Elaine Holmes

The Mediterranean diet (MD), known for its high intake of fruits, vegetables, whole grains, legumes, and healthy unsaturated fats, has been linked to a diverse and beneficial gut microbiome. However, its effect on the gut microbiome during pregnancy remains understudied. This study aimed to investigate the impact of high adherence to a Mediterranean diet on gut microbiome composition and function in pregnant women by analysing their metabolic profiles and faecal microbiome composition. Stool, serum, and urine samples were collected from 48 pregnant women at weeks 20/28 and at week 36. Participants were stratified based on MD adherence using a validated questionnaire. Stool samples underwent 16 S rRNA gene amplicon sequencing, and serum short-chain fatty acids (SCFAs) were measured using UPLC-MS. Women with high MD adherence showed significantly higher α-diversity in their faecal microbiomes at both time points. Significant differences in microbiome composition were observed between low and high adherence groups at weeks 20/28, but not at week 36. No significant differences in serum short-chain fatty acid concentrations were found between the groups. Our findings suggest that adherence to the Mediterranean diet during pregnancy is associated with changes in gut microbiome diversity and function. These results contribute to a better understanding of how dietary patterns during pregnancy may influence gut microbiome ecology.

To determine whether plasma lipid profiles predict diabetic retinopathy (DR).
DR status (none, mild non-proliferative diabetic retinopathy (NPDR), moderate NPDR, severe NPDR/worse) was categorised in 762 adults with type 2 diabetes (mean age 64.5 years, 54.3% males) at baseline and Year 4/6. Ultra-performance liquid chromatography-tandem mass spectrometry generated baseline plasma lipid profiles. Multiple logistic regression identified baseline associates of new/worsening DR and incident DR. The likelihood ratio test (LRT) evaluated incremental biomarker contribution. The net reclassification improvement (NRI) was calculated.
New/worsening DR occurred in 121 (16%). Of 495 without DR at baseline, 35 (7.0%) developed DR. For new/worsening DR, the inclusion of cholesterol ester (20:4), fatty acid (20:3) and lysophosphatidylglycerol (18:1) in addition to conventional risk factors (hypertension, HbA1c, diabetes treatment intensity, urinary albumin:creatinine) added to the conventional model predictivity (LRT, P = 0.00002). The NRI gain at a 10% cut-off was 5.7% (SE 2.8%; P = 0.043). For incident DR, fatty acid (20:3), lysophosphatidylglycerol (18:0) and phosphatidylcholine (18:0_20:3) with HbA1c improved model performance (LRT, P = 0.00001). The NRI gain at 5% risk cut-off for incident DR was 28.3% (P = 0.002).
This exploratory study identified five lipid biomarkers with the potential to improve DR risk prediction and the understanding of pathophysiological mechanisms.
•Lipidomic biomarkers could enhance prediction of incident diabetic retinopathy (DR).•Baseline plasma lipid profiles were generated in people living with type 2 diabetes.•New/worsening DR and incident DR were assessed over 4 years of follow-up.•Five lipid moieties added to conventional variables in predicting these outcomes.•Lipid biomarkers may improve DR risk prediction and provide mechanistic insights.

Systematic literature reviews are widely acknowledged to be at the top of most evidence hierarchies but are very time-consuming and labour-intensive. The advent of artificial intelligence (AI) offers possibilities to accelerate the review process. We propose a systematic literature review protocol that incorporates AI and evaluate the accuracy of AI against a human-constructed gold standard. Human researchers performed a systematic literature review on the PubMed metabolomic literature investigating Long COVID. OpenAI model o3-mini was then used via the application programming interface to carry out the three key stages of a systematic literature review: 1) screening articles for inclusion criteria, 2) evaluating articles' risk of bias (RoB) using a customised metabolomics tool based on well-established quality criteria, and 3) extracting data from articles. The performance of AI was evaluated against a manually created reference by the assessment of accuracy, negative predictive value, precision, sensitivity, specificity, F1 score, and Cohen's kappa. Out of the 320 identified articles, 33 met all inclusion criteria. AI performed screening, RoB assessment, and data extraction with mean accuracies of 92.7% +/- 4.9%, 91.9% +/- 5.9%, and 92.0% +/- 9.2%, respectively, in similar to 10 h and at a cost of similar to 15.35 USD, while human reviewers took similar to 116 h, equivalent to similar to 2000 USD at minimum wages, to complete these steps in duplicate. OpenAI model o3-mini can independently complete the three main stages of systematic literature reviews in the metabolomics domain, thus considerably reducing the time to complete such reviews, at minimal cost. Nonetheless, human reviewers should always verify the AI-generated output to ensure robustness.

