European Journal of Pharmaceutical Sciences 76 (2015) 156–164

Contents lists available at ScienceDirect

European Journal of Pharmaceutical Sciences

journal homepage: www.elsevier .com/ locate /e jps

Tanshinone IIA – loaded pellets developed for angina chronotherapy:Deconvolution-based formulation design and optimization,pharmacokinetic and pharmacodynamic evaluation

http://dx.doi.org/10.1016/j.ejps.2015.05.0120928-0987/� 2015 Elsevier B.V. All rights reserved.

Abbreviations: TA, tanshinone IIA; TA-tSD-IRPs, tanshinone IIA ternary soliddispersion immediate-release pellets; TA-SRPs, tanshinone IIA sustained-releasepellets; AA, angina attacks; IHD, ischemic heart disease; PVAc, polyvinyl acetate;PVA–PEG, polyvinyl alcohol–polyethylene glycol; NZW rabbits, New Zealand whiterabbits; SDS, sodium dodecyl sulfate; PBS, phosphate buffer solution; NO, Nitricoxide; HPLC, high pressure liquid chromatography; Tmax, time to maximumconcentration; Cmax, maximum plasma concentration; MRT, mean residence time;AUC0–t and AUC0–1, area under the plasma concentration-time curve from 0 to t hand 0 to infinity; %PE, percent prediction error.⇑ Corresponding authors at: Department of Pharmaceutics, China Pharmaceutical

University, No. 24 Tongjiaxiang, Nanjing, PR China.E-mail addresses: zwllz@163.com (W.-L. Zhang), liujianpingljp@hotmail.com

(J.-P. Liu).

Hong-Xiang Yan a, Jin Li b, Zheng-Hua Li a, Wen-Li Zhang a,⇑, Jian-Ping Liu a,⇑a Department of Pharmaceutics, China Pharmaceutical University, Nanjing, PR Chinab Department of Pharmacy, Xuzhou Medical College, Xuzhou, PR China

a r t i c l e i n f o a b s t r a c t

Article history:Received 5 December 2014Received in revised form 3 April 2015Accepted 10 May 2015Available online 11 May 2015

Chemical compounds studied in this article:Tanshinone IIA (PubChem CID: 164676)Polyvinylpyrrolidone (PubChem CID: 6917)Poloxamer 188 (PubChem CID: 24751)Talc (PubChem CID: 16211421)1,2-propylene glycol (PubChem CID: 1030)Nitric oxide (PubChem CID: 145068)Methanol (PubChem CID: 887)Sodium dodecyl sulfate (PubChem CID:3423265)Ethyl acetate (PubChem CID: 8857)Nitrogen (PubChem CID: 947)

Keywords:Tanshinone IIAPelletsDeconvolution-based designAngina chronotherapyPharmacokineticsPharmacodynamics

This paper put forward a deconvolution-based method for designing and optimizing tanshinone IIAsustained-release pellets (TA-SRPs) with improved efficacy in the treatment of variant angina. TA-SRPswere prepared by coating TA ternary solid dispersion immediate-release pellets (TA-tSD-IRPs) with theblends of polyvinyl acetate (PVAc) and polyvinyl alcohol–polyethylene glycol (PVA–PEG) using fluidizedbed technology. The plasma concentration–time curve of TA-tSD-IRPs following oral administration as aweight function was investigated in New Zealand white rabbits. The predicted/expected plasmaconcentration–time curve of TA-SRPs as a response function was simulated according to the circadianrhythm of variant angina during 24 h based on chronotherapy theory. The desired drug release profileof TA-SRPs was obtained via the point-area deconvolution procedure using the weight function andresponse function, and used for formulation optimization of TA-SRPs. The coating formulation ofTA-SRPs was optimized as 70:30 (w/w) PVAc/PVA–PEG with 5% (w/w) coating weight due to in vitro drugrelease profile of these TA-SRPs was similar to the desired release profile (similarity factor f2 = 64.90).Pharmacokinetic studies of these optimized TA-SRPs validated that their actual plasma concentration–time curve possessed a basically consistent trend with the predicted plasma concentration–time curveand the absolute percent errors (%PE) of concentrations in 8–12 h were less than 10%.Pharmacodynamic studies further demonstrated that these TA-SRPs had stable and improved efficacywith almost simultaneous drug concentration–efficacy. In conclusion, deconvolution could be employedin the development of TA-SRPs for angina chronotherapy with simultaneous drug efficacy and reduceddesign blindness and complexity.

� 2015 Elsevier B.V. All rights reserved.

1. Introduction

Tanshinone IIA (TA, Fig. 1), one of the major liposoluble bioac-tive constituents isolated from the roots of Chinese herb Salviaemiltiorrhiza Bunge (Danshen), exhibits a variety of cardiovascularactivities, including prevention and treatment of angina pectoris(Shang et al., 2012; Gao et al., 2012). TA has poorwater-solubility (2.8 ng mL�1) (Li et al., 2008), short half-life (1–2 h) (Zhang et al., 2013), substantial intestinal first pass metabo-lism (Hao et al., 2007) and low oral bioavailability (Li et al.,2005). Recently, a number of new drug delivery systems have beenemployed to resolve these issues, e.g. solid dispersion (Hao et al.,

Fig. 1. Chemical structure of tanshinone IIA.

H.-X. Yan et al. / European Journal of Pharmaceutical Sciences 76 (2015) 156–164 157

2006), solid lipid nanoparticles (Zhang et al., 2008), solid inclusioncomplex (Fan et al., 2005), intravenous lipid emulsion (Chu et al.,2012) and so on. In our previous study, TA ternary solid dispersionimmediate-release pellets (TA-tSD-IRPs) were developed toachieve complete dissolution, accelerated absorption rate andsuperior oral bioavailability (Li et al., 2012). However, clinicalresearch has demonstrated that the onset of variant angina(Prinzmetal et al., 1959) shows a significant circadian rhythm,which predominantly occurs between 02:00 and 06:00 (Moriet al., 2002) and peaks around 04:00 (Lemmer, 2006). This physio-logical phenomenon is seldom considered in most conventional orpreviously developed pharmaceutical preparations of TA.Obviously, it is extremely inconvenient for the patients to takethese drug products in the sleeping hours in order to prevent orrelieve angina attacks. Therefore, the development of novel TAcontrolled-release formulation according to angina circadianrhythm is necessary to achieve better patient compliance andimproved therapeutic efficacy.

