Research trends of uveal melanoma:a bibliometric study and visualization analysis

Hao Wang,Li-Xin Ke,Cun-Cun Lu,Wang Li,Ya-Min Zhao

1Ophthalmology Department,The Second Hospital,Lanzhou University,Lanzhou 730000,China;2The First Affiliated Hospital,Sun Yat-sen University,Guangzhou 510080,China;3School of Basic Medical Sciences,Lanzhou University,Lanzhou 730000,China;4Institute of Basic Research in Clinical Medicine,China Academy of Chinese Medical Sciences,Beijing 100700,China;5The First Hospital,Lanzhou University,Lanzhou 730000,China;6School of Public Health,Lanzhou University,Lanzhou 730000,China.

Background:Uveal melanoma is the most prevalent intraocular malignancy.In preceding decades,many studies have been published on uveal melanoma.However,so far,uveal melanoma-related bibliometric studies have not been reported.Methods:Uveal melanoma-related articles and reviews published between 2005 and 2019 were retrieved in the Web of Science Core Collection on March 15,2020.Three bibliometric tools (HistCite,VOSviewer,and CiteSpace) were employed to conduct the current study.Results:A total of 3,404 publications related to uveal melanoma were identified,involving 3015 articles (88.57%) and 389 reviews (11.43%) with 65,429 co-cited references in 9 languages,which were published by 2,875 institutions in 66 countries/regions.The United States contributed to most publications (n = 1427,41.92%),and the Thomas Jefferson University was involved in most of the studies (n = 160,5.56%).Investigative Ophthalmology & Visual Science published the majority of the papers (n= 175,5.14%),whereas Ophthalmology received the most co-citations (n = 7050).Shields CL owned the most publications and co-citations (n = 122 and 2151,respectively).Gene and molecular biology aspects,prevalence,the main prognosis factor,and treatment were the intellectual foundations of uveal melanoma research.In the field of uveal melanoma,emerging hotspots of uveal melanoma include epidemiologic characteristics,prognosis factors,mechanisms of uveal melanoma development,the use of novel predictive tools,and clinical applications of novel targeted drugs.Conclusion:This first bibliometric study focused on the integral tendency of the past 15 years,identified landmark items,and revealed current research hotspots in the field of uveal melanoma,which will provide references for clinicians and researchers,as well as an example of visualization analysis to explore research hotspots in other fields.

Key words:Bibliometric,Uveal melanoma,HistCite,VOSviewer,CiteSpace

Background

Uveal melanoma (UM),which threatens patients with painful eyes,visual handicaps,facial disfigurement,and untimely death from metastatic diseases [1],is regarded as the most prevalent type of primary intraocular malignancy in adults.UM accounts for approximately five percent of melanoma [2],the remaining include skin melanoma (95.0%),genital tract melanoma (0.6%),and oral cavity melanoma (0.1%).However,the mechanisms underlying UM differ from those of cutaneous melanoma,featuring specific chromosomal changes and gene mutations [3].The choroid is the most common site of UM (85%),followed by the ciliary body and the iris (5-8% and 3-5%,respectively) [4].Globally,UM affects approximately 4.3 people per million individuals [1].The estimated prevalence of UM changes with race and the living area.For example,the prevalence is 6 per million in Caucasians [5],and in the United States,the prevalence is 4.9-5.2 cases per million individuals [6].Interestingly,in Europe,the incidence varies from greater than 8 per million individuals in the north to no more than 2 per million individuals in the south [7,8].In the past three decades,ongoing and significant advances have been made to better understand and treat UM.However,no obvious decrease in the incidence of UM was observed [1,5].Although UM has a relatively high overall survival(around eighty percent at five years),the prognosis remains poor when metastasis develops [4,9].Metastasis mostly occurs via hematogenous dissemination [10] and mainly occurs in the liver (89%),lungs (29%),and bones (17%) [11].However,at the initial diagnosis,only 2-4% of all patients presented with metastasis [12].Furthermore,for patients with liver metastatic diseases,the median survival is approximately 2-3 months [13].In addition,UM is regarded as a systemic disease [14].In general,primary UM is treated by surgery (enucleation or globepreserving) or Iodine-125 (125I) brachytherapy,however,survival outcomes are similar [15].At present,there is still a need to monitor treatment selection continuously,and develop more effective treatment strategies in the area of UM [16].To guide clinical practice and future studies,there is an unmet need for systematically analyzing the evidence landscape,and emerging trends for UM research.

