Worldwide prevalence of hepatitis C virus (HCV) infection in human immunodeficiency virus (HIV) infected patients is about 33% 1 2 3 and this pathology represents one of the main comorbidities in this population 4 5 . Before the use of protease inhibitors for genotype 1 HCV infection, combination of ribavirin (RBV) and pegylated interferon-α (PEG-IFN) was the standard of care for treatment of chronic HCV patients and it is still the standard for non-1 genotype infection with or without HIV coinfection. All guidelines to adjust RBV doses have been defined in monoinfected patients but are also used in HIV-HCV ones. According to these guidelines and excepted for genotype 2 or 3 infected patients, RBV doses must be adapted to body weight 6 . Studies have shown that high RBV plasma concentrations could improve early virological response rates 7 8 . However, RBV daily dose, even adjusted for body weight is poorly correlated with RBV plasma concentration 9 10 due to a large inter-individual variability in exposure 11 12 13 . Different explanations have been proposed for this variability: i) the lean body weight would be a better covariate with a linear influence on RBV clearance 14 ; ii) RBV exposure depends on creatinine clearance 15 ; iii) the RBV pharmacokinetics could be impacted by the pharmacological or pathological context of HIV-HCV coinfection 16 . Loustaud-Ratti et al., showed that RBV plasma exposure after a single dose of 600 mg in HCV monoinfected patients was predictive of sustained virological response (SVR) and proposed a threshold of 1755 μg.h/L for the RBV plasma AUC from 0 to 4 h post-dose to discriminate patients with subsequent SVR from those with non-response 17 . This is currently the only data that can be used to interpret early RBV exposure and determine the optimal dose to initiate treatment. The impact of the pharmacokinetic properties of RBV in HIV patients has not been fully investigated. As for HCV monoinfected patients, we know that a large inter-individual variability in RBV plasma concentrations exists in HIV-HCV coinfected patients and that RBV exposure is associated with both efficacy and anemia 8 . Moreover, the SVR rate is known to be lower in HIV-HCV coinfected than in HCV monoinfected patients. One possible explanation could be that, as for other drugs, HIV infected patients present a lower bioavailability of RBV 18 . The aim of this study was to determine early plasma exposure evaluated by the AUC_0-4h after a first dose of RBV in a population of HIV-HCV coinfected patients and to compare the results to those obtained in HCV monoinfected patients.

Eighty-six Caucasian patients with chronic hepatitis C entered the study. Among them, 62 were males (72%) and 23 (27%) were HIV-HCV coinfected. Most of them (85%) were infected with HCV genotype 1 or 4. Baseline characteristics are presented in Table 1.

ALT, alanine aminotransferase; AST, aspartate aminotransferase; MDRD, Modification of the Diet in Renal Disease. Bold style indicates statistical significance. Quantitative variables are expressed as median and range.

Among HIV-HCV coinfected patients, 20/23 (87%) had a negative HIV RNA viral load at baseline (<50 copies/mL) and 57% had a CD4 count above 500 cells/mm³. Most of them received two nucleoside reverse transcriptase inhibitors (NRTIs) (22 patients): tenofovir and emtricitabine for 20 of them (91%) and didanosine associated with stavudine or tenofovir for two. One patient received a cART regimen without NRTI. No patient received abacavir. Nucleoside treatment was associated with a protease inhibitor in 14 patients: seven patients received lopinavir, four darunavir, two atazanavir and one saquinavir. All these patients were also treated with ritonavir as a booster. Eight patients received a non-nucleoside reverse transcriptase inhibitor (NNRTI): four nevirapine, three efavirenz and one etravirine.

Finally, four patients received raltegravir, associated with a protease inhibitor regimen in three of them and with tenofovir and emtricitabine in one. One patient received maraviroc in association with a protease inhibitor, NNRTI and raltegravir.

Medians of AUC_0-4h and AUC_0-4h normalized to the first RBV dose expressed in mg/kg BW were 1469 μg.h/L and 182 (μg.h/L)/(mg/kg) respectively. According to the threshold of 1755 μg.h/L defined in HCV monoinfected patients, 14/23 (61%) coinfected patients were found to have an AUC_0-4h below this value. No significant difference was found for cART type and duration, HIV viral load and CD4 level between the 2 sub-groups of HIV-HCV coinfected patients defined by this threshold (Table 2). All three patients found to have a detectable HIV RNA, had an AUC_0-4h below 1755 μg.h/L. With regard to the immune status, 61% of patients had a CD4 level below the retained cutoff of 500/mm³. Interestingly, the AUC_0-4h was significantly lower for these patients than for those with CD4 level greater than 500/mm³ with values of 1270 μg.h/L and 1840 μg.h/L respectively (p = 0.047). For these 2 sub-groups of coinfected patients, normalized AUC were 144 and 195 (μg.h/L)/(mg/kg) but the difference did not reach statistical significance. Table 3 summarizes these results.

