JCR Layre Busulfan

111 (2006) 271–280www.elsevier.com/locate/jconrel

Journal of Controlled Release

Novel composite core-shell nanoparticles as busulfan carriers

A. Layre a, P. Couvreur a, H. Chacun a, J. Richard b,1, C. Passirani c,D. Requier b, J.P. Benoit c, R. Gref a,⁎

a UMR CNRS 8612, Faculty of Pharmacy, Paris-Sud University, 5, rue Jean Baptiste Clément, 92296 Châtenay-Malabry, Franceb Ethypharm, 194 Bureau de la colline, 92213 Saint-Cloud, France

c INSERM U646, Faculty of Pharmacy, 10 rue A. Boquel, 49100 Angers, France

Received 23 August 2005; accepted 10 January 2006Available online 20 February 2006

Abstract

This study presents a method for the design of novel composite core-shell nanoparticles able to encapsulate busulfan, a crystalline drug. Theywere obtained by co-precipitation of mixtures of poly(isobutylcyanoacrylate) (PIBCA) and of a diblock copolymer, poly(ε-caprolactone)–poly(ethylene glycol) (PCL–PEG), in different mass ratios. The nanoparticle size, morphology and surface charge were assessed. The chemicalcomposition of the top layers was determined by X-ray photo-electron spectroscopy (XPS). 3H-labelled busulfan was used in order to determinethe drug loading efficiency and the in vitro drug release by liquid scintillation counting. Physico-chemical techniques such as Zeta potentialdetermination and XPS analysis provided evidence about a preferential surface distribution of the PCL–PEG polymer. Therefore, compositenanoparticles have a ‘core-shell’-type structure, where the “core” is essentially formed by the PIBCA polymer and the “shell” by the PCL–PEGcopolymer. The use of PIBCA to form the core of the nanoparticles leads to a 2–4 fold drug loading increase, in comparison to the single PCL–PEG nanoparticles. In addition, the complement activation results showed a significant difference between the composite nanoparticles and thesingle PIBCA nanoparticles, thus demonstrating that PEG at the surface of the nanoparticles reduced the complement consumption. The PIBCA:PCL–PEG composite nanoparticles prepared using the new co-precipitation method here described represent an original approach for busulfanadministration.© 2006 Elsevier B.V. All rights reserved.

Keywords: Nanoparticle; Poly(ethylene glycol); Drug delivery; XPS; Complement

1. Introduction

Busulfan is a bifunctional alkylating agent [1], which iswidely used at high dose as a part of myeloablative regimenbefore both allogenic and autologous bone marrow transplan-tation for the treatment of haematological malignancies [2] andnon-malignant disorders such as immunodeficiency [3]. For along time, busulfan has been available only in oral form and awide intra-patient and inter-patient bioavailability variability inboth adult and children has been reported [4]. Moreover, severe

⁎ Corresponding author. Tel.: +33 1 46 83 59 09; fax: +33 1 46 61 93 34.E-mail address: [email protected] (R. Gref).

1 New permanent address: Serono, Via di Valle Caia, 22, 00040 Ardea(Roma), Italia.

0168-3659/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.jconrel.2006.01.002

side effects were reported such as the veino-occlusive disease(VOD). This pathology has been correlated with a high systemicexposure to busulfan expressed as the area under the plasmaconcentration–time curve [5].

In order to overcome these problems, intravenous formula-tions of busulfan were developed, using cosolvent mixtures[6,7]. However, these organic solvents have their own well-documented toxicity [8,9]. Therefore, to avoid the massive useof organic solvents, injectable colloidal carriers, such as con-ventional liposomes [10] and biodegradable polymer nanopar-ticles have been elaborated [11,12] . However, these carriers hadencapsulation efficiencies lower than 1% (w/w). Indeed,successful encapsulation of busulfan into nanoparticles hasnever been described, yet.

In previous studies, we have established the ability of poly(isobutylcyanoacrylate) (PIBCA) nanoparticles to encapsulatelarger amount of busulfan than other polymers [13].

272 A. Layre et al. / Journal of Controlled Release 111 (2006) 271–280

Unfortunately, such nanoparticles displaying large hydrophobicsurface areas were rapidly recognised by the mononuclearphagocyte system (MPS), eliminated from the blood streamwithin minutes and ended up mainly in liver and spleen [14].The liver uptake of busulfan-loaded nanoparticles might run therisk to increase the occurrence of the VOD. It is our hypothesisthat by using long circulating nanoparticles, able to dramaticallyreduce liver accumulation, the probability for the VOD to occurwould be reduced.

