This article is mentioned in:

Introduction

Renal cell carcinoma (RCC) is a urological malignancy with increasing incidence in countries with a high Human Development Index (HDI) (i.e. with a HDI of >0.8) (1,2). Clear cell RCC (ccRCC) is the predominant histological type, representing 75–80% of all cases (3). A total of ~20% of patients are initially diagnosed with metastatic ccRCC (mRCC) and ~20% of primary localized cases become metastatic during follow-up (4,5). The management of mRCC has markedly changed in recent years with the introduction of novel therapies leading to substantial improvements in the survival and quality of life of patients (6). Antiangiogenic tyrosine kinase inhibitors (TKIs), immune checkpoint inhibitors (ICIs) and their combination represent novel systemic therapies in mRCC (7,8). The expanding horizons of systemic therapies suitable for mRCC require predictive and prognostic molecular biomarkers for the selection of optimal therapy approaches for patients in personalized medicine.

Long non-coding RNAs (lncRNAs) have emerged as one of the key regulators of gene expression in cancer, and they may exhibit tumor-suppressive or oncogenic functions based on their interacting partners. Their expression is highly tissue- and condition-specific, and they have been found to be involved in various cancer-associated processes, including tumorigenesis, progression and metastatic spread (9). Thus, lncRNAs hold promise as novel biomarkers and therapeutic targets for cancer.

Previous studies have shown several differentially expressed lncRNAs in ccRCC tissue, as well as in ccRCC cell lines, including CRNDE (10), H19 (11), HOTAIR (12), TUG1 (13), MEG3 (14) and GAS5 (15) as summarized by Li et al (16). Moreover, MALAT1 (17), PVT1 (18), LINC00152 (19) and LUCAT1 (20) have been suggested as lncRNAs associated with poor prognosis and decreased overall survival (OS) in patients with ccRCC. A potential predictive role of several lncRNAs has previously been suggested in experimental studies (21–23). The lncRNA ARSR promotes resistance to sunitinib by serving as competing endogenous RNA (22), whereas high expression of lncRNA SARCC increases its efficacy (23). In mRCC specifically, mostly protein-coding genes have been studied (24,25). Genes for transmembrane protein programmed death-ligand 1 and serine/threonine-protein kinase p21-activating kinase 1 serve a prognostic role in mRCC (26,27). Nevertheless, the association between lncRNA profiles and outcomes of patients with mRCC treated with a specific type of systemic therapy remains unclear.

The present study evaluated the expression profile of 84 cancer-associated lncRNAs in patients with mRCC treated with sunitinib as the first-line treatment. Differences in lncRNA expression levels between primary tumors and adjacent non-malignant tissue were analyzed. The present study also evaluated associations of lncRNA expression profile with clinical data, including objective response and patient survival. The role of clinically relevant lncRNAs in the context of mRNA transcriptome profile was investigated to reveal their mutual interactions and biological importance.

Materials and methods

Patients and tissue samples

The present study included 38 patients with mRCC treated with sunitinib (SUTENT®; Pfizer, Inc.) as first-line therapy at the Department of Oncology and Radiotherapeutics, University Hospital Pilsen (Pilsen, Czech Republic). Only patients with ccRCC histology and those with favorable or intermediate risk according to the Memorial Sloan-Kettering Cancer Center (MSKCC) prognostic model were included (28,29). All patients had distant metastases at the start of sunitinib therapy. Sunitinib was administered orally at the standard approved dosing (30) until disease progression, unacceptable toxicity or patient refusal. The clinical data, including baseline clinical characteristics, treatment course and outcomes, were obtained from medical records.

Physical examination and routine laboratory tests were performed every <6 weeks and CT was performed every 3–4 months during treatment with sunitinib. The objective response was assessed by an independent experienced radiologist using Response Evaluation Criteria in Solid Tumors version 1.1 (31). The objective response was classified in terms of complete response (CR), partial response (PR), stable disease (SD) and progressive disease (PD) (2).

Fresh-frozen tissue samples, including primary ccRCC tumor and adjacent non-neoplastic kidney tissue, were obtained during radical or cytoreductive nephrectomy surgery and were stored in RNAlater (cat. no. AM7020; Thermo Fisher Scientific, Inc.) at −80°C until processing.

All patients signed an informed consent for participation. The Ethics Committee of the Faculty of Medicine and University Hospital Pilsen, Charles University (approval no. 302/2020) approved the informed consent form and study protocol for samples collected during the project. The present study was performed in accordance with the Declaration of Helsinki.

