Abstract
There is limited data on the molecular mechanisms whereby hepatitis D virus (HDV) promotes liver cancer. Therefore, serum and liver specimens obtained at the time of liver transplantation from well-characterized patients with HDV-HCC (n = 5) and with non-HCC HDV cirrhosis (n = 7) were studied using an integrated genomic approach. Transcriptomic profiling was performed using laser capture–microdissected (LCM) malignant and nonmalignant hepatocytes, tumorous and nontumorous liver tissue from patients with HDV-HCC, and liver tissue from patients with non-HCC HDV cirrhosis. HDV-HCC was also compared with hepatitis B virus (HBV) HBV-HCC alone, and hepatitis C virus (HCV) HCV-HCC. HDV malignant hepatocytes were characterized by an enrichment of upregulated transcripts associated with pathways involved in cell-cycle/DNA replication, damage, and repair (Sonic Hedgehog, GADD45, DNA-damage-induced 14–3-3σ, cyclins and cell-cycle regulation, cell cycle: G2–M DNA-damage checkpoint regulation, and hereditary breast cancer). Moreover, a large network of genes identified functionally relate to DNA repair, cell cycle, mitotic apparatus, and cell division, including 4 cancer testis antigen genes, attesting to the critical role of genetic instability in this tumor. Besides being overexpressed, these genes were also strongly coregulated. Gene coregulation was high not only when compared with nonmalignant hepatocytes, but also to malignant hepatocytes from HBV-HCC alone or HCV-HCC. Activation and coregulation of genes critically associated with DNA replication, damage, and repair point to genetic instability as an important mechanism of HDV hepatocarcinogenesis. This specific HDV-HCC trait emerged also from the comparison of the molecular pathways identified for each hepatitis virus–associated HCC. Despite the dependence of HDV on HBV, these findings suggest that HDV and HBV promote carcinogenesis by distinct molecular mechanisms.
Implications:
This study identifies a molecular signature of HDV-associated hepatocellular carcinoma and suggests the potential for new biomarkers for early diagnostics.
Introduction
Hepatocellular carcinoma (HCC) is the fifth most common human cancer and the second leading cause of cancer-related death worldwide (1). Although the major etiologic agents and risk factors for HCC are well defined, the molecular mechanisms of hepatocarcinogenesis remain elusive (2, 3). The increasing incidence of HCC worldwide (4), along with the lack of early diagnostic markers and effective therapies, has made this disease one of the most challenging to control. Cirrhosis is the single most important risk factor, being present in 80% of individuals with HCC (5), and infection with hepatitis viruses account for over 60% of all cases globally (5). The application of genomic technologies has provided valuable tools to investigate the pathogenesis of complex diseases using a global approach (6), and many studies of gene expression on HCC-associated with hepatitis B virus (HBV; ref. 7) or hepatitis C virus (HCV; refs. 8, 9) have been reported. However, there are no data on the molecular profiling of hepatitis D virus (HDV)-associated HCC.
HDV is a unique defective RNA virus that requires the helper function of HBV for viral assembly and in vivo transmission (10).It is highly pathogenic and causes the least common but most severe and rapidly progressive form of chronic viral hepatitis, leading to cirrhosis in about 80% of the cases within 10 years (11). HDV-related cirrhosis may be a stable disease for many years, but a high proportion of patients eventually die of hepatic decompensation or HCC unless they undergo liver transplantation (LT). However, the proportion of patients who will develop each of these long-term complications remains uncertain due to the lack of large prospective studies on the natural history of HDV. In two longitudinal studies, the annual incidence rates were 2.5% and 2.7% for liver decompensation and 1% to 2.8% for HCC (12, 13).
Owing to the vital dependence of HDV on HBV, the specific role of HDV in promoting HCC remains to be fully elucidated. It is unknown whether HCC is the result of the underlying cirrhosis, of a direct oncogenic effect of HDV, or of a cumulative effect of HBV and HDV. Access to a unique collection of liver samples from well-characterized patients with HDV-associated HCC who underwent LT provided us with the opportunity to study the role of host and viral factors in HDV-associated HCC. Gene expression profiling was performed on selected laser capture–microdissected (LCM) malignant and nonmalignant hepatocytes (hereafter abbreviated as MH and NMH), and from multiple whole liver tissue (WLT) specimens obtained from the tumor and surrounding nontumorous tissue of individual livers containing HCC. Additional WLT specimens were obtained from livers with non-HCC HDV cirrhosis. Because HDV is invariably associated with HBV, we have also compared the gene expression profiles of patients with HDV-associated HCC with those from patients with HCC associated with HBV alone (7), as well as with those from patients with HCC associated with HCV, to identify distinct molecular signatures for each hepatitis virus-associated HCC, which may shed new light on pathogenesis and facilitate the discovery of new biomarkers for the early detection of these deadly tumors.
Materials and Methods
Patients
Multiple liver specimens and serum were obtained from a cohort of 5 male patients, aged 57 ± 3 years (mean ± SEM) who underwent LT for HDV-associated HCC, and 7 patients, 3 females, and 4 males, aged 56 ± 1 years, with HDV non-HCC cirrhosis who underwent LT for end-stage liver disease. For comparative purposes, the study also included liver specimens from 11 patients, 1 female, and 10 males, aged 56.7 ± 3.6 years, with HCV-associated HCC who underwent LT, and from 11 patients with HBV-associated HCC, all males, aged 60 ± 8 years, whose data were included in a recent report (7). All patients were followed between 2004 and 2009 at the Liver Transplantation Center of the Brotzu Hospital in Cagliari, Italy. The patient characteristics are described in the Results section. All patients provided written informed consent, and the protocol was approved by the ethical Committee of the Hospital Brotzu (Cagliari, Italy). The study was also approved by the Office of Human Subjects Research of the NIH, granted on the condition that all samples were deidentified.
Liver pathology
Liver biopsies were evaluated blindly by two expert hepatopathologists (S. Govindarajan and D.E. Kleiner). For each liver biopsy specimen, activity grade, and stage of fibrosis were established according to Ishak scoring system (14). The grade of tumor differentiation was evaluated according to the Edmondson and Steiner grading system (15).
Serologic and virological assays
Serologic markers of infection with hepatitis viruses were available for all patients at the time of LT or partial hepatectomy. HBsAg, anti-HBs, anti-HBc, IgM anti-HBc, HBeAg, anti-HBe, antibody to HCV (anti-HCV), and antibody to human immunodeficiency virus (anti-HIV) were measured with a commercial enzyme immunoassay (Abbott Laboratories). Antibodies against hepatitis delta antigen (HDAg), IgG, and IgM anti-HD, were measured using commercial enzyme immunoassays (Sorin Biomedica). Serum HBV DNA was quantified by a commercial assay (Amplicor, HBV Monitor test; Roche Diagnostics). Serum HCV RNA was measured by a commercial assay (Cobas Amplicor HCV Monitor 2.0, Roche Diagnostics). Serum HDV RNA was evaluated by PCR as reported previously (16).
Quantification of liver HDV RNA by real-time PCR
Total RNA was extracted from stored frozen liver specimens using QIAzol (Qiagen) reagent according to the manufacturer’s recommendations, and RNA quality and integrity were assessed using the RNA 6000 Nano assay on the Agilent 2100 Bioanalyzer. HDV RNA was measured by TaqMan, as described previously (17), with cycling conditions based on the manufacturer’s recommendations; the forward, reverse primer, and probe concentrations were 900, 900 and 225 nmol/L, respectively. The primers and probe were: forward primer GAC CCG AAG AGG AAA GAA GGA (position 894), reverse complementary primer AGA GTT GTC GAC CCC AGT GAA TAA (position 971) and MGB probe 6-FAM-CGA GAC GCA AAC CTG TGA (position 917). A secondary quantity standard was developed on the basis of a plasmid generously provided by Dr. John Taylor (Fox Chase Cancer Center, Philadelphia, PA). HDV RNA quantity was expressed as genome equivalents (GE) per ng of total RNA.
