The Future of Mesenchymal Stromal Cells in cancer – A Bibliometric Analysis

Mesenchymal stromal/stem cells have garnered much interest in the eld of cancer biology. Bibliometric analysis of a scientic eld has the potential to reveal insights pertaining to the maturity of the eld, the hotspot sources of publication and temporal shifts in research subcategories within the eld. Despite the benets of such an analysis, one has not been conducted to date. The Web of Science™database and VOSviewer software were used to analyse publication and citation data for all publications relating to Mesenchymal Stromal/Stem Cells in oncology. Contributions to the literature were also identied by country, journal and cancer type. Keyword analysis identied trends in developing research categories.


Background
In the 1980's Friedenstein and Owen described a broblastic cell that was isolated from the bone marrow of rodents and formed colonies when cultured in vitro. The described population were a heterogeneous collection of cells whose differentiation could be manipulated in vitro and were labelled "bone marrow osteogenic stem cells" (1,2). In 1991 the term "Mesenchymal Stem Cell' was coined to describe these multipotent progenitors, a term which has persisted (3). Subsequently the International Society for Cellular Therapy (ISCT) recommended an amendment to the nomenclature. In a 2006 position paper the term 'Mesenchymal Stromal Cell' (MSC) was recommended for the description of these cells. The ISCT paper was in response to inter-investigator variability in cell isolation, expansion and characterisation and aimed to standardise scienti c reporting and acknowledge the variability in cells determined by their tissue of origin (4).
Aside from their multipotent differentiation, other functions of MSCs have been uncovered that demonstrate potential for translational application particularly in tissue healing and autoimmune conditions (5)(6)(7). MSCs have also garnered much interest in the eld of cancer research. Authors have demonstrated that MSCs home to the site of tumours and can have a pro-tumorigenic in uence (8).
MSCs native to the cancer's tissue of origin have also been implicated (9,10).
Within the tumour microenvironment (TME) the pro-tumourogenic in uences of MSCs include induction of angiogenesis (11) promotion of cancer cell migration (10) and epithelieal to mesenchymal transition (EMT) (12) the latter of which constitutes enhanced resistance to chemotherapies (12,13). More recently, the immunosuppressive abilities of MSCs have been particularly interesting as they allow for the cancer to evade the immune response of the host (14,15).
Prior to the digital age any given eld of study would command a small number of print publications from which new ndings could be accessed. The coming of the digital age has provided researchers with access to a quantity of data that was heretofore unimaginable. There is now a growing expanse of print and electronic sources. The freedom and speed with which knowledge is disseminated confers an obvious bene t but also presents a challenge. Experimentation is based on a hypothesis, which in turn is generated from robust knowledge of the prior research. In an era of monumental research output, incomplete knowledge of prior research can lead to error in study design (16). Bibliometric analysis, conceived by Paul Otlet in 1934, provides a solution for a robust understanding of the ocean of print and digital data. Bibliometrics focuses on the data related to the reading and writing of books and documents rather than the content which would be the focus of a traditional literature review (17). By utilising the advancements in bibliometric and indexing software, one can access large volumes of publication metadata and draw conclusions about the trends of a particular eld of study, the untapped niches and the maturity of that scienti c eld. Employing bibliometrics, researchers have been able to analyse whole research elds to provide insights in HIV (18), cancer (19) and microRNAs (20) among others.
Despite the bene ts of bibliometric analyses and the growth of the eld of MSCs in oncology, analysis of the topic has not been conducted to date. To conduct such an analysis would highlight areas of the globe producing high quality research, sites of highest productivity, evidence of trends and topics requiring further exploration.
The search terms used were "Mesenchymal Stromal Cells" and "Mesenchymal Stem Cells" with the Boolean operator "OR". Initial results were further re ned using the WoS category "Oncology". Duplicates were identi ed and removed. All languages and years of publication were included, as were all publication types. Data extracted included; Title, Author names, abstract, place of publication, language, year of publication, publishing journal, type of publication, WoS Category, number of citations and H-Index.
All data were extracted on a single day in April 2019 as WoS is an open database which is regularly updated with new publications. The use of a single source bibliometric database was conducted due to its comprehensive indexing of journals, suitability and extensive use for similar analyses in the literature (20,(22)(23)(24).
The description of publishing journals, on their journal website was accessed. The focus of journals as either cancer speci c or general scienti c was determined. Journals were deemed to have a cancer speci c focus if their description included any of the following terms; cancer, tumour, tumor, malignancy, oncology, neoplasia, carcinoma or carcinogenesis and generalised science if they had none of the above terms in their description. Journals were also categorised based on their focus on either basic, translational or clinical research by the same process. Descriptors indicating a basic science focus included the following; basic science, basic biology, bench research, preclinical, laboratory, cell biology or molecular. Descriptors of a translational focus; translation, bench-to-bedside, translational and a clinical focus if; clinical, surgical, clinical trial or clinic were included. Journals that had more than one research focus were recorded in each of the relevant categories. The top 6 most common cancers and top 5 most common causes of cancer death as per the World Health Organisation (25) were recorded among search results.

