Pharmacokinetic Allometric Scaling of Oligonucleotides

Abstract

The objective of this study was to evaluate the predictive performance of interspecies scaling of oligonucleotides to predict clearance and volume of distribution at steady state in humans from animal data. The human pharmacokinetic parameters were predicted using 1, 2, or at least 3 animal species. The results of the study indicated that the pharmacokinetic parameters of oligonucleotides can be predicted with reasonable accuracy in humans when at least 3 animal species are employed. On the other hand, allometric scaling based on 1 or 2 species or fixed coefficient or fixed exponent can be erratic and unreliable. Further work should be conducted in this direction.

Introduction

It is now a well-established fact that many physiological processes and organ sizes exhibit a power-law relationship with the body weight of the species. This relationship is the scientific basis of allometric scaling (BOXENBAUM, 1984; Boxenbaum and D'Souza, 1990; Chappell and Mordenti 1991).

Allometric equations extrapolate quantitative parameters over orders of magnitude of body weights and provide a method to estimate or predict a physiological process [blood flow, creatinine clearance (CL), heart rate, liver weight, kidney weight, and glomerular filtration rate, etc.] of several species, including humans. The one goal of interspecies pharmacokinetic scaling is to select the first-in-human dose based on predicted pharmacokinetic parameters (especially CL) from animals. The allometric equation (power function) is as follows: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes {10} {9} {7} {6}\begin{document} \begin{align*} Y = aW^{ \rm b} \tag{1} \end{align*} \end{document}

where Y is the parameter of interest, W is the body weight, and a and b are the coefficient and exponent of the allometric equation, respectively. The log transformation of equation 1 is represented as follows: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes {10} {9} {7} {6}\begin{document} \begin{align*} \log Y = \log a + b \ \log W \tag{2} \end{align*} \end{document}

where log a is the y-intercept, and b is the slope. The slope (b) of the equation indicates the rate of change between a parameter and body weight.

Three pharmacokinetic parameters, CL, volume of distribution, and elimination half-life, are regularly predicted from animals to humans. Over the years, several methods of scaling have been developed to predict these parameters from animals to humans (Mahmood, 2005).

CL is the most important pharmacokinetic parameter and considering the importance of CL, several investigators have attempted to improve the prediction of human CL from animals (Mahmood, 2005). All these methods have advantages and disadvantages (Mahmood, 2005).

Developing antisense oligonucleotides as drugs has drawn enormous interest in recent years. An oligonucleotide is a short nucleic acid polymer, typically with fifty or fewer bases. Oligonucleotides are characterized by the sequence of nucleotide residues consisting of the entire molecule. As compared to a small molecule (molecular weight <300 Da), the length and molecular weight of the oligonucleotide depend on the number of bases and can widely vary. Pharmacokinetics of several oligonucleotides in animals as well as humans have been well established (GEARY, 2009). The pharmacokinetics of oligonucleotides with a phosphorothioate backbone can be characterized as drugs that bind to plasma protein at >90%, rapidly distribute to tissues, and less than 5% of the drug is excreted in urine unchanged (GEARY, 2009).

Interspecies scaling of pharmacokinetic parameters (CL, volume of distribution, and elimination half-life) has been conducted on hundreds of drugs (small as well as macromolecules). Considering that interspecies scaling can reasonably predict human pharmacokinetic parameters for small molecules, therapeutic proteins, monoclonal antibodies, and coagulation factors, it was reasonable to apply allometric principles to oligonucleotides. Therefore, the objectives of this report were (1) to predict human CL and volume of distribution of oligonucleotides from animals, and (2) to evaluate if 1 or 2-species allometric scaling is as predictable as 3-species allometric scaling, as sometimes 3 animal species data may not be available for scaling.

Methods

Prediction of oligonucleotides CL in humans

A literature search was conducted to obtain pharmacokinetic parameters of oligonucleotides in animal species and humans. Eight oligonucleotides were studied. For 4 oligonucleotides, CL and volume of distribution data were available in at least 3 species whereas for 4 oligonucleotides, pharmacokinetic data were available in only 2 species. For volume of distribution, scaling was only possible for 4 oligonucleotides (3 at least from 3 species and 1 from 2 species). All oligonucleotides were administered intravenously to animals and humans. Human data were not included in the scaling, and predictions were made using a human body weight of 70 kg.

It was noted that the pharmacokinetics of oligonucleotides in humans are non-linear (both CL and volume of distribution appeared to be dose dependent); therefore, where available, human CL and volume of distribution values were reported as a range (based on lowest and highest dose) and the predicted values were compared against the lowest and the highest observed values. Pharmacokinetic scaling was conducted using 1, 2, or at least 3 or more species and the predicted values were compared with the observed human values.

