The document discusses the importance of selecting sensitive patient populations for clinical trials of biosimilar monoclonal antibodies to properly assess clinical similarity to the reference product. It provides examples of indications and populations that are more sensitive for detecting differences based on effect size, such as rheumatoid arthritis patients for assessing ACR20 response to anti-TNF antibodies. The document cautions that clinical trials must be designed appropriately to demonstrate equivalence or non-inferiority and should obtain data on relevant endpoints in sensitive populations as well as long-term safety and immunogenicity follow-up to justify extrapolation to other indications. It critiques examples of biosimilar clinical trials that failed to use a sensitive population or design.

1. Biosimilars – The Clinical Development Approach
The Selection of Sensitive Patient Populations as
a one of the Prerequisites for the Extrapolation of
Clinical Data of Biosimilar Monoclonal Antibodies
Dr. Thomas Schreitmueller, Head Regulatory Policy, Biologics
F. Hoffmann – La Roche Ltd., Basel, Switzerland

2. Based on science, the Concept of Biosimilarity is
built on five indispensible pillars:
Pharmacovigilance
Proper Quality System
Clinical Similarity
Pre-clinical Similarity
Analytical Similarity
Biosimilarity
Science
The use of existing copies of biotherapeutic products that
have not gone through an adequate development program
is not recommended due to potential safety implications.

3. 3

4. Biosimilar pathways – EMA biosimilar antibody guideline
• The guideline is setting the stage for the overall stepwise development
approach having the goal “…ensuring that the previously proven
safety and efficacy of the drug is conserved.”.
• The stepwise approach at the clinical side is outlined more clearly
focusing on the main principles to be considered when establishing
clinical similarity: “The guiding principle is to demonstrate similar
clinical efficacy and safety compared to the reference medicinal
product, not patient benefit per se, which has already been shown
for the reference medicinal product.”.
• This has to be achieved by planning all studies “…with the intention
to detect any potential differences between biosimilar and
reference medicinal product and to determine the relevance of
such differences, should they occur.”.
Guideline on Similar Biological Medicinal Products Containing Monoclonal Antibodies. Non-clinical and Clinical Issues.
EMA/CHMP/BMWP/403543/2010
4

5. Biosimilar vs. innovator clinical studies in oncology.
Differences in requirements and study designs
Aspects of Development Biosimilar
Innovator
Patient Populations
Sensitive and
homogeneous
Any
Clinical Designs
Comparative versus
innovator, normally
equivalence
Superiority vs. standard of
care (SoC*)
Study Endpoints
Clinically validated PD
markers if not available
CR, ORR for example
Clinical outcomes data or
accepted/established
surrogates (e.g. OS)
Safety
Similar safety profile to
innovator
Acceptable risk/benefit
profile versus SoC*
Immunogenicity
Similar immunogenicity
profile to innovator
Acceptable risk/benefit
profile versus SoC*
Extrapolation
Possible if justified
Not allowed
5
* In some cases SoC may not exist

6. What is a sensitive and homogeneous population and
endpoints?
• The idea is to study the biosimilar in the population of patients in
whom – if there is a difference between biosimilar and reference
product – that difference will most easily be detected
– for example, we have a treatment that works in 60%
of patients. If we were able to identify who are the
“responder” patients, then we would target treating just
those patients
• Activity endpoints with a large effect size may be
considered as PFS, DFS and OS may not be suitable
– CR, ORR (also measured at a certain timepoint),
percentage change in tumour mass from baseline, or
pathological Complete Response (pCR) in certain clinical
settings

7. Overall Response Rate is not a sensitive endpoint in
Follicular Lymphoma patients treated with R-CHOP
100
By what fraction of
MabThera’s effect
size can the
biosimilar treatment
effect differ and still
be considered
clinically similar?
Responders (%)
80
50% ?
25% ?
15% ?
60
Mabthera
40
6%
Control
20
0
ORR (NHLCHOP)
ACR 20
Therefore, if the difference in ORR
responses between Mabthera and biosimilar
is statistically significantly less than 1.5%,
the biosimilar is within the comparability
margin
If 25% of the effect size is chosen
Then the comparability margin =
0.25 x 6%= 1.5%
Sample size = 4,000
per group
7

