Introduction

figure a

Plasma NEFA concentrations are higher in obese individuals [1] and this may contribute to the pathogenesis of obesity-associated insulin resistance and beta cell dysfunction [2,3,4,5]. However, the distinct and potentially opposite effects of individual NEFA on insulin action and secretion are still poorly understood.

The fatty acid palmitoleate (16:1 n-7), also known as 9-hexadecenoic acid, is the second most abundant monounsaturated fatty acid (MUFA) in human blood and adipose tissue [6]. The two main sources of circulating palmitoleate are endogenous fat synthesis (cis isomer) and dietary whole-fat dairy products (trans isomer). Both palmitoleate isomers have been associated with lower metabolic risk [7,8,9,10,11,12,13]. The enrichment of cis-palmitoleate (hereinafter simply referred to as palmitoleate) in the plasma NEFA pool depends on its endogenous synthesis by subcutaneous adipose tissue, from which palmitoleate can be readily mobilised in response to different metabolic stimuli [9]. Recent studies in rodent models and cell cultures have reported that palmitoleate can directly enhance whole-body glucose disposal [7, 14,15,16,17], attenuate hepatosteatosis induced by a high-fat diet or diabetes [7, 14, 18] and protect beta cells from palmitate-induced apoptosis [19]. These observations support the physiological relevance of palmitoleate as an adipocyte-derived lipid hormone (or ‘lipokine’) by which the adipose tissue can regulate systemic metabolism [7,8,9,10,11,12].

In humans, the potential metabolic benefits of palmitoleate are controversial. A positive relationship between palmitoleate levels and insulin sensitivity has been observed in some studies [8,9,10], but others have reported no or even negative correlations [4, 20,21,22,23]. These discrepancies may be due to small sample sizes, inadequate assessment of insulin sensitivity and beta cell function, and lack of longitudinal data. More importantly, few studies have been able to provide detailed adjustment for potential confounders, such as total NEFA concentration and composition.

The aim of this study was to investigate whether circulating palmitoleate is an independent determinant of insulin sensitivity and beta cell function in humans. To this end, plasma concentration and relative enrichment of the most abundant NEFA were measured by a targeted quantitative metabolomic approach in a large cohort of non-diabetic individuals at baseline and after a 3 year follow-up. Insulin sensitivity and beta cell function variables were determined by means of the hyperinsulinaemic–euglycaemic clamp and by mathematical modelling of plasma glucose and C-peptide concentrations during a 2 h OGTT.

Methods

Study design and participants

The Relationship between Insulin Sensitivity and Cardiovascular disease (RISC) study is a multicentre, prospective, observational study involving 1566 participants at baseline and 1059 participants at follow up between 30 and 60 years of age, recruited in 19 centres in 14 European countries [24]. Major exclusion criteria were diabetes, dyslipidaemia, hypertension, class III obesity, chronic lung, hepatic or kidney diseases and neoplastic and inflammatory diseases. Information regarding medical history, drug use and family history of diabetes (i.e. any first-degree family member with type 2 diabetes) was collected using standardised self-reported questionnaires. The examination protocol included anthropometry, blood pressure measurements, a fasting blood test, a euglycaemic−hyperinsulinaemic clamp at baseline and an OGTT at baseline and after a 3 year follow-up. Anthropometric, clinical and biochemical measurements as well as metabolic tests were performed in all centres according to standardised methods [24].

For the purpose of this study, we excluded individuals whose fasting blood samples were not available for NEFA measurement (n = 330) and those individuals with total NEFA inappropriately elevated (higher than 2000 μmol/l, n = 2). This resulted in a population of 1234 participants for the cross-sectional study and 924 participants for the prospective analysis.

The study was conducted according to the principles expressed in the Declaration of Helsinki and approved by the local ethics committee of each centre. Written informed consent was obtained before recruitment from all participants.

Metabolic tests

OGTTs were performed at baseline and at follow-up. Blood samples were collected at fasting and at 30, 60, 90 and 120 min after the ingestion of 75 g glucose for measurement of glucose, insulin and C-peptide concentrations. Normal glucose tolerance (NGT), impaired fasting glucose (IFG), impaired glucose tolerance (IGT) and type 2 diabetes were defined according to the current ADA criteria [25].