Chronotherapy is to optimize the efficiency and safety of med-ications by proportioning drug concentration during 24 h in syn-chrony with biological rhythm determinants of diseases,especially applying to the improved treatment of ischemic heartdisease (IHD) (Smolensky and Portaluppi, 1999; Lemmer, 1996;Portaluppi and Lemmer, 2007). As one of the typical symptomsof IHD, angina can be better treated based on chronotherapy the-ory. Thus, the aim of the present study was to prepare chronother-apeutic tanshinone IIA sustained-release pellets (TA-SRPs) tosynchronize plasma TA concentration with the circadian rhythmof variant angina during 24 h. TA-SRPs could be taken at an appro-priate and favorable time in a day, such as 6 pm. In this case, highplasma concentrations could be reached in 8–12 h (between 02:00and 06:00) with time to peak concentration around 10 h (at 04:00)so as to prevent the sudden angina attacks in the sleeping hoursearly in the morning.

Deconvolution (Langenbucher and Mysicka, 1985; Qi et al.,2003; Gaynor et al., 2008; Vaughan and Dennis, 1978; Yeh et al.,2001), recommended by FDA (Administration et al., 1997), hasbeen widely applied in the pharmaceutical research involved within vivo/in vitro correlation of dosage forms (Sirisuth and Eddington,2002; Sunesen et al., 2005). In the current study, we made anattempt to design and optimize chronotherapeutic TA-SRPs witha deconvolution-based method. Herein, numerical deconvolutionwas utilized from a new perspective to reduce the blindness andcomplexity throughout the chronotherapeutic formulation devel-opment process. Weight function was the plasma concentration–time curve of TA-tSD-IRPs following oral administration in normalNew Zealand white (NZW) rabbits. Response function was the pre-dicted/expected plasma concentration–time curve of TA-SRPs,which was simulated according to the incidence of variant anginaduring 24 h based on angina chronotherapy theory. The desireddrug release profile of TA-SRPs was determined by deconvolutionusing the weight function and the response function, and subse-quently used for guiding the formulation optimization ofTA-SRPs. For the formulation optimization, two coating polymers,polyvinyl acetate (PVAc) dispersion (Kollicoat� SR30D) and

polyvinyl alcohol–polyethylene glycol (PVA–PEG) graft copolymer(Kollicoat� IR), with different permeability were used in combina-tion to adjust drug release from TA-SRPs. The pharmacokinetic andpharmacodynamic studies in rabbits were performed to verifywhether TA-SRPs optimized based on deconvolution had a suitableplasma drug concentration time course with better drug efficacy.

2. Materials and methods

2.1. Materials

TA (98.63%) was purchased from Xi’an Honson BiotechnologyCo., Ltd. (Shanxi, China). TA standard was purchased from theNational Institute for the Control of Pharmaceutical andBiological Products (Beijing, China). Sugar spheres (0.75–0.85 mm) were supplied by JRS Pharma (Rosenberg, Germany).PVP-K29/32 was from China Division, ISP Chemical Company(Shanghai, China). Opadry� II was kindly donated from ColorconCoating Technology Co., Ltd. (Shanghai, China). Poloxamer 188(Pluronic� F68), polyvinyl acetate dispersion (Kollicoat� SR30D)and polyvinyl alcohol–polyethylene glycol graft copolymer(Kollicoat� IR) were obtained from BASF Chemical Company(Ludwigshafen, Germany). Talc (1200 mesh) was received fromMerck-Schuchardt (Hohenbrunn, Germany). 1,2-propylene glycolwas delivered from Shanghai Chemical Agent Co., Ltd. (Shanghai,China). Gelatin capsules were from Suzhou Capsugel Ltd.(Suzhou, China). High-fat fodder was provided by Jiangsu XietongBiological Engineering Co., Ltd. (Nanjing, China). Nitric oxide (NO)reagent box was purchased from Beyotime Institute ofBiotechnology (Shanghai, China). Medical 5% glucose solutionwas from Jiangsu Dahongying-Hengshun Pharmaceutical Co., Ltd.(Nanjing, China). All the reagents were of analytical grade exceptmethanol, which was of chromatographic grade.

2.2. Animals

Healthy male NZW rabbits (body weight 2.0 ± 0.9 kg) were pur-chased from Experimental Animal Center of China PharmaceuticalUniversity (Nanjing, China). The rabbits were housed in a temper-ature and humidity controlled room (23 �C, 55% air humidity) withfree access to water and standard rabbit chow for at least 5 days toadapt to the new environment prior to the experiments. All theexperiments were approved by the Institutional Animal Care andUse Committee of China Pharmaceutical University.

2.3. Preparation of pellets

2.3.1. Preparation of TA-tSD-IRPsTA-tSD-IRPs were prepared by a single-step fluid-bed coating

technique. A hydrophilic polymer PVP and a surfactant poloxamer188 were used as dispersing carriers of drug substance TA. Firstly,TA, PVP and poloxamer 188 at a definite ratio (1:4:1, w/w) weredissolved in a mixed solvent of ethyl acetate–anhydrous ethanol(5:1, v/v) to form a clear solution. Then the solution under contin-uous stirring was sprayed onto the sugar cores (5 g) from a bottomnozzle (0.5 mm in diameter) in a fluid-bed granulator and coater(JHQ-100; Shenyang, China). This equipment was attached to aperistaltic pump (HL-2; Shanghai, China). The preparation ofTA-tSD-IRPs was protected from light exposure and carried outunder the condition of coating temperature of 35–37 �C, spray rateof 1.0–1.2 mL min�1, atomization pressure of 1.5–1.6 bar and airblow rate of 100–150 mL min�1. After drug/carriers layering, thepellets were further fluidized for 15 min at 30–35 �C to minimizethe solvent residue. The resulting pellets were placed in a containerfor subsequent studies.

