Enhancement of oral absorption in selective 5-HT(1D) receptor agonists: Fluorinated 3-[3-(Piperidin-1-yl)propyl]indoles [2]

Full text of DOI:10.1021/jm980204e

J . Med. Chem. 1998, 41, 2667-2670
2667
Ch a r t 1
En h a n cem en t of Or a l Absor p tion in
Selective 5-HT1D Recep tor Agon ists:
F lu or in a ted
3-[3-(P ip er id in -1-yl)p r op yl]in d oles
J ose´ L. Castro,*,† Ian Collins,†
Michael G. N. Russell,† Alan P. Watt,† Bindi Sohal,†
Denise Rathbone,† Margaret S. Beer,‡ and
J osephine A. Stanton‡
Chemistry and Biochemistry Departments,
Merck Sharp and Dohme Research Laboratories,
Neuroscience Research Centre, Terlings Park,
Eastwick Road, Harlow, Essex CM20 2QR, U.K.
Received April 1, 1998
The discovery and introduction into clinical practice
of selective serotonin 5-HT1D/B receptor agonists has
revolutionized the acute treatment of migraine head-
ache.1 However, the exact mechanism of action of this
class of compounds, collectively referred to as “triptans”
(e.g., sumatriptan, zolmitriptan, naratriptan, rizatrip-
tan, eletriptan), is still a matter of debate. Because all
of the triptan agents so far developed bind with almost
equal affinities to human 5-HT1D and 5-HT1B receptors,2
both a direct vasoconstrictor effect on cranial blood
vessels3 (5-HT1B mediated) and/or an inhibition of
neurogenic inflammation4 in the dura matter (5-HT1D/B
mediated) have been proposed as potential pathways for
headache pain relief by these agents. A central anti-
nociceptive component has also been advanced as a key
contributor in their mechanism of action.5
Although the side effect profile of these compounds
is extremely good, the potential exists for coronary vaso-
constriction (5-HT1B mediated) that precludes their use
in patients with known heart disease.6 This fact,
coupled to the mechanistic uncertainties described
above, prompted us to seek the development of selective
5-HT1D receptor agonists as a potential second genera-
tion migraine therapy devoid of vasoconstrictor li-
abilities. In this regard, we have recently disclosed the
discovery of the first 5-HT1D agonists with up to 200-
fold binding selectivity over the 5-HT1B subtype.7,8
Compounds such as 1 (L-772,405) and 4 (L-775,606)
were highlighted from this work (Chart 1). While 1 was
found to have an almost ideal in vitro profile, and to be
rapidly absorbed by the subcutaneous route in rats (F,
88%; Cmax, 320 ng/mL; Tmax, 10 min; dose, 3 mg/kg), its
oral bioavailability was extremely low (<5%). The lack
of good systemic exposure after oral dosing was a
differentiating factor between compounds belonging to
the piperidine series (e.g., 1, 2, 3) and those incorporat-
ing a piperazine moiety (e.g., 4, 5; F, 27 and 21%,
respectively; rat, 3 mg/kg). Extensive pharmacokinetic
and metabolism studies with 1 and related compounds
showed that their low oral bioavailablity was most
probably due to deficient oral absorption rather than
metabolic instability. For example, concentrations of
1 in hepatic portal vein (hpv) at 0.5 and 2 h post 3 mg/
kg po in rats were extremely low (Table 1), and this
poor absorption correlated well with epithelial transport
studies carried out in Caco-2 cells.9 In addition, 1 was
not extensively metabolized in rat liver microsomes or
fresh rat and human liver slices. It was, therefore, of
particular interest to try to understand the reasons
behind the different absorption behavior of the pip-
eridines and piperazines (e.g., compare 3 vs 5, Table 1)
in order to introduce oral bioavailability into the former
series. It was speculated that in the more basic pip-
eridine series (e.g., 3, pKa 9.7 vs 5, pKa 8.2) a smaller
proportion of uncharged species would exist at the
physiological pHs encountered in the small intestine,
leading to reduced passive diffusion across the gut
wall.10 To test this idea, modulation of the pKa of the
piperidine moiety was undertaken, and here we report
on the synthesis and biological evaluation of three series
of fluorinated compounds (see structures 6, 7, and 8;
Chart 1) which indeed showed enhanced absorption
properties when compared to their unsubstituted ana-
logues.
