Global status and trends of metabolomics in diabetes: A literature visualization knowledge graph study

Abstract

BACKGROUND

Diabetes is a metabolic disease characterized by hyperglycemia, which has increased the global medical burden and is also the main cause of death in most countries.

AIM

To understand the knowledge structure of global development status, research focus, and future trend of the relationship between diabetes and metabolomics in the past 20 years.

METHODS

The articles about the relationship between diabetes and metabolomics in the Web of Science Core Collection were retrieved from 2002 to October 23, 2023, and the relevant information was analyzed using CiteSpace6.2.2R (CiteSpace), VOSviewer6.1.18 (VOSviewer), and Bibliometrix software under R language.

RESULTS

A total of 3123 publications were included from 2002 to 2022. In the past two decades, the number of publications and citations in this field has continued to increase. The United States, China, Germany, the United Kingdom, and other relevant funds, institutions, and authors have significantly contributed to this field. Scientific Reports and PLoS One are the journals with the most publications and the most citations. Through keyword co-occurrence and cluster analysis, the closely related keywords are "insulin resistance", "risk", "obesity", "oxidative stress", "metabolomics", "metabolites" and "biomarkers". Keyword clustering included cardiovascular disease, gut microbiota, metabonomics, diabetic nephropathy, molecular docking, gestational diabetes mellitus, oxidative stress, and insulin resistance. Burst detection analysis of keyword depicted that "Gene", "microbiota", "validation", "kidney disease", "antioxidant activity", "untargeted metabolomics", "management", and "accumulation" are knowledge frontiers in recent years.

CONCLUSION

The relationship between metabolomics and diabetes is receiving extensive attention. Diabetic nephropathy, diabetic cardiovascular disease, and kidney disease are key diseases for future research in this field. Gut microbiota, molecular docking, and untargeted metabolomics are key research directions in the future. Antioxidant activity, gene, validation, mass spectrometry, management, and accumulation are at the forefront of knowledge frontiers in this field.

Key Words: Diabetes; Metabolomics; Bibliometric; CiteSpace; VOSviewer

Core Tip: Metabolomics is an important method to study diabetes, and we use bibliometrics to reveal the research status and future trends of the two. Through analysis and summary, diabetic nephropathy, diabetic cardiovascular disease, and kidney disease are key diseases for future research in this field. Gut microbiota, molecular docking, and untargeted metabolomics are key research directions in the future. Antioxidant activity, gene, validation, mass spectrometry, management, and accumulation are at the forefront of knowledge frontiers in this field.

INTRODUCTION

Previously, we believed that diabetes was a disease of developed countries and wealthy people[1]. In contrast, according to a survey by the International Diabetes Federation, it is estimated that 592 million people will have diabetes by 2035, of which 80% are low and middle-income people, which will bring heavy pressure and burden to global health care[2].

Metabolomics technology became increasingly common and received widespread attention in the scientific community[3]. Metabolomics can be used to diagnose disease, explore the mechanism of disease, new targets, and determine drug treatment and treatment results[4], especially in obesity, diabetes, cardiovascular disease, cancer, and neonatal metabolic defects[5-8].

At present, some scholars have used CiteSpace6.2.2R and VOSviewer6.1.18 software to analyze the literature on the correlation between gut flora and diabetes[9], but no one has paid attention to the correlation between metabolomics and diabetes, and we try to fill this knowledge gap. Bibliometrics has become an important tool for evaluating and analyzing the cooperation between universities, the output of scholars, the proportion of funds, and other information[10]. It is widely used in the medical community and provides a reasonable vision for the future through comprehensive measurement and evaluation[11]. This study is based on bibliometrics and visual atlas information to intuitively and systematically analyze the literature, aiming to reveal the publications, countries, institutions, funds, scholars, journals, citations, and keywords related to metabolomics and diabetes in the past 20 years and provide researchers with more detailed data support and future development directions.

MATERIALS AND METHODS

Data source

This study used the Web of Science Core Collection (WoSCC) as the data source, and two researchers searched it. Because of the different retrieval times and the slight increase or decrease in the amount of data, we chose one day to complete the retrieval task and determined the literature retrieval formula as follows: TS = (diabetes mellitus or diabetes or diabetic or diabetic mellitus or hyperglycemia) and TS = (metabolomic or metabolomics or metabonomic or metabonomics). The retrieval date was October 23, 2002-2023. The literature type was articles, and the language was limited to English (Figure 1).

Figure 1  Screening of research publications related to diabetes and metabolomic.

Statistics and analysis

After checking the included documents, we selected "plain text information", which contains the full records of the documents and the references quoted and named them as "download_*" format for export, respectively.

Microsoft Office Excel 2016 was used to make a tree chart and trend chart to describe the overall trend of the field, the number of published papers, and citations in each year. VOSviewer6.1.18 was deployed to describe the cooperation network relationships among countries, institutions, authors, and keywords. Different colors represent different years, nodes represent countries, authors, institutions and keywords, and the number of connections indicates the degree of cooperation. CiteSpace6.2.2R was utilized to predict future trends in the research field. It mainly displays keyword changes through keyword clustering, emergence, and timelines. Bibliometrix was employed to describe the national network map and the top ten authors, countries, institutions, etc.