158 H.-X. Yan et al. / European Journal of Pharmaceutical Sciences 76 (2015) 156–164

2.3.2. Preparation of TA-SRPsThe fluid-bed granulator and coater (nozzle diameter 0.8 mm)

was loaded with 5 g of TA-tSD-IRPs for each run. Opadry� II wassprayed onto these pellets to produce the isolated layer ofTA-SRPs. The lower permeable PVAc and higher permeable PVA–PEG were used as main coating materials. 1,2-propylene glycol(plasticizer, 2.5% w/w, based on total polymer mass) was addedinto PVAc suspension and stirred overnight. The talc (antiadherent,25% w/w, based on total polymer mass) was then dispersed in thepolymer suspension. Subsequently, PVA–PEG was added to form asuspension containing 15% w/w total solid content. The coatingsuspension was sieved through a 80-mesh screen and then sprayedonto the previous pellets under continuous stirring. The coatingparameters were as follows: temperature 38–40 �C, spray rate0.6–0.8 mL min�1, atomization pressure 0.8–1.2 bar and air blowrate 120–180 mL min�1. Subsequent to the coating, the pelletswere further fluidized for 10 min in the coating chamber and curedfor 24 h at 60 �C in an oven.

A series of TA sustained-release coated pellets were preparedwith different PVAc/PVA–PEG ratios (90:10, 85:15, 70:30 w/w)and coating weights (3%, 5%, 10% w/w, representing coating thick-ness). The coating weight was calculated by equationF = (Wa �Wb)/Wb � 100%, where Wb and Wa are the accurateweights of pellets before and after coating, respectively.

2.4. In vitro release studies

2.4.1. Quantitative analysis of TA20 lL of release medium containing TA was injected into high

pressure liquid chromatography (HPLC, Shimadzu LC-20A, Kyoto,Japan). The system was made up of a Shimadzu SIL-20AC autosam-pler, a Shimadzu LC-20AB pump and a Shimadzu SPD-M20A diodearray detection set at 268 nm. The separation was completed at30 �C on a Synergi Hydro-RP C18 column (5 lm,250 mm � 4.6 mm, Phenomenex, USA) protected by a C18Securityguard column (5 lm, 10 mm � 4.6 mm, Kromasil,Sweden). The mobile phase was methanol/water (90:10, v/v) at aflow rate of 1.0 mL min�1. The standard curve was found to be lin-ear in the 0.05–5.00 lg mL�1 range: A = 205,386 � C + 1779,r = 0.9997 (where C is the concentration of TA and A is the corre-sponding peak area, n = 3). The recovery rates of low, middle andhigh concentration for TA were all in the range of 98–102% andthe RSD were less than 2%. The RSD of intra-day and inter-day pre-cision for TA were below 2%.

2.4.2. Drug release testsTA-SRPs equivalent to 2.5 mg TA were sealed in hard gelatin

capsules by a manual capsule-filling machine (CapsulCN,Zhejiang, China). Then the release experiments were carried outin USP34 Apparatus I (rotating basket method; ZRS-8G, Tianjin,China) at a rotation rate of 100 ± 1 rpm. The release medium was900 mL of distilled water maintained at 37 ± 0.5 �C and 0.5%(w/w) sodium dodecyl sulfate (SDS) was added for perfect sinkconditions of TA. At pre-determined time points of 0.5, 1, 1.5, 2,3, 4, 6, 8, 10, 12, 24 h, 5 mL of samples were withdrawn andreplaced with an equivalent-volume of fresh medium. The sampleswere filtered through 0.22 lm filter and then quantified for TA byHPLC as described above. All the release tests were implemented intriplicate (n = 3) and the cumulative release percents and standarddeviations were calculated.

2.4.3. Investigation of drug release stabilitiesThe drug release stability studies of TA-SRPs were conducted in

a ZRS-8G release tester (Tianjin, China) in triplicate (n = 3) to inves-tigate the influence of different pH condition, release method androtation rate on drug release of TA-SRPs. The release mediums with

different pH (900 mL) used were distilled water containing 0.5%SDS, 0.1 M HCl and pH 6.8 phosphate buffer solution (PBS), sepa-rately and were all maintained at 37 ± 0.5 �C. Two release methodswere rotating basket method and paddle method. The rotationrates were investigated at 50, 100, 150 rpm, respectively. All thesampling method and analysis were same as the descriptions inSection 2.4.2.

2.4.4. Analysis of release dataThe similarity factor (f2) is a measurement of the similarity of

the release profiles (similarity factor f2 P 50, difference 6 10%)(Jantratid et al., 2009). Equation of f2 is shown as follows:

f 2 ¼ 50 log 1þ 1n

Xn

t¼1

Rt � Ttð Þ2" #�0:5

� 100

8<:

9=; ð1Þ

where n is the number of time points, Rt is the release value of thereference at time t, and Tt is the release value of the test at time t.

2.5. Pharmacokinetic studies

2.5.1. Animal experimentsThe male NZW rabbits (body weight 2.0 ± 0.9 kg) were fasted

for 12 h but supplied with water ad libitum before the experi-ments. Capsules of TA-tSD-IRPs or TA-SRPs were orally adminis-tered (30 mg kg�1 of TA) to six rabbits (n = 6). At pre-determinedtime points of 0 (pre-treatment), 0.5, 1, 2, 3, 4, 5, 6, 8, 10, 12 16,24 and 48 h post-dosing, 1.5 mL of blood samples were collectedfrom marginal ear veins of the rabbits and put into heparinizedtubes to avoid clotting. Then the plasma samples were separatedsuccessfully by centrifugation at 3000 rpm for 10 min and storedat �20 �C until analysis.

2.5.2. Plasma sample processing and TA determinationProcessing of the thawed plasma samples were performed

under subdued light at room temperature. A single-step proteinprecipitation procedure was adopted to extract TA from the rabbitplasma. To begin with, 200 lL of plasma sample and 400 lL ofethyl acetate were pipetted into a 10 mL centrifuge tube andvortex-mixed for 3 min to precipitate protein fully. Afterwards,the mixture was centrifuged at 3000 rpm for 10 min. The super-natant was transferred into a clean centrifuge tube and evaporatedto dryness under a N2 stream at 40 �C in a water bath. Then theresidue was redissolved in 200 lL of methanol and centrifuged at12,000 rpm for 10 min to isolate the supernatant. 20 lL of thesupernatant was injected into HPLC for quantitative analysis ofTA. The chromatographic condition was same as described inSection 2.4.1. The linearity of the method was achieved in the0.005–0.5 lg mL�1 concentration range: A = 175264C + 324.5,r = 0.9976 (where C is the concentration of TA and A is the corre-sponding peak area, n = 3). The RSD of method recoveries, extrac-tion recoveries, intra-day and inter-day variabilities were all lessthan 10%, which indicated the rationality of this bioanalyticalmethod for TA determination.

