Open Air Scholar

Design:

The study employed a multi-omics approach combining bioinformatics analysis of public microarray data (GSE8586), machine learning algorithms, and experimental validation in hyperoxia-exposed rat models. Umbilical cord tissue samples from preterm infants with and without BPD were analyzed to identify differentially expressed genes (DEGs). ARGs were overlapped with DEGs to pinpoint autophagy-related candidates. Single-cell RNA sequencing (scRNA-seq) and immune infiltration analysis via CIBERSORT were used to explore cellular and immune contributions. Drug repurposing predictions and a nomogram for BPD risk stratification were also developed.

Procedures:

Microarray data from the GEO database (GSE8586) were analyzed to identify DEGs between BPD and control groups. ARGs were obtained from the Autophagy Database. Lasso regression and Cytoscape-based analyses (MCC, Degree, MCODE) were applied to select hub genes. Functional enrichment and pathway analysis (e.g., IL-17 signaling, MAPK activation) were performed using tools like DAVID. Single-cell sequencing data were used to map hub gene expression in lung cell types, including epithelial cells, fibroblasts, and pneumocytes. CIBERSORT quantified immune cell populations in BPD samples. Hyperoxia-induced rat models were exposed to oxygen for 14 days, and lung tissues were assessed for histology and gene expression. Drug-gene interactions were predicted using DrugBank and CMap databases. A nomogram integrating hub genes was validated via receiver operating characteristic (ROC) curves and decision curve analysis.

Primary Findings:

Analysis of GSE8586 identified 647 DEGs in BPD, with 73 overlapping ARGs (DE-ARGs). Among these, 43 were upregulated and 30 downregulated. Machine learning selected five hub genes (BRIX1, JUN, PES1, NR4A1, RRP9) with low expression in BPD and diagnostic accuracy (AUC >0.67). Single-cell sequencing confirmed their expression in critical lung cell types. Immune infiltration analysis revealed increased pro-inflammatory cells (e.g., M1 macrophages, eosinophils) and decreased anti-inflammatory cells (e.g., T follicular helper [Tfh] cells, natural killer [NK] cells) in BPD. Hub genes correlated with specific immune populations: JUN negatively associated with M2 macrophages, while NR4A1 linked to CD4+ T cells. Experimental validation in hyperoxia-exposed rats showed reduced hub gene expression compared to normoxic controls, alongside histological evidence of impaired alveolarization. A nomogram combining the hub genes achieved robust predictive accuracy (AUC = 0.85) for BPD risk.

Secondary Findings:

Pathway analysis highlighted roles for DE-ARGs in ribosome biogenesis, apoptosis, and immune responses. The IL-17 signaling and colorectal cancer pathways were significantly enriched. Each hub gene exhibited distinct functional roles: BRIX1 and RRP9 regulate ribosomal RNA processing, PES1 influences fibroblast proliferation via pre-RNA regulation, JUN drives pulmonary fibrosis by activating reticular fibroblasts, and NR4A1 counteracts fibrosis by suppressing TGFβ1-induced collagen deposition. Reduced RRP9 expression destabilized ribosomal components, promoting fibrosis-related cell proliferation. Autophagy dysfunction in alveolar type II (AT2) cells was linked to delayed lung maturation. The study also corroborated prior findings on H19 lncRNA’s role in MAPK signaling and IL-17’s contribution to neonatal lung injury.

Interpretation:

The study demonstrates that autophagy dysregulation and immune imbalance are central to BPD pathogenesis. The identified hub genes (BRIX1, JUN, PES1, NR4A1, RRP9) bridge autophagy dysfunction and fibrosis, suggesting their roles in impaired lung repair and inflammation. For example, PES1’s inhibition of autophagy may exacerbate ribosomal stress in epithelial cells, while JUN’s fibrogenic effects and NR4A1’s anti-fibrotic activity highlight opposing pathways. The nomogram’s strong predictive performance underscores its potential for clinical risk assessment. The interplay between autophagy, immune responses, and fibrosis provides a mechanistic framework for understanding BPD progression.

