INTRODUCTION

Hepatocellular carcinoma is a highly malignant tumor that is prevalent in Southeast Asia, China, and sub-Saharan Africa. It has a high mortality rate, and recurrence after surgical resection is common. Risk factors, including viral hepatitis Types B and C infection and dietary aflatoxin, are well recognized in hepatocellular carcinoma, but the pathways by which malignant transformation of hepatocytes occurs remains unclear.

Recurrent genetic alterations in hepatocellular carcinoma have been reported from microsatellite analysis and comparative genomic hybridization (CGH) studies (1, 2, 3). Using CGH, we and others have detected recurrent gains of 1q, 6p, 8q, 17q and 20q and losses of 1p, 4q, 8p, 13q, 16q and 17p (4, 5, 6). Among the common aberrations observed, loss of heterozygosity (LOH) of 1p36–p34 has been identified in the putative premalignant cirrhotic nodules and, by using centromeric probes, chromosome 1 polysomy has been reported in liver dysplasia by interphase cytogenetics (7, 8). Thus, deletions on distal 1p and 1q copy gains have been implicated in the early stages of hepatocellular carcinoma development.

In this study, we conducted a simultaneous CGH and microsatellite analysis study on 31 hepatocellular carcinoma cases, aiming to characterize the nature of the chromosome 1 abnormalities. Microsatellite analysis on chromosome 1 was carried out using a panel of 31 polymorphic markers that spanned approximately 5–10cM. This combined CGH and microsatellite analysis identified four distinct regions of allelic imbalances on chromosome 1, and microsatellite analysis has enabled the smallest overlapping regions (SORs) in deletions and high-level gains, as identified by CGH, to be defined.

MATERIALS AND METHODS

Patients

Tumorous liver tissues were obtained from patients (aged 34–76 years) who underwent curative liver surgery for hepatocellular carcinoma at the Prince of Wales Hospital, Hong Kong. Tumor samples from 31 hepatocellular carcinoma patients (77% male) and the corresponding peripheral blood from the same patient were collected. One patient had 2 foci, H29a and H29b. Patients recruited in the present series were predominantly viral hepatitis B carriers (84%). The clinicopathological information of hepatocellular carcinoma patients studied is summarized in Table 1. An experienced pathologist confirmed the diagnosis of hepatocellular carcinoma and evaluated all tumor specimens for the tumor cell content. High molecular weight genomic DNA from tumor tissues and blood leukocytes was extracted by QIAamp Tissue kit and QIAamp Blood kit (Qiagen, Hilden, Germany) respectively.

TABLE 1 Clinicopathologic Characteristics and CGH Findings on Chromosome 1 in 31 Cases of HCC
Full size table

Comparative Genomic Hybridization

Comparative genomic hybridization (CGH) was performed according to the method of Kallioniemi et al. (9), with modifications described in Wong et al. (6). One microgram of tumor and reference DNA was differentially labeled with biotin-16-dUTP (Boehringer Mannheim, Mannheim, Germany) and digoxigenin (dig)-11-dUTP (Boehringer Mannheim), respectively by nick translation. Labeled DNAs were co-hybridized onto normal metaphase chromosomes in excess Cot-1 DNA. After hybridization, biotin signals were detected through avidin-conjugated FITC (fluorescein isothiocyanate) antibody (Sigma, St Louis, MO), and dig-labeled DNA was visualized by TRITC (tetramethylrhodamine isothiocyanate)-conjugated antibody (Sigma). Chromosomes counterstained with DAPI (4′, 6-diamidino-2-phenylindole) were captured through a cooled CCD camera mounted on a Leitz DM RB (Leica, Wetzlar, Germany) fluorescence microscope. Three band-pass filter sets (DAPI, FITC, and TRITC) arranged in an automated filter wheel were employed for image acquisition. CGH software version 3.1 on Cytovision (Applied Imaging Ltd., U.K.) was used in digital image analysis. Average ratio profiles of 10–12 metaphases were calculated based on chromosome identification of the inverted DAPI image. Thresholds for gains and losses were defined as theoretical values of 1.25 and 0.75, respectively. High-level amplifications were considered when ratio exceeded 1.5.

