Global Research on Osteosarcoma, 1999–2019: A Bibliometric and Visual Analysis CURRENT STATUS: POSTED

Background Osteosarcoma is a primary malignant bone tumor that occurs in children and adolescents. Increasing numbers of scholars have studied its development and treatment. To fully understand the current status of osteosarcoma research and global trends therein, we performed a bibliometric and visual analysis of osteosarcoma studies published between 1999 and 2019. Methods We searched the Web of Science database for publications on osteosarcoma. The basic characteristics of this sample of publications, such as H indices, annual outputs, languages of publication, and authors, journals, institutions, and countries of origin, were determined. Co-citation, collaboration, and keyword co-occurrence were analyzed using CiteSpace software. Results The sample comprised 16,934 articles. The number of publications increased annually. H indices and total numbers of citations were far higher for articles from the United States than for those from other countries. Among institutions, the largest proportion of articles originated from Shanghai Jiaotong University. R. Gorlick was the author with the highest H index and total number of citations. Oncology Letters published the largest number of articles and Cancer Research was the most frequently cited journal. The five most frequently appearing keywords were “osteosarcoma,” “cancer,” “expression,” “apoptosis,” and “metastasis.” The analysis generated 10 major clusters of keywords and 23 clusters of co-cited references. Conclusions The findings of this study have guiding significance for researchers seeking cooperating institutions and partners for osteosarcoma research, popular journals and important literature in the field, an understanding of the knowledge base for this research, and up-to-date identification of research hotspots and trends. We performed in-depth analysis of these maps and corresponding data

Osteosarcoma (OS) is a primary bone malignancy encountered frequently in children and adolescents, commonly in the metaphyses of long bones (e.g., distal femur, proximal tibia and proximal humerus) [1]. In elderly patients, OS is usually considered to be a secondary tumor attributable to the sarcomatous transformation of Paget's disease or other benign bone lesions [2]. Distant metastasis occurs before the definitive diagnosis of OS in about 15% of patients, and more than 85% of such metastases are in the lung [3]. At present, the clinical treatment of OS is primarily surgical, with adjunctive chemoradiotherapy and biotherapy. In recent years, immunotherapy [4,5], molecular targeted therapy [6,7], and cancer stem cell therapy [8,9] have been used increasingly for the treatment of this disease. Although the 5-year survival rate of patients with OS has increased significantly with the improvement of treatment and clinical management, therapeutic effects remain unsatisfactory in patients with metastatic and recurrent OS [10]. Thus, abundant research has been conducted on the occurrence, development, and treatment of OS, and the body of literature has grown. However, no summary or analysis of the status of global OS research or changes in its focus has been published.
Bibliometrics is an observational method used to assess the status of research and quality of publications on given topics; it can be used to study growth, development, and communication, and to identify the most influential and pioneering research, in particular fields. In the medical field, it has been applied to research on spinal surgery, arthroscopy, and surgical oncology, as well as many other topics [11][12][13]. CiteSpace is a widely used bibliometrics and visual analysis tool developed recently by Chen et al. [14]. It is based mainly on co-citation analysis and pathfinding network algorithms, applied to explore key paths in bodies of literature and to identify points of inflection in the evolution of knowledge on particular topics. It is also used to analyze global research hotspots in given disciplines and explore trends in research development through the application of a series of visualization techniques [15].
In this study, we used the Web of Science Core Collection (WoSCC; http://www.webofknowledge.com) tools and CiteSpace software to analyze the literature on OS published between 1999 and 2019. The aim was to fully understand the current situation and global trends in OS research, to provide a reference and foundation for theory development for subsequent scholars.
2. Materials And Methods 2.1. Data source and retrieval strategy On 19 December 2019, we searched the WoSCC for publications related to OS using the topic (TS) keyword "osteosarcoma*" OR "osteogenic sarcoma*," file type "article," and search period of 1999 to 2019. The search generated 16,934 document records for review.

Bibliometrics and visual analysis
First, we conducted a WoS-based analysis to understand the basic features of the publications in the sample, such as the H index; annual output; all authors, journals, institutions, and countries/regions of origin; and languages of publication. The H index and total number of citations reflect the quality and academic influence of publications at the institution or author level [16].
Then, we performed a visual analysis using the CiteSpace software. This analysis generated visual knowledge networks consisting of nodes (representing, e.g., authors, countries, institutions, or references cited) connected by lines (representing cooperation, co-existence, or co-citation), with representation of temporal patterns and the degree of node centrality, or key role in collaboration and co-citation networks. Network cluster analysis of co-cited references and co-occurrence keywords was performed. The identification of frequently used keywords is often performed to determine directions of development and research foci in a given field. The results of this analysis were manifested as two indicators generated by CiteSpace -the module (Q) value and average silhouette (S) value -which reflect the network structure and cluster clarity, respectively. In general, Q values range from 0 to 1, with larger values indicating better network clustering. Q values > 0.3 indicate a significant network cluster structure. S values range from − 1 to 1, and reflect the homogeneity of network clusters, with values closer to 1 indicating greater homogeneity; S values > 0.5 are considered to reflect reasonable clustering results.
Timeline maps reflect mainly the historical evolution and interrelationship of clusters, and time zone maps show the evolution of research over time, revealing "hotspots" and developmental trends.
Research hotspots in fields with strong temporal characteristics can also be identified by analyzing high-frequency keywords. We performed in-depth analysis of these maps and corresponding data to explain the evolution of OS research and identify hotspots emerging at different times.
The knowledge base is composed of co-cited literature, and the research front is composed of the literature in which knowledge base components are cited. In CiteSpace, we identified clusters, named according to nominal terms extracted from the citing literature, which can be considered to comprise the research frontier. We analyzed the most cited and central articles in each of the top 10 co-citation clusters to determine the knowledge base for each research frontier.