Baseline clinical data for all patients with mRCC included in the present study (n=38) and the subgroup (n=20) profiled by RNA sequencing (RNASeq) are summarized in Table I. The median age of patients at the time of sunitinib initiation was 64.5±8.8 years. The study included 26 male and 12 female patients. The majority of patients (>80%) belonged to the intermediate risk group according to the MSKCC criteria, and 50% of patients were evaluated as good responders to sunitinib as first-line systemic therapy. The median progression-free survival (PFS) and OS for the entire cohort were 7 and 36 months, respectively.

Table I. Baseline clinical characteristics of patients with metastatic renal cell carcinoma.

Table I.

Baseline clinical characteristics of patients with metastatic renal cell carcinoma.

Characteristic	lncRNA profiling (n=38)	mRNA profiling (n=20)
Median age at treatment initiation, years	64.5±8.8	64.5±8.5
Sex (%)
Male	26 (68)	15 (75)
Female	12 (32)	5 (25)
Histopathological grade (%)
G1	8 (21)	4 (20)
G2	12 (32)	6 (30)
G3	16 (42)	9 (45)
Not available	2 (5)	1 (5)
Stage at diagnosis (%)
I	3 (8)	1 (5)
II	0 (0)	0 (0)
III	18 (47)	8 (40)
IV	17 (45)	11 (55)
Distant metastasis (%)
Synchronous	16 (42)	11 (55)
Metachronous	22 (58)	9 (45)
MSKCC risk (%)
Favorable	7 (18)	3 (15)
Intermediate	31 (82)	17 (85)
First-line objective response to sunitinib
Complete response	6 (16)	5 (25)
Partial response	14 (37)	5 (25)
Progressive disease	18 (47)	10 (50)

[i] Memorial Sloan-Kettering Cancer Center; lnc, long non-coding.

Isolation of total RNA and preparation of cDNA

Tissue samples were removed from RNAlater and ground to powder by mortar and pestle under liquid nitrogen. RNA was isolated using AllPrep DNA/RNA/Protein Mini kit (cat. no. 80004; Qiagen GmbH) according to the manufacturer's protocol. Total RNA was quantified with Quant-it™ RiboGreen RNA Assay kit (cat. no. R11490; Thermo Fisher Scientific, Inc.) on Microplate Reader-Infinite M200 (Tecan Group, Ltd.). RNA quality was determined by estimation of the RNA Integrity Number (32) with Agilent RNA 6000 Nano kit (cat. no. 5067-1511; Agilent Technologies, Inc.) on Bioanalyzer 2100 (cat. no. G2939BA; Agilent Technologies, Inc.). cDNA was synthesized using RT2 First Strand kit (cat. no. 330404; Qiagen GmbH) with 0.5 µg total RNA, according to the manufacturer's protocol. cDNA was stored at −20°C until quantitative PCR (qPCR) was performed.

qPCR

To quantify relative gene expression, ViiA7 Real-Time PCR System (Thermo Fisher Scientific, Inc.) was used. qPCR study design adhered to the Minimum Information for Publication of the guidelines for qPCR experiments (33). The reaction mixture contained 650 µl 2X RT2 SYBR Green Mastermix (cat. no. 330502; Qiagen GmbH), 102 µl cDNA synthesis reaction (cDNA diluted with nuclease-free water according to the manufacturer's protocol) and 548 µl nuclease-free water. PCR components were dispensed into 384-well RT2 lncRNA PCR Array (cat. no. 330721; Qiagen GmbH) according to the manufacturer's protocol. Each RT2 lncRNA PCR Array Human Cancer PathwayFinder (cat. no. 330721; Qiagen GmbH; GeneGlobe ID: LAHS-002Z) included control elements for data normalization, detection of genomic DNA contamination, RNA sample quality and general PCR performance check (Table SI). Thermocycling conditions were as follows: Initial step at 95°C for 10 min followed by 40 cycles of denaturation at 95°C for 15 sec and annealing at 60°C for 60 sec. The samples were analyzed in duplicate. The ACTB, B2M, RPLP0, R7SK and SNORA73A genes were used as reference genes for the normalization of results. For statistical analyses, the expression data were normalized and the 2−ΔΔCq method was used to determine relative expression (34).

