There is growing concern that the widespread use of protease inhibitor (PI)-based therapies is selecting resistant virus. Among heavily treated individuals the prevalence of PI mutations and reduced susceptibility to PI continues to increase. As reported from the Virco data base of clinical samples from the U.S. and Europe, the L90M mutation is the predominant PI resistance mutation, present in 33% of more than 5,000 samples in 1998-1999. (Abstract 740). Investigators in Marseilles documented an increase in PI resistance among 782 patients between 1997 and 1999 with a 3- to 4-fold increase in the prevalence of PI mutations at M46I/L, L90M, and V82A/F/T to about 30% in the same time frame (Abstract 744). There has been speculation that non-subtype B viruses may differ in their capacity to develop resistance to PI drugs. However, the G48V and V82A/F mutations (and less commonly L90M) were associated with virologic failure and PI resistance in one study of non-subtype B viruses from Brazil (Abstract 745) and another from the Ivory Coast (Abstract 746).

The prevalence of PI resistance among treated patients has lead to concern that transmission of resistant virus could jeopardize initial therapy. Michelle Roland and her colleagues showed evidence for the potential transmission of PI resistance in their post-exposure prophylaxis (PEP) study, in which they examined the resistance patterns in source patients (those to whom their index patients seeking PEP had been exposed). Primary and secondary mutations associated with PI resistance were seen in 6 and 15 of the 16 source patients who were tested, respectively (Abstract 197). Similarly, the frequency of PI resistance among source patients in the Occupational HIV Exposure Study Group cohort was 9/42 (Abstract 496). In a primary infection cohort of 85 newly infected patients in whom resistance testing was performed, the frequency of PI resistance (genotypic and phenotypic) was 12% in Canada (Abstract 567). The transmission of resistance, or its increasing prevalence among untreated patients, was examined by several groups. Susan Little presented further results from her study of primary infection, published in JAMA last year (1999; 282:1142), in which moderate, 3- to 5-fold decreased susceptibility to PI drugs was seen in nearly 10% of newly infected patients. Among those naive patients with moderate reductions in PI susceptibility, she demonstrated associations with changes at codons 93, 71, and 63, rather than primary PI resistance mutations as a possible explanation for moderate reductions in PI susceptibility (Abstract 565). In a similar vein, among 130 drug-naive patients in Italy who started a PI-containing regimen, baseline mutations at codons 10 and 36 (in 28%) were significantly associated with a 2-fold risk of virologic failure (Abstract 728).

The spectrum of genotypic changes associated with PI resistance continues to increase. Two groups documented changes in gag cleavage sites which are not a part of standard genotypic testing, and which were associated with PI resistance. Coté and colleagues at Virco found that 90% of highly PI-resistant viruses contained at least one gag cleavage site mutation (Abstract 722) and Peters used recombinant virus and passage/selection techniques to show that insertions and deletions at the p6 gag motif could impart reduced susceptibility to individual PI drugs (Abstract 724). Mark Winters presented phenotypic data on 8 clinical samples from the Quest sequencing data base, where insertions mapping between codons 35 and 38 were found in 0.1% of > 8,000 samples and were associated with increased resistance to PI drugs (Abstract 723).

Many patients who fail therapy with one or more PI combinations develop evidence of PI resistance. However, the dynamics and frequency of PI resistance in virologic failure has been widely variable. In a small number of patients who failed ritonavir, Resch and colleagues presented a two-step process with the selection of genotypic resistance following "initial rebound" from a nadir with further addition of resistance mutations during stable virus replication in the presence of drug (Abstract 725). Cross-resistance between PI was examined in a summary of genpotypic data from the ACTG 333 study, in which patients were switched from saquinavir (hard cap) therapy to either indinavir or saquinavir (soft-gel caps). Both drug susceptibility and virologic response were proportional to the number of PI mutations at baseline (Abstract 732).

There were a number of clinical trials of salvage therapy with dual PI-containing regimens following initial or multiple PI failure. ACTG 359 (where delaviradine or adefovir or both was combined with nelfinavir-saquinavir or ritonavir-saquinavir demonstrated that only 30% of patients who had failed indinavir + NRTIs could be suppressed at 16 weeks (Abstract 235). Similarly, in ACTG 398, combinations of indinavir or nelfinavir with amprenavir (with abacavir, efavirenz and adefovir) found a similar rate of 34% virologic success after PI failure (Abstract LB7 ). The most promising data on salvage of PI failures comes from the study of 70 PI failures offered therapy with ABT378/ritonavir, nevirapine and one new NRTI. The 48 week data presented by Steve Deeks showed a 70% virologic success rate in these difficult to treat patients (Abstract 532).

In summary, PI resistance is increasingly detected among heavily treated patients and is now relatively frequent in newly infected and drug-naive subjects. While the significance of moderate levels of PI resistance (associated with secondary PI-related mutations) for initial therapy with a PI-containing regimen has not been established, there is good reason to be concerned. The same frequency of moderate resistance to NNRTI drugs raises the question of the need for resistance testing at baseline to establish the optimal regimen, even in newly infected untreated patients. The most important advances in PI therapy are likely to come from the judicious use of dual-PI combinations which take advantage of pharmacologic enhancement to maximize the effectiveness of these potent drugs, and in some cases to overcome resistance. The two new PI drugs discussed at the conference, ABT378 and BMS-232632 are both likely to be once-daily drugs when combined with low dose ritonavir.