Real-World Data Included in Post-authorisation Measures: A Case Study of Approved Advanced Therapy Medicinal Products in the European Union between 2013 and 2024

Abstract

Background

Advanced therapy medicinal products (ATMPs) often require long-term monitoring to assess both safety and efficacy post-authorisation due to uncertainties identified during the approval process. This study aims to characterise the use of real-world data (RWD) in post-authorisation measures (PAMs) for ATMPs approved in the European Union.

Methods

A systematic extraction of all PAMs from publicly available European Medicines Agency (EMA) regulatory documents for ATMPs approved between January 2013 and December 2024 was performed, followed by the identification of the presence and sources of RWD. Additional databases including the HMA-EMA Catalogue of RWD studies and sources and ClinicalTrials.gov were consulted.

Results

Amongst 25 ATMPs approved by the European Commission over the study period, a total of 118 PAMs were identified, of which 49 (41.5%) involved RWD. Most RWD-PAMs were imposed by the EMA (n = 34; 69.4%), secondary data use was the most referenced data use type (n = 28; 57.1%) and registries were the main source of RWD being mentioned (n = 26; 53.1%). Further, 5 (10.2%) included a comparator group and 13 (32.5%) incorporated patient-reported outcomes.

Conclusions

This study emphasises the instrumental role of RWD in the post-authorisation monitoring of ATMPs in the European Union. PAMs reflect the regulatory flexibility for these products, shifting some efforts to the post-authorisation phase to address benefit–risk gaps. Enhancing the use of RWD in this context could improve evidence generation, minimise uncertainties and support more informed regulatory decisions.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40259-025-00737-x.

Key Points

Advanced therapy medicinal products (ATMPs) are often approved under considerable uncertainties, with submissions based on limited clinical trial data, therefore requiring post-authorisation evidence to fill benefit–risk data gaps.

Real-world data (RWD) plays a crucial role in the post-authorisation monitoring of ATMPs, with regulators imposing RWD-based post-authorisation measures to address long-term efficacy and safety.

Despite its potential, RWD face challenges in integration, requiring reliable data sources, better transparency and early stakeholder engagement to enhance its role in decision-making.

Introduction

Unlike traditional medicines, advanced therapy medicinal products (ATMPs) are derived from cells and tissues and involve complex manufacturing processes that rely on innovative technologies. Their development is particularly challenging, costly and lengthy, especially when targeting rare diseases or conditions with a high prevalence of unmet medical needs. These challenges span across manufacturing, non-clinical development, clinical trials, marketing authorisation and post-market surveillance [1, 2].

In recognition of the unique characteristics of ATMPs, the European Medicines Agency (EMA) established the Committee for Advanced Therapies (CAT) in 2009 following the implementation of Regulation 1394/2007 [3]. In 2016, the EMA introduced the PRIority MEdicines (PRIME) scheme to accelerate the development of treatments addressing unmet medical needs [4]. Whilst this framework allows for faster patient access with smaller datasets, it also needs more rigorous post-authorisation safety and efficacy monitoring [5].

Although clinical trials correspond to the major body of evidence for marketing authorisation (MA), they are inherently limited in their ability to fully represent real-world patient populations [6, 7]. Strict inclusion criteria and controlled study settings create a gap between trial-generated data and clinical practice. To bridge this gap, real-world evidence (RWE) generated through real-world data (RWD) serves as a fundamental tool for follow-up assessments post-authorisation [7]. These data could be especially important in the context of ATMPs, as their complexity, their potential to fulfil unmet medical needs, the frequent indication for rare diseases and reliance on small, uncontrolled clinical trials contribute to the challenges of having robust evidence at the time of authorisation [8].

Given these uncertainties, long-term monitoring is crucial to ensure the safety and effectiveness of ATMPs in real-world settings. To address these concerns, the EMA issued the ‘Guideline on Safety and Efficacy Follow-up and Risk Management of Advanced Therapy Medicinal Products’, emphasising the need for continued data collection beyond initial approval [9]. In line with this, the European regulatory framework foresees the implementation of post-authorisation measures (PAMs), for which marketing authorisation holders (MAHs) are required to generate additional evidence after the marketing authorisation to further support and refine benefit–risk assessments over time [10].

Assessing the integration of RWD into PAMs is then crucial, as these data provide valuable insights into long-term efficacy and safety in routine clinical care settings, thus complementing the data obtained from clinical trials [11]. A more up-to-date analysis of RWD use in the post-authorisation setting would be highly relevant in evaluating the effectiveness of efforts by regulatory agencies, MAHs, academia, patient organisations and other stakeholders in leveraging such data for decision-making. Amongst the various types of RWD sources, patient registries have gained increasing relevance. As structured systems for collecting uniform data on patients diagnosed with a particular condition or receiving specific treatments, registries are particularly useful for tracking disease progression, standard-of-care practices, and long-term outcomes [12]. The growing importance of these organised RWD infrastructures is reflected in the EMA’s Patient Registry Initiative, launched in 2015, which underscores the regulatory effort to harness registries as RWD sources for more informed decision-making, including in the post-authorisation setting [13].

Hence, this study aims to analyse the use of RWD in post-authorisation measures for ATMPs approved in the European Union between 2013 and 2024, focussing on their objectives and study designs. Moreover, we also aim to conduct an exploratory characterisation of registries used as RWD sources for ATMPs’ post-authorisation monitoring.