Quantification of liver HBV DNA by real-time PCR
HBV DNA in liver was quantified using a modification of a previously described method (7). The primers/probe were located near the 5′ end of the S gene. Each 20-μL reaction contained 50ng of DNA, 45 pmol of forward (5′- GGA CCC CTG CTC GTG TTA CA-3′) and reverse (3′- TTG AGA GAA GTC CAC CAC GAG TC-5′) primers, 12.5 pmol of nonfluorogenic-quenched probe (6FAM-TGT TGA CAA GAA TCC TCA), and TaqMan Fast Universal PCR Master Mix (Applied Biosystems). PCR was performed using an ABI PRISM 7900HT Sequence Detection System (Applied Biosystems). Conditions included incubation at 95°C for 20 seconds followed by 45 PCR cycles of 1 second at 95°C and 20 seconds at 60°C. Viral titers were expressed as log10 GE per ng of DNA.
Detection of intrahepatic HDV and HBV markers by IHC
Formalin-fixed, paraffin-embedded liver biopsy sections from five paired liver biopsies obtained from both the tumor and the surrounding nontumorous tissue of five patients with HDV-associated HCC, and 14 liver biopsies obtained from 7 patients with non-HCC HDV cirrhosis, including one biopsy from the right and one from the left lobe of the cirrhotic liver, were stained for intrahepatic HBsAg, HBcAg, and HDAg. HBsAg (Thermo predilute, clone 3E7) and HBcAg (Dako B0586, 1:500) stains were performed on a Ventana Benchmark Ultra. HDAg stains were performed by manual staining using a high titer patient serum diluted at 1:1,000 and detected using a biotinylated goat antihuman IgG (1:200, Vector Laboratories). Detection was performed using an avidin–biotin complex and diaminobenzidine chromagen.
Gene expression profiling
Whole liver tissue.
To investigate the molecular heterogeneity within and outside the tumor of HDV-associated HCC, we analyzed from each patient with HCC one liver specimen obtained from the tumor and one from the surrounding nontumorous tissue (Supplementary Fig. S1A and S1B). In addition, we had the unique opportunity to study multiple liver specimens from 2 of the 5 patients (patients 104 and 129), including 5 specimens from the tumor and 12 from the surrounding nontumorous areas taken in all 4 directions, which for simplicity we defined north, south, east, and west, and at different distances from the center of the tumor (Supplementary Fig. S1C). Specifically, 5 biopsies were taken from the tumor, one at the center (area A) and 4 in the periphery of the tumor (area B) in all 4 directions; 4 were taken from the perilesional area (area C); 4 were taken 2–3 cm from the tumor (area D); and 4 from the edges of the liver (area E). In total, the whole liver tissue (WLT) samples collected from the patients with HDV included 11 from the tumor and 24 from the nontumorous tissue. In addition, we also studied liver specimens obtained from the right and the left lobe of 7 patients with non-HCC cirrhosis for a total of 29 specimens (Supplementary Fig. S1D). Each liver specimen was divided into two pieces: one was snap-frozen for molecular studies and the other was formalin-fixed and paraffin-embedded (FFPE) for pathologic examination. Importantly, FFPE sections obtained from the tumor or the perilesional area were also observed microscopically to check the homogeneous histologic composition of the samples. A mixed population of malignant and nonmalignant hepatocytes was found only in two sections, and the corresponding frozen liver specimens were excluded from microarray analysis.
Laser capture microdissection.
Because the liver contains a heterogeneous cell population, gene expression profiling was performed on MH and NMH isolated by laser capture microdissection (LCM) in all 5 HDV-associated HCC livers. For each patient, two paired samples for LCM were obtained, one from the center of the tumor and one from the most distant nontumorous tissue (Supplementary Fig. S1E and S1F). We optimized the LCM method as recently reported (7) based on the procedure described by Erickson and colleagues (18).
HBV- and HCV-associated HCC samples
To further elucidate the role of HDV in hepatocarcinogenesis, data from HDV-associated HCC were compared with the data from 11 patients with HCV-associated HCC and 11 patients with HBV-associated HCC (7). HCV-HCC data included 44 WLT samples from the tumor and 31 from the nontumorous tissue, and 9 paired LCM samples of MH and NMH. HBV-HCC data included 39 WLT samples from the tumor and 81 from the nontumorous tissue, and 10 paired LCM samples of MH and NMH.
Gene expression profiling
All liver specimens were analyzed by microarray using Affymetrix Human U133 Plus 2.0 arrays (Affymetrix), which contain 54,675 transcripts representing approximately 27,000 unique human genes. Total RNA from WLT was extracted from frozen liver specimens as described previously (19) using TRIzol reagent (Invitrogen) according to the manufacturer’s recommendations; total RNA from microdissected hepatocytes was extracted using the Arcturus PicoPure RNA Isolation Kit (Life Technologies), as reported previously (7). The specimens used for RNA extraction and LCM were derived from the same frozen liver samples. Total RNA quality and integrity were assessed using the Agilent 2100 Bioanalyzer. To maintain comparability between LCM and WLT, gene expression profiling was performed using the same technique as reported previously (19). Total liver RNA (50 ng) obtained from WLT and microdissected hepatocytes was subjected to two successive rounds of amplification (20), and the resultant RNA was then subjected to biotin labeling, hybridization, staining, washing, and scanning procedures according to standard Affymetrix protocols. All microarray datasets are available at the following Gene Expression Omnibus link: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE107170
Statistical analysis
Microarray data were analyzed using BRB-Array Tools Version 4.2 (21). Microarray raw data (CEL files) were summarized and normalized by the RMA method. Transcripts showing minimal variation (less than 1.5-fold deviations from the median in more than 80% of the arrays) were excluded from the analysis. After filtering, only 7,000 genes were eligible for subsequent analyses. Differentially expressed genes were identified by comparing the tumor versus the nontumorous tissue, or MH versus NMH, using a t test with a FDR <10%. Gene expression fold changes were calculated as the ratio between the geometric means of tumor and nontumorous tissues, or between MH and NMH. Statistical tests and heatmaps were performed using single patients as unit data. All methods were identical to those applied to the microarray data analysis of patients with HBV-HCC (7). Multidimensional scaling was performed using individual samples from each patient. Pathway and network analysis was performed using Ingenuity Pathway Analysis (IPA, v 9.0, Qiagen Redwood City, www.qiagen.com/ingenuity). The association of genes to pathways was evaluated as the ratio between the number of genes present in the dataset and the total number of genes that map to the same pathway. The Fisher exact test was also used to calculate the probability of such association. Additional gene annotations were obtained from the Molecular Signature Database (Hallmark, Reactome and Gene Ontology BP gene sets, available from http://www.broadinstitute.org/gsea/msigdb). Clusters of coregulated genes were investigated by the t-distributed stochastic neighbor embedding (t-SNE) method (22) that enables a dimensionality reduction more defined than other classical methods such as principal component analysis or multidimensional scaling, and has been already successfully applied to transcriptomic data (23, 24). For t-SNE, we used the Rtsne function available from the CRAN repository (25), with the following parameters: perplexity = 20, number of PCA components before t-SNE = 5, and number of iterations = 10,000.
