Abstract
It is reported that lncRNA KCNQ1 opposite strand/antisense transcript 1 (KCNQ1OT1) is oncogenic in many cancers. This work aimed at probing into its expression and biological functions in retinoblastoma (RB) as well as its regulatory effects on miR-153-3p and hypoxia-inducible factor-1α (HIF-1α). In our study, RB samples in pair were collected, and quantitative real-time PCR (qRT-PCR) was employed for examining the expression levels of KCNQ1OT1, miR-153-3p and HIF-1α. KCNQ1OT1 short hairpin RNAs were transfected into SO-Rb50 and HXO-RB44 cell to inhibit the expression of KCNQ1OT1. The proliferative activity, colony formation ability and apoptosis were examined through cell counting kit-8 assay, colony formation assays, Transwell assay and flow cytometry, respectively. qRT-PCR and western blot analysis were used for analyzing the changes of miR-153-3p and HIF-1α induced by KCNQ1OT1. The regulatory relationships between miR-153-3p and KCNQ1OT1, miR-153-3p and HIF-1α were examined by dual luciferase reporter gene assay and RNA-binding protein immunoprecipitation assay. The results of our study showed that KCNQ1OT1 expression was markedly enhanced in RB tissue samples, and KCNQ1OT1 knockdown had an inhibitory effect on the proliferation, migration, invasion and viability of RB cells. There were two validated binding sties between KCNQ1OT1 and miR-153-3p, and KCNQ1OT1 negatively regulated the expression of miR-153-3p in RB cells. HIF-1α was a target gene of miR-153-3p, and could be positively regulated by KCNQ1OT1. In conclusion, our study indicates that KCNQ1OT1 can increase the malignancy of RB cells via regulating miR-153-3p/HIF-1α axis.
Significance of this study
What is already known about this subject?
Long non-coding RNAs (lncRNAs) have been reported as potential diagnostic biomarkers in various tumors including retinoblastoma (RB).
lncRNAs have been proved to be competitive endogenous RNAs to participate in tumor development by decoying miroRNAs.
KCNQ1 opposite strand/antisense transcript 1 (KCNQ1OT1) has been reported as an oncogenic factor in many cancers.
What are the new findings?
KCNQ1OT1 was aberrantly upregulated in RB tissue samples.
KCNQ1OT1 knockdown had an inhibitory effect on the proliferation, migration, invasion and viability of RB cells.
KCNQ1OT1 was able to increase the malignancy of RB cells via regulating miR-153-3p/hypoxia-inducible factor-1α axis.
How might these results change the focus of research or clinical practice?
The findings indicated that KCNQ1OT1 and miR-153-3p might be novel therapy targets for patients with RB.
Introduction
Retinoblastoma (RB) is the most common primary intraocular malignancy among children and infants, with a neonatal morbidity of 1/16,000, which is severely detrimental to patients’ vision and life.1 2 Despite significant advances in diagnosis and surgical technologies, the prognosis for patients with RB remains less than satisfactory.2 3 In such a case, there is an urgency to further clarify the molecular mechanism of RB, so that new effective diagnostic markers and therapeutic targets can be screened.
As a type of RNA molecules, long non-coding RNAs (lncRNAs) are defined as a transcript longer than 200 nucleotides, with no typical initiation codon, conserved region of promoter, open reading frame and termination codon, thus lacking protein coding ability.4 Most lncRNAs are transcribed by RNA polymerase II, spliced and then matured, and some of mature lncRNAs also possess a 5’ cap and 3’ polyadenylic acid tail.5–7 Studies show that as an important regulatory factor, lncRNA contributes to regulating gene expression in multiple aspects, that is, epigenetics, transcription and post-transcription, and exerts its biological functions in various biological processes, such as stem cell maintenance, cell proliferation, differentiation and apoptosis.8 9 In recent years, there are an increasing number of studies about lncRNAs in RB. For instance, compared with in normal tissues, colon cancer-associated transcript 1 (CCAT1) expression in RB tissues is higher, and interference of CCAT1 represses RB cell proliferation.10 In contrast, maternally expressed gene 3 (MEG3) in RB tissues is relatively underexpressed, and its expression has a negative correlation with the metastatic ability of tumor cells; in terms of mechanism, MEG3 suppresses RB cell proliferation and metastasis via suppressing Wnt/β-catenin signaling pathway.11 However, the expression, biological function and underlying mechanism of lncRNA KCNQ1 opposite strand/antisense transcript 1 (KCNQ1OT1) in RB remain obscure.