RESULTS

Analysis of publications and citations

A total of 3123 articles on metabolomics and diabetes-related literature from 2002 to 2022 were included. Metabolomics and diabetes-related literature was first published in 2002 and sorted by year from 2002 to 2022. The annual number of publications and citations displayed an overall upward trend (Figure 2A). In total, 579 articles were published in 2022, with an annual number of 18887 citations. Up to the date of retrieval, due to the continuous updating of data in 2023, there are 459 articles as of October 23, 2023, and there will be more publications in this field in the future. As displayed in Figure 2B, the number of publications increased significantly from 2011 to 2012, 2015 to 2016, 2018 to 2019, and 2021 to 2022, while the annual growth fluctuation of citations generally increased reaching a maximum from 2020 to 2021.

Figure 2 Analysis of publications related to metabolomics and diabetes. A: Quantitative analysis of publications and citations; B: Analysis of annual changes in publications and citations.

Analysis of countries

The included publications were analyzed by country. The national map is revealed in Figure 3. A total of 95 countries have output publications (Figure 3A), displaying the cooperation among countries worldwide in this field. As presented in Figure 3B, a node represents a country, different colors of the circle represent different years, and the size of the node represents centrality. This figure illustrates 92 countries, of which China and the United States have the largest nodes, indicating that they are the most central. In Figure 3C, the VOSviewer displays a national density heatmap of no less than ten articles. The redder the color, the closer the cooperation connection. The countries in darker red are the United States, the United Kingdom, Germany, the Netherlands, Switzerland, and Finland. Regarding publication citations, Figure 3D displays the literature citations of the top 10 countries, followed by the United States, China, Germany, and the United Kingdom.

Figure 3 National analysis of the relationship between metabolomics and diabetes. A: Map of cooperation networks in all countries; B: Map of national central cooperation network; C: Density map of national cooperation network; D: Ranking map of publications in the top ten countries.

Analysis of institutions

The institution map is revealed in Figure 4. According to statistics, 3787 institutions participated in this research. As illustrated in Figure 4A, the most relevant institution is Harvard University (503 articles). According to the change chart of the number of publications issued by the top five institutions with the greatest relevance each year (Figure 4B), the Helmholtz Federation was at the top before 2018, and Harvard University occupied the first place after 2018. As revealed in Figure 4C, there were 180 institutions with more than 500 citations, which can be roughly divided into eight categories. The Chinese Academy of Sciences has the highest number of citations (5413 citations), followed by Duke University (5215 citations), and the University of Helsinki (4906 citations). Figure 4D depicts all institutions with a minimum threshold of 30. Institutions with strong centrality include the Chinese Academy of Sciences and Imperial College of Technology.

Figure 4 institution analysis of the relationship between metabolomics and diabetes. A: Ranking of top ten institutional publications; B: Time chart of top five institutional publications; C: Organization cooperation network map with more than 500 citations; D: Central cooperation relationship diagram of institutions.

Analysis of journals

As depicted in Figure 5A, the top ten journal publications are sorted, among which Scientific Reports has the largest number of publications, with a total of 106 publications, followed by Metabolomics (100 publications), Metabolites (92 publications), PLoS One (81 publications), and Journal of Project Research (78 publications). Figure 5B displays the journal density heatmap with more than 200 citations, which contains 113 journals and is divided into four different color modules. Among them, PLoS One has the largest number of citations (4553 citations), followed by Journal of Project Research (3232 citations), Diabetes (3154 citations), Scientific Reports (2905 citations), and Diabetologia (2783 citations). As revealed in Figure 5C, there are 22 core journals in this research field. We used CiteSpace to superimpose double graphs in the literature. As demonstrated in Figure 5D, the left side represents the collection of citing journals, and the right side represents the collection of cited journals. Citing journals are the knowledge frontiers, and the cited journals are the knowledge base. The diagram has four main paths, two of which are formed by crossing. The research literature on molecular, biology, immunology, medical, and clinical can support the results of molecular, biology, and genetics, and the research literature on medical, clinical, immunology, molecular, and biology can support the results of health, nursing, and medicine.

Figure 5 Journal analysis of the correlation between metabolomics and diabetes. A: Ranking of publications in the top ten journals; B: Density heat map of journal publications with more than 200 citations; C: Core journal collection; D: Journal double graph overlay.

Analysis of funds

According to statistics, analyzing the funds of all publications revealed 3746 fund projects in total. As presented in Figure 6A, the top ten funds are sorted. The different colored squares represent fund projects. The larger the square, the more frequently the funds appear. The top ten fund projects appear 2468 times, accounting for 83.638% of the total funds; the was the United States Department of Health Human Services (605 times). As illustrated in Figure 6B, the United States has three of the top ten funds, with a total frequency of 1341 times, while there is only one related fund in China, with a frequency of 594 times, accounting for 18.762%.

Figure 6 Fund analysis of the correlation between metabolomics and diabetes. A: Map of the proportion of top ten fund; B: Frequency and number of countries in the top ten fund project.

Analysis of authors and co-cited authors

Table 1 and Figure 7A reveal that 18239 scholars participated in this field. According to the ranking of the top ten authors, German scholars occupy the top two, and the number and centrality of publications by Jerzy Adamski scholars are the first (63 articles, 0.11) from Helmholtz Munich Center, Germany, while Guowang Xu (19 articles, 0.2) from Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian, China is the last. A total of 78607 scholars were also co-cited in this field. In Figure 7B, we included co-cited authors with more than 100 citations. As illustrated in Table 2, the first one was Thomas J Wang from Harvard Medical School, with 547 citations. There are two authors in the top ten columns regarding the number of publications and the number of co-cited authors, respectively, Karsten Suhre and Christopher Bang Newgard. As revealed in Figure 7, the numbers of co-cited authors and the cooperative relationship are closer.

