Session 1. New agents and regimens for salvage therapy
Several new compounds active against viruses with commonly seen drug resistance patterns were presented. Dr. Robert Rando presented data on dOTC, a cytosine analog developed by BioChem Pharma, which appears to be active against clinical HIV isolates with IC50 values of 0.6 to 2.5 mM [1]. DOTC remained active against HIV resistant to ZDV, 3TC or ZDV+3TC. In vitro resistance to dOTC was associated with RT mutations at codons 65 and 184, mutations that generally produce 3TC resistance. Dr. Angelos Hatzakis presented data on 14 antiretroviral-naïve patients who received IFN-a2b (INTRON-A) at 5 MIU sq daily for 28 days prior to initiation of HAART [2]. One patient did not respond, five patients had intermediate responses (0.5 to 1.0 log10 decrease in HIV RNA) and 8 patients responded with an average 1.0 to 1.2 log10 decrease in HIV RNA over the 28 days. The authors feel that IFN-a2b should be considered for use in patients who fail HAART therapy. Dr. Michael Miller from Gilead Sciences presented clinical data from study GS-96-408 showing that the in vitro increased susceptibility of viruses with M184V mutation to adefovir translated into improved antiviral responses in the clinic [3]. Patients whose virus had a M184V mutation had showed a 0.38 log10 decrease in HIV RNA while patients with high level ZDV resistance showed a 0.62 log10 increase in HIV RNA response. This data suggests that 3TC should be continued along with adefovir when this agent is used in patients whose virus has the M184V mutation. Dr. Vicente Soriano from Madrid [4] presented retrospective data from heavily pretreated patients who received therapy with either ritonavir plus saquinavir (both 400 mg BID) or nelfinavir plus saquinavir (1250 mg plus 800 mg BID). At 6 months, viral responses (HIV RNA <50 copies/mL) for ritonavir plus saquinavir were superior to nelfinavir plus saquinavir by intent-to-treat-analysis (48.8% versus 26.5%) or on-treatment analyses (64.5% versus 34.6% respectively).

Session 2. Predictors and mechanisms of failure Dr. Richard Harrigan presented data from a cohort of patients from Vancouver who were switched to either ddI plus hydroxyurea or d4T plus 3TC [5]. The strongest predictor of a poor virologic response (HIV RNA change <0.5 log10) was not drug history but the presence of the T215Y/F mutation associated with ZDV resistance (O.R. = 14, p=0.004). This expands the number of nucleoside drug combinations that have been adversely affected by the presence of the T215Y/F mutation. Dr. Roger Paredes from Catalonia, Spain presented data from a retrospective cohort of 58 patients with virologic failure on their first PI-containing regimen who were switched to a four drug regimen containing saquinavir 800 mg BID and ritonavir 400 mg [6]. Th majority of the patients who responded to the treatment, as measured through HIV RNA response at 6 months (<200 copies/mL) had a viral load of <5000 copies/mL at time of the change in regimens (76%, versus 30% with HIV RNA >5000 copies/mL) and one or less major PI resistance mutations at the time of the switch (2/3 responders with 1 major mutation versus 0/6 responders with >2 major PI mutations). Dr. Veronica Miller presented data from an observational cohort of 558 patients followed by the Frankfurt HIV Clinic who experienced HIV RNA declines to <500 copies/mL on HAART [7]. The risk of viral rebound was associated with the CD4 cell count at baseline (p=0.001), CD4 cell count change from baseline (p=0.003) and the most recent CD4 cell count (p=0.001), with most recent CD4 cell count being the strongest predictor of viral rebound (relative hazard 5.4 for CD4 <20 versus > 500; p=0.00001).

Session 3. Responses to specific salvage therapy regimens

Dr. Nancy Shulman of Stanford reviewed the virologic response to efavirenz combined with adefovir dipivoxil in addition to other recycled antiretrovirals in a retrospective analysis of 33 highly drug-experienced patients [8]. Of the patients who were non-nucleoside reverse transcriptase inhibitor (NNRTI)-naïve, 50%, in the intent-to-treat analysis, had viral loads of <500 copies/mL at 24 weeks, including 31% with <20 copies/mL, while none of the NNRTI-experienced patients had viral loads of <20 copies/mL. Lower pre-treatment viral loads and higher CD4 counts were also predictors of a positive response. Patients who were on lamivudine or abacavir did better than those who were not, possibly by continuing M184V mutation selection pressure and therefore enhancing the adefovir antiviral activity as has been shown in vitro. Finally, looking at those patients who had received NNRTIs, those with Y181C and G190A mutations had a greater initial viral load decrease than those with the K103N mutation, but the response was short lived. This suggests that efavirenz after nevirapine or delavirdine failure is not effective.

Dr. Vicente Soriano from Madrid, Spain looked at responses to using 2 different 3- or 4-drug regimens, most often stavudine/didanosine/nelfinavir/efavirenz alternating with zidovudine/lamivudine/ritonavir/indinavir at monthly intervals [9]. All 17 patients were highly experienced, 6 patients dropped out because of toxicity/intolerance. 10/11 of the remaining had at least a 1 log drop in viral load during therapy and 6/11 had a viral load of <500 copies/mL after 6 months of treatment. It would be interesting to see the baseline resistance profiles of these patients, and to see a comparison arm of continuous therapy of each of the regimens.

Dr. Julio Montaner from Vancouver presented results of Mega-HAART using 6-9 drugs (with a median of 7 drugs) in highly experienced patients with two different cohorts of patients (n=82 & 67) [10, 11]. In an intent-to-treat analysis, about 60% of both groups achieved viral loads of <400 copies/mL at some point in the course of therapy. In the first cohort, about a quarter of the patients (also by intent-to-treat analysis) had sustained suppression for as long as 42 weeks. A substantial number of patients dropped out due to toxicities/intolerance in both cohorts. Baseline phenotypic susceptibility to three or more drugs and low pretreatment viral loads were associated with a better virologic response.

