Novartis Oncology 4x3 Template

Role of M&S in biosimilar development Finding the needle in the haystack Mohamed Elmeliegy Novartis Oncology Clinical Pharmacology

Outline  What is biosimilarity?  Impact of dose on biosimilarity assessment  Role of M&S in biosimilar development  Case study: A Population PK/PD model for a biosimilar candidate

2 | AAPS - NBC - Biosimilars | Mohamed Elmeliegy | 05.18.2016

What is Biosimilarity? Finding the differences (needle) in the haystack (the similarities)  Proving biosimilarity is about finding and explaining analytical and biological differences between the biosimilar and the originator  Differences between biosimilar and the originator are inevitable. However, these differences should be comparable to the batch-tobatch variability of the originator

 Given the high & in-depth sensitivity of in vitro chemical and biological characterization of binding, CDC, & ADCC, biosimilars advancing to clinical testing are analytically comparable. Clinical trials (both phase I & III) are merely confirmatory to resolve residual uncertainty and fulfill regulatory requirement  A clinical trial should be sensitive to detect these tough to detect residual uncertainty

3 | AAPS - NBC - Biosimilars | Mohamed Elmeliegy | 05.18.2016

Impact of dose on biosimilarity assessment  Large Phase III clinical studies typically conducted at high therapeutic doses saturating the receptors/ligand by several folds are less likely to show these small differences (i.e. find the needle)  Data from recent phase III trials are predominantly positive: • ABP 501 biosimilar candidate vs adalimumab in RA - ACR50: ABP 501 vs adalimumab (49.2 percent vs. 52.0 percent) and ACR70 - Difference mean change from baseline in DAS28-CRP between ABP 501 and adalimumab was – 0.01 at week 24

• ABP 215 biosimilar candidate vs bevacizumab in NSCLC

 In Phase I trials to demonstrate BE, studying multiple doses (often low enough and non-therapeutic) is feasible  Phase I trials looking at multiple relevant biomarkers that could be used as surrogates for hard-to-find biological differences may be more clinically meaningful than large phase III trials

4 | AAPS - NBC - Biosimilars | Mohamed Elmeliegy | 05.18.2016

Role of M&S in biosimilar development  FDA draft Clin Pharm guidance highlighted the utility of modeling and simulation tools to design PK and/or PD studies to:  Select optimally informative doses for evaluating PD similarity

 Evaluate PK/PD parameters (e.g. Emax and EC50) for similarity

 PK/PD modeling and simulation can support studies where clinical pharmacology evaluation is a major source of information to assess clinically meaningful differences and residual uncertainties between biosimilar and reference product. Ultimately, this may allow moving from large phase III trials towards more PK/PD-targeted development program. The timeline and budget implications are significant. 5 | AAPS - NBC - Biosimilars | Mohamed Elmeliegy | 05.18.2016

Study Design Dense PK sampling up to 84 days with concurrent PD collections

BSC OGN-1 OGN-2

6 | AAPS - NBC - Biosimilars | Mohamed Elmeliegy | 05.18.2016

A biosimilar candidate with similar PK/PD properties to the originator (OGN) BSC

OGN-1

OGN-2

Drug concentration

Conc. vs time

Biomarker 1 vs time

Biomarker 2 vs time

Biomarker 1

Biomarker 2

BSC

7 | AAPS - NBC - Biosimilars | Mohamed Elmeliegy | 05.18.2016

OGN-1

OGN-2

A semi-mechanistic population PK/PD model to support biosimilarity of BSC to OGN  A population PK/PD model was used to characterize the exposure-response relationship of BSC, OGN

𝑘𝑠𝑠𝑠−𝐵𝐵𝐵

𝑘𝑠𝑠𝑠−𝐵𝐵𝐵

+

BM1

+

BM2

𝑘𝑑𝑑𝑑−𝐵𝐵𝐵

𝑘𝑑𝑑𝑑−𝐵𝐵𝐵

 Drug concentrations assumed to stimulate BM1 synthesis via Emax and EC50 parameters  The turnover model allows explaining the delay observed between max PK conc. and peak PD response (hysteresis)  BM1 concentrations assumed to stimulate BM2 synthesis via Emax and EC50 parameters

8 | AAPS - NBC - Biosimilars | Mohamed Elmeliegy | 05.18.2016

BSC, OGN-1, OGN-2 induce similar BM1-induction pattern BSC A 53

OGN1 B 53

OGN2 C 53

25450 (26)

27705 (29)

25896 (30)

27.65 (14)

28.22 (15)

28.34 (31)

Kdeg (1/day)

0.55 (4)

0.54 (4)

0.55 (5)

Emax

17.08 (7)

17.64 (7)

17.14 (9)

EC50 (ug/mL)

29.94 (14)

29.2 (18)

30.28 (16)

1.59 (15)

1.62 (17)

1.59 (14)

Mean (CV%) PK Parameters Treatment N

Similar BM1 exposure Similar total BM1 synthesis (formation) and degradation rates. Indicates similar BM1 baseline across three treatment groups. Similar maximum induction Similar sensitivity for induction Similar steepness of the drug vs BM1 relation

9 | AAPS - NBC - Biosimilars | Mohamed Elmeliegy | 05.18.2016

AUEClast (pg*day/mL) Ksyn (pg/mL)

Gamma

BM1 exposure is significantly correlated to drug exposure  Subjects with high drug exposure tend to have high BM1 stimulation persistent depletion of BM1 and thus higher stimulation of a feedback loop that produces more BM1

 The model can be applied to simulate PK/PD relationship at alternative, higher clinically approved doses

BM1 AUEC (pg*hr/mL)

 This might be due to more

Drug AUC (μg*hr/mL)

10 | AAPS - NBC - Biosimilars | Mohamed Elmeliegy | 05.18.2016

Acknowledgements  Elizabeth Potocka  Didier Renard: PMX I&D  Johann-Dan Laurent  Shefali Kakar  Jerry Nedelman  Jason Liao  Yulan Li  Debra Barker  Miguel Izquierdo

11 | AAPS - NBC - Biosimilars | Mohamed Elmeliegy | 05.18.2016