As a category of small-molecule non-coding RNAs made up of 21–25 nucleotides, microRNAs (miRNAs) regulate, at the post-transcriptional level, the expression of target genes through the complementary base pairing with target genes’ 3’ untranslated region (3’-UTR).12 Aberrant expressions of miRNAs in RB are confirmed. For example, a study uses Affymetrix platform to analyze the miRNA expression profiles of 12 cases of RB samples, and the results show that 30 miRNAs are markedly overexpressed and 993 miRNAs are consistently absent in all cases.13 In addition, miR-486-3p and miR-532 are found to be significantly downregulated in RB and their overexpression in RB cells leads to apoptotic cell death.14 miR-153-3p functions in melanoma,15 medullary thyroid carcinoma16 and other diseases. Nonetheless, the role of miR-153-3p in RB is still not clear.
In this research, we demonstrated that KCNQ1OT1 and miR-153-3p were involved in regulating the malignancy of RB cells. We proved that miR-153-3p was a target of KCNQ1OT1, and KCNQ1OT1/miR-153-3p axis could regulate the expression of hypoxia-inducible factor-1α (HIF-1α). The findings indicated that KCNQ1OT1 and miR-153-3p were novel therapy targets for the RB.
Materials and methods
Tissue samples
Twenty-five paired tissue samples were available from patients with RB who had underwent surgical excision in the No. 1 People’s Hospital of Lanzhou from September 2014 to December 2018. None of patients received chemotherapy and/or radiotherapy prior to the excision. All samples were frozen in liquid nitrogen and stored. The subjects’ average age was 17.5±4.7 months, and their guardians signed the informed consent.
Cell culture
A human normal retinal pigment epithelial cell line (ARPE-19 cells) and four human RB cell lines (SO-Rb50, Weri-Rb1, Y79 and HXO-RB44) were bought from the American Type Culture Collection (Manassas, Virginia, USA). SO-Rb50, Y79, HXO-RB44 and Weri-Rb1 cells were cultured in a modified Roswell Park Memorial Institute 1640 medium (Hyclone, Camarillo, California, USA) containing 100 U/mL penicillin, 100 mg/mL streptomycin (Gibco, Rockville, Maryland, USA) and 10% fetal bovine serum (Gibco, Grand Island, New York, USA). A Dulbecco’s modified Eagle’s medium (Hyclone, Logan, Utah, USA) was used for ARPE-19 cell culture. All of these cells were cultured in a humidified incubator at 37°C in 5% CO2.
Cell transfection
When the cells reach 70% confluence, the cell transfection was carried out based on the instructions of Lipofectamine 2000 (Invitrogen, Carlsbad, California, USA). The vectors and oligonucleotides included small interfering-negative control (si-NC), small interfering RNA (siRNA) against KCNQ1OT1 (si-KCNQ1OT1#1 and si-KCNQ1OT1#2), overexpression vector plasmid pcDNA-KCNQ1OT1 and negative control plasmid (pcDNA), miR-153-3p mimic and miR-153-3p inhibitor (anti-miR-153-3p) and their respective controls (GenePharma, Shanghai, China).
RNA extraction and qRT-PCR
In line with the instructions, TRIzol reagent (Invitrogen) was employed for extracting total RNA from tissues or cells, and reverse transcriptase (Takara, Dalian, China) was used to reverse-transcribe RNA into complementary DNA. Furthermore, SYBR Premix Ex TaqTM II (Takara) was employed to prepare the reaction system for qRT-PCR experiment, and the 2-△△Ct method was used for analyzing the relative expressions of the genes. The primer sequences were: KCNQ1OT1 forward, 5’-TGCAGAAGACAGGACACTGG-3’ and reverse, 5’-CTTTGGTGGGAAAGGACAGA-3’; miR-153-3p forward, 5’-ACACTCCAGCTGGGTTGCATAGTCACAAAAGT-3’ and reverse, 5’-CTCAACTGGTGTCGTGGAGT-CGGCAATTCAGTTGAGGATCACTTT-3’; HIF-1α forward, 5’-ACTTGGACGCTCTGCCTATG-3’ and reverse, 5’-TTGCGGGGGTTGTAGA-3’; GAPDH forward, 5’-TGCACCACCAACTGCTTAGC-3’ and reverse, 5’-GGCATGCACTGTGGTCATGAG-3’; U6 forward, 5’-CTCGCTTCGGCAGCACA-3’ and reverse, 5’-AACGCTTCACGAATTTGCGT-3’. Among them, glyceraldehyde 3-phosphate dehydrogenase (GAPDH) or U6 functioned as the internal reference.
