Presentation Date

• April 3, 2024 at 12 p.m. (EDT)

Scientific Question: The null hypothesis (H0) is that there is no difference in PPD Treatment utilization based on race. The alternative hypothesis (Ha) is that at least one group differs significantly in PPD Treatment utilization.

Keywords: Maternal Health, Postpartum Depression, Racial

Introduction: This study aims to explore the disparities in mental health care utilization related to postpartum depression across different racial and ethnic groups. The research focuses on the United States, where specific demographic segments, particularly economically disadvantaged racial and ethnic minorities, are disproportionately susceptible to postpartum depression (PPD). These vulnerable groups often face challenges in accessing quality mental health services due to socioeconomic constraints. Despite the heightened risk, limited research has revealed that cultural stigma surrounding mental health issues discourages African-American women from seeking treatment. Given these evident disparities, a thorough investigation into treatment utilization for PPD among diverse racial and ethnic backgrounds is imperative. The null hypothesis (H0) of this research is that there is no difference in PPD Treatment utilization based on race.

Methods: This retrospective cohort study utilized data from the 'All of Us Research Program' database and applied descriptive statistics to characterize the study participants. The dataset comprised 63,402 postpartum women aged 18-45. Among them, 43,874 were identified as white, 14,777 as black or African American, 4,751 as Asian, and 3,330 as Hispanic. Of these participants, 437 were diagnosed with postpartum depression, with 295 identifying as white, 81 as African American, 21 as Asian, and 59 as Hispanic/Latino. We examined participants who were prescribed antidepressant medication following a PPD diagnosis. Notably, the variable of therapy was initially included but was excluded due to missing data, which had no effect on our findings. Statistical analysis was performed to establish the relationship between race/ethnicity and mental health care utilization, specifically in terms of the prescription of antidepressant medications among women diagnosed with postpartum depression, as assessed using the Chi-squared test.

Results: Among the participants, 33.3% of Asians (N=21), 60.5% of African-Americans (N=81), 73.2% of White individuals (N=295), and 18.6% of Hispanic or Latino participants (N=59) received treatment for PPD. Pearson's Chi-Square test demonstrated a significant relationship between race and the treatment group (p-value < 2.2e-16, df = 3). Regarding ethnicity and the treatment group, the p-value was 3.26e-1, df = 1; therefore, the null hypothesis was rejected, indicating a significant difference between racial and ethnic groups.

Presentation Date

• April 3, 2024 at 12 p.m. (EDT)

Scientific Question: What is the scope and the impact of (SDoH) on the risk of severe acidosis?

Keywords: disparity, SDoH, acidosis, causal inference

Introduction: Severe acidosis that needs emergency care or hospitalization is associated with a high rate of mortality. The risk of this life-threatening clinical event demonstrates health disparities. The All of Us Research Program provides multi-domain data for those of minority groups or underrepresented in clinical study, thus provide unique opportunity to comprehensively scrutinize the scope and the impact of (SDoH) on the risk of severe acidosis.

Research objectives: To assess the health disparities across various SDoH domains and the role of genetic variants on the risk of severe acidosis independent of demographic and clinical factors.

Hypothesis: SDoH factors play a crucial role in the observed health disparities in severe acidosis risk.

Methods: The primary outcome is severe acidosis incidences. The main measures are adjusted odds ratios describing associations between the SDoHs and genetic variants with the risk of acidosis events after controlling the effects of demographic features, clinical conditions, and EMR data availability patterns. A retrospective case-control study (n=13,310, 1:4 matching) is performed using electronic medical records (EMR), SDoH surveys, and genomics data from the All of Us participants. The causal inference design controls confounding effects related to real-world data, with propensity score matching for EMR availability, number of diagnoses, and AoU enrollment date. Conditional logistic regressions are utilized to adjust for demographic and clinical factors.