Methods

Definitions

PAMs, as defined by the EMA and according to the European legal framework, were considered regulatory measures to be completed by MAHs to provide additional post-authorisation data. These measures were categorised into different categories. Annex II conditions correspond to obligations imposed by the European Commission to the MAHs which are crucial to further inform on the benefit–risk of the medicinal product in question. Specific obligations are considered binding conditions imposed when the medicinal product is granted a conditional marketing authorisation (the MAH needs to provide the required data within a set timeframe) or when this marketing authorisation is granted under exceptional circumstances (when comprehensive data cannot be obtained even after authorisation). Category 3 studies of the risk management plan (RMP) correspond to additional pharmacovigilance activities required—not imposed—to address specific safety concerns or to assess the effectiveness of risk minimisation measures [10].

According to the EMA, RWD were defined as data collected in routine clinical practice that describes not only patient characteristics, but also treatment utilisation and its respective outcomes [14]. On the contrary, a registry [15] was defined as an organised system that uniformly collects data to assess pre-defined outcomes in a population according to a particular disease, condition or exposure, thereby serving as a source of RWD. The term ‘product registry’ is sometimes used to refer to data collection systems by MAHs to track patients exposed to a specific medicine. Nevertheless, the EMA guideline on registry-based studies clarified that, from a regulatory standpoint, this scenario qualifies as a clinical trial or non-interventional study, and therefore, these terms should be applied instead of ‘product registry’ [16].

Study Sample and Post-authorisation Measures Eligibility Criteria

A list of authorised medicinal products classified as ATMPs, according to Regulation (EC) no. 1394/2007 [3], was obtained from the EMA medicines data table available online [17]. After filtering specifically for ATMPs, a total of 27 products approved by the European Commission following a positive opinion from the EMA’s Committee for Medicinal Products for Human Use (CHMP) up to December 2024 was identified.

Two products (ChondroCelect and Glybera) were excluded, as they were authorised in 2009 and 2012, respectively, prior to the full implementation of the pharmacovigilance legislation [18], which came into effect in 2012. This choice would allow for a more systematic and standardised approach to data collection and analysis.

All PAM categories were included (annex II conditions, specific obligations and category 3 studies of the RMP). However, PAMs without a clinical objective, such as those related to quality, toxicology or procedural aspects, were excluded from the analysis. Whilst compassionate use programs represent a potential use case for RWD, they were excluded due to the challenge of systematically identifying information on the variables of interest in the data sources considered.

Data Sources

Detailed data on the included ATMPs were extracted from the EMA medicines data table, available on the agency’s website to characterise the sample of medicines studied. Additionally, publicly available regulatory documents—including European Public Assessment Reports (EPARs)—were consulted to complete information not present in the summarised data table provided by the agency.

EPARs (both initial and variation reports for extensions of indication) and RMPs were screened to identify PAMs across all studied ATMPs, including the identification of RWD use and registries in the retrieved studies. Further details were extracted from the Heads of Medicines Agencies-European Medicines Agency (HMA-EMA) Catalogues of RWD studies and ClinicalTrials.gov [19, 20] when such information was not available in the regulatory documents. For the HMA-EMA Catalogues of RWD studies, if study protocols were available, that information was prioritised over the abstracted study-related information published on the respective Catalogues’ study page.

Data on registries’ features was retrieved primarily from the HMA-EMA Catalogues of RWD sources and the official registry webpages when the first option was not available. Furthermore, EMA qualification opinions on registries were consulted to assess whether the identified registries had undergone this qualification process. This EMA’s initiative provides opinions on the qualification of innovative development methods and letters of support for novel methodologies that have shown promise in pharmaceutical research and development. The opinions are issued by EMA's CHMP on the basis of recommendations from the Scientific Advice Working Party and are published in the respective section of the EMA webpage. In the context of registries, these qualifications describe the circumstances in which the agency considers the use of that specific registry data suitable for regulatory purposes [21]. Electronic Supplementary Material Appendix A summarises the data sources considered for the study and illustrates how each was prioritised for the different sets of variables, mostly related to ATMP characterisation, PAM identification (including RWD use), and registry description.

Data Collection

A systematic extraction of PAM-related details from publicly available EMA regulatory documents was conducted in Microsoft Excel. Data extraction was performed independently by two researchers (D.A., D.M. or C.T.), and in case of disagreement, a third researcher (B.S.) was consulted to reach a consensus. A random sample of 10% of the products was used to pilot the extraction table.

General information about each ATMP—brand name, international non-proprietary name (INN), year and type of marketing authorisation, Anatomical Therapeutic Chemical Classification System (ATC) code, therapeutic indication and product marketing authorisation status—was obtained from the EMA medicines data table to characterise the sample of medicines included in the study.

Outcome variables were defined to characterise PAM study details and RWD use, including PAM category, study objective, study design, planned sample size, planned duration, population covered, countries where the study is conducted, RWD presence and source. Registry-related variables were adapted from pre-existing checklists such as the REQuEST tool published by EUnetHTA [22] and the EMA guideline on registry-based studies [16], which collectively address dimensions of registry quality. Information on registry holders, year of registry establishment, funding, country coverage, total number of patients enrolled, PRO collection and the availability of contacts, as well as data protection and privacy policies, were retrieved from the registry-related information sources (HMA-EMA Catalogues of RWD sources and the official registry webpages). The supplementary material presents the variables defined for this study, along with their possible values and coding strategy (Electronic Supplementary Material Appendix A). Data collection was completed in January 2025, corresponding to the cut-off date for data extraction.

