We performed a systematic search for the Science Citation Index Expanded (SCI-Expanded) of Web of Science Core Collection (WoSCC) in 2012–2021 during 1 January 2012 and 31 December 2021, and we downloaded the data on 20 September 2022 for avoiding deviations. The search terms included: TS = (autophagy) AND TS= (“pulmonary” OR “lung”). Two reviewers (FH Lin and JX Yuan) took charge of independently identifying this data search and discussing underlying differences, with the final agreement of 0.90 (17). After excluding the online publication time of 2022 and articles written in non-English and by limiting the publication type to reviews and original articles, we finally obtained 4,522 original articles and reviews that met the criteria for inclusion in the analysis (Figure 1).

First, the original data were extracted from the SCI-Expanded database of WoSCC. The recorded information included the number of papers and citations, H-index, publication year, countries/regions, affiliations, authors, journals, references, and keywords. Afterward, duplicate authors and misspelled elements were moved artificially. Although inaccurate analysis may not be avoided completely due to multiple versions of cited references, the same abbreviated names of different authors, and different forms of cited journals, we pointed out the first extracted original data from the SCI-Expanded database. The recorded information was paper and citation number, H-index, G-index publication year, countries/regions, affiliations, authors, journal, references, and keywords. Then, we removed misspelled elements and duplicate authors. Cited references have different versions, different authors may have the same abbreviated name, and cited journals have different forms, which may lead to inaccurate analysis. Despite this, most raw data were accurate. We removed the misspelled elements and duplicate authors and adopted a thesaurus file for merging duplicates into one word, deleting useless words and correcting the misspelled elements. Later, we imported the cleansed data to VOSviewer (version 1.6.18.0), CiteSpace (version 6.1. R3), and the “bibliometrix package 4.0.1” of R software (version 4.2.1) for the subsequent bibliometric analysis.

The number of papers and citations is the bibliometric indicator commonly used to represent the bibliographic material. In general, the research level can be evaluated from the perspective of productivity and impact, which, in the paper, were respectively measured via the number of publications (NP) and the number of citations without self-citations (NC). Sometimes, H-index is used to evaluate the individual academic achievements or the publication output (18). G-index refers to the highest number of papers that receive H-index or more citations (19). In addition, the article value can also be indicated by the impact factor (IF) from the latest version of Journal Citation Reports (20, 21). We visualized the results for the number of papers per year, country, organization, author, citation, and other relevant aspects. Prior to data analysis, the R4.2.1-based Bibliometrix package was used to store, count, and clean up data. As important indicators of the research, H-index, G-index, and IF were all included in the analysis. Subsequently, the following software was used for bibliometric analysis.

R (version 4.2.1) stands for the language and environment specific to the statistical computing and graphics. It presents a strong extensibility, capable of automating the analyses and creating new functions. The cleansed data received bibliometric analysis under the assistance of bibliometrix package in R (22). To further explain the annual document quantity alternation, we employed the fitting polynomial model for predicting the annual NP. Variable f (x) refers to the annual study number, and x stands for the publication year. Moreover, VOSviewer software assisted in constructing bibliometric maps for obtaining more extensive information regarding the results considering co-citation (CC) and co-occurrence (CO) (23, 24). Co-citation refers to the situation that two items are simultaneously cited by the third one. Keyword co-occurrence takes charge of measuring the keywords appearing in the same documents most frequently (25). Analysis on the keywords serves for demonstrating the autophagy-related research in lung disease.

VOSviewer, CiteSpace, and R (version 4.2.1) were used for the statistical computing and graphics. VOSviewer uses data from WoSCC for establishing the bibliometric maps (23), capable of assisting in comprehensively viewing the bibliometric maps in detail considering the collaborative data. CiteSpace focuses on analyzing the underlying knowledge in the scientific literature and visualizing the collected data (26).

The number of retrieved publications was 4,522, including 3,772 original research articles and 750 reviews, with total NC of 114,589, a mean NC/article ratio of 27.91, and a mean H-index of 130.

