Citation analysis of the most influential ependymoma research articles illustrates improved knowledge of the molecular biology of ependymoma

The history of academic research on ependymoma is expansive. This review summarizes its history with a bibliometric analysis of the 100 most cited articles on ependymoma. In March 2020, we queried the Web of Science database to identify the most cited articles on ependymoma using the terms “ependymoma” or “ependymal tumors,” yielding 3145 publications. Results were arranged by the number of times each article was cited in descending order. The top 100 articles spanned across nearly a century; the oldest article was published in 1924, while the most recent was in 2017. These articles were published in 35 unique journals, including a mix of basic science and clinical journals. The three institutions with the most papers in the top 100 were St. Jude Children’s Research Hospital (16%), the University of Texas MD Anderson Cancer Center (6%), and the German Cancer Research Center (5%). We analyzed the publications that may be considered the most influential in the understanding and treatment management of ependymoma. Studies focused on the molecular classification of ependymomas were well-represented among the most cited articles, reflecting the field’s current area of focus and its future directions. Additionally, this article also offers a reference for further studies in the ependymoma field.


Introduction
Ependymomas are rare primary tumors of the central nervous system (CNS) that affect both children and adults [40,42]. The 2016 World Health Organization (WHO) Classification of Tumors of the CNS categorizes them into four subtypes: subependymoma and myxopapillary ependymoma (grade I), ependymoma (grade II), ependymoma RELA (v-rel avian reticuloendotheliosis viral oncogene homolog A) fusion-positive (grade II or III), and anaplastic ependymoma (grade III) [12,58]. Ependymomas are more commonly found in children and at this age more likely to be located intracranially and harbor more aggressive molecular variants, leading to worse overall survival (OS) when compared to adult variants [6,14,28,30,55].
Ependymomas have been extensively studied with respect to molecular subtyping, prognostication, and clinical outcomes [4,8,14,30,56,58]. Therapy focuses on strategic surgical approaches to achieve gross total resection (GTR), and conformal radiation therapy (CRT) is the most common adjunctive treatment [6,20,21,24,38,49]. Chemotherapy has been studied primarily in children under 3 years of age due to their susceptibility to radiotherapy-induced neurotoxicity. However, chemotherapy has failed to demonstrate improved outcomes compared to CRT [53]. At present, several clinical trials are underway scrutinizing promising neoadjuvant chemotherapeutic strategies [13,32,53]. Presently, effective treatment of ependymoma requires a multi-modal approach, involving an interdisciplinary team of neurosurgeons, neurologists, oncologists, radiologists, and primary care physicians, among others [14,29]. Prognosis varies by type and location; pediatric ependymomas are more commonly intracranial with a 10-year estimated survival rate of 13-50%, while adult ependymomas have a predilection for the spine and have a 5-year survival rate of 67 to 85% and a 10-year survival rate of 72% [14,30,37]. In both populations, complete resection is the most consistent factor correlated with improved outcomes [6,14,37].
Given the diversity, volume, and interdisciplinary nature of ependymoma research, a bibliometric analysis focused on the history, recent developments, and trajectory of research can help frame our current understanding of the disease [3,17]. The objective of this study is to analyze the most influential articles on ependymoma and identify the most relevant clinical problems in the field to guide further investigation. While bibliometric analyses have been published for neoplastic lesions of the brain [2,3,7,17,25] and for the spinal cord [1,9,10], no such investigation currently exists for ependymoma.

Methods
On March 22, 2020, we performed a title-specific search of the Thomson Reuters Web of Science (WoS) database (Thomson Reuters, NY, USA) to identify the most cited articles on ependymoma. We used "ependymoma" or "ependymal tumors" as our query term for the years 1900 to 2020 selecting the "all databases" option. The results were arranged according to the number of times each article was cited in descending order. To avoid the subjective exclusion of studies from our analysis, all papers from our query were included if they were ranked 1 to 100 in terms of number of total citations. The following variables were extracted: rank of article by total citations, rank of article by average citations per year, first and last author, title of article, publication, year, total citations for each article, average citations per year for each article, article country of origin, and institution of the first author. In cases of co-first authorship, country and institution of the author listed first were used [3,7,17,31]. The average citations per year for each article was calculated as previously described [10].
We categorized the articles as either clinical, basic science, or literature review. Articles were independently classified by LRK and DH and reexamined by BVL and AP. Any inconsistencies were resolved by discussion with the senior author (IY) after careful review of full-text articles. Studies that were primarily focused on basic tumor biology or molecular classification of ependymoma were classified as basic science (e.g., involving genome sequencing) as described previously [25]. Studies that were patientfocused and reported outcomes were classified as clinical, which included histopathological studies [25].

