Advertisement

The AAPS Journal

, 20:107 | Cite as

Tissue Physiology of Cynomolgus Monkeys: Cross-Species Comparison and Implications for Translational Pharmacology

  • Danielle Mandikian
  • Isabel Figueroa
  • Amy Oldendorp
  • Hanine Rafidi
  • Sheila Ulufatu
  • Michelle G. Schweiger
  • Jessica A. Couch
  • Noel Dybdal
  • Sean B. Joseph
  • Saileta Prabhu
  • Gregory Z. Ferl
  • C. Andrew Boswell
Open Access
Research Article
  • 522 Downloads

Abstract

We previously performed a comparative assessment of tissue-level vascular physiological parameters in mice and rats, two of the most commonly utilized species in translational drug development. The present work extends this effort to non-human primates by measuring tissue- and organ-level vascular volumes (Vv), interstitial volumes (Vi), and blood flow rates (Q) in cynomolgus monkeys. These measurements were accomplished by red blood cell labeling, extracellular marker infusion, and rubidium chloride bolus distribution, respectively, the same methods used in previous rodent measurements. In addition, whole-body blood volumes (BV) were determined across species. The results demonstrate that Vv, Vi, and Q, measured using our methods scale approximately by body weight across mouse, rat, and monkey in the tissues considered here, where allometric analysis allowed extrapolation to human parameters. Significant differences were observed between the values determined in this study and those reported in the literature, including Vv in muscle, brain, and skin and Q in muscle, adipose, heart, thymus, and spleen. The impact of these differences for selected tissues was evaluated via sensitivity analysis using a physiologically based pharmacokinetic model. The blood-brain barrier in monkeys was shown to be more impervious to an infused radioactive tracer, indium-111-pentetate, than in mice or rats. The body weight-normalized total BV measured in monkey agreed well with previously measured value in rats but was lower than that in mice. These findings have important implications for the common practice of scaling physiological parameters from rodents to primates in translational pharmacology.

KEY WORDS

allometric scaling non-human primates physiologically based pharmacokinetic modeling tissue physiological parameters translational pharmacology 

INTRODUCTION

The scaling of various physiological parameters across species has been performed for several decades, largely for the purpose of translational pharmacology [1, 2, 3, 4, 5, 6]. Knowledge of physiological parameters is required for derivation of drug concentrations within compartments such as interstitial fluid (often the drug site of action, particularly for biologics) [7, 8, 9] and is an important component of many physiologically based pharmacokinetic (PBPK) models [10, 11, 12, 13, 14, 15, 16, 17, 18]. Human tissue/organ-level physiological parameters are often derived by allometric scaling of corresponding animal data ( [5, 19, 20]). For biologic drugs with minimal cell partitioning (e.g., antibodies), blood correction of whole tissue concentrations and subsequent conversion into interstitial fluid concentrations requires knowledge of the fractional composition of vascular and interstitial spaces, respectively, within each tissue [7, 8]. We previously reported measurement of tissue-level vascular volumes (Vv), interstitial volumes (Vi), and rates of blood flow (Q) in several immunocompromised mouse strains [21, 22, 23, 24, 25]. More recently, we expanded our investigations to include immunocompetent mice [26, 27] and rats [28], but comparable experimental data in primates remains unavailable for some parameters (e.g., Vi) and is reported only in limited tissues for others. The cynomolgus monkey is often the nonclinical species of choice, particularly in antibody development, for a number of reasons: the higher degree of target homology with human makes interspecies cross-reactivity more likely, the increased pharmacological and toxicological relevance in cases where target biology is similar to human, and the availability of robust, highly validated methods for predicting clinical pharmacokinetics based on monkey studies. Additionally, generating more accurate and methodologically consistent physiological parameter values will improve preclinical to clinical allometric scaling, prospective clinical PK/PD simulations, and other related calculations (e.g., predicting target-mediated drug disposition based on observed/assumed volumes of distribution), potentially resulting in decreased animal usage. Herein, we address this gap by measuring the same three parameters in cynomolgus monkeys using identical experimental approaches as were used in rodents [8, 28]. In addition, total blood volumes (BV) were calculated for monkey and rodents (using current and historical data, respectively) to gain a further quantitative understanding of vascular physiological parameters across these species commonly used in drug development.

Accurate measurement of Vv (i.e., residual blood volume) in tissues requires the use of very large molecules that are unable to extravasate (Fig. 1). Radiolabeled albumin or immunoglobulin uptake should not be used to estimate Vv as these molecules are known to exhibit transudation (i.e., extravasation) into interstitial spaces of tissues even at early time points [29]. Consequently, we selected a red blood cell (RBC) radiolabeling method that relies on stannous ion (Sn+2)-mediated reduction of technetium [30]. After priming RBCs with Sn+2 (often termed pre-tinning), technetium in the form of pertechnetate ion (99mTcO4−) is introduced and becomes reduced by Sn+2 leading to stable binding to the beta chain of hemoglobin with high labeling efficiency. A similar protocol is used for clinical blood pool imaging [31] and has been back-translated for measuring Vv in mouse xenograft models [32]. However, the portion of 99mTcO4− that does not bind hemoglobin can skew Vv measurements in tissues since extravasation of unbound 99mTcO4− into tissues exaggerates the number of 99mTc-labeled RBCs present. To address this issue, we modified the clinical in vivo procedure for quantitative Vv measurement in rodent tissues using a transfusion approach in which pre-tinning, 99mTcO4− labeling (at high specific activity), and purification (via extravasation and/or renal filtration) occur in a donor animal [21, 22, 26, 28]. The 99mTc-labeled blood from this donor can be subsequently collected at euthanasia and carefully reinjected into naïve animals for Vv determination. In addition to providing Vv measurements at the tissue level, data from RBC labeling can also be used to calculate hematocrit [33] and total BV [34, 35].
Fig. 1

Schematic of compartmental tissue measurements. Conceptual illustration of three physiological parameters—vascular volume (Vv), interstitial volume (Vi), and blood flow (Q). Red blood cells (RBCs) labeled with 99mTc are restricted to the vascular space (Vv). The intravenously infused extracellular marker, 111In-DTPA, reaches equilibrium between plasma and interstitial fluid (Vi). The intravenous bolus dose of the blood flow (Q) probe 86Rb+ quickly extravasates and is sequestered into cells via sodium/potassium ion transporters

Historically, measurement of Vi has relied on the use of several small molecules as extracellular tracers, including sucrose, inulin, chloride, sulfate, and the radiometal-polyaminopolycarboxylate complex chromium-51-ethylenediaminetetraacetic acid (51Cr-EDTA) [36, 37]. More recently, preclinical measurement of interstitial volume in tumors has been accomplished by indium-111-labeled diethylenetriamine pentaacetate (i.e., 111In-pentetate or 111In-DTPA) [38, 39]. Such measurements rely on two assumptions: (1) tracer is excluded from all living cells (an appropriate assumption for a polar, charged molecule like 111In-DTPA) and (2) tracer is in steady-state equilibrium between plasma and interstitial fluid at time of measurement (Fig. 1). In preclinical studies, others have either suppressed renal filtration (nephrectomy, renal ligature) or applied a constant infusion of tracer in order to maintain constant plasma levels and accurately derive Vi values in selected tissues [37]. We previously employed constant intravenous infusion of 111In-DTPA followed by prompt euthanasia and tissue harvest [21, 26, 28]. This approach allows us to measure Vi in most tissues except brain and eye (because 111In-DTPA is excluded) and kidneys (because 111In-DTPA is excreted).

Measurement of Q (blood flow rate) utilizes certain tracers such as the noble gas xenon (Xe) [40] and alkali metal ions (K+/Rb+) [41] that are completely extracted by tissues from circulating blood on first pass, such that extravascular tracer levels in each tissue shortly after tracer administration can be used to approximate blood flow rate to that tissue as a fraction of cardiac output (Fig. 1). This rapid extraction phenomenon has been exploited to measure Q in tumors and other tissues preclinically [32, 42, 43, 44], with further efforts for clinical translation via 81Rb positron emission tomographic imaging in patients [45]. The absence of significant Rb+ washout following extraction is due, in part, to its active transport into cells via Na-K pumps (i.e., Na+/K+-ATPases) [46], particularly in cardiac myocytes [47]. Consistent with this concept, the amount of measured 42K+ and 86Rb+ has been shown to be relatively constant from as early as 9 s to as late as 120 s post administration [41]. The two most common indicators for measuring regional blood flow rates (i.e., relative distribution of cardiac output to various tissues) are radiolabeled microspheres and the radioactive potassium ion analogue 86Rb+. A direct comparison of these two methods yielded largely similar results in rat [48]. Most preclinical reports on methods for measuring Q have been limited to xenograft tumors and a few selected tissues, although compilations of data across species are available [49]. However, the use of various techniques and/or strains has thereby motivated extension of this approach to measure Q in a much wider range of tissues in mice [26] and rats [28]; however, brain and eye should be excluded due to exclusion of 86Rb by the blood-brain and blood-ocular barriers [50].

