Emerging trends and new developments in monoclonal antibodies: A scientometric analysis (1980–2016)

ABSTRACT

This article aims to explore the intellectual landscape of the study of monoclonal antibody (mAb), mainly to identify thematic trends, landmark articles and emerging trends involving mAb. This work is based on 4 sets of bibliographic records retrieved from the Web of Science. The final data set, consisting of 7,385 bibliographic records, was combined from the 4 individual data sets. This study explores the document co-citation clusters of 7,385 bibliographic records to identify the origin of mAb and the hot research specialty of this domain by applying CiteSpace software. We examined the mAb evolution from 4 perspectives: (1) Clusters of cited references regarding mAb; (2) Cited authors as contributors to mAb research; (3) Institutions participating in mAb research; and (4) Cited journals regarding mAb. The technical development, drug development and clinical applications of mAbs were analyzed. Through data analysis, we have identified the new directions for the exploration of mAbs, interactions between mAb technologies and diseases, and evolving global collaboration among institutions.

Introduction

Over the past 3 decades the development of monoclonal antibody (mAb) has made great progress. The first therapeutic mAb, Orthoclone OKT3, was commercialized in 1986, which is a murine mAb approved for prevention of kidney transplant rejection. Subsequently, chimeric mAbs were developed to overcome the drawbacks of murine mAb in the 1990s. Since 2002, humanized and fully human mAbs were successfully developed and approved for clinical applications.¹ In addition, the development of bispecifics and antibody drug conjugates (ADC) as additional forms of licensed antibodies has advanced significantly.² By the end of May 2016, 51 therapeutic mAb products have been approved in the US or Europe for the treatment of various diseases. Moreover, there were minibodies, nanobodies and other Ab forms as new R&D developments.² Until now, mAb has made a remarkable transformation from a scientific tool to a useful drug for human therapeutics.

With the technological development of mAbs, the mAb market has grown dramatically way. According to the statistics of Firestone, the global sales of mAbs was about US$98 billion in 2015, which is about 7 times that of the sales in 2005.³ Approximately 43% of mAb sales were for the treatment of immune system diseases, while 35% of these sales were used for tumor therapy and 22% for anti-rejection.³ The sales of mAbs accounted for over half of the total sales of all biopharmaceutical products.³ As research and development investment on mAbs continues to grow, it is expected that mAb will lead the global biopharmaceutical market.⁴

Based on the mechanism of action, therapeutic mAbs can be generally divided into 2 categories, e.g., those designed to modulate immune responses by directly target immune competent cells or molecules, and those designed to target cells or molecules not belonging to the classic immune system. Nevertheless, as a major product of immune cells and an important molecule to execute the effector function of immune cells, any given mAb inevitably has some immunoregulatory effect. In fact, the Fc region of mAbs binds to the receptors expressed by various immune cells, such as natural killer (NK) cells, monocytes, macrophages, and granulocytes, this can result in a modulation of immune responses.² Moreover, although not designed to target cells in the immune system, many mAbs still have the capacity to up- or down-regulate the activation of immune cells. Their immunoregulatory activities are known to contribute to their therapeutic effect.^5–8

The rapid growth of mAb literature presented challenges regarding how to identify emerging trends and new developments of mAb research. As the accumulated mAb literature is becoming extensive, a traditional expert review is not sufficient for providing a comprehensive and deep understanding of this topic. Moreover, the research and development of mAbs has become highly interdisciplinary, and thus it has become even more difficult to acquire a firm understanding of the entire mAb field. Some bibliometric surveys on mAbs have been conducted.^9–11 Through the use of statistical analysis, these studies have reported the main subjects and topics of mAb research by co-word analysis,⁹ international research profiles about antibody drugs¹¹ and the different development stages.¹⁰ However, a scientometric analysis is required to gain a deeper understanding of mAb research. In comparison with bibliometric research that focuses on the literature itself, scientometric analysis can quantitatively measure and analyze all aspects of the literature.¹² In particular, citation data analysis of scientometric research can facilitate the exploration of the field at issue by distinguishing various areas of research and its future directions.¹³ Until now, very little scientometric research has been directed toward mAbs.