Results
Our query yielded 3145 publications on ependymoma. The top 100 most cited articles were selected for review based on overall citation count and are shown in Tables 1 and 2, organized by total citations and average citations per year, respectively. These articles were published between 1924 and 2017. They have been cited a collective 11,640 times, averaging 116.4 citations per article ( Table 1). The top 10 articles on the list were published between 1977 and 2015 and averaged 304 total citations (standard deviation [SD], 106.5; range, 206-551) ( Table 1).
The most cited article overall was a basic science article entitled "Radial glia cells are candidate stem cells of ependymoma," published in Cancer Cell in 2005 (Table 3) [51]. The second most cited article overall was a clinical article entitled "Intramedullary ependymoma of the spinal cord," published in the Journal of Neurosurgery in 1990 (Table 4) [34]. Basic science and clinical articles comprised the majority of the top 50. The first review article ranked forty-second overall was titled "Pediatric        [22].
As shown in Fig. 1, the time period from 2005 to 2009 oversaw the publication of the greatest number of articles on the list (24 papers). This was followed by 2000-2004 and 1995-1999 (16 papers each) (Fig. 1). Total citations (2870) and average citations per year (204) were also highest for papers published in 2005-2009 (Fig. 2).

Journal of publication
The top 100 cited articles on ependymoma were published in 35 unique journals. The most frequent journals featuring the top cited articles included International Journal of Radiation Oncology Biology Physics (13%), Cancer (10%), and Journal of Neurosurgery (9%) ( Table 5). Of the top 10 most cited, 3 articles were published in Cancer Cell, followed by 2 articles each in Nature and Cancer.

Countries and institutions
A total of 13 countries represented the top 100 articles published (Fig. 3). The USA (n = 63), Germany (n = 8), and the UK (n = 7) were the highest contributors of the top 100 articles. The top institutions contributing the greatest number of articles among the top 100 most cited articles were St. Jude Children's Research Hospital (n = 16), the University of Texas MD Anderson Cancer Center (n = 6), and the German Cancer Research Center (n = 5) ( Table 6). The USA contributed 5 of the top 10 most cited articles.

Article category
Each article was categorized as either basic science (19%), clinical (74%), or literature review (7%) ( Table 1; Fig. 1). Studies are separated into basic science and clinical studies and ranked by times cited in Tables 3 and 4, respectively. Of the top 10 articles, 6 were basic science articles and 4 were clinical articles. Of the top 20, 10 were basic science articles, and 10 were clinical articles.

Citations per year
Since articles published more remotely are advantaged in terms of collecting citations over time, we examined the citation frequency per year. Using this metric, the article with the greatest number of citations per year (53.5) was a basic science article entitled "Molecular classification of ependymal tumors across all CNS compartments, histopathological grades, and age groups," published in Cancer Cell in 2015 (Table 2) [42]. Comparatively, the clinical article with the most citations per year -"Conformal radiotherapy after surgery for pediatric ependymoma: a

Authors
The first and senior authors of each paper in the top 100 list were analyzed (  (Fig. 4).