A euglycaemic−hyperinsulinaemic clamp was performed on a separate day within 1 month of the baseline OGTT. Insulin was administered as a primed-continuous infusion at a rate of 240 pmol min−1 m−2. Plasma glucose was clamped at 4.5–5.5 mmol/l with a variable 20% dextrose infusion adjusted every 5 min. Blood was sampled every 20 min for determination of insulin concentrations. OGTT and clamp procedures were standardised across centres.

Insulin sensitivity and beta cell function variables

Insulin sensitivity was expressed as the ratio of the average glucose infusion rate (M value) during the final 40 min of the clamp normalised to the fat-free mass (FFM) and to the mean plasma insulin concentration measured during the same time interval (M/I, in units of μmol kgFFM−1 min−1 [nmol/l]−1). Insulin sensitivity was also measured by the oral glucose insulin sensitivity index (OGIS) calculated during the baseline and follow-up OGTTs [26]. The liver insulin resistance index was calculated in all participants with an algorithm based on OGTT insulin levels, percentage fat mass (FM%), BMI and HDL-cholesterol [27]. This index strongly correlates with the tracer-derived endogenous glucose production relative to insulin at baseline (r = 0.65, p < 0.001) and during clamp (r = 0.59, p < 0.001) in these individuals without diabetes [27].

Basal and glucose-stimulated insulin secretion rate was estimated by C-peptide deconvolution [28]. Total insulin output throughout the OGTT was calculated by integrating insulin secretion rate over the duration of the test. Variables of beta cell function were calculated by mathematical modelling of insulin secretion and glucose concentrations, as previously reported [29,30,31]. This model describes the relationship between beta cell insulin secretion and glucose concentration as the sum of two components. The first component represents the dependence of insulin secretion on absolute glucose concentration through a dose–response function relating the two variables. The slope of this dose–response function is calculated and called beta cell glucose sensitivity. The dose–response is modulated by a potentiation factor, which accounts for the physiological processes that can acutely modify insulin secretion (e.g. protracted hyperglycaemia, non-glucose substrates, gastrointestinal hormones, neural modulation). The second component of beta cell function represents the dependence of insulin secretion on the rate of change of glucose concentration and is determined by a single variable (beta cell rate sensitivity), which is related to early insulin release.

Body composition

Body weight and FFM were assessed by electrical bioimpedance using a Body Composition Analyzer (model TB-300; Tanita, Tokyo, Japan). Fat mass was obtained as the difference between body weight and FFM; FM% was calculated as the ratio of fat mass to body weight. Waist circumference was measured at the narrowest circumference between the lower rib margin and anterior superior iliac crest. Hip circumference was measured around the widest portion of the buttocks, and the waist/hip ratio was calculated.

Biochemical measurements

Blood glucose was measured at the bedside by the glucose oxidase technique. Plasma samples were divided into aliquots and stored at −80°C until analysis. Centralised biochemical analyses were performed in predefined core laboratories to minimise assay errors and variability [24]. Plasma insulin and C-peptide were measured by fluoroimmunoassay (AutoDELFIA Insulin kit; Wallac Oy, Turku, Finland). Total plasma NEFA were measured by a fluorometric method (Wako, Neuss, Germany). Plasma concentrations of the six most prevalent NEFA, including saturated fatty acids (SFA: palmitate [16:0] and stearate [18:0]), monounsaturated fatty acids (MUFA: palmitoleate [16:1 n-7] and oleate [18:1 n-9]), and polyunsaturated fatty acids (PUFA: linoleate [18:2] and linolenate [18:3]), were measured by a targeted quantitative metabolomic approach using isotope dilution ultrahigh-performance liquid chromatography coupled to tandem MS (MS/MS), as previously reported [32].