Using the aforementioned techniques, we measured Vv, Vi, and Q in vivo across a wide range of tissues in mice, rats, and monkeys for the purpose of informing translational pharmacology as well as physiologically based pharmacokinetic (PBPK) modeling efforts. A recently published PBPK model [13] compiled a list of previously measured (or scaled/assumed) Vv, Vi, and Q values from a diverse set of published literature [17, 18, 51, 52, 53] for the species considered here; however, as is typical in PBPK modeling efforts, these values were obtained using a wide variety of experimental techniques or were scaled from other species in cases where measured values were unavailable. In some cases, nominal values (e.g., 10% vascular volume for well-perfused tissues) were assumed based on little to no experimental data for the specific organ of interest [17]. In this study, we aim to address (1) the limited availability of tissue-level parameters in primates [18], where scaled or nominal values are often used when measured values are unavailable and (2) reduce the variability in these physiologic values by measuring Vv, Vi, and Q in cynomolgus monkeys using standardized methods that we have also previously applied to rats [28] and mice [26].

MATERIALS AND METHODS

Ethical Approval

All procedures were conducted in accordance with the Animal Welfare Act, the Guide for the Care and Use of Laboratory Animals, the Office of Laboratory Animal Welfare, and the IACUC at the in-life study site. Care and housing were USDA regulatory compliant.

Animal Care

A total of eight drug-naïve cynomolgus monkeys (Chinese origin) ranging from 2 to 7 years of age and from 2 to 5 kg weight from Covance Research Products, Inc., (Alice, TX) or from a Covance stock colony were used. Tissues were harvested from six monkeys following euthanasia (three each for Vv/Q and Vi), while two additional monkeys served as blood donors without requiring euthanasia. Animals were allowed to commingle during the 1-week acclimation period (prior to administration of radionuclides), but were housed individually in stainless steel cages during the < 1-day test period. Certified Global Primate Diet No. 2055 (Harlan Teklad, Inc.) and fresh water (ad libitum) were provided and supplemented with appropriate fruits and cereals as treats. Cage enrichment devices were used for environmental and psychological enrichment. Environmental controls were set to maintain a temperature of 20 to 26°C, a relative humidity of 50% ± 20%, and a 12-h light/12-h dark cycle.

Derivation of Total Blood Volumes

The total blood volume of each monkey was calculated from the total radioactive dose in counts per minute (CPM) of 99mTc-fortified blood and the radioactive (99mTc) concentration in blood at time of euthanasia in CPM per microliter as follows:
$$ \mathrm{Total}\ \mathrm{blood}\ \mathrm{volume}=\frac{\mathrm{Total}\ \mathrm{dose}\ \left({\mathrm{CPM}}_{\mathrm{Tc}-99\mathrm{m}}\right)}{\frac{{\mathrm{CPM}}_{\mathrm{Tc}-99\mathrm{m}}}{\upmu \mathrm{L}\ \mathrm{blood}}} $$
(1)

The same approach was used to calculate total blood volumes of rats and mice from our previously reported studies [21, 23, 24, 25, 26, 27, 28], and blood volumes were normalized to body weight (kg) for comparison across species.

Measurement of Organ-Level Vascular Volumes and Rates of Blood Flow

The vascular spaces of monkey tissues were measured using a previously reported indirect red blood cell (RBC) labeling/transfusion method [22] based on protocols used in both clinical blood pool imaging [31] and preclinical studies [32, 54, 55]. Using a dual radionuclide detection strategy, regional blood flow rates (Q) in various organs and tissues were determined in the same animals by prompt euthanasia after intravenous bolus injection of 86Rb chloride (86RbCl) [32, 42, 43, 44, 45]. The RBC labeling method relies on tin-mediated reduction of 99mTc pertechnetate intracellularly so that it may bind to the beta chain of hemoglobin [30]. An appropriate number of animals were pre-screened to ensure donor blood compatibility, evident by a lack of antigen-antibody reaction resulting in clumping or hemagglutination in blood samples. TechneScan PYP™ (i.e., stannous pyrophosphate kits for preparation of 99mTc pyrophosphate injection) were obtained from a radiopharmacy and used according to label instructions for the in vivo method of blood pool imaging with the appropriate human-to-monkey scaling based on relative body weight. Each reconstituted clinical kit vial of TechneScan PYP™ contains 4 mg of tin chloride (three human doses) in 3 mL. Two donor cynomolgus monkeys received bolus IV injections of stannous pyrophosphate. Approximately 30 min later, blood (approximately 3 mL) was collected into tubes containing K2EDTA anticoagulant and maintained at ambient temperature. Approximately 7 mCi of 99mTc pertechnetate saline solution (approximately 165 μL) was added to each 3 mL sample of blood. The fortified samples were gently inverted to mix and incubated with gentle rotation at ambient temperature for approximately 20 min. Three study animals (3.3–4.3 kg) received a bolus IV injection of approximately 0.5 mL of 99mTc-fortified blood (~ 1 mCi) followed by 0.1 mL of 86Rb chloride (PerkinElmer, ~ 100 μCi) approximately 30 min later. Approximately 1 min after the 86RbCl dose, blood was collected into tubes containing K2EDTA anticoagulant and placed on wet ice prior to subsampling and centrifugation to obtain plasma. Animals were euthanized within 2.5 to 3.5 min after the 86RbCl dose via an overdose of sodium pentobarbital. Following blood collection, tissues were collected as described below for prompt radioanalysis of 99mTc (140 keV gamma energy, t1/2 ~ 6 h), followed by radioanalysis of 86Rb (1774 keV gamma energy, t1/2 ~ 18.6 days) subsequent to 99mTc decay. Vv was calculated from 99mTc counts as follows:
$$ {V}_{\mathrm{v}}\ \left(\frac{\upmu \mathrm{L}}{\mathrm{g}\ \mathrm{tissue}}\right)=\frac{\frac{{\mathrm{CPM}}_{\mathrm{Tc}-99\mathrm{m}}}{\mathrm{g}\ \mathrm{tissue}}}{\frac{{\mathrm{CPM}}_{\mathrm{Tc}-99\mathrm{m}}}{\upmu \mathrm{L}\ \mathrm{blood}}} $$
(2)
By converting to milliliters per gram (i.e., Vv / 1000), one can easily derive the fractional vascular volume (γ), a unitless value if one assumes that blood and tissue have equal densities. Importantly, drug concentrations in tissues (Ct) may be corrected for circulating drug in blood based on Vv using the following equation, where Cb (μg/μL) is the concentration of drug in whole blood:
$$ {C}_{\mathrm{t},\mathrm{blood}\ \mathrm{corrected}}\ \left(\frac{\upmu \mathrm{g}}{\mathrm{g}\ \mathrm{tissue}}\right)={C}_{\mathrm{t},\mathrm{total}}-{V}_{\mathrm{v}}\times {C}_{\mathrm{b}} $$
(3)
Rates of blood flow (Q) were calculated from 86Rb counts where total cardiac outputs (COtotal) are 8000, 74,000, and 884,000 μL/min for mice, rats, and monkeys, respectively [52], as follows:
$$ Q\ \left(\frac{\upmu \mathrm{L}}{\mathrm{g}\bullet \min}\right)=\frac{\frac{{\mathrm{CPM}}_{\mathrm{Rb}-86}}{\mathrm{g}\ \mathrm{tissue}}\times {\mathrm{CO}}_{\mathrm{total}}}{\mathrm{Total}\ \mathrm{Dose}\ \left({\mathrm{CPM}}_{\mathrm{Rb}-86}\right)} $$
(4)

The cardiac output for the monkeys weighing 3.8 kg (884,000 μL/min) was derived by allometrically scaling (using an exponent of 0.75) [20] the value previously reported for monkeys weighing 5 kg [52, 56]. Values for 20 and 250 g mice and rats, respectively, were used without scaling.

Measurement of Organ-Level Interstitial Volumes

The interstitial spaces of monkey tissues were measured by continuous infusion of the extracellular marker, 111In-DTPA [38, 39]. 111In-DTPA (clinical grade, GE Healthcare) was administered (~ 3 mCi, 11–12 mL) as an intravenous infusion over approximately 1 h. Three animals (2.7–5.2 kg), previously acclimated to the procedure, were restrained in chairs throughout dose administration. Upon completion of the infusion, blood (at least 5 mL, collected into a 6-mL vacutainer) was immediately (within 1 min after the end of infusion) collected into tubes containing K2EDTA anticoagulant. Blood samples were placed on wet ice prior to subsampling and centrifugation to obtain plasma. Animals were euthanized within 1.5 min after the end of infusion via overdose of sodium pentobarbital anesthesia. Tissues were harvested as described below followed by radioanalysis for 111In (171 and 245 keV gamma energies, t1/2 ~ 2.8 days). Vi was calculated from 111In counts and Vv values (see below) as follows:
$$ {V}_{\mathrm{i}}\ \left(\frac{\upmu \mathrm{L}}{\mathrm{g}\ \mathrm{tissue}}\right)=\frac{\frac{{\mathrm{CPM}}_{\mathrm{In}-111}}{\mathrm{g}\ \mathrm{tissue}}-\frac{{\mathrm{CPM}}_{\mathrm{In}-111}}{\upmu \mathrm{L}\ \mathrm{blood}}\times {V}_{\mathrm{v}}\ \left(\frac{\upmu \mathrm{L}}{\mathrm{g}\ \mathrm{tissue}}\right)}{\frac{{\mathrm{CPM}}_{\mathrm{In}-111}}{\upmu \mathrm{L}\ \mathrm{plasma}}} $$
(5)
The above equation assumes that 111In-DTPA is in steady-state equilibrium between plasma and tissue interstitial fluid at time of euthanasia and tissue harvest. Importantly, for drugs with little to no cell partitioning (e.g., antibodies), conversion of blood-corrected whole tissue drug concentrations (Ct,blood corrected) to interstitial concentrations (Ci) may be achieved using the following equation:
$$ {C}_{\mathrm{i}}\ \left(\frac{\upmu \mathrm{g}}{\upmu \mathrm{L}}\right)=\raisebox{1ex}{${C}_{\mathrm{t},\mathrm{blood}\ \mathrm{corrected}}$}\!\left/ \!\raisebox{-1ex}{${V}_{\mathrm{i}}$}\right. $$
(6)

Blood-Brain Barrier Permeability

Blood-corrected brain levels of 111In-DTPA, normalized to plasma levels, were compared across species as indicators of blood-brain barrier permeability. The calculation is analogous to the equation for Vi as described above but is designated as a brain partitioning coefficient (Kp) to highlight the lack of physiological relevance to brain Vi.

