UEF//eRepository
DSpace https://erepo.uef.fi
Rinnakkaistallenteet Terveystieteiden tiedekunta
2016
Prediction of ocular drug distribution from systemic blood circulation
Vellonen, Kati-Sisko
American Chemical Society
article
info:eu-repo/semantics/acceptedVersion
© American Chemical Society All rights reserved
http://dx.doi.org/10.1021/acs.molpharmaceut.5b00729
https://erepo.uef.fi/handle/123456789/26766
Downloaded from University of Eastern Finland's eRepository
1
Prediction of ocular drug distribution from systemic blood circulation
Kati-Sisko Vellonena, Esa-Matti Soinib, Eva M. del Amo a, Arto Urtti a,b*
a School of Pharmacy, University of Eastern Finland, Kuopio, Finland
b Centre for Drug Research, Division of Pharmaceutical Biosciences, Faculty of Pharmacy, University of Helsinki, Finland
*Corresponding author. School of Pharmacy, University of Eastern Finland, P.O. Box 1627, 70211 Kuopio, Finland. Tel. +358 40540 2279 , Fax: +358 17 162424, Email:
arto.urtti@uef.fi
This document is the Accepted Manuscript version of a Published Work that appeared in final form in Molecular Pharmaceutics copyright © American Chemical Society after peer review and technical editing by the publisher. To access the final edited and published work see https://doi.org/10.1021/acs.molpharmaceut.5b00729
2 TABLE OF CONTENTS/ABSTRACT GRAPHIC
ABSTRACT
Systemically circulating drugs may distribute to the ocular tissues across the blood-ocular barriers. Ocular distribution is utilized in the treatment of ocular diseases with systemic medications, but ocular delivery of systemic drugs and xenobiotics may also lead to ocular adverse effects. Ocular distribution after systemic drug administration has not been predicted or modeled. In this study, distribution clearance between vitreous and plasma was obtained from a previous QSPR model for clearance of intravitreal drugs. These values were used in a pharmacokinetic simulation model to describe entry of unbound drug from plasma to vitreous.
The simulation models predicted ocular distribution of 10 systemic drugs in rabbit eyes within 1.96 mean fold error and the distribution of cefepime from plasma to vitreous in humans. This is the first attempt to predict ocular distribution of systemic drugs. Reliable predictions were obtained using systemic concentrations of unbound drug, computational value of ocular distribution clearance, and a simple pharmacokinetic model. This approach can be used in drug discovery to estimate ocular drug exposure at early stage.
KEYWORDS: QSPR, pharmacokinetic simulation, clearance, ocular pharmacokinetics, systemic administration
3
ABBREVIATIONS:AUC0-last Area under the concentration vs time curve from 0 to last time point; BCRP Breast cancer resistance protein; CLBV distribution clearance between blood and vitreous; CLivt intravitreal clearance; Cu unbound drug in plasma; fu fraction of unbound; HD hydrogen bond donors; MRP4 Multidrug resistance-associated protein 4; OAT3 Organic anion transporter 3; QSPR quantitative structure property relationships; P-gp P-glycoprotein;
Vss,ivt intravitreal volume of distribution.
1. INTRODUCTION
Systemic drugs circulate through ocular tissues, for example iris, ciliary body, choroid, and retina. More than 85% of ocular blood flow takes place in the choroid, where blood flow is 43 ml/h (1). Choroidal blood vessels are fenestrated and allow easy drug distribution from the blood stream to the extravascular choroid (2, 3). However, access of choroidal drug into the retina is limited by the retinal pigment epithelium barrier (3, 4). Blood vessel walls in the iris, and retina have tight junctions between the endothelial cells that slow down drug permeation across the vessel walls (3, 5). Small and lipophilic compounds can permeate across the blood- ocular barriers (retinal pigment epithelium, the inner non-pigmented ciliary epithelium,and posterior iris epithelium and blood vessel walls of iris, and retina), while permeation of hydrophilic compounds and large molecules is restricted (3, 6, 7).
Ocular distribution of systemic drugs is utilized in ocular drug treatment. For instance, acetazolamide tablets have been used in glaucoma treatment, systemic corticosteroids and antibodies in the treatment of uveitis, mannitol infusion in high ocular hypertension, and intravenous antibiotics are used in the treatment of endophthalmitis (8-10). Often, high doses must be used to deliver enough drug into the eye leading to systemic adverse effects.
