Overfeeding Polyunsaturated and Saturated Fat Causes Distinct Effects on Liver and Visceral Fat Accumulation in Humans

Fredrik Rosqvist; David Iggman; Joel Kullberg; Jonathan Cedernaes; Hans-Erik Johansson; Anders Larsson; Lars Johansson; Håkan Ahlström; Peter Arner; Ingrid Dahlman; and Ulf Risérus


Diabetes. 2014;63(7):2356-2368. 

In This Article


Of the 55 participants assessed for eligibility, 41 were randomized, but 2 dropped out before the study started, leaving 39 participants with baseline data. All 39 participants completed the study (Figure 1). One individual from each group was excluded from the primary analyses due to considerable and unexplained weight loss during the intervention (>;3 SD below the mean weight gain, more than can be attributed to day-to-day variation). Including those two outliers, however, did not affect the results, except for differences between groups for the Bod Pod analyses, which were no longer statistically significant in the intention-to-treat analysis. Presented data are thus based on 37 participants who were considered compliant with the intervention. The mean age (26.7 ± 4.6 vs. 27.1 ± 3.6 years) and sex distribution (5:13 vs. 6:13 women/men, respectively) were similar between the PUFA and SFA groups. Fatty acid composition of the intervention oils is shown in Table 1 . Baseline characteristics regarding body composition are shown in Table 2 .

Figure 1.

Flow diagram for the LIPOGAIN trial.

Weight Gain, Body Composition, and Fat Oxidation

Both groups gained 1.6 kg in weight; however, the MRI assessment showed that the SFA group gained more liver fat, total fat, and visceral fat, but less lean tissue compared with subjects in the PUFA group ( Table 2 ). Relative changes are shown in Fig. 2. The ratios of lean/fat tissue gained in the PUFA and SFA groups were ~1:1 and 1:4, respectively. Pancreatic fat decreased by 31% (P = 0.008) in both groups combined, but without significant differences between groups (P = 0.75, data not shown). D-3-hydroxybutyrate decreased by 0.11 (0.15) mmol/L or −70% and 0.05 (0.09) mmol/L or −45% in the PUFA and SFA groups, respectively, without significant difference between groups (P = 0.14). When total-body water content was taken into account by using a three-compartment model for assessment of fat and lean tissue, the results remained and were even strengthened (data not shown).

Figure 2.

Relative changes in liver fat and body composition by MRI during 7 weeks of overeating SFAs or PUFAs. AF: Relative changes are calculated for each individual as change during the intervention/baseline measurement. Boxes represent medians and IQRs, whiskers represent the most extreme value besides outliers, and circles represent outliers (>1.5 IQRs outside IQR). P values represent between-group t test or Mann-Whitney U test. A: Change in liver fat is in percentage. B,C, E, and F: Changes are in liters. D: VAT/SAT ratio is calculated as VAT/(TAT − VAT).

Dietary Intake and Physical Activity

Both groups consumed on average 3.1 ± 0.5 muffins/day, equaling an additional 750 kcal/day. Both groups increased their energy intake comparably, without any differences in macronutrient intake during the study (Table 3). Food craving, hunger, and satiety showed no differences between groups (data not shown). In both groups combined, energy expenditure due to physical activity was 1,039.7 ± 112.5 kcal at baseline, and the total energy expenditure at baseline was 2,683.9 ± 245.3 kcal, without differences between groups. Physical activity did not change or differ between groups (P = 0.33) during the intervention (data not shown).

Plasma and Tissue Fatty Acid Composition

Changes in fatty acid composition in plasma as well as adipose tissue reflected dietary intakes, indicating high compliance ( Table 4 ). In addition to the dietary biomarkers, the estimated SCD-1 activity in plasma cholesterol esters was decreased by PUFAs ( Table 4 ). Changes in liver fat and visceral fat and total adipose tissue (TAT) were directly associated with changes in plasma palmitic acid, whereas liver fat and TAT were inversely associated with linoleic acid. The SCD-1 index was associated with change in liver fat. Changes in lean tissue were inversely associated with changes in palmitic acid and directly with linoleic acid (Fig. 3).