RNASeq

RNASeq was performed in 20 pairs of primary tumor and adjacent non-malignant renal tissue. Libraries were prepared using 0.5 µg total RNA with QuantSeq 3′mRNA-Seq Library Prep kit FWD and PCR Add-on kit for Illumina (cat. no. 015.96 and 020.96, respectively; both Lexogen GmbH) according to the manufacturer's protocol. Bioanalyzer 2100 and High Sensitivity DNA kit (cat. no. 5067-4626; Agilent Technologies, Inc.) were used for quality control of prepared libraries. Libraries were quantified by qPCR, KAPA Library Quantification kit Illumina® Platforms (cat. no. 07960140001; Fritz Hoffman-La Roche Ltd). The equimolar pool (4 nM) of prepared libraries was sequenced on NextSeq 500 platform (Illumina, Inc.) with NextSeq 500/550 High Output kit v2.5 (1×75 bp, single read; cat. no. 20024906; Illumina, Inc.) in one run [seeding concentration 1.8 pM measured on Quibit 4.0 with dsDNA High Sensitivity Assay kit (cat. no. Q32851; Thermo Fisher Scientific, Inc.)]. RNASeq of all samples was performed in sufficient depth (~10 million reads/sample) for the detection of lowly expressed genes. Quality control of raw RNASeq data was performed with the FastQC v0.11.9 package (35).

Statistical analysis

Statistical analysis of associations between lncRNA expression and clinical data of patients were performed by SPSS (v16.0; SPSS, Inc.) or GraphPad Prism (v6.0; Dotmatics). The distribution of most lncRNAs deviated from normality, and non-parametric statistical tests were used. Kruskal-Wallis test was used for evaluation of association between lncRNA expression profile and clinical parameters such as clinical stage, primary tumor size, histopathological grading and MSKCC risk. Mann-Whitney test was used for evaluation of associations between lncRNA expression profile and response to sunitinib, sex, presence of regional lymph node metastasis, type of distant metastatic spread (synchronous or metachronous) and comparison of lncRNA profile between primary tumor and paired non-neoplastic tissue. Spearman rank test was used for evaluating the correlation between the lncRNA expression levels and age of patients. Log-rank test and Kaplan-Meier plots were used to identify associations of lncRNA expression levels with PFS and OS in months. Patients were divided according to the median expression of a given lncRNA. PFS was defined as time between sunitinib treatment initiation and first documented progression or death or patient censoring. OS was defined as time from sunitinib treatment initiation until the date of death or patient censoring. All patients with OS >60 months were censored at this time point. Cox regression was performed to assess the hazard ratio (HR) with a 95% confidence interval (CI). Two-sided P-value was calculated for all statistical analyses. The false discovery rate (FDR) test was applied according to Benjamini and Hochberg (36) and Q-values were computed for each comparison. Q<0.05 was considered to indicate a statistically significant difference. The ‘good responders group’ was defined as patients who achieved CR or PR, while ‘poor responders group’ was defined as patients who achieved PD.

mRNA expression analysis

For gene annotation, Ensembl v101 (genome assembly GRCh38.p13) was used (37). A pseudoaligment approach for gene quantification by kallisto v0.46.1 was used (38). Differential expression analysis was carried out with the edgeR v3.42.2 package in R (39). Differentially expressed genes with log fold-change (FC)>2 and Q<0.001 (Benjamini Hochberg FDR correction) were considered statistically significant for comparison of tumor vs. non-malignant tissue, while for good vs. poor responders, Q<0.05 (Benjamini Hochberg FDR correction) was considered statistically significant. Expression data in the normalized format (transcripts per million) were used for analysis of associations between mRNA expression and clinical data of patients only for significantly differentially expressed protein-coding genes. The statistical tests used for these analyses were the same as those for lncRNAs.

Spearman correlation analysis was used to evaluate correlation between mRNA and lncRNA expression. R>0.8 and Q<0.001 [Bonferroni correction (40)] were considered to indicate a statisitcally significant difference. The complete set of 84 examined lncRNAs and 11,342 protein-coding transcripts with an interquartile range of expression values >0.1, with the exception of pseudogenes and uncharacterized proteins, were incorporated into the analysis.

Pathway annotation of protein-coding genes associated with lncRNAs was performed with the Reactome database (41). Q<0.05 (Benjamini Hochberg FDR correction) was considered to indicate a statistically significant difference.

Results

lncRNA profile between the primary tumor and paired non-neoplastic tissues

The levels of lncRNAs were detected by qPCR in 20 pairs of primary tumor and adjacent non-neoplastic tissue samples. Comparison of lncRNA expression profile revealed 50 differentially expressed lncRNAs (Table II). A significantly higher expression of 13 lncRNAs in carcinoma compared with paired non-neoplastic tissue was observed. By contrast, the levels of 37 lncRNAs were significantly decreased in carcinoma. In addition, four differentially expressed lncRNAs (GACAT1, HEIH, MIR155HG and POU5F1P5) passed the P-value cut-off (P<0.05) but not the FDR correction (Benjamini and Hochberg) (Table II).

Table II. Significant differences in expression of cancer-associated lncRNAs between primary tumor and adjacent non-malignant tissue (n=20 pairs).

