Research Interest and Public Interest in Melanoma: A Bibliometric and Google Trends Analysis

Introduction Melanoma is a severe skin cancer that metastasizes quickly. Bibliometric analysis can quantify hotspots of research interest. Google Trends can provide information to address public concerns. Methods The top 15 most frequently cited articles on melanoma each year from 2015 to 2019, according to annual citations, were retrieved from the Web of Science database. Original articles, reviews, and research letters were included in this research. For the Google Trends analysis, the topic “Melanoma” was selected as the keyword. Online search data from 2004 to 2019 were collected. Four countries (New Zealand, Australia, the United States and the United Kingdom) were selected for seasonal analysis. Annual trends in relative search volume and seasonal variation were analyzed, and the top related topics and rising related topics were also selected and analyzed. Results The top 15 most frequently cited articles each year were all original articles that focused on immunotherapy (n=8), omics (n=5), and the microbiome (n=2). The average relative search volume remained relatively stable across the years. The seasonal variation analysis revealed that the peak appeared in summer, and the valley appeared in winter. The diseases associated with or manifestations of melanoma, treatment options, risk factors, diagnostic tools, and prognosis were the topics in which the public was most interested. Most of the topics revealed by bibliometric and Google Trends analyses were consistent, with the exception of issues related to the molecular biology of melanoma. Conclusion This study revealed the trends in research interest and public interest in melanoma, which may pave the way for further research.


INTRODUCTION
Melanoma is a severe skin cancer that metastasizes quickly. Cutaneous melanoma causes 55,000 deaths each year, and once the disease spreads, it rapidly becomes life-threatening (1). Cases of cutaneous melanoma account for approximately 1.7% of all newly diagnosed cases of primary malignant cancers (1). The incidence and mortality rate of melanoma vary around the world. Fair-skinned populations are particularly prone to melanoma, and the incidence of melanoma is the highest in New Zealand and Australia (2). Exposure to ultraviolet radiation, number of atypical moles, and genetic background are common risk factors for melanoma (3).
Bibliometric analysis is a method used to quantify hot topics and research interest in the research community (4)(5)(6). Bibliometric analysis can provide physicians and investigators with crucial messages in a specific field. A thorough bibliometric analysis of the most frequently cited articles may facilitate an understanding of disciplinary development and future directions of a research field (7,8). Google Trends is a commonly used tool for addressing online health issues. Infodemiological methods using Google Trends can estimate the epidemiological characteristics, explore the public interest, and monitor the dynamic variations in infectious diseases (9). Previously, some studies demonstrated positive correlations between the online search frequency of "melanoma" and that of its risk factors (10)(11)(12). However, McDonald and Bloom reported negative results on the association between the search index and the incidence of melanoma (13,14).
Compared to bibliometric analysis, which provides information on research interest, Google Trends analysis provides information on public interest. Physicians and investigators should know not only the hotspots of scientific research on melanoma but also the issues of interest for the general public. This study aimed to update the topics of research interest and public interest in melanoma using bibliometric and Google Trends analyses and compare the similarities and differences, which may pave the way for further research.

Bibliometric Analysis
We analyzed the top 15 most frequently cited articles on melanoma each year from 2015 to 2019 according to the bibliometric analysis method. These publications were retrieved from the Web of Science in descending order according to their numbers of annual citations. Two researchers (H. Zhang and Y. Wang) independently screened the abstracts and reached a consensus on the qualifying papers. Articles focusing on multiple diseases, conference articles, patents, comments, or case reports were all excluded. Original articles, reviews, and research letters were all included in this research.

