Original article Duration of first-line antiretroviral therapy in HIVinfected treatment-naive patients in routine practice

Transcription

1 Antiviral Therapy 2016; 21: (doi: /IMP3084) Original article Duration of first-line antiretroviral therapy in HIVinfected treatment-naive patients in routine practice Thomas Tesson 1, Mathieu Blot 1, Aurélie Fillion 2, Hama Djerad 3, Joséphine Cagnon-Chapalain 4, Aline Creuwels 5, Anne Waldner 1, Michel Duong 1, Marielle Buisson 1, Sophie Mahy 1, Pascal Chavanet 1,6, Lionel Piroth 1,6 * 1 Département d infectiologie, CHU de Dijon, Dijon, France 2 Service des Maladies Infectieuses, CH de Chalon sur Saône, Chalon sur Saône, France 3 Service de Médecine Interne, CH de Nevers, Nevers, France 4 Service des Maladies Infectieuses, CH de Macon, Macon, France 5 Service de Médecine Interne, CH de Sens, Sens, France 6 MERS UMR1347, University of Burgundy, Dijon, France *Corresponding author lionel.piroth@chu-dijon.fr Background: First-line antiretroviral therapy (1st ART) is an important step in a patient s management and often considered a long-term therapy at treatment initiation. Methods: To describe the duration of 1st ART and the factors associated with treatment modification in a recent real-life setting, antiretroviral-naive patients who began their 1st ART in six French hospitals in were included in a cohort. Clinical, immunological, virological and therapeutic data, as well as the reasons for therapeutic changes, if any, were retrospectively collected. Results: A total of 206 patients started 1st ART, mainly a protease inhibitor-based triple therapy (73%), with a tenofovir-including backbone (87%). Of these, 89 (43%) had their 1st ART modified after a median of 16.5 months (IQR ). Having a CD4 + T-cell count <200 cells/mm 3, being pregnant, or 1st ART including zidovudine + lamivudine or lopinavir/r were significantly associated with a higher risk for treatment modification in multivariate analysis. In 47 patients (53%), 1st ART was modified for safety reasons, with no significant association with a given antiretroviral drug or class. No significant difference in virological, immunological and clinical outcomes was observed between the patients who had their 1st ART modified and those who did not. Conclusions: The proportion of modifications of the 1st ART during the first 2 years remains high. These modifications are frequently because of safety issues and the willingness to simplify treatment, and less often driven by virological failure, thus emphasizing that 1st ART is not or is no longer a lifelong treatment. Introduction Since the late 1990s, the use of highly active antiretroviral therapy (HAART) has reduced the mortality and the morbidity related to HIV infection [1,2]. Its efficacy is based on its ability to control HIV viral load and to foster reconstitution of the immune system [3,4]. However, several factors may compromise its short- and long-term biological and clinical efficacy. In particular, the need to change HAART may be a surrogate marker of different situations and motivations that could have a negative impact on the clinical evolution of patients. Indeed, several studies that analysed HAART modifications reported a high proportion of patients (40 60%) with changes in their initial regimen within the first year of treatment, mainly because of virological failure and side effects [5,6]. With the availability of new antiretroviral drugs that are easier to take and virologically efficient, it could be speculated that first-line antiretroviral therapy (1st ART) could be less frequently modified than in the past, even though this has rarely been assessed in recent cohort studies. We thus conducted a study to determine the duration of 1st ART, the proportions and predictors of 1st ART modification, as well as the motivations and the potential medium-term consequences of such modifications International Medical Press (print) (online) 715