3-species scaling

Scaling of CL from 3 species was based on equation 1 or 2, where the CLs of at least 3 species were plotted against the body weight.

2-species scaling

The following 2 methods were used to predict human CL:

(1) Scaling of CL from 2 species was based on equation 1 or 2, where the CLs of 2 species were plotted against the body weight.

(2) Tang et al. (2007) have suggested the following equations to predict human drug CL using rat-monkey or rat-dog data. In equations 3 and 4, a is the coefficient from 2 species scaling.

\documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes {10} {9} {7} {6}\begin{document} \begin{align*} \begin{split}{ \rm CL}_{ \rm human} = a_{ \rm rat \hbox{-} { \rm monkey}} * W^{0.650}{_{ \rm human}} \end{split}\tag{3} \end{align*}\end{document}

\documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes {10} {9} {7} {6}\begin{document} \begin{align*} \begin{split} { \rm CL}_{ \rm human} = a_{ \rm rat \hbox{- } { \rm dog}} * W^{0.628}{_{ \rm human}}\end{split}\tag{4} \end{align*}\end{document}

1-species scaling

The following 2 methods were used to predict human drug CL:

(1) When only one species is available, a fixed exponent of 0.75 for CL and 1.0 for volume of distribution is used. The following equation was used to predict human drug CL from 1 species. \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes {10} {9} {7} {6}\begin{document} \begin{align*} \hbox { \rm CL in human } = { \rm Animal \ CL } * \left( \frac { human \ weight } { animal \ weight } \right)^ { 0.75 } \tag { 5 } \end{align*}\end{document}

(2) Tang et al. (2007) have suggested the following equations to predict human drug CL from a single species using a fixed coefficient.

\documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes {10} {9} {7} {6}\begin{document}\begin{align*} ( { \rm i} ) \ { \rm CL}_{ \rm human} / { \rm kg} = 0.152 * ( { \rm CL}_{ \rm rat} / { \rm kg} ) \tag { 6 } \end{align*}\end{document}

\documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes {10} {9} {7} {6}\begin{document}\begin{align*} \quad \ ( { \rm ii} ) \ { \rm CL}_{ \rm human} / { \rm kg} = 0.407 * ( { \rm CL}_{ \rm monkey} / { \rm kg} ) \tag { 7 } \end{align*}\end{document}

\documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes {10} {9} {7} {6}\begin{document} \begin{align*} ( { \rm iii} ) \ { \rm CL}_{ \rm human} / { \rm kg} = 0.410 * ( { \rm CL}_{ \rm dog} / { \rm kg} ) \tag { 8 } \end{align*}\end{document}

Prediction of oligonucleotides volume of distribution at steady state in humans

Volume of distribution at steady state (V_ss) was predicted in humans according to equation 1 or 2 using 2 or at least 3 animal species (body weight against V_ss). A fixed exponent of 1.0 for V_ss was used for 1 species scaling as shown in equation 5 (animal CL was replaced by V_ss).

Statistical analysis

Prediction error was calculated in terms of ratio: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes {10} {9} {7} {6}\begin{document} \begin{align*} { \rm Ratio} = { \rm Predicted} / { \rm observed} \tag{9} \end{align*}\end{document}

A ratio of 1.0 indicates a perfect prediction. In this study a ratio between 0.5 and 2.0 was considered acceptable. A value of 0.5 indicates a prediction error of 100% (under-predicted) and 2.0 indicates a prediction error of 100% (over-predicted). Whether a 100% prediction error in either side in the CL and volume of distribution at steady state of oligonucleotides is acceptable or of any practical value for the first-in-human dose selection is not known.

Results

Prediction of human drug CL (at least 3-species allometric scaling)

Table 1 is the summary of the drugs and species used in this study. The exponents of simple allometry, the correlation coefficient between body weight and CL, and predicted and observed CL values are summarized in Table 2. A good correlation (>0.8) between body weight and CL was observed for all 4 drugs. The exponents of the simple allometry ranged from 0.503 to 0.835. Two out of 4 drugs were within the acceptable range (Table 2).

Table 1.

Name of Oligonucleotides and Species Used in the Allometric Scaling of Pharmacokinetic Parameters

Drugs	Species	Reference (animals)	Reference (humans)
ISIS 301012	m, r, mk	Yu et al. (2001)	Yu et al. (2007)
ISIS 104838	m, d, mk	Geary et al. (2003)	Sewell et al. (2002)
ISIS 5132	m, r, mk	Geary et al. (2001)	Stevenson et al. (1999)
ISIS 3521	m, r, mk	Geary et al. (2001)	Nemunaitis et al. (1999)
ISIS 2302	m, mk	Geary et al. (2001)	Glover et al. (1999)
ISIS 2503	r, mk	Yu et al. (2001)	Cunningham et al. (2001)
G3139	m, r	Dai et al. (2005), Raynaud et al. (1997)	Marcucci et al. (2005)
GEM91	r, mk	Zhang et al. (1995a), Grindel et al. (1998)	Zhang et al. (1995b)

m, mouse; r, rat; d, dog; mk, monkey.

