DOI: https://pmc.ncbi.nlm.nih.gov/articles/PMC13626808/

Effect of a walnut-based low carbohydrate diet on fat redistribution and visceral fat accumulation in people with central obesity: A randomized controlled trial

Abstract

Central obesity remains a major cardiometabolic risk state, with excess abdominal and visceral adiposity linked more strongly to diabetes and cardiovascular outcomes than overall adiposity. This 12-week randomized controlled trial tested whether replacing staple carbohydrates with walnuts within a structured diet could improve fat distribution and reduce visceral fat accumulation in adults with central obesity in China. The trial recruited participants aged 18 to 60 years who met Chinese central obesity criteria, defined by waist circumference at least 90 cm in men and at least 85 cm in women, with BMI between 24 kg/m2 and 32.5 kg/m2. Participants were assigned to a walnut-based low carbohydrate diet or a high-carbohydrate low-fat diet after a one-week nut washout period.

The primary between-group signal came from the time course of fat distribution and visceral fat change, not endpoint separation at 12 weeks. The Group × Time interaction was significant for TLR, eta squared = 0.071, p = 0.039, VFA, eta squared = 0.093, p = 0.014, and BMI, eta squared = 0.168, p = 0.001, while WHR showed a near-significant interaction, eta squared = 0.052, p = 0.072. Within the walnut group, TLR fell with a large effect, eta squared = 0.174, p = 0.001, VFA fell with a large effect, eta squared = 0.393, p < 0.001, and BMI fell with a large effect, eta squared = 0.399, p < 0.001. At 12 weeks, GLP-1 was higher in the walnut group, 2.10 pmol/L [1.84, 2.69] vs 1.59 pmol/L [0.86, 2.14], Z = -2.343, p = 0.019, PYY was higher, 33.1 ± 5.50 pg/mL vs 26.8 ± 5.93 pg/mL, t = 3.663, p = 0.001, and C-peptide was lower, 1.11 [0.73, 1.29] ng/mL vs 1.61 [1.01, 3.64] ng/mL, Z = -2.242, p = 0.025.

Study Design and Methodology

• Design: 12-week prospective randomized controlled trial
• Population: 80 adults with central obesity randomized 1:1, 40 to w-LCD and 40 to HCD
• Final analyzed sample: 35 in w-LCD, 29 in HCD
• Duration: 12 weeks
• Recruitment period: December 2022 to October 2023
• Allocation: computer-generated random number table, concealed by a non-principal researcher
• Blinding: allocation concealment was described, participant and dietary blinding was not reported
• Washout: 1-week purge period avoiding all nut supplementation before intervention
• Diet control: structured meal plan based on 2021 Chinese Medical Nutrition Therapy guidance
• Intervention contrast: HCD used the guideline-based low-fat, high-carb regimen, w-LCD replaced 150 g/day of staple food with walnuts
• Measurements: InBody720 bioelectrical impedance for TLR, VFA, BMI, BF%, and trunk and leg fat; tape-measured WC and HC for WHR; fasting GLP-1, PYY, and C-peptide by ELISA; 3-day diet records; IPAQ for physical activity
• Follow-ups: baseline, 1 week, 4 weeks, 8 weeks, 12 weeks
• Statistics: repeated-measures ANOVA, t tests, Mann-Whitney tests, chi-square tests, ITT with LOCF

