| |
Don't Stay on Virologically Failing Regimens
|
| |
| |
Evolution of drug resistance in HIV-infected patients remaining on a virologically failing combination antiretroviral therapy regimen
AIDS: Volume 21(6) 30 March 2007 p 721-732
Cozzi-Lepri, Alessandroa; Phillips, Andrew Na; Ruiz, Lidiab; Clotet, Bonaventurab; Loveday, Clivec; Kjaer, Jesperd; Mens, Helenee; Clumeck, Nathanf; Viksna, Ludmilag; Antunes, Franciscoh; Machala, Ladislavi; Lundgren, Jens Dd; for the EuroSIDA Study Group
From the aRoyal Free and University College Medical School, London, UK
bIrsiCaixa Foundation & Lluita contra la SIDA Foundation, Badalona, Spain
cInternational Clinical Virology Center (ICVC), Buckinghamshire, UK
dCopenhagen HIV Programme, Hvidovre Hospital, Denmark
eDepartment of Infectious Diseases, Copenhagen, University Hospital Hvidovre, Denmark
fDepartment of Infectious Diseases, Chu St. Pierre, Brussels, Belgium
gRiga Stradins University, Riga, Latvia
hHospital de Santa Maria, Lisboa, Portugal
iAIDS Centra FN Bulovka, Infek_ni kliniky 2, Praze, Czech Republic.
"....The main perceived risk associated with keeping viraemic patients on cART is the accumulation of HIV drug resistance mutations (DRM) [11-16] and hence the reduction in future drug options (FDO).....In patients kept on the same virologically failing cART regimen for a median of 6 months, there was considerable accumulation of drug resistance mutations, particularly in patients with initial low level of resistance to the failing regimen.....In our study population of patients who were kept on the same virologically failing cART regimen (> 400 copies/ml) for a median of 6 months, there was considerable accumulation of DRM (drug resistance mutations).....
....Our analysis is partially consistent with the results of other previous studies: 77% of patients acquired one or more IAS mutations over an average follow-up of 6 months. The overall average crude number of acquired IAS mutations was two per 6 months...
....In conclusion, our analysis, because of the marked accumulation of DRM observed in people kept on a stable virologically failing regimen, supports the current view that therapy should be immediately modified if a regimen containing at least two active drugs can be constructed....."
the authors also said this but today there are so many new good drugs that I find it hard to find a scenario where a switch is not viable: "On the other hand, the management of patients, for whom a sound switch is not viable, remains a challenge. A possible alternative strategy to the maintenance of a failing regimen could be to recycle drugs to which patients' virus is already resistant, in order to reduce the risk of accumulating additional mutations and maximize the benefits of a reduced viral replication capacity [4,5,16]. It has also been suggested that, in such patients with no options, only the RT gene should be targeted in order to prevent the accumulation of further mutations in the protease gene [37]. However, these strategies need to be properly evaluated in randomized studies before they could be recommended in the clinical setting."
Abstract
Objective: To estimate the extent of drug resistance accumulation in patients kept on a virologically failing regimen and its determinants in the clinical setting.
Design: The study focused on 110 patients of EuroSIDA on an unchanged regimen who had two genotypic tests performed at two time points (t 0 and t 1) when viral load was > 400 copies/ml.
Methods: Accumulation of resistance between t 0 and t 1 was measured using genotypic susceptibility scores (GSS) obtained by counting the total number of active drugs (according to the Rega system v6.4.1) among all licensed antiretrovirals as of 1 January 2006. Patients were grouped according to the number of active drugs in the failing regimen at t 0 (GSS_f-t 0).
Results: At t 0, patients had been on the failing combination antiretroviral therapy (cART) for a median of 11 months (range, 6-50 months). Even patients with extensive resistance to the failing regimen were still receiving benefit from treatment. An overall 6-monthly increase of 1.96 (SD, 2.23) International Aids Society-mutations and an average loss of 1.25 (SD, 1.81) active drugs were estimated. In comparison with patients with GSS_f-t 0 = 0, the number of active drugs lost was -1.08 [95% confidence interval (CI), -2.13 to -0.03; P = 0.04] in those with GSS_f-t 0 of 0.5-1.5 and -1.24 (95% CI, -2.44 to -0.04; P = 0.04) in those with GSS_f-t 0 ≥ 2.
