Our literature search identified 310 publications focused on PTR, published between 2004 and 2021. The number of published studies increased steadily and peaked in 2020, exhibiting a strong growth trend (Figure 1A). Different colors were used to identify each publication type. The total number of publications was positively correlated with the publication year (r = 0.8197, P < 0.0001). Using curve fit, we mapped the relationship between total publications and the publication year (R² = 0.6792) (Figure 1B). The identified publications included 178 original articles (178/310, 57.42%, cited 2,482 times) and 47 reviews (47/310, 15.16%, 1,226), giving an article-to-review ratio of 3.79:1 (Supplementary Table S1). In addition, 57 meeting abstracts (57/310, 18.39%, 15) were included in this list.

The top ten most productive authors, based on the number of publications, are listed in Table 1. These authors contributed to 82 publications, accounting for 14.38% of the total number of papers. Slichter S. J. (10, 3.23%) was were the most prolific authors in the field of PTR. As shown in Table 1, the research groups led by Slichter S. J. and Santoso S. had a remarkably high h-index. Estcourt Lise had the highest m-index.

Our analysis identified 56 authors for their co-authorship, and all authors had more than three publications and ten citations, which implied the high quality of their publications. Among these authors, we mapped a set of 11 authors who were inter-connected (Figure 2A).

Supplementary Table S2 lists the ten most productive organizations, of which seven are located in the United States, two in the Netherlands and one in Canada. University of Oxford was the most productive organization (n = 16, 5.16%).

The publications in our database involved 481 organizations, with 19 institutions having more than five publications and 10 citations. We analyzed these 19 institutions to show the collaborations between them. Seven institutions were not connected to each other and were excluded (Figure 2B). The six institutions with the highest total link strength were the University of Washington (total link strength, 47 times), Puget Sound Blood Center (37), Canadian Blood Services (34), the University of Toronto (34), Sanquin Research (22), and the University of Amsterdam (22).

Global contributions to PTR research are shown on a world map according to the number of articles published, which directly shows the number of publications from different countries using different colors (Figure 3A). Higher numbers of scientific publications are represented by darker colors. Lines mark the co-author relationship between countries. The ten most productive countries are listed in Table 2. The United States contributed the greatest number of articles (94, 30.32% of all articles), followed by Japan (33, 10.65%) and China (30, 9.68%). The United States also produced the highest number of citations (1,769 citations). Japan had top citations per publication (CPP). The international collaboration rates of the three countries were over 30%.

The co-authorship analysis identified 18 countries with more than five publications, and three countries, which were not connected with each other, were excluded (Figure 3B). The top five countries with the strongest total link strength were the United States (total link strength, 12 times), Netherlands (10), England (9), France (8), and Germany (8).

In total, 310 articles were published in 108 journals. We considered the Journal Impact Factor and Journal Citation Reports (JCR) as important parameters for measuring the various journals. The journal impact factor is a measure of the citations of articles in a specific journal, and it is intended to gauge the importance of a journal in its particular field. First, we listed the top ten most popular journals for publishing articles on PTR (Table 3). Transfusion (59 records, 19.03%) had the most publications, followed by Blood (22, 7.1%) and Vox Sanguinis (21, 6.77%). The number of PTR-related publications in these journals steadily increased over time (Figure 4A), particularly in Transfusion, where the total publication number increased from 2 in 2005 to 59 in 2021 and has become the most productive journal in PTR research. The research area (based on the JCR listing) of the ten journals in hematology is the top research area in the PTR field.

Moreover, Table 4 shows the top ten cited journals. Blood had the largest number of citations (534 citations), followed by Transfusion (430), British Journal of Haematology (332), Cochrane Database of Systematic Reviews (181), and Vox Sanguinis (157). Although the total number of citations of articles published in JCR's top quartile does not show obvious advantages, the number of CPP is higher than that of journals in other partitions. These findings highlight the value of the JCR in identifying influential articles in this field.

Twenty-eight journals were highlighted in our study for documents that were co-cited in more than 50 publications (Figure 4B). Transfusion (total link strength, 44,618 times), Blood (40,329), British Journal of Haematology (17,660), Vox Sanguinis (14,906) and New England Journal of Medicine (8,977) are the top five co-cited journals. We also tried to assess the core sources of PTR research knowledge. Through Bradford's law analysis, we identified Transfusion, Blood, Vox Sanguinis, Transfusion and Apheresis Science as the core journals in PTR research.

