FormalPara Key Summary Points

Safety observations made during the clinical phases make it possible to anticipate and minimize the risks to which patients are exposed in the post-marketing era.

The need for early CT safety surveillance is globally recognized, and therefore many countries have implemented the fundamental methods proposed by the ICH.

Diversity in CT safety surveillance methods among countries mainly concerns the expedited reporting requirements and the lack of defined strategies for aggregate data analysis, leading to inconsistencies in the reported data, thus rendering their comparability low and increasing the time and cost for the development and approval of safe and efficient drug therapies.

Harmonization of the safety surveillance methodology at global level can be achieved via the development of safety surveillance plans (as directed by the CIOMS VI report) and the appointment of qualified teams to review safety data in aggregates.

A more active involvement of the regulatory authorities in CT safety surveillance is recommended, as they have access to data from different sources, including different pharmaceutical companies.

Introduction and Research Objectives

Drug development is a long process that starts with identifying a chemical substance with potential therapeutic activity and leads to approval and marketing of the drug. The assessment of the safety of the identified therapeutic candidate is an even longer process, starting very early at preclinical stages via in vitro and in vivo toxicology studies and continuing throughout the life span of the product, with the detection and analysis of adverse events (AEs) occurring during use in humans.

This review aims to investigate the existing regulations guiding the ongoing safety surveillance of medicinal products during their development. Although  the term “medicinal products” is broad and may include a wide range of therapeutic or diagnostic entities, the current review focuses only on chemical compounds. The regulations related to generic drugs, biotechnology-derived products, vaccines, and medical devices are out of scope. This review summarizes the regulations that focus on the safety of human subjects during the pre-approval phases (e.g., phase I, II, and III of clinical trials (CTs)) and before marketing authorization is granted. Post-marketing (PM) regulations are considered only for reasons of comparison or for better understanding, but they are not analyzed in detail. Preclinical toxicology studies are also excluded. Regulatory specificities concerning some types of studies (e.g., bioavailability or bioequivalence studies, trials in the geriatric or pediatric population) are not discussed (and no distinction is made between sponsor and sponsor–investigator trials, as the latter initiates and conducts the study, undertaking the responsibilities of both the sponsor and the investigator). The present review compares the differences between different countries or territories worldwide to identify potential gaps or inconsistencies, as well as opportunities for improvement and harmonization. Refinement and harmonization of surveillance methodologies will allow for more efficient use of the cumulative data (clinical and non-clinical), reinforcing the protection of human subjects during clinical investigations.

This review is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

Importance of Safety Surveillance in CTs

Ethical Codes: A Historical Perspective

Circa 400 B.C., the Greek physician Hippocrates included in his famous oath—intended for newly trained physicians—the requirement to protect patients from any deleterious, “mischievous,” or lethal treatment, according to their ability and judgment [1, 2]. This oath was one of the first policies adopted by the World Medical Association (WMA) in 1947 and is still valid nowadays in a revised version known as the modern Hippocratic oath or Declaration of Geneva, stating “The health of my patient will be my first consideration” and “I will maintain the utmost respect for human life” [3].

A series of ethical codes were developed during the twentieth century, emphasizing the importance of continuous assessment of the risk–benefit balance when conducting human studies (Fig. 1). Such codes emanated after scandals or incidents of unethical behavior by physicians who exercised their experimental approaches (often in vulnerable populations) without considering the potential harms or respectfully informing the participants [1]. The Berlin Code of 1900, enacted in the Prussian Empire (also known as the Prussian Standards), stressed the need for a proper explanation of possible adverse consequences of medical interventions to obtain informed consent from the participant [1, 4]. In the same year, Major Walter Reed, a physician of the United States (US) army performing human research on the transmission of yellow fever in Cuba, developed the first written “informed consent,” describing the risks and benefits of participating in the study [5, 6]. In 1931, the Weimar Government of the German Republic issued the Reichsgesundheitsrat Circular 1931, which remained in statute books until 1948. The “1931 Guidelines” provided guidance on the use of innovative therapies in humans and on experimentation involving human subjects, referring to the requirement for assessment of the risk (potential AEs) versus the anticipated benefits of the new therapy (risk–benefit balance) [1, 7, 8]. However, in the same year, Hitler signed a memo to exclude the Jews, the mentally disabled, and other groups of people from the Berlin Code of 1900 [9]. The medical experimentations performed by the Nazis on concentration camp captives during World War II led to the drafting of the Nuremberg Code in 1947 [8] to set the legal standards at the Doctors’ Trial in Nuremberg (December 1946 to August 1947) [10]. This Code is still considered a landmark document on the ethics of medical research on humans.

Five of its clauses refer directly to the need for the protection of participants from potential injury and unnecessary risks that are not justified by the benefits. Nevertheless, when the Nuremberg Code was written, USA researchers of the Tuskegee Study deprived patients with syphilis of treatment by studying the course of the disease in African American men, thus resulting in the death of 128 participants [11]. The revelation of the scandal in 1972 led to the establishment of the USA Office for Human Research Protections (OHRP) and institutional review boards (IRBs) for the protection of CT subjects, the compilation of the Belmont Report in 1979, and USA federal laws. The three fundamental ethical principles of the Belmont Report guidelines are respect for persons, justice, and beneficence, based on a systematic assessment of risk and benefits since “avoiding harm requires learning what is harmful” [10].

