Comparison of Gastrointestinal pH in Dogs and Humans:
`Implications on the Use of the Beagle Dog as a Model for
`Oral Absorption in Humans
`
`CHUNG Y. Lur, GORDON L. AMIDON', ROSEMARY R. BERARDI, D. FLEISHER,
`CAROLE YOUNGBERG, AND JENNIFER B. DRESSMAN
`Received April 17, 1985, from the College of Pharmacy, The University of Michigan, Ann Arbor, MI 48109-1065.
`December 16, 1985.
`
`Accepted for publication
`
`Abbtract 0 Gastrointestinal pH as a function of time was recorded for 4
`beagle dogs and 10 human subjects using radiotelemetric pH measur-
`ing equipment. Results indicated that in the quiescent phase, gastric pH
`in the dogs (mean = 1.8 f 0.07 SEM) was significantly (p < 0.05) higher
`than in humans (1.1 f 0.15). No significant difference in the time for the
`pH monitoring device to empty from the stomach was noted for the two
`species (99.8 f 27.2 min for dogs, 59.7 f 14.8 min for humans, p >
`0.05). The fasting intestinal pH in dogs was consistently higher than in
`humans, with an average canine intestinal pH of 7.3 f 0.09 versus 6.0
`2 0.14 for humans. The implication of these observations for extrapola-
`tion of drug absorption data from dogs to humans are discussed.
`
`Although it is recognized that there are many physiologi-
`cal differences between animals and humans, new drugs and
`dosage forms are frequently tested in animals to determine
`oral bioavailability. Dogs provide a particularly convenient
`animal model for testing oral dosage forms in terms of ability
`to ingest human-scale dosage forms and ease of husbandry.
`In many instances, however, bioavailability in dogs after oral
`administration of a drug differs considerably from bioavail-
`ability in humans.' Consideration of differences in gross
`physiology, e.g., motility, pH, and surface area, between the
`GI tracts of the two species may provide an explanation for
`some of the differences observed in drug absorption. In
`addition, it should enable us to make more rational extrapo-
`lations from canine absorption data to predict oral bioavail-
`ability in humans.
`The physiological parameter chosen for this study was the
`fasted-state GI pH profile. Variation in GI pH in the fasted
`state could affect drug dissolution rate and the fraction
`present in the un-ionized form, which in turn would influence
`the rate and extent of drug absorption. The fasted-state pH is
`of particular interest since most bioavailability studies are
`conducted under fasting conditions.
`In this study, pH was continuously recorded using a ra-
`diotelemetric monitoring system, the Heidelberg capsule.2.3
`Capsules were allowed to traverse the GI tract under the
`normal influences of motility, enabling pH data to be collect-
`ed in the stomach and small intestine over a 5-h period.
`Implications of similarities and differences in the GI pH
`profiles of the two species with respect to drug absorption are
`discussed using several case examples.
`
`Experimental Section
`radiotelemetric
`Radiotelemetric pH-Monitoring System-The
`technique (Electro-Medical Devices, Inc., Norcroes, GA) for GI pH
`monitoring has been described previously.2.3 Briefly, the system
`consists of a Heidelberg capsule (a nonreusable radio frequency
`signal emitter), an antenna, and a recorder. A Heidelberg capsule
`was calibrated using pH 1 and 7 standard buffer solutions before
`each experiment.
`Heidelberg Capsules-Five capsules were evaluated in vitro in
`the pH range of 1-10. A calibrated capsule was placed in a pH 7
`
`standard buffer solution and the buffer solution was titrated by the
`addition of a pH 1 or 10 buffer solution to give various pH values. The
`pH measured by the radiotelemetric system (digitized through the
`use of an analogue/digital converter (Adalab, Interactive Microware,
`Inc., State College, PA) was then compared to that of a pH meter
`(model pHM61, The London Co., Westlake, OH).
