Bibliometric Analysis of Circular RNA Research on Stem Cells: A Swiftly Expanding and Shifting Focus

Background: The important roles of circular RNA in various aspects of the stem cell eld have been gradually reported. This is the rst time in recent decades that quantitative and qualitative evidence from bibliometrics has been used to conduct an in-depth analysis of circRNA in stem cell research to reveal relative research progress, trends and hotspots. Methods: Publications from 1999 to August 7, 2021 were retrieved from the Web of Science Core Collection (WoSCC). All statistical analyses were performed using VOSviewer, CiteSpace and GraphPad Prism 7. Results: The study shows that the number of papers on circRNA in stem cells has increased signicantly since 2016. Among all types of papers, articles are the most published paper types, followed by reviews. Worldwide, published papers come mostly from China, followed by the USA, Germany, and Italy. However, Israel and Iran ranked rst per million people and per trillion gross domestic product (GDP), respectively. The keyword analysis results were stratied into four clusters by VOSviewer software: Cluster 1, “cancer treatment”; Cluster 2, “function and effect”; Cluster 3, “characteristics” and Cluster 4, “osteogenic differentiation”. Conclusions: Our results suggest that attention should be drawn to the latest hotspots, such as extracellular vesicles, ceRNAs and osteogenesis. In addition, suggestions on research directions and investment trends are provided to researchers and institutions.


Introduction
Circular RNAs (circRNAs) are a novel class of noncoding RNAs with a special covalent loop structure without a 5' cap and 3' tail [1] . CircRNAs are widely present in various organisms, including the most recent common ancestors in fungi, plants, and eukaryotes [2] . Generally, circRNAs are more stable than associated linear mRNAs in vivo. In some cases, the abundance of circular molecules exceeds more than ten times that of associated linear mRNAs [3] . In 2013, Sebastian Memczak et al. [4] detected thousands of well-expressed, stable circRNAs from human, mouse and nematode RNA. Sequence analysis indicated important biological regulatory functions for circRNAs. Since then, there has been an upsurge in studying the biological functions of circRNAs. In the same year, Thomas B Hansen et al. [5] revealed the rst circRNA with a speci c biological function-ciRS7-which contains more than 70 selectively conserved miRNA target sites. Further studies have shown that circRNAs are involved in the occurrence and development of various diseases [6][7][8][9] and play especially important roles in regulating tumor proliferation, migration, drug resistance and stemness [10][11][12][13] .
Stem cells are self-renewing cells that can differentiate into specialized cell types. [14] There are three types of stem cells in the laboratory: embryonic stem cells, adult stem cells and induced pluripotent stem cells. Due to the particularity of their functions, stem cells have long been popular research objects in regenerative medicine and clinical treatment of various diseases. Stemness refers to stem cell-like characteristics, the research related to which is mainly focused on cancer-related elds. For example, tumor stemness refers to cancer stem cells (CSCs) that have stem cell-like characteristics [15] . Studies have demonstrated that tumor stemness does have a prominent effect on regulating tumor proliferation, metastasis and drug resistance [16] . Furthermore, circRNA, as an important regulator of gene expression, has been proven to play an important regulatory role in many elds related to stemness [17] . In terms of cancer, circRNAs can be speci cally expressed in cancer tissues and regulate tumor stemness, proliferation, drug resistance and other characteristics. In terms of stem cell differentiation, Xiao Zheng et al. [18] performed an in-depth review on the essential in uence and internal mechanism of circRNAs involved in stem cell differentiation. In short, the analysis of stem cell growth and differentiation mechanisms and clinical treatment methods from the perspective of circRNA has become a new focus of research in recent years. However, due to the limitations of the search method, evaluation in the eld from a holistic perspective has been di cult to achieve. Therefore, it is necessary to nd a new method to learn more about the research status and hot trends of circRNAs in the eld.
Bibliometrics is an interdisciplinary suject based on published literature and its references through mathematical and statistical methods for quantitative analysis [19] . Research trends on speci c subjects or academic impacts of publications can be assessed and predicted by bibliometric analysis. This article will use bibliometrics to explore the correlation between circRNAs and stem cells and summarize the types of papers and the distribution of authors, publishers and research hot spot trends. The main purpose of this article is to provide suggestions on scienti c research topics, cooperative institutions or submitted journals for related researchers and institutions.

