Slow Offset Compounds

One frequently encounters the case where the equilibrium dissociation constant (Kd, see above) is defined by microconstants with "fast" rates on and off the receptor. However, any change in potency in a chemical series (affinity) must represent an increase in the on (k+:) rate or a decrease in the off rate (k_a). Occasionally, either by accident or design, the off rate is altered dramatically enough to redefine the receptor kinetics of the compound such that the rates influence the actual pharmacodynam ic


Fig. 2.13 Structures of salbutamol and salmeterol, rapid and slow offset ß2-adrenoceptor agonists.

Fig. 2.13 Structures of salbutamol and salmeterol, rapid and slow offset ß2-adrenoceptor agonists.

ics of the compound. These compounds are termed "slow offset" and their pharmacodynamic action exceeds that which would be predicted from the duration of the plasma concentrations. Often such compounds are detected during in vitro studies by increasing affinity or potency with time of incubation or persistence of activity following removal of drug by "wash out". A number of explanations for this phenomena have been advanced. Extra-receptor binding attempts to explain the slow offset of a compound by invoking a binding site removed from the actual active site domain. This site could be either protein or lipid. Salmeterol (a p2 adrenoceptor agonist) represents an agent designed in this manner [10]. The lipophilic side chain interacts with an exosite and markedly improves duration against compounds such as salbutamol (Figure 2.13).

The exosite appears to be located at the interface of the cytoplasm and the transmembrane domain of the p2-adrenergic receptor [10]. The structures of salbutamol and salmeterol are clearly different, although it is obvious both are based on the "adrenalin" pharmacophore. More subtle changes in structure leading to "slow-offset" can only be rationalized by changes in intra-receptor binding. Possibilities for such increases can include simply increased interaction per se and resultant affinity, with an effect largely confined to changes in the off rate. Thus, telenzepine is more potent than pirenzepine as well as showing slow offset from the receptor [11]. This increase in affinity may simply reflect the increased lipophilicity of the telenzepine head group (Figure 2.14).

It is possible to achieve slow offset without a change in potency. Here conformational restriction may be the mechanism. If one assumes a number of binding functions in a molecule, and that for stable binding all have to interact, then probability suggests that in a flexible molecule, association and disassociation will be occurring rapidly (fast on, fast off). With a molecule whose confirmation is restricted to one favourable to the interactions, it is likely that the rate of association and dissociation will be markedly lower (slow on, slow off). Such restrictions may be very simple molecular changes, for instance a single methyl group converts the fast offset compound carfentanil [12] to the slow offset compound lofentanil (Figure 2.15).


/ ! Fig. 2.14 Structures of pirenzepine and its more CH3 potent, slow offset Ml antimuscarinic analogue telenzepine telenzepine.



Fig. 2.15 Structures of opioid agonists carfentanil and its slow offset analogue lofentanil.



carfentanil lofentanil

Fig. 2.15 Structures of opioid agonists carfentanil and its slow offset analogue lofentanil.


Factors Governing Unbound Drug Concentration

We thus have in many cases only two parameters defining drug activity at steady state, receptor affinity and free (unbound) plasma concentration. Occasionally actual persistence at the receptor needs to be taken into account. In some cases, particularly hydrophilic drugs, there is a permeation factor that needs to be defined. The concept of steady state allows simplification of the equations and concepts of pharmacokinetics. Steady state in the context here implies a drug dosed at specific times so that the concentrations between administered doses are effectively an exact image of previous doses and that the difference between the peak and trough levels are small. The factors governing the steady state free plasma concentration Cpf for an oral drug, are the dosing rate (dose size x frequency), the fraction of the dose absorbed (F) through the g.i. tract and the free drug clearance (Clu) as shown below:

at steady state: rate in = rate out

True steady state is usually only achieved for a prolonged period with intravenous infusion. If we assume that we wish for a similar steady value after oral administration, then we need to balance our dosing frequency with the rate of decline of drug concentration and the rule of thumb referred to earlier (dosing interval equal to drug half-life) can be applied. Unbound clearance and free drug are particularly applicable to drugs delivered by the oral route. For a well-absorbed compound the free plasma concentrations directly relate to Cliu (intrinsic unbound clearance).