Results: Multiple comorbidities are associated with an increased risk of acidosis, such as renal disease (adjusted odds ratio or AORc= 1.97-2.01), liver disease (AOR = 1.25-2.58), diabetes (AOR = 28.5-64.8), dementia (AOR:1.42), chronic pulmonary disease (AOR:1.33), myocardial infarction (AOR:1.34), congestive heart failure (AOR:1.53), hemiplegia paraplegia (AOR:1.41), and metastatic solid tumors (AOR:1.33). SDoHs significantly contribute to the health disparity in the risk of severe acidosis. Those with employer-provided insurance and those with Medicaid plans show dramatically different risks (AOR: 0.761 vs 1.41). Low-income groups demonstrate higher risk (household income less than $25k, AOR: 1.3 -1.57) than high-income groups ($100k - $200k, AOR: 0.597 - 0.867). Other high-risk factors include impaired mobility (AOR: 1.32), unemployment (AOR: 1.32), renters (AOR: 1.41), other non-house-owners (AOR: 1.7), and house instability (AOR: 1.25). Education was negatively associated with acidosis risk. The genetic variants of candidate transporters include organic ion transporters OCT1, OCT2, OCTN1 and multidrug and toxin extrusion transporter MATE1 and MATE2K do not show statistically significantly associations with acidosis risk. Meanwhile, the T/T homozygous variant at 17:160122116 of gene SLC22A1 demonstrates (AOR: 0.285) and the A>C or A>T variants at 6:160249250 of SLC47A2 (AOR: 1.48 and 1.63, respectively) are potential genotypes for further analysis.

Conclusions: Social determinants of health are strongly associated with systematic health disparities in severe acidosis risk. Types of health insurance, household income levels, housing status and stabilities, employment status, educational levels, and mobility disabilities play significant roles after being adjusted for demographic features and clinical conditions. Our results provide novel knowledge and real-world evidence on the complexity of health disparities and enable future studies into specific types of acidosis, health interventions, and genotypes.

Presentation Date

• April 3, 2024 at 12:00 p.m. (EDT)

Scientific Question: The All of Us Research Program has collected extensive data from a diverse cohort of participants, encompassing various racial and ethnic backgrounds, geographic locations, and socioeconomic statuses. Leveraging this rich dataset, we plan to investigate the relationship between social determinants of health (SDOH) and race/ethnicity in individuals with severe obesity (BMI of 40 or greater), with a focus on understanding the disparities in health outcomes that is specific to this unique population.

Keywords: SDOH, disparities, preeclampsia, race, ethnicity

Introduction: Preeclampsia is one of the leading causes of maternal and perinatal morbidity and mortality. However, the relationship between preeclampsia and various social determinants of health (SDOH) is not well understood. The utilization of All of Us Research Program data allows for a thorough examination of the associations between SDOH and preeclampsia, particularly when stratified by race and ethnicity, due to the extensive range of SDOH factors.

Research Objectives: To examine the associations between SDOH and preeclampsia by race and ethnicity.

Methods: All of Us cross-sectional research program data of pregnant individuals (N=2,287) aged 18-to-44 years) was analyzed. SNOMED codes confirmed preeclampsia cases (yes/no). Race and ethnicity were self-reported (Asian, non-Hispanic Black (NHB), Hispanic, and non-Hispanic White (NHW)). Multivariable mediation analyses, using multiple additive regression trees (MART), determined the associations between SDOH and preeclampsia by race and ethnicity.

Results: Of 2,287 individuals (mean age 37.6 [SD 5] years, 67% NHW, 21% Hispanic, 8% NHB, 3% Asian) 5% (n=108) had preeclampsia. Racial disparities (NHB vs NHW) in preeclampsia were partially explained by insurance coverage (Relative Effect 0.8), marital status (16.1), and healthcare access (5.1) (p<.001 for all). Perceived discrimination (-36.2), perceived stress (-6.0), age (-7.3), and education (-2.0) magnified this disparity (p<.001 for all). Ethnic group preeclampsia disparities (Hispanic vs NHW) were partially explained by perceived stress (7.2), marital status (15.8), and income (56.9) (p<.001 for all). Neighborhood cohesion (-10.4), discrimination in medical settings (-20.0), age (-48.2), and education (-25.0) magnified this disparity (p<.001 for all).

