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 Most products remain authorised (n = 19; 76.0%), whilst 5 (20.0%) had their marketing authorisations withdrawn (Provenge (2015), Zalmoxis (2019), Zynteglo (2022), Skysona (2021) and Alofisel (2024)). All withdrawn products resulted from the request of the respective MAHs due to commercial reasons, except for Alofisel, for which the clinical benefit was no longer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).

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).

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).

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.

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.

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.

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).