Conclusions: In patients kept on the same virologically failing cART regimen for a median of 6 months, there was considerable accumulation of drug resistance mutations, particularly in patients with initial low level of resistance to the failing regimen. Randomized comparisons of maintenance treatment strategies while awaiting a new suppressive therapy to become available are warranted.
Introduction
HIV treatment guidelines recommend that, after confirmed virological failure, a new regimen containing at least two - expected to be - virologically active and tolerated drugs is started with the aim of regaining complete control of viral replication [1-3]. However, in everyday clinical practice, it may happen that patients with limited therapeutic options and a low level viral rebound (for example with a viral load < 10 000 copies/ml) are kept on a virologically failing regimen probably due to short-term virological and immunological benefit or other personal decisions [4,5]. The reasons for the continued CD4 cell count benefit of combination antiretroviral therapy (cART) in viraemic patients are unclear but both host and viral factors as well as residual viral suppression below patients' natural set-point may play a role [4,6-10].
The main perceived risk associated with keeping viraemic patients on cART is the accumulation of HIV drug resistance mutations (DRM) [11-16] and hence the reduction in future drug options (FDO). The quantification of this potential loss of options as an outcome has already been suggested and used in previous analyses [13,14]. However, the determinants of the accumulation of DRM in this setting remain poorly evaluated and, for example, there is little or no conflicting evidence about the role of the extent of resistance at the beginning of the study period, viral load, CD4 cell count and specific antiretrovirals [11-13].
Methods
A total of 110 patients of the EuroSIDA study fulfilled the following criteria: (a) had two genotypic tests performed at time points (t 0 and t 1) over a period when viral load was > 400 copies/ml and a combination therapy containing at least three antiretrovirals was unchanged (failing regimen); (b) the failing regimen had been maintained for at least 6 months prior to t 0. The test results were obtained from retrospective sequencing of plasma samples stored in the central EuroSIDA repository and identified for this purpose. If, over the course of treatment, individual patients had several episodes fulfilling the above set of criteria only the most recent was included in this analysis. Sequence analysis of HIV-1 protease and reverse transcriptase (RT) reading frames was performed in two laboratories in the UK and Spain using standardized equipment, the Trugene HIV-1 Genotyping Kit and OpenGene DNA Sequencing System (version 8.0, Visible Genetics, Barcelona, Spain), according to the manufacturer's recommendations.
The DRM considered were those listed in the International Aids Society (IAS) document of October 2005 [17]. DRM present at t 0 were assumed to be still present at t 1 as it has been shown that once a mutation has been accumulated it persists in minority viruses even if it is not detected by population sequencing [18]. A mutation was then defined as acquired between t 0 and t 1 if it was detected at t 1 but never at t 0 or before; insertions and mixtures of a mutated and wild-type HXB2 strain were counted as mutations.
Statistical analysis
The main outcome of this analysis was the genotypic susceptibility score (GSS) of a virtual regimen containing all licensed drugs available as of 1 January 2006 (GSS_t); the GSS_t at t 0 was calculated as follows: for each drug licensed by the European Agency for the Evaluation of Medical Products (EMEA) as of 1 January 2006 [19] a score of 0 was given if an isolate was classified as resistant to that drug by the Rega Institute interpretation system (version 6.4.1) [20], 0.5 if it was intermediate and 1 if it was susceptible. The GSS_t for all possible options at t 0 was calculated as the sum of the scores for each individual drug and the same was repeated at t 1. Analyses were repeated after interpreting the genotypic results with the system developed by the Agence Nacionale for Recherce sur le SIDA (ANRS, July 2005 version) [21]. Other secondary endpoints were evaluated (e.g. total number of amino-acid changes from consensus sequence, viral load and CD4 cell count). The number of amino-acid differences from HXB2 was calculated by simple count of the different positions (1-230 for reverse transcriptase and 1-99 for protease) and cross-validated by phylogenetic analysis. Phylogenetic neighbour-joining distance trees were constructed in MEGA 3.1 using Kimura 2-parameter distances, pair-wise deletion of gaps and considering all three codon positions [22].
For all endpoints the mean and standard deviations (SD) of the absolute values at t 0 and t 1 and of the changes over t 0-t 1 were calculated. These changes were divided by the duration of the follow-up over t 0-t 1 in order to obtain a standardized change expressed per 6 months. A standard linear regression analysis of these changes was used for the multivariable analysis. This approach relies on the assumption that, for example, drug options decrease in a linear fashion with the accumulation of resistance. In order to verify whether the same results could be obtained after relaxing this assumption, in a sensitivity analysis we modelled the absolute change in total number of drug options and we considered the exact months between the two tests as an additional covariate.
