Bibliometric Analysis on Potential Biomarkers for Long COVID

1. Introduction

Post-Acute Sequelae of SARS-CoV-2 Infection (PASC), commonly referred to as long COVID or long COVID-19, refers to the persistence of symptoms and health issues in some patients long after the acute phase of COVID-19 infection has resolved (Nalbandian et al., 2021; Davis et al., 2023). This condition has emerged as a significant public health challenge due to its diverse and often debilitating symptoms that affect various bodily systems, including respiratory, cardiovascular, and neurological functions (Carfì et al., 2020; Greenhalgh et al., 2020).

However, the definition of this condition has not been well described yet. For example, the U.S. Centers for Disease Control and Prevention uses the term "PASC" to describe the wide range of health consequences that can persist for four weeks or more following SARS-CoV-2 infection (Centers for Disease Control and Prevention, 2024). Furthermore, the European Society of Clinical Microbiology and Infectious Diseases defines long COVID as the persistence or recurrence of one or more symptoms beyond 12 weeks after a confirmed COVID-19 infection, which cannot be explained by alternative diagnoses (Yelin et al., 2022). Similarly, the National Institute for Health and Care Excellence in the United Kingdom differentiates between ongoing symptomatic COVID-19, occurring from 4 to 12 weeks post-infection, and post-COVID-19 syndrome, which includes symptoms persisting beyond 12 weeks (National Institute for Health and Care Excellence, 2020). These definitions emphasize that long COVID is not a single disease but rather a condition with a broad spectrum of clinical manifestations (World Health Organization, 2021).

Identifying reliable diagnostic biomarkers for long COVID is crucial for early detection, patient stratification, and the development of personalized treatment strategies. Biomarkers, which are measurable indicators of biological states or conditions, can provide valuable insights into the underlying pathophysiological mechanisms of long COVID (Espín et al., 2023; Y.-J. Lai et al., 2023; Nalbandian et al., 2021). However, it’s not well known whether there is a specific biomarker for long COVID exists or which markers are associated with long COVID.

A bibliometric analysis involves the quantitative analysis of scientific publications to measure research output, influence, and collaboration patterns. This approach can help identify high-impact studies, leading researchers, significant research institutions, and most frequently used keywords, providing a detailed overview of the research dynamics (Guler et al., 2016; Ninkov et al., 2022; Pritchard, 1969).

This study aims to conduct a comprehensive bibliometric analysis of the literature to identify available biomarkers associated with long COVID and map the current research landscape, key trends, and research gaps.

2. Methods

2.1. Initial Data Collection

2.1.1. Data Source

A comprehensive bibliometric analysis of the literature on long COVID diagnostic biomarkers involves the quantitative analysis of scientific publications to measure research output, influence, and collaboration patterns. This approach can help identify high-impact studies, leading researchers, and significant research institutions, providing a detailed overview of the research dynamics (Ninkov et al., 2022; van Eck & Waltman, 2014). The Scopus database is exclusively used for literature search due to its comprehensive coverage of global academic information across multiple disciplines and its adherence to Bradford's law, making it particularly suitable for bibliometric analysis (Burnham, 2006).

2.1.2. Keyword Search

Conduct a literature search using the following specific keyword string:

(TITLE-ABS-KEY("Long COVID" OR "Post-acute Sequelae of SARS-CoV-2 infection" OR "Post-COVID-19 Syndrome" OR "PASC" OR "Chronic COVID" OR "COVID-19 long-term effects" OR "Long haul COVID" OR "Long-term COVID" OR "Long term effects of COVID") AND TITLE-ABS-KEY("Diagnostic markers" OR "Biomarkers" OR "Markers" OR "Indicators" OR "Diagnostic indicators" OR "Prognostic markers" OR "Prognostic indicators" OR "Clinical markers"))

The initial search yields 815 documents.

Figure 1. Flow diagram of the literature search and selection process for Long COVID diagnostic biomarker studies. This flowchart illustrates the screening process starting from an initial set of 815 documents identified by the specified keywords (Step 1). Subsequent steps include selecting only documents from certain subject areas (Step 2), limiting language to English (Step 3), restricting document type (Step 4), and filtering by publication years (Step 5). Finally, a manual screening of abstracts (Step 6) was performed. A total of 398 documents fitting “Long COVID” and biomarker research requirements remained for the bibliometric analysis. In the figure, “n” indicates the number of documents remaining after each screening step.

