Serum Dehydroepiandrosterone Sulphate Levels in Patients with Non-Alcoholic Fatty Liver Disease

□ ORIGINAL ARTICLE □

Masafumi Koga 1, Hiroshi Saito 1, Mikio Mukai 1, Toshiji Saibara 2 and Soji Kasayama 3

Abstract

Backgound Dehydroepiandrosterone (DHEA) is an adrenal hormone reported to prevent body weight gain, diabetes mellitus and atherosclerosis. We hypothesized that DHEA is involved in the pathophysiology of nonalcoholic fatty liver disease (NAFLD) often associated with obesity and insulin resistance. In this study, we aimed to examine the clinical significance of serum DHEA sulfate (DHEAS) in patients with NAFLD. Methods We determined serum DHEAS, serum alanine aminotransferase (ALT), serum lipids, plasma fasting glucose and insulin levels in 158 Japanese men who had neither viral hepatic diseases nor alcohol intake exceeding 20 g/day. NAFLD was diagnosed by the presence of fatty change of the liver by echotomographic examination. Results Among the study subjects, 69 were diagnosed as having NAFLD. Their serum DHEAS levels were significantly higher than in 89 subjects without NAFLD. Serum DHEAS levels in 19 NAFLD patients with elevated ALT levels (>40 U/L) were significantly higher than in the other 50 NAFLD patients with normal ALT levels (!40 U/L). Multivariate regression analysis demonstrated that serum ALT was positively corre- lated with serum DHEAS, serum triglyceride and body mass index. Conclusion Serum DHEAS levels are increased in patients with NAFLD with elevated ALT levels. In- creased serum DHEAS may be a component of the pathophysiology of NAFLD.

Key words: dehydroepiandrosterone, non-alcoholic fatty liver disease, fatty liver, alanine aminotransferase, non-alcoholic steatohepatitis

(Intern Med 50: 1657-1661, 2011) (DOI: 10.2169/internalmedicine.50.4682)

aminotransferase (ALT) to that with elevated ALT remain Introduction undetermined. Dehydroepiandrosterone (DHEA) is a hormone mainly Non-alcoholic fatty liver disease (NAFLD) is the most produced in the adrenal cortex. In addition to its androgenic common liver disease in affluent countries (1). Sometimes properties, DHEA has been reported to exert some benefi- NAFLD progresses to non-alcoholic steatohepatitis (NASH) cial effects on obesity, diabetes mellitus, atherosclerosis and and liver cirrhosis (2-4). Some inflammatory cytokines and ageing (10, 11). We, thus, hypothesized that DHEA may be adipose tissue-derived cytokines are suggested to be in- involved in preventing the development and progression of volved in this progression (5-7) since the association of NAFLD. Recently, however, it has been shown that levels of NAFLD with obesity and/or the metabolic syndrome has serum DHEA sulphate (DHEAS), the sulphated congener of been indicated (8). This is also the case in subjects with he- DHEA, were rather reduced in patients with histologically patic steatosis. Although obesity and insulin resistance have diagnosed advanced NASH (12, 13). These observations been shown to play important roles in the development of probably reflect the reduction of DHEA sulphonation, and NAFLD (9), the precise mechanisms for the progression therefore it is also important to determine serum DHEAS from simple steatosis without elevated serum alanine levels in the earlier stage of NAFLD not progressing to

１Department of Internal Medicine, Kinki Central Hospital, Japan, ２Department of Gastroenterology and Hapatology, Kochi Medical School, Ja- pan and ３Department of Medicine, Nissay Hospital, Japan Received for publication October 14, 2010; Accepted for publication April 11, 2011 Correspondence to Dr. Masafumi Koga, [email protected]

