Programme & Abstracts
CONTINUING MEDICAL EDUCATION
This programme is sponsored by the University of California, San Diego, School of Medicine
Abstracts presented in this meeting are published as part of Antiviral Therapy Volume 8 Issue 3. The page number given for each abstract is correct for Issue 3. The correct citation for an abstract is: Authors. Title. Antiviral Therapy 2003 8:S Page number.
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications i ANTIVIRAL THERAPY
EDITORS-IN-CHIEF SECTION EDITORS
Joep MA Lange EMERGING VIRUSES PRE-CLINICAL HIV National AIDS Therapy Evaluation Centre Clarence J Peters Amalio Telenti Academic Medical Centre University of Texas Medical Branch Division of Infectious Diseases, CHUV University of Amsterdam, Meibergdreef 9 3.146 Keiller Building 1011 Lausanne, Switzerland 1105 AZ Amsterdam, The Netherlands 301 University Boulevard Fax: +41 21 314 1008 Tel: +31 20 566 4479 Galveston, TX 77555-0609, USA Fax: +31 20 691 8821 Fax: +1 409 747 2429 CLINICAL HIV E-mail: [email protected] Peter Reiss HEPATITIS VIRUSES National AIDS Therapy Evaluation Centre Douglas D Richman Stephen Locarnini Academic Medical Centre University of California, San Diego Vicotian Infectious Diseases Reference University of Amsterdam, Meibergdreef 9 Departments of Pathology and Medicine 0679 Laboratory 1105 AZ Amsterdam, The Netherlands 9500 Gilman Drive, La Jolla WHO Collaborating Centres for Virus Reference Fax: +31 20 691 8821 CA 92093-0679, USA and Research, and Biosafety Tel: +1 858 552 7439 10 Wrekyn Street PAPILLOMAVIRUSES Fax: +1 858 552 7445 North Melbourne 3051 William Bonnez E-mail: [email protected] Victoria, Australia Infectious Diseases Unit - Box 689 Fax: +61 3 9342 266 University of Rochester Medical Center 601 Elmwood Avenue HERPESVIRUSES Rochester, NY 14642, USA Richard J Whitley Fax: +1 585 442 9328 University of Alabama at Birmingham School of Medicine, Division of Clinical Virology, RESPIRATORY VIRUSES 616 Children’s Hospital Frederick G Hayden 1600 7th Avenue South, Birmingham Department of Internal Medicine AL 35233-0001, USA Box 473, University of Virginia Fax: +1 205 934 8559 Health Sciences Center, Charlottesville VA 22908, USA Fax: +1 804 924 9065 EDITORIAL BOARD
Lawrence Banks (Trieste, Italy) José M Gatell (Barcelona, Spain) Julio SG Montaner (Vancouver, Canada) Karen K Biron (Research Triangle Park, Lutz Gissmann (Heidelberg, Germany) Luc Perrin (Geneva, Switzerland) North Carolina) Scott M Hammer (Boston, Massachusetts) Thierry Poynard (Paris, France) Charles AB Boucher (Utrecht, The Netherlands) Diane Havlir (San Francisco, California) Stephen Sacks (Vancouver, Canada) Angela Caliendo (Boston, Massachusetts) Martin Hirsch (Boston, Massachusetts) Raymond F Schinazi (Atlanta, Georgia) Ann Collier (Seattle, Washington) Richard Hoetelmans (Amsterdam, Robert T Schooley (Denver, Colorado) Robert B Couch (Houston, Texas) The Netherlands) Stephen E Straus (Bethesda, Maryland) Richard D’Aquila (Charlestown, Massachusetts) John W Huggins (Frederick, Maryland) John J Treanor (Rochester, New York) Janet Englund (Houston, Texas) Brent E Korba (Rockville, Maryland) Stefano Vella (Rome, Italy) Delia Enria (Pergamino, Argentina) Anna Lok (Ann Arbour, Michigan) Teresa Wright (San Francisco, California) Joseph J Eron (Chapel Hill, North Carolina) Dennis J McCance (Rochester, New York) Patrick Yeni (Paris, France)
EDITORIAL OFFICE
MANAGING EDITOR International Medical Press Tel: +44 20 7398 0700 − Stephen Cameron 2 4 Idol Lane Fax:+44 20 7398 0701 London EC3R 5DD E-mail: [email protected] PRODUCTION EDITOR UK http://www.intmedpress.com Phil Garner
ii CONTENTS
Workshop supporters iv
General information v
Organizing and Scientific Committees vi
Programme vii
Abstract contents xi
Session 1 Resistance to New Antiretroviral Agents S3
Session 2 Mechanisms of HIV Drug Resistance S31
Session 3 HIV Pathogenesis, Fitness and Resistance S61
Session 4 New Resistance Technologies and Interpretations S93
Session 5 Epidemiology S127
Session 6 Clinical Implications of Resistance S147
Author index S171
Delegate affiliation S177
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications iii Support for the XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications has been provided by the following organizations
Principal Supporter:
Major Supporters:
Supporters: Contributors:
Abbott Laboratories Achillion Pharmaceuticals, Inc Bayer Diagnostics Celera Diagnostics Boehringer Ingelheim International Quest Diagnostics Shire Pharmaceuticals Group Tibotec Virco
General Contributions:
Pharmasset, Inc Triangle Pharmaceuticals
iv GENERAL INFORMATION
Workshop objectives Provide an update on resistance to new antiretroviral agents Review mechanisms of HIV drug resistance Present topics specific to HIV pathogenesis, fitness and resistance Expand the scope of the clinical implications of resistance Present new development in epidemiology Introduce new resistance technologies and interpretations
CME Accreditation This activity has been planned and implemented in accordance with the Essential Areas and Policies of the Accreditation Council for Continuing Medical Education ("ACCME") through the joint sponsorship of the University of California, San Diego School of Medicine and Informed Horizons, LLC. The University of California, San Diego School of Medicine is accredited by the ACCME to provide continuing medical education for physicians.
The University of California, San Diego School of Medicine designates this educational activity for a maximum of 18 CME hours in category 1 credit towards the AMA Physician’s Recognition Award. Each physician should claim only those hours of credit that he/she actually spent in the educational activity.
Date and venue The meeting is held at Hacienda del Mar, Cabo San Lucas, Mexico on 10-14 June, 2003
Language The official language of the workshop is English
Badge policy All registered delgates are provided with an identity badge. Please wear it at all times to ensure admission to the meeting.
Certificate of attendance Delegates who require a certificate of attendance should inform the workshop Secretariat. A certificate will be forwarded by e-mail after the event.
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications v ORGANIZING COMMITTEE Charles Boucher, Chair University Hospital Utrecht, the Netherlands John Mellors, Chair University of Pittsburgh, USA Brendan Larder Cambridge, UK Douglas Richman University of California San Diego/VA Medical Center, USA
SCIENTIFIC COMMITTEE Lee Bacheler Tibotec-Virco, USA Nick Cammack Roche Palo Alto, USA François Clavel Inserm U552, France Bonaventura Clotet Fundacio IRSI Caixa, Spain John Coffin National Cancer Institute-Frederick, USA Richard Colonno Bristol-Myers Squibb Company, USA Richard D'Aquila Vanderbilt University Medical Center, USA Victor De Gruttola Harvard School of Public Health, USA John Erickson Sequoia Pharmaceuticals, USA Daria Hazuda Merck Research Laboratories, USA Victoria Johnson University of Alabama at Birmingham, USA Dale Kempf Abbott Laboratories, USA Joerg-Peter Kleim GlaxoSmithKline, UK Daniel Kuritzkes Brigham and Women’s Hospital, USA Douglas Mayers Boehringer Ingelheim Pharmaceuticals, Inc., USA Thomas Merigan Stanford University School of Medicine, USA Luc Perrin Geneva University Hospital, Switzerland Rick Pesano Pfizer Global Research and Development, USA Christos Petropoulos ViroLogic, Inc, USA Deenan Pillay Royal Free and University College Medical School, UK Jonathan Schapiro Stanford University, USA Raymond Schinazi Emory University/VA Medical Center, USA Ronald Swanstrom University of North Carolina Center for AIDS Research, USA Mark Wainberg McGill University AIDS Centre, Canada Patrick Yeni Hopital Bichat, France
ORGANIZING SECRETARIAT Informed Horizons, LLC 1860 Montreal Road, Suite 212 Tucker GA 30084, USA Telephone: +1 770 946 3480 Facsimile: +1 770 897 9639 Email: [email protected] Website: www.informedhorizons.com
vi FINAL PROGRAMME TIME TITLE PRESENTING AUTHOR ABSTRACT Tuesday, June 10, 2003 Opening Welcome Reception
Wednesday, June 11, 2003 MORNING SESSIONS 08:00 Welcome Charles Boucher, University Hospital Utrecht, the Netherlands 08:15 Mechanisms of HIV-1 diversity Bette Korber, The University of Texas 1 Medical Branch at Galveston, USA RESISTANCE TO NEW ANTIRETROVIRAL AGENTS Chairs: Nick Cammack Roche Palo Alto, USA Brendan Larder Cambridge, UK
09:00 Mappicine inhibitors of HIV-1 reverse Michael Parniak, University of Pittsburgh, USA 11 transcriptase-associated ribonuclease H 09:15 The Identification of active site mutations Daria Hazuda, Merck, USA 10 that confer resistance to structurally diverse inhibitors of HIV-1 integrase strand transfer supports a general mechanism of phosphotransferase inhibition 09:30 Elvucitabine: potent antiviral activity Lisa Dunkle, Achillion Pharmaceuticals, Inc., USA 2 demonstrated in multidrug-resistant HIV infection 09:45 TMC114 binds within the substrate envelope Celia Schiffer, University of Massachusetts 16 of HIV-1 protease, which could account for its Medical School, USA efficacy against multi-protease inhibitor- resistant virus 10:00 Characterization of the impact of genotype, Doug Mayers, Boehringer Ingelheim, USA 12 phenotype and inhibitory quotient on antiviral activity of tipranavir in highly treatment-experienced patients 10:15 Panel Discussion 10:45 BREAK MECHANISMS OF HIV DRUG RESISTANCE Chairs: Dale Kempf Abbott Laboratories, USA John Mellors University of Pittsburgh, USA
11:15 Crystal structure of a multidrug-resistant Ladislau Kovari, Wayne State University, USA 45 HIV-1 protease clinical isolate reveals an expanded active site cavity and represents a novel target for the design of protease inhibitors 11:30 Structural correlates of broad-spectrum Abelardo Silva, Sequoia Pharmaceuticals, Inc., USA 46 activity for a resistance-repellent HIV protease inhibitor
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications vii 11:45 Co-evolution of the nucleocapsid-p1 Celia Schiffer, University of Massachusetts 47 cleavage site with the V82A mutation in Medical School, USA HIV-1 protease preserves substrate recognition 12:00 Mechanisms involved in zidovudine Matthias Götte, McGill University, Canada 28 hypersusceptibility in the presence of foscarnet resistance-conferring mutations 12:15 The removal of chain-terminating nucleoside Anthony Smith, University of Miami 38 analogues by HIV-1 reverse transcriptase School of Medicine, USA utilizing intracellular substrates 12:30 Panel Discussion 13:00 LUNCH AFTERNOON SESSIONS 14:00 Poster Viewing: Resistance to New Antiretroviral Agents Mechanisms of HIV Drug Resistance Discussants: Raymond Schinazi Emory University/VA Medical Center, USA Dale Kempf Abbott Laboratories, USA 15:30 Roundtable Facilitated Discussion: Resistance to New Antiretroviral Agents 16:30 Roundtable Facilitated Discussion: Mechanisms of HIV Drug Resistance Thursday, June 12, 2003 MORNING SESSIONS HIV PATHOGENESIS, FITNESS AND RESISTANCE Chairs: John Coffin National Cancer Institute-Frederick, USA Charles Boucher University Hospital Utrecht, the Netherlands 08:00 Immune-mediated suppression of virulent Koen Van Rompay, California National 70 simian immunodeficiency virus induced by Primate Research Center, USA tenofovir treatment 08:15 CCR5 cell surface density and HIV-1 envelope Michael Miller, Merck Research Laboratories, USA 61 sequences govern antiretroviral potency of CCR5 antagonists 08:30 Virus characteristics predict viraemia control Huldrych Günthard, University Hospital 56 after cessation of antiretroviral therapy Zurich, Switzerland 08:45 Co and super-infection: persistent Luc Perrin, Hopital Canonal Universitaire, 63 replication of both HIV-1 strains? Switzerland 09:00 Population Genetics in HIV-1 Superinfection Sarah Palmer, HIV Drug Resistance Program, 62 NCI, NIH 09:15 Panel Discussion 09:45 BREAK NEW RESISTANCE TECHNOLOGIES AND INTERPRETATIONS Chairs: François Clavel Inserm U552, France Douglas Richman University of California San Diego/ VA Medical Center, USA 10:15 The impact of minor populations of Akhter Molla, Abbott Laboratories, USA 85 wild-type HIV on the replication capacity and phenotype of mutant variants in a single-cycle HIV resistance assay
viii 10:30 Examination of wide variation in replication Mark Segal, Center for Bioinformatics 94 capacity of wild-type HIV-1: analysis of & Molecular Biostatistic Health, USA genotype–phenotype association via tree- structured methods 10:45 Panel Discussion 11:00 Poster Viewing: HIV Pathogenesis, Fitness and Resistance New Resistance Technologies and Interpretations 12:30 LUNCH AFTERNOON SESSIONS Discussants: Christos Petropoulos ViroLogic, Inc., USA Richard D’Aquila Vanderbilt University Medical Center, USA 13:30 Roundtable Facilitated Discussion: HIV Pathogenesis, Fitness and Resistance 14:30 Roundtable Facilitated Discussion: New Resistance Technologies and Interpretations Friday, June 13, 2003 MORNING SESSIONS 08:00 How good is the evidence that HIV specific CTL Rodney Phillips, John Radcliffe Hospital, UK 114 can protect against and control infection? EPIDEMIOLOGY Chairs: Victoria Johnson University of Alabama at Birmingham, USA Charles Boucher University Hospital Utrecht, the Netherlands 08:45 Persistence of transmitted drug-resistant Susan Little, University of California, 115 virus among subjects with primary HIV San Diego, USA infection deferring antiretroviral therapy 09:00 Prevalence of transmitted drug resistance Annemarie Wensing, University Medical 117 in Europe is largely influenced by the Center Utrecht, the Netherlands presence of non-B sequences: analysis of 1400 patients from 16 countries: the CATCH-Study 09:15 Patients with antiretroviral-resistant HIV Michael Kozal, Yale University School of 116 infections engaging in high-risk transmission Medicine, USA behaviour 09:30 Panel Discussion 09:45 BREAK CLINICAL IMPLICATIONS OF RESISTANCE Chairs: Jonathan Schapiro Stanford University, USA Brendan Larder Cambridge, UK 10:15 Low frequency non-nucleoside reverse John Mellors, University of Pittsburgh, USA 134 transcriptase inhibitor (NNRTI)-resistant variants contribute to failure of efavirenz-containing regimens in NNRTI-experienced patients with negative standard genotypes for NNRTI mutations 10:30 Short duration, single drug discontinuation to Frank Maldarelli, HIV Drug Resistance 133 assess the activity of individual drugs in Program NCI, USA patients failing antiretroviral therapy 10:45 K65R: a multi-nucleoside resistance mutation Urvi Parikh, University of Pittsburgh, USA 136 of low but increasing frequency 11:00 Characterization of resistance mutation Michael Miller, Gilead Sciences, USA 135 patterns emerging over 2 years during first-line antiretroviral treatment with tenofovir DF or stavudine in combination with lamivudine and efavirenz
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications ix 11:15 Which nucleoside and nucleotide backbone Alan Winston, Chelsea and Westminster 137 combinations select for the K65R mutation in Hospital, UK HIV-1 reverse transcriptase? 11:30 Panel Discussion 12:00 LUNCH AFTERNOON SESSIONS 13:30 Poster Viewing: Epidemiology Clinical Implications of Resistance Discussants: Deenan Pillay Royal Free and University College Medical School, UK Daniel Kuritzkes Brigham and Women's Hospital, USA 15:00 Roundtable Facilitated Discussion: Epidemiology 16:00 Roundtable Facilitated Discussion: Clinical Implications of Resistance 17:00 Closing Remarks John Mellors, University of Pittsburgh, USA 19:30 Awards Dinner
x CONTENTS
page session title and author abstract
S1 Plenary Abstract Mechanisms of HIV-1 diversity 1 B Korber
S3 Session 1: Resistance to New Antiretroviral Agents
S5 Elvucitabine: potent antiviral activity demonstrated in 2 multidrug-resistant HIV infection LM Dunkle
S6 Antiviral activity of SPD754 against clinical isolates of 3 HIV-1 resistant to other nucleoside reverse transcriptase inhibitors RC Bethell
S7 Synthesis and anti-HIV activity of enantiomerically pure D-FDOC 4 DC Liotta
S8 In vitro induction of HIV variants with reduced susceptibility to 5 elvucitabine (ACH-126,443, β-L-Fd4C) WG Rice
S9 Antiviral activity of the nucleoside Reverset following single 6 oral doses in HIV-1-infected patients L Stuyver
S10 Characterization of baseline and treatment-emergent 7 resistance mutations following 1 year of therapy on an entirely once a day regimen including emtricitabine K Borroto-Esoda
S11 Antiviral activity of TMC125, a potent next-generation 8 non-nucleoside reverse transcriptase inhibitor (NNRTI), against >5000 recombinant clinical isolates exhibiting a wide range of NNRTI resistance M-P de Béthune
S12 Characterization of resistance before and after short-term 9 therapy with TMC125 in patients with documented non-nucleoside reverse transcriptase inhibitor resistance J Vingerhoets
S13 The identification of active site mutations that confer 10 resistance to structurally diverse inhibitors of HIV-1 integrase strand transfer supports a general mechanism of phosphotransferase inhibition DJ Hazuda
S14 Mappicine inhibitors of HIV-1 reverse transcriptase-associated 11 ribonuclease H MA Parniak
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications xi S15 Characterization of the impact of genotype, phenotype, and 12 inhibitory quotient on antiviral activity of tipranavir in highly treatment-experienced patient S McCallister
S16 Characterization of treatment-emergent resistance mutations 13 in two Phase II studies of tipranavir D Hall
S17 Characterization of HIV-1 showing decreased susceptibility to 14 tipranavir and their inhibition by tipranavir-containing drug mixtures L Doyon
S18 TMC114, a potent next-generation protease inhibitor: 15 characterization of antiviral activity in multiple protease inhibitor-experienced patients participating in a Phase IIa study D De Meyer
S19 TMC114 binds within the substrate envelope of HIV-1 16 protease, which could account for its efficacy against multi-protease inhibitor-resistant virus CA Schiffer
S20 Antiviral activity of TMC114, a potent next-generation 17 protease inhibitor, against >4000 recent recombinant clinical isolates exhibiting a wide range of (protease inhibitor) resistance profiles M-P de Béthune
S21 In vitro cross resistance profile of RO033-4649 against a 18 panel of multiply-substituted protease inhibitor-resistant viruses: role of common protease resistance mutations G Heilek-Snyder
S22 HIV clinical isolates containing mutations representative of 19 those selected after first-line failure with unboosted GW433908 remain sensitive to other protease inhibitors L Ross
S23 Antiviral activity of P-1946, a novel anti-HIV protease inhibitor 20 G Sévigny
S24 Baseline and on-treatment gp41 genotype and susceptibility to 21 enfuvirtide (ENF) and T-1249 in a 10-day study of T-1249 in patients failing an ENF-containing regimen (T1249-102) GD Miralles
S25 Determinants of susceptibility to enfuvirtide map to 22 gp41 in enfuvirtide-naive HIV-1 SA Stanfield-Oakley
xii S26 Sensitivity of Env-gene recombinant viruses derived 23 from antiretroviral drug-sensitive and -resistant HIV-1 clinical isolates to the novel CCR5 antagonist, UK-427,857 M Westby
S27 ADS-J1, a non-peptidic low molecular weight HIV fusion 24 inhibitor targeting gp41, with no cross-drug resistance with peptidic HIV fusion inhibitors T-20 and C-34, and HIV binding inhibitors JA Esté
S28 Human β-defensins inhibit HIV-1 replication in vitro 25 ME Quiñones-Mateu
S29 In vitro resistance development of human immunodeficiency 26 virus type 1 towards mannose-specific plant lectins K Van Laethem
S30 Viral resistance against a candidate HIV microbicide 27 Z Ambrose
S31 Session 2: Mechanisms of HIV Drug Resistance
S33 Mechanisms involved in zidovudine hypersusceptibility in 28 the presence of foscarnet resistance-conferring mutations M Götte
S34 HIV-1 reverse transcriptase mutations that suppress 29 zidovudine resistance also increase in vitro susceptibility to tenofovir, but not stavudine NT Parkin
S35 The ∆67 complex of mutations enhances the ability of 30 HIV-1 reverse transcriptase to excise zidovudine, stavudine and PMPA from blocked primers SH Hughes
S36 The 3′-azido group is not the primary structural determinant 31 for the excision phenotype correlated with HIV-1 resistance to AZT N Sluis-Cremer
S37 Evolution of amino acid 215 in HIV-1 reverse transcriptase in 32 response to intermittent drug selection C Chappey
S38 Molecular mechanisms of resistance to tenofovir by HIV-1 33 reverse transcriptase dontaining a di-serine insertion after residue 69 and multiple thymidine analogue-associated mutations MD Miller
S39 Drug resistance and viral fitness at the molecular level: the 34 case of tenofovir B Canard
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications xiii S40 Molecular mechanism for the mutual exclusion of K65R and 35 L74V substitutions in HIV-1 reverse transcriptase-mediated dideoxynucleoside resistance J Deval
S41 Mechanism of anti-HIV activity of dioxolane nucleosides 36 against lamivudine-resistant HIV-1 reverse transcriptase-molecular modelling approach CK Chu
S42 Prevalence and quantitative phenotypic resistance patterns 37 of specific nucleoside analogue mutation combinations and of mutations 44 and 118 in reverse transcriptase in a large dataset of recent HIV-1 clinical isolates M Van Houtte
S43 Removal of chain-terminating nucleoside analogues by 38 HIV-1 reverse transcriptase utilizing intracellular substrates AJ Smith
S44 Colinearity of reverse transcriptase inhibitor resistance 39 mutations detected by population-based sequencing MJ Gonzales
S45 Identification of the minimal conserved structure of the 40 HIV reverse transcriptase under the presence and absence of drug pressure F Ceccherini-Silberstein
S46 Identification of a clinical reverse transcriptase backbone 41 that improves replication of a non-nucleoside reverse transcriptase inhibitor-resistant mutant by increasing the rate of polymerization LM Demeter
S47 Non-nucleoside reverse transcriptase inhibitor hypersusceptibility 42 can be demonstrated in multicycle phenotype assays and in inhibition assays of purified HIV-1 reverse transcriptases NS Shulman
S48 Genetic correlates of phenotypic hypersusceptibility to 43 efavirenz among 446 baseline isolates from five ACTG studies NS Shulman
S49 Rare 1 and 2 amino acid insertions in the non-nucleoside 44 reverse transcriptase inhibitor (NNRTI) binding pocket of HIV-1 reverse transcriptase affect NNRTI susceptibility MA Winters
S50 Crystal structure of a multidrug-resistant HIV-1 protease 45 clinical isolate reveals an expanded active site cavity and represents a novel target for the design of protease inhibitors LC Kovari
xiv S51 Structural correlates of broad-spectrum activity for a 46 resistance-repellent HIV protease inhibitor AM Silva
S52 Co-evolution of the nucleocapsid-p1 cleavage site with 47 the V82A mutation in HIV-1 protease preserves substrate recognition CA Schiffer
S53 Preliminary characterization of a newly described protease 48 substrate cleft mutation at position 23 RW Shafer
S54 Emergence of a novel lopinavir resistance mutation at 49 codon 47 correlates with ARV utilization RM Kagan
S55 Parameters driving the selection of nelfinavir-resistant HIV-1 50 variants F Mammano
S56 I84A and I84C mutations in protease confer high-level 51 resistance to protease inhibitors and impair replication capacity H Mo
S57 The HIV-1 protease mutation K55R is associated with the 52 presence of the M46I/L mutation E Morgan
S58 Nucleic acid differences between HIV-1 non-B and B reverse 53 transcriptase and protease sequences at drug resistance positions R Kantor
S59 Search for polymorphic sites in R5 tropic HIV-1 Env and 54 enfuvirtide drug susceptibility in baseline isolates from TORO 1 & TORO 2 C Su
S60 Subgroup analysis of baseline susceptibility and early 55 virological response to enfuvirtide in the combined TORO studies P Sista
S61 Session 3: HIV Pathogenesis, Fitness and Resistance
S63 Virus characteristics predict viraemia control after 56 cessation of antiretroviral therapy H Günthard
S64 Distinct patterns of the virological and immunological 57 response to antiretroviral therapy interpreted in terms of the dynamics of immune activation and HIV transmission Z Grossman
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications xv S65 Both stable and decaying HIV plasma viral loads are 58 observed in subjects with sustained suppression of viraemia <50 copies/ml and are uninfluenced by influenza vaccination EP Coakley
S66 HIV DNA as a predictor of residual viraemia in patients 59 treated with tenofovir+lamivudine+efavirenz or stavudine+lamivudine+efavirenz DV Havlir
S67 Stable, persistent viraemia in patients with plasma HIV-1 60 RNA suppressed to less than 75 copies/ml on potent antiretroviral therapy A Wiegand
S68 CCR5 cell surface density and HIV-1 envelope sequences govern 61 antiretroviral potency of CCR5 antagonists MD Miller
S69 Population genetics in HIV-1 superinfection 62 S Palmer
S70 Co- and super-infection: persistent replication of both HIV-1 63 strains? L Perrin
S71 HIV viral sex: inbreeding, recombination, drug resistance 64 and clinical outcome C Fraser
S72 HIV-1 genetics of intermittent low-level viraemia 65 NH Tobin
S73 Limited genotypic and phenotypic evolution after 66 interruption of a single therapeutic drug class in patients with multidrug-resistant HIV SG Deeks
S74 Clonal analysis of HIV-1 variants in proviral DNA during 67 treatment interruption in patients with multiple therapy failures B Masquelier
S75 Vaginal HIV-1 shows distinct drug resistance mutation 68 patterns compared to plasma HIV-1 and remain M-tropic despite advanced disease G Tirado
S76 Effects of failing antiretroviral therapy on the composition 69 of HIV-1 in CSF and blood JK Wong
S77 Immune-mediated suppression of virulent simian 70 immunodeficiency virus induced by tenofovir treatment KKA Van Rompay
xvi S78 A potential role for CD63 in CCR5-mediated HIV-1 infection 71 of macrophages WA O’Brien
S79 Mechanisms underlying a sustained CD4 T cell recovery despite 72 the emergence of resistance in antiretroviral-experienced patients on prolonged enfuvirtide treatment B Rodés
S80 Wide variation in pro/pol replication capacity in recently 73 transmitted HIV-1 is conferred in part by protease inhibitor resistance mutations JD Barbour
S81 Modulation of replication capacity in drug sensitive clade B 74 HIV-1 by protease and the C-terminal end of gag M Bates
S82 Drug-resistant phenotype is associated with decreased 75 in vivo T-cell activation independent of changes in viral replication among patients discontinuing antiretroviral therapy PW Hunt
S83 Effects of amino acid and synonymous polymorphisms in 76 HIV-1 protease on viral fitness AJ Leigh Brown
S84 Influence of gag and pol regions on replication efficiency of a 77 multidrug-resistant HIV-1 variant: a systematic assessment V Simon
S85 Rapid flux in non-nucleoside reverse transcriptase inhibitor 78 resistance mutations among subtype C HIV-1-infected women after single dose nevirapine D Katzenstein
S86 HIV-1 variants with diverse nevirapine resistance mutations 79 emerge rapidly after single-dose nevirapine: HIVNET 012 SH Eshleman
S87 Persistence of multidrug-resistant HIV-1 without any 80 antiretroviral treatment 2 years after sexual transmission C Delaugerre
S88 Impact of transmission of drug-resistant HIV viruses on 81 viral load, CD4 counts and CD4 decline in recent seroconverters A Corral
S89 Transmitted HIV-1 carrying D67N or K219Q evolve rapidly to 82 zidovudine resistance in vitro and show a high replicative fitness in the presence of zidovudine JG García-Lerma
S90 Male genital tract compartmentalization and transmission 83 of 215L revertant DM Smith
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications xvii S91 Analysis of virion morphology and assembly process in protease 84 inhibitor-resistant HIV-1 W Sugiura
S93 Session 4: New Resistance Technologies and Interpretations
S95 The impact of minor populations of wild-type HIV on the 85 replication capacity and phenotype of mutant variants in a single-cycle HIV resistance assay A Molla
S96 Comparison of single-genome sequencing with standard 86 genotype analysis for detection of HIV-1 drug-resistance mutations M Kearney
S97 Drug-resistant mutants of HIV-1 persist at higher concentrations 87 in peripheral blood mononuclear cells compared to plasma and are detected at a higher rate by OLA compared to consensus sequencing LM Frenkel
S98 Simultaneous assessment of association between two or more 88 HIV-1 drug susceptibility phenotypes and genotype G DiRienzo
S99 Using GENE-CODE technology to detect early emerging 89 populations of drug-resistant human immunodeficiency virus Type 1 MJ Moser
S100 Evidence for preferential genotyping of minority HIV populations 90 due to primer mismatching during PCR-based amplification K Morales-Lopetegi
S101 Quality control assessment for HIV-1 genotypic antiretroviral 91 testing (2002) F Brun-Vézinet
S102 Updated, blinded, multicentre comparison of the 92 sensitivity of different technologies to detect and quantify a minor drug-resistant HIV-1 variant E Halvas
S103 Detection of rare drug-resistant HIV-1 reverse transcriptase 93 variants using a sensitive phenotypic assay DV Nissley
S104 Examination of wide variation in replication capacity of 94 wild-type HIV-1: analysis of genotype–phenotype association via tree-structured methods MR Segal
S105 Reverse transcription fitness assay by quantitative determination 95 of intermediate HIV-1 reverse transcription cDNA products in PBMCs M Nijhuis
xviii S106 Quantitative prediction of HIV drug susceptibility from viral 96 genotype through linear regression modelling H Van Marck
S107 Tree models for the evolution of drug resistance 97 N Beerenwinkel
S108 Predictive value of drug resistance interpretation systems and 98 therapeutic drug monitoring for virological therapy response to salvage therapy H Walter
S109 Predicting phenotype from genotype: a comparison of 99 statistical methods AS Foulkes
S110 Does HIV-1 subtype influence the results of drug resistance 100 interpretation algorithms? JC Schmit
S111 Comparison of five interpretation algorithms for the prediction 101 of protease inhibitor susceptibility in HIV-1 non-B subtypes J Snoeck
S112 A neural network model using clinical cohort data accurately 102 predicts virological response and identifies regimens with increased probability of success in treatment failures BA Larder
S113 Correlation between Interpreted Genotype and Virtual Phenotype 103 for predicting drug resistance in HIV-1 O Gallego
S114 Extending the range of measurable resistance to efavirenz: 104 the effect of combinations of mutations and baseline polymorphisms F Clavel
S115 Interpreting resistance to lopinavir/ritonavir in HIV-1 105 subtype C patients Z Grossman
S116 Exploration of methodology for estimating upper clinical 106 breakpoints for lopinavir/ritonavir by analysis of the emergence of resistance during virological failure in experienced patients RC Stevens
S117 Clinically relevant interpretation of genotype for resistance 107 to ritonavir (100 mg twice daily) plus saquinavir (800 mg twice daily) in HIV-1-infected protease inhibitor-experienced patients V Calvez
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications xix S118 Exploring the effects of adherence on resistance: use of local 108 linear regression to reveal relationships between adherence and resistance in antiretroviral-naive patients treated with lopinavir/ritonavir or nelfinavir M King
S119 Genotypic and pharmacological determinants of the 109 virological response to tenofovir in nucleoside reverse transcriptase inhibitor-experienced patients B Masquelier
S120 Mutations at reverse transcriptase codon 103: phenotypic 110 resistance to non-nucleoside reverse transcriptase inhibitor and clinical correlates PR Harrigan
S121 Genotypic associations with non-nucleoside reverse transcriptase 111 inhibitor susceptibility in circulating recombinant forms of HIV-1 strains in North and South America A Florance
S122 Long-term clinical and viro-immunological outcomes of 112 Argenta, a randomized trial on the usefulness of HIV-1 drug resistance genotyping and expert advice A De Luca
S123 Clinician knowledge and attitudes towards genotypic testing 113 C Salama
S125 Plenary Abstract How good is the evidence that HIV specific CTL can protect 114 against and control infection? R Phillips
S127 Session 5: Epidemiology
S129 Persistence of transmitted drug-resistant virus among subjects 115 with primary HIV infection deferring antiretroviral therapy SJ Little
S130 Patients with antiretroviral-resistant HIV infections engaging 116 in high-risk transmission behaviour M Kozal
S131 Prevalence of transmitted drug resistance in Europe is largely 117 influenced by the presence of non-B sequences: analysis of 1400 patients from 16 countries: the CATCH-Study AMJ Wensing
S132 Trends in genotypic and phenotypic resistance among clinical 118 samples submitted for routine HIV resistance analysis L Bacheler
S133 Prevalence of mutations associated with antiretroviral drug 119 resistance among men and women newly diagnosed with HIV in 10 US cities, 1997–2001 DE Bennett
xx S134 Declining nucleoside reverse transcriptase inhibitor primary 120 resistance in San Francisco, 2000–2002 RM Grant
S135 The Canadian HIV Strain and Drug Resistance Program – a 121 population-based effort to enhance HIV surveillance GC Jayaraman
S136 Drug resistance prevalence declines in recently infected 122 subjects having sex with men but not in those using drug injections: results from the Montreal Primary HIV-Infection Cohort JP Routy
S137 French National Sentinel Survey of antiretroviral resistance 123 in patients with HIV-1 primary infection and in antiretroviral- naive chronically infected patients in 2001–2002 ML Chaix
S138 The UK HIV Drug Resistance Database: development and use 124 for national surveillance D Pillay
S139 HIV-1 drug resistance mutations in ARV-treated patients 125 from southern Brazil MA Soares
S140 L89I/V: a novel mutation selected by protease inhibitor therapy 126 in subtype G, but not in Subtype B-infected patients R Camacho
S141 Viral load decline and selection of drug resistance mutations in 127 individuals with HIV-1 group O viruses undergoing antiretroviral therapy C de Mendoza
S142 Synonomous genetic changes within subtype F HIV-1 128 may influence mutational routes to drug resistance A Tanuri
S143 Decreased rates of transmission of drug-resistant HIV-1 strains 129 containing the m184v mutation in reverse transcriptase D Turner
S144 Evidence for a different transmission efficiency of viruses with 130 distinct drug-resistant genotypes C de Mendoza
S145 Modelling the sexual synergy between HSV-2 and HIV-1 131 epidemics: implications for transmitted HIV resistance SM Blower
S146 A systematic approach that identifies 11 new mutations as 132 indicators of transmission of resistance DAMC van de Vijver
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications xxi S147 Session 6: Clinical Implications of Resistance
S149 Short duration, single drug discontinuation to assess the 133 activity of individual drugs in patients failing antiretroviral therapy F Maldarelli
S150 Low frequency non-nucleoside reverse transcriptase inhibitor 134 (NNRTI)-resistant variants contribute to failure of efavirenz- containing regimens in NNRTI-experienced patients with negative standard genotypes for NNRTI mutations J Mellors
S151 Characterization of resistance mutation patterns emerging 135 over 2 years during first-line antiretroviral treatment with tenofovir DF or stavudine in combination with lamivudine and efavirenz MD Miller
S152 K65R: a multi-nucleoside resistance mutation of low but 136 increasing frequency U Parikh
S153 Which nucleoside and nucleotide backbone combinations 137 select for the K65R mutation in HIV-1 reverse transcriptase? A Winston
S154 Thymidine analogue mutations emergence in antiretroviral- 138 naive HIV-1 patients on triple therapy including either zidovudine or stavudine L Bocket
S155 Implications of lamivudine use and the M184V mutation 139 in V118I-related multidrug resistance BG Brenner
S156 Antiviral activity of lamivudine in persons infected with 140 HIV-1 that has M184V and multiple thymidine analogue mutations TB Campbell
S157 The impact of the HIV reverse transcriptase M184V mutation 141 in combination with single thymidine analogue mutations on nucleoside reverse transcriptase inhibitor resistance in clinical samples from a large patient database L Ross
S158 Determination of capravirine inhibitory quotient in HIV-infected, 142 non-nucleoside reverse transcriptase inhibitor treatment- experienced patients AK Patick
S159 Resistance genotypes in patients failing nevirapine: co-existence 143 of majority viral populations expressing Y181C and minority populations expressing K103N AJ Hance
xxii S160 Mutations in HIV-1 reverse transcriptase associated with 144 hypersusceptibility to non-nucleoside reverse transcriptase inhibitors: impact on virological response to efavirenz-based salvage therapy V Tozzi
S161 Polymorphism and drug-selected mutations in the reverse 145 transcriptase gene of HIV-2 from patients living in southern France P Colson
S162 High frequency of selection of the Q151M mutation in HIV-2- 146 infected patients receiving nucleoside reverse transcriptase inhibitor-containing regimen D Descamps
S163 Resistance analysis of the benefit of a treatment interruption 147 before salvage therapy including seven to nine antiretroviral drugs (GIGHAART ANRS 097) C Delaugerre
S164 Baseline resistance and predictors of virological response in the 148 CPCRA 064 study: a randomized trial of structured treatment interruption for patients with multidrug-resistant HIV J Lawrence
S165 A 1-year randomized controlled trial of genotypic versus 149 genotypic plus phenotypic resistance testing to guide antiretroviral therapy (the ERA Trial) C Loveday
S166 Resistance mutations during intermittent antiretroviral 150 therapy: does PBMC genotyping before treatment interruption help predict their occurrence? L Palmisano
S167 Accumulation of antiretroviral resistance in treatment- 151 experienced patients: the impact of medication adherence R Haubrich
S168 The effect of number of mutations and of drug-class sparing on 152 virological response to salvage genotype-guided antiretroviral therapy MP Trotta
S169 Genotypic predictors of response to lopinavir/ritonavir in clinical 153 practice A Zolopa
S170 Optimizing 908/r in protease inhibitor-experienced patients: 154 a retrospective analysis of virological response based on baseline genotypic mutations and after 8 weeks of therapy RC Elston
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications xxiii
Los Cabos, Mexico, 10–14 June 2003
PLENARY ABSTRACT 1
Mechanisms of HIV-1 diversity
B Korber
Los Alamos National Laboratory, Los Alamos, N. Mex., USA; The Santa Fe Institute, Santa Fe, NM, USA
HIV has evolved many ways to vary, particularly in the envelope gene where immune escape for neutralizing antibody responses and features of cellular tropism are determined. The rate of base substitution is extraordi- narily high, and modelling the evolution of these changes enables phylogenetic reconstructions of HIV viruses. But HIV has other mechanisms of variation that can result in immune evasion including recombi- nation, shifting glycosylation patterns, insertions and deletions, and action at a distance through conforma- tional change. Each of these mechanisms of diversity will be reviewed in the context of what can be learned form the database, with particular focus on the immunological impact of HIV variation, studies where recombination was not seen (surprising in the context of some of the recent studies documenting high levels of recombination), and patterns of glycosylation site variation. New tools available at the Los Alamos HIV database will be discussed.
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S1
SESSION 1 Resistance to New Antiretroviral Agents
Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 2 in all subjects. Potent anti-HIV activity was observed in both ELV groups: mean declines at 28 days of –0.67 and –0.78 log copies/ml in the 50 and 100 mg Elvucitabine: potent antiviral 10 groups, respectively, as compared with an increase of activity demonstrated in multidrug- +0.01 log copies/ml in the 3TC group (P<0.0001 for 10 resistant HIV infection both comparisons). Baseline genotypes were not clear- ly correlated with antiviral activity. Myelosuppression LM Dunkle1, JC Gathe2, DE Pedevillano1, was the only study-related adverse event observed and HG Robison1, WG Rice1, JC Pottage Jr1 and the was generally associated with the ELV 100 mg dose ACH-006 Study Team group.
1 Achillion Pharmaceuticals, Inc., New Haven, Conn., USA; and 2 CONCLUSIONS: ELV demonstrated potent anti-HIV Houston, Tex., USA activity in patients with multidrug-resistant HIV, com- parable or superior to other potential ‘salvage’ thera- BACKGROUND: Elvucitabine (ELV) is an L-nucleo- pies, with a convenient single daily oral dose. Doses of side analogue with potent in vitro anti-HIV activity, 50 and 100 mg/day were similarly potent; the safety particularly against strains with resistance mutations profile of 50 mg daily was more desirable than 100 to numerous other drugs, including nucleosides, all mg. Further study with doses up to 50 mg daily is war- non-nucleosides (NNRTI) and HIV protease inhibitors ranted to identify the optimum dose for long-term clin- (PI). ELV exhibits reduced potency against strains har- ical development of ELV in this population with limit- bouring the M184V mutation [lamivudine (3TC)-resis- ed treatment options. tant], but the inhibitory concentration (IC ) remains 50 within the range of achievable plasma concentrations. This multicentre study was designed to assess the anti- HIV activity of ELV in an otherwise unchanged com- bination of highly active antiretroviral therapy (HAART) in individuals with the M184V mutation.
METHODS: Adults receiving triple, 3TC-containing HAART therapy whose genotype included the M184V mutation were eligible; full genotypic profiles were available for each subject. Subjects had HIV RNA lev- els of 1000–30000 copies/ml and CD4 counts >200 cells/µl. Minimum baseline safety parameters were required, together with stable clinical disease and no active AIDS-defining conditions. Subjects were ran- domized to receive blinded ELV 50 or 100 mg/day or continued 3TC 300 mg/day in a ratio of 25:25:10, while continuing the other two agents of their HAART regimen. Subjects were treated for 28 days; HIV RNA and safety determinations were measured weekly. Open-label ELV was available for subjects who wished to continue ELV after the 4 weeks.
RESULTS: Fifty-nine subjects were randomized, 56 ini- tiated treatment and 46 completed 28 days of therapy. Mean age was 43.5 years (range 21–65), mean CD4 count was 471 cells/µl (range 120–960) and mean plas- ma HIV RNA level was 10300 copies/ml (range 120–75000; Roche Amplicor 1.5 assay). Prior anti- retroviral therapies included all commercially available agents. HIV genotypes (HIV Genosure) revealed ≥ one primary nucleoside resistance mutation (in addi- tion to the required M184V) and primary resistance mutations to either the entire class of NNRTIs or PIs
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S5 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 3 sensitivity to AZT, and 4.8-, 4.5-, 1.4- and 3.6-fold resistance to 3TC, abacavir, ddI and tenofovir, respec- Antiviral activity of SPD754 tively. Pairwise comparison of separate groups of viruses with different background genotypes (wild- against clinical isolates of HIV-1 type, 41 and 215, and 67, 70 and 219) with and with- resistant to other nucleoside reverse out mutations at codon 184 showed that in all three transcriptase inhibitors cases, the addition 184 was associated with a 1.8-fold reduction in sensitivity to SPD754. Pairwise regression analysis of log-transformed FC values showed a high RC Bethell1, N Parkin2 and Y Lie2 degree of cross resistance with abacavir and ddI (r2=0.79 and 0.74, respectively), but a substantially 1 Shire BioChem, Inc., Laval, Quebec, Canada; and 2 Virologic, lower level of cross resistance with AZT and tenofovir Inc., South San Francisco, Calif., USA (r2 approximately 0.5).
BACKGROUND: SPD754 is a heterocyclic cytidine CONCLUSION: The presence of up to five mutations analogue with anti-HIV activity against a range of at codons 41, 67, 70, 210, 215 and 219 of RT confers wild-type viruses and viruses resistant to other nucleo- no more than a twofold reduction in median sensitivi- side reverse transcriptase inhibitors (NRTIs). In order ty to SPD754. Mutation at codon 184 of RT is associ- to further characterize the resistance profile of SPD754 ated with a 1.8-fold reduction in sensitivity to SPD754. antiviral activity has been assessed against a panel of 215 clinical isolates of HIV-1 with genotypes associat- ed with resistance to NRTIs.
METHODS: All viruses were selected from the Virologic library of clinical isolates. Groups of viruses were selected with defined mutation patterns at codons 41, 67, 70, 184, 210, 215 and 219 of reverse tran- scriptase, with at least 10 different viruses per group. Viruses with 69 insertion mutations, Q151M, or muta- tions at positions 65, 69, 74 or 75 were excluded. The antiviral activity of zidovudine (AZT), lamivudine (3TC), didanosine (ddI), abacavir, tenofovir and SPD754 against these viruses was assessed using the Phenosense assay.
RESULTS: Viruses with no resistance-associated RT mutations had median fold-change (FC) to SPD754 of 0.9 (range 0.7–1.1, n=20). Viruses with mutations at codons 67 and 70 of RT remained sensitive to SPD754 (median FC 1.0 vs 5.5 for AZT, n=10). Addition of mutations at codons 219, 215 and 41 progressively decreased the sensitivity of HIV-1 to SPD754, with viruses containing mutations at 41, 67, 70, 215 and 219 having a median FC of 1.3 (range 0.9–1.9, n=15). The same group of viruses showed 108-fold reduced sensitivity to AZT, and 3.2-, 3.0-, 1.3- and 2.5-fold resistance to 3TC, abacavir, ddI and tenofovir, respec- tively. Mutations at codons 41 and 215 of RT were associated with 1.2-fold reduced sensitivity to SPD754 (range 0.7-1.7 vs 33 for AZT, n=20). Addition of mutations at codons 210, 67 and 219 progressively decreased the sensitivity of HIV-1 to SPD754, with viruses containing mutations at 41, 67, 210, 215 and 219 having median FC of 1.8 (range 1.2–2.6, n=15). The same group of viruses showed 438-fold reduced
S6 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 4 thetic approach that employs a tandem kinetic resolu- tion/chiral salt crystallization protocol for preparing Synthesis and anti-HIV activity of the D-enantiomer of FDOC in high enantiopurity. In addition, we report conditions that allow for the enantiomerically pure D-FDOC racemization and recycling of the unwanted butyrate ester of the L-enantiomer of FDOC. S Mao1, M Bouygues1, C Welch3, M Biba3, J Chilenski3, RF Schinazi2 and DC Liotta1
1 Department of Chemistry, Emory University, Atlanta, Ga.; 2 Veterans Affairs Medical Center and Department of Pediatrics, Emory University School of Medicine, Decatur, Ga.; and 3 Merck Research Laboratories, Division of Merck & Co., Inc., Rahway, NJ, USA
The potential of 1′,3′-dioxolanyl nucleosides as anti- HIV drugs has been recognized for some time. In the mid-1990s, we both enantiomers of 2′,3′-dideoxy-5-flu- oro-oxacytidine (FDOC) were synthesized. Although both enantiomers exhibited good potency against HIV, they both appeared to be too toxic to be used clinical- ly. Subsequent to these initial studies, we discovered that the sample of the less toxic D-enantiomer that was tested actually contained 3–5% of its significantly more toxic L-counterpart. This then raised the interest- ing question as to whether the observed toxicity was inherent to D-FDOC or resulted from the presence of small quantities of its more toxic enantiomer. To answer this question, we used preparative chiral chro- matography to obtain several grams of optically pure D- and L-FDOC, respectively. With these two enan- tiomers in hand, we could, for the first time, unam- biguously evaluate their anti-HIV activity, cytotoxicity and resistance profile. The results of these studies indi- cated that D-FDOC not only showed excellent potency (EC and EC values in primary human lymphocytes 50 90 infected with HIV-1 are 0.04 µM and 0.26 µM, LAI respectively) and low toxicity (>100 µM in uninfected primary human lymphocytes), but also exhibited no cross resistance to lamivudine, zidovudine or nevirap- ine. In addition, in primary mouse bone marrow cells, D-FDOC showed no increase in lactic acid production even at 300 µM. In contrast, treatment with either L- FDOC and zalcitabine resulted in a >300-fold increase in lactic acid production relative to untreated control. Furthermore, in HepG2 cells (5-day assay), D-FDOC displayed no toxicity when tested up to 100 µM, whereas its L-counterpart demonstrated significant toxicity at 1.4 µM. While these data, taken in aggre- gate, clearly indicate that our original toxicity determi- nations were compromised by the presence of small quantities of the toxic L-enantiomer, they also suggest that clinical evaluations of D-FDOC should only be performed with materials that contain very little, if any, of the L-enantiomer. Herein, we describe a syn-
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S7 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 5 eration of elvucitabine resistance was examined with respect to growth fitness and resistance after they were In vitro induction of HIV variants introduced back into the wild type backbone individu- ally. A computational model was developed to explain with reduced susceptibility to the observed phenomenon that the mutation at 184 elvucitabine (ACH-126,443, confers only moderate resistance to elvucitabine but β-L-Fd4C) total resistance to lamivudine.
CONCLUSION: An in vitro resistance induction study J Fabrycki, Y Zhao, J Wearne, Y Sun, A. was performed with elvucitabine. The resulting variant Agarwal, M Deshpande, WG Rice and M Huang with reduced susceptibility carried mutations at amino acid 184 (M to I) and 237 (D to E). The D237E muta- Achillion Pharmaceuticals, New Haven, Conn., USA tion has not been described previously and its role in the generation of resistance variant is under investiga- BACKGROUND: Elvucitabine is an L nucleoside ana- tion. The double mutation conferred moderate resis- logue designed to improve on anti-HBV and anti-HIV tance to elvucitabine (approximately 10-fold shift in potency of lamivudine while maintaining the improved EC ). A computational model is proposed to explain 50 safety profile with respect to mitochondrial toxicity the phenomenon. over D-nucleoside analogues. In vitro, elvucitabine has demonstrated potent activity against a wide range of HIV-1 variants, including all clinic subtypes and mutant strains carrying common mutations resistant to nucleosides, non-nucleosides and protease inhibitors. This study was designed to evaluate in vitro resistance induction with elvucitabine.
METHODS: CEM-SS cells were infected with HIV-1 Lai and were passaged at increasing concentrations of elvucitabine. The phenotypes and genotypes of viruses collected during each passage were determined at the end of induction. The mutations were back-cloned into the wild-type HIV-1 backbone and the genotypes Lai and phenotypes associated with the mutants were iden- tified.
RESULTS: After 159 days encompassing seven pas- sages in elvucitabine concentrations ranging from 0.05 to 2.0 µM, variants of HIV with reduced suscepti- LAI bility were isolated. The EC of the resultant variant 50 was 0.2 µM, ~10-fold increase over the wild-type. Genotypic sequencing of the variant isolated from each sequential passage revealed that two mutations in the reverse transcriptase (RT) had emerged simultaneous- ly, M184I and D237E. The mutation at the 184 locus has been described after exposure to lamivudine, but is usually rapidly replaced in the presence of that drug with M184V. No ‘switch’ to M184V could be detected in the presence of elvucitabine. The contribution of the two mutations to viral resistance was confirmed after they were introduced into the wild-type backbone and the resulting virus was tested for its susceptibility to elvucitabine. Preliminary cross resistance study indi- cated that the mutant was cross resistant to lamivu- dine, as expected, but not to other nucleoside inhibitors tested. The role of each mutation in the gen-
S8 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 6 schedule, and showed a 0.61, –0.05 and 0.43 log 10 drop, respectively. The viral genotype for all subjects Antiviral activity of the nucleoside remained unchanged at the end of the treatment sched- Reverset following single oral doses ule. in HIV-1-infected patients CONCLUSIONS: RVT reduced the HIV-1 viral load after a single dose by a mean of 0.4 log for all doses 10 L Stuyver1, TR McBrayer1, RL Murphy1,2, tested. One subject infected with a mutant virus D Schürmann3, I Kravec3, A Beard1, responded as well as subjects infected with wild-type RF Schinazi4, A De La Rosa1 and MJ Otto1 virus.
1 Pharmasset, Inc., Tucker, Ga., USA; 2 Northwestern University, Chicago, Ill., USA; 3 Charité University Hospital, Berlin, Germany; and 4 Emory University School of Medicine, and Veterans Affairs Medical Center, Atlanta, Ga., USA
BACKGROUND: Reverset (RVT, β-D-2′,3′-didehydro- 2′,3′-dideoxy-5-fluorocytidine) is a nucleoside ana- logue that retains activity against lamivudine- and zidovudine-resistant HIV-1 in vitro. It is being devel- oped under a US IND for the treatment of HIV-1 in nucleoside reverse transcriptase inhibitor (NRTI)- experienced patients.
METHODS: RVT was administered as a single oral dose to antiretroviral treatment-naive HIV-1-infected males (six per cohort) at doses of 0, 10, 25 or 50 mg as buffered solutions or 0, 50, 100 or 200 mg as enteric-coated tablets. Blood samples, obtained over a 48 h period for pharmacokinetic analysis, were analysed for HIV RNA levels using quantitative real- time RT-PCR.
RESULTS: Viral loads dropped significantly over 48 h with an average reduction of 0.4 ±0.2 log for all dose 10 levels. The antiviral response over the 48 h was not dose-dependent. At the 10 mg dose, a 0.42 ±0.2 log 10 was observed (P=0.005), while at 100 mg a similar effect of 0.44 ±0.17 were noted (P=0.0007). The dose- independent significant antiviral response was also observed at the 24 h time-point, with an average reduction of 0.11 ±0.11 (P=0.03). The 12 h time-point was not significantly different from the baseline. The mean plasma C values ranged from 1 to 8 µM. A max maximal effect of viral inhibition was obtained at the lowest C (0.87 mM), which is equivalent to the in max vitro EC value for wild-type virus. All available viral 90 strains were sequenced in the reverse transcriptase gene before (n=18 strains) and after (n=18 strains) the treat- ment schedule. Wild-type viral genotype was found in all but one subject who showed at baseline the follow- ing genotype: L41+N103+C181+W210+D210, sug- gesting past exposure to zidovudine (possibly in anoth- er host) and non-nucleoside analogues. This subject received the 10 mg, placebo and 25 mg treatment
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S9 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 7 occurred only in FTC failures, 31%, while develop- ment of thymidine analogue-associated mutations Characterization of baseline and (TAMs) occurred only in d4T failures, 15%. One VF in the FTC arm (6%) and three VF in the d4T arm treatment-emergent resistance (7%) developed ddI-associated mutations. In these mutations following 1 year of antiretroviral-naive patients who failed therapy, the therapy on an entirely once a day prevalence of mutations at baseline was relatively high; 38% FTC and 32% d4T. NNRTI-associated muta- regimen including emtricitabine tions (31% FTC, 22% d4T) and TAMS (19% FTC and 15% d4T) were the most prevalent baseline muta- K Borroto-Esoda, J Waters, JB Quinn, A Shaw, tions in each group. No FTC (M184I/V) or ddI J Hinkle and F Rousseau for the FTC-301 (L74V/K65R) mutations were present in either cohort Study Team at baseline. The incidence of VF remained statistically higher in the d4T arm after censoring patients with Triangle Pharmaceuticals, Inc., a wholly owned subsidiary of baseline TAMs (d4T arm), P=0.0132, or baseline Gilead Sciences K103N (FTC arm) and K103N/TAMS (d4T arm), P=0.0182. BACKGROUND: While highly active antiretroviral therapy has significantly impacted the treatment of CONCLUSIONS: These results demonstrate that a HIV infection, complexity of administration and toxi- regimen containing once daily FTC was statistically city issues can lead to suboptimal adherence and the superior to twice-daily d4T, in a background of once- development of resistance mutations. Emtricitabine daily ddI plus EFV, with a significantly lower rate of (FTC) is a new once-daily nucleoside analogue reverse VF with fewer mutations, even after adjusting for the transcriptase inhibitor (NRTI) with potent activity prevalence of baseline mutations. against HIV. In study FTC-301 we evaluated the safe- ty and antiviral activity of a simple, non-thymidine, once-daily regimen as first-line treatment for therapy- naive HIV-infected patients.
METHODS: 571 antiretroviral-naive patients with plasma HIV-1 RNA (VL) >5000 copies/ml were ran- domized in a 1:1 ratio to receive 200 mg FTC once daily or stavudine (d4T) twice daily in combination with once-daily didanosine (ddI) and once-daily efavirenz (EFV). VL was measured at baseline and every 4 weeks to week 48. Virological failure (VF) was defined as never achieving <400 copies/ml or rebound >400 copies/ml on two consecutive measurements. The incidence of VF was compared between treatment arms using a two-sided exact test. Patients who expe- rienced VF had sequence analysis performed on the HIV RNA isolated from plasma at baseline and at the time of VF.
RESULTS: Virological failure occurred in 6% (17) of patients in the FTC arm and 15% (41) of patients in the d4T arm (P=0.0014). Genotypic data were obtained for 57/58 VF (16/17 FTC, 41/41 d4T) at baseline and at time of VF. In these 57 patients experi- encing VF, the incidence of mutations was 69% in the FTC arm and 85% in the d4T arm. Development of non-nucleoside reverse transcriptase inhibitors (NNRTI) mutations accounted for the majority of mutations in both subgroups of VF, 63% FTC and 85% d4T. Development of the M184V mutation
S10 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 8 respectively. Most (79%) of these 2630 samples were resistant to efavirenz and 69% were resistant to all Antiviral activity of TMC125, a current NNRTIs. At 10 nM, TMC125 inhibited 80% of the samples resistant to at least one NNRTI and potent next-generation 76% of the samples resistant to all current NNRTIs. non-nucleoside reverse The corresponding percentages for efavirenz were 29% transcriptase inhibitor (NNRTI), and 9%, respectively. In addition, TMC125 inhibited 78% of the EFV-resistant subset (n=2066), at 10 nM. against >5000 recombinant clinical The median EC of TMC125 for samples containing 50 isolates exhibiting a wide range of V106M, which was observed together with at least one NNRTI resistance other NNRTI resistance mutation, was 1.6 nM (n=11). Sixty-three percent of the samples harbouring four NNRTI resistance mutations still had an EC below 50 J Vingerhoets1, H Van Marck1, J Veldeman2, 10 nM for TMC125, whereas 70% of the samples 1 2 1 with only two mutations had an EC above 10 nM for M Peeters , P McKenna , R Pauwels and 50 M-P de Béthune1 efavirenz.
1 Tibotec, Mechelen; Belgium; and 2 Virco, Mechelen; Belgium CONCLUSIONS: TMC125 is a potent next-genera- tion NNRTI, with activity against most of recently cir- BACKGROUND: TMC125 is a potent next-genera- culating strains of HIV, including samples that are tion non-nucleoside reverse transcriptase inhibitor resistant to all marketed NNRTIs. The antiviral activ- (NNRTI), active against wild-type as well as NNRTI- ity against class-associated NNRTI resistance together resistant HIV-1. In vitro selection experiments have with the increased genetic barrier to development of demonstrated an increased genetic barrier to the devel- resistance, are unique features of TMC125. opment of resistance to the compound. TMC125 also showed in vivo antiviral activity in patients with docu- mented phenotypic NNRTI resistance in a 7-day Phase IIa trial. In the present study, we determined the antivi- ral activity of TMC125 in more than 5000 clinical iso- lates submitted for phenotypic resistance testing in 1999–2000 (panel A) and 2001–2003 (panel B). The antiviral activity of TMC125 on these isolates was compared to the currently approved NNRTIs.
METHODS: Recombinant clinical isolates were con- structed according to the Antivirogram method. Phenotypic and genotypic analyses were performed by the Antivirogram and VirtualPhenotype assays, respectively. Data analysis was performed using SAS and Spotfire DecisionSite software.
RESULTS: The prevalence of mutations at 15 NNRTI resistance-associated positions (98, 100, 101, 103, 106, 108, 179, 181, 188, 190, 225, 230, 236, 238 and 318) was compared between panels A (n=2065) and B (n=3545). A relative increase in frequency of mutations K101P, K103S, V106A, V179I and Y188L was observed in panel B. No significant change was observed for V106M, a recently described NNRTI resistance mutation. TMC125 inhibited 91% of all samples (n=5610) with an EC <10 nM, while 50 efavirenz only inhibited 67% at 10 nM. The number of samples resistant to at least one of the three current NNRTIs (defined as a fold change in EC >10) was 50 1050 (51%) and 1580 (45%) for panels A and B,
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S11 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 9 V189V/I). The appearance of these partial mutations was not associated with an increase in fold resistance Characterization of resistance before for TMC125 or any of the current NNRTIs. At base- line, there was no correlation between the fold resis- and after short-term therapy with tance values of nevirapine or efavirenz with TMC125. TMC125 in patients with The median (range) fold resistance values at baseline documented non-nucleoside reverse for nevirapine, efavirenz and TMC125 were 128 (58–136), 116 (5–820) and 2.2 (0.5–8.5), respectively. transcriptase inhibitor resistance At the end of therapy, the median (range) fold resis- tance values for nevirapine, efavirenz and TMC125 J Vingerhoets1, M Peeters1, H Azijn1, were 120 (46–146), 103 (4–974) and 2.6 (0.8–11.6), C Jordens2, P McKenna2, L Bacheler1, respectively. Neither NNRTI fold resistance values at G Van ’t Klooster1, R Pauwels1 and baseline nor the presence of mutations associated with M-P de Béthune1 NNRTI resistance at baseline were predictive for response in this group of patients. 1 Tibotec, Mechelen; Belgium; and 2 Virco, Mechelen; Belgium CONCLUSIONS: TMC125 is effective in suppressing BACKGROUND: TMC125-C207 was an open-label resistant HIV strains from patients failing on an Phase IIa study to evaluate the antiviral activity, safety NNRTI-containing regimen and with phenotypic evi- and tolerability of TMC125 in patients with docu- dence of resistance. No evidence has been found that mented non-nucleoside reverse transcriptase inhibitor TMC125 selected for increased resistance during 7 (NNRTI) resistance. Sixteen HIV-1-positive subjects days of treatment. By overcoming class-associated on a failing antiretroviral regimen (viral loads above NNRTI resistance, TMC125 is considered to be a 2000 HIV RNA copies/ml), consisting of two NRTIs next-generation NNRTI. and an NNRTI, and with phenotypically confirmed resistance to efavirenz, were enrolled. They received TMC125 (900 mg twice daily) for 7 days as a substi- tution for the failing NNRTI; NRTI therapy remained unchanged. TMC125 was highly active in patients infected with NNRTI-resistant HIV-1, as demonstrat- ed by a median viral load drop of 0.89 log RNA copies/ml from baseline to day 8, and was well tolerat- ed. In the present study, comparative phenotypic and genotypic resistance data from screening, baseline and end of therapy have been analysed.
METHODS: Drug susceptibility profiles were deter- mined using the Antivirogram assay and mutational patterns were determined using VirtualPhenotype. Both resistance determinations were performed on plasma samples taken at screening (within 49 days prior to treatment start), baseline (day 1) and end of therapy (day 8) time-points.