Strengths:

Key strengths include the integration of multi-omics data (microarray, single-cell sequencing), experimental validation in vivo, and drug repurposing predictions. The use of machine learning algorithms (Lasso regression, random forest, SVM) enhanced biomarker selection, while CIBERSORT provided detailed immune profiling. The study’s focus on translational applications, such as therapeutic drug candidates, aligns with clinical needs.

Limitations:

Limitations include the small sample size of the GEO dataset (n=12 BPD vs. 12 control cord samples), which may limit generalizability. While the rat model validated gene expression trends, it does not fully replicate human BPD’s complexity. Additionally, the study primarily relied on computational predictions for drug efficacy, requiring further experimental testing. The cross-sectional design of the microarray analysis precludes insights into temporal gene expression changes during BPD development.

Design:

A retrospective analysis of syndromic surveillance data from the UK Health Security Agency (UKHSA) and the Royal College of General Practitioners (RCGP) Research and Surveillance Centre (RSC). The study included 1,245 GP practices covering 11 million patients. HFMD cases were identified using SNOMED-CT clinical codes, with weekly consultation rates calculated per 100,000 population. Data were analyzed for trends from 2017 to 2022, comparing pre-pandemic (2017–2019), pandemic (2020–2021), and post-pandemic (2022) periods.

Procedures:

Weekly HFMD consultation rates were calculated using RCGP data. Seasonal patterns, age-specific incidence, and sex differences were assessed. The impact of non-pharmaceutical interventions (NPIs) during the pandemic was evaluated by comparing consultation rates before, during, and after lockdowns. Historical comparisons with a 1996 RCGP sentinel network study were conducted to contextualize current trends.

Primary Findings:

The mean weekly HFMD consultation rate from 2017 to 2022 was 1.53 per 100,000 population. Seasonal peaks occurred between September and December in non-pandemic years. Children aged 1–4 years had the highest incidence (2,411.5 per 100,000 in 2018), followed by infants under 1 year. Males consistently had 20% higher consultation rates than females. During the pandemic (2020–2021), HFMD activity dropped to near-zero levels due to lockdowns and social distancing. Post-pandemic (2022), activity rebounded but peaked later (week 46) compared to pre-pandemic years.

Secondary Findings:

Historical comparisons revealed similar seasonal peaks in the 1996 RCGP data but lower current incidence rates, possibly due to improved surveillance or reduced reporting. Emerging serotypes like CVA6 and EV-A71 B5 were noted, with potential links to severe outcomes. Meteorological factors, such as temperature and humidity, were associated with HFMD transmission in other global studies, though not explicitly analyzed in this dataset.

Interpretation:

The pandemic’s NPIs, including school closures and social distancing, drastically reduced HFMD transmission, highlighting their effectiveness. Post-pandemic shifts in seasonality suggest altered transmission dynamics, possibly due to changes in childcare attendance or immunity patterns. The male predominance may reflect behavioral factors, such as higher social interaction in young boys. The study underscores the role of schools and childcare settings in disease spread and the importance of syndromic surveillance for tracking epidemiological shifts.

Strengths:

The study leveraged a large, nationally representative dataset with long-term surveillance (2017–2022). The integration of pandemic-era data provided unique insights into the impact of NPIs on HFMD transmission. Historical comparisons with 1996 data added context to current trends.

Limitations:

Reliance on GP consultation data may underestimate true incidence due to asymptomatic cases or underreporting. SNOMED-CT coding accuracy could affect case identification. The study did not analyze viral serotypes directly, limiting conclusions about specific enterovirus strains.

Design:

Eleven recreationally trained cyclists completed two experimental visits in a randomized, counterbalanced order. Each visit included either a normal warm-up (10 minutes at 50 W) or a priming warm-up (4 minutes at 50 W followed by 6 minutes at 70% of the difference between gas exchange threshold [GET] and maximum work rate [WRmax]). Both visits involved a ramp test (30 W/min increments) and a verification phase (105% WRmax for ~120 seconds).

Procedures:

Key measures included VO₂, heart rate (HR), blood lactate (BLa), ventilation, and respiratory exchange ratio (RER). VO₂pl was identified using individualized criteria (ΔVO₂ < 2.1 mL·kg⁻¹·min⁻¹ for 30 seconds) to minimize false positives. Recovery periods were standardized to 30 minutes between warm-up and ramp test. Verification phases confirmed VO₂max via plateau, RER ≥1.15, and HR ≥90% of maximum.

