点击显示 收起
第14届国际HIV药物耐药会议
Highlights of the XIV International HIV Drug Resistance Workshop
2005年7月7－11日
加拿大魁北克
July 7 - 11, 2005, Quebec City, Quebec, Canada
Antiretroviral Drug Resistance: Much More Common Than We Thought
William O'Brien, MD
The 14th International HIV Drug Resistance Workshop advanced several important concepts relevant to clinicians; abstracts from the 6 sessions are available at http://www.intmedpress.com/General/.
The first session led off with a series of talks examining the effects of reduction in mother-to-child-transmission of HIV by single-dose nevirapine (Viramune) treatment of newborns and pregnant women at delivery, using recently developed low-frequency drug resistance assays. Conventional assays are rather insensitive and require a mutation frequency of at least 20% to 30% for detection. Newer assays used for research applications that have a sensitivity of less than 1% were applied to samples collected during studies of mother-to-child-transmission in Africa.
The problem with selection of HIV drug resistance during treatment with single-dose nevirapine was brought clearly into view during the International AIDS Conference in Bangkok, Thailand, in July 2004,[1,2] where it was shown that detectable levels of serum nevirapine could be detected in nearly half of treated women up to 3 weeks after the single dose. This allowed selection of resistance, which was detected in about 40% of mothers and essentially all newborns. Mutations diminished in frequency over time but tended to persist beyond 6 months. Resistance resulting from single-dose nevirapine treatment in mothers could pose substantial barriers to successful treatment with typical nonnucleoside reverse transcriptase inhibitor (NNRTI)-based first-line therapies.
Using an HIV-1 subtype C real-time polymerase chain reaction (PCR) assay that detects 2 common nevirapine mutations, K103N and Y181C, with established detection limits of approximately 0.2% to 0.3%, K103N was successfully identified in all 16 post-nevirapine samples in which mutations were previously detected by standard assays. The K103N mutation was also identified in 16 of 40 (40%) post-nevirapine treatment specimens that had been negative for mutations by conventional sequencing at 6-36 weeks after exposure.[3] Other studies examining other populations with similar treatment also showed detection of K103N in about 50% of samples 6 weeks after single-dose nevirapine treatment. Overall, nevirapine resistance mutations can probably be detected in 75% to 80% of nevirapine-treated HIV-infected mothers and infants,[4,5] and enrichment of NNRTI resistance frequency probably occurs uniformly. In any case, reduction in transmission is only about 50%, and this rate of protection diminishes over 6 months when infants are breastfed. In areas where single-dose nevirapine treatment is most needed, clean water often is not available and breastfeeding is the only option for infant nutrition. These discouraging data have led to studies searching for better ways to prevent mother-to-child transmission.
To limit the selection for NNRTI resistance mutations, single-dose nevirapine treatment was evaluated alone or together with 4 or 7 days of treatment with zidovudine/lamivudine (Combivir) in 226 mothers and 228 infants. Population sequencing of specimens over 6 weeks showed NNRTI resistance in 39 of 68 (57%) mothers receiving single-dose nevirapine, 9 of 67 (13%) mothers receiving single-dose nevirapine plus 4 days of zidovudine/lamivudine, and 6 of 68 (9%) mothers receiving single-dose nevirapine plus 7 days of zidovudine/lamivudine.[6] In a subgroup of 32 women, more sensitive allele-specific reverse transcriptase PCR detected K103N or V181C variants in week 6 samples from 75% of women receiving single-dose nevirapine alone, and in samples from 6 of 22 women (27%) receiving single-dose nevirapine plus 4 or 7 days of zidovudine/lamivudine.[7] Thus, short-course zidovudine/lamivudine therapy reduces but does not eliminate emergence of nevirapine resistance induced by single-dose therapy.
One approach to preventing mother-to-child transmission without engendering maternal nevirapine resistance involved withholding single-dose nevirapine from women and treating infants only with either single-dose nevirapine or nevirapine plus zidovudine twice daily for 7 days (Table).[8] Although transmission rates were similar for women who did or did not receive single-dose nevirapine, frequency of nevirapine resistance was lowest in infants who received single-dose nevirapine plus zidovudine and were born to women who did not receive single-dose nevirapine (P < .001).
SD-NVP= single-dose nevirapine; ZDV= zidovudine
This strategy might spare women almost certain emergence of NNRTI resistance, as was seen with previous use of single-dose nevirapine. In addition, nevirapine resistance was significantly reduced in infants by avoiding predelivery nevirapine therapy in mothers and by giving infants SD-NVP plus zidovudine for 7 days. Because use of SD-NVP to reduce mother-to-child transmission is fraught with so many problems, most advisory groups now appropriately discourage this practice; other affordable ways to safely prevent mother-to-child transmission in Africa are promising.