2 T Tesson et al. Methods Study population All antiretroviral-naive patients who began their first line therapy in six French hospitals (Auxerre, Chalon sur Saone, Dijon, Macon, Nevers, Sens) between January 2009 and December 2012 were included in a historical cohort. Data collection and definitions For all these patients, the clinical (including the CDC stage), immunological (CD4 + and CD8 + T-cell counts), virological (HIV viral load) and therapeutic (antiretroviral drugs) data were collected at the 1st ART initiation (M0), at 6 (M6) and 12 (M12) months after the initiation of 1st ART, at the time of ART modification (if any) and at the last visit (before 31 December 2013). When 1st ART was introduced in the first two visits (outpatients) or at the first hospital stay (inpatients), it was considered as introduced at the first visit. Modification of 1st ART was defined as the discontinuation of one or more antiretroviral drugs which were included in the 1st ART (whether replaced or not) or the addition of one or more new drugs. The 1st ART was considered unmodified when the number of pills was changed, but with strictly identical antiretroviral drugs. Dose adjustments were not considered treatment modifications either if the drug combination remained unchanged. When the treatment was modified, the reasons for the modification were collected from the medical record. The reasons for modification were classified as simplification, virological failure, safety reasons (because of side effects assessed either medically, clinically or biologically or simply reported by the patient), pregnancy or miscellaneous (including the participation in a clinical trial, for example). Simplification was defined as a change towards an easier-to-take treatment, with fewer daily intakes and/or fewer pills, or a change towards drugs with a better reported safety profile (such as the replacement of zidovudine by abacavir or tenofovir, for example). The persistence of an HIV viral load above 200 copies/ml after 6 months, or above 50 copies/ml (confirmed by a second consecutive viral load test) after 12 months of ART was considered as virological failure. These thresholds are close to but not exactly the same as those used in many recommendations and therapeutic trials. They were used to clearly characterize virological failure, so as to be sure that associated treatment modification was at least partly for that reason (independently of the reason[s] put forward by the physician in charge), since low HIV viral loads in real life do not systematically lead to ART modification. Statistical analysis Qualitative variables were compared by the Fisher s exact test or the McNemar test, when appropriate, and quantitative variables by the Mann Whitney test. Survival curves were compared by the log-rank test. To adjust differences between patients who discontinue 1st ART and those who did not, we used Cox s proportional hazard model. All the characteristics at the initiation of the treatment associated with the 1st ART duration with a P-value <0.20 in univariate analysis were first included in the model. A backward procedure was then used to eliminate variables that were not significant at the P=0.05 level. The evolution of the CD4 + T-cell count (at 6 months, 12 months and at the end of the follow-up) was compared between patients who had their 1st ART modified and those who did not by a factor of matched averages by one-way analysis of variance (with repeated measures), and the interaction was assessed using Fisher s test. For all analyses, a P-value below 0.05 was considered significant. Results Main characteristics of the patients included A total of 206 patients were included in the study with a median follow-up of 111 weeks (IQR ), 64% (152) males, with a median age of 41 years (IQR 32 49). Most of them (86.9%, n=179) contracted HIV via sexual intercourse and 63% (81) of the men were men who have sex with men. The main characteristics of the patients included at the time of their first positive HIV serology and at the time of the 1st ART initiation are summarized in Table 1. The median time between the first HIV-positive serology and the initiation of the 1st ART was 18.5 weeks (IQR 4 119). It was shorter in patients diagnosed with AIDS (2 weeks [IQR ] versus 21 weeks [IQR ] for the others). ART was initiated at the first visit in 46% (n=95) of the patients, whereas 72% (136) were asymptomatic and 20% (23) were diagnosed with AIDS at the time of 1st ART initiation. Fifteen women (28% of the women) were pregnant at this time. In keeping with the evolution of the French national recommendations, the proportion of patients starting treatment with a CD4 + T-cell count >350 cells/mm 3 was higher in than in (57% versus 39%, P<0.01). The distribution of the different regimens of 1st ART, as compared with the distribution in France in 2010, is presented in Table 2. Most of the 1st ART combinations were in accordance with the recommendations. A combination of 2 nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs) + 1 ritonavir-boosted International Medical Press

