e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 1 ( 2 0 0 7 ) 249–261
`
`a v a i l a b l e a t w w w . s c i e n c e d i r e c t . c o m
`
`j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / e j p s
`
`Conference report
`When poor solubility becomes an issue:
`From early stage to proof of concept夽
`S. Stegemann a, F. Leveiller b, D. Franchi c, H. de Jong d, H. Lind´en e,∗
`
`a Capsugel, Rijksweg 11, 2880 Bornem, Belgium
`b AstraZeneca, 22187 Lund, Sweden
`c GSK, Via A. Fleming 4, 37135 Verona, Italy
`d Servier, 14 rue de Bezons, 92415 Courbevoie, France
`e EUFEPS, Walingatan 26A, 11181 Stockholm, Sweden
`
`a r t i c l e
`
`i n f o
`
`a b s t r a c t
`
`Article history:
`Received 16 May 2007
`Accepted 16 May 2007
`Published on line 21 May 2007
`
`Keywords:
`Drug solubility
`Bioenhancement
`Formulation
`Poor water solubility
`
`Drug absorption, sufficient and reproducible bioavailability and/or pharmacokinetic profile
`in humans are recognized today as one of the major challenges in oral delivery of new drug
`substances. The issue arose especially when drug discovery and medicinal chemistry moved
`from wet chemistry to combinatorial chemistry and high throughput screening in the mid-
`1990s. Taking into account the drug product development times of 8–12 years, the apparent
`R&D productivity gap as determined by the number of products in late stage clinical devel-
`opment today, is the result of the drug discovery and formulation development in the late
`1990s, which were the early and enthusiastic times of the combinatorial chemistry and high
`throughput screening. In parallel to implementation of these new technologies, tremen-
`dous knowledge has been accumulated on biological factors like transporters, metabolizing
`enzymes and efflux systems as well as on the physicochemical characteristics of the drug
`substances like crystal structures and salt formation impacting oral bioavailability. Research
`tools and technologies have been, are and will be developed to assess the impact of these
`factors on drug absorption for the new chemical entities.
`The conference focused specifically on the impact of compounds with poor solubility on
`analytical evaluation, prediction of oral absorption, substance selection, material and for-
`mulation strategies and development. The existing tools and technologies, their potential
`utilization throughout the drug development process and the directions for further research
`to overcome existing gaps and influence these drug characteristics were discussed in
`detail.
`
`© 2007 Elsevier B.V. All rights reserved.
`
`夽 The publication is the summary of the EUFEPS Meeting prepared by the scientific committee: “When poor solubility becomes an issue:
`from early stage to proof of concept.” which took place in GlaxoSmithKline research facility in Verona (Italy) on 26–27 April 2006.
`∗
`Corresponding author. Tel.: +46 8 723 5086; fax: +46 8 411 3217.
`E-mail address: hans.linden@eufeps.org (H. Lind ´en).
`0928-0987/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
`doi:10.1016/j.ejps.2007.05.110
`
`MYLAN EXHIBIT 1011
`
`

`

`e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 1 ( 2 0 0 7 ) 249–261
`
`250
`
`Contents
`
`250
`Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`1.
`251
`Physical/chemical properties of the drug substance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`2.
`252
`Biopharmaceutical evaluation of the drug substance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`3.
`254
`4. Drug candidate selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`255
`5.
`Formulation strategies for solubility and bioenhancement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`258
`6.
`Biopharmaceutical evaluation of drug delivery systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`7.
`Future outlook. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
`8.
`Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`259
`Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`260
`References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`260
`
`1.
`
`Introduction
`
`With the introduction of combinatorial chemistry and high
`throughput screening, the properties of new chemical enti-
`ties shifted towards higher molecular weight and increasing
`lipophilicity that results in decreasing aqueous solubility
`(Lipinski et al., 1997; Lipinski, 2000).
`Solubility in different solvents is an intrinsic material char-
`acteristic for a defined molecule. To achieve a pharmacological
`activity, the molecules must in general exhibit certain sol-
`ubility in physiological intestinal fluids to be present in the
`dissolved state at the site of absorption. The aqueous solubil-
`ity is a major indicator for the solubility in the intestinal fluids
`and its potential contribution to bioavailability issues.