IHC staining for BRCA1 and H2AFX
Formalin-fixed, paraffin-embedded human liver biopsy sections were labeled with mouse monoclonal anti-BRCA1 (MS110) from Millipore-Sigma and rabbit polyclonal anti-H2AFX (Ab A11361) from ABclonal. Staining was carried out on the Bond RX (Leica Biosystems) platform according to manufacturer-supplied protocols. Briefly, 5-μm-thick sections were deparaffinized and rehydrated. Heat-induced epitope retrieval (HIER) was performed using Epitope Retrieval Solution 1, pH 6.0, heated to 100°C for 20 minutes. The specimen was then incubated with hydrogen peroxide to quench endogenous peroxidase activity prior to applying the primary antibody. Both primary antibodies were applied at a dilution of 1:200. Detection with DAB chromogen was completed using the Bond Polymer Refine Detection kit (Leica Biosystems), which included a hematoxylin counterstain. Slides were finally cleared through gradient alcohol and xylene washes prior to mounting and coverslipping.
Results
The demographic, clinical, serologic, and virologic features of the 12 individuals with HDV-associated liver disease, including 5 patients with HCC and 7 with non-HCC cirrhosis, are reported in Table 1. None of the features were statistically different between the two groups of patients, except for a lower number of platelets in those with non-HCC cirrhosis who underwent LT for end-stage HDV liver disease. Chronic hepatitis D is typically associated with splenomegaly (10). All patients were positive for serum HDV RNA and IgG anti-HDV, whereas IgM anti-HD was positive in 3 of 5 patients with HCC and in all patients with HDV non-HCC cirrhosis. Regarding the serologic HBV profile, there was no difference between the two groups with HDV-associated disease. All were positive for hepatitis B surface antigen (HBsAg), antibody to hepatitis B core antigen (anti-HBc) and antibody to hepatitis B e antigen (anti-HBe), and negative for hepatitis B e antigen (HBeAg) and antibody to HBsAg (anti-HBs). The levels of serum HBV DNA were very low in all cases (Table 1). The activity grade and stage of fibrosis of the surrounding nontumorous tissue are reported in Table 1, and there were no significant differences between the two groups. The grade of tumor differentiation was G3 in 4 patients, and G2 in the remaining patient. The tumor size was less or equal to 3 cm in 3 patients and larger than 3 cm in the remaining 2 patients. The clinical and histopathologic features of the 11 patients with HBV-associated HCC have been reported previously (7), whereas those from the 11 patients with HCV-associated HCC are shown in Table 1. The higher activity grade seen in patients with chronic hepatitis D compared with those with chronic hepatitis C (Table 1) confirmed that HDV induces the most severe form of chronic viral hepatitis (10).
Table 1.
Demographic, clinical, serologic, virological, and pathologic features of the 12 individuals with HDV-associated liver disease, including 5 patients with HCC and 7 with non-HCC cirrhosis, and the 11 individuals with HCV-associated HCC
| HDV-associated HCC | HDV-associated non-HCC cirrhosis | HCV-associated HCC | ||
|---|---|---|---|---|
| Patients, no. | 5 | 7 | Patients, no. | 11 |
| Age, y | 57 ±3 | 56 ± 1 | Age, y | 58 ± 4 |
| Male, no. (%) | 5 (100) | 4 (57) | Male, no. (%) | 10 (90.9) |
| Alanine aminotransferase, U/La | 87 ± 25 | 78 ± 23 | Alanine aminotransferase, U/La | 81 ± 13 |
| Aspartate aminotransferase, U/Lb | 82 ± 21 | 81 ± 14 | Aspartate aminotransferase, U/Lb | 86 ± 16 |
| γ-glutamyltransferase, U/Lc | 98 ± 22 | 67 ± 20 | γ-glutamyltransferase, U/Lc | 104 ± 26 |
| Prothombin time, INRd | 1.4 ± 0.0 | 1.6 ± 0.1 | Prothombin time, INRd | 1.4 ± 0.1 |
| Total bilirubin, mg/dLe | 2.3 ± 1.3 | 2.4 ± 0.4 | Total bilirubin, mg/dLe | 1.9 ± 0.5 |
| Platelets (103/mL)f | 101.4 ± 16.1j | 53.6 ± 11.5 | Platelets (103/mL)f | 138.8 ± 24 |
| α-fetoprotein, ng/mgg | 17.4 ± 10.5 | 13.0 ± 6.2 | α-fetoprotein, ng/mgg | 898 ± 495 |
| Liver pathology | Liver pathology | |||
| Nontumorous tissueh | Nontumorous tissueh | |||
| Activity grade | 9.1 ± 1.1 | 11.0 ±1.6 | Activity grade | 6.9 ± 0.8 |
| Fibrosis stage | 6.0 ± 0.0 | 6.0 ± 0.0 | Fibrosis stage | 5.2 ± 0.5 |
| F5, no. | 5 | 0 | F5, no. | 2 |
| F6, no. | 0 | 7 | F6, no. | 8 |
| Tumor gradei | Tumor gradei | |||
| G2, no. | 1 | G2, no. | 5 | |
| G2/G3, no. | 0 | G2/G3, no. | 1 | |
| G3, no. | 4 | G3, no. | 5 | |
| Tumor size | Tumor size | |||
| ≤ 2and ≥ 3 cm, no. | 3 | ≤ 2and ≥ 3 cm, no. | 6 | |
| >3 cm, no. | 2 | >3 cm, no. | 5 | |
| Serum HDV RNA positive, no. | 5 | 7 | Serum HCV RNAk | |
| IgG anti-HDag positive, no. | 5 | 7 | positive, no. | 9 |
| IgM anti-HD positive, no. | 3 | 7 | negative, no. | 1 |
| Serum HBV DNA (Log10 IU/mL) | 1.6 ± 0.4 | 1.7 ± 0.3 | Anti-HCV positive, no. | 11 |
| HCV genotypel | ||||
| 1a/1B, no. | 7 | |||
| 2a/2c, no. | 1 | |||
| 4 | 1 |
NOTE: Plus–minus values are means ± SEM.
Normal range, ≤43 UI per liter.
Normal range, ≤42 UI per liter.
Normal range, ≤38 UI per liter.
Normal range, 0.80–1.20 international normalized ratio (INR).
To convert serum bilirubin values to micromoles per liter, multiply by 17.1.
Normal values range ≥ 159–≤388 (103/μL).
Normal range, <10.0 ng/mL.
The degree of activity and stage of fibrosis were assessed according to Ishak scoring system (14)
The tumors were graded using the Edmondson–Steiner grading system (15).
Statistically significant difference between HDV-HCC and HDV-non-HCC (P = 0.032).
Determined in 10 patients out of 11.
Determined in 9 patients out of 11.
HDV RNA and HBV DNA levels in different areas of livers containing HCC and in controls with non-HCC HDV cirrhosis
The levels of HDV RNA within the tumor were lower than in the surrounding nontumorous tissue in 2 patients (patient 104 and 129) in whom up to 16 and 17 liver specimens (Supplementary Fig. S1C), respectively, were available for intrahepatic HDV RNA testing (Fig. 1A). This decrease in HDV replication was observed between the periphery of the tumor and the perilesional area, whereas no changes were documented in the different areas of the surrounding nontumorous areas. The levels of HDV RNA tended to be higher in non-HCC cirrhosis compared with patients with HCC, with no differences among different liver areas, spanning both the right and left lobes (Fig. 1B). The levels of intrahepatic HBV DNA were markedly lower in all patients with HDV, including both patients with HCC and non-HCC cirrhosis (Fig. 1C and D). In contrast, the levels of HBV DNA in patients with HBV-associated HCC were consistently higher than in patients with HDV-HCC, both within and outside the tumor (Supplementary Fig. S7).