Figure 1. Flow diagram of the literature search and selection process for Long COVID diagnostic biomarker studies. This flowchart illustrates the screening process starting from an initial set of 815 documents identified by the specified keywords (Step 1). Subsequent steps include selecting only documents from certain subject areas (Step 2), limiting language to English (Step 3), restricting document type (Step 4), and filtering by publication years (Step 5). Finally, a manual screening of abstracts (Step 6) was performed. A total of 398 documents fitting “Long COVID” and biomarker research requirements remained for the bibliometric analysis. In the figure, “n” indicates the number of documents remaining after each screening step.

2.1.3. Literature Screening and Selection

A systematic screening process was implemented to identify relevant literature. Initial exclusion removed documents from non-relevant fields including Agricultural and Biological Sciences, Engineering, Physics and Astronomy, Environmental Science, Social Sciences, Earth and Planetary Sciences, Arts and Humanities, and Mathematics (n=807). Subsequently, non-English articles were excluded (n=759), followed by the removal of non-article and non-review types such as conference papers, book chapters, comments, data papers, editorials, and short surveys (n=697). The search was then restricted to publications from January 2020 to May 2024 (n=688). Finally, manual screening excluded studies focusing on acute COVID-19, treatments, other unrelated topics, and those with follow-up periods less than three months, yielding 398 articles for final analysis.

2.2. Data Extraction and Analysis

In this study, the R package "bibliometrix" was utilized for conducting bibliometric analysis. Additionally, "biblioshiny", an app providing a web interface for bibliometric analysis, was employed to extract variable data (Aria & Cuccurullo, 2017). We analyzed and visualized various aspects such as publication trends, journal distributions, author distributions, countries and institutions, collaborative networks, citation impact, and the most frequently cited papers globally, along with performing keyword clustering analysis. Journal impact factors and category information were based on the "2024 Journal Citation Reports" (JCR, 2024).

For visualization, bibliometrix (version 4.1.2) and VOSviewer (version 1.6.19) were used to map co-authorship and keyword co-occurrence among countries, institutions, and authors to enhance the understanding of long COVID research (Z. Lai et al., 2024; Yu et al., 2020). Biomarkers were selected based on frequency of citation (>5% of articles analyzed), clinical relevance in cited studies, and confirmed associations with persistent long COVID symptoms identified in high-impact literature (citation threshold: >50 citations).

3. Results

3.1. Document Types and Source Analysis

From January 2020 to May 2024, research on diagnostic biomarkers for long COVID has shown a significant increase (Figure 2).

A total of 398 documents were published across 212 sources. The annual growth rate of publications is 208.01%, highlighting the rapidly growing interest and urgent need for understanding long COVID diagnostics.

As depicted in Figure 3, the number of articles published per year has increased steadily, with jump starting from 2021 (Figure 3). In 2023, approximately 175 articles were published, and the predicted values for 2024 indicate a continued upward trend, with an estimated 200 articles. This suggests a sustained and growing interest in the field.

3.2. Author and Authorship Analysis

A total of 4,286 authors contributed to the body of work on long COVID biomarkers. The average number of co-authors per document is 12.3, reflecting a high level of collaboration within the research community. Only 11 documents were single authored, indicating a strong preference for collaborative research efforts. International co-authorship accounted for 30.15% of the publications, underscoring the global nature of research in this field (Figure 2).

3.3. Document Age and Research Impact

The average age of the documents is 1.25 years, indicating the recent and rapidly evolving nature of research on long COVID biomarkers. This metric underscores the ongoing advancements and the up-to-date nature of the research outputs. Additionally, the high citation counts for key publications reflect the significant impact and influence of these studies in the scientific community (Figure 2).

3.4. Geographic and Institutional Contributions

The United States emerged as the leading contributor, with 114 publications and significant contributions from institutions including Harvard University, which had 11 publications and 417 citations (Table 1). Other notable contributors included Germany, Italy, and the United Kingdom, with Charité – Universitätsmedizin Berlin and University College London being key institutions in these countries.