Cell counting kit-8 assay
Cells were inoculated into 96-well plate (1000 cells/well) for 24, 48 and 72 hours, respectively. At each time point, 10 µL cell counting kit-8 (CCK-8) kit (Dojindo Laboratories, Kumamoto, Japan) was added into each well. After 1 hour incubation, a microplate reader (Bio-Rad, Hercules, California, USA) was employed for detecting the optical density value of the cells at 450 nm.
Transwell chamber model
The density of cell suspension was modulated to 1×105 cells/mL with serum-free medium; 100 µL of cell suspension was then added into the Transwell chambers (Corning, New York, New York, USA), which were placed in a 24-well plate, and each well was added with 750 µL of complete medium, and the cells were cultured at 37 ℃ for 24 hours, and then the cells on the upper surface of the membrane were removed. Subsequently, the cells on the below surface of the membrane were fixed with 4% paraformaldehyde, followed by crystal violet staining and dried by airing for photographing and counting.
Cell invasion assay
The membranes of the Transwell chambers should be coated with Matrigel (Millipore, Bedford, Massachusetts, USA) in advance before adding the cells, and the rest of the steps were the same as the migration assay.
Flow cytometry analysis
Briefly, after being transfected for 48 hours, the cells were treated with 0.25% trypsin and then stained for 15 min with annexin V-FITC/propidium iodide (PI) apoptosis detection kit (Invitrogen). Following that, an Attune NxT flow cytometer (Thermo Fisher Scientific, Waltham, Massachusetts, USA) was employed for monitoring the apoptosis rate.
Dual-luciferase reporter assay
PCR was employed to amplify the fragment of wild-type (WT) KCNQ1OT1 and Qiaquick gel extraction kit (Qiagen, Hilden, Germany) was used for its purification. Then it was inserted into pmirGLO dual luciferase miRNA target expression vector (Promega, Madison, Wisconsin, USA) to build a dual-luciferase reporter gene vector of KCNQ1OT1 (WT). The fragment of the KCNQ1OT1 mutant (MUT) was also inserted into the vector to build a reporter gene vector of KCNQ1OT1 (MUT). The WT or MUT reporter vectors were co-transfected with miR-153-3p mimic or the negative control into the cells, respectively. Cells were cultured at 37 ℃ for 48 hours, and lysate was added for lysis, and then the cell supernatant was collected. At last, based on the Dual Luciferase Assay Manual (Promega), the relative luciferase activity was detected.
RNA-binding protein immunoprecipitation assay
Following the instructions of the Magna RNA immunoprecipitation kit (Millipore, Billerica, Massachusetts, USA), the mixture of 2×107 cells and cell lysate was incubated with magnetic beads conjugated IgG or Ago2 antibodies. After that, immunoprecipitation complex was incubated with proteinase K, and co-immunoprecipitated RNA was extracted. Ultimately, the enrichment of miR-153-3p and KCNQ1OT1 was measured through qRT-PCR.
Western blot analysis
The RNA-binding protein immunoprecipitation (RIP) assay lysis buffer (Beyotime Biotcchnology, Shanghai, China) was employed for extracting the total protein from the collected tissues and cells. After the lysate was centrifuged, the supernatant was collected and mixed with loading buffer. After denaturation, the samples were taken for 8% sodium dodecyl sulfate-polyacrylamide gel electrophoresis. In addition, the protein was transferred to polyvinylidene fluoride membranes (Millipore), and then tris-buffered saline-Casein blocking buffer containing Tween 20 (TBST; 10 mM Tris-HCI, pH 7.0, 150 mM NaCI, o.1% Tween 20) solution containing 5% skim milk was used to block the unspecific antigens at room temperature for 30 min. Afterward, primary antibodies anti-GAPDH (ab8245, 1:2000; Abcam, UK) and anti-HIF-1α (ab5160, 1:200; Abcam, UK) were added for overnight incubation at 4°C, and the membrane was immersed with TBST three times (15 min/time). Subsequently, the membranes added with secondary antibody (1:1000) were incubated for 1 hour at room temperature before being rinsed three times with TBST (15 min/time). At last, ECL chemiluminescent solution’ (Beyotime Biotechnology) was used to show the bands.