`healthy 2-3 year old male beagle dogs
`Animal Studies-Four
`weighing between 10 and 20 kg were studied. The dogs were
`maintained on a Purina Laboratory canine diet and cared for by
`University Laboratory Animal Medicine personnel. Each dog was
`fasted for 12 to 18 h prior to each study day. Each study morning, a
`calibrated Heidelberg capsule was administered with 20 to 50 mL of
`water. The dog stood quietly in a sling for the ensuing 5 h, with GI
`pH being continuously monitored. Dogs were studied on three
`occasions, with a t least a week separating study days. The GI
`pWtime profiles were analyzed for gastric pH, gastric emptying time
`(defined as the time elapsed between time of capsule administration
`and the time at which sustained elevation in the pH was observed),
`and intestinal pH.
`human studies were carried out in the
`Human Studie-The
`Clinical Research Center of The University of Michigan Hospitals,
`after approval by the Human Subjects Review Committee. All
`participants gave written informed consent. Five female and five
`male adults were selected for participation. Each was within 10% of
`normal weight range, as defined by Metropolitan Life Actuarial
`Tables. The mean age was 24 2 2.2 years, with a range from 21 to 29
`years. None of the participants had historical, clinical, or laboratory
`evidence of GI disease, nor were any receiving medication on a
`chronic basis. All were nonsmokers.
`Participants refrained from all medication for 3 d prior to the
`study, and from alcohol and caffeine for 24 h prior to and during the
`study. After fasting overnight, each subject swallowed a calibrated
`Heidelberg capsule. As in the canine studies, the capsule was
`permitted to traverse the GI tract under normal motility. The pH
`was recorded for a minimum of 5 h aRer the capsule was swallowed,
`during which time the subject sat quietly. Three hours after gastric
`emptying of the capsule, the subject was fed two donuts and 8 oz of
`decaffeinated soda.
`a separate study with human volunteers, the
`Validation-In
`capsule was administered after tethering it with surgical thread.
`The tether thread was used to position the capsule in the duodenum.
`This enabled us to record duodenal pH for 30 min prior to and 4 h
`after a meal. The capsule was then recovered, and the pH response to
`pH 1 and 7 buffers was checked against precalibration values.
`Twenty-two of 23 capsules so tested remained within 0.5 pH units of
`the original values recorded.
`Intestinal Absorption of Aspirin--Rat
`intestinal perfusion ex-
`periments a t four different pHs, 4.8, 6.5, 7.3, and 8.0, were carried
`out as previously described.a Isoosmotic citrate phosphate (pH 4.8
`and 6.5) and Tris (pH 7.3 and 8.0) buffers were used. Inlet (C,) and
`exit (C,) concentrations of the perfused intestinal segment were
`measured by an HPLC method previously reported.' Experiments
`were carried out with the aspirin concentrations a t 50 pg/mL. The
`dimensionless permeability of the intestinal wall, P:, was calculat-
`ed from the concentration ratio C,,,/C,, length of perfused intestine,
`flow rate, and aqueous diffusity of aspirin as described previously,4.6
`and results were reported as the average of data from two to four
`rats.
`regression was employed to
`Statistical Analysis-Polynomial
`determine the correlation between the pH measured by the pH meter
`
`OO22-3549/86/03oO-O271$Ol .OO/O
`0 1986, American Pharmaceutical Association
`
`Journal of Pharmaceutical Sciences / 271
`Vol. 75, No. 3, March 1986
`
`Grün. Exhibit 1078
`Grünenthal v. Antecip
`PGR2017-00022
`
`

`

`and radiotelemetric system. The regression analysis was performed
`using an HP 87 microcomputer. The standard errors and deviations
`for the regressions were estimated according to Draper and Smith.a
`Equality of variances and statistical differences in gaetric emptying
`time and GI pH between dogs and humans were tested by the F and t
`tests, respectively.
`Results and Discussion
`Figure 1 shows the results of the in vitro evaluation of the
`Heidelberg capsules in the pH range of 1-8. The result of
`polynomial regression analysis (Table I) shows that a polyno-
`mial of second degree gives the best fit. However, the
`regression fit for the linear model is comparable (Table I and
`Fig. 1). Since the quadratic model is only marginally better
`than the linear model, the latter correlation was adopted. For
`pH values of approximately 2.5-6.5, the capsules tended to
`underestimate the pH of the buffer solution and from pH 7 to
`8 the readings were slightly overestimated. However, these
`under- and overestimations of pH did not exceed 20.5 units.