Figure 7 Authors analysis of the relationship between metabolomics and diabetes. A: Author collaboration network map; B: Co-citated collaboration network map.

Table 1 The order of the top ten authors involved in the correlation between metabolomics and diabetes mellitus.

Rank	Number of publications	Centrality	Earliest time of posting	Author	Country	Institution
1	63	0.11	2010	Jerzy Adamski	Germany	Hochschule Neubrandenburg
2	31	0.01	2011	Cornelia Prehn	Germany	Helmholtz Center Munich - Environmental Health
3	30	0.02	2011	Karsten Suhre	United States	Weill Cornell Medical College
4	30	0.01	2012	Christopher Bang Newgard	United States	Duke University
5	25	0.05	2012	Robert E. Gerszten	United States	Beth Israel Deaconess Medical Center
6	24	0.05	2011	Gabi Kastenmüller	Germany	Helmholtz Zentrum München
7	24	0.03	2006	Mika Ala-Korpela	Finland	University of Oulu
8	20	0.02	2013	Clary Clish	United States	Harvard University
9	19	0.04	2009	James R. Bain	United States	Duke University
10	19	0.02	2009	Guowang Xu	China	Dalian Institute of Chemical Physics Chinese Academy of Sciences

Table 2 The order of the top ten co-cited authors in the correlation between metabolomics and diabetes mellitus.

Rank	Number of publications	Centrality	Earliest time of publications	Author	Country	Institution
1	547	0.02	2011	Thomas J Wang	United States	Harvard Medical School
2	488	0.06	2010	Christopher Bang Newgard	United States	Duke University
3	371	0.04	2004	Jeremy Kirk Nicholson	United Kingdom	Imperial College of Technology
4	334	0	2008	David Scott Wishart	Canada	University of Alberta
5	312	0	2012	Anna Floegel	Germany	Hochschule Neubrandenburg
6	240	0.03	2011	Karsten Suhre	United States	Weill Cornell Medical College
7	237	0	2014	Peter Würtz	Finland	Nightingale Health
8	211	0.01	2016	Marta Guasch-Ferré	Denmark	University of Copenhagen
9	203	0.1	2003	Oliver Fiehn	United States	West Coast Metabolomics Center
10	197	0.02	2013	Rui Wang-Sattler	Germany	Helmholtz Center Munich

Analysis of highly cited references and co-cited publications

A total of 120514 references were cited in these publications, and the top ten references were counted (Table 3). "Metabolite profiles and the risk of developing diabetes", published by Wang et al[12] in Nature Medicine, has been cited the most (523 citations) in 2011. This article predicts the development of diabetes using metabolomics and correlates the age, body mass index, and fasting blood glucose of 2422 normal individuals and 201 patients with diabetes. Amino acid metabolism is potentially critical in the pathogenesis of diabetes, indicating that the amino acid spectrum can assess the risk of diabetes. The second is "A branched-chain amino acid (BCAA)-related metabolic signature that differentiates obese and lean humans and contributes to insulin resistance", published by Newgard et al[13] in Cell Metabolism in 2009, which has been cited 382 times. In this study, the metabolic characteristics of BCAA related metabolites were determined by metabonomic analysis of obese and thin people, and increased catabolism of BCAA was associated with insulin resistance (IR). It was confirmed that high-fat and BCAA fed mice induced IR with chronic phosphorylation of mTOR, JNK, IRS1 and accumulation of a variety of acylcarnitine in muscle, indicating that BCAA contributed to the development of obesity related IR. In 2013, Floegel et al[14] published " Identification of serum metabolites associated with risk of type 2 diabetes using a targeted metabolomic approach ", which has been cited 275 times in Diabetes. This publication reveals the biomarkers of type 2 diabetes risk through targeted metabolomics. The data demonstrated that glucose metabolism, amino acids, and choline-containing phospholipids are highly correlated with type 2 diabetes in the early stages.

Table 3 Top 10 references in the correlation between metabolomics and diabetes mellitus.

Rank	Ref.	Title	Journal	Source (IF)	Citation	DOI
1	Wang et al[12]	Metabolite profiles and the risk of developing diabetes	Nature medicine	82.9	523	10.1038/NM.2307
2	Newgard et al[13]	A branched-chain amino acid-related metabolic signature that differentiates obese and lean humans and contributes to insulin resistance	Cell Metabolism	29	382	10.1016/J.CMET.2009.02.002
3	Floegel et al[14]	Identification of serum metabolites associated with risk of type 2 diabetes using a targeted metabolomic approach	Diabetes	7.7	275	10.2337/DB12-0495
4	Wang-Sattler et al[60]	Novel biomarkers for pre-diabetes identified by metabolomics	Molecular Systems Biology	9.9	194	10.1038/MSB.2012.43
5	Guasch-Ferré et al[57]	Metabolomics in Prediabetes and Diabetes: A Systematic Review and Meta-analysis	Diabetes Care	16.2	186	10.2337/DC15-2251
6	Benjamini et al[61]	Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing	Journal of the Royal Statistical Society: Series B (Methodological)	5.8	158	10.1111/j.2517-6161.1995.tb02031.x
7	Nicholson et al[62]	'Metabonomics': understanding the metabolic responses of living systems to pathophysiological stimuli via multivariate statistical analysis of biological NMR spectroscopic data	Xenobiotica	1.8	156	10.1080/004982599238047
8	Suhre et al[63]	Metabolic footprint of diabetes: a multiplatform metabolomics study in an epidemiological setting	Plos one	3.7	156	10.1371/JOURNAL.PONE.0013953
9	Menni et al[54]	Biomarkers for type 2 diabetes and impaired fasting glucose using a nontargeted metabolomics approach	Diabetes	7.7	154	10.2337/DB13-0570
10	Gall et al[64]	Alpha-hydroxybutyrate is an early biomarker of insulin resistance and glucose intolerance in a nondiabetic population	Plos one	3.7	145	10.1371/JOURNAL.PONE.0010883

