Emily J. Erbelding, M.D., M.P.H.
The Hopkins HIV Report - May 1999

How Resistance Evolves Laboratory Testing Methods The Impact of Genotypic Testing in Clinical Management Summary Resistance Testing: Case Studies

The remarkable ability of HIV to evolve resistance to specific antiretroviral agents is among the major factors leading to failure of therapy. While antibiotic choice in treatment of a serious bacterial infection is often guided by antimicrobial susceptibility testing, management of HIV-infected patients with virologic failure has meant switching to as many new drugs as possible based upon the treatment history and known patterns of cross resistance. The availability of resistance testing in clinical practice brings with it the possibility that therapeutic choices can be guided and refined to ultimately improve virologic and clinical outcomes. In this article I will discuss how resistance is detected in the clinical laboratory, situations in which resistance testing makes sense, situations where it may fall short, and how it can currently be used in clinical management.

How Resistance Evolves

Variations in HIV RNA are generated, on average, at a rate of 1 nucleotide per replication cycle, meaning that if 10 billion viral particles are made daily, every possible drug mutation is generated daily. Thus, viral polymorphism (the presence of variants composed of different genetic material but which presumably have the same overall "fitness") is common in persons with established HIV infection. In the presence of selection pressure exerted by antiretroviral drugs, any pre-existing virus with mutations conferring greater fitness emerges as the predominant strain. For some antiretroviral agents (such as 3TC and the NNRTIs), a single point mutation leads to high level resistance. For other drugs (such as AZT and the protease inhibitors) high-level resistance requires the accumulation of multiple resistance mutations.

Evidence implicating a point mutation as the cause of drug resistance arises from 2 lines of data. First, in vitro passage of HIV-1 in culture in the presence of drug leads to the development of mutations. Secondly, viral genotypes from many patients receiving a specific drug who experience virologic rebound have characteristic mutations. For most reverse transcriptase (RT) and protease (Pr) mutations outlined in Figure 1, the clinical and in vitro data correlate well. In standard resistance nomenclature, the first letter denotes the wild-type amino acid for that codon, the number the codon of interest, and the ending letter represents the amino acid coded for by the mutated codon. Thus, "K103N" detected in the RT gene means that asparagine has been substituted for lysine at the 103 codon. Table 1 lists the single letter code relating to amino acids in this system.

Figure 1. Mutations in the reverse transcriptase and protease genes that have been associated with drug resistance are depicted above. Bold lines represent major mutations, while thin lines indicate secondary mutations. Dotted lines represent mutations that have been identified in vitro though have not been associated with resistance in clinical isolates. Resistance mutations, along with more complete sequencing data, can be found at http://www.viral-resistance.com and at http://hiv-web.lanl.gov.

Laboratory Testing Methods

Genotype analysis: Genotypic assays depend upon amplification by reverse transcriptase coupled PCR of either the Pr or RT genes isolated from viral RNA in plasma. The amplicons generated are then sequenced through automated DNA sequencing techniques. The amino acid sequence of the RT or Pr can then be inferred. Multiple point mutations that code for amino acid substitutions do not necessarily coincide within the same viral particle, but each mutation identified must be present in at least 20% of circulating viral particles in order to be picked up through this process. Thus, genotypic analysis can only detect mutations that predominate at the time of sampling and is insensitive to mutations that have developed in the past due to drug pressure but that are not currently predominant. Genotype analysis is also often insensitive to the T69SSS insertion,a mutation associated with resistance to all nucleoside analogues. Genotypic assays are technically difficult to perform when the viral copy number is <1,000 c/ml. The cost associated with genotypic testing varies from $360-480 per gene. The test is not currently licensed by the FDA for clinical use, which means that commercial insurers, including Medicaid, might not reimburse the cost of the test.

Phenotype analysis: The growth properties of an isolated virus in the presence of varying drug concentrations are assessed through phenotypic analysis in a fashion analogous to traditional susceptibility testing in clinical microbiology. Rather than assessing growth characteristics of primary viral isolates, however, currently available techniques insert the RT and Pr genes into a molecular HIV clone with standardized envelope and accessory genes. The phenotypic characteristics of the Pr and RT from the clinical isolate are then assessed in the presence of varying drug concentrations. The amount of drug required to inhibit viral growth by 50%, 90%, or 95% is determined. A 4-fold or higher shift in IC50 (compared to wild type Pr and RT genes) reliably correlates with drug resistance. As with genotype analysis, the assays are difficult to perform if the viral copy number is <1,000 c/ml. As is also true with genotypic testing, the phenotypic assay is insensitive to resistant strains that do not represent at least 20% of the quasispecies. Thus, clinical isolates from patients who are off of all therapy may not reflect resistance or cross-resistance acquired during time exposed to drugs in the past. Rather, as with genotypic analysis, the quasispecies off therapy is likely to be wild-type HIV. Phenotypic testing is also unlicensed for clinical use at this time. Phenotypic testing is more expensive (at about $900/test) and time consuming than genotypic testing, and commercial availability in the United States is limited to one supplier.