3 First-line HAART duration Table 1. Main clinical, immunological and virological characteristics of the 206 patients included at the time of their first positive HIV serology and at ART initiation At the first At the initiation positive serology of the treatment Median age, years (IQR) 40 (29 48) 41 (32 49) Gender, n male/female 152/54 Positive HCV serology, % (n) 4 (9) Positive HBs antigenaemia, % (n) 4 (8) HIV acquisition risk factor Men who have sex with men, % (n) 39 (81) Heterosexual intercourse, 49 (48/53) % (n male/female) Drug use, % (n male/female) 12 (23/1) CDC categories A, % B, % C, % CD4 + T-cell count >500 cells/mm 3, % <CD4 500 cells/mm 3, % <CD4 500 cells/mm 3, % <CD4 350 cells/mm 3, % cells/mm 3, % Median CD4 + T-cell count, 432 ( ) 340 ( ) cells/mm 3 (IQR) Median %CD4 (IQR) 23 ( ) 21.3 ( ) Median CD4/CD8 ratio (IQR) 0.47 ( ) 0.41 ( ) Median HIV viral load, 4.8 ( ) 4.8 ( ) log 10 copies/ml (IQR) HIV viral load >100,000 copies/ml, % ART, antiretroviral therapy; HBs, hepatitis B surface. protease inhibitor (PI/r) was more commonly used (71%) than a combination of 2 NRTIs + 1 non-nucleoside reverse transcriptase inhibitor (NNRTI) used in 23% of the patients. The NRTI backbone was very often tenofovir + emtricitabine (83%). In pregnant women, lopinavir/r was used in 87% and zidovudine + lamivudine in 53% of them. Immunovirological evolution As expected, the median HIV viral load significantly decreased between M0 and M6 (median 5.20 to 2.10 log 10 copies/ml; P<0.001). However, only 66% (105) of the patients had an HIV viral load <50 copies/ml at M6 and 80% (108) at M12. This proportion of patients with an HIV viral load <50 copies/ml at M6 was lower in those with a baseline HIV viral load 5 log 10 copies/ml (43% versus 80% in those with baseline HIV viral load <5 log 10 copies/ml; P=0.05). The median increase in the CD4 cell count was +144/mm 3 (IQR ) between M0 and M6 (P<0.001), +58/mm 3 (IQR ) between M6 and M12 (P=0.04), and +85 cells/mm 3 (IQR 0 220) between M12 and the last visit (P=0.009). Between M0 and M6, the decrease in HIV viral load was different in patients on protease inhibitors (PI), lower on lopinavir/r (1.96 log 10 copies/ml versus 3.30 log 10 copies/ml for atazanavir/r, 3.40 log 10 copies/ml for darunavir/r and 3.45 log 10 copies/ml for efavirenz; P<0.05), whereas the increase in CD4 + T-cells was higher in patients on PI/r-based 1st ART (192/mm 3 versus 149/ mm 3 in patients on NNRTI-based 1st ART; P<0.05). No other significant correlation was observed. Therapeutic evolution Of the 206 patients included, 89 (43%) had their 1st ART modified during the study period. 52 of the 150 (35%) PI/r-based 1st ART were modified, more frequently with lopinavir/r (14/28, 50%) and atazanavir (29/73, 40%) than with darunavir/r (9/49, 18%). NNRTI-based 1st ART was modified in 16 out of 47 patients (34%). NRTI backbones were modified in 30/203 (15%), more often with zidovudine (11/18, 61%) than with tenofovir (17/171, 10%) or abacavir (3/14, 21%). The median treatment duration before modification was 66 weeks (IQR ; Figure 1A). The estimated time at which 50% of the patients would have their 1st ART modified was 152 weeks (nearly 3 years). The likelihood of modification was nearly twice as high in patients with a CD4 + T-cell count <200/mm 3 (P=0.01; Figure 1B). There was also a trend towards a twofold higher proportion of 1st ART modification in pregnant women (P=0.08; Figure 1C). + In univariate analysis, having low baseline CD4 T-cell count below 200/mm 3, being pregnant and being on zidovudine + lamivudine and/or on lopinavir/r were or tended to be significantly associated with a higher likelihood of 1st ART modification (Table 3). In multivariate analysis, only a baseline CD4 + T-cell count below 200/mm 3 (P=0.01) and the presence of zidovudine and lamivudine (P=0.02) remained associated with a higher likelihood of 1st ART modification. The main reasons for 1st ART modification were the occurrence of side effects (52%, median time 26 weeks [IQR ]), simplification (18%, median time 40 weeks [IQR ]), and virological failure (17%, median time 42 weeks [IQR ]). The most frequent side effects leading to 1st ART discontinuation were neuropsychiatric discomfort (21%, mainly related to efavirenz), digestive disorders (17%, mainly related to PIs/r), renal function alteration (17%, mainly related to tenofovir) and icterus (15%, mainly related to atazanavir). Dyslipidaemia and/or hyperglycaemia motivated the 1st ART modification in only one case. A move towards fixed-dose combinations was the main reason for 1st ART simplification (56%), which was sometimes associated with the anticipation of potential safety issues not yet observed. This was the Antiviral Therapy

