www.elsevier.com/locate/jhep

Journal of Hepatology 49 (2008) 916–922

Application of transient elastometry to differentiate mild frommoderate to severe liver fibrosis in HIV/HCV co-infected patientsq

Juan Macıas1,11, Eva Recio2,11, Eugenia Vispo3, Antonio Rivero4,11,Luis F. Lopez-Cortes5,11, M Jose Rıos6,11, Dolores Merino7,11, Mercedes Gonzalez8,11,

Pablo Barreiro3, Victor de Ledinghen9, Carmen Quereda10, Juan A. Pineda2,11,*

1Clinical Unit of Infectious Diseases, Department of Internal Medicine, Seville, Spain2Clinical Unit of Infectious Diseases, Hospital Universitario de Valme, Seville, Spain

3Department of Infectious Diseases, Hospital Carlos III, Madrid, Spain4Clinical Unit of Infectious Diseases, Hospital Universitario Reina Sofıa, Cordoba, Spain

5Department of Infectious Diseases, Hospital Universitario Virgen del Rocıo, Seville, Spain6Clinical Unit of Infectious Diseases, Department of Internal Medicine, Hospital Universitario Virgen Macarena, Seville, Spain

7Department of Internal Medicine, Hospital Juan Ramon Jimenez, Huelva, Spain8Infectious Diseases Unit, Department of Internal Medicine, Hospital Universitario Virgen de la Victoria, Malaga, Spain

9Centre d’Investigation de la Fibrose Hepatique, Hopital Haut-Leveque, Bourdeaux, France10Service of Infectious Diseases, Hospital Ramon y Cajal, Madrid, Spain

11Grupo para el Estudio de las Hepatitis Vıricas de la Sociedad Andaluza de Enfermedades Infecciosas, Spain

Background/Aims: Transient elastometry (TE) is accurate for detecting cirrhosis (F = 4) in human immunodeficiency

virus (HIV)/hepatitis C virus (HCV) co-infected patients. However, this procedure is less precise to differentiate mild(F 6 1) from moderate to severe (F P 2) fibrosis using the cut-off value of 7.2 kPa, a level previously proposed by some

authors. Because of this, we elaborated and validated cut-off values of liver stiffness (LS) to better discriminate F 6 1 from

F P 2 in HIV/HCV co-infected subjects to aid therapy decisions.

Methods: One hundred and ninety-seven co-infected patients with liver biopsy and TE measurement, without prior ther-

apy against HCV infection, were included.

Results: To diagnose F P 2, a cut-off of 9.0 kPa showed a positive predictive value of 87%. To discard F P 2, a cut-off

of 6.0 kPa showed a negative predictive value of 90%. Considering all the patients, 61 (31%) patients yielded LS values

66.0 kPa and 81 (41%) patients showed LS values P9.0 kPa. There were no severe classification errors as the NPV ofLS 6 6.0 kPa for F P 3 was 100% and the NPV LS P9.0 kPa for F = 0 was also 100%.

Conclusions:The usefulness of TE can be enhanced using two different cut-off values to identify patients with F 6 1 and F P 2.

� 2008 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved.

Keywords: HCV; HIV; Transient elastometry; Liver fibrosis; Co-infection

0168-8278/$34.00 � 2008 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved.

doi:10.1016/j.jhep.2008.07.031

Received 20 February 2008; received in revised form 22 June 2008; accepted 14 July 2008; available online 1 October 2008

Associate Editor: F. Negro

The authors who have taken part in the research of this paper declared that they do not have a relationship with the manufacturers of the deviceinvolved either in the past or present and they did not receive funding from the manufacturers to carry out their research.

* Corresponding author. Tel.: +34 9550158887.E-mail address: japineda@telefonica.net (J.A. Pineda).Abbreviations: HIV, human immunodeficiency virus; HCV, hepatitis C virus; TE, transient elastometry; LS, liver stiffness; HAART, highly active

antirretroviral therapy; ROC, receiver-operating-characteristic; S, sensitivity; Sp, specificity; PPV, positive predictive value; NPV, negative predictivevalue.