1657 Intern Med 50: 1657-1661, 2011 DOI: 10.2169/internalmedicine.50.4682

Table 1. Clinical Characteristics of the Subjects with or impaired visualization of the hepatic vein borders or the dia- without NAFLD phragm (16). When the fatty change was observed in the study patients, they were diagnosed to have NAFLD. Fur- NAFLD – + p thermore, we judged the degree of fatty liver as moderate or n 89 69 – more when the findings of impaired visualization of the he- age (years) 51.5 ± 6.9 50.0 ± 7.0 0.197 patic vein borders or the diaphragm were observed. Other- BMI (kg/m2) 22.8 ± 2.5 26.6 ± 3.4 <0.001 wise we judged as mild degree of fatty liver. serum AST (U/L) 19 ± 4 27 ± 13 <0.001 Statistical analyses serum ALT (U/L) 20 ± 6 27 ± 13 <0.001 Data are shown as means ± SD. Comparisons of serum serum triglyceride (mg/dL) 116 ± 53 197 ± 99 <0.001 DHEAS levels between groups were analyzed with a Stu- serum HDL cholesterol (mg/dL) 55 ± 14 48 ± 11 <0.001 dent’s t-test. Univariate as well as stepwise multivariate re- serum LDL cholesterol (mg/dL) 125 ± 26 126 ± 25 0.782 gression analyses were performed to assess the effects of fasting plasma glucose (mg/dL) 103 ± 19 109 ± 23 0.082 variables on serum ALT levels. For the stepwise multivariate HOMA-%S (%) 187 ± 80 109 ± 52 <0.001 regression analyses, the F value for the inclusion of the vari- DHEAS (ng/mL) 1464 ± 661 1814 ± 842 0.004 ables was set at 4.0. In the univariate as well as multivariate regression analyses, serum ALT and serum DHEAS were logarithmically transformed, to correct for skewed distribu- tion. These analyses used a StatView computer program NASH. The aim of this study was to analyze the clinical (Version 5.0). The statistical differences were considered to significance of DHEAS levels in patients with NAFLD. be significant at p<0.05.

Materials and Methods Results

Of the 158 study subjects, 69 were diagnosed as having Subjects NAFLD, whereas the other 89 were not. In the patients with We initially recruited a total of 278 Japanese men who NAFLD BMI, AST, ALT, triglycerides and low density lipo- visited the Health Care Center at Kinki Central Hospital for protein (LDL) cholesterol were significantly higher than in health examinations. Of these subjects we excluded subjects the subjects without it (Table 1). HDL cholesterol and with viral hepatic diseases (positive hepatitis B surface anti- HOMA-%S were significantly lower in the patients with gen test, positive hepatitis C antibody test and/or past histo- NAFLD than in the subjects without it. Serum DHEAS lev- ries of hepatitis B or hepatitis C), diabetes mellitus, renal els were