Figure 1.
HDV RNA and HBV DNA in tumor and nontumorous tissues of liver containing HDV-associated HCC and IHC staining of HDV and HBV markers. A, HDV RNA levels in tumorous and nontumorous tissues of individual patients with HCC. B, HDV RNA in the left and right lobes of individual controls with non-HCC HDV cirrhosis. C, HBV DNA levels in tumorous and nontumorous tissues of individual patients with HDV-associated HCC. D, HBV DNA in the left and right lobes of individual controls with non-HCC HDV cirrhosis. In all plots, bars indicate the mean ± SEM. E, Staining for HBsAg and HDAg within the tumor and in the surrounding nontumorous tissue, and in a representative control with non-HCC HDV cirrhosis. The white arrow indicates a positive nuclear staining for HDAg.
Intrahepatic HDV and HBV markers
Among patients with HDV-associated HCC, IHC staining for HBsAg, HBcAg, and HDAg showed absence of HBcAg staining in all but one patient, a finding consistent with the typically low levels of HBV replication in patients with chronic HDV infection (10). Within the HDV tumor, HBsAg and HDAg were detected in 2 and 3 patients, respectively, whereas the prevalence of these markers was higher in the surrounding nontumorous area as well as in non-HCC cirrhosis (Table 2; Fig. 1E).
Table 2.
Detection of HBsAg, HBcAg, and HDAg in liver tissue of patients with HDV-associated HCC, both in tumor and nontumorous area, and in patients with non-HCC HDV cirrhosis
|
Tumor |
Nontumor |
||||||
| HCC | Patient no. | HBsAg | HBcAg | HDAg | HBsAg | HBcAg | HDAg |
| 5 | 0 | 0 | 0 | 1 | 0 | 1 | |
| 19 | 3 | 0 | 1 | 1 | 0 | 2 | |
| 27 | 1 | 0 | 1 | 1 | 0 | 1 | |
| 104 | 0 | 0 | 1 | 1 | 0 | 2 | |
| 129 | 0 | 0 | 0 | 0 | 0 | 0 | |
| Total positive patients (%) | 2/5 (40) | 0/5 | 3/5(60) | 5/5(100) | 0/5 | 4/5(80) | |
|
Left lobe |
Right lobe |
||||||
| Non-HCC | Patient no. | HBsAg | HBcAg | HDAg | HBsAg | HBcAg | HDAg |
| 6 | 0 | 0 | 0 | NA | NA | NA | |
| 73 | 1 | 0 | 3 | 1 | 0 | 2 | |
| 90 | 1 | 0 | 1 | 1 | 0 | 1 | |
| 113 | 1 | 0 | 0 | 1 | 0 | 0 | |
| 134 | 2 | 1 | 1 | 1 | 1 | 1 | |
| 152 | 2 | 0 | 1 | 1 | 0 | 1 | |
| 177 | 1 | 0 | 1 | 1 | 0 | 1 | |
| Total positive patients (%) | 6/7 (86) | 1/7 (14) | 5/7 (71) | 6/6 (100) | 1/6 (17) | 5/6(83) | |
NOTE: 0, no staining; 1, <10% positive; 2, 10%–50% positive; 3, >50% positive.
Abbreviation: NA, not available.
Differential gene expression in WLT and LCM tumor samples
An unsupervised multidimensional scaling (MDS) using all genes (about 7,000) that passed the filtering criteria showed a clear separation between tumorous and nontumorous areas from WLT samples, as well as between MH and NMH from LCM samples (Fig. 2A). However, only a relatively small number of genes were found to be differentially expressed in tumor samples from WLT (n = 385, listed in Supplementary Table with S1A) and LCM (n = 547, listed in Supplementary Table S1B) with a prevalence of downregulated genes in both cases (Fig. 2B). No significant changes were found between different nontumorous areas (Fig. 2E). Conversely, an abrupt change was detected between the tumor and nontumorous areas, just outside of the tumor boundaries (Fig. 2E). The gene expression profile of nontumorous areas was very similar to that of livers of patients with HDV cirrhosis without HCC (Fig. 2D), consistent with the presence of cirrhosis in the surrounding nontumorous tissue.
Figure 2.
A, Unsupervised multidimensional scaling (MDS) plot of WLT and LCM samples obtained from HDV-associated HCC livers. The plot shows a complete separation between tumor and nontumor liver samples. In addition, WLT samples denote a greater dispersion in both groups, which is consistent with the greater heterogeneity of WLT compared with LCM hepatocytes. B, Pie charts showing the number of upregulated and downregulated genes identified in WLT and LCM samples of HDV-associated HCC. C, Correlation of fold changes of differentially expressed genes in both LCM and WLT samples from HDV-associated HCC. Fold changes were calculated as the ratio between malignant and nonmalignant hepatocytes (LCM) or between tumor and nontumorous samples (WLT). Red and green points represent upregulated and downregulated genes, respectively. D and E, Heatmaps of the 385 genes differentially expressed in WLT samples from HDV-associated HCC. D, Tumor and nontumor tissues of 5 patients with HDV-HCC and, for comparative purposes, of 7 patients with HDV-associated cirrhosis without HCC. E, Multiple samples of two patients with HDV-HCC, obtained from the center and the periphery of the tumor (heatmap columns A and B, respectively), and from the nontumorous tissue at increasing distances from the tumor (heatmap columns C, D, E). Gene expression levels were log2-transformed and row-wise standardized. Upregulated genes are shown in shades of red; downregulated genes in shades of green.
Approximately 50% of the genes differentially expressed in WLT samples were also differentially expressed in LCM samples. Interestingly, the fold changes of all genes common to WLT and LCM samples were closely correlated (R2 = 0.97; Fig. 2C).
Molecular pathways of HDV-associated HCC
The 20 top-scored pathways associated with MH genes in HDV-associated HCC are shown in Fig. 3A, and all genes included in these 20 pathways are shown in Supplementary Fig. S2. The top-scored pathway was represented by hepatic fibrosis and hepatic stellate cell activation (P = 0.00015), with all genes downregulated, suggesting that the extracellular matrix production was indeed inhibited in the tumor. The second pathway was STAT3 (P = 0.0005), which plays an important role in the normal development, as well as in the regulation of cancer metastases (26). However, the most significant finding was represented by a group of 6 pathways (sonic hedgehog, GADD45, DNA damage-induced 14–3-3σ, cyclins, and cell-cycle regulation, cell cycle: G2–M DNA damage checkpoint regulation, hereditary breast cancer; P values ranging between 0.0008 and 0.016), with a majority (80%) of upregulated genes, involved in several interrelated functions inherent to cell cycle/DNA replication, damage, and repair (e.g., BARD1, BRCA1, CCNA2, CCNB1, CCNE2, CDK1, CDKN2C, GSK3B, H2AFX, MSH2, NPM1, PRKDC and TOP2A). We also identified additional genes not included in the 6 molecular pathways but functionally related to cell kinetics and mitotic apparatus (ANLN, ASPM, BUB1B, CASC5, CDCA3, CDK5R1, CDKN1C, CENPJ, CENPF, CEP55, CENPW, DSCC1,FEN1, HMMR, KIF20A, MAP2K3, MAP3K9, MELK, MPHOSPH9, NCAPD2, NDC80, NEK2, NUF2, PRC1, RMI1, RPS27, TAF10, TRIP13 and TTK). Some of these genes (CDK1, H2AFX, TOP2A, FEN1, KIF20A, MELK, NCAPD2, PRC1, RMI1, TRIP13 and TTK) have been also associated with chromosome instability and correlated with enhanced tumor progression, early tumor recurrence and poor survival of patients with HCC (27, 28). Considering that the majority of genes are downregulated in HDV-HCC, it is worth noting that the vast majority (88%) of this subset of genes were upregulated.