4 T Tesson et al. Table 2. Distribution of the first antiretroviral treatment initiated in the included patients, as compared with the national distribution at the time of the study [7] Antiretroviral therapy Number of patients % National distribution in 2010, % 2 NRTIs PI/r Tenofovir + emtricitabine Atazanavir/r Darunavir/r Lopinavir/r Abacavir + lamivudine Atazanavir/r Darunavir/r Lopinavir/r Zidovudine + lamivudine Atazanavir/r Darunavir/r Lopinavir/r NRTIs NNRTI Tenofovir + emtricitabine Efavirenz Etravirine Abacavir + lamivudine Efavirenz Zidovudine + lamivudine Efavirenz NRTIs II Tenofovir + emtricitabine Raltegravir NRTIs PI (unboosted) Abacavir + lamivudine Atazanavir (unboosted) NRTIs Zidovudine + lamivudine Tenofovir Double therapy PI/r II II, integrase inhibitors; NNRTI, non-nucleoside reverse transcriptase inhibitor; NRTI, nucleoside/nucleotide reverse transcriptase inhibitor; PI/r, protease inhibitor boosted with ritonavir. other main reason for 1st ART simplification. It was particularly frequent in patients on lopinavir/r (14%) and/or on zidovudine (22%). No significant difference was observed between PI/rbased and NNRTI-based 1st ART, in particular regarding modifications for virological failure (5% versus 4%, respectively), or regarding their 1st ART duration. However, the median duration of lopinavir/r-based 1st ART tended to be significantly shorter (52 weeks [IQR ] versus 71 weeks [IQR ] for atazanavir/r, 69 weeks [IQR ] for darunavir/r and 79 weeks [IQR ] for efavirenz, P=0.06; Figure 1D), as did that for 1st ART containing zidovudine and lamivudine (32 weeks [IQR ] versus 69 weeks [IQR ] for tenofovir + emtricitabine and 75 weeks [IQR ] for abacavir + lamivudine, P<0.001; Figure 1E). Virological, immunological and clinical evolutions None of the patients included stopped their 1st ART without taking a new antiretroviral therapy thereafter. At the last visit, 82% (n=164) of the patients had an HIV viral load <50 copies/ml. The CD4 + T-cell counts had significantly evolved at month 6, month 12 and at the end of the follow-up whether their 1st ART had been modified or not (P<0.05), but the evolution of the CD4 + T-cell count at month 6, month 12 and at the end of the follow-up were not significantly different between patients who had their 1st ART modified and those who did not (P>0.999). A majority of patients (71%, 140/197,) had a last CD4 + T-cell count above 500/mm 3, and 7% (13) still had a last CD4 + T-cell count below 200/mm 3. However, patients with a CD4 + T-cell count <200/mm 3 at 1st ART initiation were more likely to reach a CD4 + T-cell count >500/mm 3 when their treatment had been modified, for whatever reason (16% versus 5%; P=0.046). No change in CDC stage status was observed in any of the patients during the follow-up period. Discussion This study confirmed first that the proportion of 1st ART modification remains high (more than 40%), with a relatively short half-life (nearly 16 months), despite advances in the efficacy, simplicity and safety of recent antiretroviral drugs. These results are drawn from a study population which was representative of the population of patients living with HIV and followed at the same time in France, both for their baseline characteristics and for the therapeutic options used [7]. CD4 + T-cell count thresholds retained to introduce HAART International Medical Press

5 First-line HAART duration Figure 1. Survival curves A Patients Total population Duration of treatment, weeks B Patients CD4 + T-cell count <200 cells/mm 3 CD4 + T-cell count >200 cells/mm Duration of treatment, weeks Survival curves for (A) first-line therapy (patients still on first-line antiretroviral therapy), (B) first-line therapy in patients with a CD4 + T-cell count <200 cells/mm 3 and in patients with a CD4 + T-cell count >200 cells/mm 3 (P<0.01), (C) first-line therapy in terms of pregnancy (P=0.08), (D) first-line therapy containing atazanavir/r, lopinavir/r, darunavir/r or efavirenz (P<0.001) and (E) first-line therapy containing tenofovir + emtricitabine, abacavir + lamivudine or zidovudine + lamivudine (P<0.001). Antiviral Therapy

6 T Tesson et al. Figure 1. Continued C Patients Non-pregnant women Pregnant women Duration of treatment, weeks D Patients Efavirenz Lopinavir/r Atazanavir/r Darunavir/r Duration of treatment, weeks International Medical Press