4
The eye is exposed to such drugs and xenobiotics from plasma that are not used for ophthalmic treatment. Elderly patients are frequently using more than 10 medications in parallel. Ocular exposure to the chemicals and drugs may cause adverse ocular reactions. These have been compiled in several review articles and book chapters (11-14). For example, chloroquine, sildenafil, chlorpromazine, carmustine and vigabatrine can cause retinal side effects. Some effects can be relatively rare and transient in acute use, but may become more serious and common during long-term medications. Drug distribution to the eye across the barriers is one of the key factors behind the ocular adverse effects of systemic medications.
Scientific literature contains plenty of information about the ocular adverse effects of systemic medications, but the related information on pharmacokinetics is sparse. Even though the structure of blood-ocular barriers and the main trends of their permeability have been known for decades, no one has systematically predicted or modeled drug distribution from blood to the eye. Model for ocular drug distribution would be a useful tool in drug discovery. The simulation model could be used to estimate delivery of ophthalmic drugs and exposure and risks of other drugs and chemicals.
In this study, we built a pharmacokinetic simulation model for prediction of drug concentrations in the vitreous based on the unbound drug concentrations in the plasma and the computational estimate for the distribution clearance between the blood circulation and the eye.
The model predicted drug distribution to the eye accurately and it may become a useful tool for drug discovery and development.
5 2. MATERIALS AND METHODS
Compartmental model was developed with STELLA® Software (version 8.1.1, isee systems, USA) for simulation of drug concentrations in vitreous and blood circulation after systemic administration. The structure of the simulation model is depicted in Figure 1.
Figure 1. Compartmental model for drug distribution from the systemic circulation to the ocular vitreous.
Literature search was performed to find studies where drug has been administered into systemic circulation in rabbit or human, and drug concentrations were measured from both blood and vitreous samples. If the pharmacokinetic parameters for compartmental model (volume of distribution, clearance and micro-constants) were not reported, they were solved using curve fitting with Phoenix™ WinNonlin® (Pharsight, Certara L.P., USA) (Supplement Table 1). If the concentrations were not reported in numerical format, they were extracted from the concentration vs. time graphs with GetData Graph Digitizer (version 2.26., Germany).
Drug distribution clearance from the blood circulation into the eye was assumed to be equal to drug clearance from the vitreous to the blood circulation. This assumption is valid if there is no significant active and directional drug transport in the blood-ocular barriers. Rate of drug transfer from blood to the eye JBV = CLBV x Cu (where CLBV is the distribution clearance between blood and vitreous and Cu is the unbound drug in plasma) was calculated with previously established QSPR equation (1) for intravitreal drug clearance (CLivt) in rabbit eyes (6). We assume that CLBV = CLivt. Therefore,
6
Log CLBV = Log CLivt = -0.25269 - 0.53747 (LogHD) + 0.05189 (LogD7.4) (1)
This QSPR model has been built using all available rabbit data on intravitreal injections of small molecular drugs in solution. Extracellular proteins do not permeate across the blood- ocular barriers significantly; therefore, we used unbound drug concentrations in the simulations. Drug concentrations in plasma were multiplied by fraction of unbound (fu) drug in plasma. When available, experimental values for intravitreal clearance (CLivt) and intravitreal volume of distribution (Vss,ivt) in rabbit were used (this was the case for ciprofloxacin, fluconazole, foscarnet, methotrexate). Otherwise, the Vss,ivt in rabbit vitreous was assumed to be the median value (1.48 ml) reported in a previous study (6) and CLivt was calculated from the QSPR model (Eq. 1). Simulations were carried out using both experimental parameters and the calculated parameters (Vss,ivt = 1.48 ml and CLivt QSPR model).
Human serum and vitreous concentrations for cefepime were simulated using CLBV value for human eyes. CLBV was obtained by calculating the rabbit value with QSPR model (1) and then translating it to the human values based on Eq. 2 (7):
Human apparent CLBV = Human apparent CLivt = (Rabbit CLivt - 0.04)/1.41 (2)
Volume of cefepime distribution was assumed to be equal to the anatomical volume of human vitreous (4 ml; (15)) or two times larger (8 ml). The simulated concentrations were compared to the experimental ones obtained in human in plasma and in vitreous after systemic administration of cefepime (16).