Long-circulating nanoparticles can be obtained by coating orgrafting flexible and non-ionic polymers, such as poly(ethyleneglycol) (PEG) onto the hydrophobic nanoparticles' surface[15,16]. Several methods have been investigated to preparePEG-coated poly(alkylcyanoacrylate) nanoparticles, such as theadsorption of amphiphilic Poloxamer or Poloxamine copoly-mers onto nanoparticles' surface [17]. However, no significantbiodistribution changes were observed in comparison withuncoated nanoparticles. Indeed, it was supposed that in vivo,these copolymers might be competitively displaced from thesurface by plasma proteins having a higher affinity for PIBCA[17] or that they could be released as a consequence of poly-mer's bioerosion [18]. Another approach to design PEG-coatedPIBCA nanoparticles consisted in grafting PEG onto thenanoparticles during the emulsion polymerization process ofthe isobutylcyanoacrylate monomer [19]. However, it has beenshown that some drugs were able to react with the cyanoacrylatemonomers too, which dramatically affected both nanoparticleformation and the biological activity of the drug [20,21]. Sincebusulfan is also a very strongly reactive drug, this method islikely not adapted for the preparation of busulfan-loadednanoparticles. More recently, the synthesis of block copolymerscontaining PEG and polyalkylcyanoacrylate blocks has beenreported [22]. These copolymers could be nanoprecipitatedunder the form of nanoparticles with increased blood circulationtimes compared with the uncoated poly(alkylcyanoacrylate)nanoparticles [23]. Nevertheless, it was observed that the bloodhalf life in rats of these “PEGylated” polyalkylcyanoacrylatenanoparticles was significantly lower than that of the PEGylatedpolyester nanoparticles [23,24].

Thus, the aim of this study was to design novel compositecore-shell nanoparticles able to combine the ability of thePIBCA to efficiently encapsulate busulfan, with the excellentstealth properties of the diblock copolymer, poly(ε-caprolac-tone)–poly(ethylene glycol) (PCL–PEG), which is expected toprovide increased blood half lives to the nanoparticles [16,24].This paper describes the feasibility of these compositenanoparticles with special emphasis on their surface character-ization. Loading efficiency of busulfan and in vitro drug releasewere investigated, too.

2. Materials and methods

2.1. Materials

Poly(isobutylcyanoacrylate) (PIBCA) was synthesized byan anionic polymerization of isobutylcyanoacrylate monomersin water. The monomer (1 ml) was added in one shot to water

(15 ml). The polymerization carried on during 1 h 30 at 40 °Cunder magnetic stirring (1200 rpm). After this time, a milkysuspension was obtained together with a polymer aggregatearound the magnetic stirrer. This polymer was collected in twofractions: the milky suspension was freeze dried (fraction 1),and the aggregated polymer was dissolved in acetone anddried under vacuum at room temperature (fraction 2). Thepolymer thus obtained in fraction 2 (representing more than85% of the polymer synthesized) was further used in thenanoprecipitation process.

The poly(ε-caprolactone)–poly(ethylene glycol) diblockcopolymer was synthesized as previously described [25].Briefly, given amounts of the monomer (ε-caprolactone)freshly distilled and monomethoxy polyethylene glycol(MPEG) (weight average molar mass 2000 g/mol, Sigma-Aldrich, Germany) were dissolved in xylene. The weight ratioε-caprolactone :MPEG was 9 :1. Stannous octanoate (Sigma-Aldrich, Germany) purified by distillation was used as acatalyst in equimolar quantity with regard to MPEG. Thereaction was carried on at 110 °C for 6 h. Average molarmasses (Mw) were determined using gel permeation chroma-tography equipped with a refractometric and a multiangle lightscattering detector (Wyatt Dawn Model F, Milton Roy, WyattTechnology). The diblock copolymer symbolized PCL10k–PEG2K had a PCL block with a Mw of 10,000 g/mol and aPEG block with a Mw of 2000 g/mol.

Busulfan was purchased from Sigma-Aldrich (Germany) andTritium-labelled busulfan from RC TRITEC (Switzerland).Acetone was obtained from Carlo-Erba (France). Poloxamer188 (Synperonic PE/F68) and D-(+)trehalose were purchasedfrom Fluka (Switzerland). Rat plasma was obtained fromCharles River (USA).

