• Introduction
• Part 1: Clinical Trials in Salvage and First Treatment
• Part 2: Progress in Resistance Testing Technologies and Interpretation
• Part 3: Proficiency Testing Programs
• Key Points
• Related Articles
• References

At the 8^th Conference on Retroviruses and Opportunistic Infections (8^th CROI), data were presented from two studies evaluating the clinical utility of resistance testing in managing patients with virologic rebound on antiretroviral therapy (abstracts 433 and 434), adding to existing information from studies such as GART, VIRADAPT and VIRA 3001 (see related articles below) [1,2]. There is a substantial and growing incidence of transmitted drug resistance seen in untreated populations and guidelines suggest that resistance testing be considered in these circumstances [3], although the clinical relevance of such resistance has not been defined. Retrospective data evaluating the clinical utility of resistance testing in untreated patients were presented (abstracts 436, 439) [4,5].

While clinical trials data continue to suggest that there is some utility to using resistance tests in the management of people with HIV infection, the utility of these tests is limited in clinical practice by their poor sensitivity at low HIV RNA copy numbers and difficulty in detecting sub-populations of resistant quasispecies that make up less than 20% of the overall population. Reports on the developments of technologies to improve sensitivity were presented and will be described in Part 2 of this report. This is a crucial area of development as current technologies only enable these tests to be performed at HIV RNA levels of 500 to 1,000 copies/mL, yet resistance can develop at much lower copy numbers. Improvements in sensitivity will allow these tests to be useful in directing the clinical management of patients with virologic rebound at low copy numbers.

Interpretation of phenotypic resistance test results has been limited by a lack of understanding about the relationship between fold resistance in vitro and the activity of a given drug in vivo. Both companies providing phenotypic assays (ViroLogic, Virco) have been developing clinically relevant cutoffs; data were presented on such cutoffs for abacavir (abstract 254) [6]. Using combined clinical trials data from abacavir studies, scientists from ViroLogic and Glaxo-SmithKline were able to define a 4.5–fold increase in IC50 as the "clinically relevant" cutoff for abacavir using the PhenoSense assay. These data will be described in more detail in Part 2 of this report. Scientists from Virco presented data on their virtual phenotype technology (abstract 524), which will also be described in Part 2 of this report [7].

Many commercial laboratories and academic centers currently offer resistance tests. No products on the market in the US are currently FDA approved, and many are not regulated by the FDA at all. Laboratory proficiency testing is a standard for many assays and is becoming increasingly available to providers of resistance tests. Part 2 of this report will include descriptions and discussions of data on proficiency testing programs. In addition, specific, up-to-date information regarding individual laboratory’s participation in proficiency testing programs will be included in our improved Test Information pages, coming soon.

Part 1: Clinical Trials in Salvage and First Treatment
(Please note that the following are REPORTS on presentations at the 8^th CROI and not conference abstracts.)

Abstract 433: The ARGENTA Study of the Clinical Utility of Genotyping and Adherence Assessment in Salvage Treatment [1]

Methods: ARGENTA was a randomized trial of genotyping using Visible Genetics' (VGI) TruGene kit vs. standard of care, with treatment decisions proposed by a panel in both arms, in antiretroviral experienced patients with virologic failure. The endpoint of change in HIV RNA was measured at 6 months after the change in antiretroviral therapy.

Results: 177 patients were randomized to the two arms. The mean baseline CD4 count was 265 and the mean baseline HIV RNA level was 4.25 log₁₀ copies/mL. 47% of subjects were failing their first, 28% their second and 25% their third antiretroviral regimen. 41% were NNRTI-experienced. There was a mean of 7 genotypic mutations at baseline (7 in the standard of care arm and 8 in the genotype arm). There was a statistically significant difference in the proportion of patients reaching a plasma HIV RNA level < 500 at 3 months (12% vs. 27%; P = 0.02) that was lost by 6 months (17% vs. 21%; P = 0.10). Logistic regression models showed the following odds rations for predicting a plasma HIV RNA of < 500 at 3 months: genotyping OR, 2.6; history of viral load < 500, OR, 2.9; failing 1^st or second regimen OR, 3.2. Less than 95% adherence was associated with an OR of 0.4 for virologic success. By 6 months, the genotyping was no longer predictive of a successful virologic response.

Conclusions: The authors concluded that genotypic resistance testing to guide salvage therapy if most useful in those with lower plasma HIV RNA levels at the time of failure and greater self-reported adherence. It is also more useful in those who are less antiretroviral experienced and thus have more options for salvage therapy.

