- Email: [email protected]
Expression of xenobiotic and steroid hormone metabolizing enzymes in hepatocellular tumors of the non-cirrhotic liver
ARTICLE IN PRESS
Pathology – Research and Practice 205 (2009) 716–725 www.elsevier.de/prp
ORIGINAL ARTICLE
Expression of xenobiotic and steroid hormone metabolizing enzymes in hepatocellular tumors of the non-cirrhotic liver Susanne Haasa,, Sabine Merkelbach-Brusea, Christina Justenhovenb,c, Hiltrud Brauchb,c, Hans-Peter Fischera a
Institute of Pathology, Medical Faculty of the University of Bonn, Sigmund Freud Str. 25, D-53127 Bonn, Germany Dr. Margarete Fischer-Bosch Institute of Clinical Pharmacology, Auerbachstr. 112, D-70376 Stuttgart, Germany c University Tu¨bingen, Germany b
Received 10 March 2009; received in revised form 11 May 2009; accepted 11 June 2009
Abstract Hepatocellular adenomas (HCA) and some hepatocellular carcinomas (HCC) arise in the non-cirrhotic liver. Although the liver is involved in the metabolism of a huge number of exogenous and endogenous substances, little is known about the role of metabolic enzymes in the development of liver tumors in the absence of cirrhosis. We analyzed the expression of glutathione S-transferases (GST) and cytochrome P450 enzymes (CYP) in 23 HCA, 20 HCC, and 22 focal nodular hyperplasias (FNH) using immunohistochemistry. The liver tissue revealed consistent specific staining for GST alpha, CYP1A1, 1A2, 2E1, and 3A4. In HCA and HCC, GST alpha expression was significantly reduced (po0.001 and 0.043). Reduced GST alpha expression was significantly associated with steatosis in HCA and HCC (n ¼ 12, p ¼ 0.006), but not in non-neoplastic liver tissue. CYP3A4 expression was also reduced in HCA and HCC (p ¼ 0.03 and 0.02), and this was correlated with diabetes mellitus type 2 (p ¼ 0.02). In conclusion, HCA and HCC revealed changes in the expression of certain metabolic enzymes as compared with the non-neoplastic liver tissue or FNH. Therefore, reduced expression of GST alpha and CYP3A4 may indicate specific metabolic defects in the tumor tissue characterizing subgroups of HCA and HCC. r 2009 Elsevier GmbH. All rights reserved. Keywords: HCA; HCC; FNH; GST; CYP
Introduction The liver plays a major role in the metabolism of numerous endogenous and exogenous compounds. Main contributors of this pathway are phase I and phase II enzymes. The phase I system is composed mainly of cytochrome P450 (CYP) enzymes as well as peroxidases, oxidases, reductases, and dehydrogenases, Corresponding author. Tel.: +49 228 287 19244; +49 228 287 15030. E-mail address: [email protected] (S. Haas).
fax:
0344-0338/$ - see front matter r 2009 Elsevier GmbH. All rights reserved. doi:10.1016/j.prp.2009.06.003
which introduce a reactive group to the endogenous or exogenous substrate [18]. Some of the phase I metabolites are conjugated to polar groups by phase II enzymes. Phase II enzymes, e.g. glutathione S-transferases (GSTs), sulfotransferases, and UDP glucuronosyltransferases, catalyze conjugation reactions of endogenous and exogenous substances or of their phase I metabolites, producing usually inactive and water soluble compounds that can be excreted through urine or bile [30]. Regarding the development of liver tumors, it is of special interest that phase I and phase II enzymes are
ARTICLE IN PRESS S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
involved in the metabolism of steroid hormones, e.g. estrogen and testosterone. The vast majority of Hepatocellular adenomas (HCA) arise in women, and development of HCA is often associated with application of contraceptive or anabolic steroids. Anabolic steroids may also play a role in the development of a subset of hepatocellular carcinomas (HCC) in the absence of cirrhosis [12,21]. CYP1A1, CYP1B1, and CYP3A4 catalyze 2-, 4-, and 16-hydroxylation of 17-bestradiol and formation of 4-hydroxy estrone and 2hydroxy estrone [1,9,13]. GSTs are involved in the inactivation of catechol estrogen quinones. They catalyze glutathione conjugation of catechol estrogens quinones, the reactive intermediates of estrogen metabolism capable of binding to DNA [28]. For both GSTs and CYPs, there exist numerous genetic polymorphisms leading to differences in the individual metabolic capacity within a population [30]. In addition, activities of some CYP isoenzymes, such as CYP1A1 and CYP1B1, may be altered due to induction by environmental factors like cigarette smoke or local pollutants [22,26]. Therefore, functional alterations in GST and CYP activity may impact HCA or HCC risk in association with steroid hormone intake. Further risk factors for HCC in the non-cirrhotic liver include obesity and diabetes with non-alcoholic fatty liver disease [10,13,32]. In the recent years, subtypes of HCA were identified by genetic alterations, such as HNF1a and b-catenin mutations, and by histomorphological features [3–5,8]. Germ line mutations of the steroid hormone metabolizing enzyme CYP1B1 were found to be associated with HNF1a mutated HCA [23]. In HCC, a large number of genetic alterations and several carcinogenetic pathways have been described [6]. Based on the recently described diagnostic and molecular features of non-cirrhotic liver tumors, this study intends to systematically analyze expression of several CYP and GST enzymes in order to identify characteristic changes of enzyme expression.
Materials and methods Patients and tumor specimens Tissue sampling and use for this study were performed in accordance with standard guidelines of the ethics committee of the Medical Faculty of the University of Bonn. Paraffin-fixed tissue of 65 patients was included in this study. This collective consisted of 23 HCA, 20 HCC, and 22 focal nodular hyperplasias (FNH). Except for 8 cases of FNH, one HCA and one HCC diagnosed by punch needle biopsy, resection specimens were used for immunohistochemical staining. Sufficient amounts of non-tumorous liver parenchyma
717
were present for histological examination in 19 HCA, 18 HCC, and 14 FNH. The male to female ratio was 3:20 for HCA, 15:5 for HCC, and 5:17 for FNH. Liver cirrhosis was absent in all of the cases. There was no history of chronic hepatitis, alcohol abuse, or hepatitis B or C viral infection. Steatosis of the tumor or liver tissue was assessed if more than 10% of tumor cells or hepatocytes revealed steatotic changes. Among the 20 patients with HCC, one patient was suffering from the Alagille syndrome. Another patient was diagnosed with alpha 1 antitrypsin deficiency type PIZ. One patient had a Budd Chiari syndrome. Grading of HCC was well (G1) in 6 cases, moderate (G2) in 13 cases, and poor (G3) in one case.
Immunohistochemistry and antibodies Tissue specimens were fixed in 4% buffered formaldehyde and embedded in paraffin. Immunohistochemical stainings were performed on 4 mm sections with an immunostainer (Techmate 500; DAKO). The antigen–antibody binding was visualized by the avidin–biotin complex (ABC method) using 3-amino-9-ethylcarbazol (AEC) as chromogen. GST antibodies were purchased from Novo Castra, Newcastle upon Tyne, UK. Antibodies for CYP1A1, CYP3A4 and CYP2E1 were purchased from Natutec, Frankfurt am Main, Germany (Manufacturer Daiichi Pure Chemicals, Tokyo, Japan), and antibodies for CYP1B1 and 1A2 were purchased from Gentest, Woburn, USA. The specificity of the primary antibodies was ensured as described previously [15]. Liver fatty acid binding protein (FABP) antibody was purchased from Abcam, Cambridge, UK; anti-Glutamine Synthetase (GS) from BD Biosciences, NY, USA; monoclonal antibody to Glypican-3 (GPC3) from Biomosaics, Burlington, USA; anti-HSP-70 from Santa Cruz Biotechnology, USA; anti-b-catenin from Transduction Laboratories Lexington, KY, USA. Antibodies for determination of the estrogen receptor (ER) were derived from Novocastra (clone 6F11/2), for the progesterone receptor (PR) from DAKO (clone PgR636), for the androgen receptor (AR) from Antibodies online GmbH (Germany), clone AR441. Details about the dilutions of the primary antibodies are given in Table 1.
Evaluation of immunohistochemical stainings For hormone receptors (estrogen receptor, progesterone receptor, androgen receptor) and b-catenin, any nuclear staining was assessed positive. Cytoplasmic staining of GST and CYP enzymes was scored as follows: the percentage of stained cells (0–100%) was multiplied with the staining intensity (0–3) to give a
ARTICLE IN PRESS 718
S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
Table 1. Dilutions and staining patterns of the primary antibodies.
