Long-term aspartame and saccharin intakes are related to greater volumes of visceral, intermuscular, and subcutaneous adipose tissue: the CARDIA study

The study addresses whether habitual, long-term intake of artificial sweeteners (aspartame, sucralose, saccharin) and diet beverages is associated with greater adipose tissue (AT) deposition and adverse anthropometric outcomes, independent of total caloric intake and diet quality. Contextually, despite recommendations by major health organizations to replace added sugars with artificial sweeteners, emerging evidence suggests artificial sweeteners may not be metabolically benign and could contribute to adiposity and obesity. The purpose is to evaluate associations of total and individual artificial sweeteners and diet beverages with computed tomography–derived adipose tissue depot volumes (visceral, intermuscular, subcutaneous), anthropometric measures, their 25-year changes, and incident obesity among young adults in the CARDIA cohort. The hypothesis is that higher long-term intakes of total and individual artificial sweeteners and diet beverages are associated with higher VAT, IMAT, SAT, higher BMI/weight/waist circumference, greater increases in these measures over 25 years, and higher risk of incident obesity, independent of demographic and lifestyle factors, energy intake, and diet quality.

Observational evidence has linked artificial sweetener intake to higher BMI, waist circumference, and obesity incidence, though findings are mixed. Meta-analyses of observational studies report positive associations between artificial sweetener/diet beverage consumption and adiposity measures. Experimental cell and animal models suggest potential adipogenic effects of artificial sweeteners through mechanisms such as hyperphagia and alterations in intestinal enzyme activity. Conversely, randomized controlled trials (typically short-term and in controlled settings with caloric restriction or sugar replacement) have shown modest weight loss benefits of replacing sugars with artificial sweeteners, but these may not reflect long-term, real-world consumption. Prior cohort studies have shown diet beverage intake associated with greater waist circumference and BMI over follow-ups of up to ~10 years. No prior studies have captured long-term habitual intakes of individual artificial sweeteners while controlling for diet quality and total energy over decades together with CT-derived adipose depots, motivating the present investigation.

Design and population: Prospective analysis within the CARDIA cohort, which enrolled 5115 Black and White adults aged 18–30 years at baseline (1985–1986) from four US sites. For year 25 body composition analyses, 3276 attended; after exclusions for missing diet data (n=5), implausible energy intake (<600 or >6000 kcal/d for women; <800 or >8000 kcal/d for men; n=32), missing CT data (n=167), or other missing data (n=2), 3088 participants remained (869 Black women, 867 White women, 590 Black men, 762 White men). For incident obesity analyses, those with baseline obesity (BMI ≥30 kg/m²) were excluded, yielding 2745 participants. Exposure assessment: The validated CARDIA Diet History interview at years 0, 7, and 20 assessed usual intake over the prior month, including brand names and nutrient modifications. Nutrition Data System for Research software quantified intakes of artificial sweeteners (mg/day: aspartame, saccharin, sucralose) and food/beverage groups including diet beverages (servings/day; 1 serving = 8 fl oz). A time-averaged intake across years 0, 7, and/or 20 was computed; categories were formed as quintiles for total artificial sweetener, aspartame, sucralose, and diet beverages; saccharin was analyzed in tertiles due to many non-consumers. Diet quality was assessed using the Healthy Eating Index-2015. Covariates: Demographics (age, sex, race, field center, education), lifestyle (smoking, alcohol status, physical activity score), height, energy intake, and HEI-2015 were collected and included as adjustments. Outcome assessment: At year 25, multidetector CT quantified abdominal adipose tissue and muscle composition. VAT, SAT, and IMAT volumes were measured using MIPAV software from contiguous 1–1.25 mm slices around L4–L5 (AT) and L3–L4 (muscles). Inter-reader reliability correlations were >0.95 for psoas muscle volume and 0.99 for VAT. Anthropometry (BMI, weight, waist circumference) was measured at exams; 25-year changes were computed from baseline to year 25. Statistical analysis: SAS 9.4 was used. AT volume distributions were checked for normality. Linear regression estimated associations of time-averaged intakes (quintiles/tertiles) with year 25 AT volumes, anthropometry, and 25-year changes, adjusting for the covariates above; changes additionally adjusted for baseline values. Cox regression with time-varying exposures estimated hazard ratios for incident obesity (BMI ≥30 kg/m²) over a median 17.5-year follow-up with the same covariate adjustments. Effect modification by sex or race was tested and not significant. Correlation between artificial sweetener intake (mg/d) and diet beverage intake (servings/d) was 0.33 (p<0.001).