Microsatellite Analysis

Thirty-one fluorescent-labeled polymorphic markers, 16 on 1p and 15 on 1q, were acquired from ABI PRISM Linkage Mapping Sets (Applied Biosystems, Foster City, CA). PCR was performed in 7.5 μL reaction buffer (Perkin Elmer, Foster City, CA) that contained 30 ng of genomic DNA, 5 μm primer pair mix, 2.5 mm GeneAmp dNTP mix, 25 mm MgCl2, and 5 units AmpliTaq Gold DNA Polymerase (Applied Biosystems). The reaction was performed using the GeneAmp PCR System 9600. The PCR conditions consisted of an initial denaturation step at 95° C for 12 minutes, followed by 10 cycles of denaturation at 94° C for 15 seconds, annealing at 55° C for 15 seconds, and extension at 72° C for 30 seconds. A further 20 cycles was carried out at 89° C denaturation for 15 seconds, annealing at 55° C for 15 seconds, and extension at 72° C for 30 seconds. The final extension was performed after the 30th cycle at 72° C for 10 minutes. Amplified DNA products were analyzed on 5% Long Ranger denaturing gel in TBE buffer. Electrophoresis was run on an automated sequencer ABI 377 (PE Biosystems). Data collected were analyzed by Genescan software. Cases were classified as informative when two distinct alleles of similar intensity were found in the normal DNA and were classified as allelic loss or allelic imbalance when ≥50% intensity differences were present between tumor and normal DNAs. Cases of homozygosity were considered noninformative. Examples of allelic imbalances detected in hepatocellular carcinoma are shown in Figure 3.

FIGURE 3
figure 1

Examples of allelic imbalances detected in hepatocellular carcinoma. A, allelic loss of three markers on 1p, D1S468, D1S450, and D1S 2667, were detected in HCC cases H3, H10, and H12. B–D, cases H2, H4, H6, H22 and H24 displayed allelic imbalances at 1q markers, D1S196, D1S249, and D1S2785, which were localized at regions 1q23.1–q24.2, 1q32.1, and 1q43–q44, respectively. Allelic gains on 1q were suggested. T, tumor of the patient. N, corresponding lymphocytes of the same patient.

Full size image

RESULTS

Comparative Genomic Hybridization

Among 31 cases analyzed, partial and whole-chromosome 1p loss were detected in 14 cases (45%). A minimal overlapping region on 1p36–p35 was also determined. Gains of partial or whole chromosome 1q were found in 29 cases (94%). Furthermore, high-level gain of regional or whole 1q was suggested in 14 of 29 that displayed copy gain (48%). An overlapping region was narrowed down to 1q21–q25 and 1q32–qter. CGH imbalances on chromosome 1 were not detected in H29a and H29b. Furthermore, H29a and H29b displayed similar aberrations in CGH analysis and microsatellite analysis. This suggested that both specimens shared a similar clonal origin. In the calculation of percentage incidences, we have therefore scored H29a and H29b as one case. The CGH results obtained are summarized in Table 1, and examples of average ratio profiles on aberrant chromosome 1 are shown in Figure 1.

FIGURE 1
figure 2

Chromosome 1 imbalances detected by CGH. The mean ratio profile of 22–24 analyzed chromosomes (n = 22–24) is depicted with the 95% confidence interval (gold lines). Red and green lines represent thresholds for the chromosomal losses (0.75) and gains (1.25), respectively. A, case H3. Loss of entire p-arm and high-level gain of the entire q-arm were suggested. B, case H14 displayed regional loss of p-arm at 1p33-pter and low-level copy number gain of the entire q-arm. C, Case H30, absence of p-arm imbalances and regional amplification of 1q21–q25 were indicated.

Full size image

Microsatellite Analysis

Using 31 polymorphic markers, markers with allelic imbalances in >40% of informative cases were observed at 24 loci (Table 2). The highest incidences of allelic imbalance identified in the 31 pairs of matched hepatocellular carcinoma and leukocyte-derived DNA were D1S468 at 74% (1p36.32), D1S450 at 67% (1p36.23), D1S2667 at 65% (1p36.22), D1S2697 at 75% (1p36.21), D1S199 at 52% (1p36.13), and D1S234 at 67% (1p36.11). All loci represented regions on the distal 1p region at 1p36. On chromosome 1q, frequent allelic imbalances were detected on D1S2635 at 64% (1q23.1), D1S484 at 67% (1q23.2), D1S2878 at 65% (1q23.3), D1S196 at 70% (1q24.2), D1S249 at 64% (1q32.1), D1S2785 at 75% (1q43), D1S2842 at 73% (1q43) and D1S2836 at 74% (1q44). These regions also lay within the minimal amplified regions identified by the parallel CGH analysis. The allelic imbalances found and their corresponding locations are shown in Figure 2.