Statistical analysis
The statistical analysis of data was conducted through SPSS V.19.0 software (SPSS, Chicago, Illinois, USA). Mean±SD (x±s) functioned as the expression form of measurement data. Besides, the comparison between two groups was conducted by t-test, and the comparison among three or more groups was conducted by one-way analysis of variance. p<0.05 implied that the difference was of statistical significance.
Results
KCNQ1OT1 expression was greatly increased in human RB cell lines and tissues
First, we studied KCNQ1OT1 expression in normal retinal samples and RB tissues, and learnt that compared with normal tissues, KCNQ1OT1 expression in cancer tissues was higher (figure 1A). In addition, KCNQ1OT1 expression in RB cell lines was detected through qRT-PCR, and it was revealed that in comparison with ARPE-19 cells, KCNQ1OT1 expression was significantly increased in RB cell lines (figure 1B).
Inhibition of KCNQ1OT1 suppressed RB cell proliferation, invasion and migration, and promoted apoptosis
To examine KCNQ1OT1’s biological role in RB, we first transfected two siRNAs (si-KCNQ1OT1#1 and si-KCNQ1OT1#2) into SO-Rb50 and HXO-RB44 cells to knockdown KCNQ1OT1, and then measured cell proliferation, migration, invasion and apoptosis, the results of which displayed that both siRNAs reduced KCNQ1OT1 expression in HXO-RB44 and SO-Rb50 cells (figure 2A), and si-KCNQ1OT1#2 showed higher knockdown efficiency, so si-KCNQ1OT1#2 was used in all subsequent experiments. CCK-8 assay suggested that compared with the small interfering-negative control (si-NC) group, si-KCNQ1OT1#2 significantly suppressed HXO-RB44 and SO-Rb50 cell proliferation (figure 2B, C). Likewise, si-KCNQ1OT1#2 markedly decreased the migration and invasion capabilities of HXO-RB44 and SO-Rb50 cells (figure 2D, E), but promoted cell apoptosis (figure 2F). The above-mentioned findings revealed that downregulation of KCNQ1OT1 could repress RB cell proliferation, migration and invasion, and promote apoptosis.
KCNQ1OT1 targeted miR-153-3p in SO-Rb50 and HXO-RB44 cells
A growing body of studies indicate that lncRNAs, as competitive endogenous RNAs (ceRNAs), participate in tumor development by competing with mRNA to sponging miRNAs.17 The StarBase2.0 (http://starbase.sysu.edu.cn/) was employed for predicting the target miRNAs of KCNQ1OT1, and it was shown that KCNQ1OT1 contained three putative complementary binding sites for miR-153-3p (figure 3A). Dual-luciferase reporter gene assay suggested that miR-153-3p overexpression reduced the luciferase activity of KCNQ1OT1-wt1 and KCNQ1OT1-wt2 in SO-Rb50 and HXO-RB44 cells, but no effect was exerted on that of KCNQ1OT1-mut and KCNQ1OT1-wt3, indicating two of the three predicted binding sites were functional (figure 3B, C). RIP analysis further revealed that, compared with IgG, miR-153-3p and KCNQ1OT1 were enriched preferentially in Ago2-containing microribonucleoproteins in both HXO-RB44 and SO-Rb50 cells (figure 3D, E). Clinically, miR-153-3p expression was observably reduced in RB tissues (figure 3F), and correlation analysis displayed that miR-153-3p had a negative correlation with KCNQ1OT1 (figure 3G). Likewise, miR-153-3p expression was also significantly reduced in RB cells (figure 3H). In addition, si-KCNQ1OT1#2 knockdown markedly enhanced miR-153-3p expression in SO-Rb50 and HXO-RB44 cells (figure 3I). HXO-RB44 and SO-Rb50 cells were transfected by pcDNA-KCNQ1OT1 to build a KCNQ1OT1 overexpression model (figure 3J). As shown, overexpression of KCNQ1OT1 greatly suppressed miR-153-3p expression in RB cells (figure 3K). We concluded from the above results that KCNQ1OT1 had interaction with miR-153-3p, and repressed its expression in RB cells.