Key Findings

• TLR improved over time in w-LCD, with a significant Group × Time interaction, eta squared = 0.071, p = 0.039
• VFA improved over time in w-LCD, with a significant Group × Time interaction, eta squared = 0.093, p = 0.014
• BMI improved over time in w-LCD, with a significant Group × Time interaction, eta squared = 0.168, p = 0.001
• WHR trended lower, interaction eta squared = 0.052, p = 0.072
• Within w-LCD, TLR fell from 1.80 ± 0.19 to 1.74 ± 0.20, eta squared = 0.174, p = 0.001
• Within w-LCD, VFA fell from 113 ± 33.4 cm2 to 97.4 ± 31.7 cm2, eta squared = 0.393, p < 0.001
• Within w-LCD, BMI fell from 27.6 ± 2.49 to 26.1 ± 2.63 kg/m2, eta squared = 0.399, p < 0.001
• Between-group comparison at 12 weeks did not separate significantly for TLR, VFA, or BMI, despite the stronger trajectory in w-LCD
• GLP-1 was higher in w-LCD at 12 weeks, p = 0.019
• PYY was higher in w-LCD at 12 weeks, p = 0.001
• C-peptide was lower in w-LCD at 12 weeks, p = 0.025
• Dietary adherence showed the macronutrient shift was achieved: w-LCD maintained roughly 39% carbohydrate and 42% fat, HCD maintained roughly 57% carbohydrate and 24% fat
• Energy intake fell in both groups over time, from 1746 ± 412 to 1162 ± 224 kcal/day in w-LCD and from 1892 ± 512 to 1211 ± 326 kcal/day in HCD
• Group effect for carbohydrate intake was significant, eta squared = 0.184, p = 0.001
• Group effect for fat intake was significant, eta squared = 0.155, p = 0.002
• Fat intake had a strong Group × Time interaction, eta squared = 0.290, p < 0.001
• Within w-LCD, fat intake rose from 52.1 ± 18.5 to 55.2 ± 11.2 g/day, while HCD fell from 54.7 ± 21.3 to 31.3 ± 11.3 g/day

Limitations

• Small sample size with substantial attrition, 35 vs 29 completed
• Unequal dropout, higher in HCD than w-LCD
• Short duration, only 12 weeks
• Self-reported dietary records introduce reporting bias
• No strict isocaloric control, so energy matching was imperfect
• No baseline C-peptide
• No dynamic postprandial hormone testing
• No insulin resistance or inflammatory biomarkers
• No stratified subgroup analysis by sex, baseline obesity severity, or lipid profile
• Walnut exposure was partially self-funded, which can affect adherence and external validity

Discussion and Implications

This trial shifts the nutrition conversation from “low-fat versus low-carb” toward carbohydrate quality and macronutrient substitution. Replacing staple carbohydrate with walnuts produced a stronger reduction trajectory for trunk fat distribution and visceral adiposity than the comparator diet, while also shifting fasting GLP-1, PYY, and C-peptide in the expected direction. The data support walnut inclusion as a practical food-based strategy inside a carbohydrate-restricted framework, not as a simple calorie addition.

Conclusion

Overall, the study found that swapping some staple carbs for walnuts led to better reductions in belly fat patterning and visceral fat over 12 weeks, while also shifting appetite-related hormones in a favorable direction. The results are promising, but this is still a small, short trial, so it needs larger and longer follow-up before anyone treats it as settled practice.

DOI: https://doi.org/10.3390/nu18183072

Late-Life Dietary Patterns and Subsequent Survival Among Adults Aged 90 Years or Older in the Sardinian Blue Zone

This prospective cohort study asked a narrower question than most longevity papers: not what people ate across life to reach extreme old age, but whether diet measured after age 90 related to survival from that point forward. The authors enrolled 200 Sardinian Blue Zone residents aged 90 to 106 years and followed them for up to 7 years to address a gap in the longevity literature, which has leaned heavily on ecological and cross-sectional data. Baseline diet was captured with a qualitative food-frequency questionnaire covering 17 foods or food groups, and participants were grouped using principal component analysis into empirical dietary patterns.

The main signal was a dairy–traditional-fat pattern built around milk, cheese, and lard, with lower olive oil and coffee. Higher adherence to that pattern tracked with longer survival: per 1-SD increase, mortality hazard fell to HR = 0.839 (95% CI: 0.727 to 0.969, p = 0.017), but that pattern-level result did not survive false-discovery-rate correction (q = 0.066). In tertile analyses, the highest versus lowest tertile showed HR = 0.682 (95% CI: 0.480 to 0.969), and median survival rose from 2.3 years in the lowest tertile to 4.3 years in the highest. Of the individual foods, milk (HR = 0.735, q = 0.0014), cheese (HR = 0.775, q = 0.0077), and chicken (HR = 0.811, q = 0.0275) were linked to lower mortality, while wine was linked to higher mortality (HR = 1.336, q = 0.0022). None of the other dietary patterns, including red-meat, chicken–carbohydrate, or plant–starchy food, showed a significant association with survival.