Bibliometric analysis (also known as scientometrics)is a type of systematic analysis using mathematical and statistical methods [17].In recent years,it has become increasingly ubiquitous,and has been frequently utilized by policymakers,researchers,individual scientists,and department heads [18-20].Bibliometric analysis assesses different items (institutions,countries/regions,journals,and authors,etc.) to conduct an in-depth evaluation of the metrological characteristic structure and predict the forthcoming direction of a particular research or disease area [19,21,22].Moreover,using visualization,available information can also be excavated,intuitively displayed [20],and provide novel ideas for future research [23].For example,Tao et al.[22] employed this method to study scoliosis and found that “the development of various new orthopedic plants,artificial intelligence diagnosis,and genetic studies will be future hotspots” in the field of scoliosis.For bibliometric analysis,many tools are available,including CiteSpace [24],VOSviewer [25],and HistCite [26],etc.Since preceding decades,many UM-related studies issued were performed.However,in the field of UM,no bibliometric study has been reported.To fill this important void,outline the research foundations and trends,and provide methodological cases for other research fields,this bibliometric study was performed.

Methods

Data sources and search strategy

Because the Web of Science Core Collection (WoSCC)is widely used as the primary database for bibliometric studies [27],we searched the WoSCC on March 15,2020 at the Lanzhou University to collect UM-related studies that were published between 2005 and 2019.Databases were restricted to the Social Science Citation Index and the Science Citation Index Expanded.Publication types were confined to “article” and“review”.The search strategy was as follows:topics ="uveal melanoma" OR "uveal melanomas" OR"choroidal melanoma" OR "choroidal melanomas" OR"ciliary body melanoma" OR "ciliary body melanomas"OR "ciliochoroidal melanoma" OR "ciliochoroidal melanomas" OR "iridociliary melanoma" OR"iridociliary melanomas" OR "iris melanoma" OR "iris melanomas".All available data were retrieved from the WoSCC on March 15,2020,and were further analyzed using bibliometric tools.

Statistical analysis

The number of annual publications,publication languages and study types of UM research were identified using HistCite (12.03.07) [26].Next,VOSviewer (1.6.13) [25] was employed to confirm countries/regions,institutions,journals and co-cited journals,authors and co-cited authors,as well as cocited references.In addition,this tool was used to visualize the relationships among these items in network maps [21,25].In the network maps,nodes represented items,including countries/regions,institutions,and journals,etc.The nodes would enlarge when the number or co-occurrence frequencies of elements increased.Links between nodes represented co-occurrence relationships,and the sizes of links represented co-occurrence frequencies of 2 nodes [28].The settings for VOS viewer were as follows:counting method (full counting),thresholds (T) of items(countries/regions,institutions,journals,authors,and references) were adopted based on special conditions.Furthermore,CiteSpace (5.6.R2) [24] was employed to create a dual-map overlay of journals and to identify references representing emerging information(references with strong citation burstness).The parameters of CiteSpace were as follows:link retaining factor = 3,look back years = -1,e for top N (e = 2),time span (2005-2019),years per slice [1],links (strength:cosine,scope:within slices),selection criteria (top 50),and minimum duration (MD = 5).For data management and issuing trend analysis,Microsoft Office Excel 2019(Microsoft Corporation,Redmond,WA,United States)was chosen.The binomial model f (x) = ax2+ bx + c was used to estimate the quantity of publications in 2020.Variable x means the year of issuance,and f (x) means the number of published studies.The 2018 Journal Citation Reports (Clarivate Analytics,Philadelphia,USA) was chosen as the source of the impact factor (IF).

Results

Analysis of the annual growth trend of publications

From 2005 to 2019,there were a total of 3404 UMrelated publications,involving 3,015 articles (88.57%)and 389 reviews (11.43%).These were published in nine languages,including 3,277 (96.27%) in English,70(2.06%) in German,41 (1.20%) in French,7 (0.21%) in Portuguese,3 (0.09%) in Polish,2 in Hungarian or Italian (0.06%,respectively),and 1 in Korean or Spanish (0.03%,respectively).Figure 1A displays the yearly output of UM research between 2005 and 2019 with an increasing trend.Overall,the number of annual publications was greater than 150.In addition,the number of publications in 2017 (n = 304,8.93%) was nearly twice that in 2009 (n = 157,4.61%).Between 2017 and 2019,UM research reached a peak with a maximum number of studies in 2019 (n = 335,9.84%).As is shown in Figure 1B,a clear correlation was observed between the year and the number of studies(the coefficient of determination (R2) = 0.9384).Using the binomial curve-fitting procedure,we forecasted that the number of publications will be around 358 in the year 2020.