One crucial factor, that was a priori hypothesized to be important, was the extent of resistance to the failing regimen already present at t 0. This was calculated counting only the scores for the drugs contained in the failing regimen (GSS_f). We defined patients with a GSS_f = 0 at t 0 as patients with 'extensive resistance' (to the failing regimen), those with a GSS_f of 0.5-1.5 as patients with 'intermediate resistance' and those with a GSS_f ≥ 2 as patients with 'low resistance'. A number of other potential confounders were included in the multivariable linear regression analysis (see Table 3 of Results for a detailed list).
In a subset of patients, a measure of viral load before initiation of any antiretroviral therapy was available and it was used as an estimate of patients' viral set-point. The regression analysis in this subset of patients was repeated after adjusting for this additional parameter.
Results
Study population
The main characteristics of the study population at t 0 are shown in Table 1. Patients had been on ART for a median of 45 months (range, 6-135 months) and on the virologically failing regimen for 11 months (range, 6-50 months), 71% had previously experienced virological failure to two or more drug classes (not counting the drugs of the current failing regimen). Despite the fact that patients were not heavily pre-treated, on average, at t 0 only 1.1 drugs in the failing regimen were still active against their dominant virus populations according to the Rega interpretation system. At t 0, 20% of patients (n = 22) had extensive resistance to the failing regimen (GSS_f = 0), 57% (n = 63) had intermediate resistance (GSS_f of 0.5-1.5) and 23% (n = 25) had low resistance (GSS_f ≥ 2). Of the 25 patients with GSS_f ≥ 2, only six (24%) had a GSS_f ≥ 3, suggesting that a minority of patients were receiving a regimen at t 0 that was predicted to be able to fully suppress viral load. The median time between the two genotypic tests was 6 months (range, 2-28 months).
Failing regimen
The nucleosides most frequently used in the failing regimen were lamivudine (84%), stavudine (65%) and zidovudine (35%). The failing regimen contained efavirenz in 7% of patients and nevirapine in 12% of patients. The most frequently used protease inhibitors (PIs) were indinavir (single 25%, ritonavir-boosted 4%), nelfinavir (single 21%), ritonavir (20%) and hard-gel formulation saquinavir (single 8%, ritonavir-boosted 8%).
In the majority of patients, the failing regimen contained two nucleosides and a single PI (43%) or other combinations containing all three classes (39%). The remaining regimens were two nucleosides and a non-nucleoside reverse transcriptase inhibitor (NNRTI) (6%), ritonavir-boosted PI combinations (10%) and nucleoside-based combinations (2%, Table 1).
HIV drug resistance mutations
For 75 patients (68%) only one genotype was used to define the prevalence of resistance by t 0, for 23 (21%) two genotypes and for the remaining 12 (11%) ≥ 3 genotypes. The prevalence of DRM at t 0 was relatively high, especially for nucleoside mutations and PI mutations: the percentage of patients' viruses with thymidine analogues mutations (TAMs) was 75% (n = 82), 76% (n = 84) had 184V, 58% (n = 64) had ≥ 1 major PI mutations and 89% (n = 98) had ≥ 1 minor PI mutations. A smaller proportion of patients (19%) had resistance to NNRTIs (n = 21 with ≥ 1 NNRTI-associated mutation).
The overall prevalences of major mutations/mutational patterns at t 1, respectively, with and without the assumption about persistence of resistance were the following: TAMs (80 versus 76%), 184V (83 versus 74%), NNRTI-resistance (29 versus 26%) and major PI resistance (72 versus 67%). Considering single IAS mutations, both in the RT and protease region, the percentage of patients who acquired a new mutation (i.e. that was not already present at t 0) was similar irrespective of the mutation and it was estimated to be around the 5% mark (41L:5%, 67N:5%, 70R:4%, 184V:7%, etc.; Fig. 1). Overall, 77% of patients (85 out of 110) acquired ≥ 1 IAS mutation over the study period; 27 acquired ≥ 1 TAM, (25%), seven acquired 184V (6%), 13 ≥ 1 NNRTI mutation (12%), 40 ≥ 1 major PI mutation (36%) and 51 ≥ 1 minor PI mutation (46%; Fig. 1). This corresponded to an overall 6-monthly increases of 1.96 (SD, 2.23) IAS mutations and 5.72 (SD, 7.53) total changes from HBX2. By interpreting this increase in the number of mutations/changes using the Rega interpretation system, it translated into an overall average loss of 1.25 (SD, 1.81) active antiretroviral drugs per 6 months.