Study design and methodology

• Design: prospective cohort study
• Sample size: N = 200
• Population: Sardinian Blue Zone residents aged 90 to 106 years
• Follow-up: up to 7 years; mean follow-up 3.27 ± 1.81 years
• Outcome: all-cause mortality
• Diet assessment: face-to-face qualitative FFQ with 17 foods or food groups
• Pattern derivation: principal component analysis with Varimax rotation
• Covariate adjustment: age, sex, and comorbidity burden using a CIRS-derived count; sensitivity models also included MMSE, BADL, calf circumference, physical activity, and recruitment wave
• Controls used: no separate control group, this was an observational cohort
• Metabolic ward: no
• Blinding: no
• Washout period: no

Key findings

• Four dietary patterns emerged:
  • Dairy–traditional-fat
  • Red-meat
  • Chicken–carbohydrate
  • Plant–starchy food
• The dairy–traditional-fat pattern was the only one consistently linked to lower mortality.
  • Continuous score: HR = 0.839 (95% CI: 0.727 to 0.969, p = 0.017)
  • Highest vs lowest tertile: HR = 0.682 (95% CI: 0.480 to 0.969)
  • Median survival: 4.3 years vs 2.3 years in the highest vs lowest tertile
• The association with the dairy pattern did not meet the study’s FDR threshold:
  • q = 0.066
• No significant mortality association appeared for the other patterns:
  • Red-meat: HR = 1.043, p = 0.566
  • Chicken–carbohydrate: HR = 0.984, p = 0.821
  • Plant–starchy food: HR = 0.993, p = 0.928
• In mutually adjusted models, the dairy pattern stayed associated with lower mortality:
  • HR = 0.832
  • 95% CI: 0.718 to 0.964
  • p = 0.014
• In sensitivity analyses, the dairy-pattern result stayed broadly similar after extra adjustment for cognitive status, function, body size, physical activity, and recruitment wave:
  • Example: HR = 0.832, p = 0.012 with physical activity and recruitment wave included
• Excluding early deaths did not erase the signal, though it weakened:
  • Excluding deaths within 6 months: HR = 0.852, p = 0.031
  • Excluding deaths within 1 year: HR = 0.865, p = 0.063
• Food-level analysis showed:
  • Milk: HR = 0.735, q = 0.0014
  • Cheese: HR = 0.775, q = 0.0077
  • Chicken: HR = 0.811, q = 0.0275
  • Wine: HR = 1.336, q = 0.0022

Limitations

• Small sample size: N = 200
• Observational design, so causality can’t be inferred
• Diet was measured only at baseline
• Qualitative FFQ, so no portion sizes or total energy intake
• Residual confounding is a real issue, especially from frailty, appetite loss, swallowing problems, and functional decline
• Reverse causation is plausible in this age group
• No blinded design, no intervention, no washout
• No metabolic ward, so intake wasn’t tightly controlled
• Participants were highly selected survivors who had already reached age 90 or older
• Attrition wasn’t the issue here, since all participants died by the end of follow-up, but follow-up timing still varied

Discussion and implications

This paper adds an important distinction that longevity research often blurs: diet linked to reaching old age is not the same as diet linked to surviving after age 90. In this cohort, a dairy-heavy late-life pattern tracked with lower mortality, and the food-specific results lined up with that pattern, especially for milk and cheese.

The wine finding also matters because it cuts against the common habit of treating alcohol, especially traditional wine intake, as automatically beneficial in very old age. The authors are careful not to overclaim causality, and that caution is warranted. Still, the consistency of the dairy signal across the main, categorical, and sensitivity analyses gives the result real weight, even if multiple-testing correction weakens the formal significance.

This study points toward dairy foods, especially milk and cheese, as markers of better late-life survival in a very old Sardinian cohort. It does not prove dairy extends life, but it does show that diet still matters after age 90, and the pattern is worth following up in larger prospective studies.