More sensitive methods of resistance detection have also been applied to studies of horizontal transmission, and transmission rates of 10% to 25% for HIV drug resistance appear to be seen across various primary infection studies. In a study with allele-specific resistance detection targeting only 3 common mutations, 1 for each class (protease inhibitors: L90M; NNRTIs: K103N; nucleoside reverse transcriptase inhibitors: M184V), drug-resistant variants were detected in 10 of 49 patients (20.4%) overall. K103N was seen in 5 of 49 patients (10.2%), M184V was seen in 6 of 49 patients (12.2%), and L90M was detected in 1/49 patients (2%).[9] This suggests that the rate of transmission of resistance in primary infection is actually much higher, since many commonly seen mutations were not considered in this study.
Of great importance, resistant virus transmission can persist for much longer than previously appreciated. Earlier published reports from the primary infection research group at University of California, San Diego, reported that K103N, in particular, persisted for up to 2 years.[10] On this basis, current Department of Health and Human Services guidelines [11] suggest that baseline resistance testing should be performed before treatment if individuals are thought to have been infected in the past 2 years. In a study to evaluate persistence of resistance in paired blood and semen samples, the same group showed detection of NNRTI resistance mutations beyond 4 years in both blood and semen.[12] This suggests that guidelines should probably be revised to recommend universal baseline resistance testing, particularly in settings where acquisition of drug-resistant virus is likely.
Two of the workshop presentations caused concern about new trends in transmission.[13,14] One was from Kozal and colleagues[14] at Yale University, New Haven, Connecticut, and the other was from the Study of Consequences of PI Era (SCOPE) at the University of California, San Francisco, led by Chin-Hong and others.[13] The Kozal and colleagues conducted a longitudinal study of transmission risk behaviors in a diverse HIV-positive population. Surveys regarding self-reported risk behaviors were completed by 393 HIV-positive patients from 2000-2003. In addition, HIV viral load and genotypic resistance data were obtained. The study population was 44% female and 79% heterosexual; 38% were African American, 34% were Hispanic, and 22% were white. Of the 250 patients (64%) who had sex during the study period, 112 (45%) reported engaging in risky sexual behavior that might allow HIV transmission. Drug resistance was fairly common in this population; 21% had resistance to 1 drug class, 12% had resistance to 2 drug classes, and 2% had multidrug, triple-class resistance at the time of the risky sexual events. It is concerning that patients with drug-resistant HIV continue to engage in risky behavior for HIV transmission, and it is noteworthy that 11% of risky sexual events involved patients harboring HIV resistant to at least 2 drug classes.
In the San Francisco study, 465 men and women were followed in a clinic-based cohort of HIV-infected adults. This population was 89% male, 60% white, and 21% heterosexual. Of the 189 participants with genotypically confirmed drug resistance, 29% reported unsafe vaginal or anal sex with HIV-negative partners or with partners whose status was unknown. The 2 most prominent predictors of unprotected intercourse were methamphetamine use (odds ratio [OR], 4.2; 95% CI, 1.6-11.3; P = .004) and sildenafil use [Viagra] (OR, 3.7; 95% CI, 1.7-8.3; P = .001). Both of these drugs were associated with high-risk sexual behavior with HIV-uninfected partners, but sildenafil was particularly associated with unprotected sexual behavior between HIV-infected partners. With the potential for HIV superinfection, this behavior could result in transmission of virus strains that can worsen or accelerate disease. These studies highlight the need for improved education and community intervention to try to diminish recreational drug use that can lead to high-risk behaviors.
In considering the process of reverse transcription as a therapeutic target, our attention is naturally focused on the polymerase function that is the target of our antiretroviral drugs, the nucleoside- and nonnucleoside reverse transcriptase inhibitors (NRTIs and NNRTIs, respectively). However, ribonuclease (RNase) H, localized to the C-terminus of the p66 subunit of reverse transcriptase, performs the obligate step of degrading the RNA template after it is transcribed into DNA, permitting formation of the complementary DNA strand. RNase H activity and DNA polymerase activities reside in spatially distinct domains of the reverse transcriptase enzyme. Interactions between these 2 activities can occur according to distinct models.
