publications

2026

1. BMJGH

   Understanding Long COVID (LC) in Frontline and Essential Workers (FEWs) through Proteomic Profiling: A Longitudinal Cohort Study

   Tuo Liu, Danielle Stea, Shawn C Beitel, and 3 more authors

   in prep, 2026

   Abs DOI

   Background: Long COVID (LC) can debilitate due to lingering symptoms. However, there is incomplete information on the mechanistic understanding of LC, and if proteomic profiles of individuals can predict long-COVID status. Objective: To examine differences in proteomic profiles and cellular activities between frontline and essential workers (FEWs) who developed LC and who recovered at baseline, during COVID-19 convalescence, and at follow-up, to understand LC development in FEWs. Method: Frontline and essential workers (FEWs) in the HEROES-RECOVER cohorts were tested weekly for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection from 2020-2022. Participants with long-COVID (N = 57) (symptoms persisting for at one month) and participants with COVID-19 without LC (N = 80) were selected for untargeted proteomic profiling using the SOMAscan Assay at baseline (before COVID-19 infection), convalescence (within 60 days of COVID-19 infection), and follow-up (60 days after infection onset). We fitted a logistic regression model for each SOMAmer to extract crude risk of LC status at follow-up, at a false discovery rate (FDR) q <0.05 level. random forest model to identify important SOMAmers at each timepoint Functional enrichment analysis was applied at each timepoint to identify pathways enriched among those with LC compared to regular COVID participants, by FDR q-value p-<0.05. Results: We profiled a total of 7,659 serum SOMAmers with SOMAScan in 137 FEWs at 3 study timepoints, separately. A total of 116 SOMAmers were differentially expressed between FEWs who ultimately developed LC (N=57) and regular COVID (N=80) at adjusted p-value 0.05 level. Of these, 20 differential SOMAmers were unique to baseline, 14 were unique to convalescence and 17 were unique to follow-up. Functional enrichment analysis revealed sustained immune activities and inflammatory profiles across the entire study (FDR q<0.05), and potential contribution from occupational hazards to the sustained enrichments. Conclusion: Our findings in FEWs suggest that potential interplay of pre-existing immune variability, acute immune activation during SARS-CoV-2 infection, and persistent inflammatory signaling throughout the recovery phase may collectively contribute to the LC development in this population.

2. PLoS ONE

   Unveiling Post-Vaccination Proteomic Signatures in Infection-Naïve Individuals Associated with Omicron Breakthrough Infections

   Yiwen Liu, Eric Lu, Katherine D. Ellingson, and 6 more authors

   PLoS ONE, submitted, 2026

   Abs DOI

   Background: Given the persistence of the SARS-CoV-2 virus, it is important to understand the proteome associated with breakthrough infections among fully COVID-19 vaccinated individuals. Methods: We utilized a nested case-control study within the frontline worker HEROES-RECOVER cohorts, comparing serum proteomic profiles from SARS-CoV-2 infection-naïve participants following a third dose of COVID-19 origin strain WA-1 monovalent mRNA vaccine and subsequently experiencing Omicron breakthrough infections with the proteomic profiles of matched controls without infections. Our study leveraged proteomics data generated from the SomaScan Platform and adopted a robust feature selection method, elastic net regularized conditional logistic regression together with bootstrapping technique, to identify key proteins. Enrichment analyses were performed to investigate biological pathways.Results: We identified 28 significantly different proteins out of over 7,000 candidate proteins. Key findings included downregulated chemokines (CXCL2, CXCL3, CCL19, CCL23) and elevated cytokine IL-7 levels in breakthrough cases, with pathway analysis revealing enrichment in chemokine signaling and cytokine-cytokine interaction pathways. Other key proteins, such as LGALS1, HAVCR2, and SELE were upregulated in breakthrough cases. Discussion: These results reveal potential immune response mechanisms in breakthrough infections, characterized by viral immune evasion and compensatory T-cell regeneration. The identified biomarkers may provide valuable insights for future predictive profiles and therapeutic strategies.

