Cross-sectional associations between central and general adiposity with albuminuria: observations from 400,000 people in UK Biobank

The study investigates whether central adiposity (measured by waist-to-hip ratio) is more or less strongly associated with albuminuria than general adiposity (measured by BMI), and the extent to which glycemic status (HbA1c/diabetes) and other factors explain or modify these associations. Rising adiposity has been linked to increasing chronic kidney disease (CKD) mortality. BMI is associated with advanced CKD risk, potentially via dysglycaemia and hypertension, but central adiposity may differentially influence mediators (e.g., stronger links to insulin resistance and diabetes). Albuminuria is an early marker of CKD and offers a means to compare independent associations of central vs general adiposity with early kidney disease. The study also examines potential confounding/modification by HbA1c and dietary sodium, given recent therapies (SGLT-2 inhibitors, GLP-1 receptor agonists) that reduce weight and HbA1c and lower albuminuria risk.

A systematic review identified 46 observational studies assessing adiposity and albuminuria. Only five considered central adiposity adjusted for BMI. Two small studies (<5000 participants) found no association between central adiposity and proteinuria after BMI adjustment; two small Chinese studies reported positive associations; and a large Japanese study (>200,000) found central adiposity associated with proteinuria in men but not women after adjusting for BMI. Overall, prior evidence was inconsistent regarding whether central adiposity is independently associated with albuminuria beyond BMI.

Design and population: Cross-sectional analysis of 408,527 UK Biobank participants recruited in 2006–2010 across 22 centers. Exclusions: participants withdrawing consent (n=133); self-reported cancer (n=38,516), COPD (n=1,563), or liver cirrhosis/failure (n=323); extreme BMI (<15 or >60 kg/m²; n=77); missing adiposity measures (n=9,221), blood pressure (n=1,757), HbA1c (n=30,878), urinary albumin (n=11,649) or urinary creatinine (n=6). Final sample included 190,386 men and 218,141 women. Exposures: Central adiposity measures (waist-to-hip ratio [primary], waist-to-height ratio, waist circumference, percent trunk fat by bioimpedance) and general adiposity measures (BMI [primary], height-adjusted weight, hip circumference, percent body fat). Anthropometrics were measured using standardized equipment (Seca height rod; Wessex tape for waist/hip; Tanita BC-418 MA for weight, body fat %, trunk fat %). Adiposity measures were categorized into fifths for shape assessment. Outcome: Urinary albumin-to-creatinine ratio (uACR). Participants with undetectable albumin (<6.7 mg/L) were assigned an undetectable category. uACR categories: undetectable; low normal (0.1–<1 mg/mmol); high normal (1–<3 mg/mmol); albuminuria (≥3 mg/mmol). Sensitivity analysis treated albuminuria as binary (≥3 vs <3 mg/mmol/undetectable). Statistical analysis: Ordinal logistic regression estimated odds ratios (ORs) for higher uACR category across adiposity fifths, relative to the lowest fifth. Models: (1) confounder-adjusted (age, ethnicity, education, Townsend deprivation index, smoking, physical activity; urinary sodium-to-creatinine ratio assessed but excluded from final proportional odds models due to violation of proportional odds and lack of important confounding); (2) adiposity-adjusted (mutual adjustment of central and general adiposity, e.g., waist-to-hip ratio adjusted for BMI and vice versa) to estimate independent effects; (3) mediator-adjusted (further adjusting for diabetes status [diabetes: self-report or HbA1c ≥6.5%; pre-diabetes: HbA1c 5.7–<6.5%; no diabetes: HbA1c <5.7%], duration of diabetes, prior vascular disease [MI, angina, stroke], and systolic/diastolic blood pressure) to assess mediation. Measurement error correction: To reduce regression-dilution bias from single baseline measures, repeat measures from a resurvey (~4.3 years later) in 16,833 participants were used to obtain measurement-error–adjusted adiposity levels for baseline fifths. Sensitivity analyses without correction were conducted. Scaling and trend estimation: Associations were approximately log-linear across the top four fifths. ORs were reported per increment equivalent to ~1.1 SD: BMI per 5 kg/m²; waist-to-hip ratio per 0.06; waist-to-height ratio per 0.07; waist circumference per 12.1 cm; percent trunk fat per 7.3%; height-adjusted weight per 14 kg; hip circumference per 9.2 cm; percent body fat per 7.6%. Sex-specific analyses were conducted, with overall estimates by inverse variance–weighted average of sex-specific results. Subgroup analyses (e.g., by diabetes status) used adiposity-adjusted models. Proportional odds assumptions were checked. Software: SAS 9.4 and R 3.5.1.