Quantitative variables are expressed as median and range.

Quantitative variables are expressed as median and range. Bold style indicates statistical significance.

For the reference group of HCV monoinfected patients, median AUC_0-4h and normalized AUC_0-4h were 2030 μg.h/L and 271 μg.h/L respectively. In comparison to HIV-HCV coinfected patients, these two parameters were significantly higher (p = 0.018 and 0.001, respectively). Sixty-eight percent of monoinfected patients had an AUC_0-4h above the threshold of 1755 μg.h/L vs 39% of HIV-HCV coinfected patients (p = 0.014). The median AUC_0-4h was slightly higher in naive HCV patients than in relapsers and non-responders (2434 μg.h/L, 2025 μg.h/L and 1873 μg.h/L, respectively, although this difference did not reach statistical significance (p = 0.165).

Different factors potentially associated with RBV pharmacokinetics variability were investigated. Except for body weight, BMI, and hemoglobin level, baseline characteristics did not differ between monoinfected and coinfected patients. Median body weight and BMI were lower in coinfected patients than in monoinfected ones (69 kg vs 75 kg , p = 0.032 and 22.1 vs 25.7, p = 0.003, respectively). Conversely, hemoglobin level was lower in monoinfected patients (median 14.5 g/dl) than in coinfected patients (15.2 g/dl; p = 0.036).

Multivariate logistic regression analysis indicated that presence of HIV coinfection (OR = 3.76, 95% CI [1.21-11.6]; p = 0.022) and male gender (OR = 4.59, 95% CI [1.15-18.3]; p = 0.031) were independently associated with RBV underexposure (Table 4).

Bold style indicates statistical significance.

Using the AUC_0-4h method, our results clearly showed a plasma underexposure in HIV-HCV coinfected patients after a single dose intake of RBV in comparison with monoinfected patients. The pharmacokinetics of RBV is less documented in HIV-HCV coinfected patients than in monoinfected ones but some studies showed that the achievement of RBV cutoffs was a predictive factor of SVR independent of HIV coinfection 20 21 . The heterogeneity of our cohort of HIV-HCV coinfected patients was representative of a treated HIV population in a hospital environment (57% with CD4 > 500 cells/mm³; 87% with undetectable HIV viral load under HAART; patients mainly under 2 NRTIs + 1 boosted protease inhibitor) 22 . The choice of AUC_0-4h was made in accordance with the study of Loustaud-Ratti et al. for HCV monoinfected patients 17 . Indeed, the authors showed that the abbreviated AUC_0-4h was highly correlated to the full AUC_0-12h because it described the main inter-individual variability of RBV pharmacokinetic parameters (i.e. absorption and distribution phases). Interestingly in our study, more than 60% of coinfected patients had an AUC_0-4h below the threshold of 1755 μg.h/L defined as associated with SVR in HCV monoinfected patients 17 . In comparison, only 32% of monoinfected patients of our study had an AUC_0-4h below this value. Demographic or biological parameters were collected in order to detect confounding factors that could affect differences in early RBV exposure between both populations. Some differences were found. The first of them was a body weight significantly lower for coinfected patients. As the initial RBV dose was the same (600 mg) for all patients, irrespectively of this parameter, early exposure was normalized for RBV dose expressed as mg/kg BW. Normalized AUC_0-4h remained significantly different between both groups. In addition, hemoglobin level was higher in coinfected than in monoinfected patients. It is well known that RBV is highly concentrated in red blood cells 23 , but this parameter was not expected to influence AUC_0-4h which was determined after a single dose of RBV. Among other parameters known to have an influence on RBV exposure is the renal function, but neither median baseline serum creatinine, nor estimated glomerular filtration rate (GFR) calculated by Cockroft-Gault equation or MDRD formula, differed significantly between both groups (Table 1). We have determined early RBV exposure regardless of HCV genotype as no difference was observed in the genotype distribution in each of the two populations of this study and because we wanted to evaluate early bioavailability of RBV independently of the subsequent treatment response. In the study by Loustaud-Ratti et al. the population was exclusively constituted of monoinfected HCV genotype 1 patients. This is the main difference with our study in which genotypes 1 to 4 were present but with a majority of genotype 1. Indeed, except for this parameter and for the distribution of fibrosis score, baseline characteristics were similar in both studies. However, 32% of our HCV monoinfected patients had an AUC_0-4h below the 1755 μg.h/L threshold compared with 58% in this historical study. This difference could be explained by the presence of patients infected by HCV genotype 2, 3 or 4 because only two of these 15 patients had an AUC_0-4h below this threshold. Finally, we found a strong gender ratio in favor of male patients in our coinfected population. In order to estimate if gender could affect RBV exposure, we compared parameters which were statistically different in the whole population. For body weight the significant difference was even more important and as expected, this parameter was lower in coinfected patients (Table 1). Hence the interest of the AUC normalized to the first RBV dose deletes the weight effect. Conversely, differences in hemoglobin level were no longer significant due to the higher proportion of women in the monoinfected group. In spite of the reduced number of patients, the difference of AUC_0-4h or normalized AUC_0-4h between monoinfected and coinfected male patients was statistically significant, and the proportion of AUC under the 1755 μg.h/L threshold remained high. No other differences in demographic or biological characteristics were found between the two populations of this study.