CD4 cell count and viral load
Mean CD4 cell count remained stable around 280 cells/μl over t 0-t 1 (crude estimate of 6-monthly change: -9.79 cells/μl, SD, 113.5; P-value against mean = 0; P = 0.39) in spite of a small increase in viral load (+0.14 log10 copies/ml per 6 months; SD, 0.91; P = 0.14). The mean viral load was 3.87 log10 copies/ml (SD, 1.02) at t 0 and 3.98 log10 copies/ml (SD, 0.89) at t 1.
Predictors of HIV evolution
Number of IAS drug resistance mutations
Pylogenetic trees covering all population sequences showed that pairs of sequences from the same patient formed a tight cluster distinct from that of other patients (data not shown). A crude analysis showed that the accumulation of IAS-DRM was inversely related to the prevalence of resistance to the failing regimen present at t 0, patients with lower prevalence showing greater accumulation (Table 2). In the adjusted analysis, however, no significant differences according to the extent of resistance to the failing regimen were observed (data not shown). The only significant predictors of the change in number of IAS mutations per 6 months were an inverse association with CD4 cell count nadir (-0.41 [95% confidence interval (CI), -0.79 to -0.03] mutations per 100 cells/μl higher; P = 0.05). There was some evidence for an association between the change in viral load between t 0 and t 1 and the rate of accumulation of IAS-DRM [+0.35 (95% CI, -0.12 to 0.83) mutations per log10 larger increase in viral load; P = 0.14].
The crude trend for larger accumulation of mutations in patients with less resistance to the failing regimen seemed to be equally driven by nucleosides and major PI mutations (Table 2). However, again, the extent of resistance to the failing regimen was not independently associated with the change in IAS-nucleosides mutations or TAMs and only marginally with the change in IAS-major PI mutations [in comparison with people with extensive resistance: +0.52 (95% CI, -0.15 to 1.20) PI mutations; P = 0.13 for those with intermediate resistance and +0.71 (95% CI, 0.15 to 1.27); P = 0.01 for those with low resistance]. The presence of lamivudine in the failing regimen was a predictor of slower accumulation of nucleosides mutations and TAMs but not of PI mutations (data not shown).
In a subset of 24 patients (21.8%), residual viral suppression on therapy was associated, although not significantly, with a reduced risk of accumulation of IAS mutations [-1.32 (95% CI, -3.77 to 1.14) per log10 suppression from set-point; P = 0.23].
Future drug options
The loss of drug options seemed to be mainly explained by the loss of susceptibility to the drugs included in the failing regimen (Table 2). This was confirmed by the multivariable analysis: the adjusted differences in average number of FDO in comparison with patients with extensive resistance were -1.08 (95% CI, -2.13 to -0.03; P = 0.04) in patients with intermediate resistance and -1.24 (95% CI, -2.44 to -0.04; P = 0.04) in patients with low resistance (Table 3).
The results were similar when the absolute loss of FDO was modelled, therefore relaxing the assumption of a linear decrease in drug options over the time period: -0.57 drugs in people with GSS_f(t 0) = 0 (comparator), -1.34 in those with GSS_f-(t 0) of 0.5-1.5 (adjusted difference: -1.08; 95% CI, -2.04 to -0.12; P = 0.03) and -1.96 in those with GSS_f(t 0) ≥ 2 (adjusted difference: -1.42; 95% CI, -2.53 to -0.31; P = 0.01). In this analysis, the absolute change in total number of drug options was directly related to the length of the observation period (in comparison with 0-3 months, -0.55 for 3-6 months; 95% CI, -1.62 to 0.53; P = 0.31 and -1.26 per > 6 months; 95% CI, -2.34 to -0.20; P = 0.02).
The results were also similar when the ANRS interpretation system was used in place of the Rega Institute system (data not shown).
Patients with higher CD4 cell count nadir experienced a less severe loss of drug options (0.34 drugs less per 100 cells/μl; 95% CI, 0.03 to 0.65; P = 0.03).
