Clinicopathologic and prognostic values of apolipoprotein D alterations in hepatocellular carcinoma



We previously identified apolipoprotein D (Apo D) as a novel tumor suppressor gene based on the pharmacological unmasking of epigenetic silencing. We analyzed Apo D expression using real-time reverse transcription-PCR and evaluated its expression status based on the clinicopathological parameters of 70 patients with hepatocellular carcinoma (HCC). Immunohistochemical staining was also performed. We determined the methylation status of Apo D gene promoter by methylation-specific PCR (MSP). The Apo D gene-expression in tumor tissue was significantly lower than that in nontumor tissue (p = 0.011). A low Apo D expression significantly correlated with less-differentiated HCC (p = 0.019). Immunohistochemical studies confirmed a decreased Apo D expression in poorly differentiated tumors. The prognosis of patients with a lower Apo D gene-expression was significantly worse than that in those with a higher expression (p = 0.028). The Apo D gene-expression was an independent prognostic factor (relative risk: 2.36, p = 0.018). An MSP assay showed a low-level of methylation in well differentiated HCC and a high-level of methylation in less differentiated tumors. Apo D may be a novel tumor suppressor gene of HCC, and its expression status may be a useful biomarker for predicting the patient outcome. © 2005 Wiley-Liss, Inc.

We recently performed a comprehensive survey of commonly inactivated tumor suppressor genes in esophageal cancer cell lines based on functional reactivation by a demethylating agent.1 A total of 58 silenced genes were thus identified using this approach, and one of those genes was Apolipoprotein D (Apo D). In our report, we detected the promoter hypermethylation of Apo D gene using bisulfite DNA sequencing. The methylation status of Apo D correlated closely with the Apo D expression status. Moreover, the growth suppressive activity of Apo D was also demonstrated by a colony focus assay.1 These findings thus prompted us to analyze the clinical significance of the Apo D expression status in human cancers.


Apo D, apolipoprotein D; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; HCC, hepatocellular carcinoma; MSP, methylation-specific PCR; RT-PCR, reverse transcription-polymerase chain reaction.

Apo D, belonging to the lipocalin superfamily, is a glycoprotein of about 30 kDa found in the high-density lipoprotein fraction of human plasma.2 The functional role of Apo D remains unclear, but the observation that this protein forms complexes with lecithin-cholesterol acyltransferase suggests that Apo D may be involved in cholesterol esterification.3 More recently, the Apo D gene-expression was reported to be induced by retinoic acid in breast cancer cell lines. The induction of the Apo D expression was accompanied by an inhibition of cell proliferation and a progression through a more differentiated phenotype. These finding suggest that the mechanisms controlling retinoic acid-induced growth arrest, cell differentiation and Apo D synthesis may be directly coordinated in human breast cancer cells.4 Furthermore, a low expression of Apo D determined by an immunohistochemical study was significantly associated with a poorer prognosis in patients with breast cancer.5 In ovarian carcinoma, patients with Apo D-negative tumors had a poorer overall survival than those with Apo D-positive tumors.6 However, the clinicopathological and prognostic significance of the Apo D gene-expression status has never been previously investigated in digestive organ cancers.

Our study thus focused on identifying whether the expression of Apo D mRNA is involved in the clinicopathological features of human hepatocellular carcinoma (HCC) using real-time quantitative reverse transcription (RT)-PCR. The methylation status of Apo D promoter region by methylation-specific PCR (MSP) was also examined. Interestingly, our findings suggested that the methylation-mediated silencing of Apo D expression in tumor tissue showed a strong association with both a histologically low tumor grade and a poor prognosis for patients with HCC.

Material and methods

Tissue samples and cell lines

Seventy patients with hepatocellular carcinoma who underwent surgery at our institutes were entered in our study. The resected tumor and paired nontumor tissue specimens were immediately frozen in liquid nitrogen and kept at −80°C until analysis. Written informed consent was obtained from all patients. All patients were closely followed after surgery at regular 1-month intervals. The human liver cancer cell lines, Hep3B, HepG2 and HuH7, were provided by the Cell Resource Center for Biomedical Research, Institute of Development, Aging and Cancer, Tohoku University, Japan. All cell lines were maintained in an RPMI 1640 medium supplemented with 10% of fetal calf serum and antibiotics.