Table 2.

Predicted and Observed Clearance (mL/Min) of Oligonucleotides (3 Species)

Drugs	Exponent	R²	Observed	Predicted	Ratio
ISIS 301012	0.506	0.999	48	14	0.29
ISIS 104838	0.643	0.981	48–142	54	0.38–1.13
ISIS 5132	0.835	0.970	85–169	110	0.65–1.29
ISIS 3521	0.503	0.816	81–237	13	0.05–0.16

Ratio=predicted/observed.

Prediction of human V_ss (at least 3-species allometric scaling)

The predicted and observed V_ss values are summarized in Table 3. A good correlation (>0.9) between body weight and V_ss was observed for all 3 drugs. The exponents of the allometry ranged from 0.310 to 0.764 (Table 3). Two out of 3 drugs were within the acceptable range (Table 3).

Table 3.

Predicted and Observed Volume of Distribution (Liters) of Oligonucleotides (3 Species)

Drugs	Exponent	R²	Observed	Predicted	Ratio
ISIS 104838	0.764	0.994	4.6–20.5	2.4	0.12–0.5
ISIS 5132	0.310	0.998	6.9–13.2	1.1	0.08–0.16
ISIS 3521	0.738	0.988	9.9	6.1	0.62

Ratio=predicted/observed.

Prediction of human CL (2-species allometric scaling)

The prediction of human drug CL was performed using 2 methods as described in the Methods section. In Tables 4 and 5, the results of the 2-species allometric scaling for the prediction of human oligonucleotides CL are shown. The results of the study (based on a log-log plot) indicated that the prediction error from 2-species allometric scaling can be highly variable and will depend on the combination of the species used in the scaling.

Table 4.

Predicted and Observed Clearance (mL/min) of Oligonucleotides (2 Species) ^a

Drugs	Species	Exponent	Observed	Predicted	Ratio
ISIS 301012	m, mk	0.506	48	14	0.29
	r, mk	0.525	48	15	0.31
	m, r	0.486	48	12	0.25
ISIS 104838	m, mk	0.570	48–142	32	0.23–0.67
	d, mk	1.205	48–142	211	1.49–4.40
	r, d	0.677	48–112	75	0.67–1.56
ISIS 5132	m, mk	0.833	85–169	134	0.79–1.58
	r, mk	1.084	85–169	282	1.67–3.32
	m, r	0.574	85–169	16	0.09–0.19
ISIS 3521	m, mk	0.510	81–237	18	0.08–0.22
	r, mk	1.092	81–237	108	0.45–1.33
	m, r	−0.118	81–237	0.12	0.0015–0.0005
ISIS 2302	r, mk	0.957	90–145	82	0.57–0.91
ISIS 2503	r, mk	0.678	50–168	29	0.17–0.58
G3139	m, r	0.795	130	85	0.65
GEM91	r. mk	0.959	20–32	19	0.59–0.95

Ratio=predicted/observed.

Based on log–log plot of body weight and clearance.

m, mouse; r, rat; d, dog; mk, monkey.

Table 5.

Predicted and Observed Clearance (mL/min) of Oligonucleotides Using a Fixed Exponents (2 Species)

Drugs	Species	Observed	Predicted	Ratio
ISIS 301012	r, mk	48	25	0.52
ISIS 104838	d, mk	48–142	18	0.13–0.38
ISIS 5132	r, mk	85–169	45	0.27–0.53
ISIS 3521	r, mk	81–237	21	0.09–0.26
ISIS 2302	r, mk	90–145	22	0.15–0.24
GEM91	r, mk	20–32	5	0.16–0.25

Rat-monkey from equation 3 and dog-monkey from equation 4.

Ratio=predicted/observed.

r, rat; d, dog; mk, monkey.

Prediction of human drug CL (2 species with fixed exponent)

The predicted oligonucleotides CL values in humans from 2-species scaling (rat and monkey) using equation 3 is shown in Table 5. The predicted CL values for all drugs were underestimated.

Prediction of human V_ss (2-species allometric scaling)

The predicted and observed V_ss values (from equation 1) using 2-species are summarized in Table 6. Like CL, the results indicated that the prediction error from 2-species allometric scaling could be highly variable and will depend on the combination of the species used in the scaling.