Inhibition of miR-153-3p rescued the effects of KCNQ1OT1 knockdown on RB cell proliferation, migration, invasion and apoptosis
Next, si-KCNQ1OT1#2 and anti-miR-153-3p were transfected into SO-Rb50 and HXO-RB44 cells for rescue experiments, and it was found that anti-miR-153-3p could significantly inhibit miR-153-3p expression in si-KCNQ1OT1-transfected HXO-RB44 and SO-Rb50 cells (figure 4A). Furthermore, CCK-8 analysis showed that anti-miR-153-3p was able to significantly reverse the decrease in SO-Rb50 and HXO-RB44 cell proliferation induced by si-KCNQ1OT1 (figure 4B, C). Also, Transwell assay and flow cytometry analysis indicated that anti-miR-153-3p partly eliminated si-KCNQ1OT1-mediated effects on SO-Rb50 and HXO-RB44 cell invasion, metastasis and apoptosis (figure 4D–F). The above findings revealed that KCNQ1OT1 played a role in RB via targeting miR-153-3p.
miR-153-3p specifically modulated HIF-1α expression in RB cells
Intratumoral hypoxia is a typical characteristic of solid tumors. Previous research uncovers a bidirectional regulation between HIF-1α and hypoxic tumor microenvironment, which contributes to the progression of RB.18 Our data indicated that HIF-1α expression was increased in both RB cells and tissues (figure 5A, B), and in clinical samples, miR-153-3p expression had a significantly negative correlation with HIF-1α expression (figure 5C). Furthermore, the TargetScan (http://www.targetscan.org/), miRanda (http://www.microrna.org/microrna/home.do) and StarBase were used for predicting upstream miRNAs of HIF-1α, and it was revealed that the 3’-UTR of HIF-1α could bind with miR-153-3p (figure 5D). Next, the luciferase reporter assay was used for the predicted relationship verification, and it was discovered that compared with the control group, transfection of miR-153-3p mimics into HXO-RB44 or SO-Rb50 cells significantly decreased the luciferase activity in HIF-1α-wt group, but no significant change was observed in the luciferase activity of HIF-1α-mut group (figure 5E, F). In addition, our data revealed that miR-153-3p overexpression significantly repressed HIF-1α mRNA and protein expressions in HXO-RB44 or SO-Rb50 cells, while anti-miR-153-3p exerted the adverse effects (figure 5G–I). In summary, miR-153-3p specifically modulated HIF-1α expression in RB cells.
KCNQ1OT1 regulated HIF-1α expression in SO-Rb50 and HXO-RB44 cells through miR-153-3p
Considering that miR-153-3p specifically regulates HIF-1α and that KCNQ1OT1 also has a regulatory effect on miR-153-3p expression, we further explored whether KCNQ1OT1 could modulate miR-153-3p-mediated HIF-1α expression, the findings of which displayed that knockdown of KCNQ1OT1 significantly reduced HIF-1α mRNA expression in HXO-RB44 and SO-Rb50 cells, and the transfection with anti-miR-153-3p was able to reverse this trend (figure 6A). Besides, Pearson’s correlation analysis displayed that HIF-1α expression had a positive correlation with KCNQ1OT1 expression in clinical RB tissues (figure 6B). The above findings suggested that KCNQ1OT1 could regulate HIF-1α expression in RB through miR-153-3p.
Discussion
Numerous lncRNAs reportedly have effects on various cellular functions and act as biomarkers or regulators of RB.19 For instance, lncRNA AFAP1-AS1 expression is increased in RB cell lines and tissues, which can facilitate the cell proliferation, migration and invasion20; lncRNA MALAT1 aggravates RB by sponging miR-20b-5p and upregulating STAT3.21 In contrast, lncRNA H19 suppresses RB development via binding with the miR-17-92 cluster so as to upregulate p21 expression, inhibit STAT3 phosphorylation and downregulate Bcl2, Bcl2l1 and Birc5 expressions.22 KCNQ1OT1, an unspliced long non-coding RNA, is the antisense to the potassium voltage-gated channel subfamily Q member 1 (KCNQ1) gene.23 Previous research authenticates that KCNQ1OT1 expression in breast cancer is enhanced, which is responsible for the increase of malignancy.24 Additionally, KCNQ1OT1 is able to promote the growth and metastasis of melanoma.25 It was indicated from the present research that KCNQ1OT1 expression was increased in RB cell lines and tissues, and KCNQ1OT1 knockdown could significantly suppress RB cell proliferation, migration and invasion, and promote apoptosis. Our data suggested that KCNQ1OT1 was a potential therapy target for RB.