7
The compounds and parameter values in the simulations are compiled in supplement Table 1.
Number of hydrogen bond donors (HD) and LogD7.4 values were obtained from ChemSpider database (http://www.chemspider.com). The simulations were carried out using Runge-Kutta 4 algorithm. Simulations were continued until the time point of the last measured sample and AUC0-last was calculated.
3. RESULTS
Predictions in rabbits. The rabbit simulations were done for 10 compounds and human simulations with one drug. The requirement was that drug concentrations from plasma and vitreous must be available. All compounds obeyed two-compartment kinetics in systemic circulation. After curve fitting, the resulting kinetic parameters were incorporated to the simulation model for intravenous drugs and used to simulate the concentrations in plasma. The simulations matched the experimental data well (Fig. 2).
Figure 2. Experimental (solid line) and simulated (dashed line) concentrations of the drugs in systemic circulation (plasma or serum) of rabbit after intravenous administration. The drugs ciprofloxacin, cytarabine, fleroxacin, fluconazole and foscarnet are presented in panel A and mercaptopurine, methotrexate, ofloxacin, sparfoxacin and topotecan in panel B (see Supplement Table 1 for original references).
8
Figure 3. Experimental (solid line) and simulated (discontinuous line) drug concentrations in rabbit vitreous. The dashed lines represent simulations where CLivt is defined by (Eq. 1) (QSPR model) and Vss,ivt is the reference value of 1.48 ml. The dotted lines represent simulations using the experimental CLivt and Vss,ivt values of drugs (Sim. Exp.) The drugs ciprofloxacin, cytarabine, fleroxacin, fluconazole and foscarnet are presented in panel A and mercaptopurine, methotrexate, ofloxacin, sparfoxacin and topotecan in panel B (see Supplement Table 1 for the original references).
Pharmacokinetic model based on computational distribution clearance and free drug concentrations in plasma was capable of predicting the drug concentrations in vitreous (Fig. 3).
The vitreal concentrations of the studied compounds spanned 1000-fold range of concentrations, and yet the simulated drug concentrations were relatively close to the experimental values in each case. The simulations with computational and experimental values of distribution clearance showed similar predictability (Fig. 3).
The AUC values in the vitreous represent the total drug exposure. The simulated (based on QSPR model) and experimental AUC0-last values showed high correlation (ρ= 0.90) (Fig. 4, Table 2). The absolute differences between the simulated and experimental AUC0-last values were two-fold or less in 8/10 cases and the maximal difference was 3.8-fold.
9
Figure 4. Correlation of measured and simulated AUC0-last in rabbit vitreous after intravenous administration. Drug distribution clearance between systemic circulation and eye was defined computationally (QSPR; solid circles) (CLivt from Eq. 1 and Vss,ivt of 1.48 ml) or based on the experimental values of CLivt and Vss,ivt (Exp.; open circles) were used in the model. Solid line represents the slope of 1 and dotted lines 2-fold deviation from 1:1 relationship.
For four compounds (ciprofloxacin, foscarnet, methotrexate, fluconazole) experimental values for Vss,ivt and CLivt were available (6). These values were used in the simulation models as ocular volume of distribution and distribution clearance from the blood circulation into the eye (Fig. 4). The AUC0-last values from simulations with computational and experimental values of distribution clearance were similar. Foscarnet showed 2.1-fold difference, but for three other drugs the differences were less than 11%. Overall, the use of computational distribution clearance results in reliable predictions of ocular drug distribution from blood circulation.
Prediction in humans. The intravitreal volume of distribution in rabbit is close to the anatomical volume of vitreous with a about 2-fold range, (6) and therefore in predictions for human a two-fold range of Vss,ivt (4 ml and 8 ml) was used in simulations. Concentrations of cefepime in the human vitreous after intravenous injections were simulated at two dosing levels (1 and 2 g). The results are shown in Table 1 and Supplement Fig 1. The simulated and experimental cefepime concentrations are in the same range, but the simulated profiles decline slower (tmax between 4 – 12 h) than the experimental ones (tmax at 2 h). The AUC0-last values
10
differed between simulations and experiments modestly (less than 1.5 and 2.8 -fold difference, when Vss,ivt was assumed to be8 ml and 4 ml, respectively).