Table 6.

Predicted and Observed Volume of Distribution (Liters) of Oligonucleotides (2 Species) ^a

Drugs	Species	Exponent	Observed	Predicted	Ratio
ISIS 104838	m, mk	0.718	4.6–20.5	1.7	0.08–0.37
	d, mk	1.118	4.6–20.5	5.7	0.28–1.24
ISIS 5132	m, mk	0.309	6.9–13.2	1.1	0.08–0.16
	r, mk	0.330	6.9–13.2	1.2	0.08–0.17
ISIS 3521	m, mk	0.739	9.9	5.3	0.54
	r, mk	0.603	9.9	3.6	0.36
GEM91	r. mk	1.021	53–86	73	0.85–1.38

Ratio=predicted/observed.

Based on log-log plot of body weight and clearance.

m, mouse; r, rat; d, dog; mk, monkey.

Prediction of human CL from 1 species

The predicted and observed values of CL using 1 species with a fixed exponent or fixed coefficients are summarized in Table 7. There were uncertainties in the prediction of CL as the predictions were species dependent. For example, for ISIS 301012, the prediction error in CL from mouse, rat, and monkey was 108%, 6%, and 42%, respectively. This phenomenon was observed with all 8 drugs (Table 7). Like 1 species, with a fixed exponent, the predication of human drug CL obtained from fixed coefficients was also species dependent and unreliable (Table 7).

Table 7.

Predicted and Observed Clearance (mL/Min) of Oligonucleotides Using a Fixed Coefficient or a Fixed Exponents (1 Species)

	Observed	Fixed coefficient ^a		Fixed exponent 0.75
Drugs	Human	Predicted	Ratio	Predicted	Ratio
ISIS 301012
Mouse	48	NA	NA	100	2.08
Rat	48	32	0.67	51	1.06
Monkey	48	24	0.50	28	0.58
ISIS 104838
Mouse	48–142	NA	NA	137	0.96–2.85
Dog	48–142	58	0.41–1.21	87	0.61–1.81
Monkey	48–142	46	0.32–0.96	54	0.38–1.13
ISIS 5132
Mouse	85–169	NA	NA	68	0.40–0.80
Rat	85–169	27	0.16–0.32	44	0.26–0.52
Monkey	85–169	89	0.53–1.05	105	0.62–1.24
ISIS 3521
Mouse	81–237	NA	NA	141	0.59–1.74
Rat	81–237	15	0.06–0.19	24	0.10–0.30
Monkey	81–237	33	0.14–0.41	39	0.16–0.48
ISIS 2302
Rat	90–145	15	0.10–0.17	25	0.17–0.28
Monkey	90–145	36	0.25–0.40	43	0.30–0.48
ISIS 2503
Mouse	50–168	NA	NA	50	0.30–1.0
Monkey	50–168	28	0.17–0.56	34	0.20–0.68
G3139
Mouse	130	NA	NA	59	0.45
Rat	130	48	0.37	67	0.52
GEM91
Rat	20–132	4	0.03–0.20	5	0.04–0.25
Monkey	20–132	9	0.07–0.45	10	0.08–0.50

Ratio=predicted/observed.

From equations 6 –8.

NA, fixed coefficient has not been established for mouse.

Prediction of human V_ss from 1 species

The results of this study are summarized in Table 8. The predicted human V_ss, with a fixed exponent, was species dependent and unreliable.

Table 8.

Predicted and Observed Volume of Distribution (Liters) of Oligonucleotides Using a Fixed Exponent 1.0 ^a

Drugs	Observed (human)	Predicted	Ratio
ISIS 104838
Mouse	4.6–20.5	17.2	0.84–3.74
Dog	4.6–20.5	4.5	0.22–0.98
Monkey	4.6–20.5	4.0	0.19–0.87
ISIS 5132
Mouse	6.9–13.2	304.5	23.1–44.1
Rat	6.9–13.2	50.4	3.82–7.30
Monkey	6.9–13.2	8.6	0.65–1.25
ISIS 3521
Mouse	9.9	45.5	4.60
Rat	9.9	33.6	3.39
Monkey	9.9	11.8	1.19
GEM91
Rat	53–86	63	0.73–1.19
MK	53–86	68	0.79–1.28

Ratio=predicted/observed.

From equation 5; replacing animal clearance by animal volume of distribution and exponent 0.75 by 1.0.

Discussion

The pharmacokinetic interspecies scaling of oligonucleotides (although small sample size) indicates that it is possible to predict pharmacokinetic parameters of these compounds in humans with reasonable accuracy from animal data.