Figure 3.
Top-scored canonical pathways of genes differentially expressed in LCM samples (malignant hepatocytes, MH, top row) and genes differentially and uniquely expressed in WLT samples (bottom row) obtained from HDV-, HBV-, and HCV-HCC. A, The most significant pathways of HDV-MH are represented by hepatic fibrosis and hepatic stellate cell activation, followed by pathways mostly involved in cell-cycle/DNA replication, damage, and repair (sonic hedgehog, GADD45, DNA damage-induced 14–3-3σ, cyclins, and cell-cycle regulation, cell cycle: G2–M DNA damage checkpoint regulation, hereditary breast cancer), with a vast majority of upregulated genes, and other pathways involved in transcription, growth, and inflammation. B, Conversely, HBV-MH show pathways mainly associated with retinoic acid receptor functions and metabolic processes, with a predominance of downregulated genes, and pathways associated with cell remodeling and motility functions, with a predominance of upregulated genes. C, HCV-MH show pathways associated with retinoic acid receptor functions, in common with HBV-MH, and pathways associated to amino acid degradation, in all cases mostly downregulated. In addition, HCV-MH show two pathways (GADD45 and cell cycle: G2–M DNA damage checkpoint regulation) that were also identified in HDV-MH, although with lower percentages of upregulated genes. A detailed list of the pathways identified in HDV, HBV, and HCV-MH is shown in Supplementary Fig. S3. D, Top-scored canonical pathways of WLT-unique genes of HDV-HCC include several biosynthetic pathways (spermine, inosine-5′-phosphate, spermidine, serine, and glycine biosynthesis) with 100% of genes upregulated. Other pathways, with genes mostly downregulated, are involved in IFN signaling and in GAGs, glycerol, and ceramide degradation. E, WLT-unique pathways of HBV-HCC are mostly involved in amino acid metabolism (biosynthesis, degradation, and transformation), activation of LXR/RXR NR, prothrombin activation, coagulation, and acute-phase response signaling, with all genes downregulated. F, WLT-unique pathways of HCV-HCC include 13 pathways associated with cytokine signaling and immune response, mostly downregulated. It should be noted that the pathways associated with HDV-, HBV-, and HCV-HCC WLT-unique genes are all different except one (superpathway of serine and glycine biosynthesis I) that was common to HDV- and HBV-HCC (Supplementary Fig. S6). Columns (quoted on the left y-axes) represent the percent ratio between the number of genes present in the dataset and the total number of genes present in the database, for each pathway. The green and red portions of columns indicate down- and upregulated genes, respectively. The blue curves (quoted on the right y-axes) show the statistical significance of each pathway, expressed as the negative log of the P value of Fisher exact test. The arrows point to the significance threshold corresponding to P = 0.05 on the log scale. Pathways were obtained by Ingenuity Pathway Analysis (www.ingenuity.com).
Because HDV is a defective virus that always coexists with HBV, we also investigated the molecular pathways identified in MH derived from patients with HCC associated with HBV alone. In addition, the availability of liver specimens from well-characterized patients with HCV-associated HCC provided us with the unique opportunity to extend the analysis of the pathways also to MH derived from HCV-associated HCC. Remarkably, among the 6 pathways involved in cell cycle/DNA replication, damage, and repair detected in HDV-MH, two (GADD45 and cell cycle) were also found in HCV-MH, while none of them was found in HBV-MH. Thus, despite the dependence of HDV on HBV, our findings suggest that the molecular signature of HDV-HCC is markedly different from that of HBV-HCC and that genetic instability is a specific feature of HDV-associated HCC. The molecular pathways of HBV-MH were primarily associated with metabolic processes, retinoic acid receptor, cell remodeling, and motility functions (Fig. 3B). Pathways associated with retinoic acid receptor functions were also found in HCV-MH, but only two of them were in common with HBV-MH. Moreover, HCV-MH showed several amino acid degradation pathways, all with mostly downregulated genes (Fig. 3C). A synopsis of the pathways identified in MH derived from the three tumors is shown in Supplementary Fig. S3.
The BRCA1 network in HDV-associated HCC
Among the 20 top-scored pathways detected in HDV-associated HCC, hereditary breast cancer signaling was one of the most closely associated with various common tumors (breast, ovarian, prostate, and colorectal). Thus, we extended the analysis to investigate the network of genes functionally associated with BRCA1, one of the most studied genes in breast cancer susceptibility, which is involved in DNA damage repair, regardless of their inclusion in the hereditary breast cancer pathway. A large number (88%) of genes associated with the BRCA1 network were found to be differentially expressed by MH in HDV-associated HCC (Fig. 4). In contrast, this pathway did not emerge in the other viral associated HCC, and BRCA1 was not differentially expressed in HBV- HCC nor in HCV-HCC, highlighting its specific association with HDV-HCC.
Figure 4.
Network of genes functionally associated with BRCA1, obtained from the IPA database. The network shows genes that are connected to BRCA1 by one-node step (25 genes, forming the yellow-highlighted inner circle) or two-node steps (44 genes, forming the outer circle). Shapes represent functional categories, as explained in the bottom. A large majority of these genes (61 of 69, indicated by colored shapes) were differentially expressed in MH of HDV-HCC. Red and green colors indicate upregulated (n = 40) and downregulated (n = 21) genes, the color intensity being roughly proportional to the fold change. Fourteen of the 61 differentially expressed genes (BRCA1, ANLN, BARD1, CCNA2, CCNB1, CCNE2, CDK1, CDKN1C, FEN1, GSK3B, H2AFX, PRKCD, RPS27, and TAF10) are involved in the DNA damage and repair, according to IPA, Gene Ontology, Reactome, and Hallmark databases. Genes are clock-wise alphabetically sorted by the gene symbol.
Gene coregulation in HDV-associated HCC
Gene co-regulation was visualized using the t-SNE (22) applied to all 547 genes differentially expressed in HDV MH. The t-SNE plot showed a prominent clustering of 37 genes involved in cell proliferation and DNA damage/repair (Fig. 5). Interestingly, the cluster included BRCA1, BARD1, and MSH2, which were already found in the 6 related pathways, and HMMR, an oncogene assigned by IPA to a different top-scored pathway (glioma invasiveness signaling, see Supplementary Fig. S3), but also implicated in the progression of several tumors by promoting genomic instability in association with BRCA1 and BARD1 (29). It is worth noting that all 37 clustered genes were upregulated. The comparison with HDV-NMH showed that 11 genes, including BRCA1, BARD1, TIMELESS, DSCC1, and MSH2, were not clustered in NMH (Fig. 5, top inset). This suggests that, in addition to overexpression, coregulation of this subset of genes is a specific trait of HDV-MH. Next, the coregulation of these 37 genes was also investigated in HBV-MH and HCV-MH, although 4 important genes, BRCA1, BARD1, TIMELESS, and DSCC1, were not differentially expressed in HBV-MH, and BRCA1 was not differentially expressed in HCV-MH. The t-SNE plots showed that the 37 genes were less intensely clustered in both HBV-MH and HCV-MH (Supplementary Fig. S4). This suggests that genes involved in the maintenance of genome stability, DNA replication, and repair are more upregulated and coregulated in HDV-MH compared with HBV- and HCV-MH. A possible interpretation of these findings is that in HBV- and HCV-MH, these genes may be involved in different functions, not strictly associated to the functions identified for HDV-MH. This hypothesis is in agreement with the differences observed at the level of molecular pathways.