Statistical analysis

Continuous variables are presented as mean ± SD and nominal variables are expressed as percentages. Variables with a skewed distribution are presented as median (interquartile range) and were log-transformed in multivariable analyses. Differences between groups (i.e. tertile of plasma palmitoleate at baseline or tertile of palmitoleate change at follow-up) were tested using χ2 for nominal variables and using ANOVA or Kruskal–Wallis test for normally or non-normally distributed continuous variables, respectively. Post hoc pairwise comparisons were tested by Tukey’s honest significant difference (HSD) test or by Steel–Dwass test, as appropriate. Differences between participants’ characteristics at baseline and at follow-up were assessed by paired Student’s t test or Wilcoxon signed-rank test. Correlations were tested using Pearson or Spearman correlation coefficients. Multivariable linear regression analysis tested the effect of palmitoleate on insulin sensitivity, beta cell function and glucose tolerance while controlling for potential confounders.

Cross-sectional analyses were adjusted for age, sex, BMI, FM% and total NEFA (Model 1). They were further corrected by adding the percentage enrichments of all measured NEFA as covariates (Model 2). The effect modification by sex was examined by adding a product term to all regression models (palmitoleate × sex), which was eventually removed as it showed no significant effect. The effect of palmitoleate on the expected [33] negative correlation between insulin sensitivity and palmitate was tested by including plasma palmitate, palmitoleate tertile and a product term between the two variables as factors. Longitudinal analyses were adjusted for baseline OGIS and percentage changes in BMI, FM% and total NEFA. They were further corrected by sex, age, follow-up duration and a product term between baseline OGIS and changes in palmitoleate to estimate the effect modification by baseline insulin sensitivity. Longitudinal analyses were also repeated using baseline M/I instead of baseline OGIS as covariate. To quantify the relative contribution of palmitoleate with respect to other factors, standardised regression coefficients (std. β), which indicate how many SDs the dependent variable changes per SD change in the predictor variable, were obtained from models on standardised variables. Statistical analyses were performed using JMP Pro 13.2.1 software (SAS Institute, Cary, NC, USA). A two-sided p < 0.05 was considered statistically significant.

Results

Cross-sectional analyses

Baseline clinical and metabolic characteristics of study participants (n = 1234) stratified by tertile of percentage palmitoleate enrichment in the total NEFA pool (PO%) are shown in Table 1.

Table 1 Baseline characteristics of study participants stratified by PO% tertile
Full size table

The mean (± SD) value of total NEFA was 531 ± 210 μmol/l (range 70–1585 μmol/l). Total NEFA were higher in women than men (585 ± 214 and 465 ± 184 μmol/l, respectively, p < 0.0001) and associated with age (r = 0.09, p < 0.0001). They were also associated with FM% (r = 0.28, p < 0.0001), clamp-derived whole-body insulin sensitivity (M/I: r = −0.16, p < 0.0001) and liver insulin resistance (r = 0.11, p < 0.0001), but not with BMI (p = 0.40) or beta cell function variables (p > 0.14 for all).

Plasma palmitoleate concentration was 13 ± 8 μmol/l (range 1–61 μmol/l), being higher in women than men (15 ± 8 μmol/l vs 10 ± 6 μmol/l, respectively, p < 0.0001) and increased with age (r = 0.10, p = 0.0003). PO% was 2.3 ± 0.8% (range 0.3–5.9%) and correlated with absolute palmitoleate concentration (r = 0.71, p < 0.0001). Consistently, PO% was higher in women (2.5 ± 0.9 % vs 2.1 ± 0.7 %, p < 0.0001) and in older people (age: r = 0.10, p = 0.0003). PO% was positively correlated with total NEFA concentration (r = 0.19, p < 0.0001) and adiposity (FM%: r = 0.21, p < 0.0001), while other major NEFA decreased or remained stable as total NEFA and FM% increased (Fig. 1; subgroup analyses by sex are shown in electronic supplementary material [ESM] Fig. 1). Furthermore, PO% was associated with plasma adiponectin concentration (r = 0.21, p < 0.0001) and negatively correlated with waist/hip circumference ratio (r = −0.19, p < 0.0001) and alanine aminotransferase levels (r = −0.15, p < 0.0001).

