Superficial vs Deep Subcutaneous Adipose Tissue: Sex-Specific Associations With Hepatic Steatosis and Metabolic Traits.

CONTEXT: Subcutaneous adipose tissue (SAT) is not homogeneous, as the fascia scarpa separates the deep SAT (dSAT) from the superficial SAT (sSAT).
OBJECTIVE: The aim of this study is to evaluate the sex-specific associations of sSAT and dSAT with hepatic steatosis and metabolic syndrome in overweight individuals.
METHODS: We recruited 285 individuals with a body mass index (BMI) greater than or equal to 27 and aged 55 to 81 years. Abdominal magnetic resonance imaging was performed around level L4 to L5 to measure visceral adipose tissue (VAT), dSAT, and sSAT volumes. The amount of hepatic fat was quantified by MR spectroscopy.
RESULTS: Men had significantly higher volumes of VAT (122.6 cm3 vs 98.7 cm3, P < .001) and had only half the volume of sSAT compared to women adjusted for BMI (50.3 cm3 in men vs 97.0 cm3 in women, P < .001). dSAT correlated significantly with hepatic fat content in univariate analysis (standardized β = .190, P < .05), while VAT correlated significantly with hepatic steatosis in a multivariate model, adjusted for age, alcohol use, and other abdominal fat compartments (standardized β = .184, P = .037). Moreover, dSAT in men correlated negatively with HDL cholesterol (standardized β = -0.165, P = .038) in multivariate analyses. In women with a BMI between 30 and 40, in a multivariate model adjusted for age, alcohol use, and other abdominal fat compartments, VAT correlated positively (standardized β = -.404, P = .003), and sSAT negatively (standardized β = -.300, P = .04) with hepatic fat content.
CONCLUSION: In men, dSAT is associated with hepatic steatosis and adverse metabolic traits, such as lower HDL cholesterol levels, whereas in women with obesity sSAT shows a beneficial relation with respect to hepatic fat content.

The prevalence of obesity has nearly doubled worldwide since 1980 (1). This increase has led to a pandemic increase in obesity-related disorders such as hypertension, dyslipidemia, and diabetes mellitus, all of which are components of metabolic syndrome. The expansion of the visceral adipose tissue stores is especially related to metabolic dysregulation and subsequently leads to an increased risk for cardiovascular morbidity and mortality (2, 3). Visceral adipose tissue (VAT) is an important factor in the development of nonalcoholic fatty liver disease (NAFLD), also seen as the hepatic equivalent of metabolic syndrome, both in men and women. Approximately 75% of individuals with obesity have increased hepatic fat, defined as a hepatic fat fraction above 5.6% as measured by magnetic resonance spectroscopy (MRS) (4-6). Hepatic steatosis is the first stage of NAFLD and is usually benign and reversible(7). However, this condition can progress to nonalcoholic steatohepatitis, resulting in cirrhosis in 1% to 2% of patients (8). NAFLD and nonalcoholic steatohepatitis are both related to a higher risk for cardiovascular disease (9, 10).

Previous studies have reported that VAT is an important factor in the development of NAFLD and contributes to an increased cardiovascular risk (2, 3, 11). However, a relative increase in subcutaneous adipose tissue (SAT) compared to VAT is related to decreased severity of NAFLD (12). Currently, there are 2 hypotheses to explain the difference between VAT and SAT in the pathogenesis of NAFLD. The “portal theory” proposes that the liver is exposed to excessive free fatty acids (FFAs) and proinflammatory factors released from the visceral fat into the portal vein of patients with obesity, which promotes the development of hepatic insulin resistance and hepatic steatosis (13). However, only 5% to 30% of the delivered FFAs to the liver originates from visceral fat (14, 15), whereas at least 50% to 60% of the FFAs comes from the systemic circulation and thus from subcutaneous fat (16). The second hypothesis is the “ectopic fat hypothesis”(17), which proposes an indirect mechanism whereby increased energy storage in peripheral subcutaneous fat exerts a protective effect by decreasing fat deposition in the liver, muscle, and heart. However, although subcutaneous fat seems to be related to a more beneficial metabolic profile than visceral fat, it is important to realize that subcutaneous fat is not a homogeneous compartment, and we now hypothesize that there is a differential contribution to fatty liver disease of the 2 different subcutaneous fat compartments.