Figure 3.

Correlations between changes in outcome measures and changes in plasma cholesterol esters. White circles, PUFA group; black squares, SFA group. A, D, F, and H: 18:2 n-6 is linoleic acid (in percentage of all fatty acids by gas chromatography). B: SCD-1 index is calculated as palmitoleic/palmitic acid (in percentages of all fatty acids by gas chromatography). C: The dependent variable (change in VAT) was log transformed before analysis of Pearson r. C, E, G, and I: 16:0 is palmitic acid (in percentage of all fatty acids by gas chromatography).AI: ρ, Spearman correlation coefficient; r, Pearson correlation coefficient.


Comparison of adipose tissue gene expression between groups at baseline revealed no significant differences in gene expression (false discovery rate [FDR] 50%). Absolute differences in gene expression were calculated for each gene in each subject, comparing after with before intervention. These absolute differences in gene expression were compared between intervention groups with SAM. Twelve genes were significantly differently expressed with FDR 25% and 8 with FDR 0% ( Table 5 ). These absolute differences in gene expression were next adjusted for weight gain and compared between PUFAs and SFAs. Altogether, 20 genes were differentially regulated between groups PUFA and SFA according to SAM (FDR 25%), including the 12 genes previously discovered ( Table 5 ). Five genes that were most differently expressed between groups were selected for PCR confirmation; three genes were confirmed (carbonic anhydrase 3 [CA3]; connective tissue growth factor [CTGF]; and aldehyde dehydrogenase 1 family member A1 [ALDH1A1]), and one gene showed a trend of expression in the same direction (phosphodiesterase 8B [PDE8B]; one-sided P = 0.21). 6-Phosphofructo-2-kinase/fructose-2,6-bisphosphatase 1 could not be confirmed.

Changes in mRNA expression among several of the genes selected for PCR confirmation were associated with changes in target fatty acids in subcutaneous adipose tissue (SAT). CA3 was inversely associated with SCD-1 index (r = −0.46; P = 0.004) and directly associated with linoleic acid (r = 0.45; P = 0.006). PDE8B was directly associated with linoleic acid (r = 0.51; P = 0.002) and inversely with palmitic acid (r = −0.35; P = 0.035). CTGF was inversely but not significantly associated with linoleic acid (ρ = −0.32; P = 0.06) and directly with palmitic acid (r = 0.34; P = 0.04). ALDH1A1 was inversely associated with linoleic acid (r = −0.39; P = 0.02) and directly with SCD-1 index (r = 0.37; P = 0.03).

Glucose, Insulin, and Adiponectin

Fasting plasma glucose was 4.6 (4.4–5.0) mmol/L and 4.5 (4.3–4.9) mmol/L in PUFA and SFA groups at baseline, respectively (P = 0.69), and was virtually unchanged during the intervention: 0.06 ± 0.3 mmol/L and −0.06 ± 0.4 mmol/L in PUFA and SFA groups, respectively (P = 0.53 for difference between groups). Fasting serum insulin was 5.8 ± 2.7 and 5.0 ± 2.0 mU/L in the PUFA and SFA groups at baseline, respectively (P = 0.33), and increased to a similar extent in both groups: 0.92 ± 2.2 and 0.94 ± 1.3 in PUFA and SFA groups, respectively (P = 0.97). Homeostasis model assessment of insulin resistance was 1.23 ± 0.63 and 1.04 ± 0.43 in PUFA and SFA groups at baseline, respectively (P = 0.28), and increased to a similar extent in both groups during the intervention: 0.22 ± 0.49 and 0.18 ± 0.30 in the PUFA and SFA groups, respectively (P = 0.79). Adiponectin was 8.5 (6.1–9.6) and 6.4 (5.4–9.4) in the PUFA and SFA groups at baseline, respectively (P = 0.24), and increased by 0.92 ± 1.46 and 0.42 ± 0.94, respectively (P = 0.34).