Fig. 1
figure 1

Evolution of the PV regulatory landscape throughout modern history. AE, adverse event; CIOMS, Council for International Organizations of Medical Sciences; CT, clinical trial; DSUR, Development Safety Update Report; ICH, International Council of Harmonization; GCP, Good Clinical Practice; PM, post-marketing; PV, pharmacovigilance; WG, working group; WMA, World Medical Association

Full size image

The currently accepted ethical principles that should guide human clinical research are summarized in the Declaration of Helsinki [12], firstly adopted by the WMA in 1964 and last amended in 2013. Several of its 37 articles state the importance of safeguarding and protecting human life and health during medical research (art. 4, 7, 9, 14, 33) and of continuously monitoring and assessing the risks (art. 17, 23), which should not outweigh the importance of the research objective (art. 16) [12].

Compliance to international ethical and regulatory research standards has been questioned leading to the evolution these standards. In this context, Pfizer failed to adequately protect children with meningitis in its Nigeria study in 1996 [13, 14]. In addition, TeGenero went bankrupt after their first-in-human study of the monoclonal antibody TGN1412, which caused life-threatening multiorgan failure to all six volunteers exposed in 2006 [15, 16]. In another first-in-human study sponsored by Bial-Portela & Ca and Biotrial, after repeated administration of BIA10-2474, an experimental fatty acid amide hydrolase inhibitor, one patient died, and at least two patients suffered long-term neurological damage [15, 17, 18]. Such incidents, as well as cases of drug withdrawals early after marketing approval due to serious AEs (SAEs) (e.g., selective COX-2 inhibitors rofecoxib (Merck) and valdecoxib (Pfizer) because of cardiovascular risks and potentially fatal skin reactions respectively [19], or the anti-obesity cannabinoid receptor antagonist rimonabant (Sanofi-Aventis) because of the risk of depression and suicide [20, 21]), have emphasized the critical need to initiate rigorous monitoring of drug safety during development and before marketing approval.

Current International Research Standards: CIOMS and ICH

In 1949, the Council for International Organizations of Medical Sciences (CIOMS) was established as a result of an agreement between the United Nations Educational, Scientific and Cultural Organization (UNESCO) and the World Health Organization (WHO) [22]. The CIOMS, in collaboration with other international organizations, such as UNESCO, WHO, the Council on Health Research for Development (COHRED), and WMA, have published several versions of Ethical Guidelines related to biomedical research in humans and human health data (1982, 1993, 2002, 2009, and 2016). All those Guidelines are based on the principles of the Nuremberg Code, the Declaration of Helsinki, and the Belmont Report. The CIOMS recommendations are not binding, and their incorporation into a legislative framework depends on other bodies, such as the International Council for Harmonization (ICH). The latter was founded in 1990 by regulatory agencies and industry associations from Europe, Japan, and the USA, to harmonize the regulations governing the development and marketing of medicines [23]. Harmonization was indeed crucial in an era when the pharmaceutical industry was growing and expanding globally.

Moreover, the need to satisfy different local regulations impacted the cost of research and development and the time to bring new medicinal products to the market. In 2015, the ICH became a non-profit legal entity under Swiss law to extend the harmonization of guidelines to more world regions. Currently it has 17 members, including regulatory authorities (RAs) and industries. All regulatory ICH members are requested to implement the ICH Guidelines in their national or regional laws [24, 25].

Developing guidelines on the safety, quality, and efficacy of medicines has been the priority of the ICH. Most of those guidelines have been based on existing CIOMS reports. Since 1986, CIOMS has set up different working groups (WGs) to focus on pharmacovigilance (PV), i.e., “the science and activities relating to the detection, assessment, understanding, and prevention of adverse effects or any other drug-related problem” (WHO’s definition) [26]. The first CIOMS WG set the standards for reporting pharmaceutical manufacturers’ adverse drug reactions (ADRs) to RAs by introducing the CIOMS I form [27]. This form has been the model for the ICH Efficacy Guidelines 2A (E2A) and 2B (E2B) on the clinical safety data management and expedited reporting of ADRs [28]. The reports of CIOMS WGs II, III, and V have contributed to the development of the ICH Guideline E2C [29] on periodic safety update reports (PSURs) and periodic benefit–risk evaluation reports (PBRERs) for marketed drugs and of the ICH Guideline E2D [30], which further elaborates on post-approval expedited reporting standards. In addition, the WG II introduced the concept of developing a document containing the minimum essential and up-to-date safety information for a marketed drug, namely the “Company Core Safety Information” (CCSI), which was further analyzed in the report of the WG III and extended to the pre-approval phase as development core safety information (DCSI) by the WG V [31]. The ICH Guideline E2E on PV planning (2004) suggested initiating product safety specifications from the pre-marketing phase [32, 33]. The report of the WG VI (2005) shifted the CIOMS focus from PM PV activities to the management of CT safety information and proposed an adaptation of post-authorization PV methods and tools to the CT setting [34]. The CIOMS VI report stressed the paramount importance of early detection of safety issues during drug development through a well-established systematic process for the detection and evaluation of safety information (including clinical and non-clinical data) on an ongoing basis by the sponsors. That should enable the timely establishment of risk minimization procedures to protect trial participants from exposure to undue risk. The periodic analysis of the risk during clinical development was further elaborated in the report of the CIOMS WG VII [35], which proposed a harmonized format and content for development safety update reports (DSURs), forming the basis for the generation of the respective ICH Guideline E2F on DSURs [36]. Interestingly, additional tools for safety risk monitoring were proposed by the CIOMS WG VIII report (2010), which presented strategies for safety signal detection, analysis, and management, concerning only approved products [37].

The international ethical and scientific standards for CT management are described in the ICH E6 Good Clinical Practice (GCP) Guideline [38], which was first published by the ICH in 1996 and was revised to its current final form in 2016. The ICH GCP Guideline has been adopted by many countries worldwide and describes the quality standards for designing, conducting, recording, and reporting CTs and the primary responsibilities of all involved entities (investigators, sponsors, and ethics committees (ECs) or IRBs), so that the rights, safety, and well-being of trial subjects are protected, and that the CT data are credible. Similarly to the CIOMS WG VI report, the ICH GCP emphasizes the sponsor’s obligation to continuously assess the safety of the investigational product (ICH GCP § 5.16.1) [38].