Simultaneously, we counted the top ten high-frequency publications (Table 4), with three publications cited more than 1000 times. The first is "Gut microbiota in human metabolic health and disease ", published by Fan et al[15] in Nature Reviews Microbiology in 2021, which has been cited 1339 times. This article reviews the role of the intestinal microbiota in human metabolic health and disease over the past two decades. Through the joint analysis of intestinal microbiota and multiomics, it aims to optimize the targeted intervention of microbiota in metabolic health to treat metabolic disorders, including diabetes, and understand how the intestinal microbiota affects the host's metabolism. The second is "Sex differences in the gut microbiome drive hormone-dependent regulation of autoimmunity", published by Markle et al[16] in Science magazine in 2013, which has been cited 1262 times. The authors found that in a model of non-obese diabetic mice with type 1 diabetes, early microorganisms determined the level of sex hormones and changed autoimmunity, thus protecting non-obese diabetic male mice from the damage caused by type 1 diabetes. In addition, transferring the intestinal microbiota of adult male mice to immature female mice changes the microbiota of the receptor, resulting in changes in testosterone and metabolomics, affecting insulin and autoantibodies, and preventing the invasion of type 1 diabetes, which indicates the importance of androgen receptor activity. The third is "A purified membrane protein from Akkermansia muciniphila or the pasteurized bacterium improves metabolism in obese and diabetic mice", published by Plovier et al[17] in Nature Medicine in 2017, which has been cited 1128 times. They found that using Akkermansia muciniphila in mice can prevent the development of obesity and complications; however, the underlying mechanism is unclear. Therefore, extracting a specific protein isolated from the outer membrane of Akkermansia muciniphila is relatively stable at pasteurization temperatures and can interact with Toll-like receptor 2 to improve the intestinal barrier, thereby reducing the growth of fat and IR in mice. Finally, they demonstrated that Akkermansia muciniphila which grew in a synthetic medium or was pasteurized was safe for the human body and helped treat human obesity and related diseases.

Table 4 Top 10 co-cited literatures in the correlation between metabolomics and diabetes mellitus.

Rank	Ref.	Title	Journal	Source (IF)	Total citations	DOI
1	Fan et al[15]	Gut microbiota in human metabolic health and disease	Nature Reviews Microbiology	88.1	1339	10.1038/s41579-020-0433-9
2	Markle et al[16]	Sex differences in the gut microbiome drive hormone-dependent regulation of autoimmunity	Science	56.9	1262	10.1126/science.1233521
3	Plovier et al[17]	A purified membrane protein from Akkermansia muciniphila or the pasteurized bacterium improves metabolism in obese and diabetic mice	Nature Medicine	82.9	1128	10.1038/nm.4236
4	Rui Chen et al[65]	Personal omics profiling reveals dynamic molecular and medical phenotypes	Cell	64.5	844	10.1016/j.cell.2012.02.009
5	Wishart[4]	Emerging applications of metabolomics in drug discovery and precision medicine	Nature Reviews Drug Discovery	120.1	792	10.1038/nrd.2016.32
6	Suhre et al[66]	Human metabolic individuality in biomedical and pharmaceutical research	Nature	64.8	744	10.1038/nature10354
7	Newgard[67]	Interplay between lipids and branched-chain amino acids in development of insulin resistance	Cell Metabolism	29	740	10.1016/j.cmet.2012.01.024
8	Floegel et al[15]	Identification of serum metabolites associated with risk of type 2 diabetes using a targeted metabolomic approach	Diabetes	7.7	708	10.2337/db12-0495
9	Freemerman et al[68]	Metabolic reprogramming of macrophages: glucose transporter 1 (GLUT1)-mediated glucose metabolism drives a proinflammatory phenotype	Journal Biological Chemistry	4.8	541	10.1074/jbc.M113.522037
10	Guasch-Ferréet al[57]	Metabolomics in Prediabetes and Diabetes: A Systematic Review and Meta-analysis	Diabetes Care	16.2	535	10.2337/dc15-2251

Analysis of keywords co-occurrence

Keywords are the core of the publication and can be used to analyze the research hotspots of the correlation between metabolomics and diabetes. Using the Bibliometrix software, the network relationships of the top 50 keywords were counted (Figure 8A), which were divided into blue and red. Nodes represent keywords, node size represents frequency, and connection thickness represents the relationship between the keywords. The blue part is closely related to IR, risk, metabolics, and biomarkers, whereas the red part is closely linked to obesity, metabolism, glucose, and disease. Figure 8B displays the frequencies and proportions of the top 50 keywords. A box represents a keyword, and the top ten keywords account for 41%. These included IR (621 times), risk (487 times), metabolomics (317 times), obesity (315 times), metabolism (273 times), biomarkers (257 times), plasma (241 times), disease (239 times), glucose (236 times), and oxidative stress (229 times). In Figure 8C, a node represents a keyword. The higher the centrality, the larger the circle. The color of the circle represents each year, and different colors represent different years. Showing the co-occurrence relationship of all keywords, the red color of the outer ring represents 2022, and the largest red density of the outer ring is gut microbiota. Using VOSviewer to count keywords, the keywords with a frequency of more than 100 times are displayed in the density map (Figure 8D). The redder the color, the higher the density value. A total of 40 keywords were counted, and the color of the metabolomics was the reddest.