Studies unearth that lncRNAs can exert their biological functions as ceRNAs of miRNAs, therefore impacting the expressions of miRNA target genes.26 27 In previous studies, in Parkinson’s disease model, miR-153 protects neurons from cell death by preserving activation of mammalian target of rapamycin and stress-activated protein kinase/Jun amino-terminal kinase signaling pathways and attenuating activation of p38MAPK in the neurotoxin 1-methyl-4-phenyl-pyridinium (MPP(+))-treated neurons.28 Besides, miR-153 is found to be dramatically decreased and hypermethylated in glioblastoma multiforme tissues; moreover, demethylation of glioblastoma cells by 5-aza-2'-deoxycitidine stimulation can reduce the methylation level of miR-153 and make it re-expressed, which indicates that miR-153 serves as a tumor suppressor in glioblastoma.29 Recently, miR-153-3p is reported to regulate progression of ovarian carcinoma in vitro and in vivo by targeting MCL1 gene.30 In this study, we observed that miR-153-5p could suppress the malignant biological behaviors of RB cells via targeting HIF-1α. We also predicted the target miRNAs of KCNQ1OT1 through bioinformatics to explore the probable downstream molecular mechanism by which KCNQ1OT1 participated in the RB progression, and found that three potential binding sites existed between miR-153-3p and KCNQ1OT1, and the luciferase reporter assay and RIP assays confirmed that miR-153-3p was a direct target of KCNQ1OT1 in RB. Moreover, it was discovered that miR-153-3p expression had a negative relationship with KCNQ1OT1 expression in RB tissues. What matters was that the inhibition of miR-153-3p reversed si-KCNQ1OT1-induced effects on RB cell proliferation, migration, invasion and apoptosis. In short, KCNQ1OT1 may function as a molecular sponge for miR-153-3p in RB to repress its expression.
The nutrients and oxygen required for tumor growth depend on the formation of tumor vessels. The formation of new vessels will result in the rapid growth of tumor tissues. When the tumor grows too fast, nutrition is inadequate, or oxygen is deficient, HIF-1α signaling pathway will be activated, which in turn regulates the expression of downstream target genes, such as vascular endothelial growth factor (VEGF), and promotes the angiogenesis.31 In RB, HIF-1α promotes the proliferation, migration and invasion of RB cells via regulating multiple cancer-related genes including matrix metalloproteinases, VEGF, glucose transporter 1, survivin, lncRNA ANRIL and so on.32–34 The present study revealed that in comparison to in normal tissue samples, HIF-1α mRNA expression in RB tissues was upregulated, which was consistent with previous report. Moreover, it was negatively correlated to miR-153-3p expression in RB samples. It was also validated that there exited a binding site between HIF-1α and miR-153-3p. Furthermore, it was discovered that KCNQ1OT1 knockdown could markedly reduce HIF-1α mRNA expression, while miR-153-3p inhibition was able to reverse this trend. The above findings demonstrated that KCNQ1OT1 could modulate HIF-1α expression in RB cells through miR-153-3p.
Collectively, this study, for the first time, demonstrates that KCNQ1OT1 expression is increased in RB cell lines and tissues, and plays a tumor-promoting role in RB via regulating the miR-153-3p/HIF-1α axis. Our study helps clarify the mechanism of RB progression, and provides clues for the diagnosis and treatment of RB.
Footnotes
YW and JW are joint first authors.
Contributors YW and JW conceived and designed the experiments. YW, JW and HH performed the experiments. YW, JW and XL exercised statistical analysis. YW and JW wrote the manuscript.
Funding Innovation and Entrepreneurship Scientific Research Project of Lanzhou (Approval No. 2016-RC-5).
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval The present research was endorsed by the Ethics Committee of the No. 1 People’s Hospital of Lanzhou (Approval No. 202006-1), and conducted strictly following the Declaration of Helsinki.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement The data used to support the findings of this study are available from the corresponding author on request.