Tissue Handling and Analysis

Harvested tissues included brain, colon (without contents), eyes, fat (subcutaneous), heart, ileum (without contents), kidneys, liver, lungs, muscle (from the thigh), skin (dorsal), and spleen. Brain, heart, kidneys, and lungs were collected in entirety, weighed, divided into pieces, supplemented with reverse osmosis water (approximately twice the sample weight), homogenized in a blender, and divided into duplicate aliquots for analysis by solid scintillation (i.e., gamma) counting. Colon and ileum were collected in entirety, emptied of their contents, weighed, and divided into duplicate representative samples (~ 1 g each). Duplicate representative samples of fat, muscle, and skin were collected. Colon, eyes, fat, ileum, muscle, and skin samples were not homogenized; instead, samples were lightly rinsed and blotted dry, weighed, and placed on wet ice prior to transfer for analysis by solid scintillation counting. Blood was mixed and duplicate weighed aliquots were analyzed by solid scintillation counting. Samples were analyzed for at least 1 min or 1,000,000 counts. As applicable, one complete set of samples (one aliquot per sample) was counted; the second set was then analyzed.

Allometric Analysis

To evaluate the scalability of measured parameters across different species, we conducted an allometric analysis. Tissue weight-normalized values were multiplied by their respective organ weights (Table S1) in order to express measured values on a per-organ basis (Q in μL/min, Vv in μL, Vi in μL). Estimated tissue-specific values were related to total animal body weights (BW) by the equation:
$$ Y=A\times {\mathrm{BW}}^b $$
(7)
where Y corresponds to Vv, Vi, and Q and BW is the total body weight for each animal. A and b, the allometric coefficient and exponent, are fitted parameters (Table S2). A value of b = 1 suggests that the parameter Y scales proportionally with BW and b values lower or greater than 1 suggests that Y scales less or more than BW proportionally, respectively. Extrapolation of the linear regression lines to 70 kg was used to estimate values in human, which were subsequently normalized using human organ weights from the literature [52].

Sensitivity Analysis

We used a previously published PBPK model [10], which we scaled from mouse to cynomolgus monkey (Table S3), to evaluate the impact of uncertainty in Vv, Vi, and Q on simulated total uptake of a 150-kDa monoclonal antibody in selected fast and slow perfused tissues. Using the scaled cyno model, concentration-time curves were simulated using Vv, Vi, and Q values measured here and those compiled by Shah and Betts [13]. The area under each simulated % injected dose versus time curve from 0 to 7 days (AUC7) was calculated and compared.

RESULTS

Total Blood Volumes

Measured whole-body blood volumes are shown in Table I. Total BV values (mL) were 181 ± 24 in monkey and 10.2 ± 0.5 in rat and ranged from 1.3 ± 0.1 to 2.1 ± 0.1 in mice. Weight-normalized BV values (mL/kg) were 47.7 ± 0.8 in cynomolgus monkeys and 48 ± 2 in Sprague-Dawley rats and ranged from 59 ± 5 for C57BL-6 to 109 ± 4 for C.B-17 SCID beige mice. Higher weight-normalized BVs were measured in immunocompromised strains (108 in nude and 109 in SCID) relative to immunocompetent strains (ranging from 59 to 78) despite a similar body weight range.
Table I

Summary of Species/Strains in Which Vv, Vi, and Q Were Measured in Previous and Current Studies with Body Weights and Measured Whole-body Blood Volumes (BV)

Species

Strain/origin

Sex

Immunocompetent

Number

Body weight (g)

Total BV (mL)

Weight-normalized BV (mL/kg)

Data source

Homo sapiens (human)

Global average

M/F

Yes

n/a

70,000

5200

74

[50]

Macaca fascicularis (cynomolgus monkey)

Chinese origin

M

Yes

3

3800 ± 400

181 ± 24

47.7 ± 0.8

Current study

Rattus norvegicus (rat)

Sprague-Dawley

F

Yes

5

213 ± 4

10.2 ± 0.5

48 ± 2

[28]

Mus musculus (mouse)

DBA/2

F

Yes

5

19 ± 1

1.5 ± 0.1

78 ± 4

[26, 28]

Mus musculus (mouse)

C57BL-6

F

Yes

4

4

21.0 ± 0.6

22 ± 4a

n/a

1.3 ± 0.1

n/a

59 ± 5

[26, 27, 28]

Mus musculus (mouse)

B6SJL-Tg (SOD1) 2Gur/J

F

Yes

5

23 ± 3

1.6 ± 0.1

68 ± 7

[23]

Mus musculus (mouse)

Beige nude

F

No

5

18.8 ± 0.6

2.0 ± 0.1

108 ± 5

[21]

Mus musculus (mouse)

C.B-17 SCID.bg

F

No

5

20 ± 2

2.2 ± 0.1

109 ± 4

[24]

Mus musculus (mouse)

C.B-17 SCID.bg

M

No

4

26 ± 1

n/a

n/a

[25]

M, male; F, female; n/a, not available

aMeasured in two separate studies

Vascular Volumes

Vv in monkey tissues, derived from 99mTc labeling of red blood cells and expressed in microliters of blood per gram (μL/g) of tissue, are presented in the first column of Table II. Vv in muscle was roughly consistent across cynomolgus monkeys and our previously measured values in other species (Table II, columns 2–8), ranging overall from 5.4 ± 2.0 to 9 ± 2 (both values in mice). Muscle Vv values were 7 ± 1, 7 ± 1, and 5.5 ± 1.1 for cynomolgus monkeys, rat, and DBA/2 mice, respectively. Monkey tissue Vv values ranged from 4 ± 1 in eyes to 188 ± 22 in kidneys. One notable exception was lung, for which the Vv value in rats (538 ± 69) was significantly higher than that in monkeys (126 ± 29) and mice (ranging from 125 ± 17 to 214 ± 78). While the high rat lung Vv was reproduced in a second study, we suspect that blood pooling during euthanasia and harvest exaggerated the value as the mean rat lung weights were higher than expected.
Table II

Measured Organ-Level Vascular Volumes (Vv) in Various Laboratory Animal Species. All Values Were Measured by the Same 99mTc Red Blood Cell Labeling Method. Each Study Is N = 5 Animals per Group Unless Otherwise Noted. Values in Brackets Indicate Data from a Second Independent Study. Data from Individual Animals, for Cynomolgus Monkey Only, Are Available in Supplementary Table S7

Mean Vv (μL/g)

Tissue

Monkey

Cynomolgus

Current study, N = 3

Rat

Sprague-Dawley [28]

Study 1

[study 2],

Mouse

DBA/2 [28, 26]

Mouse

C57/Bl6

Study 1 [28, 26]

[study 2] [27]

N = 4, both studies

Mouse

B6SJL-Tg (SOD1)

2Gur/J [23]

Mouse

beige nude

XID [22, 21]

Mouse

C.B-17 SCID beige (female) [24]

Mouse

C.B-17 SCID beige (male) [25] (N = 4)

SM intestine

16 ± 2

26 ± 8

[22 ± 3]

16.4 ± 0.4

6 ± 1

[6.8 ± 0.4]

13 ± 9

8 ± 2

7 ± 3

10 ± 5

LG intestine

27 ± 7

20 ± 4

[19 ± 5]

11 ± 5

5.3 ± 0.6

n/a

n/a

n/a

13 ± 3

Lung

126 ± 29

538 ± 69

[605 ± 84]

160 ± 27

146 ± 13

[125 ± 17]

126 ± 26

192 ± 64

214 ± 78

207 ± 101

Liver

92 ± 8

124 ± 75

[92 ± 18]

45 ± 13

41 ± 10

[44 ± 4]

38 ± 4

42 ± 8

55 ± 11

55 ± 10

Kidneys

188 ± 22

190 ± 20

[186 ± 21]

110 ± 19

89 ± 10

[117 ± 20]

121 ± 10

77 ± 7

108 ± 28

111 ± 24

Heart

73 ± 10

82 ± 24

[222 ± 58]

96 ± 41

40 ± 10

[38 ± 4]

53 ± 18

30 ± 7

49 ± 9

42 ± 22

Spleen

169 ± 41

261 ± 120

[440 ± 66]

76 ± 16

92 ± 14

[94 ± 5]

87 ± 20

121 ± 29

117 ± 36

153 ± 45

Muscle

7 ± 1

7 ± 1

[9 ± 4]

5.5 ± 1.1

5.9 ± 0.9

[5.8 ± 0.9]

6 ± 1

9 ± 2

5.4 ± 2.0

9 ± 3

Fat

25 ± 19

14 ± 2

[12 ± 8]