Figure 8 Co-occurrence analysis of keywords related to metabolomics and diabetes. A: Network map of top 50 keywords; B: Block map of frequency and proportion of top 50 keywords; C: Centrality network map of keywords; D: Top 40 keywords density heat map.

Analysis of keywords clustering

All keywords were clustered, and eight types of tags were counted (Figure 9A). These include cardiovascular disease, gut microbiota, metabonomics, diabetic nephropathy, molecular docking, gestational diabetes mellitus, oxidative stress, and IR. Figure 9B depicts the broken-line peak diagram of each cluster label with the change in research intensity over time, where each color represents a cluster label. The cluster label cardiovascular disease peaked in 2014-2017, followed by a decline in popularity and a new peak in 2022. The cluster label gut microbiota has been hot continuously since 2010. The kurtosis of cluster label metabonomics was high from 2005 to 2008, after which the heat decreased. Although the clustering label diabetic nephropathy started earlier, its popularity continued to increase around 2010. Cluster label molecular docking was highly popular between 2020 and 2022. However, the relative popularity of cluster label gestational diabetes mellitus, cluster label oxidative stress, and cluster label IR were not high. Figure 9C the first 30 keywords into three clusters using multiple correspondence analysis. The first category of red markers involves important concepts in this field, which is highly consistent with the keywords in the fields related to diabetes and metabolomics. Obesity, mellitus, glucose, metabonomics, and disease are important factors. This group also emphasizes mass spectroscopy, IR, biomarkers, and serum, which have attracted considerable research interest. The category marked in blue is located in the second quadrant of the graph and is closely related to the mechanism of action in this field. The last cluster was green. Amino acids and branched chains are mainly involved in metabolomics. Figure 9D classifies the top 50 keywords into four categories: red, blue, green, and purple. Red and green emphasize the keywords related to metabolomics, purple emphasizes the keywords related to type 2 diabetes, and blue emphasizes the mechanism of action between the two.

Figure 9 Keywords cluster analysis of correlation between metabolomics and diabetes. A: Keywords clustering map label; B: Time variation map of keywords clustering; C: Multiple keywords correspondence analysis map; D: Clustering map of top 50 keywords.

Analysis of keywords bursts

CiteSpace can detect the emergence of keywords, reflecting new academic trends and research directions and revealing potential hotspots in the field. As revealed in Figure 10, it is divided into six parts: keywords, year, strength, begin, end, and time. Blue represents the timeline, and red represents the start year, end year, and duration of the burst. Through the analysis, we included 82 keywords with the highest burst intensity, and the highest burst intensity was "metabolomics (23.17 burst intensity)", followed by "systems biology (15.05 burst intensity)" and "NMR spectroscopy (13.66 burst intensity)". Among them, "metabolomics", "mass spectrometry", "urine", and "identification" have a longer duration. We were interested in burst words from 2018 onwards. It is expected that "gene", "glutathione", "validation", "microbiota", "kidney disease", "untargeted metabolomics", "management", "accumulation", and "antioxidant activity" will become the forefront of the discipline in this field. Currently, they are in an explosive stage.

Figure 10  Analysis of keywords related to metabolomics and diabetes by burst detection, including the top 82 burst words.

DISCUSSION

We searched for publications related to diabetes and metabolomics in WoSCC and retrieved 3123 publications. CiteSpace6.2.2R, VOSviewer6.1.18, and Bibliometrix were used to analyze and evaluate the temporal and spatial distribution, countries and institutions, core authors and literature, research hotspots, and knowledge frontiers. We learned that metabolomics and diabetes related publications were first published in 2002. Since then, the number of publications and citations has increased steadily, particularly since 2017, and the number of publications and citations has increased rapidly each year. It is expected that publications will continue to grow in the future, and the research in this field will continue to attract extensive attention from scholars in the future. The main conclusions are as follows.

Global knowledge structure

We know that the countries, funds, institutions, and authors are not individual entities but an organic whole that has a close relationship with each other. From the perspective of global cooperation, the countries of North America, Europe, and Asia are closely connected and have great influence, while those of South America and Africa are not connected enough, which may be linked to multiple factors such as population, economy, and healthcare.

North America is centered on the United States, which has the second-largest number of publications and the first-largest number of citations, indicating a great influence in this field, which may be closely related to the sponsorship of various funds in the United States, the contribution of major research institutions, core authors, and other factors. The number of fund sponsorships in the United States accounted for 42.356% of the global funds. It was not only strongly supported by the United States Department of Health and Human Services and the United States National Institutes of health but also independent of the Institute of Diabetes, which provided a good economic basis and special funding for scholars' research. In this field, the main research institutions in the United States include Harvard University, Harvard Medical School, Duke University, University of California, University of Michigan, and especially Harvard University, whose publications occupy the first. The comprehensive strengths of these universities provide a better research environment and platform for scholars. The strength of American scholars is relatively high, whether in terms of the number of publications by research scholars or the number of co-citations. The publications and citations of two scholars, Suhre Karsten and Newgard Christopher B, ranked in the top 10, while Thomas J Wang and Newgard Christopher B ranked first and second in terms of co-citations. Consequently, it can be explained that these factors have promoted the influence of the United States in the research on the correlation between diabetes and metabolomics, including the strong sponsorship of the United States Foundation and the high contribution of core scholars and research institutions.