This simplifies greatly the concepts of first-pass hepatic metabolism and systemic clearance referred to previously. Most importantly Cliu is directly evolved from the enzyme kinetic parameters, Vmax and Km:

Cliu Vmax/Km

When the drug concentrations are below the Km, Cliu is essentially independent of drug concentration. The processes of drug metabolism are similar to other enzymatic processes. For instance most oxidative processes (cytochrome P450) obey Michaelis-Menten kinetics:

where v is the rate of the reaction, Vmax the maximum rate, Km the affinity constant (concentration at 50 % Vmax) and s the substrate concentration. Substrate concentration (s) is equal to or has a direct relationship to Cp^. In many cases Cpf (or s) are below the Km value of the enzyme system. However, in some cases (particularly the higher affinity P450s such as CYP2D6, see Chapter 7), Cpf (or s) can exceed the Km and the rate of metabolism therefore approaches the maximum (Vmax). As such the kinetics move from first order to zero order and the elimination of the drug is capacity limited. The term saturation kinetics is applied. Under these conditions

and clearance depends on drug concentration.

These values are obtained from in vitro enzyme experiments. From the previous relationship between in vitro pharmacology measurements and free drug concentrations and those outlined here, it is reasonable to assume that clinical dose size can be calculated from simple in vitro measurements.

It is easiest to understand how clearance relates to the rate of decline of drug concentration (half-life) if we consider the model depicted in Figure 2.9. When a dose (D) is administered intravenously then the initial free concentration achieved in plasma Cp(fo) is dependant on the volume of extracellular or total body water minus plasma water and the amount of drug bound to tissues and proteins.

Free volume is calculated by equations analogous to those for total drug (see Eq. 2.1).

in which Vd(f )is an apparent volume not only including the actual fluid the drug is dissolved in but also including the drug bound to tissues and protein as if it was an aqueous compartment in direct equilibrium with the free drug. Thus the greater the amount of drug bound, the greater the apparent free volume. The clearance and volume of distribution of unbound drug are related by the equation

where fcd is the elimination rate constant. Note that this equation and others are essentially the same as those for total drug except that free (unbound) drug values are substituted for total drug values. Free volume and free clearance are always equal to or greater than the values calculated from total drug. Moreover increases in plasma protein binding increase free volume but decrease total volume.

These concepts lead to two important observations. Protein binding or tissue binding is not important in daily dose size. The daily dose size is determined by the required free (unbound) concentration of drug required for efficacy. Protein binding or tissue binding is important in the actual dosage regimen (frequency). The greater the binding the lower and more sustained the free drug concentrations are. Thus a drug with four-fold higher binding and hence free volume than another, with the same unbound (free) clearance, will have a four-fold longer half-life. This could result in a dosage regimen of 20 mg once a day compared to 5 mg four times a day, both giving rise to broadly similar profiles and fluctuations around the average steady state concentration.


1 Haller M, Akbulut C, Brechtelsbauer H, Fett W, Briegel J, Finsterer U. Peter K, Life Sci. 1993, 53, 1597-1604.

2 Burns E, Triger DR, Tucker GT, Bax NDS, Clin. Sci. 1991, 80, 155-160.

3 Wong WW, Sheng HP, Morkeberg JC, Kosanovich JL, Clarke LL, Klein PD, Am. J. Clin. Nutr. 1989, 50, 1290-1294.

4 Brans YW, Kazzi NJ, Andrew DS, Schwartz CA, Carey KD, Biol. Neonate 1990, 58, 137-144.

5 Gibaldi M, McNamara PJ, Eur. J. Clin. Pharmacol. 1978, 13, 373-378.

6 Stopher DA, Beresford AP, Macrae PV, Humphrey MJ,J. Cardiovasc. Pharmacol. 1988, 12, S55-S59.

7 Edgar B, Regardh CG, Johnsson G, Johansson L, Lundborg P, Loftberg I,

8 Smith DA, Rasmussen HS, Stopher DA, Walker DK, Xenobiotica 1992, 22, 709-719.

9 Thummel KE, Kunze KL, Shen DD, Adv. Drug Delivery Rev. 1997, 27, 99-127.

10 Green SA, Spasoff AP, Coleman RA, Johnson M, Liggett SB, J. Biol. Chem. 1996, 271, 24029-24035.

11 Schudt C, Auriga C, Kinder B, Birdsall NJM, Eur. J. Pharmacol. 1988, 145, 87-90.

12 Leysen JE, Gommeren W, Drug Dev. Res. 1986, 8, 119-131.

Pharmacokinetics and Metabolism in Drug Design 35 Edited by D. A. Smith, H. van de Waterbeemd, D. K. Walker, R. Mannhold, H. Kubinyi, H. Timmerman I