`During drug substance development the molecules are
`screened in nanomolar receptor assays. The molecules with
`the best receptor binding are selected for further pre-clinical
`studies, for which more drug substance is synthesized.
`Already at this stage, solubility is of critical importance,
`because solubility estimates for the poorly defined drug sub-
`stance serve its pharmacological and toxicological profiling.
`When going first into humans, sufficient and well charac-
`terized solubility becomes even more critical. From now on
`the solubility or dissolution of the dose ranges in the vari-
`ous biophysiological media to which the drug substance or
`formulated drug substance will be exposed is expected to be
`reproducible and remain unchanged for the final development
`and eventually marketing.
`It is well accepted today throughout the scientific com-
`munity that drug substance solubility and especially aqueous
`drug substance solubility is an issue for the drug discovery as
`well as the early and late stage pharmaceutical development
`
`process and therefore needs to be addressed very early on,
`during compound design and optimization.
`The solubility or dissolution of the drug substance can be
`mainly altered on two levels, through material engineering of
`the drug substance or through formulation approaches. What-
`ever route is taken to enhance or modify the solubility and/or
`dissolution of a lead substance, it needs to be scalable to a
`commercially viable process later on in the development.
`Besides the aqueous solubility of a drug substance, its per-
`meability is a second critical aspect for oral bioavailability.
`The Biopharmaceutical Classification System (BCS) was intro-
`duced in the mid-1990s to classify the drug substances with
`respect to their aqueous solubility and membrane permeabil-
`ity (Amidon et al., 1995). Drug substances, for which solubility
`enhancement can improve the oral bioavailability, are sub-
`stances that are classified in class 2 (poor soluble/permeable)
`and class 4 (poor soluble/poor permeable). Especially for class
`2 substances, solubility enhancement is part of the strategies
`to improve the oral bioavailability.
`The BCS classification takes into account the required dose
`since low dosed drugs will sufficiently dissolve in the intesti-
`nal fluids of the GI tract to be absorbed, while higher doses
`of drugs with similar aqueous solubility will not. To gener-
`ally describe “solubility” the Pharmacopoeia (USP) uses seven
`different solubility expressions as shown in Table 1. The Euro-
`pean Pharmacopoeia uses similar solubility definitions except
`the ‘practically insoluble’ characteristic, which is not specified
`(European Pharmacopoeia 5.0).
`When looking at the product launches between 1995 and
`2002, out of 100 substances 14 were considered class I, 12 were
`classified as class II, 28 were class III and 46 were class IV
`substances (Mehta, 2002). If a drug substance exhibits a poor
`
`Table 1 – Solubility definition in the USP
`Description forms
`Parts of solvent required
`(solubility definition)
`for one part of solute
`
`Solubility range
`(mg/ml)
`
`Solubility assigned
`(mg/ml)
`
`Very soluble (VS)
`Freely soluble (FS)
`Soluble
`Sparingly soluble (SPS)
`Slightly soluble (SS)
`Very slightly soluble (VSS)
`Practically insoluble (PI)
`
`<1
`From 1 to 10
`From 10 to 30
`From 30 to 100
`From 100 to 1000
`From 1000 to 10,000
`>10.000
`
`>1000
`100–1000
`33–100
`10–33
`1–10
`0.1–1
`<0.1
`
`1000
`100
`33
`10
`1
`0.1
`0.01
`
`

`

`e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 1 ( 2 0 0 7 ) 249–261
`
`251
`
`aqueous solubility, the product development will have to focus
`on the investigation of various other drug substance character-
`istics like its physicochemical, biopharmaceutical properties
`and the targeted dose to identify the potential impact of the
`solubility on the further product development. Today, about
`35–40% of the lead substances are known to have an aqueous
`solubility of less than 10 ␮M or 5 mg/ml at pH 7 and it is not
`expected that this figure will change in the future.
`
`Physical/chemical properties of the drug
`2.