Europe is also a hotspot in studying the correlation between diabetes and metabolomics. In terms of the number of publications and citations in all countries, the European countries accounted for the top 7/10. Germany and the United Kingdom have a significant influence. The Helmholtz Federation of Germany is highly influential. Adamski Jerzy, Prehn Cornelia, and Kasten Mueller Gabi occupy the first, second, and sixth places in the number of publications, respectively, all of which are from the Helmholtz Federation, which has promoted Germany's influence worldwide. In addition to the high proportion of funds in the United States, two of the top ten funds are funded by UK Research and Innovation and Medical Research Council. Jeremy K Nicholson, a scholar from the Imperial College of Technology, ranks third in terms of total citations and second in terms of centrality, indicating that his publications have received widespread attention worldwide.

Asia is centered on China, where the incidence of diabetes in China was observed to be increasingly higher[18,19]. Consequently, Chinese research institutions and scholars are committed to studying diabetes and its complications. The number of publications in China ranked first. Shanghai Jiao Tong University and the Chinese Academy of Sciences are major research institutions with potential. Among them, scholar Guowang Xu ranks among the top ten authors the number of publications. China's National Natural Science Foundation accounts for the second place. These favorable conditions fully indicate that China has great development potential in this field.

Similarly, when we want to deeply understand the research progress, direction, and results of the correlation between diabetes and metabolomics, we first consider the major journals, publications, co-cited references, and core authors in this field, which will bring new ideas and methods to our research. We can learn that most publications we selected first came from the top ten journals, such as Scientific Reports, Metabolomics, Metabolites, and PLoS One. Assuming that the scope is extended to a wider range, we can certainly search it from the core journals of diabetes and metabolomics correlation research (Figure 5C), which includes not only the top ten journals but also journals such as Diabetes Care and Journal of Pharmaceutical, which will provide more specific and extensive information to our research.

Global research priorities and trends

According to the co-occurrence of keywords, we can identify some hotspots and trends in the correlation between diabetes and metabolomics over the past two decades. These core keywords included IR, obesity, oxidative stress, metabolomics, risk, metabolism, and biomarkers.

The discovery of insulin was a milestone[20]. Human insulin is a peptide hormone comprising 51 amino acids and two chains. It is synthesized and released in the Langerhans pancreatic β cells and its main function is to maintain glucose homeostasis. Simultaneously, it is involved in glycogen synthesis, lipid metabolism, DNA synthesis, gene transcription, amino acid transport, protein synthesis, and degradation[21]. Insulin-mediated lack of glucose metabolic control in tissues induces IR[22]. An increasing number of studies have shown that IR is the main pathogenic factor in many metabolic diseases, including type 2 diabetes, obesity, and cardiovascular diseases[23,24]. Indeed, IR is a common feature in obesity and type 2 diabetes mellitus[25]. The prevalence of obesity has rapidly increased worldwide and has become a global public health concern[26,27]. Studies have demonstrated that obesity is not only related to an increased risk of cancer[28] but also increases the risk of type 2 diabetes, cardiovascular disease, musculoskeletal disease, and infection[29]. Obesity is the main risk factor for type 2 diabetes. Simultaneously, the high lipid content caused by obesity is the main factor that promotes IR[30]. IR, obesity, and diabetes are closely related, and they are related to inflammation, adipocyte dysfunction, oxidative stress, endoplasmic reticulum stress, aging, hypoxia, gene changes, and other mechanisms[31,32]. Oxidative stress is the imbalance between the oxidation and antioxidant systems of cells and tissues and results from excessive production of oxidative free radicals and related reactive oxygen species (ROS)[33]. Oxidative stress is associated with almost all major human diseases, including neurodegenerative diseases, cardiovascular diseases, and cancer[34]. Studies have disclosed that oxidative stress is the key mechanism of IR[35] and a key factor in the occurrence and development of diabetes and its related complications[36,37]. Under hyperglycemia, NF-κB, p38MAPK, PKC, and other signaling pathways are activated by ROS, which insulin signaling pathways, including phosphorylation of the INSR, IRS, activation of PI3K, and GLUT4, ultimately leading to IR[38]. Consequently, we can understand that the interaction between IR and oxidative stress is a potential mechanism of diabetes, and obesity and type 2 diabetes are causal to each other, providing a reasonable idea for the follow-up treatment of diabetes and complications. Combined with metabolomics, we can reveal its biomarkers and the mechanism of action behind the back, which is also a hot topic in studying the correlation between metabolomics and diabetes.

