Hepatic Fat and Macrophages Are Increased in Livers of Diabetic Patients without Non-Alcoholic Fatty Liver Disease

Introduction: Diabetes mellitus (DM), especially type 2, is strongly associated with non-alcoholic fatty liver disease (NAFLD). Recent studies indicate that particularly in DM patients, “simple” liver steatosis can progress into more severe disease. However, little is known about putative hepatic histopathological changes in DM patients without NAFLD. In this study, we therefore analysed fat content and inflammatory cell infiltration in the livers of deceased DM and non-DM patients without NAFLD, and analysed age/sex effects hereon. Methods: Hepatic fat and inflammatory cells were studied through (immuno)histochemical analysis in liver tissue from 24 DM patients and 66 non-diabetic controls, without histopathological characteristics of NAFLD. Results: We observed a 2-fold increase in fat percentage/mm2 and a near 5-fold increase in the number of fat-containing cells/mm2 in DM patients compared to non-diabetic controls. Fat content was significantly higher in patients with type 2 DM, but not type 1 DM, compared to non-diabetic controls, while the number of CD68+ cells/mm2 was significantly elevated in both DM groups. Conclusion: Hepatic fat and number of macrophages are increased in patients with DM without NAFLD, which may reflect a higher risk on development of steatosis and steatohepatitis.

Both "simple" steatosis (>5%) without inflammation and steatosis with mild inflammation (small foci of inflammatory cells, mainly lymphocytes and macrophages) have so far been regarded as relatively benign with low risk of progression [6].Recent studies however showed that steatosis with mild inflammation, although insufficient for NASH diagnosis, can potentially progress into NASH and cirrhosis [6,7].This is supported by previous findings indicating inflammation as a predictor of (progression of) fibrosis [8,9].Additionally, these studies also found that "simple" steatosis can progress into more severe disease, particularly in patients with diabetes [6,7].Determining the onset of steatosis may thus be relevant for early intervention.
Sex and age differences are known to play an important role in DM prevalence and pathogenesis and have been well described [10,11], but are only relatively recently investigated in the NAFLD spectrum [12].However, extensive research already unravelled many sex-and age-related differences in NAFLD pathogenesis to the point where sex and age are wellknown risk factors of NAFLD development [12,13].In general, men have a higher NAFLD prevalence, but when menopause is included as a factor it seems this elevation only holds true when compared to pre-menopausal women (<50-60 years), whereas NAFLD prevalence is similar or lower in men when compared to post-menopausal women [12,13].Whether these sex and age differences are already present prior to NAFLD is yet unknown.
We studied hepatic fat and inflammation in patients without diabetes and liver disease, to examine putative baseline differences due to sex and age.Subsequently, we assessed possible effects of diabetes on hepatic fat and inflammation by examining patients with diabetes without NAFLD in comparison with the non-diabetic controls.

Patients
Liver tissue was obtained during autopsy from 24 DM patients and 66 non-diabetic controls, which were later grouped per sex and age.After excision, liver tissue was immediately fixed in formalin and subsequently embedded in paraffin.This study was approved by and performed according to the guidelines of the Ethics Committee of the AUMC, location VUmc, and conforms to the principles of the Declaration of Helsinki.Use of the leftover material after the pathological examination has been completed in part of the patient contract in our hospital.All liver sections were pathologically evaluated to exclude NAFLD, meaning a steatosis lower than 5% and the absence of fibrosis and hepatocyte ballooning.

Immunohistochemistry
The paraffin-embedded liver tissue was cut into 4-μm-thick sections and mounted onto microscope slides.The slides were first deparaffinised in xylene and dehydrated in graded ethanol (100-96-70%).The haematoxylin-eosin staining was performed according to standard protocol.Slides were stained with haematoxylin (Klinipath #4085) for 5 min and blued with tap water for 5 min, and then incubated with eosin (Merck #1.15935) for 30 s. Slides were then dehydrated in 100% ethanol and covered.
For CD45 and CD68, the slides were incubated in methanol containing 0.3% H 2 O 2 for 30 min to block endogenous peroxidases.Antigen retrieval was performed by boiling the slides in citrate buffer (pH 6.0) for 20 min followed by cooling down for 15 min (for CD68 staining).All sera and antibodies were diluted in Normal Antibody Diluent Solution (ImmunoLogic), and incubations were at room temperature unless otherwise specified.The slides were incubated with either monoclonal mouse anti-human CD45 (1:100 dilution, Dako #M0701) or mouse anti-human CD68 (1:500 dilution, Dako #M0814) for 1 h.The slides were rinsed with tris-buffered saline and incubated with an HRP-conjugated antimouse (undiluted, Dako #K4001) for 30 min.After tris-buffered saline washing, staining was visualised with 3,3' diaminobenzidine (Dako) for 8-10 min.Finally, the slides were counterstained with haematoxylin, dehydrated in 100% ethanol, and covered.