Figure 2A displays the annual NP associated with autophagy in lung diseases. The annual paper number indicates a fast increase from 119 in 2012 to 785 in 2021, and the NP reached the peak in 2021. From 2012, annual NP did not change significantly in the United States (US) and Korea but indicated a fast elevation in China, and there has been a surge since 2015. The annual NP exhibited an obvious relation to the publication year, with the correlation coefficient R² of 0.9872 (Figure 2A). The rapidly increasing NP demonstrates the increasing number of scholars who placed emphasis on this field. There were more than 4,500 publications in 2021, demonstrating the increasing number of scholars who conducted relevant research and the flourish of lung disease-related autophagy research theories (Figure 2B).

The 10 high-output countries/regions with the highest output were ranked according to NP of all authors. China published the most papers (2,564, 56.7%), followed by the US (1,048, 23.18%) and South Korea (234, 5.17%) (Figures 3A, B; Table 1). NC was 23,791 for China, accounting for 38.2% of the total number of citations, followed by the US (47,846) and South Korea (5,449). However, the US enjoyed the highest H-index (181) and G-index (264), about three times that of South Korea (58 of 81).

Figures 3C, D display the publication distribution in various countries and regions. Related cooperation was mainly concentrated upon China and the US, and cooperation with other countries was relatively weak. As shown in Figure 3C, cooperation between countries and regions significantly promoted the development of scientific research. The lines in the figure indicate cooperation: The wider the lines, the stronger the cooperation. Nevertheless, lines cannot be found in most countries, meaning that these countries had no stable communication and cooperation.

According to the author visualization chart by VOSviewer, 78 authors performed autophagy-related studies. In terms of the NP, Choi Augustine M. K. (28 publications) ranked first, followed by Ryter Stefan W. (28 publications), Wang Jing (25articles), and Wang Wei (24 articles) (Figure 4A; Table 2). Obviously, there was a relatively low author centrality (≤0.04), demonstrating the necessity to enhance the autophagy impact in lung disease. Co-cited authors refer to those (at least two) who are cited at the same time (Figure 4B). Only six had a frequency of citation of over 400 times amid the 103,587 co-cited authors. The lines between nodes stand for authors’ collaboration, and the collaboration graph means that authors’ collaboration indeed existed in the field.

Figure 5 is an institutional co-occurrence map built with CiteSpace, listing the top 10 institutions considering the output and centrality in autophagy-related lung disease studies. Nodes stand for institutions: The larger the node, the more output the institution made. Links between nodes mean institutional collaboration, and the color and thickness denote the collaboration duration and intensity, respectively. China Medical University (n = 190) was in the leading position considering the output, followed by Shandong University (n =180) and Zhejiang University (n = 174). The top three institutions considering the centrality were New York University (0.37), Pittsburgh University (0.30), and Michigan University (0.26) (Table 3). As the data show, institutions were closely collaborated with each other.

Table 4 gives the top 10 journals considering the research number of lung disease–related autophagy and the impact indicators, namely, H-index, G-index, and IF eigenfactor score. These journals had a larger possibility in accepting articles regarding autophagy in lung disease, given their largest NP of relevant topics. Scholars in the autophagy area are suggested to place emphasis on these journals. ONCOTARGET (115 publications, IF = 4.147) had the highest NP, followed by INT J MOL SCI (98 publications, IF = 5.924), PLOS ONE (97 publications, IF = 3.248), CELL DEATH DIS (81 publications, IF = 8.468), and AUTOPHAGY (80 publications, IF = 16.016). Notably, the AUTOPHAGY had higher citations, H-index, and G-index, demonstrating that the quality of many studies was not high. In Figure 6A, the analysis of journals revealed that PLOS ONE (2,677 total citations) presented the highest citation frequency, followed by AUTOPHAGY (2,626 total citations) and J BIOL CHEM (2,440 citations). J BIOL CHEM (0.58) enjoyed the highest centrality, followed by PLOS ONE (0.40) and P NATL ACAD SCI USA (0.39) (Table 5), all of which had a high centrality, demonstrating their stronger impact in this field. In the domain dimension, superposition graph analysis was carried out for the fields involved in the literature (Figure 6B). Among them, the top fields of paper production were oncology, pharmacology and pharmacy, chemistry medicinal, multidisciplinary sciences, and chemistry multidisciplinary. Therefore, in the study of autophagy in lung diseases, oncology and pharmacochemistry are currently the focused areas.