Once regarded to be a single entity, the abdominal SAT is divided by the scarpa fascia into superficial (sSAT) and deep subcutaneous adipose tissue (dSAT). The adipocytes from dSAT have higher lipolytic activity than superficial SAT adipocytes and contribute substantially to FFA levels in the circulation (18). It has been suggested that abdominal dSAT exhibits an intermediate phenotype between VAT and abdominal sSAT (19). Moreover, in patients with type 2 diabetes mellitus, a higher relative distribution of abdominal fat in sSAT was associated with a lower cardiovascular risk, whereas dSAT was related to a higher blood pressure and lower heart rate variability (20).

Although different studies have shown that dSAT and sSAT are evidently distinct abdominal adipose tissue depots, in the majority of studies investigating the association between abdominal adipose tissue distribution and hepatic steatosis or metabolic syndrome, no distinction is made between dSAT and sSAT. Because adipocytes from dSAT and sSAT have a different metabolic and inflammatory profile (21), we hypothesized that dSAT, but not sSAT, is associated with hepatic steatosis. We also investigated the associations between the different components of metabolic syndrome and the various adipose tissue compartments (VAT, dSAT, sSAT), taking into account the sex differences in adipose tissue distribution.

Materials and Methods

Study Population

We enrolled 302 individuals with overweight (body mass index [BMI] > 27), aged 55 to 80 years, in the period between 2014 and 2016 (300-OB study). The majority of the individuals (n = 227) also participated in the NBS-NIMA1 study, a population-based survey of Nijmegen residents (22). We recruited another 75 participants, acquaintances of previously included individuals, who fulfilled the inclusion criteria of age older than 55 years and BMI greater than 27. Exclusion criteria were a recent cardiovascular event (myocardial infarction, transient ischemic attack, or stroke < 6 months before), history of bariatric surgery or bowel resection, inflammatory bowel disease, renal dysfunction, increased bleeding tendency, use of oral or subcutaneous anticoagulant therapy, use of thrombocyte aggregation inhibitors other than acetylsalicylic acid and carbasalate calcium, or a contraindication for magnetic resonance imaging (MRI). Participants who used lipid-lowering therapy temporarily discontinued this medication 4 weeks prior to the measurements. All women were postmenopausal and did not use hormonal replacement therapy. All participants received detailed written and oral information and provided written informed consent. The Arnhem-Nijmegen Medical Ethics Committee approved the study protocol (in accordance with the Declaration of Helsinki).

Clinical Parameters

Individuals filled out an extensive questionnaire about lifestyle, medication use, and previous medical history. Blood sampling was performed in the morning after an overnight fast. Blood glucose, triglycerides (TGs), total cholesterol, and high-density lipoprotein cholesterol (HDL-C) were measured by standard laboratory procedures. Weight and height were measured and BMI was calculated as body weight (in kilograms) divided by the square of height (in meters). Waist circumference (WC) was measured at the level of the umbilicus to the nearest 0.1 cm. Hip circumference was measured at the level of the trochanter major. Waist-to-hip ratio (WHR) was calculated by dividing the WC by the hip circumference. The diagnosis of metabolic syndrome was made using the clinical criteria of the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III) (23).

Magnetic Resonance Imaging and Magnetic Resonance Spectroscopy Data Acquisition

Abdominal fat distribution and liver fat content were determined by MRI and proton MRS, respectively. The combined MR examinations were performed on a 3.0-T Magnetom Skyra or Trio MR system (Siemens). Individuals were examined in the supine position with their arms positioned parallel to the lateral sides of the body. For the determination of visceral and subcutaneous fat distribution a series of T1-weighted fast low-angle shot (FLASH) 2-dimensional axial MR images were acquired from a region extending from 4 cm above to 4 cm below the fourth and fifth lumbar interspace (16 slices, matrix size 192/256, field of view 300-353/400-470 mm2, slice thickness 5 mm, breath-hold, repetition time/echo time 80 ms/2.46 ms). Breathing commands were used to avoid motion-induced artifacts.

For the MRS measurement, a single voxel of 27 cm3 was positioned in the right lobe of the liver, avoiding the biliary tree and large blood vessels. A stimulated echo acquisition mode (STEAM) (24) localization sequence without water suppression was used for data acquisition. To minimize relaxation effects on signal intensity, a long repetition time (3 seconds) and short echo time (20 ms) were used. Six scans were averaged during breath-holding for 15 seconds. No prescans were used.