Primary Findings:

No significant differences were observed in VO₂pl incidence (73% in both protocols) or VO₂max between warm-up conditions. Oxygen kinetics (slopes and time constants) and cardiopulmonary responses during ramp tests were comparable. The verification phase confirmed VO₂max in all participants, increasing confirmation rates from 73% (ramp test alone) to 100%.

Secondary Findings:

The priming warm-up induced greater physiological strain (higher HR, BLa, and ventilation) but did not improve VO₂pl occurrence or peak performance metrics. Ramp test design limitations, such as aggressive work rate increments (30 W/min), may have obscured kinetic changes. Grouping data by exercise domains did not resolve this issue. Post-recovery BLa levels (3.4 ± 1.4 mmol·L⁻¹) adhered to best practices, avoiding excessive acidosis.

Interpretation:

The findings suggest priming warm-ups do not enhance VO₂pl incidence in trained individuals, possibly because ramp tests themselves act as a form of priming. The verification phase’s universal success underscores its necessity for accurate VO₂max assessment. Methodological inconsistencies in prior studies, such as non-individualized VO₂pl criteria and shorter recovery periods, likely contributed to conflicting results.

Strengths:

The study employed rigorous, individualized VO₂pl criteria and controlled warm-up durations. Standardized recovery periods minimized confounding factors like elevated baseline BLa. Verification protocols were validated across all participants, reinforcing their diagnostic utility.

Limitations:

The 30-minute recovery period may have reduced priming effects, and the ramp test’s aggressive design limited kinetic analysis. Future studies should explore discontinuous protocols (e.g., step-ramp-step) to better isolate priming impacts.

Design:

A cross-sectional observational study analyzed 184 reptile samples from 160 shipments spanning 2013–2014. Reptiles originated from 27 countries, primarily the USA (284 isolates), Vietnam (135), and included Squamata (lizards/snakes) and Testudines (turtles/tortoises). Bacterial isolates were identified using MALDI-TOF MS, TYGS, and JSpec. Whole-genome sequencing (WGS) and S1-pulsed-field gel electrophoresis (S1-PFGE) were employed to characterize ESBL/AmpC genes, plasmids, and genetic lineages. Metadata from shipping documents were reviewed for sourcing claims (wild-caught vs. captive-bred/farmed).

Procedures:

Phenotypic screening for β-lactam resistance used cefotaxime-supplemented agar. ESBL/AmpC confirmation relied on WGS, identifying resistance genes (e.g., blaCTX-M, blaCMY-2, blaDHA-1) and plasmid types (IncIγ/K1, IncFIB). Multilocus sequence typing (MLST) determined bacterial clonal lineages. Phylogenetic analysis grouped isolates by geographic origin and host species. Metadata accuracy was assessed via collaboration with the German Herpetological Society (DGHT), which evaluated sourcing claims against known farming practices.

Primary Findings:

Of 668 bacterial isolates across 16 genera, Klebsiella (181), Citrobacter (131), Escherichia coli (116), Salmonella (69), and Enterobacter (52) dominated. Phenotypic screening identified 79 cefotaxime-resistant isolates, with 39 confirmed as ESBL and 39 as AmpC producers via WGS. ESBL genes were most prevalent in E. coli (29 isolates), led by blaCTX-M-15 (16 isolates). AmpC genes included blaCMY-2 (12 isolates) and chromosomal variants like blaACT-16. Salmonella lacked ESBL/AmpC genes but exhibited high resistance to ceftiofur (>90% non-wild-type), enrofloxacin (63%), tetracyclines (63.3%), and trimethoprim/sulfamethoxazole (50.6%). Colistin resistance (8.9%) included two E. coli with mcr-1.1. Plasmid analysis revealed IncIγ/K1 and IncFIB as major resistance gene carriers. MLST identified 36 E. coli sequence types (STs), including globally disseminated ST131 (CTX-M-27) and reptile-associated novel STs (e.g., Enterobacter ST182). Phylogenetic clusters linked isolates from Vietnam/Uzbekistan and specific host species.