Data Analysis

Descriptive statistics were calculated to summarise numerical and categorical variables across all ATMPs included in the study sample. All analyses were performed using R software (version 2024.04.2+764). Data management, cleaning, analysis, tabulation and visualisation were performed using the dplyr, flextable, ggplot2, gtsummary, janitor, readxl, maps, officer, stringr and tidyr packages.

Results

Characteristics of Authorised Advanced Therapy Medicinal Products

Between 2013 and 2024, 25 ATMPs were authorised (Maci, Provenge, Holoclar, Imlygic, Strimvelis, Zalmoxis, Spherox, Alofisel, Kymriah, Luxturna, Yescarta, Zynteglo, Libmeldy, Tecartus, Zolgensma, Abecma, Skysona, Breyanzi, Carvykti, Ebvallo, Roctavian, Upstaza, Hemgenix, Casgevy and Beqvez). From these, half received a standard marketing authorisation (n = 13; 52.0%), followed by conditional marketing authorisations (n = 10; 40.0%), whilst 2 (8.0%) products were authorised under exceptional circumstances. Most products remain authorised (n = 19; 76.0%), whilst 5 (20.0%) had their marketing authorisations withdrawn (Provenge (2015), Zalmoxis (2019), Zynteglo (2022), Skysona (2021) and Alofisel (2024)). All withdrawn products resulted from the request of the respective MAHs due to commercial reasons, except for Alofisel, for which the clinical benefit was no longer demonstrated, and the MAH considered it impossible to provide the EMA-required effectiveness data. One (4.0%) ATMP had its marketing authorisation expired (Maci (2018)) due to the MAH not proceeding with the product’s marketing authorisation renewal. Additionally, its marketing authorisation had been suspended since November 2014 due to the absence of an authorised manufacturing site, and it had not been lifted before the expiry.

Most ATMPs fall under the ATC code ‘L–Antineoplastic and immunomodulating agents’ (n = 12; 48.0%), have an orphan designation (n = 16; 64.0%) and were submitted to the PRIME scheme (n = 14; 56%) (Table 1).

Table 1.

Characteristics of the advanced therapy medicinal products included in the study

Product name	INN	Therapeutic area	ATC code	MA type	MA year	Product status	Orphan	PRIME
Maci	Matrix-applied characterised autologous cultured chondrocytes	Fractures, cartilage	M09AX02	Regular	2013	Expired	No	No
Provenge	Sipuleucel-T	Prostatic neoplasms	L03AX17	Regular	2013	Withdrawn	No	No
Holoclar	Ex vivo expanded autologous human corneal epithelial cells containing stem cells	Stem cell transplantation, corneal diseases	S01XA19	Conditional	2015	Authorised	Yes	No
Imlygic	Talimogene laherparepvec	Melanoma	L01XX51	Regular	2015	Authorised	No	No
Strimvelis	Autologous CD34+ enriched cell fraction that contains CD34+ cells transduced with retroviral vector that encodes for the human ADA cDNA sequence	Severe combined immunodeficiency	L03	Regular	2016	Authorised	Yes	No
Zalmoxis	Allogeneic T cells genetically modified with a retroviral vector encoding for a truncated form of the human low affinity nerve growth factor receptor (ΔLNGFR) and the herpes simplex I virus thymidine kinase (HSV-TK Mut2)	Hematopoietic stem cell transplantation, graft versus host disease	L01	Conditional	2016	Withdrawn	No	No
Spherox	Spheroids of human autologous matrix-associated chondrocytes	Cartilage diseases	M09AX02	Regular	2017	Authorised	No	No
Alofisel	Darvadstrocel	Rectal fistula	L04	Regular	2018	Withdrawn	Yes	No
Kymriah	Tisagenlecleucel	Precursor B-cell lymphoblastic leukaemia-lymphoma, large B cell, diffuse lymphoma	L01XL04	Regular	2018	Authorised	Yes	Yes
Luxturna	Voretigene neparvovec	Leber congenital amaurosis, retinitis pigmentosa	S01XA27	Regular	2018	Authorised	Yes	No
Yescarta	Axicabtagene ciloleucel	Follicular lymphoma, large B-cell, diffuse lymphoma	L01XX70	Regular	2018	Authorised	Yes	Yes
Zynteglo	Betibeglogene autotemcel	beta-Thalassemia	B06A	Conditional	2019	Withdrawn	No	Yes
Libmeldy	Autologous CD34+ cells encoding ARSA gene	Leukodystrophy, metachromatic	N07	Regular	2020	Authorised	Yes	No
Tecartus	Brexucabtagene autoleucel	Mantle-cell lymphoma	L01X	Conditional	2020	Authorised	Yes	Yes
Zolgensma	Onasemnogene abeparvovec	Spinal muscular atrophy	M09AX09	Conditional	2020	Authorised	Yes	Yes
Abecma	Idecabtagene vicleucel	Multiple myeloma	L01	Regular	2021	Authorised	Yes	Yes
Skysona	Elivaldogene autotemcel	Adrenoleukodystrophy	N07	Regular	2021	Withdrawn	No	Yes
Breyanzi	Lisocabtagene maraleucel	Large B cell, diffuse lymphoma, follicular lymphoma, mediastinal neoplasms	L01XL08	Regular	2022	Authorised	No	Yes
Carvykti	Ciltacabtagene autoleucel	Multiple myeloma	L01XL05	Conditional	2022	Authorised	Yes	Yes
Ebvallo	Tabelecleucel	Lymphoproliferative disorders	L01XL09	Exceptional circumstances	2022	Authorised	Yes	Yes
Roctavian	Valoctocogene roxaparvovec	Haemophilia A	B02BD15	Conditional	2022	Authorised	Yes	Yes
Upstaza	Eladocagene exuparvovec	Amino acid metabolism, inborn errors	A16AB26	Exceptional circumstances	2022	Authorised	Yes	No
Hemgenix	Etranacogene dezaparvovec	Haemophilia B	B02BD	Conditional	2023	Authorised	Yes	Yes
Casgevy	Exagamglogene autotemcel	beta-Thalassemia, sickle cell anaemia	B06AX05	Conditional	2024	Authorised	Yes	Yes
Beqvez	Fidanacogene elaparvovec	Haemophilia B	B02BD17	Conditional	2024	Authorised	No	Yes

