3.1 Literature screening process
We originally screened 1,359 articles in the database. As shown in Fig. 1, 82 articles were published outside the years 2009 to 2023 and 1 non-English paper was excluded through manual screening. 106 articles were excluded because of irrelevant article categories. Finally, 1,170 eligible studies (999 original articles and 171 reviews)were retained for further analysis.
3.2 Overview of publication status
It has been recognized that the development landscape in a research field closely correlates with the volume of articles published in each period. These 1,170 documents have been cited a total of 29,279 times (25,239 times without self-citations), with an H-index of 84. The trend of annual publications is illustrated in Fig. 2A. Analyses revealed that (i) in the initial 7 years (2009–2015), fewer than 5 articles were published annually; (ii) in the subsequent 3 years (2016–2018), the annual publication rate started to rise steadily; (iii) in the recent 5 years (2019–2023), the publication number has dramatically increased, reaching its peak in 2023 (n = 354). The changing trend of the citation number per year was close to that of the annual publication counts. Compared with publications in all fields, global research hotspots measured by the RRI index also rose exponentially after 2018, as shown in Fig. 2B. Since data collection was completed in December 2023, we could not infer that the number of publications would keep rising in 2024. However, we can anticipate that the development of research in this field will not stop accelerating in the future.
3.3 Contribution of countries to publications
All the documents were published in 57 countries or regions. The United States of America (USA) exhibited the highest contribution to research on scRNA-seq in CVDs (n = 553), closely followed by China (n = 434) and Germany (n = 157), as shown in Fig. 3A and Table 1. Meanwhile, the number of citations in the USA (n = 17,847) far exceeded those of other countries, followed by China (n = 6,933) and Germany (n = 6,495). The above results showed that the USA maintained its lead in the number and quality of publications. In Figure S1, we summarized the funding data in this field and identified that the United States Department of Health and Human Service funded the highest number of projects (n = 391).
As part of our investigation, we visualized the collaborations among countries or regions. In Fig. 3B, the most frequent collaborations were between the USA and China (frequency = 77). Additionally, among the top 10 countries with the highest frequency of cooperation, the USA-centered international cooperation held 6 positions. A country co-authorship analysis of all publications across the 57 countries above was also conducted by VOSviewer. As shown in Fig. 3C, the 57 countries formed 15 clusters, and the USA had closer ties to other nations than any other country, highlighting that the USA acted as a “bridge” node, particularly with certain European countries, China, and Canada. In Fig. 3D, the time-overlapping network showed that the UK and the USA, were early pioneers in this field, while there was a rapid progression in Chinese research outputs recently.
3.4 Contribution of institutions to publications
All the documents were conducted at 1,565 academic institutions worldwide. Chinese Academy of Medical Sciences in China (n = 61), Stanford University in the USA (n = 58), Harvard University in the USA (n = 49), Zhejiang University in China (n = 41), and Karolinska Institutet in Sweden (n = 36) are the top 5 institutions in terms of publication number respectively (Table 1 and Figure S2). To further investigate collaboration between institutions, we performed an institutional co-authorship analysis. Using VOSviewer with a threshold of a minimum of 10 publications per institution, we identified 75 nodes and 5 clusters. As shown in Figure S2B, Chinese Academy of Medical Sciences was located at the center of the yellow cluster, and cooperated closely internally, but further collaboration with other clusters was needed. Harvard University with the highest total link strength (n = 94), and occupied a core position in the blue group. In Figure S2C, the time-overlapping network showed that several American research institutions (For example Massachusetts General Hospital and Brigham and Women’s Hospital) contributed significantly to early development. In contrast, multiple Chinese research institutions (For example Zhejiang University) became more involved in this field in recent years.
3.5 Contribution of authors to publications
A total of 8,595 authors participated in research on scRNA-seq and CVDs in the present study. As shown in Figure S3A and Table 2, Qingbo Xu from Zhejiang University in China was the most productive scholar, with 17 articles published. Following closely behind were Klaus Ley (n = 16, La Jolla Institute for Immunology, USA) and Joseph C Wu (n = 15, Stanford University, USA). More than half of the top 10 most prolific scholars were from China. Klaus Ley had the highest value in the citation (n = 1,188), followed by Joseph C Wu (n = 1,145), and Holger Winkels (n = 1,116, La Jolla Institute for Immunology, USA).
Researchers’ collaborative relationships are shown in Figure S3B. 100 authors with five or more articles were grouped into 21 clusters. By analyzing the clustering network of co-authors, we found that the cooperative relationship among these authors presented a dispersive distribution, indicating that most of the cooperated authors were from the same units and that geographical factors influenced collaboration between authors. The time-overlapping network results are shown in Figure S3C. We observed that researchers from China were forming a new research network in this field. National and institutional collaborations are one of the future directions due to the inadequate collaboration between different research groups.
3.6 Contribution of journals to publications
Table 3 and Fig. 4A list the top 10 journals ranked by publication quantity. The top 5 prolific journals were Circulation Research (count: 52, IF: 20.1, H-index: 25), Circulation (count: 45, IF: 37.8, H-index: 25), Frontiers in Immunology (count: 44, IF: 7.3, H-index: 10), Frontiers in Cardiovascular Medicine (count: 43, IF: 3.6, H-index: 9), and Nature Communications (count: 42, IF: 16.6, H-index: 23), respectively. Among these 10 academic journals, 6 journals belonged to the Q1 category of JCR quartiles. There were 4 journals from the USA, and three from Switzerland and the United Kingdom (UK), respectively. According to Bradford’s Law, the grey box refers to core journals in this field, as shown in Fig. 4C. Figure 4D is the journal’s dual-map overlay analysis, which revealed the distribution of journal topics. The map's left section illustrates the citing journals, while the right section represents the cited journals. The labels of various color clusters suggest the discipline of the corresponding journals. The thickest path from left to right is the citation trajectory, reflecting the flow of citations. The horizontal axis of the ellipse is positively correlated with the number of contributing authors while the longitudinal axis is positively correlated with the journal's output. As shown in Fig. 4D, there were three main citation paths, containing two yellow paths, and one green path, respectively. Two yellow path suggested studies published in Molecular/Biology/Genetics journals, and Health/Nursing/Medicine journals, were generally cited by studies in Molecular/Biology/Immunology journals.
