Large language models facilitate the generation of electronic health record phenotyping algorithms

OBJECTIVES

Phenotyping is a core task in observational health research utilizing electronic health records (EHRs). Developing an accurate algorithm demands substantial input from domain experts, involving extensive literature review and evidence synthesis. This burdensome process limits scalability and delays knowledge discovery. We investigate the potential for leveraging large language models (LLMs) to enhance the efficiency of EHR phenotyping by generating high-quality algorithm drafts.

MATERIALS AND METHODS

We prompted four LLMs-GPT-4 and GPT-3.5 of ChatGPT, Claude 2, and Bard-in October 2023, asking them to generate executable phenotyping algorithms in the form of SQL queries adhering to a common data model (CDM) for three phenotypes (ie, type 2 diabetes mellitus, dementia, and hypothyroidism). Three phenotyping experts evaluated the returned algorithms across several critical metrics. We further implemented the top-rated algorithms and compared them against clinician-validated phenotyping algorithms from the Electronic Medical Records and Genomics (eMERGE) network.

RESULTS

GPT-4 and GPT-3.5 exhibited significantly higher overall expert evaluation scores in instruction following, algorithmic logic, and SQL executability, when compared to Claude 2 and Bard. Although GPT-4 and GPT-3.5 effectively identified relevant clinical concepts, they exhibited immature capability in organizing phenotyping criteria with the proper logic, leading to phenotyping algorithms that were either excessively restrictive (with low recall) or overly broad (with low positive predictive values).

CONCLUSION

GPT versions 3.5 and 4 are capable of drafting phenotyping algorithms by identifying relevant clinical criteria aligned with a CDM. However, expertise in informatics and clinical experience is still required to assess and further refine generated algorithms.

Individualized Dose-Response to Statins Associated with Cardiovascular Disease Outcomes

Background

Statins reduce low-density lipoprotein cholesterol (LDL-C) and are efficacious in the prevention of atherosclerotic cardiovascular disease (ASCVD). Dose-response to statins varies among patients and can be modeled using three distinct pharmacological properties: (1) E0 (baseline LDL-C), (2) ED50 (potency: median dose achieving 50% reduction in LDL-C); and (3) Emax (efficacy: maximum LDL-C reduction). However, individualized dose-response and its association with ASCVD events remains unknown.

Objective

We analyze the relationship between ED50 and Emax with real-world cardiovascular disease outcomes.

Method

We leveraged de-identified electronic health record data to identify individuals exposed to multiple doses of the three most commonly prescribed statins (atorvastatin, simvastatin, or rosuvastatin) within the context of their longitudinal healthcare. We derived ED50 and Emax to quantify the relationship with a composite outcome of ASCVD events and all-cause mortality.

Results

We estimated ED50 and Emax for 3,033 unique individuals (atorvastatin: 1,632, simvastatin: 1,089, and rosuvastatin: 312) using a nonlinear, mixed effects dose-response model. Time-to-event analyses revealed that ED50 and Emax are independently associated with the primary endpoint. Hazard ratios were 0.85 (p < 0.01), 0.83 (p < 0.01), and 0.87 (p = 0.10) for ED50 and 1.13 (p < 0.001), 1.06 (p < 0.001), and 1.15 (p = 0.009) for Emax in the atorvastatin, simvastatin, and rosuvastatin cohorts, respectively.

Conclusion

The class-wide association of ED50 and Emax with clinical outcomes indicates that these measures influence the risk for ASCVD events in patients on statins.

Characterizing genetic profiles for high triglyceride levels in U.S. patients of African ancestry

Hypertriglyceridemia (HTG) is a common cardiovascular risk factor characterized by elevated circulating triglyceride (TG) levels. Researchers have assessed the genetic factors that influence HTG in studies focused predominantly on individuals of European ancestry (EA). However, relatively little is known about the contribution of genetic variation to HTG in people of AA, potentially constraining research and treatment opportunities; the lipid profile for African ancestry (AA) populations differs from that of EA populations-which may be partially attributable to genetics. Our objective was to characterize genetic profiles among individuals of AA with mild-to-moderate HTG and severe HTG versus those with normal TGs by leveraging whole genome sequencing (WGS) data and longitudinal electronic health records (EHRs) available in the All of Us (AoU) program. We compared the enrichment of functional variants within five canonical TG metabolism genes, an AA-specific polygenic risk score for TGs, and frequencies of 145 known potentially causal TG variants between patients with HTG and normal TG among a cohort of AA patients (N=15,373). Those with mild-to-moderate HTG (N=342) and severe HTG (N≤20) were more likely to carry APOA5 p.S19W (OR=2.21, 95% CI [1.70-2.88], p=2.74×10 -9 and OR=4.16, 95% CI [1.39-12.45], p=0.01, respectively) than those with normal TG. They were also more likely to have an elevated (top 10%) PRS, elevated carriage of potentially causal variant alleles, and carry any genetic risk factor. Alternative definitions of HTG yielded comparable results. In conclusion, individuals of AA with HTG were enriched for genetic risk factors compared to individuals with normal TGs.

Large Language Models Facilitate the Generation of Electronic Health Record Phenotyping Algorithms

Objectives

Phenotyping is a core task in observational health research utilizing electronic health records (EHRs). Developing an accurate algorithm demands substantial input from domain experts, involving extensive literature review and evidence synthesis. This burdensome process limits scalability and delays knowledge discovery. We investigate the potential for leveraging large language models (LLMs) to enhance the efficiency of EHR phenotyping by generating high-quality algorithm drafts.

Materials and Methods

We prompted four LLMs-GPT-4 and GPT-3.5 of ChatGPT, Claude 2, and Bard-in October 2023, asking them to generate executable phenotyping algorithms in the form of SQL queries adhering to a common data model (CDM) for three phenotypes (i.e., type 2 diabetes mellitus, dementia, and hypothyroidism). Three phenotyping experts evaluated the returned algorithms across several critical metrics. We further implemented the top-rated algorithms and compared them against clinician-validated phenotyping algorithms from the Electronic Medical Records and Genomics (eMERGE) network.

Results

GPT-4 and GPT-3.5 exhibited significantly higher overall expert evaluation scores in instruction following, algorithmic logic, and SQL executability, when compared to Claude 2 and Bard. Although GPT-4 and GPT-3.5 effectively identified relevant clinical concepts, they exhibited immature capability in organizing phenotyping criteria with the proper logic, leading to phenotyping algorithms that were either excessively restrictive (with low recall) or overly broad (with low positive predictive values).

Conclusion

GPT versions 3.5 and 4 are capable of drafting phenotyping algorithms by identifying relevant clinical criteria aligned with a CDM. However, expertise in informatics and clinical experience is still required to assess and further refine generated algorithms.

Leveraging generative AI to prioritize drug repurposing candidates for Alzheimer's disease with real-world clinical validation

Drug repurposing represents an attractive alternative to the costly and time-consuming process of new drug development, particularly for serious, widespread conditions with limited effective treatments, such as Alzheimer's disease (AD). Emerging generative artificial intelligence (GAI) technologies like ChatGPT offer the promise of expediting the review and summary of scientific knowledge. To examine the feasibility of using GAI for identifying drug repurposing candidates, we iteratively tasked ChatGPT with proposing the twenty most promising drugs for repurposing in AD, and tested the top ten for risk of incident AD in exposed and unexposed individuals over age 65 in two large clinical datasets: (1) Vanderbilt University Medical Center and (2) the All of Us Research Program. Among the candidates suggested by ChatGPT, metformin, simvastatin, and losartan were associated with lower AD risk in meta-analysis. These findings suggest GAI technologies can assimilate scientific insights from an extensive Internet-based search space, helping to prioritize drug repurposing candidates and facilitate the treatment of diseases.

Longitudinal evaluation of cervical length and shear wave elastography in women with spontaneous preterm birth

OBJECTIVES

To compare longitudinal changes in cervical length (CL) and mean cervical shear wave elastography (CSWE) scores between women with singleton and twin pregnancies who experience spontaneous preterm birth (sPTB) and those who have term births (TB).

METHODS

This was a prospective longitudinal study of 1264 unselected women with singleton (n=1143) and twin (n=121) pregnancy attending a dedicated research clinic for screening of sPTB at 4 timepoints during pregnancy including 11-15+6 (visit 1), 16-20+6 (visit 2), 21-24+6 (visit 3) and 28-32+6 (visit 4) weeks of gestation. At each visit, a transvaginal ultrasound scan was conducted to measure the CL and the CSWE scores from six regions of interest (ROI) (inner, middle, and external parts of anterior and posterior lips) in the cervix. The mean of CSWE scores from the six ROIs were calculated for data analysis. Log10 transformation was applied to make the data Gaussian prior to statistical analysis. A multilevel mixed-effects analysis was performed to compare CL and CSWE longitudinally between sPTB and TB groups.

RESULTS

A total of 57 (4.99%) singleton pregnancies and 33 (27.27%) twin pregnancies were complicated with sPTB. Women with sPTB had shorter CL across gestation when controlling for history of cervical surgery, number of fetuses, gestational age at cervical assessment (GA), and the interaction between GA and sPTB. CL in the sPTB group was significantly lower than that of the TB group at 21-24+6 weeks (p=0.039) and 28-32+6 weeks (p<0.001). Twin pregnancies had significantly longer CL throughout pregnancy, compared to singleton pregnancies (coefficient=0.01864, p<0.001). Furthermore, after adjusting for maternal age, weight, height, body mass index (BMI), and GA, CSWE scores in sPTB group were significantly lower in the sPTB group across gestation, compared to the TB group (1.28265 vs 1.32832; p=0.013). However, in the individual visit analysis, CSWE scores in the sPTB group were significantly lower than that of the TB group only at 11-15+6 weeks (p=0.013). There was no difference in CSWE scores between singleton and twin pregnancies throughout pregnancy (coefficient=-0.00128, p=0.937).