Introduction of fluorine was first targeted at C-4 of
the piperidine for two reasons. First, this modification
would not introduce a chiral center in the molecule, and
in addition, it was thought that perhaps the electrostatic
potential generated by the electronegative fluorine11
could better mimic the region around N-4 of the pip-
erazine moiety in 5. Because this transformation would
probably compromise the stability of compounds bearing
a nitrogen substituent at C-4 of the piperidine (e.g., 1,
2), fluorination of the ring was also pursued at C-3,
although the achiral advantage of the previous substi-
tution was lost. Compounds of general structure 6 were
prepared from the corresponding 4-hydroxypiperidine
analogues by treatment with DAST.12 Thus, ring open-
ing of oxirane 913 with 3-fluorophenethylmagnesium
bromide afforded 10 in moderate yield, which was
† Chemistry Department.
‡ Biochemistry Department.
S0022-2623(98)00204-0 CCC: $15.00 © 1998 American Chemical Society
Published on Web 06/30/1998

2668 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 15
Communications to the Editor
Ta ble 1. Human 5-HT1D Receptor Binding and Absorption Data for Fluorinated Piperidines and Reference Compounds
a
b
Where applicable, all compounds are racemic. Displacement of [3H]-5-HT binding to cloned human 5-HT1D receptors stably expressed
in CHO cells.7 The figures are the mean of two to four independent determinations performed in duplicate. In each case the radioligand
concentration was used at the KD for the receptor. The maximum variance from the mean of the log (IC50) values was 3.5%. c Binding
selectivity for 5-HT1D over 5-HT1B receptors. Measurement of agonist induced [35S]GTPγS binding in CHO cells.7 e Efficacy relative to
d
5-HT. Values are the mean of two independent determinations. f Concentration of compound in rat plasma obtained from hepatic portal
vein (hpv) and cardiac (systemic) blood samples following po administration at 3 mg/kg (four animals per group). Measured by electrospray
mass spectrometry. Determined by potentiometric titration as previously described.17
g
Sch em e 1a
Sch em e 2a
a
Reagents: (i) TMSCl, Et3N, DMF, 80 °C; (ii) Selectfluor,
MeCN, room temperature; (iii) RNH2, NaBH(OAc)3, ClCH2CH2Cl;
or NaCNBH3, AcOH, MeOH, room temperature; (iv) TFA, CH2Cl2;
(v) 13, K2CO3, 2-propanol, reflux; (vi) CH2O, NaCNBH3, AcOH,
MeOH.
followed by reaction of the resulting piperidine with
mesylate 1314 finally afforded 6b. In a similar manner
6a was prepared from 4-hydroxybenzylpiperidine. Prepa-
ration of 3-fluoro-4-aminopiperidines 7 was achieved as
shown in Scheme 2, the key step being the treatment
of trimethylsilyl enol ether 15 with a source of electro-
philic fluorine such as 1-chloromethyl-4-fluoro-1,4-diazo-
niabicyclo[2.2.2]octane bis(tetrafluoroborate) (Select-
fluor)15 to cleanly afford 16 in 91% yield. Reductive
amination of this ketone under different conditions gave
varying mixtures16 of racemic cis and trans products 17
and 18 which, after separation, could be converted into
the required targeted compounds 7a ,b as described
above.
a
Reagents: (i) 3-fluorobenzyl bromide, Mg, Et2O; then add 9,
-30 °C; (ii) DAST, CH2Cl2, -78 to 0 °C; (iii) TFA, CH2Cl2; (iv) 13,
K2CO3, 2-propanol, reflux.