Thus, this article aims to explore the intellectual landscape of mAbs research through a scientometric analysis, mainly to identify thematic trends, landmark articles and emerging trends. In addition, the leading institutions of mAb research and development will also be studied.

In addition, based on the above analysis, we knew that not all mAbs belong to immunotherapeutic mAbs. In addition, we also found that a great preponderance and dominance of immunotherapeutic mAbs over simple-therapeutic mAbs exists in the field today. Therefore, it's difficult to separate immunotherapeutic mAbs from all mAbs, and thus we focus on the overall development of mAbs in this study.

This study is guided by a computational approach implemented using the CiteSpace software package. CiteSpace is a Java application for visualizing emerging trends and abrupt changes in the scientific literature.^14,15 This visual analytic system is readily accessible and specifically designed to meet the needs for generating a systematic review of a fast-moving and complex field.¹⁶ One of the key features of CiteSpace is its ability to facilitate the detection and interpretation of emerging trends and transition patterns.¹⁶ Therefore, it is expected that the application of CiteSpace can be used to identify and define the emerging trends and transition patterns of mAb research.

Results

The intellectual landscape

The intellectual landscape analysis was shown by a document co-citation network. This network was generated from the final data set. It contains 1,199 references from the top 100 most cited references per time slice between 1980 and 2016. As shown in Fig. 1, the node represents different cited references. The modularity Q 0.8871 and mean silhouette 0.5183 suggest a strong inter-cluster connection within the network and considerable partition of the network, which shows a significant network for intellectual landscape analysis.

Figure 1.

The landscape of mAb (a document co-citation network).

The largest cluster in the visualization is #0 antibody drug conjugate, and followed by #1 reshaping human-antibodies and #2 cytotoxic t-cell. Table 1 summarizes the top-cited terms derived from citing articles of the 10 largest clusters in the mAb domain.

Table 1.

Top 10 largest clusters of panorama (time sorting).

Cluster ID	Size	Silhouette	Mean (year)	Label (LLR)
#8	46	0.974	1979	human glioma xenograft model
#6	52	0.975	1980	mouse monoclonal-antibodies
#1	83	0.886	1984	reshaping human-antibodies
#7	52	0.903	1987	phase-I trial
#2	82	0.892	1989	cyto-toxic t-cell
#3	67	0.816	1992	single chain fv
#9	45	0.992	1999	crohns disease
#4	66	0.917	2000	endothelial growth factor
#5	59	0.897	2006	bispecific antibody
#0	98	0.96	2010	antibody drug conjugate

Table 1 shows the top 10 largest different clusters in the mAb domain. It accounted for 54.21% of the target references. The silhouette scores are all over 0.8, which correspond to a relatively reliable quality of these clusters. The publications years of articles covering mABs range from 1979 to 2010. Based on the development characteristics of the field, we can divide these clusters into 3 phases.

The first phase (#8, #6, #1, #7, #2, #3): from the first mAb appearance to the fully human mAb application, mAb continue to be improved. The human glioma xenograft model was used to localize the production and characterization of mAbs.¹⁷ Humanization is a remarkable progress of antibodies as therapeutic reagents. It's a technological breakthrough that has played a fundamental role in addressing the drawbacks of murine mAbs.¹⁸ Since the first murine mAb was introduced to clinical practice until fully human mAbs were enabled, the drawbacks, including allergic reactions, anti-drug antibodies (ADAs), relatively short half-life in humans, poor recruiters of effector function, ADCC, and complement-dependent cytotoxicity (CDC), were improved stage by stage.^19-21 It is apparent that the humanization technology played a major role to increase the human proportion in mAbs.^22,23 Furthermore, with the advent of in vitro phage display technology and the transgenic mouse technology, the fully human mAbs were enabled.^24–28 This significantly reduced immunogenic potential and showed properties similar to those of human endogenous IgGs.²⁹ However, it also should be noted that the immunogenicity cannot be completely removed by any known technology as even fully human mAbs are immunogenic (anti-Id responses) in some of subjects following repeated administrations.