TABLE 2 Frequency of Allelic Imbalances Detected on Chromosome 1
Full size table
FIGURE 2
figure 3

Summary of chromosome 1 allelic imbalances detected in 31 cases of hepatocellular carcinoma. The microsatellite markers are listed in order from telomere of p-arm to telomere of q-arm. Their corresponding locations are listed on the left of the ideogram. At a given locus, tumor-specific allelic imbalance is indicated by a filled circle; retention of heterozygosity by an open circle, and noninformative (constitutional homozygosity) by a hatched circle.

Full size image

Overall, 26/31 cases (84%) of the tumors studied displayed allelic imbalances on 1p, and 28/31 cases (90%), on 1q for one or more markers. Only 7 cases showed losses in 1p at all informative markers tested, and 3 cases displayed allelic gains in 1q at all informative markers tested. Four distinct regions of SOR were further determined from the microsatellite analysis findings, these being 1p36.13–p36.22 (D1S199-D1S2667), 1q23.1–1q23.3 (D1S2635-D1S2878), 1q25.1–q31.1 (D1S452-D1S238) and 1q43 (D1S2785-D1S2842). The SOR of allelic imbalance was defined based on the common deleted region among cases, which showed interstitial allelic imbalances. For the region 1p36, the upper boundary was defined by cases H29a and H29b that showed retention at D1S450 and allelic imbalances between D1S199 and D1S2667. The lower boundary was defined by case H21, which showed retention at locus D1S234 and allelic imbalances between D1S199 and D1S468. The SOR on 1p36 was thus defined between D1S199-D1S2667. Similar analysis was performed in defining the SORs on 1q.

CGH versus microsatellite analysis

Concordance between microsatellite analysis and CGH findings on chromosome 1 imbalances was found in >60% of the cases. On the p-arm, deletions defined as being CGH[−], microsatellite analysis[−] were found in 13 cases (H1–4, H6–10, H14, H16, H17, and H21), whereas absence of aberrant 1p defined as CGH[0], microsatellite analysis[0] was found in 7 cases (H23, H24, H26–28, H30, and H31). The overall concordant rate on the p-arm was 65% (20/31 cases).

Whole q-arm gain, as indicated by CGH, was frequent, but regional imbalances on the q-arm were suggested from microsatellite analysis. In cal culating the concordance rate on the 1q arm, we have simplified the analysis by subdividing the q-arm into three regions, namely 1q21–q25, 1q31–q32, and 1q42–q44. In 1q21–q25, CGH gain and allelic imbalances, as defined by CGH[+]microsatellite analysis[+], were found in 20 cases (H1–4, H6–8, H11, H14, H18, H19, H22–28, H30, and H31). There was absence of regional changes as defined by CGH[0]microsatellite analysis[0] in three cases (H15, H21, and H29). The overall concordance rate on 1q21–q25 was 74% (23/31 cases). In 1q31–q32, CGH[+]microsatellite analysis[+] was found in 14 cases (H2–7, H10, H15, H2O, H22, H24, H25, H28, and H31) and there were absence of aberration as defined by CGH[0]microsatellite analysis[0] in 5 cases (H12, H16, H21, H29, and H30). The overall concordance rate on 1q31–q32 was 61% (19/31 cases). In 1q42–q44, CGH[+]microsatellite analysis[+] was found in 15 cases (H1–6, H10, H11, H14, H15, H18, H24, H26, H28, and H31), and CGH[0]microsatellite analysis[0] in 5 cases (H16, H19, H21, H29, and H30). An overall concordant rate of 65% (20/31 cases) was suggested on 1q42–q44.

DISCUSSION

Transformation of normal hepatocytes to hepatocellular carcinoma will, as with other solid tumors, involve activation of oncogene and inactivation of tumor suppressor genes, all of which will lead to increasing genetic instability and loss of cell cycle control. In hepatocellular carcinoma, the role of proto-oncogene(s) and tumor suppressor gene(s) on common gains of 1q and LOH on 1p36–p34 respectively has been implicated. In this study, patients examined were predominantly chronic hepatitis B virus carriers who were derived from a racially and etiologically homogenous population in southern China. The simultaneous CGH and microsatellite analysis on chromosome 1 has enabled the consistently over- and underrepresented regions on chromosome 1 to be defined. ]Although CGH suggested minimal overlapping regions on 1p36–p35, 1q21–q25, and 1q32–qter, microsatellite analysis has further refined chromosome 1 abnormalities to four distinct SORs at 1p36.13–p36.22, 1q23.1–1q23.3, 1q25.1–q31.1, and 1q43.