In this study, the exponents of the simple allometry for CL (3-species scaling) ranged from 0.506 to 0.835. This wide range of exponent is common in interspecies scaling and has also been observed with conventional drugs, therapeutic proteins, and antibodies. Due to this wide range of exponents, especially for CL, simple allometry alone is not adequate for the prediction of drug CL in humans especially for small molecules. Therefore, to improve the prediction of drug CL for small molecules, many approaches have been suggested (Mahmood, 2005). The most commonly used method for small molecule drugs is the rule of exponents. The application of the rule of exponents for therapeutic proteins and antibodies is slightly different than the small molecule drugs and has been outlined by Mahmood (2004, 2005, 2009a). This rule may also be applicable to oligonucleotides but due to the lack of data this could not be assessed in this report.

In this study, there were 4 drugs, for which data from at least 3 animal species were available (Table 3). A very good prediction of CL for 2 drugs (ISIS 104838 and ISIS 5132) was noted but the other 2 drugs were substantially under-predicted resulting in high prediction error. The exponents of allometry for these 2 drugs were around 0.5 and as pointed out by Mahmood (2005) that when the exponents of allometry are <0.55 then there is a high likelihood that the predicted CL values will be lower than the observed CL values which was the case for 2 oligonucleotides. However, with exponent <0.55 the under-predicted CL values as compared to observed CL values may be substantial or negligible. In the case of 2 oligonucleotides, the prediction error was substantial.

For a reasonably good prediction (error <50%) of a pharmacokinetic parameter, one requires at least 3 animal species. Three animal species, however, may not be always available and one may want to conduct pharmacokinetic scaling based on 1 or 2 species. In general, 1 or 2 species scaling has not been found successful (BOXENBAUM, 1982; Caldwell et al., 2004; Goteti et al., 2010).

In this report, 2 approaches have been taken to predict human drug CL using 2 species. The use of allometric scaling and the fixed exponent (equations 1, 3, and 4) provided comparable results, but due to small sample size it was not possible to compare the predictive performance of 3-species scaling with 2-species scaling. However, in 2-species scaling, it is not possible to determine a priori which combination of 2-species will give a reasonably accurate prediction of human drug CL. For example, it can be seen that for ISIS 104838, ISIS 5132, and ISIS 3521, the predicted CLs in humans is highly variable depending on the choice of the 2-species (Table 4).

For 2-species scaling (based on the plot of CL and body weight; equation 1), the data from 8 drugs and 16 pairs of 2-species were available (Table 4). Like 3-species scaling, it was noted that when the exponents of allometry were <0.55 then the predicted CLs for these drugs were underestimated. It was also interesting to note that when the exponents of allometry were >1.0, then the predicted CL values were over-estimated. This phenomenon is not surprising and was noted with small molecules, therapeutic proteins, and monoclonal antibodies.

One species scaling has been suggested based on fixed exponents. These exponents are 0.75 and 1.0 for CL and volume of distribution, respectively. The results of the study indicted uncertainty in the prediction of human pharmacokinetic parameters as the prediction of these parameters was species dependent (Table 7). These observations were also noted with conventional drugs, therapeutic proteins, and antibodies. Overall, the data suggest that occasionally one may get a reasonable prediction of human pharmacokinetic (PK) parameters with 1 species and a fixed exponent scaling but there remains uncertainty in the predicted values because one will not know a priori which species is a suitable species and which parameter it can predict with reasonable accuracy. Further, as seen in this study and with small molecules, therapeutic proteins, and antibodies, the exponents of allometry widely vary and the use of a fixed exponent is highly unreliable.

Equations 6 –8 advocate the use of a fixed coefficient rather than a fixed exponent. The results of the analysis indicate that like the fixed exponent, the application of a fixed coefficient can be erratic, species-dependent, and unreliable (Table 7).

A single species may not predict all the human pharmacokinetic parameters (CL, volume of distribution, and half-life) with same degree of accuracy or error. For example, for ISIS 5132, there was substantial difference in the prediction error of CL and volume of distribution using mouse or rat data but monkey provided a comparable prediction error for these 2 pharmacokinetic parameters. Overall, the data suggest that occasionally one may get a reasonable prediction of human PK parameters with 1 species and a fixed exponent or a fixed coefficient, but there remains uncertainty in the predicted values because one will not know a priori which species is a suitable species and which parameter it can predict with reasonable accuracy. Further, considering the wide variability in the exponents of allometry, the application of a fixed exponent appears to be inappropriate.

For first-in-human dosing, some investigators advocate that a 100% (or 2-fold) prediction error is acceptable; nonetheless, the human margin of safety may make this criterion inadequate and dangerous.