Discussion
This study identifies the most widely cited articles related to the understanding of ependymoma. Our bibliometric analysis revealed 100 articles published across 35 distinct journals, which highlighted a broad international interest    [17]. Our hope is that this bibliometric analysis informs researchers in their efforts to understand the most relevant and significant literature relating to ependymoma. It is important to note that while overall citation number is an important indicator of an article's impact and importance, it can be misleading in older articles that have more time to be cited with each passing year. To account for this, our analysis included another important metric: citations per year ( Table 2). As an example, the article ranked sixth overall on our list -"Conformal radiotherapy after surgery for pediatric ependymoma: a prospective study" published in Lancet Oncology in 2009 -also ranked sixth in citations per year [35]. This article reported a high rate of local tumor control and event-free survival following aggressive surgical intervention and adjuvant high-dose conformal radiotherapy in pediatric patients, including those younger than 3 years of age [35]. Its presence within the top 10 in both overall and average yearly citations indicates its continued relevance in our understanding of ependymoma, particularly for pediatric patients, despite having been published over ten years ago.
In bibliometric analyses, it is not uncommon to find several articles with drastically different positions on these two lists. Such articles tend to be highly impactful articles published very recently. Two such articles on our list worth examining in closer detail are studies by Peeler et al. and Ramaswamy et al. in 2016. Ranked 81st and 82nd overall and 9th and 10th in citations per year, respectively, these two studies provided novel insights into two well-established treatment modalities. Specifically, Peeler et al. discovered that proton therapy-induced damage to normal tissue dependent on the physical radiation dose and track-averaged linear energy transfer, one of the main determinants of proton therapy's biological effectiveness [45]. Ramaswamy et al. reported that incomplete resection of molecular variant EPN_PFA (posterior fossa ependymoma A) ependymomas was associated with poor prognosis and that adjuvant radiation is preferred for patients with complete resections, while delayed external-bean radiation is preferred for relapsing cases of EPN_PFB (posterior fossa ependymoma B) tumors [46]. Both articles highlight critical discoveries in our understanding of current therapeutics for ependymoma, so it is unsurprising that they have each been given considerable attention since publication. Their place on the overall citation list is likely just a consequence of having less time to gather citations.
A closer examination of the top 20 articles in particular revealed a trend with respect to article type and publication year. Clinical articles within the top 20 tended to be published earlier (i.e., 1955 to 2009), while basic science These clinical articles tended to focus on the initial clinical presentations and pathophysiologic prognosticators of the disease, much of which is considered common knowledge today. One such article, entitled "Ependymoma: follow-up study of 101 cases," published in Cancer in 1977 (fourth most cited overall), managed to follow a cohort of patients who underwent ependymoma treatment over a considerable period of time (22 years) [37]. The authors reported favorable clinical outcomes in cases of spinal ependymoma, which more commonly affects adults (10-year survival of 72%), compared to intracranial ependymoma, which more commonly affects children (10-year survival of 13%) [37]. They also reported a survival benefit with postoperative radiation therapy but failed to find much prognostic value in tissue histopathology, an issue still under debate in current literature [27,52,57,59]. Given the extensive followup reported by the authors as well as the relatively novel findings with respect to clinical course of ependymoma at the time of publication (1977), it is not surprising that this clinical article has maintained citation prevalence to date. The article entitled "Myxopapillary ependymoma: a clinicopathologic and immunocytochemical study of 77 cases," published in Cancer in 1985 and eighth overall on our list, is another example of a clinical article that has maintained relevance despite its remote publication date [50]. This study focused on gross tumor characteristics as prognosticators for postoperative course, reporting that certain physical findings, such as the presence of a tumor capsule, were more indicative of prognosis than histological features [50]. Since Such advances have likely contributed to the recent shift in focus from ependymoma's clinical characteristics and prognostic factors to biomolecular properties of the disease. Continued scientific interest in ependymoma molecular biology, technological advancements, and new innovations may eventually give rise to novel treatments, such as small molecule and personalized precision medicine therapies [39,44,54].
The top cited clinical studies on ependymoma are most often case series describing key clinical features, diagnostic modalities, different treatment regimens, and outcomes. One common theme among studies is that GTR is the single factor most consistently associated with improved survival and reduced recurrence compared to subtotal resection (STR) [15,18,35,36,47,50]. There were no prospective randomized controlled trials in the top 100 most cited articles. The most cited clinical article (ranked second overall)titled "Intramedullary ependymoma of the spinal cord," was published in 1990 and described a retrospective series of 23 patients who underwent surgical resection of this entity [34]. All tumors were histologically benign, gross total resection was achieved in all cases, and no recurrences were reported. Other series reported outcomes in various treatment strategies combining surgical resection, radiation therapy, and chemotherapy. For instance, the 19th most cited study (by Salazar et al. published in the Journal of Neurosurgery in 1983) was one of the first studies to establish efficacy of adjunctive radiotherapy in ependymoma treatment [48]. The authors reported a 10-year overall survival of 69% in a series of patients with intracranial ependymoma treated with resection and whole-brain radiation therapy [48]. The progression from whole-brain radiation to localized radiation was demonstrated in a more recent prospective trial by Merchant et al. in 2009. These authors published a large series of 153 pediatric patients who underwent surgery and conformal radiation therapy (CRT) and reported 85% overall survival in patients who received CRT without delay [35]. In addition to their excellent outcomes, this study irradiated pediatric patients younger than 3 years old, which has been historically avoided due to concerns for delayed radiation neurotoxicity [16]. The role of adjunctive chemotherapy in ependymoma treatment was the topic of two prospective trials that were 13th [15] and 23rd [16] most cited studies overall. The 13th most cited study involved treatment of 73 children with primarily high-grade ependymoma with surgery and chemotherapy, without radiation [15]. The authors reported a low 4-year progression-free survival rate at 22% and overall survival rate of 59% [15]. The 23rd most cited study treated 89 children aged 3 years or younger with surgical resection and chemotherapy. Similarly, disease progression occurred in 62.5% of patients with non-metastatic disease, and overall survival at 5 years was 63.4% [16]. Notably, the authors did report that higher doses of chemotherapy were associated with improved 5-year overall survival compared to low doses (76% vs. 52%) [16].
Several clinical studies were lower in overall citations but higher in citations per year, suggesting that they are impactful articles published more recently. For instance, the study entitled -"Histopathological grading of pediatric ependymoma: reproducibility and clinical relevance in European trial cohorts," published in 2011 was 30th in overall citations but 11th in citations per year [11]. This study developed a novel method for ependymoma grading that demonstrated higher concordance among pathologists than the traditional WHO grading method. However, the study found little correlation between ependymoma grade and clinical outcomes, calling into question the clinical utility of histological grading of ependymoma [11]. Two studies ranked 9th [45] and 17th [26] in average citations per year, utilized proton beam radiation for adjunctive ependymoma treatment. Peeler et al. created linear regression models correlating proton beam radiation dose and linear energy transfer with post-treatment changes on imaging. This demonstrated objective clinical changes caused by proton beam radiation, although did not report patient outcomes such as overall or progression-free survival [45]. On the other hand, MacDonald et al. reported excellent 2-year overall survival (89%) and progression-free survival (80%) in 17 pediatric patients treated with proton therapy after surgical resection. These studies together may represent a promising new adjunct to GTR in the treatment of ependymoma. Finally, as previously mentioned, one study ranked 83rd overall and 10th by citations per year addressed the effect of distinct molecular profiles of posterior fossa ependymoma on outcomes after surgery and radiation [46]. The authors report EPN_PFA was a highly significant predictor of poor progression-free survival (hazard ratio [HR], 2.14; 95% confidence interval [CI], 1.31 to 3.49, P = 0.002) and overall survival (HR, 4.30; 95% CI, 1.88 to 9.87; P < 0.001). Conversely, EPN_PFB was associated with excellent 10-year overall survival of 96.1% after GTR [46]. These findings in this recent article with a high citations per year count highlight the new appreciation of ependymoma molecular subtyping in treatment prognosis.
The WHO grading criteria for ependymoma based on tumor histopathology (most recently updated in 2016) have been shown to have poor predictive value for overall survival for the disease [27]. Given the limited clinical utility of these criteria, recent research has focused on understanding the molecular biology of ependymoma to improve on our current prognostic capabilities [27]. Six of the 10 most cited articles were basic science studies aimed at addressing the issue of ependymoma subtyping. The top article overall, entitled "Radial glia cells are candidate stem cells of ependymoma," published in Cancer Cell in 2005, found that supratentorial, infratentorial, and spinal cord ependymomas are derived from radial glial cells [51]. From this, the authors suggested that histologically similar ependymomas from different regions of the central nervous system represent molecularly distinct diseases and that ependymomas have gene expression profiles that resemble regionally specific radial glial cells. More recently, the article entitled "C11orf95-RELA fusions drive oncogenic NF-kappa B signalling in ependymoma" published in Nature in 2015 elaborated upon the genetic underpinnings of a well-known oncogenic pathway (NF-κB; nuclear factor kappa-light-chain-enhancer of activated B cells), which was found to exist in twothirds of supratentorial ependymomas [43]. Subsequently, a RELA fusion-positive (grade II or III) ependymoma subtype was included in the 2016 WHO Classification of Tumors of the CNS [12].
Another 2015 study focusing on ependymal classification -"Molecular classification of ependymal tumors across all CNS compartments, histopathological grades, and age groups" published in Cancer Cellranked third overall and first in citations per year [42]. This study used DNA methylation profiling to identify nine distinct molecular subgroups of ependymoma and subcategorized each according to its location within the CNS (supratentorial, posterior fossa, and spine) [42]. The novel predictive system developed by this study outperformed previously published histopathological classifications in predicting overall and progression-free survival. The DNA-methylation-specific categorization was not included in the 2016 WHO Classification of Tumors of the CNS likely because DNA methylation profiling is only available in restricted institutions [23] and is therefore not amenable to widespread implementation [41]. Collectively, the recent momentum favoring biomolecular research in ependymoma has led to a more robust classification system for the disease, which will allow for improved prognostication and narrowed molecular targeting for therapeutic development. Such advancements are imperative given the high (40%) prevalence of incurable tumors, poor postoperative prognosis, and chemotherapyresistant properties of ependymomas [19,43,51]. Continued research will reveal the impact of these basic science investigations on the therapeutic and diagnostic landscape of these tumors.