ATC Anatomical Therapeutic Chemical Classification System, INN international non-proprietary name, MA marketing authorisation, PRIME PRIority Medicines scheme

Overview of Post-authorisation Measures

After screening the regulatory documents for the 25 included ATMPs, a total of 136 PAMs were identified. Of these, 18 were excluded due to the following reasons: 12 PAMs were related to quality questions, 3 PAMs were related to compassionate use programs, 2 PAMs were related to procedural questions and 1 PAM was related to toxicology aspects. As a result, the total number of included PAMs was 118.

Following the exclusion of non-eligible PAMs, each ATMP presented an average of 4.7 PAMs (standard deviation (SD) ± 2.3; range 1–10). On average, ATMPs with a conditional marketing authorisation had the highest number of PAMs (5.9; SD ± 2.2; range 3–10), followed by products granted a full-standard marketing authorisation (4.2; SD ± 2.1; range 1–8) and lastly, by ATMPs approved under exceptional circumstances (2.0; SD ± 0.0; range 2–2).

Category 3 studies of the RMP were the most prevalent PAM category (n = 43; 36.4%), followed by annex II conditions (n = 40; 33.9%) and specific obligations (n = 35; 29.7%).

Regarding the objectives of the studied PAMs, the majority were related to both efficacy and safety aspects (n = 74; 62.7%), with 23 (19.5%) solely being related to safety and 8 (6.8%) to efficacy. Most of the included PAMs employed a clinical trial design (n = 67, 56.8%), amongst which 47 (70.1%) were single arm, 62 (92.5%) were open label, and 55 (82.1%) were non-randomised. Phase III trials were the most prevalent (n = 30; 44.8%). In addition, 39 PAMs (33.0%) consisted of non-interventional cohort studies, from which 38 (95.0%) had a prospective study design.

As for the size of the population included in the studies, the median planned sample size for all PAMs was 84 (interquartile range (IQR) 190), and the majority of studies (n = 67; 56.8%) comprised an adult-only population. Orphan ATMPs presented PAMs with a lower median planned sample size when compared with non-orphan ATMPs (124; IQR 193 versus 73.5; IQR 174.2). The identified PAMs across all ATMPs have a mean planned duration of 7.8 (SD ± 5.9; range 1–21) years and are planned to be conducted in eight countries (SD ± 7.0; range 1–32), on average. The vast majority of studies are planned to be conducted in European countries (n = 94; 79.7%), of which 27 (28.7%) occur exclusively in Europe.

As of the data extraction cut-off date, most studies listed in the EMA RWD Catalogues and ClinicalTrials.gov were still ongoing (n = 58; 49.2%), and 36 (30.5%) were finalised. Concerning study protocols’ availability, only 12 (10.2%) protocols were uploaded in EMA RWD Catalogues at the time of data extraction (n = 7; 58.3% concerning annex II conditions and n = 5; 41.7% for category 3 studies of the RMP).

Utilisation of Real-World Data

Amongst the 118 PAMs, RWD was found to be used in 49 (41.5%) measures. For these RWD-PAMs, most were imposed by the EMA (n = 34; 69.4%), with 28 (82.4%) and 6 (17.6%) being annex II conditions and specific obligations, respectively (Fig. 1).

Fig. 1.

Distribution of the post-authorisation measures’ categories amongst the studied advanced therapy medicinal products and the inclusion of real-world data

From all the 25 ATMPs included in the study, only Spherox did not present RWD-PAM. Concretely, this medicine only had one imposed PAM (annex II condition) referring to the completion of a phase III clinical trial with both efficacy and safety objectives. As for the 24 remaining products, RWD-PAMs were present with a mean frequency of 2.0 (SD ± 1.0; range 1–4) per ATMP authorised between 2013 and 2024.

Secondary data use was the most referenced data use type amongst RWD-PAMs (n = 28; 57.1%), and registries were the main source of RWD being mentioned (n = 26; 53.1%). Assessment of both efficacy and safety remained the most common objective for RWD-PAMs (n = 20; 40.8%), followed by safety-focussed studies (n = 16; 32.6%). Another noteworthy category corresponded to studies targeted to assess risk minimisation measures’ effectiveness (n = 7; 14.3%).

