Exploring the relationship between social jetlag with gut microbial composition, diet and cardiometabolic health, in the ZOE PREDICT 1 cohort

The study investigates whether minor circadian misalignment, measured as social jetlag (SJL), is linked to gut microbiome composition, dietary patterns, and cardiometabolic health. Modern lifestyles disrupt sleep patterns, and even small irregularities like SJL (difference in mid-sleep time ≥1 h between work and free days) may affect health. Prior work associates SJL with poorer diet, adiposity, glycemic dysregulation, dyslipidemia, and inflammation, but mechanisms remain unclear and data on postprandial responses in large healthy populations are limited. As sleep and the gut microbiome are bidirectionally related and diet can mediate microbiome changes, the study aims to: (1) assess relationships between SJL and gut microbiome, diet, fasting and postprandial cardiometabolic markers; and (2) evaluate potential mediation by diet in SJL–microbiome associations.

Prior literature links SJL with unhealthy dietary patterns (lower adherence to Mediterranean diet, higher sugar-sweetened beverage intake, lower fiber) and adverse metabolic profiles (higher triglycerides, insulin resistance, adiposity, and metabolic syndrome). Sleep disturbances can increase inflammatory markers and oxidative stress. Postprandial glycemia and lipemia are associated with adverse health outcomes, yet large-scale postprandial data in relation to sleep irregularity are scarce. Sleep quality and the gut microbiome are interrelated; circadian disruption and dysbiosis are both associated with obesity, metabolic syndrome, and IBD, with potential mediation via diet and microbial metabolites along the microbiota–gut–brain axis. The PREDICT program previously reported links between sleep disturbances and impaired postprandial glycemia, and between diet quality and gut microbiome, but SJL–microbiome associations had not been explored.

Design and cohort: Secondary analysis of the ZOE PREDICT 1 study (NCT03479866), a single-arm, single-blinded intervention conducted June 2018–May 2019 in 1002 healthy UK adults aged 18–65, including twins and non-twins. Real-world data included clinic and at-home measures (genetics, metabolic markers, gut microbiome, activity, sleep, meal order/timing/composition, demographics). Ethical approval IRAS 236407. Sleep assessment: Self-reported habitual sleep and wake times for weekdays and weekend days were used to compute SJL as the difference in midpoint of sleep between weekend and weekday nights. SJL was analyzed categorically with a cut-off ≥1.5 h (SJL vs no-SJL) and continuously. Actigraphy was available but not used for SJL due to protocol-induced sleep changes on weekends. Participants also completed the Pittsburgh Sleep Quality Index (PSQI). Sleep duration categories: short <7 h, average 7–9 h, long >9 h. Chronotype estimated from weekend mid-sleep time. Microbiome: Participants collected stool samples at home prior to clinic visit (n = 1001). Shotgun metagenomic sequencing was performed (Illumina NovaSeq 6000, 300-bp paired-end reads). Taxonomic and functional profiling followed previously described protocols. Clinic meal and blood sampling: During a clinic visit, participants consumed a standardized meal (890 kcal; 86 g carbohydrate, 53 g fat, 16 g protein, 2 g fiber). Venous blood collected at multiple timepoints (0–360 min) for glucose, insulin, C-peptide, triglycerides, and NMR lipid profiling. Postprandial AUCs: glucose 2h iAUC, insulin 2h iAUC, triglycerides 6h iAUC. Inflammation: GlycA measured fasting, 4 h and 6 h postprandially by NMR metabolomics; IL-6 measured. CGM: Participants wore Freestyle Libre Pro CGM for 14 days; analyzed data from 12 h post-fitting onward. Outcomes: glucose variability (CV%), optimized time-in-range (3.9–5.6 mmol/L). Diet assessment: Habitual intake via EPIC FFQ; nutrient data and diet quality indices computed (PDI, hPDI, uPDI, HEI, aMED). Exclusions for implausible energy intake or incomplete FFQ. Free-living diet habits: On 2–4 at-home days, participants logged ad libitum intake with app support. Nutrient data from McCance and Widdowson and brand sources. Eating occasion (EO) defined as ≥50 kcal, ≥30 min apart. Main meals defined by sex-specific kcal thresholds; snacks as other EOs. Derived variables: number of EOs, number of main meals, first/last EO time, eating window, eating midpoint. Appetite/mood: Visual analogue scales collected hunger ratings post-meal and daily anxiousness (~9 PM). Averaged across study for participants with ≥7 days of ratings. Statistics: Participants with complete sleep timing for week and weekend included (n = 945); exclusions applied for implausible times/durations and negative SJL, resulting in n = 934. Group differences (SJL vs no-SJL) tested via ANCOVA adjusting for sex, age, BMI, ethnicity, education. Normality tested; transformations applied as needed. Significance at two-sided P < 0.05; Benjamini–Hochberg FDR correction applied. Additional adjustments: habitual sleep duration, sleep quality, alcohol intake, diet quality (hPDI). Partial correlations between continuous SJL and diet variables adjusted for covariates. Microbiome differential abundance tested on species prevalent in ≥20% using Mann–Whitney U with FDR; significance defined as FDR q < 0.2 and absolute Cohen’s d > 0.2. Random Forest classification using species-level genome bins with 100 bootstraps (80/20 splits). Age-matched subgroup analysis. Mediation analysis (diet quality and nut intake) on species using linear mixed-effects models, arcsine square root transformation for relative abundances, adjusting for sex, age, BMI, ethnicity, education, and family relatedness.