3.5. Author Productivity and Impact

Author productivity and impact analysis revealed significant contributions from several key researchers in the field of long COVID diagnostic biomarkers. Figure 4 illustrates the relationship between the number of publications and the total citations received by individual authors, providing insight into both the quantity and influence of their research output. This visualization helps identify the most prolific and impactful researchers in the field. The analysis revealed that a total of 4,286 authors contributed to the body of work on long COVID biomarkers, with an average of 12.3 co-authors per document. Only 11 documents were single-authored, indicating a strong preference for collaborative research efforts. International co-authorship accounted for 30.15% of the publications, underscoring the global nature of research in this field. The most productive authors in this field included Peluso Michael J. from the University of California at San Francisco, and Scheibenbogen Carmen M. from Charité – Universitätsmedizin Berlin, each with 7 publications (Table 2). These authors have significantly influenced the field, as evidenced by their high citation counts and h-indices. (Figure 4)

3.6. Journal Analysis

The top journals publishing research on long COVID diagnostic biomarkers include "Frontiers in Immunology," "International Journal of Molecular Sciences," and "Journal of Clinical Medicine" (Table 3). These journals have consistently published high-impact articles, with "Frontiers in Immunology" leading with 29 publications and a high citation score.

3.7. Highly Cited Articles

The article titled “Long COVID or Post-COVID-19 Syndrome: Putative Pathophysiology, Risk Factors, and Treatments” by Yong S.J. et al. (2021) stands out as the most cited, with an impressive 653 citations (Table 4). Other highly cited works delve into the pathophysiology, risk factors, and long -term effects of COVID-19, underscoring the pivotal role these studies play in advancing understanding and driving progress in the field.

3.8. Keyword and Citation Analysis

The documents featured 1,019 unique keywords, suggesting a diverse range of research topics and approaches within the field. The total number of references cited across all documents is 27,294, indicating extensive academic dialogue and foundational research underpinning current studies. The average number of citations per document is 22.08, demonstrating the high impact and relevance of these studies within the scientific community (Figure 2).

3.9. Network Analysis

The keyword co-occurrence network provided insights into the interconnectedness of various research themes. Central nodes such as C-reactive Protein(CRP), Interleukin-6(IL-6), and Tumor Necrosis Factor-alpha(TNF-α) indicate a predominant focus on inflammatory processes (Figure 5 and Figure 6). The network also highlighted significant connections to clinical symptoms like fatigue, dyspnea, and myalgia, emphasizing the multifaceted nature of long COVID and the need for a comprehensive diagnostic approach.

3.10. Collaborative Networks

The co-authorship network map revealed strong collaborative ties among researchers and institutions across different countries (Figure 7). Major collaborative hubs were identified in the United States and Europe, with extensive links to Asia and Australia. This global collaboration is essential for addressing the complex and multifaceted challenges of long COVID.

3.11. Analysis of the Ten Most Representative Studies in the Field of Long COVID from 2020.1 to 2024.5

The ten most representative studies in long COVID biomarker research were selected based on rigorous criteria emphasizing both scientific impact and methodological quality. These studies were chosen from the highest-cited original research articles (excluding reviews and meta-analyses) within our dataset, with impact factors ranging from 5.5 to 54.4, ensuring inclusion of only the most influential clinical investigations. The selection prioritized prospective and cross-sectional studies that provided robust evidence for biomarker associations with long COVID symptoms, with sample sizes ranging from 42 to 384 patients and follow-up periods extending from median 54 days to over 8 months post-infection.

Table 5 was constructed to systematically present the clinical characteristics and biomarker profiles from these ten pivotal studies, enabling direct comparison of findings across different research groups and methodologies. The table consolidates critical information including study design, sample demographics, symptom prevalence, and specific biomarker measurements, providing a comprehensive overview of the current evidence base.

Analysis of Table 5 reveals consistent patterns across multiple independent investigations. The most frequently reported clinical manifestations include fatigue (ranging from 51% to 100% across studies), cognitive impairment and brain fog (49% to 87%), and respiratory symptoms including breathlessness and persistent cough (34% to 80%). Regarding biomarker findings, inflammatory markers demonstrated persistent elevation, with interleukin-6, tumor necrosis factor-alpha, and C-reactive protein showing sustained abnormalities months after acute infection. Cardiovascular biomarkers including D-dimer, troponin, and natriuretic peptides indicated ongoing cardiac and vascular dysfunction. Neurological markers, particularly neurofilament light chain and glial fibrillary acidic protein, correlated with cognitive symptoms and sensory disturbances.

Based on the convergent findings from Table 5 and the broader literature analysis, we established a four-system biomarker framework that captures the multisystem pathophysiology of long COVID. This categorization was developed because the identified biomarkers clustered into distinct pathophysiological domains: immune-inflammatory, cardiovascular, neurological, and metabolic systems. This systematic approach provides a structured framework for understanding the complex, interconnected nature of long COVID pathophysiology while facilitating clinical translation and future research directions.