2.5.3. Pharmacokinetic analysisPharmacokinetic parameters such as time to maximum concen-

tration (Tmax), maximum plasma concentration (Cmax), mean resi-dence time (MRT), area under the plasma concentration–timecurve from 0 to t h and 0 to infinity (AUC0–t and AUC0–1) were cal-culated by non-compartmental analysis using WinNonlin program(version 1.5, Scientific Consulting, Inc., Cary, NC, USA). Data wereexpressed as mean values ± standard deviations (Mean ± SD).

H.-X. Yan et al. / European Journal of Pharmaceutical Sciences 76 (2015) 156–164 159

2.6. Pharmacodynamic studies

2.6.1. NZW rabbit model of anginaEighteen male NZW rabbits (body weight 2.0 ± 0.9 kg) were

administered daily with dl-homocysteine thiolactone(20 mg mL�1 in 5% glucose solution, 20–25 mg kg�1) by subcuta-neous injection, and fed by high-fat fodder (79.5% ordinary feed,10% cholesterol, 5% lard, 5% egg yolk powder, 0.5% sodium cholate,w/w) (Kan et al., 2014). Eight weeks later, the NZW rabbits withangina induced by atherosclerosis were obtained and could beused for the evaluation of the drug efficacy of pellets.

2.6.2. Animal experiments and NO analysisEighteen NZW rabbits with angina were fasted for 12 h but sup-

plied with water ad libitum before the experiments, and randomlydivided into three groups (n = 6): controlled group, TA-tSD-IRPsgroup and TA-SRPs group. The controlled group was orally admin-istered with the controlled capsules containing blank pellets with-out TA. The TA-tSD-IRPs group and TA-SRPs group were orallyadministered with respective drug pellets capsules (30 mg kg�1

of TA). At pre-determined time points of 1, 2, 3, 4, 6, 8, 10, 12, 24and 48 h after administration, 2 mL of blood samples were col-lected from marginal ear veins of these rabbits. After standing fora period of time in the test tubes, the upper serum was obtainedfor NO analysis. NO concentration as the pharmacodynamic indexwas detected by nitrate reduction method described on NO reagentbox. The average concentrations and standard deviations werecalculated.

Fig. 2. The plasma concentration–time curve of TA-tSD-IRPs in rabbits after oraladministration in a dose of 30 mg kg�1 (n = 6) and the mean predicted plasmaconcentration–time curve of TA-SRPs.

3. Results and discussion

3.1. Deconvolution method

Deconvolution method can be used to directly obtain in vivodynamic information of drug from the pharmacokinetic data.Input rate function, weight function and response functionin vivo are involved in this method. It is based on the convolutionintegral, which is defined as

RðtÞ ¼Z t

0IðhÞ �Wðt � hÞdh ð2Þ

For a sustained release preparation, R(t) is the plasma drug con-centration at time t following administration of a sustained releasedosage unit. W(t) is the unit impulse response, which is the plasmadrug concentration function following administration of the corre-sponding oral solution or immediate-release preparation. I(t) rep-resents the in vivo release rate of the sustained releasepreparation at time t. Deconvolution involves estimating thein vivo drug release rate I(t) using the observed in vivo R(t) dataand observed in vivo reference data W(t).

Providing the observed time points are t1, t2, t3 . . . ti, the corre-sponding response values are R1, R2, R3 . . . Ri and the input valuesare I1, I2, I3 . . . Ii, where Ii represents average in vivo drug input ratewithin each time interval ti�1 � ti, respectively. A working expres-sion (3) can be derived from formula (2).

Ri ¼Xi

k¼1

Ik � AUCti�tk�1ti�tk

¼ I1 � AUCti�t0ti�t1þ I2 � AUCti�t1

ti�t2þ � � � þ Ii � AUCti�ti�1

ti�tið3Þ

where AUC is area under the plasma drug concentration–time curveof oral solution or immediate-release preparation (weight function)within time interval ti � tk � ti � tk�1, which can be calculated bythe trapezoid algorithm.

Then Ii can be calculated in turn by formula (4), which is thetransformation of formula (3).

Ii ¼Ri � I1 � AUCti�t0

ti�t1� I2 � AUCti�t1

ti�t2� � � � � Ii�1 � AUCti�ti�2

ti�ti�1

AUCti�ti�1ti�ti

ð4Þ

An integration of Ii yields the cumulative drug release amountCAi at each time point ti.

CAi ¼Xi

k¼1

Ik � ðtk � tk � 1Þ ð5Þ

3.2. Design of TA-SRPs based on deconvolution

3.2.1. Evaluation of weight functionWeight function was the plasma concentration–time curve of

TA-tSD-IRPs in NZW rabbits after oral administration (shown inFig. 2). The pharmacokinetic data were processed following anon-compartmental model method and the important pharma-cokinetic parameters (Tw

max, Cwmax, MRTw, AUCw

0–t and AUCw0–1) were

presented in Table 1. Compared to the compartment model, thenon-compartmental model analysis is easier to automate, andhas least intervention decisions made by the user. It is especiallysuitable to be used in predicting the systemic drug concentrationsby convolution or estimating the time-course of absorption bydeconvolution (Gillespie, 1991). Hence, all the pharmacokineticdata obtained from animal experiments in this study were ana-lyzed following a non-compartmental model.

3.2.2. Simulation of response functionResponse function was the predicted/expected plasma concen-

tration–time curve of TA-SRPs, which was simulated and fittedaccording to the incidence of angina pectoris. Fig. 3 illustratedthe statistical relative percentage of variant angina attacks (AA)during 24 h (starting from 18:00). Given the expected plasma con-centration at the time point of the maximum AA frequency (04:00,C10h) equaled to Cw

max of weight function, a series of plasma concen-tration values at the corresponding time points (C0.5h, C1h, C2h, C3h,C4h, C5h, C6h, C8h, C10h, C12h, C16h, C24h) were calculated accordingly,which were proportional to the percentage of AA. Assuming C48h ofthis calculated curve equaled to Cw

48h of weight function, areaunder this plasma concentration–time curve (AUC0–t) was analyzedusing WinNonlin program and subsequently compared withAUCw

0–t of weight function. All the concentration values wereadjusted proportionally together, and the predicted/expected

Table 1The pharmacokinetic parameters of TA after oral administration of TA-tSD-IRPs and TA-SRPs in rabbits in a dose of 30 mg kg�1 and the pharmacokinetic parameters of thepredicted response function.