Materials And Methods
Data source and search criteria A comprehensive bibliographic retrieval was accomplished online through the Web of Science Core Collection (WoSCC) on 3 August 2021, which was performed on a single day to avoid daily updating bias since the database was still open.
All the search keywords referred to MeSH terms from PubMed except "stemness", which was irreplaceable but could not be obtained in the MeSH terms. The search terms were as follows: (((circRNA OR circRNAs) OR closed circular RNA) OR circular RNA) OR ciRNA) AND (((stemness OR stem cell) OR stem cells) OR progenitor cell). The time period of article publication was from 1 January 1999 to 3 August 2021. The export record contents were Full Record and Cited References, and the le format was Tab-delimited (Win, UTF-8).

Analysis method
VOSviewer (version 1.6.16.0) and CiteSpace (5.7. R5) were used to analyze the document types, publication years, authors, cocited authors, journal source titles, countries, keywords and research areas to form network maps [20] . The original data extracted from WoSCC were imported into Microsoft Excel 2010 to conduct data aggregation and preliminary analyses. The time trend of the number of publications was graphed with GraphPad Prism 7 software. Considering the differences in aggregate economic volume and population among countries, the latest information on population and gross domestic product (GDP) was obtained from the World Bank website [https://data.worldbank.org].

Paper type and quantity
The research ow chart of our study is shown in Figure 1. A total of 504 records that met our inclusion criteria were identi ed in this study. The records were classi ed as 9 types ( Fig. 2A). Articles (64.22%) accounted for the largest proportion of all records, followed by review articles (32.41%). Therefore, we will focus on these two types of papers in the following analysis. Moreover, we also determined the rst ve research elds with the most published papers in Table   1. Among them, the rst area was the eld of cell biology (27.63%), followed by biochemical molecular biology (20.48%), research experimental medicine (18.49%), oncology (18.29%), and genetics heredity (9.74%). The quantity of papers in other related elds, such as Biotechnology Applied Microbiology and Pharmacology Pharmacy, is also continuing to increase.
Author, organization, country and funding agency A total of 3027 authors were included in this study. Table 2 shows the top ten and cocited authors according to the numbers of papers and citations. China was the main distribution with the most researchers in the related eld, followed by the USA, Germany, and Italy. Among authors in China, and were the main contributors.
In terms of organizations, a total of 637 research institutions were included in this study. The top 3 organizations with the most contributions are displayed in Fig. 2C, which are all from China. The top one is Nanjing Medical University, with 22 related papers, followed by Shanghai Jiao Tong University (21 papers) and Sun Yat-sen University (20 papers). Furthermore, close collaborative connections between various institutions were also identi ed. The Chinese Academy of Science is ranked rst, having 16 links with other institutions, followed by Shanghai Jiao Tong University (12 links) and Sun Yat-sen University (11 links). The data also showed that the collaborations among Chinese universities were very close. Frequent exchanges occurred between Peking University, Tongji University, Central South University and China Medical University. These bibliometric network maps can provide reliable information about in uential institutions, which can help researchers or students be more convinced when making a related decision.
All the papers were from 40 countries. Regarding the research strength of the countries, China (332 papers) was the most productive country, accounting for 63.85%, followed by the USA (86 papers, 16.54%), Germany (24 papers, 4.62%) and Italy (20 papers, 3.85%). Table 3 shows that after adjustment by population, Israel was rst with 0.86 papers per million people. After adjustment by GDP, Iran ranked rst, with 41.76 papers per trillion GDP, followed by China, with 22.55 papers per trillion GDP. Fig. 2D shows the 22 most productive countries in terms of the quantity of papers and the collaboration situation. The color of the node represents a different cluster, and the size of the node represents a different number of papers, with larger nodes meaning more papers. The collaborations among various countries were very close. The connections between Western European countries and the relationship between China and other countries were relatively obvious.
The top 10 funding agencies are shown in Table 4 Table 5. The journal with the most published papers was Molecular Cancer, which published 11 related papers, with 504 citations. Four of the top 10 journals are British journals, ranked 1 and 2. The United States had three journals, followed by Switzerland (2 journals) and Greece (1 journal). The table also shows that the ranking of the number of papers has some disagreement with the ranking of the number of citations. Based on the number of citations and the IF, the authority of the journals Journal of Experimental Clinical Cancer Research and Molecular Therapy Nucleic Acids is relatively high, although they were not ranked in the top three. Furthermore, "citation/N" represents the average citations per paper. Molecular Cancer ranked rst, with 11 papers and 504 citations, meaning that each paper was cited an average of 48.51 times. Table 5 also shows the names, total citations, publishing countries, and IF of the top 10 cocited journals. Nature from the UK ranked number one with 1476 citations and an impact factor of 49.962, followed by Cell from the USA (1330 citations, 41.58 IF). Among these ten journals, 6 were based in the US, and 4 were based in the UK.

