Review of phenotypic resistance testing data presented at the XI International HIV Drug Resistance Workshop

By Mike Youle MD, for

The measurement of resistance to HIV has sped on in leaps and bounds, and with each new conference comes technologic improvements, novel technologies or modification of existing ones as well as more information on how to maximize the utility of these tests.

At the XI International HIV Drug Resistance Workshop (July 2-5, 2002, Seville, Spain) there was a wide application of phenotypic resistance tests and the derived replication capacity measure, some of which are discussed below.

Please note that unless otherwise indicated, all references are to the Program and Abstracts of the XI International HIV Drug Resistance Workshop, July 2-5, 2002, Seville, Spain. All abstracts are published in Antiviral Therapy Volume 7(2) and are available on-line.

Replication capacity

A novel replication capacity assay has been developed as a modification of the PhenoSense phenotypic resistance assay in which the capacity to replicate of the patient viral strain is expressed as a percent of that of a wild-type reference virus (NL4-3).

In a prospective study of phenotypic susceptibility testing (CCTG 575), 97 of 207 patients who failed to suppress HIV RNA at month 6, showed a significant linear correlation between lower RC and greater month 6 HIV RNA reduction (r=0.34, P=0.001) [1].

Although the baseline HIV RNA was the same in each group, patients with lower RC (<35%, n=49) had a mean HIV RNA change of  –0.54 log10 at month 6 compared to +0.08 for those with higher RC (n=48, p=0.0003).  In a multivariate model with RC, baseline HIV RNA, CD4, and the number of NRTI, PI and NNRTI to which the virus was susceptible, only the baseline HIV RNA (p=0.004) and RC (p=0.0002) were independent predictors of month 6 HIV RNA change.

In another study clinical and laboratory data from 12 studies of RC in over 800 clinical viral isolates (N>500 patients) were evaluated [2].

Data across studies were linked to define a continuous model of viral/clinical evolution during the course of HIV infection and antiretroviral treatment. During early HIV infection, wild-type virus, spanning a broad range of RC (<10% to >100%) usually predominates.  Viruses with lower RC during early infection are associated with higher CD4 cell counts, suggesting a slower rate of immune depletion.

Upon initiation of therapy, RC effects appear to be obscured by the magnitude of the antiviral drug effects.  After onset of treatment failure associated with emergence of drug resistance, the RC often declines precipitously, especially in the setting of protease inhibitor resistance but less so in the face of isolated NRTI or NNRTI resistance; and the RC drop is directly correlated with the magnitude of viral load suppression and CD4 cell increase from the pre-treatment baseline.

During prolonged virologic failure, RC tends to remain relatively stable despite a slow, progressive increase in drug resistance and viral load.  The presence of virus with higher RC is associated with CD4 cell decline, and a lower RC with CD4 increases.

If treatment is interrupted, wild-type virus with higher RC usually re-emerges over 2-12 weeks, in association with abrupt increases in viral load and declines in CD4 cell counts toward pre-treatment baselines.


There remain many unresolved issues concerning the transmission of HIV-resistance at primary infection. The ViroLogic PhenoSense HIV assay was used to assess the prevalence of transmitted drug resistance across three study periods: A=1995-1998 (n=264), B=1999-May, 2000 (n=123), and C=June 2000-March 2002 (n=122) [3].

Baseline characteristics did not differ over time and the proportion of subjects with an IC50 >10 times the reference virus (NL4-3) to one or more ARV drugs did not change significantly between period B and C (12.2% vs. 6.6%, p=0.13). This compared to an increase from A and B (3.4% vs. 12.2%, p=0.001.

Resistance to non-nucleoside RTI (NNRTI) from period B to period C (7.3% vs. 6.6%, p=1.00) was stable. However resistance to nucleoside RT inhibitors (NRTI) (5.7% to 0.8%, p=0.06), PI’s (8.1% to 0%, p=0.002) and multi-drug resistance (an IC50 >10 fold to drugs in 2 or more classes) decreased significantly between these same study periods (6.5% to 0.8%, p=0.04).

To assess the effect of transmitted drug resistance 130 persons recently infected with HIV-1 were identified in the Options Project, a study of primary HIV-1 infection based in San Francisco [4].

Drug resistance was assessed both phenotypically (PhenoSense HIV) and genotypically (TruGene HIV). CD4 T-cell counts were significantly higher in subjects with genotypic evidence of drug resistance (P=0.02) or decreased replication capacity (P=0.01).  These associations persisted after controlling for duration of infection.

RC was significantly lower in subjects with genotypic PI resistance compared with wild-type (median RC 24% vs. 41%; P=0.04) and ranged widely from 1% to 113%.  Hypersusceptibility to protease inhibitors (defined as IC50 fold change < 0.4 for any PI), was observed in 17.6% and correlated with lower replication capacity (eg: IDV IC50 fold change vs RC; Rho 0.52; P=<0.001).