Presentation Date

• April 3, 2024 at 1 p.m. (EDT)

Scientific Question: For non-European patients with colonic diverticular disease, does a polygenic risk score derived from European populations demonstrate transferability with respect to risk stratification?

Keywords: Polygenic risk scores; PRS; PheWAS 

Introduction: A clinical need for personalized stratification approaches with colonic diverticular disease has prompted investigations of underlying genetic contributions. One area of interest has been polygenic risk scores (PRS), which combine effects of linked genomic locations into a single risk score for an individual. While validations of diverticular disease PRSs have shown promise in European cohorts, assessment of potential benefit for other ancestries must be performed prior to considering clinical implementation.

Research objectives: To quantify the transferability of an existing PRS for colonic diverticular disease to non-European ancestries

Methods: All of Us Research Program adult participants with available short read whole genome sequencing data and electronic medical records were identified. A PRS derived from the UK Biobank was applied to unrelated individuals after genomic and sample quality control. Per-ancestry phenome-wide association studies (PheWAS) identified conditions associated with genetic susceptibility to diverticular disease. The cohort was then divided into no diverticular disease, asymptomatic diverticulosis, or symptomatic diverticulitis using an independently validated rule-based algorithm. For ancestries with at least 100 cases, logistic regression models were fit adjusting for age, sex, body mass index, smoking status, and the first ten genetic principal components. To assess the additional value attributable to the PRS, full models including the PRS were compared to base models without it. Metrics included area under the receiver operating characteristics curve (AUROC), Nagelkerke's R2 on the liability scale, and fraction of new information derived from model likelihood-ratio chi-square statistics.

Results: Across European, African, and Admixed American ancestries, there were 181,719 individuals for the PheWAS and 34,446 for the adjusted modeling. The PRS was associated with diverticular disease phecodes in all cohorts, though with attenuated strength in non-European ancestries (odds ratio (OR) range 1.18 to 1.28; p-value range 1.55 * 10-8 to 1.22 * 10-78). In adjusted modeling alongside other clinical covariates, the PRS was associated with higher odds of symptomatic diverticulitis (OR range 1.28 [1.11 - 1.47] to 1.66 [1.42 - 1.95] per standard deviation increase in PRS). The highest model performance was achieved in the European ancestry model with the PRS (AUROC [95% CI] 0.72 [0.71 - 0.74]; R2 0.14). The PRS provided 20% fraction of new information in the European model, with decreases in this metric observed for the African (8%) but not Admixed American (25%) cohorts. Improvement was limited in AUROC when comparing full models to base models (maximum 0.72 [0.71 - 0.74] vs 0.70 [0.69 - 0.71]).

Conclusions: A diverticular disease PRS from European populations was a positive predictor of symptomatic and severe diverticulitis in African and Admixed American cohorts. While relative improvements in model performance were observed, the small magnitude likely limits clinical utility.

Presentation Date

• April 3, 2024 at 1 p.m. (EDT)

Scientific Question: How does healthcare access and utilization differ among transgender and gender diverse adults?

Keywords: Transgender, Healthcare Access

Introduction: Transgender and gender diverse (TGD) people have reported high rates of dissatisfaction with the healthcare that they receive as well as barriers to accessing healthcare. Yet studies regarding their health care access are often limited by sample size and consider all TGD people to represent one undifferentiated category. This limits our ability to understand the potentially diverse patterns of healthcare access and health within the TGD group. To address this, we analyzed a diverse dataset through the All of Us Research Program that contains detailed information on gender identity.

Research objectives: We describe the healthcare access and utilization of TGD adults by different gender identities.

Methods: The All of Us Research Program by the National Institutes of Health aims to diversify research by creating a dataset with at least one million participants. We analyzed "The Basics Survey" and "Healthcare Access and Utilization Survey" within the All of Us Controlled Tier v7 database and described respondents' access to and use of healthcare. We created seven gender categories: Transgender women, transgender men, non-binary assigned female at birth (NB-AFAB), non-binary assigned male at birth (NB-AMAB), cisgender women, cisgender men, and gender diverse (e.g., genderfluid, genderqueer, and two spirit). We computed descriptive statistics to present access (e.g., ability to afford healthcare), use (e.g., rates of speaking to a doctor within the last year, being respected by healthcare providers), and other demographic information (e.g., race and education). We examined bivariate differences in access between the gender categories via chi-squared tests using R version 4.3.1.