Table II.

Significant differences in expression of cancer-associated lncRNAs between primary tumor and adjacent non-malignant tissue (n=20 pairs).

lncRNA	Expression change	P-value	Q-value
ADAMTS9-AS2	Downregulated	0.011	0.021
AIRN	Downregulated	0.021	0.035
BANCR	Downregulated	<0.001	<0.001
BLACAT1	Downregulated	0.021	0.035
CAHM	Downregulated	<0.001	<0.001
CBR3-AS1	Downregulated	<0.001	<0.001
CDKN2B-AS1	Upregulated	<0.001	<0.001
CRNDE	Upregulated	0.003	0.006
DGCR5	Upregulated	<0.001	<0.001
EMX2OS	Downregulated	<0.001	<0.001
FTX	Downregulated	<0.001	<0.001
GACAT1	Downregulated	0.040	0.063a
GAS6-AS1	Upregulated	0.001	0.002
H19	Downregulated	0.006	0.012
HAND2-AS1	Downregulated	<0.001	<0.001
HEIH	Downregulated	0.046	0.072a
HIF1A-AS1	Downregulated	0.020	0.035
HIF1A-AS2	Upregulated	<0.001	<0.001
HNF1A-AS1	Downregulated	<0.001	<0.001
HOTAIRM1	Downregulated	<0.001	<0.001
HOXA11-AS	Downregulated	<0.001	<0.001
HOXA-AS2	Downregulated	<0.001	<0.001
HULC	Downregulated	<0.001	<0.001
IPW	Downregulated	<0.001	<0.001
KCNQ1OT1	Downregulated	<0.001	<0.001
KRAASP1	Downregulated	0.002	0.035
LINC00152	Upregulated	0.001	0.002
LINC00887	Upregulated	<0.001	<0.001
LINC00963	Downregulated	<0.001	<0.001
LINC01233	Downregulated	<0.001	<0.001
LINC01234	Upregulated	0.007	0.014
LUCAT1	Upregulated	<0.001	<0.001
MALAT1	Downregulated	0.008	0.015
MEG3	Downregulated	<0.001	<0.001
MIR155HG	Downregulated	0.040	0.063a
MIR17HG	Upregulated	<0.001	<0.001
MIR31HG	Downregulated	0.001	0.002
MRPL23-AS1	Downregulated	0.001	0.002
NAMA	Downregulated	0.002	0.004
NBR2	Downregulated	<0.001	<0.001
NEAT1	Downregulated	<0.001	<0.001
NRON	Downregulated	0.013	0.024
POU5F1P5	Downregulated	0.038	0.063a
PRNCR1	Downregulated	<0.001	<0.001
PTCSC3	Downregulated	<0.001	<0.001
PVT1	Upregulated	<0.001	<0.001
RMRP	Downregulated	0.006	0.012
RMST	Downregulated	<0.001	<0.001
SNHG16	Upregulated	0.006	0.012
SUMO1P3	Downregulated	0.001	0.002
TUG1	Downregulated	<0.001	<0.001
WT1-AS	Downregulated	<0.001	<0.001
ZFAS1	Upregulated	0.002	0.004

a Non-significant. lncRNA, long non-coding RNA.

Associations of lncRNA expression profile with baseline clinical data

The levels of the lncRNAs in tumors from 38 patients were evaluated for their associations with the clinical data. The association of lncRNA expression profile with sex, grade, MSKCC risk, primary tumor size and synchronous or metachronous distant metastatic spread is shown in Table SII. However, only higher lncRNA TSIX and XIST levels in females were significant after FDR correction.

Associations of lncRNA expression profile with objective response and survival

The expression of 10 lncRNAs (ADAMTS9-AS2, CDKN2B-AS1, CRNDE, EMX2OS, HEIH, HNF1A-AS1, IPW, LINC00963, NRON and PTENP1) was upregulated, while the expression of LINC00261, LINC01234, TUG1 and TUSC7 was downregulated in good compared with poor responders (Table III). The downregulation of TUSC7 lncRNA, which remained significant after FDR correction, was the most important finding.

Table III. Association of expression levels of cancer-associated lncRNAs with the objective response to sunitinib in good (n=20) vs. poor (n=18) responders.

Table III.