Dr. Michael Miller from Gilead looked at adefovir and nelfinavir plus other nucleosides in 24 nucleoside-experienced children, 16 of whom were PI-naïve [12]. The investigators performed a dose finding, pharmacokinetic study at the beginning of the treatment, and, for children, a dose of 1.5mg/kg/day of adefovir was found to be equivalent to a dose of 60mg/day in adults. Six patients sustained a >1 log reduction in viral load for 24-36 weeks, all were PI-naïve. Four of the durable responders had mutations associated with AZT/3TC, suggesting that adefovir may have contributed to their durability.

Dr. Mike Youle from London also looked at mega-HAART treatment [13]. He and his colleagues examined 63 heavily PI-experienced patients, a majority of whom (86%) were NNRTI-naïve. All patients received 5-7 drugs including efavirenz in all, and most received didanosine, hydroxyurea, indinavir and ritonavir. Over half of the patients changed or stopped therapy because of choice or toxicity. Of the 20 patients who made it to 28 weeks, 85% had viral loads of <400 copies/mL, with a median CD4 increase of 120, despite the hydroxyurea. Factors associated with a better outcome were lower pre-treatment viral loads, white race, and having a month or greater drug holiday prior to salvage initiation.

Dr. Tomas Pumarola Sune from Barcelona, Spain looked at efficacy of nelfinavir/nevirapine plus nucleoside salvage therapy is 74 NNRTI naïve patients who failed PIs [14]. At 3 months only 28% had viral loads of <200 copies/mL, and at 6 months only 3 patients were suppressed, showing the inadequacy of this combination for salvage.

Dr. Roy Gulick presented data from ACTG 359, a factorial design, randomized, controlled salvage study comparing six salvage regimens in patients who had failed indinavir HAART only [15]. It compared ritonavir-saquinavir with nelfinavir-saquinavir, along with delavirdine, adefovir, or both. At 16 weeks, 30% of patients in all arms had viral loads of <500 copies/mL. The two different double PI combination pooled arms were equivalent. Delavirdine arms did better than those without, and adefovir/delavirdine arms were equivalent to the delavirdine only arms. A disturbing observation presented in the Retrovirus Conference was the reduction of delavirdine levels by 50% in the adefovir arms vs. no adefovir, and subsequent reduction of the saquinavir levels. Gilead is currently studying this unexpected interaction between adefovir and delavirdine (and potentially other NNRTIs) more closely.

Dr. Howard Grossman presented another mega-HAART study from New York, looking at 6-8 drugs including adefovir, abacavir, and efavirenz, in 35 heavily experienced patients, most of whom were NNRTI experienced [16]. Only one patient dropped out due to toxicity, and 2 discontinued treatment due to failure. Two-thirds achieved viral loads of <20 copies/mL at two points and more than a third had viral loads of <20 copies/mL at 6 months.

Session 4. The role of diagnostic/monitoring technologies in salvage therapy This session was dominated by a presentation by Dr. Veronica Miller on the use of mega-HAART following a drug holiday [17]. Data was presented for 94 patients who received mega-HAART („ 6 drugs) with or without a drug holiday and for 50 patients during their drug holiday. During a drug holiday, HIV RNA increased by a median of 0.71 log10 and CD4 cell counts decreased by a median of 89 cells/ml. 26/39 patients reverted from drug resistant virus to wild type during the drug holiday. Subsequent mega-HAART therapy was associated with a 2.9 log10 decrease at 8 weeks for patients whose virus reverted to wild type versus 0.78 log10 in patients whose virus did not revert to wild type. These observations (which are uncontrolled) are very intriguing, and will generate several randomized, controlled trials in the near future.

Session 5. Clinical trial design for salvage therapy Victor DeGruttola and Janet Anderson in the plenary talk discussed novel trial designs for salvage therapy studies [18]. They advanced a proposal to group patients by clusters of genotypic mutation patterns into factorial or fractional factorial designs to evaluate new drug combinations. They suggested the use of short-term virologic endpoints to screen for antiviral potency with "play the winner" designs to improve trial efficiency. Drug combinations that appear active would then be advanced to larger confirmatory trials. Innovative strategies such as these will be needed to assess the increasing number of drug combinations available with the heterogeneous population seen with heavily pre-treated patients. Michael Saag presented data from a retrospective analysis of 770 patients attending the University of Alabama 1917 clinic [19]. Among 51 patients who died in 1998, only 45% had ever experienced HIV RNA <500 copies/mL, and the average duration from an undetectable HIV RNA and death was 5.9 months. Among 114 prospectively followed patients, 1067 drug regimens had been utilized. Two hundred forty-two unique drug regimens had been utilized and 93% of the patients used a sequence of drug regimens unique to themselves. This study demonstrates the diversity of treatments for individual patients and the difficulty of extending clinical trial results for heavily pre-treated patients to individual HIV-infected persons. The most controversial proposal of the session was advanced by Dr. William Cameron from the University of Ottawa [20]. He proposed a three-arm trial for patients who had experienced virologic failure on at least two HAART regimens. The three arms would be: mega-HAART (at least five drugs including NRTIs and PIs), continued standard therapy with 3 to 4 drugs, or up to six months of a drug holiday. The endpoints of the study would be quality of life and survival. The study, which would address several of the key issues for heavily pre-treated patients, generated intense discussion and highlighted critical issues that will need to be evaluated in randomized trials in the near future.