Results: We identified 401,327 (out of 413,360) adults who reported gender identity information. The largest groups were cisgender: cisgender men (N=152,096) and cisgender women (N=245,133). Within the TGD group (N=4089), 46% identified as gender binary (909 transgender women and 970 transgender men), 51% as non-binary (1,581 NB-AFAB, 511 NB-AMAB), and 127 (3%) as gender diverse. Overall, 54.8% of the sample were White, 18.7% were Black, and 89.1% had at least a high school education. We observed differences in healthcare access across the groups. For example, the percentage of people unable to afford healthcare was highest for the gender diverse group (18.6%) and non-binary groups (14.2% NB-AFAB, 10.2% NB-AMAB) compared with 4.9% for cisgender men and 6.5% for cisgender women (p<0.001). Regarding experiences with healthcare providers, rates of speaking to a doctor regarding their health within the past year were lowest for the gender diverse group (72.3%), while the other groups ranged between 78.9-84.2% (p<0.001 ). Cisgender people reported higher rates of feeling respected all or most of the time by healthcare professionals (96.1% cisgender women, 97.0% cisgender men), while non-binary patients reported the lowest rates (83.0%% NB-AFAB, 89.6% NB-AMAB ) (p<0.001 ). We plan to further our analysis to include multivariate modeling.

Conclusions: Healthcare access and utilization, including rates of seeing a doctor and projectredcap.org being able to afford healthcare, varies between gender identities, and within the TGD population. Parsing out these disparities is necessary to identify ways to improve care for TGD individuals.

Presentation Date

• April 3, 2024 at 1 p.m. (EDT)

Scientific Question: What are the sociodemographic factors associated with experiencing barriers to healthcare among individuals engaging in heavy drinking?

Keywords: heavy drinking; healthcare access

Introduction: Primary care providers (PCPs) play a crucial role as the initial point of care for patients exhibiting problematic drinking. Provider roles in care for these patients might involve screening then linking patients to alcohol use treatment or intervention. PCPs are critically important; in 2021, only 8% of individuals with alcohol use disorders received treatment. Barriers to healthcare may be impeding this important first step in getting help, yet attitudinal and structural barriers limit seeking healthcare. The All of Us Workbench data provides a unique opportunity to examine barriers to care among a national sample of individuals who engage in heavy drinking and who completed the behavioral and the social determinants of health surveys.

Research objectives: To examine barriers to healthcare in five domains: sociodemographic, structural (transportation and rural living), competing social roles (work, childcare, and adult caregiving), attitudinal (nervous about seeing a provider), and financial (couldn't afford the copay, deductible, or out of pocket costs) among individuals engaging in heavy drinking.

Methods: A subsample of 3,257 participants (55.5% male, 74% white, 13% Black/African American, 13% Hispanic/Latine) from the All of Us study who engaged in heavy drinking (i.e., six or more drinks on one occasion, at least weekly) was selected to examine the association between sex at birth, race/ethnicity, annual income, educational attainment, insured status and reasons for lack of healthcare in the past 12 months or longer. Multivariate logistic regressions were implemented to estimate adjusted Odds Ratios (ORs) and their corresponding 95% confidence intervals (CI).

Results: Financial reasons (23.6%) were the most common factor acting as a possible barrier to healthcare, followed by attitudinal (18.3%), competing social roles (15.7%), and structural (13.4%) barriers. Female participants were more likely to report competing social roles (aOR=1.6, 95% CI=1.3,1.9), attitudinal (aOR=1.4, 95% CI=1.2,1.7), and financial (aOR=1.4, 95% CI=1.2,1.8) barriers to healthcare than male participants. Lower income (aOR=6.7, 95% CI=4.7,9.6), lower educational attainment (aOR=1.6, 95% CI=1.2,2.1), and being Black or African American (aOR=1.4, 95% CI=1.1,1.9) were associated with higher odds of reporting structural barriers to healthcare in the past 12 months or longer.

