Clinical trials for authorised biosimilars in the European Union: a

Session
Regulatory aspects and
review of EMA authorized biosimilars
Session “Biosimilar Development”, 15 May
Johanna Mielke, Bernd Jilma, Franz Koenig, Byron Jones
Franz König
Medical University of Vienna
Section of Medical Statistics
[email protected]
www.meduniwien.ac.at
http://www.ideas-itn.eu/
Acknowledgements & Disclaimer
This project has received funding by the Swiss State Secretariat for Education, Research and
Innovation (SERI) under contract number 999754557 (Novartis: JM, BJo) and from the European
Unions Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant
agreement No 633567 (MUW: FK, BJi).
The opinions expressed and arguments employed herein do not necessarily reflect the official views of
the Swiss Government and should not be linked to any affiliations the authors are associated with.
http://www.ideas-itn.eu/
2
Introduction: BIOSIMILAR
“A biosimilar medicine is a biological
medicine that is developed to be similar
to an existing biological medicine
(the ‘reference medicine’).
[...]
When approved, its variability
and any differences between it and
its reference medicine will have
been shown not to affect safety or
effectiveness.”
Source: Christian Schneider, Chair EMA Biosimilar Working Party:
http://www.ema.europa.eu/docs/en_GB/document_library/Medicine_QA/2009/12/WC500020062.pdf
3
Questions
• Which type of clinical trials have to be undertaken for getting
approval in Europe?
• How much and in which way do the development programs
differ?
• Is there a unified approach for biosimilars with the same
active substance?
• How close are the studies in the submitted applications
following the recommendations in the EMA guidelines?
• How is multiplicity addressed when planning and analyzing
biosimilar trials?
4
Main Data Source
European Public Assessment Reports (EPAR)
• Available online at
http://www.ema.europa.eu
• Detailed information about the
application:
– Drug: active substance, indications, ...
– Non-clinical: toxicology, ...
– Clinical development program:
study design, endpoints, sample size, ...
• EMA leading agency with 28 approved biosimilars
on 11 different biologics (April 2017)
• Review conducted in June 2016 (Mielke et al., 2016)
5
Methods: EPARs analysed
• Review based on 21 approved biosimilars
on 7 different biologics (June 2016)
• Focused on approved biosimilars
(no refused, no withdrawn products)
• Sponsors collaborated & submitted identical clinical trials,
but products were marketed separately
– Example: biosimilar to active substance epoetin zeta
Silapo (Stada Arzneimittel AG) - Retacrit (Hospira UK Ltd)
 13 different applications
6
Methods
Comparison of the submitted applications in terms of
• Sample size/Number of clinical trials
• Trial design
• Endpoints
• Statistical models
• Equivalence margins
• Approved indications, extrapolation to other indications
7
Overview of results
First Biosimilar in 2006
8
Sample size PK/PD vs. Phase III
• PK/PD trials
– 1-5 trials
– 24-269 subjects
• Phase III
Number of PK/PD trials
– 1-3 trials
– 120-1295 subjects
Number phase III trials
9
Sample size PK/PD vs. Phase III
Epoetin
Alfa/Zeta
10
Filgrastim
Infliximab
Follitropin
Alfa
Others
Sample size PK/PD vs. Phase III
Epoetin
Alfa/Zeta
11
Filgrastim
Infliximab
Follitropin
Alfa
Others
Trial design
PK/PD
• Guidelines: 2x2 crossover, mostly followed
• Exceptions:
– Remsima/Inflectra
(parallel groups design, but was allowed in product specific guideline)
– Epoetin Alfa Hexal/Abseamed/Binocrit (pivotal PK/PD is a parallel groups design,
contradicts product specific guideline)
Phase III
• Parallel groups design recommended, followed
• Exceptions
– Zarzio/Filgrastim Hexal and Grastofil/Accofil
(single arm design, but accepted in product specific guideline)
12
Endpoints, equivalence margins &
statistical models: PK
• Metrics for bioequivalence testing: AUC, Cmax
– Approach: calculation of geometric mean ratio with confidence intervals, if
confidence intervals fully lie within pre-specified limits
 bioequivalence
• Recommendation in guideline for PK studies:
– Equivalence margins: 80-125%
– 90% confidence intervals
• Mostly followed, exceptions:
– Silapo/Retacrit: wider equivalence range for Cmax
– Ovaleap, Benepali: no details given in EPAR and publication, unclear if formal
testing was done
13
Endpoints, equivalence margins &
statistical models: PK
• If criteria are not fully fulfilled, approval is possible: Example
Zarzio/Filgrastim Hexal:
– PK/PD-studies in five different doses, for lower doses and after multiple
subcutaneous doses: AUC and Cmax not within limits
– Sponsor claimed “differences in level of purity”  adjustment to doses
– Nonetheless, for three settings outside of equivalence region
– Sponsor provided modelling results and explanations for mechanism of action 
approval
14
Endpoints, equivalence margins &
statistical models: PK
Grastofil/Accofil: Study KWI-300-101
15
Endpoints, equivalence margins &
statistical models: PK
“The location and the width of the confidence interval
should also be taken into account in the interpretation of
similarity.
For example, statistically significant differences in 90%
CIs within the justified acceptance range regarding
relevant PK parameters would need to be explained and
justified as not to preclude biosimilarity.