2.2. Nanoparticle preparation

To prepare the composite nanoparticles, firstly two PIBCAand PCL10k–PEG2k solutions at 20 mg/ml were mixed indifferent volume ratios (80 / 20, 70 / 30, 60 / 40, 50 / 50).Secondly, busulfan was dissolved in acetone at 4 mg/ml.After, 500 μl of the polymer solution and 500 μl of thebusulfan solution were mixed. The resulting solution (1 ml)was then injected into water (2 ml) under magnetic stirring(1200 rpm) at room temperature leading to spontaneousformation of composite nanoparticles. Acetone was eliminatedusing a rotative evaporator (rotavapor®) at room temperature.The suspensions were purified by centrifugation (MR22i,Jouan, France) (5 min at 630 ×g), prefiltration (Acrodisc, Glassfiber membrane, 1 μm, Pall, USA), and finally filtration(Millex®-HV, 0.45 μm, Millipore, USA) in order to eliminatedrug crystals which might form during acetone evaporationstep [26]. The composite nanoparticles are further namedCNP(PIBCA:PCL10k–PEG2k). For example, CNP(80 : 20) was obtain-ed from a PIBCA:PCL10k–PEG2k mixture in 80:20 volume ratio.

Single PIBCA and single PCL10K–PEG2K nanoparticleswere prepared by the nanoprecipitation process, as previouslydescribed by Fessi and Devissaguet [27]. Briefly, an organicsolution of polymer (PIBCA or PCL10K–PEG2K) (10 mg) and

273A. Layre et al. / Journal of Controlled Release 111 (2006) 271–280

busulfan (2 mg) in acetone (1 ml) was injected into 2 ml waterunder magnetic stirring (1200 rpm) at room temperature.Nanoparticle suspensions were purified as described previously.Drug free nanoparticles were prepared according to the sameprocedure, without the purification steps.

The drug content in nanoparticles was assessed using tritium-labelled busulfan. 3H-Busulfan loaded CNP(PIBCA:PCL10k–PEG2k)nanoparticles and single PIBCA or PCL10K–PEG2K nanopar-ticles were prepared as described above, with a theoreticalactivity of 1.6 μCi/mg drug. The nanoparticles were collected bycentrifugation (30,000 ×g for 30min) (MR22i, Jouan, France) anddried in desiccators under vacuum at room temperature over24 h. After weight determination, the dried nanoparticles weredissolved in 1 ml acetone. The drug loading was determined byliquid scintillation. It was expressed as the weight of drug in thenanoparticles divided by the weight of the dried nanoparticlescollected.

2.3. Nanoparticle characterization

2.3.1. Nanoparticle size measurementThe mean particle diameters were measured by laser light

scattering using a nanosizer (Coulter® N4MD, CoulterElectronics Inc, Hialeath, USA). Each sample was properlydiluted in water, in order to maintain the number of counts persecond between 5 ·104 and 1 ·106. Water was filtered with a0.22 μm filter to remove any impurities that could affectscattering of the light. Each sample was measured three timesfor at least three minutes at 20 °C and at an angle of 90°. Bothunimodal and size distribution processor (SDP) analysis wereperformed.

2.3.2. Nanoparticle morphologyNanoparticles were observed using transmission electron

microscopy after freeze fracture. A small drop of an aqueousnanoparticle suspension was deposited into a 100 μm deepsymmetric cup. Then, the sample was frozen using a highpressure cooling device HPM 010 (Bal-Tec). Fracturing, etch-ing and shadowing, using Pt–C, were performed in a Bal-TecModel BAF 400T apparatus. The replicas of the surface werethen floated off by specimen submerging in successive baths ofwater/acetone, water, NaOH (1 M), water and acetone. Finally,the replicas were collected onto naked 400 mesh grids whichwere subsequently mounted in a transmission electron micros-copy (TEM) for inspection. TEM observations were performedon a LEO 912 Omega high resolution microscope working at120 kV.

2.4. Surface characterization

2.4.1. Zeta potential determinationThe nanoparticle zeta potential measurements were carried

out using a Zeta Sizer 4 (Malvern Instruments Ltd. UK). Eachsample was properly diluted in NaCl (1 mM), in order tomaintain the number of counts per second around 600. Threemeasurements were carried out for each sample and the meanvalues and standard deviations were calculated.