`When the capsules were tested at a pH range of 9-10 (not
`shown in Fig. l), the radiotelemetric system consistently
`registered lower pH values (-0.5 units or more). All of these
`observations were within the error specified by the manufac-
`turer, i.e., 50.5 pH units in the pH 1-8 range. These results
`indicate that the GI pH values measured in this study are
`accurate to within 0.5 pH units.
`Table I1 summarizes two of the important GI parameters
`(gastric pH and emptying time) for both dogs and humans.
`When gastric emptying time was tested for statistical signifi-
`cance between dogs and humans, no significant difference (p
`> 0.05) was found. It should be noted however, that the
`gastric emptying time in dogs was somewhat longer and
`exhibited a wider range. This is reflected by the unequal
`variances between dogs and humans. The large variances
`observed can be explained by the fasting motility pattern.
`
`9r
`
`/. 9
`
`1
`
`2
`
`3
`
`5
`4
`pH Meter
`Flgure 1-Regression plots of the pH of buffer solutions measured by a
`radiofelemetric sysfem versus a pH meter. The 45" line (---) demon-
`strates the under- and overestimation of PH.
`
`6
`
`7
`
`8
`
`I
`
`9
`
`
`
`~
`
`~
`
`Table I-Results of Polynornlal Regresslon Analyslr for In Vltro
`Evaluatlon of the Heldelberg Capsule In pH 1-8
`Coefficients'
`so
`4 P
`A0
`Model
`A,
`-
`1.0328
`0.991 0.224
`- -
`Linear
`-0.3897
`(0.0285) -
`(0.1361 )'
`0.996 0.165
`Quadratic
`0.7017
`0.0378
`0.1724
`(0.1734) (0.0884) (0.0098) - -
`.y = 4 + A,X + 4 X 2 . "Number in parentheses is the SEM
`
`272 / Journal of fharmaceufical Sciences
`Vol. 75, No. 3, March 1986
`
`Table II-The Mean f SEM and Range of Gaslrlc pH and
`Emptylng T h e of 4 Dogs and 10 Human Subjects
`Humans
`fTest FTest
`Dogs
`1.1 5 0.15' - " E.V.d
`Gastric pH
`1.8 2 0.07'
`(0.4-4 .O)
`(0.5-3.9)
`U.E.V.8
`Gastric emptying 99.8 27.2'
`59.7 ? 14.8'
`NS
`lime, min
`(16.0-137)
`(35.0-31 7)
`'n = 91. 'n = 30. "Significant at p < 0.05. dE.V.. equal variances at
`p > 0.05. 'n = 12. 'n = 10. BU.E.V., unequal variances at p < 0.05.
`The Heidelberg capsule is emptied from the stomach during
`Phase I11 contractions, which occur temporally on a cyclic
`basis.g Random timing of capsule administration relative to
`the next Phase I11 contraction results in a wide range of
`capsule emptying times. It should also be noted that since the
`capsule remains in the stomach only as long as the Phase I
`and I1 motility patterns prevail, the pH recorded pertains
`chiefly to the quiescent phase of gastric motility. There was a
`statistically significant difference (p < 0.05) in quiescent
`gastric pH between dogs and humans. While the gastric pH
`in dogs was higher and exhibited a wider range, the vari-
`ances were not significantly different (Table 11). The gastric
`pH values measured in this study are comparable to the
`values obtained by other techniques.lO,ll
`Figures 2 and 3 present the mean and range of the GI pH-
`time profiles of humans and dogs, respectively. The mean
`profiles were computed by normalizing the gastric emptying
`time of any given experiment to 99.8 min for dogs and 59.7
`rnin for humans. In humans, the intestinal pH was initially
`high, decreased to a minimum pH of 5.5 at 60 min following
`gastric emptying, then gradually returned to a pH of -6.5 at
`150 min after emptying. The overall mean intestinal pH was
`6.0 2 0.14 and was comparable to the reported values in the
`As can be seen in Fig. 2, the range in intesti-
`l i t e r a t ~ r e . ' ~ . ~ ~
`nal pH measured was widest (4-7.2) immediately &r
`gastric emptying. This is probably due to variable pancreatic
`bicarbonate secretion associated with the Phase I11 contrac-
`tions that move the capsule out of the stomach. After the
`initial 60 min in the small intestine, the range of intestinal
`pH measured became relatively small (+1.0 units). In con-
`trast, the intestinal pH-time profile of dogs (Fig. 3) increased
`to a maximum (pH = 7.7) at 120 min after gastric emptying
`and gradually declined thereafter (pH = 7.2 at 180 min after
`emptying). The overall mean intestinal pH was 7.3 5 0.09,
`
`similar to values reported in the l i t e r a t ~ r e . ~ J ~ ~ * ~ 5 The range
`in intestinal pH values measured for dogs was similar
`throughout the entire period of observation in the small
`intestine (51.5 units). The F test statistics indicate that
`variance of intestinal pH in dogs is greater than in humans.