11 ± 4

6 ± 3

n/a

14 ± 4

9 ± 4

n/a

Brain

8 ± 1

9 ± 1

[10 ± 2]

10 ± 1.3

8 ± 1

[9 ± 1]

7 ± 1

11 ± 2

11 ± 3

n/a

Skin

13 ± 2

13 ± 3

[11 ± 2]

5.3 ± 2.6

14 ± 9

[8 ± 1]

7 ± 1

n/a

8 ± 7

7 ± 2

Stomach

n/a

19 ± 4

[23 ± 5]

11 ± 1.6

9 ± 2

n/a

n/a

n/a

n/a

Thymus

n/a

18 ± 4

[17 ± 4]

140 ± 60

14 ± 8

n/a

n/a

n/a

n/a

Pancreas

n/a

30 ± 4

[30 ± 6]

15 ± 14

8 ± 2

n/a

n/a

n/a

n/a

Lymph node

n/a

25 ± 9

[6 ± 8]

17 ± 5.5

5.2 ± 0.9

n/a

n/a

n/a

n/a

Ovaries

n/a

60 ± 24

[125 ± 86]

n/a

n/a

n/a

n/a

n/a

n/a

Eyes

4 ± 1

n/a

n/a

n/a

n/a

n/a

n/a

n/a

SM, small; LG, large; n/a, not available

Interstitial Volumes

Measured tissue-level Vi in cynomolgus monkeys, derived from 111In-DTPA levels and expressed in microliters of interstitial fluid per gram (μL/g) of tissue, are presented in the first column of Table III. It is critical to note that 111In-DTPA does not provide physiologically meaningful measurements of Vi in kidney and brain/eye due to probe clearance and lack of blood-brain/ocular-barrier permeability, respectively. A remarkably good agreement in Vi on a per-gram-tissue basis was observed across most monkey tissues analyzed here and previously analyzed rodent tissues. For instance, measured Vi in muscle were fairly consistent across cynomolgus monkeys and our previously measured values in other species (Table III, columns 2–7), ranging overall from 48 ± 8 in female SCID beige mice to 108 ± 18 in monkeys. Muscle Vi values in immunocompetent mice (ranging from 92 ± 27 to 93 ± 20) and rats (93 ± 18) were similar to the value in monkeys. Monkey tissue Vi values ranged from 71 ± 40 in spleen to 268 ± 50 in skin.
Table III

Measured Organ-Level Interstitial Volumes (Vi) in Various Laboratory Animal Species. All Values Were Measured by the Same 111In-DTPA Probe Method. Each Study Is N = 5 Animals per Group Unless Otherwise Noted. Data From Individual Animals, for Cynomolgus Monkey Only, Are Available in Supplementary Table S7

Mean Vi (μL/g)

Tissue

Monkey

Cynomolgus (current study) (N = 3)

Rat

Sprague-Dawley [28]

Mouse

DBA/2 [28, 26] (N = 4)

Mouse

C57/Bl6 [28, 26]

Mouse

beige nude XID [21]

Mouse

C.B-17 SCID beige (female) [24]

Mouse

C.B-17 SCID beige (male) [25]

SM intestine

140 ± 15

163 ± 43

218 ± 129

116 ± 33

121 ± 62

112 ± 47

157 ± 48

LG intestine

81 ± 11

248 ± 69

223 ± 87

158 ± 37

n/a

n/a

150 ± 21

Lung

247 ± 24

157 ± 90

208 ± 20

206 ± 69

173 ± 71

210 ± 62

179 ± 32

Liver

100 ± 29

187 ± 82

86 ± 18

148 ± 48

90 ± 15

72 ± 9

92 ± 25

Heart

148 ± 7

105 ± 54

115 ± 25

143 ± 48

158 ± 67

184 ± 92

150 ± 20

Spleen

71 ± 40

43 ± 27

24 ± 4

22 ± 25

25 ± 10

n/a

20 ± 7

Muscle

108 ± 18

93 ± 18

92 ± 27

93 ± 20

83 ± 41

48 ± 8

55 ± 18

Fat

188 ± 45

57 ± 25

462 ± 379

97 ± 36

346 ± 259

49 ± 12

115 ± 80

Skin

268 ± 50

709 ± 139

366 ± 78

452 ± 126

n/a

259 ± 81

n/a

Stomach

n/a

240 ± 60

210 ± 99

177 ± 41

n/a

n/a

n/a

Thymus

n/a

197 ± 53

162 ± 114

204 ± 89

n/a

n/a

n/a

Pancreas

n/a

120 ± 63

93 ± 38

70 ± 31

n/a

n/a

n/a

Lymph node

n/a

178 ± 49

390 ± 178

153 ± 117

n/a

n/a

n/a

Ovaries

n/a

111 ± 24

n/a

n/a

n/a

n/a

n/a

SM, small; LG, large; n/a, not available

Rates of Blood Flow

Measured tissue-level rates of blood flow (Q) in cynomolgus monkeys derived from 86RbCl expressed in microliters of blood per gram of tissue per minute (μL/g/min) are presented in the first column of Table IV. Importantly, 86RbCl cannot freely cross the blood-brain and blood-ocular barriers so values for brain and eyes are not shown [50]. It should also be noted that 86RbCl may somewhat exaggerate Q in heart due to cardiac sequestration of Rb+ through K+ channels and Na+/K+-ATPase pumps [57]. Q in muscle for monkeys (231 ± 54) was in good agreement with previously measured rodent values (Table IV, columns 2–7), including 190 ± 38 in rats and ranging from 186 ± 47 to 300 ± 34 in mice. Monkey tissue Q values ranged from 93 ± 23 in skin to 3365 ± 961 in kidneys. The Q value in monkey spleen (471 ± 99) was comparable to previously measured values in rats (572 ± 134) but generally higher than previously measured values in mice (ranging from 124 ± 51 to 326 ± 43).
Table IV

Measured Organ-Level Blood Flow Rates (Q) in Various Laboratory Animal Species. All Values Were Measured by the Same 86Rb Probe Method. Each Study Is N = 5 Animals per Group Unless Otherwise Noted. Values in Brackets Indicate Data from a Second Independent Study. Data from Individual Animals, for Cynomolgus Monkey Only, Are Available in Supplementary Table S7

Mean Q (μL/g/min)

Tissue

Monkey

Cynomolgus (current manuscript)

N = 3

Rat

Sprague-Dawley [28]

Mouse

DBA/2 [28, 26]

Mouse

C57/Bl6

Study 1 [28, 26]

[study 2] [27] (N = 4, both studies)

Mouse

B6SJL-Tg (SOD1) 2Gur/J [23]

Mouse

Beige nude XID [22, 21]

Mouse

C.B-17 SCID beige (female) [24]

SM intestine

332 ± 94

992 ± 338

863 ± 114

388 ± 89

[464 ± 19]

320 ± 168

380 ± 136

289 ± 36

LG intestine

458 ± 209

830 ± 361

520 ± 40

263 ± 34

n/a

n/a

n/a

Lung

570 ± 115

353 ± 58

544 ± 216

590 ± 74

[843 ± 73]

721 ± 86

538 ± 150

707 ± 129

Liver

416 ± 164

273 ± 50

250 ± 28

137 ± 17

[279 ± 48]

109 ± 27

158 ± 20

219 ± 33

Kidneys

3365 ± 961

3771 ± 1509

2853 ± 551

1920 ± 196

[2492 ± 251]

1050 ± 331

1172 ± 419

1622 ± 210

Heart

1730 ± 153

1723 ± 216

1247 ± 101

904 ± 37

[1463 ± 185]

1549 ± 164

877 ± 63

1227 ± 240

Spleen

471 ± 99

572 ± 134

326 ± 43

163 ± 23

[324 ± 28]

124 ± 51

169 ± 105

134 ± 27

Muscle

231 ± 54

190 ± 38

200 ± 43

285 ± 50

[217 ± 35]

300 ± 34

186 ± 47

203 ± 38

Fat

155 ± 31

72 ± 34

86 ± 65

45 ± 18

n/a

120 ± 37

102 ± 45

Skin

93 ± 23

142 ± 9

99 ± 52

116 ± 21

[160 ± 32]

171 ± 22

n/a

101 ± 29

Stomach

n/a

561 ± 79

241 ± 39

228 ± 27

n/a

n/a

n/a

Thymus

n/a

381 ± 68

237 ± 99

141 ± 17

n/a

n/a

n/a

Pancreas

n/a

756 ± 138

337 ± 177

258 ± 45

n/a

n/a

n/a

Lymph node

n/a

219 ± 98

381 ± 229

137 ± 9

n/a

n/a

n/a

Ovaries

n/a

797 ± 43

n/a

n/a

n/a

n/a

n/a

SM, small; LG, large; n/a, not available

Blood-Brain Barrier Permeability

Brain Kp values measured in monkeys (0.003 ± 0.001) were significantly lower than previously measured values in rats (0.019 ± 0.007) and mice (ranging from 0.016 ± 0.007 to 0.035 ± 0.012) (Fig. 2).
Fig. 2

Species dependence of brain partitioning. Blood-corrected brain/plasma partitioning coefficient (Kp) for 111In-DTPA as a measure of blood-brain integrity in cynomolgus monkeys determined in this study, compared to tissues from other laboratory animal species (Sprague-Dawley rat and various mouse strains) measured previously