RESULTS: The population in this study had a wide range of mutations associated with resistance to NNRTIs, including changes at positions 98, 100, 101, 103, 108, 179, 181, 188, 190, 225 and 238. The medi- an number of NNRTI mutations was two (range 1–4), at both screening and baseline. One patient acquired a partial NNRTI mutation (Y181C/Y) between screen- ing and baseline (in the absence of TMC125). Between baseline and end of therapy, three patients acquired additional changes in the RT gene: these were all mutant/wild-type mixtures (K101Q/K, K103N/K and
S12 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 10 integrase inhibitors with distinct resistance profiles, however there appears to be a significant potential for The identification of active site cross resistance between many such compounds (such as, S-1360 and L-870810) despite their apparent dif- mutations that confer resistance to ferences in structure. Localization of the critical deter- structurally diverse inhibitors of minants for resistance to the integrase active site is HIV-1 integrase strand transfer consistent with biochemical studies that demonstrate these inhibitors function by sequestering the active site supports a general mechanism of metals in integrase and the observation that similar phosphotransferase inhibition compounds have been identified which inhibit mecha- nistically-related metal-dependent phosphotransferases DJ Hazuda and the MRL HIV-1 Drug Discovery such as HIV-1 RNase H. Team
Merck Research Laboratories, West Point, Pa., USA
Inhibitors of the integrase strand transfer reaction have been shown to be effective inhibitors of integration and HIV-1 replication in vitro and in vivo. The dike- tone S-1360 and the napthyridine L-870810 are the first compounds in this novel class to enter into clini- cal studies in HIV-1-infected patients. In the presence of human serum, S-1360 and L-870810 inhibit HIV-1 replication in cell culture with IC95s of 12000 and 100 nM, respectively. Although structurally distinct, the compounds have identical mechanisms of action and compete for binding to the same site. In an attempt to understand the potential for cross resistance within this new class of agents we have used a variety of dike- tones and naphthyridines related to S-1360 and L- 870810, respectively, to select resistant HIV-1 variants in vitro. Mutations in integrase were identified upon sequencing each of the resistant virus populations. The observed mutations were introduced into an isogenic virus background and the recombinant viruses were then used to evaluate their respective susceptibility to a panel of integrase inhibitors from each structural class. All of the inhibitors we evaluated selected for similar but not identical mutations in integrase. In each case the mutations were localized to the integrase active site proximal to the residues that coordinate the catalytic metal ions (D64, D116 and E152). Although similar mutations were selected with a variety of structurally diverse compounds, marked differences were observed in the susceptibility profiles of these inhibitors both within and between each structural class. Some com- pounds including S-1360 were affected by a wide range of different mutations, while other inhibitors, including L-870810, displayed an overlapping but more restricted resistance profile. In the course of this extensive evaluation we also identified compounds that exhibited apparently discrete resistance profiles and integrase inhibitors that were effective against all of the resistant variants we have identified to date. These analyses demonstrate it is possible to identify
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S13 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 11 virtually no cytotoxicity (CI >100 µM) as assessed in 50 several cell lines. Importantly, mappicine 756 retained Mappicine inhibitors of HIV-1 full antiviral potency against several drug-resistant HIV-1 strains, including virus with high-level resis- reverse transcriptase-associated tance to nevirapine, delavirdine and efavirenz, the ribonuclease H three clinically approved non-nucleoside RT inhibitors. HIV strains resistant to mappicine 756 developed upon MM Hossain1, W Zhang2, D Curran3 and serial passage of the virus in the presence of increasing MA Parniak1 concentrations of the drug, leading to virus with 80- fold resistance to the drug. Several mutations were 1 Viral Diseases Unit, Division of Infectious Diseases, Department found in the virus resistant to mappicine 756; these of Medicine, University of Pittsburgh; 2 Fluorous Technologies, mutations localized only in the segment of the RT gene Inc.; and 3 Department of Chemistry, University of Pittsburgh, corresponding to the RNase H subdomain. Pittsburgh, Pa., USA CONCLUSION: The good antiviral activity in the INTRODUCTION: Current therapeutics for treatment absence of significant cytotoxicity suggest that map- of HIV infection are directed at two viral targets, pro- picine analogues may represent an interesting new tease and the DNA polymerase activity of reverse tran- class of antiretroviral agents, those targeting RT-asso- scriptase (RT). Viral resistance to these therapeutics is ciated RNase H. Structural variants of mappicines are an increasingly serious problem, thus identification of readily prepared by combinatorial methods, and it is drugs directed at new HIV targets is essential. The therefore expected that improvements in antiviral ribonuclease H (RNase H) activity of RT resides in a potency may be attained. subdomain that is spatially distinct from the RT DNA polymerase active site. The viral RNase is absolutely essential for retroviral replication and thus presents a logical target for antiviral intervention. However, while numerous inhibitors of HIV-1 RT DNA poly- merase activity have been identified (including 10 drugs in current clinical use), very few inhibitors of RNase H have been identified, and none are in pre- clinical development. We have developed a fluores- cence-based assay for HIV RT RNase H that enables high-throughput screening for inhibitors of this poten- tial target. Using this screening assay, we have discov- ered that certain analogues of mappicine are potent inhibitors of HIV-1 RT-associated RNase H.
OBJECTIVE: To characterize the antiviral properties of mappicine analogue inhibitors of HIV-1 RT-associ- ated RNase H.
RESULTS: Initial screening of a small library of 110 mappicine analogues using a novel fluorescence-based high-throughput screening assay for HIV RT RNase H resulted in the identification of two compounds with reasonable antiviral potency, but unfortunately with significant cytotoxicity. A larger library of 560 map- picine analogues was prepared by fluorous-tagged combinatorial synthesis and screened, leading to the identification of 55 additional inhibitors. Of these, mappicine 756 was among the most potent (IC ≈2 µM 50 against RT-associated RNase H in vitro). Mappicine 756 showed very good antiviral activity against the wild-type HIV-1 IIIB strain (EC ≈2.5 µM) and had 50
S14 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 12 graphics and prior treatment experience. The median fold-change in susceptibility relative to wild-type (WT) Characterization of the impact of for TPV and available PIs at baseline was 1.1, 76.5, 8.7, 7.0, 12.2, 36.8 and 94.2 for TPV, lopinavir, genotype, phenotype, and amprenavir, saquinavir, indinavir, nelfinavir and riton- inhibitory quotient on antiviral avir, respectively. This study has identified four univer- activity of tipranavir in highly sal PI-associated mutations (UPAMs; mutations at codons 33, 82, 84 and 90) that are associated with treatment-experienced patients class cross resistance. Three of these mutations (33, 82 and 84) are in the protease active site, such that 16 to S McCallister1, V Kohlbrenner1, K Squires2, A 20 compensatory mutations may be required for ≥2 Lazzarin3, P Kumar4, E DeJesus5, J Nadler6, J UPAMs to coexist. Patients with 0–2 UPAMs had a 7 8 9 1 –0.97 to –1.32 log change in VL, compared with Gallant , S Walmsley , P Yeni , J Leith , C 10 1 1 1 1 –0.32 log in patients with 3 UPAMs. The relationship Dohnanyi , D Hall , JP Sabo , TR MacGregor , 10 W Verbiest10, P McKenna10 and D Mayers1 of UPAMs to phenotypic susceptibility was evaluated. In the presence of 0–2 UPAMs, the median fold-change 1 Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, Conn., in TPV IC was 0.9, 1.0 and 1.3, compared with 2.2- 50 USA; 2 University of Southern California, Los Angeles, Calif., fold in patients with 3 UPAMs. Across all three arms USA; 3 Fondazione Centro San Raffaele Del Monte Tabor, Milan, of the study, patients whose isolates had ≤1 and >one- Italy; 4 Georgetown Medical University, Washington DC, USA; 5 to twofold change in TPV IC relative to WT demon- 50 IDC Research Initiative, Altamonte Springs, Fla., USA; 6 strated a –1.23 and –1.24 log change in VL, respec- 10 Hillsborough County Health Department, Tampa, Fla., USA; 7 tively, at 2 weeks. A –0.21 log response was seen in 10 Johns Hopkins University School of Medicine, Baltimore, Md., patients whose isolates had >two- to fourfold change USA; 8 Toronto General Hospital, Toronto, ON, Canada; 9 Groupe in TPV IC , suggesting a phenotypic cut off of approx- 50 Hospitalier Bichat-Claude Bernard, Paris, France; and 10 VIRCO, imately twofold WT IC (VIRCO assay). Similarly, the 50 Mechelen, Belgium median VL responses after 2 weeks for IQs ≤5, >5–25, >25–50, >50–100, >100–150 and >150 were –0.19, BACKGROUND: Tipranavir (TPV), the first non-pep- –0.35, –0.82, –1.31, –0.96 and –1.23 log , respective- 10 tidic protease inhibitor (NPPI), has demonstrated ly. This result suggests that there is an apparent IQ robust antiviral activity against multiple protease breakpoint of roughly 50 in HTE patients, below inhibitor (PI)-resistant HIV-1, both in vitro and in clin- which there is a decrease in antiviral response. ical studies. A large Phase IIB study of TPV/r has allowed characterization of the impact of genotype, CONCLUSION: This analysis has determined that a phenotype and inhibitory quotient (IQ) on antiviral good response to TPV is maintained in the presence of activity of TPV in highly treatment-experienced (HTE) <3 UPAMs, IC
RESULTS: Two-hundred-and-sixteen patients were randomized and evenly distributed across arms with regard to baseline viral load, CD4 cell count, demo-
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S15 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 13 three or more patients across the two studies include L10I/V (5), K20M/L/T (3), M46I (3), I54V (4) and Characterization of treatment- L63A/D/T (3). Mutations occurring in three or more patients out of a total 63, at codons not generally asso- emergent resistance mutations in ciated with resistance, were I13V (3), K55R/Q (3), two Phase II studies of tipranavir H69Y (3) and T74A (3).
D Hall, S McCallister, D Neubacher, M Kraft CONCLUSION: Analysis of HIV-1 viral isolates from and DL Mayers patients enrolled in studies of treatment-experienced patients has identified several mutations that emerged Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, Conn., during TPV/r treatment of patients with a mean of USA approximately 10 baseline protease gene mutations. The most frequent mutations seen in this study at BACKGROUND: Tipranavir (TPV) is the first non- codons 33, 82 and 84 are UPAMs located in the pro- peptidic protease inhibitor (NPPI). Phase II clinical tri- tease active site. The presence of ≥2 of these mutations, als have demonstrated a sustained viral load response therefore, appears to require the accumulation of for up to 80 weeks of treatment in single and multiple- 16–20 other protease mutations, and may have impor- protease inhibitor (PI)–experienced patients. tant implications for the fitness of TPV-resistant HIV- Furthermore, these studies have demonstrated that as 1. Few patients in these studies accumulated >2 many as 16 to 20 protease gene mutations may be UPAMs. required for reduced susceptibility to TPV. This analy- sis investigates treatment-emergent mutations seen in two Phase II trials in treatment-experienced HIV-1- positive adults.
METHODS: In two Phase II, open-label, randomized trials (BI 1182.4 and BI 1182.2), 91 HIV-1-positive adults with single- or multiple-PI failure received vari- ous doses of TPV/ritonavir (TPV/r). Fifty single PI- experienced patients received two new nucleoside reverse transcriptase inhibitors (NRTIs) plus either low-dose or high-dose TPV/r twice daily; and 41 mul- tiple PI-experienced, non-NRTI (NNRTI)-naive patients received low-dose or high-dose TPV/r plus efavirenz (EFV) and one new NRTI. Genotypic testing was performed at baseline and at follow-up.
RESULTS: On-treatment genotypes were available for 24 patients from BI 1182.4 and 39 from BI 1182.2. The mean number of protease gene mutations at study entry was 10 in BI 1182.4 and 12 in BI 1182.2. The most frequent treatment-emergent protease mutations were L33I/F/V, V82L/T and I84V, occurring in 11, 9 and 9 isolates, respectively. Recent studies have shown that >2 universal PI-associated mutations (UPAMs; defined as any mutation at codons 33, 82, 84 or 90) may be required for reduced susceptibility to TPV at clinically relevant doses and reduced antiviral activity; 1 or 2 UPAMs may be sufficient for reduced suscepti- bility to available PIs. Three of five patients in BI 1182.2 who developed reduced susceptibility to TPV during treatment had 2 UPAMs at baseline, and went on to accumulate a third; one had 3 UPAMs at base- line. Other treatment-emergent mutations at codons associated with protease resistance that occurred in
S16 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 14 genotypes were found to confer five- and sevenfold decreased susceptibility to TPV, respectively: Characterization of HIV-1 showing V3I/L10V/I15V/L19I/M36L/S37N/R41K/M46I/K55R/ Q61N/I64V/I72V/T74S/V82T/I84V/I85V/I93L and decreased susceptibility to V3I/L10V/L33I/E35D/M36I/S37N/I54V/Q58E/I62V/L tipranavir and their inhibition by 63P/A71V/V82L/L90M/I93L/C95V. Drug mixture tipranavir-containing drug mixtures experiments using these clinical isolate-derived viruses or wild-type HIV-1 showed that combinations of TPV:APV and TPV:LPV displayed predominantly L Doyon, S Tremblay, E Wardrop, R Maurice, additive effects against wild-type virus replication and D Thibeault, J Archambault and MG Cordingley somewhat lower CIs against protease mutant viruses. However, only in a few instances was the deviation Department of Biological Sciences, Boehringer Ingelheim from the additive effect outside the inter-experimental (Canada) Ltd, Research and Development, Laval, Canada variability.
BACKGROUND: Tipranavir (TPV) is a non-peptidic CONCLUSIONS: Resistance to TPV involves multiple protease inhibitor that maintains potent activity mutations in the protease gene and leads to a reduced against a broad range of multiple protease inhibitor sensitivity to most other PIs and to a decreased repli- resistant human immunodeficiency virus (HIV-1) iso- cation capacity of viruses. TPV, however, maintains lates. The genotypic changes responsible for the mostly additive effects on TPV-resistant or wild-type reduced susceptibility to TPV and their effect on cross virus when used in combination with the protease resistance to other PIs are only partially understood. inhibitors APV and LPV. Here we present our characterization of TPV-resistant viruses at the genotypic and phenotypic levels includ- ing their in vitro susceptibility to inhibition by mix- tures of the protease inhibitors TPV:amprenavir (APV) or TPV:lopinavir (LPV).
METHODS: Recombinant viruses were reconstructed to represent HIV-1 selected to grow in culture in the presence of TPV or representing clinical isolates from patients undergoing therapy with TPV. Viruses con- tained between 2–17 mutations in the protease gene and in some cases an additional mutation in the CA/P2 p55gag cleavage site. Viruses were studied in antiviral assays to determine their susceptibility to protease inhibitors and in a Jurkat-LTR luciferase reporter cell line to determine their replication capacity. Susceptibility to inhibition by TPV containing drug mixtures was determined using constant ratio combi- nation experiments and the calculation of combination indices (CI). Using this model, CI values <0.9=synergy, CI between 0.9 and 1.1=additive effects (no drug inter- action) and CI>1.1=antagonism.
RESULTS: In vitro selected TPV-resistant viruses con- tain up to 10 mutations in the protease gene, including L33F, V82L and I84V. Introduction of these mutations into viral molecular clones by site-directed mutagene- sis conferred up to 69-fold resistance to TPV and two- to 118-fold decreased susceptibility to other protease inhibitors in addition to decreasing viral replication capacity. A CA/P2 cleavage site mutation observed during in vitro selection did not directly contribute to TPV resistance. The following two clinical protease
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S17 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 15 D30N, M46I/L, G48V, I50V/L, V82A/F/T/S, I84V or L90M). More than 80% of subjects had more than one TMC114, a potent next-generation primary PI mutation. All primary PI mutations, except I50L and V82S, were present at baseline in at least one protease inhibitor: characterization sample. Phenotyping of the baseline samples showed of antiviral activity in multiple that 46% of the subjects were resistant to all currently protease inhibitor-experienced approved PIs and only 27% of the subjects were sensi- tive to two or more PIs (cut-offs as defined by the patients participating in a Antivirogram). In the treatment arms, the median fold change in EC as compared to wild-type for Phase IIa study 50 TMC114 was 1.8 (range 0.3 to >21) at baseline and S De Meyer1, M Peeters1, C Jordens2, 1.5 (range 0.3–13.1) at end of treatment. There was no P McKenna2, R van der Geest1, R Pauwels1 and correlation between TMC114 susceptibility at baseline M-P de Béthune1 and virological outcome at day 14. Many genotypic changes were observed between screening, baseline 1 Tibotec, Mechelen, Belgium; and 2 Virco, Mechelen, Belgium and end of treatment. No mutation pattern could be associated with virological response to TMC114. BACKGROUND: TMC114 is a potent next-genera- tion protease inhibitor (PI), active against wild-type as CONCLUSIONS: This study demonstrates the potent well as PI-resistant HIV. The study TMC114-C207 antiviral activity of TMC114, a next-generation PI, in was a placebo-controlled Phase IIa trial to evaluate the multiple PI-experienced patients over 14 days. No antiviral activity, safety and tolerability of TMC114 mutation patterns influencing the response to treat- over 14 days treatment. Fifty multiple PI-experienced ment with TMC114 could be detected in this study. subjects (range 2–4 PIs) on a failing nucleoside reverse transcriptase inhibitor (NRTI)- and PI-containing regi- men (HIV-1 RNA >2000 copies/ml) were enrolled. They received TMC114 with low-dose ritonavir (TMC114/RTV) at one of three doses (300/100 mg twice daily, 600/100 mg twice daily or 900/100 mg q.d.) as a substitution for their current PI or remained on their current regimen (control group) for 14 days. Afterwards, all patients switched to an investigator- selected highly active antiretroviral therapy (HAART) regimen. Overall, the median change in plasma HIV-1 RNA for the three TMC114/RTV groups at day 14 was –1.35 log compared to +0.02 log for the con- 10 10 trol group. No significant difference was observed between the three TMC114/RTV treatment arms. In this study, phenotypic and genotypic resistance data from screening, baseline and end of therapy were analysed.
METHODS: Phenotypic analysis was conducted using the Antivirogram assay and genotypic analysis using the VirtualPhenotype. Both determinations were per- formed on plasma samples taken at screening (within 28 days prior to treatment start), baseline (day 1) and end of therapy (day 15) time-points.
RESULTS: Subjects in this study had a broad range of protease mutations at baseline. The median number of total protease gene mutations was 15 (range 8–26) and the median number of PI resistance-associated muta- tions was 6 (range 1–11), with a median number of primary PI mutations of 3 (range 0–5) (including
S18 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 16 inhibitor with the enzyme is very tight (Kd=10–12 M) and is extremely enthalpically driven. When the struc- TMC114 binds within the substrate ture was compared with the substrate complexes of HIV-1 protease, it was found that TMC114 occupies a envelope of HIV-1 protease, which volume that is contained within the substrate envelope, could account for its efficacy unlike most of the currently approved PIs. against multi-protease inhibitor- CONCLUSIONS: Many drug-resistant variants of resistant virus HIV protease evolve to maintain substrate recognition while compromising inhibitor binding, especially when N King1, M Prabu-Jeyabalan1, P Wigerinck2, the inhibitors extend beyond the substrate envelope. M-P de Béthune2 and CA Schiffer1 The fact that TMC114 fits well within the substrate envelope, associated with its tight binding to the 1 University of Massachusetts, Worcester, Mass., USA; and 2 enzyme, therefore, could account for why TMC114 Tibotec, Mechelen, Belgium remains active against most multi-PI-resistant variants. Hence, a mutation that affects TMC114 binding will INTRODUCTION: HIV-1 protease is the target of likely cause a dramatic change in the ability of HIV-1 very potent antiviral drugs for the treatment of HIV-1 protease to recognize its substrates. This may also infection. All current protease inhibitors (PIs) are the explain why selection of TMC114-resistant virus in successful result of structure-based drug design. vitro has proven difficult, as this might require changes Unfortunately, as the viral reverse transcriptase is high- beyond the protease gene, most probably in the cleav- ly error prone, and under the selective pressure of drug age sites. These results support our previous hypothe- therapy, many viable drug-resistant variants of HIV-1 sis that inhibitors that fit within the substrate envelope protease have emerged. These PI-resistance mutations of HIV-1 protease may be more effective and less sus- occur mostly at positions in the protease that will com- ceptible to drug resistance mutations. promise inhibitor binding whilst retaining substrate specificity. We have determined from crystal structures of substrate complexes with HIV protease that the cur- rent PIs protrude beyond the substrate envelope, this may explain why resistance mutations constrain inhibitor binding. TMC114 is a next generation PI: recent virological and clinical results indicate that it is effective against known multi-PI-resistant variants of HIV-1. Furthermore, in vitro selection of TMC114- resistant variants from wild-type HIV-1 has proven dif- ficult. In the present study we determined and com- pared the high-resolution crystal structure and ther- modynamics of TMC114 or substrate binding to wild type HIV-1 protease.
METHODS: TMC114 was crystallized in complex with wild-type HIV-1 protease, and X-ray diffraction data was collected, processed and refined, using stan- dard crystallographic techniques. The structure of the inhibitor complex was compared graphically with sub- strate complexes previously determined in our labora- tory. The thermodynamics of inhibitor binding was determined using isothermal titration calorimetry (ITC) at 25°C and compared with other inhibitors binding.
RESULTS: The structure of TMC114 in complex with wild-type protease was determined to 1.93Å (P2 2 2 ; 1 1 1 R=19.7; Rf=22.2). In addition, the binding constant determined by ITC showed that the interaction of the
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S19 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 17 The median fold change in EC for TMC114 was <4 50 for each of these subgroups, which illustrates the activ- Antiviral activity of TMC114, a ity of TMC114 against PI-resistant isolates. A geno- type was available for 498 of the 1666 PI-resistant iso- potent next-generation protease lates. The number of primary mutations (D30N, inhibitor, against >4000 recent M46I/L, G48V, I50V/L, V82A/F/T/S, I84V or L90M) recombinant clinical isolates was determined for each of these isolates. One percent had no primary mutation, 23% had 1, 41% had 2, exhibiting a wide range of (protease 31% had 3 and 4% had 4 primary mutations. The median fold change in EC for TMC114 was <4 for inhibitor) resistance profiles 50 each of these subgroups. S De Meyer1, H Van Marck1, J Veldeman2, P McKenna2, R Pauwels1 and M-P de Béthune1 CONCLUSIONS: TMC114 is a potent, next-genera- tion PI with activity against a wide range of PI-resis- 1 Tibotec, Mechelen, Belgium; and 2 Virco, Mechelen, Belgium tant recombinant clinical isolates. This activity, defined by a median fold change of <4, extended to isolates INTRODUCTION: TMC114 is a potent, next-genera- resistant to all currently-approved PIs and also to iso- tion protease inhibitor (PI), active against wild-type as lates carrying up to four primary PI mutations. well as PI-resistant HIV. Recently, TMC114 showed in vivo antiviral efficacy in a 2-week Phase IIa trial in multiple PI-experienced patients. In order to assess the performance of TMC114 against currently circulating strains of HIV, the compound was tested against >4000 clinical isolates submitted for phenotypic resis- tance testing. The antiviral activity of TMC114 on these isolates was compared to the currently approved PIs: indinavir, ritonavir, nelfinavir, saquinavir, ampre- navir and lopinavir.
METHODS: Recombinant clinical isolates were con- structed according to the Antivirogram method. Phenotypic and genotypic analyses were performed by the Antivirogram and VirtualPhenotype assays, respectively. Data analysis was performed using SAS and Spotfire DecisionSite software.
RESULTS: From the 4024 tested recombinant clinical isolates, 1666 (41%) were resistant to at least one of the currently approved PIs, defined as a change in EC 50 >fourfold as compared to wild-type. The median fold change in EC against these 1666 resistant isolates for 50 TMC114 was 1.1, corresponding to an EC of 3.5 nM. 50 Eighty percent of these PI-resistant isolates were still susceptible (defined as fold change in EC <4) to 50 TMC114. For the remaining 20% isolates, the median fold change in EC for TMC114 was 10, thus show- 50 ing that the compound can inhibit 90% of the 1666 PI- resistant isolates with a fold change ≤10. A subgroup of 1501 isolates, for which data for all six approved PIs were available, was used to determine the influence of the number of PIs with a fold change >4 on the activity of TMC114. Among these PI-resistant isolates, 67% were resistant to 4 or more PIs, with 31% resis- tant to all 6 approved PIs, 23% to 5, and 13% to 4.
S20 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 18 I50V mutation also had low RO033-4649 FC (median threefold). Cross resistance between pairs of PIs In vitro cross resistance profile of showed the highest degree of correlation with nelfinavir (r2=0.63) and lowest with amprenavir (r2=0.15). Both RO033-4649 against a panel of HXB2 and a clinical isolate G34 (M46L, G48V, I62V, multiply-substituted protease L63P, T74S, V77I, V82A) were used in passaging stud- inhibitor-resistant viruses: role of ies at increasing drug concentrations. No mutation arose in the wild-type HXB2 culture after 11 passages, common protease resistance and in the clinical isolate, V32I and later I54V appeared mutations under increasing drug pressure (120 nM and 480 nM, respectively). G Heilek-Snyder1, A Kohli1, N Cammack1 and N Parkin2 CONCLUSIONS: (1) An average of 14 mutations are required before RO033-4649 demonstrates a loss of 1 Roche, Palo Alto, Calif.; and 2 Virologic, Inc., South San susceptibility greater than 20-fold. (2) In a background Francisco, Calif., USA of multiple (10–18) resistance mutations and polymor- phisms, clinical samples carrying G73C, S or T show BACKGROUND: RO033-4649 is a potent, highly reduced susceptibility to RO033-4649. (3) In a back- selective HIV-1 protease inhibitor (PI) with activity ground of seven pre-existing resistance mutations, against PI-resistant viruses and a promising pharmaco- V32I and I54V arose under drug pressure, confirming kinetic profile. the requirement of multiple mutations to confer reduced susceptibility to RO033-4649. METHODS: Susceptibility to RO033-4649 and to amprenavir, indinavir, nelfinavir, saquinavir and lopinavir was tested in single cycle (PhenoSense) and live HIV-1 virus assays. The fold change (FC) in IC 50 from NL4-3 reference was calculated and analysed using correlation matrices to investigate relationships between protease mutations and FC, either as a con- tinuous or dichotomous variable. FC values between groups of viruses were compared using the Mann- Whitney non-parametric test. Virus isolates were pas- saged in MT-4 cells in the presence of RO33-4649 and the emergence of resistant populations was analysed by sequencing of the outgrowing virus isolates.
RESULTS: In a panel of 49 clinical isolates 39 and 28 samples showed high-level resistance (FC>20) to ≥3 or ≥4 PIs, respectively. The number of samples with FC<5 and between 5 and 20 were 11 and 30 for RO033- 4649, 11 and 26 for amprenavir, 0 and 17 for indi- navir, 0 and 5 for nelfinavir, 3 and 2 for saquinavir, and 0 and 12 for lopinavir. Isolates with RO033-4649 FC>20 (n=9) carried on average 14 protease muta- tions, with 2.67 primary mutations (primary muta- tions at positions 30, 46, 48, 50, 82, 84, 90. There was a significant positive correlation between RO033-4649 FC and the presence of mutations at positions 10, 33, 46, 54, 54, 71 and 73. For example, isolates contain- ing mutations at position 73 (C, S or T) had a median FC of 17 compared to 7.6 for isolates lacking such mutations (P=0.003). A negative effect was observed in the five samples with V82F or T: compared to samples without a mutation at position 82, median FC was lower (3.9 vs 13.5, P=0.035). The two isolates with an
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S21 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 19 (n=19). The primary mutation M46I/L had a high fre- quency for all mutational pathways (67% with HIV clinical isolates containing V32I+I47V, 37% with I54L/M and 74% with I50V). Certain secondary mutations were also present at an mutations representative of those elevated incidence of >25%, including L63P for the selected after first-line failure with V32+I47V mutations; L10V, M36I, L63P and V77I for unboosted GW433908 remain the I54L/M mutations; and L33F, L63P, A71V, V77I and V82I for the I50V mutation. For the viruses iden- sensitive to other protease tified, and in the presence of these additional protease inhibitors mutations, viruses with V32I+I47V, I54L/M and I50V, had median fold-changes to amprenavir of 3.4-, 3.6- L Ross1, N Parkin2, C Chappey2, M Tisdale3 and and 20-fold, respectively. For V32I+I47V, I54L/M and R Elston3 I50V, the median fold-change for saquinavir (0.5, 1.0 and 1.2, respectively) and indinavir (2.5, 1.3 and 1.0, 1 GlaxoSmithKline, RTP, NC, USA; 2 ViroLogic, Inc. Calif., GSK; respectively) were below the assay cut-off. The median and 3 GlaxoSmithKline, Stevenage, UK fold-changes were also below the clinical cut-off for lopinavir (2.9, 1.5 and 8.2, respectively). With respect BACKGROUND: GW433908 (908) is an investiga- to nelfinavir, samples were sensitive or had low-level tional protease inhibitor (PI) with demonstrated antivi- resistance, with fold-changes of 3.0, 3.6 and 2.2. Low- ral efficacy, durability and tolerability in antiretroviral level cross resistance was also seen for ritonavir, with (ART)-naive and -experienced subjects. Although no fold-changes of 3.4, 2.9 and 6.8. protease resistance-associated mutations were selected during 48 weeks of treatment with ritonavir-boosted CONCLUSION: Clinical isolates with mutational pat- 908 QD (SOLO trial, n=322) protease mutations were terns similar to those selected by unboosted 908 detected infrequently with unboosted 908 twice-daily remain sensitive to most other PIs, suggesting that (NEAT trial, n=166). In the NEAT trial, the predomi- viruses present after treatment failure of an unboosted nant protease mutations selected were the V32I+I47V 908 regimen will respond to second-line PI-containing (n=4) and I54L or M (n=4) mutations, consistent with therapy. an amprenavir-like resistance profile for unboosted 908. In one subject, the I54L was replaced with the L33F+I50V mutations at a later time-point, after con- tinued treatment with unboosted 908 in the presence of detectable viral load. To understand the potential cross resistance to other PIs, clinical isolates containing the V32I+I47V, I54L/M or I50V mutations were select- ed from a database of patient samples and susceptibil- ity to all marketed PIs was assessed.
METHODS: Approximately16000 HIV clinical sam- ples from the ViroLogic database, with matching geno- types and phenotypes, were examined to identify sam- ples with the following primary mutational pathways : V32I+I47V, I54L/M or I50V. Samples with mixtures at these specific amino acids were excluded, as were sam- ples that also contained certain primary protease muta- tions (D30N, G48V, V82A/T/S/F, I84V or L90M) that would indicate prior exposure to PIs. The presence of the primary protease mutation M46I/L was included because this mutation had been observed with the V32I+I47V (n=1) or I54L/M (n=1) mutations in NEAT. The median fold-change in susceptibility was calculated for samples with the specified mutation(s).
RESULTS: Fifty clinical isolates were identified: V32I+I47V (n=12), I54L/M (n=13/n=6) and I50V
S22 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 20 resistance to several PIs, including cyclic urea. Eight additional viruses were used to better define the phe- Antiviral activity of P-1946, a notypic resistance profile of P-1946 using the PhenoSense assay (ViroLogic, Inc.). The compound novel anti-HIV protease inhibitor remained fully active in the presence of mutation at position 30 or 50, which are associated with resistance G Sévigny1, B Tian1, A Dubois1, B Stranix1, to nelfinavir and amprenavir, respectively. Significant G Sauvé4, C Petropoulos2, Y Lie2, N Hellmann2, resistance to P-1946 required the presence of at least B Conway3 and J Yelle1 six mutations in the protease gene.
1 Pharmacor, Inc., Laval, Canada; 2 ViroLogic, Inc., South San CONCLUSION: P-1946 is an amino acid derivative Francisco, Calif., USA; 3 UBC, Department of Pharmacology & typical of a new family of PIs. Its antiviral activity pro- Therapeutics, Vancouver, Canada; and 4 INRS-Institut Armand- file makes it a good lead compound for the develop- Frappier, Laval, Canada ment of new potent agents that would offer therapeu- tic alternatives for individuals carrying isolates resis- BACKGROUND: The use of protease inhibitors (PIs) tant to current PIs. in the treatment of HIV infection has led to significant improvement in AIDS-related morbidity and mortality. Unfortunately, the rapid emergence of HIV strains resistant to currently available antiretroviral drugs threatens to make obsolete the current therapeutic approaches. Therefore, novel PIs with distinct resis- tance profiles are needed to help resolve this dilemma.
METHODS: PIs derived from an amino acid were pre- pared using a simple, straightforward synthesis scheme developed in our laboratory. Antiviral activity (EC ) 50 of the candidate compounds was determined for wild- type and mutant viral strains, using MT-4 as the cell line. The assay was based on the inhibition of HIV- induced cytopathic effect, measuring cell viability by MTT colorimetric assay. The threshold for phenotypic resistance to the test compound was defined as fold resistance (EC /EC )>4.0. 50test 50wt
RESULTS: A novel family of amino acid derivatives was readily obtained in few synthetic steps using clas- sic chemistry. The five compounds (P-1933, P-1935, P-1939, P-1946, P-1999) with the highest antiviral activity against the wild-type HIV-1 NL4.3 virus (EC <400 nM) also remained active against two typi- 50 cal strains carrying mutations 48/90 (saquinavir-resis- tant strain) and 10/46/63/82/84 (strain 4596). Compound P-1946 was selected as a prototype of the family and further characterized. P-1946 displayed good antiviral activity against wild-type strain NL4.3 (EC =150 nM). Cytotoxicity (CCIC ) of P-1946 was 50 50 40 µM, in the same range as currently marketed PIs. In order to define the resistance profile of P-1946, we analysed the antiviral activity in the presence of pro- tease mutations at amino acids 10, 46, 48, 63, 82, 84 and 90, using HIV strains obtained from the NIH. In addition to saquinavir-resistant and 4596 strains, P- 1946 was active (fold resistance <4) in the presence of combined mutations 82/84, which are associated with
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S23 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 21 for paired analysis. In four patients, there was a >four- fold increase in T-1249 FCIC ; plasma virus from 50 Baseline and on-treatment gp41 these patients demonstrated genotypic substitutions in amino acid 36–45. Virological response was not asso- genotype and susceptibility to ciated with viral tropism, baseline HIV RNA or base- line FCIC to ENF or T-1249, but was associated with enfuvirtide (ENF) and T-1249 in a 50 10-day study of T-1249 in patients length of time receiving a failing ENF-containing regi- men and with day 11 fold change from BL in T-1249 failing an ENF-containing regimen FCIC . (T1249-102) 50 CONCLUSIONS: T-1249 retains antiviral activity in GD Miralles1, T Melby1, R DeMasi1, Y Zhang1, most patients experiencing viral replication in the pres- R Spence1, N Cammack2, TJ Matthews1 and ence of isolates with reduced susceptibility to ENF M Greenberg1 and/or changes in the target region of ENF. Treatment emergent amino acid substitutions in gp41 and 1 Trimeris, Inc., Durham, NC; and 2 Roche, Palo Alto, Calif., USA reduced susceptibility to T-1249 were identified in some patients. BACKGROUND: T-1249 is a second-generation fusion inhibitor that has shown potent in vitro antivi- ral activity against most HIV isolates resistant to enfu- virtide (ENF). Study T1249-102 evaluated the short- term antiretroviral activity of T-1249 in patients failing a regimen containing ENF. Here we present the base- line and day 11 results of genotypic and phenotypic testing, and their correlation to treatment responses, for the first 25 patients included in the study’s planned interim analysis.
METHODS: Patients with two viral loads >5000 copies/ml while dosing on an ENF-containing regimen discontinued ENF and added 192 mg/day of T-1249 subcutaneously to the unchanged background regimen for 10 days. The intent-to-treat population included patients with amplifiable plasma virus at baseline (BL) that demonstrated ENF-resistance mutations and/or decreased phenotypic susceptibility to ENF. Resistance data were generated on Env amplified from patient plasma samples using the novel GeneSeq and PhenoSense Entry Assays. Fold changes in ENF and T-1249 IC were calculated in relation to reference 50 strains tested in parallel with the patient samples (FCIC ). 50
RESULTS: Plasma virus from 24 of 25 patients (98%) who entered T1249-102 exhibited ENF resistance- associated substitutions in gp41 amino acids 36–45 at BL. ENF IC were available for viral envelopes from 50 23 (96%) patients at BL. Geometric mean (GM) BL ENF FCIC was 150.1 (range 1.7–2041.6) and T- 50 1249 FCIC was 1.8 (range 0.14–12.6). For those 50 patients with paired samples at BL in their ENF parent study and at BL for the T1249-102 study (n=13), there was a GM increase of 70.6- and 1.8-fold in FCIC for 50 ENF and T-1249, respectively. On day 11, 22 (92%) patients had both genotype and phenotype available
S24 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 22 the other hand, the major determinants of ENF sensi- tivity tracked with the gp41 donor. Chimeric env con- Determinants of susceptibility to structs exchanging either the N-terminal ectodomain (containing HR1) or the C-terminal ectodomain enfuvirtide map to gp41 in through the end of gp41 (containing HR2) suggested enfuvirtide-naive HIV-1 that both regions can contribute to baseline suscepti- bility to ENF. In support of this assertion, we found SA Stanfield-Oakley1, J Jeffrey1, CB McDanal1, that ENF sensitivity can be modulated by SDM of S Mosier1, L Talton1, L Jin1, P Sista1, gp41 amino acids 45 (within HR1) and 135 (within N Cammack2, TJ Matthews1 and ML Greenberg1 HR2) with amino acids observed in rare FI-naive iso- lates or those more commonly found. 1 Trimeris, Inc., Durham, NC; and 2 Roche, Palo Alto, Calif., USA CONCLUSIONS: Previous studies have demonstrated BACKGROUND: Enfuvirtide (ENF, formerly T-20) is that the HR1 region of the HIV-1 gp41 is the target for the first fusion inhibitor to demonstrate efficacy in ENF. In addition, results from Phase III clinical studies controlled Phase III trials. Fusion inhibitor (FI)-naive of ENF have shown that this same region is the prima- HIV-1 isolates exhibit a range of susceptibility to ENF ry locus for development of ENF resistance. Our cur- in vitro, and several studies have suggested that HIV-1 rent results suggest that gp41 also contains the major envelope co-receptor tropism or affinity may con- determinants for baseline sensitivity to ENF of clade B tribute to this range. Co-receptor tropism and affinity FI-naive virus. are conferred by the gp120 subunit of the HIV-1 enve- lope glycoprotein, thus, those studies imply that deter- minants of susceptibility to ENF lie within gp120. Using chimeric envelope constructs and site-directed mutagenesis, we examined the envelope gp120 and gp41 subunits as loci for determinants of ENV suscep- tibility.
METHODS: Sensitivity of HIV-1 isolates to ENF was determined in a cMAGI infectious centre assay. HIV-1 env genes from R5- and X4-tropic isolates exhibiting a range of ENF sensitivities were cloned into an expres- sion vector and served as the starting point for gp120- gp41 chimeric envelope constructs. Co-receptor tro- pism and sensitivity of envelope clones and chimeras to ENF were determined from Env-deficient reporter viruses pseudotyped with these envelopes following cotransfection of envelope expression vectors and an env-deficient NL4-3-based reporter virus construct into 293T cells. The pseudotyped reporter viruses pro- duced were evaluated on U87 cells expressing CD4 and either CCR5 or CXCR4. Sequences of all cloned and chimeric envelope constructs were determined by dideoxy sequencing chemistries using a Beckman Coulter CEQ 2000XL system and DNAStar software.
RESULTS: Full-length functional envelopes were cloned from five R5-tropic and two X4 tropic clade B primary virus isolates of HIV-1. The cloned envelopes exhibited ENF IC s ranging 0.04–12.6 µg/ml in the 50 U87-based pseudotyped reporter virus assay. The cloned envelopes retained the same tropism character- istics noted with the parental virus isolates. The gp120/gp41 chimeric envelopes exhibited tropism specificity of the gp120 parental virus as expected. On
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S25 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 23 drug-resistant R5 clinical isolates, suggesting that viruses selected in vivo during HIV drug treatment Sensitivity of Env-gene retain sensitivity to UK-427,857. The study supports the continued development of this compound for the recombinant viruses derived from treatment of HIV-infected individuals. antiretroviral drug-sensitive and -resistant HIV-1 clinical isolates to the novel CCR5 antagonist, UK-427,857
M Westby1, C Napier1, R Mansfield1, D Collins1, W Huang2, N Hellmann2, Y Lie2 and M Perros1
1 Pfizer Global Research and Development, Sandwich Laboratories, Kent. UK; and 2 ViroLogic, Inc., South San Francisco, Calif., USA
UK-427,857 is a novel small molecule CCR5 antago- nist that is currently being developed for the treatment of HIV infection. Its antiviral activity was evaluated against 200 clinically-derived R5 HIV-1 isolates using the ViroLogic PhenoSense HIV Entry inhibitor suscep- tibility assay. A panel of recombinant viruses was pre- pared using HIV-1 gp160 envelope genes derived from 100 clinical isolates lacking known ‘drug-selected’ mutations in either protease (PR) or reverse transcrip- tase (RT) (‘drug-sensitive’ isolates) and 100 clinical isolates with one or more drug selected PR or RT mutations (‘drug-resistant’ isolates). The virus panel comprised 160 clade B and 40 isolates from other clades.
UK-427,857 inhibited all 200 recombinant viruses tested with a geometric mean IC of 1.6 nM (range 50 0.3–8.9 nM). The geometric mean IC s derived from 50 drug-sensitive and drug-resistant isolates were 1.3 nM (range 0.3–3 nM) and 2.1 nM (range 0.7–8.9 nM), respectively. The difference between these two groups (1.6-fold) is statistically significant (P<0.05) but less than the expected assay-to-assay variation. There was no difference in sensitivity between the clade B and non- clade B isolates, with geometric mean IC of 1.6 nM 50 (range 0.5–6.9 nM) and 1.6 nM (range 0.3–8.9 nM), respectively, consistent with a previous study demon- strating broad cross-clade activity of UK-427,857 against primary isolates grown in mitogen-activated peripheral blood lymphocytes.
These data provide further evidence that UK-427,857 is a potent antiviral compound with broad activity against recombinant viruses derived from a large num- ber of clinically-relevant isolates and diverse clades. The compound inhibited CCR5-mediated infection of Env-recombinant viruses derived from antiretroviral
S26 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 24 ly act on gp41-dependent fusion. Our results support the hypothesis that ADS-J1 binds to a hydrophobic ADS-J1, a non-peptidic low cavity region within gp41 for preventing fusion-active gp41 core formation. ADS-J1 may serve as a lead low molecular weight HIV fusion molecular weight compound to develop new anti-HIV inhibitor targeting gp41, with no agents. cross-drug resistance with peptidic HIV fusion inhibitors T-20 and C-34, and HIV binding inhibitors
M Armand-Ugón1, A Gutiérrez1, S Jiang2, B Clotet1 and JA Esté1
1 Retrovirology Laboratory irsiCaixa, Hospital Universitari Germans Trias i Pujol, Universitat Autònoma de Barcelona, Badalona, Spain; and 2 Lindsley F Kimball Research Institute, The New York Blood Center, New York, NY, USA
BACKGROUND: ADS-J1 is a low molecular weight compound selected for its ability to interfere with the association of the N- and C-terminal heptad repeat regions of HIV-1 gp41 envelope glycoprotein. Since ADS-J1 is a polysulfonic acid compound, it is interest- ing to know whether it, like other polyanionic anti- HIV compounds, blocks the binding of HIV-1 to CD4 cells through electrostatic interactions and to test the anti-HIV activity of ADS-J1 against HIV strains that have been made resistant to polyanionic HIV binding inhibitors.
METHODS: Evaluation of anti-HIV activity in MT-4 cell culture against wild-type and drug-resistant HIV-1 strains. Cell culture selection of HIV drug resistance to known inhibitors of gp41-dependent fusion (T-20 and C-34). Flow cytometry evaluation of drug interaction with HIV co-receptors.
RESULTS: Here, we show that ADS-J1 was active against T-20- and C-34-resistant HIV-1 isolates with similar potency to the wild-type HIV-1 NL4-3 strain (EC 0.6, 0.3 and 0.4 µg/ml, respectively). ADS-J1 50 (10 µg/ml) could not block the binding of an HIV strain that was made resistant to AR177, a negatively charged oligonucleotide that blocks HIV binding, and is cross-resistant (>100-fold) to dextran sulfate (DS), a known polysulfonic HIV binding inhibitor. However, ADS-J1 blocked AR177-resistant virus fusion and replication (EC 1.5 µg/ml), suggesting that ADS-J1 50 has a mechanism of action different from the polyan- ionic HIV binding inhibitor AR177.
CONCLUSION: If the activity of polyanionic com- pounds on HIV binding is ‘bypassed’ by selection of resistance, compounds such as ADS-J1 may exclusive-
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S27 Los Cabos, Mexico, 10–14 June 2003
ABSTRACTS 25 via a direct interaction with virions and through mod- ulation of the CXCR4 co-receptor. These properties Human β-defensins inhibit HIV-1 may be exploited as new strategies for mucosal protec- replication in vitro tion against HIV-1 transmission.
ME Quiñones-Mateu1,2, MM Lederman2, Z Feng2, B Chakraborty1, J Weber1, HR Rangel1, ML Marotta1, M Mirza1, B Jiang2, P Kiser1, K Medvik2 and A Weinberg2
1 Cleveland Clinic Foundation, Cleveland, OH; 2 Case Western Reserve University, Cleveland, OH, USA
BACKGROUND: Mechanisms underlying the infre- quent transmission of HIV-1 through the oral mucosa are incompletely understood. Here we describe the anti-HIV-1 activity of human β-defensins (hBDs), small cationic innate defence molecules that provide a first line of protection at mucosal surfaces.
METHODS: Recombinant hBD-1, -2, and -3 were generated and used to evaluate their anti-HIV-1 activi- ty and cytotoxicity in PBMC, CEM X4/R5 and Ghost X4/R5 cells. Real-time RT-PCR was used to quantify hBD mRNA expression in normal human oral epithe- lial cells. Flow cytometric and confocal microscopy analysis was used to determine the effect of hBDs over CCR5 and CXCR4 co-receptors. Finally, the potential interaction between hBDs, cell membrane and viral particles was analysed by immunogold transmission electron microscopy.
RESULTS: HIV-1 induced expression of hBD-2 and -3 mRNA, but not hBD-1, four- to-78-fold above base- line, in normal human oral epithelial cells. HIV-1 failed to infect normal human oral epithelial cell monolayers, even after 5 days of exposure. While recombinant hBD-1 had no antiviral activity, both rhBD-2 and -3 showed a concentration-dependent inhibition of HIV-1 replication without cellular toxicity. This antiretroviral effect was greater against the CXCR4-tropic than against CCR5-tropic HIV-1 isolates. HBD-2 and -3 bound to virions and induced a direct and irreversible effect on virion infectivity, with electron microscopy confirming direct binding of hBDs to viral particle. In addition, hBD-2 and -3, but not hBD-1, induced down-modulation by internalization of the HIV-1 co- receptor CXCR4 (but not CCR5) in peripheral blood mononuclear cells and T-lymphocytic cells, as shown by confocal microscopy and flow cytometric.
CONCLUSION: This study shows for the first time that (i) HIV-1 induces hBD expression in human oral epithelial cells; and (ii) hBDs block HIV-1 replication
S28 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 26 CONCLUSION: Resistance development of HIV-1 In vitro resistance development of against the mannose-specific plant lectins GNA and HHA is associated with a unique spectrum of resis- human immunodeficiency virus tance mutations in the gp120 envelope gene, which has type 1 towards mannose-specific not been previously observed with any of the other plant lectins known viral entry inhibitors.
K Van Laethem1,2, Y Schrooten1,2, S Hatse1, K Vermeire1, E De Clercq1, W Peumans3, E Van Damme4, D Schols1, A-M Vandamme1,2 and J Balzarini1
1 Rega Institute for Medical Research, Katholieke Universiteit Leuven; 2 AIDS Reference Laboratory, University Hospitals Leuven; 3 Department of Fytopathology, Katholieke Universiteit Leuven; and 4 Department of Molecular Biotechnology, Rijksuniversiteit Gent, Belgium
BACKGROUND: Mannose-specific plant lectins from the Amaryllidaceae family, such as Galanthus nivalis agglutinin (GNA) and Hippeastrum hybrid agglutinin (HHA), have previously been shown to exhibit pro- nounced anti-HIV activity (Balzarini et al., 1991). There is strong evidence that these drugs target the heavily glycosylated envelope glycoprotein gp120. Therefore, the aim of this study was to select for resis- tance against GNA and HHA, to determine the phe- notype and the associated genotypic changes in the HIV envelope gene.
METHODS: HIV-1 was subjected to sub-cultivations in the presence of dose-escalating concentrations of GNA and HHA. Virus isolates, recovered in the presence of different concentrations of the respective drugs, were investigated for their phenotypic susceptibility towards a subset of mannose-specific plant lectins, and a variety of viral entry and reverse transcriptase inhibitors. Their gp120 and gp41 gene sequences were also determined.
RESULTS: Many sub-cultivation steps were required to obtain virus strains that could grow in the presence of drug concentrations that were at least 100-fold higher than the initial EC (50% effective concentra- 50 tion) values. The mutations found in the different virus isolates were predominantly related to N-glycosylation sites and at the S or T residues that are part of the N- glycosylation motif. One potential O-glycosylation site in gp120 was also affected. The degree of resistance of HIV-1 to the plant lectins was correlated with an increasing number of mutated glycosylation sites in gp120. The susceptibility of the GNA- and HHA-resis- tant virus strains towards other viral entry and RT inhibitors was fully preserved.
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S29 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 27 have antiviral immune responses as compared to the controls. HIV that was selected with BCD in vitro was Viral resistance against a candidate shown to be resistant to complete inactivation by BCD. Sequences of these viruses show mutations as HIV microbicide compared to the same virus without BCD exposure.
Z Ambrose1, CJ Miller2, L Compton2, CONCLUSIONS: Should BCD continue to prevent SH Hughes1, JD Lifson3 and VN KewalRamani1 SIV transmission and not perturb mucosal tissues in this model, its current approved use in humans sug- 1 HIV Drug Resistance Program, National Cancer Institute, gests it would be an important candidate for use as an Frederick, Md.; 2 California Regional Primate Research Center, anti-HIV microbicide. Nonetheless, the possibility of Davis, Calif.; and 3 AIDS Vaccine Program, National Cancer drug resistance against microbicides should be careful- Institute, Frederick, Md., USA ly evaluated before such drugs are administered to humans. Systemic antiviral therapy to manage HIV BACKGROUND: Sexual transmission accounts for infection indicates that combinatorial approaches have greater than 90% of worldwide HIV infection. significant benefits over monotherapy. Thus, we are Moreover, the incidence and prevalence of HIV infec- also evaluating new models in which to examine the tion in women has been increasing. Vaginal microbi- efficacy of BCD in combination with other potential cides provide a female-controlled strategy to prevent microbicides. HIV transmission in women. We have evaluated an HIV inactivating agent, 2-hydroxypropyl-beta- cyclodextrin (BCD), as a potential microbicide. Significantly, BCD recently has been proven as an effective microbicide in a mouse model for intravaginal HIV-1 transmission and is used extensively for other purposes in individuals.
METHODS: First, we evaluated the efficacy of virus neutralization in vitro, using a single cycle replication assay with HIV or SIV in the presence or absence of BCD. Based on encouraging in vitro data, we adminis- tered BCD intravaginally in rhesus macaques, followed by inoculation with large doses of highly pathogenic SIV. Control animals were treated with gel alone or nothing before inoculation with the same dose of SIV. The animals have been evaluated for infection by RT- PCR of gag sequences in their plasma and nested PCR of gag sequences in their PBMC. These animals con- tinue to be monitored for antiviral humoral and cellu- lar immune responses. In addition, we selected differ- ent strains of HIV-1 in vitro in the presence of escalat- ing doses of BCD.
RESULTS: BCD was successful in neutralizing infec- tion of cells with both X4- and R5-tropic HIV as well as SIV in vitro. In fact, we could not see outgrowth of virus after 30 days post-exposure. Our in vivo data indicated that intravaginal pretreatment with BCD sig- nificantly reduced SIV mucosal transmission (1/6 ani- mals infected) relative to untreated control animals (8/10 animals infected). Currently we are performing repeated challenges with BCD and SIV in the uninfect- ed animals to assess whether they continue to be pro- tected from infection. In addition, we are determining whether or not the uninfected BCD-treated animals
S30 SESSION 2 Mechanisms of HIV Drug Resistance
Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 28 CONCLUSIONS: Enzymes containing PFA resistance Mechanisms involved in zidovudine conferring mutations alter the precise positioning of RT on its nucleic acid substrate. Such displacement hypersusceptibility in the presence diminishes the unblocking of ZDV-terminated primer of foscarnet resistance-conferring strands. The high concentrations of dNTPs that are mutations required to force translocation also helps to explain the diminished rates of DNA synthesis. These parame- ters may directly correlate with ZDV resensitization B Marchand and M Götte effects and diminished viral replication fitness associ- ated with viruses that contain the E89K mutation. McGill University/Lady Davis Institute, Montreal, Quebec, Canada
BACKGROUND: Phosphonoformate (PFA, foscarnet) binds at or in close proximity to the nucleotide binding site of HIV-1 reverse transcriptase (RT) and inhibits DNA synthesis presumably through interference with the exchange of pyrophosphate. Like other PFA-resis- tance conferring mutations, E89K is distal from the dNTP binding site and interacts with the template strand. This mutation is also associated with increased susceptibility to zidovudine (ZDV). We hypothesized that this effect might be attributable to alterations with regard to the precise alignment of ATP, that acts as a pyrophosphate donor and the primer/template sub- strate.
METHODS: To address this problem, we developed novel site-specific footprinting techniques that allowed us to monitor the position of RT on its template at sin- gle nucleotide resolution. We used different sources of hydroxyl radicals that promote site-specific cleavage on the bound template, and compared the cleavage patterns between wild-type RT and the mutant enzyme under different reaction conditions.
RESULTS: Wild-type RT promotes cleavage at posi- tions –8 and –18 in the absence of the incoming dNTP, when using a substrate that contained a ZDV-termi- nated primer. This is the configuration that allows excision of the incorporated ZDV-monophosphate. The presence of the next dNTP forces the enzyme to translocate a single position further downstream, as evidenced by cleavage at positions –7 and –17. The presence of PFA diminishes the translocation of RT, which provides a novel mechanism for drug action. The E89K mutation appears to alter the relative posi- tion of RT. The mutant promotes cleavage at positions –9 and –19 in the absence of the incoming dNTP. Excision of ZDV cannot occur in this configuration, which helps to explain earlier findings that pointed toward diminished rates of primer unblocking associ- ated with resistance to PFA. Moreover, relatively high concentrations of the next nucleotide are required to force the translocation of RT.
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S33 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 29 suppressing zidovudine and tenofovir resistance. For example, median zidovudine and tenofovir FC for HIV-1 reverse transcriptase samples with L100I+M184IV (n=28) were 3.8- and mutations that suppress zidovudine 0.8-fold, and for Y181I/C/V+M184I/V (n=174) were 10- and 1.1-fold, respectively. Lower FC in groups resistance also increase in vitro with suppressive mutations could not be explained by susceptibility to tenofovir, but not fewer thymidine analogue mutations (median number stavudine 3 to 4 for all groups). As expected, FC for didanosine, zalcitabine (ddC) and abacavir was higher in groups NT Parkin, C Chappey, CJ Petropoulos and containing L74I/V and/or M184I/V, and was not sig- N Hellmann nificantly affected by L100I or Y181C/I/V. CONCLUSIONS: M184I/V increases susceptibility to ViroLogic, Inc., South San Francisco, Calif., USA zidovudine, tenofovir and stavudine. Other suppres- sive mutations in RT affect tenofovir and zidovudine, BACKGROUND: Susceptibility of HIV-1 to zidovu- but not stavudine. Susceptibility to stavudine dine is increased by at least five mutations in reverse decreased in the presence of L74I/V and Y181I/C/V. transcriptase (RT): K65R, L74V, L100I, Y181C and Additive effects were observed when suppressive muta- M184V. In some cases these ‘suppressive mutations’ tions were present together. Combined mutations were restore susceptibility to zidovudine despite of the pres- capable of re-sensitizing tenofovir (FC<1.4) and ence of resistance mutations such as K70R or T215F/Y. zidovudine (FC<2.5) in the presence of multiple thymi- In general, susceptibility to stavudine and tenofovir is dine analogue mutations. Since genotype interpreta- modulated by RT resistance mutations in a manner tion algorithms do not account for the effects of most that qualitatively parallels that of zidovudine. Thus, suppressive mutations, these observations provide an we evaluated the effect of suppressive mutations on explanation for phenotype/genotype discordance for susceptibility to stavudine and tenofovir. zidovudine and tenofovir.
METHODS: A clinical sample database of over 16000 matched genotypes and phenotypes was queried for samples containing T215F or Y without, or with one or more, suppressive mutations (K65R, L74I and V, L100I, Y181C, I, and V, and M184I and V; different variants at each position were grouped together). Samples containing multi-nucleoside RT inhibitor (NRTI) resistance mutations (T69ins or Q151M) were excluded, as were samples with mixtures at positions that were part of the query. NRTI fold change (FC) in IC vs NL4-3 reference was compared between groups 50 of viruses using the Mann-Whitney non-parametric test.
RESULTS: Median zidovudine FC for T215Y/F sam- ples with no suppressive mutations (n=966), L74I/V alone (that is, no other suppressive mutations, n=87), L100I alone (n=54), Y181I/C/V alone (n=283) or M184I/V alone (n=1423) was 145-, 102-, 78-, 75- and 12-fold, respectively (P<0.05 for each suppressive mutation group vs no suppressive mutations). The cor- responding FC values for tenofovir were 2.8-, 2.2-, 2.1-, 2.4- and 1.3-fold, respectively (all P<0.05), and for stavudine were 2.4-, 3.3-, 2.5-, 2.8- and 1.8-fold, respectively (all P<0.05 except L100I). The number of samples with K65R and T215F/Y was too low to pro- vide meaningful comparisons. Combinations of two or more suppressive mutations were generally additive in
S34 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 30 low. The data suggest that one or more of the ∆67, T69G, L74I or K103N mutations contribute to the The ∆67 complex of mutations ability of the mutant RT to excise AZT at low ATP enhances the ability of HIV-1 concentrations. reverse transcriptase to excise zidovudine, stavudine and PMPA from blocked primers
PL Boyer1, T Imamichi2, SG Sarafianos3, E Arnold3 and SH Hughes1
1 HIV Drug Resistance Program, NCI-Frederick, Frederick, Md.; 2 Clinical Services Program, SAIC-Frederick, Inc., Frederick, Md.; and 3 Center for Advanced Biotechnology and Medicine and Rutgers University Chemistry Department, Piscataway, NJ, USA
BACKGROUND: Most patients that receive anti-HIV- 1 therapy are treated with drug combinations. Many patients are treated with multiple nucleoside analogues (NRTIs). Viruses that are resistant to multiple NRTIs are an increasing problem. Insertions in the fingers of reverse transcriptase (RT) (most often between amino acids 69 and 70), in combination with T215F/Y, caus- es enhanced excision of a number of NRTIs. The ∆67 complex (41L/∆67/T69G/K70R/L74I/K103N/T215Y/ K219Q) has been reported to cause resistance to a number of NRTIs and non-nucleoside inhibitors. We asked whether RT carrying the ∆67 complex muta- tions could efficiently excise NRTIs.
METHODS: We purified recombinant wild-type HIV- 1 RT, RT carrying the ∆67 complex mutations and RT carrying the classical zidovudine (AZT) resistance mutations. These RTs were used to investigate the mechanism(s) of NRTI resistance of the ∆67 complex RT.
RESULTS: The ∆67 complex RT was able to excise AZT, stavudine and PMPA more efficiently than either wild-type RT or AZT-resistant RT. Both the ∆67 com- plex and the AZT-resistant RT were able to excise PMPA much more efficiently than ddA. In addition, the ∆67 complex RT was able to excise AZT at much lower concentrations of ATP than AZT-resistant RT.
CONCLUSIONS: It would appear that PMPA is rela- tively susceptible to excision by RTs that carry the clas- sical AZT resistance mutations. PMPA excision is enhanced by RTs carrying the mutations in the ∆67 complex. The ability of RT carrying the ∆67 complex mutations to efficiently excise AZT at low ATP con- centrations might provide an advantage for the virus in quiescent cells, where ATP levels are expected to be
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S35 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 31 between wt and AZT-resistant RT. In contrast to the incorporation data, significant differences in the rates The 3′-azido group is not the of ATP-mediated phosphorolytic excision of the vari- ous 3′-azido-2′,3′-ddNMP were noted. The relative primary structural determinant for rates of ATP-mediated phosphorolysis by wt and AZT- the excision phenotype correlated resistant RT (in the absence of dNTP) were 3′-azido- with HIV-1 resistance to AZT 2′,3′-ddA>AZT>3′-azido-2′,3′-ddC>3′-azido-2′,3′- ddG. However, under reaction conditions that enabled multiple rounds of 3′-azido-2′,3′-ddNMP incorpora- N Sluis-Cremer1, D Koontz1, D Arion1, tion and excision, the AZT-resistant RT enzyme was U Parikh1, R Schinazi2, J Mellors1 and 15-fold more efficient at forming full length DNA MA Parniak1 products in the presence of AZTTP than the wt enzyme. In contrast, only a threefold difference 1 Division of Infectious Diseases, Viral Diseases Unit, University between mutant and wt RT was noted for reactions in of Pittsburgh School of Medicine, Pittsburgh, Pa.; and 2 the presence of 3-azido-2′,3′-ddCTP, and no differ- Laboratory of Biochemical Pharmacology, Emory ences between the two enzymes were noted for reac- University/Veterans Affairs Medical Center, Decatur, Ga., USA tions in the presence of 3′-azido-2′,3′-ddATP and 3′- azido-2′,3′-ddGTP. The antiviral activities of the vari- BACKGROUND: The phenotypic mechanism of HIV- ous 3′-azido-2′,3′-ddN were consistent with the enzy- 1 resistance to 3′-azido-2′,3′-dideoxythymidie (AZT) matic data, in that AZT-resistant virus was 10-fold has been proposed to involve the removal or excision resistant to AZT and fourfold resistant to 3′-azido- of the incorporated chain-terminating AZT molecule 2′,3′-ddC, but was not cross-resistant to either 3′- by a reverse transcriptase (RT)-mediated enzymatic azido-2′,3′-ddA or 3′-azido-2′,3′-ddG. reaction termed phosphorolysis. Previous studies have indicated that primers terminated with AZT- CONCLUSIONS: AZT resistance mutations do not monophosphate (AZT-MP) are better substrates for confer significant cross resistance of RT or virus to this reaction than those terminated with 2′-3′- other nucleosides having a 3′-azido group. dideoxynucleoside monophosphate (2′,3′-ddNMP) Furthermore, the presence of a 3′-azido group on the analogs that lack a 3′-azido moiety. This has lead to 3′-terminal nucleotide of the primer does not enhance the suggestion that the 3′-azido group may be a major phosphorolytic excision by AZT-resistant RT in vitro, structural determinant for maintaining the primer ter- suggesting that other structural factors must play a minus in the appropriate site to allow phosphorolytic role in defining the specificity of the excision pheno- excision by AZT-resistant RT. We tested this possibili- type arising from mutations correlated with AZT-resis- ty by carrying out detailed biochemical and virological tance. evaluations of the incorporation, phosphorolytic exci- sion and antiviral properties of a panel of 3′-azido- 2′,3′-ddN compounds including 3′-azido-2′,3′-ddA, 3′- azido-2′,3′-ddC, 3′-azido-2′,3′-ddG, 3′-azido-2′,3′-ddU and AZT.
METHODS: Steady-state and pre-steady-state kinetic parameters for the incorporation and ATP-mediated excision of the 3′-azido-2′,3′-ddNTPs were determined for purified wild-type (wt) RT and D67N/K70R/ T215F/K219Q AZT-resistant RT. Antiviral activities of the 3′-azido-2′,3′-ddN nucleosides were evaluated in MT2 cells using WT or D67N/K70R/T215Y/K219Q AZT-resistant virus.
RESULTS: Each of the 3′-azido-2′,3′-ddNTPs was an excellent substrate for DNA-dependent single nucleotide incorporation reactions catalysed by wt and AZT-resistant RT. The relative catalytic efficiencies of incorporation (k /K ) were 3′-azido-2′,3′-ddA > AZT pol d > 3′-azido-2′,3′-ddC, with no differences noted
S36 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 32 of these five viruses showed a complete reversion to phenotypic susceptibility to all drugs and loss of drug- Evolution of amino acid 215 in selected mutations at other positions in RT and pro- tease, strongly suggesting an outgrowth of archival HIV-1 reverse transcriptase in drug-sensitive viruses carrying a revertant codon at response to intermittent drug position 215. Twenty-eight viruses (21 T215Y and 7 selection T215F) showed complex mixtures of position 215. Mean RC was high for 215T, S, D, C (100, 120, 102, 98 respectively), and lower for 215V, L, I (62, 61, 40 C Chappey1, T Wrin1, S Deeks2 and respectively). CJ Petropoulos1 CONCLUSIONS: T215 revertants are common in 1 ViroLogic, Inc., South San Francisco, Calif.; and 2 UCSF AIDS HIV-1 patients. T215Y and F revert preferably to 215- Program, San Francisco, Calif., USA codons that are one nucleotide different (Y215S/D/C, F215S/V). The relationship between the variant at the BACKGROUND: HIV-1 reverse transcriptase (RT) first time point and at the second time point supports amino acid position 215 is monomorphic for threonine the model that the RC of the various variants deter- (T) in the ‘wild-type’ virus population. Thymidine ana- mines their relative presence in the archived virus pool logues such as zidovudine and stavudine select for drug and their subsequent emergence in the absence of selec- resistant variants containing tyrosine (215Y) or pheny- tive pressure. lalanine (215F). Two nucleotide changes are required to substitute T by Y or F. Intermediate alleles contain- ing single ‘forward’ mutations encoding for 215N/S/I are drug-sensitive. 215N/S/I variants can also appear as the result of single revertant mutations of T215Y/F when drug pressure is interrupted. Variants 215D/V/C/H/L generated by single revertant mutations of T215Y/F preserve the two ‘forward’ nucleotide changes from 215T. Our objective is to evaluate the emergence of T215 variants in (a) a large sequence database compiled from HIV-1 isolates submitted for routine antiretroviral drug resistance testing and (b) a small cohort of HIV-1-infected patients that participat- ed in a structured treatment interruption study.
