Report 1 -  4th International Workshop on HIV Drug Resistance and Treatment Strategies
Written by Jules Levin
Sitges, Spain, June 12-16 2000

ABT-378: correlates for virologic response at week 24 (phenotypic and genotypic resistance, and trough/EC50)

Genotypic Correlates of Reduced In Vitro Susceptibility to ABT-378 in HIV Isolates from Patients Failing Protease Inhibitor Therapy

We all know the strong need for effective new drugs for individuals with extensive treatment experience and resistance to the available drugs. It is difficult to design effective and ethical studies in which to test a new drug and to identify virologic response in individuals with extensive treatment experience. Abbott used new approaches to analyze the study results reported below. The early data presented on ABT-378 at this meeting was in essence the first reported data analysis of this drug's use in individuals with multiple PI experience. More information and additional analysis is expected to become available over time as more studies are conducted, and should supply more insight into the use and effectiveness of ABT-378 in individuals with extensive prior treatment experience. Additional new week 24 information will be presented at Durban.

Dale Kempf delivered an oral presentation on an analysis of HIV isolates (n=112) from patients experienced with single (study M97-765) or multiple PI therapy (study M98-957). Kempf reported that among 100 naÔve patients treated with ABT-378/r plus d4T/3TC for 72 weeks, the selection of resistance to ABT-378 has not been documented. In the absence of data on primary treatment failures, this presentation tells of their attempts to gain insight into ABT-378 resistance by identifying protease mutations which correlate statistically with certain levels (<5 fold, <10 fold, <20 fold, and <40 fold) of reduced phenotypic susceptibility to ABT-378.  In a following oral presentation from Eugene Sun, described below, he presents the response to therapy in treatment experienced patients with respect to baseline genotypic and phenotypic resistance.

In study M97-765, patients were enrolled who had experience with only 1 PI, and the range of phenotypic resistance to ABT-378 was up to 26-fold with a mean of 2.8 fold. In the second study (M98-957) which enrolled multiple PI experienced patients, the median number of protease inhibitors individuals had used at baseline was 3, and the range of phenotypic resistance was up to 96-fold with a mean of 16-fold. There were 56 isolates from each study and they were combined for the purposes of this analysis for a total of 112. To ask the question--what mutations are associated with reduced susceptibility--they used 2 complementary statistical approaches (Wilcoxon Rank Sum Test and Anova). As an example, Kempf showed how the Wilcoxon method was used. For the 54 mutation, there were 38 patient clinical isolates containing the 54 mutation, and the median fold reduction in susceptibility was 16. That compared to the 74 isolates that lacked the 74 mutation and the median fold change in susceptibility was 1.1. 

They found 11 protease mutations statistically associated with phenotypic resistance to ABT-378. In the active sight they identified the 82 and 84 mutations. In the "bowels" of the enzyme they found the 10, 20, 24, 63, 71, 90. In the flap region they found the 46 and 54 mutations, and 1 newcomer--the 53 mutation. Kempf said that to the best of his recollection the 53 has not been previously associated with protease resistance. Kempf reminded everyone that these mutations were not selected by ABT-378. All these mutations were selected by other protease inhibitors, and he is reporting which of those mutations contributereduced phenotypic susceptibility to ABT-378. Kempf said that these findings do not eliminate the possibility that there may be other unidentified mutations associated with ABT-378 resistance, or that there is some unique resistance profile or pathway leading to ABT-378 resistance.

Kempf said that because the trough levels for ABT-378 are so high above the EC50 for wild-type virus, they wanted to try to identify which mutations may be associated only with modest reductions in susceptibility versus mutations which if added on top of those may contribute to higher levels of resistance. They established 4 arbitrarily chosen cut-offs for phenotypic resistance (4-fold, 10-fold, 20-fold, and 40-fold). Mutations at 10, 54, 63, 71, 82, and 84 were more highly associated with relatively more moderate 4-fold and 10-fold resistance. The 10 mutation was also associated with 20-fold resistance. The 84 mutation was associated with 4-fold resistance. The mutation at position 20 was associated only with >20 fold resistance. The mutation at 53 was associated with >20 fold and most highly with >40 fold resistance. The 82 mutation was also associated with 20 and 40-fold resistance. Kempf said this at least begins to help find a pattern of mutations that can build in to contribute to what may be clinically significant resistance to ABT-378.

Viruses that display >20 fold reduced in vitro susceptibility to ABT-378 all contain the 10, 54, 63 and either the 82 and/or 84 mutations. On top on these mutations they contain a median of 3 mutations from the remaining 11 mutations identified above.