Significantly, there were 11 references co-cited in more than 30 articles (Figure 4C). The 11 articles with the highest number of co-citations are listed in Table 5. The five articles with the largest number of co-citations are by Slichter S. J. (2005, Blood; 73 citations; total strength, 224 times), McFarland J. (1997, New Engl J Med; 82, 206), Hod E. (2008, British Journal of Haematology; 68, 193), Doughty H. A. (1994, Vox Sanguinis; 41, 160), and Pertz Ld (2000, Transfusion; 34, 129).

We analyzed historical citations to identify significant publications that contributed to research development (Figure 5A). Figure 5B lists these documents ordered by the number of globally cited documents (GCS) and locally cited documents (LCS). GCS is the total times one document was cited, while LCS is the number of times the document was cited by all documents we collected. The combination of GCS and LCS indicates their association and contribution to the PTR research area. The top five documents are by Slichter S. J. (2005, Blood; 73 LCS, 301 GCS), Hod E. (2008, British Journal of Haematology; 68 LCS, 172 GCS), Slichter S. J. (2015, Transfusion Medical Review; 67 LCS, 104 GCS), Pavenski K. (2012, Tissue antigens; 38 LCS, 92 GCS), and Seftel M.D. (2004, Blood; 23 LCS, 83 GCS).

Citation burst detection identifies publications that receive particular attention in a specific period of time (34). The minimum duration of citation bursts in Cite Space was set to 1 year for the present study. As a result, seven articles with strong citation bursts were identified (Figure 6). The line in Figure 6 represents the period from 2004 to 2021, with the red lines indicating the time interval of the citation burst. The strongest citation burst (strength, 6.61) was associated with an article entitled “Platelet refractoriness–practical approaches and ongoing dilemmas in patient management”, which was published in the British Journal of Haematology by Stanworth et al. The citation burst of this article lasted from 2017 to 2021. Six among the top seven publications with citation bursts that ended in or after 2015 were selected for discussion regarding the latest topics in PTR research.

As the focus on different topics will greatly affect the number of citations, total citation analysis will cause some articles with high contributions and high citations to be lost. Therefore, using 40 keywords, we used factorial analysis to include more articles and highlight articles related to key topics. Multiple correspondence analysis and multidimensional scaling was performed in factor analysis, which can classify factors that underlie different topics, sorted into different clusters. Multiple correspondence analysis is one of the most widely used data dimensionality reduction algorithms, which can be considered as an extension of simple correspondence analysis in low-dimensional graphics (40). After factorial analysis, all 40 keywords were classified into five clusters marked purple, blue, red, orange and green (Figure 7A). Connection between these clusters is shown in Figure 7B. The words in purple cluster are related to “gene”. Gene frequencies determine the types of antigens. Identifying the different types of antigens can help avoid the occurrence of PTR. The red cluster represents words related to the clinical aspect of PTR. Most patients with diseases like acute leukemia often require frequent blood transfusions to increase platelet numbers; however, PTR often occurs in multi-transfused patients. HLA antibodies are the prime risk factor for PTR and the management of patient and blood components can greatly avoid the occurrence of PTR. The green cluster identifies bone marrow transplantation and stem cell transplantation. Both are at risk of PTR, caused by HLA antibodies. The blue cluster includes diseases also caused by platelet transfusion. The orange cluster reflects the PTR patient. According to the classification by factorial analysis, we determined the most cited papers (Figure 7C) and the most contributing papers (Figure 7D) in a particular research topic within PTR.