ER alpha PR AR GST pi GST mu GST alpha CYP1A1/2 CYP3A4/5 CYP2E1 CYP1A2 CYP1B1 FABP Glutamine synthetase Glypican-3 HSP-70 b-catenin a
Dilution
Staininga
1:50 1:50 1:75 1:750 1:400 1:1500 1:200 1:1000 1:200 1:100 1:1000 1:50 1:100 1:100 1:250 1:1000
N N N C C C/N C C C C C C C C/M C/N N
N ¼ nuclear, C ¼ cytoplasmic, M ¼ membrane.
maximum score level of 300. Score levels 440 were defined as positive expression. Tumor tissue and corresponding non-neoplastic liver parenchyma were assessed separately. For statistical testing, w2 analysis and Fisher’s exact test were used, if appropriate.
Sequence analysis for b-catenin mutations Extraction of genomic DNA from the tumor samples was performed using standard procedures. b-catenin exon 3 DNA was amplified by PCR using the following primers: 3-forward: 50 -tttgatggagttggacatgg-30 and 3reverse: 50 -ctgagaaaatccctgttccc-30 . Amplification of target DNA sequences was carried out in 50 ml of a reaction mixture containing an aliquot of extracted DNA, 10 mM TRIS–HCl, 1.5 mM MgCl2, 50 mM KCl, 100 mM of each dNTP, 0.1 mM of forward and reverse primer, respectively, and 1 U of Platinum Taq DNA polymerase (Invitrogen, Karlsruhe, Germany). Each PCR cycle consisted of a denaturation step at 94 1C for 40 s, primer annealing at 60 1C for 40 s, and extension at 72 1C for 35 s. An initial denaturation step was performed at 94 1C for three minutes and in the final cycle; the extension step was elongated to 5 min at 72 1C. Samples were amplified through 40 cycles. The PCR products were purified using MicroSpin columns (Amersham Biosciences, Freiburg, Germany). Template DNA concentrations for cycle sequencing were estimated by agarose gel electrophoresis. Bidirectional DNA sequencing of the entire exon and exon–intron boundaries was performed with the Big Dye Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems, Weiterstadt, Germany) on all
samples using the primers listed above. Cycle sequencing products were precipitated with 3 M sodium acetate and analyzed on a ABI PRISM 3130 capillary electrophoresis system (Applied Biosystems). The identity of the amplicon sequences was confirmed by database search (NCBI database: www.ncbi.nlm.nih.gov).
Results Expression patterns of GST and CYP enzymes in non-neoplastic liver parenchyma The majority of metabolic enzymes displayed reliable immunohistochemical staining and characteristic expression patterns in the non-neoplastic liver tissue (Table 2). A panacinar mosaic-like expression pattern of GST alpha with patchy staining was a constant finding. Expression was both nuclear and cytoplasmic. GST pi turned out to be a selective marker for bile duct epithelium of the portal tracts, while hepatocytes were always negative for GST pi (Fig. 1B). Consequently, no positive staining for GST pi was seen within HCA (Fig. 1C). Most of the CYP enzymes revealed a zonal distribution with intense positive staining in the centrilobular area and only weak staining in the periportal area. This staining pattern was most intense for CYP3A4 (Fig. 1F), to a lesser degree for CYP1A1, CYP1A2, and CYP2E1. Immunostaining of GST mu and CYP1B1 was usually weak or moderate with no zonal distribution pattern in the liver parenchyma, and was thus less significant. Table 2. Distribution of metabolic enzymes in the nonneoplastic liver parenchyma. No of cases with positive expressiona
Predominant staining
GST alpha
51 (100%)
GST mu GST pi
23 (46%) 0 (0%)
CYP1A1
51 (100%)
CYP1A2
48 (96%)
CYP1B1 CYP2E1
27 (54%) 51 (100%)
CYP3A4
51 (100%)
Panacinar, mosaic like Pancinar Bile duct epithelium Centrilobular, panacinar Centrilobular, panacinar Panacinar Centrilobular, panacinar Centrilobular
a
Scoring levels 4 40.
ARTICLE IN PRESS S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
719
Fig. 1. (A–C) Inflammatory HCA with marked steatosis and absent expression of GST alpha (A, right corner). Note the strong GST alpha expression in the steatotic non-neoplastic liver parenchyma (A, left corner). Selective staining of bile duct epithelium of the liver parenchyma by GST pi (B) and absent GST pi staining in portal tract like areas of the HCA (C). (D, E) Steatotic HCA with absent expression of CYP3A4 as compared with the non-neoplastic liver parenchyma (D). Focally positive expression of the progesterone receptor in the tumor tissue (D, inset). Absent staining of FABP in the HCA (E). In this tumor, GST alpha expression was also reduced (not shown). (F–H) Zonal expression of CYP3A4 in the liver parenchyma (left side) and patchy strong expression in the HCC (right side, F). Nuclear expression of b-catenin in the tumor tissue (bottom, G) and strong expression of glutamine synthetase (bottom, H).
Expression of GST and CYP enzymes in FNH, HCA, and HCC Table 3 demonstrates the staining intensities of the enzymes in the tumor tissue and in the non-tumorous liver parenchyma. For the majority of enzymes (GST mu, CYP1B1, 1A1, 1A2, 2E1), expression levels in the tumor tissue corresponded to the non-tumorous liver parenchyma. However, in a significant number of HCA and HCC, expression of GST alpha was absent or reduced (po0.001 and p ¼ 0.043). CYP3A4 expression
was also significantly decreased in HCA (p ¼ 0.03) and HCC (p ¼ 0.02). Reduction of GST alpha and/or CYP3A4 in HCC was independent of the tumor grading. Significant alterations in enzyme expression were not seen in FNH. In addition, the topical distribution of the CYP enzymes in FNH corresponded to the nontumorous liver tissue: expression was reduced in paraseptal areas and was stronger in the center of the portal tract equivalent. A distinct zonal expression of CYPs was also seen in most HCA. Especially CYP3A4 and CYP1A1 were expressed in the areas surrounding
ARTICLE IN PRESS 720
Table 3.
S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
Expression of metabolic enzymes in liver tumors and adjacent liver parenchyma (LP). HCA
HCA-LP
HCC
HCC-LP
FNH
FNH-LP
23
20
14
8 12
18 p ¼ 0.043 2 16
22
10 13
19 po0.001 0 19
6 16
1 13
GST mu Negative Positive
18 5
15 4
12 8
8 10
6 16
4 10
CYP1A1 Negative Positive
1 22
0 19
1 19
0 18
1 21
0 14
CYP1A2 Negative Positive
4 19
2 17
2 18
0 18
3 19
1 13
CYP1B1 Negative Positive
10 13
12 7
11 9
8 10
7 15
7 7
CYP2E1 Negative Positive
0 23
0 19
1 19
0 18
0 22
0 14
CYP3A4 Negative Positive
5 18
p ¼ 0.03 0 19
5 15
p ¼ 0.022 0 18
0 22
0 14
N GST alpha Negative Positive
Table 4.
Expression of diagnostic markers in liver tumors and adjacent liver parenchyma (LP). HCA
HCA-LP
HCC
HCC-LP
FNH
FNH-LP
23
20
18
22
14
9 11
5 13
9 13
4 10
N FABP Negative Positive
13 10
19 p ¼ 0.048 5 14
GS Negative Positive
21 2
18 1
11 9
p ¼ 0.012 17 1
15 7
12 2
GPC 3 Negative Positive
20 3
19 0
12 8
p ¼ 0.015 17 1
22 0
14 0
HSP70 Negative Positive
23 0
19 0
5 15
p ¼ 0.003 13 5
22 0
14 0
b-catenin Negative Positive
21 2
19 0
15 5
18 0
22 0
14 0
HRa Negative Positive
21 2
19 0
17 3
18 0
22 0
14 0
a
Hormone receptor positive: expression of ER and/or PR and/or AR.
ARTICLE IN PRESS S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
the efferent vein, corresponding to the center of the lobule in non-neoplastic liver parenchyma. No distinct pattern of enzyme expression was observed in HCC, usually exhibiting diffuse or patchy staining (Fig. 1F).
Expression of diagnostic markers and subtyping of HCA, HCC, and FNH Subtyping of the 23 HCA was performed based on histomorphological and immunohistochemical features. Histomorphology of 6 HCA corresponded to the type of inflammatory HCA (formerly ‘‘teleangiectatic FNH’’type) with inflammatory infiltrates, sinusoidal dilatation, and ductular reaction. The teleangiectatic FNH type HCA did not differ from other HCA regarding the expression of GST and CYP enzymes. One case revealed tumor steatosis (Fig. 1A–C). Expression of GS, GPC3, and HSP70 was significantly associated with HCC (p ¼ 0.012 for GS, 0.015 for GPC3, 0.003 for HSP70) (Table 4). Eighteen of 20 HCC expressed at least one of these markers. In the liver parenchyma, focal weak positive staining of GPC 3 and HSP 70 could be seen. GS expression was restricted to a few hepatocytic layers around the central vein. Table 5.