The second phase (#9, #4): the target expansion. Crohn's disease (CD) is a chronic inflammatory disorder of unknown etiology that can affect any portion of the gastrointestinal tract.³⁰ The proinflammatory cytokine of tumor necrosis factor α (TNF- α) plays an important role in the pathogenesis of CD.^31–33 Infliximab is a chimeric anti- TNF- α mAb that was approved by the US FDA in 1998, while adalimumab is a human anti- TNF- α mAb that was approved by the US in 2002. Both of them bind to TNF- α with high affinity, thereby neutralizing its biological activity.^34,35 Vascular endothelial growth factor (VEGF) is an important regulator of physiological angiogenesis during embryogenesis, skeletal growth and reproductive functions, the more implication is been used in pathological angiogenesis associated with tumors, intraocular neovascular disorders and other conditions.³⁶ There are 3 mAbs against VEGF had been approved by FDA. Bevacizumab, which is marketed with the brand name Avastin, was approved by the US FDA in 2004. It's the first mAb against VEGF and implicated in clinical activity against metastatic colorectal cancer.³⁷ Ranibizumab was approved for treating age-related macular degeneration in 2006. Ramucirumab was approved for treating non-small cell lung cancer.

The third phase (#5, #0): the innovation of mAb. Bispecific is a mAb-based therapeutic that interacts with 2 target antigens but not with mAb mixtures. Bispecifics have recently emerged as a novel and exciting way to target multiple antigens, or multiple epitopes within the same antigen.³⁸ The first bispecific catumaxomab (Removab) was approved in 2009 for treating malignant ascites. This bispecific simultaneously targets CD3 on T cells and epithelial cell adhesion molecule (EpCAM) on tumor cells to facilitate the killing of tumor cells.³⁹ The second bispecific blinatumomab (Blincyto) was approved in 2014 for the treatment of leukemia with targeting CD3 /CD19 T cells.⁴⁰ Antibody drug conjugate (ADC) combines a mAb with a highly potent cytotoxic chemical for the treatment of solid tumors. Thus far, only 3 ADCs have been approved for use in humans. The first ADC gemtuzumab ozogamicin (Mylotarg) was marketed in 2000 for the treatment of granulocytic leukemia, but withdrawn in 2010 as its efficacy did not differentiate from chemotherapy alone.⁴¹ Brentuximab vedotin (Adcetris) was approved for the treatment of lymphoma in 2011 with targeting CD30. Ado-Trastuzumabemtansine (Kadcyla) was approved for the treatment of breast cancer in 2013 with targeting HER2.

Emerging trends and research front of mAb: Articles with citation burst

There are a total of 532 references that have citation bursts. We summarized the ones with the strongest burst in the group of articles that started to burst at the same year in Table 2. As shown in Table 2, it summarized the 28 high citation burst references. The contents include the cited paper, the strength of the citation burst, the time frame of the citation burst, the global citation on Web of Science and the summary. According to the research characteristics and chronological order, we can divide these 28 high citation burst references into 4 different categories

Table 2.

References with the strongest citation bursts every year.