Two of the widely used methods for the first-in-human-dose selection are based on the no adverse effect level (NOAEL) or toxicokinetic studies of a drug in 1 to 3 species. There are several disadvantages to these approaches. In the NOAEL approach, it becomes necessary to select a NOAEL, which is a tedious and time consuming process. In reality, one may never find an absolute NOAEL in a given species. In toxicokinetic studies, animals are given a very high dose chronically, and if the resultant toxicity does not kill the animal, it may alter the physiology of the animal. This change in the physiology of the animal may have an impact on the pharmacokinetics of a given drug. Overall, both the NOAEL and toxicokinetic approaches for the first-time selection of a dose to humans are slow and conservative approaches. Therefore, a pharmacokinetic approach based on predicted human drug CL was suggested (Boxenbaum and Dilea, 1995; Reigner and Blesch, 2002). Mahmood et al. (2003) using CL and different approaches have shown the usefulness of the approach for conventional drugs, therapeutic proteins, and antibodies. These suggested approaches are not only time and cost effective but also provide rational alternatives to the somewhat arbitrary dose selection process often used. Having a clear set of methods rather than relying on some fuzzy approach should be an important advantage in the bigger context of drug development.

Conclusions

(1) Like conventional drugs, therapeutic proteins, and antibodies, exponents of the simple allometry for oligonucleotides also widely vary. This wide range of exponent is common in interspecies scaling. The exponents of the simple allometry have no physiological meaning and will vary depending on the number and types of species used in the scaling; therefore, a proper understanding about the nature of allometric exponents for the prediction of CL is necessary.

(2) The wide range of exponents observed for oligonucleotides indicates that the use of a fixed exponent is inappropriate, since it may be substantially inaccurate.

(3) Two-species scaling may be useful, but the predicted parameters may not be as accurate as from 3 or more-species scaling (although not established in this study due to small size).

(4) One of the most misleading concepts in allometric scaling of pharmacokinetic parameters is the notion of a fixed exponent. As seen here and elsewhere (MAHMOOD, 2009b), this method frequently leads to highly erroneous predictions.

(5) In general, one should neither use a single species nor a fixed exponent or a fixed coefficient for the prediction of human pharmacokinetic parameters. If for some unforeseen reason one has to use fixed exponent or a single species, then extreme caution in the interpretation of data is needed as the predicted pharmacokinetic parameters from a single species using a fixed exponent or fixed coefficient may be in gross error.

(6) The goodness of fit or a strong correlation between body weight and a pharmacokinetic parameter does not usually translate into a good prediction of human pharmacokinetic parameters. For example, an improvement in correlation coefficient between body weight and a pharmacokinetic parameter from 0.90 to 0.99 will not necessarily give an improved prediction of a pharmacokinetic parameter.

(7) Many external factors such as experimental design, species, analytical errors, and physicochemical properties of drugs may have impact on allometric extrapolation. The analytical method to determine plasma concentrations can have impact on the estimated pharmacokinetic parameters. One should also be careful scaling PK parameters using different biological fluids for a given drug. For example, scaling a drug in 1 species using PK parameters obtained from blood and in other species from plasma or serum.

(8) Although the number of studied oligonucleotides in this report is very small, the results mirror those of conventional drugs, therapeutic proteins, and antibodies.

(9) In short, it is possible to predict human PK parameters for oligonucleotides with reasonable accuracy using the principles of allometry and at least 3 animal species. Two-species allometry also may provide reasonable prediction of human PK parameters of oligonucleotides, but finding the combination of 2 suitable species will remain a challenge.

(10) With the availability of more oligonucleotides data, further work in this direction should be conducted.

Acknowledgments

The views expressed in this article are those of the author and do not reflect the official policy of the FDA. No official support or endorsement by the FDA is intended or should be inferred.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

BOXENBAUM

1982. Comparative pharmacokinetics of benzodiazepines in dog and man. J. Pharmacokinet. Biopharm., 10:411–426.

BOXENBAUM

1984. Interspecies pharmacokinetic scaling and the evolutionary—comparative paradigm. Drug Metab. Rev., 15:1071–1121.

BOXENBAUM

, DILEA

1995. First-time-in-human dose selection: allometric thoughts and perspectives. J. Clin. Pharmacol., 35:957–966.

BOXENBAUM

, D'SOUZA

R.W.

1990. Interspecies pharmacokinetic scaling, biological design and neoteny. Adv. Drug Res., 19:139–196.

CALDWELL

G.W.

, MASCUSSI

J.A.

, YAN

, HAGEMAN

2004. Allometric scaling of pharmacokinetic parameters in drug discovery: can human CL, Vss and t1/2 be predicted from in-vivo rat data? Eur. J. Drug Metab. Pharmacokinet., 29:133–143.