Phenomic medicine is an emerging field that involves measurement and modelling of multidimensional phenotypic data to provide a systems-level framework to understand how health and disease evolve over time and inform decision making in a clinical context . Unlike genomic medicine, which is largely predictive of predisposition, phenomic medicine captures the dynamic biochemical, physiological, and metabolic changes that occur across the disease continuum. Metabolic profiling, imaging, digital health data, and electronic health records can be used to characterise these phenotypic shifts longitudinally by constructing temporal models that map disease trajectories at both individual and population-wide scales. Phenomic trajectories allow for the stratification of patients based on the evolving pathobiology arising from gene-environment interactions that underpin disease progression, resilience, or recovery. The modelling of these trajectories can identify early markers of transition from health to subclinical dysfunction, establish critical inflection points, and distinguish reversible versus irreversible stages of disease. This capability is particularly powerful in complex, multifactorial conditions such as cardiovascular disease, diabetes, or post-viral syndromes, where a simple panel of biomarkers established in cross-sectional cohort studies lacks the required specificity for prognostic precision. Ultimately, phenomic medicine offers a route to transforming care from a reactive to a proactive mode, and can support personalised health management across the life course.

The human gut microbiome is increasingly viewed as an active regulator of host physiology, extending beyond earlier taxonomy‐centered descriptions of a complex microbial community. Accumulating evidence supports an organ‐like conceptual framework in which the gut microbiome exhibits spatially structured organization, extensive metabolic capacity, and continuous bidirectional communication with host systems. Through the production of bioactive metabolites with endocrine‐like, immunomodulatory, and neuromodulatory properties, the microbiome contributes to metabolic, immune, and neuroendocrine regulation, thereby influencing systemic homeostasis and disease susceptibility. Recent advances in multi‐omics, spatial biology, and computational modeling are moving the field from taxonomic association toward functional interpretation, mechanistic insight, and causal inference. These approaches are beginning to reveal microbiome‐derived functional modules and host–microbe signaling networks that are shaped by host genetics, diet, medications, feeding patterns, circadian rhythms, and environmental exposures. In this review, we synthesize current mechanistic and translational evidence to conceptualize the gut microbiome as an organ‐like functional system, delineate its structural and functional organization, and propose a framework for mapping, modeling, and therapeutically targeting microbiome‐derived circuits to support precision medicine in metabolic, inflammatory, and selected gut–brain axis‐related disorders. The gut microbiome can be conceptualized as a distributed organ‐like functional system with spatially structured organization, broad biochemical capacity, and continuous bidirectional communication with the host. By transforming dietary, host‐derived, and environmental substrates into bioactive metabolites with endocrine‐like, immunomodulatory, and neuromodulatory activities, the microbiome contributes to metabolic, immune, and neuroendocrine pathways that shape systemic homeostasis and disease susceptibility. Emerging spatial, multi‐omics, and computational approaches are enabling systems‐level mapping of microbiome architecture, functional modules, and host–microbe signaling networks, moving the field beyond descriptive taxonomy toward mechanistic insight and causal inference. This organ‐level framework provides a conceptual and translational basis for developing precision microbiome medicine through the identification of targetable pathways, functional circuits, and personalized therapeutic strategies.

Background
Obesity is associated with adverse alterations in lipoprotein profiles and increased cardiometabolic risk, yet standard clinical lipid measures provide limited resolution of the underlying lipoprotein structure. Here, we perform a population-level characterisation of the impact of obesity on the high-resolution blood lipoprotein subfraction profile.

Methods
Lipoprotein profiles were investigated in an Australian cohort (n = 1806) stratified by Body Mass Index (BMI) (healthy weight, overweight, obese). The composition and particle number of lipoproteins and their subfractions were quantified (n = 112 parameters) using 1H NMR spectroscopy. Associations between BMI, lipoprotein subfractions, and particle number ratios were evaluated. Key BMI-associated ratios were validated in a second Australian cohort (n = 1313)

Results
Participants with obesity had significantly higher particle concentrations of VLDL and IDL and lower HDLs, while total LDL concentration was similar between the BMI categories. Notably, obesity was associated with a marked redistribution of LDL subfractions, with a shift from larger, buoyant LDL1–3 particles to smaller, dense LDL4–6 particles, accompanied by a similar shift in HDL subfractions, with proportionally lower concentrations of HDL1–3 particles in the high BMI group. An inverse association between BMI was found for two LDL particle number ratios (LDL1–3:LDL4–6 and LDL2:LDL5) and one HDL ratio (HDL1:HDL4), which demonstrated good discriminatory power for obesity status (AUROCs 0.70–0.74).

Conclusions
The pattern of lipoprotein subfraction redistribution in participants with obesity was consistent across cohorts. These subfractional changes offer enhanced resolution compared with conventional lipid and lipoprotein measures and may contribute to improved cardiovascular risk stratification in individuals with obesity.