7 First-line HAART duration Figure 1. Continued E Patients Tenofovir + emtricitabine Abacavir + lamivudine Zidovudine + lamivudine Duration of treatment, weeks Table 3. Factors associated with 1st ART duration (univariate and multivariate analyses) Univariate analysis Multivariate analysis Variables Hazard ratios 95% CIs P-value Hazard ratios 95% CIs P-value C CDC clinical stage , CD4 + T-cell count below 200 cells/mm , , CD4, % , CD4/CD , HIV viral load, log 10 copies/ml , st ART initiation 2008 versus , Time between first HIV-positive serology , and 1st ART NNRTI versus PI/r , II , Efavirenz versus Lopinavir/r , Atazanavir/r , Darunavir/r , Tenofovir/emtricitabine versus Abacavir/lamivudine , Zidovudine/lamivudine , , Pregnancy , Positive HCV serology , Positive HBs antigenaemia , HBs, hepatitis B surface; II, integrase inhibitor; NNRTI, non-nucleoside reverse transcriptase inhibitor; PI/r, protease inhibitor boosted with ritonavir; 1st ART, first-line antiretroviral therapy. Antiviral Therapy

8 T Tesson et al. in the 2008 and 2010 French therapeutic recommendations (at the time of the study) were 350 and 500/mm 3, respectively, except for patients with significant comorbidities that are HIV-related or worsened by HIV in whom HAART had to be started earlier. Even though the 1st ART was introduced in accordance with these national recommendations in more than 94% of the cases, the frequent use of PI/r in 1st ART, which was higher than that observed in other countries, has to be noted (Switzerland [6], US [8], Canada [9], Europe and North America together [10]). This use of PI/r in 1st ART was even slightly more frequent in our cohort than in France overall, mainly because of a local trend to favour such combinations considering the uncertainty regarding adherence to 1st ART, and the relative security offered by drugs with a high genetic barrier, such as PIs. The median CD4 + T-cell count at the time of 1st ART introduction also increased throughout the study period, in line with the evolution of national recommendations, and was globally higher than in other parts of the world [6 9]. The proportion of 1st ART modification we observed (43%) is close to that observed in other studies (41.5% [6], 49% [8], 47.7% [9] and 51.1% [10]). It was also similar between PI- or NNRTI- based 1st ART in our study (around 34%), which contrasts with some previous studies, which showed a lower proportion of modification for NNRTI-based 1st ART [8,9]. It has to be noted that the more frequent use of PI/r based 1st ART in our cohort did not seem to affect the global proportion of 1st ART modification. Safety issues were involved in more than half of the cases of treatment modifications. This is relatively surprising, since this proportion is similar to that observed at the beginning of the highly active antiretroviral therapy era [5,8,9,11]. This stability was also observed in another study, which showed that the percentage of patients who switched, because of gastrointestinal symptoms in particular, was not significantly different between the , and periods (29%, 34% and 25%, respectively; P=0.27) [8]. This could be partly explained by the persistent use of older antiretroviral drugs at the time of the study, such as zidovudine and lopinavir/r, for example, which were often associated (by contrast with atazanavir or darunavir which were often prescribed together with tenofovir). The type of side effects leading to the modifications (digestive intolerance on PI/r [12,13], neuropsychiatric troubles on efavirenz [14] and renal abnormalities on tenofovir [15]) are well known. Modifications because of metabolic disorders (hyperglycaemia, dyslipidaemia) were surprisingly rather rare considering the frequent use of PIs, for example. The lower proportion of modifications with newer drugs we observed should nevertheless be interpreted with caution, since exposure to and the length of follow-up for such antiretroviral drugs could have been insufficient. The proportion of modifications because of virological failure was by contrast low (6%), as observed in the Swiss study [6]. It is thus now an infrequent reason for 1st ART modification, and is significantly less frequent than in older studies (from 20.9% [5] to 62% [9]). This proportion is in line with another study, which showed that the percentage of patients with a detectable viral load among those who switched their ART decreased over time (64%, 41% and 35% for the periods , and , respectively) [8]. It could be advocated that the less strict thresholds used in this study to define virological failure could have led us to underestimate this reason, but such thresholds help to ensure that virological failure was the main if not the only reason to modify 1st ART. Another point is that non-adherence was never used as the main reason for 1st ART modification, since according