Predictability of Genotypic Resistance Testing
CPCRA GART Study 045: A Pilot Study of the Short Term Effects of Antiretroviral
Management Based on plasma Genotypic Antiretroviral Resistance Testing (GART)
in Patients Failing Antiretroviral Therapy
authors: JD Baxter, D Mayers, T Merigan, D Wentworth, D Neaton. CPCRA
046 Study Team
John Baxter, from Cooper Hospital/UMC Camden NJ, presented results from this study which evaluates using genotypic testing in combination with expert advice from a pre-selected panel of expert virologists interpreting the test results and suggesting regimens based on the test results. The study data suggests that if a person uses a genotypic test, has expert advice on how to interpret and select a regimen based on the test results, and actually follows that advice in selecting a regimen, they should be able to select drugs that they are more sensitive to and obtain a greater viral load reduction.
Patients were included in this study if they experienced a 3-fold or greater rise in HIV-RNA while taking for greater than or equal to 16 weeks a combination therapy consisting of 2 NRTIs and a protease inhibitor. Patients also had to have a CD4 count between 50 and 500 and at least 12 months of previous antiretroviral therapy. Plasma samples were collected at baseline for viral load and genotypic testing. The study virologists reviewed the genotype results and the patient's antiretroviral history, and prepared a consensus "expert" GART report with genotype interpretations and suggested treatment regimens. 153 patients were randomized between July 1997 and December 1998-- 78 to the GART group and 75 to the no GART group. Prior to randomization doctors were asked to propose a regimen that they would prescribe in the absence of the GART report. Patients in the no GART group had these proposed regimens prescribed at the time of randomization while doctors caring for the GART group patients prescribed regimens after reviewing the GART report. They did not always follow the GART report recommendations. Patients had follow-up visits at 4, 8 and 12 weeks following randomization at which time repeat viral load tests were performed.
Baseline characteristics for the GART and no GART groups were similar: mean CD4s were about 230, median viral load was about 28,000 copies/ml in the GART group & 25,000 copies/ml in the no GART group; most patients were using AZT/3TC (about 45% in each group) or d4T/3TC (about 40% in each group) with indinavir (about 54% in each group); nelfinavir (about 33% in each group); ritonavir (about 6% in each group); saquinavir (about 5% in each group). 53% in the GART group and 44% in the no GART group were failing their first PI.
Baseline genotypic testing showed that about 73% of patient's isolates had at least one major RT and at least 1 major PI resistance mutation; 20% had at least 1 major RT mutation without any PI mutation; 4.6% had no major mutations; 2.0% had only PI mutations. The overall frequency of specific mutations were:
| M184V (3TC) | 82% |
| 215F/Y (AZT) | 61% |
| 151 MDR* | 3% |
| any NNRTI mutation | 4% |
| 30N | 14% |
| 46I/L | 31% |
| 82A/F/T | 34% |
| 90M | 31% |
| 84V | 8% |
Results
The primary endpoint of the study was change in viral load from baseline to an average of the 4 and 8 week viral load measures.
| Avg (4 & 8 wks) |
The investigator said that the effect of GART was seen across all baseline subgroups including entry drug regimen, number of previous PIs, CD4 levels, patient demographics, and genotypic patterns.
When looking at the differences in viral load responses for each group by study week, at each time point (weeks 4, 8 and 12) the treatment differences were statistically significant. At 12 weeks, the difference in viral load reduction between the two groups was 0.39 log (about -1.0 log in GART group and about -0.5 log in no GART group) with a P-value of .02; at week 4, the mean HIV-RNA reduction was about -1.3 log in the GART group and about -0.8 log in the no GART group; at week 8, the viral load reduction was about -1.1 log in the GART group, and about -0.6 log in the no GART group. I estimated these log reductions by visual observation of the graph Baxter presented on a slide. As you can see the maximal viral load reductions seen at week 4 for both groups started decreasing after week 4.