The wide variation in replication capacity of recombinant viruses containing PR-RT segments derived from recently transmitted HIV-1 may reflect the diversity of quasispecies capable of establishing new infections, or bottlenecks in the virus population as a result of poorly understood selective pressures exerted during virus transmission.

Antiretroviral resistance

To assess the utility of phenotyping resistance to protease inhibitors PR genotype (GT) and phenotype (PT) for 1418 patient samples with at least one primary PR mutation or one PI with reduced susceptibility (fold change [FC]>2) were analyzed [5].

Samples were classified as LPV resistant by GT (GT-R) if six or more LPV mutations were present, and by PT (PT-R) if the FC was over 10.  APV GT-R was defined as presence of any of primary APV mutation, and PT-R was defined as FC > 2.5.

Using a published LPV algorithm, 182 samples classified as GT-S (13%) actually tested PT-R. A comparison of the prevalence of PR mutations in the discordant GT-S/PT-R samples and the concordant GT-S/PT-S samples (n=489) identified several new mutations that are significantly associated with LPV PT-R. This finding has implications for the use of LPV-based salvage therapy after failure of APV-based regimens, or vice versa.

In a further analysis of amprenavir susceptibility data from >3000 clinical samples displaying evidence of reduced PI susceptibility by phenotype (PT) or genotype (GT) were examined to establish a clinical cut-off for APV resistance [6]. Since this is currently unknown, a conservative estimate of 2.5-fold increase in IC50 was used to define samples with reduced APV susceptibility (PT-R).

Samples were defined as resistant by GT (GT-R) if any of the following mutations were present: V32I, I50V, I54L or M, or I84V. One quarter of the initial genotypic interpretations were discordant with the observed phenotypic results. Data analysis identified novel combinations of these and other mutations that added significant predictive information, including L33F/V82A, V82F/L90M, M46I/I47V, and M46I/V82F.

It would seem that reduced susceptibility to APV can develop via genetic pathways that do not include APV-selected mutation and that using APV PT data to define mutations associated with cross-resistance leads to an improved ability to predict reduced susceptibility from genotypic data.

To assess the impact of IDV resistance on responses to RTV-enhanced regimens in patients who had had virologic failure on multiple protease inhibitors 173 patients were identified from three observational cohorts and one prospective clinical study on a variety of indinavir/ritonavir regimens [7].

A meta-analysis of these populations was performed to assess the relationship between baseline IDV fold change (PhenoSense assay) and subsequent virologic responses to the IDV-RTV regimen. Phenotypic susceptibility at baseline correlated inversely with clinical response to IDV-RTV based therapy.  The clinical cutoff for IDV-RTV 800/200 is at least five to ten fold.

More sophisticated statistical analyses (regression and/or CART) will be required to define a precise clinical cutoff.

To further describe phenotypic (PT) and genotypic (GT) patterns associated with (atazanavir) ATV resistance PT (PhenoSenseTM) and GT (GeneSeqTM) evaluations were performed on baseline and post-treatment isolates from 76 patients treated with ATV and classified as treatment failures in studies AI424-007, -008 and –009 [8].

Seventeen of the post-treatment (24 to 104 wk) isolates from patients designated as treatment failures displayed decreased susceptibilities to ATV ranging from five to 141-fold. Recombinant viruses containing I50L and the I50L/A71V combination displayed decreased susceptibility to ATV and were significantly growth impaired. PT and GT analysis of isolates from patients treated with regimens containing ATV suggest that the emergence of the amino acid substitutions I50L and A71V in treatment naïve patients may result in selective resistance to ATV.

With the Antivirogram, a tenofovir phenotypic cut-off of four-fold was previously established and this corresponded to a week 24 HIV RNA response to TDF therapy of only –0.24 log10 copies/mL [9].

Using study 907 a phase III study of TDF when added to stable background antiretroviral therapy in treatment-experienced patients with extensive resistance mutations.  Baseline HIV phenotypes were obtained for 112 randomly selected patients with >500 HIV RNA copies/mL at baseline and outcomes assessed at week 24 to establish clinical cut-off using the PhenoSense HIV assay.

The mean week 24 response was –0.63 log10 copies/mL (range –2.03 to +1.16) and using categorical response variables the biological/assay cut-off for tenofovir was 1.4-fold, i.e. 99% of wild-type viruses from treatment-naive patients fall within 1.4-fold of the drug sensitive reference virus.   There appeared to be two clinical phenotypic cut-offs for tenofovir DF in this assay, with the 1.4-fold level corresponding to a reduced response to TDF therapy (23% of patients fell above this cutoff) and a second cut-off of four-fold corresponding to no clinically significant response (7% of patients).

In ACTG 364, a randomized clinical trial in highly nucleoside experienced subjects, 131 subjects were treated with NFV as their first PI [10]. Virologic failure (VF) occurred in 48 (37%) NFV-treated subjects and 38/66 (58%) and 10/64 (16%) subjects randomized to the NFV+nRTIs and NFV+EFV+nRTIs arms, respectively.