METHODS: A database of matched genotypes and phenotypes from over 1000 clinical longitudinal sam- ples gathered from 1999 to 2003 was queried for sam- ples containing T215Y or F (not mixed) at the first time point. Correlation between the presence of T215Y or F mutation at the first and the 215-mutant at the second time point was tested using logistic regression. Mean replication capacity (RC) of each 215 substitution was calculated from protease wild-type samples. Ten longitudinal plasma samples were part of a supervised structured treatment interruption study (Deeks et al., New England Journal of Medicine 2001).
RESULTS: Fifty-five longitudinal pairs of viruses were identified with T215Y (n=44) or T215F (n=11) at the first time point. Twenty-two of these viruses (20 T215Y and 2 T215F) showed complete replacement by the wild-type 215T virus at the second time point. Five viruses showed complete replacement by revertant mutants: 215Y by S, D, and C, 215F by S and V. Two
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S37 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 33 tenofovir. The relative binding/incorporation of teno- fovir diphosphate was slightly decreased for FS-SSS RT Molecular mechanisms of resistance (2.8-fold) compared to wild-type, but not significantly for FS RT (1.7-fold). However, significant ATP-medi- to tenofovir by HIV-1 reverse ated excision of tenofovir was detected for both transcriptase containing a di-serine mutant RTs with FS-SSS > FS > wild-type, and excision insertion after residue 69 and rates of 3.5, 1.8 and 0.8% per min, respectively. The presence of physiological concentrations of the next nt multiple thymidine analogue- inhibited tenofovir excision by wild-type, slightly associated mutations inhibited excision by FS, whereas excision by FS-SSS remained high (12-fold greater than wild-type). KL White1, JM Chen1, NA Margot1, T Wrin2, Computer modelling shows that the insertion mutation CJ Petropoulos2, LK Naeger1, S Swaminathan1 could generate a more flexible β3-β4 fingers loop and MD Miller1 domain that would enhance excision by facilitating dissociation of the ATP-mediated excision product 1 Gilead Sciences, Foster City, Calif., USA; and 2 ViroLogic, Inc., and/or destabilizing the protective next nt complex. South San Francisco, Calif., USA CONCLUSIONS: Increased ATP-mediated excision of BACKGROUND: The insertion of two amino acids incorporated tenofovir without efficient inhibition by after residue 69 of HIV-1 reverse transcriptase (RT) is the next nucleotide appears to be the primary mecha- a rare mutation that may develop in viruses containing nism of tenofovir resistance for HIV-1 RT with T69 multiple thymidine analogue-associated mutations insertion mutations and multiple TAMs. Decreased (TAMs) and confers high-level resistance to all cur- binding/incorporation of tenofovir also makes a minor rently approved chain-terminating nucleoside and contribution to tenofovir resistance. The same TAMs nucleotide RT inhibitors (NRTIs). Decreased incorpo- without the insertion mutation showed detectable, but ration and increased excision are two known mecha- lower levels of excision and greater inhibition by the nisms of NRTI resistance. Excision for many NRTIs next nucleotide. Increased flexibility of the β3-β4 loop may be inhibited by binding of the nucleotide comple- by the insertion mutation may be the basis for the mentary to the next position in the template (next nt) high-level and broad NRTI cross resistance caused by to form a stable dead-end complex where the chain- the T69 insertion mutations. terminator is protected from excision. The mechanism of resistance to tenofovir for RT with an insertion mutation and multiple TAMs was examined in this study.
METHODS: A patient-derived HIV-1 virus (FS-SSS) was obtained that contained the SS insertion after residue 69 in a background of additional resistance mutations M41L, T69S, L74V, L210W and T215Y. The insertion and T69S were reverted by site-directed mutagenesis in a second virus (FS) that retained the other resistance mutations. In vitro drug susceptibility was determined by the PhenoSense assay. Incorporation of tenofovir was examined by measur- ing steady-state kinetic constants for wild-type and mutant RTs. ATP-mediated excision with or without the next nt was measured. Molecular models and mol- ecular dynamics simulations were produced using Sybyl software.
RESULTS: The multiple TAM-containing FS virus exhibited a large reduction in zidovudine susceptibility and smaller reductions in susceptibility to other NRTIs including tenofovir; the FS-SSS virus showed greater reductions in susceptibility to all NRTIs including
S38 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 34 CONCLUSIONS: Our data describe at the molecular Drug resistance and viral fitness at level both how a resistant virus is unable to resist to two drugs simultaneously, and for the first time, how the molecular level: the case of viral fitness of a resistant virus is directly linked to its tenofovir decreased ability to use natural nucleotide substrates. All together, these data predict a benefit for the combi- J Deval1, KL White2, MD Miller2, nation of tenofovir DF with 3TC, as well as open new J Courcambeck3, B Selmi1, J Boretto1 and avenues in how to drive resistant virus to reduced viral B Canard1 fitness.
1 CNRS, Marseille, France; 2 Gilead, Foster City, USA; and 3 Genosciences, Marseille, France
BACKGROUND AND METHODS: The selection of highly resistant HIV-1 mutant by tenofovir disoproxil fumarate (TDF), the most recently approved antiretro- viral drug, is unfrequent. An integrated study of resis- tance to tenofovir and its active metabolites was per- formed using infected cell, biochemical assays, and 3D- structure modelization to determine resistance mecha- nisms and their impact on viral fitness.
RESULTS: K65R, the most significant tenofovir resis- tance variant in reverse transcriptase (RT) to date shows only a limited resistance to the active intracellu- lar metabolite of tenofovir, TFV-DP (4.4-fold resis- tance), as judged by pre-steady state analysis. This resis- tance is due to a decrease in k (TFV-DP): from 7 s-1 for pol wild-type RT to 0.32 s-1 for K65R RT. On the other hand, M184V RT, the lamivudine (3TC)-resistant mutant, displays a 2.5-fold increase in susceptibility to TFV-DP. When the two mutations are combined, the susceptibility of K65R/M184V RT to TFV-DP is simi- lar to that of wild-type RT. Molecular modelling stud- ies show that the methyl group of tenofovir makes a favourable Van der Waals interaction with valine 184, explaining TFV-DP resensitization with M184V. Cell culture susceptibility data corroborate the observation of resensitization as the K65R/M184V HIV double mutant shows only 1.7-fold reduced susceptibility to tenofovir. Interestingly, the decrease in resistance to tenofovir with the double mutant is associated with a fivefold decrease in the affinity of K65R/M184V RT for the natural nucleotide substrate dATP. The incor- poration of all four natural nucleotides was thus inves- tigated. We observed that each of them was poorly incorporated by K65R/M184V RT, with a 27% global incorporation efficiency as compared to wild-type RT. The M184V substitution was found to contribute to the diminished binding of the natural nucleotides to RT, whereas K65R affects catalysis generally. In vitro replication capacity assays using recombinant HIV also show that K65R/M184V is less fit than wild-type HIV, in agreement with our enzymatic data.
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S39 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 35 Molecular mechanism for the mutual exclusion of K65R and L74V substitutions in HIV-1 reverse transcriptase-mediated dideoxynucleoside resistance
J Deval1, J-M Navarro2, B Selmi1, J Courcambeck3, J Boretto1, P Halfon3, J Sire2 and B Canard1
1 AFMB-CNRS, Marseille; 2 INSERM U-372, Marseille; and 3 Genosciences, Marseille, France
BACKGROUND: Either K65R or L74V in human immunodeficiency virus type 1 reverse transcriptase (HIV-1 RT) confer resistance to 2′,3′-dideoxyinosine (ddI) in vivo. The two substitutions never occur together, while L74V is frequently found in patients receiving ddI and K65R is not.
METHODS: This integrated study involves in vitro infection assays with recombinant viruses, combined to a molecular and structural approach based on enzy- matic assays with purified RTs. We investigated drug resistance and replication capacity with both methods.
RESULTS: Recombinant viruses carrying K65R and K65R/L74V display the same resistance level to ddI (about 9.5-fold) relative to wild-type. Consistent with this result, purified HIV-1 RT carrying K65R RT or K65R/L74V substitutions exhibits an eightfold resis- tance to ddATP. Resistance is due to a selective decrease of the catalytic rate constant k : 22-fold pol (from 7.2 to 0.33 s-1) for K65R RT and 84-fold (from 7.2 to 0.086 s-1) for K65R/L74V RT. However, the K65R/L74V virus replication capacity (viral fitness) is severely impaired relative to that of wild-type virus. This loss of viral fitness is due to a poor ability of K65R/L74V RT to use natural nucleotides relative to wild-type RT: 15% that of wild-type RT for dATP, 36% for dGTP, 50% for dTTP, and 25% for dCTP. The order of incorporation efficiency is wild-type RT>L74V RT>K65R RT>K65R/L74V RT. Processivity of DNA synthesis, however, remains unaffected.
CONCLUSION: These results explain why the two mutations K65R and L74V do not combine in the clin- ic, and give a mechanism for a decreased viral fitness at the molecular level. This study gives rational sup- port to the benefit in combining mutations that impair viral replication.
S40 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 36 dioxolane-T, conformational change in sugar moiety of the primer residue (from 3′-exo to 3′-endo) in 5-F- Mechanism of anti-HIV activity of dioxoloane-C and rotation of Val184 into the nucleo- side triphosphate binding site with concurrent confor- dioxolane nucleosides against mational change in the dioxolane moiety (from 3′- lamivudine-resistant HIV-1 reverse endo to 3′-exo) in dioxolane G. transcriptase-molecular modelling CONCLUSION: The molecular modeling studies approach show that the dioxolane moiety of D-dioxolane nucle- osides enables the nucleoside triphosphate to strongly YH Chong1, RF Schinazi2 and CK Chu1 bind to the active site of 3TC-resistant mutant reverse transcriptase without steric hindrance with Val184. It 1 College of Pharmacy, The University of Georgia, Athens, Ga.; is noteworthy that, depending upon the attached hete- and 2 Emory University/VA Medical Center, Decatur, Ga., USA rocyclic moiety, the binding modes of each dioxolane nucleoside triphosphate is different. This work would lead to the discovery of additional dioxolane with BACKGROUND: The development of viral resistance improved activity against 184V mutants (supported by to lamivudine (3TC) prompted the discovery of nucle- NIH AI32351,AI25899 and Veterans Affairs). osides with activity against HIV isolates containing the 3TC resistance mutation. Since the finding that β−D- dioxolane-2,6-diaminopurine (DAPD or amdoxovir) is active against zidovudine (AZT)- as well as 3TC-resis- tant mutants, several nucleosides with a dioxolane moiety have been synthesized. Among the series of dioxolane nucleosides, the thymidine and 5-fluorocyti- dine analogues showed potent anti-HIV activity against 3TC-resistant mutant reverse transcriptase. Thus, it is of great interest to understand the role of dioxolane moiety in the anti-HIV activity against 3TC- resistant mutant.
METHODS: Various D-dioxolane nucleoside triphos- phates and 3TC triphosphate were docked into the active site of wild-type as well as 3TC-resistant reverse transcriptase, and the resulting complexes were ener- gy-minimized. The M184V mutation imposes steric hindrances to the incoming nucleoside triphosphates as well as the nearby primer chain. Therefore, if the nucleoside triphosphate–reverse transcriptase complex cannot provide enough conformational flexibility to escape from the steric hindrance, it results in an abortive binding state.
RESULTS: Our molecular modeling studies indicated that D-dioxolane nucleoside triphosphates–reverse transcriptase complexes, unlike 3TC triphosphate– reverse transcriptase complex, are not sterically hin- dered by the bulky side chain of Val184, and maintain the favourable binding modes through the interaction of 3′-oxygen with active site residues such as Arg72 or Tyr115. However, the way in which each complexes resolve the steric hindrance of Val184 with the primer strand without deforming the binding mode, was quite different depending upon the heterocyclic bases attached: slight rotation of Val184 into Tyr183 in
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S41 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 37 tions in the 41-210-215 pathway had higher mean resistance to each of the NRTIs (for example, range Prevalence and quantitative mean fold-change for AZT: 4.0–30.6) compared to corresponding NAM combinations in the 67-70-219 phenotypic resistance patterns of pathway (range mean fold-change for AZT: specific nucleoside analogue 2.02–17.7). For each specific NAM combination stud- mutation combinations and of ied, mutation 184 was associated with increased dal- citabine (ddC), didanosine (ddI) and abacavir (ABC) mutations 44 and 118 in reverse resistance and decreased AZT, stavudine (d4T) and transcriptase in a large dataset of tenofovir (TDF) resistance. Alleles 219R and 219N recent HIV-1 clinical isolates appeared at high frequency in five-NAM combina- tions: 41L 67N 210W 215Y 219N and 41L 67N 210W 215Y 219R in 4.1 and 1.6% of isolates, respec- M Van Houtte1, P Lecocq1 and L Bacheler2 tively, whereas 41L 67N 210W 215Y 219Q/E appeared in 0.4% of isolates. 219N and R appeared at 1 Virco BVBA, Mechelen, Belgium; and 2 Virco Lab, Inc., Durham, low frequency in four-NAM combinations (0.32 and NC, USA 0.15%), but were virtually absent from two-, three- and six-NAM combinations, and rarely found in com- BACKGROUND: Two independent pathways to resis- binations with mutation 70. Mutations 44 and 118 tance to nucleoside analogue reverse transcriptase strongly associated with the 41-210-215 pathway and inhibitors (NRTIs) associated with mutations at posi- were associated with low-level 3TC resistance. 67-70- tions 41, 210, and 215 or 67, 70, and 219 (NAMs) of 219 pathway isolates never harboured 44 and 118 HIV-1 reverse transcriptase have been suggested. simultaneously, some harboured either 44 or 118. Furthermore, mutations at positions 44 and 118 in a NAM background are known to confer low-level CONCLUSIONS: Associations among mutations caus- lamivudine (3TC) resistance. We examined prevalence ing resistance to NRTIs vary, as do levels of NRTI and phenotypic resistance of specific NAM combina- resistance associated with specific NAM combinations. tions and of mutations 44 and 118 in recent clinical The predominance of isolates belonging to the 41-210- isolates. 215 pathway with higher levels of NRTI resistance may influence response to NRTI-containing therapy in METHODS: Over 31400 viral patient isolates submit- treatment-experienced patients. ted for routine testing between January 2001 and February 2003 were analysed for presence of 41L, 44D/A, 67N, 70R, 118I, 210W, 215Y, 215F, 219Q/E, 219N, 219R and 184V/I. Prevalence analyses utilized a subset (n=23616) of isolates without mutations 65, 69, 151 or 333. Phenotypic resistance profiles were obtained from Virco’s large matched genotype/pheno- type dataset.
RESULTS: 38.4% of isolates harboured ≥1 NAMs (mean 2.7). Mutation 184 was present in 58.7% of NAM-containing isolates, with similar prevalence in each NAM combination. Only 17/64 possible NAM combinations occurred at frequency >1% among NAM-containing isolates. Mean fold-change for zidovudine (AZT) in isolates with ≥3 NAMs (50% of NAM-containing isolates) ranged 3.1–35 and 6.15–45 in isolates with and without 184, respectively. Combinations in the 41-210-215 pathway were most common: 41L 215Y (12.4%), 41L 210W 215Y (11.8%), 41L 67N 210W 215Y (8.11%). Combinations 67N 70R, 67N 70R 219Q/E and 67N 70R 215F 219Q/E were present at 2.5, 8.7 and 3.0% respectively. Two-, three-, and four-NAM combina-
S42 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 38 activated CD4 and CD8 T cells were used as the source of substrate acceptors (estimated intracellular concen- trations: ATP, 2.7–2.9 and PP , 55–85 µM). Removal of chain-terminating i nucleoside analogues by HIV-1 CONCLUSIONS: The main substrate acceptors recov- ered from cells were ATP, PP and GTP. In the presence reverse transcriptase utilizing i of cell extract, mutant HIV-1 RT can transfer ddAMP intracellular substrates from the DNA primer terminus to one of these accep- tor metabolites, generating Ap ddA (ATP), ddATP 4 AJ Smith, PR Meyer, D Asthana, MR Ashman (PP ) or Gp ddA (GTP). For unstimulated T cells, i 4 and WA Scott monocytes and MDMs, ATP was the predominant substrate acceptor utilized for ddAMP removal. Upon University of Miami School of Medicine, Miami, Fla., USA activation of T cells, PP levels increased three- to eight- i fold and became the predominant acceptor substrate BACKGROUND: HIV-1 replication is inhibited by the for the removal reaction. In light of previous results incorporation of chain-terminating nucleotide ana- from our laboratory and others that the nucleotide- logues during proviral DNA synthesis. HIV-1 reverse dependent excision reaction but not the PP -dependent i transcriptase (RT) can remove these chain-terminating excision reaction is enhanced in RT containing the residues from blocked DNA chains through a reaction AZT-resistance mutations, the in vivo selection of related to pyrophosphorolysis. Our laboratory and AZT-resistance mutations requires further explana- others have demonstrated that the removal of a chain tion. It is possible that selection occurs specifically in a terminator from a blocked DNA chain by HIV-1 RT cell subpopulation or a subcellular compartment con- occurs in vitro and that the substrate acceptor for this taining low levels of PP . i excision reaction can be pyrophosphate (PP ) or any i Supported by NIH grant AI-39973 and American Heart Association nucleoside di- or triphosphate. However, even though Predoctoral Fellowship 0215087B. resistance mutations that confer enhanced excision activity are commonly selected during zidovudine (AZT) therapy, the intracellular acceptor(s) for this transfer reaction is unknown. Experiments were car- ried out to identify compounds present in cell extracts that could serve as substrate acceptors for the excision reaction.
METHODS: Cell extracts were prepared from purified primary CD4, CD8 and CD14 cells. Removal of 2′, 3′- dideoxyadenosine monophosphate (ddAMP) was assessed by incubating 3′ [32P]-labelled ddAMP-termi- nated primer/template with mutant HIV-1 RT and cell extract. Low molecular weight labelled products gen- erated from the removal reaction were separated by gel electrophoresis.
RESULTS: The major 32P-labelled products formed after incubation with H9 (T lymphoid cell line) cell extract were identified as Ap ddA, ddATP and Gp ddA 4 4 produced from the removal and transfer of ddAMP to ATP, PP and GTP, respectively. Similar results were i obtained with extracts from unstimulated primary human T cells, monocytes, and monocyte-derived macrophages (MDMs). In each case, the predominant substrate acceptor for the transfer reaction was ATP, although transfer to PP and GTP could also be i observed (estimated intracellular concentrations: ATP, 1.6–2.3 mM and PP , 10–28 µM). In contrast, the PP - i i dependent reaction predominated when extracts from
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S43 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 39 positions containing mutations that were not observed in at least one clone. Colinearity of reverse transcriptase CONCLUSIONS: In this study multi-RTI-resistant iso- inhibitor resistance mutations lates obtained from heavily treated patients generally detected by population-based consisted of viruses containing either all or nearly all of sequencing the mutations detected by population-based sequenc- ing. This suggests that most RTI-resistance mutations are colinear. The potential benefit of mega-HAART in MJ Gonzales, B Johnston, KM Dupnik, RW this population is likely to derive from the decreased Shafer replication that is often found in viruses containing multiple RTI-resistance mutations rather than from the Stanford University, Stanford, Calif., USA effects of different drugs acting on different virus sub- populations. BACKGROUND: High-level resistance to each of the available reverse transcriptase inhibitors (RTIs) is often detected by the direct sequencing of uncloned PCR products (population-based sequencing). It is not known, however, how often the mutations observed by population-based sequencing are colinear (present in the same viral genome). We sought to determine the relative frequency with which clinical multi-RTI-resis- tant HIV-1 isolates consist of clones containing each of the mutations present in the population-based sequence rather than of mixtures of clones containing different subsets of the mutations present in the popu- lation-based sequence.
METHODS: We selected multi-RTI-resistant isolates from 25 heavily treated individuals in northern California. To determine whether the mutations in these isolates were colinear, we sequenced a mean of 2.8 molecular clones per isolate (71 clones, RT posi- tions 24–310). Each clone was aligned to the popula- tion-based sequence to determine the proportion of mutations that were shared between the clones and the population-based sequences at drug resistance and non-drug resistance positions.
RESULTS: The 25 population-based sequences con- tained a mean of 5.7 nucleoside RTI (NRTI)-resis- tance mutations, 1.2 non-NRTI (NNRTI)-resistance mutations and 11.3 differences from consensus B at non-RTI-resistance positions. The 71 clones con- tained a mean of 5.3 NRTI-resistance mutations, 1.0 NNRTI-resistance mutations and 10.2 differences at non-RTI-resistant mutations. Sequences of the clones closely resembled the population-based sequences: 31 (51%) clones had each of the RTI mutations present in the population-based sequence, 25 (35%) had all but one RTI mutation, 4 (6%) had all but two RTI mutations, 3 (4%) had all but three RTI mutations, and 3 (4%) had all but four RTI mutations. The pop- ulation-based sequence contained a mixture of wild- type and mutant nucleotides at 41/54 (76%) of the
S44 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 40 mutations primarily associated with drug resistance, aa invariance was still maintained in 204 out of 320 aa Identification of the minimal (64% of conservation versus 72% in naive patients). conserved structure of the HIV The long conserved areas I, III, IV, V and VIII, were preserved in drug-treated patients, with the minor reverse transcriptase under the exceptions K11, V108, F116, Q151 (variability in presence and absence of drug ≤3% of patients). Two other large conserved areas (II pressure and VI) shrunk to smaller and fragmented domains (W71-K73, R78-K82, D185-L187 containing the cat- F Ceccherini-Silberstein1, M Santoro1, alytic active-site aa, and S191-L193), while the largest V Svicher1, F Forbici2, C Gori2, R Esnouf3, conserved region (VII) was cut into isolated residues M Ciccozzi4, M Ruiz Alvarez4, R D’Arrigo2, with two smaller conserved areas (W229-D237 and MC Bellocchi2, C Balotta5, A Bertoli2, W239-I244). In addition, a trend of RT structure con- S Giannella2, A Cenci1, MP Trotta2, servation was also suggested by the finding of a high J Balzarini6, A d’Arminio Monforte5, ratio of synonymous to non-synonymous aa substitu- A Antinori2 and CF Perno1,2 tions (ds/dn>7) in RT nucleotide sequences of selected patients, of whom both pre- and post-therapy sequences were available. 1 University of Rome ‘Tor Vergata’, Italy; 2 INMI ‘L Spallanzani’ of Rome, Italy; 3 University of Oxford, UK; 4 Istituto Superiore di CONCLUSIONS: Even in drug-treated patients, HIV- Sanita’, Rome, Italy; 5 University of Milan, Italy; and 6 University 1 RT requires the preservation of at least two-thirds of of Leuven, Belgium aa (some with still unknown function), and of large areas of its tertiary structure in order to maintain a sta- BACKGROUND: Several polymorphisms and drug- ble and functional structure. Future HIV RT inhibitors related variability occur in HIV-1 reverse transcriptase may be designed to target with these invariant (RT). However, little is known regarding its minimal domains. conserved structure. The extent of RT conservation in vivo in the absence and presence of pharmacological pressure has been studied in a large cohort of patients.
METHODS: Sequences of the first 320 amino acids (aa) of RT obtained from plasma of 463 drug-naive patients, and of 802 patients treated with an average of >3 highly active antiretroviral therapy (HAART) regi- mens, were compared and analysed. Consensus B was used as a reference strain for the definition of muta- tions. Amino acid positions with variability <1% were considered invariant.
RESULTS: In naive patients, the RT protein sequence showed no variation in 230 out of 320 RT aa sequence (72% overall conservation). Isolated and pairs of invariant residues were scattered throughout the sequence, while others were clustered to form regions (from 4 up to 25 consecutive invariant residues). The eight longest invariant regions were: I (T7-P19), II (W71-K82), III (T107-S117, containing the D110 cat- alytic site aa), IV (F124-S134), V (N147-P157), VI (Y181-E194, containing the catalytic active site aa D185 and D186), VII (K220-I244) and VIII (W252- Y271). Regarding variable residues, 41 aa were mutat- ed in ≥5% of patients, of which 16 were highly vari- able (substituted in >25% of patients). Frequency of mutations associated with drug resistance was always <1%. In treated patients, despite the appearance of
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S45 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 41 ration of dGTP was measured under burst conditions (substrate excess) with a DNA:DNA template:primer, Identification of a clinical reverse using a KinTek quench-flow apparatus. transcriptase backbone that RESULTS: Polymerase-dependent and polymerase- improves replication of a non- independent RNase H cleavages by P236L/NL4-3 nucleoside reverse transcriptase and P236L/8-26 RTs were similarly slowed relative to wild-type RT. The turnover rate for dGTP incor- inhibitor-resistant mutant by poration (kss+sd) was 0.16 ±0.018/sec for wild-type, increasing the rate of 0.04 ±0.018/sec for P236L/NL4-3 and 0.21 ±0.018/sec polymerization for P236L/8-26 RTs.
CONCLUSIONS: A clinical RT backbone that partial- R Domaoal, C Dykes, R Bambara and ly compensates for the replication defect of P236L also LM Demeter improves its reduced turnover rate (kss) during poly- merization. We postulate that this significantly con- University of Rochester, Rochester, NY, USA tributes to the ability of this clinical RT sequence to compensate for P236L’s replication defect. These stud- BACKGROUND: We previously identified a clinical ies demonstrate that pre-steady state kinetics can iden- reverse transcriptase (RT) sequence containing P236L tify the underlying biochemical mechanisms leading to (8-26), which, when placed into NL4-3, resulted in modulation of drug resistance mutations by clinical RT virus with improved replication kinetics compared to backbones. In addition, these studies indicate that the NL4-3 that contained P236L in an NL4-3 RT back- polymerization abnormalities identified for ground (Dykes et al., Virology 2001; 285:193.). We P236L/NL4-3 RT contribute significantly to the reduc- demonstrated that P236L/NL4-3 RT slows steady- tion in replication efficiency conferred by this mutant state rates of both polymerase-independent and poly- in cell culture. We believe these abnormalities account merase–dependent modes of RNase H cleavage for the infrequent occurrence of this mutant during (Gerondelis et al., Journal of Virology 1999; delavirdine therapy, despite its high level of drug resis- 73:5803.). We also recently found that P236L/NL4-3 tance, and that such studies can be used to better RT slows the turnover rate (kss) during single understand resistance patterns during clinical failure of nucleotide incorporation with a DNA:DNA other non-nucleoside RT inhibitors. primer:template, under pre-steady state conditions (Domaoal et al. 10th CROI #612). For wild-type RT, kss reflects the rate of product dissociation, and is nor- mally the rate-limiting step in polymerization. No reductions were seen in k (the maximal rate of pol nucleotide incorporation) or K (affinity) for dGTP by d P236L RT. We wanted to determine whether improve- ments in either RNase H or polymerase function accounted for the partial improvement in P236L/8-26 replication relative to P236L/NL4-3.
METHODS: Wild-type NL4-3, P236L/NL4-3 and P236L/8-26 RTs were expressed in E. coli and purified. Polymerase-dependent RNase H activity was assayed using a 5′-end-32P-labelled 41 nucleotide (nt) RNA annealed to a shorter DNA, with the DNA 3′-end recessed relative to the RNA 5′-end. Polymerase-inde- pendent RNase H activity was assayed using the same RNA hybridized to a longer DNA, with the RNA 5′- end recessed relative to the DNA 3′-end. RNase H reactions were carried out in the absence of dNTPs and RT input was normalized for steady-state polymeriza- tion specific activity on a homopolymeric RNA:DNA template:primer. kss during single nucleotide incorpo-
S46 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 42 were well correlated with r2 for EFV=0.98, NVP=0.97, n=10, and r2 for EFV=0.79, NVP=0.77 among the Non-nucleoside reverse eight clones with fold change values of <2.5. HS was also detected in the expressed RTs derived from three transcriptase inhibitor HS clones when compared with the wild-type in the hypersusceptibility can be RT activity inhibition assay. The p51–p66 het- demonstrated in multicycle erodimers and p66–p66 homodimers derived from individual clones were inhibited similarly. The p66 phenotype assays and in inhibition from the HS clones when coupled with wild-type p51 assays of purified HIV-1 reverse were HS, but p51 from HS clones with wild-type p66 transcriptases were not HS.
CONCLUSIONS: NNRTI HS is not an assay-depen- NS Shulman, J Delgado, MA Winters, E dent phenomenon, but exists in multicycle replication Johnston, DA Katzenstein, RW Shafer, T and cell-free assays as well as single-cycle phenotypic Merigan assays. As would be expected by the location of the NNRTI binding pocket, NNRTI HS is determined by Stanford University, Stanford, Calif., USA the active p66 RT subunit.
BACKGROUND: Nucleoside resistance has been shown to enhance susceptibility to the non-nucleoside reverse transcriptase inhibitor (NNRTI) class. This NNRTI hypersusceptibility (HS), defined as a fold change in susceptibility of <0.4 compared to wild-type, has been associated with improved outcomes to NNRTI therapy in several retrospective studies. To date, NNRTI hypersusceptibility has only been report- ed in the PhenoSense HIV single-cycle replication phenotypic assay (ViroLogic) and not yet been report- ed in multicycle phenotypic assays or in cell-free enzy- matic systems.
METHODS: Ten recombinant HIV-1 cloned viruses derived from different patients were evaluated for their susceptibilities to nevirapine (NVP) and efavirenz (EFV) using the PhenoSense assay and the ACTG PBMC phenotypic method. Clones were selected on the basis of multiple nucleoside resistance mutations based on their genotypic nucleoside resistance pat- terns. In addition, E. Coli-expressed purified recombi- nant reverse transcriptases (RT) were generated from four of the cloned viruses as well as the wild-type NL4- 3. The EC values of RT activity inhibition using a 50 digoxigenin-labelled nucleotide incorporation assay were determined and compared with the wild-type NL4-3. Different p66 and p51 RT subunit composi- tions were analyzed to determine their relative contri- butions to NNRTI HS.
RESULTS: In the PhenoSense assay, 6 of 10 clones were HS to EFV and 4 of 10 were HS to NVP. In the PBMC assay, four of the six EFV HS and all four of the NVP HS were also HS. The four clones not HS to EFV and six not HS to NVP were also not HS in the PBMC method. The fold change values between the assays
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S47 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 43 was performed including mutations at 41, 67, 69ins, 210, 211 and 215 as variables. The first split selected Genetic correlates of phenotypic was T215Y/F, present in 76% (117/153) of HS vs 37% of non-HS isolates. Further branching in the partition- hypersusceptibility to efavirenz ing also found the L210W and D67N mutations were among 446 baseline isolates from important. In a subgroup of 54 viruses containing five ACTG studies D67N, L210W and T215Y/F, 39 (72%) were HS to EFV. A second CART analysis, expanded to include 29 codons, again selected T215F/Y at the first split, but NS Shulman1, RJ Bosch2, JW Mellors3, found H208Y as the next branch. Of 26 isolates con- MA Albrecht4, and DA Katzenstein1 for the taining both T215Y/F and H208Y, 22 (85%) were HS DACS 217 study team to EFV. In comparing 215Y to 215F, EFV HS was pre- sent in 94/177 (53%) of isolates with 215Y and 24/55 1 CFAR, Stanford University, Stanford, Calif.; 2 SDAC, Harvard (44%) with 215F. School of Public Health, Boston, Mass.; 3 University of Pittsburgh, Pittsburgh, Pa.; and 4 Beth Israel Deaconess Medical CONCLUSION: Univariate and CART analyses of Center, Boston, Mass., USA 446 genotype/phenotype pairs identified key mutations in RT associated with EFV HS. Mutation at codon 215 BACKGROUND: Increased phenotypic susceptibility (Y>F) is most discriminatory but other mutations con- (hypersusceptibility, HS) to non-nucleoside reverse tribute to the HS phenotype including 67N, 208Y and transcriptase inhibitors (NNRTIs) is observed in ~30% 210W. In addition, polymorphisms in the NNRTI- of viral isolates with NRTI-resistance mutations and binding region (K103R and V179I) appear to be asso- has been associated with improved virological ciated with EFV HS. These results can be used to iden- response to NNRTI-based therapy in several studies. tify the biochemical and structural basis for EFV HS Although preliminary analyses have shown an associa- and to predict its presence in clinical samples. tion between zidovudine resistance mutations and NNRTI HS, the genetic basis for NNRTI HS has not been thoroughly defined. We sought to identify specif- ic mutations in RT associated with efavirenz (EFV) HS. Elucidating the genetic basis for EFV HS will improve interpretation of genotypes to optimize use of EFV- containing treatment regimens.
METHODS: Paired baseline genotypes (VGI or ABI) and phenotypes (ViroLogic) were obtained from 446 subjects entering one of five ACTG studies: 290, 359, 364, 370 or 398. All subjects were NRTI-experienced but NNRTI-naive at study entry. Fisher’s exact tests and recursive partitioning (CART) were used to identi- fy RT mutations associated with EFV HS and evaluate their relative importance.
RESULTS: Of the 446 isolates, 153 (34%) demon- strated HS to EFV, defined by an IC <0.4-fold that of 50 a wild-type reference virus. In univariate analyses, EFV HS was significantly associated (P<0.001) with the fol- lowing NRTI resistance mutations: M41L, E44D, D67N, mutations/inserts at 69, V118I, H208Y, L210W, and T215Y/F, but not associated with K70R, L74V, Q151M, M184V, or K219Q/E/N. Two poly- morphisms in the NNRTI binding site of RT, K103R and V179I, were significantly associated with EFV HS (P=0.052 and 0.028, respectively) but other substitu- tions in this region were absent in the HS group includ- ing A98G, L100I, K101E, and V106A. CART analysis
S48 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 44 samples showed 1000- to 10000-fold reduced suscep- tibility to EFV and 50- to 1000-fold reduced suscepti- Rare 1 and 2 amino acid insertions bility to NVP. Specific clones from Patient A that were genotypically matched except for the presence of the in the non-nucleoside reverse insert showed that the insert-containing clone was 50- transcriptase inhibitor (NNRTI) times less susceptible to EFV compared to the non- binding pocket of HIV-1 reverse insert containing clone, indicating a contribution of the insert to EFV resistance. Site-directed mutants created transcriptase affect NNRTI with only the 1 or 2 amino acid insertion did not pro- susceptibility duce viral particles after transfection into cell lines. Molecular modelling studies of the patient-derived MA Winters1, RM Kagan2, PNR Heseltine2, insert-containing strains showed conformation L Kovari3 and TC Merigan1 changes in the NNTRI binding that resulted in altered van der Waals interactions with the EFV and NVP. A 1 Stanford University, Stanford, Calif.; 2 Quest Diagnostics, San larger number of van der Waals contacts were lost Juan Capistrano, Calif.; and 3 Wayne State University, Detroit, between the RT and EFV compared to NVP. Mich., USA CONCLUSIONS: HIV strains possessing 1 or 2 amino BACKGROUND: Inserts in the β3–β4 region of the acid inserts in the NNRTI binding pocket can be found reverse transcriptase (RT) gene of HIV-1 have been in patients failing antiretroviral therapy and contribute shown to affect susceptibility to nucleoside RT to reduced susceptibility to NNRTI. The presence of inhibitors (NRTIs). Recently, an insert in the non- these insertions appears to be dependent on other NRTI (NNRTI) binding pocket region of the RT gene mutations and/or polymorphisms in the RT gene. The from a patient was reported (Rosenbaum et al., inserts affect molecular interactions between the Antiviral Therapy 2000; 5(Suppl. 3):29). We have NNRTI binding pocket and NNRTI. Further monitor- identified and studied two patient-derived HIV-1 ing of treated patients will determine if these insertions strains with NNRTI binding pocket inserts that display are a new, emerging mechanism of resistance. different genotypic features.
METHODS: Physician-requested population-based genotypes were obtained from plasma samples submit- ted to Quest Diagnostics. After RT-PCR amplification, RT genes from strains under investigation were trans- ferred into an NL4-3-based vector. The recombinant clones were transfected into C8166 cells to generate virus stocks. In vitro susceptibility assays were per- formed using titered virus stocks in SupT1 cells. Homology modeling of RT gene structures was per- formed using the program MODELER, and was based on published RT-efavirenz (EFV) and RT-nevirapine (NVP) crystal structures. Changes in molecular recog- nition between genotypically-matched insert-contain- ing and non-insert-containing strains were calculated using the programs HBPLUS and BONDSGL.
RESULTS: Two separate strains with inserts near codon 100 of the RT gene were identified from geno- types collected from 70000 patient samples over the last 4–5 years. Both strains were identified in late 2002. Sample A had a strain possessing a single amino acid insertion, along with the drug resistance muta- tions K103N, V179D/E, M184V and P225P/H. Sample B had a two amino acid insert along with the drug resistance mutations M184V, G190E and the Q151M complex. Recombinant viral clones from both
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S49 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 45 mode of the small molecule. The ligand binds to the ‘open’ MDR protease interacting with residues 28, 47, Crystal structure of a 48, 49 and 80 of a single monomer. multidrug-resistant HIV-1 protease CONCLUSION: The crystal structure of an HIV-1 clinical isolate reveals an expanded protease MDR clinical isolate reveals an expanded active site cavity and represents a active site cavity and provides a structural basis for novel target for the design of protease inhibitor resistance. The crystal structure of the complex represents a novel binding mode where protease inhibitors the peptide binds only to one side of the active site cav- ity and the ‘flaps’ of the protease stay wide open. This 1 1 1 LC Kovari , JF Vickrey , BC Logsdon , G crystal structure will provide structural insight to 1 2 3 Proteasa , Z Wawrzak , MA Winters and TC develop a new class of inhibitors against the ‘open’ 3 Merigan form of the HIV-1 protease. All the licensed inhibitors were designed based on the structure of the ‘closed’ 1 Wayne State University School of Medicine, Department of form of the HIV-1 protease. Biochemistry and Molecular Biology, Detroit, Mich., USA; 2 Northwestern University, DND-CAT Synchrotron Research Center, Argonne, Ill., USA; and 3 Stanford University School of Medicine, Center for AIDS Research, Stanford, Calif., USA
BACKGROUND: The goal of this work is to investi- gate the structural basis of HIV-1 protease multidrug resistance. The hypothesis tested is that HIV protease multidrug resistance is associated with the expansion of the protease active site cavity.
METHODS: We expressed, purified and crystallized a multidrug-resistant (MDR) HIV-1 protease variant (patient isolate 769). Isolate 769 is representative of the end-stage MDR form of the virus. This isolate con- tains the following amino acid changes relative to the wild-type NL4-3: L10I, M36V, M46L, I54V, I62V, L63P, A71V, V82A, I84V and L90M. The IC ratio of 50 the 769 variant to the IC value of the NL4-3 wild- 50 type is: 43 for saquinavir, 47 for nelfinavir, 14 for amprenavir, 80 for atazanavir, >100 for DG-3, >100 for palinavir and >100 for GS3333.
RESULTS: We solved the crystal structure of the HIV- 1 protease MDR isolate 769 to 1.8 Å resolution. The 1.6 Å crystal structure of the 769 MDR protease com- plexed with the tripeptide EDF was also solved. When the 769 crystal structure is compared to the wild-type crystal structure (3PHV), there is an expansion of the active site cavity as we proposed earlier based on mod- elling studies. The active site cavity expands at posi- tions 82 and 84. The active site expansion is due to mutations from larger to shorter side chains, e.g. V82A and I84V. A second observation is that the ‘flaps’ of the protease stay open wider compared to the wild- type. The distance between residues I50 and I150 of the two ‘flaps’ increases from 4 Å in the wild-type to 12 Å in the MDR variant. The crystal structure of the MDR protease – EDF complex reveals a new binding
S50 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 46 CONCLUSIONS: Our analysis has led to the discovery of a conserved substructure of the active site of HIV Structural correlates of PR that is important for the broad-spectrum activity of rrPIs. The mode of binding of UIC-94003 and APV are broad-spectrum activity for a rigidly conserved in both the wild-type and mutant resistance-repellent HIV enzymes, but APV lacks key interactions with the con- protease inhibitor served substructure. In SQV, critical interactions with the wild-type enzyme are lost due to mutations. Mutations involved in drug resistance not only affect AM Silva, SV Gulnik, B Yu, M Eissenstat, JW the interactions between inhibitor and enzyme, but Erickson also can alter intrinsic properties of the enzyme, such as active site solvation and dimer stability. Our analy- Sequoia Pharmaceuticals, Inc., Gaithersburg, MD, USA sis reveals that structural flexibility of PIs is not a nec- essary factor in their ability to exhibit broad-spectrum BACKGROUND: Cross resistance is a serious problem activity. for long-term antiretroviral therapy. In an effort to design broad-spectrum HIV protease inhibitors (PIs), we analysed the structures of mutant HIV proteases (PR) derived from highly cross-resistant HIV strains in complexes with several highly potent PIs.
METHODS: X-ray crystallography and biochemical methods were used to compare the effects of several drug resistance protease mutants on the binding of a resistant-repellant PI (rrPI), UIC-94003; a structurally- related but resistant-susceptible analogue, amprenavir (APV); and, a structurally-unrelated, resistant-suscep- tible PI, saquinavir (SQV). For this study we used wild- type PR, an I84V mutant, (M1) and two mutants that showed average increases in Ki and IC values of over 50 two orders of magnitude for all FDA-approved PIs. These mutants contained four substitutions in one case, including I50V (M2), and 14 substitutions in the other case (M3).
RESULTS: Crystallographic analysis indicates that UIC-94003 and APV bind in a nearly identical man- ner to both wild-type and mutant enzymes, with the exception of additional hydrogen bonds made by the rrPI with a structurally-conserved region of the enzyme. The I50V and I147V mutations in M2 create a larger hydrophobic S2′ cavity that results in weak- er van der Waals interactions between the protein and inhibitor. In addition, weak electron density for the characteristic flap water that typically anchors car- bonyl groups of most inhibitors is observed for M2 and M3. In the M2 and M3 mutant complexes, the intersubunit salt bridge between R8-D129 is absent. The flap water and the salt bridge are both a strong features of all known wild-type/inhibitor complexes, is broken. Comparison of the M3 and wild-type com- plexes with SQV shows that a critical hydrogen bond between the inhibitor and the D30 carboxyl group is lost.
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S51 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 47 have solved so that the P1′-Phe contacts V82, while the P2-Ala does not make extensive contact with the residues in the S2 pocket. In the AP2VNC-p1 com- Co-evolution of the nucleocapsid-p1 V82A plex the P1′-Phe no longer contacts A82, however P2- cleavage site with the V82A Val makes extensive contact with V32, I47 and I84 of mutation in HIV-1 protease the S2 pocket, and the substrate peptide backbone is preserves substrate recognition aligned with the conformation of other substrate com- plexes.
M Prabu-Jeyabalan1, N King1, E Nalivaika1, CONCLUSIONS: The NC-p1 substrate peptide R Swanstrom2 and CA Schiffer1 appears to have an unusual fit in the active site cavity of HIV-1 protease compared with the substrate pep- 1 Department of Biochemistry and Molecular Pharmacology, tides whose complexes have been solved which may University of Massachusetts Medical School, Worcester, Mass.; account for the its slow cleavage rate. The backbone is and 2 UNC Center for AIDS Research, University of North rearranged and the P1′-Phe is in an unusual conforma- Carolina at Chapel Hill, NC, USA tion contacting V82. This is most likely due to the P2- Ala that is unable to fill the S2 pocket of the active site BACKGROUND: The Nucleocapsid-p1 (NC-p1) effectively. When the drug-resistant V82A mutation cleavage site is the last, slowest and, therefore, rate occurs, this likely further destabilizes the complex as determining site to be processed in the HIV-1 Gag by the P1′-Phe contact is lost. P2-Val compensates for this the viral protease. The sequence of the cleavage site is protease mutation, as has been reflected in previously also the most unusual of the viral substrate sites, R-Q- measured kinetics, by filling the S2 pocket and stabi- A-N-*-F-L-G-K, with an asparagine at P1 (the only lizing the substrate’s conformation in the active site. substrate site with a hydrophilic residue at P1) and an alanine at P2 (the only substrate site not to have a branched amino acid at P2). A co-evolution of this site to R-Q-V-N-*-F-L-G-K has been observed in patient sequences when the drug-resistant V82A mutation occurs in HIV-1 protease. Previous studies have shown that this mutation in the NC-p1 substrate site increas- es the rate of its cleavage by more than twofold over the wild-type. In this study we have solved the crystal structures of both the wild-type HIV-1 protease and the V82A HIV-1 protease in complex with their respec- tive NC-p1 substrates and may be able to provide a structural explanation for the change in substrate sequence.
METHODS: The NC-p1 wild-type substrate, R-Q-A- N-*-F-L-G-K, and the AP2V mutant substrate, R-Q-V- N-*-F-L-G-K, were crystallized with the wild-type and V82A protease variants, respectively (referred to as WTNC-p1 and AP2VNC-p1 complexes). Diffraction WT V82A data were collected, processed and refined, using stan- dard crystallographic techniques. The two structures were compared graphically with each other and with other substrate complexes previously determined in our laboratory.
RESULTS: The structures of WTNC-p1 and AP2VNC- WT p1 complexes were refined to 2.2 Å (R=21.0%; V82A R =24.2%) and 2.0 Å (R=19.5%; R =23.1%), free free respectively. In the WTNC-p1 complex the conforma- WT tion of the backbone of the substrate peptide rearranges relative to previous substrate complexes we
S52 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 48 accessory mutations; the other had I84V and L90M with multiple accessory mutations. In both of these Preliminary characterization of a pairs of clones, L23I was associated with increased replication capacity (6 to 39% and 35 to 59%) and newly described protease substrate high-level resistance to each of the PIs. Studies of site- cleft mutation at position 23 directed mutants with and without L23I are pending.
E Johnson, MA Winters, K Vyas, TC Merigan CONCLUSION: L23I is a rare substrate cleft mutation and RW Shafer that occurs in about 1% of patients receiving multiple PIs. By itself, L23I appears to be associated with nelfi- Stanford University, Stanford, Calif., USA navir resistance and decreased replication capacity. In combination with other mutations, L23I appears to be BACKGROUND: L23I/V is a recently recognized muta- associated with multi-PI resistance and increased repli- tion associated with protease inhibitor (PI) therapy cation capacity. (Journal of Virology 2003; 77:4836–4837). However, the risks for developing this mutation and the effect of this mutation on drug susceptibility have not been studied.
METHODS: We analysed the frequency of L23I/V in 1004 untreated patients and 1240 patients receiving one or more PIs to determine whether this mutation was associated with particular PI regimens. We deter- mined the susceptibility and replication capacity of recombinant infectious molecular clones with this mutation using the PhenoSense Assay (ViroLogic, Calif., USA). Recombinant infectious clones were cre- ated by ligating amplified fragments containing the 3′ part of gag and complete protease gene into a gag/pro- deleted pNL43 vector (pNLPFB, provided by T Imamichi, NIAID). Two pairs of clones were derived from clinical isolates that contained a mixture of L and I at position 23. Each of the pairs was isogenic except for the presence or absence of the L23I mutation.
RESULTS: L23I/V occurred in 0/1004 untreated patients and in 18/1240 (1.5%) of patients receiving one or more PIs (P=0.0005; L23I in 16 patients and L23V in two patients). The median duration of PI ther- apy was 161 weeks (range 48–316). Three patients developed L23I during monotherapy with saquinavir, nelfinavir, or amprenavir; 15 patients received 2–4 PIs before L23I was detected. L23I was present at two or more time points in eight patients. All but three of the 18 isolates contained multiple other PI-resistance mutations. Seven recombinant infectious molecular clones encompassing the 3′ part of gag and complete protease gene were created from five clinical isolates containing L23I. One clone containing L23I in combi- nation with the polymorphism V82I had 6.7-fold resis- tance to nelfinavir but to be susceptible to each of the other PIs. This clone had a replication capacity of 10%. The two pairs of clones that differed only at position 23 each had other drug resistance mutations. One pair had V82A, I84V and L90M with multiple
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S53 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 49 cant for APV (r =0.33, P=0.23) but were highly signif- s icant for LPV (r =0.93, P=0.0001). I47V results from s an A→G nucleotide change, whereas I47A requires Emergence of a novel lopinavir two changes (ATA→GCA). A single substitution is resistance mutation at codon 47 required for I47V→I47A (GTA→GCA). We identified correlates with ARV utilization five I47A mutants in specimens with prior sequences containing I47V, suggesting that the I47A variant emerges via a two-step pathway. In the protease-LPV RM Kagan1, M Shenderovich2, K Ramnarayan2 model, the Ile side chain of residue 47 is positioned and PNR Heseltine1 close to the LPV dimethoxyphenyl aromatic ring and VDW interactions may contribute to binding. A V47 1 Department of Infectious Diseases, Quest Diagnostics, Inc. San model shows similar positioning with slightly Juan Capistrano, Calif.; 2 Structural Bioinformatics, Inc. San decreased VDW interaction, however these interac- Diego, Calif., USA tions are lost for the smaller A47.
OBJECTIVES: Multiple primary protease mutations CONCLUSIONS: We have identified a second path- are required to produce high-level resistance to way to high-level LPV resistance that does not result lopinavir (LPV). These mutations may be selected from cross resistance to other PIs. Increased usage of through antiretroviral therapy (ARV) with other pro- LPV correlates with increased frequencies of the I47V tease inhibitors (PIs), leading to LPV cross resistance. mutation leading to the stepwise emergence of the We identified a novel I47A LPV-resistant variant and LPV-resistant I47A variant. Surveillance of emerging found that the emergence of mutations at codon 47 is resistance in large clinical databases may be facilitated highly correlated with increasing clinical LPV utiliza- by structural phenotypic methods that enable rapid tion. computational determinations of the significance of newly identified mutational patterns. METHODS: Genotypes were obtained by DNA sequencing of clinical samples and mutational frequen- cies were obtained from the Quest Diagnostics HIV-1 sequence database. Quarterly ARV prescription data was obtained from Scott Levin Source Prescription Audit. Phenotypes were determined by the Antivirogram assay. Changes in binding energy were calculated from PR-LPV complexes built by amino acid substitutions in the wild-type complex with sub- sequent energy minimization of the ligand and binding site residues.
RESULTS: We identified the I47A mutation in 40/47797 clinical specimens genotyped after LPV became avail- able in late 2000. None of 26459 samples genotyped prior to October 2000 had the I47A mutation (χ2=20.6 P<0.0001). Phenotypic data obtained for three I47A mutants showed high levels of LPV resistance (86.2×, >91.5× and >102.5×) that was not predicted by geno- typic analysis. Molecular modeling and energy calcula- tions for these variants showed binding energy changes (∆E ) for LPV of 3.0, 3.1 and 2.8 kcal/mol, respec- bind tively, consistent with high levels of LPV resistance (>100×). I47V variants had smaller ∆E changes (2.0, bind 1.9 and 0.93 kcal/mol, respectively). I47V is associat- ed with resistance to amprenavir (APV) but not LPV. I47V frequency has increased 4× from 2000 to 2002. Spearman correlations between APV or LPV prescrip- tion utilization and I47V frequencies were not signifi-
S54 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 50 CONCLUSIONS: Different biological properties account for the replicative advantage of D30N and Parameters driving the selection of L90M mutants in the presence of nelfinavir. The high- nelfinavir-resistant HIV-1 variants er prevalence of D30N mutation in patients receiving nelfinavir reflects the higher level of resistance that can V Perrin and F Mammano be attained by viruses engaged in the D30N pathway. In addition, restoration of infectivity of D30N-L90M double mutant by a common polymorphism reinforces INSERM U-552, Recherche Antivirale, Hôpital Bichat-C.Bernard, the notion that the genetic context of the virus may Paris, France substantially influence the impact of resistance muta- tions. BACKGROUND: Resistance to nelfinavir most often begins with the selection of D30N mutation in the pro- tease. In a percentage of patients, however, the alter- native L90M mutation is found. Mutants carrying both these mutations are only rarely found in vivo. Here, we investigated the parameters that determine the emergence of mutants following the D30N or the L90M pathways. To this end, a series of mutant clones were compared for infectivity in the absence of drug, resistance to nelfinavir and replicative advantage as a function of drug concentration.
METHODS: Replication-competent viral clones that carry single or combined mutations in the protease were constructed by site-directed mutagenesis. The drug-free infectivity of virions was measured in a sin- gle-cycle assay and expressed as percentage of the parental (pNL4.3) wild-type clone. Nelfinavir resis- tance was calculated as the ratio of IC and IC val- 50 90 ues of the mutant clones to those of wild-type virus. Replicative advantage curves were drawn by com- paring the infectivity of mutant clones and of wild- type virus at several drug concentrations (from 0 nM to 2500 nM).
RESULTS: Both D30N and L90M mutations confer a selective advantage for replication in the presence of low nelfinavir concentrations. The advantage dis- played by D30N mutant was mostly due to resistance, while the advantage of L90M mutant reflects preser- vation of infectivity coupled with a minimal reduction in susceptibility. At higher nelfinavir concentrations, viruses harbouring D30N coupled to additional muta- tions of this evolutionary pathway could reach higher levels of resistance and display a higher selective advantage compared to mutants of the L90M series. A mutant carrying both D30N and L90M was character- ized by a dramatic loss of infectivity, which explains the rare appearance of this combination in vivo. Interestingly, the loss of infectivity could be efficiently rescued by the addition of compensatory mutations in the protease or by the L63P polymorphism.
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S55 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 51 respectively (differences between groups not signifi- cant). The A431V p7/p1 cleavage site mutation was I84A and I84C mutations in very common in I84A mutants and I84V (with I54V and/or I47V) isolates and uncommon in I84V mutants protease confer high-level lacking I54V and I47V. In one patient from whom lon- resistance to protease inhibitors gitudinal data were available, the I84V mutant was an and impair replication capacity apparent precursor to I84A. Structure modelling of the mutant proteases indicated that I84V, I84C and I84A mutations all create unoccupied volume in the active H Mo1, N Parkin2, KD Stewart, L Lu1, site, with I84A introducing the most greatest change T Dekhtyar1, D Kempf1 and A Molla1 from the wild-type structure. Mutation of the gag p1- p7 and p1/p6 substrates reduced the unoccupied vol- 1 Abbott Laboratories, Abbott Park, Ill.; and 2 Virologic, Inc, ume created by I84A. South San Francisco, Calif., USA CONCLUSION: In general, I84V mutants without BACKGROUND: The I84V mutation in HIV protease other major mutations display relatively modest resis- (PR) is associated with resistance to the PR inhibitor tance to PIs. Increased resistance is accomplished by (PI) class. The I84A variant has been observed follow- addition of other major mutations (common pathway) ing ritonavir (RTV)/saquinavir (SQV) therapy and in or (less commonly) by selection of I84C or I84A. vitro passage with the experimental PI BILA 1906 BS. Molecular modelling provides useful insight into the However, the clinical relevance of I84A or other I84 mechanism of resistance mutations on drug suscepti- mutations is not well characterized. bility and replicative capacity.
OBJECTIVE: To evaluate the effect of novel substitu- tions at position 84 on PI susceptibility and replication capacity (RC).
METHODS: Phenotypic susceptibility to PIs [RTV, indinavir (IDV), amprenavir (AMP), SQV, nelfinavir (NFV) and lopinavir (LPV)] and RC was evaluated using PhenoSense HIV. Molecular modelling was performed using Insight-II.
RESULTS: The I84C or I84A mutations were each identified in eight separate isolates. Several secondary mutations, as well as p7/p1 or p1/p6 cleavage site mutations were also observed. I54V and/or I47V were observed in 1/8 I84A, 0/8 I84C and 6/12 I84V isolates chosen for comparison. Compared to I84V isolates lacking I54V or I47V, I84C isolates displayed substan- tially higher (22- to 36-fold) resistance to NFV and SQV, slightly higher resistance to RTV, IDV and AMP (three- to sevenfold), but similar susceptibility to LPV (1.6-fold). I84A isolates exhibited >60-fold median resistance to all PIs except LPV. In contrast, I84V iso- lates with I54V and/or I47V were especially resistant to LPV (median IC 80-fold). The median fold IC 50 50 values for the six PIs tested were 22-fold for I84C mutants, 76-fold for I84A mutants, 2.9-fold for I84V mutants without I54V or I47V, and 36.5-fold for I84V mutants with I54V and/or I47V. Greater than 10-fold resistance was observed 64, 96, 14 and 78% of the time with the above four groups of isolates, respective- ly. RC was impaired in all mutant isolates tested: medi- an RC for I84A, C and V were 6.4, 11 and 28%,
S56 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 52 K55* was present in 49/803 (6.1%) of the clinical iso- lates. The most common non-consensus substitution The HIV-1 protease mutation K55R was K55R (30/49, 61%) or K55k/r (16/49, 33%). K55R was only rarely detected in PI-naive subjects is associated with the presence of (1/422 isolates). Of the subjects with the K55R substi- the M46I/L mutation tution, the majority also had the M46I/L substitution (29/30, 97%; M46I n=18, M46L n=11). Of note, non- E Morgan1, D Pillay2, P Cane1, JP Kleim3, consensus substitutions at positions M46*, V82* or M Tisdale3, M Maguire3, S Macmanus3, I54* were not associated with K55* indicating that P Yates3 and R Elston3 K55R may develop after these mutations or only in a specific sub-set of viruses containing these mutations. 1 University of Birmingham, Birmingham; 2 UCL, London; and 3 GSK, Stevenage, UK CONCLUSION: Analysis of the associations between non-consensus substitutions across all 99 protease BACKGROUND: The association of specific HIV-1 amino acids identified additional associations not pre- protease resistance mutations is well documented; for viously described including K55R being associated example, D30N and N88D, V32I and I47V. with either M46I or L, or V82A/S/T or I54V. K55R Additional associations between resistance mutations was only rarely detected in PI-naive subjects suggesting and either natural polymorphisms or substitutions not that this mutation could represent an accessory resis- considered as being key to resistance may exist. To tance mutation. identify additional associations, analysis was per- formed comparing the incidence of non-consensus sub- stitutions at one position with non-consensus substitu- tions at the other 98 amino acid positions in protease.
METHODS: Protease genotypes from 803 protease inhibitor (PI)-experienced subjects that had failed at least one PI were available for analysis. Non-HXB2 consensus substitutions at each of the 99 residues of the protease gene were identified. The incidence of a non-consensus substitution occuring at amino acid X was compared to the incidence of a non-consensus substitutions at each of the other amino acid positions. A total of 9604 comparisons were made. The co-asso- ciation between pairs of non-consensus substitutions was analysed, except when the incidence was low (<1%). Statistical significance was tested using the χ2 test.
RESULTS: 39/99 positions had <1% non-consensus substitutions in the 803 samples of which 13/39 posi- tions had no non-consensus substitutions (L5, W6, R8, P9, D25, T26, G27, A28, D29, G40, P44, G52, Y59). Analysis of association between non-consensus substi- tutions at the remaining positions confirmed previous- ly described associations between protease resistance mutations and identified additional associations between natural polymorphisms; for example, I47*+V32*, D30*+N88*, V82*+I54*, M36*+E35*, M46*+L10*, I54*+L10*, M46*+L63*, G73*+L90*, M36*+K20*, L90*+A71*. In addition a novel associ- ation between K55* and M46* (P<0.001) was identi- fied and, to a lesser extent, an association between K55* and V82* (P<0.001) and K55* with I54* (P<0.001). A non-consensus substitution at position
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S57 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 53 CRF02_AG (83% untreated). Synonymous NA codon differences were identified at 18 RT and seven protease Nucleic acid differences between drug resistance loci in sequences from non-B-infected untreated persons. AA at these positions in sequences HIV-1 non-B and B reverse from non-B-infected treated persons were similar to transcriptase and protease those seen in subtype B, except RT position V106 in sequences at drug resistance subtype C. Of sequences from 548 untreated subtype C-infected persons, 79% (431/548) had GTG/valine, positions and 17% (93/548) GTA/valine, while in subtype B, 95% (1580/1671) had GTA/valine at RT position 106. R Kantor1, AP Carvalho2, B Wynhoven3, Among sequences from non-nucleoside RT inhibitor MA Soares4, P Cane5, J Clarke6, J Snoeck7, (NNRTI)-treated subtype C-infected persons, 16% C Pillay8, S Sirivichayakul9, K Ariyoshi10, (22/142) had ATG/methionine and 1% (2/142) had A Holguin11 , H Rudich12, R Rodrigues13, GCG/alanine, while in subtype B 1% (4/450) had ATG MB Bouzas14, P Cahn14, LF Brigido13, /methionine and 4% (17/450) had GCA/alanine at that Z Grossman12, V Soriano11 , W Sugiura10, position. Of the 22 subtype C-infected persons with P Phanuphak9, L Morris8, A-M Vandamme7, V106M, 19 (86%) were exposed to efavirenz and 3 J Weber6, D Pillay5, A Tanuri4, PR Harrigan3, (14%) to nevirapine. All isolates harboured other R Camacho2, JM Schapiro1, RW Shafer1 and NNRTI-related mutations (G190A/S>K103N>K101E/ D Katzenstein1 P>F227L>Y188C/H>A98G=V179D>181C>M230L). Other RT valine residues (positions 75, 118, 179) har- 1 Stanford University, Stanford, Calif., USA; 2 Hospital Egas boured mainly codons GTT/GTA in sequences from B Moniz, Lisbon, Portugal; 3 BC Centre for Excellence in HIV/AIDS, or C infected untreated persons, with non-synonymous Vancouver, Canada; 4 Universidade Federal do Rio de Janeiro, G to A transitions to ATT/ATA, coding for isoleucine Rio de Janeiro, Brazil; 5 PHLS, University of Birmingham, in sequences from treated persons. In other non-B sub- Birmingham, UK; 6 Wright Fleming Institute, London, UK; 7 Rega types only 2% (21/1043) of sequences from untreated Institute for Medical Research, Leuven, Belgium; 8 National persons had GTG/valine and 0.5% (1/186) from treat- Institute for Virology, Johannesburg, South Africa; 9 ed persons had ATG/methionine. Chulalongkorn University, Bangkok, Thailand; 10 National Institute of Infectious Diseases, Tokyo, Japan; 11 Hospital Carlos CONCLUSION: There are characteristic, subtype-spe- III, Madrid, Spain; 12 Ministry of Health, Tel-Aviv, Israel; 13 cific baseline synonymous NA differences at RT and Instituto Adolfo Lutz, Sao Paulo, Brazil; and 14 Fundación protease drug resistance positions, which rarely result Huesped, Buenos Aires, Argentina in different non-synonymous substitutions with drug therapy. However, as described by Brenner et al., BACKGROUND: HIV-1 subtypes are phylogenetically V106M uniquely predominates in sequences from sub- distinct due to non-synonymous and synonymous type C-infected persons after efavirenz, and to a lesser nucleic acid (NA) differences leading to dominant extent, as shown here, after nevirapine exposure. The amino acid (AA) coding. Conservation and variation predominant use of GTG/valine and a G to A transition of codon usage among divergent subtypes may con- select ATG/methionine at V106 preferentially in subtype tribute to different mechanisms and pathways in the C. This may be a consequence of G to A hypermutation, evolution of drug resistance. seen also at other RT valine drug resistance positions. This unique codon use may contribute to a lower thresh- METHODS: RT and protease codon usage was exam- old for NNRTI resistance in subtype C. ined at non-polymorphic drug resistance positions in a large dataset of sequences from non-B and B-infected, untreated and subsequently in treated persons. Significant differences (P<0.001) in the prevalence of NA patterns at resistance positions among non-B and B sequences, were determined by Fisher’s exact test.
RESULTS: Data were available from 289 persons infected with subtype A (72% untreated), 769 with C (55% untreated), 223 with D (64% untreated), 148 with F (57% untreated), 318 with G (43% untreated), 297 with CRF01_AE (47% untreated) and 343 with
S58 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 54 higher frequencies of the polymorphisms N42S, E151A, N305D and T130T, which were associated Search for polymorphic sites in R5 with increased ENF susceptibility. Additionally, the clusters were compared using week 24 efficacy data; tropic HIV-1 Env and enfuvirtide patients in Cluster 2 experienced a larger drop in viral drug susceptibility in baseline load than patients in Cluster 1. To further explore the isolates from TORO 1 and TORO 2 genotype and phenotype relationship, binary regres- sion tree models were constructed. The optimized tree contained five terminal nodes. Tree branch points C Su1, G Heilek-Snyder1, D Fenger1, occurred at gp41 positions 3ins (insertion at 3), 24 and P Ravindran1, K Tsai1, N Cammack1, P Sista2 42, which also appeared in the ‘important’ amino acid and S Chiu1 set obtained by ANOVA.