Pharmacokinetic parameters (TA-tSD-IRPs) (n = 6) Pharmacokinetic parameters (TA-SRPs)

Predicted (mean) Observed (n = 6)

Twmax (h) 4.00 ± 0.025 Tmax (h) 10.00 10.00 ± 0.315

Cwmax (ng mL�1) 81.51 ± 17.170 Cmax (ng mL�1) 46.00 42.51 ± 10.330

MRTw (h) 9.88 ± 0.462 MRT (h) 16.37 16.94 ± 0.679AUCw

0–t (ng h mL�1) 949.16 ± 135.391 AUC(0–t) (ng h mL�1) 951.27 948.47 ± 115.234AUCw

0–1(ng h mL�1) 959.73 ± 143.245 AUC(0–1) (ng h mL�1) 961.98 986.66 ± 127.982

Fig. 3. The percentage of variant angina attacks during 24 h.

160 H.-X. Yan et al. / European Journal of Pharmaceutical Sciences 76 (2015) 156–164

plasma concentration–time curve was simulated successfully tillAUC0–t within 80–120% of AUCw

0–t for its bioequivalence.The final mean predicted plasma concentration–time curve of

TA-SRPs as a response function was depicted in Fig. 2. Followinga non-compartmental model analysis, the pharmacokinetic param-eters (Tmax, Cmax, MRT, AUC0–t and AUC0–1) (mean values) were pre-sented in Table 1. Among them, Cmax was about 0.56 times of Cw

max

value. Meanwhile, AUC0–t value was about 100.22% of AUCw0–t

value, which indicated the bioequivalence of the design forTA-SRPs. Obviously, the plasma drug concentration time courseof these newly designed chronotherapeutic TA-SRPs were synchro-nized with the circadian rhythm of variant angina during 24 h(Fig. 4).

3.2.3. Calculation of input rate functionInput rate function was the in vivo drug release rates of TA-SRPs,

which was obtained via the point-area deconvolution procedureusing the weight function and response function. Input values Ii

were calculated in turn by formula (4), where total response Ri isthe predicted plasma concentration of TA-SRPs at time point ti;

AUC is the area under the curve of weight function within time

Fig. 4. The mean predicted plasma concentration–time curve of TA-SRPs and themean percentage of variant angina attacks during 24 h.

interval ti � tk � ti � tk�1; Ii represents average in vivo drug releaserate of TA-SRPs within time interval ti�1 � ti.

3.2.4. Desired drug release profile of TA-SRPsThe in vivo cumulative release amount CAi of TA-SRPs at time

point ti could be obtained by formula (5), which is equal to thesum of TA release amount within each time interval tk�1 � tk dur-ing t0 � ti, where the release amount within each time interval isequal to average in vivo drug release rate Ik multiplied by timeinterval tk � tk�1. In vivo cumulative drug release percent CRi (%)of TA-SRPs at each time point ti is equal to the ratio of the cumu-lative release amount CAi at time point ti to the total cumulativerelease amount CAtotal. Then the desired cumulative drug releaseprofile of TA-SRPs was obtained. It was important to note thatthe calculated negative input values Ik should be replaced withzero in order to minimize the instability in the calculation(Langenbucher and Mysicka, 1985). This resulted in a stepwiseand relatively smooth appearance of the desired drug release pro-file, as shown in Fig. 5.

3.3. Formulation optimization of TA-SRPs

On one hand, pellets as a multiparticulate drug delivery systemcan overcome the poor and variable gastrointestinal tract drugabsorption and possess the ability to reduce or eliminate the influ-ence of food on bioavailability (Chopra et al., 2013). On the otherhand, coating film produced by blends of PVAc and PVA–PEG hasmany characteristics, including a drug release independent of pHand ionic strength and a high resistance to mechanical stress (Liuet al., 2012; Mies and S., 2004; Meyer and K., 2004). Therefore, sim-ilar release patterns in vivo with in vitro were readily achieved forTA-SRPs coated with PVAc/PVA–PEG due to little effect of the con-dition of gastrointestinal tract on drug release (Gaynor et al., 2008).Under this premise, the in vivo desired drug release profile ofTA-SRPs determined by deconvolution could be used for guidingthe formulation optimization of TA-SRPs. Here, all the in vitro drug

Fig. 5. The desired drug release profile of TA-SRPs determined by deconvolution.

Fig. 7. In vitro release profiles of TA-SRPs coated with 70:30 w/w of PVAc/PVA–PEGat different coating weights (n = 3) and the desired release profile determined bydeconvolution.

H.-X. Yan et al. / European Journal of Pharmaceutical Sciences 76 (2015) 156–164 161

release profiles of TA-SRPs varying PVAc/PVA–PEG blend ratios andcoating thickness were compared with the desired drug releaseprofile to screen the optimal coating formulation. The release pro-file of the optimal formulation was similar to the desired drugrelease profile (similarity factor f2 P 50). Eq. (1) was used for thecalculation of f2, where n is the number of time points, Rt and Tt

are the desired and observed cumulative drug release percent attime t, respectively (Zolnik and Burgess, 2008).

3.3.1. Ratio of PVAc/PVA–PEGThe variation of the blend ratio of coating components is an effi-

cient way to modify drug release patterns for the formulation(Ensslin et al., 2009). In vitro release behaviors of TA-SRPs coatedwith different ratios (90:10, 85:15, 70:30 w/w) of PVAc/PVA–PEGat the same coating weight of 5% (w/w) were evaluated and com-pared with the desired drug release profile by f2 analysis. As shownin Fig. 6, as the relative amounts of PVA–PEG (a soluble film coatingpolymer) increased, the release of TA from pellets became fasterand more complete. The drug release of TA-SRPs with 70:30 ratioof PVAc/PVA–PEG was close to 100% in 24 h while pellets with90:10 and 85:15 ratios of PVAc/PVA–PEG released only approxi-mate 50% of the total drug. Moreover, the drug release profile for70:30 ratio was similar to the desired release profile (f2 = 64.90),whereas the drug release profiles for 90:10 and 85:15 ratios dis-tinctly differed from the desired one (f2 = 25.10 and f2 = 29.30,respectively). Hence, 70:30 ratio of PVAc/PVA–PEG was chosen asthe optimal film coating composition and used for the furtherinvestigation.