Country-level distribution and cooperation
The articles originated from 120 countries; the 10 countries producing the most publications, with H indices and citation frequencies, are listed in Table 1. Researchers in China published the most articles (n = 4783), followed by those in the United States (n = 4533) and Japan (n = 1695). The United States had the highest H index value and total citation frequency. Figure

Institution-level distribution and cooperation
Articles in the sample originated from 10,247 institutions. Table 2

Author-level distribution and cooperation
In total, 54,423 authors published articles related to OS. Table 3

Journal-level distribution and co-occurrence
In total, 2338 journals published articles related to OS. Table 4

Keyword co-occurrence
The five most frequently occurring keywords were "osteosarcoma," "cancer," "expression," "apoptosis," and "metastasis" (Fig. 6A). Ten keyword clusters were generated, with a Q value of 0.8269 and S value of 0.8009, indicating a significant cluster structure and reasonable clustering results, respectively (Fig. 6B). Table 5 shows the main keywords in each cluster. The three keywords whose use increased most rapidly during the study period were "invasion" (2016-2019), "osteoblast" (1999-2010), and "migration" (2016-2019; Fig. 6C). Figure 6D shows keywords who use expanded rapidly in recent years, including "nanoparticle tumor," "supervisor," and "signaling pathway." Table 6 shows research hotspots according to keyword time zones. Figure 7 shows timeline and time zone maps of keyword co-occurrence.

Reference co-citation
The analysis revealed co-citation of 579 articles. Tables 7 and 8 list the top 10 most cited and most central articles, respectively. Twenty-three clusters representing leading OS research topics, including "multifocal osteosarcoma," "survival," and "osteocalcin," were identified. The generated network structure had a Q value of 0.8905 and S value of 0.5184, indicating a significant cluster structure and reasonable clustering results (Fig. 8).

Research hotspots and trends
The analysis of high-frequency keywords identified 10 clusters. Useful information can be extracted via in-depth interpretation of the data presented in Table 5. For example, the p53 mutation or p53-

Research frontiers and knowledge base
The co-citation analysis yielded 23 clusters representing the OS research frontiers. The most cited and central articles in the top 10 co-citation clusters are described here to characterize the knowledge base for these research frontiers.
A 1997 study [17] described good histological responses to preoperative chemotherapy among patients with nonmetastatic OS, with an 8-year postoperative event-free survival (EFS) rate of 81% and overall survival rate of 87%; those with poor histological responses had an 8-year postoperative EFS rate of 46% and overall survival rate of 52%. The authors concluded that EFS and survival were related directly to the histological response to neoadjuvant chemotherapy in these patients [17].
Bacci et al. [18] reported that age, the serum alkaline phosphatase level, tumor volume, surgical margins, and histological response are of independent prognostic value for nonmetastatic OS of the extremities.

Survival
Bielack et al. [1] found that tumor site and size, primary metastasis, response to chemotherapy, and surgical remission had independent prognostic value for survival in 1702 consecutive patients with newly diagnosed high-grade OS of the trunk or limbs. Meyers et al. [19] showed that the addition of ifosfamide to standard chemotherapy (cisplatin, doxorubicin, and high-dose methotrexate) did not enhance EFS, but that the addition of muramyl tripeptide to chemotherapy might improve EFS, in patients with newly diagnosed OS. Goorin and colleagues [20] reported no advantage of preoperative chemotherapy in terms of the EFS of patients with OS.

Osteocalcin
Ducy et al. [21] identified an osteoblast-specific cis-acting element, termed OSE2, in the osteocalcin promoter and cloned the cDNA encoding Osf2/Cbfa1, the protein that binds to OSE2. They demonstrated that Osf2/Cbfa1 is an osteoblast-specific transcription factor and regulator of osteoblast differentiation [21]. Using a Cbfa1-mutated mouse model, Komori et al. [22] showed that Cbfa1 is essential for osteoblasts and bone formation, and may regulate a variety of genes, including those related to osteocalcin and osteopontin.

Systems biology
Atiye et al. [23] identified OS-related amplicons and reported that 12Q amplicons seem to occur much more frequently than previously assumed. These data are valuable for further electrophoresis research. Using high-resolution array comparative genomic hybridization combined with interphase fluorescence in-situ hybridization, Selvarajah and colleagues [24] characterized the genomic imbalance and chromosomal instability associated with OS, providing important insight into the mechanisms that produce complex genomes in patients with this disease.     Journal-level co-occurrence findings. A. Top 10 journals in terms of co-occurrence counts. B.
Top 10 journals in terms of centrality.

Figure 5
Journal-level co-occurrence findings. A. Top 10 journals in terms of co-occurrence counts. B.
Top 10 journals in terms of centrality.