5-Azacytidine treatment of cells

Tumor cell lines were plated at a concentration of 1 × 106 cells /100 mm dish and treated with 2.5–5 μM 5-Azacytidine (Sigma Chemical Co., St. Louis, MO). The medium was changed 72 hr after treatment in order to maintain the concentration of 5-Azacytidine at a constant level. The total RNA was extracted 96 hr after treatment.

Total RNA extraction

Frozen tissue specimens were homogenized in guanidinium thiocyanate and total RNAs were obtained by ultracentrifugation through a cesium chloride cushion as described previously.7, 8 Cultured cells from each cancer cell line were dissolved in 350 μl of buffer RLT (Qiagen, Hilden, Germany) containing 1% b-Mercaptoethanol, and the total RNAs were extracted and purified with an RNeasy Tissue Kit (Qiagen) according to the manufacturer's protocols.

Real-time quantitative RT-PCR

The cDNA was synthesized from 8.0 μg of total RNA as described previously.9 The real-time monitoring of PCR reactions was performed using the LightCycler™ system (Roche Applied Science, Indianapolis, IN) and SYBR green I dye (Roche Diagnostics). Monitoring was performed according to the manufacturer's instructions, as described previously.10, 11 In brief, a master mixture was prepared on ice, containing 1 μl of cDNA of each gene, 2 μl of LC DNA Master SYBR Green I mix, 50 ng of primers and 2.4 μl of 25 mM MgCl2. The final volume was then adjusted to 20 μl with water. After the reaction mixture was loaded into the glass capillary tube, PCR was carried out under cycling conditions as listed in Table I. After amplification, the products were subjected to a temperature gradient from 68°C to 95°C at 0.2°C/sec, under continuous fluorescence monitoring to produce a melting curve of the products.

Table I. Real-Time RT-PCR Amplification Programs for Apolipoprotein D (Apo D) and Glyceraldehyde 3-Phosphate Dehydrogenase (GAPDH)
Gene namePrimers from 5′ to 3′ f; forward primer, r; reverse primerTemperature and duration denaturation annealing extensionCycle number
  72°C for 10 sec 
  72°C for 10 sec 

We determined the expression levels of Apo D and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) by comparisons with cDNA from Human Universal Reference total RNA (Clontech, Palo Alto, CA). The fit point method was employed to determine the cycle in which the log-linear signal was first distinguishable from the baseline, and then that cycle number was used as a crossing-point value. The standard curve was produced by measuring the crossing point of each standard value and plotting them against the logarithmic value of concentrations. The concentrations were calculated by plotting their crossing points against the standard curve and were divided by the GAPDH content. Each assay was performed 3 times to verify the results, and the mean mRNA expression was used for the analysis.


Immunohistochemical studies of Apo D were performed on surgical samples from HCC patients using the avidin-biotin-peroxydase method (LSAB2 kit; Dako, Kyoto, Japan) on formalin-fixed, paraffin-embedded tissue specimens. All sections were counterstained with hematoxylin. The primary antibodies against Apo D (Research Diagnostics, Inc., Flanders, NJ) were used at dilutions of 1:500, as described previously.5

Methylation-specific PCR (MSP)

We extracted genomic DNA from Trizol (Invitrogen)-treated tissue specimens and carried out bisulfite modification of genomic DNA.12 Bisulfite-treated DNA was amplified with either a methylation-specific or unmethylation-specific primer set for Apo D at 33 cycles: 96°C for 30 sec, 59°C (methylated) and 55°C (unmethylated) for 30 sec and 72°C for 30 sec as described previously.1 The methylation-specific primer sequences for Apo D were designed using 5′-CACACGCGCGAAAACAATAT-3′ as the forward primer and 5′-TATGTATGTTACGTTCGTCG-3′ as the reverse primer. The unmethylated-specific primer sequences were 5′-CACACAAAAACAATATCTCATTTCT-3′ and 5′-TTTTTTATGTATGTTATGTTTGTTG-3′.