Metabolomics can empower diabetes research[39]. Metabolomics has been used to explore diabetes biomarkers to determine the occurrence, development, prognosis, and evaluation of diabetes. Some reports, combined with metabolomics, are from animals, and some reports are from humans. High-fat diet-induced IR in C57BL/6J mice depicted significant differences in lysine, glycine, citrate, leucine, octanoate, and acetate levels, and these metabolites were involved in energy metabolism[40]. Other studies[41] have found that taurine, creatinine, allantoin, and α-ketoglutarate increased in the fingerprint analysis of metabolites in the urine of Zucker obese and normal Wistar rats. In the urine metabolomics experiment of Zucker and GK rats[42], compared with normal rats, the creatine/creatinine, dimethylamine, and acetylacetic acid levels of Zucker and GK rats increased. Compared with Zucker rats, the levels of trimethylamine, acetate, and choline in GK rats decreased, while alanine, citrate, 2-ketoglutarate, succinate, lactic acid, hippurate, creatine/creatinine, and acetylacetic acid increased. There are metabolic similarities between the two stages of type 2 diabetes, including a decrease in the tricarboxylic acid cycle and an increase in ketone body production; the change in energy metabolism is greater in the hyperglycemia stage. In a study of db/db mice[43], carnitine, creatine, trimethylamine-N-oxide, and phenylalanine levels were significantly increased, and these metabolites may participate in the tricarboxylic acid cycle, fatty acid metabolism, and other pathways. In addition, it was reported[44] that based on the urine metabolomics of rhesus monkeys, a new compound, L-piperidic acid, was found, in which glycine, betaine, citric acid, kynurenine, piperidic acid, and glucose were significantly increased. Some scholars also used Akita mice to conduct metabonomic studies[45] on type 1 diabetes mellitus and found that BCAA, valine, leucine, proline, citrulline, and alanine were significantly increased.

Indeed, some studies[46] compare the urine metabolomics of mice, rats, and humans with type 2 diabetes. It is found that citrate, malic acid, fumarate, and aconitate are significantly increased in rats and mice. In contrast, malic acid, fumarate, and succinic acid levels are relatively reduced in humans. The concentrations of ketones and fatty acids are significantly increased in humans, mice, and rats, and the nucleotide metabolism is significantly changed. Similarly, the early and highly cited report in human metabolomics research[13] compared the data of amino acids, acylcarnitine, organic acids, and fatty acids in the plasma and urine of subjects with thin and fat people. Principal component analysis integration disclosed that the biggest differences between the two were leucine/isoleucine, valine, methionine, glutamic acid/glutamine, aromatic amino acids, phenylalanine amino acids, tyrosine, and C3 and C5 acylcarnitine. A linear correlation was observed between the metabolites of BCAA and Homeostasis Model Assessment of IR (P < 0.0001). Through the rat experiment, it was confirmed that the supplementation with BCAA, especially in the high-fat diet, could lead to the occurrence of IR. Previous studies have suggested that BCAA and aromatic amino acids are associated with obesity and IR[47]. There are also reports[48,49] that support this view. The BCAA levels decreased with weight loss. Subsequently, there is evidence[50] that BCAA and related metabolites can distinguish IR and insulin sensitivity between non-obese Chinese and non-obese Asian Indians. BCAA and their related metabolites are correlated with IR, even without obesity. Finally, the disorder of amino acid metabolism may be an early event in IR and type 2 diabetes. In a more valuable study[12], 422 normoglycemic individuals were followed up for 12 years, of which 201 had diabetes. Amino acids, amines, and other metabolites in the baseline samples were analyzed using liquid chromatography-tandem mass spectrometry. Significant differences were found in the five BCAA and aromatic cluster amino acids. Moreover, subsequent reports[51,52] further explained that amino acid metabolism may predict the risk of early diabetes. Certainly, some studies have found metabolites different from BCAA and aromatic amino acids. This study combined liquid chromatography tandem mass spectrometry, which can analyze 70 types of intermediate organic acids, purines, pyrimidines, and other compounds, to determine the risk markers and pathways of diabetes. It was found that 2-aminoadipic acid, which was decomposed by lysine metabolism, was most closely related to diabetes mellitus. The level of 2-aminoadipic acid is not highly correlated with BCAA or aromatic amino acids and has different physiological and pathological pathways. In addition to amino acid, organic acid, and nucleotide metabolism, there are reports of risk markers related to glucose and lipid metabolism. Studies[53] have predicted diabetes risk markers through the lipid metabolism spectrum. Comparing 189 type 2 diabetes and 189 matched disease-free individuals indicated that triacylglycerol with a low carbon value and double bond content was associated with increased diabetes risk. It was also reported that the metabolites in the serum of 9398 men were measured using nuclear magnetic resonance spectroscopy. It was found that glycerol, free fatty acids, and monounsaturated fat were biomarkers for the increased risk of type 2 diabetes. Impaired fasting glucose and glucose tolerance are markers of prediabetes. There are reports based on metabonomic studies of prediabetes[54,55], and α-hydroxybutyric acid is the most relevant marker of impaired glucose tolerance. Oleoylglyceride choline phosphate and oleic acid are selective markers of impaired glucose tolerance. BCKA 3-methyl-2-oxovalerate is the strongest predictor of impaired fasting glucose, which is second only to glucose and independent of glucose. Some studies[56] have found that biomarkers are negatively correlated with diabetes. Two groups of individuals in the Finnish Diabetes Prevention study believed that indole propionic acid is a derivative compound of the intestinal microbiota. It may be mediated by β cell function, plays a role in reducing inflammation, and partially combines with lipid-related metabolites, which have a protective effect on the development of type 2 diabetes. In short, different types of research on diabetes, studies across various regions and races, and investigations into different stages and mechanisms have been conducted in combination with metabolomics. Diabetes related metabolites were also identified, including amino acid metabolism, lipid metabolism, sugar metabolism, organic acid metabolism, nucleotide metabolism, and other pathways. In particular, BCAA and aromatic amino acids are significantly correlated with diabetes. This provides more data to support for promoting clinical diagnosis, treatment, and recovery.