Safety Surveillance During Drug Development: Responsible Entities and Methods as per International Recommendations and Guidelines

Protecting the safety, well-being, and rights of CT participants over the interests of science and society is one of the fundamental principles of the ICH GCP (§ 2.3). It is a shared responsibility among all stakeholders involved in the conduct of a CT [38]. The investigators interacting directly with the participants (ICH GCP § 4.3) and the sponsors via the ongoing safety evaluation of the investigational product (ICH GCP § 5.16) are the primary entities responsible for the trial participants’ safety [34, 38, 39]. Further duties lie with third parties contracted by the sponsor, regulatory bodies (RAs and ECs/IRBs) overseeing the trial as per local laws, and even with trial participants, who are requested to provide information about their health status from the moment they consent to accept all the risks described in the informed consent form (ICF) [34, 38].

The following sections present the entities responsible for the safety surveillance of drugs under development and respective methodology, as directed by the relevant ICH Guidelines and recommended in the CIOMS reports. As defined by the ICH, those responsibilities are also summarized in Table 1.

Table 1 Safety surveillance responsibilities as per ICH
Full size table

Investigators

The investigator (including the institution in which they exert trial-related activities or their designee) is responsible for conducting the CT and, thus, is the principal entity directly following the safety of the trial subjects. Essential prerequisites to fulfil this obligation are appropriate medical experience, adequate knowledge and comprehension of the GCP and respective regulatory requirements, and sufficient training on the trial documents (especially the protocol and the investigator’s brochure, IB) (ICH GCP § 4.1) [38].

The investigator manages AEs occurring to the trial subjects and records AEs along with information that may assist in their further analysis, such as relevant medical history and concomitant medications (ICH GCP § 4.3.2) [34, 38]. As per the CIOMS WG VI, all serious (i.e., fatal, life-threatening, requiring inpatient hospitalization or prolongation of existing hospitalization, leading to persistent or significant disability/incapacity or a congenital anomaly or congenital disability, or deemed medically important) and non-serious AEs, as well as specific laboratory abnormalities, should be collected, followed up, and reported to the sponsor as per protocol (ICH GCP § 4.11.2 and 4.11.3) [38]. This allows the sponsor to conduct further analyses of AEs in aggregates and consequently draw safety conclusions that may potentially impact the trial [34]. The CIOMS WG VI suggests that AE collection continues for at least an additional five half-lives of the investigational product, following the administration of the last dose to the subject and at any time the investigator becomes aware of potential latent safety incidents [34]. To facilitate subsequent safety assessments, the CIOMS WG VI strongly recommends that the investigator also provides an opinion regarding the possible association of any AE with the investigational product. This is of particular significance for evaluating unusual or rare AEs, for which aggregate analysis is not feasible because of their low incidence of occurrence during the developmental phase [34]. Depending on the applicable regulatory requirements, the investigator may also need to submit to ECs/IRBs and RAs any SAEs assessed as possibly related to the investigational product (e.g., ADRs). That includes any SAEs that are unexpected in nature or severity based on the applicable reference safety information for the medication (e.g., IB for an unapproved investigational product or package insert/summary of product characteristics for an approved product) (ICH GCP § 4.11.1) [38].

Apart from the usual AE management, additional safety surveillance processes involving the investigator include the provision of regular (usually annual) progress reports and a final report with a summary of the trial results to ECs/IRBs and any required reports to the RAs (ICH GCP § 4.13) [38]. Written ad hoc notifications must be provided to the sponsor and the ECs/IRBs, in case of observations indicating an increase in the overall risk to the trial subjects and of urgent measures taken to eliminate potential hazards (ICH GCP § 3.3.8 and 4.5). Such actions may even include suspension or termination of the trial without prior agreement by the sponsor or ECs/IRBs (ICH GCP § 4.12) [38]. Nevertheless, a relevant notification should be sent with a detailed explanation of the individual actions.

Sponsors

The responsibility for ongoing safety surveillance of the investigational product by the sponsor is stated in the ICH GCP (§ 5.16.1) [38]. During the trial design stage, the protocol and the IB generated by the sponsor should define stopping rules or discontinuation criteria for the trial subjects and specify safety parameters that the investigator should monitor during the trial. Drafting these documents should be based on the up-to-date knowledge of the investigational product from non-clinical pharmacology, toxicology, and metabolic studies and on available data from clinical and marketing experience (if the product is already approved).

The SAEs or non-serious AEs of special interest (as defined in the protocol and/or IB) should be reported to the sponsor for further assessment upon investigator’s awareness (ICH GCP § 4.11). Subsequent expedition by the sponsor to investigators, ECs/IRBs, and RAs is required if the SAE is unexpected and is considered related to the investigational treatment, in compliance with the ICH E2A Guideline and applicable regulations (ICH GCP § 5.17). In addition, the sponsor is mandated to promptly notify concerned investigators, ECs/IRBs, and RAs of any findings that could adversely affect the subjects’ safety and potentially impact the progress of the trial (ICH GCP § 5.16.2), such as a clinically significant increase in the occurrence rate of expected serious adverse reactions (SARs), a considerable hazard to patients (lack of efficacy in life-threatening diseases), and important safety findings from newly completed animal studies (ICH E2A § III A 2) [28]. Any modification to study documents (protocol, IB, or ICF) triggered by such observations must be approved by the EC/IRB before implementation (ICH GCP § 3.1.2). Nevertheless, if assessed as necessary for the protection of the subjects against an immediate hazard, urgent safety measures can also be taken without prior approval but should be promptly communicated (ICH GCP § 3.3.7). In the case these measures lead to the suspension or termination of the trial, detailed reasoning for such a decision should also be provided (ICH GCP § 5.21).