Association of expression levels of cancer-associated lncRNAs with the objective response to sunitinib in good (n=20) vs. poor (n=18) responders.

lncRNA	Expression change	P-value	Q-value
ADAMTS9-AS2	Upregulated	0.026	0.196a
CDKN2B-AS1	Upregulated	0.019	0.196a
CRNDE	Upregulated	0.011	0.154a
EMX2OS	Upregulated	0.024	0.196a
HEIH	Upregulated	0.047	0.288a
HNF1A-AS1	Upregulated	0.028	0.196a
IPW	Upregulated	0.002	0.084a
LINC00261	Downregulated	0.010	0.154a
LINC00963	Upregulated	0.024	0.196a
LINC01234	Downregulated	0.011	0.154a
NRON	Upregulated	0.024	0.196a
PTENP1	Upregulated	0.011	0.154a
TUG1	Downregulated	0.033	0.213a
TUSC7	Downregulated	<0.001	<0.001

a Non-significant. lncRNA, long non-coding RNA.

Association of lncRNA expression profile with PFS and OS showed that patients with expression levels of HNF1A-AS1 (HR=0.193; 95% CI=0.051-0.724) and IPW (HR=0.18; 95% CI=0.03–0.95) above the median (high expression) had prolonged PFS compared with those with levels below the median (Fig. 1A and B). Expression levels of TUSC7 above the median (high expression) were associated with poor PFS (HR=4.3; 95% CI=1.35–13.70; Fig. 1C) and OS (HR=4.15; 95% CI=1.38–12.48; Fig. 1D).

Figure 1. Association of expression of lncRNA with PFS. Association of (A) HNF1-AS1, (B) IPW and (C) TUSC7 with PFS. (D) Association of TUSC7 with OS. lncRNA, long non-coding RNA; OS, overall survival; PFS, progression-free survival; CI, confidence interval; HNF1-AS1, hepatic nuclear factor 1 homeobox A antisense RNA; IPW, imprinted in Prader-Willi syndrome; TUSC7, tumor suppressor candidate 7.

mRNA transcriptome profile between primary tumor and adjacent non-neoplastic tissues

Analysis of differential expression between primary tumors and paired adjacent non-neoplastic tissues was performed in 20 patients, whose baseline clinical data are summarized in Table I. In total, 768 significantly differentially expressed genes were identified. A total of 462 and 306 genes were down- and upregulated, respectively, in tumor compared with non-neoplastic tissue (Table SIII). The volcano plot and top 10 down- and upregulated genes are shown in Fig. 2A and B.

Figure 2. Differential expression analysis of protein-coding genes between primary tumor and adjacent non-malignant tissue samples. (A) Volcano plot of the most significant results from differential expression analysis (logFC>2; Q<0.001). (B) Top 10 most up- and downregulated genes (n=20).

Association of mRNA expression profile with objective response and survival

Among all protein-coding genes, only one significant association with objective response to sunitinib was identified: Significantly upregulated CLIP4 expression was observed in primary tumor tissues of poor responders compared with good responders (logFC=−1.9; Q=0.02; Fig. 3).

Figure 3. Significantly higher CLIP4 expression is observed in poor vs. good responders (n=20) as shown in a scatter plot where the mean is indicated by a horizontal line. CLIP4, CAP-Gly domain containing linker protein family member 4.

mRNA-lncRNA co-expression networks

The complete mRNA expression profile was compared with 84 cancer-associated lncRNA in 20 primary tumor tissues. This revealed 107 significant associations (R>0.8 and Q<0.001). In total, levels of 26 lncRNAs were significantly correlated with mRNA expression levels of 65 protein-coding genes (Fig. 4). The lncRNAs with the highest number of correlations with protein-coding genes were MEG3 (14 positive correlations) and SNHG16 (7 negative correlations; Fig. 4; Table SIV). MEG3 was positively correlated with expression of ABHD2, BPIFA3, CARD8, CBS, CDK12, IPO4, MB, OSER1, POLR3F, RNF126, SLC6A1, SPHK2, TRIM60 and UNC45A (Table SV). SNHG16 was negatively correlated with the expression of ABCF3, AP2M1, CLN8, DNMT3A, IQCB1, PDE4D and SSBP4.

Figure 4. Significant correlations between expression of protein-coding genes and lncRNAs in primary tumor tissue. (A) Positive and (B) negative correlations. Only results with Spearman correlation coefficient R>0.8 and Q<0.001 are shown. lnc, long non-coding.

Among the deregulated lncRNAs in poor responders to sunitinib, lncRNA ADAMTS9-AS2 was positively correlated with expression of NKX2-1 and INTS8 mRNAs (R=0.83 and R=0.84, respectively). By contrast, CRNDE lncRNA displayed only negative correlations, namely with the expression of DPH5 and CDK11B mRNAs (R=−0.84 and R=−0.84, respectively). The expression of lncRNA EMX2OS was positively correlated with that of RHOBTB3 and UGT2A1 and negatively correlated with that of ITGA2 mRNA (R=0.84, R=0.85 and R=−0.87, respectively), while LINC01234 was positively correlated with the expression of TUBB4A mRNA (R=0.85).