J. Macıas et al. / Journal of Hepatology 49 (2008) 916–922 917

1. Introduction

Liver disease is currently the first cause of death ofHIV-infected patients with chronic hepatitis C in areaswhere highly active antiretroviral therapy (HAART) iseasily accessible [1]. Treatment of hepatitis C virus(HCV) infection with pegylated interferon plus ribavirinis associated with a decrease in liver-related mortality inthis setting [1]. However, the efficacy of anti-HCV therapyin HIV-infected patients with chronic hepatitis C is lowerthan in HCV-monoinfected subjects [2,3]. In addition, thetoxicity of interferon-based therapy in HIV/HCV co-infected patients is considerable [4]. For these reasons, itis common practice to select patients with higher risk ofdeveloping decompensated liver disease, particularly thosewith a lower chance of achieving sustained virologicalresponse, as candidates for anti-HCV therapy. Accordingto international recommendations by panels of experts,the stage of liver fibrosis has been proposed as a criterionto select patients with progressive liver disease. Thus,patients with moderate to severe fibrosis, i.e. fibrosisextending beyond the portal tracts or more advancedfibrosis, would be considered candidates to therapy [5,6].For this reason, it is important to screen for moderate tosevere fibrosis to take decisions regarding indication ofanti-HCV therapy. On the other hand, liver toxicity asso-ciated with antiretroviral therapy is more likely in patientswith higher degrees of fibrosis [7–9]. Given that acute toxichepatitis due to antiretroviral drugs can be fatal, liverfibrosis screening can help to select drugs with a better liversafety profile in co-infected patients.

Liver biopsy is the gold standard to stage fibrosis inchronic hepatitis C. However, this is an invasive proce-dure that requires expertise, may cause serious side effects[10], is poorly accepted by patients and is costly [11]. Inaddition, variability in fibrosis scoring is a concern[12,13]. Because of these drawbacks, non-invasive mark-ers as surrogates of liver biopsy are under investigation.Transient elastometry (TE) is a non-invasive procedurethat measures liver stiffness (LS), a parameter that corre-lates highly with liver fibrosis. TE achieves a high degreeof accuracy in detecting liver fibrosis in HCV infected sub-jects [14–16]. Cirrhosis can be ruled out or diagnosedaccurately in HIV/HCV co-infected patients [17,18].However, the discrimination of mild from moderate tosevere liver fibrosis is not so accurate in these patients[17,18]. Using the LS cut-off value of 7.2 kPa, previouslyelaborated in HCV-monoinfected patients, 24% patientswere misclassified as having mild fibrosis while theyshowed moderate to severe fibrosis at liver biopsy [18].If this cut-off value were used to screen for moderate tosevere fibrosis with the aim of selecting patients for anti-HCV therapy, an important proportion of patients withmoderate to severe fibrosis would not receive therapy.For these reasons, more accurate LS cut-off points areneeded.

We have elaborated and validated LS cut-off valuesto discriminate HIV/HCV co-infected patients withmoderate to severe fibrosis from those without substan-tial fibrosis, with the aim of improving the clinical appli-cability of TE.

2. Patients and methods

2.1. Study design and patients

This was a cross-sectional study performed in ten tertiary-carehospitals in Spain, France and Italy. From July 2003 to May2007, we included in this study all patients seen in the participatingcenters with detectable HCV RNA who underwent a liver biopsyand fulfilled the following criteria: (a) Older than 18 years; (b)Assessment of liver fibrosis by biopsy within 12 months before orafter TE; (c) No therapy against HCV infection between liver biopsyand TE.

2.2. Measurement of liver stiffness

Trained operators determined the stiffness of the liver with Fibro-ScanTM (Echosens, Paris, France) [19]. The operators were blinded tothe clinical data of the patients. Liver tissue was localized by A-modeimages provided by the probe transducer placed in the right intercostalspaces. A minimum of 10 measurements were obtained, the median ofwhich as expressed in kilopascal (kPa) units was assumed as represen-tative of overall LS.