to our guidelines HAART must not be changed until educational measures have been implemented and adherence improved. Exceptions are when HAART is based on drugs with a low genetic barrier (such a NNRTIs or raltegravir), which were not so frequently used in our study, or when adherence troubles are shown or thought to be directly linked to one antiretroviral drug per se. It is nevertheless likely that non-adherence is an underlying cause in patients who switch their 1st ART because of virological failure. The high proportion of 1st ART modifications towards fixed-drug combinations we observed (56% of the modifications for simplification ) is also likely to decrease, considering the higher frequency of their use as 1st ART choices. Though not considered a treatment modification per se, moving towards a smaller number of pills was often associated with a modification of at least one antiretroviral drug. Sometimes, such changes were also motivated by the fear of potential future side effects. The potential benefits of such modifications could thus not be specifically assessed. Beyond the characteristics of the different antiretroviral drugs, some particular characteristics of the patients may also affect the likelihood of 1st ART modification. Pregnancy is one such situation leading to 1st ART modification, first because of the move towards an ART combination authorized during pregnancy in women who become pregnant while on 1st ART, and second since national recommendations at the time of the study stated that there was no need to continue 1st ART after delivery in women with a good baseline immunovirological status. Since these recommendations have since evolved towards recommendations for ART for all, it is likely that the proportion of 1st ART modification for pregnancy will decrease in the future, and even more so if the possibility of a pregnancy is anticipated in choosing the 1st ART combination. In International Medical Press

9 First-line HAART duration addition, zidovudine and lamivudine were mostly used in pregnant women, as recommended at the time of this study, often associated with lopinavir/r. The increasing possibility of using similar treatments in pregnant and non-pregnant women will probably contribute to reducing the percentage of 1st ART modifications. Patients with a low baseline CD4 + T-cell count are also more likely to see their 1st ART modified, with a twofold higher probability of modification in such cases. Such an association was also observed in a recent study, even though it was not confirmed in the adjusted model [9]. The higher risk of side effects [16], of interactions with other treatments, in particular for opportunistic infections [17], or the preferential use of PI/r-based 1st ART [8,9] rather than fixed-drug co-formulations in these patients may explain this finding. Last, the modification of 1st ART was not associated with any particular clinical, immunological or virological consequences, with global response rates to therapy consistent with those reported in the literature, with two-thirds of the patients with HIV viral load below 50 copies/ml 6 months after 1st ART introduction, and 80% at the end of the follow-up period, and with no difference in virological or immunological evolution between patients initiating a PI/r- or an NNRTI-based 1st ART [18,19]. However, comparing CD4 + T-cell count changes without considering when 1st ART modification took place is only descriptive and does not allow to directly assess the impact of such modification on immunological evolution. Direct comparison of CD4 + T-cell evolutions after 1st ART modification to a comparable time period in patients who did not modify is hard to define and was not possible here. The fact that treatment modification in patients initiating 1st ART with a CD4 + T-cell count <200/mm 3 was significantly associated with an increased likelihood of transition towards CD4 + T-cell counts >500/mm 3 probably reflects a longer time spent on ART with a good evolution, and thus an increased probability for 1st ART modification. In conclusion, even though our study suffers from several limitations, such as its retrospective design and a lack of statistical power for the most recent antiretroviral combinations, it shows that the percentage of modifications of 1st ART remains high in the current therapeutic era, even though the motivations for change have evolved over time. Safety issues remain a persistent concern, not only because of the decreasing acceptance of side effects at a time when the safety profile of new antiretroviral drugs seems