Our mission as medicines regulatory agency is to protect and promote public health. We achieve our mission through regulatory science, which underlies the objective evaluation of the safety, efficacy, and quality of medical products and supports science-based decision-making. The development of standards and regulatory guidance accelerates product development and regulatory review to make innovative products available to the public in a timely manner. At the same time, regulatory agencies are confronted with emerging technologies that may have issues beyond the expertise gained from existing medical products. Novel modalities in biological or cell-based products represent a regulatory challenge in terms of efficacy, safety, and quality because of the heterogeneous nature of the product and the multifaceted mode of action. For example, the use of intestinal microbiota as biological products poses a gap to be filled, as they do not reach the systemic circulation but rather modulate mucosal immunity (1). To respond to innovation, medicines regulators worldwide, including in Europe (2) and Japan (3), explore many ways for horizon scanning and cooperate via an international framework, the International Coalition of Medicines Regulatory Authorities (ICMRA) (4). As described previously, it is common to use scientific literature, committees, expert groups, the web, and Delphi methodologies to identify innovation (5). Thus, a comprehensive and transparent methodology is required.

To complement foresight capacity, text mining of a dataset of scientific publications provides a tool for the early identification of emerging technologies, as discussed for Tools for Innovation Monitoring in Europe which makes an overall science survey (6). Text mining technique has extensively been used by policy-makers (6, 7). A combination of text mining and network analysis reported the emergence and evolution of research fronts in biomedical areas (8–11). This strategy, using bibliometric analysis, has the advantage of being supported by scientometric evidence and elucidates paradigms or key elements organizing innovation.

A caveat for searching a database is to determine the appropriate “search term,” which captures the panoramic view of how the key elements of the target field are organized. One solution is to select an encompassing search term that captures the co-evolution of related paradigms (10).

Herein, we focused on the pharmacologic interventions that have been developed for T cell immunity as a case study of bibliometric analysis for horizon scanning. T cells play pivotal roles in the immune system and have therapeutic potential against cancer, autoimmune and/or infectious diseases, and inflammatory conditions. The ability of T cells to form “memory” cells is the fundamental basis of vaccination; however, they are also responsible for harmful reactions such as graft-versus-host disease (GvHD) or donor cell rejection in allogeneic transplants. The first-in-human clinical trial of TGN1412 (monoclonal antibody to co-stimulator CD28), which caused serious adverse effects, highlighted the critical need for regulating the therapeutic ability of T cells and their destructive potential (12). Recently, however, harnessing T cell co-signaling pathways to re-ignite T cell immunity has achieved the practical use as two modalities: one is cellular modality targeting cancer antigens through highly activated chimeric antigen receptor (CAR)-T cells, the other is antibody re-activating endogenous quiescent T cells through checkpoint blockade. These new treatment paradigms prompted us to investigate the development of the core modality in T cell immunity.

To systematically identify novel modalities, we took three steps; network formation with direct citation links, followed by dividing the network into several clusters, finally extracting the characteristic keywords of each cluster. These steps allow us to grasp the overall landscape of T cell immunity, position and interpret each cluster as a distinct technical domain and analyze the targeted clusters with keywords.

This study aimed to explore the possibility of a citation network and clustering in identifying modalities and classes of therapeutics under development. We hypothesized that bibliometric analyses would reveal clusters of distinct modalities in T cell immunity, which would warrant regulatory guidance.

We analyzed a citation network of publications obtained from PubMed, and 38 clusters were formed. The clusters were arranged in descending order of the number of included constituent papers (Supplementary Figure 1); the top 20 clusters were used for subsequent analyses, which covered 95.3% of papers in the citation network.

Table 1 summarizes the cluster keywords, ranked in the top 20 TFICFs, on the recently developed modalities and research fields. Clusters 1, 2, 3, 11, 13, 16, and 20 were chosen because they contained keywords related to the use of T cells for therapeutics. Cluster 1 contained keywords related to immunotherapy, including vaccines, immune checkpoints, CARs, and cytotoxic T lymphocytes. This cluster also had “oncolytic” as a keyword with a lower TFICF. Cluster 2 consisted of keywords related to mucosal immunity, such as microbiota, intestinal, and dendritic cells (DCs). In addition, clusters 1 and 2 were sub-clustered due to the large volume of publications to extract specific topics, showing the detailed character of each cluster. Cluster 3 included keywords on regulatory T cells (Tregs), autoimmunity, and tolerance, while cluster 11 showed exosomes at the top of TFICF. The keywords for cluster 13 were characterized by unconventional T cells, such as invariant NKT (iNKT) cells and mucosal-associated invariant T (MAIT) cells. The keywords for cluster 16 included coronavirus, vaccine, and severe acute respiratory syndrome, while those for cluster 20 comprised mesenchymal stem cells (or mesenchymal stromal cells) (MSCs). Our analysis identified novel modalities classified into each cluster in the citation network.

We characterized the research trends in each cluster by selecting papers on drug development or translational research, which have been published recently (mainly in the last 5 years). The clusters, categorized by modality, are summarized in Table 2. In addition, the clinical study information for each modality was supplemented to validate drug development.

Recent studies on immune checkpoint antibodies were classified mainly into sub-clusters 1-1, 1-3, and 1-6.