Tissue Analysis
The stained slides were scanned (Philips IntelliSite Pathology Solution, v3.2) into Digipath software (Philips, v3.3).Using a grid overlay, multiple 1 mm 2 selections of the scanned images were exported to ImageJ software (v1.8.0_172) for analysis.The fat percentage was determined on 20 1 mm 2 selections/patient from HE-stained tissue using colour deconvolution (HE).For this, blood vessels were excluded using polygon selection.The number of fatcontaining cells/mm 2 was determined on these same selections using a particle analyser screen (at level 30-infinity).The validity of this screen was verified with manual cell counting.To quantify the number of inflammatory cells in the tissue, five ad random 1 mm 2 selections were used to count extravascular CD45-positive lymphocytes (manually) or CD68-positive macrophages (digitally via ImageJ software (v1.8.0_172)).The average cell size of the CD68positive macrophages was later determined via dividing the average cell surface in mm 2 (ImageJ software v1.8.0-172) by the number of cells (online suppl.Fig. 1; for all online suppl.material, see https://doi.org/10.1159/000531542).Additionally CD68positive macrophages were stratified by hepatic location, subtracting the number of cells located in the portal triad from the total number, thus presenting the number of parenchymal CD68positive macrophages (online suppl.Fig. 2).Representative histological images of all groups are provided in online supplementary Figures 3-5.

Statistical Analysis
Data analysis was performed with Prism v.4.0 (GraphPad Software, La Jolla, CA).Gaussian distribution was assessed with a Shapiro-Wilk normality test.Accordingly, a parametric unpaired t test or a Mann-Whitney U test was used to analyse differences between men and women, and between the non-diabetic and diabetic patient groups.Patients with DM1 and DM2 were compared to non-diabetic controls with a one-way ANOVA or a Kruskal-Wallis test according to normality.Further comparison between non-diabetic and diabetic men and women was analysed with a two-way ANOVA combined with Tukey's multiple comparison test.Pearson correlation coefficients were calculated for the correlation analyses.Data values are displayed as mean ± standard deviation, and p < 0.05 was considered statistically significant for all analyses.

Patient Characteristics
We included 66 non-diabetic patients and 24 diabetic patients, of whom 8 with DM1 and 18 with DM2 (Table 1).The age of the non-diabetic controls (59 ± 22 years) and of the patients with diabetes (66 ± 21 years) did not differ significantly (p = 0.2).However, the age of patients with DM2 (74 ± 14 years) was significantly higher compared to patients with DM1 (48 ± 23; p = 0.002) and non-diabetic controls (p = 0.02).The sex distribution did not differ between patients with and without diabetes (p = 0.8).

No Effect of Sex or Age on Hepatic Fat and Inflammation in Patients without Diabetes
Hepatic fat in the liver tissue of patients without diabetes could be observed on a standard HE staining (Fig. 1a).There were no significant differences between men and women in either fat percentage or number of fat-containing cells (Fig. 1b, c).Additionally, there were no differences in CD45-positive cells/mm 2 (Fig. 1d) and CD68-positive cells/mm 2 (Fig. 1e) between men and women.We also examined if these factors differed between different age groups, namely, 0-50 years, 51-70 years, and 71-100 years (Fig. 1f-i), but again found no significant differences.