Search Tool and Keyword Selection
Online search data were collected from Google Trends. Google Trends provided an index, namely, the relative search volume (RSV), to facilitate comparisons between terms, times, and locations. The RSV was restricted to a range from 0 to 100. An RSV of 100 represented the highest search count in a given period (weeks, months, or years), and the search counts were proportionally assigned lower numbers in other periods. For example, an RSV of 50 indicates that half as many searches were performed in the selected period compared to the searches indicated by an RSV of 100 (15). An RSV of 0 did not necessarily indicate 0 searches but may represent an extremely low search count compared to other periods (16). Google Trends also automatically adjusted the RSV based on population sizes to allow a comparison between populated areas and underpopulated areas (17).
The keywords were selected under the instruction of a previous guideline (18). Words or short phrases that were specific and not prone to be confused with other words or short phrases were preferable. Google Trends provided two types of query modes. One mode was the "Terms," which could be combined for exhaustive search, but the results would only be shown in the given language. The other type was "Topics," which were defined as groups of terms that shared the same concept in any language. This mode also included related searches in non-English speaking countries and might contain the most associated information (16). The mesh words of PubMed only provided "melanomas" for possible synonyms or homonyms of "melanoma" and allowed us to compare the two types of query modes by inputting different patterns of keywords, including "melanoma" alone as a term or topic, "melanomas" alone as a term, and "melanoma + melanomas" as a combination of terms in Google Trends. Both tests yielded similar fluctuations and patterns, but the topic "melanoma" produced the highest RSV. Therefore, the topic "Melanoma" was selected as the keyword in this study.

Data Query
The "Health" category was chosen to exclude unrelated information. The time range was set from January 2004 to December 2019. On 1 September 2020, the RSV data were exported to Microsoft Excel 2019. Four English-speaking countries with high RSVs were selected for seasonal variation analysis. Two countries (the United Kingdom and the United States) were located in the Northern Hemisphere, and the other two countries (Australia and New Zealand) were located in the Southern Hemisphere.

Google Trends Analysis
Topics related to the search term were also extracted from Google Trends to analyze the public interest. Google Trends provided two types of related topics: "Top related topics" and "Rising related topics." "Top related topics" are defined as the most frequently searched topics within the chosen category, time, or country. "Rising related topics" are topics with high RSV growth and are presented as a percentage of fold changes. We queried the "Top related topics" and "Rising related topics" each year from 2014 to 2019 globally to analyze the variation in the public interest over time. The results were manually examined by two searchers (H. Zhang and Y. Wang) to exclude irrelevant information.

Statistical Analysis
R software (v 3.6.2) was used for statistical analysis and plotting graphs. A diagram was plotted using the "plot" function in R to observe the trend in the annual average RSV. A cosinor model was applied for seasonal analysis according to Barnett (19). Boxplots of the seasonal variation for different countries were plotted by the "season" package in R. A p-value < 0.05 was considered statistically significant.

Ethical Requirements
This study did not involve animal experiments or clinical trials. Thus, permission from the ethical committee was not needed.

Annual trends and seasonal variation in Google Trends
The annual trends for the RSV of melanoma in Google Trends are shown in Figure 1A. The maximum value appeared in June 2005, and the minimum value was observed in December 2012. The average RSV remained relatively stable across the years. The seasonal variation curve fit with the "cosinor" model for the RSV is shown in Figure 1B (p-value < 0.05). The analysis revealed that the peak RSV of melanoma occurred in summer (January for Australia and New Zealand and June for the United States and the United Kingdom) and the valley occurred in winter (July for Australia and New Zealand and December for the United States and the United Kingdom).