Co-occurrence keywords and burst keyword analysis
All the keywords extracted from 505 papers were identi ed and analyzed. Fig. 3A shows the density visualization based on the keywords. The color of each point in a map depends on the density of items at that point. The warm red color represents hot and important areas, and the cool blue color represents cool areas. According to the order of occurrence of the keywords, the main hot keywords are circular RNA, expression, stem cells, proliferation, identi cation, cancer, differentiation, microRNA, etc. (similar keywords removed).  To make the result more convincing, the analysis of keyword co-occurrence in CiteSpace was also conducted. The visualization result is shown in Fig. 4A. Each node in the gure represents one keyword, and the node size indicates the frequency of the keywords; the larger the circle is, the higher the frequency. Careful comparison shows that the result is highly similar to the one we obtained in VOSviewer. Fig. 4B clearly shows the top 8 clusters from all the keywords: circular RNA, neural stem cell, miRNAs, lncRNAs, etc. Furthermore, Fig. 4C and Fig. 4D can make it more intuitive to see the research status of each cluster at different time nodes.

Discussion
This article conducts a preliminary analysis of keywords in the selected literature at the intersection of circRNAs and stem cells. Eighty keywords were divided into 4 clusters by the network visualization of VOSviewer. Cluster 1 had a close relationship with Cluster 2, mainly keywords related to "tumor" and "treatment" (including speci c tumor name), such as tumor suppressors and biomarkers. Cluster 3 is mainly related to the "characteristics" of circRNA and stem cells themselves, such as biogenesis, transcripts, and noncoding RNA. Cluster 4 mainly contains keywords related to the direction of "osteogenesis differentiation".
In recent years, research hot spots at the intersection of circRNAs and stem cells have generally shifted from Cluster 3 to Clusters 1 and 2. From the result of our overlay visualization, researchers focused on the source, composition, classi cation in vivo and differentiation and expression of embryonic stem cells. However, in the past two years, the main research focus has shifted to the elds of tumorigenesis, development and clinical treatment. According to our keyword co-occurrence analysis, both circRNAs and stem cells are closely related to the proliferation, metastasis and drug resistance of tumors. Among the keywords, "circRNA" and "biomarkers" have a link strength of 15, which indicates that circRNA has a promising future in the direction of emerging biological targeting drugs.
Furthermore, the connection between circRNA and stem cells is very close (link strength: 40). In addition, the study of circRNA involvement in stem cell differentiation and expression is also a new hot spot.
Cancer (here generally referring to all malignant tumors) is the main cause of death in all countries around the world, and it is also one of the high-frequency keywords screened in this study. Among all types of cancers, cancers of the female breast, lung, colorectum, and prostate comprised 62% of the total cancer burden in medium to very high HDI (Human Development Index) areas [21] . The numbers of cancer cases and deaths are expected to increase rapidly as populations grow, age, and adopt lifestyle behaviors that increase cancer risk [22] . Therefore, it is particularly important to nd new methods for the prevention, early detection, and effective treatment of the abovementioned high-risk cancers. CircRNAs have shown great potential in the treatment and prognosis of cancer because they are widely expressed in eukaryotes and highly stable [22] . On the one hand, circRNAs can be used as a new type of biomarker to monitor the progress of speci c cancers; on the other hand, they can also regulate cancer proliferation, metastasis or drug resistance through ceRNA network mechanisms. Similarly, the heterogeneity and plasticity of stem cells also make them play an important role in tumor development. They can produce organoid structures to mimic human disease models [23] . In addition, transgenic stem cell transplantation is also a potential cancer treatment [24] . From our research, it is intuitive that the keyword "cancer" has a frequency of 57 and a total connection strength of 289. It is closely related to both circular RNA and stem cells, which is more in line with the current research status. This also shows that circRNAs and stem cells have good research prospects in cancer-related elds, which are worthy of further research and exploration.