2.3. Liver histology

Most patients underwent a liver biopsy to stage fibrosis as part ofthe work up before starting HCV treatment. Liver fibrosis was scoredfollowing the METAVIR index [20]. Fibrosis is classified into five cat-egories (stages 0 to 4) according to this index. The scoring systemdefines fibrosis stage 0 as absent fibrosis, stage 1 as fibrotic portalexpansion without septa formation, stage 2 as fibrotic stellate expan-sion of portal tracts with a few septa formation, stage 3 as fibroticexpansion of portal tracts with numerous septa, and stage 4 as cirrho-sis. Fibrosis stages 0 and 1 were considered mild fibrosis (F 6 1) andstages equal or over 2 are considered moderate to severe fibrosis(F P 2) for the present study. The length of liver biopsy specimens,expressed in millimeters (mm), was recorded. Two central pathologistsstaged the liver biopsies from 134 (70%) patients. The agreementbetween them was excellent for the diagnosis of moderate to severefibrosis (j = 0.86). Discordant fibrosis staging was resolved by consen-sus reading. The agreement between the central pathologists and theoriginal staging was good for the diagnosis of moderate to severe fibro-sis (j = 0.71–0.74).

2.4. Statistical analysis

The categorical variables were expressed as numbers (percentage)and the continuous variables as median (Q1–Q3). Categorical variableswere compared using the v2 test or the Fisher’s exact test, where appro-priate. The Student’s t test or the Mann–Whitney U test, if applicable,was used to compare quantitative variables.

LS values from a randomly generated split-sample of 50% patientswere used to elaborate LS cut-off values, and data from the remainingpatients were used to validate these cut-off values. LS values and fibro-sis stages of both groups were compared to rule out any significant dif-ference. The general characteristics of the elaboration and validationgroups were also compared.

The diagnostic value of LS values was tested calculating the areaunder the receiver-operating-characteristic (ROC) curve. The cut-offpoints were selected from the ROC curve to maximize the predictive

918 J. Macıas et al. / Journal of Hepatology 49 (2008) 916–922

values and minimize the percentage of misclassifications. The closestwhole LS values were selected as cut-off points. The diagnostic accu-racy was calculated by sensitivity (S), specificity (Sp), positive predic-tive value (PPV), and negative predictive value (NPV), consideringmoderate to severe fibrosis (F P 2) as the disease. The statistical anal-ysis was carried out using the SPSS 14 statistical software package(SPSS, Chicago, Illinois, USA).

2.5. Ethical aspects

The study was performed according to the Helsinki declaration andwas approved by the Ethics committee of Hospital Universitario deValme.

3. Results

3.1. Characteristics of study population

Two hundred and twenty-two patients underwent aliver biopsy and TE during the period of study. Onehundred and ninety-seven patients fulfilled the inclusioncriteria. Seven (3%) patients were excluded because lessthan 10 successful LS acquisitions were obtained. Eigh-teen patients were excluded because the period of timefrom the liver biopsy to the LS measurement exceeded12 months. One hundred and forty-two (72%) patientswere recruited in four out of the ten participant centers.The main characteristics of the patients included in thestudy are summarized in the Table 1. One hundredand five (53%) showed F P 2 in the liver biopsy. Themedian (Q1–Q3) LS value was 7.7 (5.6–14.2). The med-ian LS values of the elaboration and validation groupswere 7.7 (5.9–13.5) and 7.7 (5.3–15.5), respectively(p = 0.6). The proportions of patients with F P 2 were53 (53.5%) for the elaboration group and 50 (51%) forthe validation group (p = 0.7). The remaining character-

Table 1

Characteristics of study population: elaboration and validation groups

Parameter All patients (n = 197) E

Age (years)b 42 (39–46) 4Male sex, n (%) 144 (85) 7BMIc (kg/m2)b 22.8 (20.7–24.9) 2HBsAg positive, n (%) 4 (2) 2HCV genotype, n (%)

1 or 4 161 (82) 72 or 3 36 (18) 2

Time between LBd and TEe (months)b 1.6 (0.9–4.4) 1Liver fibrosis, n (%)

F = 0 20 (10) 9F = 1 72 (37) 3F = 2 36 (18) 1F = 3 29 (15) 1F = 4 40 (20) 2

Size of LB (mm)b 19 (15–25) 1

a For the comparison between the elaboration and validation sets.b Median (Q1–Q3).c BMI, body mass index.d LB, liver biopsy.e TE, transient elastometry.

istics of the groups of patients did not differ significantly(Table 1).

3.2. Diagnosis of absent to mild and moderate to severe

fibrosis

The diagnostic value of TE was assessed in the elab-oration group by the ROC curve for F P 2. The areaunder the ROC curve (95% confidence interval) was0.86 (0.78–0.93) for F P 2 (Fig. 1). The whole LS valuewith highest ability to discriminate absence of F P 2was 6 kPa (Table 2). The highest yield to differentiateF P 2 was obtained with the whole LS value of 9 kPa(Table 2). The diagnostic accuracy of these cut-off pointswas tested in the validation group (Table 2).

