February 14, 2003
Several recent well-designed studies of naive patients have highlighted the efficacy of initial regimens based on non-nucleoside reverse transcriptase inhibitors. These studies include ACTG 384 despite the overall complexity of the design and analysis, clearly showed that AZT (zidovudine, Retrovir), 3TC (lamivudine, Epivir) and efavirenz (EFV, Sustiva) were superior to nelfinavir (NFV, Viracept)-based or d4T (stavudine, Zerit)/ddI (didanosine, Videx)-based regimens. Gilead study 903 showed remarkable sustained efficacy with either d4T, 3TC and EFV or tenofovir (TDF, Viread), 3TC and EFV. Similarly clean and straightforward studies have been lacking for nevirapine (NVP, Viramune). Several studies suggest similar results, but they are limited by small size or, in the case of the Atlantic study, the use of d4T and ddI as a backbone, which is now known to be more toxic than other backbones.
One of the most eagerly awaited studies of this conference was the 2NN study. It was a large, randomized study, which directly compared NVP and EFV against a backbone of d4T and 3TC. Two additional objectives were to establish if NVP could be dosed once daily (as suggested by the pharmacology of the drug) and to evaluate the use of dual non-nucleosides. This was based on an uncontrolled series using both EFV and NVP that suggested there might be additive efficacy.
This study took place in 65 centers in 17 countries on five continents. It was coordinated by the International AIDS Treatment Evaluation Center (IATEC) in the Netherlands. A total of 1,216 ART naive patients were randomized to one of four arms: NVP 400 mg QD, NVP 200 mg BID, EFV 600 mg daily, or EFV plus NVP in a 1:2:2:1 fashion.
All were used on a backbone of d4T and 3TC BID. The median CD4 count was 190 and the viral load 4.7 log. As might be expected with an international trial, the patients were heterogeneous and 35 percent were women. Hepatitis C and B co-infection were somewhat uncommon, at nine percent and five percent, respectively.
The predefined primary endpoint was treatment failure at 48 weeks, defined as failure to drop one log in 12 weeks, two viral loads greater than 50 after week 24, need to switch therapy, or an AIDS-defining event. The usual, well-known endpoints of percent less than 50 and 400 copies were also included as secondary endpoints. As is good statistical practice, the primary comparisons were specified in advance, to differentiate the primary study analysis from exploratory analysis (less kindly referred to as data-torturing).
The bottom line was that the NVP QD, NVP BID and EFV QD regimens were equivalent in overall efficacy by a number of measures, and generally superior to the dual non nucleoside regimen. It is worth looking at the details however.
Study conduct was quite successful. Eighty-three percent completed the study, evenly distributed among the arms. The proportion achieving treatment success (that is, the primary endpoint) was 56.4, 56.3, 62.3, and 46.9 for the NVP QD, NVP BID, EFV and dual NNRTI arms respectively. The only significant difference was EFV compared to the dual NNRTI arm. Virologic success was seen in 65, 63, 68, and 62 percent. There was no difference when the results were stratified by baseline viral load less than or more than 100,000 copies.
Using the well known strict analysis of percent less than 400 copies by intent to treat missing = failure, the proportions were 89, 82, 87 and 80 percent. Using less than 50 copies, the proportions were 70, 65, 70, and 63 percent. Despite the danger of comparing across trials, these results are similar to those of the 903 study and of the original DMP 006 study.
Clinical side effects were more common in the dual nucleoside arm, and were more likely to result in stopping medication in that arm (30 percent). Clinical hepatitis (gr 3-4) was uncommon: 1.4, 2.1, 0.3, and 1.0 percent. Central nervous system symptoms were more common in the EFV-containing arms. Laboratory evidence of liver inflammation was significantly more common in the NVP-containing arms, particularly the QD arm. Grade 3-4 liver test abnormalities were seen in 13.2, 7.8, 4.5, and 8.6 percent.
There were three deaths of note among the 1,200 patients with fairly advanced HIV disease. One person died of lactic acidosis, linked to d4T. One patient died of liver disease (about 0.15 percent) linked to NVP. Another patient died of a hospital complication after being hospitalized for NVP-related Stevens Johnson syndrome. Lipid changes were favorable in all groups with increased HDL cholesterol and modest changes in total cholesterol. The ratio of total cholesterol to HDL, a marker of cardiac risk, was more favorable in the NVP arm.
|Treatment arm||NVP(od)||NVP(bd)||EFV||NVP + EFV||p value for Rx Groups|
|n = 220||n = 387||n = 400||n = 209|
|pVL < 50 copies/mL,%||70.0||65.4||70.0||62.7||1.00||0.17||0.11||0.07|
|CD4 increase, cells/mm3||170.0||160.0||160.0||150.0||0.49||0.74||0.91||0.71|
|clinical AE1,%||27.7||27.1||22.3||35.4||0.13||0.11||0.09||< 0.001|
|liver associated lab AE2, %||13.2||7.8||4.5||8.6||< 0.001||0.06||0.13||0.04|
|other lab AE3, %||8.2||12.9||8.8||9.6||0.81||0.06||0.61||0.74|
What are the clinical and take home messages? It certainly appears that NVP QD or BID is equivalent to EFV in naive patients with a good and well-tolerated backbone. In this population, neither liver inflammation with NVP nor neurologic complications with EFV were a major problem.
The marketing wars will begin. However, patients and providers will need to balance the modest differences between the drugs based on underlying risk. For example, a patient with a history of depression and multiple risk factors for heart disease would do better on NVP, while someone with significant liver disease due to hepatitis B or C may be a better candidate for EFV. The risk of birth defects for EFV will need to be weighed against the somewhat higher risk of serious rash illness for NVP.