Figure 5.
t-SNE plots of the 547 genes differentially expressed in MH of HDV-associated HCC. Each gene is surrounded by genes that have similar expression, so that clusters represent groups of coregulated genes. Genes involved in the cell cycle, found in Gene Ontology (GO), are indicated by red and green points, with color tones representing the fold change. HMMR (alias RHAMM) and MLF1IP (alias CENPU) were not associated with the cell cycle in GO; however, they were included in the plot in view of their relationships with the mitotic spindle and BRCA1/BARD1 complex (3). Thirty-seven genes, about 50% of all genes associated with the cell cycle, all upregulated, form a compact cluster (enlarged in the circular inset) at the periphery of the whole cloud of points. This means that these upregulated genes are strongly coregulated, and their coregulation is independent of the coregulation of other genes. On the right, the t-SNE plot of MH is compared with that of NMH. In NMH, the same subset of genes involved in the cell cycle is conventionally identified by the same colors corresponding to the fold changes calculated for MH genes. In NMH, some genes tend to aggregate, but the clusters (encircled by the ovals) are much more scattered than among MH genes. The contrast between the two patterns is also denoted by the different location of BRCA1, BARD1, and MSH2 in the two plots.
Among the 37 genes of the HDV cluster, we also found three cancer testis antigen (CTA) genes, all closely related to the spindle-centromere complex: NUF2 (component of the NDC80 kinetochore complex), CEP55 (centrosome component), and TTK (essential for chromosome alignment at the centromere during mitosis). However, these CTA genes were upregulated in MH of all three tumors (7). A fourth CTA gene, CASC5 (required for creation of kinetochore-microtubule attachments and chromosome segregation), not comprised in the cluster, was upregulated in HDV- but not in HBV-MH nor in HCV-MH. To the best of our knowledge, this is the first time that CASC5 is found to be associated with human HCC.
None of the 5 genes (HN1, RAN, RAMP3, KRT19, and TAF9) proposed by Nault and colleagues (30) to predict patient survival after HCC resection was found to be differentially expressed in MH. However, no HCC case was related to HDV infection in that study. This finding further confirmed the presence of specific traits in the gene expression profile of HDV-associated HCC.
Twenty-five paired liver biopsies obtained from both the tumor and the surrounding nontumorous tissue of all patients with viral-associated HCC were stained for BRCA1 and H2AFX. A very weak nuclear BRCA1 positivity was found only in 1 (4%) tumor sample and in 5 (20%) nontumorous samples. Similar data were obtained for H2AFX, which showed only a weak nuclear positivity in 7 (28%) tumor samples and in 2 (8%) nontumorous samples. No significant differences were found between HDV, HBV, and HCV HCC samples.
Genes identified in WLT but not in MH (LCM)
We also investigated the genes that were expressed in WLT, but not in microdissected malignant hepatocytes, which we referred to as “WLT-unique” genes. The WLT-unique genes in HDV-HCC were nearly 50% of the total number of WLT genes (Supplementary Fig. S5A). The 20 top-scored pathways of WLT-unique genes were mainly associated with biosynthetic pathways (spermine, spermidine, inosine-5′-phosphate, serine, and glycine) with 100% of the genes upregulated, as well as with IFN signaling and degrading processes (GAGs, glycerol, ceramide), with genes prevalently downregulated (Fig. 3D). Conversely, the WLT-unique genes of HBV-HCC were only 26% of the total number of WLT genes (Supplementary Fig. S5B). Among the 20 top-scored pathways identified, only one (the superpathway of serine and glycine biosynthesis I) was in common with the pathways of HDV WLT-unique genes (Figs. 3D and 2E; Supplementary Fig. S6). However, in HDV-HCC the genes of this pathway were all upregulated, whereas in HBV the genes were all downregulated. Other pathways identified in HBVWLT-unique genes were involved in amino acid biosynthesis and degradation, LXR/RXR activation, extrinsic and intrinsic prothrombin activation, coagulation system, acute phase response signaling and amino acid (arginine, glutamate) biosynthesis, all with 100% of genes downregulated. Only two biosynthetic pathways showed a minority of upregulated genes. The WLT-unique genes of HCV-HCC were 62% of the total number of WLT genes (Supplementary Fig. S5C). The 20 top-scored pathways (Fig. 3F; Supplementary Fig. S6) included 12 pathways associated with cytokine signaling and immune response, with the vast majority of genes downregulated. These data further confirmed the differences in gene expression profiling between HDV-, HBV-, and HCV-associated HCC.
Discussion
In this study, we characterized the molecular signature of HDV-associated HCC using an integrated analysis of microarray data obtained from LCM-isolated MH and NMH, and from WLT samples. By LCM and WLT, we identified a number of genes with strongly correlated fold changes (R2 = 0.97), indicating that a significant proportion of genes expressed by MH may also be identified in WLT, despite the presence of a heterogeneous cell population within the liver tissue. On the other hand, about two thirds of MH genes were not identified in WLT samples, and this underlines the importance of using LCM to unravel genes associated with malignant cells. The different origin of genes identified in LCM (attributable to MH) and genes uniquely identified in WLT (attributable to nonmalignant hepatocytes, Kupffer cells, stromal cells, vascular cells, cholangiocytes, stellate cells, lymphocytes, polymorphonuclear leukocytes, and macrophages) is supported by critical differences between the functional profiles of LCM vs. WLT. Four independent analyses of MH genes (canonical pathways, biological functions, relational networks and coregulation clusters) indicated that the proliferation of tumoral cells is associated with dysregulation of genes that control genome instability, which modulates the mechanisms that recognize DNA damage, prevent cells with damaged DNA from entering mitosis or promote DNA repair. Several pathways (GADD45; DNA damage-induced 14–3-3σ; cell cycle: G2–M DNA damage checkpoint regulation; hereditary breast cancer) and in particular 5 genes (BRCA1, BARD1, HMMR, MSH2, and RMI1) were significantly involved in these crucial processes. BRCA1 and BARD1 form a complex that is required for arresting cells in G1/S following DNA damage (31). HMMR (alias RHAMM) is upregulated in many human tumors and is associated with BRCA1 and BARD1. Such association dysregulates the HMMR control of the normal mitotic spindle, which may promote tumor progression (29). BRCA1 also interacts with MSH2 to participate in the formation of the BASC complex, which recognizes and repairs damaged DNA structures (32). RMI1 (alias BLAPT75) is involved in the control of genome instability in association with TOP3 topoisomerase III (33). We also found that a large majority (88%) of genes in the BRCA1 network were differentially expressed in MH. A comparison with HBV-associated HCC, using data from a previous study (7), showed that BRCA1 and BARD1 were not differentially expressed in HBV-MH. Likewise, BRCA1 was not differentially expressed in MH of HCV-associated HCC. However, upregulation was not the only feature of genes associated with genomic instability in HDV-HCC. A second important feature was the coregulation of a large cluster of 37 genes, including BRCA1, BARD1 HMMR, and MSH2, associated with the cell cycle. Conversely, HBV-MH and HCV-MH exhibited a lower level of coregulation of genes associated with cell cycle.
