Introduction
Pancreatic cancer (PC) is one of the most malignant gastrointestinal tumors, with a 5-year survival rate of less than 9%, and its incident and mortality rates are rising.1,2 According to the 2020 global cancer statistics, PC has become the 7th leading cause of cancer death in both sexes, and this has become a growing public health concern worldwide.2 Unfortunately, PC does not typically present with obvious symptoms in its early stages. Thus, this cannot be diagnosed until the advanced stage of the disease, at which point treatment options are limited.1,3 Therefore, the accurate screening of early-stage PC is urgently needed.
Several international guidelines have illustrated the importance of early screening for PC.4,5 Since almost 20 years from the appearance of lesions in the pancreas to the development of metastatic PC, a sufficient time window for the early screening of PC was given.6 A large multicenter study revealed that patients with PC, who received early screening, had a significantly longer median overall survival time, when compared to patients who did not receive early screening (9.8 years vs. 1.5 years).7 It is presently generally accepted that screening is warranted and cost-effective for individuals who are at high risk for PC. However, the definition of high-risk groups remain controversial.5,8 For the screening strategies, endoscopic examination, imaging examination, and biomarkers have demonstrated the screening and diagnostic value of PC and precancerous lesions in recent years.8–10 In general, the establishment of a feasible screening scheme for PC remains at its research and exploration stage, and a broader consensus remains to be formed.1,8,9,11
At present, numerous approaches exist to systematically develop an overview of the academic field, in which bibliometric analysis has become a key instrument in assessing research trends and developing guidelines.12 Bibliometrics can not only analyze the contributions and collaborations of authors, organizations, countries and journals, qualitatively and quantitatively, but also assess the development and new trends of academic research. To date, no bibliometric analysis has focused on the early detection of PC. The present study conducted the most comprehensive search and bibliometric analysis of PC early screening literature over the past 20 years, and explored the present development direction and research hotspots in this field.
Materials and methods
Data sources and search strategies
All data employed for the present study were retrieved from the Web of Science (WoS) core database from January 2003 to December 2022. The specific screening strategy was, as follows: TS = ((((((AK = (Pancreatic Neoplasm*)) OR AK = (PC*)) OR AK = (Pancreatic tumor*)) OR AK = (Pancreatic carcinoma*)) OR AK = (Pancreatic adenoma*)) OR (((((TI = (Pancreatic Neoplasm*)) OR TI = (PC*)) OR TI = (Pancreatic tumor*)) OR TI = (Pancreatic carcinoma*)) OR TI = (Pancreatic adenoma*))) AND (((((TI = (cancer detect*)) OR TI = (cancer screen*)) OR TI = (early detection of cancer)) OR TI = (early diagnosis of cancer)) OR ((((AK = (cancer detect*)) OR AK = (cancer screen*)) OR AK = (early detection of cancer)) OR AK = (early diagnosis of cancer))) AND publication year = (2003 to 2022).
Studies published in the English language, with the theme, PC early screening, were included. The exclusion criteria were, as follows: (1) editorials, meetings, letters, news, book chapters, and corrections; (2) non-English publications; (3) studies with main topics not related to PC early screening. Two authors (GX M and YC W) independently screened the search results. Any disagreements were resolved by discussion with a third reviewer (WBZ). Figure 1 shows the specific selection process.
Data extraction
Two reviewers (GX M and YC W) independently extracted the following information from the included studies: study title, study type, name of author(s), abstract, keywords, references, journal and the impact factor (IF), citations, country(ies)/region(s), institutions, and address. These were exported in plain text format.
Bibliometric analysis and visualized analysis
An Excel spreadsheet was used to present the bibliometric indicators, which included the following: total number of publications, year of publication, study type, the top journals ranked by IF, the number of publications of each country, institution and scholar, and the citations of the studies. The sum of the citations of all articles published in the same journal were divided by the number of articles, in order to obtain the average number of citations of journals. The IF and quartile rank (Q) were retrieved from the year of publication of the journal from Journal Citation Reports (jcr.clarivate.com).