Virtually all the viruses that contain two or fewer mutations do not have mutations at 82, 84 or 90. And virtually all of the viruses that have three or more mutations have at least one of these mutations. Kempf reported that when there were 0-3 of these 11 mutations present (n=48) the median fold increase in EC50 (decrease in susceptibility) was 0.8. When 4-5 mutations were present (n=31), the median fold change was 2.7. When 6-7 mutations were present (n=26), the median fold change was 13.6, And when there were 8-10 mutations present (n=7), there was a 44-fold change in resistance (susceptibility). Kempf said that when there were up to 6-7 of these 11 mutations there was not enough phenotypic resistance or reduced susceptibility to overcome the high drug levels of ABT-378. Kempf concluded ABT-378 should have activity in individuals who have up to "at least" 7 of these mutations. Eugene Sun will present the percent of individuals in study M98-957 with <400 copies/ml in each of these categories grouped by the number of mutations they have from the 11 identified. Participants in M98-957 were NNRTI naÔve, had experience with multiple protease inhibitors, and >1000 copies/ml of viral load. The regimen they received was ABT-278, efavirenz and NRTIs. Kempf mentioned that Abbott was planning a study for individuals with NNRTI experience.

Identification of Clinically Relevant Phenotypic and Genotypic Breakpoints for ABT-378/r in Multiple PI-experienced , NNRTI-naÔve Patients

Dale Kempf and Eugene Sun, both with Abbott, discussed the unique situation with ABT-378 and its high blood levels. Since this aspect of ABT-378 has received attention previously, you may already know that Abbott reports that at a dose of 400/100 ABT-378/r (higher doses are being explored for salvage) pre-dose mean plasma trough levels are 75-fold above its serum adjusted EC50. EC50 is the drug level in plasma or blood that is necessary to inhibit 50% of HIV replication. This is a standard measure used to evaluate drug blood levels. The 75-fold level is considerably higher than the level most single protease inhibitors reach. Abbott has reported that other protease inhibitors are several fold above the EC50. Actually, there has been quite a bit of discussion and controversy at this meeting about how use of this measure is not standardized. That is, different pharmaceutical and academic researchers often use different standards when reporting EC or IC50s. This can be confusing, and since many people may not realize this comparisons of information may be difficult. Nonetheless, the high plasma levels of ABT-378 appear to have an important significance. Firstly, the effectiveness of ABT-378 in individuals with PI experience and resistance will likely be associated with the drug's high blood levels rather than a unique resistance profile, which it does not appear to have. Second, it appears that due to high blood levels ABT-378 will have higher phenotypic susceptibility cut-offs or breakpoints than other protease inhibitors. What does this mean in plain language? For example, for the sake of discussion, 12-fold reduced susceptibilty may be enough for full resistance to a given PI. But the amount of reduced susceptibility to ABT-378 may have to be significantly higher for resistance to develop since ABT-378 has high blood levels. It's in this context in which Abbott has designed and reported at this meeting their approach to analyzing the response to ABT-378 in PI experienced individuals.

Following Dale Kempf's talk on genotypic and phenotypic correlates of reduced in vitro susceptibility to ABT-378--how many mutations of the 11 identified to be associated with ABT-378 resistance development, and there correlations to reduced phenotypic susceptibility--Eugene Sun delivered an oral presentation on how these geno/pheno correlations relate to virologic response to ABT-378.

The analyses that Abbott designed are unique and gave impetus to quite a bit of discussion and controversy at this meeting. A number of researchers recommended that we need these types of analysis for all new drugs. We need to know specific phenotypic cut-offs for all drugs. Without knowing the specific cut-offs for individual drugs, results reported from HIV drug resistance assays may not be accurate enough. At the moment the same phenotypic resistance cut-offs are used for every PI. Several additional shortcomings of resistance testing received much discussion and attention: the inability of many doctors to interpret resistance test reports; their inability to apply the information appropriately; the need for education for doctors and patients about testing; there were differences of opinion about when genotypic or phenotypic testing may be more useful.

Study Design

This analysis is of participants in study M98-957 which was for individuals with multiple PI experience, NNRTI-naÔve, and HIV-RNA >1000 copies/ml. The number of median prior protease inhibitors was 3, 4 NRTIs, and 7.5 total antiretroviral drugs. Individuals received ABT-378, efavirenz, and NRTIs as chosen by treating physician. The mean HIV-RNA viral load was 4.5 log (31,600 copies/ml). All but two individual's baseline clinical viral isolates were susceptible to efavirenz. Baseline genotypic and phenotypic resistance tests were performed, and PK profiles were performed on all participants at week 5.  Additional week 24 data will be reported at Durban.