`For comparison purposes, the intestinal pH-time profiles in
`humans and dogs are shown in Fig. 4. Notice that the
`intestinal pH of the dogs was always higher than in humans,
`by as much as 1.9 pH units, and always at least 0.7 pH units
`at comparable times following gastric emptying. When the
`intestinal pH values at each time point were compared for
`the two species using a method for unequal variances, the
`canine pH was significantly higher (p < 0.05) at all times
`except at initial gastric emptying. The differences between
`canine and human GI pH profiles are consistent with report-
`ed values for gastric acid secretion and pancreatic bicarbon-
`ate secretion in the two species. Basal gastric acid secretion
`is lower in dogs than in humans, while secretin-stimulated
`pancreatic bicarbonate secretion is higher and more vari-
`able.16 Consequently, it is reasonable to expect higher gastric
`and intestinal pH values and to expect the intestinal pH to be
`more variable in dogs than in humans.
`The GI pH profiles observed suggest that absorption would
`not vary between dogs and humans as a result of pH for many
`
`

`

`..........
`
`'I
`
`8
`
`8 ,
`7 -
`6.
`
`2 5 -
`H
`
`4:
`
` .
`.
`.
`.
`.
`.
`.
`.
`...
`
`.
`-
`I
`0 30 60 90 120 150 180 210 240 270 300 330
`Minutes PcstCapsule Ingesm
`Flgure 2--The mean (-)
`and range (---) of the GI pK-lime profile of
`humans (see text).
`
`a
`
`
`
`1
`30 0 30 60 90
`120 150 180 210 240 270
`Minutes PostGastnc Emptying
`Figure &The mean and SEM of the intestinal pH-time profile of
`dogs and (0) humans.
`
`...........
`................
`......
`............ .........
`... ........
`
`F 6'
`g 5 '
`i ::
`
`1
`
`............ ...... ......
`O 2 ,
`\
`... ...
`..............
`120 150 180 210 240 270 300 33
`0 r) 60 90
`Minutes PcstCtWEuk 1-
`
`3
`
`Flgure 3-7778 mean (-)
`dogs (see rext).
`
`and range (---) of the GI pWirne profile of
`
`drugs. However, in certain cases, where the pH of half-
`maximal absorption'' falls into the pH 5-8 range, andlor the
`PK, of a poorly soluble drug falls in this pH range, there may
`be a substantial discrepancy between fractions absorbed in
`dogs and in humans. The effect of luminal pH on intestinal
`permeability (Pc) has been documented for several
`dn1gs,*.~.*7.'R and this intestinal permeability can be directly
`related to the absorption rate constant.63'9 For drugs which
`are incompletely absorbed, a change of the intestinal perme-
`ability resulting from a change in intestinal pH would be
`expected to translate into an alteration in the rate of absorp-
`tion. Aspirin provides an example of a drug which has pH
`dependent permeability. Figure 5 illustrates the bulk phase
`pH effect on the P*, of aspirin. A change of bulk phase pH
`from 7.3 to 6.5 gives rise to more than a two-fold increase in
`Pw In the absence of other physiological and drug-related
`parameters (first-pass metabolism, transit through the GI
`tract, intestinal blood flow, dose and solubility, etc.) there
`may be more than a two-fold increase in the rate of aspirin
`absorption. Although the pH at the mucosal surface is also a
`factor in drug absorption, it is obvious that the P: of aspirin
`is bulk pH dependent. This observation suggests that other
`weakly acidic and basic compounds may have similar pH-
`permeability profiles and pH-dependent fraction absorbed
`values, depending on their PK, values and permeabilities.