converted into a mixture of 11 and 12 by reaction with
DAST at low temperature (Scheme 1). Chromato-
graphic separation of the fluorinated compounds from
the tetrahydropyridine side products was facilitated by
epoxidation of the crude mixture with m-chloroperoxy-
benzoic acid. Removal of the BOC protecting group
It can be seen from the data in Table 1 that fluorina-
tion at C-4 of the piperidine is well tolerated by the

Communications to the Editor
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 15 2669
Sch em e 3a
5-HT1D receptor and both 6a and 6b bind with affinities
comparable to those of the unsubstituted analogue 3 and
piperazine 4, respectively. A similar result was ob-
served for the 3-fluoro-4-aminopiperidines 7a and 7b,
although the former 3,4-cis isomer showed a 6-fold
reduction in affinity compared to 2. In addition, 6a was
as efficacious as 3 in the [35S]GTPγS assay7 while 6b
behaved as a partial agonist when compared to 4. It
can also be seen in Table 1 that introduction of fluorine
into the piperidine ring reduced the basicity of the
molecules by aproximately one pKa unit,17 bringing it
closer to that found in the piperazines. Thus, although
6a ,b and 7a ,b had reduced selectivity for 5-HT1D over
5-HT1B receptors compared to the unfluorinated ana-
logues, it was of interest to determine the effect of this
pKa modulation on their oral absorption profiles. Rats
were dosed orally (3 mg/kg, 5 mL/kg dose volume;
aqueous formulation) with the test compound, and blood
samples were removed from the hpv and by cardiac
puncture at 0.5 and 2 h postadministration (four rats
per time point). Drug concentrations measured in
plasma samples originating from hpv blood would reflect
the levels of absorbed compound, which might be
augmented by additional drug present in the systemic
circulation (after successful avoidance of first-pass
metabolism). The concentrations arising from cardiac
blood samples would give an indication of systemic
exposure and therefore of bioavailability of the com-
pound. It is clear from the data in Table 1 that
reduction of the pKa of the piperidine results in in-
creased oral absorption. Thus, 6a and 7a showed much
improved levels of compound in hpv plasma when
compared to 3 and 2, respectively. Moreover, 6a was
also found to be present in the systemic circulation at
reasonable concentrations both at 0.5 and 2 h (57 and
37 ng/mL, respectively). The enhanced absorption seen
with 7a did not translate, however, into appreciable oral
bioavailability for this compound.
Encouraged by the above data, modulation of the pKa
was sought by fluorination of the propyl linking chain
because previous structure-affinity relationships data
indicated that substitution at this position could be less
detrimental for 5-HT1D over 5-HT1B receptor selectivity.
Ring opening of commercially available 1,2-epoxy-5-
hexene (19) with 70% HF-pyridine occurred at the most
substituted end of the epoxide18 to give 20 in 30% yield
(Scheme 3). Protection of the primary alcohol as the
TBS derivative was then followed by ozonolysis of the
olefin to afford aldehyde 22 in 54% yield.19 Fischer
indolization with 4-(1,2,4-triazol-4-yl)phenylhydrazine20
under acidic conditions gave fluorinated homotryptol 23
in low yield (15%). Activation of the primary alcohol
followed by reaction with the appropriate amines finally
afforded compounds 8a -d in moderate yields (50-55%).
It can be seen (Table 1) that fluorination of the propyl
chain or the piperidine ring led to a similar reduction
in pKa (e.g., 6a vs 8a ). Compounds 8a -d bound with
high affinity to 5-HT1D receptors and, in contrast to the
above fluorinated piperidine analogues 6a ,b and 7a ,b,
they showed high levels of selectivity over the 5-HT1B
subtype. In the case of 8b and 8d the selectivity was
even improved ∼3-fold over the unfluorinated analogues
2 and 4, and it is tempting to speculate that this might
be due to a change in binding mode for these two
compounds. Thus, one would anticipate that the pres-
a
Reagents: (i) 70% HF-Py, CH2Cl2, -10 °C; (ii) TBSCl, imida-
zole, DMF; (iii) O3, CH2Cl2; then Me2S, room temperature, 3 h;
(iv) Et3N (2 equiv), CH2Cl2, room temperature, 2 h; (v) 4-(1,2,4-
triazol-4-yl)phenylhydrazine, 4% H2SO4, reflux; (vi) MsCl, Et3N,
THF; (vii) R′2NH, K2CO3, 2-propanol, reflux.
ence of the fluorine atom in 8d would make N-1 of the
piperazine (the nitrogen atom joined to the indole
nucleus) less basic than N-4 and, therefore, 8d would
bind to helix-3 aspartate in 5-HT1D receptors through
a protonated N-4 nitrogen atom rather than N-1 as in
4. The same argument could probably be applied to the
4-aminopiperidine 8b but would be absent in the mono-
basic 4-benzylpiperidine analogue 8a . In the functional
GTPγS assay 8a behaved as a partial agonist while
8b-d were full agonists. Gratifyingly, the enhanced
oral absorption seen with the previous fluorinated
compounds was also maintained in this series and, for
example, 8a rapidly achieved high plasma levels in both
hpv and systemic blood which compared very favorably
with those obtained for 3. In full pharmacokinetic
studies, 8a and 8d were found to be orally bioavailable
in rats (8a : F, 14%; t1/2, 1.2 h. 8d : F, 7%; t1/2, 1.2 h).