Statistical methods

For continuous variables, the data were expressed as the means + standard deviation. The relationship between the Apo D mRNA expression and the clinicopathological factors were analyzed using the chi-square test and Student's t-test. The surviving curves were plotted according to the Kaplan-Meier method, and the generalized Wilcoxon test was applied to compare the survival curve. A multivariate adjustment was also made using the Cox's proportional hazard model.13 All tests were analyzed using the StatView software package (Abacus Concepts, Inc., Berkeley, CA) and the findings were considered to be significant when the p-value was less than 0.05.


Induction of Apo D by 5-Azacitidine in cell lines

Figure 1 shows the Apo D gene-expression status in the human liver cancer cell lines before and after treatment with a demethylating agent, 5-Azacitidine. HuH7 originally expressed the Apo D gene, while Hep3B or HepG2 did not. Following 96 hr of incubation with 5-Azacitidine, the gene expressions of Apo D were induced in both the methylation-silenced cell lines (Hep3B and HepG2).

Figure 1.

Induction of Apolipoprotein D (Apo D) expression by 5-Azacytidine. In each cell line, the left (−) and right (+) lanes show the status before and after treatment with 5-Azacytidine, respectively. GAPDH: glyceraldehyde-3-phosphate dehydrogenase; M: marker.

Expression value of Apo D mRNA in clinical tissue specimens

We determined the levels of Apo D mRNA expression by comparisons with a human universal reference as the quantifying standard, which express human Apo D sufficiently. The mean expression level of Apo D mRNA in tumor tissue, 0.096±0.021, was significantly lower than the level, 0.229±0.048, in the corresponding nontumor tissue (p = 0.011). The patients with values of less than the median expression level (0.038) in tumor tissue were considered to belong to the T-low expression group (n = 35), while those with values of equal to or more than 0.038 were considered to belong to the T-high expression group (n = 35). The patients with values of less than the median expression level (0.130) in nontumor tissue were considered to belong to the N-low expression group (n = 35), while those with values of equal to or more than 0.130 were considered to belong to the N-high expression group (n = 35). The clinical features of Apo D expression in patients with HCC were evaluated by comparisons among these groups.

Apo D mRNA expression and clinicopathological characteristics

The clinicopathological factors analyzed are shown in Table II in relation to the Apo D mRNA expression in nontumor tissue. No significant differences in the clinicopathological data, such as the results of liver function tests and the percentages of patients with histological liver cirrhosis, were observed between the 2 groups. Table III summarizes the clinicopathological features determined according to the criteria proposed by the Liver Cancer Study Group of Japan14 in relation to the Apo D mRNA expression in tumor tissue. The incidence of poorly differentiated tumors in the T-low expression group (13/35, 37%) was significantly higher (p = 0.019) than that in the T-high expression group (4/35, 11%). In contrast, other pathological variables, such as the tumor diameter, invasion to the portal vein and intrahepatic metastasis, were not associated with the Apo D expression status in tumor tissue specimens.

Table II. The Clinicopathological Data and Apolipoprotein D mRNA Expression i the Nontumor Tissue Specimens From 70 Patients With Hepatocellular Carcinoma
Variables1N-low expression2 (n = 35)N-high expression2 (n = 35)p value
  • 1

    HBsAg, hepatitis B surface antigen; HCV, hepatitis C virus, AST, aspartate transaminase; ICG R15, indocyanine green dye retention test at 15min.

  • 2

    The low and high expression groups were determined by a median value of Apo D mRNA in the 70 nontumor tissue specimens.