3.3.2. Coating weight (thickness)In addition to the blend ratio variation of PVAc/PVA–PEG, the

coating film thickness also plays a vital role in drug release(Ensslin et al., 2009). In vitro release behaviors of TA-SRPs coatedwith 70:30 (w/w) ratio of PVAc/PVA–PEG were investigated at dif-ferent coating weights (3%, 5%, 10% w/w) and compared with thedesired drug release profile by f2 analysis (Fig. 7). The formulationof 3% coating weight was eliminated due to a less satisfactory sim-ilarity (f2 = 46.60), which was attributed to a slightly faster releaseduring 12 h. Although the drug release profiles for 5% and 10%coating weights were both similar to the desired release profile(f2 = 64.90 and f2 = 59.49, respectively), the former was selectedas the optimal formulation allowing for a more complete drugrelease, the time and costs consumption for preparation.

It was worth noting that the disparity in the release pattern ofTA-SRPs with different coating weights (f2 = 46.60, 64.90 and59.49, respectively) was much smaller than that of different blend

Fig. 6. In vitro release profiles of TA-SRPs coated with different ratios of PVAc/PVA–PEG at the same coating weight of 5% w/w (n = 3) and the desired release profiledetermined by deconvolution.

ratios (f2 = 25.10, 29.30 and 64.90, respectively), which indicatedthat the ratio variation of coating polymers PVAc/PVA–PEG wasdominant for the controlled drug release of TA-SRPs.

3.4. Investigation of drug release stabilities

By using the TA-SRPs prepared with the optimal coating formu-lation (70:30 of PVAc/PVA–PEG with 5% coating film weight) andoperation process, the influence of different pH condition, releasemethod and rotation rate on drug release was investigated andcompared by f2 analysis according to Eq. (1).

3.4.1. Release mediumThe different release condition in vitro was used to simulate the

different gastrointestinal tract environment in vivo. Due to pHchanges in the gastrointestinal tract, the release study of the opti-mized TA-SRPs in three kinds of release mediums with different pHwas necessary to investigate the effect of different pH condition ondrug release of TA-SRPs. As depicted in Fig. 8a, the drug releasebehaviors in 0.1 M HCl and pH 6.8 PBS were similar to that in dis-tilled water containing SDS (0.5%) with f2 of 93.41 and 68.67. Thisresult indicated that TA-SRPs could keep a stable drug release indifferent release mediums. The release extent and rate were inde-pendent of pH variation.

3.4.2. Release methodThe drug release behaviors of the optimized TA-SRPs by using

rotating basket method and paddle method were compared in thisrelease study. As shown in Fig. 8b, two release profiles were extre-mely alike with f2 of 84.78. It indicated that these TA-SRPs couldkeep a stable and similar drug release regardless of the choice ofthe release method.

3.4.3. Rotation rateThree release profiles of the optimized TA-SRPs obtained at a

rotation rate of 50, 100 and 150 rpm were presented in Fig. 8c.As it could be seen in this figure, with the significant increase ofrotation rate from 50 to 150 rpm, the drug release at each timepoint increased slightly especially from 4 h. With 100 rpm as a ref-erence, f2-test values for 50 rpm and 150 rpm were 70.18 and70.72, respectively. With 50 rpm as a reference, f2-test value for150 rpm was 55.10. These results showed the differences betweenthree release profiles were all less than 10%, which indicated thatthe rotation rate had no remarkable influence on the drug releaseof TA-SRPs.

Fig. 8. The drug release profiles of TA-SRPs (n = 3): (a) different pH medium; (b) different release method; (c) different rotation rate.

162 H.-X. Yan et al. / European Journal of Pharmaceutical Sciences 76 (2015) 156–164

3.5. Pharmacokinetic studies

In Section 3.3, the coating formulation of TA-SRPs was opti-mized as follows: 70:30 of PVAc/PVA–PEG with 5% coating filmweight. However, pharmacokinetic studies were needed to validatewhether the actual plasma drug concentration time course of thisformulation was in accordance with the predicted one.

The actual plasma concentration–time curve of the optimizedTA-SRPs (70:30 of PVAc/PVA–PEG and 5% coating weight) andthe simulated mean predicted plasma concentration–time curveof TA-SRPs were plotted together in Fig. 9. Apparently, plasma drugconcentrations of two curves possessed a basically consistent trendover the entire period of time. The percent prediction error (%PE) ofTA concentration at each time point was calculated by equation%PE = (observed value � predicted value)/observed value � 100%(Qi et al., 2003). The absolute %PE values from 4 to 16 h did notexceed 15% and the absolute %PE values from 8 to 12 h were lessthan 10%. Such low %PE values suggested that the actual observedplasma concentration values were indeed very close to the pre-dicted concentration data calculated and fitted according to AA fre-quency. These results demonstrated that these optimized pelletssucceeded in producing a sustained release effect in vivo with rel-atively higher plasma concentrations in 8–12 h. In addition, theobserved pharmacokinetic parameters (Tmax, Cmax, MRT, AUC0–t

and AUC0–1) of the optimized TA-SRPs were given and comparedwith the predicted values in Table 1. These almost equivalentnumerical values further confirmed that in vivo pharmacokineticprocess of these TA-SRPs after oral administration was in accor-dance with the predicted one.

3.6. Pharmacodynamic studies

In addition to the pharmacokinetic studies, pharmacodynamicstudies were also necessary to estimate the in vivo real drugefficacy of these optimized TA-SRPs.

Fig. 9. The plasma concentration–time curve of the optimized TA-SRPs (70:30 ofPVAc/PVA–PEG and 5% coating weight, w/w) in rabbits after oral administration in adose of 30 mg kg�1 (n = 6) and the mean predicted plasma concentration–timecurve of TA-SRPs.