Conclusions: As many as 1 in 4 individuals who engage in heavy drinking experience at least one barrier to healthcare. Women, Black or African Americans, and those with lower socio-economic status (SES) experience more barriers to seeking healthcare, highlighting the need to intervene to reduce these barriers to healthcare as a strategy to improve screening and reduce alcohol use-related disparities.

Presentation Date

• April 4, 2024 at 12 p.m. (EDT)

Scientific Question: This paper investigates the association between exposure to ambient PM2.5 and NO2 and the risk of acute asthma exacerbation in this cohort

Keywords:  air quality, asthma, PM2.5, NO2

Abstract: The NIH All of Us Research Program has enrolled over 500,000 participants across the US and its territories, enabling researchers to explore relationships between exposures and health states in a national dataset of unparalleled scope. This project investigates the association between exposure to ambient PM2.5 and NO2 and the risk of acute asthma exacerbation in this cohort.

This work was performed on data from 247,885 All of Us Research Program participants using the All of Us Researcher Workbench. Acute asthma case ascertainment was performed using data from electronic health records. 5-year ambient outdoor PM2.5 data were retrieved from the Atmospheric Composition Analysis group at Washington University in St. Louis. 5-year ambient outdoor NO2 data were retrieved from the TROPOspheric Monitoring Instrument (TROPOMI). Both pollutants were assigned to participants at the 3-digit zip code prefix level. Multivariate logistic regression was used to estimate the odds ratio (OR) and 95% confidence interval (CI). Cox Proportional Hazard Regression was used to estimate the hazard ratio for acute asthma diagnosis during the follow up against per unit increase in air pollutants. 

A total of 9,660 acute asthma cases were ascertained from participant EHR data. The mean NO2 level from 2015 to 2019 was 0.41 ppb (min 0.02, max 1.05) and the mean PM2.5 level from 2017 to 2021 was 7.75 μg/m3 (min 3.69, max 13.90). In Cox proportional hazards regression, increased odds for asthma exacerbation were observed for Hispanic participants in the 4th (HR = 2.15, 95% CI:1.07-4.30) NO2 exposure quartile.

Long-term exposure to even low levels of PM2.5 and NO2 is associated with respiratory mortality, and increased exposure to PM2.5 and NO2 is associated with worse asthma symptoms as well as outpatient visits and hospitalization for acute asthma exacerbation. Further understanding the relationship between low-level PM2.5 and NO2 exposure in acute asthma, especially among the adult population, will add to the growing literature on the subject.

Presentation Date

• April 4, 2024 at 12 p.m. (EDT)

Scientific Question: How can cardiovascular risk in individuals with Type II Diabetes best be predicted using machine learning?

Keywords:  diabetes, cardiovascular risk, machine learning

Introduction: In the United States, over 10% of the population has type II diabetes, which doubles cardiovascular risk, the leading cause of death. If people who are diagnosed with diabetes have an accurate measure of their risk of developing heart disease, individuals at high risk can take helpful preventative measures. Both diabetes and cardiovascular disease disproportionately affect racial minorities. Therefore, the All of Us dataset, which upsamples for racial minorities, is ideal for developing a model to predict cardiovascular risk among people with diabetes. All of Us also contains rich EHR information, individual socioeconomic data from surveys, wearable data, and genomic data, providing impressive interpretability and predictive power across diverse minority subgroups.

Objectives: We aim to create an accurate machine learning model that predicts cardiovascular risk in individuals with type II diabetes using EHR, demographics, and social and lifestyle data. Additionally, we will determine which features contribute most significantly to the model, specifically analyzing variables from the social determinants of health dataset. We also will evaluate how the model performs across commonly underrepresented minority racial groups to assess the model's fairness.