CHAPPELL

W.R.

, MORDENTI

1991. Extrapolation of toxicological and pharmacological data from animals to humans. Adv. Drug Res., 20:1–116.

CUNNINGHAM

C.C.

, HOLMLUND

J.T.

, GEARY

R.S.

, KWOH

T.J.

, DORR

, JOHNSTON

J.F.

, MONIA

, NEMUNAITIS

2001. A phase I trial of H-ras antisense oligonucleotide ISIS 2503 administered as a continuous intravenous infusion in patients with advanced carcinoma. Cancer, 92:1265–1271.

DAI

, WEI

, LIU

, MARCUCCI

, CHAN

K.K.

2005. Characterization and quantification of Bcl-2 antisense G3139 and metabolites in plasma and urine by ion-pair reversed phase HPLC coupled with electrospray ion-trap mass spectrometry. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci., 825:201–213.

GEARY

R.S.

2009. Antisense oligonucleotide pharmacokinetics and metabolism. Expert Opin. Drug Metabol. Toxicol., 5:381–391.

10.

GEARY

R.S.

, YU

R.Z.

, LEVIN

A.A.

2001. Pharmacokinetics of phosphorothioate antisense oligodeoxynucleotides. Curr. Opin. Investig. Drugs., 2:562–573.

11.

GEARY

R.S.

, YU

R.Z.

, WATANABE

T.A.

, HENRY

S.P.

, HARDEE

G.E.

, CHAPPELL

, MATSON

, SASMOR

, CUMMINS

, LEVIN

A.A.

2003. Pharmacokinetics of a tumor necrosis factor-alpha phosphorothioate 2'-O-(2-methoxyethyl) modified antisense oligonucleotide: comparison across species. Drug Metab. Dispos., 31:1419–1428.

12.

GLOVER

J.M.

, LEEDS

J.M.

, MANT

T.G.

, AMIN

, KISNER

D.L.

, ZUCKERMAN

J.E.

, GEARY

R.S.

, LEVIN

A.A.

, SHANAHAN

W.R.

JR.

1999. Phase I safety and pharmacokinetic profile of an intercellular adhesion molecule-1 antisense oligodeoxynucleotide (ISIS 2302) J. Pharmacol. Exp. Ther., 282:1173–1180.

13.

GOTETI

, GARNER

C.E.

, MAHMOOD

2010. Prediction of human drug clearance from two species: a comparison of several allometric methods. J. Pharm. Sci., 99:1601–1613.

14.

GRINDEL

J.M.

, MUSICK

T.J.

, JIANG

, ROSKEY

, AGRAWAL

1998. Pharmacokinetics and metabolism of an oligodeoxynucleotide phosphorothioate (GEM91) in cynomolgus monkeys following intravenous infusion. Antisense Nucleic Acid Drug Dev., 8:43–52.

15.

MAHMOOD

2004. Interspecies scaling of protein drugs: prediction of clearance from animals to humans. Pharm. Sci., 93:177–185.

16.

MAHMOOD

2005. Interspecies scaling of clearance. Interspecies Pharmacokinetic Scaling: Principles and Application of Allometric Scaling. Pine House Publishers: Rockville, MD, 39–85.

17.

MAHMOOD

2009a. Pharmacokinetic allometric scaling of antibodies: application to the first-in-human dose estimation. J. Pharm. Sci., 98:3850–3861.

18.

MAHMOOD

2009b. Application of fixed exponent 0.75 to the prediction of human drug clearance: an inaccurate and misleading concept. Drug Metabol. Drug Interact., 24:57–81.

19.

MAHMOOD

, GREEN

M.D.

, FISHER

J.E.

2003. Selection of the first-time dose in humans: comparison of different approaches based on interspecies scaling of clearance. J. Clin. Pharmacol., 43:692–697.

20.

MARCUCCI

, STOCK

, DAI

, KLISOVIC

R.B.

, LIU

, KLISOVIC

M.I.

, BLUM

, KEFAUVER

, SHER

D.A.

, GREEN

, MORAN

, MAHARRY

, NOVICK

, BLOOMFIELD

C.D.

, ZWIEBEL

J.A.

, LARSON

R.A.

, GREVER

M.R.

, CHAN

K.K.

, BYRD

J.C.

2005. Phase I study of oblimersen sodium, an antisense to Bcl-2, in untreated older patients with acute myeloid leukemia: pharmacokinetics, pharmacodynamics, and clinical activity. J. Clin. Oncol., 23:3404–3411.

21.

NEMUNAITIS

, HOLMLUND

J.T.

, KRAYNAK

, RICHARDS

, BRUCE

, OGNOSKIE

, KWOH

T.J.