Our finding of a smallest deleted region on 1p36.13–p36.22 (D1S199 to D1S2667) that flanked approximately 10cM was similar to that identified by Fang et al. (10). Frequent LOH on chromosome 1p36 has been implicated in the progression of intrahepatic cholangiocarcinoma, the second most common liver malignancy (11), and has also been well documented in neuroblastomas (12). Although allelic losses on 1p36 have been suggested to be associated with poor differentiated histologic grade in epithelial ovarian carcinoma (13), deletions on 1p36.12 were found in the early dysplastic lesions and cirrhotic nodules of hepatocellular carcinoma (7). A role for 1p36 deletions in the early events of hepatocarcinogenesis was thus further enforced. Several tumor suppressor genes have been described in 1p36. These include TP73, DFFA, and RIZ (10, 14, 15). Structurally and functionally, TP73 (at 1p36.33) is related to TP53. However, the tumor suppressor function of TP73 as a target essential to the development of hepatocellular carcinoma has been excluded (16). The DFFA gene mapped to 1p36.22 encodes the active subunit of the apoptotic nuclease DNA fragmentation factor (DFF45) (17). Although DFFA has been reported to be homozygously deleted in a neuroblastoma cell line, the role of DFFA in human liver carcinogenesis remains to be investigated (12). Thus far, RIZ is the only tumor suppressor gene that has been suggested in the pathogenesis of hepatocellular carcinoma (18).

Although gain of 1q has been considered as secondary karyotypic changes in hematological malignancies (19, 20), cytogenetic studies have indicated the most frequent structural alteration in hepatocellular carcinoma to be unbalanced 1q translocations that resulted from 1q gain (21, 22). This in turn would suggest aberrant 1q to be a primary karyotypic event in the liver tumorigenesis. By CGH, we have identified, in the current hepatocellular carcinoma series, two minimal amplified regions at 1q21–q25 and 1q32–qter. High-level gain has previously been mapped to 1q12–q22 in hepatocellular carcinoma (6, 23), and common allelic imbalances suggested on 1q22–q23 (2). Gains in chromosomal region 1q23 have been proposed in drug-resistant phenotype in neuroblastoma patients (24) and suggested to be associated with metastases in well-differentiated liposarcoma (25). Within the current SOR identified, 1q23.1–1q23.3, candidate genes described included TRK proto-oncogene (26) and PBX1, a human homeobox gene involved in t(1;19) translocation in acute pre–B-cell leukemias (27).

Specific regions on 1q have been implicated in the pathogenesis of human cancers. High-level amplifications of 1q25 described in metastatic prostate cancer (28) also displayed copy number gain of candidate gene HPC1 (at 1q24–q25) in the advanced stage cancers (29). Allelic imbalances of the same region have also been reported in the progression of cervical carcinoma (30). High-density mapping on chromosome 1 indicated common allelic imbalances on 1q31–q32 in medulloblastoma and breast cancer (31, 32), and by linkage analysis on high-risk prostate cancer families, the predisposing gene for early-onset prostate cancer, PCaP, has been localized on 1q42.2–43 (33, 34). Although target genes within common SOR regions identified on 1q in liver carcinogenesis remained to be verified, our findings, and those from other research groups, strongly suggested the presence of genes residing on these chromosomal regions in cancer progression.

In the 31 primary hepatocellular carcinoma cases studied, the overall concordance between CGH and microsatellite analysis findings was 65% for 1p, 74% for 1q21–q25, 61% for 1q31–q32 and 65% for 1q42–q44. The remaining disconcordant cases can be sub-divided into two groups, the first group being allelic imbalance detected by microsatellite analysis but not by CGH. This might be due to the lower resolution of CGH analysis that precluded detection of regions smaller than 10 Mb. In addition, some cases demonstrated loss of the whole p-arm by microsatellite analysis but showed absence of imbalances by CGH. This might have been due to a loss of p-arm accompanied by reduplication of the remaining allele resulting in no net gain or loss of chromosomal material. The second group included cases that could be detected by CGH as gains or losses but not by microsatellite analysis. Although CGH has a lower resolution in detecting regional gains and losses, it has a higher sensitivity in identifying low-level copy gains or losses.

In conclusion, our combined CGH and microsatellite analysis studies on chromosome 1 have enabled deletion boundaries on 1p and regions of allelic gains on 1q to be defined. The SORs identified may have implications for inactivation of tumor suppressor gene(s) at 1p36.13–p36.22 and activation of proto-oncogenes on 1q23.1–q23.3, 1q25.1–q31.1, and 1q43 in the development and progression of hepatocellular carcinoma.