Baseline Susceptibility of Viruses to ABT-378

In study M97-765, in single PI experienced individuals, the mean reduced susceptibility was 2.8 fold, while in study M98-957 the mean reduced susceptibility was 16 fold.

Frequency of Mutations: ABT-378  Mutation Score of Baseline Clinical Isolates

Of the 11 mutations identified and reported to be associated with ABT-378 reduced susceptibility by Dale Kempf, here is a breakdown of the frequency of mutations. Nearly 80% of patients had at least 4 of the 11 mutations (10, 20, 24, 46, 53, 54, 63, 71, 82, 84, 90) in their clinical viral isolates (blood samples). There were 0 mutations in 2 isolates, 1 mutation in 6 isolates, 3 mutations in 1 isolate, 4 mutations in 7 isolates, 5 mutations in 9 isolates, 6 mutations in 11 isolates, 7 mutations in 10 isolates, 8 mutations in 3 isolates, 9 mutations in 3 isolates, and 10 mutations in 1 isolate.

Virologic Responses

At week 24, 52 of the 57 who enrolled in study M98-957 qualified for the analysis of week 24 response with 42 (81%) of these patients having had HIV RNA £400 copies/ml at week 24.

DISPOSITION OF PATIENTS (Week 24)

Analysis of Response with Respect to Baseline Phenotypic Resistance

Baseline Genotype and Week 24 Virologic Response

Discussion

I think the new data on ABT-378 was generally well received by attendees at this meeting. Doctors will be pleased to have this new drug available for individuals with extensive drug resistance. New and promising data on several additional new drugs in development was presented at this meeting--PMPA, DAPD (NRTI), BMS-232632 (PI). The data presented here suggest all three of these drugs should have antiviral activity for individuals with resistance to currently available drugs.

This early ABT-378 data suggests preliminarily a continuum of approximate phenotypic and genotypic breakpoints that may be useful in estimating the potential for a positive virologic response by a given person. By the time ABT-378 receives accelerated approval (expected near the end of 2000; application was just submitted to the FDA) both phenotypic resistance testing companies (Virco and Virologic) should be able to integrate this information on breakpoints into their reporting of test results. Abbott is in discussion now with both companies, and both companies actually have ABT-378 to use in testing with patient blood samples. However, the information and research on these cut-offs is preliminary. Continued work by Abbott is required to confirm and validate this information clinically. As more data is collected as the drug is tested and used more such validation attempts can be made. For example, it's possible that there are mutations associated with ABT-378 resistance that have not been identified yet. The influence or affect of one mutation may be greater than the others. For example, maybe the 82 mutation is more influential in causing virologic failure and reduced susceptibility. This has yet to be determined.

There was some discussion about the relative contribution of efavirenz and ABT-378 to the virologic response of individuals. I think it was generally agreed by many researchers that ABT-378 is making a substantial contribution, certainly to those individuals with possibly about 33% reduced susceptibility or less. How much contribution ABT-378 makes to virologic success for individuals with 40 fold or more reduced susceptibility appears in question. It was also estimated by Sun that this triple combination was effective for individuals with up to 7 mutations. These questions need to be continued to be addressed as they cannot absolutely be answered yet.

Parameters Associated With Virologic Response

Three stepwise regression models were performed to assess the effect of baseline phenotype and genotype in the context of other parameters that might impact virologic response. Baseline phenotype (p=0.016), genotype (p=0.024) and inhibitory quotient (trough/EC50, p=0.068) were associated indepently to be predictive of  virologic response.

In the first model, baseline phenotypic susceptibility and genotype at baseline were considered and phenotypic susceptibility remained the most closely associated with the virologic response at week 24. In model 2, in which the baseline phenotypic susceptibility was omitted, the ABT-378 mutation score remained a significant predictor of virologic response at week 24. I the third model, ABT-378 pharmacokineticparameters were considered along with the other parameters considered in models 1 and 2. Baseline phenotype was incorporated as part of 4 exploratory inhibitory quotients (IQ) representing the ratio of Ctrough, Cmin, AUC and Cmax to the serum-adjusted EC50 for each baseline isolate (IQCtrough, IQCmin, IQAUC and IQCmax, respectively). Baseline genotype entered model 3 first but was ultimately displaced by the IQC trough which remained associated with week 24 virologic response. Other parameters considered which appeared to trend to association with being predictive of response were: NRTI susceptibility , number of new NRTIs, years since HIV diagnosis, baseline patient weight.

Sun concluded that these results provide a framework for the interpretation of HIV phenotypic and genotypic testing.