The optimal use of biorelevant media & simple modelling for the prediction of in-vivo oral behaviour James Butler Product Development GlaxoSmithKline What level of complexity is needed for adequate prediction of in-vivo behaviour? The application of simple biopharmaceutics modelling to early oral formulation development Desired knowledge: – Are drug & formulation properties adequate for good exposure/low variability in early human studies? What drug particle size? Is a bio-enhancement (solubilisation) strategy needed? – If so, which approach to choose? Developing multiple formulations is inefficient – During dose escalation, above what dose is loss of linearity in AUC/dose likely? May be similar to the dose above which large solubility-related food effects occur. Available methods for the prediction of likely oral in-vivo performance in humans Animal PK data – Often trusted more than alternatives, but…. Needs justification (minimise animal use) Not always predictive of human High variability with a few replicates: difficult to interpret Outcomes may be dose dependant – Safety assessment (higher doses) and FTIH may need different formulation approaches Models and simple measurements – Simple: solubility & permeability data, BCS, Maximum Absorbable Dose – Complex: Commercially available software models such as Gastro-Plus Although the sensitivity analysis tool within Gastro-Plus is conceptually simpler Dissolution methods – Simple: standard pharmacopoeial methods optionally with biorelevant media – Complex: non-pharmacopoeial methods: dynamic media transfer methods, TNO-TIM-1, IFR-DGM, dissolution/cell-line combinations, etc Developing a modified classification system BCS – a regulatory tool – Conservative, efficacy and patient safety in mind When is there no bio-inequivalence risk? – Useful in late development and post-launch DCS – a developability tool – Aim: realistic, product development issues in mind What factors are likely to control the extent of oral absorption? – Permeability, solubility, dissolution rate – Useful in evaluating new drug candidates (early development) Predicted Peff in Humans cm/sec x10-4 Dose/solubility ratio 250 500 1000 5000 10000 Using jejunal solubility, typically [email protected]°C 10 I Good solubility and permeability IIa II (dissolution rate limited) Good permeability, poor solubility IIb (solubility limited) 1 III IV Good solubility, poor permeability Poor solubility and permeability 0.1 BCS/DCS plot with human jejunal permeability and aqueous dose solubility ratio as axes 100000 Dose/solubility ratio 250 500 1000 5000 10000 100000 Predicted Peff in Humans cm/sec x10-4 Typically in [email protected]°C SLAD 10 IIb (solubility limited) I/IIa/III (all potentially dissolution rate limited!) 1 Why? – dissolution rate is related to solubility, not dose/solubility ratio. Use a “target particle size” to express this risk IV Poor solubility and permeability 0.1 DCS plot with human jejunal permeability and aqueous dose solubility ratio as axes Some key features of the DCS Solubility limited absorbable dose (SLAD): – Assumes a 500ml volume available for drug dissolution. – Peff >1x10-4 cm/sec assumed to proportionally increase the effective volume available for dissolution of highly permeable drugs – Represents the dose above which absorption is solubility limited. i.e. beyond this: – linear exposure/dose response may be lost – Solubility related food effects are likely – Reducing particle size alone cannot achieve complete absorption Recommended particle size – Derived from “dissolution number” equation – Approach: set target dissolution number to 1, solve the equation for particle diameter, use this as the target x90. Dn concept from Oh et al, Pharm Res 1993 10 (2) 264-270 Solubility and permeability estimations Solubility – Use best estimate available of (fasted) intestinal solubility HIF solubility FaSSIF (several versions now reported in the literature!) Permeability – Correlate to obtain an estimated human jejunal permeability Fraction absorbed data (if permeability is low) Rat perfusion Cell line (CACO-2, MDCK) In-silico/part in-silico models What is the most significant solubility to estimate for early development? Early clinical studies are almost invariably performed in the fasted state. – Fasted state usually “worst case” for low solubility compounds Fasted state gastric residence times are relatively short and variable compared to the small intestine. Gastric solubility: – of greatest importance for poorly soluble weak bases. even for these, having adequate intestinal solubility offers insurance against PK variability. Therefore, in a simple oral absorption model with a single solubility input, fasted intestinal solubility is most relevant. 250 Predicted Peff in Humans cm/sec x10-4 D/S in [email protected]ºC 10 500 I 1000 5000 10000 100000 IIa IIb Neutral, max. dose 0.5mg, Fassif solubility 17μg/mL, Estimated permeability 0.9x10-4cm/sec BCS III, DCS III 1 Digoxin III Recommended max particle size: 10µm(X50), 32µm(X90), IV 80-90% fraction absorbed 0.1 100 micron particles only 39% relative bioavailability to oral solution 9 & 13 micron mean particle formulations both bio-equivalent to oral solution Predicted Peff in Humans cm/sec x10-4 250 D/S in [email protected]ºC Weak acid, Insoluble across most of physiological range, max. dose 250mg 10 500 I solubility 30μg/mL (pH Fassif 6.5), increases with pH 1000 5000 10000 100000 Mefenamic acid IIa IIb Estimated permeability 14x10-4cm/sec BCS II, DCS IIA, Recommended max particle size: 13µm(X50), 42µm(X90), 1 Borderline for being solubility limited III 0.1 Bioavailability is known to be highly IVsize/dissolution rate particle dependant Reported issues with variable efficacy of some products. Some commercial products show a food effect, others don’t