CONCLUSION

Women with sPTB have shorter CL and softer cervix across gestation when compared to those with TB. In the individual visit analysis, the reduction in CL in the sPTB group occurs from late second trimester onwards, while the reduction in cervical stiffness in the sPTB group is observed primarily in the first trimester. Additionally, our study has found that CL is significantly shorter in singleton pregnancies compared to twin pregnancies, while cervical stiffness does not differ between the two types of pregnancy. Our findings indicate that the cervix tends to undergo a softening process prior to shortening in the sPTB cases This article is protected by copyright. All rights reserved.

Selection, optimization and validation of ten chronic disease polygenic risk scores for clinical implementation in diverse US populations

Polygenic risk scores (PRSs) have improved in predictive performance, but several challenges remain to be addressed before PRSs can be implemented in the clinic, including reduced predictive performance of PRSs in diverse populations, and the interpretation and communication of genetic results to both providers and patients. To address these challenges, the National Human Genome Research Institute-funded Electronic Medical Records and Genomics (eMERGE) Network has developed a framework and pipeline for return of a PRS-based genome-informed risk assessment to 25,000 diverse adults and children as part of a clinical study. From an initial list of 23 conditions, ten were selected for implementation based on PRS performance, medical actionability and potential clinical utility, including cardiometabolic diseases and cancer. Standardized metrics were considered in the selection process, with additional consideration given to strength of evidence in African and Hispanic populations. We then developed a pipeline for clinical PRS implementation (score transfer to a clinical laboratory, validation and verification of score performance), and used genetic ancestry to calibrate PRS mean and variance, utilizing genetically diverse data from 13,475 participants of the All of Us Research Program cohort to train and test model parameters. Finally, we created a framework for regulatory compliance and developed a PRS clinical report for return to providers and for inclusion in an additional genome-informed risk assessment. The initial experience from eMERGE can inform the approach needed to implement PRS-based testing in diverse clinical settings.

Developing and evaluating pediatric phecodes (Peds-Phecodes) for high-throughput phenotyping using electronic health records

OBJECTIVE

Pediatric patients have different diseases and outcomes than adults; however, existing phecodes do not capture the distinctive pediatric spectrum of disease. We aim to develop specialized pediatric phecodes (Peds-Phecodes) to enable efficient, large-scale phenotypic analyses of pediatric patients.

MATERIALS AND METHODS

We adopted a hybrid data- and knowledge-driven approach leveraging electronic health records (EHRs) and genetic data from Vanderbilt University Medical Center to modify the most recent version of phecodes to better capture pediatric phenotypes. First, we compared the prevalence of patient diagnoses in pediatric and adult populations to identify disease phenotypes differentially affecting children and adults. We then used clinical domain knowledge to remove phecodes representing phenotypes unlikely to affect pediatric patients and create new phecodes for phenotypes relevant to the pediatric population. We further compared phenome-wide association study (PheWAS) outcomes replicating known pediatric genotype-phenotype associations between Peds-Phecodes and phecodes.

RESULTS

The Peds-Phecodes aggregate 15 533 ICD-9-CM codes and 82 949 ICD-10-CM codes into 2051 distinct phecodes. Peds-Phecodes replicated more known pediatric genotype-phenotype associations than phecodes (248 vs 192 out of 687 SNPs, P < .001).

DISCUSSION

We introduce Peds-Phecodes, a high-throughput EHR phenotyping tool tailored for use in pediatric populations. We successfully validated the Peds-Phecodes using genetic replication studies. Our findings also reveal the potential use of Peds-Phecodes in detecting novel genotype-phenotype associations for pediatric conditions. We expect that Peds-Phecodes will facilitate large-scale phenomic and genomic analyses in pediatric populations.

CONCLUSION

Peds-Phecodes capture higher-quality pediatric phenotypes and deliver superior PheWAS outcomes compared to phecodes.

Investigating the Impact of Tea Consumption on Cognitive Function and Exploring Tea-Genetic Interactions in Older Adults Aged 65-105 Years: Findings from the 2002-2018 CLHLS Data

BACKGROUND

As the global population ages, cognitive impairment (CI) becomes more prevalent. Tea has been one of the most popular drinks in the world. Several studies have demonstrated that tea consumption has an impact on cognitive function.

OBJECTIVE

This study aims to examine the association between tea consumption and cognitive function and explore the potential effect of genetics on the relationship between tea consumption and CI risk in older adults.

DESIGN

This is a prospective longitudinal study using data from the Chinese Longitudinal Healthy Longevity Survey (CLHLS).

SETTING

Six waves of data from CLHLS containing 76,270 subjects were analyzed. Generalized estimation equations (GEE) with a logit link function were adopted to estimate the effect of tea consumption on CI risk from a cross-sectional and longitudinal perspective.

PARTICIPANTS

A population-based cohort of adults aged 65-105 years.

MEASUREMENTS

The frequency and type of tea consumption were obtained by questionnaires. CI was measured based on MMSE. Polygenic risk was measured using the polygenic score approach described by the International Schizophrenia.

RESULTS

The results showed that drinking green tea had a better protective effect on cognitive function than other types of tea, the incidence of CI gradually decreased with the increase of tea consumption frequency, and men were more likely to benefit from tea consumption. Additionally, we also found a significant interaction between tea consumption and genetic risk, measured by polygenic risk score (PRS).

CONCLUSIONS

Based on current research evidence, tea consumption, may be a simple and important measure for CI prevention.

Genetic and Survey Data Improves Performance of Machine Learning Model for Long COVID

Over 200 million SARS-CoV-2 patients have or will develop persistent symptoms (long COVID). Given this pressing research priority, the National COVID Cohort Collaborative (N3C) developed a machine learning model using only electronic health record data to identify potential patients with long COVID. We hypothesized that additional data from health surveys, mobile devices, and genotypes could improve prediction ability. In a cohort of SARS-CoV-2 infected individuals (n=17,755) in the All of Us program, we applied and expanded upon the N3C long COVID prediction model, testing machine learning infrastructures, assessing model performance, and identifying factors that contributed most to the prediction models. For the survey/mobile device information and genetic data, extreme gradient boosting and a convolutional neural network delivered the best performance for predicting long COVID, respectively. Combined survey, genetic, and mobile data increased specificity and the Area Under Curve the Receiver Operating Characteristic score versus the original N3C model.

[Standardized nomenclature of oral microorganisms in Chinese: the 2023 update]

Oral microbial community, as an important part of human microbial community, is closely related to oral and general health. Oral microbiological research has become the forefront of international microbiological research. Standardized and unified nomenclature for oral microorganisms in Chinese is of great significance to support the development of oral medicine research. Standardized translation of microbial names is the basis for writing canonical and authoritative professional textbooks and reference books, which helps students to accurately acquire the characteristics and classifications of oral microbes. Unified translation of oral microorganisms is also conducive to academic communication and cooperation, and plays an important role in oral health education and science popularization, which enables oral microbiology knowledge to be accurately disseminated to the public. Therefore, in order to standardize the words in scientific research, funding application, publications, academic exchanges and science popularization within the field of oral medicine, we have fully discussed and revised the Chinese names of oral microorganisms in 2017 edition and ones of newly discovered oral microbes, finally reaching a consensus to form the 2023 edition of Chinese names of oral microorganisms.

Efficacy of Postoperative Radiotherapy for Patients with New N2 Descriptors of Subclassification in Completely Resected Non-Small Cell Lung Cancer: A Real-World Study

PURPOSE/OBJECTIVE(S)

Patients with N2 non-small cell lung cancer (NSCLC) were heterogeneous groups and required further stratification. The International Society for the Study of Lung Cancer (IASLC) added new descriptors of three sub-stages for stage N2 NSCLC: N2 at a single station without N1 involvement (N2a1), N2 at a single station with N1 involvement (N2a2), and N2 at multiple stations (N2b). This study aimed to investigate the efficacy of postoperative radiotherapy (PORT) for patients with these N2 descriptors.

MATERIALS/METHODS

Patients with histologically confirmed NSCLC after complete resection and divided into PORT group and non-PORT group. The primary endpoint was DFS. The second endpoints were overall survival (OS) and locoregional recurrence-free survival (LRFS). Propensity-score matching (PSM) of baseline characteristics between the PORT and non-PORT groups was used for validation.

RESULTS

Totally 1832 patients were enrolled, including 308 N2a1 patients, 682 N2a2 patients, and 842 N2b patients. The median follow-up time was 50.1 months. The survival outcomes of the PORT and non-PORT groups before PSM were shown in Table 1. For patients with N2a1, PORT could not improve the DFS (median DFS of the PORT group and the non-PORT group: not reached vs. 46.8 months, P = 0.41), OS (P = 0.85), or LRFS (P = 0.32), which were consistent with the multivariate analysis and data after the PSM. For patients with N2a2, PORT significantly improved the DFS (median DFS 29.7 vs. 22.2 months, P = 0.02), OS (P = 0.03), and LRFS (P = 0.01). The multivariate analysis and data after the PSM confirmed the benefits in DFS and LRFS, but no benefit was observed in OS (multivariate analysis: HR 0.79, P = 0.18; median OS after PSM: 103.7 vs. 63.1 months, P = 0.34). For patients with N2b, PORT could not improve the DFS (median DFS 20.6 vs. 21.2 months, P = 0.39) but significantly improved the OS (P<0.001) and LRFS (P<0.001). However, the multivariate analysis showed that PORT significantly improved DFS (HR 0.81, P = 0.03), consistent with the data after the PSM (median DFS 20.6 and 17.6 months, P = 0.04).

