The 64 reviewed EU-RMPs included a total of 197 unique SCs (including 70 IIRs, 78 IPRs, and 49 MI Topics) 129 of which were removed from their respective EU-RMPs, while 19 SCs were reclassified either from an IPR to an IIR (n = 15) or from MI to an IPR (n = 4) during the period covered by this review.

Of the seven medicinal products, five had an EU-RMP included in the MAA dossier in line with the applicable regulatory requirements, and two had the initial EU-RMP submitted in the post-authorisation phase. The initial EU-RMPs for the each included medicinal product were dated between 2006 to 2014.

Evolving Regulations have Resulted in Variance in Risk Presentation for Mature Products

To investigate whether the introduction of regulations have affected the number of EU-RMP SCs, we first studied the evolution of the SCs for the two products with the longest Market Authorisation (MA), Products A and B whose first EU-RMPs were submitted post-authorisation (Fig. 1a, b). Product A was authorised in European Member States via MRP in 2002 and its first EU-RMP was submitted in 2007, having a total of nine HA-approved EU-RMPs as of 01 October 2020 (Table 1). From the first submitted EU-RMP in 2007 to the EU-RMP version 4.0 in 2011, the number of IIRs ranged between 30 and 39 (Fig. 1a) reflecting all undesirable effects included in Section 4.8 of the Product A’s EU SmPC effective at the specified time points. The introduction of the GVP Module V in 2012 resulted in the demotion of all the IIRs for Product A as they did not fulfil the criteria as per this guideline [6]. After an initial decrease from 15 to 11 between 2007 and 2008, IPRs for Product A showed a stable development from 2008 until 2019. After which seven IPRs were removed in accordance with the principles outlined in GVP Module V Revision 2 [10].

Product B was authorised for use via CAP in 1997 and the first EU-RMP was submitted in 2006 in conjunction with the approval for use in an additional indication. For Product B, the period from 2006 to 2012 was characterised by a high degree of variability with the inclusion of new SCs as well as the removal, re-addition and re-classification of multiple SCs (Fig. 1b). This was followed by a period of stable development in the number of important risks. Following, the approval of RMP version 11.0 in 2019, one IIR and seven IPRs were removed to align with GVP Module V Revision 2. Furthermore, one IIR and one IPR were included in the RMP in 2019 based on new safety data from post-marketing data.

The first EU-RMP for Product A did not include any MI topics but four topics were added between 2013 to 2015 following the approval for use in a new indication. One MI was related to obligations outlined in the paediatric investigation plan (PIP) encompassing the new indication and was removed once those commitments had been fulfilled. The first EU-RMP for Product B submitted in 2006 included two MI topics and following re-evaluations to align with updated regulatory guidelines, they were first re-classified to IPRs in 2008 and 2009, respectively, and then finally removed from the EU-RMP in 2013 and 2019. In summary, these data showed that alignment with new or revised regulatory guidelines resulted in periodic variance in the risk presentation for the most mature products.

The Introduction of GVP Module V Revision 2 Resulted in a General Decrease in the Number of EU-RMP Safety Concerns

Next, we analysed the lifecycle evolution of SCs for the five more-recent products (Products C, D, E, F and G) with an initial EU-RMP submitted as a part of the MAA dossier (Fig. 2 and Fig. 3a–d). Figure 2 presents the number of SCs that were included or removed during the various stages of the product lifecycles. The majority (n = 50) of the SCs for these five products were included in the initially submitted RMP, while 36 SCs were included in the first HA-approved EU-RMPs as a results of the MAA evaluation process. Additionally, 14 SCs were added as an effect of new safety findings post-authorisation. Although, some (n = 9) SCs were removed from the EU-RMPs prior to the implementation of the GVP Module V Revision 2 in 2017, the implementation of this revision was followed by the removal of a larger number of SCs (n = 34).

The lifecycle evolution of the number of SCs is presented in Fig. 3a–d. Amongst the five products, three (Products C, D and E) saw an overall decrease in the number of SCs following the introduction of the GVP Module V Revision 2 in 2017. Product F had a stable development in the number of IPRs and MI from 2015 to 2020, whereas the number of IIRs increased in 2019. This increase in the number of IIRs was also observed for Product G, i.e., same active substance as in Product F but in a different indication in 2019 reflecting a new SC applicable for both products despite the different target population. Otherwise, for Product G one IPR was removed between 2016 to 2020 in conjunction with the renewal process to reflect that accumulated data did not support the initial categorisation. Furthermore, in 2018 one MI for Product G was removed to reflect the results from new safety data from a post-authorisation study and another MI was added in 2020 following the EMA approval for use in a new target population.

Following the authorisation of Product D in 2014, the number of SCs remained stable until 2016 when the number of IPRs for Product D varied due to parallel submissions of multiple EU-RMPs related to different variation processes. The total number of SCs for Product D, mainly IPRs and MI, decreased after 2016 and the availability of safety data from one large cardiovascular outcome trial. The renewal procedure in 2018 resulted in the removal of multiple IIRs and MI to align the RMP with the Revision 2 principles. The number of important potential and identified risks for Product E increased during the first three years post-authorisation, i.e., between 2012 and 2015, which was then followed by a period of stability until 2017, whereas for MI, this period was associated with sequential decreases and increases. The Product E renewal procedure in 2018 resulted in a decrease in the number of SCs. Product C had a stable development in the number of important risks and MI between 2011 and 2017, and the renewal in 2016 had no impact on the number of SCs. However, following the availability of safety data from two large cardiovascular outcome trials after 2017, several SCs could be removed according to the principles outlined in GVP Module V Revision 2.