On the other hand, if the 90% CI crosses the prespecified
boundaries the applicant would need to explain such
difference and explore root causes.”
Source: http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2015/01/WC500180219.pdf
16
Endpoints, equivalence margins &
statistical models: PD
• 3 applications without any PD assessment
• Endpoints are drug specific
• Equivalence margins: 80-125% or narrower
– Example Epoetin Alfa Hexal/Abseamed/Binocrit:
“the acceptance range was then changed to 97–103% by protocol
amendment based on haemoglobin (Hb) concentration changes defined
as equivalence margins in Phase III studies”
• Similarly to PK, approval is possible if not all endpoints
meet the target
17
Endpoints, equivalence margins &
statistical models: Phase III
• Endpoint, margins, statistical models are disease specific
• Margins: no explicit explanation for choice n in publications and
EPAR
• Mainly one primary endpoint
• Only few exceptions (with 2 and 4 primary endpoints)
– No multiplicity adjustment in EPARs mentioned
18
Endpoints, equivalence margins &
statistical models: Phase III
19
Endpoints, equivalence margins &
statistical models: Phase III
• Endpoint, margins, statistical models are disease specific
• Variation within a substance:
Ovaleap and Bemfola (follitropin alfa)
– Same endpoint used (number oocytes retrieved)
– Slightly different equivalence margins
– Statistical methods
– Ovaleap: Zero-inflated Poisson (ZIP) regression model
– Bemfola: Mann-Whitney TOST or Schuirmann’s TOST (data dependent)
 Flexibility for the sponsors how to analyse the data
20
Indication & Extrapolation
• Phase III in one indication only,
but approval in all indications of the reference product
 Extrapolation
• Only in 2 out of 13 applications, the approved indications are
not essential identical to the one of the reference product
• Example for not all indications granted: Silapo/Retacrit
– “Reduction of allogeneic blood transfusions in adult non-iron deficient patients
prior to major elective orthopaedic surgery” was not granted
– Reason:
lack of shown equivalence for the subcutaneous administration route in Phase III
21
Indication & Extrapolation
Example of successful extrapolation: Remsima/Inflectra
• One Phase III trial in patients with active rheumatoid arthritis
• According to the EPAR, the sponsor provided:
– A literature review of the therapeutic indication and also of the mechanism of
action of the drug
– Preliminary data on 23 patients with either Crohn's disease or ulcerative colitis
– Additional post-marketing trials were promised
• Licensed for all indications of the reference product
– (rheumatoid arthritis, ankylosing spondylitis, psoriatic arthritis, psoriasis, adult
and paediatric Crohn's disease and adult and paediatric ulcerative colitis)
22
Multiplicity
• Multiple hypothesis tests
– several doses, routes of administration (IV, SC), different populations, endpoints
(e.g.,Cmax, AUC), …
• If ALL test must be significant, NO adjustment needed
• BUT – not always all tests have been statistically significant.
• What about defining the success criteria differently?
”success”: k out of m tests are statistically significant
• What is the impact on sample sizes needed?
23
PSI Conference 2017
For k out of m:
which multiple testing strategy?
• Bonferroni (might be too conservative, but controls FWER)
• No adjustment (inflation of FWER for k<m)
• Or aim at controlling a different error rate instead of FWER:
the k-FWER
• k-FWER = P(reject at least k true null hypothesis,
irrespectively which and how many are true null)
• k-FWER adjustment:
𝛼∗
=
𝑘∙𝛼
𝑚
Lehmann & Romano (2005), Hommel & Hoffmann (1998), Victor (1982)
24
Results
25
PSI Conference 2017
Results
26
PSI Conference 2017
Results
27
PSI Conference 2017
Results
28
PSI Conference 2017
Results
29
PSI Conference 2017
EPARs
• Very helpful to evaluate regulatory standards
• Even more helpful if reports were more standardized
– Length (25p – 105p)
– Depth of information differs
– Unified structure (e.g., including overview tables on PK/PD trials)
– All crucial information should be included (e.g., explanation for choice of margins)
• EPARs should be linked to other EMA transparency initiatives
– E.g., EudraCT should be mentioned for all trials mentioned in EPAR
30
PSI Conference 2017
Conclusions
• High variability between submitted trials
• High variety also within an active substance
 case by case decision of the regulators
• Recommendation in product specific guidelines and
overarching guidelines were mostly followed, but also
exceptions
• Align sample sizes for multiple tests and success criteria
• It is possible to gain approval even though not all pre-specified
primary endpoints meet the target
31
Business Use Only
References
• Mielke, J., Jilma, B., Koenig, F., & Jones, B. (2016). Clinical
trials for authorized biosimilars in the European Union: a
systematic review. British Journal of Clinical
Pharmacology, 82(6), 1444-1457.
OPEN ACCESS: http://dx.doi.org/10.1111/bcp.13076
• Mielke, J., Jones, B., Jilma, B., & Koenig, F. (2017). Sample
Size for multiple hypothesis testing in biosimilar development.
Submitted.
32
This project was supported by the Swiss State Secretariat for
Education, Research and Innovation (SERI) under contract number
999754557. The opinions expressed and arguments employed
herein do not necessarily reflect the official views of the Swiss
Government.
The project is part of the IDEAS European training network
(http://www.ideas-itn.eu/) from the European Union’s Horizon 2020
research and innovation programme under the Marie SklodowskaCurie grant agreement No 633567.
Thank you