- Sample and outcome distribution: 408,527 participants; mean age 56.2 years; 69% had undetectable urinary albumin; 5.0% (n=20,425) had albuminuria (uACR ≥3 mg/mmol). - Correlations: Lowest age-adjusted correlation between waist-to-hip ratio and BMI (men r=0.60; women r=0.46); other adiposity markers were more strongly correlated with BMI (r>0.75). - Waist-to-hip ratio (WHR): After confounder adjustment and measurement error correction, association with uACR was J-shaped but log-linear excluding the lowest category. Per 0.06 higher WHR, odds of being in a higher uACR category increased by 75% (95% CI 71–79%) in men and 40% (38–43%) in women; combined 55% (53–57%). After adjusting for BMI, this attenuated to 32% (30–34%); further adjustment for diabetes status, duration, vascular disease, and blood pressure reduced the odds to 24% (22–26%). - BMI: Similarly J-shaped, log-linear excluding the lowest category. Per 5 kg/m² higher BMI, odds increased by 71% (68–74%) in men and 35% (33–37%) in women; combined 47% (46–49%). After adjusting for WHR, attenuated to 35% (33–37%); after mediator adjustment (with WHR in model), 23% (22–25%). - Mediation: Approximately 40% of the association between central adiposity and albuminuria appeared mediated by diabetes, vascular disease, and blood pressure (based on attenuation with mediator adjustment). - HbA1c/diabetes status: Higher HbA1c categories (diabetes > pre-diabetes > no diabetes) were associated with progressively higher odds of albuminuria (approximately 2–3-fold higher odds for diabetes vs no diabetes). Positive, log-linear associations of both WHR and BMI with uACR were evident within each HbA1c/diabetes category; there was statistical evidence for somewhat stronger associations among those with diabetes (Ptrend <0.0001 for WHR; 0.01 for BMI). - Other adiposity measures: In mutually adjusted models, waist circumference associations strengthened after adjusting for hip circumference, while hip circumference was not associated with albuminuria after adjusting for waist circumference. - Sensitivity analyses: Results were similar when modeling albuminuria as a binary outcome and when not correcting for measurement error (directionally consistent).

The study demonstrates that both central adiposity (waist-to-hip ratio) and general adiposity (BMI) are independently and positively associated with higher levels of uACR, an early marker of CKD, even after comprehensive adjustment for confounders and mutual adjustment between adiposity types. Associations persisted across glycaemic strata and were present among individuals without diabetes, indicating that pathways beyond hyperglycaemia contribute to adiposity-related albuminuria. Partial attenuation after adjustment for diabetes, vascular disease, and blood pressure suggests these factors mediate a substantial portion (~40%) of the central adiposity association, yet residual associations imply additional mechanisms (e.g., renal haemodynamic effects, inflammation). Sex-specific differences (stronger associations in men) and the differential behavior of waist vs hip measures reinforce the relevance of fat distribution. These findings address the initial uncertainty by showing both central and general adiposity contribute independently to albuminuria, informing risk stratification and highlighting potential benefits of interventions that reduce adiposity and improve cardiometabolic health.

Higher waist-to-hip ratio and BMI are independently associated with higher odds of albuminuria in UK Biobank participants, with associations evident across glycaemic categories including those without diabetes. Approximately 40% of the central adiposity association appears mediated by diabetes, vascular disease, and blood pressure. These results emphasize the importance of both central and general adiposity as targets for prevention of early kidney damage. Future research should include longitudinal analyses to establish temporality and causality, assess impacts on CKD progression and clinical outcomes, and evaluate whether targeted reductions in central adiposity confer greater renal benefit than equivalent reductions in general adiposity.

- Cross-sectional design precludes causal inference and temporality assessment. - Single baseline adiposity and uACR measurements are subject to random measurement error and biological variability; although measurement error correction using resurvey data was applied, residual error may remain. - Proportional odds assumption was violated for urinary sodium-to-creatinine ratio; while it did not materially confound adiposity associations and was omitted from final ordinal models, modeling constraints may affect precision. - Generalizability may be limited to populations similar to UK Biobank participants; exclusions and the volunteer nature of the cohort could introduce selection bias.