A search for patients diagnosed with SHCC at three institutions [Kansai Medical University Hospital (Hirakata, Japan), Osaka Metropolitan University Hospital (Osaka, Japan) and Kindai University Hospital (Osakasayama, Japan)] over ~25 years (dating back from 2019) was conducted at the time of ethical approval. Subsequently, patients diagnosed with SHCC at the three institutions between 2006 and 2015 were retrospectively selected based on their medical records. This was due to specimens collected before 2006 not being preserved at the hospitals or being too old for use. In addition, no SHCC cases were identified between 2016 and 2019. From this pool, formalin-fixed, paraffin-embedded (FFPE) tumor tissues that had been collected and stored for diagnostic purposes at the time of surgery were obtained from 7 patients with SHCC who underwent surgical resection at Kansai Medical University Hospital (n=4), Osaka Metropolitan University Hospital (n=1) and Kindai University Hospital (n=2) between March, 2006 and December, 2015. The inclusion criteria for patient selection in the present study were as follows: Patients with HCC who underwent surgical resection and had carcinomatoid and sarcomatoid tumor components as confirmed by pathological diagnosis following tumor resection. No specific exclusion criteria were applied in the present study, except for the absence of an adequate amount of FFPE specimen for analysis. All patients provided written informed consent to participate in the present study and for tumor tissue sample collection for analysis. SHCC was diagnosed during the postoperative pathological examination by pathologists at each institution. SHCC specimens were further analyzed by the same pathologist during the present study. The present study was conducted in accordance with the Declaration of Helsinki and the Ethical Guidelines for Medical and Health Research Involving Human Subjects in Japan (14). The Institutional Ethics Review Boards of Kansai Medical University Hospital (approval no. 2019074), Osaka Metropolitan University Hospital (approval no. 2020-185) and Kindai University Faculty of Medicine (approval no. 31-201) approved the present study.

The FFPE samples were subjected to RNA extraction using the Allprep DNA/RNA FFPE Kit (cat. no. 80234; Qiagen, Inc.) according to the manufacturer's instructions. The quality and quantity of the RNA were determined with NanoDrop 2000 device (Thermo Fisher Scientific, Waltham, MA) and RiboGreen dsDNA Assay Kit (cat. no. R11490; Thermo Fisher Scientific). The AmpliSeq Transcriptome Human Gene Expression Kit (cat. no. A26325; Thermo Fisher Scientific, Inc.) was subsequently used for whole transcriptome analysis according to the manufacturer's instructions. For library preparation, barcoded cDNA libraries were generated from 10 ng total RNA using the SuperScript VILO cDNA Synthesis Kit (cat. no. 11754050; Thermo Fisher Scientific, Inc.). The cDNA was then amplified for 16 cycles following addition of the PCR master mix and AmpliSeq human transcriptome gene expression primer pool (Thermo Fisher Scientific, Inc.). After multiplex PCR, IonXpress Barcode Adapters (cat. no. 4474517; Thermo Fisher Scientific, Inc.) were ligated to the PCR products and purified using Agencourt AMPure XP beads (cat. no. A63881; Beckman Coulter, Inc.). The purified libraries were quantified using the Ion Library TaqMan Quantitation Kit (cat. no. 4468802; Thermo Fisher Scientific, Inc.) and the adjusted to 50 pM with low TE buffer (10 mM Tris-HCl, pH 8.0, 0.1 mM EDTA), and then pooled and sequenced using an Ion Torrent S5 system (Thermo Fisher Scientific, Inc.) and an Ion 550 Chip Kit (cat. no. A34538; Thermo Fisher Scientific, Inc.). The Ion Torrent S5 system produces single-end short-reads with 120-bp read length. Base calling, alignment to the human reference genome (hg19) and quality control were performed using Ion Torrent Suite v5.12 software (Thermo Fisher Scientific, Inc.). The raw reads were analyzed using the AmpliSeqRNA plugin to generate gene-level expression values for all 20,802 human reference sequence (Seq) genes.

Data were processed with Transcriptome Analysis Console (ver. 4.0.3; Thermo Fisher Scientific, Inc.) to determine the differentially expressed genes (DEGs). DEGs were selected using |fold change| >2 and P<0.05 as the cut-off. Hierarchical clustering of these genes was performed using the mean linkage method with 1-Pearson correlation coefficient as the distance measure using Morpheus (software.broadinstitute.org/morpheus/). Functional and pathway enrichment analysis was performed using Metascape (ver. 3.5; accessed on 21 Feb 2023) (15).