Visceral Adipose Tissue, Subcutaneous Adipose Tissue (SAT), Superficial SAT, and Deep SAT Analysis

The images acquired were retrieved from the MR scanner and analyzed with software developed in the IDL 6.0 environment, called HIPPO FAT (version 1.3, v. Positano) (25). Owing to the T1 weighting, fatty tissues are represented with signal intensity in these images.

VAT, SAT, dSAT, and sSAT volumes were measured on 8 separate slices, with an interslice distance of 5 mm, around the L4 to L5 intervertebrate level. HIPPO FAT automatically generates 3 contour lines at each image provided by an active fuzzy clustering algorithm (26) that allowed the separation of SAT from VAT: (1) along the outer margin of the SAT, (2) along the inner margin of the SAT and (3) around the smallest possible region in the visceral region that included all VAT. A histogram of signal intensities in the VAT region was provided, in which a gaussian curve automatically fitted the high-intensity peak, which identified the visceral fat.

After automatic segmentation, the analyst (T.B.), if necessary, manually adjusted both the contour lines and the shape of Gaussian curve by eyeballing. The MRI scan allows visualization of the scarpa fascia as a fine black line. To divide sSAT from dSAT, a line was drawn manually over the scarpa fascia. Adipose tissue pixels between this line and the outer margin of the SAT were defined as sSAT. dSAT was calculated by the total subcutaneous adipose tissue pixels minus the superficial subcutaneous adipose tissue pixels (dSAT = SAT – sSAT). Interclass correlation coefficients for interobserver comparisons were 0.799 for VAT, 0.999 for SAT, 0.998 for dSAT, and 0.999 for sSAT based on n = 11.

Quantification of Hepatic Fat Content

All MR spectra were postprocessed using the jMRUI software v3.0 package and the AMARES algorithm (27) to determine water (4.7 ppm) and methylene (1.3 ppm) resonance areas. Intrahepatic TG content was expressed as the fraction of the methylene signal in the combined signal of methylene and water (%). No correction for relaxation differences was applied. Based on the European guidelines (28), we considered NAFLD to be present when the ratio of methylene to methylene and water was equal to or greater than 5.6%.

Statistical Analysis

Quantitative variables are shown as the mean ± SD. Because hepatic fat content is not normally distributed, the univariate analyses on the relations to hepatic fat content are performed by nonparametric tests (Spearman correlation), whereas all other univariate analyses are performed by Pearson correlations (Tables 1 and 2). The association between the various abdominal adipose tissue compartments and the characteristics of metabolic syndrome were investigated using linear regression analysis, adjusted for sex, age, alcohol intake, and medication use (antidiabetic, antihypertensive drugs, and statins). The statistical analysis was performed using SPSS, version 22 (IBM Corp). A P value of less than .05 was considered to indicate statistical significance.

Table 1.

Univariate correlations of hepatic fat content and the different abdominal fat compartments with metabolic syndrome traits. Relation with hepatic fat content is based on Spearman correlations; relation with abdominal fat compartment volumes is based on Pearson correlation

	Volume VAT		Volume sSAT		Volume dSAT		Hepatic fat content
	Men(n = 155)	Women(n = 123)	Men(n = 159)	Women(n = 126)	Men(n = 159)	Women(n = 126)	Men(n = 147)	Women(n = 120)
WC	0.513c	0.448c	0.558c	0.400c	0.721c	0.527c	0.249b	0.300b
Triglycerides	0.255b	0.335c	–0.096	–0.010	0.014	–0.021	0.241b	0.392c
HDL-C	–0.229b	–0.257b	–0.013	–0.075	–0.192a	0.021	–0.114	–0.361c
Systolic BP	0.105	0.168	0.080	0.128	0.092	0.200a	0.057	0.138
Glucose	0.213b	0.383c	–0.027	0.106	0.023	0.057	0.344c	0.389c

Abbreviations: BP, blood pressure; dSAT, deep subcutaneous adipose tissue; HDL-C, high-density lipoprotein cholesterol; sSAT, superficial subcutaneous adipose tissue; VAT, visceral adipose tissue; WC, waist circumference.