	Citation burst
Cited references	Strength	Begin	End	Citation (WOS)	Summary
Herlyn DM (1980)46	4.6086	1982	1991	263	The study of murine mAb provided an possibilities approach to the treatment of anti-colorectal carcinoma antibodies in humans with immunotherapeutic, but the ADCC had not been discussed whether there will be in humans.
Milstein C (1983)47	21.9359	1984	1999	371	The utility of hybrid hybridomas in immunohistochemistry and advance the mAb development.
Staerz UD (1985)48	27.4799	1986	1997	355	Hybrid antibody is the key mechanism of monoclonal antibodies against the T-cell receptor.
Lanzavecchia A (1987)49	29.6084	1987	1996	208	This study describes a general strategy to produce hybrid monoclonal antibodies that are capable of targeting human cytotoxic T lymphocytes (CTL) against any cell carrying the appropriate target antigen.
Brennan M (1985)50	19.222	1988	1997	158	This study describes the preparation of bispecific antibodies by chemical recombination of monoclonal immunoglobulin G1 fragments.
Laemmli UK (1970)51	56.923	1989	1997	237,672	The application of bacteriophage.
Lobuglio AF (1989)52	54.8754	1990	1998	519	Chimeric mAbs were better than murine mAb in clinical trial efficacy of kinetics and immune response.
Nitta T (1990)53	31.1392	1991	1998	225	This study describes the preliminary trial of specific targeting therapy against malignant glioma.
Ziegler EJ (1991)54	31.8062	1992	2000	1,212	The efficacy of human mAb was very good in treating gram-negative and septic shock.
Bird RE (1988)55	22.4002	1993	2000	1,134	This study describes single-chain antigen-binding proteins was used to improve the application of mAb and experted to be better.
Bolhuis RL (1992)56	23.4994	1994	2000	108	This study describes a bispecific mAb was used for the treatment of ovarian cancer.
Riethmuller G (1994)57	30.9766	1995	2001	574	Adjuvant therapy with mAb reduces mortality and prolongs remission in patients with colorectal cancer.
Winter G (1994)58	26.713	1996	2003	1,105	This study describes the application of phage display technology in making antibodies.
Targan SR (1997)59	50.3401	1998	2007	223	This study describes the application of chimeric mAb cA2 for the short-treatment of crohn's diseaseto by targeting tumor necrosis factor α was effective.
Present DH (1999)60	43.196	1999	2006	1,588	The chimeric mAb infliximab with the therapeutic target TNFα is effective for the treatment of Crohn's disease.
Mclaughlin P (1998)61	47.9737	2000	2010	1,864	This study describes rituximab chimeric anti-CD20 monoclonal antibody therapy for relapsed indolent lymphoma.
Rutgeerts P (1999)62	30.7928	2001	2006	777	This study describes the efficacy and safety of anti–tumor necrosis factor antibody (infliximab) being used in treating crohn's disease.
Clynes RA (2000)63	29.5272	2002	2012	1,613	This study describes the mechanism of inhibitory Fc receptors were used to against tumor targets by modulating in vivo cytoxicity.
Hanauer SB (2002)30	34.8937	2003	2012	2,126	The chimeric mAb infliximab with the therapeutic target TNFα is effective for the treatment of Crohn's disease.
Baert F (2003)64	28.7392	2004	2012	1,115	The chimeric mAb infliximab with the therapeutic target TNFα in the treatment of Crohn's disease can increase risk of infusion reactions and reduce duration of response to treatment.
Hurwitz H(2004)37	38.9387	2005	2012	6,048	The humanized mAb bevacizumab with the therapeutic target VEGF to fluorouracil-based combination chemotherapy is effective to patients with metastatic colorectal cancer.
Holliger P (2005)65	33.1524	2007	2016	864	The emerging of engineered antibody fragments and single-domain antibodies in the development of mAb.
Zeidler R (1999)66	17.716	2008	2012	130	The study describes a new class of intact bispecific antibody was effective to kill tumor cell with the simultaneous activation of T cells and accessory cells.
Phillips GDL (2008)67	40.1062	2010	2016	401	The study describes the ADC trastuzumab-DM1 targeting HER2-positive breast cancer.
Younes A (2010)68	34.9501	2011	2016	472	The study describes the ADC brentuximab vedotin (SGN-35) for relapsed CD30-positive lymphomas.
Doronina SO (2003)69	36.2901	2012	2016	410	This study describes the development of potent monoclonal antibody auristatin conjugates for cancer therapy.
Junutula JR (2008)70	59.4501	2013	2016	379	This study describes the therapeutic index can be improved by the site-specific conjugation of a cytotoxic drug to an antibody.
Strop P (2013)71	29.8272	2014	2016	118	This study describes that ADC stability and pharmacokinetics can be significant impacted by the conjugation site in a species-dependent manner.