For the visualized analysis, CiteSpace and VOSviewer were used for the present study. These are widely used in scientific and technical text mining, and information visualization, and have the most important feature of being able to provide a scientific visualization map, in order to understand the structure, pattern and distribution of relevant studies.13 YL and LH analyzed the data. The obtained data were imported into the CiteSpace (Version 6.1.R6) software, and this was used as the literature visualization application tool to analyze indicators, such as country, journal, institution, author and reference. Furthermore, CiteSpace was used to analyze the keywords and co-cited references (those that were cited by one or more publications at the same time) with the strongest citation bursts, in order to predict the trends in the field. VOSviewer (version 1.6.16) was used to plot the country collaboration, institutional collaboration, scholar collaboration, keyword co-occurrence, and literature co-citation. The nodes of the network diagram comprised of different scholars, the size of the nodes indicated the richness, the connection between the nodes indicated the degree of association, and the thickness of the connection indicated the strength of the association. The color of the element represented the cluster it belongs to, and different clusters were represented by different colors.
Results
Publications and journals
A total of 581 publications in the field of early screening of PC was included for the analysis. For study type, 448 (77.11%) were original articles, and 133 (22.89%) were reviews or guidelines. The annual number of publications exhibited a fluctuating rising trend worldwide. Furthermore, 2022 was the year with the highest number of publications (n = 71), and the year with the highest number of articles published (n = 58), while 2021 was the year with the highest number of published reviews/guidelines (n = 23) (Fig. 2). Among these studies, 319 studies had an observational design (251 studies were retrospective studies and 68 studies were prospective studies), 23 studies were meta-analysis studies, 15 studies were systematic reviews, and three studies were randomized controlled studies.
The included studies were published in 272 journals. In the field of PC early screening, the top 10 most prolific journals are listed in Table 1. Among these, the most published journal was Pancreas (n = 35), followed by the World Journal of m, Gastroenterology (n = 24), and Pancreatology (n = 18). Furthermore, among the top 10 journals, Gastroenterology had the highest IF of 33.883. In partitioning according to Journal Citation Reports, among the top 10 journals, 50% were classified as Q1, 10% were classified as Q2, and 30% were classified as Q3. The top two journals with the highest average number of citations per article were Gastroenterology (n = 104.625) and Clinical Cancer Research (n = 100.125), indicating the high quality of articles published in these two journals in this field.
Table 1Top 10 journals that published literature related to pancreatic cancer early screening
| Rank | Journal | Number | Citations | Average citations | IF2021 | Quartile in category |
|---|
| 1 | PANCREAS | 35 | 1,081 | 30.886 | 3.243 | Q3 |
| 2 | WORLD JOURNAL OF GASTROENTEROLOGY | 24 | 1,125 | 46.875 | 5.374 | Q2 |
| 3 | PANCREATOLOGY | 18 | 288 | 16.000 | 3.977 | Q3 |
| 4 | CANCERS | 15 | 98 | 6.533 | 6.575 | Q1 |
| 5 | JOURNAL OF THE PANCREAS | 11 | 150 | 13.636 | - | Q4 |
| 6 | CLINICAL CANCER RESEARCH | 8 | 801 | 100.125 | 13.801 | Q1 |
| 7 | GASTROENTEROLOGY | 8 | 837 | 104.625 | 33.883 | Q1 |
| 8 | INTERNATIONAL JOURNAL OF CANCER | 7 | 319 | 45.571 | 7.316 | Q1 |
| 9 | FAMILIAL CANCER | 7 | 206 | 29.429 | 2.446 | Q3 |
| 10 | GASTROINTESTINAL ENDOSCOPY | 7 | 199 | 28.429 | 10.396 | Q1 |
Countries, institutions, scholars and cooperation
A total of 49 countries/regions contributed to the field of PC early screening between 2003 and 2022 (Fig. 3a). The four most prolific countries are presented in Figure 3b, and the top 10 countries are detailed in Additional file 1. The United States (n = 245) was the most productive, followed by China (n = 82), Japan (n = 82), and Germany (n = 48). The five most cited countries were the United States (n = 9,700), Japan (n = 2,079), Germany (n = 2,003), Italy (n = 1,945), and China (n = 1,844). For international cooperation, the United States was the center of collaboration. Germany, Italy, France, and the United Kingdom had more intense ties with other countries in the research cooperation network (Fig. 3c). Germany and Canada started their collaborative research earlier, while China, France, and other countries started foreign cooperation in recent years (Fig. 3d).
