Characterizing the Progression of Viral Mutations Over Time

Development and spread of resistance of human immunodeficiency virus type 1 to antiretroviral therapies is a serious medical and public health concern. A wide variety of mutations have been identified that either singly or in combination reduce the susceptibility of the virus to available therapies. This paper describes methods for understanding the genetic pathways that lead to high-level drug resistance under selective drug pressure, as well as for estimating the rates at which viral populations progress along these pathways. These methods can be used to determine whether the presence of certain mutations among drug-sensitive viruses predispose a patient under a particular treatment to develop patterns of mutations that confer high-level drug resistance. Our approach assumes that viral genotypes can be characterized as belonging to discrete states, defined by patterns of viral mutations, and considers two approaches to modeling the rates of transition between these states. The first approach treats the state at a given time point as known, whereas the second treats this as a latent variable. We apply our methods to genetic sequences of viruses cloned from the plasma of 170 patients who participated in three phase II clinical studies of efavirenz combination therapy (DMP 266-003, -004, -005). Multiple viral clones are available from each plasma sample at each time of measurement, allowing for consideration of the effect of minority species on the evolution of the viral populations infecting patients; the availability of such information motivates the second analytic approach. The sequences can be found in the Stanford HIV RT and Protease Sequence Database.