The Impact of Genotypic Testing in Clinical Management

Though genotypic analyses are now relatively easy to obtain in clinical practice, many questions remain regarding their appropriate use and the amount of useful information they add to what can be inferred from the treatment history. To date, management of virologic rebound has meant switching all components of the "failing" regimen if possible. However, data generated in the past year from ACTG 343, the Trilege study, and from Merck clinical trials strongly suggest that genotypic analysis at the time of rebound may implicate only a single drug as failing rather than the whole regimen [see Gallant: HHR 11(2), 1999]. In rebound occurring on AZT/3TC/IDV (or on IDV alone following AZT/3TC/IDV), the M184V RT mutation conferring 3TC resistance was the most common mutation seen, while none had mutations associated with IDV resistance. Similarly, in patients with virologic rebound on IDV/EFV, the K103N RT mutation associated with NNRTI was the single mutation identified most commonly. Thus, genotypic testing at the time of initial virologic rebound may suggest that part of the combination is still working and that it may be more appropriate to modify or intensify the regimen rather than changing it completely. These strategies need to be evaluated prospectively in clinical trials to further assess whether resistance is present in the quasispecies but in a low prevalence (<20% of circulating viral clones).

This question of added utility was prospectively evaluated in CPCRA 046 [Baxter, et al: 6^th Retrovirus Conf, LB8]. In this randomized controlled clinical trial, 153 patients with virologic rebound on a combination of 2 NRTIs/PI were randomized to either of two management strategies for the selection of a salvage regimen: in Group I, the patients' clinician received a genotypic report along with expert recommendations for antiretroviral management; in Group II patients, the next regimen was guided by treatment history. The genotypic testing revealed a major RT and Pr mutation in 75% of patients, and a major RT mutation with no Pr mutation in 20% of patients. No major mutations were identified in 5% of cases. Less than half of patients in Group II (management guided by history) received 3 active new drugs, compared with 86% in Group I (management with genotype available, along with expert opinion). Initial viral load declines were more pronounced in Group I (-1.17 log) compared to Group II (-0.62 log; p=0.0001). Though these early results support the use of genotypic testing in designing salvage regimens, it is important to use some caution in applying these findings broadly in clinical practice. The design of the trial makes the relative contribution of the genotype analysis unclear; it is possible that expert recommendations without the genotype data would have given the same results. Viral load declines were significantly greater at sites that adhered to expert recommendations.

As with any clinical laboratory test, resistance assays will likely be most useful when ordered to answer a very specific question. Based upon current data, specific indications for the use of resistance testing included the following:

• Determining drug-resistance in newly-infected persons, especially those who are experiencing the acute retroviral syndrome. With many mutations, in the absence of selection pressure exerted by the presence of drug, there is a reversion to wild-type virus. Thus, as HIV infection becomes established, resistance testing may become less useful.
• Defining the NRTI resistance pattern in those failing their first NRTI-containing combination. The detection of multi-drug resistant virus may be an important consideration in subsequent therapy.
• Defining efavirenz resistance in patients failing on other NNRTIs. The presence of the K103N mutation in this situation indicates that resistance to EFV is also present.
• Defining the extent of PI resistance in those taking a PI-based combination. As described above, some patients failing PIs will have no Pr mutations.

Summary

Resistance assays are now commercially available and use is expanding. Resistance testing has proven to be useful in predicting therapeutic failure. However, because such testing is insensitive to resistant strains present at low quantities, the documentation of wild-type virus by resistance testing is less useful in predicting drug success. As with any clinical laboratory test, the result is likely to be more useful in therapy decisions if the test was ordered with a clear question in mind linked to a specific management strategy. Management strategies incorporating the use of resistance testing need to thoroughly be tested prospectively in clinical trials.