Copyright © 2001 Wiley-VCH Verlag GmbH ISBNs: 3-527-30197-6 (Hardcover); 3-527-60021-3 (Electronic)



AUC Area under the curve of a concentration time profile

Caco-2 Human colon adenocarcinoma cell line used as absorption model g. i. Gastrointestinal

MDCK Madin-Darby Canine Kidney cell line used as absorption model



Percentage of dose absorbed as measured in portal vein


MedChem/Biobyte log P estimation program

F %

Percentage of dose bioavailable


Fraction absorbed

1 non

Fraction non-ionised at pH 6.5


Intestinal fluid volume (250 ml)


Absorption rate constant in rats (min-1)

log D

Logarithm of distribution coefficient

log P

Logarithm of partition coefficient

log s

Logarithm of solubility in water


Average residence time in the small intestine (270 min)


Solubility in phosphate buffer at pH 6.5


Intrinsic solubility of the neutral species at 37 °C


Volume of the lumenal contents


Dose administered

The Absorption Process

The oral absorption of a drug is dependent on the compound dissolving in the aqueous contents of the gastrointestinal tract (dissolution) and then traversing the actual barrier of the gastrointestinal tract to reach the blood (Figure 3.1).

Fig. 3.1 Schematic simplified view of the absorption process.

Fig. 3.1 Schematic simplified view of the absorption process.

For a number of reasons membrane transfer may be limited (see Figure 3.2) and therefore absorption incomplete. In this chapter these processes will be discussed.

Fig. 3.2 Mechanisms of membrane permeation [1]. The total percentage of the dose absorbed may be the result of a combination of several of these processes.


Dissolution depends on the surface area of the dissolving solid and the solubility of the drug at the surface of the dissolving solid. Considering these factors separately surface area is manipulated by the processing and formulation of the compound. Milling and micronization convert the drug into smaller particles with consequently greater surface area. In actual clinical use the compaction of the particles into tablets is offset by formulation with disintegrants. Certain formulations use a co-solvent such as polyethylene glycol (PEG) which is an organic solvent with water miscible properties.

Solubility is manipulated mainly by the structure of the drug. Broadly, solubility is inversely proportional to the number and type of lipophilic functions within the molecule and the tightness of the crystal packing of the molecule. Yalkowski [2] has produced a general solubility (log S) equation, for organic non-electrolytes. The equation incorporates the entropy of melting (ASm) and melting point (m. p. in °C) as a measure of crystal packing and log P as a measure of lipophilicity.

log S = (ASm (m.p. - 25)/1364) - log P + 0.80 (3.1)

This equation can be further simplified to log S = - log P - 0.01 m.p. + 1.2 (3.2)

It can be seen from the above that increases in either crystal packing or lipophilicity will decrease solubility.

The rate of dissolution is effected by solubility as is the actual concentration of drug in the bulk of the solution (aqueous contents of gastrointestinal tract). The concentration of drug in solution is the driving force of the membrane transfer of drug

3.3 Membrane Transfer | 37

into the body and low aqueous solubility often continues to present itself as a problem even after formulation improvements.

A number of drugs have very low aqueous solubility, mainly due to very high lipophilicity, but also due to lack of ionizable centres, and also the tight crystal packing referred to above. These drugs are erratically and incompletely absorbed due to this inability to dissolve in the gastrointestinal tract following oral administration. Examples of low solubility, dissolution limited drugs include danazole, griseofulvin, halofantrine, ketoconzaole, nitrofurantoin, phenytoin and triamterene [3,4]. Poor dissolution is responsible for both intra- and inter-patient variability in drug absorption and therefore represents a major problem in drug design.