Allometric Analysis

Results from the allometric analysis are shown in Fig. 3. Qualitatively, the observed data is consistent with the allometric relationship, where values of Vv (μL), Vi (μL), and Q (μL/min) across species are largely dependent on BW. Parameter values for human are obtained by extrapolation of the regression lines to 70 kg body weight (Table V). Human parameters were all generally lower than the corresponding monkey parameters on a tissue weight-normalized basis. For instance, Vv (μL/g), Vi (μL/g), and Q (μL/g/min) were 59, 87, and 173 in human lung compared to our measured values of 126, 247, and 570 in monkey lung. Most exponents were between ~ 0.9 and 1. An exception was Vi in liver which was 0.8, Q in kidney and spleen which were ~ 0.85, and Vv in kidney which was 0.87 (Table S2). The allometric plots with 95% confidence intervals are provided in Fig. S1.
Fig. 3

Allometric plots for Vv, Vi, and Q. Vascular volume (Vv), interstitial volume (Vi), and blood flow (Q) allometric plots of selected organs in mouse (C57-Bl6 and DBA/2), rat (Sprague-Dawley), and monkey (cynomolgus, Chinese origin). Parameters (individual points) with standard deviation are shown with fitted linear regression lines. A dotted vertical line at 70 kg indicates extrapolation to human values

Table V

Human Vv, Vi, and Q Values Extrapolated from Allometric Analysis of Preclinical Data in Mouse, Rat, and Monkey

 

Lung

Liver

Kidney

Heart

Spleen

Vv (μL/g)

59

84

116

59

53

Vi (μL/g)

87

70

n/a

93

45

Q (μL/g/min)

173

366

2188

1273

169

n/a, not available

Sensitivity Analysis

The sensitivity analysis compares the impact of various values (our measured values versus previously assembled literature values [13]) of Vv, Vi, and Q on AUC7 of an antibody in muscle and liver, in cynomolgus monkeys (Table S3). AUC7 values of simulated %ID versus time curves in both tissues are most sensitive to changes in Vv, where for muscle a 131% increase in AUC7 results from using the literature value of Vv (40 μL/g) versus the value estimated here (7 μL/g) and for liver a 55% increase in AUC7 results from using the literature value of Vv (155 μL/g) versus the value estimated here (92 μL/g). The impact of Vi on AUC7 is less pronounced, with almost no increase in AUC7 for muscle when using the literature value (130 μL/g) versus the value estimated here (108 μL/g) and a modest increase in liver (which has much higher vascular permeability) of 18% when using the literature value (200 μL/g) versus the value estimated here (100 μL/g). For large molecular weight compounds such as the 150-kDa antibody considered here, extravasation from the vascular to interstitial space is the rate-limiting step for tissue uptake; therefore, AUC7 is almost completely insensitive to the range of Q values evaluated for muscle and liver (37–416 μL/min/g), consistent with previous findings [16, 17]. Although not evaluated here, one would expect small molecule tissue uptake to be sensitive to Q [58]. The sensitivity analyses discussed here and in previous works [16, 17] are, strictly speaking, valid only for each respective model/parameter combination. As good practice, parameter sensitivity analysis should be re-evaluated each time the model is fitted to a new biodistribution data set.

DISCUSSION

Prior to this study, a relatively limited amount of organ-level physiological data in primates was available; furthermore, primate and rodent data were often acquired by different methods [18]. Organ-level values used by PBPK modelers, exemplified by a recent example [13], are usually obtained from various literature sources that include compilations of previously measured and assumed nominal values. In many cases, citations for physiological parameter values used in modeling papers refer to earlier modeling or review papers, such that ascertaining the original source and, importantly, the associated in vivo measurement methods (if any) is challenging. Our experimental strategy aimed to measure three critical parameters—Vv, Vi, and Q—in cynomolgus monkeys using protocols as consistent as possible with those we previously used in rodents [28]. This approach facilitates direct comparison of parameters, estimated using the same methodology, across three species—mice, rats, and monkeys—commonly used in translational pharmacology and drug development. Since our measurements were performed in cynomolgus monkeys of Chinese origin, it should be noted that parameters may vary across other monkey populations including other cynomolgus populations (e.g., Mauritian). Furthermore, the possibility that transport and other experimental stresses may have influenced our results cannot be dismissed; however, we included an acclimation period, commingling when feasible, treats, and other elements in our study design in order to minimize experimental stress as much as possible [59, 60].

With some important exceptions, our results measured in monkeys were consistent with our previous results obtained in mice and rats, where many of the measured tissue-specific body weight-normalized parameter values were largely similar across species (Tables II, III, IV). In general, similar values were measured for poorly perfused tissues like muscle while greater interspecies variability was observed for more highly perfused organs (e.g., heart, lung, liver). Furthermore, most of the Vv, Vi, and Q values observed in this study are in reasonably good agreement with values that have been assembled by Shah and Betts for PBPK modeling [13]; however, there are certain exceptions worth noting. For instance, sizeable differences were observed between values reported by Shah and Betts [13] and those measured in this study for Vv in muscle, brain, and skin (our values lower in all three species); Q in muscle and adipose (our values higher in monkey); and Q in heart, thymus, and spleen (our values lower in mice and rats).

Our tissue-level Vv (μL/g) values measured here in monkey and previously in mice and rats [28] were mostly consistent across these species (Table II). We are aware of only one publication that reports tissue-level Vv measurements in non-human primates; our cynomolgus monkey values differ greatly from the reported rhesus monkey values (our values are all lower, some by an order of magnitude) whose method of measurement could not be ascertained [12]. Lung Vv in monkeys agreed well with our values in mice but was significantly lower than those in rats (Table II). Our murine muscle Vv ranged from 5 to 7 μL/g, consistent with previous reports of ~ 5 to 8 μL/g measured by 51Cr-labeled RBCs in murine leg muscle [38, 39]. Generally good agreement was also noted between the measured rat Vv values that we observed and those reported by Khor et al., (measured by 51Cr-labeled RBCs) in intestine, stomach, brain, muscle, kidney, liver, lung, and spleen (but with poor agreement in lung) [61]. Our measured brain Vv values in rats (9 ± 1) and mice (8–11 μL/g) agree well with previously published values measured by 14C-labeled sucrose in rats (13 ± 3) [62] and by 3H-labeled inulin in mice (9.4) [63]. Our measured rat Vv values (μL/g) in skin (13 ± 3), muscle (7 ± 1), lung (538 ± 69), and kidney (184 ± 18) agreed reasonably well, with the exception of lung, with values previously measured by 51Cr-labeled RBCs (6.7, 6.5, 218, and 183, respectively) [64]. However, in comparison to the values assembled by Shah and Betts [13], our measured Vv’s are consistently lower in muscle and skin and higher in lung (except mouse) and kidney (Table S4) across all species considered here. These disagreements may reflect the fact that Vv values in Shah and Betts [13] were gathered from various sources which include values scaled from other species (e.g., dog) [18] and/or measured by different techniques (e.g., radiolabeled albumin) [65]. Another possible source of variability in various measured Vv values may be due to circadian effects, which have been reported in nude mice where ranges of ~ 70–200, 50–100, and 60–180 μL/g in liver, kidney, and lung, respectively, have been observed [32].

We are not aware of any previous reports of tissue-level Vi measurement in non-human primates. Our measured Vi values in most tissues were remarkably consistent across species (Table III) and agreed well with values reported in Shah and Betts [13] (Table S5). Our muscle Vi ranged from 48 to 108 μL/g which is roughly consistent with previous reports of ~ 150 μL/g (measured by 111In-DTPA) in murine leg muscle [38, 39].

Importantly, blood-brain barrier permeability to 111In-DTPA (represented as brain Kp) in monkeys was much lower than that in rodents (Fig. 2). The observed 5–10-fold discrepancy is consistent with previous reports of brain antibody concentrations reaching only 0.01–0.02% of steady-state serum concentrations in cynomolgus monkeys [66, 67], in contrast with reports of 0.1% in mice [68]. Although it is possible that these interspecies differences in apparent antibody BBB permeability could be due to differences in binding affinity of the antibody Fc region for the neonatal Fc receptor, our results here suggest that more fundamental differences in BBB permeability may exist between species since we used a non-FcRn-binding small molecule probe (111In-DTPA) to measure Kp. This finding is of great significance in the development of drugs for neurological diseases, as it has significant implications with respect to our ability to translate brain PK/PD data across species. Simply scaling from mouse or rat to monkey and human in physiologically based models may exaggerate the ability of antibodies and other drugs with low brain partitioning to reach the site of action.

Our measured Q values in small intestine, large intestine, liver, kidneys, heart, and spleen agreed reasonably well with previously measured values in cynomolgus monkey (using radiolabeled microspheres) reported by Gelman et al. (Table IV, column 1) [69]. Compared to rhesus monkey results reported by Forsyth et al. (also using microspheres) [56], our Q values for skin, heart, small intestine, large intestine, spleen, and kidney also agreed reasonably well (Table IV, column 1); however, our Q values in lung and liver of 570 and 416 µL/g/min were higher than the reported rhesus values of 180 and 340 [56]. Interestingly, Q values in liver and spleen trended higher in monkey than in mice (as did rat Q values), but were comparable to rodent values in muscle and skin (Table IV). Our muscle Q values of 200 and 190 µL/g/min in mice and rats agreed well with previously reported values of 236 and 237, respectively [20]. Similarly, our skin Q values of 99 and 142 µL/g/min agreed reasonably well with previously reported values of 185 and 141 in mice and rats, respectively [20]. Compared to values assembled by Shah and Betts [13], the following were observed: (1) our heart Q values were lower in rodents, but higher in monkey (Table S6), (2) our lung Q values could not be directly compared since lung Q is defined as total cardiac output, and (3) our muscle and adipose Q values were each roughly 6-fold higher than the respective values in monkey. Unfortunately, we were unable to ascertain the method(s) by which Q values in Shah and Betts [13] were measured or derived.