`substance
`
`Solubility is an intrinsic material property that can only be
`influenced by chemical modification of the molecule as such,
`like salt complex or prodrug formation. In contrast to this,
`dissolution is an extrinsic material property that can be influ-
`enced by various chemical, physical or crystallographic means
`like complexation, particle size, surface properties, solid state
`modification or solubilization enhancing formulation strate-
`gies.
`Solubility is one of the key physicochemical parameters of
`a new molecule that needs to be assessed and understood
`very early on in drug discovery and drug candidate selec-
`tion process. Starting with the first amount of the substance
`synthesized, commonly a few milligrams, the first series of
`analytical tests include that of the equilibrium solubility of
`the substance in a given solvent, as well as the apparent or
`dynamic solubility, which reflects the concentration of the
`substance in solution under certain conditions. The value of
`equilibrium solubility is often limited by the test duration,
`which is normally between 4 and 24 h. The solubility and
`apparent solubility of the substance in an aqueous system are
`dependent on several factors (Table 2).
`Rather than single point determinations, a solubility pro-
`file of the substance is required to identify potential issues
`for drug precipitation in vivo. Especially the pH profiling for
`weakly basic salts is of critical importance as their solubility
`will vary in the intestinal pH (typically pH 1–8) and pre-
`cipitation may occur. The pH profiling should be performed
`in different bio-relevant buffer systems to mimic the high
`concentrations of the most common counter ions of phar-
`maceutical salts in GI fluids. This screen allows detection
`of a potential counter ion exchange and formation of more
`stable/less soluble salts of a molecule that will lead to precip-
`itation in vivo. The pH profile additionally provides the basic
`guidance to choose among potential solubilization strategies.
`Conversion of the molecule to other salts or hydrates needs
`to be taken into account and evaluated in the solubility profil-
`
`Table 2 – Factors influencing solubility
`
`Composition of the aqueous media
`Temperature
`pH
`Solid state (amorphous, crystalline)
`Polymorph type
`Counter ions (salt formation)
`Ionic strength
`
`ing. Different salts of the substance can be formed dependent
`on the buffer systems as well as their ionic strength. It is gen-
`erally accepted that there should be a minimum difference
`of three units between the pKa value of the ionizable group
`and of the possible counter ion (Bowker, 2002). (Ampholytes
`display a more complex solubilization process (two step reac-
`tion) including micelle formation at high ionic strength. Buffer
`salts as well as hydrates might have different solubility char-
`acteristics.)
`Therefore different analytical methods have been devel-
`oped to assess the substance in solution or solid state
`(precipitates) (Giron and Grant, 2002). Analytical methods for
`the substance in pH-controlled solutions are LC coupled with
`UV and MS. For the solid state analysis, powder X-ray diffrac-
`tion, energy dispersive X-ray (EDX), Raman spectroscopy, IR,
`microscopy (polarized light microscopy (PLM), environmental
`scanning electron microscopy (ESEM)) and thermal analysis
`are used.
`The solubility profiling should include any other biorele-
`vant dissolution media like simulated gastric fluid (SGF) with
`and without enzymes, fasted state simulated intestinal fluid
`(FaSSIF) and fed state simulated intestinal fluid (FeSSIF) at pH
`5.0 and 6.5 or human gastric or intestinal fluids (HIF’s).
`In parallel to the analytical characterization of the initial
`material of the drug substance, substantial efforts are invested
`into understanding and optimization of the crystalline struc-
`ture and to identify a potential pseudo-thermodynamic stable
`form of the substance. These investigations are looking into
`the polymorphs, solvates and salts formed by the substance
`under various conditions to identify the most suitable material
`for dosage form development, scaling up and later manufac-
`turing.
`Polymorphs appear in a number of different structures as
`non-mixed polymorphs (free base or acid) or as mixed poly-
`morphs like salts, co-crystals (Vishweshwar et al., 2006), guest
`substances, hydrates or solvates (Bernstein, 2002; Brittain,
`1999).