Fig. 1
figure 1

NEFA composition in non-diabetic individuals stratified by NEFA tertile (a) or FM% (b) (n = 1234). The relative abundance of palmitoleate increased with total NEFA and adiposity (p < 0.0001 by ANOVA), while other major NEFA remained stable or decreased

Full size image

The M/I, which provides an accurate estimate of whole-body insulin sensitivity, was significantly higher in those with higher PO% after stratification for total NEFA (Fig. 2a) or FM% (Fig. 2b). This finding was confirmed in women and men separately stratified by sex-specific tertile of total NEFA and FM% (ESM Figs 2 and 3). In bivariate analyses, PO% positively correlated with M/I in the whole cohort (r = 0.20, p < 0.0001) and in women and men separately (r = 0.18, p < 0.0001 and r = 0.09, p = 0.03, respectively), while there was no significant relationship between absolute palmitoleate concentration and M/I (p = 0.99). The association between PO% and M/I remained significant after adjustment for age, sex, BMI, FM% and total NEFA (Model 1: std. β = 0.14, p < 0.0001) (Fig. 3) and when all other major NEFA were added to the model (Model 2: std. β = 0.16, p < 0.0001). PO% was also associated with the liver insulin resistance index both in bivariate analysis (r = −0.06, p = 0.04) and multiple adjusted analyses (Model 1: std. β = −0.08, p < 0.0001; Model 2: std. β = −0.14, p < 0.0001). Furthermore, high palmitoleate concentrations attenuated the expected decrease in insulin sensitivity associated with elevated palmitate levels (palmitate by palmitoleate interaction effect: p = 0.0001) (Fig. 4). With regard to other NEFA, oleate enrichment was not correlated with insulin sensitivity in adjusted models (p = 0.14); this was also the case for the two SFAs, palmitate and stearate (p > 0.10). PUFA showed a positive correlation with insulin sensitivity, though weaker compared with palmitoleate (linoleate: std. β = 0.06, p = 0.04; linolenate: std. β = 0.08, p = 0.006).

Fig. 2
figure 2

Insulin sensitivity (a, b), beta cell glucose sensitivity (c, d), potentiation (e, f) and 2 h plasma glucose (g, h) in non-diabetic individuals stratified by tertile of plasma NEFA (a, c, e, g) or FM% (b, d, f, h) and by tertile of PO% (n = 1234). Whole-body insulin sensitivity was measured by a hyperinsulinaemic–euglycaemic clamp (M/I). Beta cell function variables and glucose tolerance were assessed during an OGTT. Data are means ± SEM. Differences were tested using ANOVA or Kruskal–Wallis test followed by post hoc pairwise comparisons. *p < 0.05 vs PO% tertile I, †p < 0.05 vs PO% tertile II

Full size image
Fig. 3
figure 3

Multivariable linear regression analyses of the relation between PO% and measures of insulin sensitivity, beta cell function and glucose tolerance in non-diabetic individuals (n = 1234). The effect of PO% remained significant after adjustment for sex, age, BMI, FM% and total NEFA. Standardised coefficients and 95% CI are shown

Full size image
Fig. 4
figure 4

Association between plasma palmitate concentration and clamp-derived insulin sensitivity in non-diabetic individuals stratified by tertile of plasma palmitoleate (n = 1234). High palmitoleate levels prevented the reduction in insulin sensitivity induced by excess palmitate (palmitate by palmitoleate interaction effect in multivariable regression analysis: p = 0.0001)

Full size image

Among beta cell function variables, beta cell glucose sensitivity and potentiation were significantly enhanced in individuals with higher PO% after stratification by total NEFA or FM% (Fig. 2c−f). In subgroup analyses, the effect of PO% on beta cell function appeared more consistent in women than in men (ESM Figs 2 and 3). In bivariate analyses in the whole cohort, beta cell glucose sensitivity and potentiation correlated with PO% (r = 0.08, p = 0.008 and r = 0.15, p < 0.0001, respectively), but not with absolute palmitoleate concentration (p = 0.72 and p = 0.12, respectively). The associations between beta cell function variables and PO% remained statistically significant after adjustment for age, sex, BMI, FM% and total NEFA (Model 1: std. β = 0.07, p = 0.01 and std. β = 0.10, p = 0.001, respectively) (Fig. 3) and, only for potentiation, after accounting for all measured NEFA (Model 2: std. β = 0.08, p = 0.045). Among other major NEFA, only linoleate enrichment was associated with beta cell glucose sensitivity (std. β = 0.08, p = 0.007), while no other NEFA was associated with potentiation besides palmitoleate.