Using 66 kPa to discriminate F 6 1, five (18%) of 28patients in the elaboration group were misclassified. Allof them were staged as F = 2 in the liver biopsy. In thevalidation group, 5 (15%) patients with LS 66 showedF = 2. No patients with LS values 66 kPa were stagedas pre-cirrhosis or cirrhosis, F P 3, in the liver biopsy.Thus, the NPV of this cut-off for F P 3 was 100%.

Using P9 kPa to diagnose F P 2, seven (17%) of 41patients in the elaboration group were misclassified. Allof them were staged as F = 1 in the liver biopsy. Five(12.5%) of 40 patients in the validation group showedLS P9 kPa and F = 1 in the liver biopsy. None of thepatients with LS values P9 kPa were staged as absentfibrosis, F = 0, in the liver biopsy. Consequently, theNPV of this cut-off for F = 0 was 100%.

Considering the elaboration and validation groupstogether, 61 (31%) patients yielded LS values 66 kPaand 81 (41%) patients showed LS values P9 kPa. Fiftyfive (28%) patients yielded LS values >6 kPa and

laboration group (n = 99) Validation group (n = 98) pa

2 (39–46) 42 (38–46) 0.53 (73) 83 (83) 0.13.1 (20.5–25.2) 22.5 (20.5–24.5)(2) 2 (2) 0.8

0.79 (80) 82 (84)0 (20) 16 (16).1 (0–5.5) 2.3 (0.8–8.1) 0.8

0.5(9,1) 11 (11.2)

5 (35.4) 37 (37.8)9 (19.2) 17 (17.3)6 (16.2) 13 (13.3)0 (20.1) 20 (20.4)8 (13–21) 19 (15–22) 0.3

Fig. 1. ROC curves of liver stiffness values measured by transient

elastometry to discriminate absent to mild (F 6 1) from moderate to

severe (F P 2) liver fibrosis in HIV/HCV co-infected patients (elabora-

tion group, dashed line; validation group, solid line).

J. Macıas et al. / Journal of Hepatology 49 (2008) 916–922 919

<9 kPa, that would be regarded as indeterminate valuesusing these cut-off points. Ten (16%) patients with LSvalues 66 kPa and 12 (15%) patients with LS valuesP9 kPa were misclassified. A sub-analysis restricted topatients with larger liver biopsy specimens and lowervariability in LS values showed lower rates of diagnosticerrors than the global population (Table 3).

4. Discussion

We found that LS cut-off values aimed at discriminat-ing HIV/HCV co-infected patients with and without

Table 2

Optimal cut-off values of liver stiffness for discriminating mild fibrosis

(F 6 1) from moderate to severe fibrosis (F P 2) in HIV/HCV co-

infected patients

Elaborationgroup (n = 99)

Validationgroup (n = 98)

F 6 1. Cut-off value (kPa) 6 6Sensitivity (%) 50 69Specificity (%) 91 85PPV (%)a 68 58NPV (%)b 82 90Misclassified patients, n (%) 5 (18) 5 (15)

F P 2. Cut-off value (kPa) P 9Sensitivity (%) 85 90Specificity (%) 64 70PPV (%)a 83 87NPV (%)b 67 74Misclassified patients, n (%) 7 (17) 5 (12.5)

a PPV: positive predictive value.b NPV: negative predictive value.

F P 2 proposed in this study yielded a better perfor-mance than previously reported values [17,18]. Particu-larly, the prediction of mild fibrosis was improvedwith lower rates of misclassification. Most importantly,none of the patients classified as F 6 1 using TE showedpre-cirrhosis or cirrhosis in the liver biopsy. In addition,no patients without fibrosis in the liver biopsy were clas-sified as F P 2 applying TE. Therefore, if these cut-offswere used to select patients with F P 2 as candidates toreceive anti-HCV therapy, no important decision mis-takes would be made.