Limitations
This study has several limitations. First, as previously discussed, our list of the top 100 most cited papers was generated based on the total number of citations, which is subject to bias towards papers published earlier [3,10]. Conversely, more recently published articles are often shown more frequently in research databases, which may also contribute to bias. To address this issue, we included data on the total number of citations (Table 1) and average citations per year in our analysis (Table 2), in order to provide a comprehensive view of ependymoma research. This analysis also demonstrated that basic science articles have enjoyed more citations on average in the last 10 years than clinical articles. Taken in combination with overall citation data, these findings suggest that the current direction of ependymoma research will focus more heavily on research examining the biomolecular characteristics of ependymoma. Second, while WoS is the most commonly used and validated resource for bibliometric analyses, it is not comprehensive of all medical literature and does not include citations from textbooks or non-English journal articles [10,17]. Our WoS search was also title-specific, which may have led to the unintentional exclusion of relevant papers in the top 100 list since abstracts and fulltext articles were not included in the search. Third, bibliometric analyses carry the inherent limitation that the citation frequency does not always correlate with impact. For instance, a basic science article published in 2016 on childhood posterior fossa ependymomas published in Science Translational Medicine determined that reduced H3K27me3 and DNA hypomethylation were associated with poor clinical outcomes [5]. However, this impactful study did not make the top 100 list. Fourth, our list is subject to inaccuracy due to the phenomenon of "obliteration by incorporation," whereby highly important articles can become less frequently cited over time as their ideas or findings become so widely accepted as to be considered common knowledge (and thus cited anonymously) [33]. As such, citation numbers may not always accurately reflect the influence or impact of studies, a limitation that is not completely addressed despite our using previously validated bibliometric analysis methodologies for CNS tumors [3,17,25]. Fifth, we categorized studies as basic science and clinical based on the focus of each article as previously performed [25]. However, this dichotomization did not account for studies that may be further subclassified as translational in nature. Despite these limitations, this article seeks to present publishing trends within the ependymoma literature and provides a categorized reference of articles and synthesis that will be helpful for future clinical trainees and scientists in the neuro-oncological and neurosurgical fields.