CONCLUSION

PORT significantly improved the DFS and LRFS in patients with N2a2 and significantly improved the DFS, LRFS, and OS in patients with N2b. Patients with N2a1 could not benefit from PORT.

Chemoradiotherapy Combined with Nab-Paclitaxel plus Cisplatin in Patients with Locally Advanced Borderline Resectable or Unresectable Esophageal Squamous Cell Carcinoma: A Phase I/II Study

PURPOSE/OBJECTIVE(S)

To evaluate the efficacy and safety of nanoparticle albumin-bound paclitaxel (nab-PTX) plus cisplatin as the regimen of conversional chemoradiotherapy (cCRT) in locally advanced borderline resectable or unresectable esophageal squamous cell carcinoma (ESCC).

MATERIALS/METHODS

Patients with locally advanced ESCC (cT3-4, Nany, M0-1, M1 was limited to lymph node metastasis in the supraclavicular area) were enrolled. All the patients received the cCRT of nab-PTX plus cisplatin. After the cCRT, those resectable patients received esophagectomy; those unresectable patients continued to receive the definitive chemoradiotherapy (dCRT). The locoregional control (LRC), overall survival (OS), progression-free survival (PFS), distant metastasis free survival (DMFS), pathological complete response (pCR), R0 resection rate and adverse events (AEs) were calculated.

RESULTS

A total of 45 patients with ESCC treated from October 2019 to May 2021 were finally included. The median follow-up time was 30.3 months. The LRC, OS, EFS, DMFS at 1and 2 years were 81.5%, 86.6%, 64.3%, 73.2% and 72.4%, 68.8%, 44.8%, 52.7% respectively. 21 patients (46.7%) received conversional chemoradiotherapy plus surgery (cCRT+S). The pCR rate and R0 resection rate were 47.6% and 84.0%. The LRC rate at 1 and 2 years were 95.0%, 87.1% in cCRT+S patients and 69.3%, 58.7% in dCRT patients respectively (HR, 5.14; 95% CI, 1.10-23.94; P = 0.021). The OS rate at 1 and 2 years were 95.2% and 84.2% in resectable patients compared to 78.8% and 54.4% in unresectable patients (HR, 3.41; 95% CI, 1.10-10.61; P = 0.024). The toxicities during chemoradiotherapy were tolerated, the most common grade 3-4 toxicities were radiation esophagitis (15.6%).

CONCLUSION

Nab-PTX plus cisplatin were effective and safe as the regimen of conversional chemoradiotherapy of ESCC. The patients receiving conversional chemoradiotherapy plus surgery (cCRT+S) were prone to have a better survival.

Association between APOL1 risk variants and the occurrence of sepsis in Black patients hospitalized with infections: a retrospective cohort study

Background

Two risk variants in the apolipoprotein L1 gene ( APOL1 ) have been associated with increased susceptibility to sepsis in Black patients. However, it remains unclear whether APOL1 high-risk genotypes are associated with occurrence of either sepsis or sepsis-related phenotypes in patients hospitalized with infections, independent of their association with pre-existing severe renal disease.

Methods

A retrospective cohort study of 2,242 Black patients hospitalized with infections. We assessed whether carriage of APOL1 high-risk genotypes was associated with the risk of sepsis and sepsis-related phenotypes in patients hospitalized with infections. The primary outcome was sepsis; secondary outcomes were short-term mortality and organ failure related to sepsis.

Results

Of 2,242 Black patients hospitalized with infections, 565 developed sepsis. Patients with high-risk APOL1 genotypes had a significantly increased risk of sepsis (odds ratio [OR]=1.29 [95% CI, 1.00-1.67; p=0.047]); however, this association was not significant after adjustment for pre-existing severe renal disease (OR=1.14 [95% CI, 0.88-1.48; p=0.33]), nor after exclusion of those patients with pre-existing severe renal disease (OR=0.99 [95% CI, 0.70-1.39; p=0.95]. APOL1 high-risk genotypes were significantly associated with the renal dysfunction component of the Sepsis-3 criteria (OR=1.64 [95% CI, 1.21-2.22; p=0.001], but not with other sepsis-related organ dysfunction or short-term mortality. The association between high-risk APOL1 genotypes and sepsis-related renal dysfunction was markedly attenuated by adjusting for pre-existing severe renal disease (OR=1.36 [95% CI, 1.00-1.86; p=0.05]) and was nullified after exclusion of patients with pre-existing severe renal disease (OR=1.16 [95% CI, 0.74-1.81; p=0.52]).

Conclusion

APOL1 high-risk genotypes were associated with an increased risk of sepsis; however, this increased risk was attributable predominantly to pre-existing severe renal disease.

Funding

This study was supported by R01GM120523 (Q.F.), R01HL163854 (Q.F.), R35GM131770 (C.M.S.), HL133786 (W.Q.W.), and Vanderbilt Faculty Research Scholar Fund (Q.F.). The dataset(s) used for the analyses described were obtained from Vanderbilt University Medical Center's BioVU which is supported by institutional funding, the 1S10RR025141-01 instrumentation award, and by the CTSA grant UL1TR0004from NCATS/NIH. Additional funding provided by the NIH through grants P50GM115305 and U19HL065962. The authors wish to acknowledge the expert technical support of the VANTAGE and VANGARD core facilities, supported in part by the Vanderbilt-Ingram Cancer Center (P30 CA068485) and Vanderbilt Vision Center (P30 EY08126).The funders had no role in design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Developing and Evaluating Pediatric Phecodes (Peds-Phecodes) for High-Throughput Phenotyping Using Electronic Health Records

Objective

Pediatric patients have different diseases and outcomes than adults; however, existing phecodes do not capture the distinctive pediatric spectrum of disease. We aim to develop specialized pediatric phecodes (Peds-Phecodes) to enable efficient, large-scale phenotypic analyses of pediatric patients.

Materials and Methods

We adopted a hybrid data- and knowledge-driven approach leveraging electronic health records (EHRs) and genetic data from Vanderbilt University Medical Center to modify the most recent version of phecodes to better capture pediatric phenotypes. First, we compared the prevalence of patient diagnoses in pediatric and adult populations to identify disease phenotypes differentially affecting children and adults. We then used clinical domain knowledge to remove phecodes representing phenotypes unlikely to affect pediatric patients and create new phecodes for phenotypes relevant to the pediatric population. We further compared phenome-wide association study (PheWAS) outcomes replicating known pediatric genotype-phenotype associations between Peds-Phecodes and phecodes.

Results

The Peds-Phecodes aggregate 15,533 ICD-9-CM codes and 82,949 ICD-10-CM codes into 2,051 distinct phecodes. Peds-Phecodes replicated more known pediatric genotype-phenotype associations than phecodes (248 versus 192 out of 687 SNPs, p<0.001).

Discussion

We introduce Peds-Phecodes, a high-throughput EHR phenotyping tool tailored for use in pediatric populations. We successfully validated the Peds-Phecodes using genetic replication studies. Our findings also reveal the potential use of Peds-Phecodes in detecting novel genotype-phenotype associations for pediatric conditions. We expect that Peds-Phecodes will facilitate large-scale phenomic and genomic analyses in pediatric populations.

Conclusion

Peds-Phecodes capture higher-quality pediatric phenotypes and deliver superior PheWAS outcomes compared to phecodes.

Clinical diagnoses associated with a positive antinuclear antibody test in patients with and without autoimmune disease

BACKGROUND

Antinuclear antibodies (ANA) are antibodies present in several autoimmune disorders. However, a large proportion of the general population (20%) also have a positive test; very few of these individuals will develop an autoimmune disease, and the clinical impact of a positive ANA in them is not known. Thus, we test the hypothesis that ANA + test reflects a state of immune dysregulation that alters risk for some clinical disorders in individuals without an autoimmune disease.

METHODS

We performed high throughput association analyses in a case-control study using real world data from the de-identified electronic health record (EHR) system from Vanderbilt University Medical Center. The study population included individuals with an ANA titer ≥ 1:80 at any time (ANA +) and those with negative results (ANA-). The cohort was stratified into sub-cohorts of individuals with and without an autoimmune disease. A phenome-wide association study (PheWAS) adjusted by sex, year of birth, race, and length of follow-up was performed in the study cohort and in the sub-cohorts. As secondary analyses, only clinical diagnoses after ANA testing were included in the analyses.

RESULTS

The cohort included 70,043 individuals: 49,546 without and 20,497 with an autoimmune disease, 26,579 were ANA + and 43,464 ANA-. In the study cohort and the sub-cohort with autoimmune disease, ANA + was associated (P ≤ 5 × 10-5) with 88 and 136 clinical diagnoses respectively, including lupus (OR ≥ 5.4, P ≤ 7.8 × 10-202) and other autoimmune diseases and complications. In the sub-cohort without autoimmune diseases, ANA + was associated with increased risk of Raynaud's syndrome (OR ≥ 2.1) and alveolar/perialveolar-related pneumopathies (OR ≥ 1.4) and decreased risk of hepatitis C, tobacco use disorders, mood disorders, convulsions, fever of unknown origin, and substance abuse disorders (OR ≤ 0.8). Analyses including only diagnoses after ANA testing yielded similar results.

CONCLUSION

A positive ANA test, in addition to known associations with autoimmune diseases, Raynaud's phenomenon, and idiopathic fibrosing alveolitis related disorders, is associated with decreased prevalence of several non-autoimmune diseases.