Given the positive correlation of PO% with both insulin sensitivity and beta cell function, we tested the association with glucose tolerance assessed by the 2 h plasma glucose levels during the OGTT. As expected, 2 h plasma glucose was lower in individuals with higher PO% (Fig. 2g−h; ESM Figs 2 and 3). The two variables were negatively correlated in bivariate analysis (r = −0.08, p = 0.008), and after adjustments for potential confounders (Model 1: std. β= −0.12, p < 0.0001; Model 2: std. β = −0.24, p < 0.0001) (Fig. 3).

Longitudinal analyses

Clinical and metabolic characteristics of study participants at baseline and at follow-up (n =924) are shown in Table 2.

Table 2 Characteristics of the longitudinal study cohort at baseline and after a 3 year follow-up (n = 924)
Full size table

During the 3 year observational follow-up, PO% decreased by 0.3% (range −3.7% to 5.3%, p < 0.0001). Percentage changes at follow-up in PO% were not associated with percentage changes in FM%, waist/hip ratio or hip circumference (p > 0.50 for both) and showed only a marginally significant correlation with percentage changes in BMI (r = 0.06, p = 0.05). Furthermore, we did not observe associations of PO% changes with either beta cell glucose sensitivity (p = 0.79) or potentiation (p = 0.43).

At follow-up, insulin sensitivity assessed by the OGIS decreased on average by 10 ml min−1 m−2 (range −283 to 533 ml min−1 m−2, p < 0.0001). The spontaneous decline in OGIS was gradually attenuated across tertiles of changes in PO% (p = 0.04), approaching the null value in individuals whose PO% increased (tertile III) (Fig. 5a). Consistently, insulin sensitivity at follow-up was lower in people who had a decrease in PO% (n = 630) compared with those whose PO% increased (n = 294, p = 0.01), despite no differences in baseline OGIS nor in changes in BMI, FM% and total NEFA between the two subgroups. Furthermore, a positive relationship was observed between percentage changes in PO% and insulin sensitivity (OGIS: r = 0.07, p = 0.05). This association remained significant after adjustment for baseline OGIS and percentage changes in BMI, FM% and total NEFA (std. β = 0.07, p = 0.03) (Fig. 5b). It also remained significant after further adjustment for sex, age, follow-up duration and an interaction factor between percentage changes in PO% and baseline OGIS (std. β = 0.07, p = 0.04). The last factor showed no significant effect (p = 0.49). The effect of PO% changes on insulin sensitivity was similar when accounting for baseline M/I instead of baseline OGIS in the fully adjusted model (std. β = 0.07, p = 0.049).

Fig. 5
figure 5

(a) Percentage change in OGIS (Δ% OGIS ) after a 3 year observational follow-up in non-diabetic individuals stratified by spontaneous percentage change in palmitoleate enrichment (Δ% PO%) (n = 924). Data are means ± SEM. Differences were tested using Kruskal–Wallis test followed by post hoc pairwise comparisons. *p < 0.05 vs Δ% PO% tertile I. (b) Multivariable linear regression analysis of the relationship between Δ% PO% and Δ% OGIS in the whole cohort, adjusted for baseline OGIS and for percentage changes in BMI (Δ% BMI), FM% (Δ% FM%) and total NEFA (Δ% NEFA). Standardised coefficients and 95% CI are shown

Full size image

Discussion

In a large cohort of non-diabetic individuals, we demonstrated that the relative abundance of circulating palmitoleate, which increases proportionally with adiposity and total NEFA concentration, is associated with enhanced whole-body and liver insulin sensitivity, beta cell function and glucose tolerance. Furthermore, we observed that increased palmitoleate enrichment attenuates the negative effects of excess palmitate and ageing on insulin sensitivity. Our findings suggest that palmitoleate participates in the crosstalk between adipose tissue and other metabolically active organs/tissues, such as liver and beta cells, and support its protective role against glucose intolerance under physiological conditions.