Data analysis and visualisation
All data were input into Microsoft Excel 2010 © for the generation of graphical data. WoS data tools were used for the generation of several elements of data analysis eg. generating citation reports, H-Indices and WoS categorisation.
The Java program VOSviewer (version 1.6.11) (26) was used to provide a keyword cluster analysis. To perform this a network visualisation was conducted using search result titles', authors, institutions and abstracts input in .txt format. Co-occurrence of key words was performed for all key words that appear in the search results a minimum of 10 times.

Results
In total, our search strategy yielded 9 Finland, Ireland and Denmark while others no longer appear in the top 25 e.g. Japan, the Peoples' Republic of China or Spain (Fig 2A and 2C). Figure 3A shows the top 15 journals in this eld during this period. The top publishing journal in this eld was Stem Cells (1156) followed by Oncotarget (837) and Cancer Research (571). Focusing exclusively on the past 15 years the publishing journals were explored in more detail. In 2004, 38% of the papers investigating MSCs in oncology were published in journals with a cancer speci c focus. From 2004 to 2018 there was a steady increase in publications in cancer speci c journals which was highest in 2018 at 84% of all publications. In total over the 15-year period 74% of all publications occurred in these cancer speci c journals (Fig. 3B). This demonstrates that there is good acceptance of papers on MSCs in oncology among the mainstream cancer research community and that its growth has been considerable between 2004-2018.

Research Focus
To further consider the nature of studies conducted in this area of research, publications were classi ed as basic, translational or clinical research based on the journal they were published in. Figure 3C demonstrates quite clearly that basic science publications are the most prevalent followed by clinical and translational. This trend was consistent over the 15 years represented in the graph. To provide insight into the most commonly researched malignancies in the eld, the 9927 publications were ltered for speci c cancers. This process was limited to the most common cancers (Fig. 4A) and the most common causes of cancer mortality (Fig. 4B), according to the World Health Organisation publications (25). The most common cancer to be published on in this eld was breast cancer (42% of publications), followed by lung (17%) and colorectal (14%) (Fig. 4C). Of these most common cancers, breast accounts for 22% of cases worldwide, lung 22% and colorectal 19% (Fig. 4A). The distribution of mortality is lung 37%, colorectal 18% and breast 13% (Fig. 4B). These results demonstrate that breast cancer is disproportionately represented in the published literature, in respect of its prevalence worldwide and contribution to cancer related mortality.