Gender (male:female)23:1222:130.80
Age64.1 ± 7.861.3 ± 10.40.196
Positive HbsAg (%)5/35 (14)10/35 (29)0.123
Positive HCV antibody (%)27/35 (77)21/35 (60)0.121
AST (international units/L)59.0 ± 25.052.0 ± 31.20.304
Albumin (g/dL3)3.9 ± 0.43.9 ± 0.40.958
Total bilirubin (mg/dL)0.90 ± 0.500.82 ± 0.310.387
ICG R15 (%)17.0 ± 9.915.3 ± 9.60.486
Child's classification (A/B/C)28/5/228/7/ 00.311
Histologic liver cirrhosis (%)14/35 (40)14/35 (40)0.999
Table III. The Clinicopathlogical Data and Apolipoprotein D mRNA Expression in the Tumor Tissue Specimens from 70 Patients with Hepatocellular Carcinoma
Variables1T-low expression2 (n = 35)T-high expression2 (n = 35)p value
  • 1

    fc, capsular formation; fc-inf; tumor,invasion to fc; vv, invasion to hepatic vein; vp, invasion to portal vein; im, intrahepatic metastasis; AFP, alpha-fetoprotein.

  • 2

    The low and high expression groups were determined by a median value of Apo D mRNA in the 70 tumor tissue specimens.

  • 3

    Histologic differentiation of the tumor.

Gender (male: female)23:1222:130.803
Age (y)62.1 ± 9.463.3 ± 9.00.572
Tumor diameter (cm)4.4 ± 2.64.3 ± 3.20.924
Positive fc (+), (%)27/35 (77)26/35 (74)0.780
Positive fc-inf (+), (%)22/35 (63)22/35 (63)0.999
Positive vv (+), (%)3/35 (9)2/35 (6)0.642
Positive vp (+), (%)18/35 (51)22/35 (63)0.333
Positive im (+), (%)12/35 (34)13/35 (37)0.803
Well /Moderately /Poorly35/17/133/28/40.019
AFP > 20 ng/mL, (%)19/35 (54)19/35 (54)0.999

We also compared the cumulative survival rate of the T-low expression (n = 35) and T-high expression (n = 35) groups as shown in Figure 2. The T-low expression group showed a significantly poorer prognosis (p = 0.028) than the T-high expression group (4-year survival rate; 49.2% and 66.4%, respectively). Furthermore, a multivariate analysis using Cox's proportional hazard model demonstrated the Apo D expression in the tumor tissue to be an independent prognostic indicator (coefficient: 1.71, relative risk: 2.36, p = 0.018).

Figure 2.

Comparison of the Kaplan-Meier survival curves in the T-high expression group (n = 35) vs. the T-low expression group (n = 35). The survival rate for patients in the T-low expression group was significantly lower than that for patients in the T-high expression group (p = 0.028).

Immunohistochemical staining

Figure 3 shows representative findings of immunohistochemical experiments in the resected HCC tissue specimens. The staining intensity of Apo D in well differentiated HCC was almost identical to that in the corresponding nontumor hepatic tissue. On the other hand, in moderately differentiated and poorly differentiated HCC, Apo D staining was markedly weaker in tumor tissue than in nontumor tissue. Especially, the staining intensity in poorly differentiated HCC was almost negative. Consequently, the results of immunohistochemical staining closely corresponded with those of quantitative real-time RT-PCR.

Figure 3.

An immunohistochemical analysis for Apolipoprotein D (Apo D) in well, moderately and poorly differentiated hepatocellular carcinoma (HCC). Apo D immunostaining was weakly positive for moderately differentiated HCC and almost negative for poorly differentiated HCC. N: nontumor tissue (lower left, high power field; × 100). T: tumor tissue (lower right, high power field; × 100).


Since the Apo D expression status in the tumor tissue specimens detected by both real-time RT-PCR and an immunohistochemical study was associated with the tumor histological grade, we performed a MSP assay to confirm the promoter methylation status in HCC tumor tissue specimens (Fig. 4). The MSP analysis revealed a low-level of methylation in the well differentiated HCC and a high-level of methylation in the poorly differentiated HCC. These results corresponded very closely to those of real-time RT-PCR and an immunohistochemical study demonstrating reduced Apo D expressions in moderately and poorly differentiated HCC.

Figure 4.

A methylation-specific PCR analysis of Apolipoprotein D (Apo D) in tumor tissue specimens from patients with hepatocellular carcinoma (HCC). The promoter hypermethylation of the Apo D gene was obvious in the less-differentiated HCC while only low-level methylation was observed in well differentiated HCC. M: marker; N: negative control (distilled water); P: positive control (KYSE30 for methylated and TE5 for unmethylated PCR products); w: well differentiated HCC; m: moderately differentiated HCC; p: poorly differentiated HCC.