A 25-gene signature associated with chromosome instability (CIN) has been identified by Carter (34) in various tumors, predominantly breast and ovarian carcinomas, and recently also in human HCC (28). The CIN25 signature was associated with enhanced tumor progression, early tumor recurrence, and poor survival of patients with HCC (28), but the etiology of HCC cases was not indicated. We found that 9 of the 25 CIN genes (MELK, TOP2A, PRC1, KIF20A, CDK1, TTK, FEN1, TRIP13, NCAPD2) are differentially expressed in HDV-HCC. Moreover, all these genes, with the only exception of NCAPD2, were present in the cluster of coregulated genes that characterizes and differentiates HDV-HCC from HBV- and HCV-HCC. Surprisingly, BRCA1 was not included in the CIN25 signature, although the general consensus that BRCA1 is associated with genomic instability (35, 36), as also reflected by its full name “BRCA1, DNA Repair Associated.” The absence of BRCA1 from the CIN25 signature cannot be attributed to lack of studies on this topic, because at the time the CIN25 signature was formulated, in 2006, over 2,000 reports on BRCA1 had already been published (37). This suggests that the genomic instability resulting from the BRCA1 dysregulation does not result in detectable chromosomal aberrations.
In contrast to the increased BRCA1 gene expression observed in HDV-HCC compared with both HBV- and HCV-HCC, the expression of BRCA1 by IHC was negative in all but one HCC sample. These negative results are consistent with a previous report in which the expression of BRCA1 in breast cancer progressively decreased from low-grade to high-grade breast carcinoma to eventually become undetectable (38). In our series, the HCC stage was G3 in 4 of 5 patients.
Insights into the levels of viral replication within the tumor of patients with viral-associated HCC may shed light on the role of hepatitis viruses in hepatocarcinogenesis. Although there is a strong epidemiologic link between chronic infection with hepatitis viruses and the risk of developing HCC (5), the mechanisms whereby these viruses promote hepatic carcinogenesis remain elusive (3). Thus, the question remains whether HBV, HCV, and HDV elicit liver cancer indirectly, through chronic inflammation, fibrosis, and liver regeneration, or directly, through the expression of tumor-promoting viral genes. Our collection of multiple liver specimens from patients with HDV-HCC allowed us to investigate the levels of HDV replication in multiple compartments of individual livers containing HCC and compared with the levels measured in control livers with non-HCC HDV cirrhosis. Analysis of liver specimens demonstrated that the level of HDV replication in the tumors tended to be lower than in the surrounding nontumorous tissues, and markedly lower than in non-HCC HDV cirrhotic livers. The levels of HBV DNA were extremely low or undetectable both in serum and in liver of all patients with HDV-HCC, as well as of non-HCC HDV cirrhotic livers, in agreement with the low HBV replication levels that are typical of chronic HDV disease, as reported previously (39). Accordingly, HBcAg was not detected in any tumor and nontumorous tissues, as well as in all but one patient with non-HCC HDV cirrhosis. The fact that HDV does not replicate well in some patients is consistent with the dramatic reduction in viral replication that we recently documented in the tumor of patients with HCV-associated HCC, where a sharp and significant drop in HCV RNA levels was observed in all of the patients with HCC when perilesional tissue was compared with tissue inside the tumor margin (40). Collectively, our data indicate that hepatitis viruses do not grow well in malignant hepatocytes in vivo, in line with the inability or limited efficiency of primary isolates to grow in hepatoma cell lines in vitro, suggesting that malignant hepatocytes express or lack factors that are critical for viral replication.
Although the number of patients that could be included in this intensive study was limited, our patients were well characterized and we emphasize the difficulties in obtaining liver samples from patients with HDV-associated HCC. As a consequence, there are no studies published to date in which the molecular profile of this type of cancer has been investigated. Despite the limited number of patients studied, however, our findings were highly consistent in all patients studied, corroborating the conclusions of our study. Moreover, comparative analysis extended to all three different viral hepatitis-associated HCC provided the first evidence that the molecular signature of HDV-HCC is distinct not only from HBV-HCC, despite the obligatory dependence of HDV on HBV, but also from HCV-associated HCC.
In conclusion, by performing an integrated comparative analysis of transcriptomics of LCM hepatocytes and WLT, our study illustrates the critical role of LCM in identifying genes that may play a role in the molecular pathogenesis of liver cancer. The activation and coregulation of genes associated with DNA damage and repair point to genetic instability as a key mechanism of HDV-induced hepatocarcinogenesis. By comparing HCC-associated with HDV versus that associated with HBV alone, a major highlight of this study is that distinct molecular mechanisms appear to be involved in HDV versus HBV hepatocarcinogenesis, even though patients with HDV-HCC also harbor HBV. Finally, the study of multiple liver specimens from both the tumor and the surrounding nontumor tissue may help to identify genes that are important for elucidating the molecular pathogenesis of HDV-associated HCC, as well as for identifying new diagnostic markers that may predict the development of HCC or permit an early diagnosis.
Supplementary Material
Acknowledgments
This research was supported by the Intramural Research Program of the NIH, National Institute of Allergy and Infectious Diseases, and National Cancer Institute.
Footnotes
Disclosure of Potential Conflicts of Interest
J. Rodriguez-Canales is a senior pathologist at MedImmune. No potential conflicts of interest were disclosed by the other authors.
Supplementary data for this article are available at Molecular Cancer Research Online (http://mcr.aacrjournals.org/).