Last year, Nikolenko and colleagues[1,2] proposed that an equilibrium exists among nucleoside incorporation into host DNA, excision of the nucleoside (mechanism for zidovudine [Retrovir] resistance), resumption of DNA synthesis, and RNase H activity. Whereas RNA degradation before resumption of DNA synthesis would lead to the dissociation of the template primer and subsequent abrogation of HIV-1 replication, mutations in RNase H that reduce the rate of degradation would promote resistance by increasing the time available for excision of incorporated NRTIs and permitting resumption of viral replication. Two mutations in the RNase H domain, H539N and D549N, did in fact appear to augment thymidine analogue mutation (TAM)-mediated zidovudine resistance. This finding would suggest that RNase H mutations will be selected in NRTI-exposed patients. Thus, Nikolenko and colleagues[3] continued their investigations this year by cloning RNase H domains obtained from 7 antiretroviral-naive or 8 NRTI-experienced individuals into HIV-1 vectors containing a wild-type or drug-resistant polymerase domain, which in turn contained TAMs. The results of these studies were presented by Dr. Vinay K. Pathak. Although the RNase H domain from unexposed virus did not alter zidovudine susceptibility, the domain from treatment-experienced individuals conferred increased zidovudine resistance in the context of the wild-type polymerase (2.4- to 5.7-fold) and substantially more augmentation when combined with the TAM-containing polymerase (96- to 1839-fold increase in zidovudine resistance) in 5 of the 8 RNase H domains, although the role of the 2 previously identified mutations (H539N and D549N) was not investigated. However, these findings support the hypothesis that NRTI treatment will select for RNase H resistance mutations. Marcelin and colleagues[4] looked at a larger group of patients (36 treatment-naive, 118 with NRTI pretreatment) and found extensive polymorphism in the RNase H sequence of treatment-naive patients, as presented by Vincent Calvez, MD, PhD. Mutations at positions 469, 470, 554, and 558 were more prevalent in pretreated patients than in treatment-naive patients, and mutations at position 558 were associated with the presence of TAMs. Of interest, mutations at positions 539 and 549 were not observed in this set. Before anyone starts adding RNase H mutations to their memory lists, it would be prudent to wait for confirmation from other studies. In the meantime, the findings raise the following question: Do we need to include the RNase H domain in genotypic and phenotypic resistance assays?
Another model for the interaction of polymerase and RNase H is based on long-range effects of NNRTI binding to the polymerase domain. Both increased and decreased RNase H activity following NNRTI binding have been reported. Hang and colleagues[5] reported NNRTI-mediated inhibition of 5'-recessed RNA-directed RNase H activity and stimulation 3'-recessed DNA-directed RNase H activity. This effect is mediated through NNRTI binding; the inhibition potency on mutant (NNRTI-resistant) reverse transcriptase correlates with the inhibition of polymerase activity, consistent with a long-range effect of NNRTI binding.
Miller and colleagues[6] addressed the question of whether RNase H inhibitors will mediate resistance to zidovudine and stavudine (Stavudine). Using an assay system that measures only RNase H activity, only polymerase activity, or both, Miller and colleagues investigated the combination of inhibitors affecting RNase H activity (a diketo acid), polymerase activity (zidovudine-triphosphate, NNRTIs), and foscarnet, which inhibits both activities. When the authors looked at RNase H activity only, zidovudine-triphosphate did not affect the potency of the diketo acid, whereas the NNRTI decreased the diketo acid potency, a finding consistent with NNRTI enhancement of RNase H activity. The mechanism for enhancement of RNase H activity appears to be based on reduction of relative mobility of the thumb and RNase H regions, causing increased accessibility of the RNase H active site to the RNA/DNA hybrid duplex upon NNRTI binding. In contrast, the diketo acid showed synergistic inhibition with each of the polymerase inhibitors in assays that require both activities, leading the authors to conclude that a small molecule inhibiting RNase H activity would not confer resistance to zidovudine and that regimens incorporating both NNRTIs and RNase H inhibitors may be therapeutically beneficial.
To return to the question raised earlier, will it be necessary to include patient-derived C-terminus reverse transcriptase domains (RNase H and integrase activities) into resistance assays? Gupta and coworkers[7] addressed this by comparing susceptibility measurements using test vectors with or without the patient-derived RNase H and integrase domains. Resistance test vectors including the standard protease and reverse transcriptase segments (PRRT, codons 1-305 of RT), RHIN (codons 305-560 of reverse transcriptase and integrase), or the entire pol gene were constructed from viruses from 27 patients. Overall, susceptibility did not differ greatly when the standard PRRT was compared with the complete pol vectors. However, the patients' treatment history and the number and nature of resistance mutations were not reported, so it is not clear whether TAMs were present. The greatest variability in fold resistance was observed for NNRTIs: One patient's virus exhibited 6- to 10-fold increased NNRTI resistance when the pol test vector was used. This was mapped to position T369 in reverse transcriptase. Further analysis using these approaches will be required to confirm these findings.