Abstract 434: The HAVANA Study of the Clinical Utility of Genotyping and Expert Advice in Salvage Treatment[2]

Methods: HAVANA was a double randomized trial of (1) genotyping using the VGI TruGene kit vs. standard of care and (2) expert advise vs. no expert advise in antiretroviral experienced patients with virologic failure. The endpoint of change in HIV RNA was measured at 6 months after the change in antiretroviral therapy.

Results: 326 patients were randomized (each of the groups had 12-42 subjects). The mean baseline CD4 count was 388 and the mean baseline HIV RNA level was 4.04 log10 copies/mL. 54% had used 2 or more PI-containing regimens. There was a statistically significant difference in the proportion of patients reaching a plasma HIV RNA level < 400 at 6 months between both the genotype vs. no genotype groups (58% vs. 42%; P = 0.01) and the expert advise vs. no expert advice groups 59% vs. 41%; P = 0.003). The group that received both genotyping and expert advise had a statistically significant virologic advantage (69% vs. 36%; P = 0.001) over the group that received neither. A multivariable model showed the following odds ratios associated with plasma HIV RNA levels < 400 copies: genotyping OR, 1.9; expert advise OR, 2.1; failure of third regimen OR, 0.3.

Conclusions: Both genotyping and expert advise were associated with superior 6 month virologic responses in this population, with those who were less antiretroviral experienced deriving the greatest benefit.

Abstract 436: Pre-Treatment Genotype Testing and an Unexpected Superior Response to First Treatment Regimen in a US Military Population with Resistant HIV[4]

Methods: A retrospective analysis of pre-treatment genotypes (using Virco’s Vircogen I assay) and 6 months of plasma HIV RNA follow-up test results.

Results: 130 seroconverters participated in a natural history study between February 1997 and June 1999; 6 month follow-up data were available for 103 of these, 80 of whom initiated HAART. 21 subjects (20%) had at least one mutation in RT plus or minus protease mutations. There were no significant differences between those with and without resistance in baseline CD4 counts, plasma HIV RNA levels, nor initial ARV therapy. Those with resistance mutations present at baseline showed greater increases in CD4 counts over 6 months ( 230 vs. 122; P < 0.05) and greater decreases in HIV RNA levels (-3.0 log₁₀ copies/mL vs. -2.1 log₁₀ copies/mL; P < 0.02). Among those on 3 or more ARV drugs, there was a trend towards increased HIV RNA declines in the resistance groups that did not reach significance.

Conclusions: The authors explain these unexpected results with the hypotheses that transmitted resistant HIV may be either be less fit than wild type virus, or show hypersensitivity to other drugs.

Abstract 439: Baseline Genotypic Resistance Does Not Predict Response to Therapy in Chronically Infected Untreated Populations[5]

Methods: Baseline genotypes were performed in a randomized trial of various NRTI combinations plus indinavir (START 1 and 2 trials) and virologic outcome over 48 weeks were compared between those with and without baseline resistance as evaluated by the RCG data analysis plan (DAP).

Results: 358 of 409 subjects had baseline genotypes completed. 3.4% had any mutation in RT, 2.5% had an NNRTI-associated mutation and there were no primary protease mutations. There was no association between virologic response and resistance present at the time of treatment initiation. Lower baseline viral load was associated with superior virologic response (P = 0.03).

Conclusions: The authors conclude that routine resistance testing should not be used prior to initiating therapy in chronically infected untreated patients.

Discussion
These four studies add to a growing yet limited body of literature and presented reports addressing the clinical utility of resistance testing. In the salvage setting, resistance testing utility is limited by available therapeutic options and adherence, and virologic benefit may be transient. There are still no data available from prospective randomized trials of the utility of resistance testing in treatment-naive patients. Surprisingly, neither of the two retrospective studies described here suggest that pre-therapy resistance in these populations is associated with poorer response to ARV therapy.

Descriptions and discussions of presentations at CROI concerning (1) progress in resistance testing technologies and interpretation and (2) proficiency testing programs will follow in Part 2 of this report.

Part 2: Progress in Resistance Testing Technologies and Interpretation

Abstract 249: Validation of the Ultrasensitive HIV-1 Genotyping Assay [8]

Methods: HIV RNA extraction using modified Qiagen technique followed by genotyping using Visible Genetics' TRUGENE kit.

Results: 77 of 84 samples with fewer than 1,000 copies/mL of HIV-1 RNA were successfully genotyped, with the lowest viral load being 116 copies/mL. 80% had resistance mutations. Five samples were tested in triplicate; the results were 100% reproducible. None of the 7 samples with fewer than 100 copies/mL were successfully genotyped.