The diagnostic accuracy of transient elastometry todiagnose F P 2 in the study patients using the LS cut-off of P9 kPa was similar to previous reports in HCV-monoinfected patients [14–16]. Indeed, this cut-off isclose to the first one reported in chronic hepatitis C[14]. However, later studies in patients with HCV infec-tion and without HIV co-infection proposed lower cut-off points [15,16]. Finally, the LS value P7.2 kPa hasbeen proposed as diagnostic of F P 2 in chronic hepati-tis C, with or without HIV co-infection [21]. We foundin a previous study that this cut-off point failed particu-larly to exclude F P 2 in HIV/HCV co-infected patients[18]. This was not surprising, as cut-off values elaboratedin HCV monoinfection to discriminate F P 2 yieldedlow negative predictive values [14–16]. In addition, thediagnostic yield of a test to detect a disease is linkedto the frequency of the disease [22]. The lower preva-lence of fibrosis stage 1 and 2 in HIV/HCV co-infectedpatients included in a previous study [18] worsened thediagnostic yield of the LS cut-off P7.2 kPa, which wasvalidated in HCV-monoinfected patients with higherfrequency of fibrosis stage 1 and 2 [15]. In this study,we aimed at detecting and discarding F P 2 with thehighest possible certainty in HIV/HCV co-infection.The best approach was to select two different cut-off val-ues, one with high NPV and another one with high PPV.The elaboration of two different cut-offs provided a bet-ter discrimination of mild and moderate to severe fibro-sis. On one hand, the exclusion of F P 2 was improvedusing the LS value of 66 kPa, compared with <7.2 kPa[18]. Thus, the NPV of LS value <7.2 kPa was 75% [18],whereas the NPV of LS value 66 kPa was 90% in thevalidation set. Moreover, the rates of misclassificationwere lower and the NPV of LS 6 6 kPa for pre-cirrhosisor cirrhosis was 100%, much higher than the cut-off<7.2 kPa [18]. On the other hand, the diagnostic yieldof P9 kPa was similar to that obtained withP7.2 kPa. However, the rate of misclassification waslower and the NPV of LS P 9 kPa for F = 0 was100%, higher than the cut-off P7.2 kPa [18].

The present study has some limitations. First, thiswas a retrospective cross-sectional study aimed at evalu-ating the diagnostic performance of TE. This designcould have affected the results if the criteria for selectingpatients who underwent liver biopsy had changed over

Table 3

Diagnostic performance of optimal cut-off values of liver stiffness for discriminating mild fibrosis (F 6 1) from moderate to severe fibrosis (F P 2) in the

global study population, in patients with liver biopsy length greater than 15 mm, and in patients with liver stiffness interquartile range (IQR) lower than

30%

All patients (n = 197) Liver biopsy P 15 mm (n = 148) Liver stiffness IQR 6 30% (n = 168)

F 6 1. Cut-off value 66 kPa

Sensitivity (%) 68 63 55Specificity (%) 84 93 92PPV (%)a 54 75 68NPV (%)b 90 88 87Misclassified patients, n (%) 10 (16) 6 (7) 7 (8)Patients with LS 66 kPa, n (%) 61 (31) 48 (32) 52 (31)

F P 2. Cut-off value P 9 kPa

Sensitivity (%) 87 88 87Specificity (%) 67 70 66PPV (%)a 85 88 84NPV (%)b 71 70 71Misclassified patients, n (%) 12 (15) 8 (12) 11 (13)Patients with LS P9 kPa, n (%) 81 (41) 65 (44) 68 (41)

a PPV, positive predictive value.b NPV, negative predictive value.

920 J. Macıas et al. / Journal of Hepatology 49 (2008) 916–922

the study period. However, recommendations for thescreening of HIV/HCV co-infected patients for anti-HCV therapy have not changed significantly duringthe study period [5,6]. Second, liver biopsy was per-formed as part of the screening before indicating pegy-lated interferon plus ribavirin in most cases.Abstinence from alcohol is advised in candidates indi-cated to anti-HCV treatment [5,6]. These patients canhave a lower frequency of advanced fibrosis and cirrho-sis than the general population of HIV/HCV co-infectedpatients. In addition, patients with clinically evidentdiagnosis of cirrhosis usually do not undergo liverbiopsy. As histological assessment of fibrosis was aninclusion criteria for the present study, these moreadvanced patients were excluded. The inclusion ofpatients with overt cirrhosis would have increased evenmore the diagnostic yield of transient elastometry todetect F P 2, as the frequency of patients with F P 2would be higher and TE performs better with higherdegrees of liver fibrosis [18]. Third, one fourth of thepatients yielded indeterminate results applying the LScut-offs proposed in this study. These patients needadditional evaluations to diagnose their liver fibrosis.However, the classification of patients is more accurateusing two LS cut-offs than a single one [17,18]. The highrate of misclassification of the LS value 7.2 kPa pre-cludes its clinical application to diagnose F 6 1 inHIV/HCV co-infected patients [18]. Finally, the propor-tion of patients with a diagnosis of fibrosis, and withouta need for further evaluations, is greater with TE thanwith other non-invasive techniques in HIV and HCVco-infection [23,24].