3. TBD

   Extreme Heat Exposure During Pregnancy and Birth Outcomes: A Review of Epidemiological Evidence and Policy Implications

   Rebecca Ramos, Paulina L. Ramirez, Rietta S. Wagoner, and 7 more authors

   in prep, 2026

   Abs DOI

   Extreme heat events are on the rise, raising concerns about their impact on public health. Pregnant individuals are particularly vulnerable to heat exposure, which has been linked to adverse birth outcomes such as low birth weight, preterm birth, and stillbirth. This literature review synthesizes current epidemiological evidence on the relationship between maternal heat exposure and low birth weight. A comprehensive literature search was conducted using PubMed, Scopus, Web of Science, and Embase. Peer-reviewed epidemiological studies published in English that assessed maternal exposure to extreme heat and associations with birth outcomes were included. Studies were evaluated for heat exposure characterization, birth outcome measures, statistical modeling approaches, and potential confounders. The review identified 15 studies from various geographical regions, with cohort and case-control study designs being the most common. Despite differences in heat exposure definitions (absolute temperature thresholds, percentiles, or heat waves), 13 studies reported a significant association between high temperatures and lower birth weights, mainly when exposure occurred during critical gestational windows. Some studies found no association, potentially highlighting the need for standardized exposure assessment methods. The evidence suggests that maternal heat exposure increases the risk of low birth weight, emphasizing the need for targeted public health interventions. Policy recommendations include integrating heat risk education into prenatal care, strengthening workplace protections for pregnant individuals in high-heat environments, and expanding access to cooling resources in vulnerable communities. Further research is needed to standardize heat exposure metrics and explore the long-term health effects of prenatal heat exposure on offspring.

2025

1. Metabolomics

   Differential metabolic profiles by Hispanic ethnicity among male Tucson firefighters

   Tuo Liu, Melissa A. Furlong, Justin M Snider, and 10 more authors

   Metabolomics, 2025

   Abs DOI

   Introduction Firefighters face regular exposure to known and probable human carcinogens, such as polycyclic aromatic hydrocarbons (PAHs), benzene, and formaldehyde, leading to an increased risk of various cancers compared to the general population. Hispanic and black firefighters are at increased risk of additional cancers not elevated in non-Hispanic white firefighters, yet biological pathways underlying these differences are unknown. Objectives The study objectives were to evaluate differences in the urinary metabolome between Hispanic and non-Hispanic firefighters, pre-and post-fireground exposure. Methods To investigate the metabolic patterns, we employed a comprehensive metabolomics pipeline that leveraged liquid chromatography coupled with high-resolution mass spectrometry. We applied linear mixed effects regression to identify the differential metabolites at an FDR < 0.05 among 19 Hispanic and 81 non-Hispanic firefighters. We also performed overrepresentation analysis using Mummichog to identify enriched pathways at FDR < 0.05. Results Out of 175 features in HILIC(−) mode and 1847 features in RP(+) mode, we found 26 and 276 differential urinary features, respectively, when comparing Hispanic and non-Hispanic firefighters. We noted pathway enrichment in tryptophan and galactose metabolism. However, post-exposure, we did not observe differences in the metabolomic response by ethnicity despite differing fireground exposures. Conclusion Dysregulation in the tryptophan and galactose pathway is an important contributor to cancer risks and may explain the increased cancer risk among Hispanic firefighters. Supplementary Information The online version contains supplementary material available at 10.1007/s11306-024-02198-9.

2. EnvHealth

   Evaluating Urine Metabolic Profiles with Wildland- Urban-Interface (WUI) Fire Exposure: A Comparison with Municipal Structure Fires (MSF)