If a drug has an ionizable centre then solubility can be improved by salt formation. In the absence of a salt basic drugs will also have increased solubility in the acidic environment of the stomach.

The incorporation of an ionizable centre, such as an amine or similar function, into a template can bring a number of benefits including water solubility. A key step [5] in the discovery of indinavir was the incorporation of a basic amine (and a pyridine) into the backbone of hydroxyethylene transition state mimic compounds (Figure 3.3) to enhance solubility (and potency).

Fig. 3.3 Structures of lead compound L-685,434 and indinavir which incorporates basic functions aiding water solubility.

L-685,434 indinavir

Fig. 3.3 Structures of lead compound L-685,434 and indinavir which incorporates basic functions aiding water solubility.

Membrane Transfer

The barrier of the gastrointestinal tract is similar to any other that involves the crossing of biological membranes. Biomembranes are composed of a lipid-bilayer [6]. The bilayer results from the orientation of the lipids (phospholipids, glycolipids and cholesterol) in the aqueous medium. Phospholipids are amphipathic with polar head groups and lipid "tails" and align so that the polar head groups orientate towards the aqueous medium and the lipid tails form an inner hydrophobic core. Because of the high flexibility of membrane lipids they are able to perform transversal/lateral movements within the membrane. A variety of proteins such as selective ion channels (Na+, K+, Ca2+, Cl-) are embedded within the membrane. Tight junctions are formed by the interaction of membrane proteins at the contact surfaces between single cells. Tight junctions are in reality small aqueous-filled pores. The dimensions of these pores have been estimated to be in the range of 3-10 A. The number and dimensions of the tight junctions depend on the membrane type. For the small intestine these tight junctions make up about 0.01 % of the whole surface. Thus the surface area of the actual biological membrane is much greater than that of the aqueous pores (tight junctions).

Compounds can cross biological membranes by two passive processes, transcellu-lar and paracellular mechanisms. For transcellular diffusion two potential mechanisms exist. The compound can distribute into the lipid core of the membrane and diffuse within the membrane to the basolateral side. Alternatively, the solute may diffuse across the apical cell membrane and enter the cytoplasm before exiting across the basolateral membrane. Because both processes involve diffusion through the lipid core of the membrane the physicochemistry of the compound is important. Paracellular absorption involves the passage of the compound through the aqueous-filled pores. Clearly in principle many compounds can be absorbed by this route but the process is invariably slower than the transcellular route (surface area of pores versus surface area of the membrane) and is very dependent on molecular size due to the finite dimensions of the aqueous pores.

The actual amount of a drug absorbed (Fa) is dependent on two rates: the rate of absorption (ka) and the rate of disappearance of the drug from the absorption site. Disappearance can be due to absorption (ka) or movement of the drug (km) through the gastrointestinal tract and away from the absorption site. The proportion absorbed can be expressed as:

Compounds crossing the gastrointestinal tract via the transcellular route can usually be absorbed throughout the length of the tract. In contrast the paracellular route is only, readily, available in the small intestine and the term "absorption window" is often applied. The calculated human pore sizes (radii) are, jejunum 6-8 A, ileum 2.9-3.8 A and colon less than 2.3 A. In practice the small intestine transit time is around 6 h whilst transit of the whole tract is approximately 24 h. For lipophilic compounds, with adequate dissolution, which have high rates of transcellular passage across membranes, ka has a high value. Moreover, since the drug is absorbed throughout the g.i. tract km is of a low value and therefor the proportion of a dose absorbed is high (complete). For hydrophilic compounds, which are dependent on the slow paracellular pathway, ka has a low value. Moreover, the "absorption window" referred to above means that the drug rapidly moves away from the absorption site and km is high. Consequently paracellularly absorbed compounds show incomplete absorption and the proportion which is absorbed is low. Table 3.1 gives examples of compounds absorbed by the paracellular route.