- Higher time-averaged total artificial sweetener intake was associated with greater CT-derived adipose tissue volumes at year 25: VAT ptrend=0.001, SAT ptrend<0.001, IMAT ptrend<0.001. Comparing extreme quintiles, VAT was 8.8% higher, SAT 14.8% higher, and IMAT 13.3% higher in the top quintile versus bottom. - Higher aspartame intake showed similar associations: VAT ptrend<0.001, SAT ptrend<0.001, IMAT ptrend<0.001; top vs bottom quintile differences were 8.4% (VAT), 11.6% (SAT), and 10.6% (IMAT) higher. - Higher saccharin intake (tertiles) was associated with higher AT volumes: VAT ptrend=0.001, SAT ptrend<0.001, IMAT ptrend<0.001; top vs bottom tertile differences were 10.3% (VAT), 14.0% (SAT), and 10.0% (IMAT) higher. - Diet beverage intake (quintiles) was positively associated with AT volumes: VAT ptrend<0.001, SAT ptrend<0.001, IMAT ptrend<0.001; top vs bottom quintile differences were 10.4% (VAT), 14.1% (SAT), and 14.8% (IMAT) higher. - Sucralose intake was not associated with AT volumes or 25-year changes in anthropometry, but was associated with higher BMI (ptrend=0.004), weight (ptrend<0.001), and waist circumference (ptrend=0.05). - Anthropometry: Total artificial sweetener, aspartame, saccharin, and diet beverage intakes were associated with higher year 25 BMI, weight, and waist circumference (all ptrend<0.001) and with greater 25-year increases in these measures (ptrend≤0.03 across exposures; saccharin showed ptrend=0.06 for BMI change). - Incident obesity: Over a median 17.5 years, risks increased across exposure categories. Top vs bottom quintiles showed 78% higher risk for total artificial sweetener, 64% higher for aspartame, and 57% higher for diet beverage intake. Sucralose showed a significant trend (ptrend=0.018). Saccharin tertiles showed modestly higher risks (13% and 19% in 2nd and 3rd tertiles) but a non-significant overall trend (ptrend=0.13). - Baseline characteristics across higher total artificial sweetener intake quintiles included higher proportions of women and White participants, higher education, higher physical activity, higher energy intake, and slightly higher BMI, waist circumference, and fasting insulin.

The findings support the hypothesis that long-term consumption of artificial sweeteners (particularly aspartame and saccharin) and diet beverages is associated with greater visceral, intermuscular, and subcutaneous adipose tissue volumes and adverse anthropometric profiles, independent of total energy intake and diet quality. The dose–response patterns across intake categories, temporality from repeated exposure measures over decades, and consistency across multiple adiposity outcomes strengthen the inference that these exposures may contribute to adipose tissue accumulation. The lack of association between sucralose and AT volumes suggests heterogeneity in metabolic effects among artificial sweeteners. These results align with prior observational studies linking artificial sweeteners to higher BMI and waist circumference, while contrasting with short-term RCTs showing modest weight loss when artificial sweeteners replace sugar under controlled conditions, highlighting potential differences between real-world long-term use and trial settings. Potential mechanisms discussed include hyperphagia (supported by higher energy intake with higher sweetener use) and compound-specific effects such as saccharin-induced adipogenesis and aspartame-mediated suppression of intestinal alkaline phosphatase leading to endotoxemia and inflammation that promote adipogenesis. Given the cardiometabolic risks linked to VAT and IMAT, these associations are clinically relevant.

Habitual, long-term intakes of total and specific artificial sweeteners (aspartame and saccharin) and diet beverages are associated with greater VAT, IMAT, and SAT volumes and with higher BMI, body weight, waist circumference, and their 25-year increases, independent of energy intake and diet quality. Sucralose was not associated with AT volumes. Higher intakes of total artificial sweeteners, aspartame, sucralose, and diet beverages were also associated with increased risk of incident obesity. These results suggest that replacing added sugars with certain artificial sweeteners may not mitigate adiposity-related risk and may have adverse consequences. Future research should clarify causal mechanisms (including gut microbiome and intestinal enzyme pathways), assess differential effects among sweetener types, evaluate objective biomarkers of long-term intake, and examine generalizability across diverse populations and dietary patterns.

- Potential for reverse causality: individuals with higher adiposity might preferentially choose artificial sweeteners or diet beverages; analyses adjusted for energy intake and diet quality, used long follow-up and repeated exposure measures, but residual reverse causality cannot be excluded. - Self-reported dietary intake introduces measurement error; lack of validated objective biomarkers for long-term artificial sweetener intake likely biases toward the null. - Microbiome-mediated effects and individualized glycemic responses to artificial sweeteners could not be assessed. - Diet instruments did not capture newer sweeteners (e.g., stevia, advantame), limiting exposure characterization. - Generalizability may be limited; findings should be examined in other populations and dietary contexts. - Missing mapping for some specific affiliations not pertinent to analysis; however, CT-based AT measures had high reliability. - Saccharin categorized in tertiles due to many non-consumers, which may reduce granularity of dose–response assessment.