`While the polymorph screening focused more on the num-
`ber of different polymorphs by crystallization in different
`solvents in the past, it now has shifted towards qualitative
`data on polymorph formation under thermodynamically con-
`trolled conditions to get more accurate data on the potential
`risk for polymorphic changes in the final dosage form under
`the expected storage conditions. These constant and con-
`trolled conditions include pressure, temperature, solvent and
`time.
`Once a polymorph is found and characterized, additional
`polymorph evaluation experiments should be performed to
`provide information on its kinetic stability.
`The evaluation starts with determining, e.g. if the sub-
`stance forms only one anhydrate or several anhydrates. Once
`different polymorphs have been identified it is important to
`characterize whether they are monotropically or enantiotrop-
`ically related. In case of an enantiotropic phase transition,
`the transition temperature as well as the temperature stabil-
`ity range must be very well characterized. Long-term stability
`is tested in slurry experiments for at least 2 months in dif-
`ferent non-solvate forming solvents and/or at temperatures
`in the stability range of the anhydrates. From this series of
`polymorph experiments the most thermodynamically stable
`
`

`

`252
`
`e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 1 ( 2 0 0 7 ) 249–261
`
`form is selected and will be further evaluated in competi-
`tive slurry experiments at room temperature with any other
`known anhydrates and solvates. Reversibility between the dif-
`ferent forms will be included in the evaluation in long-term
`slurry experiments and summarized using binary and ternary
`phase diagrams. As an example, formoterol fumarate has been
`found to form seven modifications in the polymorph screen.
`Running the polymorph experiments, the dihydrate form has
`been determined as the most stable solid form due to its favor-
`able capacity of forming efficient hydrogen bonds between the
`formoterol and the fumarate leading to a well packed crystal
`structure (Jarring et al., 2006).
`To improve physicochemical properties of the drug sub-
`stance with regard to manufacturing, isolation and long-term
`storage, as well as to improve solubility and/or dissolution
`properties of the drug substance, salt formation is tradition-
`ally preferred by medicinal chemists for weak bases or weak
`acids. Since only 20–30% of the new molecules form salts eas-
`ily, the 70–80% remain challenging (Serajuddin and Pudipeddi,
`2002).
`Salt formation is a two-step process in which proton trans-
`fer in solution has to take place followed by a crystallization
`step. Identification of a suitable solvent to achieve sufficient
`attractive forces of the substance salt as well as to overcome
`ion and molecule solvation is required. To achieve a ther-
`modynamically stable salt with sufficient aqueous solubility,
`it is important to understand the salt structure formed by
`a specific counter ion. Counter ions act as a template that
`interacts through non-covalent bonds with the molecule by
`their conformation and bonding capabilities, which directly
`influence the three-dimensional structure. Data bases on the
`possible salt structures of functional groups can provide some
`valuable predictions on the number of possible solvates and
`hydrates and their coordination number defined as the num-
`ber of non-covalent bonds formed by the ion (e.g. Cambridge
`Crystallographic Data Center). Selected salts of a molecule will
`be assessed in a salt screening following the same principle
`as polymorph screening to investigate the long term stability
`as well as its conversion to other, more stable salts and its
`precipitation in different aqueous and bio-relevant media.
`With the increasing knowledge about the implications of
`polymorphs and salts in drug discovery, automated tools
`are being developed to standardize and implement these
`experiments as a routine process in drug discovery and lead
`substance selection (York, 1999; Rohl, 2003).