Beyond the core findings presented in Table 5, comprehensive analysis of the broader literature (detailed in Supplementary Materials 2) reveals an extensive array of biomarkers spanning multiple biological systems. The immune-inflammatory system demonstrates the most extensive dysregulation, with elevated levels of numerous cytokines, chemokines, and acute-phase proteins. Complement system activation markers, including C1s-C1 inhibitor complex and terminal complement complex, indicate persistent immune activation. Adaptive immune dysfunction manifests through altered T-cell populations and autoantibody production against various self-antigens.

The four-system biomarker framework encompasses distinct but interconnected pathophysiological domains. The immune-inflammatory system, characterized by persistent elevation of IL-6, TNF-α, and CRP, correlates with fatigue, persistent cough, and depressive symptoms. Research by Peluso et al. and Phetsouphanh et al. demonstrates that these inflammatory mediators remain elevated months after acute infection, suggesting fundamental disruption of immune resolution processes.

Cardiovascular system biomarkers, including troponin, NT-proBNP, and D-dimer, reflect ongoing cardiac injury and endothelial dysfunction. Studies by Mandal et al. and Townsend et al. document persistent elevation of these markers correlating with breathlessness, chest pain, and exercise intolerance. The work of Fogarty et al. specifically demonstrates endothelial activation through von Willebrand factor and thrombomodulin elevation, indicating persistent vascular dysfunction.

Neurological system involvement is evidenced through neurofilament light chain and glial fibrillary acidic protein abnormalities, markers that directly correlate with brain fog, cognitive impairment, and sensory symptoms including anosmia and ageusia. Research by Klein et al. and imaging studies by Guedj et al. provide convergent evidence for widespread nervous system involvement extending beyond the acute phase of infection.

Metabolic system dysregulation manifests through insulin resistance markers (HOMA2-IR) and lipid metabolism abnormalities, reflecting cellular energy metabolism disruption. These findings, combined with mitochondrial dysfunction markers, suggest fundamental alterations in cellular bioenergetics that may underlie the persistent fatigue and exercise intolerance characteristic of long COVID.

4. Discussion

The findings suggest that although no specific biomarkers currently exist for long COVID diagnosis, there is strong evidence that dysregulation occurs in multiple biological systems, including immune, cardiovascular, neurological, and metabolic regulations (Figure 8).

Keyword co-occurrence network analysis reveals the centrality of inflammatory markers in long COVID research and their close association with multi-system symptoms. This multi-system biomarker pattern reflects the complex pathophysiology of long COVID, involving interactions across immune, neurological, metabolic, and cardiovascular systems (Nalbandian et al., 2021; Ceban et al., 2022).

The most consistently reported biomarkers in long COVID are inflammatory mediators, particularly IL-6, TNF-α, and CRP. These markers are closely associated with fatigue, persistent cough, and depressive symptoms (Peluso & Deeks, 2024). Their persistent elevation beyond the acute phase indicates fundamental disruption of immune resolution processes, establishing a pathological cycle where initial tissue damage promotes ongoing inflammation (Davis et al., 2023; Lai et al., 2023).

At the molecular level, IL-6 activates the JAK-STAT signaling pathway affecting hypothalamic-pituitary-adrenal axis function, potentially providing a biochemical basis for fatigue symptoms. Similarly, sustained TNF-α elevation correlates with hippocampal neurogenesis inhibition and decreased neuroplasticity, offering a neurobiological explanation for depressive symptoms (Almulla et al., 2024; Arish et al., 2023). The production of autoantibodies, such as antinuclear antibodies, may explain symptom persistence after viral clearance through "molecular mimicry" mechanisms (Chang et al., 2021; Son et al., 2023).

Neurological biomarkers including neurofilament light chain and glial fibrillary acidic protein directly reflect neuronal damage and astrocytic activation, correlating with brain fog, cognitive impairment, and anosmia/ageusia (Bark et al., 2023; Plantone et al., 2024). Abnormal brain metabolism patterns on FDG-PET imaging further corroborate these findings (Guedj et al., 2021; Gutman et al., 2024).

High-sensitivity troponin, NT-proBNP, and D-dimer as cardiovascular biomarkers reflect myocardial injury and endothelial dysfunction, associated with breathlessness, chest pain, and exercise intolerance (Aboughdir et al., 2020; Yaluri et al., 2023). Endothelial cells become direct targets for SARS-CoV-2 via ACE2 receptors, leading to microvascular dysfunction and impaired tissue perfusion (Lindner et al., 2020; Tavazzi et al., 2020). This "endotheliopathy" provides a unifying framework for multi-organ involvement, as microcirculatory dysfunction simultaneously affects cardiac, pulmonary, and cerebral tissues (Bellone et al., 2024; van den Berg et al., 2023). Elevated endothelial activation markers such as von Willebrand factor further confirm this mechanism (Ackermann et al., 2020; Tang et al., 2020).