Analysis of countries/regions

Overall,66 countries/regions contributed to the publications on UM.The top 10 countries/regions participating in UM research are shown in Table 1.Most papers were from institutions within the United States(n = 1,427,41.92%),followed by Germany (n = 376,11.05%),China (n = 311,9.14%),England (n = 279,8.20%),and Italy (n = 228,6.70%).Six out of the top 10 countries/regions are distributed in Europe.Austria(n = 97,2.85%) was the only country with a publication number less than 100.To construct the cooperation network of countries/regions,we incorporated 18 countries/regions with more than/equal to 50 studies (T= 50) (Figure 2).The United States,Germany,and China had larger sizes of nodes in the network map,indicating more publications.Moreover,there was active cooperation between countries/regions.For example,the United States collaborated with Germany,China,England,and Canada,etc.Germany had cooperation with Austria,the United States,France,and England,etc.

Analysis of institutions

The number of institutions that were involved in the 3,404 papers was 2,875.The top 10 institutions are shown in Table 1,and located in four countries/regions,among which 3/5 in the United States.Most papers were from Thomas Jefferson University published the most papers (n = 160,4.70%),followed by Leiden University(n = 123,3.61%),Memorial Sloan-Kettering Cancer Center (n = 93,2.73%),University of Liverpool (n = 85,2.50%),and University of California,San Francisco (n= 83,2.44%).Institutional collaborations were visualized by analyzing 19 institutions with over/equal to 45 publications (T = 45).Figure 3 presents the largest sub-network.In this network map,the sizes of nodes of Thomas Jefferson University,Leiden University and Memorial Sloan-Kettering Cancer Center were larger,implying more published studies.Close collaboration was observed between these institutions.For example,University of Liverpool collaborated with Royal Liverpool University Hospital and University of California,San Francisco.Memorial Sloan-Kettering Cancer Center cooperated with the University of Sydney,Duisburg-Essen University,and University of California,San Francisco.

Analysis of journals and co-cited academic journals

A total of 3404 papers regarding UM were published in 680 journals.In the top 10 journals,a total of 880(25.85%) studies were published,and three journals(InvestigativeOphthalmology&VisualScience,British JournalofOphthalmology,andMelanomaResearch)contributed to at least 100 studies (n = 175,114,and 101,respectively) (Table 2).The IF (2018) of these journals ranged from 7.732 (Ophthalmology) to 2.25(GraefesArchiveforClinicalandExperimental Ophthalmology),with an average IF of 3.943.All the top 10 journals were from either the United States (n =8) or England (n = 2).In addition,70% of the top 10 journals ranked in Quartile in Category 1,whereas others ranked in Quartile in Category 2.To visualize the citation relationship between these journals,16 journals with a publication number over/equal to 36 (T = 36)were evaluated.InvestigativeOphthalmology&Visual Science,BritishJournalofOphthalmology,andMelanomaResearchhad larger nodes sizes because of more publications (Figure 4A).Between these journals,there was close citation relationship.For instance,InvestigativeOphthalmology&VisualSciencecooperated withOphthalmology,MelanomaResearch,ClinicalcancerresearchandCancers,etc.Ophthalmologycooperated withRetina-thejournalof retinal,BritishJournalofOphthalmology,American JournalofOphthalmology,andInvestigative Ophthalmology&VisualScience,etc.

Figure 1 Annual output of UM research (A) and the binomial curve fitting of publication growth for UM research (B).UM,uveal melanoma.

Co-cited journals can be defined as journals that are usually cited together in other studies [21,25].Taken together,7,246 academic journals were co-cited.Each of the top 10 co-cited journals,except forNature(England),is from the United States (Table 2),and each had more than 2200 co-citations.Ophthalmology(n =7,050),InvestigativeOphthalmology&VisualScience(n = 5,511),andArchivesofOphthalmology(n = 5,186)were the top three co-cited journals.The IF (2018) forthese journals ranged from 43.07 (Nature) to 3.615(BritishJournalofOphthalmology),with an average IF of 12.06.Seven journals had IF of higher than 5.00.All co-cited journals ranked in Q1.Nineteen co-cited journals had greater than/equal to 1,250 co-citations (T= 1,250) and were analyzed to create the co-citation network map of the co-cited journals.As shown in Figure 4B,OphthalmologyandInvestigativeOphthalmology&VisualSciencehad larger nodes sizes because of the higher number of co-citations.Ophthalmologyhad co-cited relationships withBritish JournalofOphthalmology,InvestigativeOphthalmology&VisualScience,AmericanJournalof Ophthalmology,andArchivesofOphthalmology.Cancerresearchhad co-cited relationships withNature,ClinicalCancerResearchandInvestigative Ophthalmology&VisualScience.Furthermore,InvestigativeOphthalmology&VisualSciencehad adjacent co-cited relationships withOphthalmologyandCancerResearch,etc.ArchivesofOphthalmologyhad close co-cited relationships withBritishJournalof OphthalmologyandAmericanJournalof Ophthalmology,etc.