In summary, RNase H activity was this year's "fashion" mechanism. Whether clinically useful RNase H inhibitors will reach the clinic remains to be seen. In the meantime, we should not forget the potential impact of domains not normally included in resistance assays, especially since these are structurally "stuck" on (ie, structurally continuous with) the much more familiar polymerase activity domains.
In case anyone thought we had NRTI resistance nailed down, leaving the semantics of TAMs and nucleoside analogue resistance mutations as the only points of discussion, this year's meeting sent a loud wake-up call. It appears that resistance to zidovudine occurs via the accumulation of thymidine analogue mutations, or TAMs, via 1 of 2 possible pathways.[8] In an elegant study presented by Ceccherini-Silberstein,[9] reverse transcriptase sequences from 551 drug-naive and 1355 NRTI-treated patients were compared by using hierarchical clustering and bootstrap analysis. Several interesting observations were made.
First, 12 mutations were positively associated with NRTI treatment and clustered with NRTI mutations. T39A, K43E/Q, K122E, E203K, and H208Y clustered with the TAM 1 cluster (M41L, L210W, T215Y) were associated with a 42-fold increase in zidovudine resistance and with increased viremia in patients failing therapy. D218E clustered with the TAM 2 group (D67N, K70R, K219Q, T215F).
Second, 2 mutations (I50V, K83R) were negatively associated with NRTI treatment, displayed antagonistic correlations with NRTI mutations, and were associated with increased susceptibility of viruses at the time of failure. I50V was antagonistic with M184V and was associated with reduced lamivudine resistance in M184V-containing viruses.
Third, F214L was positively associated with the TAM 2 cluster and was negatively associated with the TAM 1 cluster.
This study was based on subtype B viruses; the implications for non-B viruses are not known. The findings have significant implications for genotype interpretation and algorithm development and illustrate the fallacy of focusing only on "well known" genotypic patterns.
Geretti and colleagues[10] presented another cohort study investigating the role of additional mutations in zidovudine resistance. The strong correlation of H208Y with drug experience confirms the finding of the study by Ceccherini-Silberstein and associates and suggests a compensatory role for this substitution. Both studies suggest significant interactions occurring between reverse transcriptase mutations. It will not be long before this type of information finds its way into the resistance interpretation system. More work on non-B subtypes is urgently needed.
For those who are unable to digest the complexity of HIV drug resistance, there may be comfort. Resistance to protease inhibitors was found without appearance of "known" mutations in a small percentage of samples in the Virologic database.[11] The mechanisms may involve accumulation of secondary mutations, changes in the C-terminal region of gag, or both. A study presented by van Maarseveen[12] showed increased nucleocapsid/p1 processing to be the underlying mechanism in gag-related HIV protease resistance in the absence of mutations. However, other studies continue to illustrate the complexity of mutations, mutation interactions, and resistance pathways for protease inhibitors. For example François Clavel, MD, described the resistance pathway in patients with protease inhibitor-resistant viruses treated with atazanavir (Reyataz).[13] The emergence of I50L was associated with significantly increased resistance to atazanavir and reduction in resistance to other protease inhibitors. Other mechanisms involved in increased atazanavir resistance included the accumulation of "minor" mutations.
Several presentations described differences in resistance pathways in non-B subtype viruses.[13-22] These will be of increasing importance as we expand treatment in the developing world and as more patients contract non-B subtype viruses in Europe and in the United States.
Costagliola[1] presented a stunning analysis on behalf of the Standardization in Clinical Relevance of HIV Drug Resistance Testing Project from the Forum for Collaborative HIV Research (The Forum). The Forum set up an initiative to evaluate the relationship between baseline genotype and virologic outcomes from treatment. The project was an attempt to bring standardization and consistency into resistance interpretation systems. The activities include the development of algorithms for interpretation of baseline drug resistance using treatment outcome-linked databases and the comparison of methodologies used for making the association (eg, neural networks, logistic regression, tree partitioning, etc). The first step consisted of comparing different algorithms for didanosine (Videx) and abacavir (Ziagen). These 2 drugs were chosen because of the need for more information on didanosine resistance and the need to confirm GlaxoSmithKline's publications on genotypic and phenotypic cutoffs for abacavir. The analysis was carried out according to the published data analysis plan.[2]
For each interpretation system and each drug, a regression model was fitted with sensitivity as a covariant (resistant-R, intermediate susceptibility-I, and sensitive-S). For abacavir, the percentage of R viruses ranged from 7.35% (ANRS) to 31.9% (VGI-Bayer). Three of the systems showed no significant association between change in viral load and sensitivity, and among other systems, all but 1 showed a larger virologic response for I viruses than for S viruses. These findings suggested problems with the various algorithms. Performance was even worse for didanosine, with failure to show significant association between change in viral load and sensitivity for most systems. Concomitant therapy (the number of additional active drugs in the regimen) was controlled for using 1 of the algorithms; however, the researchers did not account for the relative potency of the "active drugs." Nonetheless, data from these analyses will help as forward-looking interpretation systems modify their algorithms and take advantage of this external validation.
