An experienced team of statisticians will develop, implement, and refine new methods for addressing several areas of great importance in HIV research. The many available anti-HIV drugs and combinations, plus the many possible strategies for their use, have greatly complicated the study of treatment effects. This proposal will assess the effects of treatment regimens and strategies from observational studies in realistic settings using methods for simultaneously modeling longitudinal data (e.g., CD4 counts) and time-to-event data (e.g., time to a clinical event). One approach will employ models with latent classes that have different distributions of longitudinal marker trajectories, which themselves both influence and are influenced by treatment. Another approach for such data will employ counterfactual causal inference. The two approaches will be compared, although synthesis of the two may retain advantages of both. High rates of hepatitis C co-infection make HCV a potentially important cause of morbidity and mortality among HIV-infected persons. Recent methods for current status data will model uncertainty in dates of HCV infection, and proposed progression modeling will incorporate each patient's fitted distribution of possible infection times. This work will develop transition models that allow time-varying covariates, apply them to modeling fibrosis scores from liver biopsies, and adapt other methods to modeling HCV progression with uncertain infection times. With increasing recent interest in HIV transmission among subgroups of individuals characterized by differing treatment status, viral subtype, or mode of exposure, case-control and other complex sampling designs are becoming more common. Ignoring the design can lead to biased estimates. Proposed work will extend methods for modeling transmission risk in terms of covariates and per-contact infectivity to apply to doubly-censored current status data arising from such designs. Genetic determinants of HIV phenotypes, including virulence, fitness, and drug resistance, are of increasing interest. Additional work will adapt tree-structured and bump-hunting methods for such genotype-phenotype data to handle the high dimension and non-numeric nature of predictive genotypic data. Extensions will tackle more complicated situations, including quasi-speciation, time varying viral genotypes, multiple resistance phenotypes, and the availability of 3D structural information. Applications to relevant data sets will provide both realistic testing and new substantive insights. Software implementing the methods and technical papers will be available through the Internet at a project-related website. ? ?

Agency
National Institute of Health (NIH)
Institute
National Institute of Allergy and Infectious Diseases (NIAID)
Type
Research Project (R01)
Project #
5R01AI055085-03
Application #
6708024
Study Section
AIDS and Related Research 8 (AARR)
Program Officer
Gezmu, Misrak
Project Start
2002-09-30
Project End
2006-09-30
Budget Start
2004-04-01
Budget End
2006-09-30
Support Year
3
Fiscal Year
2004
Total Cost
$571,266
Indirect Cost
Name
University of California San Francisco
Department
Public Health & Prev Medicine
Type
Schools of Medicine
DUNS #
094878337
City
San Francisco
State
CA
Country
United States
Zip Code
94143
Pintye, Jillian; Bacchetti, Peter; Teeraananchai, Sirinya et al. (2017) Brief Report: Lopinavir Hair Concentrations Are the Strongest Predictor of Viremia in HIV-Infected Asian Children and Adolescents on Second-Line Antiretroviral Therapy. J Acquir Immune Defic Syndr 76:367-371
Bacchetti, Peter; Tien, Phyllis C; Seaberg, Eric C et al. (2007) Estimating past hepatitis C infection risk from reported risk factor histories: implications for imputing age of infection and modeling fibrosis progression. BMC Infect Dis 7:145
Lin, Haiqun; McCulloch, Charles E; Rosenheck, Robert A (2004) Latent pattern mixture models for informative intermittent missing data in longitudinal studies. Biometrics 60:295-305