Article Text

  1. V. Gajendran,
  2. J Kahn*,
  3. M. Mehta*
  1. University of California, Davis, Davis, CA
  2. *Stanford University, Stanford, CA


Purpose The treatment of human immunodeficiency virus (HIV) is challenging because the narrow therapeutic range of existing therapies creates a difficult tradeoff between efficacy and toxicity. Adverse side effects occur at high drug concentrations, and HIV's rapid mutation and turnover rates lead to the development of drug-resistant strains at low drug concentrations. The continual maintenance of an optimal therapeutic range over the long term requires predicting the dynamic course of infection and making timely adjustments to treatment.

Methods We constructed a prototype software application in MATLAB that allows clinicians to interactively monitor as well as predict short-term and long-term variations in HIV's key clinical outcome parameters. To accurately capture the effects of patient-specific parameters, drug-drug interactions, and the effects of different drug combinations on the activities of the various viral strains, our design uses a database of HIV drugs with their published pharmacokinetic and pharmacodynamic models, a database of the different viral strains with their viral activity parameters, the most current mathematical models for T-cell infection and replication, and a database of patients with relevant information from their medical records. The results of the analysis are presented as an interactive graph depicting time-dependent serum drug levels, CD4+ count, and viral loads of all strains. Based on this information about the expected course of the disease with the current regimen, the clinician may explore other treatment options using our model.

Results Our model was tested on real patient data involving infections by single and multiple viral strains and wild-type and mutant strains. The recommended treatment regimens using our model coincided closely with those prescribed by expert clinicians in the field. In a small number of cases, our model failed to prescribe the appropriate best course of treatment. These were due to specific pharmacogenomic factors not captured by our model, which significantly affected patient drug metabolism and response to treatment.

Conclusions An integrated pharmacokinetic-pharmacodynamic model for HIV treatment decisions can produce therapeutically relevant results as corroborated by expert physicians. The prognostic capability of such a model allows physicians to better respond to the challenge of having to constantly stay within a narrow therapeutic window. Incorporation of important pharmacogenomic factors unique to every patient is critical, however, before such a tool becomes truly useful in clinical practice.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.