From 1982 to 1987, the initial development of mAbs. The emergence of hybridoma technology directly contributed to the development of mAb, and the in vitro production of murine mAbs from hybridomas was described in 1975.^42,43 At that stage, mAbs were mainly used as clinical therapeutics. Murine mAb was the first generation of mAbs. The application of murine mAb in human cancer provided a possible approach for cancer treatment in humans with immunotherapeutics. However, the drawback of ADCC was very significant from murine mAbs.⁴⁴ It seriously affects the development prospects of mAb.²⁰

From 1988 to 1993, mAb research was in a rapid development stage. Chimeric mAbs, humanized mAbs and human mAbs were developed to overcome the inherent immunogenicity and reduce effector function of murine mAbs in humans. With genetic engineering techniques, the drawbacks of murine mAbs have been satisfactorily resolved. Chimeric mAbs and humanized mAbs performed better than murine mAb in clinical trial efficacy of kinetics and immune response. Furthermore, human mAbs exhibited the best performance.

From 1994 to 2005, the application of mAbs in cancer diagnosis, detection and therapy have made important progress. Many new therapeutic areas have been developed. Meanwhile, the blockbuster drugs rituximab, infliximab and bevacizumab were approved. In addition, it demonstrated that the target CD20, TNFα, EGFR, VEGF and HER2 were suitable therapeutic targets.

From 2007 to 2014, new technologies and applications were developed for mAbs. The emergence of engineered antibody fragments and single-domain antibodies (nanobodies), bispecifics and ADCs were developed to improve the performance of mAbs. Particularly, bispecifics are expected to revolutionize mAb therapy with the progress of the immune cell re-targeting and synergistic efficacy through engagement of multiple targets.⁴⁵ The development of bispecific and ADC were significant. The development of ADCs are significant. Much research is focused on the ADC design, especially in terms of the optimization of the ratio of cytotoxin molecules per mAb. In addition, the prophylaxis or treatment of human or avian influenza infections and HIV were the new directions explored at this stage.

In general, we can summarize the development of mAbs as proceeding through 4 stages. In the first stage, researchers attempted to find new treatment approaches for serious diseases through clinical trials. The first generation mAbs were developed, but they exhibited serious drawbacks, which limited their clinical applications. In the second stage, a new generation of mAbs was developed to overcome these drawbacks. During the third stage, researchers mainly focused on targeting diseases. Researchers explored the effectiveness of these treatments on the different targets. During the fourth stage, researchers attempted to develop new technologies and applications for mAb.

Leading institutions

Figure 2 shows the top institutions on publishing number label. There are a total of 332 institutions in the institutional cooperation network. These include 176 universities (53.01%), 67 enterprises (20.18%), 66 research institutes (19.88%), and 23 hospitals (6.93%). As the thickness of a ring reflects the number of publications, we can determine the frequency of publications for each institution. As Fig. 2 shows, Genentech (Genentech Inc.) stands out with the largest circle, but the earliest research on mAbs was not performed at Genentech. A large number of publications were produced by researchers at Genentech in recent years, which were followed by the number of publications produced by researchers at NCI (National Cancer Institute) and Harvard University.

Figure 2.

Top published institutions.

Actually, there are 62 clusters in the institutional cooperation network. Table 3 summarizes the top 5 clusters. The composition of the largest cluster #0 includes 70.45% universities, 13.64% enterprises, 9.09% research institutes and 6.82% hospitals. Universities account for the main component, the proportion of enterprise and research institute are similar in these clusters. This implies that the collaboration among universities is characterized densely and the collaboration between universities and enterprises or research institutes is the key relationship.

Table 3.

The summary of the top 5 largest clusters.

