Treatment interruptions in the salvage setting: what have we learned?

Patients with multi-drug-resistant virus are still facing the dilemma of putting together an effective treatment regimen that is tolerable. Ideally, strategies for managing treatment in such patients would be based on introducing new treatments, those specifically designed to inhibit HIV with accumulated drug-resistance mutations, such as enfuvirtide (Fuzeon), or experimental compounds like tipranavir, TMC114, and TMC125. However, such new agents usually are available one at a time because they are developed by different companies and have different timelines for research and development. This leaves clinicians and patients with the problem of how to put together a complete regimen with more than one optimally active drug.

The early reports of what happens to multi-drug-resistant virus populations when treatment is interrupted (the well-known shift to wild-type virus) raised much hope that this could be an effective strategy for dealing with the problem of resistant viruses: just let them disappear and be replaced by wild-type virus. What we have learned, however, is that this may not be such a good strategy after all. Clinically, patients may be better off staying on treatment, as shown by the CPCRA 064 study recently reported by Jody Lawrence and her co-investigators (see data review, page 5). The reasons for this notion include basic principles of drug-resistance selection, viral fitness, resulting shifts in viral population, and the benefits of treatment even for patients with "treatment failure." Perhaps much of the optimism associated with the implementation of this treatment-interruption strategy was based on a misunderstanding of what the shift in the viral population actually represents.

The population of HIV within a patient comprises quasispecies with the high diversity and rapid evolution of viral genomes characteristic of RNA viruses; this forms the basis for the efficient selection of drug-resistant mutants observed under suboptimal treatment conditions. As long as antiretroviral drugs exert selective pressure, the drug-resistant strains of HIV will have the upper hand. But, resistance mutations frequently affect enzyme activity and result in reduced fitness of the resistant HIV strains compared to wild-type viruses--when compared in drug-free conditions. This explains why, as soon as the selective pressure of the drugs is removed, the wild-type population will outgrow the resistant one(s). This phenomenon is observed in vivo, indicating that wild-type virus has been archived during the on-treatment period--just as drug-resistant strains will be archived during the off-treatment period (see also essay, page 14). The key is that neither population may ever be completely eradicated in any one individual over a lifetime of being treated; therefore, re-initiation of treatment may well lead to reselection of the resistant strains present in the archived collection of viruses. However, the virus with reduced fitness may be less harmful to the patient and thus the overgrowth of wild-type virus may not be the desired effect either. Each of these points has been documented with data derived from clinical samples.

Perhaps the most useful information that has come out of this line of research is the understanding that the "failing" treatment does provide clinical benefit, although the patient clearly may be experiencing virologic failure. One of the first indications of this premise was the observation that viral loads increase and CD4 T cell counts fall in patients stopping therapy. The association between clinical benefit and antiretroviral treatment may be indirect (attributable to changes in the fitness of mutant viruses) or direct (attributable to the residual activity of the "failing" regimen). Evidence for the "fitness" effect comes from the temporal association between loss of resistance phenotype and the steepest changes in surrogate markers. Studies by Stephen Deeks and colleagues have also given credence to this hypothesis. However, surrogate marker changes have been observed even in patients whose viral populations did not exhibit the "shift," indicating that residual antiviral activity (in spite of drug resistance) may play a role. Surprisingly, support for this notion has come from testing the effect of removing single drugs from a failing regimen--in particular, nucleoside reverse transcriptase inhibitors (NRTIs).

