Report on the XIV International HIV Drug Resistance Workshop: Parts 1-3

Following is the first of two documents that review the major data presentations at the XIV International HIV Drug Resistance Workshop held in Quebec, Canada in June. Both sections of the report are based on the official “Rapid Report” on the workshop by Mark Mascolini.

Part 1A. mother-to-child transmission

Emergence of resistance to nonnucleosides limits the value of single-dose nevirapine/NVP (Viramune) used to prevent perinatal transmission of HIV. Three reports suggested strategies that may cut resistance risks in both mothers and infants.

Adding AZT to NVP for infants

Adding zidovudine/AZT (Retrovir) to short-course NVP in newborns of HIV-infected women and eliminating the mother’s single NVP dose lowered NVP resistance rates in neonates without boosting the risk of HIV transmission.

Susan Eshleman (Johns Hopkins University, Baltimore) and colleagues in Malawi treated HIV-infected women and their infants with one of four antiretroviral regimens and found significantly different rates of NVP resistance among HIV-infected infants:

·         Single-dose NVP for mothers and infants: 87% resistance

·         Single-dose NVP for mothers and single-dose NVP plus AZT twice daily for 7 days for infants: 57% resistance

·         No NVP for mothers (presenting late in labor) and single-dose NVP for infants: 74% resistance

·         No NVP for mothers (presenting late in labor) and single-dose NVP plus AZT twice daily for 7 days for infants: 27% resistance

Further analysis revealed that no NVP for mothers and the addition of AZT for infants independently decreased the risk of resistance.  Eliminating the mothers’ risk of drug-induced NVP resistance also preserves their options for chronic antiretroviral therapy. However, Eshleman cautioned that her results require confirmation and that single-dose NVP for mothers should not be abandoned on the basis of her data.
 
Single-dose NVP plus Combivir

Adding Combivir (AZT/3TC) to single-dose NVP in both mothers and newborns reduced rates of NVP resistance from 60% to 10-12% in mothers and, in infected infants, from 88% to 14% for those receiving CBV for 4 days and 0% for those receiving CBV for 7 days. 

Monitoring resistance 2 and 6 weeks after delivery in 226 mothers and 228 infants, David Hall (Boehringer Ingelheim) and colleagues found a wide array of nonnucleoside-related mutations including K103N, Y188C, Y181C, V106M, A190G, and V106A. No 3TC-associated M184V mutations arose in either mothers or newborns. This ongoing study has not yet distinguished between the protective merits of 4-day versus 7-day Combivir.

Looking more closely for NVP resistance with a highly sensitive test in a subgroup of these women, Sara Palmer (National Cancer Institute, Frederick) learned that adding Combivir did not completely eliminate the risk of resistance. While standard sequencing failed to detect any NVP-associated mutations in those women receiving NVP plus Combivir, the ultra-sensitive assay did so in 27%. Whether the traces of resistant virus seen with this allele-specific PCR test will have a clinical impact remains to be seen.

           
Breast milk viral load with NVP versus AZT

Between delivery and 6 weeks postpartum, Dara Lehman (Fred Hutchinson Cancer Research Center, Seattle) collected 795 milk samples from 30 women treated with single-dose NVP or short-course AZT. Comparing viral loads in the two sample groups, she documented significantly lower loads in the NVP-treated group (2.3 log ands 2.25 log at 8-14 and 15-21 days postpartum) compared with AZT (2.8 and 3.1 log).

A highly sensitive genotyping test (allele-specific PCR again) detected the K103N nonnucleoside mutation in plasma samples of 36% of NVP-treated women 1 month after delivery compared with 10% by standard sequencing.  Lehman and colleagues concluded that emergence of resistance to nonnucleosides may compromise the greater antiviral activity of single-dose NVP compared with short-course AZT.

Part 1B. prediction of clinical response

Three ways to predict response to atazanavir (ATV)

Three research groups analyzed different benchmarks to predict virologic response to atazanavir/ATV (Reyataz) or ATV boosted with ritonavir (ATZ/rtv) in people with PI experience: phenotypic sensitivity cutoff; genotypic resistance score; and genotypic inhibitory quotient.

Eoin Coakley and colleagues used the PhenoSense assay to establish clinical cutoffs in 131 people who started unboosted ATV (400 mg daily) plus two nucleosides, and 111 who started ATV/rtv (300/100 mg daily) plus tenofovir/TDF (Viread) and one nucleoside, all after PI failure.   In the unboosted ATV group, a 2.2-fold cutoff appeared to distinguish people who reached a viral load below 400 copies/mL from those who did not (76% of those below the cut-off achieved <400 copies after 24 weeks, compared with 45% of those above). In the ATV/rtv trial, 5.2-fold seemed a better response cutoff.

Coakley cautioned that his statistical analysis did not take background regimens into account, or the effect of tenofovir, which lowers levels of ATV. Workshop attendees observed that the moderate level of resistance to PIs in the two studies make Coakley’s results difficult to apply to people with more resistant virus.

