The greatest frustration in current care is that many, if not most, people cannot keep their viral loads down with the current medications. Sometimes inability to take the drugs consistently is the original culprit, but HIV's acquisition of drug resistance eventually becomes the critical problem. And each set of resistance-conferring mutations makes further escapes from antiviral treatment quicker and easier. With no therapeutic breakthroughs immediately at hand, the only way to combat multidrug resistance is to employ innovative strategies with existing drugs.
Periodic treatment interruptions, or "drug holidays," have lately gained increased scrutiny as a way to turn back the clock, undoing the damage wreaked by poor past treatments that allowed the evolution of drug resistance. The theory behind such planned drug interruptions is that the mutations that confer HIV's drug resistance also impair the ability of the virus to reproduce. If the converse were true — that they increased HIV's replication rate — the principles of natural selection would cause them to be prevalent in so-called wild type HIV. These are the naturally existing viral isolates that have not yet encountered drug therapy. Going off drugs for a few months might therefore reverse evolution, leading to a loss of drug resistance mutations and the opportunity to successfully recycle drugs on which a patient has previously failed. Alternatively, the small amount of surviving residual wild type HIV might reappear, crowding out the drug- resistant HIV as it beats the slower mutated HIV in the race for infectable cells.
Some research has indicated that at least a part of this theory is correct. At the 6th Conference on Retroviruses and Opportunistic Infections last winter, French researcher Franois Clavel and colleagues reported their observations that HIV became increasingly less "fit" as the number of resistance mutations against protease inhibitors increased in patient isolates. (Fitness was defined as the ability to infect new laboratory cell cultures.)
Restoring the Natural State of Affairs
At the 2nd International Workshop on Salvage Therapy, held in Toronto on May 19–21, Veronica Miller, M.D., of Frankfurt, Germany, presented the first results from her drug interruption trial. The study involved following 50 people taken off drugs for more than two months after failing to suppress their HIV. Her patients had received at least two different protease inhibitors in the past. After the break in therapy, they were prescribed a "mega-HAART" combination consisting of six to eight drugs, including at least three nucleoside analogs, one or two NNRTIs and two to three protease inhibitors.
Resistance assay data are now available for 39 of these patients. At the time of treatment interruption, their HIV was resistant to a median of eight drugs. It had reverted to wild type by the time therapy was recommenced in 26 of the 39 individuals. In the words of Dr. Miller, the resistant virus had "disappeared." Accompanying this "disappearance," median viral load shot up from 100,000 copies/ml of plasma to almost 1 million copies/ml. Viral loads were comparatively stable in those who did not experience a reversion to wild type HIV. (Their plasma HIV levels doubled in the course of the drug holiday, from about 300,000 to 600,000 copies/ml.) Dr. Miller and her colleagues are now analyzing the genetic makeup of patients' HIV isolates when treatment was stopped to see whether particular mutational patterns correlate with a reversion to wild type HIV during drug holidays.
The "reverters" did have a better response to renewed treatment despite their rising viral load while off treatment. At week 8 following initiation of mega-HAART, viral loads in those with wild type HIV had gone down a median of 2.8 logs. The "nonreverters" experienced a median viral load decline of only 1.2 logs. Eighteen of the 26 with reverted HIV had viral loads below 500 copies/ml sometime within the first 24 weeks of mega-HAART. The same was true of only two of the twelve nonswitchers.
This last result looks impressive, but the key word is "within," Dr. Miller said just after the Workshop, "I only put this presentation together in the last two days because everyone was interested in treatment interruption. The analysis is very incomplete, and I don't yet know when individuals were below 500 copies, just how many did so sometime in the first 24 weeks." As for whether treatment interrupiton improves the long-term response to rescue therapy, not just the immediate response, Dr. Miller said, "We need to wait longer to see."
Conflicting data comes from several other studies. Michael Youle, M.D., of London, at the Salvage Therapy Workshop described a mega-HAART trial he had run. Sixty-three patients with a median of five prior anti-HIV drugs and a median baseline viral load of 63,300 copies/ml went on a regimen of hydroxyurea, efavirenz, two protease inhibitors (mainly ritonavir/indinavir) and several nucleoside analogs (mainly ddI/3TC). Each patient had stopped treatment for four weeks or more (median of eight weeks). Forty-two persons have been followed through at least 16 weeks of treatment, and, at that time point, 17 (40%) have viral loads below 400 copies/ml. At week 28, 85% are below the 400 copy/ml threshold, but data is available for only 20 of the 63 study participants. Success at reducing viral load to below 400 copies/ml was associated with the length of drug holiday — there was a 15% greater chance of success per month off treatment.