Total number of amino acid changes from HXB2
Contrary to our working hypothesis, no specific trend was observed between the accumulation of any amino acid changes from HXB2 and the extent of resistance to the failing regimen at t 0 (Table 2). From the multivariable analysis, the adjusted differences were: 3.24 (95% CI, -1.16 to 7.65, intermediate versus extensive resistance; P = 0.15) and 0.91 (95% CI, -4.23 to 6.06, low versus extensive; P = 0.72). In this last model, the presence of lamivudine in the failing regimen was again associated with lower rate of HIV change (Table 3). In addition larger increases of viral load over t 0-t 1 was associated with greater HIV change from HXB2 [1.32 (95% CI, 0.13 to 2.51) mutations per log10 larger increase in viral load; P = 0.03]. In contrast, residual viral suppression below the natural set point was not associated with this endpoint [n = 24; -1.98 (95% CI, -13.64 to 9.69) per log10 suppression from set-point; P = 0.68].
Discussion
In our study population of patients who were kept on the same virologically failing cART regimen (> 400 copies/ml) for a median of 6 months, there was considerable accumulation of DRM.
Our analysis is partially consistent with the results of other previous studies: 77% of patients acquired one or more IAS mutations over an average follow-up of 6 months. The overall average crude number of acquired IAS mutations was two per 6 months. Over our study period, considering all antiretrovirals available as of 1 January 2006 and from interpreting the genotypic results with the most recent version of the Rega Institute system, this corresponded to an average of a loss of 1.25 antiretroviral drugs. Depending on the drugs included in the failing regimen and on the extent of resistance to this therapy the accumulation of mutations may have, however, in the long term, a limit. Thus, our results cannot be extended to periods of follow-up longer than our average of 6 months.
The estimate of the risk of accumulation of DRM at the end of follow-up, in previous studies of patients experiencing virological failure on cART, ranged from 30 to 75% over a follow-up of 6-14 months in different analyses [11-13,23-25]. Another recent study showed that the GSS for FDO dropped from an average of 7.6 active drugs to 6.9 drugs over a study period of 12 months [16].
Three previous studies have shown that the rate of accumulation of DRM was higher in patients with fewer mutations detected at baseline [11-13]. This trend was also observed in our study with respect to the estimated loss of FDO and when evaluating the rate of accumulation of major PI mutations. Overall, in comparison with patients with extensive resistance to the failing regimen, those who had two or more estimated active drugs in the failing regimen at t 0, showed by t 1 an average loss of 1.24 drugs per 6 months of follow-up. This loss seemed to be mainly explained by the loss in virological activity of other antiretrovirals not included in the current failing regimens (Table 2). A recent analysis has consistently shown a rapid evolution of protease genes under the pressure of a salvage PI-containing regimen [26]. Moreover, we observed that people with extensive resistance to a PI-containing failing regimen accumulated a significantly lower number of major PI mutations in comparison with those who had low resistance to this regimen, whereas the opposite trend was seen for minor PI mutations. This is consistent with the hypothesis that if the dominant virus is the most fit under the pressure of a certain regimen the accumulation of compensatory PI mutations (instead of major DRM) may be prioritized in order to restore the replication capacity [4,27].
Patients with extensive resistance to the failing regimen were also those who accumulated less total changes from HXB2 (a mean difference of 3.2 amino acid changes in comparison with people with intermediate resistance and 0.9 with those with low resistance). The consequences of accumulation of changes that are not currently identified by IAS as conferring resistance to antiretrovirals are unknown. However, it is conceivable that an absolute increase as large as that found herein (an average of 5.7 changes per 6 months) may eventually lead to resistance to other drugs currently in the pipeline.
Another possible explanation for the slow rate of HIV change observed in patients with extensive resistance to the failing regimen is that this treatment had no virological effect on the virus. However, this is unlikely, as we have shown that, on average at t 0, patients' viral load was suppressed 0.5 log10 copies/ml below their natural set-point and their CD4 cell count was > 50 cells/μl above their nadir levels.
There are a few differences between our analysis and those of previous similar studies that need emphasis. First, our analysis was strictly conducted under the assumption that once a mutation was detected by population sequencing then it would persist in patients' viral populations from that point in time onwards. For more than 30% of our study population more than one genotype was used to establish the prevalence of resistance at t 0; furthermore, without assuming indefinite persistence of mutations over time, the estimated rate of accumulation of DRM would have been lower than that reported as a result of the lower estimated prevalence at t 1. Second, our population included a smaller proportion of patients with high baseline prevalence of resistance and prior extensive use of antiretrovirals in comparison with at least two other studies [12,16].