P less than .05.

P less than .01.

P less than .001.

Table 2.

Associations of metabolic parameters with different abdominal adipose tissue compartments, adjusted for age, alcohol, and use of antidiabetic or lipid-lowering drugs (depending on dependent variable)

	Volume VAT		Volume dSAT		Volume sSAT		Hepatic fat content
	Men(n = 149)	Women(n = 118)	Men(n = 153)	Women(n = 120)	Men(n = 153)	Women(n = 120)	Men(n = 147)	Women(n = 116)
Glucose	0.211a	0.384c	0.085	0.121	–0.005	0.161	0.179a	0.297b
Triglycerides	0.265b	0.322c	0.016	–0.013	–0.008	–0.040	0.193a	0.196a
HDL-C	–0.208a	–0.236b	–0.168a	0.003	0.015	–0.057	–0.053	–0.213a

Abbreviations: dSAT, deep subcutaneous adipose tissue; HDL-C, high-density lipoprotein cholesterol; sSAT, superficial subcutaneous adipose tissue; VAT, visceral adipose tissue.

P less than .05.

P less than .01.

P less than .001.

Results

Baseline Characteristics

In 7 individuals MRI was not performed because of claustrophobia. In 10 individuals it was not possible to calculate any abdominal adipose tissue volume because of insufficient MRI data (mostly movement artifacts). Our study cohort therefore consisted of 285 individuals (159 men and 126 women). In 7 individuals it was not possible to calculate the VAT volumes because of low-quality images, and 18 individuals could not be classified because of missing data for liver fat; MRS could not be performed in these individuals because of technical difficulties.

Individuals with missing data on VAT or liver fat were excluded from analyses including these variables. The mean age was 67.1 ± 5.3 years and mean BMI was 30.6 ± 3.3. A total of 44.6% of the study population was treated for hypertension (73 men and 54 women), 25.3% were on lipid-lowering medication (46 men and 26 women), 7.4% used oral antidiabetic medication (15 men and 6 women), and 1.8% used insulin. There was no significant correlation between the units of alcohol per day and the amount of hepatic steatosis. The baseline characteristics of the study population are shown in Table 3. A total of 153 individuals met the NCEP ATP III criteria for metabolic syndrome with no significant difference in the prevalence between men and women. The prevalence of a hepatic fat content of 5.6% of greater as measured by MRS was 54% in men and 48% in women.

Table 3.

Baseline characteristics of the study population (n = 285)

Measurement	Total cohort (n = 285)	Participants with metabolic syndrome		Participants without metabolic syndrome
	Mean ± SD	Men (n = 84) Mean ± SD	Women (n = 69) Mean ± SD	Men (n = 75) Mean ± SD	Women (n = 69) Mean ± SD
Age, y	67.1 ± 5.3	67.1 ± 5.6	67.8 ± 5.5	66.0 ± 5.0	67.5 ± 5.0
Weight, kg	90.2 ± 12.8	97.7 ± 12.1a	85.0 ± 11.0a	93.4 ± 11.0	81.2 ± 9.4
BMI	30.6 ± 3.3	31.2 ± 3.4b	31.4 ± 3.9a	29.6 ± 2.7	29.9 ± 2.8
WC, cm	106.1 ± 9.6	112.3 ± 8.3c	104.9 ± 9.7c	105.7 ± 7.8	99.2 ± 7.9
Hip circumference, cm	111.2 ± 7.4	110.7 ± 6.7	113.9 ± 9.7	109.0 ± 5.0	111.6 ± 7.2
WHR	1.0 ± 0.1	1.0 ± 0.0c	0.9 ± 0.1a	1.0 ± 0.1	0.9 ± 0.1
Glucose	5.7 ± 1.3	6.1 ± 1.6c	6.2 ± 1.5c	5.2 ± 0.7	5.0 ± 0.5
Triglycerides	1.8 ± 1.0	2.3 ± 1.4c	2.2 ± 0.8c	1.3 ± 0.4	1.3 ± 0.4
HDL-C	1.3 ± 0.3	1.1 ± 0.3c	1.3 ± 0.3c	1.3 ± 0.3	1.6 ± 0.2
Systolic BP, mm Hg	129.5 ± 13.8	131.3 ± 13.9a	131.7 ± 15.2	127.1 ± 12.3	127.3 ± 13.3
Diastolic BP, mm Hg	80.0 ± 8.9	81.1 ± 9.9	78.2 ± 9.3	82.5 ± 7.9	77.2 ± 7.2
VAT, cm3	112.0 ± 39.0	138.1 ± 39.4c	107.1 ± 34.2b	106.1 ± 35.0	88.7 ± 27.7
SAT, cm3	162.2 ± 55.3	145.6 ± 45.4b	195.7 ± 53.8	127.7 ± 38.2	191.6 ± 52.5
sSAT, cm3	70.9 ± 33.0	51.6 ± 17.7	98.7 ± 31.6	48.8 ± 14.1	95.0 ± 28.7
dSAT, cm3	91.3 ± 30.9	94.0 ± 31.0b	97.0 ± 30.4	78.9 ± 27.7	96.7 ± 31.7
VAT-to-SAT ratio	0.8 ± 0.4	1.0 ± 0.4b	0.6 ± 0.2b	0.9 ± 0.4	0.5 ± 0.2
Hepatic fat content, %	9.9 ± 11.8	12.9 ± 11.8c	12.1 ± 12.5a	6.6 ± 8.9	6.6 ± 12.3

