Human adipose tissue has been extensively characterized. Studies have analyzed the total amount, distribution, and even the phenotype of adipocytes. Many of these characteristics have been associated with health. Elevated fat percentage, as an index of adipose tissue total amount, increases the incidence of cardiovascular disease (1), and also all-cause and cardiovascular disease mortality (2). Central fat distribution, which is assessed by the waist-to-hip ratio, increases the incidence of cardiovascular disease, even after adjustment for fat percentage (1). High percentages of fat in the trunk, but not in the legs, are associated with increased mortality from cardiovascular disease (2), which is partially explained by phenotypic differences in the adipocytes within those regions (3). Finally, the volume and activity of brown adipose tissue have been proposed to influence health, yet this is still controversial (4). Many of these associations may be partially mediated by the effect that adipose tissue exerts on other nearby and distant tissues.

Adipose tissue secretes adipokines that regulate fuel oxidation in other tissues. Adiponectin, for example, stimulates lipid oxidation, glucose uptake, and lactate production in skeletal muscle cells (5). Adipose tissue also releases fatty acids, and thus influences fuel availability in other tissues. Of note, the adipokines secreted, as well as the rate and profile of released fatty acids, differ among adipose tissue depots (6–8). Thus, by influencing fuel oxidation and availability, different adipose tissue characteristics may modulate (positively or negatively) metabolic flexibility (MetF), i.e., the capacity of cells, tissues, and the organism to adapt fuel oxidation to fuel availability (9). An impaired MetF has been proposed to appear at the early stages of metabolic disturbances (10) and has been associated with several metabolic disturbances (9). This suggests that adipose tissue characteristics may influence MetF, which subsequently affects metabolic health.

Evidence for an association between adipose tissue characteristics and MetF is somewhat scattered. In the study where the concept of MetF was coined, Kelley et al. (11) compared lean subjects vs. obese subjects. MetF was assessed in response to glucose plus insulin infusions (hereafter insulin stimulation) as the change in leg respiratory quotient (RQ = VCO₂/VO₂), an index of the relative oxidation of glucose and lipid. Subjects with obesity had a blunted increase in glucose oxidation when the exogenous glucose availability increased. This response reflected an impaired MetF in obesity. Subjects with obesity also had reduced lipid oxidation during fasting (i.e., high fasting RQ), which was considered another marker of impaired MetF, although fasting RQ by itself does not necessarily reflect an adaptation of fuel oxidation to fuel availability (9, 12). Therefore, in this seminal publication, high levels of total adipose tissue (i.e., obesity) were associated with an impaired MetF. A subsequent study directly analyzed the association of MetF with some adipose tissue characteristics, namely fat percentage and the size of subcutaneous adipocytes (13); the increase in whole-body RQ during insulin stimulation was considered as the marker of MetF [mostly skeletal muscle MetF (9)]. Therein, subjects with the highest fat percentage and the largest abdominal subcutaneous adipocytes had the lowest MetF, independent of BMI. These findings suggested that impaired MetF was associated with high amounts of adipose tissue and hypertrophic adipocytes. Since the capacity for hypertrophy characterizes trunk adipocytes (3), this capacity may be somehow related to metabolic disturbances. Notably, the study also showed that subjects with the highest MetF had the highest circulating adiponectin concentrations, thus suggesting that adiponectin influences (positively) MetF in skeletal muscle (13).

The aforementioned studies suggested an association between certain adipose tissue characteristics and MetF. Nevertheless, MetF was only assessed by measuring the RQ in response to insulin stimulation. There are currently many other metabolic challenges to measure MetF in humans, e.g., meal consumption, oral glucose, exercise, prolonged fasting (9, 10, 12, 14). RQ is commonly used to determine MetF in response to these challenges. As far as we know, the evidence including measurements of the various adipose tissue characteristics and MetF in response to different challenges has not been systematically summarized. Moreover, many of the challenges to measuring MetF had been used even before the MetF concept was coined. Thus, there may exist studies that, technically, measured MetF, but without referring to it as MetF. Therefore, we aimed to systematically review the evidence for an association between adipose tissue characteristics (total amount, distribution, and phenotype) and MetF in response to different metabolic challenges in human adults. This information will allow us to gain insight into the possible role of MetF in mediating the effects of adipose tissue on health.