Previous animal and cell culture studies have shown that palmitoleate can improve insulin signalling and enhance beta cell function [7, 14,15,16,17,18,19, 34, 35], in addition to its potential anti-obesity, anti-inflammatory, anti-oxidative and anti-atherosclerotic properties [11, 12]. Our novel findings support an insulin-sensitising effect of palmitoleate in humans. In multiple cross-sectional analyses, we described a positive relationship between palmitoleate levels and whole-body insulin sensitivity, determined by a gold standard clamp procedure. Furthermore, circulating palmitoleate showed a negative correlation with hepatic insulin resistance assessed by a validated index [27]. Palmitoleate was able to predict insulin sensitivity after stratification for adiposity and total NEFA as well as after accounting for potential confounders. Consistently, longitudinal analyses confirmed that changes in palmitoleate concentration are independent determinants of changes in insulin sensitivity. A relevant finding is that high palmitoleate concentrations appear to prevent the decrease in insulin sensitivity associated with its saturated counterpart palmitate, the most abundant SFA [33] (Fig. 4). Given that palmitate is the precursor of palmitoleate and the concentrations of the two NEFA are closely associated (r = 0.79, p < 0.0001), it is possible that increased palmitoleate synthesis in response to high palmitate levels can partly compensate for the detrimental effect of excess palmitate on insulin signalling. In contrast to SFA, circulating and dietary MUFA have been associated with increased insulin sensitivity and lower risk of type 2 diabetes [36]. However, the beneficial effect of palmitoleate on insulin signalling is unlikely to be explained solely by its belonging to the class of MUFA. In fact, we observed that oleate, by far the most abundant MUFA in human blood and in the diet [36], was not correlated with insulin sensitivity in adjusted models. Together, these results are in line with previous observations by Cao et al [7], who reported improved whole-body glucose disposal during palmitoleate infusion in mice, and with mechanistic studies in isolated skeletal muscle cells [15] and adipocytes [17]. The association of palmitoleate with hepatic insulin resistance, though measured by a surrogate index, is also in agreement with previous studies in rodents, in which palmitoleate administration improved hepatic insulin sensitivity [18] and prevented the accumulation of triacylglycerol in the liver [14].

Molecular mechanisms by which palmitoleate may exert its insulin-sensitising action include an enhancement in basal and insulin-stimulated glucose uptake due to increased GLUT1 and GLUT4 expression, activation of glucose flux through aerobic and anaerobic glycolysis, stimulation of glycogen synthesis and inhibition of lipogenesis [15, 17, 18]. The modulation of these mechanisms has been attributed to the activation of AMP-activated protein kinase (AMPK) by palmitoleate [15, 17, 18]. In humans, previous evidence on the effect of palmitoleate on insulin signalling is, however, limited and conflicting. In fact, the relationship between palmitoleate and insulin sensitivity has been described as either positive [8,9,10], not significant [4] or even negative [20,21,22]. The literature prior to the present study has been hampered by inadequate study samples, lack of appropriate correction for potential confounders (e.g. parallel changes in palmitoleate and total NEFA levels), use of surrogate markers of insulin sensitivity and lack of longitudinal data. Moreover, measures of palmitoleate in lipid compartments other than NEFA, such as serum cholesteryl esters [20] or phospholipids [22], may not accurately reflect adipocyte synthesis of palmitoleate for signalling purposes.

The present analysis describes, for the first time, a cross-sectional positive relationship between palmitoleate and beta cell function in humans, with potential differences between men and women. This finding is consistent with previous studies in pancreatic islets and isolated beta cells, which exhibited increased basal and glucose-stimulated insulin secretion [35] and were protected from apoptosis induced by glucotoxicity or lipotoxicity [19] when exposed to palmitoleate. Despite this evidence, the lack of a significant association between changes in palmitoleate and beta cell function over time, in agreement with previous observations [4], warrants cautious interpretation of this finding and confirmation by intervention studies.