DISCUSSION
This study updated the topics of research interest and public interest related to melanoma and provided physicians and investigators with a detailed description of the hot issues in which scientists and the public are interested. Google Trends data are a powerful tool to monitor and evaluate public interest in melanoma. The combination of Google Trends and bibliometric analysis may allow researchers to better anticipate research interests to serve melanoma patients.
Using bibliometric analysis, we determined the 15 most frequently cited articles on melanoma with the high numbers of annual citations published from 2015 to 2019. Using annual citations instead of the total citations as bibliometric parameters for ranking yielded benefits because this ranking included newly published articles that can provide emerging insights in the analysis (35). Our analysis indicated that the majority of these articles were published in the New England Journal of Medicine, followed by Science, Cell, Nature, and Lancet Oncology, which could be attributed to the high quality of these journals or the inherent bias with which researchers tend to select high impact factor journals for citations (36,37). All the publications were original articles, reflecting the substantial demand of the community for revolutionary innovation and discoveries related to melanoma. The average numbers of citations of these most frequently cited articles were dramatically higher than those of other bibliometric analysis studies, such as those on rosacea (8), oral lichen planus (38), or psoriatic arthritis (38). This phenomenon reflects a high degree of research interest regarding melanoma. In addition, the articles were all classic with more than 400 citations, even for the articles published in 2018, showing the impact of the literature (8,39).
Furthermore, researchers might focus on other topics to provide new insights into melanoma that the public might not know. Examples include omics analysis and microbiome analysis. Genomic studies have identified activating driver mutations that stimulate the development of targeted therapies for patients (48). The overall mutational load, neoantigen load, and expression of cytolytic markers in the immune microenvironment were significantly associated with clinical benefits (29). In addition, the commensal microbiome might have a mechanistic impact on antitumor immunity in melanoma patients (23). The results suggested that patients with a favorable gut microbiome might express enhanced systemic and antitumor immunity (21).
Google Trends was particularly helpful in monitoring health information-seeking behavior and analyzing public interest. The results showed that the global average RSV for melanoma was relatively stable across the years, illustrating the continued attention given by the public to melanoma (49). Regarding seasonal analysis, in Australia and New Zealand, the peak RSV appeared in January (summer). During that time, the incidence of melanoma is predominantly high in those countries (50), and previous research has demonstrated the correlation between the RSV of sun tanning and melanoma (51). Risk factors for melanoma, including exposure to sunshine, lighter clothing, and even sun tanning, might be responsible for this result (52,53). The health prevention campaign in Australia also promisingly reduced the rates of indoor tanning among young adults and thus helped to decrease the incidence (54). For countries in the Northern Hemisphere, such as the United States and the United Kingdom, the peak RSV appeared in June (summer), and the educational campaign of public awareness month for skin cancers in May might be responsible for increasing the RSV (55).
The related topics illustrated the most concerning themes for the public. The top related topics were defined as the most frequently searched topics within the chosen category, time, or country. As a type of cancer, melanoma mostly originates from the skin; the terms "Skin," "Skin cancer," and "Metastasis" were reasonably ranked in the top 3 related topics. The differential diagnosis of melanoma from other diseases such as "Melanocytic nevus" and "nevus" also attracted attention. Even senior dermatologists had some difficulties in recognizing malignant features to distinguish melanoma from nevus in dermoscopic images (56), and the involvement of artificial intelligence in dermatology liberated dermatologists and made some contributions to solving the problem (57). The terms "Basal cell carcinoma" and "Squamous cell carcinoma" refer to common malignant tumors in the United States and hence have become hot topics (58). "Malignancy," "Prognosis," "Relapse," and "Survival rate" might be the most concerning topics for the patients and appeared in the list.
The rising related topics are of newly emerged public interest. The results marked "Breakout" represent tremendous increases of over 5,000% compared with the previous search, probably representing the rapid development of these topics. Immunotherapies are in the spotlight in this era. The systemic treatment of melanoma has completely changed since the first introduction of ipilimumab in 2011 (59). In less than 10 years, over 10 drugs have been proven or are being proven effective for treating unresectable melanoma and dramatically increase the predicted survival time of patients (60). A review recently summarized the historically published articles and guided clinicians regarding the use of systemic therapy for melanoma (40). The overall success explained the emergence of the public interest in immunotherapies in recent years. "Cancer staging," "Metastasis," "Malignancy," and "Survival rate" also attracted attention. The complete revolution of melanoma management has invigorated the public interest in the prognoses of patients. The popularization of the concept of personalized medicine caused the public to become more concerned with the outcomes of patients instead of short-term effects. Hence, it was necessary to formulate an individualized systemic medication plan according to the cancer stage and metastasis of the patients to achieve the maximum survival rate.