Three recently reported studies of partial or single-drug interruption indicate that a failing drug may not be completely devoid of activity in patients whose virus carries resistance-associated mutations to that drug. Frank Maldarelli and colleagues from the National Institutes of Health (NIH) reported that interruption of stavudine (Zerit) in 3 patients who were receiving it as part of a failing regimen resulted in significant increases in viremia. The patients' baseline viral genotypes had multiple mutations associated with stavudine resistance. Upon resumption of stavudine treatment, HIV-1 RNA levels declined to baseline, leading the authors to conclude that stavudine contributed to antiviral activity in this regimen. (1) Similarly, Tom Campbell and his colleagues from the University of Colorado demonstrated that interruption of lamivudine (Epivir) in 4 patients on a failing regimen resulted in a 0.5 log increase of HIV-1 RNA, independent of an M184V [right arrow] M switch. Thus, lamivudine appears to contribute to antiviral activity independently of the M184V effects on viral fitness and susceptibility to zidovudine (Retrovir). (2) Finally, in a study of partial treatment interruption presented by Stephen Deeks, interruption of the NRTI component of a failing regimen resulted in rapid increases of HIV-1 RNA levels. (3) These data help explain why patients on treatment (particularly those "salvage" patients who are heavily treatment experienced with several drug-resistance mutations), no matter how high the viral load or low the CD4 T cell count, appear to do better clinically than patients who are off treatment. (4,5) In the final analysis, each of the 2 mechanisms--viral fitness and residual antiviral activity--are likely to contribute to clinical benefit.

Only 1 of the 3 controlled clinical trials described above was powered to assess the clinical impact of a treatment strategy that includes a period of treatment interruption. As described in the literature review in this issue of RITA! (see page 5), the conclusions reached by investigators of treatment interruption are often in disagreement with each other.

All 3 studies described above documented changes with regard to which viral variants comprised the predominant or majority population. Interestingly, however, the CPCRA 064 study demonstrated that patients with a shift in predominant viral population towards wild-type virus did have better virologic and immunologic responses, confirming the earlier observations from observational studies and the GIGHAART study (see data review, page 5). But the surrogate marker responses did not translate into a clinical benefit. The decrease in CD4 T cell count prior to re-initiation of treatment may have been sufficient to put the patients at risk for clinical progression, regardless of the subsequent response. An estimated median of 4.9 months passed for the CD4 T cell count to exceed baseline values in the treatment-interruption group.

The study with the most positive conclusions (a significant benefit for treatment interruption in terms of surrogate marker responses) was GIGHAART, which used the shortest interruption time and the highest number of drugs in the new treatment regimen. If residual drug activity plays a therapeutic role, as indicated by the single-drug interruption studies, then the effects of such antiviral activity would have a greater chance of being detected if more drugs are used in treatment combinations. However, the clinical impact of having extremely reduced CD4 T cell counts for too long may override the effect of a superior virologic response offered by multiple agents.

The basic concepts of drug-resistance selection and viral fitness, together with the documented benefit of staying on treatment, should guide patient management decisions. In situations where treatment interruption is necessary, the interruption times should be as brief as possible, with careful monitoring of CD4 T cell counts during the interruption interval. The available data suggest that the composition of the new treatment regimen should be based on including a maximum number of truly "active" drugs.

So, where should research in this area now be focused? The best overall strategy will be one that helps us avoid situations that foster the development of multi-drug resistance and intolerable long-term toxicities. Whether this problem of multi-drug resistance will continue to be prevalent after years of HAART remains to be seen, especially considering the many current patients who are survivors from the pre-HAART treatment era and who therefore have likely been exposed to suboptimal therapy. Has HAART really been as effective in preventing resistance development as we had predicted? We should continue assessing the prevalence and characteristics of multi-drug resistance over time. Finally, the assessment of each drug's contribution to the benefit sustained by a "failing" regimen is another area that should be developed, as is the use of immune-based therapies that may provide protection during periods off treatment.

References

(1.) Maldarelli F, Palmer S, Kearney M, et al. Antivir Ther. 2003;8(Suppl. 1):S149.

(2.) Campbell TB, Young RK. Johnson SC, Lanier ER, Kuritzkes DR. Antivir Ther. 2003;8(Suppl. 1):S156.

(3.) Deeks SG, Martin JN, Hoh R, Wrin T, Petropoulos C, Grant R. 10th Conference on Retroviruses and Opportunistic Infections, February 10-14, 2003; Boston. Abstract 640.

(4.) Miller V, Mocroft A, Clotet B, et al., for the EuroSIDA Study Group. J Infect Dis. 2002;186:189-197.

(5.) Lundgren JD, Vella S, Paddam L, et al. 9th Conference on Retroviruses and Opportunistic Infections, February 24-28, 2002; Seattle. Abstract 48.

Veronica Miller, PhD

Forum for Collaborative HIV Research, George Washington University, Washington DC