on average 1,810 ± 840 ng/mL in the patients with disease and malignant disease. Subjects who ingested daily NAFLD, which was significantly higher (p=0.0039) than alcohol exceeding 20 g/day and subjects who had serum those in the subjects without it (1,460 ± 660 ng/mL) (Ta- ALT>40 U/L without fatty change of the liver were also ex- ble 1, Fig. 1A). There was a significant difference between cluded. A total of 158 men satisfied the above admission serum DHEAS levels in 29 subjects with a mild degree of criteria. Their average age was 50.8 ± 7.0 years and body fatty liver and in 43 subjects with moderate or more degree mass index (BMI) was 24.4 ± 3.5 kg/m2. The institutional of fatty liver (1,890 ± 904 ng/mL vs. 1,769 ± 810 ng/mL; committee approved the protocol of this study, and all par- p=0.5660). Serum DHEAS levels were not associated with ticipants gave their informed consent. BMI (R=0.160, p=0.0524) or fasting plasma glucose (R=- 0.022, p=0.7819), while they showed a weak but significant Laboratory methods association with body weight (R=0.155, p=0.0443) and Fasting plasma glucose, serum insulin, serum alanine HOMA-%S (R=-0.180, p=0.0216). Among 69 patients with aminotransferase, serum total cholesterol, high density lipo- NAFLD, 19 had elevated ALT levels (>40 U/L). Their se- protein (HDL) cholesterol and serum triglycerides were de- rum DHEAS levels were 2,380 ± 950 ng/mL, which were termined by means of standard laboratory assays. Serum significantly higher (p<0.0001) than those in the other 139 DHEAS was measured by radioimmunoassay (DPC DHEA- subjects with normal ALT levels (!40 U/L) (1,510 ± 670 S kit, Mitsubishi Kagaku Iatron, Tokyo, Japan). Inter- and ng/mL) (Fig. 1B). They were also significantly higher (p= intra-assay coefficients of variation in this assay were 12.0% 0.0003) than those in 50 NAFLD patients with normal ALT and 4.1% at the concentrations of 600 and 650 ng/mL (14). levels (1,600 ± 690 ng/mL). Serum ALT levels were associ- The homeostasis model assessment insulin sensitivity index ated positively with serum triglycerides (R=0.399, p< (HOMA-%S) was calculated from fasting plasma glucose 0.0001), BMI (R=0.418, p<0.0001), serum DHEAS (R= and serum insulin concentrations by using the correct 0.322, p<0.0001) and fasting plasma glucose (R=0.183, p= HOMA evaluation and a computer program (15). 0.0204), and inversely with HOMA-%S (R=-0.419, p< Fatty change of the liver was evaluated by echoto- 0.0001), age (R=-0.224, p=0.0043) and serum HDL choles- mographic examination by assessing hepato-renal contrast, terol (R=-0.186, p=0.0183) in the 160 study subjects (Ta-