Abbreviations: BMI, body mass index; BP, blood pressure; dSAT, deep subcutaneous adipose tissue; HDL-C, high-density lipoprotein cholesterol; SAT, subcutaneous adipose tissue, sSAT, superficial subcutaneous adipose tissue; VAT, visceral adipose tissue; VAT-to-SAT, visceral adipose tissue/subcutaneous adipose tissue; WC, waist circumference; WHR, waist-to-hip ratio.

P less than .05,

P less than .01,

P less than .001 for independent sample t test comparing those with and without metabolic syndrome within each sex.

Abdominal Adipose Tissue Distribution

Men had a higher weight (95.7 kg vs 83.3 kg), larger WC (109.2 cm vs 102.3 cm), and higher WHR (0.99 vs 0.91) than women, all P less than .001. There was a clear sex difference in the abdominal adipose tissue distribution. On the 8 5-mm MRI slices as described earlier, men had a significantly higher volume of VAT (122.6 cm3 vs 98.7 cm3, P < .001) and a lower volume of SAT (137.1 cm3 vs 193.8 cm3, P < .001) compared to women, despite no difference in BMI. This resulted in a higher VAT-to-SAT ratio in men compared to women (0.97 vs 0.54). When adjusted for height, the difference between men and women of VAT and SAT remained significant. In particular, there was a profound difference in the sSAT volume (50.3 cm3 in men vs 97.0 cm3 in women, P < .001), whereas the difference in dSAT volume was much smaller, but still significant (86.9 cm3 in men vs 96.8 cm3 in women, P = .007). This resulted in a higher sSAT-to-dSAT ratio in women compared with men (1.06 vs 0.61, P < .001).

Abdominal Adipose Tissue Distribution In Relation to Metabolic Syndrome

We next assessed whether the volume of the separate adipose tissue compartments and the hepatic fat content correlated with the metabolic parameters. In all individuals with metabolic syndrome, VAT and the VAT-to-SAT ratio were significantly higher compared to individuals without metabolic syndrome, both in men and in women. The hepatic fat content was twice as high in men with metabolic syndrome (12.9% vs 6.6%, P < .001), and this was also significantly higher in women with metabolic syndrome (12.1% vs 6.6%, P = .016). Furthermore, in men with metabolic syndrome dSAT was significantly higher than in men without metabolic syndrome (94.0 cm3 vs 78.9 cm3, P = .001), whereas no difference was found for sSAT.

In univariate analysis, VAT and hepatic fat content both showed a strong association with the separate components of metabolic syndrome. In men, dSAT was associated negatively with HDL-C (see Table 1). After adjustment for age, alcohol, and medication use, only in men was dSAT associated negatively with HDL-C in the plasma. In contrast, we found no association between these markers and sSAT both in men and women. As expected, VAT was associated positively both in men and women with glucose and TG levels and was negatively associated with HDL-C. Hepatic fat content was associated in women with glucose and inversely with HDL-C; in men only glucose and TGs correlated with hepatic liver fat (see Table 2).