2.4.2. X-ray photoelectron spectroscopy analysisIn order to investigate the chemical composition of the

nanoparticle top layers and to gain evidence about thepreferential localization of the PCL10k–PEG2k copolymers,XPS analysis was carried out on the dry PIBCA, PCL10k–PEG2k and CNP(50 : 50) nanoparticles. Indeed, XPS allows todetermine the elemental and average chemical composition of amaterial at its very surface (about 10 nm depth), by measuringthe binding energy of the 1s electrons emitted from the atoms inthese top layers. The spectra were recorded with a thermo VGScientific ESCALAB 250 spectrometer (VG Instrument, UK)equipped with a monochromatic Al KαX-ray source (1486.6 eV)at a spot size of 650 μm. The take-off angle relative to the sampleholder surface was 90°. The pressure in the analysis chamberwas ca. 2–3 ·10−8 mbar. The pass energy was set at 150 and40 eV for the survey and the narrow scans, respectively. The stepsize was 1.0 eV for the survey spectra and 0.1 for the narrowregions, respectively. A 4 eV flood gun combined with an argongun was used to neutralize the surface charge. The spectra werecalibrated against the N1s peak PIBCA centred at 399.7 eV.Identification of chemical functional groups was obtained fromthe high-resolution peak analysis of carbon 1s (C1s), oxygen 1s(O1s), and nitrogen 1s (N1s) envelopes. Data acquisition andprocessing software was achieved using Aventage software,version 2. The surface composition was determined using themanufacturer's sensitivity factors. The fractional concentrationof a particular element A (% A) was computed using:

% A ¼ ðIA=SAÞXn

i¼1

ðIi=SiÞ� 100

Where Ii and Si are the integrated peak areas of each of then detected elements and the sensitivity factors, respectively.

The comparison between the composition in atomic ratiosobtained from XPS analysis and from the theoretical calcula-tions should provide evidence about a preferential surfacedistribution or a uniform distribution of chemical elements. Thetheoretical atomic ratios were calculated from the chemicalformulas of PIBCA ((C8NO2)n) and PCL10k–PEG2k ((C6O2)88–(C2O)45), supposing that all the different chemical elementswere uniformly distributed within the analysed samples. Sincehydrogen is not detected in XPS, this element was not taken intoaccount in the theoretical formulas.

2.5. Complement activation

The method used to assess the interactions of thenanoparticles with complement was the haemolytic CH50test. The principle of the procedure is based on the fact that,when sensitized sheep erythrocytes are in contact with thecomplement proteins in human serum, the classical complementpathway is activated, resulting in the lysis of erythrocytes andthe release of haemoglobin. This technique consisted in thedetermination of the amount of units of serum able to lyse 50%of a fixed number of sensitized sheep erythrocytes (CH50units). When the human serum is put in contact with activating

274 A. Layre et al. / Journal of Controlled Release 111 (2006) 271–280

nanoparticles, opsonization occurs and less complementproteins remain in the serum to lyse the sheep erythrocytes,thus the CH50 units decrease.

The CH50 test was described previously [28]. Briefly,Veronal-buffered saline containing 0.15 mM Ca2+ and 0.5 mMMg2+ (VBS+) were prepared as described elsewhere [29].Human serum was a pool of forty donor blood samples providedby the ‘Etablissement Français du Sang’ (Angers, France),aliquoted and stored at −80 °C until used. Sheep erythrocytes(Eurobio, France) were sensitized by rabbit anti-sheep erythro-cyte antibodies (Haemolytic Serum, Eurobio, France) andsuspended at a final concentration of 1 ·108 cells/ml in VBS+.First, a little amount of Poloxamer 188 (2.5 ·10−1 μg per cm2 ofnanoparticle surface) was incubated with the nanoparticles, inorder to avoid the nanoparticle aggregation that occurred inpresence of VBS+. Increasing amount of nanoparticle suspen-sions were added to human serum in VBS+, so that the finaldilution of the human serum in the reaction mixture was 1 /4 (v/v)in a final volume of 1 ml. After a 60 min incubation at 37 °C withgentle agitation, the suspension was diluted 1/25 (v/v) withVBS+. Aliquots at different dilutions were added to a givenvolume of sensitized sheep erythrocytes. After a 45 min incu-bation at 37 °C, the suspension was centrifuged (2000 rpm,10 min). Then, the optical density at 415 nm of the supernatant,related to the lytic capacity of the serum, was recorded. Theamount ofCH50 units remaining in the serumwas determined andcompared to the result obtained with control serum. The resultswere expressed as consumption of CH50 units. In order to com-pare the effect of nanoparticles with different mean diameters, theamount of nanoparticles in contact with serum was expressed interms of surface area. Nanoparticle surface area was calculated asdescribed elsewhere [30].