Given its low and tightly regulated concentration, circulating palmitoleate is an ideal candidate to act as a lipokine. In the RISC and other cohorts [8, 10, 23, 37], palmitoleate represents up to 6–7% of total serum NEFA. Unlike that of most other fatty acids, dietary intake of palmitoleate is very low in Western diets [38, 39] and dietary supplementation with palmitoleate-rich food is unable to significantly increase its plasma concentration [40,41,42,43,44]. On the contrary, circulating palmitoleate levels in NEFA closely reflect its endogenous synthesis in adipocytes, which occurs through desaturation of palmitate by the enzyme stearoyl-CoA desaturase-1 (SCD-1) [45]. SCD-1 activity is enhanced by insulin and influenced by dietary intake of SFAs, sucrose, fructose and alcohol [45]. Thus, the synthesis of palmitoleate in the adipocyte can be modulated in response to different metabolic states and dietary habits. Remarkably, we observed differences between palmitoleate and other abundant NEFA in relation to measures of adiposity, in that the relative concentration of palmitoleate was increased proportionally with the expansion of adipose tissue and the parallel increase in total NEFA level, regardless of sex, while other major NEFA were decreased or remained stable. Given the plausible beneficial effects of palmitoleate on insulin sensitivity and secretion, we can speculate that increased palmitoleate synthesis may be an adaptive mechanism through which the adipocyte can counterbalance (at least in part) the detrimental effect of fat tissue expansion and increased NEFA concentrations on systemic glucose metabolism.

The gluteo-femoral subcutaneous adipose tissue has been proposed as the principal site of palmitoleate production, storage and release [9]. This observation may contribute to explain the beneficial metabolic properties of lower-body subcutaneous adipose tissue compared with abdominal subcutaneous and visceral adipose tissue [46,47,48,49]. A key role of gluteo-femoral adipose tissue on palmitoleate secretion is supported by our cross-sectional analysis, in which individuals with increased lower-body circumference feature higher palmitoleate levels. The lack of association between changes in hip circumference and palmitoleate enrichment in longitudinal analysis, however, suggests that the lower-body fat mass is only one of a complex interplay of factors regulating palmitoleate release. Still, differences in palmitoleate enrichment might aid understanding of the wide inter-individual variability in the negative impact of obesity on glucose metabolism [50].

This is the largest study to date that has investigated the role of palmitoleate on glucose homeostasis in humans. We analysed data from a thoroughly characterised cohort of non-diabetic individuals with evaluation of NEFA concentration and composition and accurate measures of insulin sensitivity and beta cell function. Whether these findings extend to individuals with diabetes remains to be examined; however, animal studies suggest that this may be the case [14]. Metabolic assessments were repeated after a follow-up that allowed the evaluation of concomitant changes of palmitoleate and metabolic variables over time. All significant relationships described in bivariate analyses were confirmed in multivariable models, accounting for important potential confounders as appropriate for the large sample.

Nonetheless, there are some limitations to our study. We determined the concentration of the six most prevalent NEFA, representative of the three classes of SFA, MUFA and PUFA, while the effect of other, less abundant, NEFA was not evaluated. The use of a validated surrogate index of liver insulin resistance allowed us to explore the association between palmitoleate and hepatic insulin sensitivity; however, this relationship should be further tested using direct measures of liver insulin resistance. In keeping with this, the lack of a direct measure of insulin sensitivity at follow-up is also a limitation of our study. Moreover, although causality is supported by the above-mentioned previous experimental evidence, intervention studies are needed to confirm the causal role of endogenous palmitoleate in preserving insulin sensitivity and/or beta cell secretion in humans.

In conclusion, our novel data suggest that palmitoleate has beneficial metabolic properties on insulin signalling, beta cell function and glucose tolerance in humans and support its key role in the endocrine network between adipose tissue and other metabolically active organs/tissues.