Several papers assessed baseline genotypic resistance and clinical response in randomized clinical trials of atazanavir (Reyataz), and suggestions were made for slight modification of phenotypic fold-change cutoffs and for genotypic resistance mutation scores. Determination of phenotypic cutoffs for atazanavir and ritonavir (Norvir)-boosted atazanavir were derived by correlation of baseline phenotype with virologic response in studies AI424-043 and AI424-045, respectively.[3] BMS AI424-043 was a study in individuals who had failed at least 1 regimen and who were randomly assigned to receive atazanavir or a fixed-dose combination of lopinavir and ritonavir (Kaletra). The mean (median) baseline atazanavir fold change for 131 individuals was 2.9 (1.2), and Fisher's exact test favored a fold-change cut-point of 2.2; an HIV RNA level < 400 copies/mL at week 24 was seen in 76% of patients with a baseline atazanavir fold-change cut-point < 2.2 and in 45% of patients with a baseline atazanavir fold change >/= 2.2. This is very close to the biologic cut-point of 2.1 derived by using the 99th percentile for atazanavir susceptibility from an analysis of nearly 10,000 plasma samples from HIV-infected individuals without recognized protease mutations.[3]
Since most individuals treated with atazanavir receive coadministered low-dose ritonavir, the analysis from the BMS AI424-045 study is more relevant. This study enrolled patients who experienced virologic failure of at least 2 previous regimens and were randomly assigned to receive ritonavir-boosted atazanavir vs coformulated ritonavir-boosted lopinavir LPV/r, with nucleoside reverse transcriptase inhibitors (NRTIs) changed to tenofovir (Viread) after 2 weeks and with another NRTI added on the basis of resistance testing. For the 111 individuals treated with ritonavir-boosted atazanavir for whom data are available, Fisher's exact test favored a fold-change cut-point of 5.2; 70% of individuals with baseline fold change < 5.2 had plasma HIV RNA levels < 400 copies/mL at week 24, compared with 12% of individuals with baseline atazanavir fold change >/= 5.2. Although this is helpful, one should think of phenotypic fold-change test results as being continuous variables, with placement of specific cutoffs merely as a guide to help determine whether virus is likely to be susceptible. It would be useful to perform similar analyses for other studies of ritonavir-boosted atazanavir to confirm limitations of response based on this cut-point.
While interpretation of phenotypic tests is more straightforward, most resistance testing relies on algorithm-based interpretation of genotype. A resistance score has been derived for lopinavir, and it has been validated and refined by subsequent analyses. However, little has yet been done to derive an atazanavir resistance score. In a retrospective cohort study,[4] a large number of protease inhibitor (PI) mutations were associated with a reduced response to ritonavir-boosted atazanavir, and while there is some overlap, many of these are different from the PI mutations initially identified by the earlier trials conducted by Bristol-Myers Squibb. In 62 PI-experienced patients switched to ritonavir-boosted atazanavir plus other antiretroviral agents, the strongest association with reduced virologic response was found with the combination of 10F/I/V, 16E, 33I/F/V, 46I/L, 60E, 84V, and 85V. A virologic response (> 1.0 log10 decrease) was observed in 100%, 100%, 80%, 32%, and 0% of patients with 0, 1, 2, 3, and >/= 4 mutations, respectively, based on an atazanavir resistance score derived from these mutations. There was also an association with virologic response and number of active drugs (P =.001): For patients with an atazanavir resistance score >/= 3, virologic response was observed in 0%, 29%, and 60% of patients with 0, 1, and 2 to 3 active drugs, respectively (P = .024). The limitation of this study is that the same data set was used to derive the resistance score and to validate it. Since this resistance score was somewhat different from others that have been proposed, it would be of interest to apply it to data from other studies of ritonavir-boosted atazanavir in patients who have experienced PI failure.