Epigenetic alterations, as well as genetic alterations, are involved in cancer development and progression.15, 16 The methylation of CpG islands in the 5' regions of tumor suppressor genes is known to inhibit transcriptional initiation and cause a permanent silencing of these genes. Gene silencing by epigenetic alterations is now regarded as one of the major mechanisms of inactivating tumor suppressor genes, along with gene mutations and deletions.17, 18 In human HCC, a number of methylation-induced inactivation of genes, such as E-cadherin, p16INK4a, 14-3-3 sigma, SOCS-1 and DLC-1, have already been documented.19, 20, 21, 22, 23, 24, 25, 26, 27 However, there is no information on the methylation-mediated silencing of Apo D in HCC. Therefore, we first analyzed the pharmacological induction of Apo D by 5-Azacitidine in human liver cancer cell lines. We confirmed that Apo D gene-expressions were reactivated in Hep3B and HepG2 after incubation with the demethylating agent, and this finding was consistent with our previous findings regarding esophageal cancer cell lines.1

We then investigated the possible correlation between the Apo D mRNA values, quantitatively determined by a real-time RT-PCR and the clinicopathological parameters of HCC patients. The Apo D expression status in nontumor tissue specimens showed no correlation with the clinicopathological factors, such as liver function tests. On the other hand, the Apo D expression values in tumor tissue specimens were significantly associated with the histological grade of the tumor, which is one of the most important prognostic factors for patients with HCC.28, 29, 30 An immunohistochemi-cal analysis also revealed a stepwise decrease in the expression at the protein level of well to moderately and poorly differentiated HCC. The prognosis of HCC patients with a low expression of Apo D was significantly worse than that in patients with a high expression. A multivariate analysis using Cox's proportional hazard model indicated the Apo D expression status to be an independent prognostic factor after curative resection for the patients with HCC. Since the Apo D expression is induced by growth arrest and cell differentiation,31 its association with the histological tumor grade might merely reflect the activities of cell growth and proliferation rather than cell differentiation.

Lopez-Boado et al.4 reported that in human breast cancer cells, retinoic acids strongly induced the Apo D expression, which was accompanied by an inhibition of cell proliferation. In addition, Diez-Itza et al.5 showed the lower Apo D values to be significantly associated with a poorer prognosis in breast cancer patients. They considered that normal breast tissue synthesizes Apo D, and a subset of Apo D expressing tumor cells, possess the required degree of differentiation to synthesize this protein. The presence of Apo D in breast carcinomas could thus be a marker for a high grade of differentiation and a consequence of growth arrest with a subsequent favorable prognosis in the corresponding patients. On the other hand, several recent studies have shown contradictory results regarding prostate32 and breast cancer,33 possibly suggesting the differential regulation of Apo D expression in malignancies with a hormonal dependence on tumor growth.

The molecular mechanisms by which such Apo D expression was silenced in breast carcinomas, especially in poorly differentiated tumors, remains unclear. In our study, we thus proceeded to perform an assay of MSP to confirm the presence of promoter methylation in primary HCC tissues. As expected, an MSP analysis revealed a remarkable hypermethylation of the Apo D gene in moderately or poorly differentiated HCC and a faint methylation in highly differentiated tumors. These findings indicate that the reduced Apo D expression in less differentiated carcinomas might be due to the inactivation of its promoter region due to hypermethylation.

We previously demonstrated the potent tumor-suppressing activity after the transfection of Apo D into esophageal cancer cell lines by a colony focus assay.1 The findings of the current study thus imply that Apo D may play a tumor suppressive role in HCC, as was observed in esophageal carcinoma. Furthermore, the induction of the Apo D expression using demethylating reagents may control the tumor aggressiveness associated with less cellular differentiation, and this action may consequently improve the outcome of patients.


We thank Miss T. Shimooka, Mrs. K. Ogata, Miss N. Kasagi and Miss N. Ando for their excellent technical assistance.