   Tuo Liu, Melissa A. Furlong, Shawn C. Beitel, and 5 more authors

   Environmental Health, 2025

   Abs DOI

   Firefighters have frequent exposure to carcinogens and an increased risk of cancer. Wildland-urban interface (WUI) fires, which involve both structures and undeveloped wildland fuels, pose unique challenges to the health of firefighters. However, the extent of health risks associated with these fires remains underexplored. This study aims to identify altered urine metabolites and metabolic processes among male firefighters that were associated with WUI fires as compared with municipal structure fires (MSF). Untargeted metabolomic profiling was applied to pre-exposure (baseline) and postfire urine samples collected from firefighters responding to WUI and MSF exposure. Differential analysis was conducted by fitting linear mixed effects regression models on preprocessed ion intensity and exposure status while adjusting for demographic covariates. Differential metabolites by post-exposure status were identified using a false discovery rate (FDR) threshold of <0.05. Enrichment analysis was performed to identify pathways that were significantly perturbed at a Bonferroni adjusted p-value <0.05 level. Eighty-five firefighters contributed paired baseline and post-fire samples from WUI events, and 98 firefighters contributed paired baseline and post-fire samples from MSF events. We performed metabolic profiling on baseline and postfire urine samples from WUI and structure fires using four modes: HILIC(-), HILIC(+), C18(-), and C18(+) and identified metabolites against an in-house library. We identified 244, 297, 320, and 266 level 1 metabolites from the four respective modes. In the statistical analysis, the main model identified a total of 176 differential metabolites from WUI fires. For MSF, the model identified a total of 652 differential metabolites from the four respective modes. Most metabolites with significant changes after a WUI fire also changed significantly after an MSF event. Two pathways were significantly enriched after WUI fires, while seven pathways were significantly enriched after MSF exposure and two pathways overlapped between the two types of fires. Fire exposure induces numerous metabolic perturbations in firefighters that may partially explain their elevated cancer risks. Although individual metabolites changed in a similar fashion across both WUI and MSF, structure fires were associated with an increased number of metabolite changes and some of the altered pathways differed between exposures to WUI fires vs. MSF. These results suggest that exposures to WUI fires and MSF present both common and unique cancer risks for firefighters.

3. EnvRes

   Per- and polyfluoroalkyl Substances (PFAS) and microRNA: an epigenome-wide association study in firefighters

   Melissa A Furlong, Tuo Liu, Alesia Jung, and 4 more authors

   Environmental Research, 2025

   Abs DOI

   The occupation of firefighting is classified as a Group 1 carcinogen. Increased cancer risk among firefighters may be partly attributable to increased occupational exposure to a range of chemicals, including per- and polyfluoroalkyl substances (PFAS). Some PFAS are known and suspect human carcinogens. Investigating epigenetic response to these PFAS exposures in firefighters may help to identify biological pathways of specific cancers, and previously unidentified health outcomes that are associated with PFAS. We therefore investigated the associations of serum PFAS concentrations with miRNA expression in firefighters. Sera collected from 303 firefighters from 6 sites across the USA were analyzed for 9 PFAS along with miRNA expression. Linear regression was used to estimate associations between log PFAS and miRNA expression, with false discovery rate (FDR) set to 0.05 for significance, and an exploratory cutoff of FDR q<0.20. Gene set enrichment analysis (GSEA) was performed using miRTarBase’s miRWalk pathways. Age, race-ethnicity, BMI, fire department, and sex were controlled for in all models. At FDR<0.05, the linear isomer of perfluorooctane sulfonic acid (PFOS) was inversely associated with hsa-miR-128-15p expression (Beta = -0.146, 95% CI -0.216, -0.076). At a relaxed FDR of 0.20, we observed inverse associations for the sum of branched isomers of PFOS with 5 miRNAs (hsa-let-7d-5p, hsa-let-7a-5p, hsa-miR-423-5p, hsa-let-7b-5p, has-mir-629-5p). Several pathways were enriched for multiple PFAS, including those correlated with certain cancers, blood diseases, thyroid disorders, autoimmune disorders, and neurological outcomes. Some PFAS in firefighters were found to be associated with alteration of miRNA consistent with increased risk for a range of chronic diseases.

4. Metabolites

   Evaluating Differential Metabolic Profile by Prostate Cancer Risk among Prostate Cancer Patients

   Tuo Liu, Jahnvi Roorkeewal, Melissa Furlong, and 6 more authors

   Metabolites, 2025

   Abs DOI

   Background: Currently there are no clinically validated biomarkers recommended for prostate cancer (PCa) risk stratification other than prostate specific antigen (PSA). Objectives: This study aimed to identify urine metabolites that are associated with presence of high-grade PCa at the time of radical prostatectomy. Methods: Urine samples were collected from patients who underwent radical prostatectomy. High-resolution metabolomics were implemented using mass liquid-chromatography mass spectrometry (LC-MS). To enhance metabolic feature identification, sample extracts were analyzed in two modes, a C18 chromatography [reverse-phase (RP)] and a hydrophilic interaction chromatography (HILIC). Results: This analysis included a total of 22 patients with PCa (10 high-grade and 12 low-grade) and identified 52 differential metabolites, 40 in RP and 12 in HILIC, at p-value 0.05 level. Among these, methyl alpha-aspartyl phenylalaninate was most significantly differentiated, while 3-methylbutanoicacid had the largest difference (slope -3.488). In pathway analysis, the histidine metabolic pathway was significantly enriched with an enrichment factor of 3.5. Our research also identified alterations in vitamins B12, B7 (biotin), B6, and B3 (niacin) pathways. Conclusion: Urine metabolites have the potential to differentiate high-grade from low-grade PCa. Our study also highlights the metabolic reprogramming that occurs as PCa becomes more aggressive.