Abdominal Adipose Tissue Distribution in Relation to Hepatic Fat Content

In all individuals there was a significant correlation between VAT and hepatic fat content. This association remained significant, even after adjustment for age, alcohol intake, and other abdominal fat compartment volumes (sSAT and dSAT in this case), with a standardized β of .184 for men (P = .037) and a standardized β of .304 for women (P = .001). Only in men however, was dSAT positively associated both with VAT and hepatic fat content in a bivariate analysis (Table 4).

Table 4.

Univariate correlations between hepatic fat content and between different abdominal fat compartments. Relation with hepatic fat content is based on Spearman correlations; relation with abdominal fat compartment volumes is based on Pearson correlation

	Hepatic fat content		Volume dSAT		Volume sSAT
	Men(n = 147)	Women(n = 120)	Men(n = 159)	Women(n = 126)	Men(n = 159)	Women(n = 126)
Volume VAT	0.322c	0.390c	0.198a	0.148	0.055	0.191a
Volume sSAT	–0.001	–0.009	0.679c	0.498c
Volume dSAT	0.190a	0.113

Abbreviations: dSAT, deep subcutaneous adipose tissue; sSAT, superficial subcutaneous adipose tissue; VAT, visceral adipose tissue.

P less than .05.

P less than .01.

P less than .001.

As the amount of total adipose tissue could influence the relation between the different adipose tissue compartments and hepatic steatosis, we investigated the associations with VAT, dSAT, and sSAT for different BMI categories (27-30 and 30-40; data not shown) after correction for age, alcohol intake, and other abdominal fat compartment volumes. In men, the association between different adipose tissue compartments and hepatic steatosis was not significant anymore, probably because of the smaller sample size. In women, in all adiposity subgroups, VAT remained most strongly positively associated with hepatic fat content. In women, no correlations were found between the volume of the dSAT and hepatic fat content, whereas only in women with a BMI between 30 and 40 was sSAT significantly negatively associated with hepatic fat content, with a standardized β of –0.300, P equal to .04. In women with overweight (BMI = 27-30), however, no associations were seen between the abdominal fat compartments and hepatic fat content, making VAT the most strongly related with hepatic fat content in men and women.

Discussion

In this study we observed profound sex differences in abdominal fat distribution and its correlations with hepatic steatosis and components of metabolic syndrome. Although men and women had similar BMI, men had significantly higher volumes of VAT and relatively more dSAT compared to women. In general, VAT was most strongly related to hepatic fat content in men and women alike. Participants with metabolic syndrome had more VAT and more hepatic fat content than individuals without metabolic syndrome. Within the SAT compartment, women had a much higher ratio of sSAT to dSAT compared to men. Importantly, only in men was dSAT positively associated with hepatic fat content and negatively associated with HDL-C, whereas in women, dSAT did not show any associations with these parameters. In contrast, only in women with obesity was sSAT negatively associated with hepatic fat content.

A sex difference in SAT distribution has been reported by Golan et al (20), who investigated VAT, total SAT, and sSAT measurements by MRI. Our results are in line with their observations that men had a higher VAT-to-SAT ratio and women had a higher proportion of abdominal sSAT. Furthermore, our results confirm those of another study, which suggested that in men adiposity is characterized by a disproportionate expansion of dSAT compared with sSAT, whereas in women both dSAT and sSAT expand similarly in obesity (29). A greater proportion of dSAT in men at any level of subcutaneous or total body fat compared with women was previously observed, in contrast with our results (30). However, in that study, single-slice computed tomography was used to evaluate abdominal adipose tissue compartments, which may explain the differences with our findings. A possible explanation for the higher VAT content and lower dSAT and sSAT content in men could be the difference in hormonal status. Hyperandrogenism in normal-weight polycystic ovary syndrome women was associated with preferential visceral fat deposition and moreover an increased proportion of small subcutaneous abdominal adipocytes, which could limit subcutaneous fat storage (31). Moreover, 6 months of testosterone therapy in aging men with a low normal bioavailable testosterone decreased subcutaneous fat on the abdomen and lower extremities, whereas visceral fat remained unchanged (32).