Resistance Testing: Case Studies

• Case 1. A 30 year old female was diagnosed to be HIV-seropositive in 1997. At that time, her CD4 cell count was 270/mm³ and HIV-RNA was 152K c/ml. She initiated treatment with AZT/3TC/IDV. Her CD4 increased to 602/mm³ within 10 months and her viral load was maintained at <400 c/ml, until 6 weeks ago when it was measured at 1,500 c/ml despite her reports of no missed doses of therapy. The viral load was repeated after 4 weeks and found to be 2340 c/ml. A genotype analysis revealed the following:
  

  RT-M184V
  Pr-Wild type sequence
  

  * Comment: This case illustrates what has been recently documented in clinical trials from patients taking similar regimens: that high-level 3TC resistance (M184V) seems to appear first in a person on a 3TC-containing regimen, along with another nucleoside and a PI. This genotype provides useful clinical management information because it suggests that no Pr or AZT-associated mutations had evolved on therapy. The regimen was intensified with abacavir, and the patient had a viral load at <40 c/ml documented at 6 weeks of follow-up.

• Case 2. A 45 year old man with HIV infection diagnosed in 1989 was started on AZT in 1991 when his CD4 cell count was measured at 310/mm³. He continued until 1995 when 3TC was added. His CD4 count at the time was 190/mm³, but then began to slowly decline. In mid-1996, HIV-RNA was measured at 92,000 c/ml, and CD4 count was 120/mm³. Therapy was switched to RTV/ddI/d4T. He developed neuropathy after 12 weeks. He was switched to RTV/SQV/NVP, but developed an exfoliating rash that was attributed to NVP; nevertheless, neuropathic symptoms improved. He continued on RTV/SAQ/d4T (reduced dose d4T), with one initial HIV RNA measured at 6 weeks follow-up showing <400 c/ml, but subsequent measurements increasing over the next 6 months to 9,600 c/ml. CD4 counts have been maintained in the range of 250/mm³. He is tolerating is medications relatively well. A genotypic analysis reveals the following:
  

  RT-Wild type sequence
  Pr-M46L, G48V, G73S, V82T, L90M
  

  * Comment: The fact that he has no RT mutations identified should come as no surprise, given that he is not currently taking any RTIs with associated mutations identified by sequence analysis. With the exception of the insertion mutation denoted as T69SSS, genotypic correlates of D4T resistance are poorly defined, and not all genotypic analyses detect insertion mutations. Therefore, it could have been predicted that an RT genotype would add nothing to the history at this point in time. The protease sequence identifies primary mutations associated with both ritonavir (46, 82 positions) and with saquinavir (48, 73, 90 positions). Though his remaining options for therapy include EFV (given the fact that NVP was discontinued due to toxicity rather than virologic failure, implying that NNRTI-resistance may not have had a chance to evolve), he has few additional potent drugs to add to it. After discussion, the patient and his physician elected to continue the current regimen as long as tolerated, waiting for new options to become available to combine with EFV in order to enhance the chances of a successful salvage.

• Case 3. A 29 year old man presents for care because he intends to initiate antiretroviral therapy. He reports a positive HIV result through a home test kit 1 year ago. He believes that he became seropositive at the time of a mononucleosis-like illness 2 years ago after ending a sexual relationship with a man known to be HIV infected. A confirmatory HIV serologic test is reactive, and his CD4 count is 490/mm³ with an HIV-RNA of 75,000 c/ml. Because he inquires about the possibility of having acquired a resistant strain of HIV, genotypic testing is ordered and shows the following:
  

  RT-wild type sequence
  Pr-wild type sequence
  

  * Comment: Because this patient has an established infection, the HIV genotype is not likely to give results that would aid in management and may not have been a prudent use of health resources. In the absence of selection pressure exerted by antiretroviral agents, the genotype shows no drug-associated mutations, even though it is possible that the patient acquired drug-resistant virus. Had the blood test been done at the time of seroconversion (when the circulating quasispecies is more likely to reflect the transmitted strain), the results may have been more likely to show any resistance characteristic of the transmitted strain.

990501
JH990501

©1999. The Johns Hopkins University AIDS Service, Division of Infectious Diseases. Permission to use and reproduce portions of this newsletter is hereby granted provided that author and publication are fully credited and both copyright and permission notice appear with reprinted material. Inquiries may be directed to Sharon McAvinue, Managing Editor. Website: Johns Hopkins AIDS Service.

The original of this article can be found at http://www.hopkins-aids.edu/publications/report/may99_1.html