Keyword analysis
Title, authors, abstracts and keywords were extracted for all of the 9927 publicaions as detailed above and a speci c focus on keyword occurrence was undertaken.
The relevant output is provided by VOSviewer in the form of a bubble plot. Each keyword that achieved the threshold for occurrences can be identi ed as a single bubble. The frequency with which any particular keyword appears corresponds to the size of the respective bubble. Proximity between bubbles is inversely related to the frequency with which they co-occur in the published literature. In line with this, VOSviewer grouped keywords into 3 separate general clusters relating to; characterisation and nomenclature (Fig. 5A red), clinical focus (Fig. 5A blue) and molecular function in cancer biology (Fig. 5A  green). The most prevalent keywords were "Mesenchymal Stem Cell", "Metastasis", "emt" (Epithelial to Mesenchymal Transition) and "Bone Marrow". The VOSviewer output for keyword analysis was also strati ed according to the temporal frequency of keywords within the literature. In Fig. 5B older terms are coloured blue and the most recent terms are coloured yellow. In considering this particular gure there is a temporal shift from keywords relating to isolation and characterisation of MSCs eg. "donor", "mesenchymal stem cell", "bone marrow", to their involvement in TME eg. "e-cadherin", "emt", "microRNA" (Fig. 5B). This nding is consistent with the original efforts by researchers to isolate, identify and culture these cells in vitro until a standardisation of this process was established. After such a point that there was general consensus about the nomenclature and characterisation of these cells, new avenues pertaining to their function within the context of cancer biology was possible. It is worth noting that the cluster pertaining to clinical focus (Fig. 5A blue), has the fewest keywords and some of the least frequently identi ed in the literature. This nding provides an insight into the current state of MSCs in clinical cancer research. The 3 most prominent keywords within the clinical focus cluster are "review", "case" and "concept", leading us to believe that clinical research in this eld is currently in its infancy with much of the published literature being of a theoretical nature, rather than applied clinical research.