In addition to analyzing keywords, we also conducted a further analysis of article types, research institutions, published journals, and the distribution of countries where articles were published. According to the results, the proportion of reviews is relatively high (32.41%), which re ects that researchers may have made many sorts and summaries on the basis of existing research to assist further studies. This proves the necessity of bibliometric analysis because great importance is attached to the summary of this eld. The high number of reviews is a major advantage for new researchers, as they can gain a more comprehensive understanding of the directions of research. However, it also means new research is scarce, and we need to bring in talent for further and in-depth research.
Currently, the most in uential institutions in related elds in China are Nanjing Medical University, Shanghai Jiao Tong University, and Sun Yat-sen University, which are mainly concentrated in southeastern China. The information is very useful for students and scholars interested in this eld so that they can choose where to study or become employed at home and abroad.
For the published sources, Molecular Cancer from the UK had published the most papers (11 papers), with the highest number of citations (504) and highest impact factor (27.401). These results indicate that researchers must have a good judgment of the journals' preferences when submitting related articles. Further studies can be guided for submission to these journals.
In regard to analysis by country, China and the USA were the most productive countries, and the H-index and total citations of China were both ranked rst. However, when considering GDP, Israel and Iran's research strengths were relatively high. This indicates that developing countries still have a certain in uence in the international arena. Developing countries should take into account the development of economic and scienti c research capabilities.

Conclusion
The number of publications regarding circRNA in stem cell research has been constantly growing since 2016, and we predict that it will continue to rise. Mainland China and the USA were found to be the most productive regions. The most popular keywords at this stage mainly focus on tumorigenesis and clinical treatment, meaning that funding agencies could increase investments in exploring the clinical diagnosis and treatment potential of circRNAs in stem cell research. It is also recommended that more attention be paid to the latest promising hotspots, such as "expression", "proliferation", "identi cation", "cancer", "differentiation", and "microRNA".
Declarations ·Ethics approval and consent to participate Table 1 The top 10 research area for circRNA in stem cell (n%).  Table 2 The top 10 authors and co-cited authors for circRNA in stem cell research.  Table 3 The top 10 countries for circRNA in stem cell research (n%).  Table 4 The top 10 funding agencies for circRNA in stem cell research (n%).  Table 5 The top 10 journals and co-cited journals for circRNA in stem cell research. Figure 1 Flow chart of the research design and the main results.  is based on the occurrence frequency of the corresponding keywords, and the node color in (B) represents the average publication year. Keywords in blue are presented earlier than those in yellow. In (D), the red line in the gure represents the time period with the strongest citation bursts.