What is noticeable is that the compounds are of low molecular weight, however, there is no simple relationship between molecular weight and percentage absorbed, probably indicating that shape and possibly flexibility are also of importance. Compounds such as propranolol (log D7A, 0.9) which are related to those in Table 3.1,

3.3 Membrane Transfer | 39 Tab. 3.1 Examples of drugs absorbed by the paracellular route.


log D7.4

Molecular weight

% Absorbed






















- G.8




- G.8




- G.6




- G.6




- G.3



show high flux rates via the transcellular route and consequently are completely absorbed. Note, however, that lipophilicity correlates with increased metabolic lability and such compounds may have their apparent systemic availabilities decreased by metabolism as they pass through the gut and the liver.

For simple molecules, like p-adrenoceptor antagonists octanol/water log D74 values are remarkably predictive of absorption potential. Compounds with log D74 values below 0 are absorbed predominantly by the paracellular route and compounds with log D74 values above 0 are absorbed by the transcellular route.

Another example of the relationship between log D values and intestinal absorption is taken from reference [1] (see Figure 3.4). Compounds with log D >0 demonstrate a nearly complete absorption. Two exceptions are compounds with a MW above 500. Whether size as such, or the accompanying increase in the number of H-bonds, is responsible for poorer absorption is not fully understood.

Fig. 3.4 Dependence of oral absorption on log D [1].

Fig. 3.4 Dependence of oral absorption on log D [1].

However, as the number of H-bonding functions in a molecule rises, octanol/ water distribution, in isolation, becomes a progressively less valuable predictor. For such compounds desolvation and breaking of H-bonds becomes the rate-limiting step in transfer across the membrane [7].

Octanoi/Cyclohexane Ratio (H-bonding)

sec Amine (4.5)

Ten Amine (2.5)

pti Amine (5.1)


Ester (2.4)

Amide (8.6)


Ether (1.8)

Carboxylate (4.7)

Halogen (<1)

Ketone (1.8)

Hydroxyl (3.2)

Nitrile (1.7)

Sulphonamide (10.0)

Nitro (0.8)

Sulphone (4.1)

Sulphoxide (3.1)

Fig. 3.5 Raevsky H-bond scores from HYBOT95 (shown in parentheses) and correlation with Dlog D (compare with Figure 1.2 in Chapter 1).

Methods to calculate H-bonding potential range from simple H-bond counts (number of donors and acceptors), through systems that assign a value of 1 for donors and 0.5 for acceptors to sophisticated scoring systems such as the Raevsky H-bond score [8]. The correlation of Raevsky H-bond scores with Alog D shown previously as Figure 1.4 in Chapter 1 (Physicochemistry) is shown as Figure 3.5.

None of these methods gives a perfect prediction, particularly because H-bonding potential needs to be overlaid over intrinsic lipophilicity. For this reason Lipinski's "rule-of-five" becomes valuable in defining the outer limits in which chemists can work [9]. Lipinski defined the boundaries of good absorption potential by demonstrating that poor permeability is produced by:

• more than five H-bond donors (sum of OHs and NHs)

• more than 10 H-bond acceptors (sum of Ns and Os)

• molecular weight over 500

• poor dissolution by log P over 5

The medicinal chemist can use these rules and understand the boundaries and work towards lowering these values. Figure 3.6 shows a synthetic strategy aimed at removing H-bond donors from a series of endothelin antagonists and a resultant increase in apparent bioavailability as determined by intra-dueodenal AUC [10]. Noticeabl CLOGP values vary only marginally with the changes in structure, values being 4.8, 5.0, 4.8 and 5.5 for compounds A, B, C and D respectively. In contrast the

Fig. 3.6 Removal of H-bond donors as a synthetic strategy for a series of azole-containing endothelin antagonists aimed at improving bioavailability by lowering H-bonding potential [10].

Fig. 3.7 Replacement of amide with acetyl in a series of amidothio-phenesulfonamide en-dothelin-A antagonists to improve oral

Fig. 3.7 Replacement of amide with acetyl in a series of amidothio-phenesulfonamide en-dothelin-A antagonists to improve oral

bioavailability [11].

number of H-bond donors is reduced by 3 and the Raevsky score from 28.9 (A) to 21.4 (D).

A similar example, also from endothelin antagonists, is the replacement of the amide group (Figure 3.7) in a series of amidothiophenesulfonamides with acetyl [11]. This move retained in vitro potency, but markedly improved oral bioavailability.