Moreover, the sponsor should periodically review the accumulated trial data and prepare periodic safety reports and a final clinical study report for submission to RAs (ICH GCP § 5.17.3) [38]. Periodic reports may have the format of simple line listings of SAEs or thorough annual reports [35, 36], as per applicable local or regional regulations [34]. The DSUR, as recommended by the CIOMS WG VII and described in the ICH Guideline E2F, represents a broadly accepted annual report format. It should present the overall safety profile of the drug based on analysis of cumulative and interval safety data (AEs) collected from any sources (clinical, PM, non-clinical data, and literature) [36] and highlight the updated benefit–risk balance, focusing on important identified and potential risks, new information on known risks, and actions to address emerging safety issues [36]. Notably, the DSUR is a means for the sponsor to inform RAs of the evolving safety profile of the investigational product and to ensure that it is adequately monitored [36].

The sponsors further require a periodic review of the safety data to update the IB and inform investigators of new safety findings [34, 38]. The ICH requests an annual IB revision, specifying that the actual frequency may depend on the drug development stage and the rhythm with which new information becomes available (ICH GCP § 7.1) [38]. To facilitate more frequent updates, the CIOMS WG VI suggests including the DCSI, containing the sponsor’s opinion on the expected AEs, as a distinct part of the IB that can be revised independently [34].

Although the exact methods for periodic safety evaluations are not defined in the ICH guidelines, the ICH GCP refers to the appointment of scientifically qualified independent data monitoring committees (IDMCs) by the sponsor for the regular evaluation of efficacy and safety trial data and subsequent suggestions regarding the continuation, modification, or suspension of the trial (§ 5.5) [38]. The CIOMS WG VI report further elaborates on the necessity and role of data monitoring committees (DMCs), using the term data safety monitoring boards (DSMBs) [34]. The DSMBs are usually trial-specific and are mainly recommended for small or large, complex (but not only), and/or high-risk studies, as judged on the basis of prior relevant safety concerns, the population involved, and the nature of the intervention. In exceptional circumstances, a DSMB monitoring drug safety across the sponsor’s entire drug development program may also be established [34]. The actual composition and operational procedures of the DSMB should be defined in a charter, including details on the safety reviews, such as timing, access to blinded or unblinded data, evaluation of aggregate and individual case reports, and consideration of relevant data from external sources [34]. Regular assessment of the evolving benefit–risk balance by the DSMB thus adds another layer of protection for the study participants beyond the traditional PV methods of SAE reporting.

Interestingly, the CIOMS WG VI goes beyond study-specific DSMBs and proposes the establishment of a safety management team (SMT) for the whole drug development program to oversee all relevant studies, assess the evolving drug safety profile, and make decisions as to whether the IB, DCSI, or ICFs need to be updated and trial amendments are required [34]. To that end, the CIOMS WG VI considers that the concept of safety signal detection and evaluation, primarily developed for the analysis of large safety data sets available in the post-approval stage, should also be adopted during the early drug development phase. However, it is acknowledged that the whole process may be more dependent on clinical judgment than statistical methods, given the usually small amount of data. Furthermore, the CIOMS WG VI also defines the safety signal as “a report or reports of an event with an unknown causal relationship to treatment that is recognized as worthy of further exploration and continued surveillance.” The major tools for the detection of safety signals include (a) the medical evaluation of individual cases (especially SAEs and AEs of special interest) on an ongoing basis, (b) the routine periodic reviews of safety data from the whole development program in aggregate, and (c) reviews triggered by study or program milestones, such as the completion and unblinding of a study. Since a signal may concern not only a new AE but also new information concerning a known ADR (e.g., altered AE intensity or frequency, population vulnerability), meaningful information can be obtained by the comparison of occurrence rates of known events with predefined background rates in the treated population, and subsequent analyses between treatment groups, or by subgroup comparisons with stratification on additional factors (dose, gender, age, concomitant medication), if sufficient data exist.

To facilitate the process, it is further proposed that the sponsor drafts a development risk management plan (dRMP) serving as a guide for safety surveillance [34]. Essential components of the dRMP would be the existing information on the investigational product and epidemiological data on the targeted disease. Such data enable the identification of known or anticipated risks due to the drug or the disease and the estimation of background rates and intensities of expected ADRs and events that manifest the underlying condition. The specification of potential risks for any new product (e.g., QT prolongation and hepatotoxicity, high-risk populations, medication errors, abuse, and off-label use), as well as the exact methodology for risk monitoring and mitigation actions, should also be part of the dRMP. Notably, the proposals have not been incorporated in any ICH Guideline; thus, no country is bound to implement them in their respective laws. Interestingly, however, the latest draft revision of the ICH GCP principles emphasizes the need for periodic safety reviews, as new information accumulates during a trial to make appropriate adjustments if needed [39].

Finally, the sponsor, being ultimately responsible for the trial, should also indirectly survey drug safety by keeping an oversight on the conduct of the trial, the compliance of all stakeholders with the trial documents and applicable regulations, and the quality of the accumulated trial data (ICH GCP § 5.2) [38]. Basic means to that end are the regular monitoring of the clinical sites and the auditing of all relevant processes, by qualified, experienced monitors and auditors, as per risk-based monitoring and auditing plans developed by the sponsor. Safety parameters monitored or audited include the data’s integrity and compliance on AE recording, assessment, and regulatory submissions (ICH GCP § 5.18 and 5.19). Any detected non-compliance impacting the safety of the subjects should be thoroughly investigated by root cause analysis and followed by appropriate preventive and corrective actions to mitigate the risks (ICH GCP § 5.20). In extreme situations, the termination of an investigator’s participation in the trial may be considered (ICH GCP § 5.21). In addition, the sponsor should ensure that any trial-related data and documents are readily accessible for EC/IRB reviews and inspections by RAs when requested (ICH GCP § 6.10).