The papers on engineered T cells and bispecific antibodies were predominantly compiled in sub-cluster 1-8. T cells genetically engineered to express artificial receptors, such as CARs, have been the subject of intense scrutiny (52, 53). The mechanism of bispecific antibodies is similar to that of CARs: it involves bridging two target cells, thereby bringing immune effector cells into close contact with particular tumor-associated antigens to facilitate cell killing (54). Compared to CAR-T cells in B-cell malignancies, the treatment of solid tumors with CAR-T cells is less effective. CAR-T cell treatment targeting EGFRvIII in glioblastoma resulted in antigen escape because of selection pressure favoring expansion of a subset of tumor cells that lacked the targeted antigen in the clinical trial (55). NY-ESO-1-specific T cell receptor-engineered T (TCR-T) cells have generated clinical responses in patients with synovial cell sarcoma and have received Sakigake and Orphan regenerative medical product designation in Japan (56, 88, 89). Clinical studies on the treatment of solid tumors with TCR-T cells targeting MAGE-A4 (57) or CAR-T cells targeting glypican 3 (GPC3) have been reported (58). Cluster 1 showed research trends on enhancing the antitumor activity of immunotherapy and expanding disease targets, while minimizing adverse events based on the molecular mechanism of immune checkpoint blockade and engineered T cells.

We focused on sub-cluster 2-3 in cluster 2, since it contained unique papers on mucosal immunity, including studies involving intestinal microbiota and commensal bacteria. Although this sub-cluster did not have many recent publications, it included those relevant to the clinical development of microbiota-based products. The top-cited papers describe how commensal microbiota affect specific host T cells (59, 60). A subsequent study reported a preclinical study on the isolation of Treg-inducing bacterial strains from human microbiota (61). Together with the study by Sivan et al. (38), studies on the mechanism of microbiota-host interaction provided evidence regarding the therapeutic potential of selected microorganisms for inflammatory disease and cancer immunotherapy (62). Clinical studies designed to assess the efficacy of microbiota in addressing specific diseases have also been reported (63–66).

Cluster 3 involved studies on Tregs. Sharabi et al. summarized clinical trials of therapies administering Tregs to treat autoimmune diseases, transplantation, and cancer (67). Practical issues related to the isolation and manufacture of Tregs for cell therapy have been noted (68). Clinical studies, on the use of Tregs in treating type 1 diabetes (69) and kidney transplantation (70), have been reported. This cluster revealed clinical applications and hurdles for Treg-based cell therapy.

Cluster 13 comprised papers on iNKT cells that recognize specific glycolipid antigens (alpha galactosylceramides) presented by CD1d protein (71). Innate-like or unconventional T cells include iNKT, MAIT, and γδ T cells, which recognize lipids, vitamin B2 metabolites, and specially modified peptides, respectively. The properties of these cells encompass innate and adaptive immune responses against cancer and infectious diseases (72, 73). Notably, unconventional T cells are considered as non-traditional adjuvants to improve vaccine efficacy and are capable of stimulating a wide array of immune cells (74). Phase I clinical studies on iNKT cells have also been reported (75, 76).

Cluster 16 was distinct in that it consisted of papers on coronavirus and vaccines. The number of papers in this cluster reached a maximum in 2020 (Supplementary Figure 1). Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)-specific T cells in patients with acute respiratory distress syndrome have been characterized (77). The kinetics of immune responses, in relation to the clinical and virological features of a patient with mild-to-moderate coronavirus disease 2019 (COVID-19), have been reported (78). Kim et al. discussed recent evidence on the adaptive immune response against SARS-CoV-2 and its potential implications for the generation of memory responses from the vaccine viewpoint (79).

Clusters 1, 11, and 20 contained papers on extracellular vesicles (EVs), including exosomes which have been the subject of intense scrutiny, with respect to therapeutic applications, because of their capacity for intercellular communication in modulating immune responses (82). Plasma-derived exosomes were found to be predictive of non-invasive biomarkers of immune dysfunction in head and neck cancer (83). Exosomes secreted by DCs have been sought as therapeutic antitumor vaccines in clinical studies (80, 86), while engineered tumor cell-derived exosomes potentiated DC immunogenicity and long-lasting antitumor immunity in preclinical models (81).

Cluster 20 contained papers on MSCs. Stem/progenitor cell-derived EVs exert immuno-regulatory effects on immune cells, such as natural killer (NK) cells, DCs, and T cells (84). The immuno-modulatory activity of MSC-derived exosomes was compared with that of parental MSCs (85). Respiratory diseases were the most common indication in clinical trials registered for MSC-derived EVs therapeutics (90). Clinical studies of exosomes carrying siRNA (87) have also been reported.

Table 3 lists the regulatory guidance documents issued for each modality identified in the present study, as well as the time of approval of the first product. Guidance documents were available for cancer vaccines, oncolytic viruses, microbiota, CAR-T cells and bispecific antibodies, and unavailable for immune checkpoint inhibitors and exosomes (as of July 2021).

Our investigation revealed citation network and clustering captured the structure of T cell immunity field as distinct clusters. Subsequently, our review of knowledge in each cluster brought understanding of research fronts of major modalities. These steps allowed us to assess the needs to develop regulatory guidance for each modality. Our method provides an effective tool for regulators to identify state-of-the-art research fronts to develop guidance documents in a timely manner, minimizing the gap between scientific innovation and product review.

The present bibliometric analysis captured a set of innovative modalities targeted for drug development and revealed several classes of therapeutics of importance. The keywords in the clusters highlight the roadmap for the timely development of regulatory guidance as well as features of research trends that provide important perspectives for subsequent consideration. The citation network offered an efficient and transparent exploratory analysis for horizon scanning that could be considered a starting point for further review and evaluation.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

RW contributed to conception, design of the study, and wrote the manuscript. HS organized the database. AF-S and KO collected data. AF-S performed data analysis. AF-S and MS reviewed and edited the manuscript. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.