Secondary Findings:

Metadata discrepancies highlighted mislabeling: 257 isolates were classified as wild-caught, but 151 'farm-bred' claims were questionable (e.g., Ecuadorian leopard tortoises). ESBL/AmpC-positive isolates included Klebsiella pneumoniae ST307 (CTX-M-15) from Vietnamese lizards and E. coli ST131 from U.S. chameleons. Resistance patterns differed by sourcing: wild-caught and captive-bred reptiles harbored comparable AMR genes, suggesting environmental or anthropogenic spread. Plasmid types like IncC and IncX, common in human pathogens, were detected in reptile isolates.

Interpretation:

Reptiles serve as reservoirs for AMR Enterobacterales, with plasmid-mediated resistance enabling cross-species transmission. High-risk clones like E. coli ST131 and K. pneumoniae ST307, linked to human infections, were detected in imported reptiles, underscoring zoonotic potential. The absence of resistance in Galápagos land iguanas contrasts with imported reptiles, implicating anthropogenic factors in AMR spread. Environmental contamination and farming practices likely contribute to resistance dissemination, exacerbated by inaccurate metadata complicating regulatory enforcement.

Strengths:

The study’s large sample size (668 isolates), WGS-based analysis, and integration of metadata and phylogenetic data provided robust insights into AMR dynamics. Collaboration with herpetological experts enhanced sourcing claim evaluations.

Limitations:

Cross-sectional design precluded causal inferences about resistance acquisition. Metadata inaccuracies limited precise sourcing analysis. Limited temporal scope (2013–2014) may not reflect current trends in reptile trade or AMR prevalence.

Design:

The study employed a retrospective analysis of air quality and meteorological data from Tehran. A dataset spanning 10 years was used, with preprocessing steps including interpolation for missing values and standardization of features. Lasso regression was applied to penalize model complexity, reduce overfitting, and identify key predictors. Cross-validation was used to optimize the regularization parameter (λ), and performance was assessed using metrics such as R², mean squared error (MSE), root mean squared error (RMSE), and normalized mean squared error (NMSE).

Procedures:

Lasso regression was implemented to select features and balance model simplicity with predictive accuracy. The optimal λ value of 0.1 was determined via cross-validation. Key pollutants were analyzed separately, and feature importance was derived from coefficient magnitudes. Meteorological variables and pollutant interactions were evaluated for their influence on model outcomes. Challenges such as inconsistent sensor data and missing values in gaseous pollutants were addressed through interpolation and standardization.

Primary Findings:

Lasso regression achieved strong performance for particulate matter: PM2.5 (R² = 0.80) and PM10 (R² = 0.75). Gaseous pollutants showed lower accuracy, with O3 (R² = 0.35) and SO2 (R² = 0.65) underperforming due to high temporal variability and nonlinear interactions. Key predictors included O3 influencing CO (coefficient 3.632), PM2.5 linked to PM10 (17.405) and humidity (5.283), and PM10 driven by PM2.5 (19.422) and temperature (15.004). Meteorological variables like dew point and air pressure were often excluded due to multicollinearity.

Secondary Findings:

Ozone exhibited the highest MSE (277.998) and RMSE (16.673), while CO had the lowest (MSE = 57.574, RMSE = 7.589). NMSE highlighted SO2 as the most variable (0.718) and PM2.5 as the most stable (0.165). Cross-validation confirmed model robustness, with optimal R² achieved at 7–8 features. Lasso consistently prioritized temperature, wind speed, and pollutant interactions over less influential variables.

Interpretation:

Lasso effectively enhanced interpretability and generalization for stable pollutants like PM2.5 and PM10, where linear relationships dominated. However, its limitations in capturing nonlinear dynamics and complex chemical processes hindered performance for gaseous pollutants. The findings underscore the need for hybrid models combining linear and nonlinear approaches to address ozone and NO2, which are influenced by photochemical reactions and meteorological extremes.

Strengths:

The study’s strengths include rigorous cross-validation, clear feature selection insights, and the use of multiple performance metrics. The dataset’s temporal breadth (10 years) provided robust evidence for pollutant trends and model stability.

Limitations:

The linear framework of Lasso may have oversimplified interactions for gaseous pollutants. Missing data in gaseous measurements and sensor inconsistencies introduced uncertainties. Additionally, the analysis did not account for spatial variations within Tehran, limiting real-world applicability.