`To overcome poor aqueous solubility or erratic bioavail-
`ability, chemical modification leading to a prodrug has
`successfully been used for several substances. The most com-
`mon used prodrug approach is the incorporation of a polar
`or ionizable moiety into the molecule. The incorporation of
`N-acyloxyalkyl moieties of different chain length leads to
`a reduced crystal lattice interaction and decreasing melting
`point with increasing number of methylene groups (Stella
`et al., 1998). In vivo studies in dogs with the N-acyloxyalkyl
`derivatives of phenytoin confirmed a higher bioavailability in
`the fed state that did not correlate with decreasing water sol-
`ubility (Stella et al., 1999). Prodrugs also might reduce the
`pre-systemic metabolism of the substance in the GI tract
`or the release of the drug substance itself by enzymatic
`cleavage of the prodrug moiety close to the site of drug
`
`absorption. Fosphenytoin is a phosphate prodrug of pheny-
`toin that is metabolized by phosphatases to release its active
`moiety phenytoin. In vivo studies in dogs and humans have
`demonstrated a better bioavailability after oral, intramuscular
`and intravenous administration. Additionally fosphenytoin
`has shown a better safety profile compared to its original
`form phenytoin sodium (Stella, 1996). Examples for success-
`ful phosphate prodrugs are fosphenytoin (CerebyxTM) and
`amprenavir phosphate (VX-175/GW 908). Another successful
`prodrug development is bortezomib, a boronic acid pro-
`drug (VelcadeTM) (Sanchez-Serrano, 2006). Current research
`is focusing on the development of new prodrug moieties
`like N-glycines, sulfenamides and cysteamines to improve
`bioavailability by better solubility or metabolic stability.
`Once the most favorable thermodynamically stable form
`of a lead substance is determined and its solubility profile is
`characterized, investigations into its biopharmaceutical char-
`acteristics using in vitro, in silico or in vivo trials in animals
`will provide the necessary information on the substances
`predicted in vivo performance and for the design of a drug
`delivery system.
`
`Biopharmaceutical evaluation of the
`3.
`drug substance
`
`The biopharmaceutical evaluation and prediction is another
`crucial part of the lead drug candidate selection process. It
`is well established today that drug substances that do not
`meet at least certain biopharmaceutical criteria are returned
`to the medicinal chemist for lead optimization (Clark and
`Grootenhuis, 2003).
`One of the most critical aspects of in vitro screening assays
`is the determination of the true concentration of the free drug
`in the assay. The concentration of free drug in the assay is
`normally calculated rather than measured. Unexpected low
`solubility and precipitation of the drug in the media due to
`the dilution sequence (e.g. dilution from an aqueous buffer
`solution or from DMSO solution) has been reported. When
`the calculated drug concentration in the assay is not reached,
`wrong conclusions are most likely to be drawn regarding effi-
`cacy, toxicity or permeability (Di and Kerns, 2004).
`The main in vitro tools used for the assessment of drug
`absorption and permeability are listed in Table 3.
`The CaCo-2 cell monolayers are well-established in vitro
`systems that are used in various stages of the drug develop-
`ment process to assess drug absorption and the underlying
`processes (Shah et al., 2006). CaCo-2 cell lines provide valuable
`information about the permeability and absorption potential
`
`Table 3 – In vitro systems to identify permeability and
`absorption
`
`CaCo-2 cell lines
`Madin-Darby canine kidney (MDCK) cell lines
`Parallel artificial membrane permeability assay (PAMPA)
`Immobilized artificial membrane chromatography (IAM)
`Advanced compartmental absorption and transit (ACAT)
`Lipophilicity (log D and log P)
`
`

`

`e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 1 ( 2 0 0 7 ) 249–261
`
`253
`
`of a drug substance. They can further be used in later devel-
`opment to provide information about the potential impact of
`metabolic or efflux systems on the drug substance and to clas-
`sify the drug according to the BCS.
`When using CaCo-2 cell monolayers to measure perme-
`ability of poorly soluble substances attention must be drawn
`to possible drug accumulation within the cells, binding to
`proteins or adhesion to plastic surfaces. To avoid variability
`beyond ±10% in the permeability data obtained with CaCo-2
`cell tests, standardization and calibration are critical.
`The parallel artificial membrane permeability assay
`(PAMPA) is an artificial, intrinsic permeability measurement
`that is used for drug lead substance selection and for pre-
`clinical studies in a high throughput manner. The PAMPA
`system is based on synthetic phospholipids and fatty acids
`that mimic physiological conditions. PAMPA is used as a low
`cost alternative to cell based systems for early ADME screening
`(Avdeef, 2005). High confidence in the predictions using PAMPA
`has been reported for drugs with good effective permeability
`−6 cm/s) and the poor effective permeability (<2.5
`−6 cm/s),
`(>10
`while the results for drugs in between are difficult to interpret.
