July 9, 2001
After several presentations of interim results, this presentation provided the final 48-week data of this important head-to-head comparison of Combivir with nevirapine (Viramune) and Combivir with nelfinavir (Viracept) in antiretroviral-naive patients. The purpose of this study was to directly compare a non-nucleoside inhibitor-based regimen with a non-boosted protease inhibitor, both with standard nucleoside support as first-line therapy. During the Dupont 006 study, one of the registrational studies for efavirenz, the efavirenz-based regimen was at least as potent as an indinavir-based regimen with follow-up that was recently extended to three years. Other studies such as BI 1090 provided support for the activity of nevirapine in patients with high viral load and low CD4, but these types of direct comparisons had not been available for nevirapine.
This open label, randomized study assigned 70 patients to each of the arms. To recap the key results, a slightly greater proportion of patients on the nevirapine arm were suppressed below 200 copies (75% vs. 60%; P = 0.056 by intent to treat, NC=F) and below 20 copies (65% vs. 50%; P=0.065, ITT NC=F). Similar results were seen in the on treatment analysis (92% vs. 81% less than 200). Importantly, in the subgroup with >100,000 copies at entry, the trends also favored nevirapine (76% vs. 53% less than 200 copies, P=0.11). Toxicities were moderate and as expected, with GI toxicity predominating in the nelfinavir arm and rash and hepatitis in the nevirapine arm.
Those familiar with statistics will see the issues at once. The magnitude of difference favoring nevirapine over nelfinavir is similar to the difference between efavirenz and indinavir in the DMP 006 study, but this study was woefully underpowered. Thus, the study suggests that nevirapine outperformed nelfinavir, but there is a 5.6% chance that a difference this large or larger could occur by chance if the drugs were similar. Note that studies with a P value of 0.049 do not prove something that is not proven with a P value of 0.055. Rather, significance testing (i.e., P values) really is a continuous measure that tells us whether, for a given size study and difference, what the likelihood is the difference is due to chance. The problem here is that an otherwise well-designed study looking at an important question was just too small to answer the question.
Another interesting point is what happens when a study is presented many times with interim analyses. At earlier presentations, several of the comparisons were statistically significant; in the final analysis they are not.
Does this change the overall interpretation? I think not. For now, we can say that nevirapine and Combivir is at least as potent as nelfinavir and Combivir in naive patients, including those with high viral load. A direct comparison of nevirapine and efavirenz will not be available until next year when the 2NN study nears completion.
|Please note: Knowledge about HIV changes rapidly. Note the date of this summary's publication, and before treating patients or employing any therapies described in these materials, verify all information independently. If you are a patient, please consult a doctor or other medical professional before acting on any of the information presented in this summary. For a complete listing of our most recent conference coverage, click here.|