Both in men and women, VAT correlated most strongly with hepatic fat content. As previous studies have shown that fat volumes could be influenced by medication use (33-35), we also adjusted the analysis of VAT and hepatic fat content for antidiabetic, antihypertensive, and lipid-lowering medication; however, VAT still remained positively associated with hepatic fat content in all individuals. It would be interesting to investigate whether the duration of medical therapy would influence this relation. Importantly, in men, dSAT also correlated positively with hepatic fat content, whereas in women with obesity sSAT correlated negatively with hepatic fat content. Yaskolka Meir et al found in a group of 275 individuals selected for high WC or dyslipidemia only a significant positive association for VAT with hepatic fat content, whereas sSAT was inversely associated with hepatic fat content in an age-, sex-, and WC-adjusted model (36). However, they did not perform analysis stratified by sex, which could explain the differences with our results. Another study related abdominal adipose tissue volumes in a small group of type 2 diabetes mellitus patients with peripheral and hepatic insulin resistance. They found that in men only was dSAT associated with peripheral and hepatic insulin resistance (37). An explanation for the clear difference in women with obesity between dSAT and sSAT in its association with hepatic steatosis and metabolic syndrome may be the contrasting morphological cellular aspects of dSAT and sSAT. As mentioned earlier, dSAT is more lipolytically active than sSAT, which can contribute to elevated levels of FFAs in the circulation (18). This would also explain the correlation with hepatic fat content in men. A potentially protective effect of sSAT in the development of hepatic steatosis is its high level of adiponectin compared to dSAT (21). Adiponectin has been considered to be able to protect hepatocytes from TG accumulation, probably by increasing β-oxidation of FFAs and/or decreasing de novo FFA production within hepatocytes (38, 39). Adipocytes in sSAT may have an increased cell size because they have an increased ability for fatty acid storage. In contrast, the presence of smaller cells in dSAT may be a reflection of their relative inability to store fat, subsequently leading to inflammation and stress signals (40). In addition, dSAT and sSAT have different mRNA expressions of inflammatory, lipogenic, and lipolytic target genes (21, 29); ADIPOR1, an mRNA expression of adiponectin, is higher in sSAT compared with dSAT, whereas the mRNA expressions of leptin, hormone-sensitive lipase, and lipoprotein lipase showed higher expression levels in dSAT compared with sSAT. The expression of genes involved in inflammatory pathways, such as interleukin-6 and MCP1, and genes involved in fatty acid synthesis, such as FASN, are higher in dSAT compared with sSAT (21, 29). In addition, Walker and colleagues also found an intermediate phenotype of dSAT with respect to the gene expression profiles of 11β-HSD1, leptin, and resistin (41).

Based on the multivariate correlations between the abdominal adipose tissue compartments and the circulating concentrations of the metabolic parameters glucose, TGs, and HDL-C, dSAT seems to be an intermediate phenotype between sSAT and VAT, especially in men. This finding in men has been described earlier by Marinou et al (29). In their study, fasting glucose was associated with dSAT only in men. Another study showed that both VAT and dSAT were associated with insulin-stimulated glucose use, HDL-C, and TGs, whereas sSAT showed weaker association with all these parameters (42). This study also performed no sex-stratified analysis.

The main strength of our study is the large population size, which enabled us to investigate the differences between men and women. The assessment both of adipose tissue compartment volumes and hepatic fat content by MRI is another strength of the present study. Our study also has a number of limitations. Because of the cross-sectional design, no causal interpretations can be made of the associations observed. We included participants only of Western European ancestry, which precludes extrapolation of our results to other ethnic groups. As we included individuals who were aged 55 to 80 years and all women were postmenopausal, our results cannot be extrapolated to other age groups. Next to this, we included only individuals with overweight and obesity, which could weaken the strength of the associations found in comparison to analyses of cohorts including individuals across a wide range of BMI.

In conclusion, VAT as measured by MRI is positively correlated with hepatic steatosis in men and women. Men have relatively more dSAT compared to women, and only in men was dSAT correlated positively with hepatic fat content. Moreover, dSAT in men was associated negatively with HDL-C in multivariate analyses, whereas in women with obesity sSAT was correlated negatively with hepatic fat content. dSAT seems to be an intermediate metabolic phenotype between sSAT and VAT. Therefore in future studies it is important to distinguish between dSAT and sSAT and between men and women, as we have shown a different impact on metabolic traits.