- Prevalence and characteristics: 16% (n = 145) had SJL (≥1.5 h). SJL group had more males (39% vs 25%), were younger (38.4 ± 11.3 y vs 46.8 ± 11.7 y), and slightly more short sleepers (<7 h: 5% vs 3%). Mean difference in mid-sleep between week and weekend was 0.83 ± 0.61 h. - Microbiome: Seventeen species differentially abundant between SJL and no-SJL (q < 0.2, |Cohen’s d| > 0.2); 9 higher in SJL, 8 higher in no-SJL. Three species (Clostridia bacterium SGB14263, Clostridia bacterium SGB3940, Peptococcaceae bacterium GB49168) were prevalent (≥20%) only in no-SJL. Whole microbiome did not discriminate classes (median AUC = 0.572). Age-matched analyses confirmed 13 of the 17 species and identified 2 new species (SGB1877, SGB15154). - Diet and mediation: SJL associated with slightly poorer diet quality (lower hPDI; higher uPDI), higher intakes of sugar-sweetened beverages and potatoes, lower fruit and nut intakes (P < 0.05 before FDR). Nut intake inversely correlated with SJL (r = −0.13, P < 0.05). Diet quality partially mediated associations between SJL and two species (SGB71759: 9% mediation, P = 0.036; Ruminococcaceae bacterium: 4%, P < 0.05). Nut intake mediated SJL–microbiome associations to a greater extent (SGB71759: 15%, P = 0.01; Ruminococcaceae bacterium: 5%, P = 0.02). Additional nut-mediated species included Flavonifractor plautii (9%, P = 0.02) and Clostridia bacterium SGB14253 (12%, P = 0.036). - Eating patterns and chronotype: SJL group had fewer eating occasions (4.7 ± 1.2 vs 5.2 ± 1.3) and fewer main meals (2.4 ± 0.8 vs 2.6 ± 0.8) (P < 0.001), later chronotype (weekend mid-sleep 04:30 [IQR 04:00–05:15] vs 03:30 [03:00–04:00]) and delayed first main meal. - Inflammation and metabolic markers: SJL associated with slightly higher fasting GlycA (1.35 ± 0.19 vs 1.32 ± 0.18 mmol/L) and IL-6 (0.95 ± 2.62 vs 0.69 ± 0.65 ng/L) (P < 0.05), but not significant after FDR. Interaction analyses showed higher GlycA particularly in males with SJL; perimenopausal women more susceptible than pre- or post-menopausal. No differences in body composition, fasting/postprandial glucose, insulin, triglycerides, or CGM-derived metrics after covariate adjustments.

The study shows that even modest circadian misalignment (SJL) is associated with differences in specific gut microbial species, slightly poorer diet quality, altered eating patterns, and small increases in systemic inflammation within a largely healthy cohort with adequate sleep duration. While overall microbiome composition did not classify SJL status, species-level differences were evident and independent of age. Mediation analyses suggest diet quality—especially nut intake—partly explains SJL–microbiome associations, supporting a pathway whereby SJL influences diet, which in turn shapes the microbiome. Findings align with prior literature linking SJL to poorer diet and metabolic disturbances and suggest that behavioral timing (later chronotype and delayed first main meal) may contribute to observed dietary and microbiome differences. Although increases in inflammatory markers were modest and not significant after multiple test correction, sex- and menopause-related interactions indicate subgroups may be more sensitive to SJL-related inflammation. The absence of differences in fasting or postprandial metabolic biomarkers likely reflects the cohort’s overall health and adequate sleep, indicating that SJL’s effects may be subtle or require longer-term exposure or higher-risk populations to manifest more robustly.

Minor circadian misalignment in the form of social jetlag is associated with distinct gut microbiome species, partly mediated by dietary differences, and with modest, subclinical indications of increased inflammation in a healthy cohort. Encouraging sufficient, consistent sleep aligned to individual chronotype and maintaining a healthy diet—particularly higher nut and fruit intake and lower sugar-sweetened beverages and refined starches—may mitigate potential adverse effects. The results raise the possibility that socially imposed sleep–wake shifts influence the gut microbiota and could inform microbiome-targeted interventions for sleep-related and metabolic diseases. Future research should include larger and more diverse cohorts, objective sleep measures, and longitudinal designs to determine causality and long-term health implications of SJL.

- Cross-sectional analysis precludes causal inference and directionality. - No data on shift work or employment status. - Sleep based on self-report and time-in-bed (no sleep latency), not polysomnography; actigraphy not used for SJL due to protocol effects. - Week vs weekend days used as proxies for work vs free days. - Inability to determine use of sleep medications. - Cohort predominantly healthy, white, and adequate sleepers, limiting generalizability and possibly attenuating effects; relatively low prevalence of SJL (≥1.5 h) and short sleep compared with other populations. - Multiple testing corrections attenuated several associations, indicating small effect sizes.