Cluster	Size	University	Enterprise	Research institute	Hospital
#0	44	70.45%	13.64%	9.09%	6.82%
#1	36	63.89%	19.44%	11.11%	5.56%
#2	35	45.71%	22.86%	25.71%	5.71%
#3	25	48.00%	16.00%	32.00%	4.00%
#4	18	61.11%	11.11%	27.78%	0.00%

Top cited journals

Table 4 depicts the high specialty from the view of top cited journals. This includes the top 20 journals and all had a cited frequency of over 1,400. Based on the categories of the journals, it is apparent that immune system diseases, tumor therapy and anti-rejection are the central research themes involving mAbs.

Table 4.

Top 20 cited journals.

Cited frequency	Centrality	Year	Cited journal
5025	0.54	1981	Proc Natl Acad Sci USA
4475	0.46	1982	Cancer Res
4012	0.32	1981	J Immunol
3817	0.09	1982	New Eng J Med
3782	0.1	1982	Nature
3461	0.07	1982	Science
3159	0.19	1985	J Biol CHem
3084	0.08	1982	Blood
2891	0.15	1986	J Clin Oncol
2631	0.05	1998	Clin CANCER RES
2588	0.1	1981	Lancet
2312	0.04	1981	J Exp Med
2300	0.04	1981	Int J Cancer
2123	0.04	1985	J Immunol Methods
1955	0.1	1981	J Clin Invest
1662	0.01	1982	Brit J Cancer
1617	0.23	1998	Nat Biotechnol
1608	0.02	1986	Cell
1574	0	1998	Nat Med
1440	0.03	1981	Cancer

Discussion and conclusion

In this study, we explored the structure and dynamics of citation networks and thematic trends involving mAbs. We examined the mAbs from multiple perspectives: (1) Clusters of cited references in mAbs; (2) Cited authors as contributors to mAbs; (3) Institutions of mAbs; and (4) Cited journals of mAbs.

Based on the analysis of 9,827 bibliographic records, we found that research on mAbs has mainly been focused on their technical development, drug development and clinical application. With the integration of hybridomas, genetic engineering techniques, phage display techniques, transgenic mouse strains, the emergence of engineered antibody fragments, and single-domain antibodies (nanobodies) in mAbs, it appears that therapeutic mAbs will become more accessible than ever before.

During the early stage, research was concentrated on human mAbs, with particular emphasis on their transformation from scientific tools to powerful human therapeutics. The Hybridomas technique is the basic and original technology for mAbs. However, the serious drawbacks of murine mAbs have severely limited their development. Genetic engineering techniques are used to graft the entire antigen-specific variable domain of a mouse Ab onto the constant domains of a human mAb.⁷² This effectively addressed the problem of immunogenicity, but the ADAs of chimeric mAbs still cannot be ignored. Therefore, the grafting of murine hypervariable regions onto a human mAb framework was performed via genetic engineering techniques to further improve their effectiveness.⁷³ The application of a phage display technique and transgenic mouse strains technique enabled the use of fully human mAbs. The transformation from scientific tools to powerful human therapeutics achieved landmark success. The transformation process with related technologies and its practical effect assessment were the research front. These can be verified by the cited references clusters from 1982 to 1993.

At a later stage, researchers attempted to expand the therapeutic usage of mAbs and enhance their effectiveness. The prevention of kidney transplant rejection was the first application of mAbs, which was followed by the treatment of colorectal carcinoma, breast cancer, gastric cancer, lung cancer, Crohn's disease and other tumors. The treatment of diseases associated with immune disorders, tumor and transplant rejection were the mainly therapeutic area of mAbs. Moreover, bispecifics and ADCs are novel mAb-based therapeutics which are expected to revolutionize mAb therapy with the significant development in clinical. Bispecifics and ADCs are likely to expand the spectrum of disease areas, and will hopefully be able to address significant untreated diseases. In addition, the prophylaxis or treatment of human or avian influenza infections and HIV represented new directions of mAbs research. These can be verified by the cited references clusters from 1994 to 2014.