1 Roche, Palo Alto, Calif.; and 2 Trimeris, Inc., Durham, NC, USA CONCLUSIONS: Statistical analyses have revealed an association between polymorphic sites in HIV Env and BACKGROUND: Whilst it is evident that the principal baseline variability of ENF susceptibility in HIV-1 R5 determinants for reduced enfuvirtide (ENF) suscepti- tropic recombinants. Cluster analysis of gp41 viral bility reside in HIV-1 gp41, the possibility remains that sequences identified combinations of polymorphisms other determinants within baseline (BL) HIV-1 enve- that were associated with higher ENF susceptibility in lope amino acid sequences may also influence ENF the 12 non-B R5 recombinants and a regression tree susceptibility for virus from fusion inhibitor-naive model illustrated amino acid interactions associated patients. We explored such relationships by further with ENF susceptibility. The significance of these asso- mining the TORO 1 and TORO 2 clinical trial data- ciations is under study. bases using univariate and multivariate statistical methods.
METHODS: Paired BL genotype and phenotype data were analysed for 377 R5 tropic recombinants from patients receiving ENF and an optimized background regimen. HIV-1 envelope (complete gp160) amino acid sequences and ENF susceptibility were generated using the GeneSeq and PhenoSense HIV Entry assays. The relationship of genotype (JRCSF reference) and phenotype at each individual gp160 position was explored using analysis of variance (ANOVA) models. The joint relationship between the entire gp160 sequence and ENF susceptibility was studied using cluster analysis and regression tree modelling.
RESULTS: The fold change in IC was approximately 50 log-normally distributed with a geometric mean (GM) of 1.58 (range 0.04-37.71) and a standard deviation of 2.64. ANOVA identified ‘important’ amino acids (P<0.05) in both gp41 (for example, G3GT, N42S and V69L) and gp120 (for example, T50I and L444M) that were all associated with >1.5-fold change in sus- ceptibility from the GM. To explore differences among gp41 genotypes, cluster analysis was used to identify two clusters of 365 (Cluster 1) and 12 (Cluster 2) recombinants. Recombinants in Cluster 2 were identi- fied as non-B subtypes. Cluster 2 (GM=0.78, range 0.04-4.44) showed a higher ENF susceptibility (P=0.01) than Cluster 1 (GM=1.61, range 0.05–37.71); Cluster 1 had a GM and range similar to the complete dataset. Recombinants in Cluster 2 had
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S59 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 55 cal response at week 2, or nadir, and also after adjust- ment for prognostic factors. Covariance analyses of Subgroup analysis of baseline virological response by BL ENF susceptibility category defined by the geometric mean (GM) IC ±1 or 2 SD 50 susceptibility and early virological yielded no significant relationships between BL ENF response to enfuvirtide in the viral susceptibility and virological response (P>0.10 combined TORO studies for all analyses across both subgroups and all virolog- ical response metrics). In an analysis of the relationship between BL GSS and GM decrease in susceptibility to P Sista1, T Melby1, ML Greenberg1, R DeMasi1, ENF at VF, we found that patients with lower BL GSS D Kuritzkes2, M Nelson3, C Petropoulos4, had significantly greater decreases in susceptibility to M Salgo5, N Cammack6 and TJ Matthews1 ENF than those with higher GSS (r=–0.21, P=0.003).
1 Trimeris, Inc., Durham, NC, USA; 2 Brigham and Women’s CONCLUSION: No significant association between Hospital, Boston, Mass., USA; 3 Chelsea and Westminster BL susceptibility to ENF and early virological response Hospital, London, UK; 4 ViroLogic, Inc., South San Francisco, in patients with the lowest GSS was observed, suggest- Calif., USA; 5 Roche, Nutley, NJ, USA; and 6 Roche, Palo Alto, ing that other factors may be influencing virological Calif., USA response. The inverse correlation seen between BL GSS and reduced susceptibility to ENF at VF suggests that BACKGROUND: In the TORO 1 & 2 Phase III clini- the antiviral activity of ENF is preserved when com- cal studies, similar virological suppression occurred bined with additional active agents in the optimized across a range of baseline (BL) viral susceptibility to background enfuvirtide (ENF) through 24 weeks (Greenberg et al, 10th CROI). Viruses from patients who met virologi- cal failure (VF) had a 21-fold geometric mean decrease in susceptibility to ENF from BL to VF and exhibited concomitant genetic changes in gp41 amino acids (aa) 36–45. To minimize the confounding effects of the optimized background (OB) regimen, we have further examined the relationship between ENF BL IC and 50 early virological response for patients with a genotyp- ic sensitivity score (GSS) of 0 or 1. We also determined the correlation between BL GSS and decreases in sus- ceptibility to ENF at VF.
METHODS: TORO 1 & 2 are randomized, open- label, controlled, multicentre, Phase III studies of patients receiving 90 mg twice daily of ENF by subcu- taneous injection in combination with an OB regimen. The intent-to-treat population included 661 patients randomized to ENF+OB. Resistance data were gener- ated using the ViroLogic GeneSeq for GSS and the experimental PhenoSense Entry Assay for ENF suscep- tibility. Correlation between baseline ENF susceptibili- ty and virological response was assessed using non- parametric linear correlation and multiple linear regression analyses.
RESULTS: Of the 612 patients with BL viral pheno- type and complete virological response data, 98 had BL GSS=0 and 167 had a GSS =1. There was no corre- lation with BL ENF IC and log plasma HIV-1 RNA 50 10 change from BL to week 4 in both subgroups (r=0.15, P=0.13; and r=0.06, P=0.47, respectively). Similar results were obtained for both subgroups for virologi-
S60 SESSION 3 HIV Pathogenesis, Fitness and Resistance
Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 56 (P=0.0062, Mann-Whitney). In addition, low replica- tion capacity of patient isolates strongly correlated Virus characteristics predict with low pre-highly active antiretroviral therapy (HAART) viral load levels (r2=0.2774; P=0.0017). viraemia control after cessation of Pretreatment intrapatient HIV env nucleotide sequence antiretroviral therapy diversities (Dnt) ranged from 0.7 to 5.5%, amino acid diversities (Daa) from 1.0 to 9.4%. Patients controlling H Günthard1, B Joos1, H Kuster1, M Fischer1, viraemia after STIs had significantly lower pretreat- C Leemann1, J Böni3, A Oxenius4, C Fagard2, ment diversities (Dnt 1.7 ±0.9% vs 3.1 ±1.3%, B Hirschel2, J Wong5, R Phillips6, P=0.03; Daa 3.1 ±1.7% vs 5.9 ±2.4%, P=0.02). Taken S Bonhoeffer4, R Weber1, A Trkola1 and the together, our findings suggest that these low replicating Swiss HIV Cohort Study viruses might have been present before HAART was initiated and, therefore, are not a direct consequence of 1 Division of Infectious Diseases, University Hospital Zurich; and STI. Virus isolates from the two patient groups were 2 Geneva, Switzerland; 3 Swiss National Center for Retroviruses, indistinguishable in terms of co-receptor usage and University of Zurich, Zurich, Switzerland; 4 ETH Zentrum, Zurich, genetic subtype. Patients controlling viraemia showed Switzerland; 5 University of California San Diego, La Jolla, Calif., significantly higher neutralization activity than non- USA; and 6 University of Oxford, Oxford, UK controllers (P=0.02), and plasma neutralization activi- ty against autologous virus was found to be associated BACKGROUND: Spontaneous control of plasma with lower pretreatment diversity (P=0.06 for Dnt and viraemia in untreated HIV-infected patients has often Daa). However, no association between diversity, anti been attributed to better immunological responses. In gp120 titres, HIV-specific CTL- and T-help responses many studies, however, virus characteristics were not were found. investigated. Here we studied virological and immuno- logical characteristics longitudinally in 21 patients to CONCLUSION: Viral characteristics such as in vitro assess their potential roles in control of plasma replication capacity and infectivity predicted control of viraemia. plasma viraemia after last cessation of therapy in patients undergoing STI. Low pretreatment env diver- METHODS: Analysis of in vitro infectivity, growth sity in patients controlling viraemia after STI suggests kinetics, co-receptor usage, genetic subtype and sensi- that those viral characteristics were not influenced by tivity to inhibition by monoclonal antibodies and STI but were already present years ago before any ther- chemokines from viral isolates obtained from 20 chron- apy was started. SSITT-baseline neutralizing activity ically infected patients participating in the Swiss Spanish but not HIV-specific CTL- or T-help responses were Intermittent Therapy Trial (SSITT). Patients underwent associated with control of viraemia. four consecutive STI cycles (2 weeks off, 8 weeks on treatment) followed by a fifth long interruption. Sequential virus isolates were obtained from patients PBMC collected during the first and the fifth inter- ruption cycle. From 21 patients pretreatment plasma HIV RNA env diversity was determined by RT-PCR, cloning (16 clones/sample), sequencing (env C2V3C3) and phylogenetic analysis (PHYLIP/ Mega). For analysis, patients were grouped according to spon- taneous control of plasma viraemia between weeks 40 and 76 (controllers: <5000 RNA copies/ml for at least 2 months).
RESULTS: Viruses from patients who controlled viraemia upon STI-treatment had a significantly lower in vitro replication capacity (P=0.0175). In vitro growth-rates of patient isolates correlated strongly with their in vitro infectivity (TCID ) (r2=0.3679; 50 P=0.0046). High control of viraemia after the STI- treatment, was predicted by lower viral load levels before any initiation of antiretroviral treatment
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S63 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 57 level of T cell activation and the rate of CD4 T cell depletion for any given level of viraemia are lower in Distinct patterns of the virological treated than in untreated patients. and immunological response to This same conceptual model may also explain why antiretroviral therapy interpreted in resistance mutations are rarely observed below terms of the dynamics of immune 50–100 copies RNA/ml despite clear evidence of ongo- ing viral replication. We suggest that below this thresh- activation and HIV transmission old, virus generated in the course of an infection burst rarely sparks new bursts in unrelated target cells, thus Z Grossman1, Z Grossman2, PW Hunt3 and precluding gradual increase in the repertoire of infect- SG Deeks3 ed memory cells; viral replication remains local and resistant variants that may emerge are prevented from 1 National Institutes of Health, Bethesda, Md., USA/Tel-Aviv spreading. However, above threshold, HIV-specific University, Tel Aviv, Israel; 2 National HIV Reference Lab, PHL, memory cells are generated and are capable of spread- MOH, Tel-Hashomer, Israel; and 3 University of California/San ing new resistant variants within the memory pool. Francisco General Hospital, San Francisco, Calif., USA Thus, both direct spreading and anti-HIV response- dependent ‘mixing’ become insignificant when infec- Recent data indicate that highly active antiretroviral tion bursts are sufficiently reduced. Consistent with therapy (HAART) does not stop virus dissemination but mixing, we previously observed that resistance muta- rather results in a lower quasi-steady state, and that the tions did develop in patients with intermittent viraemia new viraemia level is a strong predictor of subsequent episodes, who showed limited evidence of generalized disease outcome. Treated patients with low to moderate T cell activation but had large numbers of HIV-specif- levels (for example, 100–10000 copies RNA/ml) exhib- ic T cells. Importantly, PAT predicts that in the absence it no or delayed immunological progression, while of rapid spreading and mixing, wild-type HIV and patients with lower viraemia levels exhibit no or resistant variants can coexist, actively replicating in delayed viral evolution. Here, we interpret these data relative isolation. in terms of ‘proximal immune activation and virus transmission’ (PAT), a previously described model. This conceptual model posits that in the low-to-mod- erate range of viraemia, efficient HIV dissemination depends on the activation and reseeding of latently infected memory cells, which spark isolated cell-to-cell infection bursts in the context of localized immune responses. Such responses are likely to be recurrently triggered in lymphoid tissues by self and foreign anti- gens under conditions of chronic inflammation, which is presumably induced continuously at the lymphocyte- APC interaction level by secondary anti-HIV respons- es. Chronic inflammation and immune activation exceeding a certain host-dependent level drive progres- sive CD4 depletion.
The main effect of antiretroviral therapy is to reduce the amount of virus produced per burst, not the num- ber of bursts, since the latter is determined by the anti- gen-recognition diversity of infected memory cells. When the virus is drug resistant, such reduction still occurs, both because resistance is incomplete and because replication capacity is reduced. Theoretically, for any given level of viraemia, a small frequency of uncontrolled bursts may be associated with a higher level of inflammation as compared to a larger frequen- cy of partially controlled bursts. Consistent with this model, we and others previously observed that the
S64 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 58 In regard to viraemia at >1 copy/ml the mean viral load for those with and without blips was 9.4 and Both stable and decaying HIV 12.2 copies/ml, respectively (P=NS) at a mean follow- up of 22 and 14 months, respectively (P=0.023). From plasma viral loads are observed in this cohort with stable viral load suppression 19/60 subjects with sustained suppression subjects received influenza vaccination as standard of of viraemia <50 copies/ml and are care. The mean (median, range) plasma viral load at vaccination and 1 week later was 4.7 (3, 8–16) and 3.3 uninfluenced by influenza (3, 1–6) copies/ml, respectively (P=NS). vaccination CONCLUSIONS: A quantifiable viraemia was EP Coakley1, A Tiro1, A Wurcel2, R D’Aquila3 observed in 61% of subjects with apparent sustained and DR Stone2 suppression of the viral load below the limit of detec- tion by current assays. Transient viraemic episodes may 1 Tufts-New England Medical Center and Tufts University School be markers of such ongoing viraemia irrespective of the of Medicine and Tufts/Brown Center for AIDS Research; 2 The duration of time <50 copies/ml. Conversely, among Center for AIDS Research Partners/Fenway/Shattuck, Boston, those with no documented blips lower viral loads are Mass..; and 3 Vanderbilt University School of Medicine, Nashville, correlated with longer periods of time <50 copies/ml, Tenn., USA suggesting that total body viral burden may be decreas- ing in those lacking blips. Despite prior reports of tran- BACKGROUND: There are few data describing the sient viraemia after influenza vaccination in those with natural history of viraemia at very low levels. We stable viraemic suppression we did not observe any employed a highly sensitive plasma HIV-1 RNA quan- short-term change in viral load below 50 copies in this titation assay to determine the level of viraemia on sin- population. gle time-point samples from a cohort of subjects on stable antiviral therapy and with stable suppression of plasma viraemia <50 copies/ml. We also assessed the short-term impact of influenza vaccination on the level of viraemia in a subset of this cohort.
METHODS: HIV RNA was harvested from 2 ml of plasma by a guanidine-isopropanol-ethanol method. Competitive RT-PCR included an internal quantitative standard with colorimetric detection. The lower limit of quantitation is 1 copy/ml.
RESULTS: The 59 subjects had viral loads <50 copies/ml for a mean (range) of 19.9 (2–81) months. Notably, 19/59 (35%) of these subjects had at least one episode of intermittent viraemia or ‘blip’ to levels >50 copies/ml. Among the 40 subjects without prior recorded blips the mean plasma viral load was 6.7 copies/ml at a mean follow-up time of 20.4 months. For the 19 subjects with prior blips the mean viral load was 7 copies/ml at a mean follow-up of 19 months. Comparing those with and without prior blips there were no significant differences in mean viral loads or time <50 copies/ml. Among those 40 subjects with no prior blips a correla- tion was noted between the viral load and the duration of time <50 copies/ml, R=–0.45, P=0.003. Conversely, among the 19 subjects with prior blips no correlation was observed. Comparing those with and without prior blips the proportions with viral loads <1 copy/ml were 5/19 (26%) and 18/40 (45%) at a mean follow- up of 9.2 and 29.5 months, respectively (P=0.025).
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S65 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 59 CONCLUSIONS: Lower baseline proviral HIV DNA levels and randomization to the TDF arm were associ- HIV DNA as a predictor of residual ated with lower levels of residual viraemia, indepen- dent of baseline HIV RNA levels. In this exploratory viraemia in patients treated with analysis, lower levels of residual viraemia in the TDF tenofovir+lamivudine+efavirenz or arm suggests greater antiviral potency of this regimen stavudine+lamivudine+efavirenz as compared to the d4T regimen. Higher proviral HIV DNA at baseline may either directly contribute to residual viraemia through subsequent activation of the DV Havlir1, M Strain2, MD Miller3, C Ignacio2, latent reservoir, or alternatively, the proviral HIV DNA B Lu3 and J Wong2 for 903 Study Team level may be a surrogate for host factors that sustain residual HIV infection during therapy. 1 University of California, San Francisco; 2 University of California, San Diego; and 3 Gilead Sciences, Foster City, Calif., USA
BACKGROUND: In patients successfully treated with potent antiretroviral therapy, HIV RNA levels estab- lish a steady state (residual viraemia) after 6–9 months of therapy that persists for years and reflects low-level viral replication. We evaluated the predictors of resid- ual viraemia among 100 treatment-naive patients ran- domized to tenofovir (TDF)+lamivudine (3TC)+ efavirenz (EFV) (TDF arm) or stavudine (d4T)+3TC+ EFV (d4T arm) in Gilead study 903.
METHODS: A cohort of 100 sequentially enrolled patients with <50 copies/ml of HIV RNA from week 48 to 72 and available baseline PBMCs were selected from study 903 for analysis. Residual viraemia was determined by measuring HIV RNA levels at weeks 48, 64 and 72 using a modification (limit of detection 2.5 HIV RNA copies/ml) of the Roche Amplicor assay. HIV DNA was measured from stored baseline PBMCs using the Roche Monitor assay. The outcome variable of undetectable residual viraemia was defined as all three measures of HIV RNA below 2.5 copies/ml. Predictors of residual viraemia were examined in uni- variate and multivariate logistic regression analyses.
RESULTS: Baseline mean HIV RNA levels (4.8 ±0.6 log 10 copies/ml), HIV DNA levels (1.9 ±0.5 log copies/µg cel- 10 lular DNA) and CD4 cell counts (320 cells/ml) were similar among patients in the TDF arm (n=55) and d4T arm (n=45) of this cohort. Residual viraemia between 2.5 and 50 copies/ml was detectable in 29/55 (53%) in the TDF arm and 32/45 (71%) in the d4T arm. In the stepwise logistic regression model, there were three independent predictors of residual viraemia – higher baseline HIV RNA (P=0.02), higher baseline HIV DNA (P=0.05) and treatment assignment (P=0.05). HIV RNA ‘blips’ above 50 copies/ml after 72 weeks, but not change in CD4 cell count from week 48 to 96, were associated with detectable levels of residual viraemia.
S66 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 60 median value of 16. Slopes derived from linear regres- sion analysis of HIV-1 RNA values were not significant- Stable, persistent viraemia in ly different from zero, indicating no evidence of decline. A single blip in viraemia (from 4 to 130 copies/ml) was patients with plasma HIV-1 RNA observed in one patient. We investigated baseline and suppressed to less than on-treatment factors for associations with the level of 75 copies/ml on potent persistent viraemia. A trend towards lower levels of viraemia in patients on four or more antiretroviral antiretroviral therapy drugs was evident. No associations were found between the level of viraemia and the baseline CD4 cell A Wiegand1, F Maldarelli1, S Palmer1, M Polis2, count, change in CD4 cell count or duration of anti- J Falloon3, J Mican3, R Davey3, D Rock3, retroviral therapy. S Liu3, A Planta3, J Shen3, R Burke3, JA Metcalf3, J Mellors4 and J Coffin1 CONCLUSIONS: Most patients on antiretroviral therapy with plasma HIV-1 RNA suppressed to less 1 HIV Drug Resistance Program, NCI, NIH; 2 Laboratory of than 75 copies/ml have stable persistent viraemia of Immunoregulation, NIAID, NIH; 3 NIAID/CCMD Clinic, NIH; and 4 0.3–41 copies RNA/ml, as measured by our single Division of Infectious Diseases, University of Pittsburgh, copy assay. Viraemia plateaued within 4–15 months of Pittsburgh, Pa., USA initiating antiretroviral therapy. Additional studies are in progress to assess the relationship between regimen BACKGROUND: Current antiretroviral therapy is effec- potency and the level of persistent viraemia. tive in suppressing HIV-1 RNA to below the limit of detection of FDA-approved assays (50–75 copies/ml). HIV-1 decay and persistence at lower levels of viraemia are poorly understood, because assays with adequate sensitivity, accuracy and precision are not available. To characterize persistent HIV-1 viraemia and its decay in patients receiving antiretroviral therapy, we applied a new HIV-1 RNA assay with single copy sensitivity.
METHODS: Plasma samples were obtained from patients on antiretroviral therapy with HIV-1 RNA less than 75 copies/ml (bDNA version 3.0). HIV-1 RNA levels were measured using a single copy assay (SCA). HIV-1 is pelleted from 7 ml of plasma and RNA is extracted. Reverse transcription and real-time PCR are used to amplify and quantify a specific, conserved HIV gag sequence. Recovery of HIV-1 from plasma is monitored using an avian retrovirus as internal standard. The SCA has a dynamic range of 1–106 RNA copies, a %CV of 30 at 10 copies/ml, a limit of quantification of 1 copy/ml and a limit of detection of 0.3 copies/ml.
RESULTS: Of 25 patients with less than 75 copies HIV-1 RNA/ml by bDNA assay for at least 131 days, 20 had detectable HIV-1 RNA using the SCA. HIV-1 RNA levels ranged 0.3–41 copies/ml, with a median value of 3.9. Analysis of viral decay kinetics in five patients revealed that HIV-1 RNA plateaued a mean of 273 days (range 119–427 days) after initiating anti- retroviral therapy. To investigate the stability of the per- sistent viraemia, we analysed serial samples from six patients over 9–35 months. HIV-1 RNA levels remained stable in each patient during the period of observation; viraemia ranged 3.8 ±2 to 22 ±6.5 copies/ml, with a
XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S67 Los Cabos, Mexico, 10–14 June 2003
ABSTRACT 61 surface density was kept constant and the envelope glycoprotein was varied. Altering CCR5 surface densi- CCR5 cell surface density and ty significantly affected the antiviral potency of CCR5 antagonists against R8.BaL. In the spreading infection, HIV-1 envelope sequences govern IC s for the CCR5 antagonists varied over a range of 95 up to 25-fold, while in the entry assay the IC s varied antiretroviral potency of CCR5 50 antagonists over a range of up to sevenfold. In contrast, the poten- cy of compounds blocking HIV-1 entry at a different step (gp120-CD4 binding or gp41 conformational MD Miller1, JE Lineberger1, G Dornadula1, rearrangement) did not vary significantly in either set- RC Danzeisen1, CR Blau1, RM Danovich1, ting ( METHODS: Potencies of entry inhibitors (DP-178, an anti-CD4 MAb, the small-molecule CCR5 antagonists compounds A and B, and AOP-RANTES) were mea- sured in three settings: 1) spreading infection of a sin- gle virus (R8.BaL, m.o.i. ~0.01) on a panel of SupT1 clones with different CCR5 cell surface densities (rang- ing from ~4000/cell to ~60000/cell); 2) direct viral entry assay using a single virus (R8.BaL) on the panel of SupT1 clones; 3) single-cycle infection of U87-CD4- CCR5 cells by a panel of test viruses bearing envelopes from 415 CCR5-dependent primary HIV isolates. Inhibitors were titrated into each assay and the IC s or 50 IC s were calculated from inhibition curves. 95 RESULTS: Entry inhibitor potencies were measured in a setting where the virus was kept constant and CCR5 surface density was varied, or in a setting where CCR5 S68 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 62 in the 5-month sample was nearly homogeneous by SGS (0.007% sequence diversity), indicating recent Population genetics in HIV-1 infection with wild-type virus. In the same sample, allele-specific PCR revealed a substantial population of super-infection virus with 103N (~20%). The wild-type virus popula- tion became more varied in the 13- and 17-month sam- S Palmer1, M Kearney1, V Boltz1, F Maldarelli1, ples, showing 0.062 and 0.18% sequence diversity, G Achaz3, J Mellors2, E Daar4 and J Coffin1 respectively, while the K103N mutant virus declined to barely detectable levels (0.1%). There was no evi- 1 HIV Drug Resistance Program, NCI, NIH, Frederick, Md.; 2 dence of recombination between the wild-type and Division of Infectious Diseases, University of Pittsburgh, multidrug-resistant populations. Pittsburgh, Pa.; 3 Department of Organismic and Evolutionary Biology, Harvard University, Cambridge, Mass.; and 4 Harbor- CONCLUSIONS: Our results indicate that this patient UCLA Medical Center, UCLA School of Medicine, Torrance, Calif., was initially infected with a multidrug-resistant virus, USA and was subsequently super-infected with a wild-type virus, which dominated the replicating virus popula- BACKGROUND: In recent reports of HIV-1 super- tion, leaving only a very small proportion of resistant infection, both patients had received antiretroviral virus (0.1% or less). Longitudinal samples showed the therapy for primary infection. In the present case, the diversification of the wild-type super-infecting virus patient had primary infection with a multidrug-resis- from a monomorphic population to a more heteroge- tant HIV-1 (Daar et al. 9th CROI). Within 2 months, neous one by 1–1.5 years after infection. The diversifi- viraemia declined to 1000 copies/ml without antiretro- cation of the super-infecting wild-type virus occurred viral therapy. Approximately 4 months later, viraemia slowly by de novo mutation rather than recombination increased to 10000 copies/ml and showed no drug with the initially infecting virus. resistance by standard genotype analysis. We used two different analytical approaches to characterize the pop- ulations of virus and determine their origin. METHODS: Plasma samples were obtained from the patient at 1, 5, 13 and 17 months after the onset of acute retroviral syndrome. DNA sequences from about 20 individual viral genomes were obtained for each sample using single genome RT-PCR sequence (SGS) analysis of the p6 region of gag, protease and the first 900 nucleotides of reverse transcriptase (RT). For minority species quantification, a target region con- taining mutant or wild-type sequence at position 103 was amplified, diluted and used as template for a round of allele-specific PCR, which was performed using either non-discriminating primers to quantify the total virus population or mutant discriminating primers specific for 103N. RESULTS: Phylogenetic analysis revealed two geneti- cally distinct virus populations (6% difference; both subtype B) at different times after infection. SGS analy- sis showed a nearly monomorphic multidrug-resistant viral population 1 month after infection (0.025% sequence diversity) with all sequences containing drug resistance mutations, including the 69SS insertion and K103N. Allele-specific PCR detected wild-type 103K virus at levels indistinguishable from assay background. Analysis of subsequent samples (5, 13 and 17 months after infection) showed wild-type virus with increasing genetic diversity over time. The wild-type virus present XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S69 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 63 retroviral syndrome (ARS). CFR-11 was the only detectable subtype in the plasma at the time of super- Co- and super-infection: persistent infection and later on in follow-up samples. Both sub- types were detectable in proviral DNA after the super- replication of both HIV-1 strains? infection and during the follow-up. Genetic trees per- formed on B and CRF-11 amplicons indicate that the L Perrin1, S Yerly1, M Monnat2, A Telenti3, patients were initially infected with different B strains, A Cavassini3, P Burgisser4 and the Swiss HIV whereas results for CRF-11 were inconclusive due to Cohort Study the high degree of homogeneity of sequences of the CRF-11 amplicons for both the super-infected IVDUs 1 Laboratory of Virology Geneva University; 2 Centre Saint and other IVDUs infected with CRF-11 only. Martin; 3 AIDS Center and Division of Immunology; and 4 Lausanne University, Switzerland CONCLUSIONS: In co-infected patients both HIV-1 subtypes persist during the follow-up, whereas in super- BACKGROUND: In co-infected individuals two or infected patients, despite massive immune activation more HIV-1 strains established close to the time of associated with the ARS at the time of super-infection, PHI, whereas in super-infection a second strain estab- only the second strain was detectable in plasma. lished several months to years after PHI. Does this lead over time to differential replication of the strains? PATIENTS AND METHODS: Three intravenous drug users (IVDUs) co-infected with B subtype and CRF-11, two IVDUs initially infected with B subtype and later on infected with CRF-11. Population sequencing of reverse transcriptase, protease and gag p24, subtype- specific nested PCR with a first generic PCR (amplifi- cation of a 1397 bp gag/pol fragment –1872–3251 in ref to HXB2–) followed by a nested B and CRF-11- specific PCR. Population sequencing of the B and CRF-11 amplicons. RESULTS: Using subtype-specific PCR and limiting dilutions with constant amount of the heterologous subtype (250000 HIV-1 RNA copies) and decreasing amount of the homologous subtype (1000–10 HIV-1 RNA copies), the detection limit was <10 copies of the homologous subtype in 250000 copies of the heterol- ogous subtype, and there was no amplification of the heterologous subtype in presence of 250000 copies. In the three co-infected patients, both CRF-11 and B sub- types were detected in plasma and proviral DNA analysed over a follow-up of 14, 20 and 24 months, respectively. Two of the three patients had viraemia >400000 copies/ml during the follow-up. A genetic tree based on sequences of the B-specific amplicons of 10 IVDUs infected with B subtypes and of the three co- infected patients indicated that two over three co- infected patients were infected with different B strains. The two super-infected patients, initially infected by a B subtype, can be referred as LTNPs since they control their viraemia without treatment to <50 copies/ml and had <500 CD4/mm3 for, respectively, 3 and 5 years before becoming infected with CRF-11. In these two patients, super-infection with CRF-11 was associated with high viraemia, steep drop in CD4 and an acute S70 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 64 numerically for a sample of well-documented resis- tance mutations. HIV viral sex: inbreeding, CONCLUSIONS: Recombination acts on the preva- recombination, drug resistance lence of multipoint-resistant mutants in two opposing and clinical outcome directions, by re-assorting lesser resistant genotypes to increase resistance, and by re-assorting multipoint- C Fraser and RM Anderson resistant mutants with lesser resistant genotypes to reduce resistance. The kinetics of this balance are such Faculty of Medicine, Imperial College London, UK that fitness differences are smoothed, boosting the fre- quency of resistance mutations with high fitness costs, BACKGROUND: Antiretroviral treatment failure is but suppressing the frequency of mutants with com- caused by the accumulation of multiple point muta- pensatory fitness gains. This revised calculation, tions, each conferring a degree of resistance to one or including recombination, makes firmer the prediction more drugs. In previous work, it was estimated that, that all one, two and three but not higher point depending on the relative fitness of resistant viral mutants are present in untreated drug-naive patients. strains, all one- and two-point mutants, and most We predict that treatment regimens that can suppress three-point mutants are already present at low fre- all strains with three or fewer point mutations should, quencies in untreated drug-naive patients. These esti- if adhered to, not fail due to emergent resistance. mates, however, were based on the incorrect assump- tion that all resistant strains are generated by multiple point mutations, not by retroviral recombination. We revise these estimates using published data on rates of recombination in HIV-1, and consider the implications of these new results for the emergence and persistence of drug-resistant viral variants. METHODS: Due to the limited ability of genotypic resistance assays to detect very infrequent genotypes, it is necessary to use a mathematical model to estimate the frequencies of resistant viral strains. We develop a new model of replication for diploid retroviruses, with an arbitrary multiplicity of cell infection and an arbi- trary number of point resistance mutations. In general, this model is complex, but when we assume that wild- type virus is the fittest, we can calculate the equilibri- um frequencies of multipoint-resistant mutants in untreated drug-naive patients. RESULTS: The estimated frequency of any M-point- resistant mutant depends on: 1) the point mutation rate; 2) the relative fitness of all (homozygous and het- erozygous) 1, 2, … M-point mutant viral strains; 3) the M-1 linkage disequilibria between neighbouring resis- tance loci; and 4) the coefficient of inbreeding for the viral population. The point mutation rate (=3.4×10–5 per base) and the recombination rate (=2% per kb) have been empirically estimated. Linkage disequilibria are calculated from genomic distances between resis- tance loci. From published data on the multiplicity of cell infection, we estimate at 45% the coefficient of inbreeding. We explore realistic ranges of linkage dise- quilibria and relative fitness values: recombination reduces the effect of fitness differences between strains. Quantitative results will be illustrated graphically and XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S71 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 65 tion of nonamplifiable virus, possibly due to false-pos- itive tests. Distinguishing between these processes may HIV-1 genetics of intermittent be clinically important. Additionally, this study docu- ments clonal expression of virus in LLV, supporting low-level viraemia proliferation of latently infected cells. NH Tobin1, Y Wang1, GH Learn1, JL McKernan1, GM Ellis1, KM Mohan1, SE Holte1, DM Pawluk1, AJ Melvin1, PL Lewis2, LM Heath1, IA Beck1, WE Naugler1, JI Mullins1 and LM Frenkel1 1 University of Washington, Seattle, Wash.; and 2 Oregon Health and Science University, Portland, Oreg., USA BACKGROUND: Intermittent low-level viraemias (LLV) (50–400 copies/ml) of unclear significance occur during highly active antiretroviral therapy (HAART) of HIV-1 infection. This descriptive cohort study of HIV-1-infected children with LLV at an academic med- ical centre was undertaken to gain insight into the ori- gin and significance of LLV. METHODS: Multiple viral sequences from end-point dilution of LLV plasma and pre- and post-HAART peripheral blood mononuclear cell (PBMC) were examined for drug resistance mutations, genetic diver- gence from the inferred most recent common ancestor of infection (MRCA), genetic diversity and groupings in phylogenetic analysis. RESULTS: Twenty-one (57%) of 37 children on HAART with plasma HIV-1 RNA <50 copies/ml for >1 year had 56 episodes of LLV between 50 and 379 copies/ml (median 77) during 79 patient years. Multiple sequences were derived from 23 of 26 LLV specimens from 11 children. Virus did not amplify from an addi- tional four specimens from three children. Genetic analyses demonstrated that PBMC virus evolved dur- ing HAART in five children. Plasma viruses grouped with the PBMC viruses from early infection in phylo- genetic analysis of virus from nine children. Plasma viruses were monophyletic in env and/or pol encoding reverse transcriptase (RT) in three children. Plasma virus had increased divergence from the MRCA com- pared to PBMC virus in two children, both with fre- quent LLV (32–45% of measurements), and one with increasingly drug-resistant mutants. CONCLUSIONS: These genetic studies of HIV-1 sug- gest three phenomena lead to LLV during effective HAART: 1) viral expression from long-lived proliferat- ing HIV-1-infected cells, without evidence of full cycles of viral replication; 2) viral replication, including the selection of new drug-resistant mutants; and 3) detec- S72 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 66 was due to escape of an archived virus). Three subjects who interrupted lamivudine lost the M184V muta- Limited genotypic and phenotypic tions. Phylogenetic data suggest ‘back’ evolution of virus populations after PTI of reverse transcriptase evolution after interruption of a inhibitors, and an apparent absence of such evolution single therapeutic drug class in within after PI PTI. Finally, loss of resistance was tem- patients with multidrug-resistant porally associated with increasing RC and increasing HIV viraemia in most subjects. CONCLUSION: Interrupting PI therapy in patients SG Deeks, EE Paxinos, T Wrin, R Hoh, with multidrug-resistant HIV is associated with stable F Aweeka, JN Martin, CJ Petropoulos and viremia and persistent levels of PI resistance. Several RM Grant mechanisms may have accounted for the failure of PI resistance to wane. First, the close proximity of RT and University of California, San Francisco, Calif.; ViroLogic, South protease makes genetic recombination unlikely. San Francisco, Calif.; and Gladstone Institute of Virology and Second, viral evolution in presence of PI therapy is Immunology, Calif., USA associated with the accumulation of compensatory mutations that increase viral fitness. Back mutations BACKGROUND: We hypothesized that among treat- may require remodelling within these regions and an ed patients with multidrug-resistant HIV, interruption early decrease in relative fitness; this fitness ‘valley’ of all drugs from a single therapeutic class (‘partial prevents reversion. Third, archived RTI-resistant and treatment interruptions’, PTI) could: 1) maintain par- PI-susceptible virus likely contains fewer RTI-associat- tial viral suppression and its associated immunological ed mutations than more recent variants. Thus, the benefit; 2) prevent overgrowth of wild-type HIV; 3) archived virus may be relatively more susceptible to delay viral evolution; and 4) reduce drug-toxicity and RTIs. Collectively, these data suggest that decreases in drug costs. We have previously reported that interrup- replication capacity associated with PI resistance will tion of protease inhibitor (PI) therapy was often asso- persist in the absence of the inhibitor. ciated with stable viraemia, while interruption of reverse transcriptase inhibitor (RTI) therapy was often associated with rapid rises in viraemia. Here, we describe evolution of viral populations after interrup- tion of a single therapeutic class. METHODS: Eligible subjects had >90% adherence, persistent viraemia (>400 copies/ml) and detectable PI levels. Replicative capacity and resistance testing was performed retrospectively with stored plasma. Phylogenies were constructed using parsimony meth- ods. RESULTS: Nineteen subjects interrupted PI therapy and six interrupted RTI therapy. All patients had evi- dence of a treatment-mediated benefit prior to PTI (median 1.2 log copies/ml decrease in viral load rela- tive to pretreatment levels). Wild-type HIV did not emerge during the PTI period. Among the patients who interrupted PIs, we observed no loss of PI mutations during 16–24 weeks of observation. Although most subjects had been exposed to sequential RTI therapy in the pre-HAART era (and potentially harboured RTI- resistant, PI-susceptible virus), such an archived virus population did not emerge during the first 16–24 weeks of observation; we did, however, observe the emer- gence PI-susceptible virus after week 36 in one subject (phylogenetic analyses suggest that loss of resistance XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S73 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 67 the clones from the two other patients. Two patients responded to the salvage therapy, the two latter Clonal analysis of HIV-1 variants in patients had a virological rebound (including the patient with MDR clones) with reappearance of pre- proviral DNA during treatment interruption mutations in the plasma viruses. interruption in patients with multiple therapy failures CONCLUSION: In patients with multiple virological failure having a TI before a salvage therapy, archiving of pre-interruption resistance mutations in proviral S Boucher1, P Recordon-Pinson1, D Neau2, DNA can be responsible for virological failure after J-M Ragnaud2, M Faure1, H Fleury1 and treatment resumption. B Masquelier1 1 Laboratoire de Virologie, CHU de Bordeaux; and 2 Département des Maladies Infectieuses, CHU de Bordeaux, France BACKGROUND: Treatment interruptions (TI) are being evaluated in antiretroviral (ARV)-treated patients on multiple virological failure. The goal of our study was to determine the importance of archiving of mutated proviral DNA during treatment interruption and its relationship with the virological response to the subsequent salvage regimen. METHODS: Patients on multiple treatment failure but with plasma genotype suggesting some potentially active drugs (≥2) and with a CD4 cell count >300 cells/µl were enrolled in the study. Treatment was interrupted and protease and reverse transcriptase (RT) mutations reversal was checked every 2 months by sequencing the bulk plasma virus. At the day of treatment resumption, DNA was extracted from peripheral mononuclear cells and a fragment corresponding to the protease and the 5′ part of the RT was PCR-amplified and cloned. Twenty clones from each patient were submitted to sequence analysis. Each patient was followed-up 1 year after treatment resumption, and an HIV-1 plasma genotype was processed in case of virological failure (HIV-1 RNA rebound >500 copies/ml). RESULTS: Four patients were enrolled. At the time of treatment interruption, the mean CD4 cell count was at 494/µl and the mean plasma HIV-1 RNA at 4.1 log copies/ml. The mean number of potentially 10 active drugs was 2.6 according to the plasma genotype. After a mean duration of TI of 6.2 months, the mean CD4 cell loss was –225/µl, the mean increase in plas- ma HIV-1 RNA was 5.6 log copies/ml, and RT resis- 10 tance mutations and protease major resistance muta- tions had reversed to wild-type. Clonal analysis of proviral DNA showed an archived D30N protease mutation in 33% of the clones from one patient, mul- tidrug-resistant (MDR) patterns in 14% of the clones from a second patient, and absence of resistance in all S74 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 68 T215Y was identified. Records noted discontinuation of AZT and 3TC, 4 years prior. Additionally, a naive Vaginal HIV-1 shows distinct drug female with mutation M184V present in vaginal HIV-1 but not in plasma was identified. Analysis of HIV-1 V3 resistance mutation patterns sequences showed more variability in plasma com- compared to plasma HIV-1 and pared to vaginal virus regardless of concordance or remain M-tropic despite advanced discordance of drug resistance mutations. In addition, despite the presence of T-tropic viruses in 9–50% of disease clones derived from plasma, all vaginal HIV-1 clones appeared as M-tropic regardless concordance or dis- G Tirado, GR Jove and Y Yamamura cordance of drug resistance patterns. Ponce School of Medicine, Ponce, Puerto Rico CONCLUSION: This cross-sectional study suggests that local selective forces allow distinct viral lineages BACKGROUND: Infection with human immuno- to emerge and evolve independently in the plasma and deficiency virus type 1 (HIV-1) becomes compartmen- the vaginal compartment. Delayed clearance of drug talized in the body and the male/female genital tract resistance mutants was observed in vaginal compart- may serve as viral reservoir. We determined drug resis- ment and these viruses remained M-tropic despite pres- tance mutations in HIV-1 from plasma and vaginal ence of T cell tropism in plasma and advanced HIV-1 secretions to compare the response of HIV-1 to anti- disease thus suggesting the vaginal tract could serve as retroviral therapy and assess the role of vaginal tract as reservoir for M-tropic drug resistant mutants and per- reservoir for drug-resistant variants. V3 loop haps contribute to the transmission of drug resistance. sequences were used to determine T cell tropism and the relationship between plasma and vaginal HIV-1. METHODS: We collected 45 paired blood and vaginal samples from HIV-1-positive females. Viral loads were assessed using Amplicor HIV Monitor. Paired HIV-1 samples were analysed for drug resistance mutations using the TruGene Genotyping Kit. The envelope V3 loop was cloned and sequenced with a home-brew method. RESULTS: Of 45 females, 21 had detectable RNA in both plasma and vaginal secretions. Discordant drug resistance mutation patterns between plasma and vagi- nal HIV-1 were observed in 83.3% (10/12) of cases. Two of 12 (16.6%) females showed identical mutation patterns in both compartments. Three of 10 (30.0%) ‘discordant’ cases, had vaginal HIV-1 showing more resistance mutations than plasma virus. Two of 10 (20.0%) had vaginal HIV-1 presenting fewer muta- tions than plasma virus and appeared to show delayed emergence of drug resistance. Sequences showed lower homology (91.7–97.5%) between plasma and vaginal compartments of females with ‘discordant’ mutation patterns compared to ‘concordant’ females (98.9–99.9%). Treatment history of ‘discordant’ females showed two had mutation T215Y absent from plasma but present in vaginal HIV-1 for 2 and 4 years after stopping zidovudine (AZT). One also had muta- tion M184V absent from plasma but present in vaginal HIV-1 months after cessation of lamivudine (3TC). A third female showing mutation M184V in vaginal HIV-1 but absent in plasma and the contrary for XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S75 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 69 lations in CSF. The composition of three codons with- in V3 was 80% predictive of the anatomic source of Effects of failing antiretroviral virus and suggests that these positions may be impor- tant for virus selection in the CNS. ART and the emer- therapy on the composition of gence of drug resistance was accompanied by co-min- HIV-1 in CSF and blood gling of CSF and plasma sequences. If, as is suspected, HIV env plays a central role in neurotropism and neu- JK Wong1,2, S Letendre1,3, MC Strain1, S Pillai1, rovirulence, selection of drug resistance during failing TM Russell1, CC Ignacio1, HF Gunthard4, ART may have biological consequences beyond those DD Richman1,2, I Grant1,3, JA McCutchan1,3 predicted by virological failure alone. and R Ellis1,3 1 UC San Diego, Calif., USA; 2 VA San Diego Healthcare System, San Diego, Calif., USA; 3 HIV Neurobehavioral Research Center, San Diego, Calif., USA; and 4 University Hospital of Zurich, Switzerland BACKGROUND: HIV env sequence can differ between CSF and blood virus reflecting divergent selec- tive pressures or founder effects in these two compart- ments. We examined evidence for such differences in a cohort of 17 patients who were either untreated or who had developed drug resistance on antiretroviral therapy (ART). METHODS: Paired CSF and plasma were studied from 17 patients (median CD4=145/mm3, CSF RNA=4.3 log copies/ml, Plasma RNA=5.2 log copies/ml). Nine 10 10 to 15 clonal sequences of C2-V3 of env from each spec- imen and population sequencing of pol was performed. Phylogenetic relationships were inferred using a maxi- mum likelihood method. Amino acid sequences differ- ing between CSF and plasma virus were studied by step- wise logistic regression to identify positions in V3 that contributed to compartmentalization. RESULTS: Seven patients were on stable therapy and 10 patients were treatment-naive or off therapy for >3 months. All seven treated patients demonstrated drug resistance in one or both compartments with dis- cordant patterns between CSF and plasma virus pre- sent in two patients. Two of three patients who received prior treatment but were off ART at time of study showed discordant resistance mutations. V3 sequences from eight of 10 patients not receiving ART exhibited partitioning of CSF and plasma sequences while only one of seven patients on ART did so (P=0.0037, Fischer’s exact test). Logistic regression demonstrated that three codons in the V3 loop, partic- ularly codon 308, were predictive of the source of the V3 sequence. Nearly all sequences, regardless of source, were predicted to utilize CCR5. CONCLUSIONS: Among patients not on ART, the majority of patients demonstrated distinct viral popu- S76 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 70 (by bDNA), infectious virus and flow cytometry (CD3, CD4, CD8, CD20). Immune-mediated suppression of RESULTS: Three days after administration of the cM- virulent simian immunodeficiency T807 antibody, CD8 cells were no longer detectable in virus induced by tenofovir blood. One to two weeks later, a transient 20- to 5000- fold increase in viral RNA was observed (peak levels treatment up to ~6 log RNA copies/ml). Virus that was isolated during this time had K65R and the expected fivefold KKA Van Rompay1, R Singh1, C Wingfield2, reduced susceptibility to tenofovir in vitro. Virus levels DL Sodora3, JR Lawson1, B Pahar1, returned to baseline levels upon the return of CD8 KA Reimann4, ML Marthas1 and lymphocytes. At a later stage, when tenofovir treat- N Bischofberger5 ment was withdrawn for 7 weeks for three animals, viraemia increased slowly but returned to baseline lev- 1 California National Primate Research Center, University of els upon re-initiation of tenofovir treatment. California, Davis, Calif.; 2 Bayer Diagnostics Reference Testing Lab, Bayer Diagnostics, Berkeley, Calif.; 3 University of Texas CONCLUSION: In the absence of CD8 lymphocytes, Southwestern Medical Center, Dallas, Tex.; 4 Division of Viral K65R viral mutants are highly replication-competent Pathogenesis, Beth Israel Deaconess Medical Center, Boston, despite concomitant tenofovir treatment. This suggests Mass.; and 5 Gilead Sciences, Foster City, Calif., USA that suppression of replication of K65R SIV mutants is under relatively little direct control of tenofovir, but is BACKGROUND: Infection of rhesus macaques with mediated mainly by CD8-mediated immune responses. the virulent uncloned SIVmac251 isolate is a highly Continued tenofovir treatment is required to maintain relevant animal model of HIV infection. Unlike SHIV optimal suppression of viraemia. Our results help to isolates, long-term control of SIVmac251 infection has explain the absence of viral rebound that has been been very difficult to achieve. Accordingly, the efficacy described in tenofovir-treated humans who developed of the reverse transcriptase (RT) inhibitor tenofovir to K65R HIV-1 mutants. suppress viraemia in SIVmac251-infected macaques has been unprecedented. During prolonged tenofovir monotherapy, some animals showed a progressive decrease of viraemia to undetectable levels despite the emergence of viral mutants with K65R (and other, pre- sumably compensatory mutations) in RT, and ~five- fold reduced in vitro susceptibility to tenofovir. Once viraemia is low or undetectable, this status can be maintained for a long time (2 to >7 years), but only as long as tenofovir treatment is continued. This efficacy has not been achieved with any other drug in this ani- mal model. Our previous experiments demonstrated that K65R SIV mutants are not sufficiently attenuated to explain these results. Since these tenofovir-treated animals have CD8 cell-mediated antiviral immune responses, we decided to investigate the role of these immune responses on the efficacy of tenofovir to main- tain low viraemia. METHODS: Four SIV-infected rhesus macaques had been on long-term tenofovir treatment (9 months to 6 years), had K65R viral mutants, but had reached low or undetectable viraemia (<125 copies/ml). These four animals were depleted of CD8 cells while tenofovir treatment was continued. In vivo depletion of CD8 lymphocytes was performed through administration of the well-characterized cM-T807 monoclonal antibody. Animals were monitored for plasma viral RNA levels XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S77 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 71 observed upon treatment with both anti-CD63 and the CXCR4 inhibitor AMD3100 together. CXCR4-utiliz- A potential role for CD63 in ing (X4) isolates were not inhibited by anti-CD63 and inhibition of infection was not observed in PBL or cell CCR5-mediated HIV-1 infection lines. The block to productive infection of Mφ of macrophages occurred prior to reverse transcription, as determined by quantitative HIV DNA PCR assay at 24 h. JJ von Lindern1, D Rojo1, K Grovit-Ferbas2, Transfection of CD63 along with CD4 and chemokine C Cheng Deng1, MR Ferguson1, JM Decker3, receptors into low CD63-expressing quail QT6 cells A Singh4, RG Collman4 and WA O’Brien1 did not affect fusion of these cells with QT6 cells expressing HIV Env. In addition, fusion of primary Mφ 1 University of Texas Medical Branch, Galveston, Tex.; 2 with HIV Env-expressing 293T cells was not inhibited University of California at Los Angeles and the UCLA AIDS by anti-CD63 mAb. Treatment of Mφ with anti-CD63 Institute, Los Angeles, Calif.; 3 Howard Hughes Medical Institute, did not affect levels of CD4 or CCR5 expression or β- University of Alabama at Birmingham, Birmingham, Ala.; and 4 chemokine secretion. CD4-CD63 membrane associa- University of Pennsylvania, Philadelphia, Pa., USA tions were not detected in uninfected primary Mφ, as assessed by coimmunoprecipitation-western blot. BACKGROUND: HIV cell tropism is largely related to chemokine receptor usage, but the molecular details of CONCLUSIONS: It is likely that tetraspan membrane infection may differ between macrophages (Mφ) and glycoprotein CD63 is involved in HIV infection of CD4 lymphocytes. Identification of cellular factors macrophages with R5 viruses, either through associa- that are involved specifically in HIV infection of Mφ tions with CCR5 or at a post-fusion step of virus entry. (in addition to CD4 and chemokine receptors) may A potential mechanism of action is CD63-mediated lead to novel antiretroviral therapies. stabilization of membrane protein organization impor- tant for Mφ infection. Further studies may reveal suit- METHODS: Since culture-adherent Mφ become more able therapeutic targets. susceptible to HIV infection over 1 week, Mφ-specific infectivity factors were sought by screening a myeloid cell-specific monoclonal antibody (mAb) library (n=120) for mAb with increased binding to 6 day cul- tured Mφ, as compared with fresh monocytes. Such antibodies (n=15) were assessed for their ability to inhibit HIV infection by quantitative PCR at 24 h and by p24 ELISA at day 7. Quail QT6 cell–cell fusion was assessed by luciferase expression 3 days after co-trans- fection of T7-driven luciferase along with CD4, chemokine receptor and CD63 plasmids into one set of cells, and transduction of T7 polymerase and HIV Env into a second set of cells. To assess fusion in macrophages, 293T cells were transduced with recom- binant vaccinia virus expressing T7 polymerase, trans- fected with SP6-driven luciferase and T7-driven Env plasmids, and then mixed with primary Mφ transduced with SP6 polymerase-expressing recombinant vaccinia virus, in the presence or absence of anti-CD63 or con- trol mAb, and fusion was measured by luminometry. Cell membrane associations between CD63 and CD4 or chemokine receptors were assessed by coimmuno- precipitation. RESULTS: Anti-CD63 pretreatment inhibited infection of primary Mφ by seven different Mφ-tropic (R5) strains. Partial or complete inhibition of infection by dual-tropic (R5X4) strains was observed with anti- CD63 pretreatment alone, with consistent inhibition S78 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 72 observed in viruses from all these subjects. All patients were wild-type homozygous for ccr5. A net gain in the Mechanisms underlying a sustained CD4 count of 69, 71, 221 and 510 CD4 cells/µl was seen in all of them despite virological failure. A signif- CD4 T cell recovery despite the icant lower immune response (low level of activation, emergence of resistance in CTL reactivity and T cell turnover) was found in ENF- antiretroviral-experienced patients treated patients compared to controls. Moreover, a higher percentage of naive T cells was found in sub- on prolonged enfuvirtide treatment jects on ENF. A patient showed a shift from X4 to R5 viruses, being on ENF, and experienced the greatest E Poveda1, B Rodés1, JM Benito1, CD4 T cell rise. At the end of follow-up, all ENF-treat- MA Muñoz-Fernández2, M López1, ed patients harboured R5 NSI viruses. J González-Lahoz and V Soriano1 CONCLUSIONS: A single mutation within HR1 (aa 1 Instituto de Salud Carlos III, Madrid; and 2 Hospital General 36–45) results in virological failure to ENF. Individuals Universitario Gregorio Marañón, Madrid, Spain on ENF may experience CD4 T cell rises despite viro- logical failure due to a redistribution of CD4 T cells, BACKGROUND: Enfuvirtide (ENF) blocks fusion of hypothetically driven by low levels of T cell activation. viral and cellular membranes by binding to the HR1 Furthermore, selection of R5 viruses by ENF may result domain of viral gp41. Changes within the HR1 region in a lower HIV deleterious effect on thymic function. (aa 36–45) have been associated with resistance to ENF. In vitro studies have shown that R5 viruses are less susceptible to ENF, whereas low R5 co-receptor expression favours response to ENF. We previously found in patients on ENF a significant CD4 T cell recovery despite virological failure and selection of resistant viruses (AIDS 2002; 16:1959–1961). Herein, we report the evolution of gp41-genotypes along 80 weeks of ENF therapy in these patients and determine the mechanisms underlying their CD4 T cell gain. METHODS: Four heavily antiretroviral-experienced patients with genotypic/phenotypic resistance to both protease/perverse transcriptase inhibitors enrolled in the Phase III clinical trial of ENF (TORO 2) were stud- ied. Pol and gp41 sequences from each patient were analysed at baseline and every 2 months thereafter. Plasma viraemia, CD4 T cell counts, ∆32-ccr5, sin- cytium-induced (SI) virus phenotype, proportion of clones harbouring ENF resistance mutations, activa- tion markers and apoptosis in naive and memory T cells, T cell turnover, HIV-specific CD8 responses and TRECs were examined. Cellular origin of plasma HIV particles was investigated. Patients with similar char- acteristics not exposed to ENF were used as controls. RESULTS: All patients experienced a significant viral load drop after beginning ENF therapy (>1 log) but rebounded shortly thereafter; high levels of plasma viraemia have persisted since then. HR1 mutations were selected at failure in all cases (N43D in three sub- jects and G36V/D in another) and persisted in most of the clones examined along 80 weeks. No other changes associated with resistance appeared within HR1 dur- ing follow-up. Multiple polymorphisms in HR2 were XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S79 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 73 follows: PI 0.036, NRTI 0.007 and NNRTI 0.002. Adjustment of the multivariable model for baseline Wide variation in pro/pol viral load and the duration of infection, as indexed by LS EIA, did not alter the analysis’ inferences or signif- replication capacity in recently icance trends. Among primary drug resistance PR transmitted HIV-1 is conferred in mutations, the V82A mutation was found to be most part by protease inhibitor strongly associated with variation in RC (RC was an average 31 RC points lower when V82A was present, resistance mutations P=0.03). Among RT mutations, M184IV was margin- ally associated with lower RC (18 point lower RC, JD Barbour1,2, T Wrin3, FM Hecht2, P=0.07). In a model including both V82A and NS Hellmann3, CJ Petropoulos3, MR Segal2 M184IV, M184IV and V82A each remained associat- and RM Grant1,2 ed with lowered RC. 1 Gladstone Institute of Virology and Immunology; 2 University CONCLUSIONS: Resistance mutations accounted for of California, San Francisco, San Francisco, Calif.; and 3 a small fraction of the variation in RC in this popula- ViroLogic, Inc., So. San Francisco, Calif., USA tion of newly infected patients (although the assay uses patient-derived pro/pol gene segments in an isogenic BACKGROUND: Multidrug-resistant viraemia during background). Genetic interactions, and mutations not combination therapy has been associated partial sup- associated with drug resistance, may also contribute to pression of virus replication and decreased replication variations in RC. Low RC was associated most strong- capacity, which may decrease transmissibility. Wide ly with PI resistance. Impaired transmissibility of PI- variation in pro/pol replication capacity (RC) has been resistant viruses could explain the relative scarcity of observed among persons recently infected with HIV-1. PI resistance among recently infected persons. The viral genetic determinants of RC are not known. METHODS: Evolving serological responses (LS EIA) were used to identify recent infections among San Francisco Options study participants. All patients were antiretroviral drug treatment-naive. Genotypic evi- dence of drug resistance was identified using popula- tion sequencing (TRUGENE). RC was measured using a single-cycle assay (ViroLogic). The number of muta- tions associated with resistance to each class of drug [nucleoside reverse transcriptase inhibitor (NRTI), non-NRTI (NNRTI) or protease inhibitor (PI)] was counted for each genotype and entered into multivari- ate regression analysis. Stepwise multivariable regres- sion analysis was used to assess the contribution of individual sites to variation in RC. RESULTS: At study entry, RC varied widely (range 0.5, 113), averaging 41%. Thirty-five of 191 (18.3%) sub- jects had genotypic evidence of drug resistance to at least one class: 7 of 191 to PIs, 22 of 191 to NRTIs and 16 of 188 to NNRTIs. In a multivariable regression model, the number of primary PI resistance mutations was significantly associated with lowered RC (RC was an average 16.5 points lower for each additional pri- mary PI mutation, P=0.004), while the primary NRTI (P=0.2) and NNRTI (P=0.6) resistance mutations were not. The r2 for this multivariable model was 0.045, indicating that only 4.5% of the variation in RC was accounted for by the combination of terms. The rela- tive contributions of each term to the model r2 were as S80 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 74 low RC values included K14R (18.8%, 3.8%, 0.21, P=0.002) and I93L (35.9%, 21.2%, 0.59, P=0.048). Modulation of replication capacity Analysis of a much larger set of >1000 WT viruses looking at Pr and RC correlations, K14, R41N and in drug sensitive clade B HIV-1 by K43 were associated with low RC and L63P, L63T and protease and the C-terminal end of L72V were associated with high RC. However, only gag R41N (P=0.049) and L63P (P=0.001) carried associa- tions that were statistically significant. M Bates, C Chappey, S Bates and N Parkin CONCLUSIONS: Univariate analyses of a database of WT viruses demonstrates that selected mutations in ViroLogic, Inc. South San Francisco, Calif., USA gag and in Pr are associated with high or low RC val- ues. Multivariate analyses utilizing larger datasets may BACKGROUND: Using a modified PhenoSense Assay provide additional insights. (ViroLogic, Inc.), which generates a recombinant virus including 3′ gag (aa 418–500), protease (Pr), reverse transcriptase (RT) (aa 1–305); it is possible to measure the replication capacity (RC) of HIV-1 isolates. It has been shown that wild-type (WT) HIV-1, defined as viruses that have neither genotypic nor phenotypic evi- dence of resistance, exhibit a very broad range of RC values. We asked whether mutations in the C-terminal region of gag derived from patient sequences may account for the extremely broad range of RC values seen in WT viruses. METHODS: Utilizing the ViroLogic phenotype-geno- type database (~17000), we selected WT viruses for which RC data were available (1063). The population was normally distributed with a range of 0.8–251.7%. We further selected viruses which had RC values less than the tenth percentile (n=64), greater than the 90th percentile (n=80), and close to the median (97.4% ±1.4%, n=24) of the RC distribu- tion for those WT viruses. Non-clade B viruses were excluded. We determined the gag sequences and per- formed univariate analyses (Fisher’s exact test) to look for specific mutations in gag or protease that were associated with either high or low RC. RESULTS: Mutations in gag that were associated with high RC values included insertions at 458 (% with insertions, low RC=6.3, % with insertions, high RC=25, high/low=4.0, P=0.002), and mutations at 418 (% mt, low=10.9, % mt, high=35.6, h/l=3.25, P=0.0005), 470 (12.5%, 30.8%, 2.46, P=0.009), and 473 (34.4%, 51.9%, 1.51, P=0.038). Gag mutations associated with low RC values included those at posi- tions 429 (12.5%, 2.9%, 0.23, P=0.022), 483 (37.5%, 20.2%, 0.54, P=0.019), and 484 (15.6%, 2.9%, 0.19, P=0.005). Mutations in Pr that were associated with high RC values were L10I (4.7%, 16.3%, 3.49, P=0.027), I72 (14.1%, 27.9%, 1.98, P=0.039), I64V (17.2%, 31.7%, 1.85, P=0.047), and L63 (67.2%, 82.7%, 1.23, P=0.025). Pr mutations associated with XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S81 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 75 changes in VL and time after treatment interruption. There was no evidence for an increase in the percent- Drug-resistant phenotype is age of activated CD8 T cells after phenotypic switch. We next assessed change in T-cell activation in eight associated with decreased in vivo subjects who interrupted just the PI component of a T-cell activation independent of partially suppressive regimen. Phenotypic evidence of changes in viral replication among PI resistance persisted in all subjects for 24 weeks. PTI of PIs was associated with a mean increase of 0.32 log patients discontinuing antiretroviral CD38 molecules/CD8 T cell (P=0.03) in an unadjusted therapy analysis. However, after adjustment for VL, there was only a trend toward increased CD38 expression with PW Hunt1, JN Martin1, E Sinclair1, PTI of PIs (mean increase of 0.27 log molecules/cell, TB Neilands1, RM Grant1, NS Hellmann2 and P=0.12). SG Deeks1 CONCLUSION: The observed reduction in T-cell acti- 1 University of California, San Francisco, Calif.; and 2 ViroLogic, vation among patients with MDR viraemia is mediat- South San Francisco, Calif., USA ed through both partial viral suppression and a change in the viral phenotype, though we cannot exclude an BACKGROUND: Many antiretroviral-treated patients independent drug-effect on T-cell activation. These maintain treatment-mediated CD4 gains despite the data suggest that suppression of viraemia below a pre- continued presence of drug-resistant viraemia. The therapy baseline and the maintenance of a drug-resis- decreased levels of T-cell activation associated with tant phenotype are necessary to preserve the immuno- drug-resistant viraemia may explain this immunologi- logical benefit of therapy. Moreover, these data pro- cal benefit. However, it is unclear whether this reduc- vide the most direct in vivo evidence that the drug- tion in T-cell activation is mediated through a direct resistant variant is less capable of causing generalized drug effect, partial viral suppression, or a change in T-cell activation. viral phenotype. METHODS: Changes in T-cell activation, plasma HIV RNA levels (VL) and phenotypic drug susceptibility were prospectively assessed in patients with high-level multi-class drug-resistant (MDR) viraemia undergoing either structured treatment interruption of all anti- retrovirals (STI) or partial treatment interruption (PTI) of only protease inhibitors (PIs). The extent of T-cell activation was determined by the percentage of HLA- DR+CD38+CD4 and CD8 T cells and the density of CD38 on CD8 T cells. Changes in T-cell activation were analysed using generalized estimating equations. RESULTS: The 20 subjects undergoing STI had a medi- an of 7% (IQR: 5–15%) activated CD4 T cells, 13% (IQR: 7–28%) activated CD8 T cells and 3.5 log copies 10 of HIV RNA/ml of plasma at baseline. Phenotypic switch to wild-type drug susceptibility occurred after a median of 8 weeks (IQR: 4–11 weeks). After treatment interruption and prior to phenotypic switch, there was a mean increase of 0.2 log% activated CD4 T cells (P=0.12) and 0.3 log% activated CD8 T cells (P=0.05). These increases were no longer significant after adjust- ing for concurrent increases in VL (P=0.54 and 0.95 after adjustment, respectively). However, phenotypic switch to wild-type drug susceptibility was associated with an independent mean increase of 0.3 log% acti- vated CD4 T cells (P=0.004), even after adjustment for S82 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 76 evidence that amino acid sites 12, 37 and 63 are posi- tively selected in both subtypes B and C. Amino acid Effects of amino acid and site 19 is selected in subtype C, while amino acid 93, known to lie within a predicted HLA A3-restricted epi- synonymous polymorphisms in tope, is under selection in subtype B only. Amino acid HIV-1 protease on viral fitness 15 appears to be selected only in the Ethiopian popu- lation. SDW Frost1, SL Kosakovsky Pond2, Z Grossman3, E Daar4, J Condra5, CONCLUSIONS: Amino acid targets of CTL selection DD Richman1, SJ Little1 and AJ Leigh Brown6 are expected to vary between human populations that are genetically distinct due to differences in HLA allele 1 University of California San Diego, San Diego, Calif., USA; 2 frequencies. We find evidence of both: 1) population- University of Arizona, Tucson, Ariz., USA; 3 Natl. HIV Reference specific; and 2) subtype-specific selection. 