When comparing the percent of patients below 500 copies/ml in the two groups, the patients in the GART group had a statistically significant greater number below the LOD (level of detection) at weeks 4 and 8. Not all patients have completed follow-up, but with 80% having reached their 12-week visit (n=121) 29% are undetectable in the GART group vs 17% in the no GART group, with a P-value of 0.15. Obviously, 29% <500 copies/ml is still less than a desirable outcome. At week 4 (n=148), about 45% and 20% had <500 copies/ml in the GART and no GART groups, respectively; at week 8 (n=137), about 50% and 20% had <500 copies/ml, respectively.
GART Group Received a Greater Number of Drugs They Were Not Resistant To. To better understand the differences in response observed between the 2 groups investigators looked at viral changes by the number of active drugs prescribed. Active drugs were defined as drugs for which the protocol virologists interpreted as not resistant for the given individuals. Viral load responses in each group were associated with the number of active drugs prescribed. Combining patients in both groups, for those patients receiving less than or equal to 1 active drug the average viral load decline was -0.10 log. In contrast, for patients receiving greater than or equal to 4 active drugs viral load reduction was on average -1.25 log. Adjusting for baseline viral load and entry regimen each additional active drug prescribed was associated with a 0.26 log reduction.
Patients in the GART group received a greater number of active drugs. 86% in the GART group received 3 or more active drugs while less than half of the patients in the no GART group received 3 or more active drugs. Baxter said he believes this information explains the differences in viral load response between the 2 arms. Of those in both groups who received 3 active drugs the average viral load reduction was about 1 log; about 45% in the GART group and about 35% in the no GART group received 3 active drugs; about 10% in the GART group and about 45% in the no GART group received 2 active drugs; about 40% in the GART group received 4 or more active drugs while only about 10% in the no GART group received 4 or more active drugs.
Not All Doctors Used Expert Gart Suggested Regimens. Study investigators found that doctors in the GART group changed the regimens they had proposed to use for 83% of their patients after review of the GART report. But only for 54% of the patients did doctors report using one of the regimens suggested by the study virologists. Among the 15 sites participating in this study this percentage ranged from 0 to 100%.
The investigators measured the viral load response by how much the doctors at units adhered to treatment suggestions. At units who adhered to suggested regimen recommendations by >80% the GART group had average viral load reductions of -1.47 log and the no GART group at these sites had an average viral load reduction of -0.47 log. The P-value was .0006. The treatment difference was less prominent for units that adhered less frequently. At units where doctors adhered to the virologists' recommendations 60-80% of the time the GART group had an average viral load reduction of -1.21 log vs a -0.55 viral load reduction in the no GART group, with a P-value of .01. At the units where doctors adhered to the recommended regimens less than 60% the GART group average viral load reduction was -0.98 log vs -0.74 log in the no GART group, with a P-value of .12. P-values less than .05 are statistically significant. This data suggests that if all doctors had followed the suggested regimens the differences in the treatment groups may have been greater and individuals randomized to the GART group may have had greater reductions in viral load.
Baxter concluded that GART with expert advice in patients failing antiretroviral therapy was superior to no GART, as measured by short-term viral load responses. One limitation of the study, however, is that the study did not compare the effect of expert advice guided by only a person's drug history, or the effect of genotypic test results without expert advice. The study only compared genotypic test results plus interpretation of those results by an expert panel to not using a genotypic test. However, it is my understanding that an algorithm is being developed to interpret genotypic test results by an expert panel. I think this tool may be available in the near future at sites making certain genotypic testing available. The utility of such a method for interpreting test results would have to be further evaluated.
At last Summer's resistance meeting at Lake Maggiori several studies were presented showing that results of genotypic and phenotypic resistance testing may be helpful in selecting regimens which are more effective. The results of these studies were detailed in the recent NATAP newsletter which is available on the NATAP web site or a hard copy can be obtained by contacting the NATAP office (212 219-0106). At that meeting several leading researchers expressed that they were impressed with the results of those studies, but they still had reservations. It is generally felt that the development of resistance testing is in the early stages and that further studies must be conducted to confirm the utility and effectiveness of using these tests. Still, the study being reported here suggests that genotypic testing may be helpful in selecting regimens.