`Due to the more than 1 pH unit difference in intestinal pH
`between dogs and humans, the potential exista for a differ-
`ence in absorption of weakly acidic and basic compounds in
`these two species. A literature search, however, revealed that
`there is limited information available providing a direct
`
`i
`
`:t.
`
`'
`'
`'
`'
`4.5 5.0 5.5 6.0 6 5 7.0 7 5 8.0 8 5 90
`Perfuson pH
`Figure 5-Plot of mean jejunal wallpermeability of raf versus perfusion
`pH. Vertical bar is the SEM.
`
`comparison of drug absorption in humans and dogs. Table I11
`shows the bioavailabilities of some compounds in these two
`species. As can be seen in Table 111, there are differences in
`the systemic availability of aspirin,*o,21 chlorothiazide,22.23
`and mefenamic acid2* between dogs and humans, while the
`
`bioavailability of c l o n a ~ e p a m z ~ . ~ ~ is similar in these two
`species. The bioavailability of chlorothiazide was estimated
`from urine recovery of intact drug while those of aspirin and
`clonazepam were obtained from the ratio of intravenous
`versus oral area under the plasma concentration-time
`curves. The percent of mefenamic acid absorbed was estimat-
`ed from the total amount of drug eliminated into the urine,
`bile, and stool. For clonazepam, the similar bioavailability in
`dogs and humans is expected, since it is a nonpolar low dose
`compound and its pK, values (1.5 and 10.5) are not in the
`physiological range. For mefenamic acid, the permeability-
`pH profiles would be expected to be similar to that of aspirin
`(Fig. 5). Consequently, the fraction absorbed in dogs would be
`expected to be less than (or at best equal to) that absorbed in
`humans, consistent with the observed bioavailability results.
`For aspirin, the observed lower bioavailability in dogs is
`consistent with its permeability-pH profile, Fig. 5. However,
`the explanation for the difference in bioavailability of mefen-
`amic acid and aspirin may also be partially due to higher
`first-pass metabolism in dogs. For chlorothiazide, the ob-
`served result is the reverse of what would be expected based
`on the expected permeability-pH profile for this compound.
`However, a previous simulation of the bioavailability of
`chlorothiazidel* suggests that the limited and dose-depen-
`dent bioavailability is due to solubility-ionization-perme-
`
`Journal of Pharmaceutical Sciences / 273
`Vol. 75, No. 3, March 1986
`
`

`

`9. Weisbrodt, N. W. “Ph siology of the Gastrointestinal Tract’’;
`Raven Press: New Yo&, 1981; p 411 438.
`10. Bueno, L.; Fioramonti, J.; RuckeguschrY. Physiology 1981,316,
`319322.
`11. Mann, F. C.; Bollman, J. L. J. Am. Med. Assoc. 1930,95, 1722-
`* rrnr
`11LW.
`12. Aynaciyan, A. V.; Bingham, J. R. Gastroenterology 1969, 56,
`476-482.
`- . . - - -.
`13. Dutta, Sudhir K.; Russell, Robert, M.; Iber, Frank L. Dig. Dis.
`Sci. 1979,24, 529534.
`14. Brooks, A. M.; Grossman, M. I. Gastroenterology 1970, 59, 90-
`OA
`15. hTown, J. C.; Johnson, L. P.; Magee, D. F. Gastroenterology
`1966,50,333337.
`16. Anderson, N. V. “Veterinary Gastroenterology”; Lea & Febiger:
`17. Winne, 6. J. Pharmacokinet. Blopharm. 1977,5, 53-94.
`Philadel hia, PA, 1980; pp 247:262.
`18. Dressman, J. B.; Fleisher, D.; Amidon, G. L. J. Pharm. Sci. 1984,
`7.?. 1274-1 279.