1) is most Lab, PHL, MOH, Tel Hashomer, Israel; 4 Harbor-UCLA, Los likely to be HLA-associated, while 2) could be both Angeles, Calif., USA; 5 Merck Research Laboratories, West Point, HLA-associated and influenced by other factors, such Pa., USA; and 6 University of Edinburgh, Edinburgh, Scotland, UK as subtype-specific differences in RNA folding. The large difference among codons in synonymous varia- BACKGROUND: Several amino acid sites that are tion suggests some of these changes are not selectively polymorphic in baseline samples are known to be asso- neutral and that selection may act at the RNA as well ciated with the response to protease inhibitor (PI) sus- as the protein level in the pol gene. ceptibility. Some of these polymorphisms differ between subtypes. Recently it has been suggested from cross-sectional studies of variation in HLA-restricted CTL epitopes, that baseline variation in pol has arisen in response to CTL selection for escape mutants. To examine this hypothesis, we have directly tested for selection on individual codons in HIV-1 protease (PR) in sequences from individuals from four different pop- ulations who have never received antiretrovirals. METHODS: Datasets: 1) Merck 035: baseline consen- sus sequences generated from an average of six clones from 89 chronically infected patients recruited during 1995 to the Merck 035 trial; 2) a baseline direct con- sensus sequence of 135 primary HIV cases recruited in San Diego and Los Angeles between 1997 and 2002; 3) 92 subtype C sequences from southern Africa (KwaZulu Natal, Zimbabwe and Zambia); and 4) 51 sequences from antiretroviral-naive Ethiopian immi- grants to Israel. Analysis: maximum likelihood phylo- genetic method for estimation of synonymous and non-synonymous evolutionary rates for each of the 99 codons in PR. The fitted models permitted multiple classes of substitution rate for synonymous sites as well as non-synonymous substitutions. RESULTS: Synonymous substitutions are usually assumed to be selectively neutral and, therefore, to occur at a constant rate. Selection in nucleotide sequences is detected by comparing rates of synony- mous and non-synonymous substitution. We have found variation between codons in synonymous rates of over 20-fold in PR, indicating synonymous substitu- tions cannot be assumed to be neutral. Taking account of the variation in synonymous rates, we find strong XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S83 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 77 (NL4-3 or heterologous patient-derived regions) into MDR2 resulted in chimeras with a modest but gag/pol/vif Influence of gag and pol regions on reproducible reduction of infectivity and slower growth kinetics in CD4 T cell culture. Conversely, replication efficiency of a insertion of p6-PR-RT of MDR2 into NL/YU2 result- multidrug-resistant HIV-1 variant: ed in a similar reduction of replication with respect to a systematic assessment NL/YU2. V Simon1, N Padte1, I Driver1, NC Wang1, CONCLUSION: These results suggest that relevant M DiMascio2 and M Markowitz1 compensatory changes occur both within and outside the drug resistance conferring regions, and that the interplay of these MDR regions results in optimal 1 Aaron Diamond AIDS Research Center, The Rockefeller replication efficiency as demonstrated by University, New York, NY; and 2 National Institute of Allergy and MDR2 . Furthermore, the replacement of MDR2 gag/pol/vif Infectious Diseases, NIH, Bethesda, Md., USA gag and PR genes by wild-type sequences results in a loss of replication capacity providing a possible expla- BACKGROUND: Some HIV-1 isolates with reduced nation for the observed genetic stability and persis- susceptibility to two or more drug classes (multidrug- tence of some drug-resistant isolates in vivo. resistant, MDR) replicate in a drug-free environment to levels comparable to those of wild-type viruses. In order to delineate the underlying compensatory mech- anisms that contribute to restored replication capacity of an MDR isolate, we examined the relative contribu- tion of gag and pol regions. METHODS: The primary viral isolate MDR2 was obtained from a newly infected individual 79 days after the onset of symptoms of acute HIV-1 infection. Despite a number of resistance-associated mutations in protease (PR: K20I, M36I, M46I, A71I, G73S, L90M) and reverse transcriptase (RT: M41L, E44D, T69D, V118I, L210W, T215Y), MDR2 was found to repli- cate in the absence of drugs to high levels both in vivo and in vitro. We constructed MDR2 encoding gag/pol/vif the 5′ region of the viral genome (gag, pol, vif) of iso- late MDR2 in a NL4-3-derived CCR5-tropic molecu- lar backbone (NL/YU2), as well as a panel of chimeric viruses encoding different segments of MDR2 . gag/pol/vif Infectivity was assessed using single-cycle infectivity assays, and replication kinetics were studied using par- allel infections of mitogen-stimulated T-lymphocytes. In addition, reciprocal chimeras encoding wild-type PR-RT in the MDR2 molecular background were gag/pol/vif analysed for infectivity and replication efficiencies. RESULTS: Single-cycle infectivity and growth kinetics of MDR2 were comparable to those observed gag/pol/vif for NL/YU2. Introduction of NL4-3 gag into MDR2 resulted in a fivefold decrease of infec- gag/pol/vif tivity. The level of infectivity was further reduced when both gag and PR of MDR2 were replaced by the gag/pol/vif corresponding NL4-3 sequences. These differences were amplified over multiple rounds of replication as demonstrated by parallel infections of CD4 T-lympho- cytes. The introduction of wild-type PR-RT regions S84 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 78 (WT) was seen in 12/27 (44%) samples with week 2 mutations where a week 8 sample was available, with Rapid flux in non-nucleoside emergence of new mutations in 2/27 (7%). Most com- mon mutations at week 8 included K103N in 9/32 reverse transcriptase inhibitor (28%), V106A and Y181C in 2/32 (6%) each. Of resistance mutations among seven women with mutations at week 8 for whom late subtype C HIV-1-infected women samples were available, all, save one, with K103N had after single dose nevirapine reverted to WT. CONCLUSION: Samples obtained within 2 weeks of R Kantor, E Lee, E Johnston, P Mateta, SD NVP demonstrated a high frequency of NNRTI L Zijenah, Y Maldonado and D Katzenstein for resistance mutations. Dual mutations, observed in the HPTN 023 Study Team some samples may be mixtures of competing single mutants. There is evidence for rapid reversion to WT Stanford University, Stanford, Calif., USA and University of of plasma RT by 8 and 24 weeks. Although Y181C Zimbabwe, Harare, Zimbabwe predominates in early samples, K103N is retained as the predominant mutation detected at 8 weeks. BACKGROUND: A single dose (SD) of 200 mg of Sequencing multiple clones from sequential time- nevirapine (NVP) to reduce vertical transmission has points will be useful for quantification and linkage of been shown to select for non-nucleoside reverse tran- mutations. The change over time in the proportion of scriptase inhibitor (NNRTI)-associated reverse tran- specific mutations may serve as an indication of in vivo scriptase (RT) mutations in subtypes A and D fitness relative to WT RT in subtype C HIV-1. (Eshleman) and in subtype B (Cunningham). In sub- type C-infected women, specific NNRTI mutations, the rate of selection, their persistence and fitness have not been defined. METHODS: We analysed RT sequences from 34 women from Chitungwiza, Zimbabwe, who partici- pated in HPTN 023, a Phase I study where women received SD NVP at onset of labour. Plasma was col- lected at 0, 2, 8 and 20–32 weeks post-partum. HIV-1 plasma RNA was quantified by ultrasensitive Roche 1.5 Amplicor Monitor and converted to log copies/ml. 10 HIV-1 RNA was isolated with the Qiagen RNA Kit and sequenced with the TruGene (Visible Genetics) kit using version 1.5, research-only, primers. Sequences were examined for subtype and recombination by bootscanning and assessed for NNRTI-associated mutations codons 98, 100, 101, 103, 106, 108, 179, 181, 188, 190, 225, 227, 230 and 238. RESULTS: Plasma RNA levels, a median of 4.4 log 10 copies/ml, were not significantly different at 0, 2 and 8 weeks post-NVP. Overall, sequences from 25/33 (76%) women harboured NNRTI-associated muta- tions at any time-point, 21/28 (75%) available sam- ples at 2 weeks, 11/32 (34%) at 8 weeks and 1/8 (13%) available samples at 24 weeks. In 2 and 8 week post-NVP samples, two or more mutations were found in 11/28 and 5/32 sequences, and a single mutation in 10/28 and 6/32, respectively. At 2 weeks, prevalence of specific mutations included Y181C in 16/28 (57%), K103N in 7/28 (25%), V106A/M in 5/28 (19%) and Y188C in 4/28 (14%). By week 8 reversion to wild-type XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S85 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 79 that were not detected by population sequencing. As many as seven different NVPR mutations were detect- HIV-1 variants with diverse ed in plasmids from a single sample. Three of the women had plasmids with more than one NVPR muta- nevirapine resistance mutations tion, confirming that those mutations were genetically emerge rapidly after single-dose linked. nevirapine: HIVNET 012 CONCLUSION: Cloning and characterization of indi- vidual HIV-1 variants reveals rapid selection of diverse SH Eshleman1, D Jones1, L Guay1, P Musoke2, subpopulations of HIV-1 after single-dose NVP. Some F Mmiro2 and JB Jackson1 variants contained more than one NVPR mutation and many contained NVPR mutations that were not detect- 1 The Johns Hopkins Medical Institutions, Baltimore, Md., USA; ed by population sequencing. Detection of variants and 2 Makerere University, Kampala, Uganda with NVPR mutations only 7 days after single-dose NVP suggests that those variants were present at low BACKGROUND: The HIVNET 012 trial in Uganda levels prior to NVP administration. Further studies of showed that single-dose nevirapine (NVP) can prevent NVPR in women and infants receiving single-dose HIV-1 mother-to-child transmission (MTCT). Women NVP may help optimize use of NVP for prevention of received a single dose of NVP in labour and infants MTCT. received a single dose of NVP within 72 h of birth. The safety, efficacy, simplicity and low cost of the HIVNET 012 regimen make it attractive for prevention of MTCT in resource-limited settings. In HIVNET 012, we detected NVP resistance (NVPR) mutations in 21 (19%) of 111 women 6–8 weeks after NVP using a commercial genotyping system based on population sequencing. NVPR mutations were also detected in some women as early as 7 days after single-dose NVP. Some samples had more than one NVPR mutation detected by population sequencing. In this study, we cloned and sequenced individual HIV-1 variants to examine the spectrum and genetic linkage of NVPR mutations in plasma 7 days after single-dose NVP administration. METHODS: Samples collected 7 days after single-dose NVP were selected from five women in HIVNET 012 for analysis of individual HIV-1 variants. Those sam- ples had 1–4 NVPR mutations detected by population sequencing using the ViroSeq HIV-1 Genotyping System. DNA amplified from those samples using the ViroSeq system was cloned and plasmids were isolated following electroporation into uracil-N-glycosylase negative (UNG-negative) Escherichia coli. At least 10 plasmids were isolated and sequenced from each plas- ma sample. RESULTS: Sequencing of cloned plasmids from the five women revealed diverse patterns of NVPR mutations. Some plasmids lacking NVPR mutations (wild-type) were isolated from each plasma sample. The other plasmids contained NVPR mutations, including K103N, V106A, V108I, V179D, Y181C, G190A, G190S, Y188C and Y188L. In three of the five women, NVPR mutations were detected in plasmids S86 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 80 DISCUSSION: In the source patients, only MDR virus- Persistence of multidrug-resistant es were detected and therefore transmitted. In the index patients, an expansion of predominant MDR HIV-1 without any antiretroviral quasispecies and the ‘archival’ of all resistance muta- treatment 2 years after sexual tions were observed. These results explain the persis- transmission tence of mutations and suggest the high difficulty to return to a wild-type viral population sensitive to an antiretroviral treatment. The treatment of index C Delaugerre1, L Morand-Joubert2, ML Chaix3, patients is limited and the major risk is the transmis- O Picard2, AG Marcelin1, V Schneider4, sion of these MDR viruses. A Krivine5, A Compagnucci6, C Katlama1, PM Girard2 and V Calvez1 1 Pitié-Salpêtrière Hospital; 2 Saint-Antoine Hospital; 3 Necker- Enfants Malades Hospital; 4 Rothschild Hospital; 5 Saint Vincent-de-Paul Hospital; and 6 Hôtel-Dieu Hospital, Paris, France BACKGROUND: The prevalence of transmission of multidrug-resistant (MDR) virus, rarely observed before, has increased recently in primary HIV-1 infec- tion. Contrasting, the prevalence of resistance muta- tions was lower at distance from seroconversion than in primary infection. We reported two cases of sexual transmission of MDR virus that persisted for at least 2 years without any selective antiretroviral pressure in HIV-1-infected patients. PATIENTS AND METHODS: Two patients were cont- aminated recently with their HIV-1-infected partners, who had received, before the contamination, all avail- able antiretroviral drugs and who died within 6 months after the contamination. The resistance mutations analysis was performed by clonal sequencing of 1.2 kb of polymerase gene (complete protease and RT codons 1–230) in plasma of index and sources patients. A phy- logenetic analysis was done. Bulk sequencing of HIV-1 DNA was performed in the PBMC of index patients. RESULTS: Genotypic testing performed in index patients at time of seroconversion showed resistance mutations to three classes of drugs. All mutations were present on the same genome viral and all quasispecies carried all mutations. No wild-type virus was detected. The same results were found in source patients shown that all mutations were transmitted. Phylogenetic analysis confirmed the epidemiological linkage between source and index patients and showed a high homogeneity of viral population in each sample. In the index patients, all mutations persisted at least 2 years without antiretroviral treatment. Moreover, the resis- tance mutations were all archived in cellular reservoir. Viral load and CD4 count of index patients remained unchanged during 2 years of follow-up. XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S87 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 81 patients over time in the absence of therapy. On aver- age they lost 46 cells/year. Impact of transmission of CONCLUSIONS: Transmission of drug-resistant HIV drug-resistant HIV viruses on viral viruses does not seem to affect initial CD4 cell counts load, CD4 counts and CD4 decline nor viral load values in recent HIV seroconverters. in recent seroconverters Moreover, no effect is seen on CD4 decline overtime along 3 years follow-up in the absence of antiretroviral therapy. C de Mendoza1, M Ortiz2, S Pérez-Hoyos3, A Corral1, A García-Saínz2, J del Amo4, J del Romero5, V Soriano1, C Rodríguez5 and the Study Group of Seroconverters of Madrid 1 Service of Infectious Diseases, Hospital Carlos III, Madrid; 2 Centro Nacinal de Microbiología, ISCIII, Majadahonda; 3 EVES, Valencia; 4 Universidad Miguel Hernández, Alicante; and 5 Centro Sandoval, Madrid, Spain BACKGROUND: Drug-resistant HIV viruses have been associated with a partial preservation of CD4 cell counts in subjects failing antiretroviral therapy. Moreover, a recent report (Grant et al. JAMA 2002) has highlighted that drug-naive patients infected with resistant strains may harbour higher CD4 counts, hypothetically due to a lower replicative capacity of these viruses. We assessed the impact of transmission of drug-resistant viruses on CD4 counts and viral load at the initial establishment of HIV infection and on CD4 decline overtime. METHODS: Persons recently infected with HIV-1 and having HIV genotype at the time of diagnosis were selected from an established seroconverter cohort in Madrid. Demographics, virological and immunologi- cal data were collected at the beginning and periodi- cally thereafter. Random effects models were used to assess CD4 decline overtime. RESULTS: Forty-six drug-naive recent seroconverters with available HIV genotyping at the time of first HIV diagnosis were identified between 1991 and 2001. Overall, 91% were men, 74% were infected through homosexual contact and the median time of follow-up after seroconversion was 3.5 years. Nine (20%) of them showed drug-resistant strains at baseline. The distribution of resistant genotypes was: M41L (3), T69N (1), K70R (1), L210W (2), T215Y (4), T215D/L (3), K219Q (2) at the RT; and M46L (1) at the PRO. No differences were observed in median CD4 counts and viral loads at the beginning comparing individuals infected with resistant and non-resistant HIV viruses (716 vs 715 cells/µl and 12340 vs 15440 copies/ml, respectively). Moreover, no differences in the loss of CD4 cells were recognized comparing both groups of S88 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 82 zidovudine that is not detectable by phenotypic testing. In the absence of drug, HXB2 was 4.7% less fit D67N than HXB2 , which is similar to the fitness difference Transmitted HIV-1 carrying D67N WT of 6.2% seen among WT viruses and a persistent virus or K219Q evolve rapidly to carrying the 215D mutation. Our results show that zidovudine resistance in vitro and viruses with unique patterns of NAMs including D67N show a high replicative fitness in and/or K219Q/E are commonly found among newly diagnosed patients. The low fitness cost conferred by the presence of zidovudine D67N in the absence of drug supports persistence of this mutation in the untreated population and high- JG García-Lerma1, H MacInnes1, D Bennett2, lights the potential for secondary transmission of virus- H Weinstock2 and W Heneine1 es carrying these unique RT genotypes. The faster evo- lution of these mutants toward zidovudine resistance is 1 HIV and Retrovirology Branch, Division of AIDS, STD, and TB consistent with the higher viral fitness observed in the Laboratory Research, National Center for Infectious Diseases, presence of zidovudine and may have clinical implica- Atlanta, Ga.; and 2 Prevention Services Research Branch, tions. Our findings demonstrate that transmitted HIV-1 Division of HIV/AIDS Prevention, National Center for HIV, STD strains with D67N and/or K219Q/E are phenotypical- and TB Prevention, Centers for Disease Control and Prevention, ly different from WT viruses. Atlanta, Ga., USA Infection with drug-resistant HIV-1 has been associat- ed with a longer time to viral suppression and a short- er time to virological failure following initiation of antiretroviral therapy compared to infection with wild- type (WT) HIV-1. Less is known about the significance of secondary resistance mutations found in transmitted HIV-1 with drug-sensitive phenotypes. From surveil- lance of drug resistance in 1082 treatment-naive, recently diagnosed HIV-1-infected persons, we found that 56 persons had viruses with nucleoside analogue mutations (NAMs), of whom nine (16.1%) were infected with HIV-1 containing secondary NAMs only. We assessed the evolution of drug resistance in these viruses. We generated 13 recombinant viruses using cloned reverse transcriptase (RT) sequences from treat- ment-naive persons or HXB2. Of these viruses, four were WT, three had D67N, four had D67N and K219Q/E and two had K219Q/E. All viruses were found to be sensitive to nucleoside analogues and repli- cated efficiently in the absence of drug. However, viruses carrying D67N and/or K219Q/E had an increased ability to acquire the K70R mutation during selection with zidovudine in vitro. K70R was selected in these viruses after a median of 37 days in culture (range 19–74), compared with 63 days (range 41 to >108) in WT viruses (P=0.028). To investigate the fac- tors associated with the rapid selection of K70R in viruses carrying D67N, we evaluated fitness differ- ences among HXB2 and HXB2 in mixing exper- WT D67N iments done in the presence of zidovudine. HXB2 D67N out-competed HXB2 in the presence of low or high WT zidovudine concentrations, indicating that the D67N mutation confers a fitness gain in the presence of zidovudine. The higher fitness of HIV-1 in the pres- D67N ence of zidovudine likely reflects low-level resistance to XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S89 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 83 phenotypic high-level resistance was also noted to non- nucleoside reverse transcriptase inhibitors and nelfi- Male genital tract navir in all samples. Length polymorphism analysis of HIV derived from the blood and semen of the index compartmentalization and subject revealed multiple HIV quasispecies, which transmission of 215L revertant were more similar to the quasispecies detected in the semen of the source partner than the quasispecies DM Smith, KK Koelsch, JK Wong, detected in the blood of the source subject. GK Hightower, CI Ignacio, DD Richman and SJ Little CONCLUSIONS: These investigations show compart- mentalization of 215 revertants in the male genital University of California San Diego, San Diego, Calif., USA tract. This may be explained by the isolation of a founding virus or the selection of such variants in the INTRODUCTION: HIV develops resistance to genital tract. Since antiretrovirals differentially pene- zidovudine (ZDV) with the primary mutation of trate the blood and male genital compartments, it may 215F/Y, which is a two-base mutation in the codon. In facilitate the production and/or selective retention of the absence of ZDV selective pressure HIV often revertants. This has significant public health implica- undergoes a single base mutation to a ‘revertant’ tions, as these revertants represent highly fit viruses (215D, C, N, D, S, E, L). These revertants have a that can become resistant to ZDV more readily than greater fitness advantage in the absence of ZDV than wild-type virus. the 215F/Y virus and only require a one-base change in the codon from the 215F/Y virus. They are long-lived in the blood in the absence of ZDV and are associated with rapid emergence of 215F/Y when ZDV is re-instat- ed. Transmission of ZDV resistance has been docu- mented both with 215F/Y and 215 revertants. To inves- tigate the nature of revertant transmission, we evaluat- ed the first documented case of 215L transmission. METHODS: A transmission partner pair was identi- fied in the UCSD Primary Infection cohort. Population (Viroseq, Applied Biosystems) and dye- primer sequencing of HIV pol, length polymorphism analysis of V12 and V45 HIV env regions (GeneScan, Applied Biosystems), and phenotypic resistance test- ing (Phenosense, Virologic) were performed on HIV RNA isolated from blood and seminal plasma of both subjects. RESULTS: The source subject was chronically infected with HIV and had a long antiretroviral treatment his- tory but had not been treated for 16 months when identified with a CD4 count of 139 cells/ml and HIV viral load of 50702 copies/ml. The index subject was acutely infected and naive to antiretroviral therapy with an initial CD4 of 472 cells/ml and HIV viral load of 36962 copies/ml. Population sequencing identified a 215F/L mixture in the blood of the source partner but only 215L was found in the blood of the receptive partner. Using both population and dye-primer sequencing (10% detection of minor species) only 215L virus was identified in the semen of both sub- jects. Phenotype testing revealed high-level resistance to ZDV in the blood of the source but only moderate- level resistance in the index subject. Genotypic and S90 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 84 site mutations was selected for the study. There were significant differences in the replication capacities of Analysis of virion morphology and the four recombinant virus types constructed from the gag-protease fragment of this patient, and the replica- assembly process in protease tion competence was, in order from most active to inhibitor-resistant HIV-1 least active, GP>P+c>P>GP-c. We observed prominent unprocessed Pr55Gag and 41 Kd intermediate gag-prod- L Myint, M Matsuda, T Chiba, H Yan, uct in replication reduced virus types P+c, P and GP-c. In J Kakizawa, A Okano, M Hamatake, EM analysis, immature viruses were observed in P+c, P M Nishizawa and W Sugiura and P-c. Interestingly, we found intracellular virus bud- ding in the P-type as well as the P+c type virus. AIDS Research Center, National Institute of Infectious Diseases, Consistent with this EM results, unprocessed Gag pro- Japan tein was found prominently at the perinuclear area in P and P+c type transfected cells when analysed by con- BACKGROUND AND OBJECTIVE: Drug-resistant focal microscopy. viruses often demonstrate reduced viral replication compared to wild-type viruses. In protease inhibitor CONCLUSION: Immature virion morphologies and (PI)-resistant viruses, this reduced replication is due to intracellular budding were observed in PI-resistant impaired Pr55Gag processing into mature proteins by viruses with impaired Gag processing. Our results sug- the protease with drug-resistant mutations within its gest aberrant interaction between virus proteins and active site. In this study, to understand the pathogene- host factors in PI-resistant viruses. sis of PI-resistant viruses, we analysed the impact of impaired Gag processing on the virion assembly process and virion morphology. METHODS: A PI-resistant case was selected from patient samples sent to NIID for routine drug resis- tance genotyping. A gag-protease fragment derived from the patient virus was amplified by RT-PCR and inserted into an HXB2-based virus expression vector. The following four types of recombinant viruses were prepared: GP-type (patient gag and protease), P-type (HXB2 gag and patient protease), GP-c type (gag cleav- age site mutations removed from GP-type) and P+C (patient protease and the cleavage site mutations found in the patient gag). Four approaches were employed to analyse characteristics of the recombinant viruses. First, replication capacities of the four recombinant viruses were evaluated using competition cultures. Second, the Pr55Gag processing pattern of each recom- binant virus was analysed using western blot analysis. Third, the morphology of the virus particles was analysed by electron microscopy (EM). COS7 cells or 293T cells were transfected with the recombinant virus DNA, and 48 h after the transfection, cells were har- vested carefully and subjected to EM analysis. Fourth, the localization of Gag proteins inside the host cells was analysed by confocal microscopy. Gag proteins were stained by anti-p17 and/or anti-p24 monoclonal antibodies (mAbs), and cellular organelles were stained by anti-ER, anti-Golgi, anti-late endosome, or anti- Tsg101 mAbs. RESULTS: A case with D30N/M46I/N88D/L90M PI- resistant mutations and A431V /L449F gag cleavage XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S91 SESSION 4 New Resistance Technologies & Interpretations Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 85 RC and susceptibility of the WT clone was not signifi- cantly affected by co-transfection with the mutant The impact of minor populations of clones. In contrast to the above, coculture of cells con- taining the individually transfected mutant and WT wild-type HIV on the replication clones at a 1:1 ratio only slightly affected the RC of the capacity and phenotype of mutant mutant constructs. Similarly, co-infection of a 1:1 ratio variants in a single-cycle HIV of mutant and WT viruses did not change the RC of resistance assay the mutants. CONCLUSION: Due to the unique co-transfection H Mo, L Lu, D Kempf and A Molla step inherent to single-cycle HIV resistance assays, even relatively small amounts of WT virus within a Abbott Laboratories, Abbott Park, Ill., USA viral population can significantly impact the apparent RC and phenotype of mutant strains. The RC and sus- BACKGROUND: In a typical single-cycle HIV pheno- ceptibility of plasma isolates from patients who are off typic assay, the ‘pool’ of DNA generated from patient therapy or not adherent to treatment, in which WT plasma is transfected into target cells in order to cap- virus may expand to significant levels, should be inter- ture and preserve the protease (PR) and reverse tran- preted with caution. scriptase (RT) sequence heterogeneity of the plasma virus. However, co-transfection of different viral vari- ants into same cell might provide the opportunity for genetic recombination or complementation, or both. The impact of cotransfection of mixtures of mutant and wild-type (WT) virus on the observed phenotype and replication capacity (RC) has not been elucidated. METHODS: Mutant HIV clones were constructed to contain a firefly luciferase expression gene in the nef region and a frameshift in envelope. The WT clone was engineered to contain a renilla luciferase gene. Firefly and renilla luciferase activities were simultaneously measured in a single cycle assay after co-transfecting the mutant constructs alone or as mixtures with the WT construct (10:1, 2:1 and 1:1, respectively) with a VSV envelope expression vector. RESULTS: Four mutant constructs with different geno- types derived from dual protease inhibitor-experienced subjects receiving lopinavir/ritonavir (LPV/r) therapy displayed <5% RC when transfected alone. Co-trans- fection of as little as 9% of the WT clone increased the RC of the mutant clones to up to 14%. Co-transfec- tion of a higher proportion of the WT clone further enhanced the RC of the mutants to 31–81%. In con- trast, the RC of four mutant clones that expressed a modest to high level of firefly luciferase activity when transfected alone did not change significantly upon cotransfection of the WT clone (one- to threefold increase in RC when co-transfected with up to 50% of WT). The LPV susceptibility of four mutant clones with sufficient RC for phenotypic evaluation when transfected alone ranged from 44- to 302-fold, com- pared to the WT clone. Incremental cotransfection of 9–50% of the WT clone decreased the LPV IC of the 50 mutant clones by up to 96%, compared to WT. The XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S95 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 86 tained linked PI mutations (L10V, M46I, I84V, L90M, I93L), none of which was detected in the composite Comparison of single-genome sequence. In a second sample example, 33% of genomes contained five linked RT mutations associat- sequencing with standard genotype ed with NNRTI resistance (A98S, K101E, Y181C, analysis for detection of HIV-1 G190A, T215Y), none of which were detected in the drug resistance mutations composite sequence. A third sample contained two NRTI resistance mutations: one (D67N) present in 30% of the genomes was not detected in the compos- M Kearney1, S Palmer1, F Maldarelli1, C Bixby4, ite sequence; and the other (V118I) present in 21% H Bazmi4, D Rock2, J Falloon3, R Davey3, was detected in the composite. All plasma samples test- R Dewar2, J Metcalf2, J Mellors4 and J Coffin1 ed contained at least one drug resistance mutation that was identified by SGS but not in the composite geno- 1 HIV Drug Resistance Program, NCI, NIH; 2 NIAID/CCMD Clinic, type. NIH; 3 Laboratory of Immunoregulation, NIAID, NIH; and 4 Division of Infectious Diseases, University of Pittsburgh, CONCLUSIONS: These results illustrate the inade- Pittsburgh, Pa., USA quacy of composite genotype analysis for detecting drug resistance mutations present in less than 35% of BACKGROUND: More sensitive analyses of HIV-1 the plasma virus population. Mutations present in populations are needed to understand viral diversity in <10% of the virus population were almost never drug-experienced patients. To investigate the extent to detected and those present in 10–35% were inconsis- which drug-resistance mutations are missed by con- tently detected. In addition to its greater sensitivity, ventional methods, we performed single-genome SGS permits detection of linked mutations that confer sequencing and standard composite genotype analysis high-level drug resistance. Such linkage cannot be on the same plasma samples from patients with known detected in composite genotypes. multidrug-resistant HIV-1. METHODS: Plasma samples were obtained from 24 patients failing antiretroviral therapy or known to be infected with multidrug-resistant HIV-1. Composite genotypes were obtained by RT-PCR and sequencing of bulk PCR product. For single-genome RT-PCR sequence (SGS) analysis, cDNA derived from plasma RNA was serially diluted to a single copy, and a region encompassing p6, protease and the first 300 codons of reverse transcriptase (RT) was amplified and sequenced. Sequences of 15–46 single viral genomes were obtained from each sample. Mutations included in the analysis were those defined as conferring resis- tance by the Stanford database. RESULTS: All mutations present in composite geno- types were also detected by SGS. By contrast, not all mutations in single genomes were detected in compos- ite genotypes of the same plasma sample. With one exception, all drug resistance mutations present in less than 10% of single genomes were not detected in the composite genotype, whereas mutations in 10–35% of single genomes were detected in only 25% of cases. For example, in one patient, 10 mutations conferring resistance to protease inhibitors (PIs), nucleoside RT inhibitors (NRTIs) and non-NRTIs (NNRTIs) escaped detection in the composite genotype. Each of these mutations was present in 5–20% of the 20 genomes analysed. 15% of the genomes in this sample con- S96 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 87 Drug-resistant mutants of HIV-1 persist at higher concentrations in peripheral blood mononuclear cells compared to plasma and are detected at a higher rate by OLA compared to consensus sequencing GM Ellis1, M Mahalanabis1, IA Beck1, G Pepper2, A Wright2, S Hamilton2, S Holte3, WE Naugler1, DM Pawluk1, C-C Li1 and LM Frenkel1,2 1 Department of Pediatrics; 2 Laboratory Medicine, University of Washington, Seattle, Wash.; and 3 Fred Hutchinson Cancer Research Center, Seattle, Wash., USA Drug-resistant mutants of human immunodeficiency virus type 1 (HIV-1) recede below the limit of detection of most assays applied to plasma when selective pres- sure is altered due to changes in antiretroviral treat- ment (ART). Viral variants with different mutations are selected by the new ART when replication is not suppressed or wild-type variants with greater replica- tion fitness outgrow mutants following the cessation of ART. Mutants selected by past ART appear to persist in reservoirs even when not detected in the plasma, and when conferring cross resistance can compromise the efficacy of novel ART. An oligonucleotide ligation assay (OLA) and consensus dideoxynucleotide-chain terminator sequencing were compared on virus from plasma and peripheral blood mononuclear cells (PBMCs) for the detection of 91 drug resistance muta- tions that had receded below the limit of detection, or were ‘lost’, from plasma assessed by sequencing. The OLA detected 27.5% (95% CI: 19–39%) of mutant genotypes ‘lost’ from plasma. Consensus sequencing of the PBMC amplicon from the same specimen detected 23.1% (95% CI: 14–34%) of the ‘lost’ mutations, while OLA of PBMC detected 53.8% (95% CI: 44–64%) of the ‘lost’ mutations. These data indicate that concentrations of drug-resistant mutants were greater in PBMCs than in plasma after changes in ART, and that the OLA was more sensitive than consensus sequencing in detecting low levels of select drug-resis- tant mutants. XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S97 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 88 I84I, L90M), where the notation M46L* means that the corresponding mutation was in group (iii) for Simultaneous assessment of codon 46, the most common mutation in that group being L. Nine genotype patterns were significantly association between two or more associated with resistance to RTV but not IDV, one of HIV-1 drug susceptibility which was (L10I, M46M, I54V, A71V, V82A, I84V, phenotypes and genotype L90M). Six genotype patterns were significantly asso- ciated with resistance to IDV but not RTV, one of which was (L10I, M46I, I54I, A71A, V82V, I84V, G DiRienzo L90M). Harvard School of Public Health, Boston, Mass., USA CONCLUSION: Several groups of antiretroviral drugs are considered to be similar with respect to the geno- BACKGROUND: Some specific HIV-1 genotype pat- type patterns associated with resistance, which may terns are considered to be associated with resistance to lead to the belief that available treatment options are more than one antiretroviral drug, greatly reducing the limited for patients with such patterns; however, it may number of drug options thought to be available for be the case that specific genotype patterns are in fact patients with such patterns. We present analyses that associated with sensitivity to one or more of these investigate the existence of specific genotype patterns drugs in a given group. The methods used may accom- that retain sensitivity to at least one drug in the set of modate a wide range of outcome variables, including drugs thought to not be viable options for a given virological and immunological response, and viral patient. replication capacity. METHODS: We used data from the Stanford University HIV Database, which included measure- ments of IC fold changes for several antiretroviral 50 drugs, along with a corresponding HIV-1 genotype sequence, for approximately 1000 samples. We invoked the statistical methods of DiRienzo et al. (2003) for testing association between genotype sequence and one or more outcome variables of inter- est. Several different labs contributed to this dataset and the methods adjust for a possible lab effect. At each codon, these methods split the possible amino acid values into three categories: (i) wild-type; (ii) the most common mutation at that codon; and (iii) all other mutations. For the vast majority of codons in this dataset, the percentage of the number of mutations other than the two most common ones was very small. Permutation-based statistical methods, which make no assumptions about the data-generating mechanism, were then used to identify those genotype sequences whose corresponding IC fold changes (of the one or 50 more drugs of interest) significantly differed from those corresponding to the wild-type sequence at any desired global type-I error level (we used 5%). RESULTS: We display results for only one pair of drugs, indinavir (IDV) and ritonavir (RTV), which are believed to be very similar with respect to the corre- sponding genotype patterns that induce resistance. The following genotype patterns in the protease region were significantly associated with resistance to both IDV and RTV: (L10I, M46M, I54V, A71V, V82A, I84I, L90M) and (L10I, M46L*, I54V, A71V, V82A, S98 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 89 in a reaction containing 1.5×107 total target molecules. The GENE-CODE technology can identify samples Using GENE-CODE technology to that contain very small fractions of a particular drug- resistant HIV sequence against a background of close- detect early emerging populations ly related sequences. To our knowledge, no other tech- of drug-resistant human nology even approaches such a level of sensitive speci- immunodeficiency virus Type 1 ficity. Additional development of GENE-CODE to include a larger number of HIV variants could lead to a simple-to-run and widely-used method for research MJ Moser1, PL Sharma2, V Nurpeisov2, and clinical diagnostics. RF Schinazi2 and JR Prudent1 1 EraGen Biosciences, Inc., Madison, Wis.; and 2 Emory University/VA Medical Center, Decatur, Ga., USA Treatment of human immunodeficiency virus Type 1 (HIV) infection with antiretroviral agents invariably leads to selection and accumulation of drug-resistant mutants that arise due to the intrinsically low fidelity of HIV polymerase. Early detection of these growing variants may be clinically relevant for a number of rea- sons. Yet before the level of clinical importance can be ascertained, technologies that reliably detect small genetic changes in a large sea of closely related viral quasi-species need to be available. Perhaps the most well known mutation observed in HIV-RT is the M184V. Genetically, M184V arises from a single base change of A→G. GENE-CODE technology was used to detect and quantitate mixed populations of wild- type and M184V viral sequences. GENE-CODE is a novel, closed-tube, single-vessel RT-PCR method that is made possible through the use of an expanded genet- ic alphabet. GENE-CODE uses two separate-fluo- rophore-labelled forward primers that contain novel bases and one common reverse primer to analyse mixed populations of closely related genetic materials. During extension, quenchers attached to novel bases are incorporated, and a change in fluorescence deter- mines the cycle threshold (C ). The percent composi- t tion of an unknown sample can be determined by mea- suring the difference between the C of the two chan- t nels (defined as ∆C ) and plotting that ∆C onto a pre- t t viously established standard curve. The standard curve is generated by plotting percent or ratio of one target to the other versus ∆C for a set of known standard t samples. The mixed population test system for M184V analyses the percentage of wild-type to mutant RNA. RNA transcripts used in these tests were derived from cloned M184V mutant and wild-type HIV sequences, as well as viral RNA from human PBM cells transfect- ed with wild-type and M184V pNL4-3 HIV molecular clones. The data shows a large change in ∆C with a t dynamic range from 100 to 0.01% M184V sequence in a standard mixture. We can detect one copy of mutant 184V per 10000 wild-type 184M RNA copies XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S99 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 90 the clones were recombinant A/B. However when the PRO+RT fragment was cloned and sequenced, 87.5% Evidence for preferential of the clones matched the sequence of virus A, while the other 12.5% resulted to be virus B. When the tar- genotyping of minority HIV get sequences for annealing of the primers used in the populations due to primer nested PCR were analysed, we observed that the for- mismatching during PCR-based ward primer had three mismatches respect to the virus A template, but only two respect to virus B. This obser- amplification vation would explain why the nested PCR sequence diverged from the major viral population. K Morales-Lopetegi1, C Gutierrez2, N Izquierdo1, S Marfil1, B Clotet1, L Ruiz1 and CONCLUSION: The extremely high variability of the J Martínez-Picado1 HIV genome may result in a preferential annealing of the PCR-based amplification primers for virus variants 1 IrsiCaixa Foundation, Badalona; and 2 Hospital 12 de Octubre, present as minority populations in the sample. This Madrid, Spain observation might explain unexpected PRO or RT genotypes, as those being wild-type during viral break- BACKGROUND: Genotyping HIV protease (PRO) through to antiviral therapy. and reverse transcriptase (RT) has become a common methodology for monitoring antiretroviral drug resis- tance in HIV-infected patients with virological failure. Due to the high genetic variability of HIV, a perfect annealing of the primers used for genotyping PRO and RT is not guaranteed. This situation might result in either a lack of amplification or in a sequence that will not be representative of the majority viral population in the sample. This irregularity might mislead thera- peutic decisions that could be inferred from such geno- types. We have investigated discordances between the population- and the clonal-based genotypes in PRO coding region. METHODS: HIV-1 RNA was reverse transcribed and PCR-amplified in one step (primers 2146–2165/PRO and 3387–3415/RT), nested-PCR-amplified (primers 2212–2230/PRO and 2571–2591/PRO) and sequenced on an automated sequencer (same primers as in nested PCR). Both, the nested PCR product (only PRO) as well as the product of the first PCR (PRO+RT), were independently cloned into a pGEM vector. Twelve clones of each were sequenced. Then, we compared the population-based sequence, PRO- cloned-derived sequences and PRO-RT-cloned derived sequences. The later sequences allowed us to identify the target sequence for the primers used in the nested PCR. RESULTS: The PRO genotype derived from popula- tion-based sequence showed a major proportion of one virus (termed here virus A) compared to a second virus (virus B). The sequences obtained when the PRO was cloned after nested PCR amplification did not match the results obtained by direct sequencing of the virus mixture. Thus, 67% of the clones resulted virus B and only 8% resulted virus A, while the remaining 25% of S100 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 91 (283 and 1190 copies/ml) were not amplified for pro- tease by three and four laboratories and for RT by two Quality control assessment for and one, respectively. The global rate of false-positive and -negative results was 0.31% (36/11639) and 1.5% HIV-1 genotypic antiretroviral (70/4524), respectively. The frequency of false-posi- testing (2002) tives was comparable in the RT (0.26%) and protease (0.38%) genes (P=0.3037), but the rate of false-nega- F Brun-Vézinet1, B Masquelier2, J Izopet3, tive was higher in the protease (1.95%) than in the RT V Calvez4, ML Chaix5, A Ruffault6, C Tamalet7, (1.09%) (P=0.0215). As regards the specificity and S Yerly8, D Descamps1, D Costagliola9 and the sensitivity, there were no differences between primary ANRS Resistance Group and secondary mutations in the protease gene. The specimen in duplicate was scored identically by 89% 1 CHU Bichat-Claude Bernard, Paris, France; 2 CHU Bordeaux, (34/38) of the laboratories. France; 3 CHU Toulouse, France; 4 CHU Pitié-Salpétrière, Paris, France; 5 CHU Necker-Enfants Malades, Paris, France; 6 Hopital CONCLUSION: The use of supernatants from co-cul- Pontchaillou, Rennes, France; 7 CHU La Timone, Marseille, tivated viruses does not improve the definition of the France; 9 INSERM E 0214, Paris, France; and 8 Hôpital de consensus amino acid sequences as compared to the Genève, Switzerland use of diluted plasma specimens. Quality control assessment for HIV drug resistance sequencing must BACKGROUND: To evaluate the performances of include an HIV-negative control, specimens with low academic and non-academic virology laboratories for viral load, specimen in duplicate and samples with HIV genotyping. high number of resistance mutations. METHODS: A coded panel of 10 samples was distrib- uted to 38 laboratories and consisted of four diluted plasma samples from HIV-1-infected patients, five supernatants from co-cultivated viruses and one plas- ma HIV-negative control. Participants were asked to report the amino acid at codons associated to nucleo- side reverse transcriptase inhibitor (NRTI), non-NRTI (NNRTI) and protease inhibitor (PI) resistance listed by the IAS-USA expert panel (2002). The reference amino acid sequences were defined as the most fre- quently reported by the participants (>80%). RESULTS: Thirty-eight laboratories reported datasets. By phylogenetic analysis of the RT gene the nine posi- tive specimens were subtype B. Viral load ranged between 283 and >750000 copies/ml. Seven samples had a high number of mutations in the protease (medi- an 10, range 6–10) and in the RT (median 8, range 6–11). Two specimens had two and three mutations in the protease and no mutation in the RT genes. All lab- oratories used automated sequencing technologies: ANRS consensus technique (20), Trugene (11), ViroSeq (2), in-house assays (5). The reference amino acid sequence remained undetermined at 11 codons of the protease (in four viral supernatants and two plas- ma samples) and 13 codons of the RT genes (in two viral supernatants and two plasma samples). The ambiguities of the reference consensus sequences were only observed in samples with high number of muta- tions. The negative control was amplified in three lab- oratories, for both genes in two and for the protease gene only in one. Two specimens with low viral load XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S101 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 92 RESULTS: ASRT assays developed by three different Updated, blinded, multicentre laboratories quantified 103N down to a frequency of 0.1, 0.4 or 25%, respectively. The Ty1 system quanti- comparison of the sensitivity of fied 103N to 0.4%, with direct confirmation of the different technologies to detect and mutant allele by DNA sequencing of individual yeast quantify a minor drug-resistant colonies. SGS and Pyro Seq assays detected 103N at 2% frequency but were not quantitative below 10%. HIV-1 variant The OLA and bulk sequencing assays quantified 103N to 5–10%. The LiPA detected 103N at 2% frequency E Halvas1, G Aldrovandi2, JP Balfe3, I Beck4, but was not quantitative. V Boltz5, L Frenkel4, J Hazelwood2, V Johnson2, M Kearney5, A Kovacs6, D Kuritzkes7, CONCLUSIONS: Blind testing of a panel of plasmid- K Metzner8, D Nissley5, M Nowicki6, derived virus mixtures showed that two of three ASRT R Ziermann9, Y Zhao10, C Jennings11 , assays and the Ty1 assay could detect and quantify the J Bremer11 , D Brambilla12 and J Mellors1 103N variant at frequencies <1%. One of the ASRT assays could quantify as low as 0.1% and possibly 1 University of Pittsburgh, Pittsburgh, Pa.; 2 Birmingham VA 0.01%. A potential advantage of Ty1 assay over ASRT Medical Hosp. and Univ. of Alabama at Birmingham, is that the mutant allele is cloned and its presence can Birmingham, Ala.; 3 Columbia University, New York, NY; 4 be confirmed by DNA sequencing. The other methods University of Washington, Seattle, Wash.; 5 National Cancer evaluated had detection and quantification limits rang- Institute, Frederick, Md.; 6 University of Southern California, Los ing from 2 to 10% mutant. FDA-approved bulk Angeles, Calif.; 7 Brigham and Women’s Hosp., Boston, Mass.; 8 sequencing technologies (ViroSeq and TruGene) had University of Erlangen-Nuremburg, Erlangen, Germany; 9 Bayer detection and quantification limits of 5–10% for Health Care-Diagnostics, Emeryville, Calif.; 10 Children’s 103N. These findings indicate that several technologies Memorial Hospital, Chicago, Ill.; 11 Rush Med. Coll., Chicago, Ill.; are available to address the role of minor drug-resis- and 12 New England Res. Inst., Watertown, Mass., USA tant variant in failure of antiretroviral therapy. BACKGROUND: The role of minor (low frequency) drug-resistant variants in antiretroviral treatment fail- ure is uncertain. To address this question, methods must be developed and validated for their ability to identify minor drug-resistant variants. We therefore evaluated the ability of several methods to detect and quantify a non-nucleoside reverse transcriptase inhibitor (NNRTI)-resistant variant of HIV-1. METHODS: Infectious plasmid clones of HIV-1 , LAI either wild-type or encoding the NNRTI resistance mutation K103N (AAA to AAC), were used to gener- ate a panel of virus mixtures in seronegative human plasma with the % mutant (103N) being 0, 0.01, 0.1, 0.4, 1, 2, 5, 10, 25, 50 or 100%. The mixtures were blinded and distributed to 15 labs that responded to a solicitation of interest. Methods used to test the panel were: 1) allele-specific real-time RT-PCR (ASRT, three labs); 2) a Ty1/HIV-1RT yeast hybrid assay (Ty1, one lab); 3) RT-PCR with allele-specific hybridization [LiPA assay (Bayer Health Care-Diagnostics), one lab]; 4) oligonucleotide ligase-based assays (OLA, two labs); 5) single genome RT-PCR and sequencing (SGS, one lab); 6) RT-PCR with pyrosequencing (Pyro Seq, one lab); and 7) RT-PCR with direct sequencing of bulk PCR product (Applied Biosystems ViroSeq v2.0, three labs; Visible Genetics HIV-1 TruGene, three labs). S102 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 93 robust RT activity and illustrate the usefulness of the system for identifying novel drug-resistant HIV-1. Detection of rare drug-resistant CONCLUSION: The TyHRT system is a sensitive phe- HIV-1 reverse transcriptase variants notypic assay that detects HIV-1 drug-resistant vari- using a sensitive phenotypic assay ants present at less than 1%. The system can be used to identify novel drug-resistant variants. The simplici- DV Nissley1,3, E Halvas2, D Garfinkel3, ty and sensitivity of this assay makes it possible to J Mellors2 and J Strathern3 monitor the emergence of drug resistance in basic and clinical research environments. 1 SAIC-Frederick, Frederick, Md.; 2 University of Pittsburgh, Pittsburgh, Pa.; and 3 NCI-Frederick, Frederick, Md., USA Funded in part by DHHS #NO1-CO-12400 BACKGROUND: Detection of rare drug-resistant variants may be important for choosing the most effec- tive antiretroviral therapy. Optimal technologies for the detection of drug resistance should be highly sensi- tive, non-specific (able to detect known and novel vari- ants) and applicable in any research environment. A yeast-based chimeric Ty1/HIV-1 reverse transcriptase (TyHRT) retrotransposon system has been used to develop a sensitive phenotypic assay for the detection of rare drug-resistant RT variants. METHODS: In TyHRT elements, Ty1 RT is replaced by HIV-1 RT. RT activity is monitored via a genetic assay in Saccharomyces cerevisiae and is sensitive to RT inhibitors. Efficient homologous recombination in yeast facilitates the assembly of RT domain libraries in vivo without conventional cloning. RT-PCR is carried with HIV-1 RNA to generate RT domain DNA that is co-transformed along with HART elements deleted for RT codons 35–250. Homologous recombination incorporates the amplified RT domains into TyHRT elements resulting in a library of clonal isolates from the RTs present in HIV-1 RNA. RT isolates are assayed for activity in the absence of RT inhibitors and for drug resistance in the presence of RT inhibitors. The activity and resistance profiles are determined for sev- eral hundred isolates from each sample. Drug-resistant variants are confirmed by DNA sequencing. RESULTS: 1) Varying mixtures of wild-type and efavirenz-resistant virus (K103N) were used to deter- mine the sensitivity of the TyHRT system. HIV-1 encoding K103N was accurately detected when pre- sent at 2 and 0.4% of the total virus population. 2) The TyHRT system is being used to detect efavirenz- resistant variants in plasma samples from non-nucleo- side RT inhibitor (NNRTI)-naive and NNRTI-experi- enced patients with negative standard genotypes for NNRTI resistance mutations. Many known efavirenz- resistant variants have been detected; some are present at lower than 1%. 3) Previously unidentified efavirenz- resistant variants are also detected. These variants have XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S103 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 94 were identified as influencing RC variation; RT135, RT177 and RT178. These three sites withstood 10- Examination of wide variation in fold cross-validation. Random forests, a recently devised technique developed to overcome the predic- replication capacity of wild-type tive shortcomings of TSM while retaining advantages, HIV-1: analysis of genotype– did not give improved performance. phenotype association via CONCLUSIONS: TSM are effective techniques for tree-structured methods genotype–phenotype association problems where genotype is represented by amino acid sequence. Due JD Barbour1,2, T Wrin3, SG Deeks2, FM Hecht2, to the potential for overfitting, use of cross-validation NS Hellmann3, CJ Petropoulos3, RM Grant1,2 or test sample data for evaluation of predictive perfor- and MR Segal2 mance is crucial. Recently proposed methods for enhancing performance (bagging, random forests) 1 Gladstone Institute of Virology and Immunology, San showed only modest improvement in performance over Francisco, Calif.; 2 University of California, San Francisco, San standard TSM. We identified three RT sites, not previ- Francisco, Calif.; and 3 ViroLogic, Inc., South San Francisco, Calif., ously associated with drug resistance, associated with USA variation in RC of wild-type virus. RT135, which sits behind the RT active site, may restrict rotation of the RT BACKGROUND: Wide variation in HIV-1 replication ‘thumb’ as the enzyme binds to, and extends its ligand. capacity (RC) has been observed for wild-type isolates, RT177 and RT178 sit on a loop containing two key RT the genetic basis for which has not been dissected. We catalytic residues, D185 and D186, which coordinate sought to determine genetic correlates of RC variation two Mg2+ cations required for RT function. Subtle in a wild-type HIV-1 population via tree-structured changes in the positioning of these residues may alter RT methods (TSM). The problem of relating viral activity, impacting on viral replication capacity. sequence to phenotype, such as replication capacity (RC), differs from standard regression problems by the unordered, categorical nature of genetic sequence data. A variety of data analytic methods have been proposed to examine genotype–phenotype associations, includ- ing TSM. Virtues of TSM include: (i) flexible, exhaus- tive and automated handling of groups of amino acids bypassing the need for computing, examining and grouping individual regression coefficients correspond- ing to indicator representations for amino acid levels; (ii) readily accommodating between-codon interac- tions, which are frequently important; and (iii) easy interpretation and prediction via associated tree schema. METHODS: We critique two approaches that have been used for sequence-phenotype association, neural networks and prediction-based classification, prior to a detailed evaluation of the strengths and weaknesses of TSM. We determined the consensus nucleotide sequence of protease (PR) from codons 4 to 99, and reverse transcriptase (RT) from codons 38 to 223. RC was assessed via a single-cycle recombinant assay based on a segment of the patient HIV-1 pro/pol gene (PR aa 1–99; RT aa 1–305). RESULTS: We compiled 336 paired viral genotypes and RC assays. Of these records, 224 (66.7%) had no genotypic evidence of drug resistance. Among these 224 isolates, three non-drug resistance-associated sites S104 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 95 typically observed in between the wild-type and the M184T. For wild-type HIV-1 about 30% of the initi- Reverse transcription fitness assay ated R/U5 cDNA products were elongated to end, whereas of the M184T variant only 10% was elongat- by quantitative determination of ed to the end product. Currently, interesting patient intermediate HIV-1 reverse derived RT variants are being investigated for their transcription cDNA products in reverse transcription efficiency. PBMCs CONCLUSION: This rapid real-time Taqman PCR approach to quantify the formation of HIV-1 cDNA M Nijhuis, L de Graaf, D Eggink, intermediates during the natural reverse transcription J den Dunnen, C Boucher and R Schuurman process in primary cells can be used to determine dif- ferences in reverse transcription efficiency for drug- Department of Virology, University Medical Center, Utrecht, the resistant RT variants. Clear defects in the reverse tran- Netherlands scription process were observed for the lamivudine- resistant M184V, I and T variants, which is in concor- BACKGROUND: HIV-1 resistance-associated muta- dance with results obtained from enzymatical and viro- tions can affect the replication capacity of the virus. logical data. While the impact of protease resistance-associated mutations on viral replication capacity is extensively investigated, less is known about reverse transcriptase (RT) resistance-associated mutations. We have devel- oped an assay in which the mechanism of RT-based fit- ness changes can be investigated by quantitative analy- sis of different parts of the HIV-1 cDNA product. This approach enables us to measure the formation of the HIV-1 cDNA intermediates during the natural reverse transcription process inside peripheral blood mononu- clear cells (PBMCs). Use of these primary cells is in particular important, since they have low deoxynu- cleotide (dNTP) pools as compared to T cell lines and the reverse transcription process is known to be sensi- tive to these dNTP levels. METHODS: PBMCs were infected with either wild- type or mutant strains containing the M184V, M184I or M184T mutation in the RT gene. At several time- points after start of infection, DNA was isolated and real-time Taqman PCR was performed to quantify the HIV-1 cDNA products: 1) initiation and elongation into the R/U5 (start); 2) first template switch and minus strand elongation into the gag region (gag); 3) second template switch and elongation over U5/PBS (end). The levels of end cDNA products were expressed relative to the amount of start cDNA formed. RESULTS: For wild-type HIV-1, elongation into the gag region was first observed after 2 h of infection. The formation of end cDNA products was observed 0.5 h later. The M184T variant demonstrated a clear delay in the kinetics of reverse transcription since gag tran- scripts were observed after 3 h of infection, which is a delay of 1 h compared to the wild-type RT. The cDNA transcripts for the M184V and M184I variants are XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S105 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 96 Proceedings of the National Academy of Sciences, USA 2002; 99:8271–8276) [10.2% (sensitivity=89.8%, Quantitative prediction of HIV drug specificity=89.7%)]. The robustness of the algorithm as a function of the size of the input dataset was susceptibility from viral genotype assessed using smaller subsets of data. Nine of 22 IAS through linear regression modelling resistance-associated mutations for RTV could be identified with subsets ≥5% (1600 isolates) of the orig- H Van Marck1, T Van den Bulcke1, inal data. However, the accuracy of the predicted con- M Van Houtte2, P Lecocq2 and L Bacheler3 tribution of the mutations improved with increasing dataset sizes up to 50% of the original database (medi- 1 Tibotec-Virco, Mechelen, Belgium; 2 Virco, Mechelen, Belgium; an standard error of the predicted contributions and 3 VircoLab, Durham, NC, USA decreased 50%). Some secondary mutations (for example, 10R, 32I, 82S) were identified as having a BACKGROUND: Several approaches for predicting significant contribution to resistance only when the the drug susceptibility profile of an HIV-1 clinical iso- subset size reached a similar 50% level. late from in vitro tests exist. However, there is a con- tinuous need to perfect systems that generate objective CONCLUSIONS: Linear regression modelling is a data-driven analyses to predict drug resistance by mak- promising new technique for the analysis of drug resis- ing integrated use of genotypic and phenotypic infor- tance in HIV-1. It is an attractive tool for identifying mation. primary and secondary resistance-associated muta- tions for new and existing drugs and for calculating the METHODS: Linear regression modelling was evaluat- contribution of mutations and combinations of muta- ed as a tool for identifying HIV-1 resistance-associated tions to resistance. The power of the method is most mutations and for predicting drug susceptibility from fully exploited when applied to large datasets of viral genotypes utilizing, encrypted and anonymized, matched genotype/phenotype results. Virco’s large database of matched genotypes/pheno- types from 28974 clinical isolates. Log fold-resistance was modelled as the sum of contributions from single mutations and/or combinations of mutations. The strength of the contributions was determined by an extension of standard linear regression, allowing the use of genotypes containing mixtures of mutations and of phenotypes with censored values. In order to identi- fy relevant mutations/combinations, a first regression analysis using single mutations resulted in a list of sig- nificant mutations. Single and paired significant muta- tions were then used in a second regression. RESULTS: In an initial test, genotypes and corre- sponding phenotypes determined for ritonavir (RTV) for 28540 HIV-1 clinical isolates were used. The linear regression analysis identified 20/22 RTV resistance- associated mutations described in the IAS mutation list (all except 10F and 77I). Additional mutations whose effect on RTV susceptibility had been previously described (for example, 73S/T/C, 84A/C and 88D) were also identified. Overall, 53 single mutations and 96 pairs of mutations were identified as having signif- icant effect on susceptibility to RTV. The predicted phenotype was compared to the measured phenotype in a leave-one-out cross-validation, demonstrating a root mean square error of 0.31 (logFR). The error rate of the linear modelling method [5.62% (sensitivi- ty=93.0%, specificity=95.4%)], compared favourably to a decision tree-based model (Beerenwinkel, S106 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 97 early events, the latter of which shows a strong link to mutation Y181C/I. In the protease, we estimated two Tree models for the evolution of preferred pathways for ritonavir resistance, both fol- lowing the initial event V82A/F/T/S: (M46I/L, I84V, drug resistance V32I) and (I54V/L, A71V/T, L10F/I/R/V, L90M, L33F). We observed less pronounced pathways in the N Beerenwinkel1, R Kaiser2, J Rahnenführer1, development of resistance to indinavir, which is reflect- M Däumer2, D Hoffmann3, J Selbig4 and ed in a star-like topology of the tree. Among the 13 T Lengauer1 mutations associated with indinavir resistance, we found significant relations only in (A71V/T, L90M, 1 Max Planck Institute for Informatics, Saarbrücken; 2 Institute G73S/A) and (V82A/F/T, I54V). Mutation I84V of Virology, University of Cologne, Cologne; 3 Center of appears to evolve neither frequently nor early under Advanced European Studies and Research, Bonn; and 4 Max indinavir monotherapy. Planck Institute of Molecular Plant Physiology, Golm, Germany CONCLUSIONS: The proposed tree models capture BACKGROUND: Understanding the evolution of HIV important features of viral evolution. They provide a strains towards drug-resistant phenotypes is an impor- probabilistic framework for the appearance of new tant requirement for designing therapeutic strategies mutations under drug pressure. The model class of against the virus. It has been shown that many muta- trees is computationally feasible and readable by tions do not appear independently of each other, and humans, yet considerably richer than that of linear for some drugs preferred mutational pathways have paths. Moreover, the approach is not limited to been observed. Generalizing from the linear path monotherapies, but also allows for analysing model, here we apply a certain class of tree models to sequences under selective pressure of a combination describe the accumulation of resistance-associated therapy. mutations in the viral genome. METHODS: We assume that a set of sequences has evolved under similar selective pressure, for example, when samples are derived from patients under the same therapy. For a given set of resistance-associated mutations (either taken from the literature or derived from the sequence set) we estimate all pairwise joint probabilities. The optimal tree that covers all muta- tions and best represents the estimated probabilities is found by translating the problem into the language of graph theory, where an efficient algorithm exists for solving the resulting discrete optimization problem. We applied the method to sequences from the Stanford HIV RT and Protease Sequence Database that have been derived from patients under monotherapy with zidovudine (364 sequences), abacavir (50 sequences), efavirenz (382 sequences), ritonavir (112 sequences) and indinavir (691 sequences). RESULTS: We recovered characteristics of viral evolu- tion under monotherapy such as the ordered accumu- lation of reverse transcriptase mutations (K70R, K219Q/E, D67N) and (T215F/Y, M41L, L210W) for zidovudine. Furthermore, we estimated L74V, K70R and M184V to be early events in the development of abacavir resistance, followed by K65R and Y115F and later other nucleoside-associated mutations. Evolution of resistance to efavirenz was found to usually begin with the appearance of K103N, but Y188L and G190S/A – although less frequently observed – are also XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S107 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 98 analyses correlated better than with ADS alone. Interestingly, it has been found that 1.3–1.9 active Predictive value of drug resistance drugs would be necessary according to the IS to avoid viral load increase. These values were slightly lower for interpretation systems and TDM-ADS analyses (1.2–1.7). To induce viral load therapeutic drug monitoring for decrease additional 0.5–0.8 active drugs/delta log virological therapy response to (TDM-ADS: 0.5–0.7) would be indicated. This value increased for IS with higher R, but decreased in all salvage therapy TDM-ADS analyses. H Walter1, M Helm2, R Ehret3, B Schmidt4, CONCLUSION: 1) All IS were predictive for therapy K Korn1, H Knechten3 and P Braun3 response. However, differences could be found in par- ticular for ANRS_AC11, that showed to be most 1 Institute for Clinical and Molecular Virology, Erlangen, predicitve. 