Studying 62 French and Swiss patients starting ATV/rtv after a history of treatment with a median of 7.5 antiretrovirals including 2 PIs, Anne-Geneviève Marcelin (Pitié-Salpêtrière Hospital, Paris) defined a mutation score that predicted virologic response.  She found that a score based on mutations at 7 positions proved the best predictor of a 1 log drop in viral load after 3 months: 10F/I/V, 16E, 33I/F/V, 46I/L, 60E, 84V, and 85V.

The mutations at positions 16, 60, and 85 have not been described as correlates of response to PIs, although the Stanford HIV resistance database lists them as potential contributors. Virologic response correlated closely with the number of these mutations ranging from 100% response among those with none to 0% response among those with 4 or 5.

Genotypic inhibitory quotient (GIQ) and an ATV mutation score predicted virologic success in 90 people starting ATV/rtv after virologic failure. But, in this analysis by Isabelle Pellegrin (Bordeaux University Hospital), ATV drug levels by themselves did not.

Most study participants had PI experience (91%) and nonnucleoside experience (62%), and they had tried a median of 5 antiretroviral regimens. The group had a baseline median of 2 ATV mutations. Pellegrin calculated GIQ as ATV minimum concentration (Cmin) divided by the number of ATV/rtv resistance mutations: L10F/I/V, K20I/M/R, L24I, L33F/I/V, M36I/L/V, M46I/L, G48V, I54L/V, L63P, A71I/V/T, G73A/C/S/T, V82A/F/T/S, I84V, and L90M.

After 6 months of ATV/rtv, the median viral load fell 1.2 log copies/mL and 66% had at least a 1.7-log drop in viral load. ATV levels did not predict response but two factors did: GIQ and Cmin in people with ATV resistance score below 6. Univariate analysis identified three correlates of 6-month virologic failure: baseline number of protease and reverse transcriptase mutations, ATV/rtv resistance score and L10I, K20R, M46I/L, 54L/T/V, or L90M at baseline.

An overriding concern with these types of analyses, attendees argued, is that they arbitrarily define virologic response. Different definitions would likely yield different cutoffs in all three studies. As a result clinicians may be reluctant to consider such cutoffs in practice. One attendee questioned the very concept of response cutoffs when weighing resistance to antiretrovirals. A continuum of variability would more accurately reflect the continuum from fully susceptible virus through fully resistant virus.

Survival with multidrug-resistant virus. A study in 628 Britons with multidrug-resistant (MDR) virus found that 9% died within 24 months of the MDR detection date. Changing therapy and adding more active drugs to the regimen reduced the risk of death compared with maintaining the same regimen.

Deenan Pillay (Mortimer Market Centre, London) and coworkers tracked mortality and viral load changes in people whose virus had at least one major mutation conferring resistance to a nucleoside, a nonnucleoside, and a protease inhibitor. Cohort members had tried a median of 8 antiretrovirals when MDR virus was detected. Estimated probability of death measured 3% 1 year after MDR diagnosis, 8% after 2 years, and 13% after 3 years. Changing therapy and maintaining or increasing the genotypic sensitivity score (GSS) lowered the risk of death by 70% compared with continuing the same regimen.

Discordant genotypic interpretation systems for abacavir and ddI.

A 9-cohort collaboration raised questions over the consistency of genotypic interpretation systems for responses to abacavir (ABC) or didanosine (ddI) after antiretroviral failure. The analysis involved 583 people starting ABC and 400 starting ddI, for the first time. Dominique Costagliola (Pierre and Marie Curie University, Paris) and coworkers evaluated 4 rules-based interpretation systems (ANRS-V12, Detroit Medical Center-3, Stanford HIV RT and PR Sequence Database-8, and Rega-6.3) plus, for ABC only, CHL-4.4, Retrogram-1.6, São Paulo-2, and VGI-5.0.

People starting ABC had a median viral load of 4.4 log copies/mL and an average 1.6-log decrease after 8 weeks. Estimates of the percentage of resistant viruses ranged from 7.3% with the ANRS algorithm to 31.9% with the VGI system. A univariate analysis of the genotypic rules systems tested indicated highly disparate correlations between intermediate or sensitive virus and changes in viral load. The ddI group switched to the drug with a median viral load of 4.2 log copies/mL and had an average 1.8-log decline at week 8. None of the interpretation systems produced results for ddI sensitivity that correlated with viral load change with ddI.

One attendee suggested the widely disparate algorithm results imply that such genotypic rules-based systems are unreliable guides in clinical practice. Costagliola argued that continued work with larger data sets may yield more consistently reliable algorithms.

Part 2. RESISTANCE to new antiretroviral agents

CCR5 antagonist mutations and coreceptor switching.

Studies of two Schering CCR5 antagonists, SCH 351125 and SCH 417690 (vicriviroc), indicate that resistance does not always involve V3 loop mutations. Schering-Plough’s Julie Strizki and collaborators reported results of serial passage studies in different cell lines indicating that resistance emerged slowly with accumulation of multiple mutations in gp120, mostly in the V3 loop. Further analysis, however, showed that one resistant virus had no changes in the V3 loop, suggesting that mutations in other gp120 regions contribute to resistance. Virus resistant to the Schering antagonists proved cross-resistant to other CCR5 antagonists but remained susceptible to AZT, emtricitabine/FTC (Emtriva), and indinavir/IDV (Crixivan).

