Analysis of tumor tissue samples from patients with HB (n=47) who were all part of the German Cooperative Pediatric Liver Tumor Registry Study HB99 and its subsequent Register for Pediatric Liver Tumors was performed. The two registries were multicentric and were initiated by the German Society for Pediatric Oncology and Hematology. They were open to registration for patients from Germany, Austria and Switzerland up to the age of 20 years with untreated HB. The registry protocols were assigned by the institutional Ethical Committee of the University Children's Hospital Basel (Basel, Switzerland) and the University of Bonn (Bonn, Germany), and written consent was obtained from the parents for treatment, data collection and analysis.

Clinical information, including demographic, therapeutic, tumor and clinical outcome variables, were retrieved from the two clinical studies. The treatment protocol consisted of preoperative chemotherapy followed by delayed surgery and postoperative chemotherapy according to two risk groups (standard- versus high-risk). The two risk groups were based on the International Childhood Liver Tumor Strategy Group risk criteria (23). Standard risk patients received two or three courses of neoadjuvant ifosfamide, cisplatin and doxorubicin (IPA) chemotherapy prior to surgery (1 g/m² ifosfamide every 72 h, days 1–3; 20 mg/m² cisplatin every 1 h, days 4–8; and 60 mg/m² doxorubicin every 48 h, days 9–10). Radical surgery was conducted after the second or third course depending on the resectability. Postoperatively, another course of IPA was applied. In case of microscopically incomplete resection, two adjuvant courses of IPA were administered. Patients with small-extended tumors (PRETEXT stage I) (24) could be resected without neoadjuvant chemotherapy and were treated with two courses of IPA postoperatively. High-risk patients received up to seven courses of carboplatin-based chemotherapy preoperatively depending on tumor shrinkage and resectability. Patients were initially treated with two courses of carbo/VP16 (200 mg/m² carboplatin every 24 h, days 1–4; and 100 mg/m² etoposide every 24 h, days 1–4) followed by stem cell collection. In case of tumor response, the therapy was continued with high dose carboplatin and etoposide (500 mg/m² per 24 h, days 8–5) following autologous stem cell transplantation. Patients without tumor response were treated with IPA. Resection was scheduled as soon as the tumor was determined to be completely resectable. In case of persisting non-resectability, liver transplantation was recommended. Lung metastases were resected if residual metastases were still observed on radiological images after chemotherapy.

Tumor specimens were reviewed by the local institution as well as the Institute of Pediatric Pathology, University of Kiel (Kiel, Germany), which served as a reference center. Matched adjacent liver tissue samples from the surgical specimens without macroscopic or microscopic tumors served as tumor-free controls (n=9). Clinical and molecular data, such as gender, age at diagnosis, PRETEXT staging (24), including vascular invasion and multifocality, metastatic disease, histology, CTNNB1 mutation, 16-gene signature and overall survival, were retrieved from the HB99 database and our recent exome sequencing study, respectively (25).

RNA extraction, complementary (c)DNA synthesis and quantitative polymerase chain reaction (qPCR) analysis were performed as previously described (26). Briefly, total RNA was isolated from all the samples using TRIzol® reagent (Invitrogen Life Technologies, Carlsbad, CA, USA) and dissolved in RNase-free water. The purity and quality of the RNA was checked using a Nanodrop® 2000 spectrophotometer (Thermo Fisher Scientific, Inc., Waltham, MA, USA). RNA (2 µg) was reverse transcribed using SuperScript™ II reverse transcriptase (Invitrogen Life Technologies), according to the manufacturer's instructions. The amplification reactions were performed with 40 ng complementary DNA, 500 nM forward and reverse primers and iTaq SYBR®-Green Supermix (Bio-Rad Laboratories, Hercules, CA, USA) in a final reaction volume of 20 µl and were incubated at 95°C for 7 min, subjected to 40 cycles of 95°C for 30 sec, 60°C for 30 sec and 72°C for 30 sec, followed by a final extension cycle at 72°C for 7 min. All the reactions were conducted on ice to minimize the risk of RNA degradation. cDNA obtained was stored at −80°C.