Table 1. Experimental and simulated cefepime concentrations and AUC0-12h valuesin human serum and vitreous after intravenous administration of cefepime.
a) Cefepime dose 1 g Time
(h)
Experimental Serum Conc (µg/ml)*
Simulated Serum Conc (µg/ml)
Experimental Vitreous Conc (µg/ml)*
Simulated Vitreous Conc (µg/ml)**
Simulated Vitreous Conc (µg/ml)***
0.5 71.76 ± 7.26 71.54 0.76 ± 0.08 1.06 0.53
1 40.83 ± 6.12 41.27 1.7 ± 0.19 1.63 0.82
2 18.24 ± 3.87 17.93 1.91 ± 0.13 2.17 1.10
4 7.13 ± 1.66 7.26 1.22 ± 0.29 2.52 1.31
12 0.71 ± 0.36 0.69 0.89 ± 0.14 2.43 1.39
AUC0-12h
(µg∙h/ml)
164 148 14 28 15
b) Cefepime dose 2 g Time
(h)
Experimental Serum Conc (µg/ml)*
Simulated Serum Conc (µg/ml)
Experimental Vitreous Conc (µg/ml)*
Simulated Vitreous Conc (µg/ml)**
Simulated Vitreous Conc (µg/ml)***
0.5 140.55 ± 13.22 140.47 1.13 ± 0.43 2.26 1.13
1 76.81 ± 6.32 76.89 2.41 ± 0.55 3.34 1.69
2 38.53 ± 4.81 38.35 2.86 ± 0.37 4.38 2.23
4 19.43 ± 5.12 19.54 2.22 ± 0.26 5.28 2.75
12 2.01 ± 0.87 1.99 0.97 ± 0.3 5.38 3.08
AUC0-12h
(µg∙h/ml)
355 326 22 60 32
* From ref. 16
**Vss, ivt = 4 ml
*** Vss, ivt = 8 ml
11 4. DISCUSSION
The simulation model in this study assumes that the distribution clearance is similar in both directions (from blood to eye; from eye to blood). Based on this assumption, we used the CLivt
QSPR model to derive the values for CLBV in the model. CLivt includes clearances both via aqueous humor turnover and permeation through the blood-ocular barriers. However, aqueous humor outflow pathway is not expected to have major impact on CLivt. Typical values for outflow mediated clearance can be estimated to be roughly 0.01-0.03 ml/h (6), while the predicted CLivt values were much higher (0.1-0.6 ml/h) (Supplement Table 1). The correlation of predicted AUC and experimental AUClast values in the vitreous (Fig. 4) supports this notion as no systematic overestimation of ocular drug distribution is seen in the model-based estimates.
The computational estimates of ocular distribution clearance values were reliable and they could be used in the simulation model. These values can be calculated conveniently based on the chemical structure for new compounds and incorporated to the model. However, the dataset that was used to build the original QSPR model did not include protein drugs or other large molecules (6). Therefore, this approach is applicable at the moment only for small molecular compounds. Furthermore, the model of del Amo et al. (6) for CLivt in the rabbit vitreous is based on compounds with range of LogD7.4 values from -10.59 – +2.79. For ten compounds used in the simulations of this study LogD7.4 varies between -7.19 and +0.7. Thus, the chemical space of previous QSPR model for CLivt covers the lipophilicity range of compounds used here.
However, our data set does not include very lipophilic compounds and we cannot be sure if the predictability holds in the case of such compounds.
12
Lipophilicity alone (LogD7.4) does not predict the compound distribution from blood into the eye as the correlation is poor (Supplement Fig. 2). Likewise, fu fails to predict ocular distribution (defined as AUCvitreous/AUCplasma) (Supplement Fig. 3). However, at steady-state situation the correlations might be better. Somewhat better correlation was seen between AUC0-last of unbound drug in plasma and AUC0-last in the vitreous (Supplement Fig. 4), but not as good as the correlation between experimental and simulated AUC0-last values in vitreous (Fig.
4). The simulations not taking into account fu provided a less accurate explanation of the data.