Predicted Peff in Humans cm/sec x10-4 D/S in [email protected]ºC 250 500 1000 5000 10000 100000 Neutral, highly insoluble, max. dose 1000 mg 10 FaSSIFI solubility 19μg/mL IIa IIb Estimated permeability 8.7x10-4cm/sec Griseofulvin BCS II, DCS IIB, Highly solubility limited Recommended max particle size: 11µm(X50), 34µm(X90), 1 III Solid dispersion approach used for a commercial oral product (GRISIV PEG) with enhanced bioavailability Less than linear increase in exposure with increasing dose 0.1 With or without with the bioenhanced approach, sizable food effects are seen Solubility estimation - Is FaSSIF solubility a reasonable estimate of actual human intestinal solubility? +/- Standard deviation Reasonable estimate for most unionised drugs, but more variable for acids/bases Data from: Pharm Res 22 12 2005 2141-2151, Pharm Res 17 2000 183-189 & 891-894, Eur J. Pharm Sci 39 2010 15-22, Pharm Res 26 2009 1456-1466, Pharm Res 23 2006 1373-1381, Int. J Pharm 376 2009 7-12, Pharm Res 22 2005 2141– 2151, Int J. Pharm 336 2007 302-309, Drug Met Dispos 38 2010 1407-1410 J Pharm Sci 99 2010 4525-4534, Mol Pharm 7 2010 1498–1507, J Pharm Pharmacol 62 2010 1656-1668 250 D/S in [email protected]ºC 500 1000 5000 10000 100000 Predicted Peff in Humans cm/sec x10-4 Mefenamic acid Ibuprofen 10 I Halofantrine IIa Nitrendipine Carvedilol Aprepitant IIb Troglitazone Etravirine Paracetamol Itraconazole Fenofibrate Danazol 1 Atovaquone Amprenavir Megestrol Acetate HO-221 Digoxin III Furosemide IV 0.1 DCS plot for developability assessment: Approximate position for selected drugs - Size control (microns+) adequate for complete dissolution in-vivo Proven significant bioavailability advantage for: Wet-milled/nano-milled formulations, Solid dispersion Liquid filled capsule IIa /(I/III)- dissolution rate limited, IIb/IV- solubility limited IR oral drug products and BCS When are IVIVRs likely? – The established view (and a regulators view?): Provided dissolution is rapid BCS 1 – IVIVRs unlikely: gastric emptying controls oral absorption BCS 3 – IVIVRs unlikely as permeability controls oral absorption BCS 2– IVIVRs likely to be possible BCS 4 – IVIVRs possible on a case by case basis only In practice though…. Many BCS II drugs as IR products are difficult candidates for IVIVR either because: – Adequate solubility in the stomach or small intestine means little actual in-vivo sensitivity to dissolution rate Or: – Drug absorption in-vivo is sensitive to multiple interacting factors in addition to dissolution rate solubility, supersaturation, precipitation, boundary layer diffusion – The use of solubilising forms (e.g. salts) and formulations further complicates these interactions How rapidly the drug dissolves has an impact on saturatable processes – E.g. first pass metabolism, efflux, active uptake DCS: a better rationale for the appropriate use of IVIVRs for IR products? Dose/ intestinal solubility ratio DCS I Human jejunal permeability In-vitro sensitivity > in-vivo sensitivity DCS IIa* IVIVRs from simple in-vitro methods likely to be possible DCS IIb IVIVRs from simple in-vitro methods less likely, may need to use more complex in-vitro methods/ modelling DCS III DCS IV IVIVRs unlikely IVIVRs from simple in-vitro methods unlikely, but may be possible with complex in-vitro methods /modelling * Including low dose, low solubility (DCSI/III) drugs like digoxin Biopharmaceutical Classification System Developability Classification System Uses minimum solubility in the pH range 1-7.5 Uses an estimate of jejunal solubility (FaSSIF or HIF) Assumes 250ml available to dissolve the drug Assumes 500ml available to dissolve the drug Single class 2 box Additional IIa/ IIb classification to distinguish between dissolution rate and solubility limited drugs High/low permeability cut-off is 90% fraction absorbed (FA) (equivalent to human jejunal Peff of ~1.0 x10-4 cm/s) Various methods used for permeability or FA - Predicted human jejunal permeability, usually calibrated cell line, whole body radiolabelled study predicted from a from cell line measurement or from etc, but method must follow regulatory guidance in-silico estimate based upon structure and LogD Used to support biowaivers (Class I compounds) Used to assess developability, identify factors limiting oral absorption, aid formulation choice Complex versus simple In a much more cost-aware industry, there is a desire to minimise costs of early development. The development of drugs with poor solubility/ poor oral absorption is a particular challenge to “keep it simple in early development” thinking. Complex: multiple formulation approaches developed for potential human use, all of which are fully characterised by the best available invitro dissolution methods and animal in-vivo studies. Simple: single formulation developed for early human studies, essential characterisation only. Simple models (e.g. DCS, Gastro-Plus sensitivity analysis) used to identify the most appropriate formulation and particle size. – With biorelevant media solubility as key input data Conclusions The use of biorelevant media solubility, in conjunction with a simple classification system (DCS) to assess the developability of oral drugs is attractive as: – Key factors influencing drug absorption can be simply represented by a few key parameters (SLAD and a target particle size) and visualised by plotting position on a BCS-like 2D graph. Tools like DCS are potentially valuable in reducing resource pre-FTIH in rational formulation development. Improved estimates of key input data (e.g. intestinal solubility, permeability) are key to the reliable prospective prediction of oral absorption. Acknowledgements The former “Predictive Technologies” group at GSK, in particular, Paul Connolly & Richard Lloyd Prof. Jenny Dressman (Uni. of Frankfurt) Questions?