State of open science in cancer research

The impact of cancer on individuals, society, and the economy is significant. By 2040, the worldwide burden of cancer will rise to 30 million cases, with the largest increases in low- and middle-income countries [1]. Globally, it is estimated that there is a prevalence of cancer at 5 years’ post-diagnosis of more than 44 million. In 2020, prostate and breast cancer ranked as the leading diagnoses in men and women, respectively, while lung cancer followed with 2.2 million cases [2]. The European Cancer Information System (ECIS) has reported a total of 2.74 million cancer patients in 2022, representing a 2.3% increase in new cancer cases compared to 2020 [3]. In Spain, over 270,000 people are diagnosed each year [4], concretely in 2023, an estimated 158,544 cases of cancer are expected to occur in men and 12,715 in women [5].

To address this situation, advances in oncology are ushering in a new era of personalized and precise medicine, transforming everyday cancer care. However, unleashing the full potential of these approaches requires sound policies to ensure their regular application in patient care, including changes and new practices in science researching directives. To contribute, the Open Science (OS) movement, whose impact on science has been remarkable in the last two decades, offers a series of useful and necessary practices to accelerate the publication and dissemination of scientific results. Starting with the Berlin Declaration on Open Access (OA) [6], today is commonly accepted by institutions, funders, and publishers that the open access contributions must necessarily include, along with the article published in OA, also the primary data and their metadata [7, 8]. Thus, after an initial phase in which the demand for publicly funded research focused on the OA publication of scientific papers, the practice of sharing the raw data is now widely recognized as a means of ensuring honesty and robustness through its role in accountability and its ability to replicate experiments, as well as being cost-effective through the re-use and improvement of existing data [9]. Particularly in cancer research, institutions such as the National Cancer Institute [10] explicitly state that “improved treatment options for cancer patients will result when researchers share their data widely with investigators in the research community”, along with Cancer Research UK [11], which directly advocates a culture of sharing research data for re-use across sectors. Similarly, the Lancet Oncology Commission “European Groundshot-addressing Europe’s cancer research challenges”, highlights the importance of “collaborative research through data sharing is essential to ensure rapid improvement in cancer care, from diagnosis to therapeutic application” [12].

Considering the significance and promptness of effective cancer research dissemination, as well as the value provided by Open Science resources such as data sharing and open access papers, this study aims to accomplish the following objectives: i) to analyse the evolution of oncology research with Spanish participation in terms of articles and research data production during the decade of 2011–2021 and ii) to assess the impact on the scientific literature of open published articles in terms of citations, journal impact factor and scientific collaborations; and the characteristics of the deposited research data through their repositories and thematic categories.

We collected scientific publications in the field of cancer from the Science Citation Index-Expanded (SCIE) database of the Web of Science (WoS) for the period 2011–2021, signed by at least one Spanish institution. Documents were retrieved using a search equation conformed by search terms representative of cancer along with the different typologies described by the NCI cancer types (for example sarcoma, glioblastoma or leukaemia) combined with those papers published by journals belonging to the “Oncology” category of the SCIE (see Supplementary file 1). Finally, a total of 50,822 scientific documents (42,326 articles and 8496 reviews) were selected. The following variables were extracted from each document in an Access database: title, journal, publisher, year, authors, international collaboration, number of citations and open access publication route, following the classified model from WoS description: Green Accepted, Green published, Free to read (Bronze), Gold hybrid, and Gold [13]. Green submitted filter was excluded since it refers to documents that have not undergone peer review.

To study the impact of the journals, the quartiles (Q) extracted based on the JIF of the WoS categories from Journal Citation Reports database (JCR) in the respective year of publication were used. To assess the impact of the papers in the scientific literature, the total number of citations received for the articles was calculated referring to the weighted citations: number of citations received/years since publication.

To describe the dissemination of datasets derived from the oncology studies developed by at least one Spanish institution, between 2011–2021, a bibliographic search was performed in the Data Citation Index (DCI) to obtain an overview in terms of thematic categories, cites and repositories used.

A descriptive analysis of the variables was performed to obtain the frequencies and percentages. All charts, tables and figures were created using Microsoft Excel. VOSviewer software was used to represent the authorship networks related to OA publications and datasets, after normalizing the signatures and including those authors who had at least 5 documents retrieved for the period 2011–2021 and were signed by a maximum of 1000 authors per article.

The aim of this study was to address the state of the art on cancer research with Spanish participation and its relationship with open science, both at the level of publications and research data dissemination. Numerous papers have been found in the literature based on the use of bibliometric techniques to investigate research on different types of cancer, however, this study relates for the first time the evaluation of oncological research production under the Open Science scenario, analysing its impact through received citations and international collaborations between researchers.

The 50,822 publications retrieved, and the 2693 datasets obtained showed, in addition to the increment of scientific production in cancer with Spanish participation, the relevance of detected open access articles and data sharing during 2011–2021.