Pathway analysis was performed with the Reactome database for the aforementioned lncRNAs (Fig. 5). Genes correlated with the lncRNA MEG3 were significantly enriched in metabolism (SPHK2 and CBS), immune system/response (CARD8 and POLR3F), apoptosis (CARD8), gene expression (POLR3F and CDK12), neuronal system (SLC6A1) and transport (MB) pathways. The function of other genes correlated with MEG3 (BPIFA3, OSER1, ABHD2, UNC45A, TRIM60 and IPO4) has not been identified to date (Fig. 5A). SNHG16 lncRNA-correlated genes were significantly enriched in GTPase cycle (CLN8), signaling (PDE4D and AP2M1), DNA methylation (DNMT3A) and plasma membrane (IQCB1), and two genes had unknown function (SSBP4 and ABCF3; Fig. 5B). Genes correlating with ADAMTS9-AS2 lncRNA were significantly enriched in gene expression (INTS8) and surfactant metabolism (NKX2-1; Fig. 5C). MUCL3 gene correlating with CDKN2B-AS1 lncRNA has an unknown function (Fig. 5D). TUBB4A correlating with LINC01234 lncRNA was significantly enriched in protein transport and folding (Fig. 5E). Genes correlating with CRNDE lncRNA were significantly enriched in metabolic pathways (DPH5) and cell cycle regulation (CDK11B; Fig. 5F), while genes correlating with lncRNA EMX2OS were enriched in drug metabolism (UGT2A1), ATPase cycle (RHOBTB3) and signaling pathways/cell interactions (ITGA2; Fig. 5G). Results are summarized in Table SVI.

Figure 5. Pathway enrichment analysis using Reactome database for selected lncRNA-mRNA correlations. Genes correlated with (A) MEG3, (B) SNHG16, (C) ADAMTS9-AS2, (D) CDKN2B-AS1, (E) LINC01234, (F) CRNDE and (G) EMX2OS lncRNAs are presented. lncRNA, long non-coding RNA.

Discussion

lncRNAs play a complex role in cancer biology (9). Although the value of lncRNAs as potential prognostic or predictive biomarker in patients with mRCC has already been suggested (16), associations between lncRNA profile and outcome focused on specific types of systemic targeted therapy remain underexplored.

The present study analyzed associations between the expression profile of cancer-specific lncRNAs selected using Human Cancer PathwayFinder and the outcome of patients with mRCC treated with sunitinib as first-line therapy. The results suggested a potential prognostic and/or predictive role of HNF1A-AS1, IPW and TUSC7 among 84 cancer-specific lncRNAs. Moreover, full transcriptome analysis protein-coding genes was performed; CLIP4 was associated with objective response. Furthermore, MEG3 and SNHG16 lncRNAs were not only dysregulated in mRCC, but also strongly associated with the expression levels of several protein-coding genes, suggesting a complex functional significance and potential use in targeted therapies.

Thus, according to the present study, downregulated expression of the TUSC7 lncRNA may serve as a negative prognostic and predictive biomarker candidate in follow-up studies on mRCC and other malignancies. It was downregulated in tumors compared with non-malignant renal tissue and upregulated in tumors of poor responders and patients with worse survival (both PFS and OS). To the best of our knowledge, TUSC7 has not been previously reported in connection with renal malignancies. In non-malignant tissues, TUSC7 expression is upregulated in testes (42). TUSC7-regulated cellular processes play a tumor-suppressor function in various types of cancer, for example, inhibiting the proliferation rate and migration of tumor cells in epithelial-mesenchymal transition (EMT) in colorectal cancer cell lines and tissue (43) or osteosarcoma cells (44). According to a previous study, TUSC7 downregulation is an independent biomarker of poor prognosis in patients with triple-negative breast cancer (45), which contradicts the observations of the present study; this is probably due to the different nature of tumors of individual tissue types. TUSC7 is regulated by p53 in vitro (45) and thus, the functional status of the p53 pathway may affect the TUSC7 prognostic significance. Patients with triple-negative breast cancer have a high prevalence (~75%) of TP53 mutation (46). Thus, TUSC7 may have different tissue-specific functions reflecting the p53 status in a specific type and histological subtype of carcinoma, which may partly explain discordant results.