Napravnik and colleagues also showed that the rate of accumulation of resistance was greater in people who had a steeper increase in viral load over follow-up [12]. In our analysis, the slope of viral load between t 0 and t 1 was associated with an increase in number of total changes from HXB2. However, since such a viral load change occurred concomitantly with the increase in number of amino acid changes, the interpretation of this correlation analysis, in terms of determining which was the cause of which, is problematic. Interestingly, patients whose viral load was more suppressed below natural set point tended to accumulate less resistance. However, because of the small number of patients included, our analysis is unfortunately inconclusive and further studies are needed to test the biological hypothesis that the accumulation of resistance is a direct function of the residual level of HIV replication under the pressure of treatment [28]. A documented date of HIV seroconversion and repeated viral load measurements before treatment initiation may be needed for an accurate estimate of patients' virological set point. Conversely, Kantor and colleagues, showed an association between the accumulation of resistance and the duration of time spent on the virologically failing regimen [11].
In a sensitivity analysis in which we modelled the absolute change in total number of drug options we found that the rate of loss was directly related to the length of the observation period.
Another interesting finding of our analysis was the fact that if lamivudine was contained in the failing regimen then patients tended to accumulate a lower number of total amino acid changes from a consensus sequence. Other studies have suggested that the maintenance of lamivudine in a failing regimen may provide particular benefit by increasing the fidelity of RT via the persistence of the 184V mutation [29,30]. However, in our analysis, we found little evidence for an association between the detection of 184V and the outcome and this is consistent with the results of a recent trial [31]. In addition, our analysis suggests that, if such effect exists, it seems to be explained by factors other than the presence of 184V.
Moreover, ours is the first analysis to suggest a possible role of patients' history of immuno-suppression, in that a higher CD4 cell count nadir would confer protection against the accumulation of resistance/loss of drug options. Of course, we cannot rule out that this association is simply confounded by patients' adherence that was not measured in our study.
As also shown by other studies of patients remaining on non-fully suppressive regimens [11,12,32], in our analysis viral load increased at a small rate (0.14 log10 copies/ml per 6 months) and CD4 cell count remained stable. The possible explanatory factors of the variability in CD4 trajectories in patients retained on non-fully suppressive regimens will be studied in separate future analyses of EuroSIDA patients.
Before drawing firm conclusions a few limitations of our analysis need to be mentioned. First of all, the fact that the losses of FDO were calculated by using rule-based interpretation systems that are known to be imperfect [33]. However, our results were remarkably similar whether the Rega or ANRS system were used.
Furthermore, our estimate of the rate of accumulation of amino-acid changes is potentially either an under or an over-estimate of the true rate as there could be additional changes that were not detected by population sequencing assays both at t 0 and t 1 [26]. A phylogenetic analysis supported the absence of potential contamination during PCR amplification and suggested that these population averages did contain useful information about the underlying sequences. In order to study this issue in more detail, the measurement of resistance in minor populations would be needed. In addition, a measure of patients' viral set point that would help to control for a crucial confounding factor (i.e. the amount of residual virological suppression at t 0) was available only in a minority of patients (21.8%). Finally, the estimate of the loss of FDO was based on standard full sequence of RT and protease genes only. It has recently been shown that mutations detected in other genes (i.e. C-terminus of RT and RnaseH) could affect phenotypic resistance of a number of antiretrovirals [34-36].
In conclusion, our analysis, because of the marked accumulation of DRM (and other changes from HXB2) observed in people kept on a stable virologically failing regimen, supports the current view that therapy should be immediately modified if a regimen containing at least two active drugs can be constructed. On the other hand, the management of patients, for whom a sound switch is not viable, remains a challenge. A possible alternative strategy to the maintenance of a failing regimen could be to recycle drugs to which patients' virus is already resistant, in order to reduce the risk of accumulating additional mutations and maximize the benefits of a reduced viral replication capacity [4,5,16]. It has also been suggested that, in such patients with no options, only the RT gene should be targeted in order to prevent the accumulation of further mutations in the protease gene [37]. However, these strategies need to be properly evaluated in randomized studies before they could be recommended in the clinical setting.
|
|
| |
| |
|
|
|