Ethics Committees/Institutional Review Boards

The ICH GCP guidelines reference the role of ECs and IRBs in safety monitoring. Although there may be slight differences between IRBs and ECs [34], this review uses the terms synonymously. IRBs (in accordance with USA Food and Drug Administration (FDA) regulation) and ECs consist of medical professionals and non-medical members, scientific and non-scientific. They are responsible for ensuring “the protection of the rights, safety, and well-being of the human subjects involved in a trial” and for providing “public assurance for that protection” (ICH GCP definition § 1.27) by reviewing and monitoring biomedical research involving human subjects [38]. Their exact composition, operations, legal status, and regulatory requirements for IRBs and/or ECs depend on the country or region. The favorable opinion of the EC/IRB is a prerequisite for the initiation of a study (ICH GCP § 3.3.6).

Moreover, it is based on the scientific, ethical, and regulatory review of the trial protocol, IB, ICF, and the participant recruitment procedures, as well as the curriculum vitae of the investigators (ICH GCP § 3.1.2) [38]. Any update in the protocol or the ICF also requires the ECs/IRBs’ approval before becoming effective, except for emergencies, when the immediate implementation of an amendment is necessary to protect the trial subjects from hazards. In any case, the ECs/IRBs should be promptly informed of any protocol deviation or new information that suggests increased risks and may adversely impact the subjects’ safety or the trial conduct (ICH GCP § 3.3) [38]. Although the currently effective ICH GCP specifies that all serious unexpected ADRs should be expedited to ECs/IRBs by the investigators, this requirement depends on the national or regional rules. Interestingly, the CIOMS WG VI report questions the necessity of such reporting because ECs/IRBs are usually not appropriately equipped to database, manage, and assess those reports. Instead, it suggests that the sponsor submits regular updates on the risk–benefit profile and only occasionally notifies the ECs/IRBs of single case reports, should they impact the CT or the drug development program [34]. Notably, the ECs/IRBs should implement a risk-based approach to review the ongoing trial continuously and at regular intervals (at least annually) to verify that subjects’ safety is safeguarded (ICH GCP § 3.1.1) [38].

Regulatory Authorities

The role of other (national and regional) RAs is not described thoroughly in the ICH Guidelines. Nevertheless, most RAs rather have the role of a supervisor in the ongoing safety surveillance. Apart from reviewing and approving CT applications and amendments, as per applicable local or regional regulatory requirements, they receive safety updates from the sponsor throughout the study’s progress in the form of expedited individual case reports, emerging safety issues, and periodic safety reports [38]. In addition, they can conduct inspections of CTs at any time if necessary, during which they are entitled to monitor any document relevant to the study [38].

Safety Surveillance During Drug Development: Country/Territory-Specific Requirements

To gain insight into the variety of different safety surveillance processes, this review focuses on the relevant regulations from selected countries representing various territories worldwide, including the USA, European Union (EU), Switzerland, United Kingdom (UK), Saudi Arabia, Japan, and Mexico, as representatives of ICH regulatory members (e.g., engaged in the ICH processes and application of ICH guidelines) [25] (Fig. 2). In addition, the regulations of Australia and South Africa were selected as examples of non-ICH members that are nevertheless ICH observers (e.g., they can attend ICH assemblies and appoint experts for ICH WGs without voting rights or the obligation to implement ICH guidelines) [25].

Fig. 2
figure 2

Regulatory authorities and CT regulations across reviewed countries/territories. UK, United Kingdom; MHRA, Medicines and Healthcare products Regulatory Agency; EU, European Union; EMA, European Medicines Agency; TGA, Therapeutic Goods Administration; DSMB, data and safety monitoring board; SFDA, Saudi Food and Drug Authority; SAHPRA, South African Health Products Regulatory Authority; COFEPRIS, Comisión Federal para la Protección contra Riesgos Sanitarios

Full size image

So far, the ICH GCP has already been incorporated into the local laws of most members (USA, Canada, EU, Switzerland, Turkey, Singapore, Japan, China, and Chinese Taipei), whereas the full implementation of the ICH E2F is pending in Brazil, Republic of Korea, and Saudi Arabia [24]. On the contrary, Mexican HA has not implemented any of the above guidelines, especially in CTs, but has adopted the ICH E2A and E2B on the management and reporting of safety data [40]. The influence of the ICH and CIOMS principles has extended non-ICH countries, including Australia and South Africa, which are reviewed as non-ICH members with representative observers at the ICH. Nevertheless, the surveillance methodology is not harmonized. Differences exist between the ICH regulatory members owing to specificities and additional requirements included in local regulations and/or the level at which the ICH guidelines have been implemented or followed. The following paragraphs only summarize the similarities and focus instead on the points differentiating each country or region from one another and on strategies that diverge from the ICH-defined frame described in the previous paragraphs. Detailed information per country is further provided in Table 2.