In spite of elaborating two different cut-off values, therates of misclassification were 7–15%. Discordant resultsbetween liver biopsy and TE could be due to the well-known limitations of biopsy or to lack of precision of

TE [12,13]. Liver fibrosis staging errors with liver biopsytend to occur in intermediate stages [13], like those thatwere tested in the present study. In this study, diagnosticerrors were less frequent in the subgroup of patientswith larger biopsy specimens, particularly for the classi-fication of patients with F = 1 in the biopsy. Regardinglack of precision of TE, we found that patients withlower variability in LS measurements showed lowerrates of disagreement between biopsy and TE. Thiswas also observed specifically among patients withF = 1 in the biopsy. LS values of patients with F = 1and F = 2 in the biopsy are very close [18]. Thus, theeffect of variability of LS in misclassifications wouldbe more pronounced in a LS cut-off value aimed to dis-tinguish F = 1 from F = 2.

The diagnostic errors using the two cut-offs to discardand detect moderate to severe fibrosis proposed in thisstudy were less relevant than with the LS value of7.2 kPa. With the latter value some patients withF P 3 were classified as mild or absent fibrosis. Withthe values elaborated in this study, patients misclassifiedas having F P 2 by TE were all staged as F = 1 in theliver biopsy. If patients with LS P9 kPa were selectedfor anti-HCV treatment, the decision to treat a fewpatients with F = 1 would not be a major managementmistake. On the other hand, a small percentage ofpatients with LS 66 kPa showed F = 2. If anti-HCVtreatment were deferred in patients with LS 66 kPa,some patients with F = 2 would not initially receive thistherapy. However, TE is easily repeated, unlike liverbiopsy, and these patients could be detected by follow-up TE.

Therapy decisions in HIV-infected patients withchronic hepatitis C, especially those with predictors ofnon-response, could be managed using the LS cut-offvalues elaborated in the present study (Fig. 2). Patients

Treat

Liver biopsy

Differ therapySequential

transient elastometry

Candidatefor anti-HCV therapy

Transient elastometry

6 kPa(Asume F 1)

6.1-8.9 kPa(Indeterminate result)

9 kPa(Asume F 2)

APRI +Forns index

APRI <1.5 andForns index<6.9

APRI 1.5 orForns index 6.9

Other Genotype 2 or 3 Genotype 1 and

HCV RNA<6x105

UI/mL

≥≥

≥≥

≤≤

Fig. 2. Proposal of stepwise algorithm for the diagnosis of fibrosis and management of HCV/HIVco-infected patients. In the first step, patients are

classified using transient elastometry (TE). Those with indeterminate results after applying TE could be evaluated using simple blood indexes. Finally,

factors associated with sustained virologic response could be taken into account.

J. Macıas et al. / Journal of Hepatology 49 (2008) 916–922 921

with LS values P9 kPa, diagnostic of moderate tosevere fibrosis, should be offered anti-HCV therapy. Inpatients with LS values 66 kPa, indicative of absent tomild fibrosis, anti-HCV treatment could be deferredand LS could be reassessed sequentially. On the otherside, in patients with indeterminate results, i.e. LS value>6 kPa and <9 kPa, non-invasive blood tests for thediagnosis of liver fibrosis could be used [23]. We havefound that using this approach 8% of the individualswith indeterminate TE results can be classified [25].Patients with indeterminate results of these bloodindexes could be offered liver biopsy, especially if theycarry high HCV RNA load or genotype 1 or 4, predic-tors of non-response to anti-HCV therapy. If this step-wise algorithm were applied, then only 20% patientswould need a liver biopsy to be screened for anti-HCVtherapy [25].

