SSSandra E. Safo

Sandra E. Safo

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Sandra E. Safo has a total of 202 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • A score based on MRI imaging variables can predict time to moderate progression in mild cognitive impairment: Multimodal data integration study

    by Elzbieta Jodlowska-Siewert, Limeng Liu, Kaifeng Yang, Denis Ostroushko, Pamela L. Lutsey, Thierry Chekouo, Sandra E. Safo, For The Alzheimer'S Disease Neuroimaging Initiative

    Background: Mild cognitive impairment (MCI) is an intermediate stage between normal aging and Alzheimer's disease (AD). Because there is no treatment for AD, it is vital to establish which patients are at a higher risk of disease progression, and might benefit from early intervention. Objective: This study investigates whether a score based on clusters of MRI and lipidomic variables can predict the time to moderate progression in MCI. Methods: We analyzed MRI and lipidomics data from 480 individuals with MCI (mean age 72.14 years, range 55–90 years). Moderate progression was defined as a ≥ 2.5-point increase in the Clinical Dementia Rating Scale Sum of Boxes (baseline mean CDR-SB of 1.2, range 0–10 points). Multi-source data integration and clustering methods were used to derive predictive scores. Results: The MRI-based score outperformed the lipidomics score (c-index: 0.754 versus 0.666). A 1-unit increase in the MRI score was linked to a 13% higher risk of moderate progression. Individuals in the top 25% of MRI scores had a 3-fold greater risk of decline. Conclusions: An MRI-based score offers a simple, effective tool to predict moderate MCI progression and identify high-risk individuals for AD.

  • Open Access

    Proteomic and genetic predictors and risk scores of cardiovascular diseases in persons living with HIV

    by Tanvi Mehta, Lillian Haine, Jason Baker, Cavan Reilly, Daniel Duprez, Shweta Sharma Mistry, Brian T. Steffen, Mamta K. Jain, Alejandro Arenas-Pinto, Mark Polizzotto, et al.

    Background: Cardiovascular disease (CVD) prediction models for persons living with HIV (PLWH) depend on traditional CVD risk factors, but these underestimate true risk. We aimed to identify proteins and genetic variants and create proteo-genomic risk scores for CVD in PLWH. Methods: We analyzed genetic and protein data from participants involved in trials for PLWH. We used state-of-the-art statistical methods for data integration, identified correlated signatures, and developed a protein score (PS) and a genetic score (GS) to predict CVD. We conducted functional enrichment analysis to explore biological functions of signatures identified in relation to CVD. Results: A panel of 14 proteins and a set of 15 genetic variants were found to be better at distinguishing between CVD cases and controls than individual proteins or genetic variants. The PS or GS was each independently associated with a higher risk of CVD (OR for PS: 2.36, CI: 1.78–3.19; OR for GS: 4.59, CI: 3.21–6.80). Combining CVD-, HIV-related factors, genetics, and protein scores resulted in the most powerful discrimination with an AUC of 0.86 (CI: 0.82–0.90). Having a PS in the top 25% compared to the bottom 75% resulted in a 3.9 times higher risk of CVD. Having a GS in the top 25% compared to the bottom 75% resulted in a 7.3 times higher risk of CVD. For individuals with both PS and GS in the top 25% compared to others, the risk of CVD was 7.9 times higher. Functional enrichment analysis showed an upregulation of the cytokine tumor necrosis factor (TNF) and strong enrichment for inflammation related pathways such as the pathogen induced cytokine storm. Conclusions: A panel of protein biomarkers, some new (IGFBP7, HGF) and some previously known in PLWH (CLEC6A), could help identify PLWH at higher risk of developing CVD. If confirmed, these scores could be used with CVD and HIV-related factors to identify PLWH at risk for CVD who would benefit from proactive risk reduction strategies.


