Most discouraging was a presentation (late breaker LB-2) by Steven Deeks, M.D., of San Francisco General Hospital, which reported on a retrospective look at incidence and risk factors for protease inhibitor failure. Dr. Deeks recounted the experiences of 136 patients who had taken protease inhibitor-containing regimens for longer than six months. The data were derived from a chart review and patients were not specifically tested or interviewed for this study. Therapy was deemed unsuccessful if a patient's two most recent viral loads were above 500 copies/ml (the limit of quantification for the standard viral load assay). By these standards, protease inhibitor therapy failed in 53% of the patients, surprising the ICAAC audience. The media immediately picked up the story, with prominent articles appearing in The New York Times, USA Today and the San Francisco Examiner, among others. Although the significance of the low levels of viral replication present in many of these patients (resulting in viral loads ranging between 500 and 10,000 copies/ ml) has yet to be determined, the public interpretation was that the drugs had stopped working altogether.
Less attention was given to more reassuring follow-up reports from several trials: Merck's 035 study (oral presentation I-89) found that most of the 28 volunteers in the original AZT/3TC/indinavir study had greater than 2 log (99%) reductions maintained for two years. Eighty to 85% had viral loads below 500 copies/ml and 60 to 70% had below 50. Also, a further look at Agouron Pharmaceuticals' trial 511 (poster I-101) found that 80% of patients on nelfinavir/AZT/3TC had HIV viral load below 500 copies/ml after one year of therapy.
A multivariate regression analysis conducted as part of the San Francisco General Hospital study found that high baseline viral loads (over 100,000 copies/ml); low CD4 counts (below 200 cells/mm3); the addition of the protease inhibitor to previous nucleoside analog therapy; and a suspicion of non-adherence to the prescribed drug schedules all were significantly associated with protease inhibitor failure. Unfortunately, the data on adherence came from a "guesstimate" accomplished by reviewing comments recorded in patients' files. Unreported nonadherence could account for much of the viral breakthrough since most patients claim to be more compliant than they actually are.
Then too, the San Francisco General figures agree considerably with the Merck 035 record in the two trial arms receiving what turned out to be inferior therapy. Only 45% of those in 035 who started out on AZT/3TC alone for the first six months and then added indinavir ever achieved the 500 copies/ml viral load limit. Similar results were obtained in the trial arm that first received indinavir monotherapy and added AZT/3TC after six months. In the real world, as in the trials, sequentially adding new drugs to piecemeal regimens yields poor results, presumably due to the emergence of progressively broader HIV drug resistance.
A presentation from the California Collaborative Treatment Group (poster I-128b) further confirmed the inappropriateness of such piecemeal strategies: Only 31% of 48 volunteers taking protease inhibitor combinations had HIV viral loads under 400 copies after six months of treatment. The volunteers, with extensive past treatment, were taking part in a study of the usefulness of viral load monitoring. Their median baseline viral load was around 60,000 and their CD4 count was 140, even with therapy. The predictors of failure in this study were having taken prior protease inhibitors and taking fewer than three drugs.
Other reports of failure came from various groups that examined the usefulness of ritonavir/saquinavir combinations in advanced patients. One small study, also conducted by Dr. Deeks (oral presentation I-205), followed indinavir failures who had taken the drug for an average of eight months. The saquinavir/ritonavir combination was ineffective in maintaining viral loads below 500 after 24 weeks of treatment. Patients with prior saquinavir experience were excluded from this study, yet the mutation at codon 90 of the protease gene that is associated with saquinavir resistance was present in 9 of 12 patients. It was not present in the two patients who had a durable response. Similar observations were seen in groups from Australia, Spain, France, Switzerland, Canada and the United States. The 54-person U.S. study, presented as a poster (I-104) by Michael Sampson of St. Vincent's Hospital in New York, along with colleagues at other sites, found that patients adding ritonavir to saquinavir h
Another report from St. Vincent's, which is my own hospital, garnered considerable attention with its observation that patients initially prescribed nelfinavir also had poor responses when switching to other protease inhibitors -- either indinavir or ritonavir/saquinavir. (See late breaker oral presentation LB-5.) Most experienced rises in viral load after initially achieving less than maximal suppression. A few patients did receive sustained benefit, but this 12-person study was too small to determine what differences accounted for the discrepant results, nor was any formal study of compliance performed.
By comparison, Keith Henry, M.D., of Regions Hospital in St. Paul, found that all of 12 nonresponding patients initially treated with nelfinavir-containing combinations had their viral loads plunge to unquantifiable levels (below 500 copies/ml of plasma) when they switched to a combination of ritonavir/ saquinavir/d4T/3TC. (See oral presentation I-204.) Six of seven remained undetectable after 16 weeks of therapy. Patients with more advanced disease and longer treatment history did not fare as well with only three of seven dropping below 500 copies/ml.
The implication of all these observations is that the choice of initial protease inhibitor-containing regimen is extremely important in the long-term virologic outcome of the patients, with adherence probably being the most significant factor. Since many of the patients in "real life" situations added a protease inhibitor to a failing regimen, the actual failure rate for new patients taking new three-drug combinations is not known, but presumably is significantly less than 53%. The choice of a second protease regimen after the initial fails is still up in the air, and no compelling data seemed to favor one regimen over another.
Whatever is prescribed, the minimum viral load achieved in a patient seems to be predictive of the overall long-term response. Ultrasensitive viral load assays may be useful to differentiate complete from nearly complete responses.