8. ACR20 is a sensitive endpoint in AR patients treated with
MabThera (TNF IR)
100
By what fraction of
MabThera’s effect
size can the
biosimilar treatment
effect differ and still
be considered
clinically similar?
Responders (%)
80
50% ?
25% ?
15% ?
60
Mabthera
40
Control
33%
20
0
ORR (NHL)
ACR 20( RA)
Therefore if difference in ACR20 between
Mabthera and the biosimilar is statistically
signicantly less than 10%, the biosimilar is
within the comparability margin
If 30 % of the effect size is chosen
Then the comparability margin =
0.30 x 33%= 10 %
Sample size = 250 per group
8

9. Case study adalimumab:
Immunogenicity of therapeutic Mabs
Page 9
R. Niebecker et. al Current Drug Safety, 2010, 5, 275-286 275
Safety of Therapeutic Monoclonal Antibodies

10. Case study adalimumab:
The dependency of Immunogenicity on Co-medications
Methotrexate reduces immunogenicity in adalimumab treated
rheumatoid arthritis patients in a dose dependent manner
DCharlotte L Krieckaert Ann Rheum Dis published online May 14, 2012
Page 10
ownloaded from ard.bmj.com on July 12, 2012 - Published by group.bmj.com

11. When is extrapolation justified?
• The biosimilar development needs to manage the risk
associated with extrapolation of clinical data to
indications not practically studied during the similarity
assessment which means:
– A step wise approach with clinical trials assessing the
different clinical parameters in the most sensitive
population is the basis.
– The mode of action has to be the same in the
indication to be extrapolated
– The risk for immunogenicity in different patient
populations has to be assessed critically
Wrong patient selection leads to wrong clinical similarity
conclusion. This would mean a high risk for extrapolation

12. Case study trastuzumab:
What is the right patient population to establish
similarity to a reference product?
Topic
PK
Metastatic Population
 Affected by patient’s health status &
tumour burden
Neoadjuvant/Adjuvant population
 Homogeneous population could
be selected

Variability is also observed
 Healthy Volunteers
PD
 Clinically validated PD marker not available
Clinical
efficacy/safety

•Population with heterogeneous
characteristics affecting final clinical
outcome.

•Populations less likely to be
confounded by baseline
characteristics and external factors.
Immunogenicity
?
?
12

13. Case study trastuzumab:
What is the most sensitive indication/patient population
to establish similarity in immunogenicity
Trastuzumab treatment regimens are
different in different patient populations

14. Case study trastuzumab:
HannaH Phase III Study
IV Herceptin®
pCR
Clinical stage Ic to
IIIc including IBC
Neoadjuvant
Trastuzumab SC 600
mg/5 mL q3w (fixed
dose)
Trastuzumab IV
6 mg/kg q3w
(8 mg/kg loading dose)
Adjuvant
Docetaxel
75 mg/m2
Objective:
Show non-inferiority of SC vs. IV based on co-primary endpoints
 PK: observed trastuzumab Ctrough pre-dose Cycle 8
 Efficacy: pathological complete response (pCR) in the breast
FEC, 5-fluorouracil, epirubicin and cyclophosphamide. IBC, inflammatory breast cancer
FEC
500/75/500
Follow-up: 24 mo
R
1:1
SURGERY
HER2positive
EBC
(N=596)
18 cycles / 1 year
SC Herceptin®

15. Case study trastuzumab:
Sensitivity of the neoadjuvant-adjuvant setting to
detect differences in immunogenicity
• Data: HannaH (BO22227) is a pivotal Phase III trial (N~600) to compare
Herceptin SC (subcutaneous administration by syringe) to Herceptin IV
(Herceptin SC EU approval Sept 2013)
• Observed ADA rates (anti-drug antibody against Herceptin)*:
– Herceptin IV:
7.1% (21/295)
– Herceptin SC: 14.6% (43/295)
• Sensitive setting: Difference between drugs (formulations) could be
found if there is one
• No correlation of ADA to efficacy/safety/PK was detected for
Herceptin
*Definition: ADA rates (all patients who tested positive for ADAs at least once post-baseline)

