Hepatic insulin resistance and muscle insulin resistance are characterized by distinct postprandial plasma metabolite profiles: a cross-sectional study

Overweight, obesity and related metabolic complications and chronic diseases, such as cardiovascular disease (CVD) and type 2 diabetes (T2DM), pose a massive burden to public health [1]. Insulin resistance (IR) is one of the earliest metabolic disturbances that underlies the development of many obesity-related metabolic complications [2]. Apart from its central role in glucose homeostasis, insulin is a major regulator of lipid and protein metabolism. As such, IR is commonly accompanied by lipid and lipoprotein abnormalities, although the causal and temporal relationships of these links are unclear [3].

The pathophysiology of whole-body IR is characterized by great heterogeneity, with inter-individual differences in IR severity in the various metabolic organs, including the liver and skeletal muscle. Tissue-specific IR in liver and skeletal muscle has previously been linked to distinct plasma metabolite and lipidome profiles [4, 5]. These findings may indicate that either the mechanisms causing tissue-specific IR differ between the affected tissues or that tissue-specific IR results in different metabolic disturbances. More specifically, muscle IR has been associated with lower fasting plasma concentrations of lysophosphatidylcholines, while liver IR has been associated with higher fasting plasma levels of triacylglycerols (TAG) and ketogenic amino acids, lower levels of ketone bodies, and higher diacylglycerols (DAG), the latter in women, but not in men [4, 5].

Furthermore, liver IR has been characterized by elevated postprandial total TAG in response to an oral fat load, compared to muscle IR and insulin-sensitive individuals [6]. Changes in postprandial metabolite levels reflect the complex interplay of the production, secretion, and clearance by the various metabolic organs, in particular the liver, adipose tissue, and skeletal muscle. Therefore, postprandial metabolite concentrations may provide more insights into the metabolism and functioning of these key metabolic organs than fasting metabolite levels. Importantly, early metabolic perturbations are more likely to become apparent in the postprandial state, when complex processes in these tissues act to maintain or regain homeostasis. Accordingly, postprandial metabolites, including TAGs, in the circulation are important predictors of risk for future CVD and related metabolic diseases, independent of fasting measures [7‐9].

To gain a better understanding of fasting and postprandial metabolism in tissue-specific IR, we compared fasting and postprandial plasma metabolite profiles, including lipoproteins, apolipoproteins, cholesterol, triglycerides, ketone bodies, and amino acids, in response to a high-fat mixed meal in individuals with predominant muscle IR or liver IR.

Data on plasma metabolomics were available from 230 participants: 142 individuals with muscle IR and 88 with liver IR. Sixteen participants were excluded from analyses due to statin use (muscle IR, 7.7%, n = 11; liver IR, 5.7%, n = 5), resulting in 131 individuals with muscle IR and 83 individuals with liver IR that were included in the analyses.

We investigated fasting and postprandial plasma metabolite profiles in tissue-specific IR. To this end, we measured 164 plasma metabolites, including lipoproteins, apolipoproteins, cholesterol, TAG, ketone bodies, and amino acids, for four hours after a high-fat mixed meal in individuals with predominant muscle IR or liver IR. Compared to individuals with muscle IR, individuals with liver IR had greater postprandial increases in concentrations of very large and large VLDL particles and medium-sized LDL particles, and lower postprandial plasma concentrations of IDL particles, while fasting lipoprotein profiles did not differ between IR phenotypes. In addition, individuals with liver IR had greater postprandial increases in TAG in very large, large, and small VLDL particles, all LDL subclasses, and small HDL, compared to those with muscle IR. Furthermore, postprandial plasma SFA and MUFA were higher, and total FA consisted of a larger percentage of SFA, and a lower percentage of PUFA postprandially in liver compared to muscle IR.

Elevated postprandial total TAG concentrations in liver compared to muscle IR have been reported previously [6]. To our knowledge, this study is the first to examine circulating lipoprotein subclasses and their composition in response to a high-fat mixed meal in tissue-specific IR. Compared to muscle IR, liver IR was characterized by a larger postprandial increase in plasma particle concentrations of large and very large VLDL particles, which was paralleled by larger increases in the TAG and cholesterol content of these VLDL subclasses. This greater postprandial increase can result from higher VLDL production, reduced clearance, or both. Studies that measured VLDL kinetics using stable isotope tracers have previously shown that IR, as assessed by HOMA-IR, which mainly reflects hepatic IR, was associated with increased hepatic production of large, TAG-rich VLDL [27, 28]. Hence, the greater postprandial increase in large and very large VLDL particles that we observed in people with liver IR may be due to larger hepatic VLDL production. There may be several mechanistic explanations for the greater production of large and very large VLDL particles in liver IR. Firstly, in healthy, insulin-sensitive individuals, insulin can directly inhibit VLDL production, partly by promoting the hepatic degradation of ApoB [29‐32]. Hence, in liver IR, impaired insulin-mediated suppression of VLDL assembly and secretion may contribute to elevated postprandial VLDL levels, while this suppression is likely better maintained in individuals with predominant muscle IR. Secondly, insulin can increase VLDL production by inducing de novo lipogenesis (DNL) via activation of sterol regulatory element binding protein-1c (SREBP-1c), thereby promoting lipid synthesis [3]. This hepatic insulin action appears to be (largely) preserved in liver IR [33]. Higher hepatic IR, as measured with the gold-standard hyperinsulinemic-euglycemic clamp method, has indeed been positively associated to the relative contribution of hepatic DNL to plasma TAG in TAG-rich lipoproteins, indicating elevated DNL [34]. In the present study, individuals with liver IR had greater insulin excursions in the first hour after the meal, which may have contributed to greater VLDL production by increased DNL.