These neurological biomarkers are closely linked to microglial activation and neuroinflammatory responses, particularly in cognitive-critical regions including the hippocampus, prefrontal cortex, and insula (Díez-Cirarda et al., 2023; Frontera et al., 2022). Notably, these patterns share similarities with those observed in certain neurodegenerative disorders, suggesting long COVID may share common neural injury mechanisms (Douaud et al., 2022; Rogers et al., 2020).

Metabolic dysregulation biomarkers HOMA2-IR and triglycerides reflect cellular energy metabolism disruption in long COVID patients, associated with glucose and lipid metabolism abnormalities (Al-Hakeim et al., 2023; Szögi et al., 2024). Mitochondrial dysfunction represents a core mechanism, manifesting as reduced ATP production, altered membrane potential, and increased oxidative stress (Ayola-Serrano et al., 2021).

Metabolomic studies further reveal alterations in tricarboxylic acid cycle intermediates and lipid peroxidation products, patterns that significantly overlap with those reported in Chronic Fatigue Syndrome/Myalgic Encephalomyelitis (Fernández-Lázaro et al., 2021; Haque & Pant, 2022), explaining the prevalent fatigue and post-exertional malaise in long COVID.

Autonomic nervous system dysfunction is common in long COVID patients, associated with Postural Orthostatic Tachycardia Syndrome (POTS) and other orthostatic symptoms. Biomarkers including altered catecholamine levels, reduced heart rate variability, and abnormal vasomotor responses reflect sympathetic-parasympathetic imbalance (Akbarialiabad et al., 2021; Phetsouphanh et al., 2022).

This autonomic dysfunction interacts with endothelial dysregulation, explaining orthostatic intolerance, dizziness, and palpitations (Su et al., 2022; K. Yin et al., 2024). Autonomic dysregulation may also represent a common mechanism underlying multiple long COVID symptoms.

Coagulation abnormalities constitute another key feature of long COVID, with alterations in D-dimer, fibrinogen, and platelet factor 4 reflecting a persistent hypercoagulable state (Gameil et al., 2021; Mohd Zawawi et al., 2023). This coagulation imbalance may lead to microthrombi formation, further exacerbating tissue hypoxia and organ dysfunction.

Complex interactions exist between coagulation and inflammatory responses, forming an "immunothrombosis" phenomenon (Sollini et al., 2021; Son et al., 2023). This coagulation-inflammation interplay explains the coexistence of multiple symptoms including fatigue, dyspnea, and cognitive impairment (Bellone et al., 2024; Guedj et al., 2021).

5. Limitations

This bibliometric analysis is subject to several limitations. First, our investigation was confined to articles indexed in the Scopus database, potentially excluding pertinent publications—especially those in non-English languages or in journals not covered by Scopus (Falagas et al., 2008; Mongeon & Paul-Hus, 2015). Second, given the rapidly evolving nature of long COVID research, with continuous updates and an influx of new publications, our analysis represents only a snapshot of the research landscape as of the search date. Third, although bibliometric analysis provides valuable quantitative insights into research trends and collaborative networks, it does not capture qualitative dimensions, such as the clinical significance of the findings or the methodological rigor of the study designs. Our analysis, while comprehensive, relies on bibliometric data limited to published and highly cited literature, potentially overlooking emerging biomarkers or those only recently gaining attention. Future reviews could overcome these limitations by integrating bibliometric methods with systematic review approaches, thereby offering a more comprehensive appraisal of biomarker research in long COVID. Moreover, expanding the scope to include additional databases and employing more advanced analytical techniques may further elucidate the evolving research landscape in this field (Davis et al., 2023; Espín et al., 2023; Z. Lai et al., 2024). Future primary research studies should validate biomarkers identified here, especially those showing strong theoretical support but limited current citation impact.

6. Conclusion

Based on our systematic bibliometric analysis, we found that although no specific biomarkers currently exist for definitive long COVID diagnosis, there is compelling evidence of sustained dysregulation across multiple biological systems. Notably, the persistent elevation of inflammatory markers (including IL-6, TNF-α, and CRP), together with abnormalities in cardiovascular and neurological injury biomarkers, suggests a complex pathophysiological process involving multiple organ systems. These findings advance our understanding of long COVID's underlying mechanisms and may inform future clinical investigations.