In summary, three related series of fluorinated 3-[3-
(piperidin-1-yl)propyl]indoles were developed as selec-
tive 5-HT1D agonists with the aim of enhancing oral
absorption by modulation of the amine pKa. Particu-
larly interesting are compounds where fluorine was
introduced on the propyl chain, some of which did
indeed show improved absorption profiles while retain-
ing or enhancing 5-HT1D over 5-HT1B receptor selectivity
and agonist properties. In addition, fluorination of the
piperidine ring at C-4 (e.g., 6a ) led to a similar reduction
in basicity of approximately one pKa unit, bringing it
closer to that of the corresponding piperazine analogue
5. This latter modification, which also generates a
region of electrostatic potential around C-4 of the
piperidine, may indeed provide a useful piperazine
replacement for drug discovery.
Su p p or tin g In for m a tion Ava ila ble: Experimental pro-
cedures for the preparation of 6b, 7a , and 8a (14 pages).
Ordering information is given on any current masthead page.
Refer en ces
(1) (a) Chustecka, Z. New Drugs for Migraine: 5-HT1 Agonists and
Beyond. Pharmaprojects Magazine, March 1997; pp 16-18. (b)
Yevich, J . P.; Yocca, F. D. Second Generation Antimigraine
5-HT1B/D Agonists: Structure Activity Relationship and Pre-
clinical Pharmacological Distinctions. Curr. Med. Chem. 1997,
4, 295-312.

2670 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 15
Communications to the Editor
(2) 5-HT1D and 5-HT1B receptors were formerly known as 5-HT1DR
and 5-HT1Dâ, respectively. (a) Weinshank, R. L.; Zgombick, J .
M.; Macchi, M. J .; Branchek, T. A.; Hartig, P. R. Human
Serotonin 1D Receptor is Encoded by a Subfamily of two Distinct
Genes: 5-HT1DR and 5-HT1Dâ. Proc. Nat. Acad. Sci. U.S.A. 1992,
89, 3630-3634. (b) Leysen, J . E.; Gommeren, W.; Heylen, L.;
Luyten, W. H. M. L.; van de Weyer, I.; Vanhoenacker, P.;
Haegeman, G.; Schotte, A.; van Gompel, P.; Wouters, R.; Lesage,
A. S. Alniditan, a New 5-Hydroxytryptamine1D Agonist and
Migraine-Abortive Agent: Ligand-Binding Properties of Human
5-Hydroxytryptamine1DR, Human 5-Hydroxytryptamine1Dâ, and
Calf 5-Hydroxytryptamine1D Receptors Investigated with [3H]5-
Hydroxytryptamine and [3H]Alniditan. Mol. Pharmacol. 1996,
50, 1567-1580. (c) Pauwels, P. J .; Colpaert, F. C. Selective
Antagonism of Human 5-HT1D and 5-HT1B Receptor-Mediated
Responses in Stably Transfected C6-Glial Cells by Ketanserin
and GR 127, 935. Eur. J . Pharmacol. 1996, 300, 141-145. (d)
Connor, H. E.; Beattie, D. T. 5-Hydroxytryptamine Receptor
Subtypes and Migraine. In Migraine: Pharmacology and Genet-
ics; Sandler, M., Ferrari, M., Harnett, S., Eds.; Chapman and
Hall: London, 1996; pp 18-31.
(10) (a) Mirchandani, H. L.; Chien, Y. W. Intestinal Absorption of
Dideoxynucleosides: Characterization Using a Multiloop in Situ
Technique. J . Pharm. Sci. 1995, 84, 44-48. (b) Dennis, M. J .