Genotype (Stanford ABI) and phenotype were retrospectively determined from plasma at study entry (n=86) and initial failure (n=39 genotype, n=32) both or continued failure (n=22). Failure was frequently associated with reduced susceptibility to NFV and the emergence of ‘secondary’ mutations.

Continued NFV use after VF selected for progressively increasing NFV resistance in a majority of subjects. Cross-resistance to other protease inhibitors, however, was infrequent and low-level in these NFV-treated VFs.

A Spanish group presented another study, which compared phenotype with virtual phenotype. They randomised 300 subjects to one or the other and showed that the virtual phenotype gave a slightly better ability to detect resistance using the virtual phenotype with some drugs [11]. However this may have been related to the choice of cut-off compared to the constantly changing algorithm of the virtual phenotype. Overall no significant differences in outcome were seen.

Treatment selection patterns appear to be influenced by availability of information. In the CPCRA 064 study of structured treatment interruption followed by a salvage regimen 245 patients were enrolled and their physicians were asked to decide on a salvage regimen based on initially genotype only and then later in the STI on genotype plus phenotype (VIRCO) [12].

Seventy two percent changed their assessment when the additional information was available.

A new class of compounds requires a new set of resistance assays and Bristol Myers Squibb Research Institute has developed an entry inhibitor assay to parallel their development program of BMS-806 [13]. It is based on generation of transcriptionally active fragments of HIV envelope and a viral envelope mediated fusion assay. This results in a rapid (around five days) relatively simple assay for this new group of drugs.

In summary, there was a plethora of information on phenotypic assays of various types at the Seville Resistance Meeting, which moved the field forward to allow better monitoring, and evaluation of HIV drug resistance.


Unless otherwise indicated, all references are to the Program and Abstracts of the XI International HIV Drug Resistance Workshop. July 2-5, 2002. Seville, Spain.

  1. R Haubrich, T Wrin, N Hellmann, et al. Replication Capacity (RC) as a Predictor of Immunologic and Virologic Benefit Despite Virologic Failure of an Antiretroviral Regimen. Seville Abs 121
  2. NS Hellmann, T Wrin, M Bates, S Deeks et al. Modeling the Effect of HIV Replication Capacity on Treatment Outcomes. Seville Abs 63
  3. SJ Little, S Holte, JP Routy et al. Longitudinal Analysis of Transmitted Drug Resistance among Recently HIV Infected Subjects in North America. Seville Abs 174
  4. RM Grant, JD Barbour, T Wrin et al. Transmission of Drug Resistant HIV-1 Exhibiting Lower Replication Capacity is Associated with Higher CD4 Cell Counts. Seville Abs 46
  5. NT Parkin, C Chappey, and CJ Petropoulos. Mutations in HIV-1 Protease Associated with Resistance to Amprenavir Contribute towards Phenotypic Resistance to Lopinavir. Seville Abs 24
  6. Nidtha, C Chappey, C Petropoulos et al. Analysis of HIV-1 Susceptibility to Amprenavir: Phenotypic Data Improves the Predictive Value of Genotype S, and NT Parkin. Seville Abs 116
  7. J. Szumiloski, H. Wilson, E. Jensen et al. Relationship between Phenotypic Susceptibility to Indinavir and Virologic Response to Indinavir-Ritonavir-Containing Regimens Following Failure of a Previous Protease Inhibitor. Seville Abs 155
  8. RJ Colonno, J Friborg, RE Rose, et al. Identification of Amino Acid Substitutions Correlated with Reduced Atazanavir Susceptibility in Patients Treated with Atazanavir Containing Regimens. Seville Abs 4
  9. B Lu, NS Hellmann, M Bates et al. Determination of Clinical Cut-Offs for Reduced Response to Tenofovir DF Therapy in Antiretroviral-Experienced Patients. Seville Abs 67
  10. D Katzenstein, Gerald Downey, Ronald Bosch et al. Evolution of Protease Phenotype and Genotype Changes at Initial and Continued Virologic Failure among Nucleoside-Experienced Subjects Receiving Nelfinavir (NFV) in ACTG 364. Seville Abs144
  11. M Perez-Elias, I Garcia-Arata, S Moreno et al. Baseline Testing information given by a real phenotype (real Ph) or a Virtual Phenotype (Virtual Ph) test in a Randomised Study (Realvirfen study):influence in final outcome. Seville Abs 109
  12. J Lawrence, D Mayers, Huppler Hullsiek et al. Simultaneous Genotypic and Phenotypic HIV Resistance Testing and Influence on Treatment Selection in the CPCRA 064 (MDR) Study. Seville Abs 103
  13. R Rose, R Fridell, J Fang et al. A Novel and Rapid Phenotypic Assay for Monitoring Viral Susceptibility to Entry Inhibitors. Seville Abs 104.

Copyright 2002 by HIV and All Rights Reserved.

Links to other websites are current at date of posting but not maintained.