Leveraging Generative AI to Prioritize Drug Repurposing Candidates: Validating Identified Candidates for Alzheimer's Disease in Real-World Clinical Datasets

Drug repurposing represents an attractive alternative to the costly and time-consuming process of new drug development, particularly for serious, widespread conditions with limited effective treatments, such as Alzheimer's disease (AD). Emerging generative artificial intelligence (GAI) technologies like ChatGPT offer the promise of expediting the review and summary of scientific knowledge. To examine the feasibility of using GAI for identifying drug repurposing candidates, we iteratively tasked ChatGPT with proposing the twenty most promising drugs for repurposing in AD, and tested the top ten for risk of incident AD in exposed and unexposed individuals over age 65 in two large clinical datasets: 1) Vanderbilt University Medical Center and 2) the All of Us Research Program. Among the candidates suggested by ChatGPT, metformin, simvastatin, and losartan were associated with lower AD risk in meta-analysis. These findings suggest GAI technologies can assimilate scientific insights from an extensive Internet-based search space, helping to prioritize drug repurposing candidates and facilitate the treatment of diseases.

Leveraging Generative AI to Prioritize Drug Repurposing Candidates: Validating Identified Candidates for Alzheimer's Disease in Real-World Clinical Datasets

Drug repurposing represents an attractive alternative to the costly and time-consuming process of new drug development, particularly for serious, widespread conditions with limited effective treatments, such as Alzheimer's disease (AD). Emerging generative artificial intelligence (GAI) technologies like ChatGPT offer the promise of expediting the review and summary of scientific knowledge. To examine the feasibility of using GAI for identifying drug repurposing candidates, we iteratively tasked ChatGPT with proposing the twenty most promising drugs for repurposing in AD, and tested the top ten for risk of incident AD in exposed and unexposed individuals over age 65 in two large clinical datasets: 1) Vanderbilt University Medical Center and 2) the All of Us Research Program. Among the candidates suggested by ChatGPT, metformin, simvastatin, and losartan were associated with lower AD risk in meta-analysis. These findings suggest GAI technologies can assimilate scientific insights from an extensive Internet-based search space, helping to prioritize drug repurposing candidates and facilitate the treatment of diseases.

[Deep learning-based dose prediction in radiotherapy planning for head and neck cancer]

OBJECTIVE

To propose an deep learning-based algorithm for automatic prediction of dose distribution in radiotherapy planning for head and neck cancer.

METHODS

We propose a novel beam dose decomposition learning (BDDL) method designed on a cascade network. The delivery matter of beam through the planning target volume (PTV) was fitted with the pre-defined beam angles, which served as an input to the convolution neural network (CNN). The output of the network was decomposed into multiple sub-fractions of dose distribution along the beam directions to carry out a complex task by performing multiple simpler sub-tasks, thus allowing the model more focused on extracting the local features. The subfractions of dose distribution map were merged into a distribution map using the proposed multi-voting mechanism. We also introduced dose distribution features of the regions-of-interest (ROIs) and boundary map as the loss function during the training phase to serve as constraining factors of the network when extracting features of the ROIs and areas of dose boundary. Public datasets of radiotherapy planning for head and neck cancer were used for obtaining the accuracy of dose distribution of the BDDL method and for implementing the ablation study of the proposed method.

RESULTS

The BDDL method achieved a Dose score of 2.166 and a DVH score of 1.178 (P < 0.05), demonstrating its superior prediction accuracy to that of current state-ofthe-art (SOTA) methods. Compared with the C3D method, which was in the first place in OpenKBP-2020 Challenge, the BDDL method improved the Dose score and DVH score by 26.3% and 30%, respectively. The results of the ablation study also demonstrated the effectiveness of each key component of the BDDL method.

CONCLUSION

The BDDL method utilizes the prior knowledge of the delivery matter of beam and dose distribution in the ROIs to establish a dose prediction model. Compared with the existing methods, the proposed method is interpretable and reliable and can be potentially applied in clinical radiotherapy.

Selection, optimization, and validation of ten chronic disease polygenic risk scores for clinical implementation in diverse populations

Polygenic risk scores (PRS) have improved in predictive performance supporting their use in clinical practice. Reduced predictive performance of PRS in diverse populations can exacerbate existing health disparities. The NHGRI-funded eMERGE Network is returning a PRS-based genome-informed risk assessment to 25,000 diverse adults and children. We assessed PRS performance, medical actionability, and potential clinical utility for 23 conditions. Standardized metrics were considered in the selection process with additional consideration given to strength of evidence in African and Hispanic populations. Ten conditions were selected with a range of high-risk thresholds: atrial fibrillation, breast cancer, chronic kidney disease, coronary heart disease, hypercholesterolemia, prostate cancer, asthma, type 1 diabetes, obesity, and type 2 diabetes. We developed a pipeline for clinical PRS implementation, used genetic ancestry to calibrate PRS mean and variance, created a framework for regulatory compliance, and developed a PRS clinical report. eMERGE's experience informs the infrastructure needed to implement PRS-based implementation in diverse clinical settings.

[General anesthesia versus deep sedation for dental treatment in children: comparison of parental acceptance, oral health-related quality of life, and treatment efficacy]

OBJECTIVE

To compare the parental acceptance of dental treatment under general anesthesia and deep sedation in children and assess the changes in postoperative oral health-related quality of life and treatment efficacy.

METHODS

The parents of 131 children undergoing dental treatment in the Department of Stomatology of Sichuan Provincial People's Hospital from January, 2022 to June, 2022 were surveyed using a questionnaire of children's advanced oral behavior management, and 83 children receiving general anesthesia or deep sedation for dental treatment between January, 2018 and December, 2021 were also investigated for changes in quality of life after the treatment using a questionnaire. The treatment efficacy was assessed at the 1-year follow-up visit in 149 children who received dental treatment under general anesthesia or deep sedation during the same period.

RESULTS

The survey of perantal acceptance showed that 62.6% of the parents preferred deep sedation, 29.01% preferred general anesthesia, and 8.4% preferred compulsory treatment. Dental treatments under general anesthesia and deep sedation both significantly improved oral health-related quality of life of the children. While dental surgeries under general anesthesia resulted in the most significant improvement of pain symptoms, deep sedation was associated with both obvious relief of the children's pain symptoms and reduction of the parents' pressure level. No significant difference was found in the efficacy of treatments under general anesthesia and deep sedation at the 1-year follow-up.

CONCLUSION

Dental treatment in children under deep sedation has the highest parental acceptance, followed by treatment under general anesthesia, and the acceptance of compulsory treatment is the lowest. The treatments under general anesthesia and deep sedation significantly improve the quality of life of the children and their parents and both have good treatment efficacy.

Observation of Monoenergetic Electrons from Two-Pulse Ionization Injection in Quasilinear Laser Wakefields

The generation of low emittance electron beams from laser-driven wakefields is crucial for the development of compact x-ray sources. Here, we show new results for the injection and acceleration of quasimonoenergetic electron beams in low amplitude wakefields experimentally and using simulations. This is achieved by using two laser pulses decoupling the wakefield generation from the electron trapping via ionization injection. The injection duration, which affects the beam charge and energy spread, is found to be tunable by adjusting the relative pulse delay. By changing the polarization of the injector pulse, reducing the ionization volume, the electron spectra of the accelerated electron bunches are improved.

The relationship between high density lipoprotein cholesterol and sepsis: A clinical and genetic approach

Sepsis accounts for one in three hospital deaths. Higher concentrations of high-density lipoprotein cholesterol (HDL-C) are associated with apparent protection from sepsis, suggesting a potential therapeutic role for HDL-C or drugs, such as cholesteryl ester transport protein (CETP) inhibitors that increase HDL-C. However, these beneficial clinical associations might be due to confounding; genetic approaches can address this possibility. We identified 73,406 White adults admitted to Vanderbilt University Medical Center with infection; 11,612 had HDL-C levels, and 12,377 had genotype information from which we constructed polygenic risk scores (PRS) for HDL-C and the effect of CETP on HDL-C. We tested the associations between predictors (measured HDL-C, HDL-C PRS, CETP PRS, and rs1800777) and outcomes: sepsis, septic shock, respiratory failure, and in-hospital death. In unadjusted analyses, lower measured HDL-C concentrations were significantly associated with increased risk of sepsis (p = 2.4 × 10-23 ), septic shock (p = 4.1 × 10-12 ), respiratory failure (p = 2.8 × 10-8 ), and in-hospital death (p = 1.0 × 10-8 ). After adjustment (age, sex, electronic health record length, comorbidity score, LDL-C, triglycerides, and body mass index), these associations were markedly attenuated: sepsis (p = 2.6 × 10-3 ), septic shock (p = 8.1 × 10-3 ), respiratory failure (p = 0.11), and in-hospital death (p = 4.5 × 10-3 ). HDL-C PRS, CETP PRS, and rs1800777 significantly predicted HDL-C (p < 2 × 10-16 ), but none were associated with sepsis outcomes. Concordant findings were observed in 13,254 Black patients hospitalized with infections. Lower measured HDL-C levels were significantly associated with increased risk of sepsis and related outcomes in patients with infection, but a causal relationship is unlikely because no association was found between the HDL-C PRS or the CETP PRS and the risk of adverse sepsis outcomes.