Absorption Processes. Comprehensive Med. Chem. Vol 5, pp
1-19. (c) Smith, D. A.; J ones, B. C.; Walker, D. K. Design of
Drugs Involving the Concepts and Theories of Drug Metabolism
and Pharmacokinetics. Med. Res. Rev. 1996, 16, 243-266. (d)
Chan, O. H.; Stewart, B. H. Physicochemical and Drug-Delivery
Considerations for Oral Drug Bioavailability. Drug Discovery
Today 1996, 1, 461-473.
(11) (a) Resnati, G. Synthesis of Chiral and Bioactive Fluoroorganic
Compounds. Tetrahedron 1993, 49, 9385-9445. (b) For a whole
issue devoted to fluorine chemistry, see: Chem. Rev. 1996, 96
(5).
(12) Hudlicky, M. Fluorination with Diethylaminosulfur Trifluoride
and Related Aminofluorosulfuranes. Org. React. 1988, 35, 513-
637.
(13) 6-Aza-6-tert-butyloxycarbonyl-1-oxaspiro[2.5]octane (9) was pre-
pared from 1-tert-butyloxycarbonyl-4-piperidone and trimethyl-
sulphoxonium iodide: Corey, E. J .; Chaykovsky, M. Methyle-
necyclohexane Oxide. Organic Syntheses; Wiley: New York,
1973; Collect. Vol. V, 1973; pp 755-757.
(3) Humphrey, P. P. A.; Goadsby, P. J . The Mode of Action of
Sumatriptan is Vascular? A Debate. Cephalalgia 1994, 14, 401-
410.
(4) (a) Buzzi, M. G.; Bonamini, M.; Moskowitz, M. A. Neurogenic
Model of Migraine. Cephalalgia 1995, 15, 277-280. (b) Bouche-
let, I.; Cohen, Z.; Case, B.; Seguela, P.; Hamel, E. Differential
Expression of Sumatriptan-Sensitive 5-Hydroxytryptamine Re-
ceptors in Human Trigeminal Ganglia and Cerebral Blood
Vessels. Mol. Pharmacol. 1996, 50, 219-223.
(14) Mesylate 13 was prepared from the corresponding alcohol using
standard conditions (MsCl, Et3N, THF). The precursor alcohol
was synthesized by Fischer indole reaction between 4-(1,2,4-
triazol-4-yl)phenylhydrazine20 and 3,4-dihydro-2H-pyran in re-
fluxing 4% H2SO4: Castro, J . L.; et al. Substituted Piperidine
Derivatives as Selective Agonists of 5-HT Receptors. World
Patent Appl. WO 97/18202, 1997.
(15) Lal, G. S. Site-Selective Fluorination of Organic Compounds
Using 1-Alkyl-4-fluoro-1,4-diazabicyclo[2.2.2]octane Salts (Se-
lectfluor Reagents). J . Org. Chem. 1993, 58, 2791-2796.
(16) Reductive alkylation of 16 with benzylamine gave predominantly
the cis-isomer 17 (ca 8:1) when sodium triacetoxyborohydride
in dichloroethane was used. This selectivity was, however, lost
(17:18 ca. 1:1) when sodium cyanoborohydride in MeOH-AcOH
was utilized. The relative stereochemistry of these compounds
could be unambiguously assigned by 500 MHz NMR spectros-
copy, details of which will be given in a forthcoming publication.
(17) pKa determinations were carried out by potentiometric titration
as previously described: Castro, J . L.; Broughton, H. B.; Russell,
M. G. N.; Rathbone, D.; Watt, A. P.; Ball, R. G.; Chapman, K.
L.; Patel, S.; Smith, A. J .; Marshall, G. R.; Matassa, V. G.
5-(Piperidin-2-yl)- and 5-(Homopiperidin-2-yl)-1,4-benzodiaze-
pines: High-Affinity, Basic Ligands for the Cholecystokinin-B
Receptor. J . Med. Chem. 1997, 40, 2491-2501.
(18) Sattler, A.; Haufe, G. High Regioselectivity in the Alternative
Cleavage of Terminal Epoxides with Different Sources of Nu-
cleophilic Fluoride. J . Fluorine Chem. 1994, 69, 185-190.
(19) It is noteworthy that the intermediate primary ozonide was quite
resistant to reductive decomposition under the usual dimethyl
sulfide conditions. However, this could be cleanly and rapidly
achieved by treatment with triethylamine at low temperature,
which follows a different mechanistic pathway: Hon, Y.-S.; Lin,
S.-W.; Chen, Y.-J . A Convenient and Efficient Workup of
Ozonolysis Reactions Using Triethylamine. Synth. Commun.