better than ever, but also because the burden of side effects may be underestimated in clinical trials compared with reallife long-course exposure. Last, the fact that no significant negative impact on virological, immunological and clinical outcome was observed also supports the rationalized use of ART for all. 1st ART no longer has to be prescribed as and considered a unique and lifelong therapy, but rather the first step towards a more individualized and ideally easier therapeutic future. Acknowledgements The authors thank Nadia Buthiot and Sandrine Gohier (Infectious Diseases Department, CHU Dijon, France) for their help in collecting the data, Philip Bastable (Research Department, CHU Dijon, France) for reviewing this article, and all the patients who participated. Disclosure statement The authors declare no competing interests. References 1. Lesko CR, Cole SR, Hall HI, et al. The effect of antiretroviral therapy on all-cause mortality, generalized to persons diagnosed with HIV in the USA, Int J Epidemiol 2016; 45: May MT, Vehreschild JJ, Trickey A, et al. Mortality according to CD4 count at start of combination antiretroviral therapy among HIV positive patients followed for up to 15 years after start of treatment: collaborative cohort study. Clin Infect Dis 2016; 62: Jacobson LP, Li R, Phair J, et al. Evaluation of the effectiveness of highly active antiretroviral therapy in persons with human immunodeficiency virus using biomarker-based equivalence of disease progression. Am J Epidemiol 2002; 155: Casado JL, Perez-Elias MJ, Antela A, et al. Predictors of long-term response to protease inhibitor therapy in a cohort of HIV-infected patients. AIDS 1998; 12:F131 F Yuan Y, L Italien G, Mukherjee J, Iloeje UH. Determinants of discontinuation of initial highly active antiretroviral therapy regimens in a US HIV-infected patient cohort. HIV Med 2006; 7: Elzi L, Marzolini C, Furrer H, et al. Treatment modification in human immunodeficiency virus-infected individuals starting combination antiretroviral therapy between 2005 and Arch Intern Med 2010; 170: Mary-Krause M, Grabar S, Lievre L, et al. Cohort profile: French hospital database on HIV (FHDH-ANRS CO4). Int J Epidemiol 2014; 43: Slama L, Li X, Brown T, et al. Increases in duration of first highly active antiretroviral therapy over time ( ) and associated factors in the Multicenter AIDS Cohort Study. J Acquir Immune Defic Syndr 2014; 65: Gonzalez-Serna A, Chan K, Yip B, et al. Temporal trends in the discontinuation of first-line antiretroviral therapy. J Antimicrob Chemother 2014; 69: Abgrall S, Ingle SM, May MT, et al. Durability of first ART regimen and risk factors for modification, interruption or death in HIV-positive patients starting ART in Europe and North America AIDS 2013; 27: Piroth L, Grappin M, Waldner A, et al. [Discontinuing protease inhibitor treatment of HIV-1 patients for intolerance. Longitudinal study of 309 patients]. Presse Med 1999; 28: French. 12. Molina JM, Andrade-Villanueva J, Echevarria J, et al. Once-daily atazanavir/ritonavir compared with twice-daily lopinavir/ritonavir, each in combination with tenofovir and emtricitabine, for management of antiretroviral-naive HIV-1-infected patients: 96-week efficacy and safety results of the CASTLE study. J Acquir Immune Defic Syndr 2010; 53: Antiviral Therapy

10 T Tesson et al. 13. Mills AM, Nelson M, Jayaweera D, et al. Once-daily darunavir/ritonavir vs. lopinavir/ritonavir in treatmentnaive, HIV-1-infected patients: 96-week analysis. AIDS 2009; 23: Clifford DB, Evans S, Yang Y, et al. Long-term impact of efavirenz on neuropsychological performance and symptoms in HIV-infected individuals (ACTG 5097s). HIV Clin Trials 2009; 10: Casado JL, Banon S, Santiuste C, et al. Prevalence and significance of proximal renal tubular abnormalities in HIV-infected patients receiving tenofovir. AIDS 2016; 30: When To Start Consortium, Sterne JA, May M, et al. Timing of initiation of antiretroviral therapy in AIDS-free HIV-1- infected patients: a collaborative analysis of 18 HIV cohort studies. Lancet 2009; 373: Ryom L, Boesecke C, Gisler V, et al. Essentials from the 2015 European AIDS Clinical Society (EACS) guidelines for the treatment of adult HIV-positive persons. HIV Med 2016; 17: Riddler SA, Haubrich R, DiRienzo AG, et al. Class-sparing regimens for initial treatment of HIV-1 infection. N Engl J Med 2008; 358: Sax PE, Tierney C, Collier AC, et al. Abacavir-lamivudine versus tenofovir-emtricitabine for initial HIV-1 therapy. N Engl J Med 2009; 361: Accepted 21 August 2016; published online 7 September International Medical Press