Table 1 The top 10 countries and institutions related to UM research

Figure 2 The network map of countries/regions (T = 50) related to UM research.UM,uveal melanoma.

Figure 3 The network map of institutions for UM research (T = 45).UM,uveal melanoma.

Table 2 The top 10 journals and co-cited journals related to UM research

Figure 4 The network map of academic journals (A,T = 36) and co-cited academic journals (B,T = 1,250) for UM research.UM,uveal melanoma.

Figure 5 shows the dual-map overlay of journals.The nodes on the left represent citing journals,and the nodes on the right nodes represent cited journals.Labels above the nodes display different fields.The direction of the paths between the nodes is from right to left.The paths can be viewed as the connections of citations [21,25].In this map,4 chief paths were identified,involving 1 orange path,2 green paths,and 1 pink path.The orange path indicated that studies of Molecular/Biology/Immunology journals cited from publications in Molecular/Biology/Genetics journals.Studies of Molecular/Biology/Genetics journals and Health/Nursing/Medicine journals were citations of Medicine/Medical/Clinical journals in 2 green ties.Importantly,we found that UM research published in Neurology/Sports/Ophthalmology fields usually cited papers published in Molecular/Biology/Genetics fields in the pink path.

Analysis of authors and co-cited authors

In total,there were 13,952 authors.Among the authors who ranked in the top 10 number of publications (Table 3),each published more than 35 papers,and Shields CL ranked first (n = 122,3.58%),followed by Shields JA (n= 94,2.76%),Jager MJ (n = 78,2.29%),Harbour JW (n= 57,1.67%),and Coupland SE (n = 53,1.56%).The publication numbers of the remaining 5 authors were between 37 and 49.Figure 6A outlines the largest subnetwork map of the collaboration among productive authors by analyzing 25 authors with greater than/equal to 25 papers (T = 25).The sizes of nodes of Shields CL and Shields JA were larger because of their higher number of publications,and a close research collaboration existed between them.

Figure 5 The dual-map overlay of journals related to UM research.UM,uveal melanoma.

Figure 6 The network map of authors (A,T = 25) and co-cited authors (B,T = 315) related to UM research.UM,uveal melanoma.

If different authors are cited by other researchers at the same time,these authors can be defined as co-cited authors [21,25].A total of 44,095 co-cited authors were obtained.As is shown in Table 3,Shields CL (n =2,151),Singh AD (n = 1,288),Damato B (n = 1,257),Onken MD (n = 957),Van Raamsdonk CD (n = 921),and Harbour JW (n = 911) received the top six most cocitations,and other authors were co-cited less than 900 times.We brought 19 authors with more than/equal to 315 co-cited times (T = 315) into the visualization of cocited authors (Figure 6B).Shields CL had the largest size of nodes due to the highest number of co-citations,and had active co-cited relationships with Shields JA,Singh AD,Damato B,and Finger PT,etc.

Analysis of co-cited references

Co-cited references are defined as references co-cited in a set of publications [21,25].A total of 3,404 publications received 65,429 co-cited references,and Table 4 presents the top 10 co-cited references.Among them,“Van Raamsdonk CD,2010,NewEngland JournalofMedicine[29]” was co-cited the most (n =437),followed by “Van Raamsdonk CD,2009,Nature[30]” (n = 431),and “Harbour JW,2010,Science[31]”(n = 382),etc.Twelve references that were co-cited for at least 155 times (T = 155) were used for the visualization of co-cited publications.As shown in Figure 7,“Van Raamsdonk CD,2010,NewEnglandJournalofMedicine[29]”;“Van Raamsdonk CD,2009,Nature[30]”,and “Harbour JW,2010,Science[31]”received larger nodes sizes.Notably,2 studies [29,30]with a larger number of co-citations shared the same first author,namely Van Raamsdonk CD.The co-cited publications had close co-cited relationships,for example,“Van Raamsdonk CD,2010,NewEngland JournalofMedicine[29]” had co-cited relationships with “Harbour JW,2010,Science[31]”,and “Van Raamsdonk CD,2009,Nature[30]”,whereas “Onken MD,2004,CancerResearch[32]” had co-cited relationships with “Prescher G,1996,Lancet[33]”,“Harbour JW,2010,Science[31]”,and “Kujala E,2003,Investigativeophthalmology&visualscience[12]”,etc.