Ocular drug distribution is a multi-factorial process where unbound drug concentration and distribution clearance are the key factors. The predictions errors are related to the errors in the computation of distribution clearance and uncertainty in fraction of unbound drug in plasma.
Sometimes fu values show concentration dependence, for example in the case of fluoroquinolones such as ciprofloxacin, ofloxacin and sparfloxacin (17).
At steady-state rough estimates of drug concentrations in the vitreous may be obtained based on the free drug concentration in plasma. For example, ocular penetration tamoxifen has been measured in ocular surgery patients after oral administration tamoxifen at steady state (18).
Tamoxifen has elimination half-life of 5-7 days (DrugBank database) and at the steady state only low concentration fluctuation is seen in systemic circulation after tablet administration.
Serum levels were measured from 5 patients taking 20 mg tamoxifen/day and concentration varied between 82.8 – 284.7 ng/ml. The range of vitreous concentration of these patients was 0.5 – 7.8 ng/ml and for 2 patients < 0.3 ng/ml). Tamoxifen binds about 99% in plasma protein meaning that Cu would be about 0.8-2.8 ng/ml, which is in the range of measured vitreous concentrations of the drug even though the tamoxifen concentrations in plasma and vitreous did not correlate among the individuals. In addition, cefepime concentrations in the human vitreous after systemic delivery could be predicted well using the QSPR model and up-scaling
13
to the bigger size of human eye. The data suggests that the approach based on unbound drug concentrations in the plasma, QSPR based CLivt and volumes of distributions of 1-2 times anatomical volume of vitreous is suitable for prediction of ocular drug distribution from systemic circulation.
Clearance values are independent of the direction in the case of passive diffusion, but not in the case of transporter mediated permeation. Based on the previous report, CLivt could be predicted with LogD7.4 and hydrogen bonding capacity (Eq. 1) without significant outliers (6).
This indicates that the transporters do not have major influence on drug clearance from the vitreous across blood-ocular barriers. Likewise, we did not see substantial deviations when simulated ocular distribution was correlated with experimental data (Fig. 4).
Several compounds in the dataset are known to be substrates of transporters and there may be some directional transport in the blood-ocular barriers. For example, influx transport on the blood side or efflux transport in the ocular side of the barrier should lead to higher ocular concentrations in the eye than expected on the basis of QSPR derived distribution clearance.
Influx on the ocular side and efflux on the blood side would lead to opposite outcome. Results of some compounds may be explained with transporter effects. For example, ocular elimination of ciprofloxacin, fleroxacin, sparfloxacin, and oxafloxacin is inhibited with probenecid suggesting that efflux proteins may be involved (19). This is in line with the simulated AUC values greater than the experimental values (ciprofloxacin, fleroxacin, oxafloxacin; Table 2), but not in line with the data of sparfloxacin (simulated AUC lower than experimental).
Mercapropurine shows also higher simulated than experimental AUC in the vitreous and this might be explained by the expression of influx transporter OAT3 (ocular side) and efflux transporter MRP4 (blood side) (20, 21). Simulated AUC of topotecan in the vitreous was higher
14
than the experimental values; this may be due to the BCRP and/or P-gp efflux (22).
Methotrexate is substrate for several transporters with different functions (e.g. efflux transporters, folate transporter and organic anion transporters (23, 24), and therefore, it is difficult to interpret the result (simulated AUC < experimental AUC) but folate transporter may contribute to its transfer from blood to retina (20).
Table 2. Experimental and simulated AUC0-last values of the drugs. Also, the ratios of
simulated / experimental values are presented. Drug distribution clearance between systemic circulation and eye was defined computationally (CLivt from Eq. 1 and Vss,ivt of 1.48 ml) or based on the experimental (Exp.) values of CLivt and Vss,ivt of the drugs.