Significantly More Hepatic Fat and Inflammation in Patients with Diabetes
Liver tissue of patients without diabetes contained only 0.4 ± 0.9% of fat/mm 2 and 19.9 ± 47.8 fat-containing cells/mm 2 .Both the fat percentage/mm 2 (0.7 ± 1.1, p = 0.0007) and the number of fat-containing cells/mm 2 (94.3 ± 113.5, p < 0.0001) were significantly higher in the patients with diabetes compared to the non-diabetic controls (Fig. 2a, b), even with all patients having a steatosis presence below the 5% threshold value of NAFLD diagnosis.While there was no difference in CD45-positive cells/mm 2 , the number of CD68positive cells/mm 2 was significantly increased in the patients with diabetes (552.5 ± 309.5, p = 0.0002) in comparison with the non-diabetic controls (345.9 ± 164.5) (Fig. 2c, d).Both CD45-and CD68-positive cells were diffusely present in the liver tissue.
Notably, the number of fat-containing cells/mm 2 was increased in the livers of women with DM (120.8 ± 145.5, p < 0.0001), but not in men (Fig. 2i, j).However, the concomitant increase in average percentage of fat/mm 2 in the livers of diabetic women did not reach statistical significance (0.8 ± 1.0, p = 0.2).The number of CD45positive cells/mm 2 did not differ between non-diabetic and diabetic men and women (Fig. 2k).In contrast, significantly more CD68-positive cells/mm 2 were counted in the livers of both men (538.8 ± 308, p = 0.03) and women with DM (570.2 ± 327.1, p = 0.05) compared to non-diabetic controls (Fig. 2l).Notably, there were no significant differences found between men and women within the diabetic and non-diabetic groups.Additionally, no correlations between liver fat and CD68-positive cells were found in either the control group or the DM1 and DM2 groups.

Discussion
In this study, we aimed to determine putative effects of age, sex, and diabetes on steatosis and inflammation of the liver in patients that were without clinical diagnosis of NAFLD and who did not meet the histopathological criteria of NAFLD at autopsy.We found that hepatic fat and macrophages were increased in DM patients compared to non-diabetic controls.While this increase Liver Fat is Higher in Diabetic Patients without Liver Disease in hepatic fat was observed only in DM2 patients and not in DM1 patients, the number of macrophages in the liver was significantly elevated in both DM1 and DM2.No differences in hepatic fat or inflammation were observed between men and women, nor between patients of different ages, in either DM patients or non-diabetic controls.
The relationships between hepatic steatosis, hepatic inflammation, and insulin resistance are not entirely clear.Firstly, although non-alcoholic hepatic steatosis is generally considered to be benign, previous studies using paired serial biopsies taken over time have shown that a substantial proportion of patients with simple hepatic steatosis with no or mild inflammation progress to more active forms, including steatohepatitis, that can have significant clinical consequences [6,9].Secondly, patients with DM, especially DM2, and to a lesser extent also DM1, have a significantly higher prevalence of NAFLD than non-diabetics [1,2], suggestive of a causal relationship between DM and NAFLD development.Indeed, NAFLD and DM2 share key pathological factors such as obesity and insulin resistance [14].Our observation of more liver fat and more inflammation in DM2 patients without NAFLD may thus suggest that these patients may be more prone to develop NAFLD.Alternatively, hepatic steatosis can be disconnected from insulin resistance, and thus DM2, as has been shown in patients with familial hypobetalipoproteinaemia [15] and may simply be a response to increased caloric consumption.However, although DM2 generally is associated with obesity and physical inactivity, we unfortunately lack The age distribution of patients with diabetes was compared to the non-diabetic controls (indicated with *) and compared between DM1 and DM2 (indicated with † ) with a t test, displayed as mean ± standard deviation.The distribution of sex was analysed with a Fischer's exact test.p < 0.05 was considered statistically significant.
information on these characteristics from our patients and could therefore not relate our findings in the liver to lifestyle and body weight.
In addition to an increase in fat, we found an increase in the number of CD68+ macrophages in the livers of DM2 patients.This increase in macrophages, which presumably in majority are Kupffer cells (liver-resident macrophages), but may include infiltrated macrophages as both express CD68, may be a consequence of increased liver fat.Indeed, expansion of Kupffer cells has been shown to be an early feature of liver steatosis in NAFLD patients [16], and there is ample evidence that Kupffer cells are critically involved in the pathogenesis of NAFLD and NASH [17].Moreover, depletion or inhibition of Kupffer cells lowered insulin resistance and hepatic steatosis, and exerted anti-obesity effects in mice subjected to high-fat or high-sucrose diets [18,19], pointing to a key role of Kupffer cells herein.Interestingly though, we also found an increase in Liver Fat is Higher in Diabetic Patients without Liver Disease macrophages in the livers of DM1 patients, who had a hepatic fat content similar to the non-diabetic controls.Furthermore, we did not find an association between fat content and macrophage numbers in the liver in any of the study groups (DM1, DM2, controls).These results thus argue against a direct causal link between increased hepatic fat content and the increase in liver macrophages in our DM patients.
Besides increasing in number, Kupffer cells can also increase in size in response to an abundance of fat and cholesterol, through receptors such as macrophage scavenger receptor 1 [20,21].Excess uptake can result in the formation of hepatic foamy macrophages, which can further lead to formation of Kupffer cell aggregates in NAFLD [22].However, despite the increased fat content and number of CD68+ macrophages in patients with DM, we did not find a difference in average CD68+ cell size compared to non-DM patients, suggesting these foamy macrophages do not manifest themselves in our DM patients without NAFLD.
We found no association between age or sex and the hepatic fat content and inflammation in patients with or without diabetes.The livers of men and women differ profoundly in gene expression and metabolism and may be considered as metabolically distinct organs [23].This is also reflected in the sex differences found in the prevalence and severity of NAFLD.Both are higher in men than in pre-menopausal women, while postmenopausal women showed a higher prevalence than men [12,24].Additionally, histological analysis showed that pre-menopausal women with NAFLD had more severe lobular inflammation and hepatocyte injury, but less fibrosis than both men and post-menopausal women [12].Our results now show that these sex differences apparently do not manifest itself in the fat content and inflammation extent of the liver within adult patients, with and without DM, without liver disease such as NAFLD.Albeit, we could not adequately study putative associations between hepatic fat/inflammation and age due to the limited number of patients.These results are in line with previous studies that found no sex differences in liver fat, as assessed using proton spectroscopy, in healthy persons (without diabetes) [25] or in the prevalence of fatty liver, as assessed using CT scan, in DM2 patients without clinical liver disease [26].However, a recent study showed that while magnetic resonance-assessed hepatic fat levels were similar between men and women with DM2, plasma levels of liverderived triacylglycerol were lower in women than in men.Moreover, in non-diabetic participants both hepatic fat levels and plasma liver-derived triacylglycerol levels were lower in women than in men [27].Similarly, our observed significant increase in the number of fatcontaining cells in diabetic women but not in diabetic men may point to a sex-related difference prior to liver disease development in patients with DM.In conclusion, in our cohort patients with DM and without liver disease had more fat and CD68+ macrophages in the liver than non-diabetic controls, which may reflect a higher risk on development of steatosis and steatohepatitis.