721
Nuclear expression of b-catenin in association with strong expression of GS was seen in five cases (4 HCC, 1 male HCA) (Fig. 1G, H). In three of these tumors, mutation or deletion of the b-catenin gene could be verified by genomic sequence analysis (p.S33P, p.S45P, p.Q28_H36del). In the remaining two cases, molecular analyses were not feasible. Four other HCC cases revealed faint indeterminate nuclear b-catenin staining and only weak positivity for GS. They were tested negative for b-catenin mutations.
HCC and HCA with special clinical, diagnostic, or histomorphologic features
Steatosis of the tumor was present in 1 FNH, 10 HCA (including one case of inflammatory HCA), and 2 HCC. Steatosis of non-tumorous liver parenchyma was observed in 17 cases, independent of the steatotic grade of the tumor tissue. Expression of GST alpha in steatotic HCC and HCA was significantly reduced compared to the surrounding liver tissue (p ¼ 0.006, Table 5). The adjacent liver parenchyma was always positive for GST alpha regardless of steatosis (Fig. 1A). Two of the steatotic HCA revealed positive expression of steroid hormone receptors (inset Fig.
Expression of metabolic enzymes and hormone receptors in subgroups of HCA and HCC. n
FABP negative
b-catenin positive
GST alpha negative
CYP3A4 negative
43
5
18
1 4
–
10 p ¼ 0.039 7 3
n(HCA+HCC) GST alpha Negative Positive
18 25
22 p ¼ 0.008 12 10
CYP3A4 Negative Positive
10 33
5 17
0 5
–
–
CYP1B1 Negative Positive
21 22
11 11
3 2
8 10
p ¼ 0.0244 8 2
GS Negative Positive
32 11
17 5
po0.001 0 5
16 2
HRa Negative Positive
38 5
18 4
4 1
15 3
Steatotic tumor Yes No
12 31
7 15
1 4
p ¼ 0.006 9 9
Diabetes Yes No
6 37
3 19
0 5
p ¼ 0.067 5 13
a
Hormone receptor positive: expression of ER and/or PR and/or AR.
9 1 p ¼ 0.073 7 3 1 9 p ¼ 0.02 4 6
ARTICLE IN PRESS 722
S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
Fig. 2. A 16-year-old male patient with Fanconi’s anemia developed multiple hepatic nodules up to 10 cm in diameter under androgen therapy (A). Resection specimen revealed one HCC with steatosis (B) and absence of GST alpha staining (not shown). There were also found HCA0 s without steatosis but positive expression of the androgen receptor (C, D). Analysis for b-catenin mutation was not feasible in this case.
1D). Lipofuscinosis of the tumors was seen in two further cases of steatotic HCA. FABP expression was absent or reduced in a significant number of HCA (n ¼ 13, p ¼ 0.048), as well as in 9 HCC and 9 FNH (Table 5). In HCA and HCC, absent FABP expression was significantly correlated with reduction of GST alpha expression (p ¼ 0.008, Fig. 1E). Although complete absence of FABP staining (compared to the positive staining of corresponding non-neoplastic liver tissue) was frequent in steatotic tumors, especially in HCA, there was no significant correlation with tumor steatosis. Diabetes mellitus type 2 was documented in 5 HCC, 1 HCA, and 2 FNH patients. In 4 male HCC patients with type 2 diabetes, CYP3A4 expression was reduced or absent in the tumors compared to the liver tissue (p ¼ 0.02). In 4 HCC and one HCA of diabetic patients, GST alpha expression was also reduced (p ¼ 0.07, Fisher0 s exact two-tailed p-value). Diabetes mellitus was not related to steatosis of the tumor tissue. Steroid hormone receptor expression was rare, with three HCC and two HCA expressing ER, PR, or AR. A remarkably strong expression of the estrogen receptor alpha, progesterone and androgen receptor was detected in one case of HCC. An additional special feature of this HepPar1-positive tumor was negativity
for GS, HSP 70, and Glypican C. This male patient had no history of exogenous hormone application. Two further patients with AR-positive HCC’s had undergone therapeutic androgen application. One case is demonstrated in Fig. 2. There was a trend towards reduced expression of CYP3A4 in hormone receptorpositive tumors (p ¼ 0.073, Fig. 1D). Expression of steroid hormone receptors was not related to specific diagnostic features in HCA or HCC (i.e., b-catenin expression, tumor steatosis, lack of FABP, inflammatory HCA). The liver parenchyma of all cases did not show any nuclear hormone receptor expression.
Discussion In this immunohistochemical study, we were able to demonstrate specific expression of several GST and CYP enzymes in hepatocellular tumors and in nonneoplastic liver tissue. The general strong cytoplasmic and nuclear staining of GST alpha in hepatocytes, as well as the restriction of GST pi expression to the bile duct epithelium, has been described before [19,20]. Inconsistent expression of GST mu may be explained
ARTICLE IN PRESS S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
by the high rate of the GSTM1 null polymorphism within the population [30]. Most of the CYP enzymes were expressed in zones 2 and 3 in accordance with the increased metabolic activity in the center of the liver lobule [14]. Weak or moderate panacinar staining was seen for CYP1B1, which is predominantly an extrahepatic enzyme [16]. In general, there were no discernible differences in enzyme expression regarding gender or age of the patients, nor was there an association with hepatic steatosis, diabetes mellitus or application of anabolic or contraceptive steroids. In FNH, distribution pattern and staining intensity of GST and CYP enzymes resembled the normal liver tissue. Zonality was imitated by weaker expression in the periseptal parenchyma and stronger expression in the expansive margins of the micronoduli. These findings are in line with the fact that FNH develop as benign hyperregenerative nodules of the liver tissue. In HCA and HCC, GST mu, CYP1A1, 1A2, and 2E1 expression corresponded to the non-neoplastic liver tissue. A faint zonality of CYP expression was also observed in HCA, but not in HCC. However, noticeable alterations in expression were detected for GST alpha and/or CYP3A4. Reduced expression of one or both enzymes in HCA and HCC could be assigned to distinct histological or clinical features. The first group of tumors with altered enzyme expression comprised steatotic HCA and HCC. In the majority of these tumors, there was a significant reduction in GST alpha expression as compared with the non-neoplastic liver tissue. Recently, steatosis and absence of FABP staining in HCA were found to be significantly associated with HNF1a mutations [3,4,33]. Steatosis in HNF1a-mutated HCA is supposed to be caused by increased lipogenesis activity [29]. There also was an association of lack of FABP staining in steatotic HCA in our study group, but we could not confirm a general, significant correlation of tumor steatosis (including inflammatory HCA and HCC) with absence of FABP. It is of note that FABP staining was significantly reduced in the GST alpha-reduced steatotic HCA and HCC. Therefore, it has to be discussed whether reduced GST alpha expression is a feature of tumor steatosis in general or whether it is an indicator for HNF1a mutation. The latter assumption may be supported by the finding that GST A transcriptional activity is regulated by HNF-1 [27]. A second small group of HCA and HCC was characterized by the expression of steroid hormone receptors. In two HCC, previous application of androgen was documented. One young patient developed multiple HCA and one HCC following androgen therapy for Fanconi0 s anemia. This patient remained free of disease for more than 6 years after cessation of hormone application. This indicates that ER, PR, or AR
723
expression in the tumor tissue may be induced by exogeneous hormone intake, especially by androgen application in men. However, steroid hormone receptor mediated signaling seems to promote tumor growth in a minority of HCC and HCA only. In part of the steroid hormone receptor-positive tumors, expression of the steroid hormone metabolizing enzyme CYP3A4 was reduced. Although this correlation did not reach significance due to the limited number of cases, there generally might be an inverse correlation between the presence of nuclear steroid hormone receptor expression on the one hand and the expression of steroid hormone metabolizing enzymes on the other hand. In breast cancer, we could demonstrate absence of ER/PR expression to be associated with strong overexpression of CYP1B1, which is one of the most important steroid hormone metabolizing enzymes in extrahepatic tissues [15]. Distinct polymorphisms of the CYP1B1 genotype (CYP1B1_1358_GG), leading to increased activity of CYP1B1, were found to be associated with ERa negativity in breast cancers [24]. The third group of liver tumors with altered enzyme expression is represented by HCA and HCC of patients with type 2 diabetes mellitus. Obesity and diabetes mellitus have been linked to HCC in several epidemiological studies [7]. In diabetic rats, combined hyperinsulinism and hyperglycemia leading to increased intracellular insulin signaling and growth stimulation was identified as a carcinogenic mechanism for the development of HCC [11]. Activity of CYP enzymes in diabetes may be altered by insulin itself [31] by several factors associated with diabetes such as elevated plasma ketone levels and altered growth hormone expression [2], or due to metabolization of anti-diabetic drugs [17]. In vitro experiments on rat hepatocytes showed that GST alpha expression is regulated by insulin and glucagon in an opposite manner [25]. Interestingly, diabetes of the patients was associated with mild steatosis of the liver parenchyma, but, except for one HCA case, no steatosis of the tumor tissue was seen. We could not demonstrate altered expression levels of CYP and GST enzymes in the non-neoplastic liver tissue of diabetic patients, indicating insulin effects. However, all diabetic tumors revealed reduced expression of either CYP3A4 or GST alpha or of both enzymes, and therefore, CYP3A4 and/or GST alpha expression are obviously down-regulated in HCC of diabetic patients without other risk factors. Finally, in spite of their different biologic behavior, HCA and HCC reveal similar alterations in enzyme expression, indicating that there are equal metabolic changes in these tumors independent of their dignity. Therefore, GST and CYP enzymes are no immunohistochemical markers discriminating benign from malignant liver neoplasms.