  • Multivariate functional linear discriminant analysis with feature selection: an application to inflammatory bowel disease classification

    by Limeng Liu, Guannan Wang, Sandra E Safo

    Summary: Inflammatory Bowel Disease (IBD), including Crohn’s Disease (CD) and Ulcerative Colitis (UC), presents significant public health challenges due to its complex etiology. Motivated by the IBD study of the Integrative Human Microbiome Project, our objective is to identify microbial pathways that distinguish between CD, UC, and non-IBD over time. Most current research relies on simplistic analyses that examine 1 variable or time point at a time, or address binary classification problems, limiting our understanding of the dynamic interactions within the microbiome over time. To address these limitations, we develop a novel functional data analysis approach for discriminant analysis of multivariate functional data that can effectively handle multiple high-dimensional predictors, sparse time points, and categorical outcomes. Our method seeks linear combinations of functions (ie discriminant functions) that maximize separation between 2 or more classes over time. We impose a sparsity-inducing penalty when estimating the discriminant functions, allowing us to identify relevant discriminating variables over time. Applications of our method to the motivating data identified microbial features related to mucin degradation, amino acid metabolism, and peptidoglycan recognition, which are implicated in the progression and development of IBD. Furthermore, our method highlighted the role of multiple vitamin B deficiencies in the context of IBD. By moving beyond traditional analytical frameworks, our innovative approach holds the potential for uncovering clinically meaningful discoveries in IBD research.

  • Open Access

    HIV Status and COVID-19 Treatment Disparities in the US National Clinical Cohort Collaborative

    by Emmanuel Nazaire Essam Nkodo, Pooja Maheria, Eric Hurwitz, Alfred Jerrod Anzalone, Dongmei Li, Jessica Y Islam, Jing Sun, Cara D Varley, Zachary Butzin-Dozier, Sandra E Safo, et al.

    While disparities in COVID-19 therapeutic access have been documented, the effect of HIV status on treatment access and how it intersects with other sociodemographic factors has not been well explored. Using data from the National Clinical Cohort Collaborative (N3C), we investigated disparities in COVID-19 therapeutic prescription among persons with HIV and without HIV. This was a retrospective cohort study of patients’ data from January 2020 to November 2024. The study included 7 806 412 patients with a COVID-19 diagnosis, of whom 45 508 (0.58%) were persons with HIV. We employed logistic and linear regression models to assess associations between therapeutic receipt and patient characteristics. Persons with HIV had significantly higher adjusted odds of receiving COVID-19 therapeutics compared to persons without HIV (remdesivir, aOR 1.26 [95% CI: 1.20, 1.33]; nirmatrelvir/ritonavir, aOR 2.86 [95% CI: 2.77, 2.95]). Despite this, significant racial/ethnic inequities were observed. American Indian or Alaskan Native persons with HIV (estimated coefficient 0.997) and Hispanic/Latinx persons with HIV (estimated coefficient 0.992) had a lower estimated prevalence of remdesivir receipt compared to White Non-Hispanic individuals. For nirmatrelvir/ritonavir, Black/African American individuals (persons with HIV, estimated coefficient 0.947; persons without HIV, estimated coefficient 0.943), American Indian or Alaskan Native persons with HIV (estimated coefficient 0.996), and Hispanic/Latinx individuals (estimated coefficient 0.992) showed a lower estimated prevalence of receipt compared to their White counterparts. Persons with HIV demonstrated higher odds of receiving COVID-19 therapeutics than persons without HIV. However, persistent racial and ethnic inequities in treatment uptake were evident.

  • Open Access

    A Bayesian Integrative Mixed Modeling Framework for Analysis of the Multi-Site Adolescent Brain and Cognitive Development Study

    by Aidan Neher, Apostolos Stamenos, Mark Fiecas, Sandra E. Safo, Thierry Chekouo

    Integrating high-dimensional, heterogeneous data from multi-site cohort studies with complex hierarchical structures poses significant variable selection and prediction challenges. We extend the Bayesian Integrative Analysis and Prediction (BIP) framework to enable simultaneous variable selection and outcome modeling in data of a multi-view nested hierarchical structure. We apply the proposed Bayesian Integrative Mixed Modeling (BIPmixed) framework to the Adolescent Brain Cognitive Development (ABCD) Study, leveraging multi-view data, including structural and functional MRI and early life adversity (ELA) metrics, to identify relevant variables and predict the behavioral outcome. BIPmixed incorporates 2-level nested random effects to enhance interpretability and make predictions in hierarchical data settings. Simulation studies illustrate BIPmixed’s robustness in distinct random effect settings, highlighting its use for complex study designs. Our findings suggest that BIPmixed effectively integrates multi-view data while accounting for nested sampling, making it a valuable tool for analyzing large-scale studies with hierarchical data.