HCV infection is one of the main risk factors for livertoxicity associated with antiretroviral therapy [26].Among HIV/HCV co-infected patients, pre-therapyliver fibrosis could influence the likelihood of liverevents during antiretroviral treatment. Thus, the severityof liver damage is associated with the probability of hep-atotoxicity [7,8]. Particularly, the risk of liver toxicity fornon-nucleoside analogs and protease inhibitors seems tobe similar among patients with mild to moderate fibro-sis, but this risk increases for non-nucleoside analogsamong patients with more advanced fibrosis [7]. In thisregard, baseline assessment of liver fibrosis using TE

could help us to select of antiretroviral drug combina-tions for HIV/HCV co-infected patients. As a non-inva-sive test, TE could be applied as a screening tool todetect patients with moderate to severe fibrosis, in whomantiretroviral drugs with the highest liver safety profileare clearly preferred.

In conclusion, transient elastometry can be applied toHIV/HCV co-infected patients to select those with LSP9.0 kPa for anti-HCV therapy, and to defer the indi-cation of treatment against HCV in those with LS66.0 kPa. Most co-infected patients can be classifiedusing these LS cut-off values and, consequently, sparedliver biopsy. Patients with indeterminate results needadditional evaluations with serum markers of fibrosisor liver biopsy, if anti-HCV treatment guided by liverfibrosis is intended.

Acknowledgements

This study was partly supported by a grant from Ab-bott Laboratories and from Consejerıa de Salud de laJunta de Andalucıa (PI-0048/2007). The authors wishto thank the Ministerio de Sanidad y Consumo, Institu-to de Salud Carlos III, Red de SIDA from Spain fortheir support (ISCIII-RETIC RD06/006).

References

[1] Pineda JA, Garcıa-Garcıa JA, Aguilar-Guisado M, Rıos-VillegasMJ, Ruiz-Morales J, Rivero A, et al. Clinical progression of

922 J. Macıas et al. / Journal of Hepatology 49 (2008) 916–922

hepatitis C virus-related chronic liver disease in human immuno-deficiency virus-infected patients undergoing highly active anti-retroviral therapy. Hepatology 2007;46:622–630.

[2] Torriani FJ, Rodrıguez-Torres M, Rockstroh JK, Lissen E,Gonzalez-Garcıa J, Lazzarin A, et al. Peginterferon Alfa-2a plusribavirin for chronic hepatitis C virus infection in HIV-infectedpatients. N Engl J Med 2004;351:438–450.

[3] Carrat F, Bani-Sadr F, Pol S, Rosenthal E, Lunel-Fabiani F,Benzekri A, et al. Pegylated interferon alfa-2b vs standardinterferon alfa-2b, plus ribavirin, for chronic hepatitis C in HIV-infected patients: a randomized controlled trial. JAMA2004;292:2839–2848.

[4] Mauss S. Treatment of viral hepatitis in HIV-coinfected patients-adverse events and their management. J Hepatol2006;44:S114–S118.

[5] Alberti A, Clumeck N, Collins S, Gerlich W, Lundgren J, Palu G,et al. Short statement of the first European consensus conferenceon the treatment of chronic hepatitis B and C in HIV co-infectedpatients. J Hepatol 2005;42:615–624.

[6] Nelson M, Matthews G, Brook MG, Main J. BHIVA CoinfectionGuideline Committee; British HIV Association. BHIVA guide-lines on HIV and chronic hepatitis: coinfection with HIV andhepatitis C virus infection (2005). HIV Med 2005;6 (Suppl. 2): 96–106.

[7] Aranzabal L, Casado JL, Moya J, Quereda C, Diz S, Moreno A,et al. Influence of liver fibrosis on highly active antiretroviraltherapy-associated hepatotoxicity in patients with HIV andhepatitis C virus coinfection. Clin Infect Dis 2005;40:588–593.

[8] Mira JA, Macıas J, Giron-Gonzalez JA, Merino D, Gonzalez-Serrano M, Jimenez-Mejıas ME, et al. Incidence of and riskfactors for severe hepatotoxicity of nelfinavir-containing regimensamong HIV-infected patients with chronic hepatitis C. J Anti-microb Chemother 2006;58:140–146.

[9] Labarga P, Soriano V, Vispo ME, Pinilla J, Martin-Carbonero L,Castellares C, et al. Hepatotoxicity of antiretroviral drugs isreduced after successful treatment of chronic hepatitis C in HIV-infected patients. J Infect Dis 2007;196:670–676.