Analysis of references with citation burstness

Studies with citation burstness received the attention of researchers in a specific field for a period of time,and can characterize hotspots of a field during an interval of time [21,24].We set the MD of burstness for 5 years using CiteSpace and 42 references (Figure 8) with the strongest citation burstness were identified.The straight line is composed of red or green bars,representing the time interval and the red bar also represents citation burstness [21].In brief,the burstness strength of the top 42 references ranged from 4.6488 [34,35] to 20.9983[36].“Kath R,1993,Cancer” [36] received the strongest burstness strength,followed by “Chapman PB,2011,NewEnglandJournalofMedicine” [37],and“Martin M,2013,NatureGenetics” [38],etc.The burstness strength was 20.9983,20.7420,and 19.0877,respectively.Among the top 42 references,the endurance strength was between 5 to 8 years,and“Scholes AGM,2003,InvestigativeOphthalmology&VisualScience[39]” and “Maniotis AJ,1999,American JournalofPathology[40]” had the longest endurance strength (8 years,both from 2005 to 2012),followed by“Egan KM,1988,SurveyOpthalmology” (7 years,from 2005 to 2011).The citation burstness of nine references ended after 2015 or later,including “Chang AE,1998,Cancer [41]”,“Chapman PB,2011,NewEngland JournalofMedicine[37]”,“Wiesner T,2011,Nature Genetics[42]”,“Damato B,2010,ClinicalCancer Research[43]”,“Hodi FS,2010,NewEnglandJournal ofMedicine[44]”,“Abdel-Rahman MH,2011,Journal ofMedicalGenetics[45]”,“Martin M,2013,Nature Genetics[38]”,“Harbour JW,2013,NatureGenetics[46]” and “Onken MD,2012,Ophthalmology[47]”.Especially,the last three references [38,46,47] were detected with ending citation burstness after 2019.

Table 3 The top 10 authors and co-cited authors related to UM research

Table 4 The top 10 co-cited references related to UM research

Figure 7 The network map of co-citation references for UM research (T = 155).UM,uveal melanoma.

Discussion

General characteristics

Figure 8 Top 42 references with strong citation burstness (MD = 5).MD,minimum duration.

In our study,the bibliometric review related to UM was represented.In general,in the past 15 years,2,875 institutions from 66 countries/regions published a total of 3,404 papers in 680 peer-reviewed journals with 65,429 co-cited references in nine languages.The results of this bibliometric analysis will provide references for clinicians and researchers in the field of UM.Popularity in a certain field can be reflected,in part,by the number of publications [21].In our study,the number of UM publications showed an increase from 157 to 335 between 2005 and 2019,thereby indicating that UM received significant attention and had great potential for development.This was also supported by our binomial fitting forecasting analysis,which showed that the number of UM studies would be about 358 in 2020.Regarding countries/regions,on the one hand,the United States was in an advanced position in the field of UM research and played leading roles in producing and publishing scientific profiles.On the other hand,there were no developing countries among the top 10 countries,except for China.This indicated that,in general,in the field of UM,developed countries/regions played leading roles,and had close collaborations with other countries/regions,such as Germany,China,and England,etc.and the research capacity of developing countries/regions remained relatively weak.However,China,as a developing country,showed a good trend in development.The possible reasons why these developed countries/regions contributed to more publications may be as follows:first,these countries/regions are usually considered as the world-leading engines in medical concepts and techniques [21].Second,scientific publications are regarded as one of the components of the knowledgebased economy,and the economy of these developed countries/regions is better when compared to that of developing countries [48].Third,there is sufficient financial support in these developed countries/regions and abundant expenditure funds for scientific research plays a positive role in research productivity and quality[49].Finally,the risk of UM is higher in these developed countries/regions,since most people in these countries/regions are Caucasians and the morbidity of UM in some of these countries/regions is higher when compared to the global average [5-8].These reasons may also explain,in part,why active collaboration could be observed within the top 10 countries,especially among developed countries.Moreover,active and extensive collaboration may contribute to productivity and research quality,therefore,developing countries could start to collaborate with developed countries.The following institutions with over 90 publications,ranked the top 3:Thomas Jefferson University published the most publications,followed by Leiden University,and Memorial Sloan-Kettering Cancer Center,thereby indicating their significant contribution to UM research.In addition,we found that some institutions not only had more publications,but also had close collaborations,such as University of Liverpool,Royal Liverpool University Hospital,and University of California,San Francisco.To improve research ability and academic levels,researchers can develop broad cooperation with these institutions.It should be noted that all top 10 institutions were from developed countries/regions (3/5 of them from the United States),which,to some extent,supported our findings that developed countries/regions played leading roles in the UM research field.