In addition to elevated postprandial VLDL-TAG in liver IR, liver IR was also characterized by higher postprandial TAG content in smaller LDL and HDL particles. As this was not paralleled by larger increases in particle concentrations, these lipoprotein subclasses were likely enriched in TAG in liver IR. In line with this, HIRI was positively associated to the TAG fraction of total lipid content in the smaller LDL and HDL subclasses. A potential explanation might be enhanced transfer of TAG from TAG-rich large and very large VLDL particles to LDL and HDL particles by increased activity of the enzyme cholesteryl ester transfer protein (CETP). CETP facilitates the transfer of TAG from large VLDL to LDL and HDL in exchange for CE, resulting in TAG-enriched LDL and HDL [35]. CETP activity has been shown to be mainly determined by plasma TAG levels, rather than by IR [36, 37], which indicates that potentially increased CETP action in individuals with liver IR may be primarily attributable to the increased postprandial TAG in the circulation, and not to the hepatic IR itself.

Furthermore, liver IR was characterized by higher postprandial TAG fraction in IDL and smaller VLDL particles as compared to muscle IR. A higher TAG fraction in these particles, which are the VLDL remnants, may point towards reduced lipoprotein lipase (LPL)-mediated lipolysis of TAG in peripheral tissues in liver compared to muscle IR. IR has indeed been associated to lower LPL activity or expression in adipose tissue and skeletal muscle [38‐40]. As far as we know, LPL activity has not been investigated in tissue-specific IR yet.

In liver compared to muscle IR, the postprandial plasma FA profile was characterized by greater total FA, SFA, and MUFA concentrations, a lower percentage of PUFA of total FA, and a higher percentage of SFA of total FA after mixed-meal ingestion. These results were independent of habitual dietary intake of fat, SFA, or LA, as assessed by FFQ. DNL produces mainly SFA, which can subsequently be desaturated to MUFA in the liver [41‐43]. The higher early postprandial insulin response in individuals with liver IR may have promoted DNL as described above, thereby contributing to a greater postprandial increase in SFA. Recently, the proportion of SFA in VLDL has been reported to strongly correlate to the hepatic SFA fraction and, in turn, a higher hepatic SFA fraction to correlate to more severe hepatic IR [44]. Hence, higher hepatic SFA availability might also have contributed to the elevated postprandial plasma SFA in liver IR.

Using multivariate analysis of the postprandial metabolite curves, we additionally identified differences in postprandial metabolite curve shapes between the IR phenotypes for GlycA. GlycA is a relatively novel inflammatory biomarker that reflects the abundance of glycan groups of several acute-phase proteins [45]. Individuals with liver IR had larger postprandial GlycA increases than those with muscle IR. This difference may possibly be (partly) attributed to the observed greater postprandial TAG in liver as compared to muscle IR, since GlycA response to a mixed meal has recently been shown to strongly correlate to postprandial TAG peak [46].

Both liver and muscle IR have previously been associated with elevated fasting levels of the amino acids alanine, valine, and isoleucine, while liver IR, but not muscle IR has been additionally associated with higher circulating leucine and tyrosine, as well as lower circulating ketone bodies [5]. In the present study, we observed similar inverse associations between MISI and fasting plasma isoleucine and alanine, although these were no longer statistically significant after adjustment for multiple testing. Other associations could (also) not be replicated. An important explanation for these incongruencies may be differences in the study population: we selected individuals with predominant muscle or liver IR, thus excluding insulin-sensitive individuals and individuals with combined muscle and liver IR, resulting in a smaller range of MISI and HIRI. This may also explain why we found more metabolites to significantly differ between individuals with liver or muscle IR than in the associations with MISI and HIRI. In addition, the cohorts used by Vogelzangs et al. [5] were much larger (n = 634 and n = 540) than our study population.