HNF1A-AS1 and IPW served as positive predictive biomarkers in the present study as their upregulation in tumors was associated with good response and prolonged PFS. HNF1A-AS1 lncRNA is upregulated mainly in gastrointestinal, liver and kidney tissues (47). HNF1A-AS1 expression is often deregulated in cancer and it serves roles in cell proliferation, invasion, migration and apoptosis primarily via cooperation with microRNAs (miRs) or by regulating the EMT process (48–51). HNF1A-AS1 serves as a tumor promoter, but also as a tumor suppressor, as shown by Zhang et al (51). Upregulation of HNF1A-AS1 is demonstrated in numerous tumors, such as osteosarcoma, gastrointestinal, breast, lung or cervical carcinoma, while it is downregulated in gastroenteropancreatic and neuroendocrine neoplasm and oral squamous cell carcinoma (48,51). In the present study, HNF1A-AS1 was significantly downregulated in patients with mRCC. In connection with development of tumors and tumor progression, HNF1A-AS1 promotes lung cancer cell proliferation and invasion via regulating miR-17-5p (49). Expression of miR-149-5p negatively correlated with HNF1A-AS1 in tissue of patients with non-small cell lung cancer (NSCLC) and in NSCLC cell lines (50). In a meta-analysis focusing on the usefulness of HNF1A-AS1 as a prognostic marker in malignant tumor, high HNF1A-AS1 expression correlated with poor OS and disease-free survival in patients with colorectal, bladder and lung cancer and osteosarcoma (52). On the other hand, HNF1A-AS1 serves as a tumor suppressor in other studies (53,54), in accordance with the present study. Dang et al (55) showed that downregulation of HNF1A-AS1 in gastric cancer is associated with tumor size and concentration of the protein serum biomarkers carcinoembryonic antigen and carbohydrate antigen 19–9, as well as with the protein expression of ribonucleotide reductase subunit M1 in tissue samples. In liver cancer, HNF1A-AS1 is downregulated, and could inhibit the proliferative and metastatic abilities of hepatocellular carcinoma xenograft tumors (54). Thus, the function and mechanism of action of HNF1A-AS1 depends on cell specificity and tumor type.

IPW is a nuclear lncRNA with tissue-specific expression. It has been shown to regulate genomic imprinting, a subject for a study of transcriptional and post-transcriptional-based gene regulation (56). The highest levels of IPW are identified in the nervous system, based on estimation by BioGPS microarray (57). IPW forms part of a six-lncRNA prognostic signature in gastric cancer (58) and recently it was reported to be downregulated in head and neck squamous cell cancer (HNSCC) cells in comparison with normal keratinocyte cells in vitro (59). In addition, downregulation of expression of IPW is associated with worse OS in patients with HNSCC (59). The present study found an association between downregulation of IPW expression and poor objective response and worse PFS in patients with mRCC, but not with OS, suggesting predictive, rather than prognostic, value. To the best of our knowledge, the role of IPW in ccRCC has not been investigated to date.

To address the complexity of lncRNA-mRNA interacting networks, the present study complemented the targeted lncRNA analyses with assessment of the coding transcriptome in a subset of patients with mRCC. The analysis revealed upregulation of CLIP4 in patients with mRCC with poor response to sunitinib. CLIP4 encodes the intracellular CAP-Gly domain containing linker protein family member 4, a protein involved in cytoplasmic microtubule organization (Gene Ontology:0031122) (60). Park et al (61) analyzed the transcriptome of patients with early-stage ccRCC (n=24) using RNASeq and subsequently suggested and validated the association of CLIP4 upregulation with poor prognosis. In addition, CLIP4 mutations are enriched 3-fold in patients with aggressive ccRCC defined as tumors exhibiting synchronous metastasis, early recurrence or cancer-specific mortality, compared with patients without aggressive ccRCC (61). Ahn et al (62) noted that upregulation of CLIP4 expression was associated with synchronous metastasis in ccRCC, and an in vitro functional study showed that CLIP4 significantly increases cell migration and viability in ccRCC (62). Taken together, several studies, including the current one, suggest CLIP4 upregulation as a poor prognosis biomarker in patients with ccRCC.

In the present study, two lncRNAs (MEG3 and SNHG16) were significantly associated with individual gene expression profile of tumors from patients with mRCC. MEG3 expression was positively correlated with expression of 14 protein-coding genes, and pathway enrichment analysis suggested an involvement of genes from biological processes such as cell metabolism, apoptosis, transport and immune system regulation. A network involving MEG3 lncRNA may serve a role in prognosis and therapy response. Gong et al (63) revealed a positive correlation of ST3 β-galactoside α-2,3-sialyltransferase 1 (ST3Gal1) expression with MEG3 in ccRCC, and suggested a potential role of the MEG3/ST3Gal1/epidermal growth factor receptor axis in ccRCC progression. Upregulation of MEG3 induces apoptosis via the reduction of Bcl-2 and procaspase-9 protein and the promotion of cytochrome c release into the cytoplasm (14). The present study found a significant downregulation of MEG3 in ccRCC, confirming the results of previous studies (14,63,64) reporting downregulation of MEG3 in tumors of patients with ccRCC and ccRCC cell lines compared with non-malignant renal tissues.