For the present selected publications, 783 institutions were involved in the study. The top five research institutions with the highest number of publications were all from the United States (detailed in Additional file 2). The institution with the most publications was the University of Texas MD Anderson Cancer Center (n = 41), and the institution with the highest average number of citations for each article was Mayo Clinic (n = 71.33). The close connections between these different units indicate the extensive cooperation among various agencies, and demonstrate the presence of several institutional cooperation clusters. The top three institutions for collaboration were the University of Texas MD Anderson Cancer Center, Johns Hopkins University, and Harvard University. These three institutions had the highest density of cooperative cluster nodes, and were at the center of the relationship graph (Fig. 4a).
In the past 20 years, 3,883 scholars contributed to the research of PC early screening. Among the 10 scholars with the highest number of publications, nine scholars were from the United States, and one scholar was from Germany (detailed in Additional file 3). Furthermore, three scholars were from Johns Hopkins Medicine, and two scholars were from UTMD Anderson Cancer Center. Anirban Maitra was the scholar with the most publications (n = 15), and Ralph H. Hruban was the scholar with the most citations per article on average (n = 125.455). A cluster of 16 academic groups was obtained through the cooperative statistical analysis of scholars who published three or more papers. As presented in Figure 4b, Marcia Irene Canto of Johns Hopkins Medicine and Randall E. Brand of the University of Pittsburgh had a large number of nodes in the whole layer, and were surrounded by several scholars.
Highly cited literature
The density visualization of the top 100 co-cited literature is presented in Figure 5a. Each literature had 20 or more co-citations. Most of the co-cited literature was from the top journals, and the main disciplines involved were oncology and gastroenterology. The basic information of the 10 most co-cited manuscripts, in the order of total citations and average citations, is listed in Table 2. The highest number of citations was 99, which were published by Canto et al. in Clin Gastroenterol Hepatol in 2006, and in Gastroenterology in 2012.14,15
Table 2Top 10 publications for pancreatic cancer early screening
| Title | First author | Country | Journal | Year | Total citations | Type |
|---|
| Screening for early pancreatic neoplasia in high-risk individuals: a prospective controlled study | Marcia Irene Canto | USA | Clin Gastroenterol Hepatol | 2006 | 99 | Article |
| Frequent detection of pancreatic lesions in asymptomatic high-risk individuals | Marcia Irene Canto | USA | Gastroenterology | 2012 | 99 | Article |
| International Cancer of the Pancreas Screening (CAPS) Consortium summit on the management of patients with increased risk for familial pancreatic cancer | Marcia Irene Canto | USA | Gut | 2013 | 98 | Review/Guideline |
| Prospective Risk of Pancreatic Cancer in Familial Pancreatic Cancer Kindreds | Klein AP | USA | Cancer Res | 2004 | 95 | Article |
| Screening for pancreatic neoplasia in high-risk individuals: an EUS-based approach | Marcia Irene Canto | USA | Clin Gastroenterol Hepatol | 2004 | 88 | Article |
| Projecting cancer incidence and deaths to 2030: the unexpected burden of thyroid, liver, and pancreas cancers in the United States | Lola Rahib | USA | Cancer Res | 2014 | 81 | Article |
| Very high risk of cancer in familial Peutz-Jeghers syndrome | F M Giardiello | USA | Gastroenterology | 2000 | 73 | Article |
| The yield of first-time endoscopic ultrasonography in screening individuals at a high risk of developing pancreatic cancer | J W Poley | Netherlands | Am J Gastroenterol | 2009 | 69 | Article |
| Early diagnosis and treatment of pancreatic dysplasia in patients with a family history of pancreatic cancer. | Brentnall TA | USA | Ann Intern Med | 1999 | 68 | Article |
| Pancreatic cancer screening in a prospective cohort of high-risk patients: a comprehensive strategy of imaging and genetics | Elizabeth C Verna | USA | Clin Cancer Res | 2010 | 63 | Article |
After analyzing the 25 co-cited publications with the strongest outbreaks (Fig. 5b), it was found that these publications focused on research advances related to screening technologies, high-risk groups, genetic testing, and cancer morbidity and mortality. The first citation burst in the field began in 2006, and this continued until 2009. Seven literature citation bursts were still in progress. Five of the 25 co-cited publications were from Marcia Irene Canto, with outbreaks in 2006,16 2007,14 2012,15 2014,17 and 2019.18 The practice guideline published by Canto et al. had the highest outbreak intensity of 20.42.17
Keywords
The keywords involved in the selected literature were extracted, and the synonyms were combined. Figure 6a summarizes the top 60 keywords with the highest frequency. Adenocarcinoma was located in the center of the keywords, which is consistent with the present research theme. Adenocarcinoma, biomarkers, family-history, ultrasound endoscopy, and papillary mucinous neoplasms were the keywords that most frequently appeared. For the markers, a table of markers associated with PC, according to the MeSH word frequencies (Additional file 4), was presented. The most frequent marker was CA-19-9 antigen (n = 52), followed by microRNAs (n = 20), proto-oncogene proteins p21 (ras) (n = 17), exosomes (n = 15), and DNA methylation (n = 14). According to the chronological order of keyword occurrences, it was found that extracellular vesicles, cell-free DNA, nanoparticles, microRNA, DNA, blood, and glypican-1 were the emerging keywords (Fig. 6b).