InvestigativeOphthalmology&VisualSciencewas the most productive journal,andOphthalmologyhad the most co-citations.Among these top 10 co-cited journals,journals which were co-cited more frequently,in general,had higher IF,butNature(IF = 43.07) only ranked ninth in the top 10 co-cited journals.One of the reasons might be that UM papers published inNaturewere fewer than those in ophthalmology or oncology journals,however,Naturewas still in the top 10 co-cited journals.This phenomenon also showed the high value of the studies regarding UM published inNature.Five journals (InvestigativeOphthalmology&Visual Science,BritishJournalofOphthalmology,Ophthalmology,AmericanJournalofOphthalmology,andInternationalJournalofRadiationOncology BiologyPhysics) ranked in the top 10 in productive and co-cited journals,therefore,these journals can be considered principal journals in the UM domain.These five journals can be sources of references when writing papers,and can be target journals for submission.Moreover,the dual-map overlay of journal analysis represented the field distribution of UM studies,which could provide indispensable references for beginners to conduct UM-related studies.Five authors (Shields CL,Harbour JW,Singh AD,Finger PT,and Damato B) were not only the top 10 productive authors but also the top 10 co-cited authors,thus,these five authors may play important roles in UM research,and could potentially be collaborative partners.Noticeably,Shields CL was the author with the most publications and co-citations,who had twice as many co-citations when compared to the second most co-cited author,Singh AD.This indicated the most pivotal position of Shields CL in the UM field,and more attention should be given to his findings.There seemed to be sparse collaboration network among fruitful authors,which indicated that authors needed to strengthen the UM-related research partnership.In addition,both Singh AD and Harbour JW were the top 10 productive authors,the top 10 cocited authors,and the first authors of three references[31,50,51] in the top 10 co-cited references in the following section.

Intellectual foundations

Co-cited references reveal how frequently 2 studies are quoted together by other publications.In addition,they can be considered as the intellectual foundations in a particular area [24].Thus,we highlighted the top ten cocited papers to establish UM-related intellectual foundations.In 2010 and 2009,Van Raamsdonk CD et al.published the first and second most co-cited studies inNewEnglandJournalofMedicine[29] andNature[30].In the former,the authors concluded that the constitutive activation of the pathway,including guanine nucleotide-binding protein G (q) subunit alpha(GNAQ) and guanine nucleotide-binding protein subunit alpha-11 (GNA11),mainly contributed to the progress of UM.In the latter,it was found that the heterotrimericGNAQwas a potential target in UM.The third co-cited publication was published in 2010 by Harbour JW et al.who discovered the somatic mutations ofBRCA1-associated protein-1 (BAP1) in UM metastasis,which implied theBAP1gene might be a relevant and useful target for treatment and diagnosis[31].In 2003,InvestigativeOphthalmology&Visual

Scienceissued the fourth most co-cited study [12].The authors,Kujala E et al.reported that UM metastasis was the main reason for death.The fifth most co-cited paper was published in 2011 by Singh AD et al.[50].In this systematic review of the Surveillance,Epidemiology,and End Results program database,it was found that UM incidence in the United States was about 5.1 per million and did not change from 1973 to 2008.Furthermore,the survival rate did not improve during this time-period,despite a decreasing trend in radiation and other eye-sparing treatments,which gradually focused on by the Collaborative Ocular Melanoma Study (COMS) centers in the treatment of smaller tumors.In 1996,Prescher G et al.published the sixth most co-cited paper in theLancetwith 313 co-citations[33].In this study,it was shown that monosomy 3 testing effectively predicted overall and relapse-free survival,which may play a significant role in the chromosome and gene detection of UM metastasis.Subsequently,in 2004,the seventh most commonly cocited study was published inCancerResearchby Onken MD et al.[32].In this study,it was suggested that primary UM could be clustered into two class (low and high risk of metastasis) based on the gene expression profile (GEP) (downregulated genes on chromosome 3 and upregulated genes on chromosome 8q).This molecular classification of UM could strongly predict metastatic death.The eighth most co-cited study (the No.18 COMS report) was published by Diener-West M et al.in 2001 [52].In the COMS,patients with choroidal melanoma were randomly assigned to the enucleation group or the125I brachytherapy group to compare these two treatment strategies.The authors did not find significant survival differences between these 2 treatment groups.In 2003,the ninth most co-cited study was issued [51].This study was an early version of the fifth most co-cited paper [50] and shared the same first author and journal (Singh AD andOphthalmology),with a consistent purpose but different time periods.The authors found that the incidence of UM in the United States was 4.3 per million and remained stable from 1973 to 1997.The study issued by the COMS Group(the No.28 COMS report) in 2006 received the tenth most co-citations [53].This publication can be considered an update of the eighth most co-cited study,which further confirmed and clarified that,during longer follow-up,there are no survival differences between the enucleation group and the125I brachytherapy group.