SNHG16 lncRNA was upregulated in tumors and negatively correlated with the expression of seven protein-coding genes in the present study. Functionally, SNHG16 promotes cell proliferation and suppresses apoptosis via interaction with miR-1301-3p, leading to the upregulation of STAR expression in ccRCC cells (65). In agreement with a previous study (65), the present study confirmed that SNHG16 may serve a role as an oncogene in ccRCC.

There are limitations to the present study, including a small sample size and a retrospective design. Nevertheless, the current study focused on the metastatic stage of ccRCC, which is not as common as the early stages of ccRCC and there are limited options to obtain fresh frozen tissue samples from patients with mRCC, particularly those with synchronous metastatic disease. The next limitation is that qPCR for lncRNA profile measures only the expression of a limited number of pre-selected lncRNAs. Another limitation is that sunitinib monotherapy is no longer the first choice of first-line treatment and it has been replaced with immunotherapy combination regimens, represented by combinations of TKI plus ICI or ICI plus ICI. Finally, the lack of functional studies of the identified candidate lncRNA biomarkers is another limitation.

On the other hand, TKIs are still widely used in the treatment of mRCC, and a search for candidate predictive biomarkers for these agents could bring progress in the personalized use of TKIs in monotherapy, even in combination with immunotherapy (7,8). Moreover, the followed-up group of patients with mRCC was clinically well-characterized, particulary during first-line systemic treatment and represented a uniquely homogenous group of patients with mRCC, coupled with the prospectively updated outcome data. Furthermore, high-throughput RNASeq methodology was used for estimation of the whole coding transcriptome in 20 patients with mRCC, and the data of the current study may serve as a hypothesis-generating screening for larger functional and replication studies in independent cohorts of patients with mRCC to confirm the observations of the present study. Functional studies of the candidate lncRNA biomarkers identified in the present study are ongoing.

In conclusion, the present study provided novel information within the lncRNA field and their clinical role as molecular biomarkers of therapeutic response in patients with mRCC. Among 84 cancer-associated lncRNAs, HNF1A-AS1, IPW and TUSC7 dysregulation was associated with outcome of patients with mRCC treated with sunitinib. Moreover, the predictive association was revealed for the CLIP4 protein-coding transcript. Additionally, significant associations of MEG3 and SNHG16 with several protein-coding transcripts, creating complex interactive networks, were identified and confirmed by in silico predictions of molecular and biological function. The aforementioned molecules represent putative candidates for predictive and prognostic biomarkers in precision and personalized therapy of mRCC.

Supplementary Material

Supporting Data

Acknowledgements

Not applicable.

Funding

The present study was supported by the Ministry of Education, Youth and Sports of the Czech Republic (project INTER-ACTION; grant no. LUAUS23164), National Institute for Cancer Research (program EXCELES; grant no. LX22NPO5102) funded by the European Union (EU)-Next Generation EU, Grant Agency of Charles University (project GAUK; grant no. 1074120), Charles University (project COOPERATIO Surgical Disciplines; grant no. 207043) and by the EU Horizon 2020 research and innovation program (grant no. 856620).

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. RNASeq data are available in the National Center for Biotechnology Information database's Sequence Read Archive repository, BioProject no. PRJNA932699 (ncbi.nlm.nih.gov/bioproject/PRJNA932699/).

Authors' contributions

RV, OF, PS and TT conceptualized the study. TT, KK and KS performed experiments. TT and KS performed data analysis. MH, OH and KP provided and characterized tissue resources, and described all clinical parameters and characteristics included in this paper. KS and TT visualized the data. TT, KK, KS and OF wrote the manuscript. OF, RV and PS reviewed and edited the manuscript. PS supervised the study and created essential parts of the discussion section. TT and KS confirm the authenticity of all the raw data. All authors have read and approved the final manuscript.

Ethics approval and consent to participate

The present study involving human subjects was conducted according to the guidelines of the Declaration of Helsinki and approved by The Ethics Committee of the Faculty of Medicine and University Hospital Pilsen, Charles University (approval no. 302/2020). Written informed consent was obtained from all subjects involved in the study.

Patient consent for publication

Not applicable.

Competing interests

Ondrej Fiala received payment or honoraria for lectures, presentations, speakers' bureaus, or educational events from Roche, Janssen, GSK, MSD, Pierre Fabre, BMS and Pfizer unrelated to this project.

References