VOS viewer was used to generate a keyword map that marked phrases mentioned in the titles and abstracts of the publications. Multiple occurrences in a single publication were counted as one event. To clarify the results of the analysis, we merged synonyms. Using co-occurring analysis, which is a statistical analysis method for standardized keywords extracted from publications, all 30 keywords that occurred more than ten times were distributed across four areas, which are shown in different colors, with the node size representing the frequency of occurrence (Figure 8A). Keywords such as “alloimmunization,” “management,” “thrombocytopenia,” “antibodies,” “HLA,” “antigens,” and “platelet transfusion refractor” were found to be significant words in PTR research. The colors in the overlay visualization indicate the average publication year (Figure 8B). Words with green/yellow colors, such as “selection,” “impact,” “reduction,” and “identification,” were published after 2015. The distance between two circles represents how closely two phrases co-occurred in PTR-related publications. Density visualization shows the same keywords mapped by their frequency of appearance (Figure 8C). Words such as “alloimmunization,” “management,” “thrombocytopenia,” and “antibodies” were significantly in PTR-related research.

BICOMB and gCLUTO are tools to classify all keywords shown in hot spots (Figure 9A). All terms were classified into 8 clusters, and connections within clusters are also shown by dendrograms on the axes. The darker the red, the more times the terms occurred. A three-dimensional landform map was also created to reveal the inter-cluster standard deviation (Figure 9B). The curve of each mountain peak is a Gaussian curve, which approximately reflects the distribution of the data in the associated cluster. Position and height reflect the inter-cluster similarity, and the volume reflects the number of terms inside the cluster. The most meaningful information is the color of the peak, which reveals the inter-cluster standard deviation. Red indicates low deviation, while blue denotes high variance.

To better understand the evolution of the research focus in PTR, we mapped keywords in Figure 10A. Keywords such as “antibodies”, “alloimmunization”, “management”, “gene-frequencies”, “CD36 deficiency,” “donors,” and “storage” emerged in PTR research before 2019. New keywords, like “antibodies,” “increments”, and “activation,” emerged in 2019–2021, thus creating a new focus vocabulary. Platelet increment is the criterion for the occurrence of PTR. Furthermore, keywords regarding mechanisms such as antibodies and alloimmunization have been a topic of concern for a long time. Diseases that are prone to PTR, such as acute leukemia and stem-cell transplantation, have always been keywords in the field of PTR. As progress has been made in managing patients and blood products, they were no longer as common as important topics. Platelet activation has been a new hot topic in recent years.

A thematic map was used to determine the significant topics in PTR research (41) (Figure 10B). The minimum frequency (per thousand docs) of each cluster is at least over 10 times. The horizontal axis represents the importance of the topic to the field, while the vertical axis represents whether this topic has been well developed. Topics were sorted into 8 clusters. These 8 clusters were sorted into four themes, including motor, basic, niche, and emerging or declining themes. Density measures (y-axis) the development of the keywords or thematic network, while centrality (x-axis) refers to the degree of importance. In the high-density and high-centrality quadrant, motor themes are the primary research themes that attract the most scientific output and citations. By contrast, emerging themes or those fading out in the lower-left quadrant have less research value. High density and low centrality in the upper left quadrant are highly developed themes but less relevant to PTR. In the lower right quadrant are basic themes; these words are core or essential themes and the topic of intense research.

Here, we combined network visualizations with bibliometric analyses to describe the present situation of PTR research; analyze the contributions of countries, institutions, journals, and authors to this field; and predict topics for continued research interest in the coming years. Since 2004, the annual publication output in the field has shown a steady growth trend (Figure 1A).

Author analysis is based on the h-index and m-index, which are widely used to measure an author's importance (Table 1). The h-index named the “Hirsch core” (37) is considered a significant gauge for quantifying an individual's scientific research output. Additionally, the m-index is divided h-index by academic years (37, 38), namely the number of years since the author first published a paper. Notably, the m-index is unavailable when scientists suspend their productivity, while the h-index remains useful (42). Therefore, when evaluating the influence of an author, it is best to assess both the h-index and the m-index.

The most productive author, the author with the highest CPP and h-index, was Sherrill J. Slichter, who leads a research team (MD, the Blood Works Northwest Research Institute, University of Washington) and is the leading scholar in the PTR field. Based on the m-index results, the most impactful newcomer to the field was Lise Estcourt (MA, Vitalant Research Institute, University of Oxford) and her research group. The United States has the largest number of publications (94 articles) and total citations (1,769), and the fourth CPP rank (18.82). Although the total number of publications (33 articles) and total citations (361) both ranked second, Japan had the highest CPP (30.08). Furthermore, seven of the top ten institutions with the largest number of publications are from the United States, which further confirms the influence of the United States in this field. Transfusion, the most productive journal, and Blood, the most cited journal with the highest CPP, are the most influential journals in PTR, and both are American journals. These results all indicate that the United States is the leader in PTR research, with Japan making a significant impact. We suggest that scholars should pay close attention to the articles from these authors, countries and journals to keep abreast of the cutting-edge research in this field.