, GEARY

, DORR

, VON HOFF

, ECKHARDT

S.G.

1999. Phase I evaluation of ISIS 3521, an antisense oligodeoxynucleotide to protein kinase C-alpha, in patients with advanced cancer. J. Clin. Oncol., 17:3586–3595.

22.

RAYNAUD

F.I.

, ORR

R.M.

, GODDARD

P.M.

, LACEY

H.A.

, LANCASHIRE

, JUDSON

I.R.

, BECK

, BRYAN

, COTTER

F.E.

1997. Pharmacokinetics of G3139, a phosphorothioate oligodeoxynucleotide antisense to bcl-2, after intravenous administration or continuous subcutaneous infusion to mice. J. Pharmacol. Exp. Ther., 281:420–427.

23.

REIGNER

B.G.

, BLESCH

K.S.

2002. Estimating the starting dose for entry into humans: principles and practice. Eur. J. Clin. Pharmacol., 57:835–845.

24.

SEWELL

K.L.

, GEARY

R.S.

, BAKER

B.F.

, GLOVER

J.M.

, MANT

T.G.

, YU

R.Z.

, TAMI

J.A.

, DORR

F.A.

2002. Phase I trial of ISIS 104838, a 2'-methoxyethyl modified antisense oligonucleotide targeting tumor necrosis factor-alpha. J. Pharmacol. Exp. Ther., 303:1334–1343.

25.

STEVENSON

J.P.

, YAO

K.S.

, GALLAGHER

, FRIEDLAND

, MITCHELL

E.P.

, CASSELLA

, MONIA

, KWOH

T.J.

, YU

, HOLMLUND

, DORR

F.A.

, O'DWYER

P.J.

1999. Phase I clinical/pharmacokinetic and pharmacodynamic trial of the c-raf-1 antisense oligonucleotide ISIS 5132 (CGP 69846A) J. Clin. Oncol., 17:2227–2236.

26.

TANG

, HUSSAIN

, LEAL

, MAYERSOHN

, FLUHLER

2007. Interspecies prediction of human drug clearance based on scaling data from one or two animal species. Drug Metab. Dispos., 35:1886–1893.

27.

R.Z.

, GEARY

R.S.

, LEEDS

J.M.

, WATANABE

, MOORE

, FITCHETT

, MATSON

, BURCKIN

, TEMPLIN

M.V.

, LEVIN

A.A.

2001. Comparison of pharmacokinetics and tissue disposition of an antisense phosphorothioate oligonucleotide targeting human Ha-ras mRNA in mouse and monkey. J. Pharm. Sci., 90:182–193.

28.

R.Z.

, KIM

T.W.

, HONG

, WATANABE

T.A.

, GAUS

H.J.

, GEARY

R.S.

2007. Cross-species pharmacokinetic comparison from mouse to man of a second-generation antisense oligonucleotide, ISIS 301012, targeting human apolipoprotein B-100. Drug Metab. Dispos., 35:460–468.

29.

ZHANG

, DIASIO

R.B.

, LU

, LIU

, JIANG

, GALBRAITH

W.M.

, AGRAWAL

1995a. Pharmacokinetics and tissue distribution in rats of an oligodeoxynucleotide phosphorothioate (GEM 91) developed as a therapeutic agent for human immunodeficiency virus type-1. Biochem. Pharmacol., 49:929–939.

30.

ZHANG

, YAN

, SHAHINIAN

H.K.

, AMIN

, LU

, LIU

, SAAG

M.S.

, JIANG

, TEMSAMANI

, MARTIN

R.R.

1995b. Pharmacokinetics of an anti-human immunodeficiency virus antisense oligodeoxynucleotide phosphorothioate (GEM 91) in HIV-infected subjects. Clin. Pharmacol. Ther., 58:44–53.

Pharmacokinetic Allometric Scaling of Oligonucleotides

Abstract

Introduction

Methods

Prediction of oligonucleotides CL in humans

3-species scaling

2-species scaling

1-species scaling

Prediction of oligonucleotides volume of distribution at steady state in humans

Statistical analysis

Results

Prediction of human drug CL (at least 3-species allometric scaling)

Prediction of human Vss (at least 3-species allometric scaling)

Prediction of human CL (2-species allometric scaling)

Prediction of human drug CL (2 species with fixed exponent)

Prediction of human Vss (2-species allometric scaling)

Prediction of human CL from 1 species

Prediction of human Vss from 1 species

Discussion

Conclusions

Acknowledgments

Footnotes

Author Disclosure Statement

References

Prediction of human V_ss (at least 3-species allometric scaling)

Prediction of human V_ss (2-species allometric scaling)

Prediction of human V_ss from 1 species