  • Open Access

    Proteomic and Genetic predictors and risk scores of cardiovascular diseases in persons living with HIV

    by Tanvi Mehta, Lillian Haine, Jason Baker, Cavan Reilly, Daniel Duprez, Shweta Mistry, Brian Steffen, Mamta K. Jain, Alejandro Arenas-Pinto, Mark Polizzotto, et al.

    Background Cardiovascular diseases (CVD) prediction models for persons living with HIV (PLWH) depend on traditional CVD risk factors, but these underestimate true risk. We aimed to identify proteins and genetic variants and create proteo-genomic risk scores for CVD in PLWH. Methods We analyzed genetic and protein data from participants involved in trials for PLWH. We used state-of-the-art statistical methods for data integration, identified correlated signatures, and developed a protein score (PS) and a genetic score (GS) to predict CVD. We conducted functional enrichment analysis to explore biological functions of signatures identified in relation to CVD. Results A panel of 14 proteins and a set of 15 genetic variants were found to be better at distinguishing between CVD cases and controls than individual proteins or genetic variants. The PS or GS was each independently associated with a higher risk of CVD. Combining CVD-, HIV-related factors, genetics, and protein scores resulted in the most powerful discrimination. Functional enrichment analysis showed an upregulation of the cytokine tumor necrosis factor (TNF) and strong enrichment for inflammation related pathways such as the pathogen induced cytokine storm. Conclusions A panel of protein biomarkers, some new (IGFBP7, HGF) and some previously known in PLWH (CLEC6A), could help identify PLWH at higher risk of developing CVD. If confirmed, these scores could be used with CVD and HIV-related factors to identify PLWH at risk for CVD who would benefit from proactive risk reduction strategies.


  • Open Access

    Supervised Integrative Biclustering with applications to Alzheimer's Disease

    by Kaifeng Yang, Thierry Chekouo, Sandra E. Safo

    Multiple types or views of data (e.g. genetics, proteomics) measured on the same set of individuals are now popularly generated in many biomedical studies. A particular interest might be the detection of sample subgroups (e.g. subtypes of disease) characterized by specific groups of variables. Biclustering methods are well-suited for this problem since they can group samples and variables simultaneously. However, most existing biclustering methods cannot guarantee that the detected sample clusters are clinically meaningful and related to a clinical outcome because they independently identify biclusters and associate sample clusters with a clinical outcome. Additionally, these methods have been developed for continuous data when integrating data from different views and do not allow for a mixture of data distributions. We propose a new formulation of biclustering and prediction method for multi-view data from different distributions that enhances our ability to identify clinically meaningful biclusters by incorporating a clinical outcome. Sample clusters are defined based on an adaptively chosen subset of variables and their association with a clinical outcome. We use extensive simulations to showcase the effectiveness of our proposed method in comparison to existing methods. Real-world applications using lipidomics, imaging, and cognitive data on Alzheimer's disease(AD) identified biclusters with significant cognitive differences that other methods missed. The distinct lipid categories and brain regions characterizing the biclusters suggest potential new insights into pathology of AD.

  • Open Access

    Extensions of Heterogeneity in Integration and Prediction (HIP) With R Shiny Application

    by Jessica Butts, Leif Verace, Christine Wendt, Russell Bowler, Craig P. Hersh, Qi Long, Lynn Eberly, Sandra E. Safo

    Multiple data views measured on the same set of participants are becoming more common and have the potential to deepen our understanding of many complex diseases by analyzing these different views simultaneously. Equally important, many of these complex diseases show evidence of subgroup heterogeneity (e.g., by sex or race). HIP (Heterogeneity in Integration and Prediction) is among the first methods proposed to integrate multiple data views while also accounting for subgroup heterogeneity to identify common and subgroup‐specific markers of a particular disease. However, HIP is applicable to continuous outcomes and requires programming expertise by the user. Here we propose extensions to HIP that accommodate multi‐class, Poisson, and Zero‐Inflated Poisson outcomes while retaining the benefits of HIP. Additionally, we introduce an R Shiny application, accessible on shinyapps.io at https://multi‐viewlearn.shinyapps.io/HIP_ShinyApp/, that provides an interface with the Python implementation of HIP to allow more researchers to use the method anywhere and on any device. We applied HIP to identify genes and proteins common and specific to males and females that are associated with exacerbation frequency. Although some of the identified genes and proteins show evidence of a relationship with chronic obstructive pulmonary disease (COPD) in existing literature, others may be candidates for future research investigating their relationship with COPD. We demonstrate the use of the Shiny application with publicly available data. An R‐package for HIP is available at https://github.com/lasandrall/HIP.

