virus infection or
HIV-hepatitis C virus coinfection. Clin Infect Dis 2009;48:963–972.
28. O’Driscoll D, Foster A, Davey M, Nixon G, Horne R. Can actigraphy
measure sleep fragmentation in children? Arch Dis Child 2010; pub-
lished online 1 June.
29. Grimes D, Schulz K. Refining clinical diagnosis with likelihood ratios.
30. Altman D, Vergouwe Y, Royston P, Moons K. Prognosis and prognostic
research: validating a prognostic model. BMJ 2009;338:b605.
31. Blanco F, Barreiro P, Ryan P, et al. Risk factors for advanced liver
fibrosis in HIV-infected individuals: role of antiretroviral drugs and
insulin resistance. J Viral Hepat 2010; published online 18 January.
32. Moucari R, Ripault M, Martinot-Peignoux M, et al. Insulin resistance
and geographical origin: major predictors of liver fibrosis and response
to peginterferon and ribavirin in HCV-4. Gut 2009;58:1662–1669.
33. Dahari H, Layden-Almer J, Kallwitz E, et al. A mathematical model
of hepatitis C virus dynamics in patients with high baseline viral loads
or advanced liver disease. Gastroenterology 2009;136:1402–1409.
34. Labarga P, Vispo E, Barreiro P, et al. Rate and predictors of success in
the retreatment of chronic hepatitis C virus in HIV/HCV-coinfected
patients with prior nonresponse or relapse. J Acquir Immune Defic
35. Tuma P, Jarrin I, Del Amo J, et al. Survival of HIV-infected patients
with compensated liver cirrhosis. AIDS 2010;24:745–753.
36. Altman D, Royston P. The cost of dichotomising continuous variables.
37. Royston P, Moons K, Altman D, Vergouwe Y. Prognosis andprognostic
research: Developing a prognostic model. BMJ 2009;338:b604.
38. Nunes D, Fleming C, Offner G, et al. Noninvasive markers of liver fi-
brosis are highly predictive of liver-related death in a cohort of HCV-
infected individuals with and without HIV infection. Am J Gastroen-