Frost SD, Holte S, Wang L, Daar ES, Richman DD, Little SJ; Conference on Retroviruses and Opportunistic Infections.

Abstr 10th Conf Retrovir Oppor Infect Feb 10 14 2003 Hynes Conv Cent Boston Mass USA Conf Retrovir Oppor Infect 10th 2003 Boston Mass. 2003 Feb 10-14; 10: abstract no. 509.

Univ of California at San Diego

BACKGROUND: "Viral setpoint" is a term used to describe the quasi-steady viral load during the asymptomatic period of HIV-1 infection with high setpoints correlated with rapid disease progression, high infectiousness, and poorer responses to treatment. Previous studies have used different ad hoc definitions of setpoint based on time since infection or shifts in viral load. We have modeled the viral dynamics during early HIV infection with a nonparametric mixed effects model, which allows us 1) to describe both population- and individual-level viral dynamics; 2) to fit and test for nonlinear dynamics; 3) to identify outlying individuals who show shifts in viral setpoint; and 4) to generate simple, objective rules of thumb for defining setpoint.METHODS: Viral load data was obtained from individuals enrolled in the San Diego and Los Angeles sites of the Acute Infection and Early Disease Research Program from January 1995 to April 2002. Analyses were restricted to time points prior to treatment and up to 2 yrs following the estimated date of infection. A nonparametric mixed effects model, where the mean viral load and individual deviations from the mean are modeled using smoothing splines, was fitted using maximum likelihood. Nonlinear viral dynamics were tested by comparing the likelihood under the full model with a restricted model where viral loads are assumed to change linearly over time.RESULTS: There was statistically significant evidence for nonlinear viral dynamics both at the population level and at the individual level when all viral loads up to 2 yrs following infection were analyzed (p < 0.05). A linear model provided a good fit when viral loads obtained during the first 5 months were excluded. However, a number of outlying individuals were identified that showed dramatic increases in viral load during early infection.CONCLUSIONS: Nonparametric mixed effects models offer a powerful and flexible way to model viral dynamics without assuming a priori that individuals have reached setpoint. These models permit identification of individuals with dramatic changes in setpoint that may warrant further study. Our model suggests the following quantitative rules of thumb for identifying viral loads not at setpoint; 1) high (> 5.5 log) viral loads; 2) viral loads obtained during the first 5 months of infection; and 3) a difference > 0.5 log10 in viral loads taken > 3 wks apart.

Publication Types:

• Meeting Abstracts

Keywords:

• Communicable Diseases
• Disease Progression
• HIV Infections
• Linear Models
• Los Angeles
• Models, Biological
• Nonlinear Dynamics
• Viral Load
• organization & administration
• virology

Other ID:

• GWAIDS0021496

UI: 102261120

From Meeting Abstracts