From the analysis of articles with citation burst, we also find the evidence of interactions between technology development and diseases treatment. At first, researchers chose the targets in the process of the use of mAbs for treating tumors. However, when researchers found a variety of targets, they attempted to explore what diseases can be treated with specific targets.

This is a scientometric analysis to explore the intellectual landscape of the study of monoclonal antibodies (mAbs), however, some research limitations are notable. Firstly, in this study, we only chose bibliographic records of Web of Science as analysis sample. We realized that PubMed has the most extensive coverage of scientific literature on biomedicine, but the references are not available. Therefore, we only retried Web of Science core database to obtain uniform references. In the future research, we will try to integrate scientific literature of PubMed database to the analysis, the co-occurrence keywords are a promising direction. Secondly, this study only focused on the bibliographic records, but did not involve cross analysis with the R&D projects of drug development. A future study could extend the current findings to incorporate a cross analysis of mAb research with the data from the discovery stage research to availability on the market.

In a summary, this study analyzed the structure and the evolution trend of the mAb field over time through an analysis of scientific publications and document co-citation clusters using the CiteSpace program. Our findings have revealed that mAb research and development has reached a mature and diversified stage. While researchers have more opportunities to utilize accumulated mAb science and technology to develop therapeutic mAbs, they need to further specify their research target and integrate interdisciplinary technologies to achieve breakthroughs in mAb research.

Method

Data collection

For data collection, we first used literature search strategies by considering mAb types. Now there are 4 types of mAbs, including murine mAbs (suffix: -omab), chimeric mouse-human antibodies (suffix: -ximab), humanized mAbs (suffix: -zumab), and human mAbs (suffix: -umab).^2,45 Moreover, bispecifics and ADCs are the emerging mAb-based therapeutics and have been an important component of mAbs.^2,45 Accordingly, we constructed 4 different search strategies. D_A was developed to retrieve murine mAbs articles as the data set D_A, while D_B was developed to retrieve chimeric mouse-human antibodies as the data set D_B. Meanwhile, D_C was developed to retrieve humanized mAbs as the data set D_C, D_D was developed to retrieve human mAbs as the data set D_D, D_E was developed to retrieve bispecifics as the data set D_E, and D_F was developed to retrieve ADCs as the data set D_F. All of the data sets were retrieved by a topic search (TS) on the ISI Web of Science Core Citation Database with a time span ranging from 1980/01/01 to 2016/04/09. Table 5 summarized the search strategies.

Table 5.

The search strategies used in the database.

Dataset	Types	Search strategies
DA	murine mAbs	TS=((“mouse antibody” or “murine antibody” or “mouse antibodies” or “murine antibodies” or “mouse monoclonal antibody” or “murine monoclonal antibody” or “mouse monoclonal antibodies” or “murine monoclonal antibodies”) and (“therapeutic” or “immunotherapeutic” or “clinical” or “clinic” or “therapy” or “therapies” or “therapeutics” or “immunotherapy” or “treatment” or “immunotherapeutics”)) AND Type=(article) AND Language=(English)
DB	chimeric mouse-human antibodies	TS=((“chimeric human mouse antibody” or “chimeric human mouse antibodies” or “chimeric mouse human antibody” or “chimeric mouse human antibodies” or “chimeric antibody” or “chimeric antibodies” or “chimeric monoclonal antibody” or “chimeric monoclonal antibodies”) and (“therapeutic” or “immunotherapeutic” or “clinical” or “clinic” or “therapy” or “therapies” or “therapeutics” or “immunotherapy” or “treatment” or “immunotherapeutics”)) AND Type=(article) AND Language=(English)
DC	humanized mAbs	TS=((“humanized antibody” or “humanized antibodies” or “humanized monoclonal antibody” or “humanized monoclonal antibodies”) and (“therapeutic” or “immunotherapeutic” or “clinical” or “clinic” or “therapy” or “therapies” or “therapeutics” or “immunotherapy” or “treatment” or “immunotherapeutics”)) AND Type=(article) AND Language=(English)
DD	human mAbs	TS=((“human antibody” or “human antibodies” or “human monoclonal antibody” or “human monoclonal antibodies”) and (“therapeutic” or “immunotherapeutic” or “clinical” or “clinic” or “therapy” or “therapies” or “therapeutics” or “immunotherapy” or “treatment” or “immunotherapeutics”) not (polyclonal or serum)) AND Type=(article) AND Language=(English)
DE	bispecific	TS=(“bispecific antibody” or “bispecific antibodies” or “bispecific monoclonal antibody” or “bispecific monoclonal antibodies”) AND Type=(article) AND Language=(English)
DF	antibody drug conjugate	TS=(“antibody drug conjugate” or “antibody drug conjugates”) AND Type=(article) AND Language=(English)