ARTICLE IN PRESS 724
S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
However, altered expression of GST alpha and CYP3A4 was associated with distinct tumor features of which tumor steatosis is the most prominent. Further studies will show whether reduced GST alpha expression in association with tumor steatosis is a result of HNF1a mutation, and whether there is a true relationship between enzyme expression and expression of steroid hormone receptors or diabetes mellitus in small HCA/ HCC subgroups.
References [1] A.F. Badawi, E.L. Cavalieri, E.G. Rogan, Role of human cytochrome P450 1A1, 1A2, 1B1 and 3A4 in the 2-, 4-, and 16alpha-hydroxylation of 17 beta estradiol, Metabolism 50 (2001) 1001–1003. [2] G.D. Bellward, T. Chang, B. Rodrigues, et al., Hepatic cytochrome P-450 induction in the spontaneously diabetic BB rat, Mol. Pharmacol. 33 (1988) 140–143. [3] P. Bioulac-Sage, S. Rebouissou, A. Sa Cunha, et al., Clinic, morphologic, and molecular features defining socalled teleangiectatic focal nodular hyperplasias of the liver, Gastroenterology 128 (2005) 1211–1218. [4] P. Bioulac-Sage, S. Rebouissou, C. Thomas, et al., Hepatocellular adenoma subtype classification using molecular markers and immunohistochemistry, Hepatology 46 (2007) 740–748. [5] O. Bluteau, E. Jeannot, P. Bioulac-Sage, et al., Bi-allelic inactivation of TCF-1 in hepatic adenomas, Nat. Genet. 32 (2002) 312–315. [6] S. Boyault, D.S. Rickman, A. de Reynies, et al., Transcriptome classification of HCC is related to gene alterations and to new therapeutic targets, Hepatology 45 (2007) 42–52. [7] S.H. Caldwell, D.M. Crespo, et al., Obesity and hepatocellular carcinoma, Gastroenterology 127 (2004) S97–S103. [8] Y.W. Chen, Y.M. Jeng, S.H. Heh, et al., P53 gene and wnt signalling in benign neoplasms: beta catenin mutations in hepatic adenoma but not in focal nodular hyperplasia, Hepatology 36 (2002) 927–935. [9] A.E. Cribb, M.J. Knight, D. Dryer, et al., Role of polymorphic human cytochrome P450 enzymes in estrone oxidation, Cancer Epidemiol. Biomarkers Prev. 15 (2006) 551–558. [10] J.A. Davila, R.O. Morgan, K.A. McGlynn, et al., Diabetes increases the risk of hepatocellular carinoma in the United States: a population based case control study, Gut 54 (2005) 533–539. [11] F. Dombrowski, C. Mathieu, M. Evert, Hepatocellular neoplasms induced by low-number pancreatic islet transplants in autoimmune diabetic BB/Pfd rats, Cancer Res. 66 (2006) 1833–1843. [12] H.A. Edmondson, B. Henderson, B. Benton, Liver-cell adenomas associated with use of oral contraceptives, N. Engl. J. Med. 294 (1976) 470–472. [13] H.B. El-Serag, T. Tran, J.E. Everhart, Diabetes increases the risk of chronic liver disease and hepatocellular carcinoma, Gastroenterology 126 (2004) 460–468.
[14] R. Gebhardt, Metabolic zonation of the liver: regulations and implication for liver function, Pharmacol. Ther. 53 (1992) 275–354. [15] S. Haas, C. Pierl, V. Harth, et al., Expression of xenobiotic and steroid hormone metabolizing enzymes in human breast carcinomas, Int. J. Cancer 119 (2006) 1785–1791. [16] J. Hakkola, M. Pasanen, O. Pelkonen, et al., Expression of CYP1B1 and CYP1A1 in human adult and fetal tissues and differential inducibility of CYP1B1 and CYP1A1 by Ah receptor ligands in human placenta and cultured cells, Carcinogenesis 18 (1997) 391–397. [17] V. Hatorp, K.T. Hansen, M.S. Thomsen, Influence of drugs interacting with CYP3A4 on the pharmacodynamics, and safety of the prandial glucose regulator repaglinide, J. Clin. Pharmacol. 43 (2003) 649–660. [18] J.D. Hayes, J.U. Flanagan, I.R. Jowsey, Glutathione transferases, Annu. Rev. Pharmacol. Toxicol. 45 (2005) 51–88. [19] P.C. Hayes, D.J. Harrison, I.A.D. Bouchier, et al., Cytosolic and microsomal glutathione-S-transferase isoenzymes in normal human liver and intestinal epithelium, Gut 30 (1989) 854–859. [20] P.C. Hayes, L. May, J.D. Hayes, D.J. Harrison, Glutathione S-transferases in human liver cancer, Gut 32 (1991) 1546–1549. [21] K.G. Ishak, H.J. Zimmerman, Hepatotoxic effects of the anabolic/androgenic steroids, Sem. Liver Dis. 7 (1987) 230–236. [22] M. Iwanari, M. Nakajima, R. Kizu, et al., Induction of CYP1A1, CYP1A2, and CYP1B1 mRNA’s by nitropolycyclic aromatic hydrocarbons in various human tissuederived cells: chemical-, cytochrome P450 isoform-, and cell-specific differences, Arch. Toxicol. 76 (2002) 287–298. [23] E. Jeannot, K. Poussin, L. Chiche, et al., Association of CYP 1B1 germ line mutations with hepatocyte nuclear factor 1a-mutated hepatocellular adenoma, Cancer Res. 67 (2007) 2611–2616. [24] C. Justenhoven, B.P. Pierl, S. Haas, et al., The CYP1B1_1358_GG genotype is associated with estrogen receptor-negative breast cancer, Breast Cancer Res. Treat. 111 (2008) 171–177. [25] S.K. Kim, K.J. Woodcroft, F.N. Novak, Insulin and glucagon regulation of glutathione-S-transferase expression in primary cultured rat hepatocytes, J. Pharmacol. Exp. Ther. 311 (2004) 99–108. [26] D.W. Nebert, Role of genetics and drug metabolism in human cancer risk, Mutat. Res. 247 (1991) 267–281. [27] L. Ng, K. Nichols, K. O’Rourke, et al., Repression of human GSTA1 by interleukin-1b is mediated by variant hepatic nuclear growth factor-1C, Mol. Pharmacol. 71 (2007) 201–208. [28] R. Raftogianis, C. Creveling, R. Weinshilboum, et al., Estrogen metabolism by conjugation, J. Natl. Cancer Inst. Monogr. 27 (2000) 113–124. [29] S. Rebouissou, S. Imbeaud, C. Balabaud, et al., HNF1a inactivation promotes lipogenesis in human hepatocellular adenoma independently of SREBP-1 and carbohydrate-response element-binding protein (ChREBP) activation, J. Biol. Chem. 282 (2007) 14437–14446.
ARTICLE IN PRESS S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
[30] C.A. Smith, G. Smith, G.R. Wolf, Genetic polymorphisms in xenobiotic metabolism, Eur. J. Cancer 30A (1994) 1921–1935. [31] K.J. Woodcroft, R.F. Novak, Insulin effects in CYP2E1, 2B, 3A, and 4A expression in primary cultured rat hepatocytes, Chem. Biol. Interact. 107 (1997) 75–91.
725
[32] M.C. Yu, J.M. Yuan, Enviromental factors and risk for hepatocellular carcinoma, Gastroenterology 127 (2004) S72–S78. [33] J. Zucman-Rossi, E. Jeannot, J.T. Van Nhieu, et al., Genotype-phenotype correlation in hepatocellular adenoma: new classification and relationship with HCC, Hepatology 43 (2006) 515–524.
