Developing Protocols & Procedures for CT Data Integrity
In vitro-in-vivo correlation
1. On Modeling Methods and Predictability of
In-Vitro-In-Vivo Correlation (IVIVC)
of Oral Controlled Release Products
Presented at BIOBIO 2010, Hyderabad, India, March 1-3, 2010
Dr. Bhaswat S. Chakraborty
Sr. Vice President, R&D, Cadila Pharmaceuticals
2. Outline
• Relevance and definition of IVIVC
• Biopharmaceutics classification system (BCS)
• Levels of IVIVC
• Generation of in-vitro release profile
• In-vivo PK profile
• Generation of in-vivo release profile
– Compartmental
– Linear Systems
– Other Methods
• Predictability Error
• Issues
• Conclusion
3. Dissolution in CR Formulation Development
For Market
Retig et al. Diss Tech, Feb. 2008, 6-8
4. Definition of IVIVC
• In-vitro-in-vivo correlations (IVIVC) are the
predictive, mathematical models relating an in vitro
property such as dissolution and an in vivo response,
e.g., amount of drug absorbed, thus allowing an
evaluation of the QC specifications, change in
process, site, formulation and application for a
biowaiver etc.
• Valid in vitro and in vivo methods valid
IVIVC
6. • Level A – point-point; first
Levels of IVIVC
deconvolution to get in vivo
%drug absorbed, then Level B
compare with %dissolved
• Level B – Statistical
moments; MRT or MDT in
vivo vs. MDT in vitro
• Level C – single point; PK
parameter vs. %dissolved
Level AC
Level
Level A
Malinowski and Marroum, Encyclopedia of Contr. Drug Deliv.
7. Overall Approach
IVIVC
Scale factor
API –
1 Physicochemi- BCS Class PK Data IVIVR
cal Properties
2 Dosage Form Biorelevent
Properties Dissolution
3 Computer Modeling Using Convolution including Transporters, PK Models,
and PK Parameters, API properties or Drug Release Data
Wang et al (2009) Diss Tech, 8, 6-12
8. Generation of In-Vitro Release Profile
• USP apparatus 1 (basket, 100 rpm) or 2
(paddle, 50&75 rpm)
• Aqueous dissolution medium, 900 ml
– pH 1-1.5, 4-4.5, 6-6.5 & 7-7.5 at 370C
– A surfactant may be required
• In-vitro food effect
– Rotating dialysis cell method
– Effects of oils, enzymes and pH
9. Dissolution Specifications
• Without IVIVC
– ± 10% of the label claim from mean dissolution profile of the bio or
clinical batch
– Can be >10% but range not >25% in certain cases
• With IVIVC
– All batches should have dissolution profiles with upper and lower
predicted bioequivalence
• Proper or Biorelevant Dissolution conditions
– Consider medium, volume, duration, apparatus (hydrodynamics)
– pH 1 – 7.4
– Predictive of bioavailability
• Similar conditions, similar dissolution and similar bioavailability
10. Mean Doxazocin Concentrations from
CR Formulations; n = 24
8mg fed SD
8mg fasted SD
2mg fasted SD
Chung et al. Br J Clin Pharmacol. (1999) 48, 678–687.
11. Oral CR of Diltiazem with a Clinically Proven
IR (first market entry)
Steady State
Single Dose Fasting
Malinowski and Marroum, Encyclopedia of Contr. Drug Deliv.
12. Limits to Oral Drug Absorption
Rate-limiting Conditions Comments
Steps
Dissolution limiting Tdiss > 199 min The absolute amount of
absorbed drug increases with
Peff > 2 × 10-4 cm/sec
the increased dose.
Dabs >> Dose
Permeability Tdiss < 50 min The absolute amount of
limiting absorbed drug increases with
Peff < 2 × 10-4 cm/sec
the increased dose.
Dabs >> Dose
Solubility Tdiss < 50 min The absolute amount of
limiting absorbed drug does not
Peff > 2 × 10-4 cm/sec
increase with the increased
Dabs < Dose dose.
(Yu, Pharm. Res. 16:1884-1888 (1999))
13. Generation of In-Vivo Release Profile
• Compartmental Models
– Wagner-Nelson
– Loo-Riegelman
• Linear Systems Models
– Deconvolution
– Convolution
• Mathematically they all yield the same result
16. Convolution
Where,
C(t) = Plasma drug concentrations after oral dose
Cδ(t) = Plasma concentrations after an IV dose or a dose of oral solution
Upon taking the derivative of C(t) wrt time:
When Cδ(0) = 0
18. Other Methods of Generating
In-Vivo Release Profiles
• Macroscopic Mass Balance
• Where An is absorption number and Cb* is the lumen drug concentration
• Inverse Gaussian
Where MIT is mean input time and CV2I is a normalized variance
22. IVIVC Model Predictability
(Weak acid; highly lipophilic; bioavailability 60-70%)
Lobenberg R. www.aapspharmaceutica.com/meetings/files/126/lobenberg.pdf
23. IVIVC Bench Issues
• Reliable and biorelevant dissolution method and apparatus
suitability
– Qualification and calibration of equipment, sink conditions
– Ability to discriminate non-BE lots
– Apparatus and media for continuous IVIVC (minimum 3 lots) and
tuning with gi conditions
• Accurate deconvolution of the plasma concentration-time
profile
– e.g., %absorbed in-vivo may be reflective of processes other than
release; absorption rate limitation is common for CR products
• Dissolution Specifications
– Based on biological findings rather than pharmacopeial or mechanistic
24. IVIVC Modeling Issues
• Intra- and Inter-subject variation
– High variations can distort the mean data and in turn the
deconvolution
– Enterohepatic recycling or second peak
– Reproducibility of reference profiles
• Modeling
– Smoothness of input and response functions
– Stability of numerical methods
– Jumps in input rate functions, e.g., delayed release or
gastric emptying
– Statistical properties of the models
25. Conclusions
• Biorelevant and reliable dissolution profiles can predict the in-vivo
absorption of drugs from CR formulations
• Batches with similar dissolution will be BE and dissimilar dissolution will
be non-BE
• Several methods exist for estimating in-vivo absorption
– Mainly mass balance (compartmental) and superposition (convolution)
• Level A (point-to-point) or B (mean dissolution times) correlation can be
obtained for BCS class 1 or 2 drugs
• At least 3 lots (desirable, fast and slow) must be established with IVIVC
and proper reference
• IVIVC is useful in
– QBD, SUPAC and biowaivers…
• Both practical and modeling issues must be addressed