5. IJHEH

   Differential Metabolic Profile by Training Fire Exposure in Female Firefighters

   Tuo Liu, James Hollister, Krystal J. Kern, and 15 more authors

   International Journal of Hygiene and Environmental Health, submitted, 2025

   Abs DOI

   Background: Women firefighters face serious health risks including elevated risk for cancer and reproductive conditions, although underlying metabolic mechanism(s) are not fully understood. Objectives: This study aimed to identify urinary metabolites and metabolic functions associated with training fire exposure among women municipal firefighters. Methods: High-resolution metabolomics (HRM) was applied to urine sample collected at baseline and after live-fire burn room/tower or flashover training fire exposure from women firefighters in the Fire Fighter Cancer Cohort Study (FFCCS). To identify differentially expressed metabolites (DEMs), differential analysis was performed using linear mixed-effects models adjusting for demographic confounders including age, socioeconomic factors, cancer history, dietary and medication behaviors, with false discovery rate adjustment. Functional enrichment analysis (FEA) was carried out using metabolite-set enrichment analysis (MSEA) from MetaboAnalyst to identify enriched metabolic processes. A secondary stratified analysis was carried out to investigate the effect of training fire type on metabolome after fire exposure using a linear regression model while adjusting for covariates. Results: One hundred women firefighters participated, resulting in a total of 200 urine samples (100 baseline, 100 postfire). The 200 samples underwent HRM analysis in four separation-ionization modes including HILIC(+), HILIC(-), C18(+), and C18(-), annotating against an in-house library of  1200 standards. We identified 200, 300, 280, and 306 metabolites and 10, 9, 23, and 19 post-training fire DEMs from the four modes, respectively. The FEA process identified that glycerophospholipid metabolism was significantly enriched at a p-value 0.05 level. Stratified analysis identified a total of 17 DEMs by fire type and increased relative ion intensities across all DEMs during burn room/tower fires compared to flashover fires. Conclusion: Women firefighters exposed to training fires exhibited a set of metabolic changes, particularly related to cellular damage from oxidative stress. These observations suggest a potential pathway for chronic inflammation with long-term fire exposure, which may help explain the higher prevalence of certain health conditions observed in women firefighters. Increased intensity DEMs were found following burn room/tower as compared with flashover fires.

2023

1. SciRep

   Evaluating changes in firefighter urinary metabolomes after structural fires: an untargeted, high resolution approach

   Melissa A Furlong, Tuo Liu, Justin M Snider, and 12 more authors

   Scientific Reports, 2023

   Abs DOI

   Firefighters have elevated rates of urinary tract cancers and other adverse health outcomes, which may be attributable to environmental occupational exposures. Untargeted metabolomics was applied to characterize this suite of environmental exposures and biological changes in response to occupational firefighting. 200 urine samples from 100 firefighters collected at baseline and two to four hours post-fire were analyzed using untargeted liquid-chromatography and high-resolution mass spectrometry. Changes in metabolite abundance after a fire were estimated with fixed effects linear regression, with false discovery rate (FDR) adjustment. Partial least squares discriminant analysis (PLS-DA) was also used, and variable important projection (VIP) scores were extracted. Systemic changes were evaluated using pathway enrichment for highly discriminating metabolites. Metabolome-wide-association-study (MWAS) identified 268 metabolites associated with firefighting activity at FDR q < 0.05. Of these, 20 were annotated with high confidence, including the amino acids taurine, proline, and betaine; the indoles kynurenic acid and indole-3-acetic acid; the known uremic toxins trimethylamine n-oxide and hippuric acid; and the hormone 7a-hydroxytestosterone. Partial least squares discriminant analysis (PLS-DA) additionally implicated choline, cortisol, and other hormones. Significant pathways included metabolism of urea cycle/amino group, alanine and aspartate, aspartate and asparagine, vitamin b3 (nicotinate and nicotinamide), and arginine and proline. Firefighters show a broad metabolic response to fires, including altered excretion of indole compounds and uremic toxins. Implicated pathways and features, particularly uremic toxins, may be important regulators of firefighter’s increased risk for urinary tract cancers.