Treatment methods ranked second among the results. Terms associated with surgical methods including "Exeresis" and "Mohs surgery" refer to effective treatment modalities for early-stage noninvasive melanoma and therefore attract public interest (65,66). Consistent with the bibliometric analysis, immunotherapies and risk genes attracted attention. In addition to those we discussed above, CTLA-4 was recently the focus of the public and appeared on the list. CTLA-4 is an immune checkpoint molecule that downregulates pathways of T cell activation (67), and "Ipilimumab" can inhibit CTLA-4 to improve survival in patients with metastatic melanoma (68).
Risk factors that had been discussed above, including sun tanning and melanin, illustrated the importance of public educational campaigns (69,70). The evolution and broad adaption of dermatoscopy in clinical examinations also improved the diagnosis of benign and malignant cutaneous neoplasms compared with diagnosis with unaided eyes. Dermatoscopy also improved the ability of expert readers to make appropriate management decisions (71). Cell cultures can contribute to the diagnosis and development of melanoma management plans and function as an experimental tool to facilitate the development of new drugs (72). Interestingly, American Joint Committee on Cancer and a celebrity, Bob Marley, who died of the disease, also appeared on the list. The former association formulates the guidelines for the cancer staging of melanoma, and the latter reflects the celebrity effect, which can stimulate the recognition of the disease among the public.
Our study revealed the consistency between the research interest and the public interest. Both interests focused on the risk genes of melanoma and their inhibitors or blockers. These included PD-1, BRAF, CTLA-4, ipilimumab, nivolumab, dabrafenib, and trametinib. The use of social media has substantially increased among researchers and the public and could explain this corresponding relationship (73). In Australia, the SunSmart skin cancer prevention program has been demonstrated to contribute to the reduction of melanoma among younger cohorts (74). In addition to Australia, the Euromelanoma campaign also organized a yearly media campaign, which targets the public and focuses on different aspects of melanoma prevention. Euromelanoma Day has been held each year in May, both in university-based and hospital-based outpatient clinics and private dermatology surgeries (75). Patients and even the normal population can enhance their knowledge through these campaigns and become familiar with the latest research interest (76). In addition, the research interest might be influenced by social media, as reported by Pemmaraju (74), and the types of tweets about skin cancer have changed rapidly over time. The number of pharmaceutical companies that is discussed has been increasing, and the topic tags transitioned from "melanoma" to "immunotherapies" from 2011 to 2016 (74).
However, some differences still exist. The public did not show interest in the omics and microbiomes of melanoma that the research community studied. This was comprehensive because the public might not be familiar with these academic terms. More importantly, patients were mostly concerned with the symptoms, differential diagnosis, metastasis, and treatment of melanoma, especially newly emerged targeted drugs, which might improve prognosis and predict survival time. These aspects might become future directions for research and the popularization of science. Mechanisms, pathogenesis, pathophysiology, and epidemiological features were probably less important for patients because the complete elucidation of such factors could not alleviate symptoms, cure the disease, and decrease the high treatment expenses. Although these research fields might not provide patients and their families with hope in this era, they remain valuable for researchers. The development of new techniques and the discovery of key molecules in melanoma are crucial to guide future management. The prognosis of melanoma patients with regional metastases is influenced by the genomic classification, offering insights to further personalize therapeutic decision making (20). In addition, the commensal microbiome might have a mechanistic impact on antitumor immunity in melanoma patients (23). Such research findings might be included in educational campaigns in the future.
There are several limitations to the study. First, the public interest is restricted to Internet users who are conducting Google searches in English. There may be selection bias because the disease might not attract enough attention in underdeveloped areas. Although English remains the most popular official language worldwide, different languages and cultures could have different interests. In addition, other search engines could also be more popular than Google Trends in certain countries. For example, the Baidu engine is the main search engine in China. To compensate for the loss of data, we tried to use "topics" instead of "terms" as keywords, which may include some synonyms of melanoma in other languages. Second, only the Web of Science database was used to search for eligible articles, and some articles may be missed. Notably, fewer citations do not mean that an article is unimportant because it may lack the ability to be accessed by scholars.

CONCLUSION
This study used bibliometric and Google Trends analyses to update the topics and to compare the differences and similarities of research interest and public interest in melanoma. Regarding research interest, the top 15 most frequently cited articles each year focused on immunotherapy (n=8), omics (n=5), and the microbiome (n=2). Regarding public interest, diseases associated with or manifestations of melanoma, treatment options, risk factors, diagnostic tools, and prognosis were of the greatest interest to the public. The results revealed the trends in research interest and public interest in melanoma, which may pave the way for further research.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.

AUTHOR CONTRIBUTIONS
HZ and QS conceived and designed the study. HZ and YZW prepared the manuscript, and had equal contribution to the study. QZ and KT prepared the tables and figures. QZ, KT, RF, YCW, and QS reviewed and revised the manuscript. All authors contributed to the article and approved the submitted version.

FUNDING
The publishing fee was funded by Peking Union Medical College Hospital.