Pathology – Research and Practice 205 (2009) 716–725 www.elsevier.de/prp
ORIGINAL ARTICLE
Expression of xenobiotic and steroid hormone metabolizing enzymes in hepatocellular tumors of the non-cirrhotic liver Susanne Haasa,, Sabine Merkelbach-Brusea, Christina Justenhovenb,c, Hiltrud Brauchb,c, Hans-Peter Fischera a
Institute of Pathology, Medical Faculty of the University of Bonn, Sigmund Freud Str. 25, D-53127 Bonn, Germany Dr. Margarete Fischer-Bosch Institute of Clinical Pharmacology, Auerbachstr. 112, D-70376 Stuttgart, Germany c University Tu¨bingen, Germany b
Received 10 March 2009; received in revised form 11 May 2009; accepted 11 June 2009
Abstract Hepatocellular adenomas (HCA) and some hepatocellular carcinomas (HCC) arise in the non-cirrhotic liver. Although the liver is involved in the metabolism of a huge number of exogenous and endogenous substances, little is known about the role of metabolic enzymes in the development of liver tumors in the absence of cirrhosis. We analyzed the expression of glutathione S-transferases (GST) and cytochrome P450 enzymes (CYP) in 23 HCA, 20 HCC, and 22 focal nodular hyperplasias (FNH) using immunohistochemistry. The liver tissue revealed consistent specific staining for GST alpha, CYP1A1, 1A2, 2E1, and 3A4. In HCA and HCC, GST alpha expression was significantly reduced (po0.001 and 0.043). Reduced GST alpha expression was significantly associated with steatosis in HCA and HCC (n ¼ 12, p ¼ 0.006), but not in non-neoplastic liver tissue. CYP3A4 expression was also reduced in HCA and HCC (p ¼ 0.03 and 0.02), and this was correlated with diabetes mellitus type 2 (p ¼ 0.02). In conclusion, HCA and HCC revealed changes in the expression of certain metabolic enzymes as compared with the non-neoplastic liver tissue or FNH. Therefore, reduced expression of GST alpha and CYP3A4 may indicate specific metabolic defects in the tumor tissue characterizing subgroups of HCA and HCC. r 2009 Elsevier GmbH. All rights reserved. Keywords: HCA; HCC; FNH; GST; CYP
Introduction The liver plays a major role in the metabolism of numerous endogenous and exogenous compounds. Main contributors of this pathway are phase I and phase II enzymes. The phase I system is composed mainly of cytochrome P450 (CYP) enzymes as well as peroxidases, oxidases, reductases, and dehydrogenases, Corresponding author. Tel.: +49 228 287 19244; +49 228 287 15030. E-mail address: [email protected] (S. Haas).
fax:
0344-0338/$ - see front matter r 2009 Elsevier GmbH. All rights reserved. doi:10.1016/j.prp.2009.06.003
which introduce a reactive group to the endogenous or exogenous substrate [18]. Some of the phase I metabolites are conjugated to polar groups by phase II enzymes. Phase II enzymes, e.g. glutathione S-transferases (GSTs), sulfotransferases, and UDP glucuronosyltransferases, catalyze conjugation reactions of endogenous and exogenous substances or of their phase I metabolites, producing usually inactive and water soluble compounds that can be excreted through urine or bile [30]. Regarding the development of liver tumors, it is of special interest that phase I and phase II enzymes are
ARTICLE IN PRESS S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
involved in the metabolism of steroid hormones, e.g. estrogen and testosterone. The vast majority of Hepatocellular adenomas (HCA) arise in women, and development of HCA is often associated with application of contraceptive or anabolic steroids. Anabolic steroids may also play a role in the development of a subset of hepatocellular carcinomas (HCC) in the absence of cirrhosis [12,21]. CYP1A1, CYP1B1, and CYP3A4 catalyze 2-, 4-, and 16-hydroxylation of 17-bestradiol and formation of 4-hydroxy estrone and 2hydroxy estrone [1,9,13]. GSTs are involved in the inactivation of catechol estrogen quinones. They catalyze glutathione conjugation of catechol estrogens quinones, the reactive intermediates of estrogen metabolism capable of binding to DNA [28]. For both GSTs and CYPs, there exist numerous genetic polymorphisms leading to differences in the individual metabolic capacity within a population [30]. In addition, activities of some CYP isoenzymes, such as CYP1A1 and CYP1B1, may be altered due to induction by environmental factors like cigarette smoke or local pollutants [22,26]. Therefore, functional alterations in GST and CYP activity may impact HCA or HCC risk in association with steroid hormone intake. Further risk factors for HCC in the non-cirrhotic liver include obesity and diabetes with non-alcoholic fatty liver disease [10,13,32]. In the recent years, subtypes of HCA were identified by genetic alterations, such as HNF1a and b-catenin mutations, and by histomorphological features [3–5,8]. Germ line mutations of the steroid hormone metabolizing enzyme CYP1B1 were found to be associated with HNF1a mutated HCA [23]. In HCC, a large number of genetic alterations and several carcinogenetic pathways have been described [6]. Based on the recently described diagnostic and molecular features of non-cirrhotic liver tumors, this study intends to systematically analyze expression of several CYP and GST enzymes in order to identify characteristic changes of enzyme expression.
Materials and methods Patients and tumor specimens Tissue sampling and use for this study were performed in accordance with standard guidelines of the ethics committee of the Medical Faculty of the University of Bonn. Paraffin-fixed tissue of 65 patients was included in this study. This collective consisted of 23 HCA, 20 HCC, and 22 focal nodular hyperplasias (FNH). Except for 8 cases of FNH, one HCA and one HCC diagnosed by punch needle biopsy, resection specimens were used for immunohistochemical staining. Sufficient amounts of non-tumorous liver parenchyma
717
were present for histological examination in 19 HCA, 18 HCC, and 14 FNH. The male to female ratio was 3:20 for HCA, 15:5 for HCC, and 5:17 for FNH. Liver cirrhosis was absent in all of the cases. There was no history of chronic hepatitis, alcohol abuse, or hepatitis B or C viral infection. Steatosis of the tumor or liver tissue was assessed if more than 10% of tumor cells or hepatocytes revealed steatotic changes. Among the 20 patients with HCC, one patient was suffering from the Alagille syndrome. Another patient was diagnosed with alpha 1 antitrypsin deficiency type PIZ. One patient had a Budd Chiari syndrome. Grading of HCC was well (G1) in 6 cases, moderate (G2) in 13 cases, and poor (G3) in one case.
Immunohistochemistry and antibodies Tissue specimens were fixed in 4% buffered formaldehyde and embedded in paraffin. Immunohistochemical stainings were performed on 4 mm sections with an immunostainer (Techmate 500; DAKO). The antigen–antibody binding was visualized by the avidin–biotin complex (ABC method) using 3-amino-9-ethylcarbazol (AEC) as chromogen. GST antibodies were purchased from Novo Castra, Newcastle upon Tyne, UK. Antibodies for CYP1A1, CYP3A4 and CYP2E1 were purchased from Natutec, Frankfurt am Main, Germany (Manufacturer Daiichi Pure Chemicals, Tokyo, Japan), and antibodies for CYP1B1 and 1A2 were purchased from Gentest, Woburn, USA. The specificity of the primary antibodies was ensured as described previously [15]. Liver fatty acid binding protein (FABP) antibody was purchased from Abcam, Cambridge, UK; anti-Glutamine Synthetase (GS) from BD Biosciences, NY, USA; monoclonal antibody to Glypican-3 (GPC3) from Biomosaics, Burlington, USA; anti-HSP-70 from Santa Cruz Biotechnology, USA; anti-b-catenin from Transduction Laboratories Lexington, KY, USA. Antibodies for determination of the estrogen receptor (ER) were derived from Novocastra (clone 6F11/2), for the progesterone receptor (PR) from DAKO (clone PgR636), for the androgen receptor (AR) from Antibodies online GmbH (Germany), clone AR441. Details about the dilutions of the primary antibodies are given in Table 1.
Evaluation of immunohistochemical stainings For hormone receptors (estrogen receptor, progesterone receptor, androgen receptor) and b-catenin, any nuclear staining was assessed positive. Cytoplasmic staining of GST and CYP enzymes was scored as follows: the percentage of stained cells (0–100%) was multiplied with the staining intensity (0–3) to give a
ARTICLE IN PRESS 718
S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
Table 1. Dilutions and staining patterns of the primary antibodies.