Overall,half of the top 10 co-cited references[29-33] involved gene and molecular biology studies(revolving aroundGNAQ,GNA11,BAP1,and chromosome 3,etc.).Two studies [50,51] were systematic reviews that focused on the incidence rate.All remaining publications [12,52,53] were clinical studies involving the leading prognosis factor of UM(metastasis of UM),and exploration of the effectiveness and safety of treatment strategy (such as125I brachytherapy) for choroidal melanoma.Combined,these topics formed the basis of the UM research field.

Emerging information

Papers with citation burstness are papers that are closely focused on by peers over a period of time.With them as a basis,investigators could better understand research hotspots and development trends [54].Therefore,we used CiteSpace to identify the top 42 references with stronger citation burstness.Currently,there is no clear selected threshold [21,54],therefore,we further defined seven references [36-38,40,44,55,56] (over 17.5,overall endurance strength from 2005 to 2019) as the strongest citation burstness strength publications.Among them,the study with the strongest burstness strength was issued inCancerin 1993 by Kath R et al.[36].In this paper,it was found that metastatic UM could result in poor prognosis and limited treatment,especially when metastasis was diagnosed in the liver.In 2011,Chapman PB et al.published the second strongest burstness study inNewEnglandJournalof Medicine[37].In this phase 3 randomized clinical trial,when compared with dacarbazine,the vemurafenib strategy improved overall and progression-free survival in melanoma patients (including UM) with a V600E proto-oncogeneBRAFmutation.The third strongest burstness publication issued in 2013 belonged to Martin M et al.[38].In this publication,the somatic mutations inEIF1AXandSF3B1and chromosome 3 copy number variation in UM were explored,thereby confirming and extending the established classification model of UM.In 1999,the fourth strongest burstness study was published by Maniotis AJ et al.[40].The authors demonstrated the relationship between the aggressive melanoma cell phenotype and the underlying mechanisms for the generation of special patterned matrix-associated vascular channel features in aggressive uveal and cutaneous melanomas,and showed that aggressive melanoma cells may generate vascular channels that facilitate tumor perfusion independent of tumor angiogenesis.The fifth strongest burstness study was published by Diener-West M et al.(1992,ArchivesofOphthalmology) [55].This metaanalysis summarized the 5-year death rates following enucleation of choroidal melanoma,and supported that tumors with a larger size were related to higher mortality rates.In 2007,Damato B et al.published the sixth strongest burstness study (Ophthalmology) [56].From this nonrandomized case series study,it was concluded that combined with tumor diameter and cell type,monosomies 3 and 8 were effective predictors of disease-specific mortality in UM.The publication with the seventh strongest burstness belonged to Hodi FS et al.(2010) [44].This team concluded that compared with glycoprotein 100 (gp100) alone,the ipilimumab strategy improved overall survival in metastatic melanoma patients (such as UM),with or without a gp100.

As for endurance strength,3 of the top 42 references[39,40,57] had a longer endurance strength that was more than/equal to seven years,thereby indicating that the topics that these studies focused on had continuity.Two papers [39,40] received the longest endurance strength (8 years,both from 2005 to 2012).Among them,the fourth strongest burstness study published by Maniotis AJ et al.was discussed in the previous section[40].The other publication with the longest endurance strength was issued in 2003 [39].In this publication,Scholes AGM et al.found that,in most UM patients,it was not reliable to predict monosomy 3 based on tumor size and histology,except for patients with large epithelioid tumors and small,spindle-cell tumors.The study with a 7-years endurance strength was published by Egan KM et al.in 1988 [57].They mainly summarized the epidemiologic characteristics of UM,including incidence,mortality,and risk factors,etc.We found that the fourth strongest burstness strength study[40] also had the longest endurance strength,suggesting that vascular channel-related research may be an area of focus.