Association Between Genetically Predicted Expression of TPMT and Azathioprine Adverse Events

Polymorphisms thiopurine-S-methyltransferase (TPMT) and nudix hydrolase 15 (NUDT15) can increase the risk of azathioprine myelotoxicity, but little is known about other genetic factors that increase risk for azathioprine-associated side effects. PrediXcan is a gene-based association method that estimates the expression of individuals' genes and examines their correlation to specified phenotypes. As proof of concept for using PrediXcan as a tool to define the association between genetic factors and azathioprine side effects, we aimed to determine whether the genetically predicted expression of TPMT or NUDT15 was associated with leukopenia or other known side effects. In a retrospective cohort of 1364 new users of azathioprine with EHR-reported White race, we used PrediXcan to impute expression in liver tissue, tested its association with pre-specified phecodes representing known side effects (e.g., skin cancer), and completed chart review to confirm cases. Among confirmed cases, patients in the lowest tertile (i.e., lowest predicted) of TPMT expression had significantly higher odds of developing leukopenia (OR=3.30, 95%CI: 1.07-10.20, p=0.04) versus those in the highest tertile; no other side effects were significant. The results suggest that this methodology could be deployed on a larger scale to uncover associations between genetic factors and drug side effects for more personalized care.

Implicating genes, pleiotropy, and sexual dimorphism at blood lipid loci through multi-ancestry meta-analysis

BACKGROUND

Genetic variants within nearly 1000 loci are known to contribute to modulation of blood lipid levels. However, the biological pathways underlying these associations are frequently unknown, limiting understanding of these findings and hindering downstream translational efforts such as drug target discovery.

RESULTS

To expand our understanding of the underlying biological pathways and mechanisms controlling blood lipid levels, we leverage a large multi-ancestry meta-analysis (N = 1,654,960) of blood lipids to prioritize putative causal genes for 2286 lipid associations using six gene prediction approaches. Using phenome-wide association (PheWAS) scans, we identify relationships of genetically predicted lipid levels to other diseases and conditions. We confirm known pleiotropic associations with cardiovascular phenotypes and determine novel associations, notably with cholelithiasis risk. We perform sex-stratified GWAS meta-analysis of lipid levels and show that 3-5% of autosomal lipid-associated loci demonstrate sex-biased effects. Finally, we report 21 novel lipid loci identified on the X chromosome. Many of the sex-biased autosomal and X chromosome lipid loci show pleiotropic associations with sex hormones, emphasizing the role of hormone regulation in lipid metabolism.

CONCLUSIONS

Taken together, our findings provide insights into the biological mechanisms through which associated variants lead to altered lipid levels and potentially cardiovascular disease risk.

Genome-wide association analyses of common infections in a large practice-based biobank

INTRODUCTION

Infectious diseases are common causes of morbidity and mortality worldwide. Susceptibility to infection is highly heritable; however, little has been done to identify the genetic determinants underlying common infectious diseases. One GWAS was performed using 23andMe information about self-reported infections; we set out to confirm previous loci and identify new ones using medically diagnosed infections.

METHODS

We used the electronic health record (EHR)-based biobank at Vanderbilt and diagnosis codes to identify cases of 12 infectious diseases in white patients: urinary tract infection, pneumonia, chronic sinus infections, otitis media, candidiasis, streptococcal pharyngitis, herpes zoster, herpes labialis, hepatitis B, infectious mononucleosis, tuberculosis (TB) or a positive TB test, and hepatitis C. We selected controls from patients with no diagnosis code for the candidate disease and matched by year of birth, sex, and calendar year at first and last EHR visits. We conducted GWAS using SAIGE and transcriptome-wide analysis (TWAS) using S-PrediXcan. We also conducted phenome-wide association study to understand associations between identified genetic variants and clinical phenotypes.

RESULTS

We replicated three 23andMe loci (p ≤ 0.05): herpes zoster and rs7047299-A (p = 2.6 × 10^-3) and rs2808290-C (p = 9.6 × 10^-3;); otitis media and rs114947103-C (p = 0.04). We also identified 2 novel regions (p ≤ 5 × 10^-8): rs113235453-G for otitis media (p = 3.04 × 10^-8), and rs10422015-T for candidiasis (p = 3.11 × 10^-8). In TWAS, four gene-disease associations were significant: SLC30A9 for otitis media (p = 8.06 × 10^-7); LRP3 and WDR88 for candidiasis (p = 3.91 × 10^-7 and p = 1.95 × 10^-6); and AAMDC for hepatitis B (p = 1.51 × 10^-6).

CONCLUSION

We conducted GWAS and TWAS for 12 infectious diseases and identified novel genetic contributors to the susceptibility of infectious diseases.

[FUT8 modulates galectin-3 expression to regulate TGF-β1-mediated fibrosis of lung fibroblasts]

OBJECTIVE

To investigate the regulatory role of α-1, 6-fucosyltransferase (FUT8) in TGF-β1-induced proliferation, migration and fibrosis of human embryonic lung fibroblasts (MRC-5 cells) and explore the underlying molecular mechanism.

METHODS

C57/BL6 mice were randomized into 4 groups for treatment with saline (control group), bleomycin, bleomycin+sh-NC or bleomycin+sh-FUT8, and pulmonary fibrosis was observed using Masson staining.MRC-5 cells were transfected with si-NC, FUT8 siRNA (si-FUT8), or both si-FUT8 and a galectin-3(Gal-3) overexpression plasmid (pcDNA3.1-Gal) prior to TGF-β1 treatment, and the changes in cell proliferation and migration were assessed using CCK-8 assay, BrdU assay, and wound healing assay; the changes in the expression levels of α-SMA, collagen I (COLIA1) and extracellular matrix fibronectin (FN) were detected with real-time quantitative PCR (RT-qPCR) and Western blotting.The interaction of FUT8 and Gal-3 was tested using coimmunoprecipitation (Co-IP) assay, and the effect of FUT8 silencing on Gal-3 and FAK/Akt signaling pathways was analyzed.

RESULTS

FUT8 knockdown significantly reduced bleomycin-induced extracellular collagen deposition in the lung tissues of the mice.Silencing FUT8 obviously inhibited cell proliferation (P < 0.05) and migration mediated by TGF-β1.FUT8 knockdown down-regulated the mRNA and protein levels of α-SMA, COLIA1 and FN (P < 0.05) in the cells.Coimmunoprecipitation analysis showed that FUT8 interacted with Gal-3.Silencing FUT8 significantly down-regulated Gal-3 expression and inhibited the activation of the FAK/Akt signaling pathway (P < 0.05).Overexpression of Gal-3 obviously reversed the effects of FUT8 silencing on cell proliferation, migration and fibrosis (P < 0.05).

CONCLUSION

FUT8 regulates TGF-β1-induced proliferation, migration and fibrosis of MRC-5 cells by modulating Gal-3 expression, in which the FAK/Akt pathway may play a role.

A multi-layer functional genomic analysis to understand noncoding genetic variation in lipids

A major challenge of genome-wide association studies (GWASs) is to translate phenotypic associations into biological insights. Here, we integrate a large GWAS on blood lipids involving 1.6 million individuals from five ancestries with a wide array of functional genomic datasets to discover regulatory mechanisms underlying lipid associations. We first prioritize lipid-associated genes with expression quantitative trait locus (eQTL) colocalizations and then add chromatin interaction data to narrow the search for functional genes. Polygenic enrichment analysis across 697 annotations from a host of tissues and cell types confirms the central role of the liver in lipid levels and highlights the selective enrichment of adipose-specific chromatin marks in high-density lipoprotein cholesterol and triglycerides. Overlapping transcription factor (TF) binding sites with lipid-associated loci identifies TFs relevant in lipid biology. In addition, we present an integrative framework to prioritize causal variants at GWAS loci, producing a comprehensive list of candidate causal genes and variants with multiple layers of functional evidence. We highlight two of the prioritized genes, CREBRF and RRBP1, which show convergent evidence across functional datasets supporting their roles in lipid biology.

Prediction of spontaneous preterm birth and preterm prelabor rupture of membranes using maternal factors, obstetric history and biomarkers of placental function at 11-13 weeks

OBJECTIVES

To determine whether first-trimester biomarkers of placental function can be used to screen for spontaneous preterm birth (sPTB), and to develop prediction models using maternal factors, obstetric history and biomarkers of placental function at 11-13 weeks for the calculation of patient-specific risk for sPTB.

METHODS

This was a retrospective secondary analysis of data derived from a prospective cohort study on first-trimester screening for pre-eclampsia in singleton pregnancies attending for routine Down syndrome screening at 11 + 0 to 13 + 6 weeks' gestation at a tertiary obstetric unit between December 2016 and September 2019. A split-sample internal validation method was used to explore and develop prediction models for all sPTB at < 37 weeks and for PTB at < 37 weeks after preterm prelabor rupture of membranes (PPROM) using maternal risk factors, uterine artery Doppler indices, serum placental growth factor (PlGF), pregnancy-associated plasma protein-A (PAPP-A) and β-human chorionic gonadotropin (β-hCG). Screening performance was assessed using receiver-operating-characteristics (ROC)-curve analysis, with calculation of the areas under the ROC curves (AUCs).

RESULTS

A total of 9298 singleton pregnancies were included in this study. sPTB at < 37 weeks occurred in 362 (3.89%) cases, including 231 (2.48%) cases of PPROM. sPTB at < 34 weeks occurred in 87 (0.94%) cases, including 39 (0.42%) cases of PPROM. Identified maternal risk factors for sPTB at < 37 weeks included chronic hypertension, conception using in-vitro fertilization and history of PTB. Maternal risk factors for PPROM at < 37 weeks included conception using in-vitro fertilization and history of PTB. Median PlGF multiples of the median (MoM) and PAPP-A MoM were significantly reduced in women with sPTB at < 37 weeks, as well as in those who had PPROM, compared to those who delivered at term. Screening by a combination of maternal risk factors, PAPP-A and PlGF achieved better performance in predicting sPTB at < 37 weeks (AUC, 0.630 vs 0.555; detection rate (DR), 24.8% vs 16.6% at a false-positive rate (FPR) of 10%; P ≤ 0.0001) and PPROM at < 37 weeks (AUC, 0.643 vs 0.558; DR, 28.1% vs 17.0% at a FPR of 10%; P ≤ 0.0001) than using maternal risk factors alone. Both models were successfully applied to the internal validation dataset, with AUCs of 0.628 and 0.650, respectively.