Background
Previous findings for the association of fish intake and its related metabolites on risk of metabolic syndrome (MetS) are inconsistent and limited. The prevalence of fish and shellfish consumers are high in Japan where it contributes to one fourth of the protein intake. We aimed to investigate the relationship of fish and fish-related metabolites with MetS risk factors in a community-dwelling Japanese population.

Methods
In this cross-sectional study, 1047 Japanese men and women aged 40–59 years were included. Four in-depth multiple-pass 24-h dietary recalls, two timed 24-h urine collections, blood collection, extensive questionnaire and anthropometric measurements were collected for each participant. Urinary and serum profiles were obtained using 1H Nuclear magnetic resonance (NMR) spectroscopy.

Results
Mean fish intake was 47.0 (SD 28.8) g per 1000 kcal. Urinary trimethylamine-N-oxide (TMAO) and taurine, and serum creatine and threonine were significantly correlated with fish intake and homarine was associated with shellfish intake (p < 0.01). Significant inverse associations were found between triglycerides and total fish intake, low omega-3 fish, urinary TMAO, taurine and between urinary taurine with systolic blood pressure. Direct association between urinary TMAO and serum threonine with high-density lipoprotein cholesterol were found.

Conclusion
We found several metabolites related to fish intake and risk factors of MetS in habitual fish-consuming country, Japan. When stratified by seafood type, raw and high omega-3 fish intake showed the strongest associations with urinary and serum metabolomes, while fried fish had weaker or non-significant correlations, suggesting that cooking methods may modulate the bioavailability or metabolic impact of fish nutrients. The combined panel of urinary TMAO and taurine with plasma creatine and threonine, reflecting seafood intake, was associated with a favorable lipid profile and lower blood pressure.

Background
Evidence increasingly suggests a connection between cardiovascular disease and brain health in later life; however, the mechanistic pathways from human studies remain unclear. This study aimed to investigate whether urinary metabolites account for part of the association between cognition and cardiometabolic risk.

Methods
Data from 606 participants (aged 48–60; 55% female; 45.5% Black/African American) in the Year 30 follow-up of the Coronary Artery Risk Development in Young Adults Study were analyzed. Urinary metabolites were profiled using nuclear magnetic resonance spectroscopy and liquid chromatography-mass spectrometry; brain magnetic resonance imaging data were available for 281 participants. Structural equation models were used to assess pathways linking cardiometabolic factors to cognitive outcome, with urinary metabolites and brain MRI-derived parameters as mediators.

Results
Fasting glucose showed a negative association with cognition. Valine, isoleucine, leucine, and phenylalanine were positively associated with fasting glucose. Valine and aminoadipic acid also showed positive associations between fasting glucose and cognition, while tryptophan was correlated with both fasting glucose and cognition. Indole-3-acetic acid showed negative associations with systolic blood pressure and fasting glucose. Brain MRI-derived parameters in memory-related medial temporal areas were associated with waist circumference.

Conclusions
Urinary metabolites and brain imaging markers were linked with hyperglycemia, obesity, and cognitive performance, highlighting multimodal biomarkers relevant to global cognitive function in individuals with cardiometabolic risk.

Broad-spectrum viral biomarkers offer a promising approach to distinguishing viral from bacterial infections, thereby reducing inappropriate antibiotic use and improving diagnostic response during emerging infectious disease outbreaks. Among these, the deoxydidehydronucleoside (ddhN) class of nucleoside derivatives has emerged as a potential tool for early detection of viral infections in settings where pathogen-specific diagnostics are unavailable. To assess the clinical utility of these compounds, we investigated the metabolism and excretion rates of four principal ddhN metabolites, 3'-deoxy-3',4'-didehydrocytidine (ddhC), 3'-deoxy-3',4'-didehydrocytidine-5'-carboxylate (ddhC-5'CA), 3'-deoxy-3',4'-didehydrouridine (ddhU), and 3'-deoxy-3',4'-didehydrocytidine-5'-homocysteine (ddhC-5'Hcy), in the Sprague-Dawley rat model following a single intravenous dose. Time-resolved biological sampling was used to characterize urinary excretion and downstream biotransformation. All four metabolites exhibited rapid urinary clearance, ranging from approximately 3 to 8 h, consistent with a transient acute-phase profile. Notably, ddhC-5'Hcy underwent extensive biotransformation, with key metabolites produced via functionalization and conjugation identified following integration of nuclear magnetic resonance (NMR) spectroscopy and mass spectrometry (MS) analyses. No adverse clinical signs were observed in any treatment group at any time point. These findings support further research into the ddhN series as markers of active viral infection for clinical application, particularly in critical care environments, where timely differentiation of infectious etiology is essential.