| |
Atazanavir/ritonavir versus lopinavir/ritonavir: equivalent or different efficacy profiles?
|
| |
| |
AIDS: Volume 19(17) 18 November 2005 p 2054-2055
Hill, Andrew
Department of Pharmacology, University of Liverpool, Liverpool, UK.
Received 25 May, 2005
Accepted 21 June, 2005
Johnson and colleagues [1] presented results from a randomized clinical trial of atazanavir/ritonavir (ATV/r), lopinavir/ritonavir (LPV/r) and atazanavir/saquinavir in a total of 358 treatment-experienced patients. The conclusion, that ATV/r is as effective as LPV/r, does not seem to be supported by further analysis of the trial data.
The most common primary efficacy analysis of HIV clinical trials is the proportion of patients with HIV RNA under 50 copies/ml, using an intent-to-treat population, in which patients discontinuing randomized medication or with missing data are classified as treatment failures. Clinical trials for drug licensing are normally powered, using this HIV RNA endpoint, to detect differences of at least 10-12% between treatment arms, or to show equivalence or non-inferiority within these limits [2]. A normal regulatory trial powered within these limits would have 250-400 patients per arm. The BMS-045 trial, with approximately 120 patients per arm, is clearly underpowered to assess efficacy by the normal endpoint of HIV RNA 50 copies/ml.
For the BMS-045 trial, the intent-to-treat analysis of the 50 copy HIV-RNA endpoint showed responses of 38% for ATV/r and 46% for LPV/r at week 48; the ATV/r arm was 8% inferior to LPV/r at week 48, with 95% confidence intervals of -20.4% to +4.4%. These 95% confidence intervals fall outside the 10-12% limits that would normally be used to define equivalent or non-inferior efficacy.
Intent-to-treat analyses in open-label trials typically include only those patients who receive at least one dose of study drug; patients who are randomly assigned but are never treated are normally excluded from this type of analysis (intent-to-treat exposed method). Of the patients randomly assigned to the BMS-045 trial, five patients in the LPV/r arm but only one patient in the ATV/r arm were assigned but never treated. Those untreated patients had been included as non-responders in the published results. If an intent-to-treat exposed analysis is conducted, including only patients who received study medication, the percentage with HIV RNA less than 50 copies/ml at week 48 would be 38% for the ATV/r arm versus 48% for LPV/r, a difference of 9.6% (95% confidence intervals -22.1 to +2.8%). Differences in efficacy between arms in the region of 8-10% have been found to be statistically significant in larger randomized clinical trials, as shown in Table 1 [1,3,4].
Although the results from the as-treated analyses show more similar findings for the two arms, the discontinuation rate in the ATV/r arm (22%) is higher than for the LPV/r arm (11%), and it is not clear what proportion of patients withdrew with detectable versus undetectable HIV-RNA levels across the treatment arms.
The primary efficacy endpoint for the BMS-045 trial was the log10 reduction in HIV RNA using a time-averaged difference (TAD) method. This method has been used for the analysis of other trials in treatment-experienced patients in which rates of HIV-RNA undetectability are expected to be low (for example the TORO clinical trials of enfuvirtide). However, for more recent Food and Drug Administration product labels, the log10 reduction in HIV RNA by TAD is no longer included in the summary of treatment efficacy [5].
There are two main problems with using log10 reductions in HIV RNA to compare treatment groups. First, when a significant proportion of treated patients show HIV-RNA reductions under the assay detection limit, the log10 reduction cannot easily be measured because the lower detection limit has already been achieved: this will tend to make treatment groups appear more similar. Second, there are not standardized methods for including data from early withdrawals. With the TAD method quoted in the report, it is not apparent what assumptions are made for patients who withdraw prematurely or have missing data. A patient who withdraws with undetectable HIV-RNA levels might have these values carried forward, censored, or reset at a zero change from baseline.
In summary, the question remains whether the study's conclusion that ATV/r is as effective as LPV/r is truly describing the overall efficacy of the treatment arms, or is the conclusion rather a result of the chosen statistical analysis in conjunction with a relatively small sample size for a study of this type?
References
1. Johnson M, Grinszstejn B, Rodriguez C, Coco J, DeJesus E, Lazzarin A, et al. Atazanavir plus ritonavir or saquinavir, and lopinavir/ritonavir in patients experiencing multiple virological failures. AIDS 2005; 19:685-694.
AIDS: Volume 19(7) 29 April 2005 p 685-694
Atazanavir plus ritonavir or saquinavir, and lopinavir/ritonavir in patients experiencing multiple virological failures
Johnson, Margareta; Grinsztejn, Beatrizb; Rodriguez, Claudiac; Coco, Jeffreyd; DeJesus, Edwine; Lazzarin, Adrianof; Lichtenstein, Kennethg; Rightmire, Annah; Sankoh, Serapi; Wilber, Richardh
From the aRoyal Free Hospital, London, UK
Abstract
Objective: To evaluate atazanavir/ritonavir (ATV/RTV) (300/100 mg) once daily, atazanavir/saquinavir (ATV/SQV) (400/1200 mg) once daily, and lopinavir/ritonavir (LPV/RTV) (400/100 mg) twice daily, each with tenofovir (300 mg) once daily and a nucleoside reverse transcriptase inhibitor in treatment-experienced HIV-infected patients.
Methods: Randomized, open-label, 48-week multicenter trial of 358 randomized adult patients who had failed two or more prior HAART regimens with baseline HIV RNA ≥ 1000 copies/ml and CD4 cell count ≥ 50 _ 106 cells/l.
Results:
The primary efficacy endpoint [plasma HIV RNA reduction assessed by time-averaged difference (TAD)] was similar for ATV/RTV and LPV/RTV [TAD 0.13; 97.5% confidence interval, -0.12 to 0.39] at 48 weeks. Mean reductions from baseline for ATV/RTV and LPV/RTV were comparable at 1.93 and 1.87 log10 copies/ml, respectively. Mean CD4 cell count increases were 110 and 121 _ 106 cells/l for ATV/RTV, and LPV/RTV, respectively. The efficacy of ATV/SQV was lower than LPV/RTV by both these parameters. Declines in total cholesterol and fasting triglycerides were greater with ATV/RTV and ATV/SQV than with LPV/RTV (P ≦ 0.005). Lipids in the LPV/RTV arm at week 48 generally increased from baseline. Lipid-lowering agents were used more frequently in the LPV/RTV arm than in the ATV arms (P < 0.05 versus ATV/RTV), as were antidiarrheal agents (P ≦ 0.04 versus both ATV treatments). No new or unique safety findings emerged.
Conclusions: ATV boosted with RTV is as effective and well tolerated as LPV/RTV in treatment-experienced patients, with a more favorable impact on serum lipids. Pharmacokinetically enhanced ATV provides a suitable choice for therapy of treatment-experienced HIV-infected patients.
|
|
| |
| |
|
|
|