`- - , -
`19. Amidon, G. L.; Topp, E. “Abstracts”, 39th National Meetin
`the
`Academy of Pharmaceutical Sciences, vol. 15; Oct. 1985, &he-
`apolis, MN; American Pharmaceutical Association: Washington,
`DC. 1985: D 6.
`20. Rowland ’ h.; Riegelman, S.; Harris, P. A.; Sholkoff, S. D. J.
`Pharm. dci. 1972, 61, 379-385.
`21. Hams, P. A.; Riegelman, S. J. Pharm. Sci. 1969, 58, 71-75.
`22. Shah, V. P.; Lee, J.; Hunt, J. P.; Prasad, V. K.; Cabana, B. E.;
`Foster. T. Curr. Ther. Res. 1981.29. 823-827.
`23. Resetarits, D. E.; Bates, T. R. ‘J. Pharmacokinet. Bwpharm.
`1979, 7,463-470.
`24. Glazko, A. J. Ann. Phys. Med. Su
`1966,8, 23-26.
`25. Berlin, A.; Dahlstrom, H. Eur. J. k%n. P h a r m o l . 1975,9,155-
`159.
`26. Kaplan, S. A.; Alexander, K.; Jack, M. L.; Puglisi, C. V.; de
`Silva, J. A. F.; Lee, T. L.; Weinfeld, R. E. J. Phurm. Sci. 1974,
`63,527-532.
`
`Acknowledgments
`This work was supported by the SmithKline Beckman Corpora-
`tion. The technical assistance of Mr. John Wlodyga is gratefully
`acknowledged.
`
`Table llCBlosvallablllty (F) of Some Compounds Estlmated In
`Dogs and Humans
`
`F, Yo
`Humans
`
`Dogs
`
`~
`
`~~~~~~~~
`
`~
`
`PK,
`
`~
`
`~
`
`Acid
`
`Base
`Compound
`-
`Aspirin”
`45
`3.5
`68
`-
`Chlorothiazide
`26-70
`15-35
`6.7
`Clonazepam‘
`100
`-
`1.5
`10.5
`98
`Mefenamic acidd
`35-60
`4.2
`80-90
`” Refs. 20 and 21. Dose dependent; refs. 18, 22, and 23. ‘ Refs. 25
`and 26. dRef. 24.
`
`1.
`
`ability effects. Given the results of the present study, a
`possible explanation of the results of the doghuman chlor-
`othiazide bioavailability comparison is that the reduced
`bioavailability in humans is due to their lower intestinal pH,
`leading to incomplete dissolution at pH values below the pK,.
`This is also consistent with the explanation of observed dose
`dependency in humans. ‘ 8
`References and Notes
`Drug Delivery in Man: fi Situ and In Vivo AP roaches”;
`Crouthamel, William; Sara u, Allen C. “Animal Models for Oral
`American Pharmaceutical Association: Washington, k 1980
`Dressman, J. B.; Amidon, G. L. J. Pharm. Sci. 1984, 73, 9351
`938.
`Am. f Vet. Res. 1985,46, 1516-1521.
`Youn berg, C. A.; Wlodyga, J.; Schmaltz, S.; Dressman, J. B.
`Elliott, R. L.; Amidon, G. L.; Lightfoot, E. N. J. Theor. Bwl.
`1980,87, 757-771.
`Amidon, G. L.; Leesman, G. D.; Elliott, R. L. J. Pharm. Sci.
`1980,69, 1363-1368.
`Amidon, G. L.; Chang, M.; Fleisher, D.; Allen, R. J. Pharm. Sci.
`6.
`1982, 71, 1138-1141.
`I.
`Bakar, S. K.; Niazi, S. J. Pharm. Sci. 1983, 72, 1020-1023.
`8.
`Draper, N. F.; Smith, H. “Applied Regression Analysis”; John
`Wiley & Sons: New York, 1966; pp 13-26.
`
`2.
`
`3.
`
`4.
`
`5.
`
`274 / Journal of Pharmaceutical Sciences
`Vol. 75, No. 3, March 1986
`
`