Adipose tissue total amount, distribution, and phenotype have been all associated with human health (1–4). The effect would be partially mediated by the release of adipokines and fatty acids that influence other tissues' functions, including their MetF (6–8, 13). If so, an association between adipose tissue characteristics and MetF is expected. To explore whether such association exists, we systematically gathered the evidence that measured adipose tissue characteristics and MetF in humans. We found that the associations depended on the method used to measure MetF. Our main findings were that: (a) MetF to insulin stimulation seems inversely associated with adipose tissue total amount, waist circumference, and VAT; and (b) MetF to dietary challenges does not seem to be associated with adipose tissue total amount or distribution. These results suggest that adipose tissue characteristics influence only certain determinants of whole-body MetF.

MetF has been defined in different ways, but the core idea remains the capacity to adapt fuel oxidation to fuel availability (9, 41–44). This definition has been operationalized in a wide diversity of methods. There is a certain agreement that the method should include a metabolic challenge requiring an adaptation in fuel oxidation (9, 41–44). The classical challenge is glucose plus insulin infusions during a euglycemic–hyperinsulinemic clamp, but other challenges have been also used (9). Since the RQ reflects the relative oxidation of glucose and lipid, the ΔRQ in response to the selected metabolic challenge has been often considered as the marker of MetF (9). Importantly, each challenge will test the capacity of specific tissues to adapt fuel oxidation to a change in the availability of different exogenous or endogenous substrates (9). Note that fasting RQ has been sometimes considered as a maker of MetF by itself; however, fasting RQ does not reflect an adaptation (i.e., a change) in fuel oxidation and depends on the preceding diet (45, 46). The same holds true for other measures that, by themselves, do not necessarily adhere to the concept of MetF (e.g., sleep RQ, exercise RQ, and absolute fat oxidation). Notably, we found over 100 reports wherein adipose tissue characteristics and any of these other putative markers of MetF were measured. The main characteristics of these reports are summarized in Supplementary Table 4. Nevertheless, given a large amount of evidence gathered, we focused only on the reports that measured ΔRQ and that had moderate-to-high methodological quality.

Several reports used insulin stimulation during a euglycemic–hyperinsulinemic clamp as the challenge to measure MetF. During the clamp, insulin stimulation is accompanied by infusions of glucose to maintain glycemia constant. Since high insulin infusions inhibit hepatic glucose production, the clamp mostly challenges the capacity of skeletal muscle to increase exogenous glucose oxidation (9). Any step from glucose uptake to mitochondrial oxidation will thus determine MetF to insulin stimulation (9, 47). In this context, most cross-sectional evidence suggested that adipose tissue total amount, waist circumference, and VAT were inversely associated with MetF. The agreement among these adipose tissue characteristics is expected considering their direct association (48). Thus, compared with lean subjects, those with higher adipose tissue or waist circumference showed lower MetF (20, 31, 33–35). Also, inverse associations of MetF with adipose tissue total amount (38, 39) and VAT (37, 38) were observed in reports that included subjects with a wide range of BMI. Only three reports suggested a different association pattern. First, Sparks et al. (37) found that women had higher adipose tissue total amount and MetF than men, but adipose tissue distribution seems to be more relevant when comparing between sexes. Indeed, the same report showed an inverse association between VAT and MetF in both men and women (37), and since women had lower VAT than men, VAT potentially explains the proposed sexual dimorphism in MetF. Second, Karczewska-Kupczewska et al. (36) found no difference in MetF between lean women and women with anorexia nervosa, thus suggesting that the association does not occur at very low adipose tissue. Finally, Amador et al. (19) showed that exercise training decreased fat mass, but had no effects on MetF. Note that this report was the only one with a clinical trial design. Perhaps, changes in adipose tissue total amount take some time to impact MetF. This may explain that the association between adipose tissue total amount and MetF to insulin stimulation mostly appears in cross-sectional comparisons of subjects with established nutritional statuses. Yet future studies are required to test these temporal associations. Note that only one study analyzed the association between adipose tissue phenotype (adipocytes volume) and MetF (37), which we considered insufficient to draw conclusions.