  • Open Access

    Deciphering Colorectal Cancer–Hepatocyte Interactions: A Multiomics Platform for Interrogation of Metabolic Crosstalk in the Liver–Tumor Microenvironment

    by Alisa B. Nelson, Lyndsay E. Reese, Elizabeth Rono, Eric D. Queathem, Yinjie Qiu, Braedan M. McCluskey, Alexandra Crampton, Eric Conniff, Katherine Cummins, Ella Boytim, et al.

    Metabolic reprogramming is a hallmark of cancer, enabling tumor cells to adapt to and exploit their microenvironment for sustained growth. The liver is a common site of metastasis, but the interactions between tumor cells and hepatocytes remain poorly understood. In the context of liver metastasis, these interactions play a crucial role in promoting tumor survival and progression. This study leverages multiomics coverage of the microenvironment via liquid chromatography and high-resolution, high-mass-accuracy mass spectrometry-based untargeted metabolomics, 13C-stable isotope tracing, and RNA sequencing to uncover the metabolic impact of co-localized primary hepatocytes and a colon adenocarcinoma cell line, SW480, using a 2D co-culture model. Metabolic profiling revealed disrupted Warburg metabolism with an 80% decrease in glucose consumption and 94% decrease in lactate production by hepatocyte–SW480 co-cultures relative to SW480 control cultures. Decreased glucose consumption was coupled with alterations in glutamine and ketone body metabolism, suggesting a possible fuel switch upon co-culturing. Further, integrated multiomics analysis indicates that disruptions in metabolic pathways, including nucleoside biosynthesis, amino acids, and TCA cycle, correlate with altered SW480 transcriptional profiles and highlight the importance of redox homeostasis in tumor adaptation. Finally, these findings were replicated in three-dimensional microtissue organoids. Taken together, these studies support a bioinformatic approach to study metabolic crosstalk and discovery of potential therapeutic targets in preclinical models of the tumor microenvironment.


  • Open Access

    Sleep as a possible mediator in the association of mental health parameters with cardiovascular health indices in women: exploratory analyses from the Heart SCORE Study

    by Ida T. Fonkoue, Milena Silva, Susan B. Racette, Sandra E. Safo, Lisa de Las Fuentes, Dawn Lowe, Imo A. Ebong, Daniel Buysse, Steven E. Reis, Anum Saeed

    Objective: This exploratory study aimed to determine the possible role of sleep in the relationships of depression and anxiety, with early surrogate markers of subclinical atherosclerosis, such as brachial artery (BA) diameter and carotid intima media thickness (CIMT) in women. Methods: We included 1,075 self-reported postmenopausal women, 45 to 75 years from the Heart Strategies Concentrating on Risk Evaluation Study. Exposure variables were depression and anxiety assessed using the Center for Epidemiologic Studies Depression Scale and the State-Trait Anxiety Inventory, respectively. Outcome variables were BA diameter and CIMT measured using ultrasonography. The mediator, sleep, was assessed with the Pittsburgh Sleep Symptom Questionnaire–Insomnia. Ordinary least squares regression was used for mediation analyses. Results: Of the 1,075 participants, 56.3% were White and 43.7% were Black. Our analyses revealed significant associations of depression and anxiety with sleep ( P < 0.001 for all). After adjusting for confounders, depression was associated with max CIMT (R 2 = 0.15, P = 0.004), but not BA diameter (R 2 = 0.09, P = 0.083). Although the mediating role of sleep in the association between anxiety and BA diameter was not statistically significant [proportion mediated (CI); 0.41 (−2.77, 4.06); P = 0.219], we observed differential results within each racial group. Sleep appeared to partially mediate the association of anxiety with BA diameter in White [0.21 (0.54, 0.80); P = 0.044] women only. Conclusions: We found preliminary indications that sleep might mediate the association of anxiety with BA diameter in White women but does not appear to serve as a mediator in all the other relationships we examined.