It is well established that sex differences in lipid metabolism exist, which may contribute to differences in the aetiology of chronic cardiometabolic diseases between men and women [47]. Interestingly, we also observed sex differences, with a more pronounced link between liver IR and postprandial lipid profile in women compared to men. In line with this, women with liver IR had more VAT and lower whole-body insulin sensitivity compared to women with muscle IR, while these parameters did not differ between IR phenotypes in men. Although women generally have a more favourable plasma lipid profile than men, various studies indicate that in impaired metabolic health, i.e. obesity or T2DM, women have greater abnormalities in lipid and lipoprotein metabolism than men [48‐53]. Similarly, hepatic IR has previously been associated with plasma lipid abnormalities in the fasting state—including elevated TAG, DAG, and BCAA—in women, and not in men [4, 5]. Lipid metabolism in women is also affected by sex hormones and menopausal state [47]. Most women in our study (84%) were postmenopausal, and the limited number of premenopausal women (n = 21) did not allow for performing stratified analyses. The observed sex differences in the relationship between tissue-specific IR and postprandial lipoprotein profile highlight the importance of taking sexual dimorphism into account and warrant further research to elucidate underlying mechanisms.

The lipoprotein profile we observed in liver compared to muscle IR—elevated postprandial TAG in larger VLDL, LDL, and smaller HDL particles—is common in IR and T2DM [30, 31, 54, 55]. Such a lipid profile is considered to be highly atherogenic and has consistently been associated with increased CVD risk [7‐9, 56, 57]. Our findings show that this postprandial lipid profile is specifically related to liver IR, and less so to muscle IR, despite similar body fat percentage, liver fat, and whole-body insulin sensitivity, and lower systemic low-grade inflammation, as indicated by lower circulating CRP levels. Humans typically spend the majority of the day in the postprandial state due to frequent eating occasions. Even in healthy, insulin-sensitive individuals, plasma TAG levels progressively rise throughout the day upon repeated meal consumption, returning only to fasting levels during sleep [58]. Thus, individuals with liver IR may be at increased risk of developing cardiometabolic disease, compared to individuals with muscle IR.

Interestingly, the more unfavorable lipoprotein profile in liver compared to muscle IR was not observed in the fasting state. Differences between IR phenotypes only became apparent after challenging homeostasis with a high-fat mixed meal. In both individuals with liver and muscle IR, the majority had fasting TAG concentrations in the normal range (< 1.7 mmol/L): 72% in muscle IR and 64% in liver IR. In addition, a large majority was normoglycemic: 77% in muscle IR and 78% in liver IR. These findings thus indicate that liver IR, in particular, is accompanied by early perturbations in postprandial lipid metabolism that are not evident in the fasting state yet compared to muscle IR. Detection of metabolic perturbations at this early stage—before the onset of overt metabolic disease—provides an opportunity for timely prevention of progression to cardiometabolic disease by lifestyle interventions such as dietary modification, exercise, and weight loss.

A major strength of this study is the extensive metabolic profiling of tissue-specific IR in both the fasting and the postprandial state, thereby broadening and deepening the characterization of plasma lipid profiles in tissue-specific IR and providing more insights into the metabolic abnormalities that are related to muscle or liver IR. It is as of yet unknown whether dysregulated lipid metabolism is a cause or consequence of hepatic IR. Due to the cross-sectional design of this study, we cannot make any causal inferences about the nature of the observed relationship between tissue-specific IR and postprandial plasma lipid profiles. Another limitation of this study is the use of OGTT-derived measures for the assessment of tissue-specific IR. Postprandial glycemic and insulin responses are also affected by gastrointestinal factors such as gastric emptying and the incretin response [59]. Therefore, MISI and HIRI provide a less precise estimation of tissue-specific IR compared to the gold-standard two-step hyperinsulinemic-euglycemic clamp. Nevertheless, these indices have been validated against the gold-standard clamp [14] and we have previously shown that using these OGTT-derived measures, we could identify distinct metabolic phenotypes in various cohorts [4, 5, 60]. In addition, our study population consisted only of individuals with some degree of tissue-specific IR. Therefore, we cannot conclude anything on postprandial metabolic profiles in tissue-specific IR as compared to healthy, insulin-sensitive controls. Finally, we sampled blood until four hours after consumption of the meal, at which many plasma lipids are at their peak. Future studies would benefit from longer sampling times until 6–8 h post-meal to allow for examination of the (rate of) return to fasting levels or the effects of a second meal.

In conclusion, individuals with liver IR or muscle IR have distinct postprandial plasma metabolite responses after a mixed meal, despite similar fasting metabolite profiles, body fat percentage, and whole-body insulin sensitivity. Liver IR was characterized by greater postprandial increases in larger VLDL particles, as well as in TAG in the larger VLDL and the smaller LDL and HDL subclasses, compared to muscle IR, which points towards more impaired hepatic lipid metabolism in liver compared to muscle IR. Therefore, improving postprandial lipid metabolism with lifestyle modifications to prevent the development of cardiometabolic disease may be particularly important for individuals with predominant liver IR.