Overall, most removals were prompted after the introduction of GVP Module V Revision 2 as shown by the general decrease in the number of EU-RMP safety concerns after 2017.

Mixed Impact of Post-authorisation Safety Studies (PASS)

We then examined the influence of PASS on the lifecycle dynamics of SCs for all seven studied products during the study period. In total, 50 EU-RMP SCs (seven IIRs, 27 IPRs, 16 MI) were associated with additional (a)PV activities such as PASS, whereas less than half (n = 23) of these SCs were removed or reclassified based on the result from PASS (Fig. 4a). Among the 23 aforementioned important risks and MI, 12 were removed or reclassified prior to the implementation of Revision 2 in 2017 (Fig. 4b). For the SCs associated with PASS, 16 had a PASS initiated prior to the introduction of the GVP Module VIII guidelines on post-authorisation studies in 2012 [8]. For IPRs, the outcome of aPV activities lead to the removal or reclassification of eight important potential risks. In contrast, almost all the aPV activities (13 of 16) associated with MI topics resulted in removal or reclassification. Overall, seven IIRs needed aPV activities while only the removal of two IIRs was related to the outcome of aPV activities. While dedicated post-authorisation studies have supported the characterisation of some SCs they did not have an impact on the risk presentation of many SCs.

Preceding Guidelines are Still Having an Influence on the Risk Presentation of Many EU-RMP Safety Concerns

Subsequently, we studied the number and characteristics of the SCs included in the latest HA-approved EU-RMP for each of the seven medicinal products as of 01 October 2020 (Fig. 5a). The two products, Products F and G, for which no decrease was observed after 2017 (Fig 3a–d) had the highest number of SCs (n = 20 and n = 18, respectively). The oldest products, Products A and B, both had five SCs, which is a significant decrease compared to the highest observed number of SCs for the respective products (n = 50 and n = 21) during the study period (Fig. 1a, b). Of the seven products, Product C had the lowest number of SCs (n = 3). The total number of IIRs and MI included in the EU-RMP for Product D had decreased from the maximum 18 to only 8. A similar trend was observed for Product E for which the number of SCs had decreased from a maximum of 17 to 8. None of the current 68 SCs needed any aRM measures (Fig. 5b), 13% (n = 9) were associated with an ongoing or planned aPV activity, whereas 87% (n = 59) did only require routine PV activities. Among the seven medicinal products just two, Products D and G, had SCs (n = 8 and n = 1, respectively) that required aPV activities 3 of which were associated with IIRs, four with IPRs and two with MI topics (Fig. 5c). Thus, while stakeholders have begun to implement the changes introduced by Revision 2, many SCs are still influenced by previous applicable guidelines.

Diverse Sources Influence Dynamics of EU-RMP Safety Concerns

Finally, one assessed how the knowledge gained from various sources (Box 3) has contributed to the characterisation of the safety profile (Fig. 6a–c). The sources of information that prompted the inclusion of SCs (Fig. 6a) were clinical trial data (n = 40), post-marketing data (n = 24) and non-clinical data (n = 9). Furthermore, the inclusion of 55 IIRs and MI was related to available data on the pharmacological class effect and 37 SCs were included based on target population characteristics. Information that triggered the removal (Fig. 6b) of SCs was derived from post-marketing data (n = 40), clinical trial data (n = 30) and PASS (n = 21). The most frequent sources associated with the reclassification of SCs were post-marketing data and clinical trial data, which influenced 15 and 11 reclassifications, respectively (Fig. 6c). Additionally, data from PASS formed the basis of the reclassification of one MI to IPR and one IPR to IIRs. In summary, diverse sources of data contributed to the increasing knowledge of the safety profile during the various stages of the lifecycle for marketed medicinal products.

The results also showed that the introduction of the new GVP Module V in 2012, as well as the revision in 2017, had an impact on the removal of 45 SCs. Note that the introduction of the GVP Module V in 2012 also prompted the removal of 37 identified risks for Product B (Fig. 1b) not included in Fig. 6b. The implementation of new or revised regulations (Volume 9A in 2008 and the GVP Module V in 2012) had an impact on the reclassification of six IPRs to IIRs and three MI to IPRs. Of the reclassified safety concerns, 47% (seven IPRs to IIRs and two MI to IPRs) were related to pharmacological class effect and/or target population characteristics (Fig. 6d).

Feedback During Assessment of the Market Authorisation Application

The last step was to assess whether the process to update (i.e., inclusion, removal, or re-classification) the risk presentations was prompted by HAs or by BI. HAs prompted the removal of 49 % (n = 45) of the SCs, whereas BI proactively proposed the removal of 51 % (n = 46) SCs (Fig. 6e). Feedback from HA during the assessment of the MA application dossier, resulted in the inclusion of 34 new SCs to the EU-RMPs. Of these SCs, 50 % (2 IIRs, 10 IPRs and 5 MI) were associated with pharmacological class effect and/or target population characteristics (Fig. 6f). Thus, the results showed that BI and HAs played an equal role in proposing the removal of important risks and MI topics, whereas HAs made a larger contribution to the inclusion of SCs, in particular those associated with class effect and target population.