Our measured monkey BV of 47.7 ± 0.7 (Table I) is lower than the range of 55–75 mL/kg reported for cynomolgus monkey [70]. It is, however, within the measured range for rhesus monkeys (44–67 mL/kg) [70] despite being considerably lower than a value of ~ 149 mL/kg previously used for PBPK modeling [12]. Although weight-normalized BVs were identical between monkeys and rats, they were consistently higher for mice (Table I). Furthermore, immunocompromised strains (nude and SCID) had higher BVs than immunocompetent strains, a trend that has not previously been noted to our knowledge. Our measured rat BV of 48 ± 2 mL/kg is lower than previous reports of ~ 64 mL/kg (measured, range 58–70) [34, 70] and ~ 74 mL/kg (previously used for PBPK modeling) [12]. Importantly, BV values are highly dependent on measurement method as values of ~ 94 and ~ 65 mL/kg were reported for rats using 131I-labeled albumin and 51Cr-labeled RBCs, respectively [35]. Our measured values in mice (60 ± 9 to 109 ± 4 mL/kg) are roughly consistent with a previously reported range of 85–95 mL/kg in CSI and CBA mice [71]. Others have noted differences in mouse BV depending on methodology, with values of 50–59 and 73–103 mL/kg using 59Fe-labeled RBCs and 59Fe-labeled transferrin, respectively [72]. These two 59Fe-labeling-based techniques also led to distinct ranges for venous (0.47 to 0.52) and whole-body (0.30 to 0.34) hematocrits in mice [72], with similar trends in rats [35] and humans [73, 74, 75, 76]. Differences between venous and whole-body hematocrits have been attributed in part to the spleen acting as a reservoir for RBCs [77], and a similar mechanism may influence BV measurements. Blood has been reported to constitute 4.9, 7.4, and 7.9% of body weight in mice, rats, and humans, respectively, equivalent to 49, 74, and 79 mL/kg (assuming density of 1 g/mL) [18]. However, our data seem to indicate an opposite trend, with lower body weight-normalized volumes of blood in higher species, roughly consistent with values of approximately 78 and 55 mL/kg reported for mouse and rat, respectively [20]. Furthermore, our results are roughly consistent with a previously reported allometric scaling method in which blood weight is equal to 0.055 multiplied by body weight to the power of 0.99 [1].

Overall, our allometric analysis revealed the expected body weight dependence of Vv, Vi, and Q (Fig. 3). Scaling exponents for Vi and Vv suggest that lung and liver vary near body weight proportionally, while spleen has a more shallow slope with respect to BW (Table S2). The extrapolated human Q values of 366, 2188, 1273, and 169 μL/g/min in liver, kidney, heart, and spleen (Table V) were lower (except heart) than human values of 806, 4000, 727, and 428 reported in Davies and Morris [52]. Note that Q values were calculated from CO values (8, 74, and 1086 mL/min for a 0.02-kg mouse, 0.25-kg rat, and 5-kg monkey, respectively) obtained from a separate report [52] in which CO (at least one of which is a measured value [56]) scales with an exponent of ~ 0.9.

One technical challenge we faced was that the size of monkey tissues limits the ability to analyze whole tissues by solid scintillation (i.e., gamma) counting as was previously done for most mouse and rat organs. One possible approach was to collect representative sections; however, we were concerned that the complex structures of organs like brain, heart, kidneys, and lungs might introduce bias into our measurements if the section analyzed did not accurately represent the organ as a whole. As such, these entire organs were homogenized so that aliquots could be counted. In contrast, the representative sectioning approach (in duplicate, to minimize bias) was applied to intestines (ileum and colon), muscle, fat, and skin. It should be noted that representative sections of these tissues were also used to derive our previous parameter values in rodents.

A second scaling challenge was the large blood volume of monkeys (approximately 180 mL), which diluted the 99mTc activity in our in vivo RBC labeling/transfusion protocol to such an extent that this method would have required working with excessive amounts of radioactivity to maintain sufficient gamma counting sensitivity. To address this challenge, we adapted an alternative in vivo/in vitro clinical protocol [78] in which pre-tinning is accomplished in vivo while 99mTcO4− labeling is accomplished in vitro but without requiring centrifugation to preserve the integrity of RBCs. To ensure that our data in monkeys can be compared to our previous data in rodents, we ensured that non-RBC-associated levels of 99mTc were low by centrifuging aliquots and counting plasma and pellets separately. Furthermore, any damage to RBCs would likely result in elevated uptake in spleen (the sequestration site for damaged or aging RBCs); however, spleen Vv values were similar between monkey and rodents (Table II), suggesting that RBC damage did not occur.

CONCLUSION

In summary, our work addresses the limited availability of tissue-level physiological parameters in primates in the literature, and our allometric analysis approach allowed extrapolation of human Vv, Vi, and Q values based on preclinical knowledge. While the majority of our measured values agree well with previously published values [13], some differed by as much as 6-fold which, particularly in the case of large tissues, may be impactful as illustrated by sensitivity analysis which shows that PBPK model simulations are very sensitive to changes in Vv. The finding that the blood-brain barrier was more impervious to intravenously infused 111In-DTPA in monkey than in rodents was surprising and warrants further investigation, as does the observed trend towards higher weight-normalized blood volumes in immunocompromised mice relative to immunocompetent strains. Overall, the data herein support previous assumptions that Vv, Vi, and Q in tissues are roughly body weight proportional across species, but that tissue-specific deviations warrant the use of measured, species-specific values.

Notes

Acknowledgements

All studies were internally funded by Genentech, a member of the Roche Group. We thank Ruedi Port and Daniela Bumbaca Yadav for scientific discussions and Bernadette Johnstone, Cynthia Young, Elizabeth Torres, Jason Ho, Jose Imperio, Kirsten Messick, Nicole Valle, Nina Ljumanovic, Roxanne Kyauk, Shannon Stainton, and Eduardo Mundo for technical assistance with animal studies.

Supplementary material

12248_2018_264_MOESM1_ESM.pdf (150 kb)
Supplementary Materials Organ weights, allometric scaling exponents, and organ level vascular volumes, interstitial volumes and blood flow rates in various laboratory animal species measured in this study and compared to those that have been reported previously. (PDF 149 kb)