Barriers to Membrane Transfer

The cells of the gastrointestinal tract contain a number of enzymes of drug metabolism and also various transport proteins. Of particular importance in the attenuation of absorption/bioavailability are the glucuronyl and sulphotransferases which metabolize phenol-containing drugs (see below) sufficiently rapidly to attenuate the passage of intact drug across the gastrointestinal tract. Cytochrome P450 enzymes are also present, in particular CYP3A4 (see Chapter 7) and again certain substrates for the drug may be metabolized during passage across the tract. This effect may be greatly enhanced by the action of the efflux pumps, in particular P-glycoprotein. The range of substrates for P-glycoprotein is large but includes a number of relatively large molecular weight drugs which are also CYP3A4 substrates. Cyclosporin A is one example. This drug shows significant attenuation of absorption across the gastrointestinal tract due to metabolism. Metabolism by the gut is greater than many other examples of CYP3A4 substrates. It can be postulated that in effect absorption of the drug is followed by secretion back into the lumen of the gut by P-glycoprotein. This cyclical process effectively exposes cyclosporin A to "multi-pass" metabolism by CYP3A4 and a resultant reduced appearance of intact cyclosporin A in the circulation.

Detailed structure-activity relationships of P-glycoprotein are not yet available. Some understanding is provided by Seelig [12] who has compared structural features in P-glycoprotein substrates. This analysis has indicated that recognition elements are present in structures and are formed by two (type I) or three electron donor groups (type II) with a fixed spatial separation. The type I element consists of two electron donor groups separated by 2.5 A, whilst the type II elements has a spatial

separation of the outer groups of 4.6 A. All molecules that are P-glycoprotein substrates contain at least one of these groups and the affinity of the substrate for P-gly-coprotein depends on the strength and number of electron donor or hydrogen bond acceptor groups. For the purpose of this analysis all groups with an unshared electron pair on an electronegative atom (O, N, S or F and Cl), or groups with a n-elec-tron orbital of an unsaturated system, were considered as electron donors. However, this analysis did not account for the directionality of the H-bonds.

The dramatic effect of a single unit is shown in Figure 3.8 for a series of beta-adrenoceptor antagonists.

Fig. 3.8 Structures of propranolol (1), betaxolol (2), metoprolol (3) and talinolol (4) and their respective extraction by the gastointestinal tract (E(g. i.)) and liver £(h)

All these compounds are moderately lipophilic and should show excellent ability to cross biological membranes by transcellular absorption. Propranolol, betaxolol and metoprolol all have minimal gut first-pass metabolism, as shown by the low value for E(g. i.). Metabolism and first pass effects for these compounds are largely confirmed to the liver as shown by the values for E(g. i.). In contrast talinolol shows high extraction by the gastrointestinal tract with low liver extraction [13]. These effects are illustrated graphically in Figure 3.9 which shows the bioavailability predicted from hepatic extraction contrasted with that seen in vivo in man.

Noticeably propranolol, betaxolol and metoprolol are close or on the borderline for hepatic first-pass effects, whereas talinolol falls markedly below it. Talinolol has been shown to be a substrate for P-glycoprotein [14]. The effect of the urea function is of key importance within this change, as urea lacks a strong type I unit in terms of Seel-

Fig. 3.9 Bioavailability (F) of propranolol (1), betaxolol (2), metoprolol (3) and talinolol (4) found in vivo in man compared to that predicted based solely on hepatic extraction.

Fig. 3.9 Bioavailability (F) of propranolol (1), betaxolol (2), metoprolol (3) and talinolol (4) found in vivo in man compared to that predicted based solely on hepatic extraction.

ig's classification. Other changes in the molecule, such as the tertiary butyl, rather than isopropyl N-substituent are not so important since the related compounds pafenolol and celiprolol (Figure 3.10) also show similar bioavailability.

Fig. 3.10 Structures of pafenolol and celiprolol, derivatives of the talinolol (see Figure 3.6) structure which show similar bioavailability characteristics.

Fig. 3.10 Structures of pafenolol and celiprolol, derivatives of the talinolol (see Figure 3.6) structure which show similar bioavailability characteristics.