Burst keywords were considered as another important indicator of research frontiers, and these can determine the hot trends in the field. Figure 6c presents the top 15 keywords with the highest number of citation bursts from 2003 to 2022. These keywords were artificially divided into three stages, according to the chronological order of appearance in the outbreak. The first stage was from 2003 to 2009. The earliest keyword burst was “gene expression” (2003–2011), followed by “juice” (2004–2006), “computed tomography” (2006–2009), “brca2 mutation” (2009–2015), “intraeputhelial neoplasia” (2009–2014), and “papillary mucinous neoplasm” (2009–2012). The second stage was from 2010 to 2015, in which only one new burst keyword emerged, “differential diagnosis” (2011–2017). There were no new burst keywords during the period of 2012–2015. The third stage was from 2016 to 2022, in which there were eight new burst keywords. Keywords, such as “extracellular vesicle” (2016–2022), “dna” (2019–2022), “circulating tumor cell” (2020–2022), and “blood” (2020–2022), continued to burst with citations until 2022. The keyword that lasted for the longest duration of nine years was “gene expression”, while the keywords that lasted for a long duration were “brca2 mutation”, “cell free dna” (2016–2020), and “cystic neoplasm” (2017–2020). The keywords with a burst strength of >3 were “dna” (n = 4.03), “computed tomography” (n = 3.32), and “extracellular vesicles” (n = 3.15).
Discussion
The present study was the first to use a bibliometric approach to provide the quantitative characterization of the global field of PC early screening at the macro level over the past two decades, which was complemented by qualitative interpretation. The present results suggest that the number of publications in this area remains modest, with fluctuations between years, and an overall increasing trend. The research institutions and scholars in Europe and the United States had a high academic influence in the field of PC early screening. The keywords indicated that the present focus on imaging, conventional biomarkers, and liquid biopsy were the three main test methods used in PC early screening, and that circulating tumor cells in the liquid biopsy would lead to new research directions, and provide new vitality.
Overall, according to the annual number of publications, the early screening of PC has received increasing attention, but the difficulty of research in related fields remains quite high. The journals that published related articles were dominated by the fields of pancreatic diseases, oncology, and digestive diseases. Researchers would likely prioritize journals with the highest number of published studies that focused on PC early screening. Articles published in journals with higher IF, or in Q1 and Q2, were usually considered to be the more important in the subject area. The analysis of the top 10 journals with the highest number of publications revealed that three journals with an IF of >10, and Q1 and Q2, accounted for 60%. This indicates that PC early screening is valued by top international scientists, while more high-quality articles are still needed in this field.