Table 2 Safety surveillance methods exercised by sponsors in different countries
Full size table

Overall, there is a general alignment among international guidelines and national regulations on the necessity for continuous safety surveillance of investigational products throughout their development program as a prerequisite to continuing any clinical study to ensure the safety and well-being of the trial participants. The fundamental methods proposed by the ICH and CIOMS, and implemented in all countries reviewed in this work through local laws and guidelines, include (1) routine assessment and follow-up (FU) of individual SAEs and of non-serious AEs of special interest that occur during the CT, (2) periodic evaluation of aggregated AEs at least during the compilation of periodic safety reports, and (3) assessment of any safety-relevant information arising from any source, such as findings from in vitro or in vivo, epidemiological investigations, scientific literature, and PM experience. Although the primary responsibility for the actions mentioned above lies with the sponsor, the investigators, IRBs/ECs, and RAs also contribute significantly to the safety surveillance: (1) the investigators by collecting AEs (as required by the protocol and the locally applicable rules), assessing them in terms of seriousness and causal relationship with the product under investigation, and reporting them to the sponsor, to IRBs/ ECs and RAs (as applicable according to local regulations), and (2) the IRBs/ECs and RAs by reviewing and assessing safety information received from sponsors and/or investigators in the form of expedited individual cases and periodic aggregate reports, as per national and regional mandates.

Diversity among the countries and regions reviewed is observed with regards to (1) the local expedited reporting requirements (origin and types of reportable AEs, the responsible party for causality assessment for reporting purposes (for instance, the USA validates the sponsor’s assessment more than the investigator one for reporting purposes), reporting responsibilities and timelines—the latter not detailed in this work, (2) the need to prepare and submit DSURs as a means of informing RAs of the most recent safety profile of the drug, and, most importantly, (3) the requirements and strategies for aggregate CT data and analysis of safety information. Those differences are also presented in detail in Table 2.

Even if a single safety incident (usually a suspected unexpected serious adverse reaction, SUSAR) may provide information regarding drug safety and lead to alteration of its benefit–risk profile, further investigations considering relevant available safety, pharmacological and epidemiological data are required to corroborate the initial observation. As sufficient information on the drug safety profile is usually lacking during development, accurate causality assessment may not always be possible. Thus, often, it is based on the medical judgement of the investigator. In addition, identification of potential unknown effects of the drug on already known (expected) AEs (e.g., occurrence rate changes, population specificities) or on medical events otherwise anticipated in the targeted population (due to the population or the treated disease characteristics) is not feasible by single case analysis. Hence, reviewing accumulated safety data aggregates is an indispensable component of drug safety surveillance. Indeed, the importance of aggregate safety data review is emphasized in the respective documentation about safety surveillance across different countries, and results of such a review are requested by all countries and regions examined. In addition, specific methods for aggregated data reviews have been proposed by the CIOMS WG VI [34]. Nevertheless, regulations in most countries do not elaborate on establishing specific sponsor processes dedicated to safety data review and leave this responsibility with DSMBs. DSMBs are usually trial-specific and meet at regular intervals to perform independent interim efficacy and safety monitoring to assess the CT progress and provide an opinion on its continuation [34, 38].

To our knowledge, the most comprehensive guidance on aggregate data analysis in the frame of drug development is available in the USA regulations (21 code of federal regulations (CFR) § 312.32) and relevant FDA guidance documents [41,42,43,44]. First of all, the FDA requires an analysis of all previously reported similar AEs or other relevant information together with any new investigational new drug (IND) application report and evaluation of the latest report in the light of that information (21 CFR § 312.32(c)(1)) [42,43,44]. A similar requirement is effective in Saudi Arabia [45] and South Africa [46], thanks to worldwide coordination through cooperative programs. However, in the latter, it applies to any suspected ADR presented in the local safety progress reports [46]. In addition, in contrast with most other countries, the USA federal law and supplementary documents include explicit requirements on the timelines and methods for expediting safety information (SUSARs) that arise from the review of aggregate data (21 CFR § 312.32(c)(1)(iv), § 312.32(c)(1)(i)(C)) [43, 44]. In principle, by adopting the CIOMS WG VI recommendations on establishing an SMT and drafting a dRMP, the USA FDA recommends developing a safety surveillance plan to guide the sponsor’s safety signal detection and management processes from the drug development phase. Parameters addressed in the plan include identification of (i) the responsible bodies, (ii) the events to be monitored as well as estimated occurrence rates (if applicable and feasible), (iii) the frequency of reviews, (iv) the analysis methods, and (v) the risk management strategies [43]. Of note, the EU, Australia, Japan, and South Africa require timely reporting if increases in the incidence rate of expected AEs are observed [46,47,48,49]. In this context, aggregate analysis is the only method that detects these changes. In addition, the latest EU, UK, and Canada regulations imply that the sponsor should implement a signal detection system by requiring that the results of signal detection activities be included in specific sections of the DSUR [48, 50, 51]. However, those regulations do not describe the methods for analyzing signal detection activities. Only the South African legislation has explicit—albeit more simplified—requirements for regular reviews of safety data via the compilation of 6-monthly safety progress reports by the sponsor [46].

Besides the aggregate analysis of safety cases, another point that is not adequately addressed in most national and regional legislations is the review and reporting of safety findings that impact the benefit–risk profile of the drug under development but stem from other sources, such as from in vitro and animal studies and literature. In the USA, such reviews may be foreseen and described in the safety surveillance plan [43]. Moreover, the FDA specifies that data suggesting a significant risk should be expedited in a narrative format (21 CFR § 312.32(c)(1)(v), § 312.32(c)(1)(ii), § 312.32(c)(1)(iii)) [42,43,44]. Saudi Arabia and South Africa also request the submission of those findings as IND or expedited reports, respectively [45, 46]. In Australia, they are notified to the regulatory bodies and investigators as significant safety issues, while in the EU, as events not falling into the definition of SUSAR, often in the frame of urgent safety measures [48]. Japan requests expedited reporting of specific events and incidents identified in the literature. Similarly, in Switzerland, they may be notified to Swiss RA as a rationale for significant CT changes [52, 53]. However, the exact review methods and submission format are not described in the regulatory documents of any of the above countries (excluding the USA).