Regression analyses have shown that skeletal muscle insulin sensitivity is the major determinant of MetF to insulin stimulation, explaining ~50% of MetF variance (9, 49, 50). The rationale is that the higher the glucose uptake, the higher the intracellular glucose available for oxidation (9). Considering this, several pieces of evidence suggest that skeletal muscle insulin sensitivity explains the association of adipose tissue total amount, waist circumference, and VAT with MetF to insulin stimulation. First, subjects with high adipose tissue total amount and waist circumference had lower insulin sensitivity than their lean counterparts (31, 33–35). Second, insulin sensitivity was directly associated with MetF (38, 39). Third, insulin sensitivity was similar in lean women and women with anorexia nervosa, which were groups with different fat mass yet similar MetF (36). Fourth, Bak et al. (20) showed that men with normal weight had higher insulin sensitivity and MetF than men with obesity after a 12-h fast; however, after a 72-h fast, differences in insulin sensitivity vanished and essentially the same occurred with MetF. Fifth, abdominal obesity (i.e., high waist circumference and VAT) has been long known to associate directly with insulin resistance (51). Note that the factors explaining the remaining MetF variance are yet to be determined. Fasting RQ, race, and circulating triglycerides and fatty acids have been implicated, but seem to play a minor role (49, 50). Identifying these determinants may help to understand, for example, the observation that exercise training can increase skeletal muscle insulin sensitivity without affecting MetF (19).

Other set of reports analyzed MetF in response to dietary challenges. In these challenges, fuel availability is at the intake level (9). MetF, therefore, depends on every level from the digestive system to mitochondrial oxidation, and also on the hormonal (insulin) responses. Moreover, the relative influence of each tissue on substrate oxidation is unknown (9), and during challenges including mixed meals, glucose, lipid, and proteins are oxidized simultaneously. Thus, MetF to dietary challenges represents a more physiological challenge, yet depends on a complex interaction of many systems. In this context, most reports suggested that adipose tissue characteristics were not associated with MetF, independent of the study design (21, 22, 24–28, 40). There were two exceptions. First, Huda et al. (23) showed that subjects with obesity had higher fat mass and waist circumference, but lower MetF to a breakfast, than their lean counterparts. The reason for the discrepancy is unknown but may depend on factors such as the specific dietary challenge (European style breakfast), among others. Second, Rudwill et al. (24) found that a 21-day bed rest decreased SAT, and this was accompanied by decreases in MetF to a meal and by increases in calf fat. This agrees with our recent observations of higher SAT in subjects with high vs. low MetF to a 75-g glucose intake (14). Together, this evidence suggests that SAT may regulate MetF to dietary challenges by storing fatty acids and thus preventing their ectopic accumulation. This could protect subjects from developing peripheral insulin resistance, but future studies should test this hypothesis.