3.6.1. Pharmacodynamic indexTA can stimulate NO production in vascular endothelial cells

(Huang et al., 2007). NO is a powerful vasodilator and plays a signif-icant role in the treatment of the cardiovascular diseases. As animportant regulator of the cardiovascular system, NO can dilatecoronary artery, increase coronary blood flow, increase myocardialhypoxia tolerance, protect vascular endothelial cells and preventmyocardial ischemia. In this study, NO concentration variation inthe serum was used as the pharmacodynamic index to investigatethe drug efficacy of TA pellets. The equation is (DNO)t =(NO)t � (NOcontrol)t, where (DNO)t is serum NO concentration varia-tion at time t, (NO)t is NO concentration after administration ofTA-tSD-IRPs or TA-SRPs at time t, (NOcontrol)t is NO concentrationof the controlled group at time t. The DNO concentration–timecurves of the TA-tSD-IRPs and the optimized TA-SRPs after oraladministration were presented in Fig. 10. TA-tSD-IRPs could gener-ate a fast and high drug efficacy in vivo, but they lost efficacy quickly.In contrast, TA-SRPs produced drug efficacy slowly, and maintainedthe stable efficacy for a long time, which could reduce the drug sideeffects effectively.

3.6.2. Relationship between drug concentration and efficacyThe drug efficacy-concentration curves of the TA-tSD-IRPs and

the optimized TA-SRPs were plotted together in Fig. 11. It couldbe found that DNO concentration versus TA plasma concentrationfollowing administration of TA-tSD-IRPs showed a clockwise hys-teresis loop, the arrows indicating the direction of time(Fig. 11a). With the increase of drug concentration from about 46to 80 ng mL�1, the drug efficacy of TA-SRPs was increased accord-ingly, while time to maximal efficacy was earlier than time to peakdrug concentration. This result indicated that their drug efficacydid not always have a positive relationship with the plasma

Fig. 10. The DNO concentration–time curves of TA-tSD-IRPs and TA-SRPs (70:30 ofPVAc/PVA–PEG and 5% coating weight, w/w) in rabbits after oral administration in adose of 30 mg kg�1 (n = 6).

Fig. 11. The drug efficacy-concentration curves in rabbits after oral administrationin a dose of 30 mg kg�1 (n = 6): (a) TA-tSD-IRPs; (b) TA-SRPs (70:30 of PVAc/PVA–PEG and 5% coating weight, w/w).

H.-X. Yan et al. / European Journal of Pharmaceutical Sciences 76 (2015) 156–164 163

concentration. In addition, with the decrease of drug concentra-tion, the drug efficacy of TA-SRPs significantly decreased, com-pared with the drug efficacy of same drug concentration in therising phase of drug concentration. These results manifested thatafter administration of TA-tSD-IRPs, although a high drug efficacycould be achieved quickly, the NZW rabbits with angina also pro-duced drug tolerance for TA rapidly in vivo. In contrast, DNO con-centration versus TA plasma concentration followingadministration of TA-SRPs showed a counter-clockwise hysteresisloop with almost simultaneous drug concentration–efficacy(Fig. 11b). Time to maximal efficacy was equal to time to peak drugconcentration. And the higher drug efficacy could be obtained inthe declining phase of drug concentration, compared with that inthe rising phase. These were because TA-SRPs released TA slowlyin vivo and then TA could continuously stimulate NO secretion invascular endothelial cells, not generating drug tolerance.Therefore, TA-SRPs had better efficacy in treating variant angina,compared with TA-tSD-IRPs.

4. Conclusion

In this research, the formulation design and optimization ofTA-SRPs was carried out based on deconvolution and variant ang-ina chronotherapy theory. The pharmacokinetic and pharmacody-namic studies in NZW rabbits confirmed these optimized TA-SRPshad a suitable plasma drug concentration time course with betterdrug efficacy. In comparison to the common design method, thispattern could markedly reduce blindness and complexity in thedevelopment of chronotherapeutic modified-release preparations.As a promising approach, it has a great potential to be adopted inthe design and optimization of other drugs or dosage forms inthe further.

Acknowledgements

This study was financially supported by National NaturalScience Foundation of China (No. 81473151) and the Priority

Academic Program Development of Jiangsu Higher EducationInstitutions. Thanks to JRS, ISP, Colorcon and BASF corporationsfor providing the excipients and sugar spheres.

References

Administration, F.a.D., 1997. Guidance for industry: extended release oral dosageforms: development, evaluation, and application of in vitro/in vivo correlations.Rockville, MD: Food and Drug Administration.

Chopra, S., Venkatesan, N., Betageri, G.V., 2013. Formulation of lipid bearing pelletsas a delivery system for poorly soluble drugs. Int. J. Pharm. 446, 136–144.

Chu, T., Zhang, Q., Li, H., Ma, W.C., Zhang, N., Jin, H., Mao, S.J., 2012. Development ofintravenous lipid emulsion of tanshinone IIA and evaluation of its anti-hepatoma activity in vitro. Int. J. Pharm. 424, 76–88.

Ensslin, S., Moll, K.P., Metz, H., Otz, M., Mäder, K., 2009. Modulating pH-independentrelease from coated pellets: effect of coating composition on solubilizationprocesses and drug release. Eur. J. Pharm. Biopharm. 72, 111–118.

Fan, Y.X., Li, J.F., Dong, C., 2005. Preparation and study on the inclusion complexes oftwo tanshinone compounds with b-cyclodextrin. Spectrochim. Acta A Mol.Biomol. Spectrosc. 61, 135–140.

Gao, S., Liu, Z.P., Li, H., Little, P.J., Liu, P.Q., Xu, S.W., 2012. Cardiovascular actions andtherapeutic potential of tanshinone IIA. Atherosclerosis 220, 3–10.

Gaynor, C., Dunne, A., Davis, J., 2008. A comparison of the prediction accuracy of twoIVIVC modelling techniques. J. Pharm. Sci. 97, 3422–3432.

Gillespie, W.R., 1991. Noncompartmental versus compartmental modelling inclinical pharmacokinetics. Clin. Pharmacokinet. 20, 253–262.

Hao, H.P., Wang, G.J., Cui, N., Li, J., Xie, L., Ding, Z.Q., 2006. Pharmacokinetics,absorption and tissue distribution of tanshinone IIA solid dispersion. PlantaMed. 72, 1311–1317.

Hao, H.P., Wang, G.J., Cui, N., Li, J., Xie, L., Ding, Z.Q., 2007. Identification of a novelintestinal first pass metabolic pathway: NQO1 mediated quinone reduction andsubsequent glucuronidation. Curr. Drug Metab. 8, 137–149.

