Abdominal Fat and Metabolic Health Markers but Not PNPLA3 Genotype Predicts Liver Fat Accumulation in Response to Excess Intake of Energy and Saturated Fat in Healthy Individuals

Abstract

Background: Saturated fat (SFA) has consistently been shown to increase liver fat, but the response appears variable at the individual level. Phenotypic and genotypic characteristics have been demonstrated to modify the hypercholesterolemic effect of SFA but it is unclear which characteristics that predict liver fat accumulation in response to a hypercaloric diet high in SFA. Objective: To identify predictors of liver fat accumulation in response to an increased intake of SFA. Design: We pooled our two previously conducted double-blind randomized trials (LIPOGAIN and LIPOGAIN-2, clinicaltrials.gov NCT01427140 and NCT02211612) and used data from the n = 49 metabolically healthy men (n = 32) and women (n = 17) randomized to a hypercaloric diet through addition of SFA-rich muffins for 7-8 weeks. Associations between clinical and metabolic variables at baseline and changes in liver fat during the intervention were analyzed using Spearman rank correlation. Linear regression was used to generate a prediction model. Results: Liver fat increased by 33% (IQR 5.4-82.7%; P < 0.0001) in response to excess energy intake and this was not associated (r = 0.17, P = 0.23) with the increase in body weight (1.9 kg; IQR 1.1-2.9 kg). Liver fat accumulation was similar (P = 0.28) in carriers (33%, IQR 14-79%) and non-carriers (33%, IQR -11 to +87%) of the PNPLA3-I148M variant. Baseline visceral and liver fat content, as well as levels of the liver enzyme γ-glutamyl transferase (GT), were the strongest positive predictors of liver fat accumulation-in contrast, adiponectin and the fatty acid 17:0 in adipose tissue were the only negative predictors in univariate analyses. A regression model based on eight clinical and metabolic variables could explain 81% of the variation in liver fat accumulation. Conclusion: Our results suggest there exists a highly inter-individual variation in the accumulation of liver fat in metabolically healthy men and women, in response to an increased energy intake from SFA and carbohydrates that occurs over circa 2 months. This marked variability in liver fat accumulation could largely be predicted by a set of clinical (e.g., GT and BMI) and metabolic (e.g., fatty acids, HOMA-IR, and adiponectin) variables assessed at baseline.

Keywords: NAFLD; fatty acids; liver fat; overfeeding; saturated fat.

Conflict of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Rosqvist, Orho-Melander, Kullberg, Iggman, Johansson, Cedernaes, Ahlström and Risérus.

Figures

Figure 1

(A) Correlation between the change in body weight and change in liver fat accumulation (n = 49), (B) changes in liver fat content and body weight at the individual level, sorted according to the relative change in liver fat content (n = 49).

Figure 2

Clinical and metabolic predictors of liver fat accumulation in univariate analyses. ALT, alanine aminotransferase; BMI, body mass index; HOMA-IR, homeostatic model assessment of insulin resistance; LDL, low density lipoprotein; HDL, high density lipoprotein; Apo, apolipoprotein; VAT, visceral adipose tissue; SAT, subcutaneous adipose tissue; BP, blood pressure; FGF21, fibroblast growth factor; TSH, thyroid stimulating hormone; PCSK9, proprotein convertase subtilisin/kexin type 9; CRP, C-reactive protein.

Figure 3

Fatty acid abundances at baseline as predictors of liver fat accumulation in univariate analyses. (A) Plasma cholesterol esters, (B) plasma phospholipids, (C) subcutaneous adipose tissue. The bars represent the Spearman rank correlation (written beside the bars). SCD; stearoyl-CoA desaturase; D6D, delta-6 desaturase; D5D, delta-5 desaturase.

Figure 4

Actual by predicted plot (linear regression) for change in liver fat content and the eight individual predictors used in the regression model. All variables significantly associated with liver fat accumulation in univariate analyses were initially included in the regression model. To avoid multicollinearity, the variable list was sequentially culled based on the highest variance inflation factor (VIF) until all included variables had a VIF P-value until all included variables had a P < 0.01. AT, adipose tissue; BP, blood pressure; VIF, variance inflation factor; HOMA-IR, homeostatic model assessment of insulin resistance; BMI, body mass index.