Method: We collect all the (non-COVID) survey and EHR All of Us data for individuals with type II diabetes. The XGBoost model is employed to build a risk prediction model, which can leverage thousands of features as input, demonstrating state-of-the-art performance in many machine learning applications. Individuals with cardiovascular disease prior to diabetes diagnosis are excluded. The XGBoost model predicts whether an individual with type II diabetes will develop cardiovascular disease within 5 years,

where cardiovascular disease includes myocardial infarction, stroke, and other major vascular diseases. We will conduct model interpretation analysis to identify important factors, particularly social factors, contributing most to cardiovascular risk prediction. Additionally, the fairlearn python package will be used to evaluate the model's fairness across different minority groups.

Results: A preliminary model has been created using solely survey data. The XGBoost model achieved an AUC ROC of 0.68, representing moderately high accuracy, which will increase when EHR data is added to the model. The primary anticipated result of this study will be an effective model that can accurately predict a diabetic individual's cardiovascular risk regardless of the individual's ethnicity and race. Predictive features identified among the survey data are employment status and financial availability of medicine, and predictive features among the EHR data will be reported when EHR data is added to the model. The model’s accuracy will be reported both for the total cohort and within specific racial and gender groups. The data processing and modeling methods that we optimize for use with All of Us researchers to create predictive models for a wide range of diseases that disproportionately impact minorities.

Presentation Date

• April 4, 2024 at 12 p.m. (EDT)

Scientific Question: Our aim is to elucidate the distinct risk profiles that exist among various ancestries, ethnicities, races, and populations, shedding light on whether the manifestations and severity of the disease exhibit notable variations across these diverse groups.

Keywords:  Crohn's Disease, Ulcerative Colitis

Abstract: Inflammatory Bowel Disease (IBD) is a systemic autoimmune disorder that includes two major subsets known as Crohn's Disease (CD) and Ulcerative Colitis (UC). This project aims (1) to explore and determine risk factors (demographics, genetics, and environment) associated with both forms of IBD, and (2) to identify genetic loci associated with increased manifestation and risk of IBD in a diverse group of people from different ancestries. In our study, we used the All of Us Researcher Workbench supported by the National Institutes of Health (NIH) to create cases and control cohorts for each of the IBD forms. Data Analysis was performed in Jupyter Notebooks in the All of Us Researcher Workbench using R programming language. Our results showed a significantly stronger risk factor for White populations in comparison to other populations (p-value < 10e-15, according to a chi-square goodness-of-fit test), and no sex bias was observed (p-value = 0.20). At this stage, we assess the significance of other risk factors, such as smoking, among others, and perform a trans-ancestry genome-wide association (GWAS) study to identify loci underlying disease risk in diverse populations. We anticipate that our findings will reveal the contribution of race/ethnicity, gender, sex at birth, smoking, and other demographic or environmental factors to IBD risk and that the results of the GWAS will highlight genetic loci associated with IBD in this diverse American cohort. To conclude, we hope this disease-focused project can help improve the lives of patients who suffer from IBD and provide a more accurate understanding of IBD risk factors helping precision medicine with this disease progress.

Presentation Date

• April 4, 2024 at 1 p.m. (EDT)

Scientific Question: Do manifestations and severity of SLE differ amongst diverse populations? Do genetic components contribute to increased risk for SLE amongst these populations?

Keywords: Genetic Exploration of SLE Epidemiology

Abstract: Systemic lupus erythematosus (SLE) is a female-biased autoimmune disorder characterized by the immune system mistakenly attacking the body's own tissues and organs resulting in inflammation and damage. SLE has heterogeneous manifestations, ranging from skin and joint damage to more severe complications of the kidneys, brain, or heart. Higher incidences of SLE have been observed among Hispanics and populations of African ancestry. However, prior studies on SLE have predominantly been conducted among individuals of European ancestry, highlighting the need for more inclusive research efforts to understand SLE complexities amongst different ethnic groups. To contribute in tackling this disparity, and broadening the knowledge on the epidemiology and genetic liability of SLE in diverse populations, we are implementing two complementary approaches. First, we seek to identify SLE risk factors (e.g., age, sex at birth, gender, race, ethnicity, etc.) and assess disease severity across a diverse cohort of 3,808