16. ADA and Nabs rates during trial
Detection time point
Percentage ADA positive
patients*
Only positive on treatment
53.1%
(34/64)
Only positive on treatmentfree follow up
37.5%
(24/64)
Positive on treatment and in
treatment-free follow up
9.4%
(6/64)
*Data pooled (similar rates in categories)
•37.5% of ADA-positive patients tested positive only in the treatmentfree phase
•Positive Nabs (neutralizing antibodies): Only three patients (1 IV arm
and 2 SC arm) and all were in the treatment-free phase.

17. Case study trastuzumab:
Key conclusions on extrapolation of immunogenicity
data
• Immunogenicity of a biosimilar trastuzumab candidate has to be
thoroughly investigated and characterized in the most sensitive
setting prior to approval.
• The adjuvant setting is considered to be sensitive and allows
the inclusion of data from a treatment-free follow-up phase which
is crucial for the comprehensive characterization of the
immune response of trastuzumab.
• Therefore extrapolation of immunogenicity data obtained in the
EBC setting to MBC is possible while extrapolation of immunogenicity data from MBC to the EBC population represents a
major risk if no safety and efficacy data are available.
EBC = Early Breast Cancer; MBC = Metastatic Breast Cancer

18. Case studies rituximab: Sensitive Populations for Efficacy
Indications approved for
rituximab
ORR Control
ORR Active
Effect Size
Reference
NHL follicular Induction
(CHOP)
90%
96%
6%
SPC (GLSG)
Hiddemann
NHL follicular Induction
(CVP)
10 %
41%
31%
SPC (CR)
NHL follicular relapsed
(CHOP)
74%
87%
13%
SPC
NHL DLBCL Induction
76%
84%
8%
SPC (CR)
CLL
72 %
86 %
14%
SPC
Rheumatoid Arthritis (TNF-IR)
18%
51%
33%
SPC (ACR20)

19. Reality: Intas/Mabtas Phase I/III trial on 100 DLBCL patients
using R-CHOP regimen for 2 cycles
(Previously approved Rituximab copy)
From our perspective the clinical study is inadequate to demonstrate
clinical bio-similarity between Rituximab-RBP and this product as:
• The clinical trial population mixes two types of populations which have
different clinical outcomes (i.e., diffuse large B cell lymphoma and
follicular lymphoma)
• ORR may not be considered a sensitive endpoint for diffuse large B
cell lymphoma nor for follicular lymphoma using CHOP chemotherapy
(GELA LNH-985, updated: Feugler et al, JCO 2005; Hiddeman et al,
Blood 2005; Marcus, Blood 2005)
• The study is severely underpowered to demonstrate equivalence of
rituximab-RBP with the copied product (the study description doesn't
mention if this is an equivalence, non-inferiority, or other type of design)
• Two cycles of therapy are not enough to demonstrate efficacy
(RECIST guideline 1.1, Eisenhauer, EJC, 2009) nor for safety.

20. Reality: Probiomed “Evaluation of Clinical Behaviour”
in NHL large B-cell (NHLCGB) CD20 +patients using
R-CHOP regimen

21. Summary
• MAb products, have and will provide essential and safe
treatment opportunities for many diseases.
• The application of proper risk mitigation strategies during the
development and marketing of biosimilar MAb’s is fundamental.
• Comparative clinical testing is a key part of these strategies and
has to be done in the relevant setting(s) most sensitive to detect
potential differences in safety, efficacy and immunogenicity.
• Unfortunately the concept of sensitive populations in the context
of the clinical development of biosimilar MAb’s often is not yet
well understood by manufacturers.
• Taking into consideration these strategies will not only minimize
the risk for the patient, only those strategies will actually make
the development of true biosimilar MAb’s feasible.
•

22. Thank You !
22