In another observational cohort[5] in which regimens containing ritonavir-boosted atazanavir were initiated in 90 antiretroviral-experienced patients (PI use = 91%), virologic response was correlated with atazanavir resistance-related mutations at baseline. These included mutations at positions 10, 20, 24, 33, 36, 46, 48, 54, 63, 71, 73, 82, 84, and 90, which are derived from the French ANRS genotype interpretation guidelines. Overall, the 24-week median decrease in plasma HIV RNA level was -1.2, 25th to 75th confidence interval: -2.8; -0.3) log10 copies/mL; 66% of patients had a viral load of < 50 copies/mL. When the atazanavir score was < 6 vs >/= 6 mutations, the median change in plasma HIV-1 RNA level was -1.74, 25th to 75th confidence interval:-2.8; -0.4) vs 0.01 (-1; 0.3) log10 copies/mL (P = .01), and virologic suppression was seen in 75% vs 20% (P = .002), respectively. Clearly, the mutation score needed to determine the likelihood of virologic success depends on the pool of codons analyzed for potential resistance mutations. These initial assessments help to define the likelihood of drug activity and will be useful in refining existing genotypic algorithms.
Virologic response was correlated with baseline resistance in several clinical trials of tipranavir (Aptivus), including the phase 3 RESIST studies; the companion trial (1182.51) for patients with > 2 mutations at codons 33, 82, 84, and 90 (who were therefore not eligible for RESIST); and the phase 2 dose-finding trial (1182.52).[6] A tipranavir mutation score was developed by using uni- and multivariance regression analysis to correlate baseline genotypes with viral load reduction at week 2 or week 24. Reduced tipranavir susceptibility or reduced response was associated with 21 mutations in 16 positions: 10V, 13V, 20M/R/V, 33F, 35G, 36I, 43T, 46L, 47V, 54A/M/V, 58E, 69K, 74P, 82L/T, 83D, and 84V. Virologic response in the RESIST trials was generally excellent at all mutation scores, but correlation with phenotypic susceptibility showed a substantial jump for a tipranavir mutation score > 7. The tipranavir fold change in susceptibility was < 2.0 for </= 4 mutations, 3.1 to 3.9 for 5 to 7 mutations, and over 14 for >/= 8 mutations. This correlated with week 2 virologic response in that patients with 7 or fewer mutations had a decrease > 1.2 log10 copies/mL in viral load, whereas with >/= 8 mutations, viral load reduction was 0.33 log10 copies/mL. The surprisingly good response in the companion trial (1182.51), with >/= 3 mutations at codons 33, 82, 84 and 90, might be explained by the new mutation score, where 90 is really not correlated with virologic response and the mutations at position 82 that are associated with resistance to other drugs (82A/F) do not appear to be important for tipranavir susceptibility; for tipranavir resistance, the mutations are 82L/T. This extensive analysis supports use of ritonavir-boosted tipranavir in patients with extensive resistance in whom other PI-based regimens have failed. The magnitude and durability of the response to ritonavir-boosted tipranavir are improved with coadministration of enfuvirtide (Fuzeon).
Several reports served to caution us about drug combinations that may not be as effective as hoped and to alert us to other combinations that may offer advantages in salvage therapy because previously selected resistance mutations may enhance susceptibility of certain other drugs. In the first category, Jemsek and colleagues[7] showed that a regimen of tenofovir + didanosine + lamivudine (Epivir) once daily was not able to suppress the viral load to < 50 copies/mL in any of 24 treatment-naive patients. Other triple-NRTI studies with tenofovir/abacavir/lamivudine had markedly suboptimal response rates, with frequent emergence of M184V and K65R. Although the mechanism for the low response rates when using tenofovir with either didanosine or abacavir in 3-drug regimens consisting solely of NRTIs is still not well elucidated, tenofovir/didanosine has been a fairly popular combination and is thought to be adequate when used with an active third drug. Two studies presented at this meeting, however, suggest that tenofovir/didanosine may be inferior to other commonly used dual nucleoside backbones when used with efavirenz (Sustiva). Gatell and colleagues[8] compared antiretroviral regimens containing tenofovir/didanosine in treatment-naive patients with other regimens that had different backbones. After a median follow-up of 23 months, 19 of 92 (21%) patients failed a regimen that included tenofovir/didanosine; 7 of 19 (37%) had received an NRTI as the third drug, 11 (12%) were receiving an NNRTI, and 1 (1%) received ritonavir-boosted lopinavir. In addition to the relatively high failure rate for treatment-naive patients, resistance mutations were detected in 18 of 19 of patients (95%).