3.7 High-cited publications and co-cited references
In Table 4, we list the top 10 high-cited publications, with all articles cited over 320 times. There was one article entitled ‘Single-cell RNA-seq data analysis on the receptor ACE2 expression reveals the potential risk of different human organs vulnerable to 2019-nCoV infection’ that received the most citations (n = 1,426) by the primary author of Xin Zou in 2020 [22]. This study provided an overview of 2019-nCoV infection-related vulnerable organs according to ACE2 expression using state-of-art single-cell techniques [22]. The article that ranked second was ‘Cells of the adult human heart’ with 659 citations by the primary author Monika Litviňuková in 2020 [23]. They characterized 6 anatomical adult heart regions and provided a human cardiac cell atlas by combining single-cell and single-nuclear RNA-seq data with machine learning and in situ imaging techniques [23].
In Table 5, we list the top 10 co-cited references. We found that the majority of these co-cited references focused on continuous technological advances and data analysis of scRNA-seq itself. As shown in Figure S4, Cluster “#0 Congenital Heart diseases”, Cluster “#1 vascular smooth muscle cell (SMC)”, and Cluster “#7 myocardial infarction” were still hotspots in this field. Based on references cited frequency in a certain period, we set at least 2 years of burst duration in CiteSpace and detected the top 50 references with the strongest citation bursts, as shown in Figure S5. The article with the strongest burstness (strength = 21.94) was entitled “Highly Parallel Genome-wide Expression Profiling of Individual Cells Using Nanoliter Droplets” [24]. This paper described Drop-seq, a strategy for quickly profiling thousands of individual cells by separating them into nanoliter-sized aqueous droplets, associating a different barcode with each cell's RNAs, and sequencing them all together [24]. “Wang Y, 2019, ARTERIOSCL THROM VAS, V39, P876” [25], “Pedroza AJ, 2020, ARTERIOSCL THROM VAS, V40, P2195” [26], and “Miao YF, 2020, CELL STEM CELL, V27, P574” [27] were the recent high-citation references. We found that two out of the three articles focused on the exploration of SMC.
3.8 Frequency and clustering analysis of keyword
A total of 4,751 keywords were identified by VOSviwer. If these keywords had similar meanings, they were merged for subsequent analysis. Table S1 lists the top 20 keywords by frequency of occurrence. The high frequency of keywords can be classified: mechanism-related keywords (expression [n = 470], inflammation ༻n = 183༽, differentiation ༻n = 110༽, activation ༻n = 109༽, proliferation ༻n = 80༽, mechanisms༻n = 66༽, heterogeneity༻n = 54༽), three cardiovascular diseases (atherosclerosis༻n = 200༽, myocardial infarction༻n = 55༽, and heart failure ༻n = 54༽, three cell types (macrophages༻n = 102༽, fibroblasts ༻n = 59༽, SMC༻n = 55༽), and scRNA-seq (n = 174). Among these keywords, 70 met the threshold of 20 occurrences and were grouped into 4 groups (Fig. 5). Group 1, represented in green, focused on the exploration of the mechanism of atherosclerosis, especially the immune system, and was composed of the following keywords: atherosclerosis, dendritic cell, macrophage, monocytes, t-cell, apoptosis, inflammation, mechanism, and vascular SMC. Group 2, represented in red, focused on cell behavior and molecular biology in CVDs and was composed of the following keywords: activation, angiogenesis, dysfunction, proliferation expression, inhibition, growth, identification, receptor, gene, protein, and cell. Group 3, represented in blue, focused on the application of scRNA-seq in CVDs and was composed of the following keywords: cardiomyocyte, atlas, differentiation, heterogeneity, mutation, phenotype, pluripotent stem cell, progenitor cell, and reconstruction. Group 4, represented in yellow, focused on heart failure and myocardial infarction and was composed of the following keywords: heart failure, myocardial infarction, repair, fibroblast, fibrosis, injury, extracellular matrix, repair, and regeneration. Figure 5B shows the visualization of the time-overlapping of keywords. Immune cells(macrophages, monocytes, and t-cells), heart failure, and inflammation have been emerging topics recently.
As we can see in Fig. 5C, 7 of 10 clusters (Cluster #0 atherosclerosis, Cluster #1 differentiation, Cluster #2 precision medicine, Cluster #3 myocardial infarction, Cluster #4 sars-cov-2, Cluster #6 oxidative stress, Cluster #8 association) were still ongoing. In the keyword burst analysis (Fig. 5D), we detected ‘gene expression’ with the most vigorous citation burst (strength = 7.37) had its citation burstiness from 2014 to 2019. Notably, 3 keywords (low-density lipoprotein, neural crest, and macrophage-like cells) were still in burstness until 2023. Figure 5E shows the trend topics and their prevailing year; we can see that ‘Macrophage’, ‘Fibroblast’, and ‘Aging’ were still hot topics in this field. Additionally, the thematic map of keywords (Fig. 5F) showed that heart development and smooth muscle cells were motor themes, signifying these topics were well-developed and crucial for structuring a research subject.