2) Including of TDM was more predictive Germany; 2 Gemeinschaftspraxis Abelein/Helm, Nuremberg, than resistance interpreted by IS alone, indicating that Germany; 3 PZB Aachen, Aachen, Germany; and 4 Department there is additional benefit by performing TDM, of Medicine, University of California, San Francisco, Calif., USA although the differences were not high. 3) The finding that a substantial part of the new therapy would be BACKGROUND: Antiretroviral therapy response has necessary to avoid an viral load increase could repre- been shown to be dependant of HIV drug resistance sent the remaining activity of the pretreatment. and sufficient drug plasma levels. Since a various num- Therefore, the influence of the actual pretreatment ber of interpretation systems (IS) are available for needs further to be evaluated. years, their predictive power on therapy response could be shown already. However, despite these find- ings comparative analyses including additional essential factors like therapeutic drug monitoring (TDM) and quantifying the effect of each IS are hard to perform. In this study we re-analysed a dataset of 131 clinical iso- lates from pretreated patients combining drug resistance interpretation of nine IS and TDM to evaluate the pre- dictive power of the IS in a clinical setting. METHODS: For 131 patients genotypic drug resis- tance testing was performed before the antiretroviral treatment was changed. All analyses has been inter- preted retrospectively by the following nine IS: Retrogram_v1.4 (RG), Rega_v5.5, ANRS_AC11, CHL_v3.2, Grupo de Aconselhamento Virologico (GAV), Detroit Medical Center 2000 (DMC), VGI_5.0, Beta Test of Stanford database (SDB-β) and geno2pheno (g2p). According to the IS an active drug score (ADS) was given for each drug [from inactive (0) to fully active (1)]. Linear regression was performed to analyse the correlation betwenn ADS and viral load. For a subset of 66 samples TDM could be performed. The ADS for each drug were corrected by the results from the TDM analyses and linear regression analyses were performed a second time resulting in a TDM- ADS representing the remaining activity of the drug. RESULTS: Correlation coefficients (R) varied from 0.44–0.61 for ADS and from 0.47–0.59 for TDM- ADS. The best correlation was found for ANRS_AC11, respectively, although it was lower for the TDM-ADS analysis. For all other IS the TDM-ADS S108 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 99 ROC approach performed better than clustering and recursive partitioning in 10/10 of the cross-validated Predicting phenotype from samples. On the other hand, the combination of clus- tering and recursive partitioning had a higher specifici- genotype: a comparison of ty than the ROC approach in all cases. In general, as statistical methods sensitivity of a prediction rule increases, specificity decreases. For patient management, maximizing sensi- AS Foulkes1, L Zhou1 and V DeGruttola2 tivity, that is the probability of correctly classifying someone as resistant, may be particularly important, 1 Division of Biostatistics, University of Pennsylvania School of since optimizing the effectiveness of a new drug Medicine, Philadelphia, Pa.; and 2 Department of Biostatistics, requires that it be combined with other active drugs. Harvard School of Public Health, Boston, Mass., USA BACKGROUND: Several analytic approaches have been proposed recently to predict phenotypic out- comes from viral genotype. These include cluster analysis and recursive partitioning, which are designed to identify structure in complex, high-dimensional data settings. We present an alternative receiver operator curve (ROC)-based approach and apply all three meth- ods to 1185 protease sequences from the Stanford Sequence Database and corresponding IDV fold resis- tance. A comparison of the findings is presented. METHODS: We consider three statistical methods. In each case, IDV fold resistance is dichotomized as sen- sitive or resistant, and a prediction rule is defined. The corresponding sensitivity (the probability of predicted resistant when a sequence is observed to be resistant) and specificity (the probability of predicted sensitive when a sequence is observed to be sensitive) are report- ed. The approaches considered are: 1) K-means clus- tering based on an unweighted Euclidean distance; 2) a classification tree using clusters and amino acid sites as predictor variables; and 3) an ROC approach involv- ing a comparison of the distributions of distances of each sequence in the sensitive group and each sequence in the resistant group to the centroid of the sensitive group. In all cases, 10-fold cross-validation (CV) is employed to arrive at accurate estimates of the error associated with the prediction rule. RESULTS: K-means clustering resulted in a sensitivity of 0.798 and a specificity of 0.898. Amino acid sites that most distinguish the resulting clusters are 10, 20, 46, 54, 71, 73, 82, 84 and 90. The classification tree approach resulted in a median CV sensitivity of 0.896 and specificity to 0.823. The first split of the classifi- cation tree is on cluster membership. Subsequent splits are on indicators for a mutation at sites 46, 82, 88 and 90, respectively. The ROC approach yielded a median CV sensitivity of 0.947 and a specificity of 0.686. CONCLUSIONS: All three approaches resulted in rea- sonably high sensitivity while in this data setting the XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S109 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 100 with didanosine and stavudine, but also with the pro- tease inhibitors amprenavir, indinavir and ritonavir Does HIV-1 subtype influence the (K 0.19–0.51). Subtype J scored poorly (only in the I=R analysis) for stavudine and the four protease results of drug resistance inhibitors, indinavir, nelfinavir, ritonavir and interpretation algorithms? saquinavir (K 0.08–0.55). Subtype D performed slight- ly less well for lamivudine in the I=R analysis JC Schmit1, D Gonzalez2 and R Boulmé2 (K=0.60), but not for protease inhibitors as previously reported. Finally, poor K values (0.28–0.45) were 1 Retrovirology Laboratory, CRP-Santé; and 2 Advanced found for F1 subtype (I=S only) for abacavir, lopinavir, Biological Laboratories, Luxembourg nelfinavir and saquinavir. BACKGROUND: A preliminary study by Snoeck et al. CONCLUSION: Taken together our results show that (1st European HIV Drug Resistance Workshop, a larger variability in genotypic drug resistance inter- Luxembourg, 2003. Abstract 40) suggested on a small pretations is found in subtypes D to K. Low K values data set a possible influence of HIV subtype on con- for drug/subtype combinations may point to specific cordance of interpretation algorithms. Largest dis- problems in algorithms, which should be further inves- agreements were found for subtype J and D (PI only) tigated at a mutational level. viruses. METHODS: In order to assess the influence of viral subtype in a large data set using appropriate statistics, we interpreted 2176 protease (PRO) and 2265 reverse transcriptase (RT) sequences, respectively [N (PRO/RT) for subtype A=145/2002, B=1200/1200, C=318/312, D=131/188, F1=107/79, G=191/218, H=30/20, J=28/21, K=26/25], with five algorithms [ANRS(09/2002), DMC(01/2003), Brazil(10/2001), CHL4.4, Rega5.5], and computed Light’s kappa statistics for 16 drugs in the different subtype groups. For each drug and subtype we calculated the kappa (K) value by algorithm pairs and then the median for all pairs. For statistical purposes, the intermediate answer class (that is, possible resis- tance or I) was considered either as resistant (I=R) or in a second analysis as not resistant (I=S). Low K val- ues for an I=R scenario point to disagreements in dif- ferentiating between sensitive and non-sensitive virus, low K values for an I=S scenario indicate non-con- cordance to differentiate between resistant and non- resistant virus. RESULTS: Overall, there was no univocal trend show- ing that some subtypes constantly performed less well using these five interpretation algorithms. However, looking at results from both I=S and I=R analysis, sub- types D to K (especially H and K) yielded more fre- quently the poorest K scores for a given drug. At the same time, the highest K scores were also found, in other drug/subtype combinations, in this same cluster of subtypes D to K. Subtypes A to C gave generally less extreme K values. Subtype H showed large disagree- ments for all three non-nucleoside reverse transcriptase inhibitors (K 0.59–0.69), didanosine (K=0.29), aba- cavir (K=0.39), tenofovir (K=0.50), indinavir and amprenavir (K 0.29–0.49). Subtype K had problems S110 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 101 ble, intermediate and resistant. In case the algorithm scored more categories, these were pooled (the lowest Comparison of five interpretation and highest categories were taken together). Drugs considered in the analysis included amprenavir, algorithms for the prediction of lopinavir, nelfinavir, ritonavir, saquinavir and indi- protease inhibitor susceptibility in navir. Statistical analysis included Students’ t-test and HIV-1 non-B subtypes one-way ANOVA using Tukey confidence intervals. RESULTS: 1993 protease sequences were available, J Snoeck1, R Kantor2, RW Shafer2, 1544 (77%) from untreated persons. A wide variety of I Derdelinckx1, AP Carvalho3, B Wynhoven4, subtypes was included: A (281), C (601), CRF01_AE MA Soares5, P Cane6, J Clarke7, C Pillay8, (148), CRF02_AG (353), D (212), F (107), G (242), H S Sirivichayakul9, K Ariyoshi10, A Holguin11 , (7), J (30) and K (12). We found a better concordance H Rudich12, R Rodrigues13, MB Bouzas14, for drug-naive patients than for therapy-experienced K Van Laethem1, F Brun-Vézinet15, C Reid16, patients (P<0.0001). For untreated patients, we found P Cahn14, LF Brigido13, Z Grossman12, the least concordance for predicting ritonavir suscepti- V Soriano11 , W Sugiura10, P Phanuphak9, bility and the best concordance for lopinavir. For treat- L Morris8, J Weber7, D Pillay6, A Tanuri5, ed patients the least concordance was also found for PR Harrigan4, R Camacho3, JM Schapiro2, ritonavir and the best for nelfinavir. Furthermore, we D Katzenstein2 and AM Vandamme1 found that the discordances between the interpretation systems is subtype dependent for untreated patients 1 Rega Institute for Medical Research, Leuven, Belgium; 2 (P<0.0001). Based on Tukey confidence intervals, we Stanford University, Stanford, Calif., USA; 3 Hospital Egas Moniz, could determine that the concordance between the dif- Lisbon, Portugal; 4 BC Centre for Excellence in HIV/AIDS, ferent systems is the highest for subtypes G and Vancouver, Canada; 5 Universidade Federal do Rio de Janeiro, CRF_02. For therapy-experienced patients, (excluding Rio de Janeiro - RJ, Brazil; 6 PHLS, University of Birmingham, subtypes H, J and K due to a limited number of Birmingham, UK; 7 Wright Fleming Institute, London, UK; 8 sequences), the mean number of drugs with discordant National Institute for Virology, Johannesburg, South Africa; 9 scores was highest for subtypes A and F (1,9) and low- Chulalongkorn University, Bangkok, Thailand; 10 National est for the two CRFs (1,5), although this was not sta- Institute of Infectious Diseases, Tokyo, Japan; 11 Hospital Carlos tistically significant. III, Madrid, Spain; 12 Ministry of Health, Tel-Aviv, Israel; 13 Instituto Adolfo Lutz, Sao Paulo, Brazil; 14 Fundación Huesped, CONCLUSION: Interpretation systems are increasing- Buenos Aires, Argentina; 15 Laboratory of Virology, Bichat, ly used to predict phenotype and/or therapy response Claude Bernard Hospital, Paris, France; and 16 Bayer HealthCare from genotypic information. Caution should be taken Diagnostics Division, Toronto, Canada when applying these algorithms, especially when deal- ing with non-B subtypes, where the discordance BACKGROUND: Several interpretation systems are between different systems is significant. We found dif- available to predict phenotype and/or therapy response ferences in predicting drug susceptibility especially for from genotypic information of HIV strains. Most of untreated patients for subtypes G and CRF_02 and for these interpretation systems have been evaluated main- subtypes A and F for treated patients. Whether these ly on subtype B viruses, and their performance on non- discordances will translate into differences in their per- subtype B sequences has not yet been evaluated. Here, formance to predict therapy response still has to be we investigated the level of concordance between five evaluated. different interpretation systems when used for non- subtype B viruses. METHODS: We analysed protease sequences gathered by the international non-subtype B workgroup, a worldwide effort to establish a database of non-B sequences. Subtyping was done by bootscanning. For all sequences, drug susceptibility was evaluated using the following interpretation algorithms: Rega 5.5, Stanford Database Algorithm, ANRS (AC11), Viradapt rules and Visible Genetics/Bayer Diagnostics 6.0. Viruses were scored in three categories: suscepti- XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S111 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 102 to evaluate alternative regimens in patients who had experienced treatment failure. The model was used to A neural network model using predict response to 30 common highly active anti- retroviral therapy regimens for each of 118 TCEs clinical cohort data accurately where VL increased after a change to a regimen of ≥3 predicts virological response and antiretroviral drugs (3540 evaluations). In 116 cases identifies regimens with increased (98%), one or more regimens were identified that the model predicted would cause a reduction in VL. This probability of success in treatment could translate into an additional 101 cases where failures patients would have benefited from treatment selected using this model (based on the 87% accuracy of tra- D Wang, BA Larder, A Revell, R Harrigan, and jectory prediction established above). J Montaner on behalf of the HIV Resistance Response Database Initiative (RDI) CONCLUSIONS: NN modelling using the narrowest baseline VL window gave the most accurate prediction of VL response, underlining the importance of the BACKGROUND: We have previously demonstrated quality of the dataset. This model also showed superi- the utility of neural networks (NN) in predicting viral or predictive accuracy compared to data obtained from load (VL) response using a clinical trial dataset. Here a NN model trained with clinical trial data, establish- we compare the reliability of using ‘real life’ clinical ing the feasibility of using treatment cohort data to cohort data to model treatment response. model VL response. Furthermore, the model identified potentially beneficial treatment regimens in almost all METHODS: Three-hundred-and-fifty-one patients patients who experienced treatment failure, demon- from the BC HIV treatment cohort were analysed. Each strating the utility of using this approach to perform had multiple genotypes and irregular VL measurements. evaluations of multiple treatment permutations. Treatment change episodes (TCEs) were selected from patients having a genotype up to 12 weeks before treat- ment change and baseline VL up to 16, 12 or 8 weeks before treatment change. Follow-up VL was within 4–40 weeks after treatment change. 652, 602 and 539 TCEs, respectively, obtained using the three VL win- dows, were used for NN training (10% of each was partitioned for independent testing). Results were com- pared to those previously obtained using clinical trial data comprising 700 TCEs, with baseline VL and genotype at week 0 (time of treatment change), and follow-up VL at 8, 16 or 24 weeks post-treatment change. RESULTS: The correlation between the predicted and actual VL change for the 16-, 12- and 8-week baseline VL models gave the following R2 values: 0.42, 0.45 and 0.55, respectively. The model with the narrowest baseline VL window (8 weeks) most accurately pre- dicted VL response. The R2 value for this model was significantly different to those for the other models (P<0.05) and also to that of a previous model derived using clinical trial data (R2=0.50, P<0.05). The NN models were also used to predict VL trajectory. The rates of correct trajectory prediction were: 74, 72 and 87% (16-, 12- and 8-week models, respectively). Again, the 8-week model was significantly more accu- rate than the two other models (P<0.05) and compa- rable to that obtained using the clinical trial model (75%, ns). We utilized the 8-week baseline VL model S112 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 103 were noticed for tenofovir (TDF) (r=0.71; P<0.0001), ddI (r=0.71; P<0.0001) and abacavir (r=0.73; Correlation between Interpreted P<0.0001). In most cases, this discordance occurred informing GT-resistant/vPT-sensitive (96% for TDF, Genotype and VirtualPhenotype for 96% for ddI and 100% for abacavir). predicting drug resistance in HIV-1 CONCLUSION: vPT did not provide results in up to O Gallego, L Martín-Carbonero, C de Mendoza, 21% of samples (20/93) in our study. Samples with a A Corral and V Soriano high number of drug resistance mutations seem to be missing in the VircoNET database. The GT/vPT cor- Instituto de Salud Carlos III, Madrid, Spain relation is good for PIs and NNRTIs; however, is low for NRTIs, especially for TDF, ddI and abacavir. New BACKGROUND: HIV drug resistance testing has sequences belonging to heavily pretreated patients become an important rule in the clinical management and/or failing recent approved drugs need to be includ- of HIV-infected patients. However, expert interpreta- ed in the VircoNET database, in order to provide tion is essential for clinicians, mainly when facing com- helpful information. plex genotypes. The development of several algorithms for drug resistance genotypic interpretation and soft- ware to predict phenotype from genotype have both demonstrated to be helpful in the design of rescue interventions. OBJECTIVES: To investigate the concordance between an algorithm interpretation of genotypes (GT) and the VirtualPhenotype (vPT) (VircoNet). METHODS: Ninety-three HIV sequences obtained from patients failing antiretroviral therapy were inter- preted following the last IAS rules and, in parallel, by vPT. Concordance of the GT and vPT was accom- plished by categorizing each sample as either ‘sensitive’ or ‘resistant’. RESULTS: Genotypic results were obtained from 93 samples, whereas complete vPT results for all drugs were available in 78.5% (73/93). vPT did not provide results in samples with less than 10 matches within the VircoNET database. The 20 samples without vPT results affected lopinavir (n=11; 55%), didanosine (ddI) (n=9; 45%) and abacavir (n=8, 40%). Considering different drug families, vPT results were obtained in: nucleoside reverse transcriptase inhibitors (NRTIs) (88.2%; 82/93), non-NRTIs (NNRTIs) (100%) and protease inhibitors (PIs) (88.2%; 82/93). For NRTIs, samples that did not present vPT results had viruses carrying ≥5 NAMs or multinucleoside-resis- tant genotypes (complex Q151M and inserts 67–69). In the case of PIs, lack of vPT was associated with high number of PI resistance mutations (median ≥7). The overall concordance between the two assays was high- er for PIs and NNRTIs (r=0.95; P<0.0001 and r=0.94; P<0.0001, respectively) than for NRTIs (r=0.81; P<0.0001). All drugs belonging to this class, except lamivudine (r=0.93; P<0.0001), presented a GT/vPT concordance below 90%. The lowest concordances XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S113 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 104 G190A, while all single mutants, including K103N and Y188L, displayed only low to intermediate resistance Extending the range of measurable levels. resistance to efavirenz: the effect of CONCLUSION: During NNRTI escape, the develop- combinations of mutations and ment of HIV-1 resistance may be more gradual and baseline polymorphisms may reach higher levels than previously thought. Both the accumulation of known NNRTI resistance muta- tions and baseline RT polymorphisms appear to play a E Dam1 and F Clavel2 critical role in this process. 1 Viralliance, Paris; and 2 Inserm U552, Hopital Bichat, Paris, France BACKGROUND: Single mutations selected in HIV-1 reverse transcriptase (RT) following failure of non- nucleoside reverse transcriptase inhibitors (NNRTIs) are often believed to confer optimal levels of resistance. However, some viruses appear to accumulate multiple NNRTI resistance mutations over time, suggesting that in reaction to the strong pharmacological pressure exerted by NNRTIs in vivo, resistance must evolve to very high levels, which are not measured by current phenotypic assays. METHODS: We have modified the Phenoscript assay to measure higher levels of resistance to efavirenz by using a wider range of concentrations of this drug. A panel of 40 patient plasma samples carrying HIV sequences with different combinations of NNRTI resis- tance mutations was tested, along with selected site- directed mutants in NL4-3. Each virus was tested at least three times, in independent experiments. RESULTS: The modified assay was able to detect lev- els of resistance to efavirenz (fold-increase in IC rela- 50 tive to NL4-3) above 9000. Viruses with the K103N mutation alone displayed a surprisingly wide range of resistance values, from 174 to 3700. The values for each individual virus were highly reproducible, supporting a strong role for baseline polymorphisms in determining this wide range of resistance. Mutation Y181C alone yielded low efavirenz resistance, ranging 20–126. The combination of Y181C with G190A slightly increased resistance up to a range of 40–400. Combinations of Y181C with K103N only yielded moderate resistance, ranging 40–800. Stronger resistance was seen with combinations of Y181C, G190A and K103N, although they did not reach the highest levels of resis- tance detectable by the assay. Maximal resistance was seen with K103N combined with L100I, V108I or V188L, and with viruses carrying mutations G190S/E/Q, often combined with K101E or K103N. With site-directed mutants, maximal resistance was seen with combinations of Y188L with K103N or S114 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 105 LPN/RTN algorithm (codons 10, 20, 24, 46, 53, 54, 63, 71, 82, 84, 90) with the exception of 53, were found Interpreting resistance to among these patients with the following prevalence: 63 (56%), 20 (50%), 10, 46, 54, 71, 90 (31%), 82 (25%), lopinavir/ritonavir in HIV-1 84 (19%), 24 (6%). The mean LPN/RTN algorithm subtype C patients score was 3.1 (range 0–9). Mutations M36I and I93L were present in all patients but are not included in the Z Grossman1, D Torten2, E Shahar3, I Levy4, score. These mutations had been maintained through- K Risenberg5, M Chowers6, V Istomin7, out successive PI regimens. A small comparator group D Averbuch8, M Burke9, D Ram1, S Maayan7, of B patients from the same clinics showed a mean M Lorber3 and JM Schapiro10 lopinavir score of 4.3, which does not reach statistical significance. Analysis of a larger cohort is ongoing. 1 National HIV Reference Lab, PHL, MOH, Tel-Hashomer, Israel; 2 Kaplan Medical Center, Rehovot, Israel; 3 Rambam Medical CONCLUSIONS: Characteristic lopinavir-associated Center, Jerusalem, Israel; 4 Sheba Medical Center, Tel-Hashomer, mutations are present in subtype C patients failing Israel; 5 Soroka Medical Center, Beer Sheva, Israel; 6 Meir therapy similar to mutations in subtype B. Multiple Medical Center, Kfar Saba, Israel; 7 Hillel Yafe Medical Center, mutations were also commonly present, consistent Hadera, Israel; 8 Hadassah Medical Center, Jerusalem, Israel; 9 with patterns seen in B. The rate of these mutations, as Sourasky Medical Center, Tel Aviv, Israel; and 10 Stanford in position 20, may differ between subtypes as a result University, Calif., USA of baseline polymorphism. Other mutations contribut- ing to lopinavir resistance are present in subtype C BACKGROUND: Resistance mutations, conferring (such as M36I and or I93L) but were not included in reduced susceptibility and virological response to the LPN/RTN algorithm due to their low prevalence in lopinavir/ritonavir (LPN/RTN) are well documented. subtype B. This may result in potential underscoring of A specific algorithm for predicting LPN/RTN response LPN/RTN resistance in subtype C. The clinical signifi- is commonly used to interpret resistance assay results cance of this may, however, be limited due to the mod- and assist in therapeutic decisions. These results have est contribution to resistance, and should not preclude been derived predominantly from subtype B-infected LPN/RTN use in subtype C. Although differences may patients. Since mutational pathways and statistical be subtle, subtype B-based interpretation algorithms derivation of algorithms may be influenced by the sub- require validation, and possibly adjustments, in non- type-specific baseline prevalence of mutations, we subtype B populations. analysed LPN/RTN resistance in subtype C-infected patients. METHODS: All samples from subtype C-infected patients receiving LPN/RTN with HIV RNA >1000 copies/ml were included. Population sequencing was performed (TRUGENE, Visible Genetics). Virological and clinical information were collected in a central database. Mutational patterns were determined, spe- cific LPN/RTN algorithm scores calculated and longi- tudinal evolution of mutations, tracked by comparison with sequencing results from previous patient assays, performed. Results were compared with current sub- type B data. RESULTS: Twenty-eight samples from 16 LPN/RTN- failing patients were analysed. Patients had been treated earlier for 14–64 months. Mean HIV RNA was 177235 copies/ml (range 1360–884000; median 35500) and mean CD4 cell count was 212 cells/µl (range 25–734; median 183). Mean number of previ- ous protease inhibitors (PIs) was two (range 0–4). Patients were receiving ≥2 nucleoside reverse transcrip- tase inhibitors (NRTIs). All mutations included in the XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S115 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 106 substantial drop in selective pressure beginning at 40- to 60-fold reduced baseline susceptibility to LPV. Exploration of methodology for Among patients with ≥4 baseline PI mutations, addi- tional resistance was selected in 13/19, 2/4 and 1/6 estimating upper clinical patients with <40-fold, 40- to 60-fold, and >60-fold breakpoints for lopinavir/ritonavir baseline reduced susceptibility to LPV. The mean fold changes in LPV IC between baseline and rebound by analysis of the emergence of 50 were 15.2-, 2.4- and 1.2-fold for the above three resistance during virological failure groups, respectively (median changes 2.1-, 2.1- and in experienced patients 1.1-fold, respectively). RC Stevens, M King, H Mo, A Molla, S Brun CONCLUSIONS: In PI-experienced patients receiving and D Kempf LPV/r, the likelihood of emergence of additional resis- tance during virological failure appears to be depen- Abbott Laboratories, Abbott Park, Ill., USA dent upon both baseline genotype and phenotype. Evidence of selective pressure during viral rebound BACKGROUND: Clinical breakpoints for protease may be a useful indicator for defining upper genotypic inhibitors (PIs) are typically determined by analysis of and phenotypic breakpoints for antiretroviral agents. virological response with respect to baseline phenotype and/or genotype. However, definition of an upper breakpoint (the degree of resistance above which there is little evidence of antiviral activity) can be complicat- ed by the need to recruit a sufficient number of study subjects with highly resistant virus in order to define a ‘no-effect’ level of resistance. In this investigation, we have explored the analysis of the selection of addition- al resistance in PI-experienced patients during failure of lopinavir/ritonavir (LPV/r) therapy as an alternate method for estimating an upper breakpoint for this boosted PI. METHODS: The emergence of additional resistance in single or multiple PI-experienced patients receiving LPV/r plus a non-nucleoside reverse transcriptase inhibitor (NNRTI) and NRTIs in two Phase II and one Phase III study was examined by analysis of baseline (pre-LPV/r treatment) and post-rebound genotype and phenotype. Selection of additional resistance, defined as a ≥twofold change in IC between baseline and 50 rebound and/or emergence of a new primary PI muta- tion, was analysed with respect to baseline genotype and phenotype. RESULTS: Baseline and rebound genotype and phe- notype were available from 40 single and 13 multiple PI-experienced patients. A logistic regression model (P-value for model <0.001) suggested maximal selec- tive pressure at 4–6 baseline PI mutations, with little selective pressure <2 or >7 PI mutations. Thus, no resis- tance emerged in the rebound isolates from 13 patients with 0–1 baseline PI mutations, while in contrast, the selection of additional resistance was evident in 3/11, 9/11, 6/14, 1/4 patients with 2–3, 4–5, 6–7 and ≥8 baseline PI mutations, respectively. A logistic regres- sion model (P-value for model =0.012) suggested a S116 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 107 was –1.23 log and –0.27 log in those with one and at least two mutations, respectively. In the multivariate Clinically relevant interpretation of analysis this score was an independent predictor of the response. The bootstrap analysis showed the robust- genotype for resistance to ritonavir ness of the score. (100 mg twice daily) plus saquinavir (800 mg twice daily) in CONCLUSION: We developed a clinically relevant interpretation of genotype for resistance to ritonavir HIV-1-infected protease (100 mg twice daily) plus saquinavir (800 mg twice inhibitor-experienced patients daily). This genotypic score was strongly linked to viral response in PI-experienced patients receiving AG Marcelin1, C Dalban2, G Peytavin3, saquinavir as part of a boosted regimen and should be C Delaugerre1, R Agher2, C Katlama2, validated in other datasets. D Costagliola2 and V Calvez1 1 Pitie-Salpetriere Hospital; 2 INSERM EMI 0214, Paris; and 3 Bichat-Claude Bernard Hospital, Paris, France BACKGROUND: The combination of saquinavir with ritonavir results in the increase of saquinavir drug lev- els that may be sufficient to overcome partially pro- tease inhibitor (PI)-resistant virus. The aim of this study was to identify and validate a clinically relevant genotypic score for resistance to ritonavir (100 mg twice daily) plus saquinavir (800 mg twice daily)-con- taining regimen in PI-experienced patients. METHODS: Seventy-two PI-experienced patients who received ritonavir (100 mg twice daily) plus saquinavir (800 mg twice daily)-containing regimen were analysed retrospectively. No non-nucleoside reverse transcriptase inhibitor or PI other than saquinavir or ritonavir were used in the antiretroviral combinations. The impact of each mutation in the protease gene on the virological response to boosted saquinavir regimen was studied in non-parametric univariate analyses. Mutations with a P-value below 0.10 were retained for further analysis. According to the number of muta- tions, in the best combination, three levels of resistance were defined. A multivariate analysis accounting for confounding variables assessed whether the genotypic score was an independent predictor of the response. The robustness of the score was analysed using the bootstrap re-sampling method. RESULTS: Seventy-two patients (mean baseline plas- ma HIV-1 RNA 4.38 log copies/ml, CD4 cell count 292 cells/mm3) were included in the study. The median duration of prior PI treatment was 24 months. In the 72 patients exposed to ritonavir plus saquinavir, the strongest association between the decrease in viral load and the number of mutations was observed with a set of five mutations at codons 24, 62, 82, 84, 90 of the protease gene. In patients with 0 mutation the median decrease in viral load was –2.2 log copies/ml, while it XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S117 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 108 PI resistance or secondary PI mutations increased con- sistently with higher adherence. For example, for NFV- Exploring the effects of adherence treated patients with genotype data, at 80, 90 and 100% adherence, probabilities of primary PI resistance on resistance: use of local linear were 33, 49 and 63%, respectively. No primary PI regression to reveal relationships resistance was observed in LPV/r-treated patients. A between adherence and resistance model describing the bell-shaped relationship observed between adherence and primary PI resistance in NFV- in antiretroviral-naive patients treated patients was explored. The model suggested treated with lopinavir/ritonavir or that at low adherence, many patients have detectable nelfinavir viral load, but selective pressure is low, resulting in lower resistance rates. At intermediate adherence, fewer patients have detectable viral load, but selective M King, S Brun, J Tschampa, J Moseley and pressure is higher, resulting in maximal resistance D Kempf rates. At near-perfect adherence, selective pressure is highest among patients who have detectable viral load, Abbott Laboratories, Abbott Park, Ill., USA but resistance rates remain lower because more patients maintain undetectable viral loads. BACKGROUND: The relationship between adherence and virological response has been extensively studied, CONCLUSIONS: Local linear regression revealed a but the relationship between adherence and resistance bell-shaped relationship between adherence and resis- is less well understood and may differ for different tance, with maximal probability of resistance at inter- drugs. Local linear regression techniques may reveal mediate resistance levels. The bell-shaped relationship relationships more difficult to discern using standard between adherence and primary PI resistance for NFV- statistical methods. treated patients appeared to be driven by lower selec- tive pressure at lower adherence, despite low rates of METHODS: In a Phase III study of antiretroviral-naive viral suppression, and by better viral suppression at patients receiving lopinavir/ritonavir (LPV/r) (n=326) higher adherence, despite maximal selective pressure. vs nelfinavir (NFV) (n=327), each with stavudine Risk of viral rebound or resistance was substantially (d4T)/lamivudine (3TC), for up to 2 years, the proba- lower for LPV/r vs NFV across all adherence levels. bility of resistance development by adherence level was analysed. Adherence was measured by pill count, which may overestimate adherence but is correlated with electronic adherence measures. The probability of primary protease inhibitor (PI) resistance, secondary PI mutations/polymorphisms or 3TC resistance was assessed by local linear regression, among (A) all enrolled patients and (B) only patients with detectable viral load and genotype data. RESULTS: (A) Among all enrolled patients, a bell- shaped relationship between resistance and adherence was observed, with the highest probability of resis- tance at intermediate adherence. Among NFV-treated patients, through 2 years of therapy: at 75, 85 and 95% adherence, probabilities of primary PI resistance were 12, 20 and 16%, respectively. A similar relation- ship was observed for the probability of secondary PI mutations/polymorphisms or of 3TC resistance in each treatment group. For example, at 70, 80 and 95% adherence, probabilities of 3TC resistance were 38, 49 and 23%, respectively, for NFV-treated patients, and 2, 13 and 5%, respectively, for LPV/r-treated patients. (B) In patients with detectable viral load and genotype data, in NFV-treated patients, probability of primary S118 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 109 mutation score <3 predicted the absence of resis- tance to TDF and >5 mutations resistance to TDF Genotypic and pharmacological and corresponded to reductions in VL of –1.3 ±1.1, and +0.4 ±0.7 log copies/ml, respectively. In patients 10 determinants of the virological with a mutation score comprised between 3 and 5, inclu- response to tenofovir in nucleoside sive, the decrease in VL was –0.8 ±1.0 log copies/ml. In reverse transcriptase the multivariate analysis, a TDF mutation score >5, previous use of amprenavir, indinavir and lopinavir, inhibitor-experienced patients and co-presciption of didanosine (ddI) were associated with a worse virological response, while a higher base- B Masquelier1, C Tamalet2, B Montès3, line VL was associated with a better response. The D Descamps4, G Peytavin4, L Bocket5, bootstrap analysis showed the robustness of the score. M Wirden6, J Izopet7, V Schneider8, V Ferré9, The median tenofovir plasma level was at 71 ng/ml A Ruffault10, P Palmer11 , A Trylesinski12, (range 5–417; n=135). The median TDF plasma level M Miller13, F Brun-Vézinet4, D Costagliola14 was lower in patients receiving ddI than in patients and the ANRS AC11 Resistance study group without ddI (58 vs 74 ng/ml, respectively; P=0.023), however, both median values were within normal 1 Laboratoire de Virologie, CHU de Bordeaux, France; 2 CHU La ranges. Most patients were considered as highly adher- Timone, Marseille, France; 3 CHU de Montpellier, France; 4 ent to TDF medication (13% METHODS: RT and protease genotype was assessed at baseline in a subgroup of 161 patients of the French expanded access programme with a baseline viral load (VL)≥4000 copies/ml and receiving a stable TDF-includ- ing regimen during 3 months or more. RT mutations associated with the virological response (HIV-1 RNA decrease at month 3) with a P-value <0.15 were retained for the construction of a mutation score. The score was then validated using a multivariate analysis accounting for confounding variables and bootstrap resampling method. The tenofovir plasma levels were estimated from a single plasma sample taken after 1 month of dos- ing using HPLC (LOQ=5 ng/ml). RESULTS: The strongest association between the decrease in VL and the number of mutations was observed with a set of seven mutations (TDF mutation score) including M41L, E44D, D67N, T69D/N/S, L74V, L210W and T215Y/F RT mutations. A TDF XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S119 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 110 EFV Del NVP N Preferred codons Mutations at reverse transcriptase K103R <2 <2 <2 115 AGA>AGG K103Q <2 <2 <2 5 AGC>AGT codon 103: phenotypic resistance K103S 10 19 >50 16 CAA>CAG to non-nucleoside reverse K103T 15 >50 >50 2 ACA transcriptase inhibitor and K103H >50 >50 >50 4 CAT clinical correlates Site-directed mutagenesis confirmed that K103R con- ferred S120 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 111 ciation of mutation 135 being present in non-B sub- types (41% B, 59% non-B, P=0.038). Mutations at Genotypic associations with codon 135 were less likely to be present with changes at codon 179 (P=0.010) and codon 98 (P=0.040). non-nucleoside reverse Other observations included an association of changes transcriptase inhibitor susceptibility at codon 178 with changes at codon 98 (P=0.047); in circulating recombinant forms mutations at codon 122 were associated with no of HIV-1 strains in North and change at codon 188 (P=0.017). South America CONCLUSION: This naive population expressed virus with some decreased susceptibility to EFV and A Florance, C Vavro, M St Clair and D Irlbeck NVP. There is some evidence of an association between changes at position 135 and non-B subtypes. While GlaxoSmithKline, Research Triangle Park, NC, USA changes at position 135 occurred in subtype B, there was a larger proportion of subjects in non-B subtypes BACKGROUND: The HIV epidemic in North and with polymorphisms at this position. There was an South America has been dominated by subtype B HIV-1 association between changes at position 135 in the infection, while reports from other geographic areas absence of other known NNRTI mutations with indicate increasing percentages of circulating recombi- decreased susceptibility to NVP but not to EFV. These nant forms (CRFs). This study examined the phylo- results support recent observations suggesting changes genetic analysis of the reverse transcriptase (RT) cod- at position 135 impact NNRTI susceptibility. ing region in HIV-1 from subjects in North and South America to determine the incidence of CRF strains in this population and the potential genetic association with non-nucleoside RT inhibitor (NNRTI) suscepti- bility. METHODS: Matched HIV genotypes and phenotypes were evaluated using plasma samples from 194 anti- retroviral (ART)-naive subjects from a subset of two controlled clinical trials. HIV-1 genotyping was per- formed using either the ViroSeq HIV-1 Genotyping or the TRUGENE HIV-1 Genotyping Systems. Phenotypic drug susceptibility was determined using the PhenoSense HIV Drug Resistance Assay. HIV subtype was determined by analysing the HIV-1 RT nucleotide sequence from each subject using the National Center for Biotechnology Information (NCBI) Retroviruses HIV Subtyping Tool. RESULTS: Of the 194 subjects, 24% possessed a non- B HIV-1 subtype. Thirteen percent had viruses with decreased susceptibility (≥2.5-fold resistance) to efavirenz (EFV) and/or nevirapine (NVP). Seventy-six percent of the subjects did not have any known NNRTI mutations. Any single NNRTI resistance mutation was represented by fewer than 8% of the viruses. In the absence of known NNRTI mutations, reduced susceptibility was observed with changes at codon 135. A mutation at codon 135 was accompa- nied by a decrease in susceptibility to NVP (82% sus- ceptible with mutation, 92% susceptible without mutation, P=0.045). This decrease was not evident for EFV (92% susceptible with mutation and 93% suscep- tible without mutation). There was a significant asso- XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S121 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 112 line VL (P=0.017), protease substitution L20M (P=0.076). Independent predictors were number of Long-term clinical and daily administrations of initial salvage HAART [OR 0.64 (95% CI: 0.42–0.97) P=0.05], and 3-month viro-immunological outcomes of change from baseline VL [for each log less OR 0.66 Argenta, a randomized trial on the (0.46–0.93) P=0.017]. Independent predictors of clini- usefulness of HIV-1 drug resistance cal progression were CDC class C at entry (HR 2.79, P=0.02), higher baseline VL (each log more HR 2.25, genotyping and expert advice P=0.04) and less profound 3-month VL reduction (for each log less HR 1.73, P=0.03). In a separate Cox's A Cingolani, S Di Giambenedetto, A Bacarelli, model using baseline virological co-variates only, pro- R Murri, A Ammassari, R Cauda and tease substitutions L20M (HR 3.05, P=0.03) and I84V A De Luca (HR 3.31, P=0.03) independently predicted clinical progression. Catholic University, Rome, Italy CONCLUSION: Salvage therapy guided by G with EA BACKGROUND: To evaluate predictors of long-term conferred a sustained virological and immunological virological outcomes and clinical progression in benefit independently from the number of failed regi- patients from a drug resistance genotyping trial. mens and immediate or deferred use of genotyping. The long-term virological and clinical outcome was influ- METHODS: Argenta was a single-centre randomized enced by the presence of selected protease substitutions trial comparing resistance genotype-guided treatment and by the initial virological response. Although existing (G) versus SOC, both with expert advice (EA) in trials of resistance testing have limited follow-up, early patients failing highly active antiretroviral therapy (3-month) virological response may predict virological (HAART) [viral load (VL)>2000]. After the end of the and clinical outcome over 3 years. trial (month 6), all patients received G+EA. All patients have reached 36 months of follow-up. Predictors of virological success (HIV RNA<500) were analysed by logistic regression; association with time to clinical progression (new AIDS-defining events or death) by Cox's models. RESULTS: From 04/99 to 02/00, 174 patients were randomized (85 G, 89 SOC). Baseline characteristics were homogeneous except for a higher proportion of IDU in SOC (42 vs 22%). Median antiretroviral exposure was 24 months (range 19–35); 25% had failed ≥3 HAART regimens. At 36 months drop-outs were 23.6%. By intent-to-treat analysis (dropouts=fail- ures), the proportion of patients with virological success differed between arms only at 3 months [12% SOC, 27% G (P=0.01)]; virological responders increased over time in both arms, with 27 and 33% in SOC and G, respectively, at 36 months. HIV RNA levels were reduced over time in both arms: at 36 months mean change from baseline was –1.21 log. CD4 counts homo- geneously increased in both arms over time: at 36 months mean change from baseline was +86l cells/µl. Twenty-four patients showed clinical progression (6.4/100 PY), which was not associated with initial randomization arm. Univariate predictors of virologi- cal success at 36 months were: virological suppres- sion before trial entry (P=0.002), number of daily administrations of initial salvage HAART (P=0.05), baseline VL (P=0.069), 3-month change from base- S122 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 113 in 41%, pan-nucleoside resistance in 33% and aba- cavir resistance in 31%. In all, 43% of providers were Clinician knowledge and attitudes able to correctly identify resistance-associated muta- tions for five out seven drug groups, 18% for all seven towards genotypic testing drug groups and 27% were unable to recognize muta- tions for any one drug group. Number of patients C Salama under care was associated with improved ability to rec- ognize resistance-associated mutations: clinicians with Elmhurst Hospital Center, NY, USA ≤50 patients could recognize an average of 1.38 muta- tions (56% answered 0 correctly), those with 51–100 OBJECTIVE: In the Spring of 2001 we distributed a patients identified 3.47 mutations correctly, those with questionnaire at the International AIDS Society – USA >100 patients identified 4.50 mutations correctly and New York City meeting in order to evaluate clinician those with >250 patients were able to identify an aver- attitudes and knowledge with respect to genotypic test- age of 5.80 mutations correctly (out of seven drug ing (GT). We distributed a similar questionnaire at the groups). same meeting 2 years later to see how clinician atti- tudes and knowledge have changed. CONCLUSION: Although still somewhat limited, ability to identify resistance-associated resistance METHODS: An anonymous survey was distributed by mutations has improved over the past 2 years among a conference staff to all attendees. The knowledge por- selected group of HIV providers in New York City. tion of the questionnaire asked clinicians to match list- ed mutation(s) with any one of seven drug groups: lamivudine, zidovudine/stavudine, abacavir, pan-nucle- oside resistance, nelfinavir, non-nucleosides, tenofovir. RESULTS: Four-hundred-and-twenty-eight HIV providers attended the meeting: 336 (79%) physicians, 92 (21%) nurse practitioners and physician’s assis- tants. One-hundred-and-twenty-six (29%) question- naires were collected. Ten were excluded (did not use GT). Of the remaining 116, 76 (66%) were physicians (24 ID, 52 non-ID) and 40 (34%) were non-physi- cians. Provider patient load varied greatly: 32 (28%) ≤50 patients, 32 (28%) 51–100 patients, 52 (44%) over 100 patients, 30 (26%) physicians with over 100 patients, 9 (8%) with >250 patients. Clinicians were asked their level of expertise: no experience (none), lit- tle to some expertise (20; 17%), moderate to high expertise (66; 57%), and HIV experts (26; 22%). Eighteen percent send GT for naive, chronically infect- ed patients, 25% for pregnant patients, 33% for patients with acute infection, 79% for first failures, 67% for second failures and 67% for salvage situa- tions. Seventy-two (62%) considered themselves able to interpret GT results and 45 (39%) utilize an HIV resistance expert to help with interpretation. Provider knowledge was assessed by counting the number groups for which at least one correct mutation was identified. Sixty-six percent were able to match M184V with lamivudine, 53% correctly identified non-nucleoside reverse transcriptase inhibitor (NNRTI) mutation(s), and 52% were able to correctly identify both lamivudine and NNRTI mutations. At least one TAM was correctly recognized in 50% of cases, nelfinavir resistance in 47%, tenofovir resistance XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S123 Los Cabos, Mexico, 10–14 June 2003 PLENARY ABSTRACT 114 How good is the evidence that HIV specific CTL can protect against and control infection? R Phillips John Radcliffe Hospital, UK During 15 years of research the reputation of CD8 cytotoxic T cells in retroviral infections has been mixed. These cells were initially thought to be an epiphenomenon and unlikely to be the major force in immunity, which prevented and controlled viral repli- cation. There has been accumulating evidence that these cells appear at a time when the viraemia is first brought under control. The depletion of these cells in simian immunodeficiency virus (SIV) infection leads to a rapid rise in viraemia; their natural restoration after this transient depletion brings SIV under control again. Mounting evidence from simian models such as those used in these depletion experiments and from human studies have clearly shown that sites restricted by HLA Class 1 molecules within HIV are under selection pres- sure. There remains serious uncertainty as to which events first trigger loss of HIV control in chronic phase infection. Exposed but uninfected individuals do have detectable cytotoxic T lymphocytes. These cells appear to wane when intense exposure to HIV stops. We have exam- ined both vertical and sexual transmission of HIV in HLA Class 1-matched and -mismatched individuals. There is clear evidence from these studies that Class 1- restricted antigens that have escaped T cell responses are transmissible and in an individual who bears the same Class 1 molecule this immune escape may dictate the subsequent course of infection. Antigens favoured by T cells induced in an uninfected stage of disease anti- genic variation was not detectable in virus that caused seroconversion in this heavily exposed individual. We remain remarkably ignorant as to the fundamental requirements for immunological protection and con- trol in HIV infection. XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S125 SESSION 5 Epidemiology Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 115 CONCLUSIONS: Transmitted NRTI, NNRTI and PI drug resistance persisted during the entire period of Persistence of transmitted follow-up for 9/10 subjects with complete reversion of a transmitted K103N variant documented in only one drug-resistant virus among subjects subject, nearly 3 years after infection. The durable per- with primary HIV infection sistence of transmitted drug-resistant virus is consistent deferring antiretroviral therapy with the establishment of widespread infection with a pure population of resistant clone(s), in contrast to the rapid reversion observed in chronically infected sub- SJ Little1, K Dawson2, NS Hellmann2, 1,3 1 jects who discontinue therapy after virological failure. DD Richman and SDW Frost Reversion of resistance is gradual and usually incom- plete, resulting in the persistence of mixtures of wild- 1 University of California, San Diego, Calif.; 2 ViroLogic, Inc., type and resistant variants in plasma HIV RNA. No South San Francisco, Calif.; and 3 VA Medical Center, San Diego, reversion of protease resistance was observed (n=3), Calif., USA suggesting that the time to reversion is comparable to or even greater than that observed for K103N. The OBJECTIVES: To assess the persistence of transmitted continued presence of detectable K103N/K mixtures drug-resistant variants in the absence of antiretroviral despite the short mutational distance between these drug (ARV) treatment among subjects with primary mutants suggests that the mutation confers a small fit- HIV infection. ness cost consistent with high RC values. The persis- tence of all classes of drug-resistant variants has signif- METHODS: Baseline nucleotide sequence analysis of icant implications for the treatment of ARV-naive sub- pol was used to identify primary drug resistance muta- jects and subsequent secondary transmission of drug- tions among 10 subjects with primary HIV infection resistant variants. who chose to defer ARV. ARV drug susceptibility and replication capacity (RC) were measured using a single replication cycle assay (ViroLogic). RESULTS: All 10 subjects (mean time from estimated date of infection: 65 days) had at least one major drug resistance mutation identified at presentation with cor- responding phenotypic resistance. The median baseline viral load was 5.5 log copies per ml (range 2.5–7.4) 10 and mean RC was 87% of wild-type (WT). Longitudinal samples were collected for a median of 177 days (range 82–1019 days) after infection and analysed for persistence of transmitted drug-resistant variants. Seven subjects were identified with non- nucleoside reverse transcriptase inhibitor (NNRTI) resistance (103N ±181C), one with NRTI (70R, 74I, 215Y) and protease inhibitor (PI) resistance (46I, 84V, 90M), one with NNRTI (188L) and PI resistance (30N), and one with three-class drug resistance (103N, 215Y, 84V, 90M). The average time to rever- sion of 103N variants to mixed 103N/K populations was 196 days following the estimated date of infection (153-238 days, 95% CI). In the three patients with protease resistance mutations, no reversion was detect- ed at 64, 191 and 342 days after infection. Complete reversion of genotypic resistance was observed in only one patient at 1019 days after infection. In this patient, viral load remained high (5.2 log copies/ml) during 10 reversion. Many persistent drug-resistant viruses had high RC. XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S129 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 116 of 159 HIV-negative or status-unknown partners. Although only 4.7% (16/338) of all patients had both Patients with antiretroviral- ART(R) virus and HRTB, this group accounted for 23% (171/728) of all HRTB events and exposed 19% resistant HIV infections engaging in of HIV-negative partners. Among source patients with high-risk transmission behaviour detectable VL (>400 copies), 49% (171/350) of all HRTB events were with patients with ART(R) virus, M Kozal1, KR Amico2, J Chiarella1, resulting in the exposure to ART(R) virus in 39% of T Schreibman1, J Fisher2, D Cornman2, HIV-negative partners engaging in HRTB. W Fisher3 and G Friedland1 CONCLUSION: The prevalence of ART(R) in patients 1 Yale University School of Medicine, New Haven, Conn.; and 2 receiving care was 28% and was similar among sexes University of Connecticut, Conn., USA and HIV transmission route risk groups, but differed among racial groups. A significant proportion (23%) BACKGROUND: Previous studies have demonstrated of patients in care continue to engage in high-risk that 10–20% of new HIV infections are with anti- transmission behaviour. A small subset of these also retroviral therapy (ART)-resistant (R) strains. The have resistant virus (<5% of patients). However, this major source of new ART(R) infections are patients core group of patients in care accounted for a large receiving ART in care who engage in high-risk HIV number of high-risk sexual HIV transmission events sexual transmission behaviours with HIV-negative or with resistant virus and potentially expose a substan- status-unknown partners. However, little is known tial number of at risk partners. These data may help about the characteristics and relationship of ART(R) explain the rising prevalence of new ART(R) infec- and the high-risk HIV transmission behaviours tions, assist in determining transmissibility of ART(R) (HRTB) among HIV-positive patients in care. strains and in developing prevention strategies to reduce ART(R) transmission. OBJECTIVE: To define the characteristics and rela- tionship of ART(R) and continuing HRTB, in patients in care, and the potential partner risk of ART(R) virus transmission. METHODS: ART genotypic resistance testing was per- formed on viral load (VL) detectable specimens using automated sequencing and linked to patient demogra- phy, sexual behaviour, adherence, ART history and CD4/VL information collected as part of an ongoing assessment of HIV-positive patients enrolled in the Options Project during 2000–2002 (study of a clini- cian based intervention to reduce HIV transmission risk). HRTB was defined as unprotected oral, vaginal or anal sex with partner believed to be HIV-negative or unknown HIV status. RESULTS: HIV genotypic data and sexual risk behav- ioural data was available on 338 patients. Of these, 28% (93/338) had ART(R). No difference was found in the prevalence of ART(R) by gender (male 28% vs female 26%), age (44 vs 43) or HIV transmission route (MSM 28%, heterosexual 26%, IDU 29%) but ART(R) differed by race: white 15% (11/71), African- American 27% (35/132), Latino 35% (39/111); P=0.014. In a 3 month time period, 23% (77/338) of HIV-positive patients reported a total of 1151 unprotected sexual events involving 159 HIV-nega- tive and 36 HIV-positive partners. 63% (728/1151) of these were HRTB events and exposed a minimum S130 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 117 sequences were received from the following countries: Austria (60), Belgium (61), Denmark (132), Finland (8), Prevalence of transmitted drug Germany (62), Greece (40), Israel (104), Italy (296), Luxembourg (163), the Netherlands (25), Norway (23) resistance in Europe is largely Poland (35), Portugal (124), Serbia-Montenegro (10), influenced by the presence of Spain (23) and Switzerland (262). Fifty-eight subjects non-B sequences: analysis of 1400 were excluded, because the sequences were obtained outside the study period or because the subjects were patients from 16 countries: the younger than 18 years of age. The prevalence of resis- CATCH-Study tance was assessed over the period of 1996–2002 based on the IAS resistance table (update of December AMJ Wensing1, DAMC van de Vijver1, B Asjo2, 2002). Sequences were subtyped by comparing them to C Balotta3, R Camacho4, C de Mendoza5, the reference sequences of known subtypes. Statistical S Deroo6, I Derdelinckx7, Z Grossman8, analyses were performed with the χ2 test or logistic O Hamouda9, A Hatzakis10, A Hoepelman1, regression analysis when appropriate. A Horban11 , K Korn12, C Kuecherer9, C Nielsen13, V Ormaasen14, L Perrin15, RESULTS: Primary drug-resistant mutations were D Paraskevis10, E Puchhammer16, F Roman6, detected in 11% (147/1369) (95% CI: 9.1–12.5) of M Salminen17, JCC Schmit6, V Soriano5, antiretroviral-naive subjects; RT 9% (126/1352) G Stanczak11 , M Stanojevic18, A-M Vandamme7, (95% CI: 7.8–11.0) and protease inhibitor (PI) 2% K Van Laethem7, M Violin3, S Yerly15, (30/1343) (95% CI: 1.5–3.2). Population characteris- M Zazzi19 and CAB Boucher1 on behalf of the tics were available for 975 sequences. The prevalence SPREAD Programme of primary drug resistance-related mutations was 11 % (63/596) in seroconverters (infected <1 year) versus 1 University Medical Center Utrecht, Utrecht, the Netherlands; 2 8% (30/379) in newly diagnosed chronically infected University of Bergen, Bergen, Norway; 3 University of Milan, subjects (P=0.17). 31% (423/1369) of the sequences Milan, Italy; 4 Hospital Egas Moniz, Lisbon, Portugal; 5 Hospital were classified as non-subtype B. In all countries, Carlos III, Madrid, Spain; 6 Centre Hospitalier de Luxembourg, except for Israel, the prevalence of resistance was much Luxembourg; 7 Katholieke Universiteit Leuven, Leuven, Belgium; higher in subtype B sequences compared to non-B. 8 Sheba Medical Center, Tel-Hashomer, Israel; 9 Robert Koch After adjustment for Israel, mutations conferring resis- Institute, Berlin, Germany; 10 Athens University Medical School, tance were found in 12% (108/922) of subtype B Athens, Greece; 11 Hospital for Infectious Diseases & AIDS sequences and in 5% (18/343) of non-B [OR=1.99, Diagnosis and Therapy Center, Warsaw, Poland; 12 University of (95% CI: 1.27–3.12), P=0.003]. Revertant mutants Erlangen, Erlangen, Germany; 13 Statens Serum Institute, possibly emerging from mutation 215Y/F of RT were Copenhagen, Denmark; 14 Ullevaal University Hospital, Oslo, identified in 2% (26/1352) of the sequences. Norway; 15 Geneva University Hospital, Geneva, Switzerland; 16 University of Vienna, Vienna, Austria; 17 National Public Health CONCLUSIONS: The frequency of primary resistance Institute, Helsinki, Finland; 18 University of Belgrade, Belgrade, in this large European survey is higher in subjects with Serbia-Montenegro; and 19 University of Siena, Siena, Italy B subtype as compared to subjects with non-B subtype. Most likely this is caused by a longer history of treat- BACKGROUND: The reported frequency of transmit- ment in the subtype B HIV-infected population in ted drug resistance varies widely between Europe and Europe as a whole. However, the frequency of resis- the USA. It is unknown whether this variation is based tance in subtype B both in RT and protease is signifi- on differences in resistance within the treated HIV- cantly lower in Europe compared to the USA. infected population and risk-related behaviour, or whether it can be explained by the expansion of non-B sequences in Europe. The CATCH-study (Combined Analysis of resistance Transmission over time of Chronically and acute infected HIV patients in Europe) combines results from recently and chronically infect- ed patients of 16 European countries. METHODS: Baseline sequences of 1427 newly diag- nosed subjects were collected as part of the CATCH study. Reverse transcriptase (RT) and protease XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S131 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 118 lates were resistant to all three drugs. Among PI-resis- tant isolates, most isolates demonstrated reduced sus- Trends in genotypic and phenotypic ceptibility to all PIs (42%) or to one PI (20%), usually nelfinavir (18%). If an alternate cut-point of 10 FR for resistance among clinical samples PIs other than nelfinavir was explored as an approxi- submitted for routine HIV mation for resistance to boosted PI regimens, 38% of resistance analysis isolates were classified as resistant to one PI, 40% to four or five PIs and only 5.5% resistant to all PIs. Changes in phenotypic resistance patterns since 1998 L Bacheler1, H Vermeiren2, P McKenna2, included declines in three-class resistance (41%) and to M Van Houtte2 and P Lecocq2 NRTIs plus PIs (104%), and increases in resistance to NNRTIs (57%) and NNRTIs plus NRTIs (53%). The 1 VircoLab, Inc., Durham, NC, USA; and 2 Virco BVBA, Mechelen, proportion of isolates without resistance increased Belgium modestly (21%). BACKGROUND: Resistance to antiretroviral drugs is CONCLUSION: Resistance remained extensive a major factor contributing to failure of HIV-1 thera- throughout the period 1998–2002. NAMS and PI py. Patterns of drug resistance are likely to evolve over resistance declined, while NNRTI resistance increased. time with changes in treatment strategies and the intro- Changing treatment practices, expanding utilization of duction of new antiviral agents. resistance testing and evolution of the patient popula- tion may have contributed to changes in patterns of METHODS: Genotypes of >60000 routine clinical sam- genotypic and phenotypic resistance. ples (excluding clinical trials) submitted to Virco between 1998 and 2002 were surveyed for trends in resistance-associated mutations and predicted phenotyp- ic resistance assessed utilizing the VirtualPhenotype. RESULTS: Among routine samples submitted in 2002 (>12000), reverse transcriptase (RT) gene mutations 184V (43%) and 103N (29%) were most common, followed by the nucleoside analogue resistance muta- tions (NAMs) 41L (22%), 215Y (22%), 67N (20%), 70R (14%) and 210W (12%). Between 1998 and 2002, the proportion of isolates with NAM mutations declined by 9–70% (219E and 215Y, respectively), 184V remained relatively constant (11% decline) and 103N increased by 25%. The largest increases (>50%) included mutations associated with non-nucleoside RT inhibitor (NNRTI) resistance (225H, 101P, 190Q and S, 103S, and 106I), as well as the 65R mutation. In the protease gene, mutations 77I, 36I, 10I and 90M were most common in 2002 (33–27% of isolates), although 10I and 90M had declined by 62 and 82%, respective- ly, since 1998. Increases of >50% were observed for mutations 50L, 54M and L, 33F, and 47V. In 2002, 70% of samples were predicted to be phenotypically resistant to one or more drugs [60% to NRTIs, 45% to NNRTIs and 36% to protease inhibitors (PIs)], with 18% of samples resistant to drugs in all three classes. Two-class resistance was observed in 12% (NRTIs and PI) to 16% (NRTIs and NNRTIs) with resistance to a single class in 12% (NRTIs), 9% (NNRTIs) and 1.2% (PIs). Among NRTI-resistant isolates, resistance to lamivudine (40%) or zidovudine (15%) was most common. Eighty-three percent of NNRTI-resistant iso- S132 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 119 sex with women (P<0.02 for all associations). Using logistic regression, independent associations with Prevalence of mutations associated MAR in men were seen for white race (OR=1.5; 95% CI 1.1–3.1) and MSM (OR=2.4, 95% CI 1.1–4.7). with antiretroviral drug resistance Univariate and multivariate analyses showed no signif- among men and women newly icant differences associated with MAR in female sub- diagnosed with HIV in 10 US cities, groups. Among the 182 recently infected, MAR preva- lence was 11.5%. Respective prevalences of MAR to 1997–2001 NRTI, NNRTI and PI were 8.8, 3.4 and 2.8%, respec- tively, 3.4% had MAR to >1 drug class. Among the DE Bennett, IF Zaidi, W Heneine, T Woods, recently infected, 13.2% of the 152 males and 3.3% of JG Garcia-Lerma, AJ Smith, L McCormick and the 30 females had MAR. MAR proportions in sub- HS Weinstock groups of >25 persons were: white males (n=74) 14.9%, black males (n=37) 16.2%, Hispanic males Centers for Disease Control and Prevention, Atlanta, Ga., USA (n=31) 3.2%; MSM (n=111) 15.3%, MSW (n=27) 7.4%; FSM (n=28) 3.6%. BACKGROUND: Studies report that resistance to antiretroviral drugs is high in some areas among indi- CONCLUSIONS: Overall MAR prevalence is lower in viduals newly diagnosed with HIV. Many US studies our sample than in some US studies in which white are small; most have limited demographic and risk fac- MSM predominate. Reported prevalences partly tor variation. reflect the demographic and risk factor composition of the sample. For each sub-group, MAR prevalence METHODS: We prospectively evaluated the preva- reflects access to antiretroviral drugs, adherence and lence of mutations associated with antiretroviral drug other factors in transmitting partner groups. resistance (MAR) among newly diagnosed untreated persons with HIV (not AIDS) sequentially enrolled in 10 US cities 1997–2001. MAR were defined as those included in the 12/02 International AIDS Society drug resistance mutations list and footnotes. Plasma speci- mens were analysed using conventional sequencing. Recent infection was evaluated using the serological testing algorithm for recent HIV infection (STARHS). Information on men and women was analysed sepa- rately for factors associated with MAR. RESULTS: Of 1082 participants, 802 (75%) were male and 280 were female. MAR prevalence was 8.3%. Respective prevalences of MAR to nucleoside reverse transcriptase inhibitor (NRTI), non-NRTI (NNRTI) and protease inhibitor (PI) were 6.4, 1.7 and 1.9%, respectively, 1.3% had MAR to >1 drug class. 