To establish an objective and complementary knowledge base of PTR research, we used four bibliographic methods to identify the most significant publications. Co-cited articles (Figure 4C) are two articles that are cited together in another publication (43). Firstly, it measures the association and the proximity of content between publications (44). Moreover, to be cited means this publication is recognized by other researchers (43). Historiography analysis (Figure 5A) is a supplementary analysis that helps visualize significant contributing publications and shows their connection by citation analysis to help us understand the chronological development of the research field. Secondly, compared with GCS, LCS can eliminate the interference of citation times in other fields and better show the value of the article in the field of PTR. Due to the accumulation of publication years, the total citation analysis may cover some newly published important documents. Thus, we also analyzed PTR-related papers for high citation bursts as a supplement by using Cite Space (Figure 6), which does not only combine the number of citations but also reflects the decline of attractiveness over time, resulting in some highly attractive articles (45, 46). Factorial analysis can classify topics of publications and show contributing publications in each cluster. The next part of the discussion explains the literature analysis findings.

The topics of these articles include PTR identification, prevention, and underlying mechanisms, as well as management of blood production and patients. The publication of Forest and Hod (47), Slichter et al. (6) (n = 73), and Hod E. et al. (strength, 6.49) introduced the definition of PTR and described the predominant cause of platelet refractoriness, including transfusion service factors (5). and antigens like HLA (5, 15, 18, 48), CD36 (20, 49, 50) and HPA (51). In fact, studies by Kiefel et al. (52) (n = 37) revealed that platelet-specific antibodies encountered in patients who underwent multiple transfusions were mostly anti-HPA-1b and anti-HPA-5b.

The review by Pavenski et al. (15) in 2012 (strength, 5.36) and Stanworth et al. (10) in 2015 (strength, 6.61) summarized several points regarding the mechanisms of PTR. Overall, PTR mechanisms remain unclear, but key elements are related to the transfused donor product composition and the immune status of the recipient. Using in vivo experiments in mice, Pavenski et al. (15) determined that the platelet-related immune response has at least two independent levels. The first level occurs within the recipient and is related to antigen processing/presentation events and CD8+ T cell-mediated immunosuppression. The second level of immune response relates to the donor product and includes donor antigen-presenting cell levels as well as age-induced changes in donor cells and/or platelets (15). Besides, a new hypothesis suggests that antibody-free platelet clearance can be mediated by CD8+ T cells (15, 53–55). As the relationship between platelet and complement is very close (56–59), some articles also pointed to complement as the main participator in PTR research (60, 61). Kerkhoffs et al. found that patient survival was not correlated with thrombocytopenia but could be related to the level of endothelial cell damage (62). Progress has been made in antibody testing and the management of various blood components; however, there is still a lack of understanding of the mechanisms.

Various techniques help prevent the occurrence of PTR. These were included under the keyword “prevention,” such as using single-donor platelets, matching the antibody of platelets, and using leukocyte-reducing blood components. Firstly, Kopko et al. (63) reported some general methods that can be utilized to identify alloantibodies, with the primary difference between them is sensitivity. The most sensitive method is Luminex or flow cytometry, followed by enzyme-linked immunosorbent and complement-dependant cytotoxicity assays. Patients with hematologic malignancies or solid tumors undergo frequent platelet transfusions, leading to a higher probability of platelet refractoriness. Kerkhoffs et al. (62) (n = 31) reported that about 25–70% of multiple transfused patients had a PTR experience. Comparatively, Pertz et al. (64) (n = 31) revealed that crossmatched platelets must be performed frequently, as platelet has a shelf life of only 5–7 days (65), leading to considerable inconvenience. For patients in need of frequent platelet transfusions, the use of antibody specificity prediction would be useful as it can identify the type of HLA antibodies. The article with the most co-citations (8) (n = 82) discussed leukocyte reduction, which blocks the direct pathway by the removal of the donor antigen-presenting cells (8) so as to effectively prevent alloantibody-mediated refractoriness to platelets; this technique is now routinely used. In contrast, leukocyte reduction has no effect in most alloimmune individuals (14) because the indirect pathway still operates efficiently in these individuals, such as most female patients with a history of pregnancy or in highly immunized patients (10). Patients' HLA antibodies have individual differences, and not every patient with HLA antibodies presents with PTR. Pilar Solves and his group (66) (strength, 3.35) reported that a lower dose of CD34+ cells, excessive amounts of antibiotics, and anti-HLA I antibodies are risk factors for the occurrence of PTR. Furthermore, splenectomy, ABO compatibility (8), standardized plans of platelet transfusion and management of patients are important to avoid the occurrence of PTR. Juskewitch et al. (67) developed an algorithm for diagnosing and managing platelet transfusion-refractory patients (strength, 3.84), which enables overall resource savings to improve the efficacy of transfusions.