In the top 42 references,nine studies [37,38,41-47]received the citation burstness that ended in 2015 or later.Thus,these nine publications could be considered research hotspots in recent years,and are discussed in detail below.Among them,Chang AE et al.[41]published a study of which the citation burstness ended in 2015.In this study,authors analyzed 84,836 cases of melanoma (involving 4.53% UM) between 1985 and 1994 from the National Cancer Data Base (NCDB),and found that patients with early treatment had good prognosis,especially for cutaneous melanoma patients,and that survival was related to the socioeconomic factors.Interestingly,this study was issued in 1998,but focused on 2011 to 2015,with 5-year citation burstness.It may be because melanomas were gradually focused on,and this research reported the epidemiologic features,case-mix characteristics,management,and prognosis of many different types of melanoma cases in the United State with the time span of ten years,which are more likely to be cited by other publications.Two publications received a citation burstness that ended in 2016.Among them,the second strongest burstness publication presented by Chapman PB et al.[37] was discussed in the previous section.The other study was published by Wiesner T et al.with a 5-year citation burstness [42].In this publication,it was discovered that patients with germline mutations inBAP1had a higher risk of UM and cutaneous melanoma.The citation burstness of three publications [43-45] ended in 2017.One of these three studies,the seventh strongest burstness publication represented by Hodi FS et al.was discussed in the previous section [44].Of the remaining two studies,one was published by Damato B et al.in 2010 with a 6-year citation burstness [43],the other was published by Abdel-Rahman MH et al.in 2011 with a 5-year citation burstness [45].Damato B et al.used multiplex ligation-dependent probe amplification(MLPA) to test chromosome 1p,3,6p,6q,8p,and 8q abnormalities in UM,and supported that MLPA can be used as a routine detection tool for metastatic UM patients in clinical prognostication [43].Furthermore,Abdel-Rahman MH et al.discovered that the germlineBAP1mutation was associated with an increased risk of cancers,including UM and meningioma [45].Especially,we found the citation burstness of three studies [38,46,47] that ended in 2019,therefore,these three studies could be viewed as continued research hotspots in the future.Among these three studies,the third strongest burstness study published by Martin M et al.was explored in the previous section [38] and the citation burstness of the remaining two papers started in 2014.Harbour JW et al.[46] described “mutations occurring exclusively at codon 625 of theSF3B1gene,encoding splicing factor 3B subunit 1,in low-grade UMs with good prognosis” [46].Onken MD et al.[47]examined the prognostic performance of the GEP assay and concluded that the prognosis of a UM patient could be most accurately estimated using the GEP assay [47].Our findings also showed that the second [37],third [38]and seventh [44] strongest burstness strength publications received a citation burstness ending in 2015 or later (from 2012 to 2016,from 2014 to 2019,from 2013 to 2017,respectively).This illustrated that genetic studies (such asEIF1AXandSF3B1),and the use of new target-specific drugs (ipilimumab and vemurafenib) were paid attention to in the field of UM.

Taken together,these nine studies of which the citation burstness ended in 2015 or later revolved mainly around epidemiologic characteristics,prognosis factors,mechanisms of UM development (such asBAP1,EIF1AX,andSF3B1mutation),the use of new predictive tools (such as MLPA and the GEP assay),and clinical applications of novel therapeutic targeted drug(such as ipilimumab and vemurafenib).In recent years,these topics were the research foci in the field of UM.Besides the third strongest burstness strength publication [38],there were two studies of which the citation burstness ended in 2019.These three studies mainly focused on mutations inEIF1AXandSF3B1and the prognostic performance of GEP assay for UM,which could point out research directions in the future.

Strengths and limitations

Our study has quite a few strengths.First,to the best of our knowledge,the present study is the first to use the bibliometric method to evaluate trends of UM-related studies in the last 15 years,and established baseline data for future analysis and comparison.Second,as a bibliometric study,this analysis provided a holistic and developmental view of UM studies when compared to traditional expert reviews,thereby providing a more objective view [58].Third,3 verified bibliometric tools were used to conduct the study,and all have been widely used in different subject fields [21,22,54,58],which makes our study relatively impartial.Moreover,our study provided a classical practice example of visualization analysis,which was also available to explore research trends in other fields.

However,similar to other bibliometric studies [21,22,28,58],our study has some limitations.On one hand,we only searched the WoSCC as the literature source,because it is widely used as the primary database for bibliometric studies [22,28,58],and it is extremely difficult to integrate and analyze data from different databases at the same time by current bibliometric software.On the other hand,the bibliometric software extracted all information automatically,which is different from meta-analysis or overview where data extraction was retrieved by reviewers manually[59-62].Thus,there might be sample bias,such as duplicate names for authors and renamed journals.

Conclusion

UM studies published between 2005 and 2019 were analyzed by the bibliometric software.The United States have been in the leading position of UM-related publications.The Thomas Jefferson University contributed to most of the publications,whereas the journalsInvestigativeOphthalmology&VisualScienceandOphthalmologywere paramount in the UM field.In recent studies,Shields CL may play a critical role in the development of UM research.Gene and molecular biology aspects (involvingGNAQ,GNA11,BAP1,and chromosome 3,etc.),incidence rate,leading prognosis factor (metastasis of UM),and exploring the effectiveness and safety of treatment strategy (such as125I brachytherapy) for choroidal melanoma were the intellectual foundations of UM studies.Emerging hotspots included epidemiologic characteristics,prognosis factors,mechanisms underlying UM development,the use of novel predictive tools,and clinical applications of novel targeted drugs,especially mutations inEIF1AXandSF3B1,and prognostic performance of the GEP assay.