The median planned sample size for all RWD-PAMs was 200 (IQR 436.2), with most studies (n = 29; 59.2%) focussing on adults. Orphan medicines had a lower median sample size than non-orphan ATMPs (186; IQR 388.8 versus 230; IQR 517.8). The average planned study duration was 11.8 (SD ± 6.0, range 1–21) years, with studies being conducted in an average of nine countries (SD ± 8.8, range 1–32). Most studies were planned to occur in Europe (n = 41; 83.7%), with 46.3% (n = 19) occurring exclusively in Europe (Fig. 2). According to the EMA RWD Catalogues and ClinicalTrials.gov, 53.1% of studies were ongoing (n = 26) and 12.2% were finalised (n = 6). The finalised studies corresponded to six distinct medicines (Maci (2013), Provenge (2013), Holoclar (2015), Imlygic (2015), Strimvelis (2016) and Yescarta (2018)), all of which were authorised in or before 2018. The same 12 (25%) protocols mentioned to be available in the previous section also correspond to PAMs including RWD.

Fig. 2.

Distribution of advanced therapy medicinal products’ post-authorisation measures presenting real-world data, per country

Comparators were applied in RWD-PAM studies for Alofisel, Casgevy, Beqvez and Zalmoxis, with Casgevy contributing with two studies (n = 5; 10.2%). Most RWD-PAMs including comparators were imposed by the EMA (3 annex II conditions and 1 specific obligation), except for Alofisel, whose PAM corresponded to a category 3 study of the RMP. In all cases, the comparators used were defined as the standard of care for each condition, according to currently available treatment strategies.

The use of patient-reported outcomes (PROs) was low (n = 13; 26.5%). From these 13 studies, 17 distinct patient-reported outcome measures (PROMs) were identified, and 2.4 PROMs (SD ± 1.8; range 1–6) were used on average per study. PROs were mainly obtained via primary data collection (n = 7; 53.8%). Quality of life was the most assessed domain by these PROMs (n = 8; 61.5%), with EQ-5D and its variations being the most used instrument (n = 5; 38.5%). Table 2 summarises RWD-PAMs information per PAM category.

Table 2.

Information on real-world data-post-authorisation measures (n = 48) identified for the included advanced therapy medicinal products

Variable		Imposed measures		Required measures
		Annex II condition n = 28c	Specific obligation n = 6c	Category 3 studies n = 15c
Real-world data source
Primary use	Prospective patient-based data collection	5 (17.9%)	1 (16.7%)	6 (42.8%)
	Questionnaire	0	0	7 (46.7%)
Secondary use	Healthcare records	1 (3.6%)	1 (16.7%)	0
	Registrya	21 (75.0%)	4 (66.7%)	1 (7.1%)
Objective
Efficacy		1 (3.6%)	0	0
Safety		12 (42.8%)	0	4 (28.6%)
Efficacy and safety		13 (46.4%)	4 (66.7%)	3 (21.4%)
Efficacy, safety and disease epidemiology/drug utilisation		2 (7.1%)	2 (33.3%)	1 (7.1%)
Risk minimisation measures effectiveness		0	0	7 (46.7%)
Planned sample size (mean (± SD) | median (IQR))		605 (800) | 300 (630)	131 (148) | 65 (273)	593 (1658) | 100 (145)
Planned duration in years (mean (± SD) | median (IQR))		14 (5) | 15 (2)	15 (0) | 15 (0)	6 (5) | 5 (2)
Number of countries (mean (± SD) | median (IQR))		12 (10) | 9 (17)	6 (3) | 5 (4)	5 (4) | 5 (5)
Use of comparator		3 (10.7%)	1 (16.7%)	1 (7.1%)
Patient-reported outcome collection		7 (25.0%)	2 (33.3%)	4 (28.6%)
Status (as of January 2025)b
Finalised		1 (3.6%)	0	5 (35.7%)
Ongoing		17 (60.7%)	2 (33.3%)	7 (46.7%)
Planned		3 (10.7%)	1 (16.7%)	0
Terminated		1 (3.6%)	0	0
Withdrawn		2 (7.1%)	0	0

^aThe numbers presented include misclassified registries

^bThe variable ‘Status’ corresponds to the combined status of both HMA-EMA RWD Catalogues and ClinicalTrials.gov at the cut-off date for data extraction, January 2025

^cSome sums do not correspond to the total n because, for certain measures, not all variables of interest are available

Registries as Real-World Data Sources

A total of 26 (54.2%) RWD-PAMs referenced registries as the main source of RWD. However, after a more thorough and critical analysis, 8 out of 26 PAMs mistakenly applied the term ‘registry’ (1 for Carvykti, 1 for Imlygic, 1 for Luxturna, 2 for Provenge, 1 for Strimvelis, 1 for Upstaza and 1 for Zolgensma). In all these scenarios, the term registry was misapplied in the context of ‘product registries’. The detailed PAM information revealed that the use of the term ‘registry’ did not align with the EMA guideline on registry-based studies, as all instances referred to a single cohort study design involving the respective ATMP. Additionally, two PAMs (concerning Beqvez and Kymriah, respectively) did not provide enough information to perform a similar analysis, making it impossible to determine whether the term ‘registry’ was correctly applied considering the relevant guideline. As a result, of the 26 RWD-PAMs citing registries as the primary data source, 16 were further assessed to gather more detailed information on registry usage.

From the 16 appropriately classified registry-based PAMs, all measures were imposed by the EU regulatory agency. These PAMs were imposed on 11 distinct ATMPs (Abecma, Breyanzi, Carvykti, Casgevy, Hemgenix, Kymriah, Roctavian, Skysona, Tecartus, Yescarta and Zynteglo), collectively accounting for 44% of the studied sample which relied on RWD obtained from registries in the post-authorisation setting.

