Type 2 diabetes (T2D) is a complex metabolic disease influenced by genetic and environmental factors, with lifestyle choices, particularly diet, playing a significant role. While the association between high consumption of sugar-sweetened beverages and increased T2D incidence is well-established, the relationship between dietary sugar intake (as a nutrient) and T2D remains controversial. Epidemiological studies have reported inconsistent findings, showing inverse, null, or even positive associations between total or sucrose intake and T2D risk. These inconsistencies might be partly attributed to the reliance on self-reported dietary data, which is prone to bias, especially in overweight and obese individuals who tend to underreport their sugar consumption. Objective biomarkers, such as urinary sucrose, offer a potential solution for more accurately assessing sugar intake. Previous research using a subset of the EPIC-Norfolk cohort found a positive correlation between urinary sucrose and overweight/obesity, contrasting with null or inverse associations observed with self-reported sugar intake. Given that high dietary sugar intake might contribute to caloric overload, leading to obesity, a known risk factor for T2D, this study aims to investigate whether the association between sucrose intake and diabetes incidence is mediated through obesity. The study utilized data from a sub-cohort of the EPIC-Norfolk study to explore the mediating effects of BMI and waist circumference on the association between dietary sugar intake (measured via urinary sucrose biomarker, FFQ, and 7DD) and the incidence of diabetes. This investigation sought to clarify whether this relationship is primarily indirect (via obesity) or involves other direct pathophysiological mechanisms. A previous global ecological mediation analysis suggested a substantial mediating role of BMI in the association between per capita sugar intake and diabetes prevalence, but this study aims to investigate this relationship with individual-level data, providing a more refined understanding of the causal pathways involved.

Numerous observational studies have explored the relationship between dietary sugar intake and type 2 diabetes incidence, yielding inconsistent results. Some studies reported inverse associations between total sugar or sucrose intake and T2D risk, while others found null associations. A systematic review and meta-analysis suggested an overall inverse association between high sucrose intake and T2D incidence, but the certainty of this evidence was rated as low or very low. These inconsistencies highlight the limitations of relying solely on self-reported dietary data, which is susceptible to reporting bias, particularly among overweight and obese individuals who tend to underreport their consumption of unhealthy foods. The use of objective biomarkers, such as urinary sucrose and fructose, has been proposed as a more accurate method for assessing sugar intake. Previous studies using these biomarkers have revealed differing associations compared to self-reported data, suggesting that reporting bias might be a significant factor in the inconsistencies observed in previous research. The potential mediating role of obesity in the relationship between sugar intake and T2D incidence has also been discussed but not thoroughly investigated using individual-level data and mediation analysis. This study intends to address these gaps in the literature.

This prospective cohort study utilized data from a sub-cohort of the EPIC-Norfolk study. Initially, 6000 participants with baseline spot urine samples were included. After excluding participants with prevalent diabetes (n=159), missing data (n=217), implausible calorie intake (n=321), and urinary sucrose concentrations outside the acceptable range (5-500 µM) (n=2292), the final study population consisted of 2996 participants. Sucrose intake was assessed via three methods: 1. A validated food frequency questionnaire (FFQ); 2. A 7-day diet diary (7DD); and 3. An objective urinary sucrose biomarker. Incident diabetes was determined by self-reports confirmed by external sources, including general practice records, hospital data, and mortality data. Baseline covariates included age, sex, total energy intake (from FFQ and 7DD), education level, smoking status, physical activity level, family history of diabetes, BMI, and waist circumference (WC). Cox proportional hazard models were used to analyze the association between sucrose intake and diabetes incidence, adjusting for covariates. Mediation analysis, using the mediator package in R, was conducted to assess the mediating effects of BMI and WC on the relationship between urinary sucrose and diabetes incidence, using VanderWeele's method. Linear and logistic regression analyses were performed to examine the associations between sucrose intake and BMI/WC. Sensitivity analyses were performed using multiple imputation for participants with urinary sucrose levels outside the acceptable range and data from a second health check for BMI/WC. The assumption of proportional hazards was tested using the Schoenfeld residuals method and Kaplan-Meier plots. The study is reported according to the STROBE-nut checklist.

The study included 2996 participants (mean age 60.6 ± 9.5 years, 52.5% women) with a mean follow-up of 11.2 ± 2.9 years. During this period, 97 participants developed incident diabetes. Cox proportional hazard models (model 2, adjusted for age, sex, total energy intake, education, smoking status, physical activity level, and family history of diabetes) revealed inverse associations between self-reported sucrose intake (via FFQ and 7DD) and diabetes incidence. Specifically, for sucrose intake assessed via 7DD, a 37% lower incidence of diabetes was observed per 50g/d increase [HR (95% CI): 0.63 (0.43, 0.91)]. The FFQ showed a similar trend but with wider confidence intervals. In contrast, urinary sucrose showed a positive association with incident diabetes [HR (95% CI) per 100 µM increase: 1.14 (0.95, 1.36)]. This positive association was attenuated after adjusting for BMI and WC. A smoothing spline analysis indicated a positive dose-response relationship between urinary sucrose and diabetes incidence. Mediation analysis showed that BMI mediated 16% and WC mediated 22% of the association between urinary sucrose and diabetes incidence. Sensitivity analyses with imputed values for urinary sucrose outside the detection limit revealed similar findings for self-reported sucrose, but the association with urinary sucrose was less clear. In a sensitivity analysis using data from the second health check three years later, the proportion mediated for BMI was 10% and WC 12%.

This study's findings highlight the inconsistencies in the association between sucrose intake and diabetes incidence, depending on the method of assessment. The inverse association found with self-reported sucrose intake aligns with some previous studies, but might be influenced by reporting bias. The positive association observed with urinary sucrose suggests that this objective biomarker might provide a more accurate reflection of the relationship. The mediation analysis suggests that obesity plays a partial mediating role, accounting for approximately 16-22% of the association between urinary sucrose and diabetes. This implies that other direct mechanisms, such as effects on liver fat content, insulin resistance, and glycemic response, might also contribute to the association. This is in line with existing evidence linking high dietary sugar intake, particularly fructose, with increased liver fat, non-alcoholic fatty liver disease (NAFLD), and insulin resistance. The findings emphasize the limitations of relying solely on self-reported dietary information in nutritional epidemiology studies. The relatively small number of incident diabetes cases (n=97) and the substantial number of participants who died during follow-up (n=1358) may affect the precision of the estimates. The use of spot urine samples to assess urinary sucrose, which has not been thoroughly validated for predicting usual intake, also adds to the uncertainties.

This study demonstrates that sucrose intake, measured using an objective urinary biomarker, shows a positive association with incident type 2 diabetes. Obesity appears to mediate a portion of this association, but other direct mechanisms are likely involved. Larger studies with more robust assessments of sucrose intake and longer follow-up periods are needed to confirm these findings and further elucidate the complex relationship between sucrose consumption, obesity, and diabetes risk. This research highlights the importance of employing objective biomarkers to improve the accuracy and interpretation of dietary intake in epidemiological studies.

The relatively small number of incident diabetes cases (97) limited the precision of the findings. The high number of deaths during follow-up (1358) raises concerns about potential bias due to competing risks and non-independence of time to death and time to diabetes diagnosis. The use of spot urine samples for urinary sucrose assessment may not accurately reflect usual intake. A significant proportion of urinary sucrose values were below the detection limit, necessitating imputation, which might have affected the results. The single measurement of urinary sucrose at baseline also limits the ability to draw definitive conclusions about long-term effects.