SLE-affected individuals from the United States using NIH's All of Us Research Program data. Disease severity was assessed based on perceived health condition, SLE remission status, and comorbidity prevalence. These data were analyzed across populations, considering self-reported race/ethnicity, in order to show whether SLE severity varies across race/ethnicity. We used this approach to compare subjective and clinical assessments of health in SLE patients to understand both socioeconomic and clinical risk factors in these diverse populations. Notably, we observed a substantial prevalence of females among SLE-affected individuals, as opposed to controls, which is consistent with prior research findings. This observation is further corroborated by a striking gender ratio of 9:1, indicating a pronounced female bias among SLE patients (chi squared p-value < 2.2e-16). SLE cases showed approximately 15% fewer individuals of White/European ancestry compared to the control cohort, aligning with prior research indicating a lower prevalence of SLE among individuals with White/European ancestry. Further exploration of this demographic uncovered a higher proportion of African-Americans affected by SLE in comparison to the control cohort (chi squared p-value < 2.2e-16). Noticeably, the All of Us cases cohort displayed an increased representation of non-White individuals affected by SLE. Our analysis of comorbidity severity uncovered notable distinctions in the severity of SLE among individuals with African American, Middle Eastern, Asian, and Pacific Islander ancestries compared to those with European ancestry (p-value = 9.225e-08). Furthermore, the analysis identified significant variations in the severity across all manifestations when comparing the mentioned racial groups (p-value = 9.225e-08).

Secondly, we are currently conducting a trans-ancestry, sex-stratified Genome-Wide Association Study (GWAS) leveraging the NIH's All of Us Research Program biobank (n=222,951) to identify SLE-associated variants. Results from this part of the project are expected to show genetic loci associated with an increased risk of SLE and highlight differences in the genetic architecture of this trait across populations of different ancestries. Because SLE has been understudied in populations of backgrounds other than European, we anticipate these results will contribute to ensuring more comprehensive and inclusive healthcare solutions that address the diverse genetic backgrounds within the All of Us database.

Presentation Date

• April 4, 2024 at 1 p.m. (EDT)

Scientific Question: How does the prevalence of heart failure with preserved ejection fraction (HFpEF) vary in patients with systemic lupus erythematosus (SLE) compared to healthy controls, and what are the underlying genetic risk factors that contribute to this variation, particularly in different racial and gender groups?

Keywords: systemic lupus erythematosus, heart failure

Introduction: Heart failure (HF), a critical global public health concern, is linked to inflammation, a key element in its pathogenesis. Systemic lupus erythematosus (SLE), an autoimmune disease characterized by widespread inflammation, predisposes a significant subset of patients to heart failure. Notably, heart failure with preserved ejection fraction (HFpEF), or diastolic HF, emerges as the predominant heart failure type within the SLE patient population. Despite this, current research lacks comprehensive studies on the prevalence of HFpEF and its associated genetic risk factors among SLE patients, particularly within a diverse and nationwide American cohort. The risk factors contributing to the development of HF in SLE patients are multifaceted, encompassing genetic, environmental, immunological, lifestyle-related, racial, and medical treatment aspects. Our research focuses on testing the hypothesis that African American females with SLE are at a higher genetic risk of inflammation-induced HFpEF compared to their Caucasian counterparts. This hypothesis stems from the need to understand the complex interplay of genetic and demographic factors in the manifestation of HFpEF in SLE patients.

Research objectives: Our study aims are three-fold: (1) to quantify the prevalence of HFpEF in SLE patients compared to non-SLE control groups; (2) to identify and analyze the frequency of genetic variants associated with inflammatory biomarkers in SLE patients and examine their distribution across different racial and gender groups; and (3) to investigate the correlation between these genetic markers and the risk of developing HFpEF in SLE patients, with an emphasis on high-risk subgroups such as African American females.