Conclusion
This study used a validated bibliometric analysis to identify the top 100 most cited articles on ependymoma. Careful examination of the list, in conjunction with another important metric -average number of citations per year -helps paint a picture of the history and behavior of ependymoma research over the last 50 years, as its focus migrated from clinical correlates and histopathologic prognosticators to genetic and molecular underpinnings of the disease. That we observe a high proportion of recently published basic science articles in the top 20 papers of our list points to a tendency to improve upon what were once widely accepted histopathological grading criteria. Ependymomas are chemotherapy-resistant, and a large proportion of tumors are incurable even with surgery and radiotherapy. Our results suggest that the field of ependymoma research is moving towards a more robust basic biological understanding and molecular classification system to guide clinical decision-making and future research endeavors into potential therapeutic options.
Author contribution Nolan J. Brown composed the original draft, revised the draft, gathered data, and approved the final draft.
Bayard Wilson composed the original draft, revised the draft, gathered data, and approved the final draft.
Brian V. Lien composed the original draft, revised the draft, gathered data, and approved the final draft.
Alexander Himstead revised the draft, gathered data, and approved the final draft.
Ali R. Tafreshi revised the draft, gathered data, and approved the final draft.
Shane Shahrestani revised the draft, gathered data, and approved the final draft.
Jack Birkenbeuel revised the draft and approved the final draft. Katelynn Tran revised the draft and is responsible for formatting and editing.
David Horton is responsible for data collection. Anushka Paladugu is responsible for data collection. Lydia R. Kirillova is responsible for data collection. Chen Yi Yang revised the draft and is responsible for approval of the final draft.
Seth C. Ransom revised the draft and is responsible for approval of the final draft.
Alvin Y. Chan revised the draft and is responsible for approval of the final draft.
Ronald Sahyouni revised the draft and is responsible for approval of the final draft.
Isaac Yang is responsible for the original conception and approval of final draft. Data Availability All data and materials support published claims and comply with field standards.
Code availability Not applicable.

Declarations
Ethics approval Not applicable.

Conflict of interest
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.