ER alpha PR AR GST pi GST mu GST alpha CYP1A1/2 CYP3A4/5 CYP2E1 CYP1A2 CYP1B1 FABP Glutamine synthetase Glypican-3 HSP-70 b-catenin a
Dilution
Staininga
1:50 1:50 1:75 1:750 1:400 1:1500 1:200 1:1000 1:200 1:100 1:1000 1:50 1:100 1:100 1:250 1:1000
N N N C C C/N C C C C C C C C/M C/N N
N ¼ nuclear, C ¼ cytoplasmic, M ¼ membrane.
maximum score level of 300. Score levels 440 were defined as positive expression. Tumor tissue and corresponding non-neoplastic liver parenchyma were assessed separately. For statistical testing, w2 analysis and Fisher’s exact test were used, if appropriate.
Sequence analysis for b-catenin mutations Extraction of genomic DNA from the tumor samples was performed using standard procedures. b-catenin exon 3 DNA was amplified by PCR using the following primers: 3-forward: 50 -tttgatggagttggacatgg-30 and 3reverse: 50 -ctgagaaaatccctgttccc-30 . Amplification of target DNA sequences was carried out in 50 ml of a reaction mixture containing an aliquot of extracted DNA, 10 mM TRIS–HCl, 1.5 mM MgCl2, 50 mM KCl, 100 mM of each dNTP, 0.1 mM of forward and reverse primer, respectively, and 1 U of Platinum Taq DNA polymerase (Invitrogen, Karlsruhe, Germany). Each PCR cycle consisted of a denaturation step at 94 1C for 40 s, primer annealing at 60 1C for 40 s, and extension at 72 1C for 35 s. An initial denaturation step was performed at 94 1C for three minutes and in the final cycle; the extension step was elongated to 5 min at 72 1C. Samples were amplified through 40 cycles. The PCR products were purified using MicroSpin columns (Amersham Biosciences, Freiburg, Germany). Template DNA concentrations for cycle sequencing were estimated by agarose gel electrophoresis. Bidirectional DNA sequencing of the entire exon and exon–intron boundaries was performed with the Big Dye Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems, Weiterstadt, Germany) on all
samples using the primers listed above. Cycle sequencing products were precipitated with 3 M sodium acetate and analyzed on a ABI PRISM 3130 capillary electrophoresis system (Applied Biosystems). The identity of the amplicon sequences was confirmed by database search (NCBI database: www.ncbi.nlm.nih.gov).
Results Expression patterns of GST and CYP enzymes in non-neoplastic liver parenchyma The majority of metabolic enzymes displayed reliable immunohistochemical staining and characteristic expression patterns in the non-neoplastic liver tissue (Table 2). A panacinar mosaic-like expression pattern of GST alpha with patchy staining was a constant finding. Expression was both nuclear and cytoplasmic. GST pi turned out to be a selective marker for bile duct epithelium of the portal tracts, while hepatocytes were always negative for GST pi (Fig. 1B). Consequently, no positive staining for GST pi was seen within HCA (Fig. 1C). Most of the CYP enzymes revealed a zonal distribution with intense positive staining in the centrilobular area and only weak staining in the periportal area. This staining pattern was most intense for CYP3A4 (Fig. 1F), to a lesser degree for CYP1A1, CYP1A2, and CYP2E1. Immunostaining of GST mu and CYP1B1 was usually weak or moderate with no zonal distribution pattern in the liver parenchyma, and was thus less significant. Table 2. Distribution of metabolic enzymes in the nonneoplastic liver parenchyma. No of cases with positive expressiona
Predominant staining
GST alpha
51 (100%)
GST mu GST pi
23 (46%) 0 (0%)
CYP1A1
51 (100%)
CYP1A2
48 (96%)
CYP1B1 CYP2E1
27 (54%) 51 (100%)
CYP3A4
51 (100%)
Panacinar, mosaic like Pancinar Bile duct epithelium Centrilobular, panacinar Centrilobular, panacinar Panacinar Centrilobular, panacinar Centrilobular
a
Scoring levels 4 40.
ARTICLE IN PRESS S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
719
Fig. 1. (A–C) Inflammatory HCA with marked steatosis and absent expression of GST alpha (A, right corner). Note the strong GST alpha expression in the steatotic non-neoplastic liver parenchyma (A, left corner). Selective staining of bile duct epithelium of the liver parenchyma by GST pi (B) and absent GST pi staining in portal tract like areas of the HCA (C). (D, E) Steatotic HCA with absent expression of CYP3A4 as compared with the non-neoplastic liver parenchyma (D). Focally positive expression of the progesterone receptor in the tumor tissue (D, inset). Absent staining of FABP in the HCA (E). In this tumor, GST alpha expression was also reduced (not shown). (F–H) Zonal expression of CYP3A4 in the liver parenchyma (left side) and patchy strong expression in the HCC (right side, F). Nuclear expression of b-catenin in the tumor tissue (bottom, G) and strong expression of glutamine synthetase (bottom, H).
Expression of GST and CYP enzymes in FNH, HCA, and HCC Table 3 demonstrates the staining intensities of the enzymes in the tumor tissue and in the non-tumorous liver parenchyma. For the majority of enzymes (GST mu, CYP1B1, 1A1, 1A2, 2E1), expression levels in the tumor tissue corresponded to the non-tumorous liver parenchyma. However, in a significant number of HCA and HCC, expression of GST alpha was absent or reduced (po0.001 and p ¼ 0.043). CYP3A4 expression
was also significantly decreased in HCA (p ¼ 0.03) and HCC (p ¼ 0.02). Reduction of GST alpha and/or CYP3A4 in HCC was independent of the tumor grading. Significant alterations in enzyme expression were not seen in FNH. In addition, the topical distribution of the CYP enzymes in FNH corresponded to the nontumorous liver tissue: expression was reduced in paraseptal areas and was stronger in the center of the portal tract equivalent. A distinct zonal expression of CYPs was also seen in most HCA. Especially CYP3A4 and CYP1A1 were expressed in the areas surrounding
ARTICLE IN PRESS 720
Table 3.
S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
Expression of metabolic enzymes in liver tumors and adjacent liver parenchyma (LP). HCA
HCA-LP
HCC
HCC-LP
FNH
FNH-LP
23
20
14
8 12
18 p ¼ 0.043 2 16
22
10 13
19 po0.001 0 19
6 16
1 13
GST mu Negative Positive
18 5
15 4
12 8
8 10
6 16
4 10
CYP1A1 Negative Positive
1 22
0 19
1 19
0 18
1 21
0 14
CYP1A2 Negative Positive
4 19
2 17
2 18
0 18
3 19
1 13
CYP1B1 Negative Positive
10 13
12 7
11 9
8 10
7 15
7 7
CYP2E1 Negative Positive
0 23
0 19
1 19
0 18
0 22
0 14
CYP3A4 Negative Positive
5 18
p ¼ 0.03 0 19
5 15
p ¼ 0.022 0 18
0 22
0 14
N GST alpha Negative Positive
Table 4.
Expression of diagnostic markers in liver tumors and adjacent liver parenchyma (LP). HCA
HCA-LP
HCC
HCC-LP
FNH
FNH-LP
23
20
18
22
14
9 11
5 13
9 13
4 10
N FABP Negative Positive
13 10
19 p ¼ 0.048 5 14
GS Negative Positive
21 2
18 1
11 9
p ¼ 0.012 17 1
15 7
12 2
GPC 3 Negative Positive
20 3
19 0
12 8
p ¼ 0.015 17 1
22 0
14 0
HSP70 Negative Positive
23 0
19 0
5 15
p ¼ 0.003 13 5
22 0
14 0
b-catenin Negative Positive
21 2
19 0
15 5
18 0
22 0
14 0
HRa Negative Positive
21 2
19 0
17 3
18 0
22 0
14 0
a
Hormone receptor positive: expression of ER and/or PR and/or AR.
ARTICLE IN PRESS S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
the efferent vein, corresponding to the center of the lobule in non-neoplastic liver parenchyma. No distinct pattern of enzyme expression was observed in HCC, usually exhibiting diffuse or patchy staining (Fig. 1F).
Expression of diagnostic markers and subtyping of HCA, HCC, and FNH Subtyping of the 23 HCA was performed based on histomorphological and immunohistochemical features. Histomorphology of 6 HCA corresponded to the type of inflammatory HCA (formerly ‘‘teleangiectatic FNH’’type) with inflammatory infiltrates, sinusoidal dilatation, and ductular reaction. The teleangiectatic FNH type HCA did not differ from other HCA regarding the expression of GST and CYP enzymes. One case revealed tumor steatosis (Fig. 1A–C). Expression of GS, GPC3, and HSP70 was significantly associated with HCC (p ¼ 0.012 for GS, 0.015 for GPC3, 0.003 for HSP70) (Table 4). Eighteen of 20 HCC expressed at least one of these markers. In the liver parenchyma, focal weak positive staining of GPC 3 and HSP 70 could be seen. GS expression was restricted to a few hepatocytic layers around the central vein. Table 5.