1993, 23, 1543-1553.
(20) Sternfeld, F.; Baker, R.; Broughton, H. B.; Guiblin, A. R.; J elley,
R. A.; Matassa, V. G.; Reeve, A. J .; Beer, M. S.; Stanton, J . A.;
Hargreaves, R. J .; Shepheard, S. L.; Longmore, J .; Razzaque,
Z.; Graham, M. I.; Sohal, B.; Street, L. J . The Chemical Evolution
of N,N-Dimethyl-2-[5-(1,2,4-triazol-4-yl)-1H-indol-3-yl]ethylamine
(L-741,604) and Analogues: Potent and Selective Agonists for
5-HT1D Receptors. Bioorg. Med. Chem. Lett. 1996, 6, 1825-1830.
(5) Martin, G. R. Inhibition of the Trigemino-Vascular System with
5-HT1D Agonist Drugs: Selectively Targetting Additional Sites
of Action. Eur. Neurol. 1996, 36, 13-18.
(6) Coronary vasoconstriction is principally mediated through ac-
tivation of 5-HT2A receptors, although a small 5-HT1B component
is also present. Because the triptans show good selectivity over
5-HT2A receptors, the potential for coronary vasoconstriction with
these agents at clinically relevant antimigraine doses is very
low. (a) Ishida, T.; Sakoda, T.; Hirata, K.; Inoue, N.; Kawashima,
S.; Yokoyama, M. Molecular Analysis of Serotonin (5-Hydrox-
ytryptamine; 5-HT) Receptor Subtypes in Human Coronary
Arteries. J pn. Circ. J . 1997, 61, 621. (b) Ottervanger, J . P.;
Stricker, B. H. Ch. Cardiovascular Adverse Reactions to Sumatrip-
tan. Cause for Concern? CNS Drugs 1995, 3, 90-98. (c) Ferro,
A.; Longmore, J .; Hill, R. G.; Brown, M. J . A Comparison of the
Contractile Effects of 5-Hydroxytryptamine, Sumatriptan and
MK-462 on Human Coronary Artery in vitro. Br. J . Clin.
Pharmacol. 1995, 40, 245-251. (d) Longmore, J .; Boulanger, C.
M.; Desta, B.; Hill, R. G.; Schofield, W. N.; Taylor, A. A. 5-HT1D
Receptor Agonists and Human Coronary Artery Reactivity in
vitro: Crossover Comparisons of 5-HT and Sumatriptan with
Rizatriptan and L-741, 519. Br. J . Clin. Pharmacol. 1996, 42,
431-441.
(7) Castro, J . L.; Street, L. J .; Guiblin, A. R.; J elley, R. A.; Russell,
M. G. N.; Sternfeld, F.; Beer, M. S.; Stanton, J . A.; Matassa, V.
G. 3-[2-(Pyrrolydin-1-yl)ethyl]indoles and 3-[3-(Piperidin-1-yl)-
propyl]indoles: Agonists for the h5-HT1D Receptor with High
Selectivity over the h5-HT1B Subtype. J . Med. Chem. 1997, 40,
3497-3500.
(8) MacLeod, A. M.; Street, L. J .; Reeve, A. J .; J elley, R. A.;
Sternfeld, F.; Beer, M. S.; Stanton, J . A.; Watt, A. P.; Rathbone,
D.; Matassa, V. G. Selective, Orally Active 5-HT1D Receptor
Agonists as Potential Antimigraine Agents. J . Med. Chem. 1997,
40, 3501-3503.
(9) (a) Hillgren, K. M.; Kato, A.; Borchardt, R. T. In Vitro Systems
for Studying Intestinal Drug Absorption. Med. Res. Rev. 1995,
15, 83-109. (b) Brayden, D. Human Intestinal Epithelia Cell
Monolayers as Prescreens for Oral Drug Absorption. Pharm.
News 1997, 3, 11-15. (c) Stewart, B. H.; Chan, O. H.; J ezyk,
N.; Fleisher, B. H. Discrimination Between Drug Candidates
Using Models for Evaluation of Intestinal Absorption. Adv. Drug.
Delivery Rev. 1997, 23, 27-45.
J M980204E