Department of Biostatistics
HIV Working Group
2015 - 2016
ABSTRACT: Randomized treatment studies with time-to-event outcomes often require long term follow-up of patients in order to observe a sufficient number of events to estimate and test for a treatment effect. The availability of a surrogate marker that could be used to estimate the treatment effect and could potentially be observed earlier than the primary outcome would allow researchers to make conclusions regarding the treatment effect with less required follow-up time and resources. Previous research on identifying and validating surrogate markers has often focused on estimation of the proportion of treatment effect explained by a surrogate marker since a valid surrogate marker should capture a large proportion of the true treatment effect on the primary outcome. However, current methods to estimate the proportion of treatment effect explained usually require restrictive model assumptions that may not hold in practice. In addition, current methods to handle censoring or the occurrence of the primary outcome before the time of surrogate marker measurement are not available. We propose a novel definition of the proportion of treatment effect explained by surrogate information collected up to a specified time in the setting of a time-to-event primary outcome. We propose a robust nonparametric procedure to estimate the defined quantity using censored data and use a perturbation-resampling procedure for variance estimation and to obtain confidence intervals. Simulation studies demonstrate that the proposed procedures perform well in finite samples. We illustrate the proposed procedures by investigating two potential surrogate markers for diabetes using data from the Diabetes Prevention Program.
ABSTRACT: When the world committed to funding mass HIV treatment for millions of HIV-infected people in sub-Saharan Africa more than a decade ago, the efficacy of HIV treatment in reducing HIV-related mortality had been firmly established in clinical trials and cohort studies. However, the population health and economic impacts of mass HIV treatment, delivered through resource-poor public-sector health systems to poor populations with generalized epidemics, was poorly understood. Population-based HIV research has quantified many of the positive impacts of HIV mass treatment, but also demonstrated important impediments to achieving the full population health benefits of HIV treatment. I will present new results on interventions to increase access to mass HIV treatment, and health and economic treatment impacts, using cohort data from a large HIV treatment program in rural KwaZulu-Natal, South Africa, linked to population-based longitudinal health and demographic surveillance data.
ABSTRACT: Clinical trials have demonstrated that early initiation of HAART will decrease HIV transmission, while other studies have demonstrated the efficacy of the use of pre-exposure prophylaxis (PrEP) can decrease HIV incidence in at risk persons. Ken Mayer has been an investigator in several of these pivotal studies (HPTN 052, iPrEX) and will discuss the reasons why PrEP may be an important part of a comprehensive HIV prevention strategy, even while scaling up earlier initiation of HIV treatment. He will discuss the current status of first generation PrEP, and the challenges to wider implementation. He will also discuss new concepts of antiretroviral and immuno-prophylaxis, that may obviate the challenges of daily dosing.
ABSTRACT: We increasingly recognize that the intergeneric space between protein coding genes (PCGs) contains highly ordered regulatory elements that control expression and function of PCGs and in themselves can be actively transcribed molecules. Indeed, over 50% of genome-wide association (GWA) studies of complex traits identify single nucleotide polymorphisms (SNPs) that fall in intergenic regions and it is only recently becoming apparent that these regions are highly organized to perform specific functions. At the same time, understanding the complex interplay among PCGs and regulatory elements, such as long intergeneric non-coding RNAs, requires rigorous interrogation with analytic tools designed for discerning the relative contributions of overlapping genomic regions. This talk describes two post-analytic strategies to address the complexity of this investigation by moving beyond single point analysis to the application of sound analytic strategoes for integrating and combining information from emerging genomic taxonomies. The first is a genetic class association test and the the second is a novel application of Bayesian variable selection using an expectation maximization algorithm. Several large GWA meta-analysis data resources in HIV and cardiometabolic disease (CMD) research are leveraged and the findings are consistent with previous reports, while providing some new insight into the genetic architecture of these complex diseases. As genomic taxonomies continue to evolve, such as refined maps of enhancer elements and splicing regions, additional classes can easily be layered into the proposed analysis framework. Moreover, application of these approaches to alternative publicly available meta-analysis resources, or more generally as post-analytic strategies to further interrogate regions that are identified through single point analysis, is straightforward. In conclusion, it would be prudent to include class-level interrogation as standard practice in GWA analysis. The methods described can be applied as simple, complementary and efficient strategies for class-level testing that leverage existing data resources, require only summary level data in the form of test statistics and linkage disequilibrium structure, and add significant value with respect to the potential for identifying multiple novel and clinically relevant trait associations.
ABSTRACT: Implementation science research often involves a natural clustering by social network or community. Only some members of the group randomized to the intervention are actually exposed. In such studies, estimation of individual and spillover effects is of interest. The individual effect is the effect on the participants who directly received the intervention and the spillover effect is the effect on the participants who shared a network with the directly treated participant. We discuss the causal inference framework and assumptions for this setting and define estimators of individual and spillover effects. Using data from the HIV Prevention Trials Network (HPTN) 037 trial, we estimate individual and spillover effects of a network-randomized intervention. This trial was a Phase III, randomized controlled HIV prevention intervention targeting injection drug users and their risk network members in the United States and Thailand. Lastly, we extend these methods to estimate individual and spillover effects of a package of interventions using causal inference methods to adjust for time-varying confounding.
ABSTRACT: There is a growing need for assays that accurately identify recent HIV-1 infections, as these are crucial for estimating HIV incidence as well as for clinical purposes. Viral genetic diversity-based assays are a promising venue but their accuracy needs to be improved. We compare the predictive power of our new biomarkers with that of the serologic assay BED. To further improve prediction, we also developed multivariate predictive models exploring age, gender, CD4 count, plasma viral load and other viral-diversity-based biomarkers. We found that a generalized-entropy approach focusing on HIRs improves predictive accuracy compared to other viral-diversity based biomarkers. A multivariate approach incorporating both HIR entropy measures within env and gag, skewness of the pair-wise Hemming distance distribution, plasma viral load and BED significantly improved predictive accuracy (AUC=0.88) compared to any single predictors. In conclusion, a multi-assay approach incorporating HIR-based entropy from multiple genes, as well as other viral-based and serological biomarkers substantially improved the accuracy for identifying recent HIV-1 infections.
ABSTRACT: None Given
ABSTRACT: None Given
ABSTRACT: None Given
ABSTRACT: None Given
ABSTRACT: None Given
|Back to SPH Biostatistics||
Maintained by the
Last Update: February 4, 2016