721
Nuclear expression of b-catenin in association with strong expression of GS was seen in five cases (4 HCC, 1 male HCA) (Fig. 1G, H). In three of these tumors, mutation or deletion of the b-catenin gene could be verified by genomic sequence analysis (p.S33P, p.S45P, p.Q28_H36del). In the remaining two cases, molecular analyses were not feasible. Four other HCC cases revealed faint indeterminate nuclear b-catenin staining and only weak positivity for GS. They were tested negative for b-catenin mutations.
HCC and HCA with special clinical, diagnostic, or histomorphologic features
Steatosis of the tumor was present in 1 FNH, 10 HCA (including one case of inflammatory HCA), and 2 HCC. Steatosis of non-tumorous liver parenchyma was observed in 17 cases, independent of the steatotic grade of the tumor tissue. Expression of GST alpha in steatotic HCC and HCA was significantly reduced compared to the surrounding liver tissue (p ¼ 0.006, Table 5). The adjacent liver parenchyma was always positive for GST alpha regardless of steatosis (Fig. 1A). Two of the steatotic HCA revealed positive expression of steroid hormone receptors (inset Fig.
Expression of metabolic enzymes and hormone receptors in subgroups of HCA and HCC. n
FABP negative
b-catenin positive
GST alpha negative
CYP3A4 negative
43
5
18
1 4
–
10 p ¼ 0.039 7 3
n(HCA+HCC) GST alpha Negative Positive
18 25
22 p ¼ 0.008 12 10
CYP3A4 Negative Positive
10 33
5 17
0 5
–
–
CYP1B1 Negative Positive
21 22
11 11
3 2
8 10
p ¼ 0.0244 8 2
GS Negative Positive
32 11
17 5
po0.001 0 5
16 2
HRa Negative Positive
38 5
18 4
4 1
15 3
Steatotic tumor Yes No
12 31
7 15
1 4
p ¼ 0.006 9 9
Diabetes Yes No
6 37
3 19
0 5
p ¼ 0.067 5 13
a
Hormone receptor positive: expression of ER and/or PR and/or AR.
9 1 p ¼ 0.073 7 3 1 9 p ¼ 0.02 4 6
ARTICLE IN PRESS 722
S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
Fig. 2. A 16-year-old male patient with Fanconi’s anemia developed multiple hepatic nodules up to 10 cm in diameter under androgen therapy (A). Resection specimen revealed one HCC with steatosis (B) and absence of GST alpha staining (not shown). There were also found HCA0 s without steatosis but positive expression of the androgen receptor (C, D). Analysis for b-catenin mutation was not feasible in this case.
1D). Lipofuscinosis of the tumors was seen in two further cases of steatotic HCA. FABP expression was absent or reduced in a significant number of HCA (n ¼ 13, p ¼ 0.048), as well as in 9 HCC and 9 FNH (Table 5). In HCA and HCC, absent FABP expression was significantly correlated with reduction of GST alpha expression (p ¼ 0.008, Fig. 1E). Although complete absence of FABP staining (compared to the positive staining of corresponding non-neoplastic liver tissue) was frequent in steatotic tumors, especially in HCA, there was no significant correlation with tumor steatosis. Diabetes mellitus type 2 was documented in 5 HCC, 1 HCA, and 2 FNH patients. In 4 male HCC patients with type 2 diabetes, CYP3A4 expression was reduced or absent in the tumors compared to the liver tissue (p ¼ 0.02). In 4 HCC and one HCA of diabetic patients, GST alpha expression was also reduced (p ¼ 0.07, Fisher0 s exact two-tailed p-value). Diabetes mellitus was not related to steatosis of the tumor tissue. Steroid hormone receptor expression was rare, with three HCC and two HCA expressing ER, PR, or AR. A remarkably strong expression of the estrogen receptor alpha, progesterone and androgen receptor was detected in one case of HCC. An additional special feature of this HepPar1-positive tumor was negativity
for GS, HSP 70, and Glypican C. This male patient had no history of exogenous hormone application. Two further patients with AR-positive HCC’s had undergone therapeutic androgen application. One case is demonstrated in Fig. 2. There was a trend towards reduced expression of CYP3A4 in hormone receptorpositive tumors (p ¼ 0.073, Fig. 1D). Expression of steroid hormone receptors was not related to specific diagnostic features in HCA or HCC (i.e., b-catenin expression, tumor steatosis, lack of FABP, inflammatory HCA). The liver parenchyma of all cases did not show any nuclear hormone receptor expression.
Discussion In this immunohistochemical study, we were able to demonstrate specific expression of several GST and CYP enzymes in hepatocellular tumors and in nonneoplastic liver tissue. The general strong cytoplasmic and nuclear staining of GST alpha in hepatocytes, as well as the restriction of GST pi expression to the bile duct epithelium, has been described before [19,20]. Inconsistent expression of GST mu may be explained
ARTICLE IN PRESS S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
by the high rate of the GSTM1 null polymorphism within the population [30]. Most of the CYP enzymes were expressed in zones 2 and 3 in accordance with the increased metabolic activity in the center of the liver lobule [14]. Weak or moderate panacinar staining was seen for CYP1B1, which is predominantly an extrahepatic enzyme [16]. In general, there were no discernible differences in enzyme expression regarding gender or age of the patients, nor was there an association with hepatic steatosis, diabetes mellitus or application of anabolic or contraceptive steroids. In FNH, distribution pattern and staining intensity of GST and CYP enzymes resembled the normal liver tissue. Zonality was imitated by weaker expression in the periseptal parenchyma and stronger expression in the expansive margins of the micronoduli. These findings are in line with the fact that FNH develop as benign hyperregenerative nodules of the liver tissue. In HCA and HCC, GST mu, CYP1A1, 1A2, and 2E1 expression corresponded to the non-neoplastic liver tissue. A faint zonality of CYP expression was also observed in HCA, but not in HCC. However, noticeable alterations in expression were detected for GST alpha and/or CYP3A4. Reduced expression of one or both enzymes in HCA and HCC could be assigned to distinct histological or clinical features. The first group of tumors with altered enzyme expression comprised steatotic HCA and HCC. In the majority of these tumors, there was a significant reduction in GST alpha expression as compared with the non-neoplastic liver tissue. Recently, steatosis and absence of FABP staining in HCA were found to be significantly associated with HNF1a mutations [3,4,33]. Steatosis in HNF1a-mutated HCA is supposed to be caused by increased lipogenesis activity [29]. There also was an association of lack of FABP staining in steatotic HCA in our study group, but we could not confirm a general, significant correlation of tumor steatosis (including inflammatory HCA and HCC) with absence of FABP. It is of note that FABP staining was significantly reduced in the GST alpha-reduced steatotic HCA and HCC. Therefore, it has to be discussed whether reduced GST alpha expression is a feature of tumor steatosis in general or whether it is an indicator for HNF1a mutation. The latter assumption may be supported by the finding that GST A transcriptional activity is regulated by HNF-1 [27]. A second small group of HCA and HCC was characterized by the expression of steroid hormone receptors. In two HCC, previous application of androgen was documented. One young patient developed multiple HCA and one HCC following androgen therapy for Fanconi0 s anemia. This patient remained free of disease for more than 6 years after cessation of hormone application. This indicates that ER, PR, or AR
723
expression in the tumor tissue may be induced by exogeneous hormone intake, especially by androgen application in men. However, steroid hormone receptor mediated signaling seems to promote tumor growth in a minority of HCC and HCA only. In part of the steroid hormone receptor-positive tumors, expression of the steroid hormone metabolizing enzyme CYP3A4 was reduced. Although this correlation did not reach significance due to the limited number of cases, there generally might be an inverse correlation between the presence of nuclear steroid hormone receptor expression on the one hand and the expression of steroid hormone metabolizing enzymes on the other hand. In breast cancer, we could demonstrate absence of ER/PR expression to be associated with strong overexpression of CYP1B1, which is one of the most important steroid hormone metabolizing enzymes in extrahepatic tissues [15]. Distinct polymorphisms of the CYP1B1 genotype (CYP1B1_1358_GG), leading to increased activity of CYP1B1, were found to be associated with ERa negativity in breast cancers [24]. The third group of liver tumors with altered enzyme expression is represented by HCA and HCC of patients with type 2 diabetes mellitus. Obesity and diabetes mellitus have been linked to HCC in several epidemiological studies [7]. In diabetic rats, combined hyperinsulinism and hyperglycemia leading to increased intracellular insulin signaling and growth stimulation was identified as a carcinogenic mechanism for the development of HCC [11]. Activity of CYP enzymes in diabetes may be altered by insulin itself [31] by several factors associated with diabetes such as elevated plasma ketone levels and altered growth hormone expression [2], or due to metabolization of anti-diabetic drugs [17]. In vitro experiments on rat hepatocytes showed that GST alpha expression is regulated by insulin and glucagon in an opposite manner [25]. Interestingly, diabetes of the patients was associated with mild steatosis of the liver parenchyma, but, except for one HCA case, no steatosis of the tumor tissue was seen. We could not demonstrate altered expression levels of CYP and GST enzymes in the non-neoplastic liver tissue of diabetic patients, indicating insulin effects. However, all diabetic tumors revealed reduced expression of either CYP3A4 or GST alpha or of both enzymes, and therefore, CYP3A4 and/or GST alpha expression are obviously down-regulated in HCC of diabetic patients without other risk factors. Finally, in spite of their different biologic behavior, HCA and HCC reveal similar alterations in enzyme expression, indicating that there are equal metabolic changes in these tumors independent of their dignity. Therefore, GST and CYP enzymes are no immunohistochemical markers discriminating benign from malignant liver neoplasms.