CONCLUSIONS

We demonstrated that low levels of maternal serum PAPP-A and PlGF in the first trimester are associated with increased risks of sPTB and PPROM at < 37 weeks. However, further research is needed to identify additional biomarkers to improve the screening performance of the combined model that includes maternal risk factors, PAPP-A and PlGF before clinical application. © 2022 International Society of Ultrasound in Obstetrics and Gynecology.

Race, Genotype, and Azathioprine Discontinuation : A Cohort Study

BACKGROUND

Thiopurines are an important class of immunosuppressants despite their risk for hematopoietic toxicity and narrow therapeutic indices. Benign neutropenia related to an ACKR1 variant (rs2814778-CC) is common among persons of African ancestries.

OBJECTIVE

To test whether rs2814778-CC was associated with azathioprine discontinuation attributed to hematopoietic toxicity and lower thiopurine dosing.

DESIGN

Retrospective cohort study.

SETTING

Two tertiary care centers.

PATIENTS

Thiopurine users with White or Black race.

MEASUREMENTS

Azathioprine discontinuation attributed to hematopoietic toxicity. Secondary outcomes included weight-adjusted final dose, leukocyte count, and change in leukocyte count.

RESULTS

The rate of azathioprine discontinuation attributed to hematopoietic toxicity was 3.92 per 100 person-years among patients with the CC genotype (n = 101) and 1.34 per 100 person-years among those with the TT or TC genotype (n = 1365) (hazard ratio [HR] from competing-risk model, 2.92 [95% CI, 1.57 to 5.41]). The risk remained significant after adjustment for race (HR, 2.61 [CI, 1.01 to 6.71]). The risk associated with race alone (HR, 2.13 [CI, 1.21 to 3.75]) was abrogated by adjustment for genotype (HR, 1.13 [CI, 0.48 to 2.69]). Lower last leukocyte count and lower dosing were significant among patients with the CC genotype. Lower dosing was validated in an external cohort of 94 children of African ancestries prescribed the thiopurine 6-mercaptopurine (6-MP) for acute lymphoblastic leukemia. The CC genotype was independently associated with lower 6-MP dose intensity relative to the target daily dose of 75 mg/m2 (median, 0.83 [IQR, 0.70 to 0.94] for the CC genotype vs. 0.94 [IQR, 0.72 to 1.13] for the TT or TC genotype; P = 0.013).

LIMITATIONS

Unmeasured confounding; data limited to tertiary centers.

CONCLUSION

Patients with the CC genotype had higher risk for azathioprine discontinuation attributed to hematopoietic toxicity and lower thiopurine doses. Genotype was associated with those risks, even after adjustment for race.

PRIMARY FUNDING SOURCE

National Institutes of Health.

Phenome-Wide Association Study of UMOD Gene Variants and Differential Associations With Clinical Outcomes Across Populations in the Million Veteran Program a Multiethnic Biobank

Introduction

Common variants in the UMOD gene are considered an evolutionary adaptation against urinary tract infections (UTIs) and have been implicated in kidney stone formation, chronic kidney disease (CKD), and hypertension. However, differences in UMOD variant-phenotype associations across population groups are unclear.

Methods

We tested associations between UMOD/PDILT variants and up to 1528 clinical diagnosis codes mapped to phenotype groups in the Million Veteran Program (MVP), using published phenome-wide association study (PheWAS) methodology. Associations were tested using logistic regression adjusted for age, sex, and 10 principal components of ancestry. Bonferroni correction for multiple comparisons was applied.

Results

Among 648,593 veterans, mean (SD) age was 62 (14) years; 9% were female, 19% Black, and 8% Hispanic. In White patients, the rs4293393 UMOD risk variant associated with increased uromodulin was associated with increased odds of CKD (odds ratio [OR]: 1.22, 95% CI: 1.20-1.24, P = 5.90 × 10-111), end-stage kidney disease (OR: 1.17, 95% CI: 1.11-1.24, P = 2.40 × 10-09), and hypertension (OR: 1.03, 95% CI: 1.05-1.05, P = 2.11 × 10-06) and significantly lower odds of UTIs (OR: 0.94, 95% CI: 0.92-0.96, P = 1.21 × 10-10) and kidney calculus (OR: 0.85, 95% CI: 0.83-0.86, P = 4.27 × 10-69). Similar findings were observed across UMOD/PDILT variants. The rs77924615 PDILT variant had stronger associations with acute cystitis in White female (OR: 0.73, 95% CI: 0.59-0.91, P = 4.98 × 10-03) versus male (OR: 0.99, 95% CI: 0.89-1.11, P = 8.80 × 10-01) (P interaction = 0.01) patients. In Black patients, the rs77924615 PDILT variant was significantly associated with pyelonephritis (OR: 0.65, 95% CI: 0.54-0.79, P = 1.05 × 10-05), whereas associations with UMOD promoter variants were attenuated.

Conclusion

Robust associations were observed between UMOD/PDILT variants linked with increased uromodulin expression and lower odds of UTIs and calculus and increased odds of CKD and hypertension. However, these associations varied significantly across ancestry groups and sex.

Ionization states for the multipetawatt laser-QED regime

A paradigm shift in the physics of laser-plasma interactions is approaching with the commissioning of multipetawatt laser facilities worldwide. Radiation reaction processes will result in the onset of electron-positron pair cascades and, with that, the absorption and partitioning of the incident laser energy, as well as the energy transport throughout the irradiated targets. To accurately quantify these effects, one must know the focused intensity on target in situ. In this work, a way of measuring the focused intensity on target is proposed based upon the ionization of xenon gas at low ambient pressure. The field ionization rates from two works [Phys. Rev. A 59, 569 (1999)1050-294710.1103/PhysRevA.59.569 and Phys. Rev. A 98, 043407 (2018)2469-992610.1103/PhysRevA.98.043407], where the latter rate has been derived using quantum mechanics, have been implemented in the particle-in-cell code SMILEI [Comput. Phys. Commun. 222, 351 (2018)0010-465510.1016/j.cpc.2017.09.024]. A series of one- and two-dimensional simulations are compared and shown to reproduce the charge states without presenting visible differences when increasing the simulation dimensionality. They provide a way to accurately verify the intensity on target using in situ measurements.

[Study on the dynamic changes of oral microbiota in type 2 diabetes patients with periodontitis after initial periodontal therapy]

Objectives: To clarify the effect of initial periodontal therapy on the dynamic changes of oral (saliva, dorsal tongue and subgingival plaque) microbiota in periodontitis patients with or without type 2 diabetes mellitus (T2DM). Methods: A total of 14 patients with chronic periodontitis (CP group) and 14 CP patients with T2DM (CP-T2DM group) were included from Department of Periodontology, School and Hospital of Stomatology,Cheeloo College of Medicine, Shandong University. The microbial samples were collected from saliva, dorsal tongue and subgingival plaque of first molars at baseline, 1.5 and 3 months after initial periodontal therapy, and were detected by 16S rRNA (V3-V4 region) gene sequencing. The sequencing data were analyzed to obtain microbial distribution and community structure information. The same professional periodontist evaluated the periodontal status of patients according to periodontitis detection indices before and after initial periodontal therapy. Meanwhile, patients' blood samples were collected and related metabolic indices were evaluated. Results: After initial periodontal therapy, the glycosylated hemoglobin levels [(7.46±1.69)%] in CP-T2DM group were significantly improved than that at baseline [(7.65±1.34)%] (t=0.52,P=0.610). The probing depth of the sampling sites [CP group: (2.94±0.46) mm, CP-T2DM group: (2.95±0.35) mm] and bleeding index (CP group: 1.91±0.42, CP-T2DM group: 1.67±0.49) at 3 months after treatment were significantly decreased than the probing depth [CP group: (3.99±0.77) mm, CP-T2DM group: (3.80±0.76) mm] (F=25.61, P<0.001; F=17.63, P<0.001) and bleeding index (CP group: 3.03±0.52, CP-T2DM group: 2.54±0.65) (F=28.43, P<0.001; F=20.21, P<0.001) at baseline. The flora analysis showed that the α and β diversity indices of the same sites in the CP and CP-T2DM groups did not change significantly before and after the initial therapy, but the bacterial abundance at each site changed. There were commonalities and differences in the microbial composition of each site in the CP and CP-T2DM groups. Among them, the relative abundance of Proteobacteria in saliva and dorsal tongue samples of the two groups after treatment was basically consistent with the change trend in the subgingival plaque microbes. In the subgingival plaque of the CP group, the relative abundance of Proteobacteria showed a gradual increase with the prolongation of initial periodontal therapy; while in the CP-T2DM group, it showed a trend of first increase and then decrease. Syntrophy, Dethiosulfate, Methanobacteriaceae and TG5 in CP and CP-T2DM groups were all significantly dominant bacteria in subgingival plaque at baseline (P<0.05). Moreover, in the CP-T2DM group Spirochetes also showed a significant advantage. At 1.5 months after treatment, Rhizobacteria, Alcaligenes, Comamomons, Delftia, Blautella, etc. were dominant in subgingival plaque (P<0.05). Firmicutes, Clostridia/Clostridiales, Enterococci and Ruminococci showed significant differences at 3 months (P<0.05). Conclusions: Plaques in saliva and tongue dorsal could reflect the effects of initial periodontal therapy on the dynamic changes of microorganisms to a certain extent. CP and CP-T2DM patients had differences in microbial composition and responses to initial periodontal therapy.