`For efficient and rapid investigation into mechanisms of
`drug absorption, CaCo-2 cell monolayers are combined with
`the PAMPA assay (Kerns et al., 2004). Drugs that correlate in
`the comparison are mainly absorbed through passive diffu-
`sion. Substances, which show higher permeability in PAMPA
`than in a CaCo-2 model, are substrates for efflux or for reduced
`passive diffusion of acids under CaCo-2 pH gradient set-
`up. Substances with a higher CaCo-2 cell permeability when
`compared to the PAMPA system have underlying absorptive
`mechanisms like paracellular, active transport or increased
`passive diffusion in a CaCo-2 cell pH gradient.
`Immobilized artificial membranes (IAM) are solid phase
`models of liposome membranes for partitioning studies of
`drugs into membranes. IAM chromatography is a simple
`experimental tool that allows for large number screening
`tests in the drug discovery phase to identify the potential
`of a drug substance of being absorbed, independently from
`other factors involved in poor bioavailability like pre-systemic
`metabolism, efflux systems or transporters (Pidgeon et al.,
`1995).
`Poor prediction still persists for the drug absorption in the
`colon. Due to the reduced paracellular or carrier-mediated
`absorption, efflux mechanisms, the reduced surface area, the
`colonic solubility environment (pH 6–7), the reduced volume
`and mixing and the unknown chemical metabolism of the sub-
`stance along the GI tract, the contribution to drug absorption
`still remains unclear. However, poorly soluble substances and
`drugs from extended release formulations will reach the colon
`at a considerable concentration and stay there for a long time.
`Even if absorbed slowly, this may contribute substantially to
`absolute bioavailability.
`One of the first prediction models used a mathemati-
`cal approach to estimate the fraction dose absorbed (Oh
`et al., 1993). This mathematical calculation uses only four
`parameters to estimate the dose that can be absorbed: initial
`saturation, absorption number, dose number and dissolution
`number.
`For the predictions of human drug absorption today the
`specific absorption rate (SAR) or human absorbable dose (Dabs)
`
`model is frequently used. The SAR plots the absorption num-
`ber versus the dose number (Collins and Rose, 2004). The Dabs
`model is based on the permeability of a substance, its sol-
`ubility in simulated intestinal fluids, the surface area of the
`human GI tract defined as 800 cm2 and a GI transit time of
`3.3 h (Yu, 1999). Plotting the Dabs versus the expected thera-
`peutic dose already gives some hint whether drug absorption
`will be the rate-limiting step for a conventional dosage form.
`For example, pioglitazone, a weak base with a good solubility
`at low pH will not require an enabling formulation due to its
`solubilization in the stomach; neither would tadalafil, a non-
`ionizable substance over the physiological pH range and a low
`solubility in all relevant media, because of its low dose (20 mg).
`Another model proposed is using the maximum absorbable
`dose (MAD) concept taking into account a small intestinal
`transit time of 4.5 h, a small intestinal fluid volume of 250 ml,
`the substance solubility and an absorption constant for the
`drug substance (Johnson and Swindell, 1996). The MAD val-
`ues provide good predictions if the projected human dose can
`be absorbed. Evaluating a series of substances it can serve
`as a guiding tool for rank ordering of the potential lead drug
`candidates in a specific series of substances (Curatolo, 1998).
`Using the MAD model as a starting point, computational
`systems have been developed to simulate drug absorption in
`vivo (Johnson, 2003). Additional input parameters like water
`absorption from the GI tract and changing gastrointestinal
`permeability have been introduced to improve the simula-
`tions.
