Analyzing Evidence in HIV Clinical Trials
Submitted by shaunakanel on June 18, 2009 - 13:06
Ida Sim, University of California, San Francisco
Human Immunodeficiency Virus/Acquired Immunodeficiency Syndrome (HIV/AIDS) continues to be a worldwide pandemic. Two critical but controversial issues in HIV/AIDS care include how to prevent mother-to-child transmission (MTCT) of the virus, and whether AIDS patients on chronic treatment can benefit from structured treatment interruptions (STI) of anti-retroviral drugs.
As with clinical research in general, the highest quality evidence on these topics comes from randomized clinical trials (RCTs), of which there are approximately 40 addressing MTCT and 13 addressing STI in chronic HIV infection. However, no clear clinical conclusions arise from the sum of these trials in part because the trials differ so much in the treatment regimens tested, the patients enrolled, and the outcomes assessed. These differences among trials severely complicate the task of discovering what the overall evidence is on MTCT prevention or the use of STI.
Typically, the overall evidence from a set of related trials is discovered through meta-analysis, a statistical method for combining results from "similar" trials to increase statistical power for detecting effects. Newer meta-analytic methods include extensive exploration of how trial results vary with differences in study design and clinical characteristics across trials. For example, combining results across all STI trials may show that STI works better in patients with a particular HIV genotype, even though no one trial had sufficient numbers of patients to show such a finding. Such meta-analytic explorations are currently labor-intensive and inflexible because very few software programs support robust exploration of heterogeneity for meta-analysis. Because of the many and varied differences among MTCT and STI trials, meta-analysis of these trials is particularly difficult. As a result, unrecognized insights into MTCT prevention and STI are likely hiding in these valuable randomized trials.
This Driving Biological Project aims to increase understanding of the trials that have been done on MTCT prevention and STI by developing a principled way to structure, summarize, and visualize evidence from these trials. We are creating annotations for HIV/AIDS clinical trials, and abstracting and annotating MTCT and STI trials and using the Center's technologies to provide two ontology-driven Web tools called CTSearch and CTExplorer for visualizing, understanding, and analyzing the trials. We thus are using the Center's resources to help answer whether, how, when, and to whom to administer MTCT or STI preventive strategies in hopes of decreasing the HIV/AIDS burden worldwide.