POS0393 ASSOCIATION BETWEEN GENETICALLY PREDICTED EXPRESSION OF TPMT AND AZATHIOPRINE ADVERSE EVENTS IN PATIENTS WITH INFLAMMATORY CONDITIONS

Background

Azathioprine is a widely used immunosuppressant for the treatment of inflammatory conditions such as systemic lupus erythematosus (SLE), systemic vasculitis, dermatomyositis, and inflammatory bowel disease. However, its use is often limited by myelotoxicity. Variants in the gene encoding thiopurine-S-methyltransferase (TPMT), an enzyme in the metabolic pathway of azathioprine, increase the risk for myelotoxicity1. We know little about the relationship between the genetically predicted expression of TPMT and side effects of azathioprine.

Objectives

To examine whether genetically predicted expression of TPMT in liver tissue is associated with azathioprine adverse effects.

Methods

We assembled a retrospective cohort of new users taking azathioprine for inflammatory conditions at a tertiary care center. We performed genotyping with Illumina Infinium Expanded Multi-Ethnic Genotyping Array plus custom content data, and we then used Michigan Imputation servers for genetic imputation and PrediXcan models trained with GTEx/Genotype-Tissue Expression Project version 8 data to impute TPMT expression in liver tissue. We prespecified nine groups of phecodes (comprised of ICD9 and ICD10 codes) corresponding to known adverse effects of azathioprine. We then tested the association between the predicted expression of TPMT and these adverse events; for outcomes significant in the Wilcoxon ranksum tests (p<0.05), each case was reviewed in clinical records for confirmation. Finally, we grouped the predicted expression of TPMT in liver tissue into tertiles and conducted logistic regressions to assess the associations between predicted expression and side effects. We conducted a sensitivity analysis restricted to patients with EHR-reported White race.

Results

The cohort included 1034 patients (Table 1). Phecodes for 3 side effects—leukopenia (n=29), skin cancer (n=13), and rash (n=52)—were identified as associated with predicted TPMT expression in liver tissue. Of these, cases of side effects attributed to azathioprine were validated by chart review: leukopenia (96.6%; n=28), skin cancer (92.3%; n=12), and rash (9.6%; n=5) and used for analysis. When assessed by tertile of predicted TPMT expression, patients in the highest tertile had lower odds of having leukopenia (OR=0.35, 95%CI: 0.12-0.98, p=0.045) and a trend towards higher odds for skin cancer, but the number of cases was small (OR=3.56, 95%CI: 0.73-17.27, p=0.115). Confirmed cases of rash attributed to azathioprine were too few for meaningful analysis. We found similar results when restricted to patients with reported White race.

Table 1.

Characteristics of patients by TPMT expression

Lowest TertileMiddle TertileHighest TertileN=345N=345N=344Female sex, n (%)228 (66.1)244 (70.7)238 (69.2)EHR-reported White race, n (%)306 (88.7)293 (84.9)290 (84.3)Age, median [IQR]42 [29-58]43 [30-55]46 [30-56]Indication, n (%)Systemic lupus erythematosus38 (11.0)42 (12.2)38 (11.0)Inflammatory bowel disease191 (55.3)185 (53.6)190 (55.2)Other connective tissue disorder/autoimmune92 (26.7)96 (27.8)100 (29.1)Other24 (7.0)22 (6.4)16 (4.7)Verified leukopenia attributed to azathioprine, n (%)14 (4.1)9 (2.6)5 (1.5)Verified skin cancer attributed to azathioprine, n (%)2 (0.6)3 (0.9)7 (2.0)Verified rash attributed to azathioprine, n (%)2 (0.6)3 (0.9)0 (0)

Conclusion

This analysis suggests that PrediXcan may be useful for examining the association between gene expression and side effects of medications. Moreover, this approach successfully identified leukopenia as a side effect associated with predicted TPMT expression.

Acknowledgements

None to declare.

Disclosure of Interests

None declared.

POS1444 FLT1 AND EPHB2 ARE NOVEL GENETIC MARKERS ASSOCIATED WITH PANCREATITIS IN PATIENTS TAKING AZATHIOPRINE FOR IMMUNE-MEDIATED CONDITIONS: INTEGRATING GENOME- AND TRANSCRIPTOME-WIDE ASSOCIATION STUDIES

Background

Azathioprine (AZA) is a thiopurine immunosuppressant medication used to treat a variety of immune-mediated diseases. Unfortunately, its use is limited by adverse effects. Pancreatitis, a potentially severe, life-threatening side effect is independent of dose and necessitates AZA discontinuation given the high risk of recurrent pancreatitis with continued use or re-challenge. The mechanisms driving pancreatitis are unclear. While classic thiopurine-induced acute pancreatitis (TIAP) has been associated with HLA haplotypes, most patients taking AZA and presenting with pancreatitis do not fulfill the stringent criteria for TIAP.

Objectives

To identify genetic risk factors for pancreatitis in patients taking azathioprine for immune-mediated conditions.

Methods

Using a biobank linked to electronic health records (EHR) from a tertiary center, we identified new users of AZA. Patients were excluded if the primary indication for AZA was organ transplant or if there was a history of pancreatitis prior to AZA use. The analysis was restricted to patients with EHR-reported race as White due to insufficient case counts for the non-White group. We then identified patients with amylase or lipase values that exceeded twice the upper limit of normal (“>2x ULN”) or with ICD-9/ICD-10 codes for acute pancreatitis. Each record was manually reviewed to confirm the timing of AZA use in relation to laboratory derangements or ICD coding, as well as to further classify patients into three increasingly strict, but not exclusive categories: 1) pancreatic injury (amylase or lipase >2x ULN); 2) acute pancreatitis1, or 3) TIAP2. We completed genotyping with Illumina Infinium Expanded Multi-Ethnic Genotyping Array plus custom content data, employed Michigan Imputation servers for genetic imputation, and used PrediXcan (GTEx v8) to impute gene expression. We then conducted genome-wide association and transcriptome-wide association studies (GWAS, TWAS). Acknowledging the relatively small overall cohort, and possible imbalance of cases vs controls, we used the Firth logistic regression method, which is a penalized likelihood-based method.

Results

We studied 2127 AZA users (35.4% male; mean 44.5+/-17.2 years). The median AZA dose was 100mg/day (IQR: 50-125mg/day). Rheumatologic conditions (56.9%) and inflammatory bowel disease (40.4%) comprised the most common primary indications for AZA. Pancreatic injury, pancreatitis, and TIAP were diagnosed in 42 (2.0%), 16 (0.8%), and 9 (0.4%) patients, respectively. GWAS identified several significantly associated genes, many with overlapping TWAS findings in the pancreas and liver (Figure 1). From these, the two protein-encoding genes Fms Related Receptor Tyrosine Kinase-1 (FLT1) and Ephrin type-B receptor-2 (EPHB2) overlapped in two or more pancreatitis phenotypes in the TWAS and GWAS, respectively. EPHB2 was associated with a 8.6-fold (P=1.84 x 10-8) and a 31.4-fold (P=2.87x 10-8) higher likelihood of pancreatic injury and TIAP, respectively.

Figure 1.

Conclusion

FLT1—a gene that encodes a receptor tyrosine kinase and is a member of the vascular endothelial growth factor receptor (VEGFR) family—and EPHB2—a gene that encodes a member of the Eph receptor family, which is the largest subgroup of the receptor tyrosine kinase family—are novel genetic markers associated with pancreatitis in patients taking AZA. VEGF can potentiate inflammation and the pancreas microenvironment is known to promote VEGF expression, which has been linked to pancreatic cancer development; anti-VEGF treatments have been investigated both for mitigating inflammation and also anti-pancreatic cancer treatment. Future studies validating our findings in AZA-induced pancreatitis are warranted.

References

[1]Crockett et al. Gastroenterology (2018). 154(4):1096-1101.

[2]Heap et al. Nature Genetics (2014). 46:1131-1134

Disclosure of Interests

Shailja Shah Consultant of: ad hoc consultant for Phathom pharmaceuticals, Tyler Reese: None declared, Jacy Zanussi: None declared, Alyson Dickson: None declared, Laura Daniel: None declared, Ran Tao: None declared, Tyne Miller-Fleming: None declared, Peter Straub: None declared, Adriana Hung: None declared, Puran Nepal: None declared, Wei-Qi Wei: None declared, Elizabeth Phillips: None declared, Nancy Cox: None declared, Charles M. Stein: None declared, QiPeng Feng: None declared, Cecilia P. Chung: None declared

Predicted expression of genes involved in the thiopurine metabolic pathway and azathioprine discontinuation due to myelotoxicity

TPMT and NUDT15 variants explain less than 25% of azathioprine-associated myelotoxicity. There are 25 additional genes in the thiopurine pathway that could also contribute to azathioprine myelotoxicity. We hypothesized that among TPMT and NUDT15 normal metabolizers, a score combining the genetically predicted expression of other proteins in the thiopurine pathway would be associated with a higher risk for azathioprine discontinuation due to myelotoxicity. We conducted a retrospective cohort study of new users of azathioprine who were normal TPMT and NUDT15 metabolizers. In 1201 White patients receiving azathioprine for an inflammatory disease, we used relaxed Least Absolute Shrinkage and Selection Operator (LASSO) regression to select genes that built a score for discontinuing azathioprine due to myelotoxicity. The score incorporated the predicted expression of AOX1 and NME1. Patients in the highest score tertile had a higher risk of discontinuing azathioprine compared to those in the lowest tertile (hazard ratio [HR] = 2.15, 95% confidence interval [CI] = 1.11-4.19, p = 0.024). Results remained significant after adjusting for a propensity score, including sex, tertile of calendar year at initial dose, initial dose, age at baseline, indication, prior TPMT testing, and the first 10 principal components of the genetic data (HR = 2.11, 95% CI = 1.08-4.13, p = 0.030). We validated the results in a cohort (N = 517 non-White patients and those receiving azathioprine to prevent transplant rejection) that included all other patients receiving azathioprine (HR = 2.00, (95% CI = 1.09-3.65, p = 0.024). In conclusion, among patients who were TPMT and NUDT15 normal metabolizers, a score combining the predicted expression of AOX1 and NME1 was associated with an increased risk for discontinuing azathioprine due to myelotoxicity.