Amongst the 16 registry-based PAMs, data were sourced from five distinct registries—the American Thrombosis and Hemostasis Network (ATHN) Transcends, the European Society for Blood and Marrow Transplantation (EBMT) Registry, the Center for International Blood and Marrow Transplantation Research (CIBMTR) Registry, the World Federation of Hemophilia (WFH) Gene Therapy Registry (GTR) and the Gesellschaft für Pädiatrische Onkologie und Hämatologie–German Society for Pediatric Oncology and Hematology–(GPOH) Rare Anemia Registry. Table 3 summarises for which ATMPs RWD-PAM was identified and which ATMPs included registries in RWD-PAMs, and specifies the registries found in line with those mentioned previously.

Table 3.

Number of real-world-post-authorisation measures and registry use amongst all advanced therapy medicinal products

Product name	MA type (MA year)	No. RWD-PAMs	Registry use
Maci	Regular (2013)	1(50.0%)	No
Provenge	Regular (2013)	3 (75.0%)	Yes (product registries)
Holoclar	Conditional (2015)	2 (66.7%)	No
Imlygic	Regular (2015)	2 (25.0%)	Yes (product registry)
Strimvelis	Regular (2016)	2 (50.0%)	Yes (product registry)
Zalmoxis	Conditional (2016)	1 (25.0%)	No
Spherox	Regular (2017)	0	No
Alofisel	Regular (2018)	1 (25.0%)	No
Kymriah	Regular (2018)	3 (50.0%)	Yes (CIBMTR, EBMT)
Luxturna	Regular (2018)	2 (100.0%)	Yes (product registry)
Yescarta	Regular (2018)	2 (25.0%)	Yes (EBMT)
Zynteglo	Conditional (2019)	3 (60.0%)	Yes (EBMT, GPOH Rare Anemia Registry)
Libmeldy	Regular (2020)	1 (25.0%)	No
Tecartus	Conditional (2020)	4 (57.1%)	Yes (EBMT)
Zolgensma	Conditional (2020)	2 (33.3%)	Yes (product registry)
Abecma	Regular (2021)	1 (25.0%)	Yes (CIBMTR, EBMT)
Skysona	Regular (2021)	2 (50.0%)	Yes (CIBMTR)
Breyanzi	Regular (2022)	1 (25.0%)	Yes (CIBMTR, EBMT)
Carvykti	Conditional (2022)	3 (50.0%)	Yes (CIBMTR, product registry)
Ebvallo	Exceptional circumstances (2022)	1 (50.0%)	No
Roctavian	Conditional (2022)	4 (40.0%)	Yes (WFH GTR)
Upstaza	Exceptional circumstances (2022)	1 (50.0%)	Yes (product registry)
Hemgenix	Conditional (2023)	3 (5)	Yes (ATHN Transcends)
Casgevy	Conditional (2024)	3 (9)	Yes (CIBMTR, EBMT)
Beqvez	Conditional (2024)	1 (4)	Yes (not available)

Percentages are calculated on the basis of the total number of post-authorisation measures for each product

MA marketing authorisation, PAMs post-authorisation measures, RWD real-world data

For all registries mentioned, registry data were used to collect outcomes of interest relevant to the PAMs under study. For eight (50.0%) studies, registry data were used for outcomes related exclusively to safety, whilst the other half (n = 8; 50.0%) corresponded to studies where registry-based outcomes pertained to both efficacy and safety. Additionally, amongst the 13 PAMs that collect PROs, only 1 (7.7%) sourced these data from a registry—an annex II condition imposed for Roctavian.

Across the five different registries mapped, only the WFH GTR was registered in the HMA-EMA RWD Catalogues database, being first published on 1 February 2024 and last updated on 17 October 2024. ATHN, CIBMTR and EBMT were registered in the database but only under the registry holder institutions (institution profile). For the GPOH Rare Anemia Registry, neither the registry nor the respective holder organisation were indexed in the EMA database. Regarding data source qualification, EBMT was the only registry included in our sample that was qualified by the EMA, a qualification that occurred in 2019. Supplementary material comprises a more systematic representation of the information regarding the five registries identified (Electronic Supplementary Material Appendix B).

Discussion

ATMPs’ innovativeness, complexity and novelty contribute to the uncertainties regulators face during the marketing authorisation process. PAMs are valuable tools in the regulatory framework, especially for ATMPs, by enabling additional evidence generation that is not available at the pre-authorisation stage [18]. This mechanism illustrates how European regulators navigate the balance between prioritising access to these technologies whilst requiring additional data to support the decisions made despite often-limited pre-authorisation evidence on the benefit–risk balance [23].

Our study highlights that a total of 118 PAMs were attributed to the 25 ATMPs authorised in the European Union between 2013 and 2024. The PAM fingerprint for the studied ATMPs may reflect the marketing authorisation pathway these advanced therapies undergo. The significant prevalence of conditional marketing authorisations and authorisations granted under exceptional circumstances amongst these medicinal products highlights the existing uncertainties at the time of approval, which in turn, often lead to a higher frequency of PAMs compared with ATMPs granted a standard marketing authorisation [10, 23].