According to the modified method of Bigioni et al, the prepared cDNA (2 µl) was used in a PCR with specific primers, based on the common sequence of the TACR1 (NK1R) human isoforms, which yield a 186-bp fragment (27). Specific primers were as follows: Forward, 5′-AACCCCATCATCTACTGCTGC-3′ and reverse, 5′-ATTTCCAGCCCCTCATAGTCG-3′ for fl-TACR1 (NM_001058.3); forward, 5′-GGGCCACAAGACCATCTACA-3′ and reverse, 5′-AAGTTAGCTGCAGTCCCCAC-3′ for tr-TACR1 (NM_015727.2); and forward, 5′-GCCCGAAACGCCGAATAT-3′ and reverse, 5′-CCGTGGTTCGTGGCTCTCT-3′ for the TBP housekeeping gene. The amplification reactions were performed with iTaq SYBR®-Green Supermix (Bio-Rad Laboratories, Hercules, CA, USA) in a final reaction volume of 20 µl and were incubated at 95°C for 7 min, subjected to 40 cycles of 95°C for 30 sec, 60°C for 30 sec and 72°C for 30 sec, followed by a final extension cycle at 72°C for 7 min. PCR was performed using a Mastercycler ep Gradient S (Eppendorf, Hamburg, Germany) and the transcript numbers were normalized according to the expression of the housekeeping gene. Relative quantification of gene expression was performed using the 2^−∆∆Ct method, as described by Pfaffl (28).

Data are presented as mean ± SD. Mean and individual relative expression values of tumor and control samples are expressed in dot plots for each group. Statistical comparisons were performed with a standard t-test and Mann-Whitney U test using GraphPad Prism biostatistics software (version 5.0d; GraphPad Software, Inc., La Jolla, CA, USA). P<0.05 and P<0.01 indicated a statistically significant difference for all the comparisons. To differentiate between a high and low expression of TACR1 and its components, 3-fold of the mean of 9 tumor-free control samples was used as a cutoff for high expression. Kaplan-Meier estimates of specific survival time in the two groups were compared using the log-rank Mantel-Cox test.

To address the aforementioned hypotheses, the gene expression pattern of fl-TACR1 and tr-TACR1 were analyzed in tumor tissue samples of HB and non-tumorous liver tissue. A significantly higher expression of tr-TACR1 was observed in HB compared with the control specimens (P=0.0301; Fig. 1A). Although not statistically significant, the expression of fl-TACR1 also tended to be higher in tumor specimens (P>0.05; Fig. 1B). These results correlated with our previous findings, in which in vitro and in vivo models of HB were used to demonstrate that tr-TACR1 is overexpressed in malignant HB cells. In turn, malignant HB cells were correlated with responsiveness to treatment with NK1R antagonists, such as aprepitant (8).

Due to the wide range of gene expression, the ratio of tr-TACR1 versus fl-TACR1 expression was presented. It was found that the ratios were comparable between the tumor and control samples (Fig. 1C), suggesting a positive correlation of the two splice variants. When analyzed in-depth, a statistically significant weak correlation was identified between tr-TACR1 and fl-TACR1 expression (r=0.3542), potentially indicating a mutual dependency (Fig. 1D).

To improve understanding of whether splice variants or their ratio correlate with the biological features of the tumor, TACR1 expression was analyzed accordingly (Table I; Figs. 2–4). First, the truncated variant was investigated, due to its significance in HB as a potential therapeutic target (8). In order to accomplish this, the relative expression of tr-TACR1 was correlated with a recently described 16-gene molecular signature known to be associated with prognosis (16). Similar to the original description of this signature, the current cohort was separated into 29 patients with HB that exhibited the C1 signature (poor prognosis; 61.7%) and 18 exhibited the C2 signature (improved prognosis; 38.3%) (Table I). Relative gene expression analysis of tr-TACR1 revealed no significant difference between patients with C1 and C2 signatures (Fig. 2A). The same features were then analyzed in correlation to the gene expression of fl-TACR1 (Fig. 3A). A significant correlation was identified between low fl-TACR1 expression and the C2 population of the 16-gene signature (P=0.0222; Fig. 4A). It is of note that 55.6% of the specimens grouped into the C2 population exhibited extremely low levels of fl-TACR1.