Study Limitations
In this study, we retrospectively studied the livers of autopsied patients.Although this allowed detailed histopathological analysis of liver fat content and quantitative assessment of inflammation, there was little additional patient information available, such as duration of diabetes, BMI, type, and duration of medicine use.We thus could not analyse possible effects of these factors on the results.Liver Fat is Higher in Diabetic Patients without Liver Disease

Fig. 1 .
Fig. 1.Hepatic fat and inflammation do not differ in non-diabetic patients of different sex and age.a Example of a standard HE staining visualising lipid droplets in liver tissue, a CD45 staining visualising leukocytes, and a CD68 staining visualising macrophages.Black arrows indicate the lipid droplets and CD45+ lymphocytes in the HE and CD45 staining, respectively.Hepatic fat was quantified as percentage fat/mm 2 (b) and number of fat-containing cells/mm 2 (c) in patients without diabetes, divided by sex (b, c) or age (f, g).Hepatic inflammation was assessed through presence of CD45+ lymphocytes and CD68+ macrophages, also per sex (d, e) and age (h, i).An unpaired t test or Mann-Whitney U test was used for analysis, and data are presented as mean ± standard deviation.*p < 0.05.Original magnification is ×200.

Fig. 2 .
Fig. 2. Patients with diabetes have more hepatic fat and macrophages than non-diabetic controls.Hepatic fat and inflammation were assessed in liver tissue of patients without (CTRL) and with diabetes (DM), both in total (a-d) and divided by type of diabetes (e-h) or sex (i-l).The hepatic fat was quantified as percentage fat/mm 2 and number of fat-containing cells/mm 2 , while inflammation was assessed through presence of CD45+ lymphocytes and CD68+ macrophages.An unpaired t test or Mann-Whitney U test was used for analysis, and data are presented as mean ± standard deviation.*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.