References

  1. 1.
    Adolph EF. Quantitative relations in the physiological constitutions of mammals. Science. 1949;109(2841):579–85.CrossRefGoogle Scholar
  2. 2.
    Dedrick RL. Animal scale-up. J Pharmacokinet Biopharm. 1973;1(5):435–61.CrossRefGoogle Scholar
  3. 3.
    McMahon T. Size and shape in biology. Science. 1973;179(4079):1201–4.CrossRefGoogle Scholar
  4. 4.
    Mordenti J. Man versus beast: pharmacokinetic scaling in mammals. J Pharm Sci. 1986;75(11):1028–40.CrossRefGoogle Scholar
  5. 5.
    Huang Q, Riviere JE. The application of allometric scaling principles to predict pharmacokinetic parameters across species. Expert Opin Drug Metab Toxicol. 2014;10(9):1241–53.CrossRefGoogle Scholar
  6. 6.
    Mordenti J, Chen SA, Moore JA, Ferraiolo BL, Green JD. Interspecies scaling of clearance and volume of distribution data for five therapeutic proteins. Pharm Res. 1991;8(11):1351–9.CrossRefGoogle Scholar
  7. 7.
    Eigenmann MJ, Karlsen TV, Krippendorff BF, Tenstad O, Fronton L, Otteneder MB, et al. Interstitial IgG antibody pharmacokinetics assessed by combined in vivo- and physiologically-based pharmacokinetic modelling approaches. J Physiol. 2017;595(24):7311–30.CrossRefGoogle Scholar
  8. 8.
    Boswell CA, Bumbaca D, Fielder PJ, Khawli LA. Compartmental tissue distribution of antibody therapeutics: experimental approaches and interpretations. AAPS J. 2012;14(3):612–8.CrossRefGoogle Scholar
  9. 9.
    Benias PC, Wells RG, Sackey-Aboagye B, Klavan H, Reidy J, Buonocore D, et al. Structure and distribution of an unrecognized interstitium in human tissues. Sci Rep. 2018;8(1):4947.CrossRefGoogle Scholar
  10. 10.
    Ferl GZ, Wu AM, DiStefano JJ 3rd. A predictive model of therapeutic monoclonal antibody dynamics and regulation by the neonatal Fc receptor (FcRn). Ann Biomed Eng. 2005;33(11):1640–52.CrossRefGoogle Scholar
  11. 11.
    Gerlowski LE, Jain RK. Physiologically based pharmacokinetic modeling: principles and applications. J Pharm Sci. 1983;72(10):1103–27.CrossRefGoogle Scholar
  12. 12.
    Hall C, Lueshen E, Mosat A, Linninger AA. Interspecies scaling in pharmacokinetics: a novel whole-body physiologically based modeling framework to discover drug biodistribution mechanisms in vivo. J Pharm Sci. 2012;101(3):1221–41.CrossRefGoogle Scholar
  13. 13.
    Shah DK, Betts AM. Towards a platform PBPK model to characterize the plasma and tissue disposition of monoclonal antibodies in preclinical species and human. J Pharmacokinet Pharmacodyn. 2012;39(1):67–86.CrossRefGoogle Scholar
  14. 14.
    Zhao J, Cao Y, Jusko WJ. Across-species scaling of monoclonal antibody pharmacokinetics using a minimal PBPK model. Pharm Res. 2015;32(10):3269–81.CrossRefGoogle Scholar
  15. 15.
    Garg A, Balthasar JP. Physiologically-based pharmacokinetic (PBPK) model to predict IgG tissue kinetics in wild-type and FcRn-knockout mice. J Pharmacokinet Pharmacodyn. 2007;34(5):687–709.CrossRefGoogle Scholar
  16. 16.
    Baxter LT, Zhu H, Mackensen DG, Butler WF, Jain RK. Biodistribution of monoclonal antibodies: scale-up from mouse to human using a physiologically based pharmacokinetic model. Cancer Res. 1995;55(20):4611–22.PubMedGoogle Scholar
  17. 17.
    Baxter LT, Zhu H, Mackensen DG, Jain RK. Physiologically based pharmacokinetic model for specific and nonspecific monoclonal antibodies and fragments in normal tissues and human tumor xenografts in nude mice. Cancer Res. 1994;54(6):1517–28.PubMedGoogle Scholar
  18. 18.
    Brown RP, Delp MD, Lindstedt SL, Rhomberg LR, Beliles RP. Physiological parameter values for physiologically based pharmacokinetic models. Toxicol Ind Health. 1997;13(4):407–84.CrossRefGoogle Scholar
  19. 19.
    Hu TM, Hayton WL. Allometric scaling of xenobiotic clearance: uncertainty versus universality. AAPS PharmSciTech. 2001;3(4):E29.Google Scholar
  20. 20.
    Lindstedt SL, Schaeffer PJ. Use of allometry in predicting anatomical and physiological parameters of mammals. Lab Anim. 2002;36(1):1–19.CrossRefGoogle Scholar
  21. 21.
    Boswell CA, Ferl GZ, Mundo EE, Bumbaca D, Schweiger MG, Theil FP, et al. Effects of anti-VEGF on predicted antibody biodistribution: roles of vascular volume, interstitial volume, and blood flow. PLoS One. 2011;6(3):e17874.  https://doi.org/10.1371/journal.pone.0017874.CrossRefPubMedPubMedCentralGoogle Scholar
  22. 22.
    Boswell CA, Ferl GZ, Mundo EE, Schweiger MG, Marik J, Reich MP, et al. Development and evaluation of a novel method for preclinical measurement of tissue vascular volume. Mol Pharm. 2010;7(5):1848–57.CrossRefGoogle Scholar
  23. 23.
    Boswell CA, Mundo EE, Johnstone B, Ulufatu S, Schweiger MG, Bumbaca D, et al. Vascular physiology and protein disposition in a preclinical model of neurodegeneration. Mol Pharm. 2013;10(5):1514–21.CrossRefGoogle Scholar
  24. 24.
    Pastuskovas CV, Mundo EE, Williams SP, Nayak TK, Ho J, Ulufatu S, et al. Effects of anti-VEGF on pharmacokinetics, biodistribution, and tumor penetration of trastuzumab in a preclinical breast cancer model. Mol Cancer Ther. 2012;11(3):752–62.CrossRefGoogle Scholar
  25. 25.
    Boswell CA, Mundo EE, Zhang C, Stainton SL, Yu SF, Lacap JA, et al. Differential effects of predosing on tumor and tissue uptake of an 111In-labeled anti-TENB2 antibody-drug conjugate. J Nucl Med. 2012;53(9):1454–61.  https://doi.org/10.2967/jnumed.112.103168.CrossRefPubMedGoogle Scholar
  26. 26.
    Deng R, Bumbaca D, Pastuskovas CV, Boswell CA, West D, Cowan KJ, et al. Preclinical pharmacokinetics, pharmacodynamics, tissue distribution, and tumor penetration of anti-PD-L1 monoclonal antibody, an immune checkpoint inhibitor. MAbs. 2016;8(3):593–603.CrossRefGoogle Scholar
  27. 27.
    Bien-Ly N, Boswell CA, Jeet S, Beach TG, Hoyte K, Luk W, et al. Lack of widespread BBB disruption in Alzheimer’s disease models: focus on therapeutic antibodies. Neuron. 2015;88(2):289–97.CrossRefGoogle Scholar
  28. 28.
    Boswell CA, Mundo EE, Ulufatu S, Bumbaca D, Cahaya HS, Majidy N, et al. Comparative physiology of mice and rats: radiometric measurement of vascular parameters in rodent tissues. Mol Pharm. 2014;11(5):1591–8.CrossRefGoogle Scholar
  29. 29.
    Manzone TA, Dam HQ, Soltis D, Sagar VV. Blood volume analysis: a new technique and new clinical interest reinvigorate a classic study. J Nucl Med Technol. 2007;35(2):55–63 quiz 77, 9.CrossRefGoogle Scholar
  30. 30.
    Rehani MM, Sharma SK. Site of Tc-99m binding to the red blood cell: concise communication. J Nucl Med. 1980;21(7):676–8.PubMedGoogle Scholar
  31. 31.
    Pavel DG, Zimmer M, Patterson VN. In vivo labeling of red blood cells with 99mTc: a new approach to blood pool visualization. J Nucl Med. 1977;18(3):305–8.PubMedGoogle Scholar
  32. 32.
    Blumenthal RD, Osorio L, Ochakovskaya R, Ying Z, Goldenberg DM. Regulation of tumour drug delivery by blood flow chronobiology. Eur J Cancer. 2000;36(14):1876–84.CrossRefGoogle Scholar
  33. 33.
    Klein HG. Mollison’s blood transfusion in clinical medicine. In: Mollison PL, Anstee DJ, Mollison PL, editors. 11th ed. ed. Malden, Mass. Blackwell Pub.; 2005.Google Scholar
  34. 34.
    Lee HB, Blaufox MD. Blood volume in the rat. J Nucl Med. 1985;26(1):72–6.PubMedGoogle Scholar
  35. 35.
    Berman I, Carr R, Malone E. Determination of total blood volume from measurements of total red blood cell mass and plasma volume, using simultaneously injected isotopes. Nature. 1964;202:1013–5.CrossRefGoogle Scholar
  36. 36.
    Aukland K, Reed RK. Interstitial-lymphatic mechanisms in the control of extracellular fluid volume. Physiol Rev. 1993;73(1):1–78.CrossRefGoogle Scholar
  37. 37.
    Levitt DG. The pharmacokinetics of the interstitial space in humans. BMC Clin Pharmacol. 2003;3(3):3.CrossRefGoogle Scholar
  38. 38.
    Shockley TR, Lin K, Sung C, Nagy JA, Tompkins RG, Dedrick RL, et al. A quantitative analysis of tumor specific monoclonal antibody uptake by human melanoma xenografts: effects of antibody immunological properties and tumor antigen expression levels. Cancer Res. 1992;52(2):357–66.PubMedGoogle Scholar
  39. 39.
    Sung C, Youle RJ, Dedrick RL. Pharmacokinetic analysis of immunotoxin uptake in solid tumors: role of plasma kinetics, capillary permeability, and binding. Cancer Res. 1990;50(22):7382–92.PubMedGoogle Scholar
  40. 40.
    Kallman RF, DeNardo GL, Stasch MJ. Blood flow in irradiated mouse sarcoma as determined by the clearance of xenon-133. Cancer Res. 1972;32(3):483–90.PubMedGoogle Scholar
  41. 41.
    Sapirstein LA. Regional blood flow by fractional distribution of indicators. Am J Phys. 1958;193(1):161–8.Google Scholar
