Dissociation between liver fat content and fasting metabolic markers of selective hepatic insulin resistance in humans

Abstract

Objective: Fasting hyperglycemia and hypertriglyceridemia are characteristic of insulin resistance (IR) and rodent work has suggested this may be due to selective hepatic IR, defined by increased hepatic gluconeogenesis and de novo lipogenesis (DNL), but this has not been shown in humans.

Design: Cross-sectional study in men and women across a range of adiposity.

Methods: Medication-free participants (n = 177) were classified as normoinsulinemic (NI) or hyperinsulinemic (HI) and as having low (LF) or high (HF) liver fat content measured by magnetic resonance spectroscopy. Fractional gluconeogenesis (frGNG) and hepatic DNL were measured using stable isotope tracer methodology following an overnight fast.

Results: Although HI and HF groups had higher fasting plasma glucose and triglyceride concentrations when compared to NI and LF groups respectively, there was no difference in frGNG. However, HF participants tended to have lower frGNG than LF participants. HI participants had higher DNL compared to NI participants but there was no difference observed between liver fat groups.

Conclusions: Taken together, we found no metabolic signature of selective hepatic IR in fasting humans. DNL may contribute to hypertriglyceridemia in individuals with HI but not those with HF. Glycogenolysis and systemic glucose clearance may have a larger contribution to fasting hyperglycemia than gluconeogenesis, especially in those with HF, and these pathways should be considered for therapeutic targeting.

Keywords: de novo lipogenesis; gluconeogenesis; hyperinsulinemia; insulin resistance; liver fat.

Graphical Abstract

Compared to participants with normoinsulinemia, those with hyperinsulinemia had higher de novo lipogenesis which may explain their elevated plasma triglyceride concentration. An increase in both gluconeogenesis and glycogenolysis (as well as a possible decrease in glucose clearance) may drive the higher plasma glucose levels in those with hyperinsulinemia compared to normoinsulinemia. Participants with high liver fat content had no change in either de novo lipogenesis or gluconeogenesis compared to those with low liver fat content, and the elevated plasma glucose and triglyceride concentrations may be driven by increased glycogenolysis and triglyceride secretion from lipid droplets, respectively.

Figure 1.

The effect of fasting plasma insulin and liver fat content on fasting GNG, GLY, and DNL. Participants (n = 177) were classified as normoinsulinemic or hyperinsulinemic (NI or HI) and as having low or high liver fat content (LF or HF). (A and B) Fractional gluconeogenesis (frGNG) was measured across the groups, and (C and D) the absolute contributions of gluconeogenesis (GNG) and glycogenolysis (GLY) to fasting plasma glucose concentration were calculated. (E and F) Fractional de novo lipogenesis across groups (frDNL) and (G and H) absolute contributions of newly synthesized 16:0 to very-low-density lipoprotein-triglyceride (VLDL-TG). *P < .05, **P < .01, ***P < .001, ****P < .0001, or P-value stated.

Figure 2.

Participants were grouped as normoinsulinemic with low liver fat content (NILF), normoinsulinemic with high liver fat content (NIHF), hyperinsulinemic with low liver fat content (HILF), or hyperinsulinemic with high liver fat content (HIHF). (A) Fractional gluconeogenesis (frGNG) was measured, and the absolute contributions of (B) glycogenolysis (GLY) and (C) gluconeogenesis (GNG) to fasting plasma glucose concentration were calculated. (D) Fractional de novo lipogenesis (frDNL) across groups and (E) absolute contributions of newly synthesized 16:0 to very-low-density lipoprotein-triglyceride (VLDL-TG). (F) A correlation was performed across all participants between fractional gluconeogenesis (frGNG) and fractional de novo lipogenesis (frDNL). *P < .05, **P < .01, ***P < .001, and ****P < .0001 compared to either NILF or LGLD group and ^##P < .01, ^###P < .001, and ^####P < .0001 for all other comparisons shown or P-value stated.