Through keyword clustering, keyword clustering time change, and keywords emergence analysis, we can find the core and knowledge frontiers of future development. Cardiovascular disease, intestinal flora, diabetic nephropathy, and molecular docking in cluster analysis have demonstrated an increasing trend in recent years and are the core of this field in the future. The words "gene", "microbiota", "validation", "kidney disease", "antioxidant activity", "untargeted metabolomics", "management", and "accumulation" are burst words in recent years and are expected to be the knowledge frontiers of the future development of diabetes. Based on the algorithm (log likelihood ratio, p-level) in CiteSpace, closely related popular terms were obstained. The clustering labels for cardiovascular disease included metaeconomics, risk, and coronary artery disease. The clustering labels for gut microbiota included IR, type 2 diabetes, obesity, and high-fat diet. Clustering labels for diabetic nephropathy include chronic kidney disease, type 1 diabetes, kidney disease, and diabetic kidney disease. The clustering labels for molecular docking included antioxidant, identification, mass spectrometry, and alpha-glucosidase inhibition. In the future, research in this field may be conducted with the core keywords of cardiovascular disease, intestinal flora, diabetes nephropathy, and molecular docking cluster. Related popular terms under the cluster label can be cross-studied. Based on the hotspots of past research, combined with the explosive words of diabetes in recent years, we can try to conduct research from a new perspective.

Advantages and limitations

Compared with the previous systematic meta-analysis[57,58] and narrative evaluation[59] of diabetes and metabolomics-related research, this research is the first bibliometric research that has mapped research in this field into a knowledge map in the past two decades. We also used a different bibliometric software for visual analysis, which provided visual feedback and data support. It also combines past hotspots, future core, and knowledge frontiers in this field to provide more possibilities for new topics and directions. It is undeniable that WoSCC is currently the most commonly used and authoritative bibliometric database, but its data resources can only represent a certain field within a certain range. Moreover, due to the continuous updating of the database, the selection of database formats, manual screening and merging methods, and the built-in algorithms of bibliometric software, this inevitably brings about biases in our research. Finally, we believe our results can effectively represent diabetes and metabolomics research from the perspective of bibliometrics.

CONCLUSION

This study downloaded the publications on the relationship between diabetes and metabolomics from 2002 to 2022 through the WoSCC. Based on the bibliometric methods and perspectives, the number of publications in this field has continued to increase. The United States, China, Germany, the United Kingdom, and other relevant scholars, institutions, and funds have significantly contributed to this field. Through keyword co-occurrence analysis, we determined that IR, obesity, oxidative stress, metabolomics, risk, metabolites, and biomarkers were at the core of this field in the past. Diabetes involves the metabolism of amino acids, lipids, sugars, organic acids, nucleotides, and other pathways. BCAA and aromatic amino acids are strongly correlated with diabetes. Through keyword clustering, keyword clustering time change, and keyword emergence analysis, diabetes nephropathy, diabetes cardiovascular disease, and kidney disease are key diseases for future research in this field, with gut microbiota, molecular docking, and untargeted metabolomics being the focus of future research. Antioxidant activity, gene, validation, mass spectrometry, management, and accumulation are at forefront of knowledge reserves in this field.

ARTICLE HIGHLIGHTS

Research background

Diabetes is a metabolic disease characterized by hyperglycemia, which has increased the global medical burden and is also the main cause of death in most countries.

Research motivation

Metabolomics is an important method to study diabetes.

Research objectives

The purpose of this study is to understand the knowledge structure of global development status, research focus, and future trend of the relationship between diabetes and metabolomics in the past 20 years.

Research methods

We downloaded relevant data from Web of Science Core Collection, used CiteSpace6.2.2R, VOSviewer6.1.18, and Bibliometrix software under R language, and summarized the data.

Research results

A total of 3123 publications were included from 2002 to 2022. In the past two decades, the number of publications and citations in this field has continued to increase. The United States, China, Germany, the United Kingdom, and other relevant funds, institutions, and authors have significantly contributed to this field. Scientific Reports and PLoS One are the journals with the most publications and the most citations. Through keyword co-occurrence and cluster analysis, the closely related keywords are "insulin resistance", "risk", "obesity", "oxidative stress", "metabolomics", "metabolites", and "biomarkers". Keyword clustering included cardiovascular disease, gut microbiota, metabonomics, diabetic nephropathy, molecular docking, gestational diabetes mellitus, oxidative stress, and insulin resistance. Burst detection analysis of keyword depicted that "Gene", "microbiota", "validation", "kidney disease", "antioxidant activity", "untargeted metabolomics", "management", and "accumulation" are knowledge frontiers in recent years.

Research conclusions

Diabetic nephropathy, diabetic cardiovascular disease, and kidney disease are key diseases for future research in this field. Gut microbiota, molecular docking, and untargeted metabolomics are key research directions in the future. Antioxidant activity, gene, validation, mass spectrometry, management, and accumulation are at the forefront of knowledge reserves in this field.

Research perspectives

This study used bibliometric methods to explore the application of metabolomics in diabetes, and summarized the global trends and progress.

ACKNOWLEDGEMENTS

To provide the raw data for this work, we thank the Web of Science Core Collection.