The FDA held a two-day workshop as part of its Antiviral Drugs Advisory Committee meeting on November 2 and 3, 1999 to consider requirements for the characterization of genotypic and phenotypic resistance patterns during new drug development. In addition, evaluation of the association between resistance mutations, phenotypic sensitivity and differences in virologic responses in antiretroviral-naive and -experienced patient populations may be required for a manufacturer to claim activity against drug-resistant virus. Some of the specific scenarios considered during the workshop included questions about what data would be required to support claims of slow development of resistance to a drug, lack of cross-resistance, and activity against drug-resistant virus.

These policy changes were proposed in an attempt to assist clinicians and antiretroviral-experienced patients by including resistance and cross-resistance data in the product label, and to ensure that pharmaceutical promotions are based on the claims for which the drug is approved, as reflected in the label. This effort is being undertaken by the FDA to avoid the difficult transition that viral load testing experienced as it moved from research into clinical practice, and became an important surrogate marker in new drug development, by proactively addressing the potential clinical uses of the rapidly evolving technologies used to measure and qualify drug resistance. Major contributors to the workshop included members of the Resistance Collaborative Group (RCG), a group of academicians, industry and government staff from the United States and Europe who have been working on several important resistance initiatives addressing standardization and clinical validation of HIV drug resistance testing methodologies in the past year.

Some new data that have not been presented elsewhere (including data from an interim analysis of the first prospective study evaluating the clinical utility of phenotypic testing), a large amount of re-analyzed existing data (including data from retrospective and prospective studies evaluating the predictive value of baseline mutations or reduced phenotypic susceptibility and subsequent virologic response to salvage antiretroviral regimens and a re-analysis of the published or presented studies describing the prevalence of resistance mutations or reduced phenotypic susceptibility in untreated patient populations) were presented to the committee for its consideration in addressing the first set of question. These questions included:

1. What are the limitations of current data correlating genotypic or phenotypic resistance with clinical outcome and what studies are needed to resolve these partially-addressed questions?
2. What are the relative strengths and limitations of genotypic and phenotypic assays and what are the potential roles of these two assays at various stages of new drug development (pre-clinical, early clinical (phase I/II), later clinical (phase III) and post-marketing (phase IV) studies)?
3. How well defined are performance characteristics (sensitivity, specificity, accuracy, reproducibility, etc) of resistance assays and what further studies and/or quality control measures should be developed to ensure these performance characteristics?
4. How should limitations in the understanding of many genotypic mutations and the lack of agreement about clinically relevant phenotypic cutoffs or breakpoints, and the resulting lack of consensus in these areas, be handled and resolved when using resistance assays in the drug development process?

The RCG re-analysis of existing data sets looked at 8 retrospective and 2 prospective studies with a standardized 24-week virologic outcome. Genotypic and/or phenotypic sensitivity scores were developed based on the following point system: resistant = 0 and sensitive = 1, with some minor exceptions. For the phenotyping studies, analyses were performed based on both the assay cut-off and 10-fold reduced susceptibility. Multiple logistic regression models were performed based on a number of confounding factors (e.g., baseline viral load, CD4+ T-cell count, PI and NNRTI history, number of new drugs, etc). The bottom-line result was that most but not all of the analyses demonstrated an association between the sensitivity scores and virologic response.

Interim analysis of the VIRA 3001 study evaluating virologic response in treatment-experienced patients randomized to receive a phenotypic resistance test (N=141) or no phenotypic resistance test (N=13) prior to switching therapy showed a statistically significant difference in reduction in HIV-1 RNA at 16 weeks (P=0.01). The proportion of patients achieving an undetectable HIV-1 RNA level at 16 weeks was 62% in the phenotyping arm vs 33% in the control arm (both arms showed larger proportions achieving this end-point than seen in the VIRADAPT or GART studies). This is the first prospective study to look at the use of phenotypic tests in salvage situations.

The second set of questions to the committee addressed the implementation of the use of resistance testing in new drug development. These questions included:

1. What types and amounts of data should be required at each stage of pre-clinical and clinical development in each of two scenarios: a company seeking an indication for the treatment of drug-resistant virus, and a company not seeking such a specific indication?
2. In what patient populations should resistance characterization be required, e.g., antiretroviral-naive patients, patients experiencing virologic rebound on their first regimen, and/or the very antiretroviral-experienced?
3. Should resistance tests be used as part of inclusion or exclusion criteria in some clinical trial situations, e.g., in antiretroviral-experienced populations when some level of certainty exists about cross-resistance that is likely to confer reduced efficacy to the experimental agent?
4. When should resistance testing be performed (e.g., in real time or batched) and when should samples be collected (e.g., at baseline and/or with virologic rebound)? If real time resistance testing in clinical trials is desirable, is it feasible?
5. What are the important variables likely to affect antiviral response and how should they be factored in to resistance characterization studies in the drug development process, e.g., adherence and pharmacologic data?

Many specific recommendations were provided by the committee to the FDA in response to these questions. The committee was enthusiastically in support of this process, which was viewed as a turning point in the use of resistance evaluation in new drug development analogous to the use of viral load testing in clinical trials. "We can't let excellence be the enemy of good," Doug Richman concluded after presenting a daunting list of unanswered questions and issues that need to be addressed. This list included (1) the need to evaluate the specificity of each mutation or susceptibility result at every HIV RNA level and for minority species comprising very small proportions of the viral swarm, (2) evaluation and characterization of potential interfering substances in the various assays, (3) evaluation of assays' performance characteristics with non-clade B viruses and HIV-2, as well as in a variety of specimen tissue types (e.g., plasma and genital secretions), (4) investigation of potential cellular (non-viral) resistance mechanisms (e.g., p-glycoprotein efflux pumps and MDR genes), and (5) determination of relevant drug exposure definitions for each class of drugs (e.g., Cmin, AUC, intracellular triphosphate levels, degree of protein binding and free drug) and the relationship between in vitro and in vivo susceptibility.

The final call in the midst of so many unanswered questions was for the development of collaborative surveillance projects with the CDC and NIH in both treatment-naive and treatment-experienced patients in order to more fully characterize the clinical relevance of changes in mutational patters and drug susceptibility. Other important areas of current and future investigation include evaluation of relative fitness and virulence of resistant viruses, and better characterization of, and tools to measure, the pharmacologic relationship of achievable drug levels and reduced susceptibility. The development of assays with improved sensitivity characteristics is in process. Mutational algorithms and susceptibility assay breakpoints will need to be established during new drug development and modified based on post-marketing evaluation and an ongoing consensus mechanism.

Transcripts of the meeting are available on the FDA Website (Dockets). Please feel free to send any questions about this meeting to Dr. Roland through our "Ask the Experts" section.

Related HIVresistanceWeb articles:

FDA's CBER Defining Regulatory Requirements for Genotypic Resistance Tests

Clinical Availability and Reimbursement Status of Antiretroviral Drug Resistance Assays