The ratio of fl-TACR1 to tr-TACR1 was then calculated to determine whether it could be correlated with the 16-gene signature (16). Notably, extremely low ratios were identified in the favorable C1 population and extremely high ratios were evident for the C2 population; however, this effect was not statistically significant (data not shown). Populations with a C2 signature were previously demonstrated to be associated with a poor prognosis of HB (16).

The expression patterns of fl-TACR1 and tr-TACR1 were correlated with clinical, biological and histological characteristics, including metastasis (Figs. 2B and 3B), the preoperative classification PRETEXT (Figs. 2C and 3C), vascular invasion (Figs. 2D and 3D), histology (Figs. 2E and 3E), onset period (Figs. 2F and 3F), multifocality (Figs. 2G and 3G), CTNNB1 mutations (Figs. 2H and 3H) and gender (Figs. 2I and 3I). Of the entire cohort, it was found that 63.8% exhibited no metastasis at the time of diagnosis, 82.9% had no vascular invasion and only 27.7% were multifocal. Gender was equally distributed (51.1% female vs. 48.9% male), the majority tumors had a fetal histology (74.5 vs. 25.5% embryonal) and, as expected, 70.2% of tumors possessed a β-catenin (CTNNB1) mutation. The age of diagnosis was predominantly within the first 24 months of life (68.1%) and the specimens were classified as PRETEXT 1–2 (36.2%), PRETEXT 3 (40.4%) and PRETEXT 4 (23.4%).

When analyzing tr-TACR1 expression in detail, no statistically relevant differences were identified with respect to the aforementioned clinical features. Notably, a high expression of tr-TACR1 correlated with a better overall survival (P=0.0551; Fig. 2J), although this was only a trend as it did not reach statistical significance.

Similarly, when analyzing the pattern of fl-TACR1 expression with metastasis, PRETEXT, vascular invasion, histology, age at diagnosis, multifocality, CTNNB1 mutations or gender, no significant correlation was observed (P>0.05; Fig. 3B–I). When clustered into groups of high versus low expression of fl-TACR1, overall survival curves did not deviate from each other (P>0.05; Fig. 3J), contrary to the finding for tr-TACR1 (Fig. 2J).

Use of the ratio of the two variants (tr-TACR1:fl-TACR1), revealed no statically significant differences with regard to the majority of the characteristics (Fig. 4A–I). The only exception to this was that a higher tr-TACR1:fl-TACR1 ratio, which occurred predominantly in PRETEXT 1–2 compared with PRETEXT 3 (P=0.0459; Fig. 4C). Similar to the analysis with tr-TACR1 alone, overall survival was worse with a low ratio of tr-TACR1:fl-TACR1 (P>0.05; Fig. 4J).

As the original description of the 16-gene signature by Cairo et al (16) suggested a worse prognosis for the C2 signature, the present study aimed to investigate whether either factor (tr-TACR1, fl-TACR1 or the ratio thereof) could refine the predictive value in our set of tumors. Therefore, overall survival within the C2 HB tumors was re-analyzed, and their outcome with respect to high versus low expression of TACR1 or its ratio was investigated. It was identified that low tr-TACR1 predicted a poor prognosis for C2 tumors with a higher significance than tr-TACR1 alone (P=0.0377; Fig. 2K). Although not significant, high fl-TACR1 suggested a worse outcome (P>0.05; Fig. 3K), and the ratio of the two variants had the same trend as truncated alone and as analyzed in the whole cohort, but with a clear tendency towards a worse prognosis for low tr/fl-TACR1 in the C2 group (P>0.05; Fig. 4K).

Taken together, no strong correlation of tr-TACR1, fl-TACR1 or tr-TACR1:fl-TACR1 gene expression with clinical and histological data was identified. However, low tr-TACR1 or a low ratio was associated with a worse prognosis, particularly when associated with the C2 signature. By contrast, no significance was identified in fl-TACR, with a marginal trend towards a worse outcome in C2 and high fl-TACR1 expression.