It should be highlighted that the number of papers in 2021 has doubled compared to 2011, which could be related to the shocking data on the prevalence of cancer both in Spain and worldwide, and the effort being made by all institutions to promote basic and translational cancer research. Along with the significant rise in the number of publications, there has been an increase in the OA publications. This OA modality has steadily grown over the past decade and now comprises over half of the total publications in the most recent year studied. This means that cancer research with Spanish participation has not been alien to the commitment of the scientific community to Open Access, which crystallized two decades ago with the publication of the so-called three “B’s” (Budapest, Bethesda and Berlin Declaration), that have celebrated their 20th anniversary this year 2023 [14], and has been evidenced over time by the requirements of numerous institutions to Open Science [7, 15]. Observing the studied decade, it is in 2014 when the trend towards publishing in Open Access journals was produced, which is not surprising in the European context, since it was precisely that year when the H2020 projects (in force from 2014–2020) established the open publication as a precept of the works derived from funded projects, a trail followed by the Member States in their national policies.

The evolution observed in our results, is also consistent with the recent creation of the Coalition for Advancing Research Assessment (CoARA) in Europe, by which more than 400 institutions commit to supporting OS, including both data and publications and to recognize this commitment as a merit to the research staff [16].

Although it is true that the predominant access via is the Green route specially in the early years of the study (by which the author deposits the postprint of an article in a repository, once the embargo period has expired), a notable modification in the trend is detected. This change shows information that is key for understanding the relevance of OA in the current context in cancer research, which is represented by the large increase in publications in the so-called Golden way (public, immediate, permanent and free access to the final article for readers, usually after payment of APC—article processing charge—by the authors), and the decrease in the Bronze way (only access to reading the work, but without any type of open license). Due to its characteristics, the golden way is defended in the framework of open science, since it represents the essence of the OA philosophy in terms of the opportunity it offers for immediacy, access and reuse, elements that are even more important in an area such as cancer research [17]. Alongside its benefits, potential drawbacks or adverse consequences of the Gold model have also been recognized. In the studied period, there has been a tendency to disseminate cancer articles in journals of publishers that publish solely and exclusively in open access by paying APC. In particular, controversial issues relating to the APC model have been cited, as it has the potential to place authors at the forefront of journals’ commerce, resulting in a reduction in the standard of quality criteria to favour quantity and increase revenue [18].

However, this argument of declining quality is not supported by our study. In fact, our results have shown, both in terms of weighted citations received by the articles and JIF quartile distributions of the journals, that papers published through the OA modalities obtain better scores than those non-OA. This result agrees with other studies such as that of Gumpenberger et al. (2013) [19], that almost a decade ago, when there were still fewer OA journals, did already mention the positive relationship between Gold OA and positive impact, as well as the fact that open articles published were more cited comparing to other OA models. On this specific aspect, there are numerous papers that have investigated the positive impact in terms of citations received from OA [20]. It is interesting to highlight the study of Levin et al. (2023) [21] and AlRyalat et al. (2019) [22], since both examine the case of oncology research and OA publications and conclude that the number of citations received is higher when the publications are open, either in Gold OA journals or in Hybrid journals. Regarding collaborations, our study show that international collaborations are not only beneficial because they promote diversity of views among researchers and countries when it comes to research and therapeutic approaches [23]; but they also (or because of this) generate higher impact publications, information that is consistent with studies such as that of Kohus et al. (2022) [24]; moreover, within the publications with international collaboration, those that are in OA have an even greater presence in Q1 journals.

Regarding the practice of data sharing, it has historically followed a different trajectory than open publishing. While open publications belong to the very specific context of the OA movement of the late 1990s and early 2000s, the practice of sharing data among researchers is much older [25]. The way in which the practice of data sharing has been included in the orbit of open science has changed, making it more than just sharing limited to “face to face” researchers [26]. In the case of the research data in our study they are mostly genetics and molecular biology related data and were deposited in the Gene expression Omnibus repository, followed by far by the European Nucleotide Archive. The area of oncology has an important basic research component and, within this, genetics, which proved to be the type of research within the field of health sciences in which the greatest number of datasets are shared, and with a greater culture of data sharing [27, 28]. An example of the relevance is found in the study of Birney et al. (2017) [29], where they refer that, by 2030, 83 million genomes of rare diseases will be sequenced, as well as almost 250 million for cancer diagnosis. Another interesting work on the relationship between genetic data, data sharing practice and cancer research is that of Knoppers & Joly (2018) [30], that after reviewing two initiatives related to shared data in health, emphasizes that if there are no common policies and genomic data are not linked to daily clinical practice through coordinated and interoperable systems, it is difficult to improve clinical decisions.

The works retrieved through WoS SCIE and DCI represent the total of existing publications and datasets derived from journals or repositories, respectively, that have been indexed in these databases.

Our study shows the growth in the number of cancer publications in the last decade, developed by at least one researcher from a Spanish institution, accompanied by an increase in the habit of sharing their studies, which marks a change in the attitude of cancer researchers towards a more open stance and closer to open science. In addition, the approach to more accessible OA models provides the opportunity to generate higher impact publications, just as it does when conducted between international research teams. Regarding research data, it has been observed that genetics area has the greatest sharing, which is a positive aspect that allows further progress in this line. However, oncology research is much more than genetics; there is a whole field of clinical and even social research that does not seem to be sharing research data at the same rate, therefore it would be very interesting to find out more about why this is happening. On the other hand, our study has allowed us to know what is shared in terms of repositories and the thematic categories where belongs, but it would be very interesting to delve more deeply into the content and structure of these datasets and their metadata in a more qualitative way, as well as to inquire about their quality in relation to compliance with the FAIR principles.