1658 Intern Med 50: 1657-1661, 2011 DOI: 10.2169/internalmedicine.50.4682

A B

4,000 4,000 p<0.0001

p=0.0039 3,000 3,000

2,000 2,000 DHEAS (ng/mL) DHEAS (ng/mL) 1,000 1,000

0 0 (-) (+) ≦40 U/L >40 U/L NAFLD Serum ALT

Figure 1. (A) Serum DHEAS levels in 69 patients with NAFLD (hatched column) and in 89 subjects without it (open column). (B) Serum DHEAS levels in 19 patients with NAFLD who had elevated ALT levels (>40 U/L) (closed column) and in 139 men with normal ALT levels (≤40 U/L) (dotted column).

4.0 with NAFLD than in those without it. In addition, serum r = 0.325 DHEAS levels were higher in who had elevated serum ALT p < 0.0001 levels. Thus, it is suggested that serum DHEAS levels in- 3.5 crease in patients with NAFLD with high levels of serum ALT. 3.0 Serum DHEAS levels depend on adrenal DHEA production and its hepatic metabolism mediated by DHEA sul- log-DHEAS (ng/mL) fotransferase catalyzing sulfonation of DHEA to form 2.5 DHEAS. Although the activity and concentration of DHEA 0.5 1.0 1.5 2.0 2.5 sulfotransferase has been shown to be reduced in hepatic tis- log-ALT (U/L) sues derived from primary biliary cirrhosis, primary scleros- Figure 2. Correlation between serum log ALT levels and se- ing cholangitis, chronic active hepatitis and alcoholic cirrho- rum log DHEAS levels in 158 studied men. Dotted line repre- sis compared with normal liver (18), there is no report on sents the upper limit of control ranges (40 U/L) of serum ALT the content and activity of DHEA sulfotransferase in that level. derived from NAFLD. Thus, it is not known whether or not the increased serum DHEAS levels in NAFLD result from the increased sulfonation of DHEA. ble 2 and Fig. 2). Stepwise multivariate regression analysis Yoneda et al (19) studied the effects of DHEA treatment showed that serum DHEAS in addition to serum triglyc- on hepatic injury by concanavalin A-induced T lymphocytes erides and BMI were independently associated with serum in mice. They showed that DHEA reduced hepatic injury by ALT (Table 2). inhibiting several inflammatory mediators such as tumor necrosis factor α and macrophage mitogen inhibitory factor, Discussion and prevented the increase of serum ALT levels. Thus DHEA might have a protective effect against hepatotoxicity In the present investigation, we examined serum DHEAS in this mouse model. Although the roles of DHEA in levels in 158 men without viral hepatic diseases, diabetes NAFLD have not yet been proven, the increased serum mellitus, renal disease, malignant disease, or excess alcohol DHEAS levels in patients with NAFLD may reflect a com- drinking habits. Since serum DHEAS levels significantly pensatory increase of DHEA production to protect against differ between men and women (17), this study was per- hepatic damage. Recently, it has been reported that serum formed in men. The results clearly showed independent as- DHEAS levels were rather decreased in patients with his- sociation of serum DHEAS levels with serum ALT levels. tologically diagnosed NASH (12, 13). They have also shown Serum DHEAS levels were significantly higher in patients that serum DHEAS levels were lower in NASH patients