ARTICLE IN PRESS 724
S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
However, altered expression of GST alpha and CYP3A4 was associated with distinct tumor features of which tumor steatosis is the most prominent. Further studies will show whether reduced GST alpha expression in association with tumor steatosis is a result of HNF1a mutation, and whether there is a true relationship between enzyme expression and expression of steroid hormone receptors or diabetes mellitus in small HCA/ HCC subgroups.
References [1] A.F. Badawi, E.L. Cavalieri, E.G. Rogan, Role of human cytochrome P450 1A1, 1A2, 1B1 and 3A4 in the 2-, 4-, and 16alpha-hydroxylation of 17 beta estradiol, Metabolism 50 (2001) 1001–1003. [2] G.D. Bellward, T. Chang, B. Rodrigues, et al., Hepatic cytochrome P-450 induction in the spontaneously diabetic BB rat, Mol. Pharmacol. 33 (1988) 140–143. [3] P. Bioulac-Sage, S. Rebouissou, A. Sa Cunha, et al., Clinic, morphologic, and molecular features defining socalled teleangiectatic focal nodular hyperplasias of the liver, Gastroenterology 128 (2005) 1211–1218. [4] P. Bioulac-Sage, S. Rebouissou, C. Thomas, et al., Hepatocellular adenoma subtype classification using molecular markers and immunohistochemistry, Hepatology 46 (2007) 740–748. [5] O. Bluteau, E. Jeannot, P. Bioulac-Sage, et al., Bi-allelic inactivation of TCF-1 in hepatic adenomas, Nat. Genet. 32 (2002) 312–315. [6] S. Boyault, D.S. Rickman, A. de Reynies, et al., Transcriptome classification of HCC is related to gene alterations and to new therapeutic targets, Hepatology 45 (2007) 42–52. [7] S.H. Caldwell, D.M. Crespo, et al., Obesity and hepatocellular carcinoma, Gastroenterology 127 (2004) S97–S103. [8] Y.W. Chen, Y.M. Jeng, S.H. Heh, et al., P53 gene and wnt signalling in benign neoplasms: beta catenin mutations in hepatic adenoma but not in focal nodular hyperplasia, Hepatology 36 (2002) 927–935. [9] A.E. Cribb, M.J. Knight, D. Dryer, et al., Role of polymorphic human cytochrome P450 enzymes in estrone oxidation, Cancer Epidemiol. Biomarkers Prev. 15 (2006) 551–558. [10] J.A. Davila, R.O. Morgan, K.A. McGlynn, et al., Diabetes increases the risk of hepatocellular carinoma in the United States: a population based case control study, Gut 54 (2005) 533–539. [11] F. Dombrowski, C. Mathieu, M. Evert, Hepatocellular neoplasms induced by low-number pancreatic islet transplants in autoimmune diabetic BB/Pfd rats, Cancer Res. 66 (2006) 1833–1843. [12] H.A. Edmondson, B. Henderson, B. Benton, Liver-cell adenomas associated with use of oral contraceptives, N. Engl. J. Med. 294 (1976) 470–472. [13] H.B. El-Serag, T. Tran, J.E. Everhart, Diabetes increases the risk of chronic liver disease and hepatocellular carcinoma, Gastroenterology 126 (2004) 460–468.
[14] R. Gebhardt, Metabolic zonation of the liver: regulations and implication for liver function, Pharmacol. Ther. 53 (1992) 275–354. [15] S. Haas, C. Pierl, V. Harth, et al., Expression of xenobiotic and steroid hormone metabolizing enzymes in human breast carcinomas, Int. J. Cancer 119 (2006) 1785–1791. [16] J. Hakkola, M. Pasanen, O. Pelkonen, et al., Expression of CYP1B1 and CYP1A1 in human adult and fetal tissues and differential inducibility of CYP1B1 and CYP1A1 by Ah receptor ligands in human placenta and cultured cells, Carcinogenesis 18 (1997) 391–397. [17] V. Hatorp, K.T. Hansen, M.S. Thomsen, Influence of drugs interacting with CYP3A4 on the pharmacodynamics, and safety of the prandial glucose regulator repaglinide, J. Clin. Pharmacol. 43 (2003) 649–660. [18] J.D. Hayes, J.U. Flanagan, I.R. Jowsey, Glutathione transferases, Annu. Rev. Pharmacol. Toxicol. 45 (2005) 51–88. [19] P.C. Hayes, D.J. Harrison, I.A.D. Bouchier, et al., Cytosolic and microsomal glutathione-S-transferase isoenzymes in normal human liver and intestinal epithelium, Gut 30 (1989) 854–859. [20] P.C. Hayes, L. May, J.D. Hayes, D.J. Harrison, Glutathione S-transferases in human liver cancer, Gut 32 (1991) 1546–1549. [21] K.G. Ishak, H.J. Zimmerman, Hepatotoxic effects of the anabolic/androgenic steroids, Sem. Liver Dis. 7 (1987) 230–236. [22] M. Iwanari, M. Nakajima, R. Kizu, et al., Induction of CYP1A1, CYP1A2, and CYP1B1 mRNA’s by nitropolycyclic aromatic hydrocarbons in various human tissuederived cells: chemical-, cytochrome P450 isoform-, and cell-specific differences, Arch. Toxicol. 76 (2002) 287–298. [23] E. Jeannot, K. Poussin, L. Chiche, et al., Association of CYP 1B1 germ line mutations with hepatocyte nuclear factor 1a-mutated hepatocellular adenoma, Cancer Res. 67 (2007) 2611–2616. [24] C. Justenhoven, B.P. Pierl, S. Haas, et al., The CYP1B1_1358_GG genotype is associated with estrogen receptor-negative breast cancer, Breast Cancer Res. Treat. 111 (2008) 171–177. [25] S.K. Kim, K.J. Woodcroft, F.N. Novak, Insulin and glucagon regulation of glutathione-S-transferase expression in primary cultured rat hepatocytes, J. Pharmacol. Exp. Ther. 311 (2004) 99–108. [26] D.W. Nebert, Role of genetics and drug metabolism in human cancer risk, Mutat. Res. 247 (1991) 267–281. [27] L. Ng, K. Nichols, K. O’Rourke, et al., Repression of human GSTA1 by interleukin-1b is mediated by variant hepatic nuclear growth factor-1C, Mol. Pharmacol. 71 (2007) 201–208. [28] R. Raftogianis, C. Creveling, R. Weinshilboum, et al., Estrogen metabolism by conjugation, J. Natl. Cancer Inst. Monogr. 27 (2000) 113–124. [29] S. Rebouissou, S. Imbeaud, C. Balabaud, et al., HNF1a inactivation promotes lipogenesis in human hepatocellular adenoma independently of SREBP-1 and carbohydrate-response element-binding protein (ChREBP) activation, J. Biol. Chem. 282 (2007) 14437–14446.
ARTICLE IN PRESS S. Haas et al. / Pathology – Research and Practice 205 (2009) 716–725
[30] C.A. Smith, G. Smith, G.R. Wolf, Genetic polymorphisms in xenobiotic metabolism, Eur. J. Cancer 30A (1994) 1921–1935. [31] K.J. Woodcroft, R.F. Novak, Insulin effects in CYP2E1, 2B, 3A, and 4A expression in primary cultured rat hepatocytes, Chem. Biol. Interact. 107 (1997) 75–91.
725
[32] M.C. Yu, J.M. Yuan, Enviromental factors and risk for hepatocellular carcinoma, Gastroenterology 127 (2004) S72–S78. [33] J. Zucman-Rossi, E. Jeannot, J.T. Van Nhieu, et al., Genotype-phenotype correlation in hepatocellular adenoma: new classification and relationship with HCC, Hepatology 43 (2006) 515–524.