These considerations are important in pro-drug design and add to the complexity referred to earlier. Many active principles in pro-drug programmes are non-lipophilic compounds, possessing a number of H-bond donor and acceptor functions (amide or peptide linkages). Addition of a pro-moiety will raise the lipophilicity and molecular weight. In doing so the final molecule may have the required structure to traverse the lipid core of a membrane, but this advantage is lost by it becoming a substrate for efflux. An example (Figure 3.11) of this is the fibrinogen receptor antagonist L-767,679, a low lipophilicity compound (log P < - 3) with resultant low membrane flux. The benzyl ester (L-775,318) analogue (log P 0.7) also showed limited absorption, and studies in Caco-2 cells (see Section 3.5) showed the compound to be effluxed by P-glycoprotein [15].

Fig. 3.11 Structures of the fibrinogen receptor antagonist L-767,679 and its benzyl ester (L-775,318) analogue.

Models for Absorption Estimation

A number of models have been suggested to estimate the absorption potential in humans (see Table 3.2) [16, 17]. These vary from low throughput (in situ rat model) to high throughput (in silico) models. Most companies will use a combination of these approaches. The human colon adenocarcinoma cell lines Caco-2 and HT-29 are widely used as screening models for absorption [18, 19]. An alternative is offered by the MDCK cell line which is a faster growing cell [20]. These cell lines express typical impediments for absorption such those mentioned above for P-glycoprotein and CYP3A4 isoenzyme. They are thus believed to by a good mimic of the physicochem-ical and biological barrier of the g. i. tract.

Tab. 3.2 Models for absorption estimation.

• In vitro (Caco-2 and other cell lines; Ussing chamber)

• Physicochemical properties

Estimation of Absorption Potential

A simple dimensionless number, absorption potential (AP), has been proposed to make first approximation predictions of oral absorption (Eq. 3.3) [3].

In this equation, log P is the partition coefficient for the neutral species, log Fnon the fraction of non-ionized compound, So the intrinsic solubility, VL the lumenal vol-

3.7 Computational Approaches 45

ume and Xo the given dose. By extending this approach the effect of particle size on oral absorption has also been modelled [21].

An approach to estimating the maximum absorbable dose (MAD) in humans is based on Eq. (3.4) [22,23].

S is the solubility in phosphate buffer at the pH 6.5 (in mg mL-1), ka the absorption rate constant in rats (min-1), IFV is the intestinal fluid volume (250 mL), and RT is the average residence time in the small intestine (270 min).

Computational Approaches

As mentioned above, hydrogen bonding and molecular size, in combination with lipophilicity have an important influence on oral absorption. A number of methods are available to compute these properties. A further example of the correlation between H-bonding, expressed as polar surface area, is found in Figure 3.12 [24,25]. Such a sigmoidal relationship is found for compounds which are absorbed by passive diffusion only and not hindered by efflux or metabolism, and which are not involved in active uptake. Otherwise deviations will be found [25].

Combination of several descriptors believed to be important for oral absorption have been used in various multivariate analysis studies [26]. The general trend is that a combination of size/shape and a hydrogen bond descriptor, sometimes in combination with log D, has good predictive value. At present such models do not account for the biological function of the membrane, such as P-gp-mediated efflux.

Fig. 3.12 Correlation between polar surface area (PSA) and intestinal absorption [24, 25].

Fig. 3.12 Correlation between polar surface area (PSA) and intestinal absorption [24, 25].


1 Van de Waterbeemd H, In: Oral Drug Absorption. Prediction and Assessment, Dekker, New York, Eds. Dressman J, Lennernäs, H, pp. 31-49.

2 Yalkowski SH, Valvani SC, J. Pharm. Sci. 1980, 69, 912-922.

3 Dressman JB, Amidon GL, Fleisher D, J. Pharm. Sci. 1985, 74, 588-589.

4 Amidon GL, Lennernas H, Shah VP, Crison JR, Pharm. Res. 1995, 12, 413420.

5 Vacca JP, Dorsey BD, Schleif WA, Levin RB, McDaniel SL, Darke PL, Zugay J, Quintero JC, Blahy OM, Roth E, Sardana VV, Schlabach AJ, Graham PI, Condra JH, Gotlib L, Holloway MK, Lin J, Chen I-W, Vastag K, Ostovic D, Anderson PS, Emini EA, Huff JR, Proc. Natl Acad. Sci. USA 1994, 91, 4096-4100.