Discussion
By utilising the techniques of bibliometric analysis, it is possible to access a large repository of publications and make certain judgements regarding the eld of interest based on bibliometric parameters. The unique strength of bibliometric analysis over a traditional literature review is that the analysis facilitates the input of a far greater number of articles, which in turn constitutes more meaningful recommendations to researchers. From the outset of this paper, the aim of the authors was to make a commentary on the current state of the eld of MSCs in cancer research and provide insights into which research streams are worthy of attention in the future.
Our analysis has included almost 10,000 publications from which the eld of MSCs in oncology has been assessed and the publication data shows considerable growth within the eld over the past 20 years, peaking in 2017. Less publications in 2018 however, does not preclude the possibility that the eld is still undergoing continued growth. It has been noted previously that delayed indexing of published articles on the WoS platform can result in an underrepresentation of the publication statistics for more recent years (20). Another reason for the drop in 2018 can be attributed to the natural progression of the eld commonly seen in bibliometric analyses. As research elds become more mature and publications of a rudimentary nature saturate the eld, research naturally shifts towards more novel niches of investigation. Turning points such as this have been documented in numerous elds of study previously (29,30). When considering the publications from 2018, one must also make a note of nomenclature. Mesenchymal Stromal Cells are closely related to the alternatively named Cancer Associated Fibroblasts (CAFs), both of which are broblastic cells that support cancer proliferation, survival, chemoresistance and immune-evasion in the TME (31). An increase in the use of the "CAF" terminology may also have contributed to the number of MSC publications in 2018.
Development of research elds is said to follow four stages; initially only a small number of contributors publish on a topic until the eld receives wider acceptance and enjoys exponential growth. Upon saturation of the eld, publications plateau at maturity until an ultimate decline in publications (32,33).
Based on the data we have presented and the strong R 2 value of 0.97, the eld of MSCs in cancer research are currently undergoing the exponential phase of growth similar to that seen in other elds of study (33). This identi ed trend in the eld of MSCs in cancer research is encouraging news for prospective cell biologists and cancer researchers with an interest in the TME, as it indicates a continued appetite for this type of research among the scienti c community.
The citation data also aligns itself with the aforementioned 4 phases of research eld maturity. In the early phases, the small number of publications were highly cited as interest in the eld grew, resulting in more publications. The consequence of this high output however corresponded with an exponential decrease in the number of citations per item, the inverse of publication volume. While this citation bias is understandable as the eld matures, researchers need to have an awareness of such phenomena when considering the merit of prior research and should also seek out the high citation articles that provided the foundation for the eld's subsequent exponential growth.
Geographically, this eld receives contribution from many jurisdictions, with over 90 countries contributing. The United States and the Peoples' Republic of China account for a signi cant volume. Both of these countries have been identi ed previously as power houses for publishing research in the eld of cancer and others (34)(35)(36)(37)(38). Despite their high overall output, correction for population has highlighted research hotspots such as Luxembourg, Switzerland and Singapore who lead the world in publications per capita but also have the highest average citations among all countries. In considering collaborative working, these ndings can inform researchers within this eld of study. Particularly, when collaborations are sought within Europe, the impressive performances of Luxembourg and Switzerland make institutions from those countries worthy of research partnerships.
In relation to the journals that publish most proli cally, there is a mix of dedicated cell biology journals such as "Stem Cells" but also those with a broader cancer research focus such as "Oncotarget". Over time, publications on MSCs in oncology were increasingly seen in journals with a deliberate cancer focus (Fig. 3B), this suggests that while publications were originally in journals dedicated to studying cells eg. "Stem Cells", future research will be most appropriately published in cancer journals. This nding is a positive development as it demonstrates a wider acceptance of MSC research in the cancer community rather than exclusively the interest of cell biologists.
As a eld matures, it is expected that publications will shift from preclinical to translational and ultimately clinical. This process has been studied in detail by Weber et al. who describe the challenge of accurately determining what stage a eld is at by bibliometric analysis (39). In our attempt to make this determination about MSCs in cancer research we have taken a similar approach to Narin et al, whereby publishing journals were designated as either basic, translational or clinical and their publication data was subsequently analysed over time (40) (Fig. 3C). Our ndings demonstrate strong predominance of basic science/preclinical research which increased over a 15 year period. One may expect that basic science research would decrease overtime while translational and clinical publications increased. At this point that is not the case and there seems to be a continued appetite for bench research on MSCs in cancer. In the interest of developing future projects this indicates that while, basic science studies are still in demand, the aspiring cancer researcher who incorporates translational or clinical components will ll a relatively unmet need within the literature.
Of the cancers analysed, studies on breast cancer were most frequently published. The proportion of publications relating to breast cancer exceeded both its incidence and rate of mortality relative to the other cancers examined. This nding is not particularly novel as other bibliometric analyses indicate that breast cancer predominates various cancer research elds (41,42) and also enjoys considerable funding (43). For those embarking on research involving MSCs in oncology, the implications of this nding are twofold. Firstly, institutions publishing on breast cancer have an established process and a strong track record of published outputs, making them promising collaborators. Secondly, given the near saturation of publications relating to breast cancer, the lesser studied cancers may contain novel ndings that remain uncovered.
The keyword analysis (Fig. 5A&B) provides an insight into the content of published articles relating to MSCs in cancer research. There are a number of familiar keywords among the more prevalent in the literature, for example "Mesenchymal Stem Cell" and "emt". Epithelial to Mesenchymal Transition (emt) has long been identi ed as a mechanism by which MSCs support tumour progression and cancer cell survival within the TME (44). Furthermore, EMT has previously been identi ed as a research hotspot in cancer research, by bibliometric analysis (45). While its frequency within the literature supports its importance in MSC research and the expertise that exists within the scienti c community, the search for novelty is likely to be found elsewhere within the cluster analysis. The cluster relating to clinical research ( Fig. 5A blue), is noticeably underdeveloped compared to the other two. Prominent keywords such as "case", "concept" and "review" suggest that original, applied MSC research in this area is underrepresented. Consistent with the underdevelopment of translational research, this cluster demonstrates a need for future research on MSCs in oncology that has a translational or clinical component. The cluster analysis also demonstrates the change in keyword prevalence over time within the literature (Fig. 5B). The older terms along this spectrum include "Mesenchymal Stem Cell", "Media", "Osteoblast" and "Bone Marrow" and are almost exclusively situated within the cluster relating to the identi cation and characterisation of these cells. This is not peculiar as earlier research will be expected to focus on identi cation of MSCs in tumours and determining their characteristics. Conversely, the more current keywords include the aforementioned "emt" along with "microRNA", "Metastasis" and "e cadherin" all of which reside within the cluster pertaining to molecular function (Fig. 5A green). The temporal shift towards MSC function in the TME alongside our other data certainly suggest that future directions in the eld will be on the functional elements of MSCs and how these can be manipulated to clinical and therapeutic advantage.

Conclusion
Bibliometric analysis has allowed for the assessment of over 9000 publications relating to the eld of MSCs in cancer research. It has demonstrated exponential growth in the outputs of the eld over the past 2 decades that persists. Analysis of research focuses and publishing countries have highlighted global hotspots of activity but also research quality.