But treatment breaks were not always salubrious in this study. Dr. Youle reported on one of his patients who had a viral load below 50 copies/ml on the mega-HAART regimen. This individual interrupted treatment for a month; his viral load soared to over 100,000 copies/ml and refused to go down when therapy recommenced. Dr. Youle theorized that interruptions of a few weeks can sometimes spawn increased drug resistance in a patient's HIV because some drugs, such as efavirenz, have lengthy half-lives in the body. As the concentration of these drugs tapers off, the virus is exposed to suboptimal levels for a time, allowing it to reproduce while exposed to the drug. These conditions promote the natural selection of resistance-conferring mutations.
Howard Grossman, M.D, a private HIV specialist in New York City, described the experience of 33 patients with long treatment histories (an average of nine prior drugs). Eight of these patients interrupted therapy one to three months before starting a mega-HAART rescue therapy. The 33, whose mean baseline viral load was 21,000 at the onset of mega-HAART, received a regimen consisting of abacavir, adefovir, efavirenz, hydroxyurea and two protease inhibitors (most commonly ritonavir or nelfinavir plus saquinavir, later switched to ritonavir plus indinavir). The patients also took two or three nucleoside analogs recycled from regimens that they had been on in the past. The burden of taking these many medications was eased by the fact that dosing was almost entirely twice a day.
The results were comparable to the other two studies: 23 patients (70%) had viral loads below 40 copies/ml at two or more time points, with 17 patients (40%) still below 40 copies/ml after seven to ten months. Notably, though, the eight with drug holidays fared no better than the others over the long run. At seven months, the average viral load drop was about two logs (99%) in both groups.
Since the number on drug holiday was small, this negative observation could have been due to random error and may not be significant, but Dr. Grossman is skeptical of the value of treatment interruptions. "Drug holidays may yield a better initial response, but the durability of that response is unproven," he said in an interview. "In my mind, the only sound rationale right now for drug holidays is to give the patient a rest from all the drugs' side effects."
Yesterday's Virus Is Always with Us
There are several theoretical objections to the treatment interruption approach to eliminating drug-resistant HIV. Exposing HIV to antiviral medications sets it down evolutionary pathways that become irreversible as mutations accumulate. It is not true that drug-resistance mutations consistently yield an HIV that is less fit than wild type. The later mutations that arise in response to AZT monotherapy are notorious for creating a faster growing HIV, at least in the test tube. Other primary resistance mutations have their deleterious effects compensated for by secondary mutations. For example, a mutation in codon (unit) 151 on the reverse transcriptase gene triggers the evolution of HIV that resists virtually every nucleoside analog to some extent. After the 151 mutation appears, further mutations occur at other points on the gene, and some of these slightly increase the virus's sensitivity to drugs while at the same time allowing it to replicate more quickly.
A similar phenomenon occurs for protease inhibitors. At the end of May, Richard D'Aquila, M.D., and a team at Boston's Massachusetts General Hospital issued the latest report on this subject (see Journal of Virology, May 1999, pages 3744–52). The group first tested HIV isolates with a single protease mutation. HIV with a nelfinavir-triggered mutation at codon 30 of the protease gene or a saquinavir-triggered one at codon 90 replicated more slowly than drug-sensitive normal HIV. But a mutation at codon 63 (frequently present even in wild type HIV as a natural variant) partially compensated for this loss of fitness, especially in the case of the codon 90 mutation. Meanwhile, indinavir-resistant isolates that had developed four or five mutations (including the one at codon 63) were just as fit as wild type HIV.
Individual strains of HIV most likely do not "disappear" (to use Dr. Miller's term) during either treatment or drug interruption. The genetic sequencing and virus culture assays used to determine HIV drug resistance have limits to their sensitivity. They are not able to detect HIV subpopulations existing below about 10% of the total viral population in someone's body. The overgrowth of "wild type" HIV after therapy is stopped may merely cover up the survival of the more slowly replicating drug-resistant strains. The converse is probably what happens during treatment, when the drug-resistant HIV replicates more rapidly than the drug-sensitive, wild type variety.
If the eclipsed subpopulations do not exist in actively replicating form, they certainly exist in a dormant form, hidden in quiescent white blood cells with latent HIV infection. These cells can last for years, even decades, and the HIV within them will suddenly spring forth whenever the cells are activated to fight infection.
Until further information arrives, drug holidays are a speculative venture at best. As Dr. Grossman remarked, the alleviation of drug side effects is their one self-evident attraction. But that siren song may be just a diversion from the sophisticated treatment strategies and agents needed to overcome multi-drug resistant HIV.