Drug
Experimental AUC0-last (µgxh/ml)
Simulated
AUC0-last (µgxh/ml)
Simulated/
Experimental ratio
Ciprofloxacin 8.7 15.2 1.74
Ciprofloxacin Exp. 8.7 17.1 1.96
Cytarabine 90.8 66.7 0.73
Fleroxacin 2.1 7.7 3.76
Fluconazole 68.7 54.1 0.79
Fluconazole Exp. 68.7 53.0 0.77
Foscarnet 59.5 42.7 0.72
Foscarnet Exp. 59.5 20.5 0.34
Mercaptopurine 2.5 4.1 1.65
Methotrexate 84.0 26.8 0.32
Methotrexate Exp. 84.0 29.4 0.35
Ofloxacin 14.5 22.8 1.57
Sparfloxacin 14.3 7.1 0.50
Topotecan 0.11 0.19 1.69
Disease state may have influence on blood-ocular barriers. Such changes were not taken into account in the simulation model. These effects may not be substantial, because the apparent drug clearance from the vitreous of diseased patients correlated well with the clearance values in the healthy rabbits (7).
5. CONCLUSION
Distribution of drugs from blood circulation to the ocular vitreous was simulated successfully.
The simulations showed high correlation with experimental data and typically ocular drug
15
exposure in rabbits and humans could be predicted within two-fold error margin. Ocular drug concentrations can be predicted when concentration levels in blood, fraction unbound in plasma and chemical structure are known. The simulation model is expected augment the estimation of ocular drug exposure at early phases of drug discovery.
ACKNOWLEDGEMENTS
This study was supported by the Academy of Finland (Kati-Sisko Vellonen, Arto Urtti). The authors thank Dr. Veli-Pekka Ranta for valuable comments and practical help.
Supporting information
Supplement Figure 1. Cefepime concentrations in human vitreous after intravenous administration of 1 g dose and 2 g dose.
Supplement Figure 2: AUC0-last in vitreous/AUC0-last in systemic circulation vs LogD7.4
Supplement Figure 3: AUC0-last in vitreous/AUC0-last in systemic circulation vs fraction unbound (fu)
Supplement Figure 4: AUC0-last in vitreous vs unbound AUC0-last in systemiccirculation Supplement Table 1: Parameter values used for pharmacokinetic simulations
This material is available free of charge via the Internet at http://pubs.acs.org.
REFERENCES
1. Sebag, J.; Tang, M.; Brown, S.; Sadun, A.A.; Charles, M.A. Effects of pentoxifylline on choroidal blood flow in nonproliferative diabetic retinopathy. Angiology 1994, 45, 429-433.
2. Bill, A. Capillary Permeability to and Extravascular Dynamics of Myoglobin, Albumin and Gammaglobulin in the Uvea. Acta Physiol. Scand. 1968, 73, 204-219.
3. Maurice, D.M. and Mishima, S. Ocular pharmacokinetics, In Pharmacology of the Eye, 1st ed.; Sears, M.L., Ed.; Springer-Verlag: 1984; pp. 19-116.
4. Cunha-Vaz, J. The blood-ocular barriers. Surv. Ophthalmol. 1979, 23, 279-296.
16
5. Bill, A. Blood circulation and fluid dynamics in the eye. Physiol. Rev. 1975, 55, 383-417.
6. Del Amo, E.M.; Vellonen, K.S.; Kidron, H.; Urtti, A. Intravitreal clearance and volume of distribution of compounds in rabbits: In silico prediction and pharmacokinetic simulations for drug development. Eur. J. Pharm. Biopharm. 2015, 95(Pt B), 215-226.
7. Del Amo, E.M. and Urtti, A. Rabbit as an animal model for intravitreal pharmacokinetics:
Clinical predictability and quality of the published data. Exp. Eye Res. 2015, 137, 111-124.
8. Yanoff, M. and Duker, J.S. Ophthalmology. Elsevier, 2014.
9. Schwartz, K. and Budenz, D. Current management of glaucoma. Curr. Opin. Ophthalmol.
2004, 15, 119-126.
10. Bartlett, J.D. and Cullen, A.P. Clinical administration of ocular drugs, In Clinical ocular pharmacology, 2nd ed. ed.; Bartlett J.D., J.S.D., Ed.; Butterworth-Heinemann: USA, 1989;
pp. 29-66.
11. Blomquist, P.H. Ocular complications of systemic medications. Am. J. Med. Sci. 2011, 342, 62-69.
12. Miguel, A.; Henriques, F.; Azevedo, L.F.; Pereira, A.C. Ophthalmic adverse drug reactions to systemic drugs: a systematic review. Pharmacoepidemiol. Drug Saf. 2014, 23, 221-233.