Identifying Potential Therapeutic Applications and Diagnostic Harms of Increased Bilirubin Concentrations: A Clinical and Genetic Approach

Bilirubin has antioxidant and anti-inflammatory properties in vitro and in animal studies and protects against inflammatory, cardiovascular, and other diseases in observational studies; therefore, bilirubin has potential as a therapeutic agent. However, observational studies could be confounded by many factors. We used a genetic (n = 61,281) and clinical (n = 234,670) approach to define the association between bilirubin and 19 conditions with a putative protective signal in observational studies. We also tested if individuals with genetically higher bilirubin levels underwent more diagnostic tests. We used a common variant in UGT1A1 (rs6742078) associated with an 26% increase in bilirubin levels in the genetic studies. Carriers of the variant had higher bilirubin levels (P = 2.2 × 10-16 ) but there was no significant association with any of the 19 conditions. In a phenome-wide association study (pheWAS) to seek undiscovered genetic associations, the only significant finding was increased risk of "jaundice-not of newborn." Carriers of the variant allele were more likely to undergo an abdominal ultrasound (odds ratio = 1.04, [1.00-1.08], P = 0.03). In contrast, clinically measured bilirubin levels were significantly associated with 15 of the 19 conditions (P < 0.003) and with 431 clinical diagnoses in the pheWAS (P < 1 × 10-5 adjusted for sex, age, and follow-up). With additional adjustment for smoking and body mass index, 7 of 19 conditions and 260 pheWAS diagnoses remained significantly associated with bilirubin levels. In conclusion, bilirubin does not protect against inflammatory or other diseases using a genetic approach; the many putative beneficial associations reported clinically are likely due to confounding.

Integrating gene expression and clinical data to identify drug repurposing candidates for hyperlipidemia and hypertension

Discovering novel uses for existing drugs, through drug repurposing, can reduce the time, costs, and risk of failure associated with new drug development. However, prioritizing drug repurposing candidates for downstream studies remains challenging. Here, we present a high-throughput approach to identify and validate drug repurposing candidates. This approach integrates human gene expression, drug perturbation, and clinical data from publicly available resources. We apply this approach to find drug repurposing candidates for two diseases, hyperlipidemia and hypertension. We screen >21,000 compounds and replicate ten approved drugs. We also identify 25 (seven for hyperlipidemia, eighteen for hypertension) drugs approved for other indications with therapeutic effects on clinically relevant biomarkers. For five of these drugs, the therapeutic effects are replicated in the All of Us Research Program database. We anticipate our approach will enable researchers to integrate multiple publicly available datasets to identify high priority drug repurposing opportunities for human diseases.

TPMT and NUDT15 Variants Predict Discontinuation of Azathioprine for Myelotoxicity in Patients with Inflammatory Disease: Real-World Clinical Results

Azathioprine is used frequently to treat several inflammatory conditions. However, treatment is limited by adverse events-in particular, myelotoxicity. Thiopurine-S-methyltransferase (TPMT) and nudix hydrolase-15 (NUDT15) are enzymes involved in azathioprine metabolism; variants in the genes encoding these enzymes increase the risk for azathioprine myelotoxicity. The Clinical Pharmacogenetics Implementation Consortium (CPIC) has recommended dose adjustments based on the results of TPMT and NUDT15 genotyping. However, little is known about the importance of this genetic information in routine clinical care. We hypothesized that in patients with inflammatory diseases, TPMT and NUDT15 genotype data predict the risk of discontinuing azathioprine due to myelotoxicity. This was a retrospective cohort study in 1,403 new adult azathioprine users for the management of inflammatory conditions for whom we had genetic information and clinical data. Among patients who discontinued azathioprine, we adjudicated the reason(s). Genotyping was performed using the Illumina Infinium Expanded Multi-Ethnic Genotyping Array plus custom content. We used CPIC guidelines to determine TPMT and NUDT15 metabolizer status; patients were grouped as either: (i) poor/intermediate, or (ii) normal/indeterminate metabolizers. We classified 110 patients as poor/intermediate, and 1,293 patients as normal/indeterminate metabolizers. Poor/intermediate status was associated with a higher risk for azathioprine discontinuation due to myelotoxicity compared to normal/indeterminate metabolizers (hazard ratio (HR) = 2.90, 95% confidence interval (CI): 1.58-5.31, P = 0.001). This association remained significant after adjustment for race, age at initiation, sex, primary indication, and initial daily dose of azathioprine (adjusted HR (aHR) = 2.67, 95% CI: 1.44-4.94, P = 0.002). In conclusion, TPMT and NUDT15 metabolizer status predicts discontinuation due to myelotoxicity for patients taking azathioprine for inflammatory conditions.

The power of genetic diversity in genome-wide association studies of lipids

Increased blood lipid levels are heritable risk factors of cardiovascular disease with varied prevalence worldwide owing to different dietary patterns and medication use1. Despite advances in prevention and treatment, in particular through reducing low-density lipoprotein cholesterol levels2, heart disease remains the leading cause of death worldwide3. Genome-wideassociation studies (GWAS) of blood lipid levels have led to important biological and clinical insights, as well as new drug targets, for cardiovascular disease. However, most previous GWAS4-23 have been conducted in European ancestry populations and may have missed genetic variants that contribute to lipid-level variation in other ancestry groups. These include differences in allele frequencies, effect sizes and linkage-disequilibrium patterns24. Here we conduct a multi-ancestry, genome-wide genetic discovery meta-analysis of lipid levels in approximately 1.65 million individuals, including 350,000 of non-European ancestries. We quantify the gain in studying non-European ancestries and provide evidence to support the expansion of recruitment of additional ancestries, even with relatively small sample sizes. We find that increasing diversity rather than studying additional individuals of European ancestry results in substantial improvements in fine-mapping functional variants and portability of polygenic prediction (evaluated in approximately 295,000 individuals from 7 ancestry groupings). Modest gains in the number of discovered loci and ancestry-specific variants were also achieved. As GWAS expand emphasis beyond the identification of genes and fundamental biology towards the use of genetic variants for preventive and precision medicine25, we anticipate that increased diversity of participants will lead to more accurate and equitable26 application of polygenic scores in clinical practice.

Methods for extremely sparse-angle proton tomography

Proton radiography is a widely fielded diagnostic used to measure magnetic structures in plasma. The deflection of protons with multi-MeV kinetic energy by the magnetic fields is used to infer their path-integrated field strength. Here the use of tomographic methods is proposed for the first time to lift the degeneracy inherent in these path-integrated measurements, allowing full reconstruction of spatially resolved magnetic field structures in three dimensions. Two techniques are proposed which improve the performance of tomographic reconstruction algorithms in cases with severely limited numbers of available probe beams, as is the case in laser-plasma interaction experiments where the probes are created by short, high-power laser pulse irradiation of secondary foil targets. A new configuration allowing production of more proton beams from a single short laser pulse is also presented and proposed for use in tandem with these analytical advancements.

Genome-wide association study of susceptibility to hospitalised respiratory infections

Background: Globally, respiratory infections contribute to significant morbidity and mortality. However, genetic determinants of respiratory infections are understudied and remain poorly understood. Methods: We conducted a genome-wide association study in 19,459 hospitalised respiratory infection cases and 101,438 controls from UK Biobank. We followed-up well-imputed top signals from the UK Biobank discovery analysis in 50,912 respiratory infection cases and 150,442 controls from 11 cohorts. We aggregated effect estimates across studies using inverse variance-weighted meta-analyses. Additionally, we investigated the function of the top signals in order to gain understanding of the underlying biological mechanisms. Results: In the discovery analysis, we report 56 signals at P<5×10-6, one of which was genome-wide significant (P<5×10-8). The genome-wide significant signal was in an intron of PBX3, a gene that encodes pre-B-cell leukaemia transcription factor 3, a homeodomain-containing transcription factor. Further, the genome-wide significant signal was found to colocalise with gene-specific expression quantitative trait loci (eQTLs) affecting expression of PBX3 in lung tissue, where the respiratory infection risk alleles were associated with decreased PBX3 expression in lung tissue, highlighting a possible biological mechanism. Of the 56 signals, 40 were well-imputed in UK Biobank and were investigated in the 11 follow-up cohorts. None of the 40 signals replicated, with effect estimates attenuated. Conclusions: Our discovery analysis implicated PBX3 as a candidate causal gene and suggests a possible role of transcription factor binding activity in respiratory infection susceptibility. However, the PBX3 signal, and the other well-imputed signals, did not replicate when aggregating effect estimates across 11 independent cohorts. Significant phenotypic heterogeneity and differences in study ascertainment may have contributed to this lack of statistical replication. Overall, our study highlighted putative associations and possible biological mechanisms that may provide insight into respiratory infection susceptibility.