To better understand the emerging and hot topics, we used co-occurrence analysis to identify all keywords (Figure 8), classifying these words by co-word bi-clustering analysis (Figure 9), and displayed the most popular recent words in thematic evolution analysis to explore the change of focus in keywords (Figure 10A). However, there are some limitations. Although popular topics are important, these topics may not have enough value for further study. Because of their popularity, some more important trend topics may be ignored. Whereas, analyses such as thematic map, which is based on co-word analysis and the h-index, can help us excavate potential research trends (41) (Figure 10B). In comparison, some antigens and platelet activation, as motor themes, deserve further research.

Platelets express many antigens that have been shown to cause PTR, including HLA-I, HPA, and CD36. As motor themes (Figure 10B), antigens expressed on platelets are important findings in PTR. HPA antibodies are a recognized cause of PTR in 1–10% of patients (68). HPA-15 alloantigen determinants reside on CD109 and can cause ~3% (69) incidence of PTR, exceeded only by HPA-1 system antigens, particularly in multi-transfused patients (69–71). The coexistence of HLA-I can occlude HPA-15 assays (72). CD36, also named glycoprotein IV, has two types of deficiencies (20). Patients with type I CD36 deficiency are transfused with platelets would cause PTR. Beta (3) integrin has several single nucleotide polymorphisms, and when it is encoded by Leu33Pro, it becomes a strong alloimmunization cause of PTR (73). Two further topics also pointed to the activation of platelets, a key process of PTR. P-selectin (CD62P) is expressed after the activation of platelets and can be used as a platelet activation marker (74). Glycoprotein Ib is one of the components of GPIb-IX-IV, which is expressed in platelets and megakaryocytes and is responsible for Von Willebrand factor-mediated platelet activation and aggregation (75).

The mechanism of PTR is still unclear and controversial but is clearly related to the activation (the other motor topic) of the platelet and complement system. One proposed mechanism suggests that when platelets are transfused to refractory patients with incompatible antigens, antibodies can induce platelet activation, then active modulation and passive opsonization of platelets combine to attract phagocytes (76, 77). Another theory is that the binding of C1q, an initial trigger of the classical complement activation pathway, leads to direct killing action and platelet activation (60, 61). The last potential mechanism is based on the hypothesis that complement components may directly contribute to the clearance of platelets, either by directly binding to complement-binding receptors on the platelet surface or by binding to HLA antibodies, leading to phagocytosis by mononuclear cells in the spleen (60). The role of complement is also supported by clinical studies; for example, C1q can better predict the occurrence of PTR (78, 79), and anti-C5 named Eculizumab can help overcome PTR (80, 81).

Although bibliometrics is useful for research performance analysis, it has some limitations. First, analysis is performed based on the publications collected from a database, which means that it might not be comprehensive. Second, citation analysis is a significant part of bibliometric analysis, which supports quantifiable items. This type of analysis is based on a theoretical assumption that scientific quality has a simple linear relationship with citation counts (33). However, the relationship can be influenced by several factors, including the self-citation rate and journal bias. We attempted to use a more comprehensive analysis, including historiography analysis and citation bursts, to provide valid results, but there are still limitations. Finally, we believe that even with these limitations, bibliometrics can greatly assist in acquiring greater knowledge of particular PTR research areas.