With our search strategy and eligibility criteria, we identified few reports that measured MetF to other challenges. Two reports measured MetF in response to sleep, as the change from either 24-h RQ (29) or awake RQ (30) to sleep RQ. This method challenges the capacity of tissues to increase the oxidation of endogenous lipids. Thus, a larger drop in RQ during sleep indicates higher MetF (9). In this context, MetF depends on the capacity of adipose tissue to release lipids, and of other tissues to uptake and oxidize those lipids. Both reports suggested no association between adipose tissue characteristics and MetF to sleep (29, 30). One report measured MetF as the RQ variance to daily living activities, thus challenging the capacity of skeletal muscle to adapt the oxidation of endogenous substrates to variable energy demands (32). Another report measured MetF to epinephrine stimulation, thus challenging the capacity of beta-adrenergic-sensitive tissues to increase lipid oxidation (28). Given the few evidence for these methods, it may be too speculative to draw conclusions. Nevertheless, the study by Júdice et al. (32) revealed a fact worth considering. They observed that fat mass and trunk fat were not linearly associated with MetF; instead, an inverse regression model better fitted the associations. This suggests that models more complex than linear regression may describe the association between adipose tissue characteristics and MetF.

A limitation of our review process was that we only searched in PubMed, thus probably missing evidence from other databases. Note, however, that PubMed has been long recognized as an optimal tool for biomedical electronic research (52), and the most influential research on MetF is indexed therein [see (9, 41–44, 53) and their references]. Another limitation is that we focused only on reports in adult humans, thus excluding children, adolescents, and older adults, in whom several insightful studies have been conducted [see for example: (54–57)]. We also excluded in vitro studies because of the difficulty in operationalizing MetF in that model, but many of those studies have been described in previous narrative reviews (43, 44). The evidence included also has some limitations. First, although we included reports that measured adipose tissue characteristics and MetF, no study was specifically designed to test the association between these variables. The selection of participants, sample size, statistical analyses, among other factors may, thus, obscure the real association between adipose tissue characteristics and MetF. We used a rather simple strategy to ascertain whether the reports suggested an association (see section Analysis), and we used vote counting to summarize the information. But this strategy is sensitive to flaws, as revealed by the report by Rynders et al. (30). Therein, obesity-prone subjects had higher fat mass and lower MetF than obesity-resistant subjects, which may suggest an association; however, further analyses showed that the differences in MetF remained after adjusting for fat mass. Third, evidence suggesting an association between adipose tissue characteristics and MetF is mostly cross-sectional, therefore, a causal relationship cannot be inferred. The idea that the profile of adipokines and/or fatty acids plays an intermediary role between adipose tissue and MetF is thus only speculative at this point. Fourth, only one report remained for the analysis of adipose tissue phenotype (37), which we considered insufficient to draw conclusions. Finally, the search strategy was designed to identify reports that showed RQ values. Yet there may be reports mostly focused on other markers (e.g., lipid oxidation) that secondarily reported RQ and may have not been retrieved by the search engine [for example (58, 59)].

Our review also has some strengths worth mentioning. First, although several narrative reviews about MetF have been published (9, 41–44, 53), as far as we know, this is the first systematic review regarding MetF. Second, we only considered reports with moderate-to-high methodological quality to draw our conclusions; yet we presented the data for the reports with lower methodological quality as well (Supplementary Table 3). Finally, although we established additional criteria to select reports for analysis, all studies meeting our initial criteria are summarized in Supplementary Table 4. This table thus represents a valuable repository for researchers interested in other aspects of MetF.

In conclusion, we have shown that some adipose tissue characteristics are associated with MetF to insulin stimulation. Thus, evidence suggests that adipose tissue may directly or indirectly influence the capacity to adapt fuel oxidation to fuel availability in skeletal muscle, an effect most probably explained by insulin sensitivity. Our results have also highlighted the relevance of specifying the method used to measure MetF in future studies. This will allow us to better understand the role that MetF has on metabolic health along with the mechanisms involved.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

AG conceived the study, acquired the data, processed the data, analyzed the data, and interpreted the data. FD-C, JF, and RR-R acquired the data and processed the data. JG conceived the study and interpreted the data. RF-V conceived the study, acquired the data, analyzed the data, interpreted the data, and drafted the manuscript. All the authors revised critically the manuscript and approved the final version.

This study was funded by ANID/CONICYT FONDECYT Iniciación 11180361 to RF-V.

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.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.