Conversely, Mike Westby reported that virus resistant to Pfizer’s CCR5 antagonist maraviroc, remained susceptible to three other CCR5 drugs—Schering SCH-C and SCH-D and Glaxo’s 873140 as well as to enfuvirtide/T-20 (Fuzeon).

In Strizki’s study resistant virus generally remained CCR5 tropic but one escape mutant showed signs of dual CCR-5/CXCR4 tropism.

Whether CCR5-resistant virus will jump to X4 receptors in humans—and the clinical consequences of that jump—continue to vex CCR5 antagonist developers. Donald Mosier (Scripps Research Institute, La Jolla) used site-directed mutagenesis to recreate all 32 possible mutation intermediates in the transition from R5 to X4-tropic viruses.

He found only 4 operational mutation pathways among the 120 such routes. Work by Kathryn Kitrinos and Glaxo associates also reduced switch fears: they detected dual tropic virus in one patient during a 10-day monotherapy trial of the experimental CCR5 antagonist 873140 but this was due to the emergence of dual-tropic virus that was already present at low levels at baseline.

TMC114 hangs tight to protease

TMC114, the Tibotec protease inhibitor in phase 3 trials, binds to viral protease more avidly than current PIs, according to results of a new Tibotec study. And once bound, it stays bound. Tibotec’s Inge Dierynck measured the protease binding affinity of TMC114, a dozen closely related compounds, and 7 licensed PIs. TMC114’s binding affinity to wild-type protease proved 3-4 orders of magnitude higher than current PIs, reflecting both a high association rate and low dissociation rate.  Although amprenavir ‘grabbed’ protease as firmly as TMC114, the latter was 1,000 times harder to shake off than all commercially available PIs.

New thymidine analog

An experimental thymidine nucleoside analog, 1-(beta-D-dioxolane) thymine, DOT, proved effective against a host of mutant viruses according to Raymond Shinazi (Emory University, Decatur, Georgia). These included those carrying M184V, the T69S multinucleoside-resistant insert, and multiple thymidine analog mutations (TAMS) including D67N, K70R, T215Y, and K219Q - particularly noteworthy because DOT appears to be a thymidine kinase-dependent nucleoside.  DOT exerted its activity without apparent cellular, mitochondrial, or human bone marrow toxicity and elimination is primarily non-renal, so it seems unlikely to be toxic to the kidney.  The new nucleoside can be detected in cerebrospinal fluid, suggesting that it crosses the blood-brain barrier.

Part 3. MECHANISMS OF HIV DRUG RESISTANCE

TAMs antagonization of K65R

Work by Urvi Parikh (University of Pittsburgh) revealed more detail of the mechanisms by which thymidine-analogue mutations (TAMS) antagonize the K65R mutation, most commonly associated with tenofovir resistance. These mutually exclusive antagonistic resistance mechanisms are of more than academic interest in that they may point the way to more durable nucleoside/nucleotide combinations, as preliminary clinical work at the meeting suggested.  For example, Schlomo Staszewski (J.W. Goethe University, Frankfurt) found that adding AZT to failing regimens of 3 people with K65R but no TAMs regained viral control, even though AZT appeared to be the only active drug in their regimen.

RT level affects susceptibility to nonnucleosides

Higher levels of reverse transcriptase (RT) in viral particles reduce viral susceptibility to nonnucleosides but not to 3TC, according to results of experiments by Zandrea Ambrose (National Cancer Institute, Frederick). Those findings, Ambrose noted, raise the possibility that HIV-1 may be able to soften the antiviral punch of nonnucleosides by pumping up RT levels.

ATP shows two faces in antiviral activity.

Current nucleoside analogs work by ‘chain termination’: HIV RT incorporates the inhibitor into the growing chain of viral DNA during replication, terminating the process. Resistance to nucleoside analogs depends on one of two mechanisms—reduced incorporation of the drug into the viral DNA (as with 3TC) or adenosine triphosphate (ATP)-mediated excision (removal) of the incorporated drug as with AZT). 

Stephen Hughes (National Cancer Institute, Frederick) compared structural and biochemical properties of HIV-1 and HIV-2 and found several differences in the potential ATP binding sites of RT. These could explain why HIV-1 RT binds ATP better and is more likely to develop AZT resistance through this mechanism than HIV-2.

While ATP looks like the bad guy when it comes to AZT resistance, it may trade its black hat for white if Tibotec succeeds in developing a novel class of RT inhibitors: the nucleotide-competing RT inhibitors (NCRTIs). Unlike familiar ‘chain-terminating’ nucleosides the NCRTIs ‘plug’ RT binding sites and ward off incoming nucleotides. And more ATP, Tibotec’s Dirk Jochmans found, appears to make NCRTIs better binders.

Parts 4-5 of the report on the XIV HIV Drug Resistance Workshop will be posted on Friday, July 8, 2005.

07/06/05

Source
A Revell. The HIV Resistance Response Database Initiative (RDI). www.hivrdi.org

Return to Conference Main Page