9.1% of men and 6.1% of women had MAR (OR=1.6 95% CI 0.9–2.8.). MAR proportions in subgroups of >25 persons were as follows: white males (n=262) 13.0%, black males (n=292) 5.8%, Hispanic males (n=210) 7.6%; white females (n=30) 13.3%; black females (n=206) 4.9%; Hispanic females (n=30) 10.0%; men who had sex with men (MSM) (n=482) 11.6%; male injection drug users (IDU) (n=75) 8.0%; men who had sex with women (MSW) (n=245) 4.5%; females who had sex with men (FSM) (n=227) 4.4%; female IDU (n=35) 5.7%. In univariate analysis, white men were significantly more likely than black or Hispanic men to have MAR; MSM and male IDUs were significantly more likely to have MAR than men exposed through XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S133 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 120 resistance to other drug classes. For comparison, isolat- ed PI or NNRTI resistance was much less common Declining nucleoside reverse among treated persons referred for clinical testing: 19 of 269 (7%) PI-resistant mutants and 17 of 69 (25%) of transcriptase inhibitor primary NNRTI-resistant mutants had no evidence of resis- resistance in San Francisco, tance to other classes. Among recently infected per- 2000–2002 sons, seven of nine viruses with isolated PI resistance were PR L90M mutants and 11 of 16 viruses with iso- lated NNRTI resistance were RT K103N mutants. RM Grant, T Liegler, G Spotts and FM Hecht CONCLUSIONS: The prevalence of primary drug Gladstone Institute of Virology and Immunology; University of resistance remains high in San Francisco, although California, San Francisco, San Francisco, Calif., USA transmission of viruses with three-class multidrug resistance is rare. Although NRTIs are prescribed near- BACKGROUND: Increasing prevalence of primary, or ly universally in combination regimens, we observe transmitted, drug resistance has been reported in San declining proportions of transmitted viruses with Francisco, other North American cities and the UK, NRTI resistance, and isolated PI or NNRTI resistance while decreasing trends in primary resistance have without NRTI resistance was common. Persons with been observed in other settings. Interpreting divergent NRTI resistance may have diminished infectiousness trends in primary drug resistance requires additional due to partial viral suppression, which is associated contemporaneous epidemiological information from with continued NRTI activity and reductions in the same geographical region. pro/pol replication capacity. In contrast, persons infected with RT K103N or PR L90M viral mutants 1 Jan 00– 1 July 01– Exact test may have relatively preserved infectiousness if NRTI 30 June 01 31 Dec 02 significance resistance is absent, as might occur with use of NRTI- n=91 n=89 sparing regimens arising due to partial adherence or to avoid toxicity. Any ART resistance 25 (27.4%) 22 (24.7%) Any NRTI resistance 19 (20.9%) 9 (10.1%) P=0.023 Any NNRTI resistance 12 (13.2%) 11 (12.4%) Any PI resistance 7 (7.7%) 12 (13.5%) P=0.088 Resistance to ≥2 classes 12 (13.2%) 9 (10.1%) Resistance to 3 classes 1 (1.2%) 1 (1.1%) METHODS: The Options project identifies persons in the San Francisco Bay Area with acute and early HIV-1 infection based on evolving serology. Drug resistance is assessed genotypically and phenotypically (TRUGENE, PhenoSense). Genotypic resistance was defined as any major protease inhibitor (PI) resistance-associated mutation and any nucleoside reverse transcriptase inhibitor (NRTI)- or non-NRTI (NNRTI)-associated mutation, including variants of T215. RT V118I, a common polymorphism, was not considered in this analysis. RESULTS: During an 18-month period between 1 July 2001 and 31 December 2002, 89 previously unreport- ed recently infected and drug-naive persons were iden- tified, and compared with 91 recently infected subjects identified during the previous 18 month period in the same geographical region. Among 314 genotypes from recently infected persons from 1996 to 2002, 9 of 25 (36%) of viruses with PI resistance and 16 of 28 (57%) of viruses with NNRTI resistance had no evidence of S134 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 121 have been received by Health Canada as a part of this programme. Of these samples, 1189 have been The Canadian HIV Strain and analysed for major mutations associated with drug resistance and 847 (71.2%) were successfully ampli- Drug Resistance Program – a fied for sequence analyses. Transmission of all three population-based effort to enhance classes of drugs and multidrug resistance has been HIV surveillance identified across Canada. Primary drug resistance has also been observed in both genders, across different age groups, ethnicities and exposure categories, in GC Jayaraman1, T Gleeson1, P Sandstrom1 and HIV-1 subtype A, B and C infections, and among CP Archibald1 for the Canadian HIV Strain and recent and older, prevalent HIV infections. Drug Resistance Surveillance Program2 CONCLUSIONS: The combination of epidemiological 1 Centre for Infectious Disease Prevention and Control, Health and laboratory information using a representative, Canada, Ottawa, Canada; and 2 BC Centre for Disease Control; population-based approach provides a unique oppor- Alberta Provincial Lab for Public Health and Departments of tunity to improve our understanding of the evolving Medicine of U Alberta & U Calgary; Saskatchewan Health; HIV epidemic in Canada and worldwide. The data Manitoba Health; Ontario Public Health Lab; Nova Scotia Public suggest that the prevalence of primary HIV drug resis- Health tance may be becoming more widespread in Canada. BACKGROUND: We describe here the Canadian HIV Strain and Drug Resistance Surveillance Program, which was initiated to characterize and monitor the genetic diversity of the HIV epidemic in Canada (including primary drug resistance). This programme is a the result of a collaborative effort between the Provinces and Territories, Health Canada, affected communities, public health authorities, primary care physicians and researchers. The programme encom- passes all HIV-positive cases diagnosed in participating provinces for whom diagnostic sera were available for analyses. METHODS: Through a series of consensus workshops the goals and objectives of the programme were defined, standard operating procedures for the collec- tion and shipment of archived diagnostic sera for HIV strain, drug resistance, and incidence testing were determined, and a core set of epidemiological variables to help interpret laboratory results were identified. An integrated, enhanced database was developed to ‘house’ laboratory and epidemiological information collected through the programme. Ongoing consulta- tions with technical advisory groups have helped to refine laboratory-testing algorithms to identify HIV strain, major mutations conferring drug resistance, and incident infections across Canada. Similarly, consulta- tions with key stakeholders have informed data analy- ses and enhanced the utility of surveillance data at all levels. RESULTS: A total of 2242 sera samples from all indi- viduals with available archived diagnostic sera, who were newly diagnosed across Canada between 1984 and 2001 and corresponding epidemiological data XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S135 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 122 non-NRTIs (NNRTIs), protease inhibitors (PIs)] before January 2001 represented 17% (n=20) of the cohort as Drug resistance prevalence declines opposed to 8% (n=3) of such patients after 2001 (P=0.2). Similarly, no significant decrease in DR preva- in recently infected subjects having lence for each drug class was observed [NRTIs (4 vs sex with men but not in those 2.6%), NNRTIs (11.0 vs 2.6%), PI (8.7 vs 2.6%) and using drug injections: results from two or three classes multi-DR (11 vs 3%)]. However, when risk factors were analysed a significant decrease the Montreal Primary HIV-Infection in DR prevalence was evidenced only among men hav- Cohort ing sex with men (MSM) (16 vs 0%, P=0.05), injecting drug users (IDU) (20 vs 9%, P=0.4) and heterosexuals JP Routy1, B Brenner2, D Rouleau3, R Thomas4, (8 vs 40%, P=0.12). Among patient characteristics B Trottier4, P Côté4, JG Baril4, C Tremblay3, only IDU predicts DR (OR=2.5 [95% CI=1.4–4.5], R LeBlanc1,4, M Edwardes1, N Machouf1,4, P=0.002) and this higher ratio may be explained in B Spira2, RP Sekaly5 and M Wainberg2 part by low drug compliance and by free drug assess in Canada. Likely, decrease in DR prevalence observed 1 McGill University Health Centre; 2 McGill University AIDS after 2001 is not explained by an abrupt change in the Centre, McGill University; 3 Centre Hospitalier de l’Université de proportion of chronically infected patients (n=2560) Montréal; 4 Private Medical Clinics; and 5 Department of who discontinued therapy or were drug-naive for anti- Microbiology and Immunology, University of Montreal, Montreal, retroviral in the two Montreal clinics. Canada CONCLUSION: Since 2001, DR prevalence signifi- BACKGROUND: Transmission of drug-resistant (DR) cantly decreases only in MSM. Conversely, IDU repre- strain is associated with treatment failure in newly sents a predictive risk factor for DR. Behaviours and infected individuals and represents an important ther- virological determinants associated with current apeutical challenge. Incidence of DR transmission in changes in DR prevalence among recently infected newly infected individuals may reflect the DR preva- individuals in Montreal have to be identified in source lence in chronically infected population. As antiretro- persons. viral drug discontinuation is associated with reversion to wild-type virus we should expect a decrease in DR among newly infected individuals after 2001, where therapeutic recommendations have been modified. The objectives of this study were: 1) to monitor the changes in DR prevalence in the Montreal Primary HIV- Infection Cohort; and 2) to assess if decrease in DR prevalence may be explained by postponing or inter- ruption of antiretroviral therapy among chronically infected patients followed in HIV clinics in Montreal. METHODS: From May 1996 to December 2002, 159 newly infected patients were studied for DR using genotyping (Visible Genetics TRUGENE) and pheno- typing analyses (Virologic PhenoSense HIV assay). DR prevalence was compared before and after 2001, when the therapeutic recommendations were introduced. Assessments of antiretroviral usage were performed in chronically infected patients followed in two down- town HIV clinics, together representing an estimated 50% of the total followed Montreal HIV population. RESULTS: Patient characteristics such as age, gender, risk factors and time elapsed between infection and study enrolment remained similar over time. Patients harbouring any mutation by genotypic analysis [either nucleoside reverse transcriptase inhibitors (NRTIs), S136 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 123 primary infection and 385 cells/mm3 (2–1280) and 4.5 log copies/ml (<1.70–6.94) for patients with chron- French National Sentinel Survey of ic infection, respectively. Among the 269 primary infected patients, 31 (12%) (95% CI: 8–15) had base- antiretroviral resistance in patients line genotypic resistance to at least one antiretroviral with HIV-1 primary infection and drug. For patients with resistance mutations to one in antiretroviral-naive chronically class of antiretroviral drug, the distribution was as fol- lows: to nucleoside reverse transcriptase inhibitors infected patients in 2001–2002 (NRTIs) in 15/269 [6%, (95% CI: 3–9)], to non- NRTIs (NNRTIs) in 5/269 [2%, (95% CI: 1–4)] and to ML Chaix1, D Descamps2, S Mouajjah3, protease inhibitors (PIs) in 6/269 [2%, (95% CI: 1–5)]. C Deveau4, P André5, J Cottalorda6, D Ingrand7, Five patients [2%, (95% CI: 1–4)] presented with virus J Izopet8, E Kohli9, B Masquelier10, mutations associated with multidrug resistance to two K Parmentier3, C Poggi11 , S Rogez12, or three classes of antiretroviral drugs. Among the 363 A Ruffault13, V Schneider14, A Schmück15, chronically infected patients with a median duration C Tamalet16, M Wirden17, C Rouzioux1, of known seropositivity of 6 months, the prevalence L Meyer4, F Brun-Vezinet2, D Costagliola3 and of virus harbouring at least one major PR substitu- the ANRS AC11 Resistance Group, Cohort tion or one RT mutation was estimated as 6.1% PRIMO, INTERPRIM and PRIMSTOP Study (95% CI: 3.6–8.5). For PR gene, prevalence of Groups patients with at least one major mutation was 0.8% (95% CI: 0.0–1.7). Prevalence of patients with muta- 1 CHU Necker EA-MRT 3620, Paris; 2 CHU Bichat-Claude tions associated to NRTI and NNRTI were 4.6% Bernard, Paris; 3 EMI 0214 INSERM; 4 INSERM U569, Kremlin- (95% CI: 2.4–6.7) and 1.4% (95% CI: 0.2–2.5), Bicêtre; 5 CHU Lyon; 6 CHU Nice; 7 A. Béclère Hospital, Clamart; respectively. Mutations associated with resistance to 8 CHU Toulouse; 9 CHU Dijon; 10 CHU Bordeaux; 11 Toulon one and two classes of ARV were seen in 20 (5.4%) Hospital; 12 CHU Limoges; 13 CHU Rennes; 14 Tenon Hospital, and 2 (0.6%) patients, respectively, while no patient Paris; 15 CHU Grenoble; 16 CHU Marseille; and 17 CHU La Pitié, had virus with mutations associated with resistance to Paris, France the three classes of ARV. Phylogenetic analysis based on 427 RT sequences revealed that 24% (95% CI: 17–31) OBJECTIVE: To survey the frequency of genotypic of primary infected patients (n=125) harboured non-B antiretroviral resistance and the spreading of non-B subtype strains compared to 19% in 1999–2000 and subtypes in primary infected patients (2001–2002) and that 33.2% of chronically infected patients (n=302) in antiretroviral-naive chronically infected patients harboured non-B subtypes isolates compared to 10% (2001), and to compare it to results of previous years. in 1998 (P<0.0001). METHODS: Plasma samples of 284 patients with CONCLUSION: In France during the years acute HIV-1 infection and of 379 naive patients with 2001–2002, there was no significant progression of the chronic infection were tested for genotypic resistance. frequency of resistant virus in these studied popula- Patients with acute infection were recruited within the tions compared to previous surveys while the preva- French network on HIV-1 primary infection national lence of non-B subtype was increasing. survey, the PRIMO Cohort study (n=125), the differ- ent laboratories of the AC11 Resistance group (n=128), the INTERPRIM study (n=20) and the PRIMSTOP study (n=11). Patients with chronic infec- tion were recruited in 23 AIDS care centres. Mutations were identified from the IAS-USA resistance panel (2002) and resistant virus was defined according to the ANRS algorithm (www.hivfrenchresistance.org). For chronic infection, weighted statistical analyses were used to derived estimates of the percentage of patients with resistance mutations. RESULTS: At inclusion, median CD4 cell counts and plasma HIV-1 RNA were 500 cells/mm3 (38–1516) and 5.3 log copies/ml (1.78–7.57) for patients with XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S137 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 124 ishing. Further, the similarity in prevalence between acute and non-acute drug-naive patients suggests that The UK HIV Drug Resistance testing in drug-naive patients/new diagnoses may be sufficient for monitoring transmission of drug resis- Database: development and use for tance. Future studies from this database will include national surveillance the resistance correlates of treatment response, the impact of HIV-1 subtypes on emergence of resistance D Pillay1 and H Green2, on behalf of the UK and the identification of novel resistance-associated Collaborative Group on HIV Drug Resistance mutations. 1 Department of Virology, UCL, London; and 2 MRC Clinical Trials Unit, UK Determining the clinical correlates of HIV drug resis- tance and estimation of resistance surveillance para- meters require large databases. We have therefore established a UK-wide non-proprietary collaboration of all virology laboratories and major clinical centres to pool resistance data and interpretation patterns, together with pre- and post-test virological, clinical and therapeutic information. This clinical interface includes patients, represented within the UK Collaborative HIV Cohort study. Resistance test data is generated by ‘in house’, VGI (Bayer) and ABI (Abbott) sequencing methodologies within UK diag- nostic laboratories, as well as by VIRCO. As of March 2003, 9808 test results obtained since 1996 from 6692 patients have been entered. key (pri- mary) resistance-associated mutations are based on IAS-USA (2002). Prevalence data has been ascribed to the time periods 1996–1998, 1999–2000 and 2001–2003. Provisional findings are as follows. Of all antretroviral-experienced patients, nucleoside ana- logue resistance (defined as at least one key mutation) remained stable (66, 71, 69%, respectively), whereas protease inhibitor (PI) resistance peaked in the 1999–2000 period (26, 32, 27%, respectively) and non-nucleoside reverse transcriptase inhibitor (NNRTI) resistance has continued to rise (20, 40, 48%). The time trends for PI and NNRTI resistance are statistically significant (P<0.001). Resistance in drug-naive individuals (n=1968), defined as any key mutation, appears to be levelling out after an initial rise (10, 16, 17%). When analysing the small subset reported as acute/recent infections (n=157), no clear differences with non-acute drug-naive infections were noted. We describe the first surveillance outputs from this UK database. Changes in the prevalence of PI and NNRTI resistance over time in treated patients mirror the changing use of these classes of drugs. In untreated patients we demonstrate continuing transmission of resistant viruses, the prevalence of which is not dimin- S138 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 125 Finally, non-NRTI (NNRTI) mutations included K103N (18.9%), L100I (10.8%) and V106IA (5.4%). Mutation M41L was more frequent in subtype B than HIV-1 drug resistance mutations in in C (P=0.038). Differences in mutation L210W did ARV-treated patients from southern not reach statistical significance between both sub- Brazil types, but when non-B subtypes were pooled together a trend was found towards subtype B (P=0.0554). Finally, mutation V82A/F/T/S of PR was more frequent EAJM Soares1, RP Santos2, JA Pellegrini2, in subtype F1 than in B (P=0.0209). E Sprinz2, A Tanuri1 and MA Soares1 CONCLUSIONS: Increasing evidence accumulates 1 Laboratório de Virologia Molecular, Universidade Federal do that HIV-1 subtypes may differ in response to current- Rio de Janeiro, Rio de Janeiro; and 2 Hospital de Clínicas de ly used ARV drugs. This study describes the epidemio- Porto Alegre, Porto Alegre, RS, Brazil logical and DRM profile of viruses infecting individu- als from Porto Alegre, the city in Brazil with the high- BACKGROUND: Subtype C of HIV-1 is currently the est prevalence of a non-B subtype characterized to most prevalent subtype in the world. Despite its impor- date. This population constitutes a valuable target for tance on HIV vaccine development, differences in sub- studying non-B subtype responses to ARV regimens, as type C antiretroviral (ARV)-targeted enzymes due to it represents one of a very few locales in the world polymorphic signatures of those viruses may impact on where large numbers of subtype C-infected individuals the long-term treatment of HIV-1 infections world- are treated. Differences in drug resistance acquisition wide. We wanted to characterize the HIV-1-infected profiles have been suggested in this study, adding evi- population from Porto Alegre, the southernmost state dence to recent investigations where those differences of Brazil, which has an endemic infection by subtype C have been initially shown. strains. Epidemiological parameters and HIV-1 sub- type prevalence in that city were measured. Drug resis- tance mutation (DRM) profiles for both protease (PR) and reverse transcriptase (RT) were also analysed. METHODS: Seventy-seven HIV-1-positive individuals followed at Hospital de Clínicas de Porto Alegre had their plasma collected and their viruses’ PR and RT genomic regions were PCR-amplified and sequenced for subtyping and genotyping analysis. Significance in differences of DRM among different subtypes was assessed by Fisher’s exact tests. RESULTS: Sixty-six percent of the individuals were female. Mean age was 39.1 years, median CD4 cell counts was 335.5 per mm3 and median HIV RNA viral load was 1101 copies/ml. Eighty percent of patients were under antiretroviral treatment. Fifty-eight patients had both viral genomic regions successfully amplified and available for analysis. HIV-1 subtype prevalences were 45% for subtype B, 40% for subtype C, 12% for subtype F1 and 2% for mosaic viruses. Stratification of samples by year of diagnosis showed that subtype C prevalence is increasing in the region. Among treated patients, the most prevalent primary PR inhibitor (PI)-associated mutations were V82A/F/T/S (20.9%), M46I (11.6%) and L90M (9.3%). Mutations associated with nucleoside RT inhibitors (NRTIs) with highest prevalences were M184V (27%), followed by M41L, L210W, and T215Y/F (18.9%) and V75I and K219Q/E (16.2%). XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S139 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 126 76/105, 72.4%) and this was significantly different compared to the mutations detected in naive subtype G L89I/V: a novel mutation selected samples (P<0.0001) as well as compared to the group of subtype G samples at failure with no major PI muta- by protease inhibitor therapy in tions present (P<0.0001). When comparing the muta- subtype G, but not in subtype tions in the naive subtype G group with the subtype G B-infected patients samples failing a non-PI based regimen or failing with no major PI mutation, no difference was detected. A Abecasis1, P Gomes1, I Derdelinckx2, AP CONCLUSIONS: While in subtype B no mutation is Carvalho1, I Diogo1, F Gonçalves1, J Cabanas1, selected at protease codon 89 under pressure of a PI- MC Lobo1, A-M Vandamme2 and R Camacho1 including regimen, on subtype G the prevalence of mutations (I/V) is high in patients failing a PI regimen, 1 Virology Laboratory, Hospital Egas Moniz, Lisbon, Portugal; and but only in the presence of major PI-mutations. These 2 Rega Institute for Medical Research, Katholieke Universiteit differences may be related to the different wild-type Leuven, Leuven, Belgium amino acid in both subtypes at this position (L in sub- type B compared to M in subtype G). These findings INTRODUCTION: Differences in genotypic resistance support the possibility that in non-B subtypes therapy pathways have been previously reported in HIV-1 non- failure may be related to mutations at codons not relat- B subtypes compared to B subtypes. So far these only ed to resistance in subtype B, with possible implica- concerned differences in prevalence of mutations at tions for the interpretation of genotypic resistance tests known resistance related positions. in this particular setting. OBJECTIVE: To compare the prevalence of new muta- tions at protease codon 89 in patients infected with subtypes B and G, both naive and at therapy failure. METHODS: Five-hundred-ninety-six sequences from 577 patients with known treatment status were identi- fied belonging to subtype B and G. Genetic sequence analysis was carried out using an automatic sequencer (ABI Prism 3100, Applied Biosystems, Foster City, Calif., USA). Resistance interpretation and subtyping was done using the Stanford HIV RT and Protease Sequence Database (http://hivdb.stanford.edu/hiv/). Sequences were stratified in groups according to subtype and treatment status. For subtype B: drug-naive (29) and failing a protease inhibitor (PI)-based regimen (226), for subtype G: drug-naive (52), failing a non-PI based regimen (125), failing a PI-based regimen with- out major PI resistance mutations (59) and failing a PI- based regimen with major PI resistance mutations (105). The presence of mutations methionine (M), isoleucine (I), valine (V) or leucine (L) on codon 89 in protease were evaluated and their prevalences com- pared between groups by Fisher Exact statistic. RESULTS: In drug-naive samples subtype B sequences displayed a L on protease codon 89, (28/29, 96.6%) in contrast to subtype G, where mainly M was present (50/52, 96.2%) (P<0.0001). Most subtype B samples at therapy failure (213/226, 94.2%) displayed L, with 13 samples showing M, I or V. Among subtype G sequences from patients failing a PI regimen with major PI mutations, M, I and V were detected (I/V in S140 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 127 ciated with drug resistance in clade B were recognized: 98G, 181C (RT) and 10I, 36I, 71V and 93L (PRO). Viral load decline and selection of The two individuals failing therapy showed a subse- quent genetic evolution with particular changes. drug resistance mutations in Patient A failed on stavudine (d4T)+lamivudine individuals with HIV-1 group O (3TC)+indinavir (RT: K70N, V75A, M184V; and viruses undergoing antiretroviral PRO: I84V) and patient B who was previously pre- treated and failed on d4T+3TC+lopinavir/rit (RT: therapy M41L, E44D, V75M, M184V, T215Y; and PRO: G48M, F53L, I54V, V82A, L90M). Changes at codons 1 1 2 2 C de Mendoza , B Rodés , M Rodgers , P Riley , 70 and 75 at the RT and 48 at the PRO genes appeared 1 1 1 V Jimenez , R Lopez-Brugada and V Soriano with different amino acids than usual in subtype. Interestingly, both subjects showed a shift C→Y at 1 Service of Infectious Diseases, Instituto de Salud Carlos III, codon 181. Phenotypic resistance to protease Madrid, Spain; and 2 Mayday University Hospital, Surrey, UK inhibitors (PIs) in these viruses was assessed using a phage-lambda-based recombinant virus assay (Rodés et al. JCM 2002), and results confirmed their expected BACKGROUND: Despite increasing recognition of loss of PI susceptibility. HIV-1 non-B infections worldwide, there is limited information on viral response and selection of drug CONCLUSIONS: HIV-1 group O-infected subjects resistance patterns to antiretrovirals in this population. develop drug resistance mutations at the same posi- HIV-1 group O strains seem to be susceptible to nucle- tions than subtype B. However, natural polymor- oside analogues and protease inhibitors, but they are phisms at either RT or PRO may drive selection of dis- naturally resistant to non-nucleoside inhibitors. tinct pathways in HIV-1 group O viruses in patients Herein, we present data from drug resistance evolution failing antiretroviral therapy. in HIV-1 group O-infected patients failing antiretrovi- ral therapy. METHODS: Plasma viraemia was measured using RT- PCR (LCx, Abbott) in longitudinal samples collected from six individuals known to be infected with HIV-1 group O, at baseline and throughout the course of antiretroviral therapy. Taking as reference clade B genetic sequences, primary and accessory drug resis- tance mutations were examined, following the latest IAS resistance guidelines. RESULTS: All viruses belonged to HIV-1 group O sub- type I (Brennan et al. AIDS Res Human Retroviruses 2002). Overall, all patients but one initiated treatment with two nucleoside inhibitors and one protease inhibitor. The other one had been treated during 5 months with dual nucleoside therapy before highly active antiretroviral therapy (HAART). Mean baseline viral load and CD4 cell counts were 31473 (6530–74132) HIV RNA copies/ml and 82 (13–170) cells/µl, respec- tively. Four patients reached undetectable viral load (<200 copies/ml) 3 months after beginning therapy and all but one have remained aviraemic since then (mean follow-up 12 months). The other patient admitted poor treatment adherence when he presented viral rebound. The remaining two subjects showed an initial viral drop but did not reach undetectable plasma viraemia, and they rebounded thereafter to baseline levels. At baseline, several changes known to be asso- XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S141 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 128 Synonomous genetic changes within subtype F HIV-1 may influence mutational routes to drug resistance A Tanuri1, MA Soares1, AT Dumans1, S Hue2 and D Pillay2 1 Department of Genetics, Federal University of Rio de Janeiro, Brazil; and 2 Department of Virology, University College London, London, UK The well recognized drug resistance-associated amino acid changes within reverse transcriptase and protease usually require one or two nucleotide changes from wild-type virus. We postulated that synonymous genetic differences at these positions between HIV-1 subtypes may underpin subtype-specific routes to resistance. This has recently been shown for the V106M non-nucleoside reverse transcriptase inhibitor (NNRTI) mutation in reverse transcriptase associated with subtype C virus. Scrutiny of consensus sequences for all the major virus subtypes identified two potentially important positions within subtype F. Firstly, the RT position 210 leucine (L) was encoded by CTG, rather than TTG in subtype B. Thus, generation of W(TGG) would require two steps (one transversion and one transition) rather than one (transversion). Secondly, RT position 151 gluta- mine (Q) was encoded by CAA, rather than CAG in most other subtypes, such that generation of the mul- tiresistance mutation Q151M would require three nucleotide changes in subtype F. Assessment of sequences from 23 drug-naive, subtype F-infected patients confirmed these synonymous differences. In order to test if a higher genetic barrier for the L210W mutation would lead to differences in rate of emergence, we calculated the frequency of this muta- tion among 131 subtype B and 30 subtype F-infected patients receiving zidovudine or stavudine containing regimens. The prevalence of L210W was higher in sub- type B viruses (31%) than subtype F (10%) (P=0.012). The prevalence of Q151M was too low to assess sub- type specific differences. We provide preliminary evidence that the differential genetic barrier between the RT of subtypes B and F may lead to altered rates of emergence of resistance- associated mutations at position 210. These data require confirmation in larger patient datasets. S142 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 129 the CI group, the prevalence of these mutations was 4.6, 4, 11.4, 7.1, 6, 14.7 and 21%, respectively. Thus, Decreased rates of transmission of the relative risks of transmission of NNRTI mutations and TAMs (PHI/CI of 0.78 and 0.75) were markedly drug-resistant HIV-1 strains greater than comparable rates of transmission of virus- containing the M184V mutation in es harbouring M184V with or without TAMs and/or reverse transcriptase NNRTI mutations (PHI/CI ranging 0.04–0.1) (P<0.001, ϕ contingency test, P<0.05, post-hoc Fisher D Turner1, BG Brenner1, J-P Routy2, D Moisi1 exact test). The specific transmission of viruses har- and MA Wainberg1 bouring mutations at codon 215 and K103N was also significantly higher than that observed for M184V. Viraemia observed in CI patients (n=552) harbouring 1 McGill University AIDS Centre; and 2 McGill University Health NNRTI mutations and TAMs alone or in combination Centre, Montreal, Quebec, Canada (median values of 34561–38000 copies/ml) were sig- nificantly higher than those observed for infections BACKGROUND: Although elevated viraemia is harboring M184V drug-resistant variants (median lev- known to increase the risk of new HIV-1 infection, lit- els of 7895–11000 copies/ml) (P<0.001 ANOVA tle is known about the potential benefit of combina- analysis). Thus, reduced transmission risk of M184V tion therapies and the influence of emergent drug resis- drug-resistant viruses could also be related to the effect tance in reducing rates of viral transmission. This of this mutation on viral replicative capacity and lower study evaluated the differential transmission of HIV-1 levels of viraemia. strains harboring mutations in reverse transcriptase (RT) in newly infected persons compared with preva- CONCLUSION: These findings suggest that reduc- lence in the potential transmitter population. tions in viraemia and viral fitness of M184V-contain- ing HIV-1 strains may significantly impact on rates of METHODS: We compared drug resistance profiles in viral transmission. patients enrolled in the Montreal Primary HIV-1 Infection (PHI) cohort (n=163) compared to the poten- tial transmitter population of chronically infected (CI) individuals having viraemia >1000 copies/ml (n=552) receiving lamivudine (3TC), stavudine (d4T) or zidivu- dine (ZDV), or non-nucleoside RT inhibitors (NNRTIs). Genotypic analysis of plasma viral RNA revealed a differential prevalence of the 3TC-associat- ed mutation M184V, thymidine analogue mutations (TAMs) M41L, D67N, K70R, L210W, T215C/D/F/S/Y and K219Q, and NNRTI-associated mutations L100I, K101E. K103N, V106A, Y181C, Y188L, G190A, P225H, M230L and P236L in both PHI and CI popu- lations. Statistical analysis of the independent and interactive effects of M184V, TAMs and NNRTIs were evaluated. The transmission of resistant variants was related to observed viraemia in the CI population receiving different combination regimens. RESULTS: Of a total of 163 PHI patients followed, transmission of RT drug resistance was observed in 11.7% (19/163) with a prevalence of M184V, TAMs and NNRTI mutations of 3, 6.7 and 7.4%, respective- ly. Independent analysis of different variables in the PHI group revealed that mutations in the following categories, that is, NNRTIs, TAMs, NNRTIs/TAMs, M184V, NNRTIs/M184V, TAMs/M184V and NNRTIs/TAMs/M184V were present at 3.6, 3, 1.8, 0.6, 0.6, 0.6 and 1.2%, respectively. In contrast, for XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S143 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 130 estimated ratio ‘potential transmitter/seroconverter’ for drug resistance genotypes was 4.8 for 41L, 5.2 for Evidence for a different 215Y/F (including revertants) and 33.8 for 184V among NRTI. For NNRTI was 7.2 for 181C. Finally, transmission efficiency of viruses for PIs was 8.9 for 46L, 16 for 90L and 21.2 for 82S. with distinct drug-resistant Transmission of some genotypes was not documented genotypes despite their presence among pretreated potential transmitters: 70R (26.8%), 74V (13%), 103N (36.8%) at the RT; and 84I (14%) and 30N (5%) at the pro- C de Mendoza1, C Rodriguez2, A Corral1, tease. No evidence of clusters or of non-B subtype O Gallego1, J del Romero2 and V Soriano1 infections were found among our population of recent seroconverters. 1 Instituto de Salud Carlos III; and 2 Centro Sandoval, Madrid, Spain CONCLUSIONS: Transmission of drug-resistant viruses has rebounded in 2002. Viruses carrying NRTI- BACKGROUND: Viruses carrying drug resistance resistant mutations are more frequently transmitted mutations may have an impaired replication capacity. than HIV with resistance to NNRTI or PI. Moreover, The impact of different resistant genotypes on trans- some mutations (such as, 41L and 215Y/F for NRTI; mission efficiency is not well known. 181C for NNRTI; and 46L for PI) may be more effi- ciently transmitted than others (184V and 74V for PATIENTS AND METHODS: Drug resistance profiles NRTI; 103N for NNRTI; and 30N and 84I for PI). were compared in 74 recent seroconverters attending in Madrid from 1997 to 2002 in a case-control study (ratio 1:5) and potential transmitters of drug-resistant viruses. This group was represented by subjects with viral load >1000 HIV RNA copies/ml and prior expo- sure to antiretroviral drugs. Controls for each serocon- verter were matched for transmission risk category (homosexual, intravenous drug use, heterosexual con- tact) and period (1997–1999, 2000–2001 and 2002). Only primary drug resistance mutations following the latest IAS rules were considered. RESULTS: The overall rate of drug resistance muta- tions in recent seroconverters in Madrid was 18.9% (14/74). It was 26.6% during 1997–1999, declined to 4.3% during 2000–2001 and rebounded to 22.7% in 2002. According to drug classes, the resistance rate was: 16.2% for nucleoside reverse transcriptase inhibitors (NRTIs), 2.7% for non-NRTIs (NNRTIs) and 4.1% for protease inhibitors (PIs). Drug resistance genotypes recorded were: 215Y/F (n=2), 215 revertant forms (n=5), 41L (n=6) and 184V (n=1) for NRTI; 181C (n=2) for NNRTI, and 46L (n=2), 82S (n=1) and 90L (n=2) for PI. The rate of drug-resistant mutations in pretreated potential transmitters was 88.8% for NRTI, 49.7% for NNRTI and 60.6% for PI. Therefore, the estimated ratio ‘potential transmitter/ seroconverter’ for drug resistance genotypes was 5.5 for NRTI, 18.4 for NNRTI and 14.8 for PI (P<0.05). When considering specific mutations, their rate in potential pretreated transmitters were: 215Y/F (49.4%), 41L (39.1%), 184V (45.6%) and 181C (19.6%) at the RT, and 46I/L (24.1%), 82A/T/S (28.6%) and 90L (43.2%) at the protease. Thus, the S144 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 131 upon our mathematical model. We will discuss the implications of HSV-2-induced HIV epidemic dynamic Modelling the sexual synergy for the expected transmission of drug-resistant strains between HSV-2 and HIV-1 of HIV. epidemics: implications for CONCLUSION: HSV-2 epidemics can substantially transmitted HIV resistance increase the speed and severity of HIV epidemics, including the transmission of drug-resistant strains of HIV. SM Blower1, T Porco2, F Wang3, A Wald4 and L Corey5 1 UCLA AIDS Institute & Department of Biomathematics, David Geffen School of Medicine at UCLA, Los Angeles, Calif.; 2 San Francisco Department of Public Health, San Francisco, Calif.; 3 Department of Computer Science UCLA, Los Angeles, Calif.; and 4 Fred Hutchinson Center, University of Washington, Seattle, Wash., USA BACKGROUND: HSV-2 is the most prevalent sexual- ly transmitted disease worldwide. In many risk groups in many geographic locations the HSV-2 and the HIV epidemics show substantial overlap. Many laboratory and epidemiological studies have shown that HSV-2- infected individuals are more susceptible to HIV infec- tion. Results from other studies have suggested that individuals that are co-infected with both HSV-2 and HIV may be more capable of transmitting HIV than individuals infected with only HIV. However, it is not clear what effects these biological changes in suscepti- bility and infectiousness will cause at the epidemic level. Here we have developed a mathematical model of both epidemics; our model simultaneously tracks individuals infected with only HSV-2, only HIV and co-infected with both viruses. We analyse this model in order to determine the impact of HSV-2 epidemics on HIV transmission dynamics. METHODS: We analyse our mathematical model: (i) to predict and (ii) to quantify how HSV-2 epidemics alter the speed and severity of HIV epidemics in both the short-term and the long-term. We quantify the impact of HSV-2 epidemics on: (i) the doubling time of the HIV epidemic; (ii) the increase in the basic repro- duction number (R ) of HIV; and (iii) the population 0 attributable fraction (PAF; that is, the number of HIV cases that are solely due to the HSV-2 epidemic). RESULTS: Our analyses reveal that HSV-2 epidemics significantly increase the speed and the severity of the HIV epidemic in both the short-term and the long- term. The degree of increase in both epidemic severity and speed is complex and changes over time, as it is a multifactorial process. Hence, we will show graphical temporal relationships that we have derived based XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S145 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 132 itive predictive value (PPV) were calculated and com- pared with the validation sample. A systematic approach that RESULTS: Included were 2655 subtype B protease identifies 11 new mutations as sequences (including 1463 treated with at least one PI indicators of transmission of and 1192 who had not received PI). 1751 PI sequences resistance (973 treated/778 untreated) were selected for model development. The following mutations in PI were iden- tified as MIRTs (12 mutations that are not part of the DAMC van de Vijver1, R Schuurman1, IAS guidelines are underlined): K20I/T, A22V, L23I, C Nielsen2, M Salminen3, J Albert4, L24I, D30N, V32I, L33F, K45R, M46I/L, G48V, I50V, A-M Vandamme5, S Coughlan6, R Shafer7 F53L, I54V, K55R, A71I, G73C, L76V, V82A/F/T, and CAB Boucher1 on behalf of the I84V, I85V, N88S, L90M and C95F. In both sample SPREAD-network sets, the sensitivity for the detection of MIRTs was 70%, the specificity was 99% and the PPV was 98%. 1 Eijkman Winkler Institute for Medical Microbiology, Inclusion of combinations of mutations increased the Department of Virology, University Medical Centre Utrecht, the sensitivity to 71% in both samples. The specificity and Netherlands; 2 Retrovirus Laboratory, Department of Virology, the PPV, however, differed between both sets (98% in Statens Serum Institut, Copenhagen, Denmark; 3 National Public the development sample and 96% in the validation Health Institute, HIV Laboratory and Department of Infectious sample; P<0.001, and the PPV 99 and 96%, respec- Disease Epidemiology, Helsinki, Finland; 4 Division of Clinical tively; P<0.001). The major mutations as defined by Virology, Department of Immunology, Microbiology and the IAS guidelines, had a sensitivity of 66%, a speci- Pathology (IMPI), Karolinska Institute, Huddinge University ficity of 99% and a positive predictive value of 99%. Hospital, Stockholm, Sweden; 5 Rega Institute for Medical Inclusion of major and minor mutations as defined by Research, Katholieke Universiteit Leuven, Leuven, Belgium; 6 the IAS guidelines increased the sensitivity to 85%, but National Virus Reference Laboratory, University College Dublin, lead to a specificity of 54% and a positive predictive Dublin, Ireland; and 7 Division of Infectious Diseases, Stanford value of 70%. A similar approach has been developed University, Stanford, Calif., USA for reverse transcriptase. BACKGROUND: Monitoring programmes have been CONCLUSION: A systematic approach was devel- established to study the transmission of antiretroviral oped to identify protease mutations associated with PI drug resistance in treatment-naive HIV patients. These treatment. The MIRTs approach demonstrated a better programmes, however, do not use approaches that sys- sensitivity compared to the major mutations that are tematically define markers for drug resistance. The aim defined by the IAS. MIRTs can be used to assess if a of this study is to develop a systematic approach to virus has been in contact with antiretroviral drugs in a detect mutations that mark transmission of resistance. previous patient. This approach can be applied for Therefore, the protease genes from protease inhibitor epidemiological studies on transmission of resistance, (PI)-treated and -untreated individuals were compared, but do not aim to be applied for clinical determination with the aim of identifying mutations that correlate of resistance. The MIRTs approach will be validated in with PI exposure and that can be used in epidemiolog- a dataset of almost 1500 newly diagnosed therapy- ical studies as molecular indicators of resistance trans- naive patients. mission (MIRTs). METHODS: MIRTs were defined as mutations, and combinations thereof, with no natural variation (<0.5% in untreated sequences) that were statistically significantly associated with treatment. Statistical sig- nificance was assessed using χ2-statistics and the Benjamini Hochberg to correct for multiple hypothesis testing. Protease HIV subtype B sequences from treat- ed and untreated individuals were obtained from the Stanford HIV reverse transcriptase and PI database. Two-thirds of the sequences were randomly selected for model development. The remaining sequences were used for validation. The sensitivity, specificity and pos- S146 SESSION 6 Clinical Implications of Resistance Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 133 Baseline genotypes in two of the patients in this group had multiple mutations associated with d4T resistance Short duration, single drug (M41L, D67N, V118I, L210W, T215Y). HIV-1 RNA levels were still increasing at the end of the discontinu- discontinuation to assess the ation period, so the full impact of d4T discontinuation activity of individual drugs in could not be determined. HIV-1 RNA levels declined patients failing antiretroviral promptly to baseline values in the two patients restart- ing d4T. No significant changes in CD4 T cell counts therapy were observed during the trial and no adverse events related to drug interruption occurred. F Maldarelli1, S Palmer1, M Kearney1, J Falloon3, RT Davey3, A Powers3, S Vogel3, CONCLUSIONS: Persistent antiviral activity of d4T, A Pau4, R Dewar2, JA Metcalf2, J Mellors5 and despite the presence of resistance-conferring muta- J Coffin1 tions, was revealed by interruption of drug. By con- trast, EFV showed no persistent activity in patients 1 HIV Drug Resistance Program, NCI, NIH; 2 NIAID/CCMD Clinic, with NNRTI resistance mutations. The single drug dis- NIH; 3 Laboratory of Immunoregulation, NIAID, NIH; 4 CC continuation strategy may be a useful approach to Pharmacy, NIH; and 5 Division of Infectious Diseases, University assess persistent antiretroviral activity in patients fail- of Pittsburgh, Pittsburgh, Pa., USA ing therapy. BACKGROUND: The relationship of in vivo drug activity, and HIV-1 genotype and phenotype is incom- pletely understood. For non-nucleoside reverse tran- scriptase inhibitors (NNRTIs), loss of therapeutic activity in vivo is strongly associated with specific mutations. In contrast, for certain NRTIs, such as stavudine (d4T), the effect of specific mutations on in vivo activity is poorly defined. To investigate the extent to which resistance mutations affect antiviral activity in vivo, we discontinued single drugs for short dura- tions in patients failing combination antiretroviral therapy. METHODS: Patients with HIV RNA greater than 5000 copies/ml plasma, CD4 T cell count greater than 50 cells/µl and adherent to a d4T- or efavirenz (EFV)- containing multidrug regimen were enrolled. Following a 10-day baseline sampling period, d4T or EFV was interrupted, maintaining the remainder of the regimen. Longitudinal samples were obtained during the 14 day (d4T) or 21 day (EFV) interruption period. Following drug interruption, patients had the option of restarting the discontinued antiretroviral with addi- tional sampling. Analysis included HIV-1 RNA mea- surements by bDNA assay and composite genotype determination. RESULTS: Of the five patients enrolled, three discon- tinued d4T and two discontinued EFV. There were no significant increases in viraemia during the 21-day observation period following EFV interruption. Baseline genotypes in both patients revealed NNRTI resistance mutations (K103N and Y188L). By con- trast, interruption of d4T for 14 days resulted in sig- nificant increases in viraemia in all three patients. XII International Workshop on HIV Drug Resistance & Treatment Strategies S149 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 134 Standard genotyping did not detect NNRTI mutations in baseline samples from 50 of 216 (23%) NNRTI- Low frequency non-nucleoside experienced patients. Virological outcome in this group, however, was not better than in the group reverse transcriptase inhibitor (n=166) with baseline NNRTI mutations. By contrast, (NNRTI)-resistant variants among all patients with negative baseline genotypes for contribute to failure of NNRTI mutations, virological outcome was signifi- cantly better at weeks 24 (P=0.015) and 48 (P=0.02) in efavirenz-containing regimens in NNRTI-naive patients (n=237) compared with NNRTI-experienced patients with NNRTI-experienced patients (n=50). These findings negative standard genotypes for suggested that standard genotyping may not have ade- quately detected NNRTI-resistant variants. Baseline NNRTI mutations plasma from a random sample of 10 NNRTI-experi- enced and eight NNRTI-naive patients who had viro- J Mellors1, S Palmer2, D Nissley3, M Kearney2, logical failure despite a negative baseline genotype for E Halvas1, C Bixby1, L Demeter4, S Eshleman5, NNRTI mutations were tested for minor NNRTI-resis- K Bennett6, S Hart7, F Vaida6, M Wantman6, tant variants. Variants encoding NNRTI-resistance J Coffin2 and S Hammer8 for the ACTG 398 mutations were identified by SGS in six of 10 NNRTI- Study Team experienced patients with the following frequencies per positive patient: 181C and 190A (5 of 15 sequences); 1 University of Pittsburgh, Pittsburgh, Pa.; 2 Drug Resistance 181C (3 of 19); 181C (3 of 22); 108I (2 of 35); 103N Program, NCI, Frederick, Md.; 3 SAIC, Frederick, Md.; 4 University (1 of 33); and 103N (1 of 34). By comparison, of Rochester, Rochester, NY; 5 Johns Hopkins University, NNRTI-resistant variants were found in only one of Baltimore, Md.; 6 Harvard School of Public Health, Boston, eight NNRTI-naive patients: 100I (1 of 33 sequences). Mass.; 7 Frontier Science, Amherst, NY; and 8 Columbia The Ty1/HIV-1 RT assay detected efavirenz-resistant University, New York, NY, USA yeast colonies in 8 of 10 NNRTI-experienced patients with the following frequencies: 10.9, 6.7, 6.4, 3.3, 2.0, BACKGROUND: The role of minor (low frequency) 1.6, 1.3 and 0.8%. In NNRTI-naive patents, resistant drug-resistant variants in failure of antiretroviral ther- colonies were found in two of eight patients with fre- apy is not defined. In ACTG 398, 212 non-nucleoside quencies of 0.6 and 0.3%. Phylogenetic analysis reverse transcriptase inhibitor (NNRTI)-experienced showed close clustering of baseline NNRTI-resistant and 269 NNRTI-naive patients were randomized to variants identified by SGS with the genotype at viro- efavirenz, abacavir, adefovir and amprenavir with a logical failure in five of six NNRTI-experienced second protease inhibitor (PI) or placebo. This study patients. In the NNRTI-naive patient, the L100I provided the opportunity to examine relations between mutant was not evident at failure and did not cluster NNRTI experience, baseline NNRTI resistance, viro- with the failure genotype. logical response and the emergence of efavirenz resis- tance. CONCLUSIONS: Prior NNRTI experience selects minor NNRTI-resistant variants that are often missed METHODS: Genotypes of baseline plasma were by standard genotyping and can lead to failure of obtained in 452 of 481 patients by a standard method efavirenz-based regimens. [ViroSeq version 2.0 kit (Applied Biosystems)]. Mutations were classified according to the IAS-USA table. Minor NNRTI-resistant variants were sought with two methods: single genome RT-PCR and sequenc- ing (SGS) and a yeast-based chimeric Ty1/HIV-1 RT retrotransposon system that measures the frequency of efavirenz resistance. Phylogenetic analyses were per- formed with the neighbour joining method (PHYLIP v3.573c). RESULTS: Virological failure (confirmed HIV RNA >200 copies/ml) was associated with NNRTI experi- ence (P<0.001), baseline NNRTI mutations (P<0.001), and development of efavirenz resistance (P<0.001). S150 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 135 range 0.9–2.2), increased zidovudine susceptibility (0.5-fold), no change for d4T (0.9-fold) and low-level Characterization of resistance changes for didanosine and abacavir (4/8 and 5/8 below clinical cut-offs, respectively). The mean RC of mutation patterns emerging over HIV from these patients was 45% of wild-type virus RC and their mean plasma HIV RNA remained 0.9 log 2 years during first-line 10 antiretroviral treatment with below baseline, consistent with a replication defect for HIV with K65R. Among these K65R viruses, A62V tenofovir DF or stavudine in and S68G RT mutations later developed in two and combination with lamivudine and three patients, respectively. Site-directed recombinant efavirenz viruses K65R+A62V and K65R+S68G showed no changes in drug susceptibility compared to K65R alone. Therefore, A62V and S68G may represent par- MD Miller1, NA Margot1, DJ McColl1, T Wrin2, tial compensatory mutations as RCs remained below DF Coakley1 and AK Cheng1 wild-type. All eight patients began new regimens with protease inhibitors and nucleoside RT inhibitors; two 1 Gilead Sciences, Inc., Foster City, Calif.; and 2 ViroLogic, Inc., S. patients remained on TDF. 5/8 achieved <50 copies/ml San Francisco, Calif., USA (median follow-up 76 weeks); two were without fol- low-up and one was non-adherent. BACKGROUND: The K65R mutation in HIV reverse transcriptase (RT) was selected in vitro by tenofovir CONCLUSIONS: Among treatment-naive patients, and has developed at low incidence (3%) in treatment- virological failure occurred similarly for patients on experienced patients adding tenofovir DF (TDF). TDF+3TC+EFV or d4T+3TC+EFV, and was associat- Mutant HIV with K65R exhibits decreased replication ed most commonly with high-level EFV and/or 3TC capacity (RC) in vitro. resistance. The K65R mutation occurred in 2.7% of TDF-treated patients and was associated with low- METHODS: Study 903 is a 3 year, randomized, dou- level tenofovir susceptibility changes, decreased HIV ble-blind study in treatment-naive patients. Patients replication capacity, and reduced viral load from base- received either TDF (n=299) or stavudine (d4T) line. Successful virological outcomes were achieved for (n=301) with lamivudine (3TC) and efavirenz (EFV). patients who failed with K65R upon initiation of sec- HIV from patients that experienced virological failure ond-line therapy. (>400 copies/ml of HIV RNA at week 96 or early dis- continuation) was analysed genotypically (Virco) and phenotypically for RC and drug susceptibilities (ViroLogic). RESULTS: At week 96, 12.3% of patients were viro- logical failures (36 TDF, 38 d4T; P=0.90). Resistance to EFV or 3TC (M184V) occurred most frequently (6.5 and 4.5%, respectively) with no significant differ- ence between arms. 27/39 patients with EFV resistance (69%) developed K103N with no difference between arms. Non-K103N patterns included V106M+V179D (n=3), V179I+G190S, G190Q, G190E and V179G+Y181C+G190E; all associated with >50-fold EFV resistance and subtype B. K65R developed in eight TDF patients versus two d4T patients (2.7 vs 0.6%, P=0.06). For one K65R developer taking d4T, K70R was present at baseline and K65R+K70R result- ed in sixfold d4T resistance. As M184V was also pre- sent, 3TC selection for K65R is unlikely. EFV resis- tance always preceded or accompanied K65R, either with (n=6) or without (n=4) M184V. HIV isolates from TDF-treated patients with K65R (n=8) showed low- level changes in tenofovir susceptibility (mean 1.2-fold, XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S151 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 136 occurred with 65R, a strong negative relation was noted with AZT resistance mutations in both databas- K65R: a multi-nucleoside resistance es. This data suggests that 65R is antagonistic to AZT resistance mutations. Indeed, the addition of 65R to mutation of low but increasing two different AZT-resistant clones (41L/210W/215Y frequency and 67N/70R/215F/219Q) reduced AZT resistance >10-fold (>30-fold to ~threefold). Similarly, the addi- U Parikh1, D Koontz1, J Hammond2, tion of 65R to AZT-resistant RT reduced primer L Bacheler3, R Schinazi4, P Meyer5, W Scott5 unblocking activity 10-fold to wild-type levels. and J Mellors1 CONCLUSIONS: 65R is a multi-NRTI resistance 1 University of Pittsburgh, Pittsburgh, Pa., USA; 2 Agouron mutation, reducing susceptibility to all D-, L- and Pharmaceuticals, La Jolla, Calif.; 3 VircoLab, Inc. Durham, NC; 4 acyclic NRTIs tested except those containing a 3′-azido Emory University, Atlanta, Ga.; and 5 University of Miami, Miami, moiety (AZT and AZA). 65R appears to be counter- Fla., USA selected by AZT because 65R antagonizes ATP-catal- ysed primer unblocking, which is the primary mecha- BACKGROUND AND OBJECTIVES: The K65R nism of AZT resistance. 65R can be an important mutation in the fingers domain of HIV-1 reverse tran- mechanism of resistance to NRTIs for which primer scriptase (RT) is commonly selected in vitro by D- unblocking is not required. NRTIs that can select 65R nucleosides but is paradoxically rare in vivo, although may need to be given with AZT to prevent 65R from recent reports suggest that its frequency may be emerging. increasing. In the GS-903 trial, 24% of virological fail- ures in the tenofovir/lamivudine(3TC)/efavirenz arm developed 65R (Miller et al., 6th Intl Congress on Drug Therapy in HIV Infection). The effect of 65R on susceptibility and response to approved and investiga- tional nucleoside RT inhibitors (NRTI) is uncertain. To gain further insights about 65R, we examined recent trends in the frequency of 65R, determined the resis- tance profile of 65R alone and in combination with zidovudine (AZT) mutations, and analysed the effect of 65R on the primer unblocking activity of RT. METHODS: We searched the Virco and Stanford data- bases for the frequency of 65R and its association with other NRTI mutations. We examined the effect of 65R alone and with different AZT resistance mutations on HIV-1 susceptibility to a large panel (n=36) of D- , L- and acyclic NRTIs using a single cycle replication assay (P4/R5 cells) and/or a multiple cycle assay (MT-2, p24). The primer unblocking activity of RTs was deter- mined by measuring the formation of [32P]Ap ddA 4 from ATP-catalysed removal of ddAMP from ddAMP- terminated primer. RESULTS: Among the >60000 samples submitted to Virco for testing from 1998 to 2003 that contained any NRTI mutation, the frequency of 65R increased from 0.8% in 1998 to 2.1% in 2002. Among samples received in 2003, 3.8% had 65R. A molecular clone of HIV-1 encoding 65R showed 2.5-fold to >>10-fold reduced susceptibility to all D-, L- and acyclic NRTIs tested except those containing a 3′-azido moiety (AZT and AZA), which showed wild-type susceptibility. In examining the frequency of other NRTI mutations that S152 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 137 2-year period. Although not associated with a specific drug, it is more common in those receiving concomi- Which nucleoside and nucleotide tant TDF and DDI, and is less common in those on thymidine analogues. In individuals failing ABC, DDI, backbone combinations select for TDF there is a high prevalence of the K65R, which the K65R mutation in HIV-1 may be due to all components of this regimen selecting reverse transcriptase? for the development of this mutation. A Winston, A Pozniak, B Gazzard and M Nelson Chelsea and Westminster Hospital, London, UK BACKGROUND: The K65R mutation has been reported to be associated with the use of abacavir (ABC), didanosine (DDI) and tenofovir (TDF). With the advent of TDF we have assessed the change in prevalence of the K65R mutation and its association with TDF, DDI and ABC therapy. METHODS: In Chelsea and Westminster Hospital individuals failing therapy have a resistance test. Genotypes were analysed on individuals failing nucle- oside and nucleotide containing regimens from October 2000 to October 2002 (time 2). This was compared to genotypes pre-October 2000 (time 1) in TDF-naive individuals. RESULTS: The K65R was identified in viruses from 28 of 705 individuals having resistance tests for virologi- cal failure during time 2 (prevalence 4.0%). The preva- lence of this mutation during time 1 was significantly lower at 1.7% (P=0.004). When adjusting for the use of other drugs, neither ABC, DDI or TDF alone was associated with the development of the K65R muta- tion. However certain combinations were associated with the mutation. The use of TDF and DDI was asso- ciated with the development of the K65R variant (14% of isolates with the K65R were on dual use of TDF, DDI versus 3.8% of isolates, which did not develop the mutation, P=0.007). Current thymidine analogue use protected from the development of the mutation, those on thymidine analogues were 76% less likely to devel- op the mutation (odds ratio 0.24, 95% CI: 0.1–0.54). The combination of ABC, DDI and TDF was associat- ed with the mutation. 32% of individuals who devel- oped the K65R were on this combination, those not on this combination were 96% less likely to develop the K65R (odds ratio 0.04, 95% CI: 0.01–0.11). A total of 22 individuals receiving this combination had a resis- tance test. The K65R developed in nine isolates (41%). CONCLUSION: The prevalence of the K65R muta- tion remains low but has significantly increased over a XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S153 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 138 sis we considered that TAM resistance has occurred in patients who were lost to follow-up and in those who Thymidine analogue mutations died. TAM resistance occurrence was still more fre- quent in patients who received an AZT-based regimen emergence in antiretroviral-naive (50%) than in those who received a d4T-based regimen HIV-1 patients on triple therapy (26%) (P<0.01). including either zidovudine or CONCLUSIONS: Results of this observational study stavudine suggest that AZT-based triple therapy is associated more frequently with the emergence of TAMs when L Bocket, Y Yazdanpanah, F Ajana, Y Gerard, compared to d4T-based triple therapy. These results N Viget, A Goffard, I Alcaraz, P Wattre and may be of benefit for an optimal sequencing of AZT or Y Mouton d4T in antiretroviral therapy for treatment-naive HIV- infected patients. Hospital and University Center, Lillie, France BACKGROUND: To assess the incidence of thymidine analogue mutations (TAMs) occurrence on stavudine (d4T)- and zidovudine (AZT)-based triple therapy in HIV-1-infected antiretroviral treatment-naive patients. METHODS: Four-hundred-and-five HIV-1-infected antiretroviral treatment-naive patients from a French clinical cohort who started on a AZT- or d4T-contain- ing regimen from January, 1996 to December, 2001, were enrolled in this study. The incidence of TAM occurrence was estimated when the first-line antiretro- viral therapy was interrupted either for failure or adverse event. Incidence was defined as the number of patients on each regimen (d4T- and AZT-based thera- py) in whom TAMs occurred divided by the total num- ber of patients on that regimen. To measure resistance occurrence Visible Genetics TRUGENE HIV-1 geno- type kit was used. RESULTS: Two-hundred-and-fifty-two patients were started on a AZT- and 153 on a d4T-based regimen, respectively. Baseline median lymphocyte CD4 count and viral load were as follow: AZT group, 317/mm3 and 4.9 log copies/ml vs d4T group 174/mm3 and 5.4 log copies/ml, respectively. At the study end-point, after a median follow-up of 26 months, 123 (49%) patients on a AZT-based regimen and 66 (43%) patients on a d4T-based regimen remained on their initial regimen with a median viral load of 2.3 and 2.6 log copies/ml, respectively. Ninety-three (37%) patients on a AZT-based regimen and 72 (47%) patients on a d4T-based regimen interrupted their treatment after a median duration of 10 months on therapy. Other patients were either lost to follow-up or died during the study period. The incidence of TAM occurrence was 23.8% in the AZT group vs 5.5% in the d4T group (P<0.01). Moreover, T215Y/F was detected in 10.4% of patients in AZT group vs 1.8% of those in d4T group (P<0.01). In a sensitivity analy- S154 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 139 to NRTIs, including 3TC; 2) V118I/M184V without TAMs in a rare ZDV/3TC-treated patient was resistant Implications of lamivudine use and to 3TC, but sensitive to all other NRTIs; 3) V118I with TAMs in d4T-treated patients (not receiving 3TC) the M184V mutation in (n=2) showed ZDV, ABC and TFV resistance but V118I-related multidrug resistance remained sensitive to 3TC; 4) V118I/M184V/TAMs in d4T/3TC-treated patients showed resistance to BG Brenner1, M Oliveira1, V Micheli2, D Moisi1, 3TC/ABC (n=4) but retained varying degrees of sensi- A Cargnel2, M Petrella1 and MA Wainberg1 tivity to ZDV and TFV; 5) V118/184V/TAMs in ZDV/3TC-treated patients (n=2) showed cross resis- 1 McGill University AIDS Centre Montreal, Quebec, Canada; and tance to all NRTIs. 