Challenges in Safety Surveillance in the Clinical Pre-approval Phase of Drug Development

CTs offer a unique opportunity for identifying safety signals, as they are usually blinded and randomized and as the cases received from CT are of high quality and provide data that are often lacking in PM (e.g., lab data, cardiovascular data, etc.). Treatment blinding reduces the bias in reporting of AEs [54]. Randomization, on the other hand, in combination with the rigorous criteria for subject enrolment and the strict definition of medical interventions, reduces the number of confounding factors that need to be considered when comparing the data between study arms [54,55,56]. Knowledge of CT exposure numbers further facilitates estimating AE occurrence rates [56]. In addition, the quality (and quantity) of collected information is under strict monitoring; therefore, CT safety cases usually include sufficient data to enable medical assessments of causality, which is rarely the situation in safety cases collected spontaneously during the PM period.

Nevertheless, the CT characteristics also raise many limitations when monitoring the product’s safety. As a result of the limited number of multimorbid subjects exposed, rare AEs, which become apparent only when tens of thousands or even millions of patients have used the product, will not be observed until after the drug is approved [34, 41]. The difficulty further hinders the detection of safety signals in assessing the causal relationship of AEs with the investigational medicinal product because of the lack of sufficient data on the drug’s safety profile [54]. More participants are recruited at later stages of clinical drug development (usually in phase III studies). However, these pivotal trials are often designed to establish drug effectiveness [32, 41, 54].

Consequently, respective statistical analyses are generally planned on the basis of critical efficacy endpoints and may lack sufficient power to detect statistically significant differences in AE rates between treatment groups [41]. In addition, except if there are existing safety concerns, multiple endpoints of potential interest need to be checked for safety determinations [57]. Therefore, when designing a study, it is critical to corroborate data from previous CTs and other sources (e.g., preclinical toxicity and epidemiological data) to predefine expected and anticipated AEs of interest and estimate respective expected occurrence rates [54, 55]. Pre-specifying hypotheses for risks that can be predicted facilitates appropriate sample size calculation, allows for better statistical planning, and increases the trial power for testing specific safety concerns [55]. Such approaches cannot be applied for unknown or unexpected AEs, the detection of which will still depends on thorough investigations of imbalances of AE rates between the different treatment arms [55]. On the other hand, diversity regarding the safety endpoints in CTs poses a statistical challenge, which can be partially addressed by applying specific adjustments and statistical models on hierarchically structured coded AE terms, as long as the size of the CT database and computational resources permit such analyses [56].

To overcome challenges due to the limited size of CT databases, analyses of pooled data from different studies across the drug development program can increase the statistical power and allow both for verification of consistency between different CTs and for the detection of significant differences in AE rates [55]. Meta-analyses extended across CTs from various sponsors can provide additional insights into the drug’s safety profile [54]. Systematic reviews may precede the meta-analyses to determine the validity and assess the heterogeneity of the CTs, both of which need to be considered during data pooling [54]. To extract meaningful results from pooled data analyses and meta-analyses, the confounding should be minimized when integrating data, and stratifications should be based on parameters like clinical subgroups of patients, type of findings (exploratory vs. pre-specified analyses), study characteristics (long-term vs. short-term, randomized vs. open-label) [55, 57]. Such pooled analyses would further benefit from the global harmonization of CT safety reporting requirements, as this would maximize accumulated data’s availability, consistency, and comparability.

Consistent and accurate coding of AEs is another crucial parameter that facilitates safety assessments by grouping similar events when analyzing data from multiple studies and maximizes the likelihood of detecting important safety information [41]. To that end, the FDA recommends that sponsors review the SAEs reported by the investigators, verify the coding accuracy of all AEs, and ensure that standardized preferred terms are used [57]. A prerequisite for accurate coding is high-quality and complete data [54] obtained through comprehensive and robust data collection systems [55], the latter being often a challenge as different clinical databases may be used for different CTs on the same investigational product. Correct AE identification is of paramount importance, as both broad classification (by using generic terms) and/or misclassification (e.g., due to insufficient information relying on patient reporting rather than systematic monitoring of the subject) may mask a signal [54, 55].

The relatively short duration of the studies compared to the actual duration of treatment in the general population is an additional limiting factor, as it prevents the detection of risks associated with long-term use of the drug or that appear long after treatment (e.g., for drugs that have a long half-life or that deposit in specific organs). [41, 54]. Indeed, CTs are better suited for monitoring AEs close to therapy initiation [54]. Long-term FU of the subjects, even after the end of the study or even if the patient has withdrawn from it, can assist in collecting relevant information [55]. Especially in the case of discontinuation, every effort should be made to obtain the reasons for withdrawal to identify potential AEs that may have led to such a decision [41].

Another challenge in safety monitoring arises from the highly selected population in a CT. Vulnerable populations (such as patients with renal or hepatic impairment, pregnant women, and children) are thus often excluded, reducing data generalizability [54]. The FDA has suggested broadening selection criteria to overcome this, especially in phase III studies [41]. Usually, the studies not sufficiently designed to allow the collection of dose–response data further contribute to the lack of generalizability of CT data [54]. In more detail, the phase II trials, designed to investigate different drug doses, often fail to provide adequate information on that aspect. This is due to various factors, including the short duration of these studies, the small number of subjects, and the common use of pharmacodynamic measurements instead of clinical outcomes as an endpoint [41].