`Another computational system for the simulation of drug
`absorption is based on the advanced compartmental absorp-
`tion and transit (ACAT) model (Agoram et al., 2001). The ACAT
`model is an extended version of the compartmental absorp-
`tion and transit (CAT) model that uses seven small intestinal
`compartments for prediction. The ACAT takes into account
`the colon as another compartment for drug absorption, which
`cannot be neglected for poorly soluble and poorly permeable
`substances as well as for sustained release formulations. The
`selection of the dissolution media for each compartment is
`an important factor that affects the accuracy of the prediction
`in the ACAT model. Using FaSSIF media for example, provides
`better correlations than simple buffer systems at pH 6.5 for
`soluble substances, however, for poorly soluble substances
`under-predictions are more likely for pH 6.5 buffer, while over-
`predictions are the case for the FaSSIF media. The output of
`the ACAT model has been improved by including other param-
`eters on the drug substance like particle size/radius, density,
`diffusion coefficient, log D, pKa and molecular weight.
`The simulations for both systems are relatively good, but
`tend to be less accurate for the poorly soluble substances due
`to the difficulty in modeling the colonic drug absorption.
`The major limitations seen today in the in silico models
`concern the potential drug precipitation in the GI tract (e.g.
`weak bases, salts of poorly soluble substances), the impact
`of colonic drug absorption, especially of poorly soluble sub-
`stances when at high dose and the effect of the drug particle
`size on absorption. While further validation work is ongoing,
`the existing in silico tools already provide important informa-
`tion for lead candidate selection and/or optimization and can
`be developed further to improve the accuracy of the simulation
`(Kuentz et al., 2006).
`
`

`

`254
`
`e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 1 ( 2 0 0 7 ) 249–261
`
`Following the biopharmaceutical evaluation of the sub-
`stance’s solubility/dissolution behavior in aqueous or bio-
`relevant media, permeability, chemical and metabolic stability
`are assessed, in vitro. Thereafter, the absorption, food effect,
`first pass metabolism, PK profiles are finally quantified in
`different animal species (rats, dogs, monkeys) in a dose-
`escalating manner. These studies are designed to establish
`first in vivo oral exposure rate and dose linearity. For these
`studies the drug is administered as a solid, suspension or
`solution. These studies should also provide information on
`any dependency of absorption on particle size as well as on
`the upper limit of exposure, the clearance and potential food
`effects of a standard formulation.
`Later on, in particular when potential for extended release
`formulation development is considered, more in depth inves-
`tigations into the mechanism of drug absorption can be
`performed using, for instance, a regional perfusion model that
`can be applied to animals (e.g. dogs) or humans (Lennern ¨as
`et al., 1992). The regional perfusion model is also used to
`establish a correlation between the human permeability and
`the CaCo-2 cell permeability (Lennern ¨as et al., 1997) and
`provides important information for the design of the drug
`delivery system. To facilitate and improve the investigation
`into regional drug absorption and its underlying mechanisms,
`special capsules have been developed that permit a time- and
`position-controlled release of the drug substance or formula-
`tion in defined areas of the GI tract (Wilding et al., 2000)
`To get at least some data on a substance performance in
`humans before lead substance selection or entering into the
`clinical development, the European Medicines Agency (EMEA)
`recently accepted the concept of a single micro-dosing study
`in humans (EMEA, 2003). Micro-dosing studies allow one single
`dose of less than 1/100th of the calculated pharmacologi-
`cal dose and not more than 100 ␮g in total dose for small
`molecules. Even so there is only very limited experience of
`micro-dosing studies and their value in the drug development
`process so far. Micro-dosing is expected to become a useful
`tool in identifying substances with critical biopharmaceutical
`properties as well as improving significantly in silico predic-
`tions of drug absorption and deposition (Wilding and Bell,
`2005).
`While some tools provide highly valuable data for lead sub-
`stance selection for one substance, they might not for others.
`Therefore the series of different in vitro, in vivo and predictive
`methods and tools provide a good base for putting together a
`meaningful evaluation program that needs to be designed for
`each individual substance based on the information gathered
`during drug discovery and substance characterization.
`
`4.
`
`Drug candidate selection
`
`Multidisciplinary teams are becoming an essential part of
`the lead candidate selection process in the pharmaceutical
`industry. Their objective is to address early on the various
`characteristics of the drug substances from the chemical as
`well as the pharmacological, toxicological and biopharma-
`ceutical point of