In another study, the EFADITE trial from Spain,[9] patients receiving stable highly active antiretroviral therapy (HAART) for more than 6 months with plasma HIV RNA levels < 50 copies/mL were switched to tenofovir/didanosine/efavirenz once daily or continued to receive the same HAART regimen they had been taking. Although plasma HIV RNA levels < 400 copies/mL were achieved in 95% of patients in the once-daily group and 92% of controls, the CD4+ cell count in the once-daily tenofovir/didanosine/efavirenz group was actually 25 cells/mcL lower at 12 months, as compared with 46 cells/mcL higher in the control group (P = .007). These 2 new studies[8,9] increase concern about the efficacy of the tenofovir/didanosine combination as part of any HAART regimen, and studies to try to understand why this combination may be less efficacious should be pursued.
Finally, there was new information about treating individuals who have the K65R resistance mutation. This is selected by tenofovir, and although K65R is still relatively rare, its frequency has increased from < 1% to > 4% over the past few years. There is an apparent incompatibility of K65R with L74V, and perhaps also with thymidine analogue mutations (TAMs). In patients with multitreatment failure, the mutation pattern seen on the genotypic resistance test may be complex, but all of the mutations may not be seen in the same genome, that is, they are not colinear mutations. Since K65R tends to increase zidovudine (Retrovir) susceptibility, use of zidovudine following tenofovir failure is logical. Three patients who experienced virologic failure associated with K65R and other resistance mutations had the regimen intensified with zidovudine without changing other drugs.[10] In all 3 patients, viral load decreased to < 50 copies/mL, with reductions of viral load ranging from 1.4 to 3 log10 copies/mL. Zidovudine together with tenofovir is an increasingly popular approach in salvage therapy and is typically coadministered with either lamivudine or emtricitabine (Emtriva). This may be effective since zidovudine is highly active against HIV with the K65R mutation, tenofovir is typically is active against virus strains with multiple TAMs, and the M184V mutation from lamivudine or emtricitabine tends to increase activity of zidovudine and tenofovir in the presence of TAMs and of tenofovir in the presence of K65R. In this way, it may be possible to obtain a virologic response by using zidovudine/tenofovir with either lamivudine or emtricitabine, since there is likely to be activity in patients whose HIV has developed multiple mutations.
CCR5 antagonists represent one of the most exciting new HIV drug classes under investigation. The concept of targeting a host receptor molecule rather than a viral target is not new, but this drug class has traveled the longest along the development pathway. Three companies are developing CCR5 antagonists, with all compounds currently in phase 2 or 2/3 development.
The mechanisms whereby HIV can generate resistance to entry inhibitors include competitive inhibition, noncompetitive inhibition (the virus uses an inhibitor-receptor complex), and receptor switch (the virus uses the CXCR4 [X4] receptor rather than the CCR5 [R5] receptor). Real concerns exist regarding the long-term impact of tropism switch from R5 to X4 tropic viruses; thus, resistance and escape are being followed very closely. X4 viruses (previously known as syncytium-inducing viruses) are associated with faster disease progression, although it is not known whether the switch from R5 to X4 prompts disease progression or whether severe immunodeficiency prompts a switch to X4 viruses.
During last year's Workshop, Westby[1] reported slow or no development of resistance to Pfizer's CCR5 antagonist maraviroc (UK-427,857) in vitro. Only 1 of 3 viruses used the receptor switch pathway. At this year's Workshop, Westby and coworkers[2] reported further analysis of clones generated from the resistant isolates used in the earlier studies. Resistance involved the accumulation of mutations in the V3 loops of viral gp120, with A315T apparently playing a critical role.
Strizki and others[3] confirmed the slow evolution of resistance in culture for the Schering-Plough compounds SCH 351125 and SCH 417690 (vicriviroc). Resistant viruses emerged in only some cultures after 12 to 16 weeks of continuous passaging and generally did not use the receptor switch mechanism; only 1 viral isolate was able to replicate in CXCR4-positive cells. Genetic correlates for resistance revealed unique mutation patterns in the V3 loop and clustering in the V3 cap region.
As we learn more about the resistant genotypes, we need more information about the clinical relevance of the natural sequence variation in the relevant gp120 viral regions. Natural variation at positions 318 and 319 in the V3 loop, for example, appears to have significant effects on sensitivity to entry inhibitors, efficiency of host cell entry, and viral replicative capacity.[4] How the natural variation relates to the resistance selection will be a subject for future meetings.
