Speed of viral load decline is a sign of more effective antiretroviral regimen

A substudy of a large clinical trial has shown that viral load declined more quickly in people treated with AZT, 3TC and efavirenz (Sustiva) than in those treated with AZT, 3TC and abacavir (Ziagen). The faster viral load decline corresponded with the better virologic response in the efavirenz group, leading to suggestions that measurements of viral load decline might be an early predictor of longer-term response in clinical trials.

Successful antiretroviral therapy in an HIV-positive individual all but stops the production of new virus. Remaining latently-infected cells result in an initially measurable viral load, which declines as the pool of HIV-infected cells dies off. Several distinct phases of viral load decline (or “decay”) occur as cells die off at different rates.

A first, rapid phase of decline (lasting seven to ten days after the start of therapy) occurs as short-lived cells, such as activated CD4 cells, die off. A second phase, corresponding to longer-lived infected cells, shows slower decline and lasts several months. Finally, with only latently infected resting cells remaining, viral load continues to decay at a very slow rate (if antiretroviral therapy continues to suppress HIV production).

The well-known clinical trial ACTG A5095 compared three treatment regimens: AZT/3TC/efavirenz, AZT/3TC/abacavir, and a four-drug regimen of AZT/3TC/abacavir/efavirenz. A5095 established the superiority of the efavirenz-containing regimen to the triple-nucleoside abacavir-containing regimen (Gulick 2004; Gulick 2006). A substudy (A5166s) of A5095 examined the rates at which viral load dropped (decayed) among people in the different treatment arms: the results were published in the Journal of Infectious Diseases on April 15^th.

This substudy involved 64 treatment-naive participants in the ACTG A5095 trial (42% black, 30% Hispanic, and 25% white; median baseline CD4 count of 261 cells/mm³ and viral load 4.7 log10 copies/mL). Overall, the first phase of viral load decline was faster for those on the three-drug efavirenz regimen than for the triple-nucleoside combination. Viral load measurements for this substudy were taken prior to and at study entry, and at days 2, 7, 10 and weeks 2, 4 and 8 (by the Roche HIV-1 Monitor Assay).

Viral load fell at a faster rate in the efavirenz (EFV) group than in the abacavir (ABC) group, during both the first phase and the second; the four-drug group was intermediate between the two. Viral load decay was modelled by a curve with two smooth exponential phases (“biexponential”); decay rates were estimated by fitting the observed data to the curve. (In the model, the viral load decreases by a fixed percentage each day during each phase: this also yields a figure for the “half-life”, or time it takes for half an existing population of cells to disappear.) This resulted in the following estimates:

(p=.02 for the first phase and p=.09 for the second, for the differences between the ABC and EFV groups, Differences between the four-drug group and the other groups were not statistically significant.)

An alternate modeling method was also applied, resulting in slightly different absolute values but similar differences between the treatment arms. Also, although women are known to exhibit generally lower viral loads than men, no differences in the rate of viral load decline was seen between men and women.

The researchers concluded that the “slower decay rate … among recipients of the triple-nucleoside regimen suggests that this regimen resulted in less complete inhibition of HIV-1 replication, compared with an efavirenz -based regimen. A comparison of the median decay rates … suggests that the triple-nucleoside regimen was ~83% as effective as the three-drug efavirenz regimen at interrupting the production of infectious virions.” Similarly, “a lack of difference in viral decay between three- and four-drug efavirenz -based regimens also paralleled the lack of [clinical] difference over the course of three years of follow-up.”

In an accompanying editorial, Martin Markowitz and Alan Perelson note that the differences in viral load decay rates detected very early in this study accurately predicted the longer-term virologic outcomes of the treatments in the clinical trial. Since such “true differences in antiviral activity can be discerned” by monitoring these viral dynamics, they suggest that early analysis of such data might “prevent the undesired prolongation of inferior treatment arms”. The editorial does note that “long-term treatment success was not predicted” by these early changes, since adherence, tolerability, “and perhaps additional factors” influence the long-term clinical outcomes.

The findings from ACTG A5095 have been reinforced by a recent analysis of the ACTG 5142 study, which compared efavirenz or Kaletra-based regimens with an NRTI-sparing combination of Kaletra and efavirenz.

That study found that day 7 viral load was significantly correlated with week 48 virological outcome. Phase 1 viral decay half-life was significantly shorter for efavirenz than Kaletra (1.1 vs 1.3 days, p = 0.03),.