According to the results for the distribution of publications, a high number of interactions were identified among developed countries, while developing countries had less cooperation with other countries, indicating that the economy may have influenced the degree of research and development in the field of PC early screening between different countries. Although China has published a considerable number of articles in this field (14.11%), the proportion of international cooperative articles was significantly lower than that of the United States or Germany, and there was a lack of authoritative research institutions. All of these show the lack of influence of China in this field. For the outstanding scholars in this research field, Marcia Irene Canto and Randall E. Brand had numerous nodes in the whole layer, indicating that the publications of these scholars are hot topics in this field. Marcia Irene Canto (from Johns Hopkins Medicine) is a core leader in the field of PC early screening, and contributes to the use of endoscopic ultrasonography (EUS) and computed tomography (CT) to detect early PC and its precancerous lesions,14,16 laying the foundation for a series of subsequent studies. Randall E. Brand (from the University of Pittsburgh) focused on the genetic research of pancreatic cancer, conducted intensive research in the screening of familial PC, and is one of the major contributors in this field.19,20
Literature co-cited for multiple times was considered to constitute the knowledge base of a particular field. A significant proportion of the co-cited articles in this area focused on the exploration of high-risk groups.15,18,21–25 At present, in addition to patients with a family history of PC and mutation carriers,17 the recognition of high-risk groups further concentrated on new-onset diabetes,26 Peutz-Jeghers syndrome,27 and other groups, such as BRCA2 carriers.28 However, the need for PC early screening in patients with BRCA2 mutations remains controversial.8,29 An early study concluded that the screening of this population has little relevance,30 while the present consensus of the International Consortium for PC Screening suggested that screening should be offered to BRCA2 carriers with at least one first-degree relative, or at least two second-degree relatives.31 The citation outbreaks of seven pieces of literature remained in progress, and the strongest value was identified in the prediction model published by Sharma et al. in 2018, which clearly defined those who warrant the clinical early screening of PC.26
The present keywords analysis revealed that imaging, conventional biomarkers, and liquid biopsies are predominant in the early screening of PC. Imaging techniques, including CT, magnetic resonance imaging (MRI) and EUS, play important roles in PC early screening for high-risk people.8 Pancreatic CT is a low-cost technique, and is easy to operate, but the resolution of CT is inferior to MRI and EUS, in terms of small lesions of the pancreas, and its radiation is a disadvantage.9,11 With the characteristics of absence of ionizing radiation and better soft tissue resolution, MRI combined with endoscopic retrograde cholangiopancreatography (MRCP) contributes to better detect tumors at the early stage.32–35 EUS can provide high-resolution images of the pancreas, and is important for detecting small pancreatic lesions.36 It is noteworthy that EUS-guided fine-needle aspiration (EUS-FNA) has high sensitivity and specificity, and is considered as the gold standard for PC diagnosis.11,37 However, as an invasive procedure, EUS-FNA may lead to pancreatitis or bleeding, and is not recommended for routine screening.7,8 Due to the different pathological staging of PC, the sensitivity and specificity of different imaging techniques lead to great differences, and the combined application of multiple methods can significantly improve diagnostic accuracy.36,38
There is a lack of biomarkers for early stage PC.8,11 Among the single marker tests, the serum CA19-9 test has sensitivity (approximately 70%), and is the only biological marker presently used in PC early screening, but this lacks sufficient accuracy and specificity in screening for early stage PC.39–41 The combination of CA19-9 with other biomarkers (such as CEA, CA125 and CA242) can increase the sensitivity to 85–90%, and specificity to more than 90%, presenting as a good application prospect.42,43 The use of metabonomics to discover differential metabolites as new tumor markers for the early diagnosis of PC has become a research approach.44 Meanwhile, proteomics has made considerable progress in PC early screening. Some recent studies have revealed that the tumor-specific variants of protein glycosylation may contribute to the early diagnosis of PC.45,46 The application of omics techniques has facilitated the rapid increase in the number of biomarkers, which accelerate the entry of effective markers in clinical practice, as soon as possible.
Liquid biopsy has become a research focus in recent years. This is minimally invasive, rapid, and suitable for dynamic surveillance, and is represented by extracellular vesicle and circulating tumor cells. Sonia et al. reported that GPC1+ crExos can be a pathway for the detection and isolation of exosomes in the circulation of cancer patients for further study.47 A research team from China identified that plasma EV-long RNA analysis may be a new marker for the detection of PC.48 The Moores Cancer Center research team developed a blood biomarker classifier based on EV, which can detect 95.5% of stage I PCs.49 However, existing studies on a single marker have not provided the high sensitivity and specificity required for the early detection of PC.50 Thus, combining multiple biomarkers for the early screening of PC may be the direction of development in this field.1,51
There were some limitations in the present study. First, merely WoS was selected as the database, and several relevant articles might have been missed. Second, this may have omitted important non-English articles, leading to bias, and the reduction of its credibility. Finally, since the literature selection focused on the period of 2003-2022, some high-quality articles published earlier in the field may have not been included, which may lead to missing hot terms, and the underestimated number of citations of recently published high-quality articles.