Huang, K.J., Wang, H., Xie, W.Z., Zhang, H.S., 2007. Investigation of the effect oftanshinone IIA on nitric oxide production in human vascular endothelial cellsby fluorescence imaging. Spectrochim. Acta A Mol. Biomol. Spectrosc. 68, 1180–1186.

Jantratid, E., De Maio, V., Ronda, E., Mattavelli, V., Vertzoni, M., Dressman, J.B., 2009.Application of biorelevant dissolution tests to the prediction of in vivoperformance of diclofenac sodium from an oral modified-release pelletdosage form. Eur. J. Pharm. Sci. 37, 434–441.

Kan, S.L., Li, J., Liu, J.P., He, H.L., Zhang, W.J., 2014. Evaluation of pharmacokineticsand pharmacodynamics relationships for Salvianolic Acid B micro-porous osmotic pump pellets in angina pectoris rabbit. Asian J. Pharm. Sci. 9,137–145.

Langenbucher, F., Mysicka, J., 1985. In vitro and in vivo deconvolution assessment ofdrug release kinetics from oxprenolol Oros preparations. Br. J. Clin. Pharmacol.19, 151S–162S.

Lemmer, B., 1996. The clinical relevance of chronopharmacology in therapeutics.Pharmacol. Res. 33, 107–115.

Lemmer, B., 2006. The importance of circadian rhythms on drug response inhypertension and coronary heart disease—from mice and man. Pharmacol. Ther.111, 629–651.

Li, J., Wang, G.J., Li, P., Hao, H.P., 2005. Simultaneous determination of tanshinoneIIA and cryptotanshinone in rat plasma by liquid chromatography-electrosprayionisation-mass spectrometry. J. Chromatogr. B 826, 26–30.

Li, C.D., Liu, J.P., Zeng, Z.Z., Zheng, Y., 2008. Study on the solubility and permeabilityof tanshinone II A and on the excipients increasing the solubility andpermeability. Lishizhen Med. Mater Med. Res. 19, 1724–1726.

Li, J., Liu, P., Liu, J.P., Zhang, W.L., Yang, J.K., Fan, Y.Q., 2012. Novel Tanshinone II Aternary solid dispersion pellets prepared by a single-step technique: in vitroand in vivo evaluation. Eur. J. Pharm. Biopharm. 80, 426–432.

Liu, P., Li, J., Liu, J.P., Yang, J.K., Fan, Y.Q., 2012. Release behavior of tanshinone IIAsustained-release pellets based on crack formation theory. J. Pharm. Sci. 101,2811–2820.

Meyer, K., Kolter, K., 2004. Reliability of drug release from an innovative single unitKollicoat� drug delivery system, CRS 2004. In: 31st International Symposium onControlled Release of Bioactive Materials.

Mies, S., Meyer, K., Kolter, K., 2004. Correlation of drug permeation through isolatedfilms and coated dosage forms based on Kollicoat� SR 30D/IR. iN: 2004 AAPSAnnual Meeting and Exposition.

Mori, H., Nakamura, N., Tamura, N., Sawai, M., Tanno, T., Narita, T., Singh, R.B.,Otsuka, K., 2002. Circadian variation of basal total vascular tone andchronotherapy in patients with vasospastic angina pectoris. Biomed.Pharmacother. 56, 339–344.

Portaluppi, F., Lemmer, B., 2007. Chronobiology and chronotherapy of ischemicheart disease. Adv. Drug Deliv. Rev. 59, 952–965.

Prinzmetal, M., Kennamer, R., Merliss, R., Wada, T., Bor, N., 1959. Anginapectoris I. A variant form of angina pectoris: preliminary report. Am. J. Med.27, 375–388.

Qi, X.H., Liu, R.R., Sun, D.X., Ackermann, C., Hou, H.M., 2003. Convolution method topredict drug concentration profiles of 2, 3, 5, 6-tetramethylpyrazine followingtransdermal application. Int. J. Pharm. 259, 39–45.

Shang, Q.H., Xu, H., Huang, L., 2012. Tanshinone IIA: a promising naturalcardioprotective agent. Evid. Based Complement. Alternat. Med.

164 H.-X. Yan et al. / European Journal of Pharmaceutical Sciences 76 (2015) 156–164

Sirisuth, N., Eddington, N.D., 2002. The influence of first pass metabolism on thedevelopment and validation of an IVIVC for metoprolol extended releasetablets. Eur. J. Pharm. Biopharm. 53, 301–309.

Smolensky, M.H., Portaluppi, F., 1999. Chronopharmacology and chronotherapy ofcardiovascular medications: relevance to prevention and treatment of coronaryheart disease. Am. Heart J. 137, S14–S24.

Sunesen, V.H., Pedersen, B.L., Kristensen, H.G., Müllertz, A., 2005. In vivo in vitrocorrelations for a poorly soluble drug, danazol, using the flow-throughdissolution method with biorelevant dissolution media. Eur. J. Pharm. Sci. 24,305–313.

Vaughan, D.P., Dennis, M., 1978. Mathematical basis of point-area deconvolutionmethod for determining in vivo input functions. J. Pharm. Sci. 67, 663–665.

Yeh, K.C., Holder, D.J., Winchell, G.A., Wenning, L.A., Prueksaritanont, T., 2001. Anextended point-area deconvolution approach for assessing drug input rates.Pharm. Res. 18, 1426–1434.

Zhang, W.L., Liu, J.P., Li, S.C., Chen, M.Y., Liu, H., 2008. Preparation and evaluation ofstealth Tashinone IIA-loaded solid lipid nanoparticles: Influence of Poloxamer188 coating on phagocytic uptake. J. Microencapsul. 25, 203–209.

Zhang, W.L., He, H.L., Liu, J.P., Wang, J., Zhang, S.Y., Zhang, S.S., Wu, Z.M., 2013.Pharmacokinetics and atherosclerotic lesions targeting effects of tanshinone IIAdiscoidal and spherical biomimetic high density lipoproteins. Biomaterials 34,306–319.

Zolnik, B.S., Burgess, D.J., 2008. Evaluation of in vivo–in vitro release ofdexamethasone from PLGA microspheres. J. Control. Release 127, 137–145.