[10] Van Thiel DH, Gavaler JS, Wright H, Tzakis A. Liver biopsy. Itssafety and complications as seen at a liver transplant center.Transplantation 1993;55:1087–1090.

[11] Wong JB, Koff RS. Watchful waiting with periodic liver biopsyversus immediate empirical therapy for histologically mild chronichepatitis C. A cost-effectiveness analysis. Ann Intern Med2000;133:665–675.

[12] Bedossa P, Dargere D, Paradis V. Sampling variability of liverfibrosis in chronic hepatitis C. Hepatology 2003;38:1449–1457.

[13] Rousselet MC, Michalak S, Dupre F, Croue A, Bedossa P, Saint-Andre JP, et al. Sources of variability in histological scoring ofchronic viral hepatitis. Hepatology 2005;41:257–264.

[14] Ziol M, Handra-Luca A, Kettaneh A, Christidis C, Mal F,Kazemi F, et al. Noninvasive assessment of liver fibrosis bymeasurement of stiffness in patients with hepatitis C. Hepatology2005;41:48–54.

[15] Castera L, Vergniol J, Foucher J, Le Bail B, Chanteloup E, HasserM, et al. Prospective comparison of transient elastography,fibrostest, APRI, and liver biopsy for the assessment of fibrosisin chronic hepatitis C. Gastroenterology 2005;128:343–350.

[16] Foucher J, Chanteloup E, Vergniol J, Castera L, Le Bail B,Adhoute X, et al. Diagnosis of cirrhosis by transient elastography(FibroScan): a prospective study. Gut 2006;55:403–408.

[17] De Ledinghen V, Douvin C, Ketttaneh A, Ziol M, Roulot D,Marcellin P, et al. Diagnosis of hepatic fibrosis and cirrhosis bytransient elastography in HIV/hepatitis C virus-coinfectedpatients. J Acquir Immune Defic Syndr 2006;41:175–179.

[18] Vergara S, Macıas J, Rivero A, Gutierrez-Valencia A, Gonzalez-Serrano M, Merino D, et al. The utility of transient elastometry inassessing liver fibrosis in patients with HIV/HCV coinfection. ClinInfect Dis 2007;45:969–974.

[19] Sandrin L, Fourquet B, Hasquenoph JM, Yon S, Fournier C, MalF, et al. Transient elastography: a new non-invasive method forassessment of hepatic fibrosis. Ultrasound Med Biol2003;29:1705–1713.

[20] The French METAVIR Cooperative Study Group. Intraobserverand interobserver variations in liver biopsy interpretation inpatients with chronic hepatitis C. Hepatology 1994; 20: 15–20.

[21] Barreiro P, Martın-Carbonero L, Nunez M, Rivas P, Morente A,Simarro N, et al. Predictors of liver fibrosis in HIV-infectedpatients with chronic hepatitis C virus (HCV) infection: assess-ment using transient elastometry and the role of HCV genotype 3.Clin Infect Dis 2006;42:1032–1039.

[22] Irwig L, Bossuyt P, Glasziou P, Gatsonis C, Lijmer J. Designingstudies to ensure that estimates of test accuracy are transferable.BMJ 2002;324:669–671.

[23] Macıas J, Giron-Gonzalez JA, Gonzalez-Serrano M, Merino D,Cano P, Mira JA, et al. Prediction of liver fibrosis in humanimmunodeficiency virus/hepatitis C virus coinfected patients bysimple non-invasive indexes. Gut 2006;55:409–414.

[24] Myers RP, Benhamou Y, Imbert-Bismut F, Thibault V, BochetM, Charlotte F, et al. Serum biochemical markers accuratelypredict liver fibrosis in HIV and hepatitis C virus co-infectedpatients. AIDS 2003;17:721–725.

[25] del Valle J, Macıas J, Barreiro P, et al. Improving the differen-tiation of mild from significant liver fibrosis in HIV/HCV-co-infected patients using transient elastometry. In: 15th conferenceon retroviruses and opportunistic infections. Boston, USA, 2008.[Abstract 1053]..

[26] Nunez M. Hepatotoxicity of antiretrovirals: incidence, mecha-nisms and management. J Hepatol 2006;44:S132–S139.