Interventional PAMs were the most prevalent amongst all PAMs included in the study, with clinical trials accounting for 56.8% of all PAMs and phase III studies being the most common. These trials were also mostly single arm, open label and non-randomised, reflecting the typical characteristics of clinical trials usually submitted as evidence for marketing authorisation in the ATMP and the orphan medicine landscapes [8, 24, 25]. Thus, the higher prevalence of this design could indicate that some trials were ongoing at the time of approval, with final results yet to be submitted, or that new studies were required to address the benefit–risk gaps remaining from the marketing authorisation moment. This also mirrors the consequences of the high prevalence of PRIME ATMPs and the inherent regulatory flexibility of this framework [25]. Amongst the ATMPs presenting more frequently imposed PAMs (Tecartus, Carvykti, Casgevy, Kymriah, Roctavian), all were submitted to the PRIME scheme and are orphan medicines (except for Zynteglo, which was only submitted for the PRIME scheme).

Nevertheless, these frequently required interventional PAMs might not fully clarify the role of these products, nor the impact they have in routine clinical practice, highlighting the need for conducting RWD-based studies [26]. In fact, RWD contributes significantly to the generation of evidence in the context of PAMs, as it provides relevant additional post-authorisation data, such as long-term efficacy and safety data and a better characterisation of the target population [27, 28]. For example, RWD from anti-CD19 CAR T cell therapy has shown that, in the post-authorisation context, patients tend to have more advanced disease, exhibit greater heterogeneity, and experience longer manufacturing periods compared with the tightly controlled clinical trial setting [5].

Thus, the relatively low use of RWD in PAMs (41.5%) that we found may seem somehow unexpected. However, this likely reflects the greater emphasis placed on interventional-derived data due to the characteristics of the clinical trials submitted as evidence for the marketing authorisation of ATMPs, as previously highlighted. Additionally, RWD presents various challenges—particularly operational, technical and methodological—that may hinder its full integration into the post-authorisation decision-making process [29]. Engaging in early dialogue with regulators is essential to ensuring that all stakeholders are aligned and that efforts to incorporate RWD are recognised and valued by regulators during decision-making [9, 29].

Most RWD-PAMs were actually imposed by regulators, reflecting their interest in RWD to ensure the benefit–risk balance post-approval, particularly for long-term efficacy and safety in larger populations, as outlined in a specific guideline for ATMP follow-up management [9]. Our results also highlight that RWD-PAMs had larger populations and longer mean duration compared with the overall average results for the PAMs investigated.

A survey conducted amongst various stakeholders, including regulators, to assess their perception of registry-based RWD use shows that long-term effects are the second most relevant aspect for which these data could be particularly valuable, surpassed only by disease epidemiology [30]. Registries are an essential data source for RWD, yet their potential in ATMPs’ post-authorisation surveillance may be underexplored, as only 16 PAMs reported using such a source. Whilst this may not be a case of outright underutilisation, it highlights significant opportunities for broader application, particularly for continuous evaluation of ATMPs’ benefit–risk balances. This limited uptake could be attributed not only to concerns about data quality, protection and sharing but also to a lack of effective dialogue between stakeholders [31]. A similar trend was observed by Bouvy et al., whose study identified 31 registry-based post-authorisation studies for only 30 (7.6%) medicinal products out of a total sample of 392 that received a positive opinion from CHMP between 2005 and 2013 [32, 33]. Likewise, Flynn et al. found low use of registry-derived data in imposed post-authorisation studies, with only 14 of 158 new marketing authorisation applications (8.9%) and 1 of 153 extensions of indication (0.7%) incorporating such studies between 2018 and 2019 [28].

However, the numbers reported in the above-mentioned studies may be even overestimated due to a lack of conceptual harmonisation. For instance, at the time of publication of both studies, product registries were also perceived as patient registries [28, 32]. In fact, the findings of those studies contributed to the development of the EMA guideline on registry-based studies, published in 2022. This guideline provides a clear definition of patient registries and explains why product registries are not considered as such in the regulatory context. It is considered that ‘recruitment and follow-up of these patients with the aim to evaluate the use, safety, effectiveness or another outcome of this exposure typically falls outside of normal routine follow-up of patients and therefore corresponds to a clinical trial or non-interventional study in the targeted population’, with these terms being preferred over ‘product registry’ [16]. This framework was applied in the present study, uncovering that in eight ATMP post-authorisation measures, the term ‘registry’ was incorrectly used. These eight measures corresponded to seven different ATMPs which were mostly authorised before 2022, that is, before the guideline’s publication, with Carvykti and Upstaza being authorised that year. This timing could explain the misuse of the term, as the protocols were referring to common non-interventional studies. Thus, the publication of guidelines (including the one issued by the EMA) and other initiatives may help harmonise understanding amongst stakeholders, reducing instances of misclassification when referring to registries in the future.

In addition to concept harmonisation, improving data discoverability and ensuring transparency are also crucial for promoting the use of registries [34, 35]. Amongst the small sample of registries identified, only the WFH GTR was listed in the HMA-EMA RWD Catalogues [36]. This database contributes to transparency and discoverability, particularly for RWD sources, including registries, in alignment with the Findable, Accessible, Interoperable and Reusable (FAIR) data principles [13]. Therefore, efforts could be made to index more registries, especially those from institutions already registered, such as the ATHN, EBMT and CIBMTR [37–39]. To support this, registry holders should proactively index their datasets to enhance visibility amongst MAHs and other stakeholders. In parallel, these catalogues should be actively monitored by the responsible managing institutions to identify new registration opportunities, fostering a coordinated effort amongst all involved parties. Beyond this, the EMA also conducts a qualification procedure for registries. Whilst the absence of qualification does not imply a lack of data quality, it serves to define the regulatory purposes for which the EMA considers qualified registries suitable [13]. The EBMT registry, the only EMA-qualified registry in the studied sample, is recognised by regulators as suitable for use in drug utilisation research, post-authorisation efficacy studies (PAES) and post-authorisation safety studies (PASS) [40]. In October 2023, a letter of support was issued for the WFH GTR, as full qualification was not possible. Assessors determined that qualification would depend on the registry’s ‘demonstrated ability to collect and report data within the context of a study, which has yet to be established’. The ongoing qualification procedure concerns the use of the WFH GTR for PAES and PASS purposes [41].