1659 Intern Med 50: 1657-1661, 2011 DOI: 10.2169/internalmedicine.50.4682

Table 2. Univariate and Multivariate Regression Analyses on Serum Log ALT Levels in 160 Studied Men

Univariate regression analysis Multivariate regression analysis Correlation Partial regression Variable coefficient p value coefficient F p value serum triglyceride 0.394 <0.0001 0.283 14.5 0.0084 BMI 0.418 <0.0001 0.270 13.4 0.0235 serum log-DHEAS 0.322 <0.0001 0.243 12.6 0.0055 HOMA-%S –0.419 <0.0001 Not entered age –0.224 0.0043 Not entered serum HDL cholesterol –0.186 0.0183 Not entered fasting plasma glucose 0.183 0.0204 Not entered serum LDL cholesterol 0.138 0.0839 Not entered Explanatory variables in multivariate regression analysis included are age, BMI, serum triglyceride, serum high density lipoprotein (HDL) cholesterol, serum low density lipoprotein (LDL) cholesterol, fasting plasma glucose, HOMA-%S and serum DHEAS. R2=0.307, F=22.4, and p<0.0001. with incremental fibrosis stage (12, 13). In the present in- endothelial cells, adrenal medulla cells, which are thought to vestigations, we failed to perform histological diagnosis of present in membrane fractions (30, 31). At present, there is the liver of patients with NAFLD and thus it was not deter- no data implying a molecular mechanism as to the relation mined how many of these patients represented NASH. We, of DHEAS with NAFLD. however, propose that most of them did not suffer from In conclusion, our results indicate that serum DHEAS lev- NASH since the frequency of NASH patients among els are increased in patients with NAFLD with increased se- NAFLD patients is suggested to be around 10 % (20). In- rum ALT levels. Serum DHEAS was found to be associated creased DHEAS in patients with NAFLD may reflect in- with serum ALT levels, independent of BMI, HOMA-%S creased adrenal secretion of DHEA for the prevention of the and serum triglycerides. Therefore, DHEAS may be a com- development and progression of hepatic damage (hepato- ponent of NAFLD, in addition to obesity and its related adrenal axis), while reduced DHEAS levels in patients with metabolic disorders. advanced stage of NASH may result mainly from the reduced sulphonation of DHEA. Further clinical and experi- The authors state that they have no Conflict of Interest (COI). mental studies are necessary to prove these hypotheses. There is an increasing number of reports demonstrating References that insulin resistance and the metabolic syndrome are involved in the development and progression of 1. Farrell GC, Larter CZ. Nonalcoholic fatty liver disease: from stea- NAFLD (5, 9). These are also supported by our results in tosis to cirrhosis. Hepatology 43 (2 Suppl 1): S99-S112, 2006. that the insulin sensitivity index HOMA-%S was inversely 2. Bugianesi E, Leone N, Vanni E, et al. Expanding the natural history of nonalcoholic steatohepatitis: from cryptogenic cirrhosis to associated with serum ALT. DHEA is known to have poten- hepatocellular carcinoma. Gastroenterology 123: 134-140, 2002. tial to improve insulin sensitivity in vivo (21-24). It also has 3. Adams LA, Lymp JF, St Sauver J, et al. The natural history of properties to increase insulin sensitivity in hepatocytes (25). nonalcoholic fatty liver disease: a population-based cohort study. In addition, it has been shown that DHEA can inhibit 11β- Gastroenterology 129: 113-121, 2005. hydroxysteroid dehydrogenase 1 expression in liver and adi- 4. Anguro O. Nonalcoholic fatty liver disease. N Engl J Med 346: 1221-1231, 2002. pose tissues (26). It is proposed that the overexpression of 5. Valenti L, Fracanzani AL, Dongiovanni P, et al. Tumor necrosis β 11 -hydroxysteroid dehydrogenase 1 in adipose tissues in- factor alpha promoter polymorphisms and insulin resistance in creases the local cortisol production and thereby leads to the nonalcoholic fatty liver disease. Gastroenterology 122: 274-280, metabolic syndrome-like status (27, 28). In the present 2002. study, serum DHEAS was a factor associated with serum 6. Kaser S, Moschen A, Cayon A, et al. Adiponectin and its receptors in non-alcoholic steatohepatitis. Gut 54: 117-121, 2005. ALT, independent of the insulin sensitivity index. Further- 7. Kamada Y, Tamura S, Kiso S, et al. Enhanced carbon more, serum DHEAS was inversely associated with the insu- tetrachloride-induced liver fibrosis in mice lacking adiponectin. lin sensitivity index HOMA-%S. Therefore, it is not plausi- Gastroenterology 125: 1796-1807, 2003. ble that insulin sensitivity per se is involved in the associa- 8. Marchesini G, Brizi M, Bianchi G, et al. Nonalcoholic fatty liver tion of serum DHEAS with NAFLD. disease: a feature of the metabolic syndrome. Diabetes 50: 1844- Steroid hormones are known to exert their biological ef- 1850, 2001. 9. Sanyal AJ, Campbell-Sargent C, Mirshahi F, et al. Nonalcoholic fects via specific nuclear receptors (29). Some effects of steatohepatitis: association of insulin resistance and mitochondrial DHEA are not exactly the same as those of androgens, and abnormalities. Gastroenterology 120: 1183-1192, 2001. thus the presence of DHEA receptors different from classi- 10. Techernof A, Labrie F. Dehydroepiandrosterone, obesity, and car- cal steroid receptors is postulated. High-affinity binding sites diovascular disease risk: a review of human studies. Europ J En- of DHEA have been shown in T lymphocytes, liver, vascular docrinol 151: 1-14, 2004.