CiteSpace assisted in constructing the subject distribution regarding academic journals by virtue of the biplot overlay function (Figure 7). According to the journal biplot overlay, there were mainly two citation paths. Journals publishing articles were basically in medical sciences covering the pharmaceutical, clinical, molecular, biological, and immunological fields, and most cited articles were published in journals in molecular, biological, immunologic, genetic, dermatological, health, nursing, rehabilitation, dental, surgical, sports, neurological, pharmaceutical, and ophthalmological fields.

A co-citation network refers to a reference network simultaneously co-cited by at least one paper. Conceptual clusters were created when a set of manuscripts was cited together repeatedly (27). As shown in Figures 8A, B, the most cited article was written by Basant A. Abdulrahman who pointed out that rapamycin, an autophagy stimulator, enhanced the clearance of C. coronans by inducing autophagy and significantly reduced the infection of C. coronans in vitro.

Rapamycin was found capable of lowering the bacterial burden in the mouse lung in vitro and markedly eliminating lung inflammation signs (28). Figure 8C demonstrates a network visualization map regarding cited references, with nodes denoting the cited reference, where the connecting lines stand for the co-citation relationship. The different colors varying from purple to yellow mean different years in the range of 2012 to 2021. Node size denotes the number of co-citation number. Figure 8D displays the 16 relevant clusters divided by CiteSpace: cancer, lung cancer, signaling pathway, anticancer immunity, ferroptosis, particulate matter (PM2.5), COPD, asthma, ER stress, idiopathic pulmonary, acute lung injury, pulmonary disease, COVID-19, oxidative stress, alveolar type II, respiratory immunity, and autophagy. Citation burst reference refers to the sudden citation increase of specific articles within a certain period, which serves for finding the latest high-profile research topics in relative fields (29). We obtained 674 references with the highest citation burst and selected the top 25 (Figure 8E). The blue line denotes the timeline, and the red sections denote the burst interval, respectively, showing the beginning and end of the year, and the burst duration.

Keywords demonstrate a paper’s principal ideas and theme concepts, briefly describing the certain research hot spots. VOSviewer and CiteSpace were used for drawing charts. We identified words that appeared over 13 times in the process of analysis as keywords and finally obtained 155 keywords (Figure 9A; Table 6). As shown in Figure 9C, the closer the keyword was closer to yellow, the more frequently it appeared in the year. According to Figure 9A, clusters 1 and 5 were mainly composed of autophagy, apoptosis, and death, which paid attention to relevant mechanisms in the programmed cell death. Cluster 2 stands for “oxidative stress”, “mitochondrial dysfunction”, and “lung disease”. Oxidative stress, the oxidation-antioxidant imbalance in the body, usually causes mitochondrial dysfunction and lung disease. Cluster 3 emphasizes the function of autophagy in lung diseases caused by COVID-19. Clusters 4 and 5 mainly reflect the cellular mechanism and molecular pathway of lung diseases caused by autophagy. In overlay visualization, the color of keywords depends on the average publication year (APY). As shown in Figure 9B, the VOSviewer colors all keywords on the basis of APY. The most recent keyword was “sars-cov-2” (cluster 3, APY: 2020.53), followed by “COVID-19” (cluster 3, APY: 2020.45), both closely related to COVID-19. The color division confirms that the most recent keywords in the research field were “pyroptosis“, “immunity”, “COVID-19”, and “lung cancer”, which are proven to be the recent research hot spots.

Following the COVID-19 epidemic outbreak in December 2019, research on lung disease increased markedly in 2020 (Figure 10A). Our study screened out and analyzed 76 papers on lung diseases and autophagy related to COVID-19 from the retrieved literature in an attempt to reveal the research focuses and trends in the association between severe lung diseases and autophagy resulting from COVID-19 (Figure 10B). The terms inflammation, immunology, cytokine storm, oxidative stress, and macrophages all had the highest frequency as revealed in the keyword visualization analysis, in addition to COVID-19, SARS-COV-2, and lung-related diseases, explaining how autophagy exerted its function in severe lung disease caused by the pathogenesis of COVID-19. In addition, the APY-based keyword color classification by VOSviewer confirmed “inflammation” and “oxidative stress” as the top two recent keywords, meeting previous findings on autophagy in lung diseases. In addition, Figure 10 shows the involvement of autophagy in the pathogenesis of COVID-19. We also found that the mechanism of anti-inflammatory treatment and immunomodulation therapy for COVID-19–related lung disease was through inhibiting autophagy activation (Figure 10).