6 Macheras P, Reppas C, Dressman JB (Eds) Biopharmaceutics of Orally Administered Drugs, Ellis Horwood, London, 1995.

7 Conradi RA, Burton PS, Borchardt RT, In: Lipophilicity in Drug Action and Toxicology (Eds Pliska V, Testa B, Van de Waterbeemd H), pp. 233-252. VCH, Weinheim, 1996.

8 Raevsky OA, Grifor'er VY, Kireev DB, Zefirov, NS, Quant. Struct. Activity Relat. 1992, 14, 433-436.

9 Lipinski CA, Lombardo F, Dominy BW, Feeney PJ, Adv. Drug Del. Rev. 1997, 23, 3-25.

10 Von Geldern TW, Hoffman DJ, Kester JA, Nellans HN, Dayton BD, Calzadilla SV, Marsch KC, Hernandez L, Chiou W, J. Med. Chem. 1996, 39 982-991.

11 Wu C, Chan MF, Stavros F, Raju B, Okun I, Mong S, Keller KM, Brock T, Kogan TP, Dixon RAF, J. Med. Chem. 1997, 40, 1690-1697.

13 Travsch B, Oertel R, Richter K, Gramatt T, Biopharm. Drug Dispos. 1995, 16, 403-414.

14 Spahn-Langguth H, Baktir G, Rad-schuweit A, Okyar A, Terhaag B, Ader P, Hanafy A, Langguth P, Int. J. Clin. Pharmacol. Ther. 1998, 36, 16-24.

15 Prueksaitanont T, Deluna P, Gorham LM, Bennett MA, Cohn D, Pang J, Xu X, Leung K, Lin JH, Drug Metab. Dispos. 1998, 26, 520-527.

16 Borchardt RT, Smith PL, Wilson G, (Eds) Models for Assessing Drug Absorption and Metabolism. Plenum Press, New York, 1996.

17 Barthe L, Woodley J, Houin G, Fund. Clin. Pharmacol. 1999, 13, 154-168.

18 Hidalgo IJ, In: Models for Assessing Drug Absorption and Metabolism (Eds Borchardt RT, Smith PL, Wilson G), pp. 35-50. Plenum Press, New York, 1996.

19 Artursson P, Palm K, Luthman K, Adv. Drug Deliv. Rev. 1996, 22, 67-84.

20 Irvine JD, Takahashi L, Lockhart K, Cheong J, Tolan JW, Selick HE, Grove JR, J. Pharm. Sci. 1999, 88, 28-33.

21 Oh D-M, Curl RL, Amidon GL, Pharm. Res. 1993, 10, 264-270.

22 Johnson KC, Swindell AC, Pharm. Res. 1996, 13, 1794-1797.

23 Lombardo F, Winter SM, Tremain L, Lowe III JA, In: Integration of Pharmaceutical Discovery and Development: Case Studies (Eds Borchardt RT et al.), pp. 465-479. Plenum Press, New York, 1998.

24 Palm K, Luthman K, Ungell A-L, Strandlund G, Beigi F, Lundahl P, Artursson P, J. Med. Chem. 1998, 41, 5382-5392.

26 Van de Waterbeemd H, In: Pharmaco-kinetic Optimization in Drug Research: Biological, Physicochemical and Computational Strategies (Eds Testa B, Van de Waterbeemd H, Folkers G, Guy R), Verlag HCA, Basel, 2001, pp. 499-511.

Pharmacokinetics and Metabolism in Drug Design 47 Edited by D. A. Smith, H. van de Waterbeemd, D. K. Walker, R. Mannhold, H. Kubinyi, H. Timmerman I

Copyright © 2001 Wiley-VCH Verlag GmbH ISBNs: 3-527-30197-6 (Hardcover); 3-527-60021-3 (Electronic)

Was this article helpful?

0 0

Post a comment