13. Potts, A.M. The Reaction of Uveal Pigment in Vitro with Polycyclic Compounds. Invest.
Ophthalmol. 1964, 3, 405-416.
14. Potts, A.M. Drug toxicity as related to ocular melanin. Psychopharmacol. Bull. 1974, 10, 40-41.
15. Ruby, A.J.; Williams, G.A.; Blumenkranz, M.S. Vitreous humor, In Duane's
Ophthalmology, Tasman, W. and Jaeger, E.A., Eds.; Lippincott Williams & Wilkins, 2006.
16. Aras, C.; Ozdamar, A.; Ozturk, R.; Karacorlu, M.; Ozkan, S. Intravitreal penetration of cefepime after systemic administration to humans. Ophthalmologica 2002, 216, 261-264.
17. Shandil, R.K.; Jayaram, R.; Kaur, P.; Gaonkar, S.; Suresh, B.L.; Mahesh, B.N.; Jayashree, R.; Nandi, V.; Bharath, S.; Balasubramanian, V. Moxifloxacin, ofloxacin, sparfloxacin, and ciprofloxacin against Mycobacterium tuberculosis: evaluation of in vitro and
pharmacodynamic indices that best predict in vivo efficacy. Antimicrob. Agents Chemother.
2007, 51, 576-582.
18. Flaxel, C.J.; Mulholland, B.; Haynes, B.; Gregor, Z.J. Intraocular penetration of tamoxifen. Ophthalmology 2000, 107, 2006-2009.
19. Liu, W.; Liu, Q.F.; Perkins, R.; Drusano, G.; Louie, A.; Madu, A.; Mian, U.; Mayers, M.;
Miller, M.H. Pharmacokinetics of sparfloxacin in the serum and vitreous humor of rabbits:
physicochemical properties that regulate penetration of quinolone antimicrobials. Antimicrob.
Agents Chemother. 1998, 42, 1417-1423.
17
20. Hosoya, K.; Tomi, M.; Tachikawa, M. Strategies for therapy of retinal diseases using systemic drug delivery: relevance of transporters at the blood-retinal barrier. Expert Opin.
Drug Deliv. 2011, 8, 1571-1587.
21. Tagami, M.; Kusuhara, S.; Honda, S.; Tsukahara, Y.; Negi, A. Expression of ATP- binding cassette transporters at the inner blood-retinal barrier in a neonatal mouse model of oxygen-induced retinopathy. Brain Res. 2009, 1283, 186-193.
22. Asashima, T.; Hori, S.; Ohtsuki, S.; Tachikawa, M.; Watanabe, M.; Mukai, C.; Kitagaki, S.; Miyakoshi, N.; Terasaki, T. ATP-binding cassette transporter G2 mediates the efflux of phototoxins on the luminal membrane of retinal capillary endothelial cells. Pharm. Res. 2006, 23, 1235-1242.
23. Tomi, M. and Hosoya, K. The role of blood-ocular barrier transporters in retinal drug disposition: an overview. Expert Opin. Drug Metab. Toxicol. 2010, 6, 1111-1124.
24. El-Sheikh, A.A.; Greupink, R.; Wortelboer, H.M.; van den Heuvel, J.J.; Schreurs, M.;
Koenderink, J.B.; Masereeuw, R.; Russel, F.G. Interaction of immunosuppressive drugs with human organic anion transporter (OAT) 1 and OAT3, and multidrug resistance-associated protein (MRP) 2 and MRP4. Transl. Res. 2013, 162, 398-409.