2 L Sacco Hospital, Milan, Italy CONCLUSIONS: V118I without M184V is generally BACKGROUND: A new pattern of mutations involv- associated with use of d4T and/or ddI, conferring resis- ing V118I with or without E44D/A can confer moder- tance to all NRTIs except 3TC. The presence of ate resistance to lamivudine (3TC), as well as cross V118I/184V/TAMs in d4T/ddI-experienced patients resistance to other nucleoside reverse transcriptase receiving 3TC conferred 3TC resistance but also facil- inhibitors (NRTIs). Recent studies suggest that thymi- itated retention of ZDV/TFV sensitivity. In contrast, dine analogues (ZDV/d4T) and didanosine (ddI), but V118I with M184V, arising in ZDV failure, rendered not 3TC, promote development of this V118I resis- viruses resistant to all NRTIs. V118I appears to arise tance pathway. We characterized the phenotypic and as a compensatory mutation that improves fitness of virological characteristics of V118I arising from differ- multi-NRTI-resistant viruses. ent NRTI regimens. METHODS: Genotypic analysis of plasma viral RNA was monitored in 559 individuals having viraemia >1000 copies/ml. Prevalence of V118I with or without other NRTI mutations was related to usage of zidovu- dine (ZDV), stavudine (d4T), ddI, abacavir (ABC) and/or 3TC. NRTI mutations included M184V, TAMs (M41L, D67N, K70R, L210W, T215Y/F, K219E/Q), NAMs (Q151M, F77L, F116Y, A62V, V75I), as well as V118I (±E44A/V and TAMs). Clinical viral isolates were amplified from select patients with known thera- peutic histories. Cell-based phenotypic assays were used to determine IC values of viruses with V118I. 50 These viruses were grown in varying doses of 3TC or tenofovir (TFV) to confirm drug susceptibility. RESULTS: A high prevalence of viruses harbouring M184V, V118I and both mutations were observed in individuals undergoing treatment failure (37.7, 6.4 and 11.7% of cases, respectively). Of note, viraemia was significantly lower in the M184V group than the M184V/V118I group, with even higher viraemia in the V118I and wild-type groups (P>0.001, ANOVA analy- sis). The presence of V118I in the absence of M184V was restricted to those individuals receiving d4T and/or ddI without 3TC. The combination of V118I and M184V was more prevalent in 3TC-treated indi- viduals receiving d4T and/or ddI (n=39) than those receiving ZDV (n=14). A subset of the samples were isolated and five distinct V118I phenotypic patterns were observed: 1) V118I alone without TAMs/M184V in untreated patients (n=2) did not result in resistance XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S155 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 140 baseline) and increased replicative capacity (1.2- to 2.7-fold increase above baseline) occurred in the three Antiviral activity of lamivudine in subjects who had 184V→M. In two of three subjects with 184V→M, ZDV susceptibility decreased nine- persons infected with HIV-1 that and 10-fold from baseline. CD4 lymphocyte counts were has M184V and multiple thymidine stable during the period of 3TC withdrawal in the three analogue mutations subjects with 184V→M, but decreased 97 cells/mm3 (19%) in the subject who did not have 184V→M. 1 M184V reappeared after reinitiation of 3TC in all TB Campbell , RK Young, SC Johnson, three subjects who had 184V→M, but plasma HIV-1 ER Lanier and DR Kuritzkes RNA did not return to baseline values in two subjects. Additional resistance mutations in protease and 1 University of Colorado Health Sciences Center, Denver, Col., decreased lopinavir susceptibility occurred in two sub- USA jects during 3TC interruption. BACKGROUND: M184V in HIV-1 reverse transcrip- CONCLUSIONS: 3TC contributes to partial suppres- tase (RT) confers ≥1000-fold resistance to lamivudine sion of HIV-1 replication despite presence of the 184V (3TC), increased thymidine analogue [zidovudine mutation. Although maintenance of decreased virus (ZDV) and stavudine (d4T)] susceptibility and replication fitness and increased thymidine analogue decreased replicative fitness. Thus, maintenance of susceptibility contribute to the persistent antiviral M184V may be an important component of treatment activity of 3TC in the presence of M184V, there is also strategies for multidrug-resistant HIV-1 infection. This residual direct antiviral activity of 3TC in this setting. study tested the hypothesis that continued 3TC treat- These data support the use of 3TC in salvage therapy ment provides antiviral benefit by maintaining of multidrug-resistant HIV-1 infection even when M184V, and that reversion to wild-type (184V→M) is M184V is present. associated with increased plasma HIV RNA. METHODS: This was an open-label pilot study of 3TC interruption in patients failing 3TC-containing regimens. Entry criteria included plasma HIV-1 RNA >1000 copies/ml, use of 3TC, and ZDV or d4T in the current regimen. Persons taking abacavir, didanosine or zalcitabine were excluded. M184V, at least three thymidine analogue mutations and mutations associat- ed with resistance to all other drugs in the current anti- retroviral regimen were required at study entry. Subjects stopped 3TC but continued all other anti- retrovirals. RESULTS: Four subjects were enrolled (entry CD4 count range: 163–517 cells/mm3; entry plasma HIV-1 RNA range: 4.26–5.53 log copies/ml). All subjects 10 were on three-drug antiretroviral regimens that includ- ed an HIV-1 protease inhibitor, in addition to d4T or ZDV, and 3TC at study entry. Three subjects complet- ed 24 weeks of follow-up and one subject completed 20 weeks. Within 12 weeks after 3TC withdrawal, plasma HIV-1 RNA increased >0.5 log copies/ml 10 above baseline in all subjects. Increased plasma HIV-1 RNA occurred concomitantly with 184V→M in one subject, but occurred prior to 184V→M in two sub- jects who had either complete or partial 184V→M during 3TC withdrawal. One subject had increased plasma HIV-1 RNA despite continued M184V. Increased 3TC susceptibility (>34-fold increase above S156 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 141 lower FC to zidovudine, stavudine or tenofovir com- pared to viruses without M184V (P<0.05). Reduced The impact of the HIV reverse susceptibility to abacavir (FC>4.5) was only seen with less common T215F+M184V (4.6-fold), and the FC to transcriptase M184V mutation in abacavir was 4.3-fold when T215Y/F+M184V were combination with single thymidine analysed together. Susceptibility (FC<1.7) was main- analogue mutations on nucleoside tained to didanosine with M184V plus any single TAM. When genotypic algorithms were compared for reverse transcriptase inhibitor the same mutation patterns, concordance across all resistance in clinical samples from a algorithms (ANRS, HIVDB, RCG, REGA, VGI and large patient database GeneSeq) was not observed. Most algorithms predict- ed decreased susceptibility to zidovudine for every TAM pattern and did not account for the presence of L Ross1, N Parkin2, M Bates2, R Fisher1, M184V, which decreased phenotypic resistance for M St Clair1, M Tisdale3 and R Lanier1 zidovudine below the cut-off for all groups. Most algo- rithms predicted decreased susceptibility to abacavir 1 GlaxoSmithKline, RTP, NC, USA; 2 ViroLogic, Inc., Calif., USA; for M184V and any TAM. Over-prediction of reduced and 3 GlaxoSmithKline, Stevenage, UK susceptibility was also observed for didanosine and zalcitabine. BACKGROUND: Previous studies have assessed the impact on multiple thymidine analogue mutations CONCLUSIONS: In vivo, individual TAMs do not (TAMs) in specific combinations on phenotypic sensi- have an equivalent impact on NRTI resistance. tivity and have demonstrated for some nucleoside Concordance across algorithms for these mutation pat- reverse transcriptase inhibitors (NRTIs), a re-sensitiza- terns was not observed. In the presence of the M184V, tion of the virus in the presence of the M184V muta- re-sensitization to some drugs is observed despite the tion. However, assessment of the impact of the M184V presence of a single TAM. These findings suggest that mutation in combination with single TAMs on pheno- retaining lamivudine in those treatment regimens typic susceptibility to NRTIs may be useful in devising where TAMs can be selected may provide therapeutic rules for genotypic-based drug resistance algorithms. benefit by maintaining M184V and concomitant sup- pression of resistance to zidovudine, stavudine and METHODS: Approximately 10000 HIV clinical sam- tenofovir. ples containing matching genotypes and phenotypes were queried to identify samples with a single TAM (M41L, D67N, K70R, L210W, T215F or Y, and K219E, H, N, Q or R). Samples with mixtures at queried positions, more than one TAM, or any other NRTI resistance-associated mutation other than M184I or V were excluded. Samples were divided into those with or without M184I or V, in addition to a spe- cific TAM. The median fold-change (FC) for each NRTI was compared with predictions made by several genotypic algorithms for HIV drug resistance using the Stanford HIValg Resistance Algorithm Comparison version 3.1.1. RESULTS: Sensitivity to all NRTIs was observed for any single TAM without M184V except for zidovudine with K70R (2.8-fold reduced susceptibility), T215Y (5.6-fold) or T215F (6.3-fold). The T215Y/F muta- tions, without M184V, were also associated with increases in FC values to stavudine (1.4- to 1.5-fold, Mann-Whitney P<0.0001 vs wild-type) and TDF (1.2- to 1.3-fold; P<0.01 vs wild-type), but were below reduced-susceptibility cut-offs. However, M184V plus K70R, T215Y/F was significantly associated with XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S157 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 142 (range <1 to 188 ng/ml). A median IQ of 33 was achieved in the 45 patients (range <1 to 2645) and IQ Determination of capravirine values >1 were achieved in 39 of the 45 (87%) patients. The IQ values in various genotypic subgroups inhibitory quotient in HIV-infected, were likewise evaluated. The highest IQ values were non-nucleoside reverse achieved in patients infected with wild-type HIV (16% transcriptase inhibitor of patients; median IQ of 509, range 10–2645) or with HIV variants containing the K103N mutation with treatment-experienced patients and without other NNRTI resistance-associated substi- tutions (58% of patients; median IQ of 43, range <1 to J Hammond, S Raber, M Amantea, 2261). A median IQ of 2.5 (range <1 to 65.2) was H Grettenberger, B Shetty, P Hawley observed in patients infected with NNRTI-resistant and AK Patick HIV variants that did not contain the K103N substitu- tion (27% of patients). Agouron Pharmaceuticals, Inc., a Pfizer Company, La Jolla, Calif., USA CONCLUSION: Results from this study demonstrate potent in vitro antiviral activity of CPV against HIV BACKGROUND: Capravirine (CPV; formerly isolates from NNRTI-experienced patients. AG1549) is a non-nucleoside reverse transcriptase Furthermore, the CPV IQ values achieved in NNRTI- inhibitor (NNRTI) currently in Phase II development experienced patients suggest that patients failing an for the treatment of HIV infection in NNRTI treat- NNRTI-based regimen may derive virological benefit ment-experienced patients. Previous studies have from subsequent treatment with CPV. Additional stud- demonstrated potent in vitro antiviral activity of CPV ies evaluating CPV in combination with nelfinavir or against strains of HIV-1 resistant to the currently lopinavir/ritonavir are currently underway. approved NNRTIs. The purpose of the current study was to determine the baseline inhibitory quotient (IQ) achieved in NNRTI-experienced patients enrolled in a Phase II study of CPV 1400 mg in combination with nelfinavir and two nucleoside analogues. METHODS: Phenotypic and genotypic analyses were conducted using the Virco Antivirogram and Vircogen assays, respectively. CPV plasma trough concentra- tions were determined at week 4 of the study and IQ values were calculated as the ratio of the free trough concentration (adjusted for protein binding in 100% human serum) to the IC value (adjusted for protein 50 binding in tissue culture medium). Complete baseline genotype and phenotype data and week 4 trough con- centrations were available for 45 of the 56 patients randomized to the CPV arm of the study. RESULTS: Genotypic analyses identified K103N as the predominant NNRTI resistance-associated substitu- tion, occurring alone or with up to two additional NNRTI resistance-associated substitutions in 26 of the 45 (58%) patient isolates examined. Phenotypic analy- ses demonstrated that 37 of the 45 (84%) isolates demonstrated high-level (>10-fold) resistance to at least one approved NNRTI. Despite the presence of resis- tance to the approved NNRTIs, 34 of the 45 (76%) isolates demonstrated <10-fold resistance to CPV. The median protein binding adjusted CPV IC was 50 0.26 ng/ml (range <0.036 to >9.4 ng/ml) and the medi- an free CPV trough concentration was 11.4 ng/ml S158 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 143 detected as having >1% of their total viral population expressing the K103N mutation (AAA→AAT=2 and Resistance genotypes in patients AAA→AAC=3). Clones containing the K103N muta- tion were detected by selective PCR in all five patients, failing nevirapine: co-existence of representing 1, 5, 7, 33 and 76% of total clones. majority viral populations Sequencing of the pol fragment in representative clones expressing Y181C and minority confirmed the results of sequence selective real-time PCR in all cases. For all patients, the sequences with populations expressing K103N and without the K103N mutation usually contained the same polymorphisms and appeared to be closely D Lecossier1, N Shulman2, AR Zolopa2, related. The G102K mutation occurred both with and F Clavel1 and AJ Hance1 without the K103N mutation, but K101E and K101R were only identified in clones without K103N. 1 INSERM U552, Hôpital Bichat – Claude Bernard, Paris, France; and 2 Department of Medicine, Stanford University, Stanford, CONCLUSIONS: Although the majority population Calif., USA of viruses found in patients failing nevirapine fre- quently express the Y181C mutation, minority popu- BACKGROUND: Patients failing nevirapine frequent- lations expressing the K103N mutation can also be ly develop mutation Y181C in reverse transcriptase present, and may contribute to the rapid viral rebound (RT), a mutation that is usually associated with limit- experienced by these patients following introduction of ed cross resistance to efavirenz. These patients, how- efavirenz. ever, often fail subsequent treatment by efavirenz, due to the rapid emergence of viruses expressing the K103N mutation. This rapid selection suggests that even in cases where viruses expressing Y181C are the dominant viral population following nevirapine fail- ure, minority populations expressing the K103N muta- tion may also be present. METHODS: Sixteen patients meeting the following criteria were evaluated: i) patients experiencing viro- logical treatment failure while receiving a combination of antiretroviral agents that included nevirapine; ii) presence of the Y181C mutation, but not the K103N mutation, on standard population-based sequence genotype. Patients failing treatment with regimens that did not include a non-nucleoside RT inhibitor (NNRTI) were used as controls (n=4). Viral RNA was extracted from plasma obtained at the time of geno- typing, RT-PCR was performed, and the products were screened for the presence of the K103N mutation using sequence selective real-time PCR. Each sample was tested for both the AAA→AAT and AAA→AAC codon changes producing this mutation. In cases where the initial screening revealed the presence of >1% of sequences containing the K103N mutation, a 257 bp fragment of the pol gene that included this codon was amplified using primers recognizing conserved sequences in RT, and the products cloned. Individual colonies were screened by sequence selective real-time PCR, and representative clones with and without the K103N mutation were sequenced. RESULTS: In the initial screening, 5/16 patients failing nevirapine, but none of the control subjects, were XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S159 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 144 L210W 24%, T215Y 43%, K219Q 15%, K219E 5%. The efavirenz-based regimens contained also at least Mutations in HIV-1 reverse two NRTIs. Overall, 63% of the efavirenz-treated patients reached undetectable viral load, after a medi- transcriptase associated with an of 7 weeks (IQ range 4–17). At multivariate analy- hypersusceptibility to sis, the baseline RT mutations significantly (P<0.05) non-nucleoside reverse associated with virological success were: M41L (OR=3.0; 95% CI: 1.2–7.4), M184V (OR=3.0; 95% transcriptase inhibitors: impact CI: 1.3–7.0), L210W (OR=3.3; 95% CI: 1.1–10.5), on virological response to T215Y (OR=3.1; 95% CI: 1.3–7.5), and the associa- efavirenz-based salvage therapy tion M41L/T215Y (OR=3.1; 95% CI: 1.2–8.1) and M41L/T215Y/M184V (OR=4.8; 95% CI: 1.4–16.4). By contrast, in the control group only the M184V V Tozzi, M Zaccarelli, F Ceccherini-Silberstein, mutation was associated with virological success P De Longis, G D’Offizi, F Forbici, R D’Arrigo, (OR=2.7; 95% CI: 1.5–5.1), while the L210W one was E Boumis, R Bellagamba, S Bonfigli, C Carvelli, associated with failure to achieve undetectable viral P Narciso, A Antinori and CF Perno load (OR=0.4; 95% CI: 0.2–0.97). INMI L Spallanzani, Roma, Italy CONCLUSIONS: Data from an observational clinical database suggest that among mutations conferring BACKGROUND: Hypersusceptibility to non-nucleo- resistance to NRTI, the presence, at the baseline geno- side reverse transcriptase inhibitors (NNRTI) has been type resistance testing, of the M41L, M184V, L210W described in association with RT mutations conferring and T215Y mutations is associated to a better virolog- resistance to NRTIs. We have evaluated the impact on ical outcome in efavirenz-based salvage regimens, as virological response to efavirenz-based salvage therapy previously suggested by in vivo and in vitro data. of RT mutations known to be associated with hyper- susceptibility to NNRTIs. METHODS: Retrospective analysis of an observation- al database containing genotypic sequence, clinical, virological and immunological data from all patients who received genotypic resistance testing since June 1999 at the INMI L. Spallanzani of Rome, Italy. Patients failing stable highly active antiretroviral ther- apy (HAART) without previous exposure to NNRTIs, who received efavirenz-based, genotypic resistance testing-guided, salvage therapy were included into the analysis. Patients who did not receive any NNRTI as savage therapy served as control group. Primary end- point was the virological success to therapy, defined as achievement of plasma HIV RNA <80 copies/ml. A logistic regression was used assess the independent effect of each mutation considered, after adjustment for duration of follow-up, number of viral load deter- minations, CD4 cell count and HIV RNA at baseline. RESULTS: A total of 357 patients were studied (109 efavirenz-treated and 248 controls). At baseline the efavirenz-treated patients (men 72%, median age 39 years, CDC stage C 56%) had a mean of 39 months of prior antiretroviral treatment, mean HIV RNA of 4.6 log copies/ml, mean CD4 cell count of 357/mm3. Overall, the baseline prevalence of RT mutations asso- ciated with NRTI resistance was: M41L 46%, E44D 6%, D67N 33%, T69D 4%, T75I 2%, M184V 67%, S160 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 145 both wild-type and mutated codons, while substitu- tions at positions 5, 151 and 184 were confirmed in Polymorphism and drug-selected longitudinal analysis. This suggests that mutations 5I→V, 151Q→M, 184M→V and 223K→R were mutations in the reverse selected under NRTI-pressure. Moreover, HIV-2 transcriptase gene of HIV-2 from strains of subtype A and B from untreated patients dis- patients living in southern France played different amino acids at the following positions: subtype A: 8I, 21L, 27T, 32E, 90I, 107T, 163T, 166Q/N, 169E, 194T, 211G/S, 212L, 218D; subtype P Colson1, M Henry1, N Tivoli1, H Gallais2, B: 8V, 21I, 27S, 32L, 90V, 107S, 163S, 166K, 169D, JA Gastaut3, J Moreau4 and C Tamalet1 194S, 211N/D, 212M/V, 218E. 1 CHU Timone; 2 Hôpital Conception; 3 Hôpital Sainte- CONCLUSION: In the present study, we identified 12 Marguerite; and 4 Hôpital Nord, Marseille, France positions in HIV-2 RT with a high level of variability in the absence of therapy. Six codons in HIV-2 from BACKGROUND: The susceptibility of human RTI-naive patients were homologous to those confer- immunodeficiency virus type 2 (HIV-2) to nucleosidic ring drug resistance in HIV-1 (three concerned reverse transcriptase inhibitors (NRTIs) is largely NNRTI-resistance). Interestingly, two amino acids (75I unknown. We studied HIV-2 reverse transcriptase (RT) and 151M) frequently observed in HIV-2 from most gene from 23 HIV-2-infected patients who were NRTI-treated patients are involved in a multi-NRTI exposed or not to NRTI. We aimed (i) to characterize resistance complex in HIV-1. We confirmed the high the polymorphism of HIV-2 RT in the absence of drug; prevalence of mutations Q151M and M184V in HIV- (ii) to know whether it naturally harbour codons asso- 2 infecting NRTI-treated patients. Finally, we found ciated with drug-resistance in HIV-1 and; (iii) to iden- two mutations (I5V, K223R) never described neither in tify mutations emerging under NRTI-selective pres- HIV-2 nor in HIV-1. sure. PATIENTS AND METHODS: Twenty-three HIV-2- infected patients followed-up in Marseille were studied for a median duration of 40 months. Fourteen patients had never undergone antiretroviral therapy and nine received NRTIs: zidovudine, seven patients; lamivu- dine, nine; didanosine, five; stavudine, five; abacavir, three; tenofovir, one. Median duration of therapy was of 59 months. Eight out of the nine NRTI-treated patients also received protease inhibitors. Median baseline and lowest CD4 cell-counts were, respective- ly, 579 and 504 cells/mm3 for untreated and 156 and 96 cells/mm3 for NRTI-treated individuals. Sixty-eight HIV-2 plasma RNA or PBMC DNA were directly sequenced. RESULTS: In untreated patients, the highest amino acid variability in HIV-2 RT was observed at positions 10, 11, 20, 43, 104, 121, 135, 162, 176, 180, 200 and 227. Moreover, six amino acids conferring resistance to NRTI (75I, 118I, 219E) or NNRTI (181I, 188L, 190A) in HIV-1 were naturally found in more than 74% of HIV-2 RT sequences. At four positions (5, 151, 184, 223), amino acid frequencies were signifi- cantly different in NRTI-treated vs -untreated patients. Notably, amino acids 184V and 151M were observed in, respectively, 61 and 0% of HIV-2 infecting NRTI- treated and -untreated patients. In one or more sequence, positions 5, 151, 184 and 223 harboured XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S161 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 146 ferent patterns of mutations alone or in combination at positions: 184 (n=16), 70 (n=5), 67 (n=3), 215 (n=3), High frequency of selection of the 65 (n=2), 74 (n=1) and 219 (n=1). No differences were observed between the patients with or without Q151M Q151M mutation in HIV-2-infected mutation as regards neither the number of sequential patients receiving nucleoside lines nor the type of NRTIs previously received (medi- reverse transcriptase an: 3). In the nine remaining isolates, as in all the iso- lates, others substitutions of unknown impact were inhibitor-containing regimen observed. D Descamps1, F Damond1, S Matheron1, CONCLUSION: Compared to HIV-1 infection, there G Collin1, P Campa1, P Foiny1, S Delarue1, was a high frequency of selection of Q151M mutation S Pueyo2, G Chêne2, F Brun-Vézinet1 and the in HIV-2-infected patients receiving various combina- French ANRS HIV-2 Cohort Study Group tions of NRTIs. Q151M mutation was often associat- ed with K65R substitution but never with the 151 1 Hôpital Bichat, Claude Bernard, Paris; and 2 INSERM U330, complex (A62V, V75I, F77L and F116Y). Although Bordeaux, France AZT and d4T were prescribed in a high number of patients, mutation at codon 215 was selected at a low BACKGROUND: To determine retrospectively which frequency while no substitutions were found at codons substitutions in the reverse transcriptase (RT) gene are M41 and N210. The selection of M184V/I mutation selected in vivo during nucleoside RT inhibitors was observed in all patients receiving 3TC. In this (NRTI)-containing regimen in HIV-2-infected patients highly thymidine analogues pretreated patients the included in the French ANRS HIV-2 cohort. selection of thymidine analogues mutations was low, suggesting a different pathway to resistance as com- PATIENTS AND METHODS: Thirty-four HIV-2- pared to HIV-1. infected patients having received NRTI-containing reg- imen with available specimens and amplifiable RT gene were studied. All patients received various sequential NRTIs except tenofovir [zidovudine (AZT): 87%, stavudine (d4T): 69%, didanosine (ddI): 75%, lamivudine (3TC): 84%, zalcitabine (ddC): 28%, and abacavir: 25%] alone and/or in combination. Plasma HIV-2 RNA was quantified using real-time PCR. RT sequences were compared to HIV-2 subtypes A and B consensus sequences. Analyses of RT gene were per- formed after a median NRTIs exposure of 54 months (6–156). Phylogenetic analyses based on RT sequences were performed for HIV-2 subtype determination. RESULTS: Median viral load at time of sequencing was 3240 copies/ml (<250, 200, 000). Thirty-one patients were infected by subtype A, two by subtype B and one by subtype C. All the 34 patients had viruses with RT mutations when compared to HIV-2 consen- sus sequences. Mutations at positions known to be involved in HIV-1 resistance were observed in 25/34 patients. Selection of Q151M mutation was observed in nine out of 34 isolates (26%) after a median NRTIs exposure of 49 months (12–71). In 8/9 cases, Q151M mutation was associated with other substitutions at positions known to be involved in HIV-1 resistance: K65R (n=6), D67N (n=1), N69S or T (n=2), K70R (n=3), M184V (n=3), S215T or Y (n=2). Among the 16 remaining patients exposed to NRTIs for a median duration of 54 months (6–156) viruses harboured dif- S162 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 147 sequencing after TI, in one case 30% of viral clones became susceptible to one more drug including in the Resistance analysis of the benefit of salvage regimen and in the other case, 13% of viral clones became susceptible to at least one therapeutic a treatment interruption before class. salvage therapy including seven to nine antiretroviral drugs CONCLUSION: In these patients with multiple thera- peutic failure, the viral population before TI was high- (GIGHAART ANRS 097) ly mutated and very homogeneous. After the TI, the analysis of molecular clones revealed the presence of C Delaugerre1, S Dominguez1, K Gourlain1, circulating quasispecies susceptible to one or more C Duvivier1,3, B Amellal1, G Peytavin2, antiretroviral drugs, while the population based- AG Marcelin1, D Costagliola3, C Katlama1,3 and sequencing showed few or no reversion of resistance V Calvez1 mutations. These results suggest that the ‘reversion phenomenon’ is an emergence of quasispecies more 1 Pitié-Salpêtrière Hospital; 2 Bichat-Claude Bernard Hospital; susceptible and is probably importantly involved in the and 3 INSERM EMI 0214, Paris, France success of TI approach before salvage therapy. BACKGROUND: Treatment interruption (TI) followed by multidrug salvage therapy induces a significant viro- logical and immunological benefit up to 48 weeks in patients with multiple failure in a context of very advanced HIV disease (Katlama et al. CROI 2003). Three major factors were independently associated to virological success: treatment interruption with resis- tance reversion (defined as the loss of at least one major resistance mutation), adequate drug concentra- tion and the use of lopinavir. The aim of the study was to understand the virological mechanisms of the suc- cess of TI before salvage therapy in patients with mul- tiple therapeutic failure. PATIENTS AND METHODS: Patients were random- ized to either immediate or deferred GIGHAART ther- apy after 8 weeks of TI. Genotypic analyses were per- formed by population-based sequencing (=bulk sequencing) at baseline (n=61) and after 8 weeks of TI (n=30). Clonal sequencing of polymerase gene (com- plete protease and RT codons 1 to 230) was done in plasma sample of four patients at baseline and in three patients at the end of TI (30 clones each). RESULTS: Median number of resistance mutations (IAS panel) at baseline within the three drugs classes was 13 (range 6–18). Clonal analysis at baseline, for the four patients, showed that the viruses harboured all resistance-associated mutations linked on the same viral genome and that all quasispecies carried all muta- tions. By population based sequencing, reversion of mutations occurred in 48% of patients after TI. In one patient with only two reversion mutations in bulk sequencing, clonal analysis showed 75% of quasi- species susceptible to at least one therapeutic class of which 40% of quasispecies completely wild-type. For patients without reversion by population-based XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S163 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 148 (82%) amprenavir (APV) (57%), lopinavir (LPV) (25%), zidovudine (95%), lamivudine (99%), stavu- Baseline resistance and predictors dine (97%), didanosine (87%), abacavir (59%), dal- citabine (41%), at least one NNRTI (97%). Mean of virological response in the baseline resistance summaries (STI vs no-STI): MS CPCRA 064 study: a randomized (NRTI)=3.6 vs 3.3, P=0.08; MS (PI)=4.8 vs 4.6, trial of structured treatment P=0.12; MS (NNRTI)=1.5 vs 1.4, P=0.24; GSS=2.0 vs 2.4, P=0.08; PSS=4.7 vs 5.2, P=0.13. Overall virologi- interruption for patients with cal response: 47% (STI) vs 59% (no-STI), P=0.05. multidrug-resistant HIV Combining both treatment arms, significant predictors of virological response include no prior exposure to J Lawrence1, K Huppler Hullsiek2, D Mayers3, APV and LPV, and the MS for both NRTI and PI class- D Abrams1, J Baxter4 and the CPCRA 064 es. Based on 95% confidence intervals for the odds Study Team for the Terry Beirn Community ratios from the model predicting virological response, Programs for Clinical Research on AIDS each 1 unit decrease in baseline MS for NRTIs results (CPCRA) in a 1.02- to 1.69-fold greater likelihood of virological response, while each 1 unit decrease in baseline MS for 1 University of California, San Francisco, Calif.; 2 University of PIs results in a 1.2- to 2.2-fold greater likelihood of Minnesota, Minneapolis, Minn.; 3 Henry Ford Hospital, Detroit, virological response. No prior exposure to APV and Mich.; and 4 Cooper Hospital/RWJ Medical School, Camden, NJ, LPV is associated with a 1.06- to 3.41- and 2.18- to USA 8.9-fold greater likelihood of virological response, respectively. BACKGROUND: CPCRA 064 is a randomized trial of structured treatment interruption (STI) for patients CONCLUSIONS: The treatment arms have similar with virological failure (HIV RNA >5000 copies/ml) baseline GSS, PSS and MS for each drug class. and multidrug-resistant (MDR) virus. Shown previous- Significant baseline predictors of virological response ly, the STI arm has decreased CD4 count, increased are lower baseline NRTI and PI mutation scores and clinical events and no benefit in virological response no prior exposure to APV or LPV. compared to the no-STI arm. The objective of these analyses is to compare baseline levels of resistance and determine the predictors of virological response. METHODS: Patients were randomized to either a 4- month STI prior to changing therapy (STI arm) or immediately changing therapy (no-STI arm). Virological response is defined as at least a 1 log drop in HIV RNA from baseline or a viral load <400 copies/ml at least once during follow-up. Multivariate logistic regression analyses were performed to predict viral load response using baseline characteristics, including treatment arm, HIV RNA, CD4 count, prior AIDS, prior antiretrovi- ral drug (ARV) use, presence of specific mutations, ARV resistance mutations score (MS), genotypic sensi- tivity score (GSS) and phenotypic sensitivity score (PSS). MS (sum of all major mutations plus half the sum of all minors) was calculated for each class [nucle- oside reverse transcriptase inhibitor (NRTI), protease inhibitor (PI) and non-NRTI (NNRTI)]. GSS and PSS are defined as the number of sensitive drugs in the entire panel of drugs by genotype and phenotype. RESULTS: n=270. Baseline characteristics: median CD4=145 cells/mm3, mean log HIV RNA=5.0. Mean number of prior ARVs=10.6. Prior ARV use: indinavir (84%), ritonavir (84%), saquinavir (89%), nelfinavir S164 Los Cabos, Mexico, 10–14 June 2003 ilar in the two arms (G: 1.37 log reduction; G+P: ABSTRACT 149 10 1.28 log reduction; difference=0.08, SE=0.27, P=0.77). 10 A 1-year randomized controlled Thirty-five percent of subjects in the G arm and 27% in the G+P arm had a VL of less than 50 copies/ml. There trial of genotypic versus genotypic was no significant differences between G vs G+P arms for plus phenotypic resistance testing mean AUC [–0.55 (SE: 0.05) vs –0.54 (SE: 0.06)], and to guide antiretroviral therapy (the change in CD4 counts [+36 (SE: 13) vs +35 (SE: 14)]. The study arms were equivalent in terms of the num- ERA Trial) ber of drugs, the number of ‘active’ drugs (as assessed by genotypic resistance), or the individual prescribed C Loveday1, DT Dunn2, H Green2, A Rinehart3 drugs following the initial test. and P McKenna4 on behalf of the ERA Steering Committee CONCLUSIONS: The ERA Trial found no clear evi- dence of added value of phenotypic resistance testing 1 International Clinical Virology Centre, UK; 2 MRC Clinical Trials against a background of genotypic resistance testing in Unit, UK; 3 VircoLab, Inc., USA; and 4 Virco BVBA, Belgium patients with limited therapeutic options. BACKGROUND: Clinical guidelines recommend the routine use of HIV drug resistance testing to support treatment changes, but they do not prescribe the choice between genotypic and phenotypic testing. We report on the clinical utility of phenotypic resistance testing in addition to genotypic resistance testing in patients with limited therapeutic options and followed for 1 year. METHODS: In the ERA (Evaluation of Resistance Assays) Trial, patients in whom therapy was failing were recruited between February 2000 and July 2001 and randomized to one of two parts (A or B). In Part B (reported here), the clinician was not able to select a regimen of three or more drugs that, with reasonable expectation, would have potent antiviral activity to which each drug contributed. Thus, all subjects had access to genotypic testing with rules-based interpreta- tion (VIRCOGEN I) at baseline and during follow- up and were randomized to have or not have access to phenotypic testing (ANTIVIROGRAM prior to the introduction of drug-specific cut-offs). The primary end-point was change in plasma HIV-1 RNA viral load (VL) and the secondary end-points changes in propor- tion of patients with undetectable VL, the AUC and CD4 changes, between randomization and 12 months by an ‘intention-to-treat’ analysis. RESULTS: 311 patients were enrolled in Part B, of whom 152 were allocated to genotypic testing alone (G arm) and 159 to genotypic plus phenotypic testing (G+P arm). At baseline, mean VL was 4.23 log copies/ml, 10 mean CD4 count was 275 cells/mm3, and subjects had previous exposure to a mean of 7.7 antiretroviral ther- apy drugs. The primary end-point was evaluable in 283 (91%) subjects; absent data was primarily due to non-attendance at the clinic around 12 months. The mean reduction in plasma VL at 12 months was sim- XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S165 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 150 the contrary, most of the NRTI-related mutations were already detectable in the baseline PBMC samples. Resistance mutations during Baseline PBMC were also studied in 23 subjects of the STI arm not developing mutations during STI. intermittent antiretroviral therapy: Mutations were present in 2/23 (8.7%) of them. There does PBMC genotyping before was a significant difference in the frequency of baseline treatment interruption help predict positivity between patients with mutations during STI their occurrence? and subjects retaining a wild-type virus. CONCLUSION: Mutations in the rebounding virus L Palmisano, M Giuliano, CM Galluzzo, were detected in 25% of subjects receiving intermittent R Amici, R Bucciardiniand and S Vella therapy. Most of the NRTI-related mutations were already present in an archival form in baseline PBMCs, Istituto Superiore di Sanità, Rome, Italy whereas mutations related to NNRTI and protease inhibitor (PI) appeared to be newly acquired. Even BACKGROUND: The emergence of resistance muta- though the overall response to therapy re-institution tions is one of the major safety concerns of intermittent did not differ in subjects with or without mutations highly active antiretroviral therapy (HAART) in chron- during STI, a prolonged follow up is needed to assess ic HIV infection. However, their impact on virological their impact on long term response to HAART. response is still unclear. We monitored the occurrence Additional data are also required to confirm that sub- of mutations during the PART study, an ongoing, ran- jects with mutations in PBMC are more likely to devel- domized multicentre clinical trial comparing continu- op resistance when they undergo intermittent therapy. ous (arm A) versus intermittent (arm B) HAART in 273 subjects with chronic stabilized HIV infection. METHODS: Study population consisted of 134 patients of arm B. One hundred and two had com- pleted at least three cycles of intermittent therapy (three STI of 1, 1 and 2 months, respectively, sepa- rated by 3 months on treatment). At enrolment, all were on their first-line HAART regimen and had CD4 >350/mm3 and viraemia <400 copies/ml. HIV-1 geno- type was analysed on plasma samples during viral rebound and on baseline PBMC, using the TrueGene assay. RESULTS: Overall, mutations were found in 34/134 patients (25.3%) at any time during STI (first STI: 13.4%, second STI: 17.5%, third STI: 17.6%). The most frequent individual mutations were: M184V (12.3% of all 3TC-treated subjects); K103N [6.9% of non-nucleoside reverse transcriptase inhibitor (NNRTI)-treated], T215Y (3.7% of NRTI-treated), L90M (10.7% of subjects receiving saquinavir or nel- finavir). The response to therapy re-institution in patients with mutations was 91.2, 88.9 and 90% after the first, second and third STI, respectively. These fig- ures do not differ from those in the general population. For 21/34 subjects HIV DNA genotype was performed on a baseline PBMC sample. Mutations were detected in 9/21 (42.8%). In general, the baseline samples did not show the NNRTI-related mutations seen during STI (V108I: two cases; V188I: two cases; K103N: three cases). The same was true for the L90M (no pos- itivity at baseline, two cases emergent during STI). On S166 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 151 (R=–0.08, P=0.31). Among all patients, there was a non-significant trend between higher month 6 adher- Accumulation of antiretroviral ence and percent of baseline antiretrovirals developing resistance by month 12 (ρ=0.14, P=0.07) and between resistance in treatment-experienced adherence and increased SRPSS (ρ=0.11, P=0.16). patients: the impact of medication Among the subgroup of 94 patients with measurable adherence phenotype at month 12, the associations were more significant, (ρ=0.25, P=0.01 and ρ=0.27, P=0.008, respectively). Among these 94 patients with month 12 L Miller1,2, JA McCutchan3, P Keiser4, phenotypes, predictors of new drug resistance were C Kemper5, M Witt1, J Leedom6, D Forthal7, assessed in multivariate models that included month 6 8 3 3,8 adherence, log baseline and month 12 HIV RNA, N Hellmann , D Richman , E Seefried , 10 A Rigby3,9, R Haubrich3,9 and the California baseline CD4, extensive (>60 months) pre-baseline Collaborative Treatment Group (CCTG) exposure nucleoside reverse transcriptase inhibitors (NRTIs) and protease inhibitors, non-NRTI (NNRTI)- 1 Harbor-UCLA Medical Center, Torrance, Calif.; 2 UCLA AIDS naive status, and number of NRTIs in baseline regi- Institute, Los Angeles, Calif.; 3 University of California San Diego, men. In these models, predictors of higher percent of San Diego, Calif.; 4 University of Texas, Dallas, Tex.; 5 Santa Clara drugs with new resistance at month 12 or higher Valley Medical Center, San Jose, Calif.; 6 University of Southern SRPSS included more extensive pre-baseline NRTI California, Los Angeles Calif.; 7 University of California, Irvine, exposure (P=0.07 and 0.008, respectively), NNRTI- Irvine, Calif.; 8 ViroLogic, South San Francisco, Calif.; and 9 CCTG naive status (P=0.05 and 0.001), number of NRTIs in Data Unit, San Diego, Calif., USA baseline regimen (P=0.04 and 0.0007) and higher adherence (P=0.03 and 0.005) BACKGROUND: The impact of antiretroviral adher- ence on the evolution of resistance is poorly under- CONCLUSIONS: While improved adherence by stood. We studied this relationship in CCTG 575, a treatment experienced patients was associated with multicentre investigation of phenotypic testing versus better virological response, it may have promoted standard-of-care. selection of more resistant HIV. Thus, interventions to improve adherence in such patients may be effec- METHODS: Patients failing antiretroviral therapy and tive at preventing virologic failure but may have little initiating a new regimen were investigated prospective- or a negative impact upon acquisition of new anti- ly. HIV RNA and phenotype to 15 antiretrovirals were retroviral resistance. measured at baseline and month 12 using the ViroLogic PhenoSense assay. Drug susceptibility was expressed as fold change in IC compared to a wild- 50 type control virus. Continuous phenotypic sensitivity scores (AIDS 2003; 17:821–830) for each drug were calculated at baseline and month 12 and modified from a 0 to 1 scale to a 1 to 2 scale. We scored new resistance at month 12 two ways: 1) percent of base- line antiretrovirals that developed new resistance by month 12; and 2) for each drug the month 12 divided by baseline modified continuous phenotypic sensitivity score was calculated and then summed for all drugs to create a summary ratio phenotypic sensitivity score (SRPSS) quantifying global newly acquired drug resis- tance at month 12. Patients with low HIV RNA at month 12 that precluded phenotype were considered to have developed no new resistance. Adherence was obtained using a standardized survey. RESULTS: Among the 205 patients with month 6 vis- its, mean (SD) adherence was 71.6% (26.4%). Month 6 adherence was associated with HIV RNA change at month 6 (R=–0.22, P=0.001) but not at month 12 XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S167 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 152 tions to NRTIs and ≥2 mutations to NNRTIs were less likely to reach the virological success [RR=0.48 (95% The effect of number of mutations CI: 0.31–0.75)], [RR=0.37 (95% CI: 0.13–1.02)] and [RR=0.48 (95% CI: 0.30–0.76)], respectively. At mul- and of drug-class sparing on tivariate analysis ≥7 mutations to PI had still no effect virological response to salvage upon chance of virological success in pts with no pre- genotype-guided antiretroviral vious history of NNRTI-treatment [RR=1.12 (0.59–2.12)]. Viral suppression in this subgroup of therapy patients was maintained for a mean time of 32.9 weeks (SD 22.6). The presence of ≥7 mutations to PI signifi- BC Ciancio, MP Trotta, P Lorenzini, F Forbici, cantly impaired the chances of virological success in U Visco-Comandini, P Marconi, C Gori, NNRTI-exposed pts [RR=0.25 (0.10–0.58)]. Similarly, S Bonfigli, MC Bellocchi, M Zaccarelli, the effect of ≥2 NNRTI-mutations upon chance of R D’Arrigo, V Tozzi, E Boumis, P Narciso, virological success was still not significant in pts with- CF Perno and A Antinori out previous PI-exposure [RR=0.91 (0.33–2.50)]. Viral suppression in this subgroup of patients was main- National Institute for Infectious Diseases Lazzaro Spallanzani, tained for a mean time of 27.7 (SD 11.3) weeks. IRCCS, Roma, Italy CONCLUSIONS: Our data show a non-linear associ- BACKGROUND: A significant association between ation between resistance-conferring mutations and the number of mutations conferring drug resistance at virological outcome. For the protease, a virological baseline genotype and subsequent virological response failure developed only when a high number of muta- was established in retrospective studies. Only few pre- tions were selected, whereas only two NNRTI-associ- liminary data from longitudinal studies has been ated mutation impaired treatment efficacy. GRT-guid- reported. The objective was to assess longitudinally the ed therapy still provides remarkable chances of durable impact of number of drug-associated mutations and virological success even in pts with ≥7 mutations to PI history of previous exposure to antiretrovirals on the and in pts with ≥2 mutations to NNRTI, when the pts virologic response to genotype-guided antiretroviral did not have a three-class exposure. GRT and as-long- therapy. as-possible sparing of a drug class could be a conve- nient strategy for long-term management of drug-fail- METHODS: Patients (pts) failing highly active anti- ing patients. retroviral therapy (HAART) who underwent geno- type resistance testing (GRT) between June 1999 and March 2002 were enrolled. Genotype was performed by Viroseq-2 with expert advice offered to physi- cians. Baseline variables and longitudinal data were recorded. Main outcome was reaching undetectable (<80 copies/ml) HIV-1 RNA level at follow-up (viro- logical success). Mutations considered in the analysis were all those recently reported by IAS. Multivariate analysis was performed by Cox proportional hazard model. RESULTS: 470 pts were enrolled and followed-up for a median of 15 months (IQR 9–19) after GRT. Forty-two percent (188 pts) reached HIV-1 RNA <80 copies/ml after a median follow-up time of 8 months (IQR 4.44–15.15). Undetectable HIV-1 RNA plasma levels were maintained for a mean of 31.3 (SD 20.8) weeks. For each single protease inhibitor (PI)-, nucleoside reverse transcriptase inhibitor (NRTI)- and non-NRTI (NNRTI)-associated mutation, there was a reduction of, respectively, 8% (P=0.002), 8% (P=0.009) and 29% (P=0.001) in the likelihood of reaching virologi- cal success. Pts carrying ≥7 mutations to PIs, ≥6 muta- S168 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 153 response in this cohort – and a ‘weighted’ genotype score appeared to be a better predictor compared to a Genotypic predictors of response simple summation of mutations. to lopinavir/ritonavir in clinical practice H Rice1, J Nadler2, J Schaenman1, T Hawkins3, C Cohen4, R Rode5, D Kempf5 and A Zolopa1 1 Stanford University, Palo Alto, Calif.; 2 University of South Florida, Tampa, Fla.; 3 University of New Mexico, Santa Fe, NM; 4 Community Research Initiative of New England, Boston, Mass.; and 5 Abbott Laboratories, Chicago, Ill., USA INTRODUCTION: Genotypic correlates of HIV-1 resistance to lopinavir/ritonavir (LPV/r) have previous- ly been identified, but continue to evolve with broader clinical experience. We sought to determine the geno- typic predictors of viral load (VL) response to LPV/r- based regimens from a multicentre clinical cohort. METHODS: The Clinic-Based Investigator Group (CBIG) is comprised of 20 academic, private and public clinics in the USA. Patients (pt) were retrospectively iden- tified who received LPV/r-based regimens, had baseline (BL) VL and genotype and had at least two follow-up VL over 16 weeks. Regression analysis was performed using an ordinal response variable, with response defined as achieving VL less than 500 copies/ml (full), at least a 0.5 log VL reduction (partial) or less than 0.5 log VL 10 10 reduction (none) at week 4 (days 14–56), week 12 (days 57–112), week 24 (days 113–224) or any of the three time points. RESULTS: Seventy-seven pt had data available for analy- sis. At baseline, median age was 43 years, CD4 count was 240 cells/mm3 and log VL was 4.16 copies/ml. 10 Median number (range) of prior regimens, prior pro- tease inhibitors (PIs) and BL PI mutations were 4 (1–24), 2 (0–6) and 4 (0–15), respectively. Overall, 72% achieved full response and 15% achieved partial response at some time during the study. Significant (P<0.05) genotypic predictors included mutations 20M, 36I, 46L, 54S/V, 71V, 73S and 82A. Stepwise regression revealed that the sum of these mutations was highly associated with VL response at weeks 4, 12 and 24. In particular, mutations 46L, 71V and 82A were the strongest predictors of VL response. The LPV mutation score and French ATU mutation set were also predictive of VL response over 24 weeks. CONCLUSIONS: This cohort analysis confirms prior analysis from clinical trials. It appears that mutations 46L, 71V and 82A were the strongest predictors of VL XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S169 Los Cabos, Mexico, 10–14 June 2003 ABSTRACT 154 virological response (>1000 copies/ml) due to the pres- ence/absence of mutations at BL. 908/r and LPV/r had Optimizing 908/r in protease a different effect depending on the presence of V82A/F/T/S (41/278, 15%) or I84V (18/278, 7%) at inhibitor-experienced patients: a BL. Subjects harbouring virus with a V82A/F/T/S retrospective analysis of virological mutation at BL were more likely to have <1000 response based on baseline copies/ml at week 8 if they received 908/r (18/25, 72%) versus LPV/r (6/16, 38%). Whereas, subjects genotypic mutations and after harbouring virus with an I84V mutation at BL were 8 weeks of therapy more likely to have >1000 copies/ml at week 8 if they received 908/r (7/12, 58%) vs LPV/r (0/6, 0%). In con- P Yates, S MacManus, S Sharp, W Snowden, trast, only 1/45 subjects (2%) entering the study har- M Tisdale and RC Elston bouring virus containing D30N had viral load >1000 copies/ml at week 8. Mutations with decreased inci- GlaxoSmithKline, Stevenage, UK dence in all treatment arms included D30N (16 to 2%), V77I (35 to 19%) and N88D/S(14 to 4%). BACKGROUND: GW433908(908) is an investiga- tional protease inhibitor (PI) with demonstrated CONCLUSION: In this study, 908/r and LPV/r had antiviral efficacy, durability and tolerability in anti- good virological effect on virus resistant to NFV retroviral (ART)-naive and -experienced subjects. (D30N, N88D). Each had a different degree of viro- CONTEXT is a 48-week Phase III study comparing logical success at week 8 against isolates with 908/r QD (1400 mg/200 mg QD) or 908/r twice daily V82A/F/T/S or I84V at BL. Subjects presenting with (700 mg/100 mg twice daily) with lopinavir/r (433 V82A/F/T/S (41/278, 15%) mutation at baseline were mg/100 mg twice daily). To be eligible subjects must more likely to respond to 908/r therapy than LPV/r have failed a PI-containing regimen and be able to con- therapy, conversely subjects with I84V(18/278, 7%) at struct an active backbone of two nucleoside reverse BL were more likely to respond to LPV/r than 908/r. transcriptase inhibitors (NRTIs) based upon a VGI These data support the use of BL genotypic testing to resistance test. Non-NRTI (NNRTI) use was not per- optimize treatment of PI-experienced subjects. mitted. The current analysis describes the development and incidence of protease resistance mutations present in virus from subjects with viral load >1000 copies/ml at week 8. METHODS: Genotypic analysis was performed on virus from subjects at Baseline (BL, n=278) and those having a viral load >1000 copies/ml at week 8 (n=57). Genotyping was performed by ViroLogic, Inc. RESULTS: Fifty-seven subjects had viral load >1000 copies/ml at week 8 (908/r QD n=25, 908/r twice daily n=14, LPV/r n=18). Resistance mutations selected in virus from subjects receiving 908/r (n=39) consisted of: I84V (n=6), I54L/M (n=5), M46I/L (n=4), L10F/I/R/V (n=3), L33F (n=2), V32I (n=1), L63P (n=1) and A71V/T (n=1). Resistance mutations selected in virus from subjects receiving LPV/r (n=18) consisted of K20M/R (n=2), M36I (n=2), M46I (n=1), I50V (n=1), V82A (n=1) and I84V (n=1). Protease mutations with increased incidence (>3-times BL) at week 8 were V32I, L33F, M46I/L, F53L, I54L/M/V and I84V for 908/r-treated subjects and K20M/R, V32I, G48V, I54V and V82A/T/S/F for LPV/r-treated subjects. Increased incidence was dependent upon the development of new mutations not detected at BL, re-selection of archived mutations and/or a subject having a less favourable S170 AUTHOR INDEX PRESENTING AUTHOR ABSTRACT PAGE Z Ambrose 27 S30 L Bacheler 118 S132 J Barbour 73 S80 M Bates 74 S81 N Beerenwinkel 97 S107 D Bennett 119 S133 RC Bethell 3 S6 S Blower 131 S145 L Bocket 138 S154 K Borroto-Esoda 7 S10 B Brenner 139 S155 F Brun-Vézinet 91 S101 V Calvez 107 S117 R Camacho 126 S140 TB Campbell 140 S156 B Canard 34 S39 F Ceccherini-Silberstein 40 S45 M-L Chaix 123 S137 C Chappey 32 S37 CK Chu 36 S41 F Clavel 104 S114 E Coakley 58 S65 P Colson 145 S161 A Corral 81 S88 M-P de Bethune 8, 17 S11, S20 A De Luca 112 S122 C de Mendoza 127, 130 S141, S144 S De Meyer 15 S18 S Deeks 66 S73 C Delaugerre 80, 147 S87, S163 L Demeter 41 S46 D Deschamps 146 S162 J Deval 35 S40 G DiRienzo 88 S98 XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S171 PRESENTING AUTHOR ABSTRACT PAGE L Doyon 14 S17 L Dunkle 2 S5 R Elston 154 S170 S Eshleman 79 S86 J Esté 24 S27 A Florance 111 S121 AS Foulkes 99 S109 C Fraser 64 S71 L Frenkel 87 S97 O Gallego 103 S113 JG Garcie-Lerma 82 S89 M Gonzales 39 S44 M Götte 28 S33 R Grant 120 S134 Z Grossman 57, 105 S64, S115 H Günthard 56 S63 D Hall 13 S16 E Halvas 92 S102 A Hance 143 S159 R Harrigan 110 S120 R Haubrich 151 S167 D Havlir 59 S66 D Hazuda 10 S13 G Heilek-Snyder 18 S21 S Hughes 30 S35 P Hunt 75 S82 G Jayaraman 121 S135 R Kagan 49 S54 R Kantor 53 S58 D Katzenstein 78 S85 M Kearney 86 S96 M King 108 S118 B Korber 1 S1 L Kovari 45 S50 M Kozal 116 S130 S172 PRESENTING AUTHOR ABSTRACT PAGE B Larder 102 S112 J Lawrence 148 S164 A Leigh Brown 76 S83 D Liotta 4 S7 S Little 115 S129 C Loveday 149 S165 P Lucia F Maldarelli 133 S149 F Mammano 50 S55 B Masquelier 67, 109 S74, S119 S McCallister 12 S15 J Mellors 134 S150 MD Miller 33, 135 S38, S151 MD Miller 61 S68 GD Miralles 21 S24 H Mo 51 S56 A Molla 85 S95 K Morales-Lopetegi 90 S100 E Morgan 52 S57 M Moser 89 S99 M Nijhuis 95 S105 DV Nissley 93 S103 WA O'Brien 71 S78 S Palmer 62 S69 L Palmisano 150 S166 U Parikh 136 S152 N Parkin 29 S34 M Parniak 11 S14 A Patick 142 S158 L Perrin 63 S70 R Phillips 114 S125 D Pillay 124 S138 M Quinones-Mateu 25 S28 W Rice 5 S8 B Rodés 72 S79 L Ross 19, 141 S22, S157 XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S173 PRESENTING AUTHOR ABSTRACT PAGE J-P Routy 122 S136 C Salama 113 S123 C Schiffer 16, 47 S19, S52 JC Schmit 100 S110 M Segal 94 S104 G Sévigny 20 S23 R Shafer 48 S53 N Shulman 42, 43 S47, S48 A Silva 46 S51 V Simon 77 S84 P Sista 55 S60 N Sluis-Cremer 31 S36 DM Smith 83 S90 AJ Smith 38 S43 J Snoeck 101 S111 M Soares 125 S139 S Stanfield-Oakley 22 S25 R Stevens 106 S116 L Stuyver 6 S9 C Su 54 S59 W Sugiura 84 S91 A Tanuri 128 S142 G Tirado 68 S75 N Tobin 65 S72 V Tozzi 144 S160 MP Trotta 152 S168 D Turner 129 S143 D van de Vijver 132 S146 M Van Houtte 37 S42 K Van Laethem 26 S29 H Van Marck 96 S106 K Van Rompay 70 S77 J Vingerhoets 9 S12 H Walter 98 S108 A Wensing 117 S131 M Westby 23 S26 A Wiegand 60 S67 A Winston 137 S153 S174 PRESENTING AUTHOR ABSTRACT PAGE M Winters 44 S49 J Wong 69 S76 A Zolopa 153 S169 XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S175 DELEGATE ORGANISATION Ambrose, Zandrea National Cancer Institute Tel: +1 301 846 1184; E-mail: [email protected] Bacheler, Lee Virco Lab, Inc. Tel: +1 919 313 2664; E-mail: [email protected] Barbour, Jason Gladstone Institute of Virology & Immunology Tel: +1 415 695 3810; E-mail: [email protected] Bates, Michael ViroLogic Tel: +1 650 624 4219; E-mail: [email protected] Beerenwinkel, Niko Max Planck Institute for Informatics Tel: +49 681 9325 304; E-mail: [email protected] Bennett, Diane Centers for Disease Control and Prevention Tel: +1 404 639 5349; E-mail: [email protected] Bethell, Richard C Shire BioChem, Inc. Tel: +1 450 978 7873; E-mail: [email protected] Blower, Sally David Geffen School of Medicine at UCLA Tel: +1 310 206 3052; E-mail: [email protected] Bocket, Laurence Hospital and University Center Tel: +33 3 20 44 69 30; E-mail: [email protected] Borroto-Esoda, Katyna Triangle Pharmaceuticals Tel: +1 919 493 5980; E-mail: [email protected] Boucher, Charles University Hospital Utrecht Tel: +31 30 250 6526; E-mail: [email protected] Braun, Patrick PZB Tel: +49 241 4709725; E-mail: [email protected] Brenner, Bluma McGill University AIDS Centre Tel: +1 514 340 7536; E-mail: [email protected] Brun, Scott Abbott Laboratories Tel: +1 847 935 1293; E-mail: [email protected] Brun-Vézinet, Françoise Laboratoire de Virologie, Hopital Bichat-Claude Tel: +33 1 4025 8895; E-mail: [email protected] Calvez, Vincent Pitie-Salpetriere Hospital Tel: +33 142 177 401; E-mail: [email protected] Camacho, Ricardo Virology Laboratory, Hospital Egas Moniz Tel: +351 964613490; E-mail: [email protected] Cammack, Nick Roche Palo Alto Tel: +1 650 852 1764; E-mail: [email protected] Campbell, Thomas University of Colorado Health Sciences Center Tel: +1 303 315 8311; E-mail: [email protected] Canard, Bruno AFMB-CNRS Tel: +33 4 91 82 86 44; E-mail: [email protected] Ceccherini-Silberstein, Francesca University of Rome Tor Vergata Tel: +39 06 72596551; E-mail: [email protected] Chaix, Marie-Laure Laboratoire de Virolgie, Hopital Necker Tel: +33 1 44 49 49 61; E-mail: [email protected] Chappey, Colombe ViroLogic Tel: +1 650 866 75438; E-mail: [email protected] XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S177 Chu, CK University. of Georgia Tel: +1 706 542 5379; E-mail: [email protected] Clarke, John Imperial College Tel: +44 207 594 3909; E-mail: [email protected] Clavel, Francois Inserm U552 Tel: +33 1 4025 6363; E-mail: [email protected] Clevenbergh, Philippe CHU Bichat Tel: +33 1 44 87 01 41; E-mail: [email protected] Clotet, Bonaventura Fundacio IRSI Caixa Tel: +34 93 465 6374; E-mail: [email protected] Coakley, Eoin Tufts-NEMC Tel: +1 617 636 2819; E-mail: [email protected] Coffin, John National Cancer Institute-Frederick Tel: +1 301 846 5944; E-mail: [email protected] Collins, Peter Shire BioChem, Inc. Tel: +1 450 978 7850; E-mail: [email protected] Colson, Philippe CHU Timone Tel: +33 0491385519; E-mail: [email protected] D'Aquila, Richard Vanderbilt University School of Medicine Tel: +1 615 343 1743; E-mail: [email protected] de Bethune, Marie-Pierre Tibotec Tel: +32 015 401 240; E-mail: [email protected] De Gruttola, Victor Harvard School of Public Health Tel: +1 617 432 2820; E-mail: [email protected] De La Rosa, Abel Pharmasset Tel: +1 678 395 0025; E-mail: [email protected] De Luca, Andrea Catholic University Tel: +39 630154945; E-mail: [email protected] de Mendoza, Carmen Hospital Carlos III Tel: +34 914532500; E-mail: [email protected] De Meyer, Sandra Tibotec Tel: +32 015 401 226; E-mail: [email protected] Deeks, Steven UCSF Tel: +1 415 476 4082; E-mail: [email protected] Delaugerre, Constance Dept Virology Tel: +33 142177401; E-mail: [email protected] Demeter, Lisa University of Rochester Tel: +1 585 275 4764; E-mail: [email protected] Descamps, Diane Hopital Bichat-Claude Bernard Tel: +33 1 40 25 61 54; E-mail: [email protected] Deval, Jerome AFMB-CNRS Tel: +33 4 91 82 86 47; E-mail: [email protected] DiRienzo, Gregory Harvard School of Public Health Tel: +1 617 432 2831; E-mail: [email protected] S178 Doyon, Louise Boehringer Ingelheim (Canada) Ltd., R&D Tel: +1 450 682 4640; E-mail: [email protected] Dunkle, Lisa Achillion Pharmaceuticals, Inc. Tel: +1 203 752 5403; E-mail: [email protected] Elston, Robert GlaxoSmithKline Tel: +44 1438 763904; [email protected] Erickson, John Sequoia Pharmaceuticals, Inc. Tel: +1 240 632 0094; E-mail: [email protected] Eshleman, Susan Johns Hopkins University Tel: +1 410 614 4734; E-mail: [email protected] Este, Jose Fundacio IrsiCaixa Tel: +34 934656374; E-mail: [email protected] Fisher, Robin GlaxoSmithKline Tel: +1 919 483 5023; E-mail: [email protected] Florance, Allison GlaxoSmithKline Tel: +1 919 483 8989; E-mail: [email protected] Foulkes, Andrea S UPenn School of Medicine Tel: +1 215 573 5174; E-mail: [email protected] Fraser, Christophe Imperial College London Tel: +44 20 7594 3397; E-mail: [email protected] Frenkel, Lisa University of Washington Tel: +1 206 987 5140; E-mail: [email protected] Garcia-Lerma, J Gerardo Centers for Disease Control and Prevention Tel: +1 404 639 4987; E-mail: [email protected] Gatell, Jose Hospital Clinic Tel: +34 93 2275430; E-mail: [email protected] Gatell, Jose María Hospital Clinic i Provincial de Barcelona Tel: +34 32275430; E-mail: [email protected] Gonzales, Matthew Stanford University Tel: +1 650 725 8826; E-mail: [email protected] Grant, Robert Gladstone/UCSF Tel: +1 415 695 3809; E-mail: [email protected] Greenberg, Michael Trimeris, Inc. Tel: +1 919 408 5183; E-mail: [email protected] Grossman, Zehava National HIV Reference Laboratory, PHL, MOH Tel: +972 3 530 2458; E-mail: [email protected] Grossman, Zvi Tel-Aviv University & NIH/NIAID Tel: +1 301 496 5046; E-mail: [email protected] Gulnik, Sergei Sequoia Pharmaceuticals, Inc. Tel: +1 240 632 0094, ext. 201; E-mail: [email protected] Günthard, Huldrych University Hospital Zurich Tel: +41 1 255 34 50; E-mail: [email protected] Götte, Matthias McGill University Tel: +1 514 340 8222 3299; E-mail: [email protected] Halima, Yasmin NAM Publications Tel: +44 20 7729 7525; E-mail: [email protected] XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S179 Hall, David Boehringer Ingelheim Tel: +1 2037984290; E-mail: [email protected] Halvas, Elias University of Pittsburgh Tel: +1 412 624 8509; E-mail: [email protected] Hance, Allan INSERM U552 Tel: +33 1 40 25 63 55; E-mail: [email protected] Haubrich, Richard University of California, San Diego Tel: +1 619 543 8080; E-mail: [email protected] Havlir, Diane University of California, San Francisco Tel: +1 415 476 4082; E-mail: [email protected] Hazuda, Daria Merck Tel: +1 215 652 7918; E-mail: [email protected] Heilek-Snyder, Gabrielle Roche Tel: +1 650 855 5468; E-mail: [email protected] Hendricks, David Bayer Health Care Tel: +1 510 705 5840; E-mail: [email protected] Heneine, Walid Centers for Disease Control and Prevention Tel: +1 404 639 0218; E-mail: [email protected] Hodder, Sally Bristol-Myers Squibb Tel: +1 609 897 3242; E-mail: [email protected] Hughes, Stephen NCI-Frederick Tel: +1 301 846 1619; E-mail: [email protected] Hunt, Peter University of California, San Francisco Tel: +1 415 221 4810 x3392; E-mail: [email protected] Jayaraman, Gayatri Health Canada Tel: +1 403 944 2493; E-mail: [email protected] Johnson, Victoria University of Alabama at Birmingham Tel: +1 205 934 4472; E-mail: [email protected] Kagan, Ron Quest Diagnostics Tel: +1 949 728 4018; E-mail: [email protected] Kantor, Rami Stanford University Tel: +1 650 736 0436; E-mail: [email protected] Katzenstein, David Stanford University Tel: +1 650 725 8304; [email protected] Kearney, Mary NCI-HIV-DRP Tel: +1 301 846 6796; E-mail: [email protected] Kempf, Dale Abbott Laboratories Tel: +1 847 937 0324; E-mail: [email protected] Keulen, Wilco Virology Education Tel: +31 30 230 7140; E-mail: [email protected] King, Martin Abbott Laboratories Tel: +1 847.938.3775; E-mail: [email protected] Kleim, Joerg Peter GlaxoSmithKline Tel: +44 20 8966 2270; E-mail: [email protected] Koelsch, Kersten University of California, San Diego Tel: +1 858 552 8585 x 2624; E-mail: [email protected] S180 Korber, Bette Los Alamos National Laboratory Tel: +1 505 665 4453; E-mail: [email protected] Korn, Klaus Institute of Clinical and Molecular Virology Tel: +49 9131 8522762; E-mail: [email protected] Kovari, Ladislau Wayne State University Tel: +1 313 993 1335; E-mail: [email protected] Kozal, Michael Yale University School of Medicine Tel: +1 203 688 2685; E-mail: [email protected] Kuritzkes, Daniel Brigham and Women's Hospital Tel: +1 617 768 8371; E-mail: [email protected] Lafeuillade, Alain General Hospital Tel: +33 494227741; E-mail: [email protected] Lanier, Randall GlaxoSmithKline Tel: +1 919 946 0194; E-mail: [email protected] Larder, Brendan RDI Tel: +44 0 223 264192; E-mail: [email protected] Lawrence, Jody University of California, San Francisco Tel: +1 415 514 0550 Ext. 402; E-mail: [email protected] LE Houillier, Guy Shire Pharmaceuticals Group Tel: +1 450 978 7860; E-mail: [email protected] Leigh Brown, Andrew University of Edinburgh Tel: +44 131 650 5523; E-mail: [email protected] Letendre, Scott University of California, San Diego Tel: +1 619 543 4730; E-mail: [email protected] Levin, Jules NATAP Tel: +1 212 219 0106; E-mail: [email protected] Liotta, Dennis Emory University Tel: +1 4047276602; E-mail: [email protected] Little, Susan University of California, San Diego Tel: +1 619 543 8080; E-mail: [email protected] Loeliger, Edde GlaxoSmithKline Tel: +44 20 8966 4707; E-mail: [email protected] Loveday, Clive International Clinical Virology Centre Tel: +44 01494 866717; E-mail: [email protected] Lucia, Palmisano Istituto Superiore di Sanità Tel: +39 06 49384021; E-mail: [email protected] Maldarelli, Frank HIV Drug Resistance Program NCI Tel: +1 301 435 8019; E-mail: [email protected] Mammano, Fabrizio INSERM Tel: +33 1 40256368; E-mail: [email protected] Martin-Carpenter, Louise GlaxoSmithKline Tel: +1 919 483 1861; E-mail: [email protected] Martinez-Picado, Javier irsiCaixa Foundation Tel: +34 93 4656374; E-mail: [email protected] Mascolini, Mark Freelance writer Tel: +1 570 588 0989; E-mail: [email protected] XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S181 Masquelier, Bernard Laboratoire de Virologie CHU de Bordeaux Tel: +33 5 56 79 55 10; E-mail: [email protected] Mayers, Douglas Boehringer Ingelheim Tel: +1 2037985052; E-mail: [email protected] McCallister, Scott Boehringer Ingelheim Tel: +1 203 798 5508; E-mail: [email protected] Mellors, John University of Pittsburgh Tel: +1 412 624 8512; E-mail: [email protected] Merigan, Thomas Stanford University School of Medicine Tel: +1 650 725 3925; E-mail: [email protected] Miller, Michael Gilead Sciences Tel: +1 650 522 5584; E-mail: [email protected] Miller, Michael Merck Research Laboratories Tel: +1 215 652 0480; E-mail: [email protected] Miller, Veronica Forum for Collaborative HIV Research Tel: +1 202 530 2311; E-mail: [email protected] Miralles, Gines Diego Trimeris Tel: +1 919 408 5046; E-mail: [email protected] Mo, Hongmei Abbott Laboratories Tel: +1 847 937 5933; E-mail: [email protected] Molla, Akhter Abbott Laboratories Tel: +1 847 938 1094; E-mail: akhter.m.molla@abbott .com Morales-Lopetegi, Kristina IrsiCaixa Foundation Tel: +34 934656374; E-mail: [email protected] Morgan, Elizabeth University of Birmingham/GSK Tel: +44 1438 763894; E-mail: [email protected] Moser, Michael EraGen Biosciences Tel: +1 6086629000; E-mail: [email protected] Najera, Rafael Instituto de Salud Carlos III Tel: +34 91 509 7937; E-mail: [email protected] Nijhuis, Monique Department of Virology, UMC Utrecht Tel: +31 (0)30 2506526; E-mail: [email protected] Nissley, Dwight V SAIC-Frederick; GRCBL, NCI-Frederick Tel: +1 301 846 1181; E-mail: [email protected] O'Brien, William A UTMB at Galveston Tel: +1 409 747 2361; E-mail: [email protected] Palmer, Sarah HIV DRP-NCI Tel: +1 301 846 5599; E-mail: [email protected] Parikh, Urvi University of Pittsburgh Tel: +1 412 624 8647; E-mail: [email protected] Parkin, Neil ViroLogic Tel: +1 650 866 7438; E-mail: [email protected] S182 Parniak, Michael University of Pittsburgh Tel: +1 412 648 1927; E-mail: [email protected] Patick, Amy Pfizer Global Research and Development Tel: +1 858 622 3117; E-mail: [email protected] Perno, Carlo Federico INMI Lazzaro Spallanzani IRCCS Tel: +39 0655170280/300; E-mail: [email protected] Perrin, Luc Hopital Canonal Universitaire Tel: +41 22 37 42 991; E-mail: [email protected] Pesano, Rick Pfizer Global Research and Development Tel: +1 858 622 7393; E-mail: [email protected] Petropoulos, Christos ViroLogic, Inc. Tel: +1 650 866 7439; E-mail: [email protected] Phillips, Rodney The Peter Medawar Building for Pathogen Research Tel: +44 1865 281230; E-mail: [email protected] Pillay, Deenan University College London Tel: +44 020 7679 9490; E-mail: [email protected] Quinones-Mateu, Miguel Cleveland Clinic Foundation Tel: +1 216 444 2515; E-mail: [email protected] Revell, Andrew RDI E-mail: [email protected] Rice, William Achillion Pharmaceuticals Tel: +1 203 624 7000; E-mail: [email protected] Richman, Douglas University of California, San Diego Tel: +1 858 552 7439; E-mail: [email protected] Riva, Chiara Institute of Infectious and Tropical Diseases Tel: +39 02 38200319/349; E-mail: [email protected] Rodes, Berta Hospital Carlos III-ISC III Tel: +34 91 453 2586; E-mail: [email protected] Ross, Lisa GlaxoSmithKline Tel: +1 919 483 6325; E-mail: [email protected] Routy, Jean-Pierre McGill University Health Centre Tel: +1 514 843 1558 ext. 35883; E-mail: [email protected] Rozenbaum, Willy Hopital Tenon-Paris VI University Tel: +33 156 01 74 12; E-mail: [email protected] Ruiz, Lidia IRSICAIXA Foundation Tel: +34 934656374; E-mail: [email protected] Rusconi, Stefano University. of Milan, Inst. Mal. Inf. & Trop. Tel: +39 02 39043350; E-mail: [email protected] Ryan, Kirk Bristol-Myers Suibb Tel: +1 609 897 3931; E-mail: [email protected] Salama, Carlos Elmhurst Hospital Center Tel: +1 718 334 2283; E-mail: [email protected] Schapiro, Jonathan Stanford University Tel: +1 972 9749 3942; E-mail: [email protected] XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S183 Schiffer, Celia University of Massachusetts Medical School Tel: +1 508 856 8008; E-mail: [email protected] Schinazi, Raymond Emory University/VA Medical Center Tel: +1 404 728 7711; E-mail: [email protected] Schmit, Jean-Claude CRP-Sante Tel: +352 44113091; E-mail: [email protected] Segal, Mark Center for Bioinformatics & Molecular Biostatistic Tel: +1 415 476 4553; E-mail: [email protected] Sévigny, Guy Pharmacor, Inc. Tel: +1 450 973 1710; E-mail: [email protected] Shafer, Robert Stanford University Tel: +1 650 725 2946; E-mail: [email protected] Shulman, Nancy CFAR, Stanford University Tel: +1 650 736 2442; E-mail: [email protected] Silva, Abelardo Sequoia Pharmaceuticals, Inc. Tel: +1 240 994 6270; E-mail: [email protected] Simon, Viviana Aaron Diamond AIDS Research Center Tel: +1 212 4485128; E-mail: [email protected] Sista, Prakash Trimeris, Inc. Tel: +1 919 408 5030; E-mail: [email protected] Sluis-Cremer, Nicolas University of Pittsburgh Tel: +1 412 383 8525; E-mail: [email protected] Smith, Anthony University of Miami School of Medicine Tel: +1 305 667 1258; E-mail: [email protected] Smith, Davey University of California, San Diego Tel: +1 858.552 8588 x 2624; E-mail: [email protected] Snoeck, Joke Rega Institute for Medical Research Tel: +32 16332160; [email protected] Soares, Marcelo Universidade Federal do Rio de Janeiro Tel: +55 21 2562 6384; E-mail: [email protected] Soriano, Vincent Instituto de Salud Carlos III Tel: +34 91 4532500; E-mail: [email protected] Stanfield-Oakley, Sherry Trimeris, Inc. Tel: +1 919 408 5187; E-mail: [email protected] Su, Cheng Roche Palo Alto Tel: +1 650 855 6510; E-mail: [email protected] Sugiura, Wataru National Institute of Infectious Diseases Tel: +81 42 561 0771; E-mail: [email protected] Tamalet, Catherine CHU Timone Laboratoire de Virologie Tel: +33 0491385522; E-mail: [email protected] Tirado, Grissell Ponce School of Medicine Tel: +1 787 984 2881; E-mail: [email protected] Tobin, Nicole University of Washington Tel: +1 206 987 3478; E-mail: [email protected] S184 Tozzi, Valerio INMI L. Spallanzani Tel: +39 06 55170420; E-mail: [email protected] Trotta, Maria Paola INMI Lazzaro Spallanzani IRCCS Tel: +39 0655170280/300; E-mail: [email protected] Turner, Dan McGill University AIDS Centre Tel: +1 514 340 7536; E-mail: [email protected] Ussery, Michael NIAID Tel: +1 301 402 0134; E-mail: [email protected] van de Vijver, David Dept. of Virology Tel: +31 30 250 6526; E-mail: [email protected] Van Houtte, Margriet Virco BVBA Tel: +32 15 401 243; E-mail: [email protected] Van Laethem, Kristel Rega Institute for Medical Research Tel: +32 016/332160; E-mail: [email protected] Van Marck, Herwig Tibotec-Virco Tel: +32 15 293 107; E-mail: [email protected] Van Rompay, Koen California National Primate Research Center Tel: +1 530 752 5281; E-mail: [email protected] Vandamme, Anne-Mieke Rega Institute for Medical Research Tel: +32 16 332160; E-mail: [email protected] Vingerhoets, Johan Tibotec Tel: +32 15 286341; E-mail: [email protected] Wainberg, Mark McGill University AIDS Centre Tel: +1 514 340 7536; E-mail: [email protected] Walter, Hauke Institute of Clinical and Molecular Virology Tel: +49 9131 8522762; E-mail: [email protected] Wensing, Annemarie University Medical Center Utrecht Tel: +31 30 2506526; E-mail: [email protected] Westby, Mike Pfizer Global Research and Development Tel: +44 1304 649876; E-mail: [email protected] Wiegand, Ann HIV DRP-NCI Tel: +1 301 846 6266; E-mail: [email protected] Winston, Alan Chelsea and Westminster Hospital Tel: +44 208 746 8000, x1581; E-mail: [email protected] Winters, Mark Stanford University Tel: +1 650 723 5715; E-mail: [email protected] Womack, Chad NIH Vaccine Research Center Tel: +1 301 594 8555; E-mail: [email protected] Wong, Joseph University of California, San Diego Tel: +1 858 552 7439; E-mail: [email protected] XII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications S185 Yeni, Patrick Hopital Bichat Tel: +33 1 40257807; E-mail: [email protected] Ziermann, Rainer Bayer Healthcare Tel: +1 510 705 5843; E-mail: [email protected] Zolopa, Andrew Stanford University Tel: +1 650 498 5881; E-mail: [email protected] S186