References
- 1.Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer statistics. CA Cancer J Clin 2011;61:69–90. [DOI] [PubMed] [Google Scholar]
- 2.Levrero M.Viral hepatitis and liver cancer: the case of hepatitis C. Oncogene 2006;25:3834–47. [DOI] [PubMed] [Google Scholar]
- 3.McGivern DR, Lemon SM. Virus-specific mechanisms of carcinogenesis in hepatitis C virus associated liver cancer. Oncogene 2011;30:1969–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Centers for Disease Control and Prevention. Hepatocellular carcinoma - United States, 2001–2006. MMWR Morb Mortal Wkly Rep 2010;59: 517–20. [PubMed] [Google Scholar]
- 5.El-Serag HB. Hepatocellular carcinoma. N Engl J Med 2011;365:1118–27. [DOI] [PubMed] [Google Scholar]
- 6.Schena M, Shalon D, Davis RW, Brown PO. Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 1995;270:467–70. [DOI] [PubMed] [Google Scholar]
- 7.Melis M, Diaz G, Kleiner DE, Zamboni F, Kabat J, Lai J, et al. Viral expression and molecular profiling in liver tissue versus microdissected hepatocytes in hepatitis B virus-associated hepatocellular carcinoma. J Transl Med 2014; 12:230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Shirota Y, Kaneko S, Honda M, Kawai HF, Kobayashi K. Identification of differentially expressed genes in hepatocellular carcinoma with cDNA microarrays. Hepatology 2001;33:832–40. [DOI] [PubMed] [Google Scholar]
- 9.Ueda T, Honda M, Horimoto K, Aburatani S, Saito S, Yamashita T, et al. Gene expression profiling of hepatitis B- and hepatitis C-related hepatocellular carcinoma using graphical Gaussian modeling. Genomics 2013;101:238–48. [DOI] [PubMed] [Google Scholar]
- 10.Taylor JM, Purcell RH, Farci P. Hepatitis D (Delta) virus. In:Knipe DM, MaH PM, editors. Fields virology. Philadelphia, PA: Lippincott Williams & Wilkins; 2013. [Google Scholar]
- 11.Rizzetto M, Verme G, Recchia S, Bonino F, Farci P, Arico S, et al. Chronic hepatitis in carriers of hepatitis B surface antigen, with intrahepatic expression of the delta antigen. An active and progressive disease unresponsive to immunosuppressive treatment. Ann Intern Med 1983;98:437–41. [DOI] [PubMed] [Google Scholar]
- 12.Niro GA, Smedile A, Ippolito AM, Ciancio A, Fontana R, Olivero A, et al. Outcome of chronic delta hepatitis in Italy: a long-term cohort study. J Hepatol 2010;53:834–40. [DOI] [PubMed] [Google Scholar]
- 13.Romeo R, Del Ninno E, Rumi M, Russo A, Sangiovanni A, de Franchis R, et al. A 28-year study of the course of hepatitis Delta infection: a risk factor for cirrhosis and hepatocellular carcinoma. Gastroenterology 2009;136: 1629–38. [DOI] [PubMed] [Google Scholar]
- 14.Ishak K, Baptista A, Bianchi L, Callea F, De Groote J, Gudat F, et al. Histological grading and staging of chronic hepatitis. J Hepatol 1995;22:696–9. [DOI] [PubMed] [Google Scholar]
- 15.Edmondson HA, Steiner PE. Primary carcinoma of the liver: a study of 100 cases among 48,900 necropsies. Cancer 1954;7:462–503. [DOI] [PubMed] [Google Scholar]
- 16.Farci P, Roskams T, Chessa L, Peddis G, Mazzoleni AP, Scioscia R, et al. Long-term benefit of interferon alpha therapy of chronic hepatitis D: regression of advanced hepatic fibrosis. Gastroenterology 2004;126: 1740–9. [DOI] [PubMed] [Google Scholar]
- 17.Engle RE, Bukh J, Alter HJ, Emerson SU, Trenbeath JL, Nguyen HT, et al. Transfusion-associated hepatitis before the screening of blood for hepatitis risk factors. Transfusion 2014;54:2833–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Erickson HS, Albert PS, Gillespie JW, Rodriguez-Canales J, Marston Linehan W, Pinto PA, et al. Quantitative RT-PCR gene expression analysis of laser microdissected tissue samples. Nat Protoc 2009;4: 902–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Farci P, Diaz G, Chen Z, Govindarajan S, Tice A, Agulto L, et al. B cell gene signature with massive intrahepatic production of antibodies to hepatitis B core antigen in hepatitis B virus-associated acute liver failure. Proc Natl Acad Sci U S A 2010;107:8766–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Wang E, Miller LD, Ohnmacht GA, Liu ET, Marincola FM. High-fidelity mRNA amplification for gene profiling. Nat Biotechnol 2000;18: 457–9. [DOI] [PubMed] [Google Scholar]
- 21.Simon R, Lam A, Li MC, Ngan M, Menenzes S, Zhao Y. Analysis of gene expression data using BRB-ArrayTools. Cancer Inform 2007;3:11–7. [PMC free article] [PubMed] [Google Scholar]
- 22.van der Maaten L, Hinton G. Visualizing data using t-SNE. J Mach Learn Res 2008;9:2579–605. [Google Scholar]
- 23.Bushati N, Smith J, Briscoe J, Watkins C. An intuitive graphical visualization technique for the interrogation of transcriptome data. Nucleic Acids Res 2011;39:7380–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Grun D, Lyubimova A, Kester L, Wiebrands K, Basak O, Sasaki N, et al. Single-cell messenger RNA sequencing reveals rare intestinal cell types. Nature 2015;525:251–5. [DOI] [PubMed] [Google Scholar]
- 25.RCoreTeam. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2014. [Google Scholar]
- 26.Huang S. Regulation of metastases by signal transducer and activator of transcription 3 signaling pathway: clinical implications. Clin Cancer Res 2007;13:1362–6. [DOI] [PubMed] [Google Scholar]
- 27.Carloni V, Lulli M, Madiai S, Mello T, Hall A, Luong TV, et al. CHK2 overexpression and mislocalisation within mitotic structures enhances chromosomal instability and hepatocellular carcinoma progression. Gut 2018;67:348–61. [DOI] [PubMed] [Google Scholar]
- 28.Weiler SME, Pinna F, Wolf T, Lutz T, Geldiyev A, Sticht C, et al. Induction of chromosome instability by activation of yes-associated protein and Forkhead box M1 in liver cancer. Gastroenterology 2017;152:2037–51e22. [DOI] [PubMed] [Google Scholar]
- 29.Maxwell CA, McCarthy J, Turley E. Cell-surface and mitotic-spindle RHAMM: moonlighting or dual oncogenic functions? J Cell Sci 2008; 121:925–32. [DOI] [PubMed] [Google Scholar]
- 30.Nault JC, De Reynies A, Villanueva A, Calderaro J, Rebouissou S, Couchy G, et al. A hepatocellular carcinoma 5-gene score associated with survival of patients after liver resection. Gastroenterology 2013;145:176–87. [DOI] [PubMed] [Google Scholar]
- 31.Fabbro M, Savage K, Hobson K, Deans AJ, Powell SN, McArthur GA, et al. BRCA1-BARD1 complexes are required for p53Ser-15 phosphorylation and a G1/S arrest following ionizing radiation-induced DNA damage. J Biol Chem 2004;279:31251–8. [DOI] [PubMed] [Google Scholar]
- 32.Wang Y, Cortez D, Yazdi P, Neff N, Elledge SJ, Qin J. BASC, a super complex of BRCA1-associated proteins involved in the recognition and repair of aberrant DNA structures. Genes Dev 2000;14:927–39. [PMC free article] [PubMed] [Google Scholar]
- 33.Raynard S, Bussen W, Sung P. A double Holliday junction dissolvasome comprising BLM, topoisomerase IIIalpha, and BLAP75. J Biol Chem 2006;281:13861–4. [DOI] [PubMed] [Google Scholar]
- 34.Carter SL, Eklund AC, Kohane IS, Harris LN, Szallasi Z. A signature of chromosomal instability inferred from gene expression profiles predicts clinical outcome in multiple human cancers. Nat Genet 2006;38:1043–8. [DOI] [PubMed] [Google Scholar]
- 35.Jang ER, Lee JS. DNA damage response mediated through BRCA1. Cancer Res Treat 2004;36:214–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Negrini S, Gorgoulis VG, Halazonetis TD. Genomic instability–an evolving hallmark of cancer. Nat Rev Mol Cell Biol 2010;11:220–8. [DOI] [PubMed] [Google Scholar]
- 37.PubMed. National Center for Biotechnology Information. Bethesda, MD. Available from: www.ncbi.nlm.nih.gov/pubmed. [Google Scholar]
- 38.Wilson CA, Ramos L, Villasenor MR, Anders KH, Press MF, Clarke K, et al. Localization of human BRCA1 and its loss in high-grade, non-inherited breast carcinomas. Nat Genet 1999;21:236–40. [DOI] [PubMed] [Google Scholar]
- 39.Farci P, Niro GA. Clinical features of hepatitis D. Semin Liver Dis 2012;32:228–36. [DOI] [PubMed] [Google Scholar]
- 40.Harouaka D, Engle RE, Wollenberg K, Diaz G, Tice AB, Zamboni F, et al. Diminished viral replication and compartmentalization of hepatitis C virus in hepatocellular carcinoma tissue. Proc Natl Acad Sci U S A 2016; 113:1375–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.