Conclusions: This RNA extraction technique appears to be useful for improving the sensitivity of the TRUGENE assay to below 1,000 copies/mL and may be useful with other genotyping assays.

Discussion: Researchers have raised the point that If someone's viral load is below 1000 copies/mL, why is there a need to know the resistance pattern? Perhaps drug resistance detected in this setting will predict frank virological failure or will predict the outcome of intensification. These questions, though, are quite open.

Abstract 524: The Virtual Phenotype is an Independent Predictor of Clinical Response [7]

Methods: Retrospective analysis of 191 baseline genotypes interpreted using the resistance collaborative group data analysis plan (DAP), phenotypes (Antivirogram), Virtual Phenotypes derived from a relational database, and HIV-1 RNA response at 16 weeks in a prospective study of the utility of phenotypic resistance testing (VIRA 3001).

Results: In analyses where drop-outs were censored, the Virtual Phenotype provided a more significant predictor of virologic response than the genotype in both univariate and multivariate models (adjusted for baseline RNA and number of new drugs) of failure to achieve an HIV-1 RNA level of less than 400 copies/mL. This difference was seen when using either an arbitrary cutoff of 4-fold reduced susceptibility or drug-specific, biologically defined cutoffs for phenotypic resistance. Phenotype was also a stronger predictor of response than genotype in the multivariate model using a 400 copy/mL cutoff for HIV-1 RNA, but only the Virtual Phenotype was a significant predictor when a 50 copy/mL cutoff was used. When the analyses were performed with dropouts considered failures, the trends continued in the same direction, but were less significant. An analysis of the positive predictive value showed that both phenotype and Virtual Phenotype were better than genotype, but there was no significant difference between the two.

Conclusions: The Virtual Phenotype interpretation of genotypic tests may be superior to the rules-based DAP system.

Part 3: Proficiency Testing Programs

Abstract 251: Worldwide Evaluation of Genotypic Resistance Analysis Using Clinical HIV-1 Isolates with Complex Drug Resistance Profiles [9]

Methods: Seven reference sequences derived from clinical isolates with complex resistance patterns were separately tested by 7 and 9 independent laboratories over two test periods.

Results: Errors at the nucleotide level were seen in 0.04% to 7.78% of the tests. Errors at the amino acid level were seen in 0 to 1.71%. The most common error was a lab indicating a mixture when the nucleotide consensus sequence indicated a single nucleotide. There was a very high concordance between labs (kappa > 0.99).

Conclusions: The authors concluded that such proficiency testing should become routine clinical practice.

Abstract 252: A Quality Control Trial for HIV-1 Drug Resistance Testing Using Clinical Samples [10]

Methods: Four reference sequences derived from diluted plasma samples were tested by 20 independent laboratories in 5 countries. 15 used commercial systems (10 Applied Biosystems and 5 Visible Genetics) and 5 used in-house systems. 22 sets of data were analyzed, including both test results and interpretations.

Results: The detection rate of key mutations was approximately 90%. The interpretation of mutations led to even more heterogeneous results, especially for the protease inhibitors. Two data sets missed the K103N mutation. One missed T215Y, M184V, Q151M and others. Minority quasi-species presented a significant problem.

Conclusions: The authors conclude that the biggest technical difficulty remains detection of resistance in minority quasispecies, but that consistent interpretation of results is an even bigger problem.

Key Points

1. While the clinical trials data continue to suggest that there is some utility to using resistance tests in the management of people with HIV infection, the utility of these tests is limited in clinical practice by their poor sensitivity at low HIV RNA copy numbers and difficulty in detecting resistant quasispecies sub-populations making up less than 20% of the overall population.
2. Phenotypic resistance testing interpretation has been limited by a lack of understanding about the relationship between fold resistance and response to the drug in vivo.
3. Laboratory proficiency testing is a standard for many assays and is becoming increasingly available to providers of resistance tests.

Interpretive Systems for HIV Drug Resistance Testing: A Report From the 5th International Workshop on HIV Drug Resistance and Treatment Strategies
(Robert Shafer, July 26, 2001)

Interlaboratory Variability in HIV Drug Resistance Testing
(Robert Shafer, July 13, 2001)

The Impact of Genotypic Antiretroviral Resistance Testing (GART) in Patients Failing Antiretroviral Therapy: CPCRA 046 (The GART Study)
(Michelle Roland, April 1, 1999)

Improving HIV Therapy with Drug Resistance Genotyping: The VIRADAPT Study
(Jonathan Schapiro, February 25, 1999)

References