Methods: Utilizing the National Institutes of Health (NIH) All of Us database, we analyzed a cohort of adults aged 18 and older, encompassing individuals diagnosed with Systemic Lupus Erythematosus (SLE) and a control group without SLE. We calculated the lifetime prevalence of heart failure with preserved ejection fraction (HFpEF) in the SLE group. This involved stratifying data across various racial/ethnic groups and gender identities. We processed whole genome sequencing data from the All of Us database to identify variants (SNPs) of inflammatory biomarkers (SCARF1, CRP, HMGB1, IL6, TNF, NFKB1, AGT) in SLE patients. We then computed a genetic risk score to estimate the risk of developing HFpEF based on genotypes at the identified inflammatory biomarker variants.

Results: The prevalence of HFpEF in SLE patients was significantly higher than in control groups, with the highest rates observed in African American females. Genetic analysis revealed a significant correlation between certain inflammatory biomarker variants and increased HFpEF risk, particularly among African American and female patients. Statistical analyses indicated these associations were robust, with potential implications for targeted screening and preventive strategies in high-risk groups.

Conclusions: Our study provides significant insights into the intricate relationship between SLE and HFpEF, underscoring the influence of genetic and demographic factors. The findings advocate for enhanced awareness and tailored management approaches for HFpEF in SLE patients, especially among populations at greater risk. Future research will aim to deepen the understanding of the genetic underpinnings contributing to HFpEF in SLE using proteomics, paving the way for personalized treatment protocols and preventive measures.

Presentation Date

• April 4, 2024 at 1:00 p.m. (EDT)

Scientific Question: Are there genetic risk factors associated with post-acute sequelae of SARS-CoV-2 infection (PASC)?

Keywords: Long-Covid, PASC, GWAS, COVID-19

Introduction: A subset of individuals infected with SARS CoV-2 can experience long-term effects from their infection, known as Long Covid or post-acute sequelae of SARS COV-2 infection (PASC). While genetic risk factors associated with COVID-19 severity have been described, there has been little research regarding genetic variants associated with the development of PASC in general or the specific long-term component conditions of PASC such as joint pain, palpitations, dizziness, confusion, anemia, etc. Since these PASC conditions may develop even in the absence of COVID-19, it is important to explore if genomic variants associated with these conditions differ in the presence or absence of COVID-19. This study leverages the diversity and breath of the All of Us dataset to identify populations with special risk for developing PASC through genome-wide association studies (GWAS).

Research objectives: The main research objective of this study is to identify genetic risk factors for PASC-associated conditions.

Methods: We identified all patients with COVID-19 based on receipt of the U07.1 ICD10 code or evidence of a COVID-19 positive antigen or PCR test. Within this group, we stratified the patients into those who developed any component conditions of PASC, as derived by the Adult RECOVER EHR Cohort, within 30 to 180 days of the onset of COVID infection, and controls who never developed the conditions. We also developed cohorts of patients with and without the PASC-like conditions who did not have a known history of COVID-19. This study explored two main comparisons on the ClinVar dataset. First, among patients with COVID-19, we conducted a GWAS analysis to assess genomic risk of developing PASC or its component conditions. Second, for each PASC condition of interest, we conducted a GWAS of the likelihood of developing PASC conditions among patients with COVID-19 versus negative and positive population control cohorts in the absence of COVID-19 infection.

Results: This study identified a total of 16,762 COVID-19 patients of which 5920 were identified with PASC in the All of Us dataset. Limiting these cohorts to the subset with available genomic data left 12,117 patients with COVID-19 and 4280 patients with at least one PASC condition. and 413,457 population controls. No genetic risk factor was found for PASC overall, however multiple potential genetic risk factors were found for select component conditions. Specifically, among patients with COVID-19, SNYE1 was found as a potential genetic risk factor for developing abdominal pain after infection and TEK was found as a potential genetic risk factor for patients developing joint pain after infection. We look to further validate these results by comparing the PASC component condition cohort to population control cohorts in the absence of COVID-19 infection.

Conclusions: This study explores the robust genomic dataset All of Us has to offer on COVID-19 and PASC patients. Identifying potential genomic risk factors and their subsequent known associations with other diseases furthers research into the genomic components of PASC and what populations are most at risk.