Isolated cases of an apparent viral receptor use switch (from R5 to X4 receptor use or to R5/X4 mixed or dual-receptor use) associated with CCR5 antagonist treatment have been reported from early phase 2 clinical studies. Kitrinos and colleagues[5] reported on 1 such case, a patient participating in the 10-day monotherapy study of the GlaxoSmithKline compound 873150. Dual-tropic virus was detected at day 10, whereas only R5-tropic viruses were detected at screening and at days 1 and 5 and again on day 24. Viral load decrease in this patient was comparable to the median decrease in the treatment group. An extensive clonal analysis (48 clones analyzed for each time point) revealed the presence of dual-tropic viruses at low frequency (4% and 8%) at days 1 and 24, respectively, which were below the population assay threshold of 10%. The fact that the V3 loop sequences of the R5 and R5/X4 clones were distinct further substantiates the conclusion that the detection of dual-tropic virus at day 10 was due to emergence (shift in the viral population) rather than a receptor use switch. This should not come as a surprise; it is reminiscent of the minor variant populations in other drug resistance studies leading to rapid shifts in dominant populations that depend on the presence or absence of selective drug pressure. However, he present results, which are based on a single case, do not negate the possibility that viral escape using the receptor tropism switch mechanism could indeed take place over longer periods.
Cross-resistance between CCR5 antagonists. Intraclass cross-resistance, unfortunately, is a familiar theme for those involved in drug discovery or clinical treatment. So far, we have had few data regarding cross-resistance in the CCR5 antagonist drug class. Several groups addressed this issue at this year's Workshop. The amount of cross-resistance will depend on the binding site of the individual inhibitors within the CCR5 receptor and the nature of the mutations in the viral gp120. The 4 maraviroc-resistant clones described earlier were not cross-resistant to the other CCR5 antagonists.[2] Maraviroc and other CCR5 antagonists are thought to bind to a pocket in the transmembrane region of the CCR5 receptor, but the compounds appear to occupy a slightly different molecular area within the binding pocket, which may explain the lack of cross-resistance. However, in the study by Strizki and colleagues,[3] nonclonal isolates resistant to SCH 351125 and SCH 417690 were cross-resistant to other CCR5 antagonists. Of interest, both studies used viruses generated from the same original CC1/85 primary isolate. Thus, as with cross-resistance in other drug classes, cross-resistance within the CCR5 antagonist class may not be initially evident in vitro, yet this is no guarantee that it will not occur in vivo.
Cross-resistance between CCR5 antagonists and enfuvirtide. Of note, there appears to be no cross-resistance between the CCR5 antagonists and the fusion inhibitor enfuvirtide (Fuzeon).[3,6] Enfuvirtide-resistant viruses, including clinical isolates, were fully susceptible to 873140. In addition, the 2 drugs in combination appear to act synergistically.[6] Additional resistance and cross-resistance studies, especially with virus escape isolates from patients receiving treatment, are needed to expand our understanding of how to best use this new drug class in combination regimens.
Tibotec scientists first presented this new drug class with potent anti-HIV activity at the 12th Conference on Retroviruses and Opportunistic Infections.[7] Nucleotide-competing reverse transcriptase inhibitors (NcRTIs) differ from nucleoside reverse transcriptase inhibitors (NRTIs) in terms of structure and the lack of requirement for phosphorylation, and differ from nonnucleoside reverse transcriptase inhibitors (NNRTIs) in terms of mechanism (competitive). The compounds bind to the active site of reverse transcriptase and compete with the next incoming nucleotide. Of interest, NcRTIs retain full activity against NNRTI-resistant strains; thus, they are potentially of great clinical interest. Jochmans and others[8] presented data demonstrating the enhancement (approximately 20-fold) of NcRTI activity by physiologic concentrations of adenosine triphosphate (ATP) for both HIV-1 and HIV-2 reverse transcriptase. While the mechanism for this effect requires further elucidation, it may involve trapping the primer/template complex in its post-translocation conformation with ATP bound to the pyrophosphate site.[8,9] Stay tuned for further advances, including information on the effect of reverse transcriptase mutations affecting ATP binding and the activity of the NcRTIs in combination with other reverse transcriptase inhibitors.
TMC-114 is a potent protease inhibitor inhibiting both wild-type and drug-resistant HIV-1. A palpable buzz followed the presentation of the 24-week clinical data at the 12th Conference on Retroviruses and Opportunistic Infections. The researchers demonstrated a reduction of > 1.8 log10 copies/mL in levels of HIV-1 RNA (for the highest-dose arm) in triple-class-experienced patients with extensive resistance to protease inhibitors.[10] Dierynck and colleagues[11] have begun to explore the molecular basis for the drug's high potency and broad-spectrum activity. The compound binds with exceptionally high affinity (KD = 4.11 × 10−13, or 3 to 4 orders of magnitude higher than other protease inhibitors) to wild-type protease. The high affinity is due to a very slow dissociation rate combined with a fast association rate, and appears to be dependent on the benzenesulfonamide substituent in the P2' pocket of TMC-114. It will be of interest to see whether the kinetics for drug-resistant protease remain the same.