As there were duplicate bibliographic records in the 4 individual data sets, we cleared the duplicates with the “remove duplicates” function of native CiteSpace. Finally, we combined all the 4 individual data sets together as the final data set D_ABCD for the scientometric analysis, which included 7,385 bibliographic records. Table 6 summarizes the final data sets collected.

Table 6.

A summary of the data sets collected.

Dataset	Collection	Duration	Results	References	Keywords	Authors	Institutions
DA	TS	1980–2016	2,507	62,703	10,737	14,127	1,917
DB	TS	1980–2016	1,278	29,310	5,164	7,250	1,018
DC	TS	1980–2016	1,689	45,804	7,275	10,314	1,285
DD	TS	1980–2016	2,376	64,099	9,801	14,663	1,700
DE	TS	1980–2016	1,726	36,840	5,904	7,332	1,667
DF	TS	1980–2016	874	30,684	2,473	3,038	959
DABCDEF	Combined	1980–2016	9,827	202,274	26,500	44,929	7,757

Data analysis

This study used several scientometric and visual analytic methods with CiteSpace. The time span of the bibliographic record is approximately 37 y. We specify 3 y as the length of a single time slice. The Minimize Spinning Tree (MST) was chosen for network pruning.

This study mainly focused on a document co-citation network and an institutional cooperation network based on the scientific literature. The top 100 most cited references per time slice were selected to map the document co-citation network in a standard graph view. In the document co-citation network, the link indicates how frequently 2 articles are cited together by other articles in a data set, such as the data set D_ABCD. The interconnectivity nodes can be aggregated into clusters. The quality of the document co-citation network is measured by the modularity Q score from 0 to 1. A well-structured network would thus exhibit a high modularity.⁷⁴ The homogeneity of the clusters is measured by the silhouette score ranging from −1 to 1. If the silhouette score is closer to 1, it would correspond to a highly distinct or homogenous group in comparison with other clusters.

A different cluster represents a unique and distinct specialty or a different thematic concentration. This clustering analysis can identify the salient conceptual structures. Based on the concept of burst detection, it can analyze the thematic trends.⁷⁵ Citation bursts of papers imply that the appearance of the paper increases sharply with reference to its peers, and it used to identify emerging trends in terms of highly cited landmark articles.⁷⁶ Persson stated that “the citing articles form a research front, and the cited articles constitute an intellectual base.”⁷⁷ However, another scholar claimed that the features of a research front are the emerging trends and abrupt changes.¹⁴ Therefore, when we explored the high frequency of cited references among document co-citation network, it is of practical significance to highlight the salient theme and contributors of the mAb research field, and provides evidence to explain how the focus of a domain changes over time.

The institutional cooperation network can be used to trace the collaboration among a variety of institutions. These institutions are the main body of bibliographic record. In other words, these institutions are the main body of mAb technical activity. By analyzing the institutional cooperation network, we can gain a further understanding of the cooperation among different institutions, even the cooperation contents and cooperation directions. In additional, analysis of highly cited journals helps to provide an understanding of the main application areas.

Disclosure of potential conflicts of interest

The authors declare that there is no conflict of interests regarding the publication of this paper.

Funding

This work was supported by the University of Macau (MYRG2016–00055-ICMS-QRCM).