Given that there was a strong gender ratio in favor of male patients and in order to see if gender could affect our results, we compared AUC_0-4h between monoinfected and coinfected male patients. It is noteworthy that 43% of monoinfected and 65% of coinfected male patients had AUC_0-4h below the 1755 μg.h/L threshold, although this difference did not reach statistical significance (p = 0.103). As expected, among HCV monoinfected patients, the AUC_0-4h was slightly higher in naive patients than in relapsers and non-responders to a previous treatment although this difference did not reach statistical significance probably because of a limited number of patients in each subgroup.

Unexpectedly, RBV plasma underexposure seems to be associated with the immunological level of HIV patients. RBV exposure was indeed better if CD4 cell count was greater than 500/mm³. In the RIBAVIC study, using PEG-IFN/RBV therapy in HIV co-infected patients, a trend for a lower SVR was noted in patients with a CD4 cell count under 500 versus patients upper this threshold (21% vs 33%, respectively; p = 0.051) 24 . Then, in the PRESCO trial, a higher dosing of ribavirin (weight-based dosing with 1000 to 1200 mg) was associated with better SVR rates but higher risk of anemia 25 . In the present study, when the immunorestoration under cART was considered complete, i.e. with a CD4 cell count > 500/mm³, RBV exposure was similar in this HIV sub-group in comparison with HCV monoinfected population.

Food intake is known to potentially impact RBV bioavailability. Both the AUC_tf (from time zero to the time of the final quantifiable sample) and C_max of a single oral dose of ribavirin were indeed shown to be increased by co-administration of a high fat meal 26 . However, in the present study all patients received RBV with light breakfast or snack which suggests no bias between mono- and coinfected patients.

In conclusion, HIV-HCV coinfected patients present a lower RBV exposure which could at least in part explain the lower response rate of these patients to HCV treatment. This hypothesis should be further investigated on a larger number of coinfected patients focusing on the immune status of these patients. In the past, recommendations were to use low dose of RBV in HIV patients in order to avoid toxicity associated with antiretroviral drugs, such as zidovudine, stavudine or didanosine 27 28 . However, these drugs are progressively neglected to the benefit of less cytotoxic new antiretroviral drugs. So, higher doses of RBV can be used to improve SVR in coinfected patients, as suggested in several studies 7 29 . Measurement of early RBV exposure by determination of AUC_0-4h possibly followed by RBV dose adjustment could help in optimizing antiviral therapy in HIV-HCV coinfected patients, particularly in the more immunosuppressed ones because RBV underexposure was noted in this specific sub-group.

AUC: Area under the curve; BMI: Body mass index; cART: Effective combination antiretroviral therapy; GFR: Glomerular filtration rate; HAART: Highly active antiretroviral therapy; HCV: Hepatitis C virus; HIV: Human immudeficiency virus; MDRD: Modification of the diet in renal disease; RBV: Ribavirin; SVR: Sustained virological response.

Fabien Zoulim received consulting/speaker fees from Gilead Science, Bristol Myers Squibb, and Roche. François Bailly received speaker fees from Janssen. The other authors have no conflict of interests to declare. No financial support.

FB, PP, MM, FP, FZ, MCG participated in the conception and design of the study. GH, FB, EP, PP, PM, MM, FP, PC, JML participated in data collection. EP and PP performed the statistical analysis. GH, FB, EP, PP, PM and MCG helped to draft the manuscript. All authors read and approved the final version of the manuscript.