1660 Intern Med 50: 1657-1661, 2011 DOI: 10.2169/internalmedicine.50.4682

11. Yanase T. Physiological significance of replacement therapy of de- 22. Kimura M, Tanaka S, Yamada Y, Kikuchi Y, Yamakawa T, Seki- hydroepiandrosterone. Intern Med 43: 156-158, 2004. hara H. Dehydroepiandrosterone decreases serum TNFα and re- 12. Charlton M, Angulo P, Chalasani N, et al. Low circulating of de- stores insulin sensitivity: independent effect from secondary hydroepiandrosterone in histological advanced nonalcoholic fatty weight reduction in genetically obese Zucker fatty rats. Endocri- liver disease. Hepatology 47: 484-492, 2008. nology 139: 3249-3253, 1998. 13. Sumida Y, Yonei Y, Kanemasa K, et al. Lower circulating levels of 23. Mukasa K, Kaneshiro M, Aoki K, et al. Dehydroepiandrosterone dehydroepiandrosterone, independent of insulin resistance, is an (DHEA) ameliorates the insulin sensitivity in older rats. J Steroid important determinant of severity of non-alcoholic steatohepatitis Biochem Mol Biol 67: 355-358, 1998. in Japanese patients. Hepatol Res 40: 901-910, 2010. 24. Kawano H, Yasue H, Kitagawa A, et al. Dehydroepiandrosterone 14. Kasayama S, Morita S, Otsuki M, et al. Independent association supplementation improves endothelial function and insulin sensi- of insulin-like growth factor-I with dehydroepiandrosterone sul- tivity in men. J Clin Endcrinol Metab 88: 3190-3195, 2003. phate in women in middle adulthood. Clin Endocrinol 22: 1701- 25. Yamashita R, Saito T, Satoh S, Aoki K, Kaburagi Y, Sekihara H. 1706, 2005. Effect of dehydroepiandrosterone on gluconeogenic enzymes and 15. Levy JC, Matthews DR, Hermans MP. Correct homeostasis model glucose uptake in human hepatoma cell line, HepG2. Endocr J 52: assessment (HOMA) evaluation uses the computer program. Dia- 727-733, 2005. betes Care 21: 2191-2192, 1998. 26. Apostolova G, Schweizer RAS, Balazs Z, Kostadinova RM, Oder- 16. Saadeh S, Younossi ZM, Remer EM, et al. The utility of radio- matt A. Dehydroepiandrosterone inhibits the amplification of glu- logical imaging in nonalcoholic fatty liver disease. Gastroenterol- cocorticoid action in adipose tissue. Am J Physiol Endocrinol Me- ogy 123: 745-750, 2002. tab 288: E957-E964, 2005. 17. Laughlin GA, Barrett-Connor E. Sexual dimorphism in the influ- 27. Stadeep TC, Walker BR. Pathophysiology of modulation of local ence of advanced aging on adrenal hormone levels: the Rancho glucocorticoid levels by 11β-hydroxysteroid dehydrogenase. Bernardo Study. J Clin Endocrinol Metab 85: 3561-3568, 2000. Trends Endocrinol Metab 12: 446-453, 2001. 18. Elekima OT, Mills CO, Ahmad A, et al. Reduced hepatic content 28. Stewart PM, Krozowski ZS. 11 beta-hydroxysteroid dehydroge- of dehydroepiandrosterone sulphotransferase in chronic liver dis- nase. Vitam Horm 57: 249-324, 1999. eases. Liver 20: 45-50, 2000. 29. O’Malley BW. Results of a search for the mechanisms of steroid 19. Yoneda M, Wada K, Katayama K, et al. A novel therapy for acute receptor regulation of gene expression. Ann NY Acad Sci 1038: hepatitis utilizing dehydroepiandrosterone in the murine model of 80-87, 2004. hepatitis. Biochem Pharmacol 68: 2283-2289, 2004. 30. Okabe T, Haji M, Takayanagi R, et al. Up-regulation of high af- 20. Japan Society of Hepatology. Epidemiology of NAFLD. In: A finity dehydroepiandrosterone binding activity by dehydroepian- Guidance for Clinicians on NASH/NAFLD. 2nd ed. Bunkodo, To- drosterone in activated human T lymphocytes. J Clin Endocrinol kyo, 2010 (in Japanese). Metab 80: 2993-2996, 1995. 21. Ladriere L, Laghmich A, Malaisse-Lagae F, Malaisse WJ. Effect 31. Liu D, Dillon JS. Dehydroepiandrosterone activates endothelial of dehydroepiandrosterone in hereditary diabetic rats. Cell Bio- cell nitric-oxide synthase by a specific plasma membrane receptor chem Funct 15: 287-292, 1997. coupled to Galpha(i2,3). J Biol Chem 277: 21379-21388, 2002.

1661