FFPE sections of 4-µm thickness were stained for polybromo 1 (PBRM1) using a validated and published IHC method (16,17). Briefly, FFPE tissue samples were sectioned and placed on positively charged slides. Sections were deparaffinized, hydrated and pretreated in DAKO target retrieval solution (cat. no. S1699; Agilent Technologies, Inc.) for 20 min in a steamer. The VECTASTAIN ABC-HRP Kit, Peroxidase (Rabbit IgG) (cat. no. PK-4001; Vector Laboratories, Inc.) was used and includes the blocking serum and the 2nd HP antibody. The procedure was carried out according to the manufacturer's instructions. The slides were blocked in goat blocking serum for 30 min at room temperature, washed in phosphate buffered saline (PBS) and then incubated with rabbit anti-PBRM1 monoclonal antibody (1:50; cat. no. 38439; Cell Signaling Technology, Inc.) overnight at 4°C. The slides were then washed in PBS and incubated with the supplied biotinylated anti-rabbit secondary antibody for 30 min at room temperature. The slides were washed in PBS and incubated with the supplied VECTASTAIN Elite ABC Reagent for 30 min at room temperature. The slides were washed in PBS and developed in DAB (cat. no. 8801-4965-72, Invitrogen; Thermo Fisher Scientific, Inc.) and counterstained with hematoxylin for 2 min at room temperature. Assessment of immunohistochemical staining was performed on scanned sections captured at magnification, ×10 using the Keyence BZ-X810 All-in-One light microscope. H scores were calculated for PBRM1 positivity in the sarcomatoid and carcinomatoid regions using QuPath v0.2.0-m4 image analysis software (https://qupath.github.io/) (18,19).

PBRM1 gene expression was compared by unpaired t test. Protein expression was compared using the Wilcoxon signed-rank test. All statistical analyses were performed using GraphPad Prism software (version 8.4; Dotmatics). P<0.05 was considered to indicate a statistically significant difference.

A total of 7 tumor specimens, of which 5 contained sarcomatoid components and 7 contained carcinoma components, were obtained from 7 patients (Table I). Representative images of hematoxylin and eosin-stained sections showing histological differences between the sarcomatoid and carcinomatoid components in SHCC are presented in Fig. 1. Following whole transcriptome analysis of the samples, 336 genes were found to be differentially expressed between the sarcomatoid and carcinomatoid components. The top 10 downregulated and top 10 upregulated differentially expressed genes are shown in Table II. The sarcomatoid components had increased ZRANB1 (zinc finger RANBP2-type containing 1), IGF2BP1 (Insulin-like growth factor 2 mRNA binding protein 1), DHRSX (dehydrogenase/reductase X-linked), TMEM19 (Transmembrane protein 19), DNAJC9 (DnaJ homolog subfamily C member 9), UBA7 (Ubiquitin-activating enzyme 7), OR10K2 (Olfactory receptor family 10 subfamily K member 2), DEPDC1B (DEP domain containing 1B), ATP10D (ATPase phospholipid transporting 10D (putative)) and HSPA7 (Heat shock protein family A (Hsp70) member 7 (pseudogene)) expression levels, and decreased ADRA2B (Adrenoceptor alpha 2B), NHLRC1 (NHL repeat containing E3 ubiquitin protein ligase 1), PBRM1, DHCR7 (7-dehydrocholesterol reductase), SPATA7 (Spermatogenesis associated 7), WRB (Tryptophan-rich basic protein), PAFAH1B1 (Platelet activating factor acetylhydrolase 1B regulatory subunit 1), PET112, (Glutamyl-tRNA amidotransferase subunit B), GDA (Guanine deaminase) and MRPS5 (Mitochondrial ribosomal protein S5) expression levels. Moreover, visualization of the differentially expressed genes using a volcano plot indicated that decreased expression of PBRM1 in the sarcoma component of SHCC tissues may serve as a biologically relevant marker of the sarcoma subtype (Fig. 2).

Next, the biological relevance of genes differentially expressed between the sarcomatoid and carcinomatoid components of SHCC was investigated. Subsequently, 336 genes were identified as differentially expressed genes between the sarcomatoid and carcinomatoid components of the tumors. Unsupervised hierarchical clustering analysis of the 336 differentially expressed genes revealed two distinct clusters, clusters I and II, which showed a strong association to the sarcomatoid and carcinomatoid components, respectively (Fig. 3A). To determine the biological relevance of the genes downregulated in the sarcomatoid components, a gene ontology analysis was performed using Metascape, and biological pathways enriched in genes from cluster II were identified. The pathways enriched in this gene cluster included ‘histone methylation’ and ‘mitotic DNA replication checkpoint signaling’, which indicated that these processes were disrupted in the sarcomatoid components (Fig. 3B). These results supported our hypothesis that impaired function of PBRM1 may be associated with changes resulting in sarcomatoid differentiation.

The aforementioned findings of the present study indicated the presence of altered PBRM1 expression in SHCC, which indicated that PBRM1 may serve as a marker of sarcomatoid differentiation. To evaluate the potential of PBRM1 as a marker, PBRM1 protein expression levels in sarcomatoid and carcinomatoid components of SHCC tissues were examined by quantitative IHC. The tumors were categorized as PBRM1⁺ when the cells had strong diffuse nuclear staining, and as PBRM1⁻ when the cells had an absence of diffuse nuclear staining (Fig. 4A) (19). Quantitative analysis demonstrated that the PBRM1 protein expression level was markedly lower was lower in the sarcomatoid than the carcinomatoid component, although the difference was not statistically significant (P=0.0781; Fig. 4B). Quantitative analysis revealed a noticeable reduction in PBRM1 protein expression levels within the SHCC, although statistical significance was not achieved (P=0.0781; Fig. 4B). This discovery implies a potential link between diminished PBRM1 protein expression and sarcomatoid tumors, offering preliminary insights into the biological role of PBRM1 in these tumors. Nonetheless, further investigations employing larger datasets are imperative to ascertain the utility of PBRM1 in IHC as a biomarker for distinguishing sarcomatoid components.