Proposals Towards a Unified Approach

The comparative analysis of regulations guiding safety surveillance at the pre-approval phase revealed two major points to be considered for a harmonized approach: (a) the definition of the types of AEs requiring submission via an expedited or periodic manner to HAs (national and regional RAs and ECs/IRBs), and (b) the establishment of strategies for reviewing aggregate data and information from sources other than CTs and harmonized methods for communicating the results of such reviews to investigators and RAs. The adoption of the CIOMS VI report recommendations, in corroboration with the respective USA initiatives [44, 55], may constitute the first step toward this direction. Emphasis should be given to (a) the development of a safety surveillance plan (as described in the 2021 FDA Guidance) or of a dRMP (as per CIOMS VI recommendations) for the detection of safety signals that covers the whole drug development program and is updated as data accumulate, and (b) the appointment of appropriately qualified safety teams by the sponsor to conduct the reviews (SMTs or DSMBs with broad responsibilities as per CIOMS VI, or DMCs and/or other individuals as per 2021 FDA Guidance) [34, 43, 55]. Important aid in addressing the challenges in the evaluation of clinical data would be the acquired relevant experience from countries like the USA, South Africa, but also the EU since clinical data are already considered in the evaluation of safety signals arising from spontaneous sources [56, 58]. The EU’s PM PV legislation includes well-established safety signal detection and assessment processes; it elaborates on the periodicity of data monitoring, the sources of data, the information to be reviewed, and the exact steps to be followed for managing a detected signal, e.g., validation and further assessment to conclude if it is “false” or if it leads to a new/changed risk for the drug [58]. Adapting similar procedures in the clinical phase of drug development may also help the sponsors establish efficient safety surveillance plans. Notably, the respective EU legislation addresses the active involvement of regulatory bodies, a parameter poorly described in the pre-approval regulations of any countries reviewed. In particular, the EU regulatory network is responsible for confirming validated signals, performing additional analyses, and making appropriate recommendations to the marketing authorization holders [58]. Analogous participation of RAs in the clinical phase could further assist safety monitoring by sponsors and ensure the safe development of drugs, counteracting any possible bias introduced by investigators or sponsors tending to provide proof of safety (and efficacy) in their CTs.

Conclusion

Safety surveillance during the pre-approval clinical drug development phase is of paramount importance as it directly impacts the clinical care of the trial participants [59]. In the longer term, it contributes to the development of a comprehensive safety profile of the drug and makes it possible to anticipate the risks to which the patients are exposed in the post-marketing period. Safe drug development is achieved by identifying potential harms, relevant risk factors, and associations between dose and durations of exposure [55, 59]. Past tragedies have been primarily attributed to poor safety monitoring during the preclinical or the clinical phase (e.g., fialuridine, TGN1412 study, and BIA 10-2474 studies leading to fatal, life-threatening, or debilitating events at early-stage trials) [15, 16, 18]. Insufficient safety planning and surveillance in the clinical phase has further led to significant public health threats with marketed products, often resulting in safety warnings or subsequent withdrawals from the market (e.g., thalidomide, COX-2 inhibitors, rimonabant) [19,20,21, 60]. The aforementioned, in combination with the growing interest (or even suspicion sometimes) of the public in CT results [54], have led to an increased emphasis of both the industry and the regulators on identifying safety signals early in the drug development process [55, 56, 61]; the Pharmaceutical Research and Manufacturers of America formed the Safety Planning, Evaluation, and Reporting Team (SPERT) to propose standards regarding safety planning from the first-in-human use of a drug to the pharmaceutical industry [55]. Those standards were considered to develop the most current USA FDA guidance [44]. The European Medicines Agency (EMA), on the other hand, coordinated a public–private partnership project, the Innovative Medicines Initiative (IMI) Pharmacoepidemiological Research on Outcomes of Therapeutics by a European ConsorTium (PROTECT [62]), to address fundamental research questions on safety signal detection, including—among others—relevant challenges in the CT setup [56].

Although the need for early drug safety monitoring is widely recognized, a harmonized approach at a global level is still lacking. There is a high divergence in individual regulatory requirements across different countries and regions regarding the process for AE causality determination, the nature of events that must be expedited, and the methods and approaches for aggregate data reviews. Despite sponsors having an in-depth knowledge of their investigational product, the contribution of investigators and RAs to the benefit–risk assessment of the drug in development cannot be overlooked; investigators have direct access to patients’ data and expertise on the investigated disease, whereas RAs receive data and oversee medications from various sponsors [63]. This regulatory divergence leads to inconsistencies in reported data, thus reducing the comparability of safety data and complicating communication between stakeholders [63]. As a direct impact, there is an overall increase in the cost and time until an investigational product becomes ready for approval; more resources and time are required to satisfy regulatory requirements and to produce comparable data. This impact is more pronounced nowadays because of the globalization of the clinical landscape over recent decades; CTs for the same or similar drugs are conducted in many countries, involving diverse populations with significant underlying comorbidities and often involving new technologies [63]. To address this issue, TransCelerate [64], a non-profit organization, was formed as a collaboration between 19 pharmaceutical companies working with RAs in compliance with ICH guidelines to promote—among others—the harmonization of global PV processes and requirements [63].

Corroboration of the accumulated experience on drug safety surveillance in the clinical phase at global level and comparative critical review of the respective national/regional legislations and international standards requires coordinated and rigorous efforts from regulatory bodies and the pharmaceutical industry worldwide. Nevertheless, it is essential for reducing the previously described legislation divergencies and for directing the development of safety surveillance plans, the latter provisioning the establishment of qualified teams for aggregate data reviews and defining respective strategies. The robust involvement of RAs by establishing relevant regulations and actively participating in the analysis of safety signals originating form CTs would further assist in that direction. The harmonization of CT safety surveillance would eventually enable the performance of multinational CTs at a lower cost and facilitate the collection of adequate data required to establish the safety and efficacy profile of new drugs. Thus, it would enable the fast approval of safe and efficacious treatments internationally.