  42. 42.
    Gullino PM, Grantham FH. Studies on the exchange of fluids between host and tumor. III. Regulation of blood flow in hepatomas and other rat tumors. J Natl Cancer Inst. 1962;28:211–29.PubMedGoogle Scholar
  43. 43.
    Hammersley PA, McCready VR, Babich JW, Coghlan G. 99mTc-HMPAO as a tumour blood flow agent. Eur J Nucl Med. 1987;13(2):90–4.CrossRefGoogle Scholar
  44. 44.
    Zanelli GD, Fowler JF. The measurement of blood perfusion in experimental tumors by uptake of 86Rb. Cancer Res. 1974;34(6):1451–6.PubMedGoogle Scholar
  45. 45.
    Cherry SR, Carnochan P, Babich JW, Serafini F, Rowell NP, Watson IA. Quantitative in vivo measurements of tumor perfusion using rubidium-81 and positron emission tomography. J Nucl Med. 1990;31(8):1307–15.PubMedGoogle Scholar
  46. 46.
    Mossberg KA, Mullani N, Gould KL, Taegtmeyer H. Skeletal muscle blood flow in vivo: detection with rubidium-82 and effects of glucose, insulin, and exercise. J Nucl Med. 1987;28(7):1155–63.PubMedGoogle Scholar
  47. 47.
    Hagemann CE, Ghotbi AA, Kjær A, Hasbak P. Quantitative myocardial blood flow with Rubidium-82 PET: a clinical perspective. Am J Nucl Med Mol Imaging. 2015;5(5):457–68.PubMedPubMedCentralGoogle Scholar
  48. 48.
    Mendell PL, Hollenberg NK. Cardiac output distribution in the rat: comparison of rubidium and microsphere methods. Am J Phys. 1971;221(6):1617–20.Google Scholar
  49. 49.
    Gabrielsson J, Weiner D. Pharmacokinetic and pharmacodynamic data analysis: concepts and applications 2016.Google Scholar
  50. 50.
    Brooks DJ, Beaney RP, Lammertsma AA, Leenders KL, Horlock PL, Kensett MJ, et al. Quantitative measurement of blood-brain barrier permeability using rubidium-82 and positron emission tomography. J Cereb Blood Flow Metab. 1984;4(4):535–45.CrossRefGoogle Scholar
  51. 51.
    Urva SR, Yang VC, Balthasar JP. Physiologically based pharmacokinetic model for T84.66: a monoclonal anti-CEA antibody. J Pharm Sci. 2010;99(3):1582–600.CrossRefGoogle Scholar
  52. 52.
    Davies B, Morris T. Physiological parameters in laboratory animals and humans. Pharm Res. 1993;10(7):1093–5.CrossRefGoogle Scholar
  53. 53.
    Graf JF, Scholz BJ, Zavodszky MI. BioDMET: a physiologically based pharmacokinetic simulation tool for assessing proposed solutions to complex biological problems. J Pharmacokinet Pharmacodyn. 2012;39(1):37–54.CrossRefGoogle Scholar
  54. 54.
    Sands H, Jones PL, Shah SA, Palme D, Vessella RL, Gallagher BM. Correlation of vascular permeability and blood flow with monoclonal antibody uptake by human Clouser and renal cell xenografts. Cancer Res. 1988;48(1):188–93.PubMedGoogle Scholar
  55. 55.
    Sands H, Shah SA, Gallagher BM. Vascular volume and permeability of human and murine tumors grown in athymic mice. Cancer Lett. 1985;27(1):15–21.CrossRefGoogle Scholar
  56. 56.
    Forsyth RP, Nies AS, Wyler F, Neutze J, Melmon KL. Normal distribution of cardiac output in the unanesthetized, restrained rhesus monkey. J Appl Physiol. 1968;25(6):736–41.CrossRefGoogle Scholar
  57. 57.
    Jilkina O, Xiang B, Kuzio B, Rendell J, Kupriyanov VV. Potassium transport in Langendorff-perfused mouse hearts assessed by 87Rb NMR spectroscopy. Magn Reson Med. 2005;53(5):1172–6.CrossRefGoogle Scholar
  58. 58.
    Evans MV, Dowd SM, Kenyon EM, Hughes MF, El-Masri HA. A physiologically based pharmacokinetic model for intravenous and ingested dimethylarsinic acid in mice. Toxicol Sci. 2008;104(2):250–60.CrossRefGoogle Scholar
  59. 59.
    Kim C-Y, Han JS, Suzuki T, Han S-S. Indirect indicator of transport stress in hematological values in newly acquired cynomolgus monkeys. J Med Primatol. 2005;34(4):188–92.CrossRefGoogle Scholar
  60. 60.
    Rosso MC, Badino P, Ferrero G, Costa R, Cordero F, Steidler S. Biologic data of cynomolgus monkeys maintained under laboratory conditions. PLoS One. 2016;11(6):e0157003.CrossRefGoogle Scholar
  61. 61.
    Khor SP, Bozigian H, Mayersohn M. Potential error in the measurement of tissue to blood distribution coefficients in physiological pharmacokinetic modeling. Residual tissue blood. II. Distribution of phencyclidine in the rat. Drug Metab Dispos. 1991;19(2):486–90.PubMedGoogle Scholar
  62. 62.
    Lockman PR, McAfee G, Geldenhuys WJ, Van der Schyf CJ, Abbruscato TJ, Allen DD. Brain uptake kinetics of nicotine and cotinine after chronic nicotine exposure. J Pharmacol Exp Ther. 2005;314(2):636–42.CrossRefGoogle Scholar
  63. 63.
    Park S, Sinko PJ. P-glycoprotein and mutlidrug resistance-associated proteins limit the brain uptake of saquinavir in mice. J Pharmacol Exp Ther. 2005;312(3):1249–56.CrossRefGoogle Scholar
  64. 64.
    O’Connor SW, Bale WF. Accessibility of circulating immunoglobulin G to the extravascular compartment of solid rat tumors. Cancer Res. 1984;44(9):3719–23.PubMedGoogle Scholar
  65. 65.
    Triplett JW, Hayden TL, McWhorter LK, Gautam SR, Kim EE, Bourne DW. Determination of gallium concentration in “blood-free” tissues using a radiolabeled blood marker. J Pharm Sci. 1985;74(9):1007–9.CrossRefGoogle Scholar
  66. 66.
    Yadav DB, Maloney JA, Wildsmith KR, Fuji RN, Meilandt WJ, Solanoy H, et al. Widespread brain distribution and activity following i.c.v. infusion of anti-beta-secretase (BACE1) in nonhuman primates. Br J Pharmacol. 2017;174(22):4173–85.CrossRefGoogle Scholar
  67. 67.
    Yu YJ, Atwal JK, Zhang Y, Tong RK, Wildsmith KR, Tan C, et al. Therapeutic bispecific antibodies cross the blood-brain barrier in nonhuman primates. Sci Transl Med. 2014;6(261):261ra154.CrossRefGoogle Scholar
  68. 68.
    Atwal JK, Chen Y, Chiu C, Mortensen DL, Meilandt WJ, Liu Y, et al. A therapeutic antibody targeting BACE1 inhibits amyloid-beta production in vivo. Sci Transl Med. 2011;3(84):84ra43.CrossRefGoogle Scholar
  69. 69.
    Gelman S, Fowler KC, Bishop SP, Smith LR. Cardiac output distribution and regional blood flow during hypocarbia in monkeys. J Appl Physiol (Bethesda, Md : 1985.). 1985;58(4):1225–30.CrossRefGoogle Scholar
  70. 70.
    Diehl KH, Hull R, Morton D, Pfister R, Rabemampianina Y, Smith D, et al. A good practice guide to the administration of substances and removal of blood, including routes and volumes. J Appl Toxicol. 2001;21(1):15–23.CrossRefGoogle Scholar
  71. 71.
    Riches AC, Sharp JG, Thomas DB, Smith SV. Blood volume determination in the mouse. J Physiol. 1973;228(2):279–84.CrossRefGoogle Scholar
  72. 72.
    Vacha J. Blood volume in inbred strain BALB/c, CBA/J and C57BL/10 mice determined by means of 59Fe-labelled red cells and 59Fe bound to transferrin. Physiologia Bohemoslovaca. 1975;24(5):413–9.PubMedGoogle Scholar
  73. 73.
    Chaplin H Jr, Mollison PL, Vetter H. The body/venous hematocrit ratio: its constancy over a wide hematocrit range. J Clin Invest. 1953;32(12):1309–16.CrossRefGoogle Scholar
  74. 74.
    Giannelli S Jr, Ayres SM, Fleming P, Conrad W, Schwartz MS, Gould H. Peripheral vascular volumes and whole body hematocrit during human heart lung bypass. Circulation. 1970;41(4):629–40.CrossRefGoogle Scholar
  75. 75.
    Kirsch KA, Johnson RF, Gorten RJ. The significance of the total-body hematocrit in measurements of blood compartments. J Nucl Med. 1971;12(1):17–21.PubMedGoogle Scholar
  76. 76.
    Fairbanks VF, Klee GG, Wiseman GA, Hoyer JD, Tefferi A, Petitt RM, et al. Measurement of blood volume and red cell mass: re-examination of 51Cr and 125I methods. Blood Cells Mol Dis. 1996;22(2):169–86 discussion 86a-86g.CrossRefGoogle Scholar
  77. 77.
    Fudenberg H, Baldini M, Mahoney JP, Dameshek W. The body hematocrit/venous hematocrit ratio and the “splenic reservoir”. Blood. 1961;17:71–82.PubMedGoogle Scholar
  78. 78.
    Bauer R, Haluszczynski I, Langhammer H, Bachmann W. In vivo/in vitro labeling of red blood cells with 99mTc. Eur J Nucl Med. 1983;8(5):218–22.CrossRefGoogle Scholar

Copyright information

© The Author(s) 2018

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Authors and Affiliations

  • Danielle Mandikian
    • 1
  • Isabel Figueroa
    • 1
  • Amy Oldendorp
    • 2
  • Hanine Rafidi
    • 1
  • Sheila Ulufatu
    • 2
  • Michelle G. Schweiger
    • 2
  • Jessica A. Couch
    • 2
  • Noel Dybdal
    • 2
  • Sean B. Joseph
    • 1
  • Saileta Prabhu
    • 1
  • Gregory Z. Ferl
    • 1
    • 3
  • C. Andrew Boswell
    • 1
    • 3
  1. 1.Preclinical and Translational Pharmacokinetics, Genentech Research and Early DevelopmentSouth San FranciscoUSA
  2. 2.Safety Assessment, Genentech Research and Early DevelopmentSouth San FranciscoUSA
  3. 3.Genentech Inc.South San FranciscoUSA

Personalised recommendations