18
Erratum in:
Prediction of ocular drug distribution from systemic blood circulation
Kati-Sisko Vellonena, Esa-Matti Soinib, Eva M. del Amo a, Arto Urtti a,b*
a School of Pharmacy, University of Eastern Finland, Kuopio, Finland
b Centre for Drug Research, Division of Pharmaceutical Biosciences, Faculty of Pharmacy, University of Helsinki, Finland
*Corresponding author. School of Pharmacy, University of Eastern Finland, P.O. Box 1627, 70211 Kuopio, Finland. Tel. +358 40540 2279 , Fax: +358 17 162424, Email: arto.urtti@uef.fi
In the present article the equation 2 in page 2907 was wrongly written and used:
Human apparent CLBV = Human apparent CLivt = (Rabbit CLivt - 0.04)/1.41 (2)
The correct equation is:
Human apparent CLBV = Human apparent CLivt = 1.41 x Rabbit CLivt + 0.04 (2) The equation was used to calculate distribution clearance (CLBV) of the only drug investigated in humans presented in the article, cefepime (see the below chart):
The new QSPR CLBV value of cefepime in human (0.3168 ml/h) was applied to the
simulation model to predict cefepime concentrations in human vitreous. The corresponding modifications in the supporting info and the article text are presented below:
• In the supporting information, in the Supplemental Table 1 the value for the QSPR CLBV in human is not 0.111 ml/h but 0.3168 ml/h.
• In the supporting information, the Supplemental Figure 1. Cefepime concentrations in human vitreous after intravenous administration of 1 and 2 g doses is as follows:
19
• In page 2909, the table 1 is corrected with the new values in bold red:
Table 1. Experimental and simulated cefepime concentrations and AUC0-12h valuesin human serum and vitreous after intravenous administration of cefepime.
Cefepime dose of 1 g Time
(h)
Experimental Serum Conc (µg/ml)*
Simulated Serum Conc (µg/ml)
Experimental Vitreous Conc (µg/ml)a
Simulated Vitreous Conc (µg/ml)b
Simulated Vitreous Conc (µg/ml)c
0.5 71.76 ± 7.26 71.54 0.76 ± 0.08 3.00 1.51
1 40.83 ± 6.12 41.27 1.7 ± 0.19 4.54 2.32
2 18.24 ± 3.87 17.93 1.91 ± 0.13 5.80 3.06
4 7.13 ± 1.66 7.26 1.22 ± 0.29 6.22 3.49
12 0.71 ± 0.36 0.69 0.89 ± 0.14 5.39 3.45
AUC0-12h
(µg∙h∙ml-1)
164 148 14 63 38
20 c) Cefepime dose 2 g
Time (h)
Experimental Serum Conc (µg/ml)*
Simulated Serum Conc (µg/ml)
Experimental Vitreous Conc (µg/ml)a
Simulated Vitreous Conc (µg/ml)b
Simulated Vitreous Conc (µg/ml)c
0.5 140.55 ± 13.22 140.47 1.13 ± 0.43 6.36 3.22
1 76.81 ± 6.32 76.89 2.41 ± 0.55 9.25 4.74
2 38.53 ± 4.81 38.35 2.86 ± 0.37 11.68 6.16
4 19.43 ± 5.12 19.54 2.22 ± 0.26 13.05 7.29
12 2.01 ± 0.87 1.99 0.97 ± 0.3 11.89 7.53
AUC0-12h
(µg∙h∙ml-1)
355 326 22 135 81
a From ref. (16)
bVss, ivt = 4 ml
cVss, ivt = 8 ml
• In page 2908, the text “The simulated and experimental cefepime concentrations are in the same range, but the simulated profiles decline slower (tmax between 4 – 12 h) than the experimental ones (tmax at 2 h). The AUC0-last values differed between simulations and experiments modestly (less than 1.5 and 2.8 -fold difference, when Vss,ivt was assumed to be 8 ml and 4 ml, respectively )” is corrected into:
The simulated profiles decline similarly to the experimental ones (tmax between 3-6 h for simulated and around 2 h for experimental profile). The AUC0-last values differed between simulations and experiments from 2.7 to 3.7-fold difference when assuming Vss,ivt of 8 ml and from 4.5 to 6.1 -fold difference when Vss,ivt was assumed to be4 ml.
• In page 2910, the text “The simulations showed high correlation with experimental data and typically ocular drug exposure in rabbits and humans could be predicted within two-fold error margin” is corrected into:
The simulations showed good correlation with experimental data and typically ocular drug exposure in rabbits could be predicted within two-fold error margin, while in the case of the only human study with cefepime the error was higher (between two to six- fold).
In conclusion, the predicted concentration profiles in human vitreous are similar to the experimental concentration profiles with similar tmax though higher AUC0-last values. The
21
predictions are still in acceptable range, within 2 to 6-fold error range, but data for more compounds investigated in human are still needed to draw final conclusions.