Efforts have been made to include patient experience in real-world evidence research [42]. Specifically, this can be achieved through the capture of PROs, which reflect patients’ direct experiences with treatment [43]. A review of publicly available PASS protocols, regardless of whether they included RWD (2012–2015) registered in the EU-PAS register—a predecessor of the EMA RWD Catalogues—found that only 14% included PRO data, primarily assessing disease burden, symptoms and quality of life [44]. The low use of PROs observed in that study is also evident in our findings, possibly due to the incomplete implementation of PROMs in RWD sources, which limits the routine collection of PROs. This may be attributed to the additional burden of data collection and the associated costs, particularly in clinical settings [45–47]. Integrating PROMs effectively into registries could enhance PRO assessment in real-world settings [13, 48]. Regulators also recognise the importance of this, as seen in their criticism of the EBMT registry for not collecting quality-of-life data in the EMA qualification opinion of the registry in question [40]. Whilst disease-specific measures seem to be more relevant for inclusion in registries, generic measures offer broader applicability, making this an important area for future research [35, 49].

This study is subject to certain limitations. The reliance on publicly available information may have resulted in missing some details. Additionally, for RWD-PAMs, only 12 protocols were accessible, limiting the depth of further analysis that would have been ideal. However, the approach of combining multiple information sources may have mitigated this information bias. The inability to analyse final results or reports from PAMs prevented the identification of protocol deviations, particularly for RWD-PAMs. Nevertheless, given the average duration of studies containing RWD (11.8 years) and the fact that the earliest authorised ATMPs in our sample were approved in 2013, it would not be feasible to have many final reports available. Concretely, only six RWD studies had a finalised status, all related to ATMPs authorised in 2018 or earlier. Furthermore, although compassionate use programs could have provided valuable insights, they were excluded due to challenges in systematically identifying relevant information. However, with only three such programs identified, the impact on the overall results might be minimal. Lastly, two ATMPs—Casgevy and Beqvez—were recently authorised, limiting the availability of information in the databases searched and requiring us to rely primarily on their respective EPARs. Despite these limitations, our study has various strengths. It includes all PAM categories, both imposed (annex II conditions and specific obligations) and voluntary (category 3 studies of the RMP), ensuring a broad and comprehensive assessment. Additionally, the use of multiple data sources alongside EPARs provides a more detailed analysis of ATMPs’ PAMs.

Overall, regulators recognise the critical importance of assessing these highly innovative products in real-world conditions to maintain a balanced benefit–risk profile, complementing evidence from controlled clinical trials. The tailored legal framework and specific guidance have been positively received by stakeholders, as these instruments address key challenges effectively [45]. As a result, the reform of the European Union’s pharmaceutical legislation reflects this advanced understanding [50], potentially shaping a new era for advanced therapies within the healthcare ecosystem.

Conclusions

This study emphasises the potential role of RWD in the post-authorisation monitoring of ATMPs in the European Union, particularly in a landscape marked by uncertainty at the time of approval. PAMs appear to reflect the orphan and PRIME framework, where regulatory flexibility shifts the burden to the post-authorisation phase, requiring additional clinical-trial-based evidence to address benefit–risk gaps. However, regulators also recognise the importance of real-world experience with ATMPs by imposing RWD-PAMs, acknowledging the limitations of single-arm, non-randomised, small-scale trials used in decision-making at the time of approval. Registries are emerging as a key source of RWD for medicine decision-making. Establishing early stakeholder engagement and reliable data sources, addressing practical challenges and ensuring transparency are essential to fully harnessing RWD’s potential. Strengthening these aspects can provide valuable insights to reduce uncertainties and support regulatory decision-making.

Supplementary Information

Below is the link to the electronic supplementary material.

Declarations

Funding

Open access funding provided by FCT|FCCN (b-on). This project has received funding from the European Union’s Horizon Europe Research and Innovation Actions under grant no. 101095479 (More-EUROPA). Views and opinions expressed are, however, those of the author(s) only and do not necessarily reflect those of the European Union nor the granting authority. Neither the European Union nor the granting authority can be held responsible for them.

Conflict of interest

D.M. is employed by Infosaúde-Instituto de Formação e Inovação em Saúde, but all research was conducted at Faculdade de Farmácia, Universidade de Lisboa.

Availability of data and material

The data analysed during this study are included in this published article and its supplementary information file.

Ethics approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Code availability

Not applicable.

Author contributions

All authors contributed to this research according to the International Committee of Medical Journal Editors (ICMJE) criteria to qualify as a listed author. B.S. and C.T. contributed to the conceptualisation and formulation of the research question. D.A., D.M. and C.T. performed data collection. D.A. drafted the manuscript. All authors read and approved the final version of the manuscript.

Disclaimer

The views expressed in this article are the personal views of the author(s) and may not be understood or quoted as being made on behalf of or reflecting the position of the regulatory agency/agencies or organisations with which the author(s) is/are employed/affiliated.