The distance between two terminally attached PEG chains atthe surface was calculated, assuming that all the PEG chainsmigrate at the nanoparticles' surface during their preparation[15] and that all the prepared nanoparticles are homogeneous interms of composition and size. Thus, it is possible to calculatethe surface SPEG that would occupy each PEG chain on thesurface of the nanoparticles, taking into account the nanopar-ticle mean radius (r).

SPEG ¼ 4pr2=NPEG

where NPEG is the total number of PEG chains in onenanoparticle. Or,

NPEG ¼ mNPfN =MPEG

where MPEG is the molar mass of the block PEG (2000 g/mol),N is the Avogadro number, f is the weight fraction of PEG inthe nanoparticles and mNP is the mass of one nanoparticle. mNP

is given by:

mNP ¼ 4pr3dNP=3

where the density of the nanoparticle (dNP) was equal to 1.2 g/cm3 [31]. We finally obtain:

SPEG ¼ ð3MPEGÞ=ðr � dNP � f �NÞ

Assuming that each PEG chain would occupy the centre of asquare at the nanoparticles' surface, the distance between twoPEG chain is dPEG ¼ ffiffiffiffiffiffiffiffiffi

SPEGp

.It results that in the cases studied:

CNPð80 : 20Þ: r ¼ 70d10−7 cm; f ¼ 0:03 and dPEG ¼ 20 A°

CNPð50 : 50Þ: r ¼ 60d10−7 cm; f ¼ 0:08 and dPEG ¼ 13 A°

PCL10k�PEG2k: r¼ 40d10−7cm; f ¼ 0:16 and dPEG ¼ 11 A°

2.6. Freeze-drying process

Aliquots of composite nanoparticles (400 μl) at a concen-tration of 5 mg/ml were added to sugar and surfactant solution(400 μl) at various concentrations before freeze-drying. Themass ratio polymer : cryoprotectant in the nanoparticle suspen-sion ranged from 1:2 to 1 :10. Freezing was performed in aconventional freezer (−20 °C). The frozen nanoparticles werethen lyophilized using a freeze-drying system (Christ-Alpha 1-4,Bioblock Scientific, France) over 24 h. Temperature cycle was−30 °C: +30 °C and the vacuum was 8·10−3 mbar.

2.7. In vitro release study

The drug release experiments were carried out for theCNP(50 : 50) suspension. These experiments were performed at37 °C in water and in rat plasma. Freshly prepared tritiumlabelled busulfan-loaded nanoparticle suspensions (5 μCi/mgbusulfan) were diluted with the release medium studied. Then,the nanoparticle suspension was separated in 1 ml aliquots andplaced on a shaker (Titramax 101, Heidolph, Germany) at 37 °C.At each given time-point, one of those aliquots was centrifuged(10 min at 30,000 ×g). Busulfan in the supernatant was assessedby liquid scintillation. Nanoparticles were collected and dried ina desiccator during 24 h under vacuum at room temperature.After weight measurement, the nanoparticles were dissolved in1 ml acetone. The busulfan quantity in the nanoparticle fractionwas determined by liquid scintillation.

3. Results and discussion

Feasibility and characterization of drug-free compositenanoparticles.

3.1. Size determination

The mean diameter of single PIBCA, single PCL10k–PEG2k

nanoparticles and composite nanoparticles (weight ratios 80 :20to 50 :50) are reported in Table 1. The co-precipitation ofPIBCA and PCL10k–PEG2k provided nanoparticles with a sizelower than 200 nm, whatever the polymers weight ratio.However, the nanoparticle mean diameter was affected by thePIBCA:PCL10k–PEG2k ratio, the mean diameter of purePIBCA nanoparticles was 143 (±37 nm) and it decreaseduntil 106 (±33 nm) for the CNP(50 : 50) nanoparticles. Thus, anincrease in the amount of PCL10k–PEG2k in the PIBCA:PCL10k–PEG2k mixtures lead to a decrease of the resultingnanoparticle mean diameter. These data might be explained by

Table 1Z-average mean diameter and zeta potential values of single PIBCA, singlePCL10k–PEG2k and CNP composite nanoparticles (weight ratios from 80:20 to50 :50)

Particle type Mean diameter (nm) Zeta potential (mV)

PIBCA 143±37 −39±0.1CNP(80 : 20) 139±43 −42±0.6CNP(70 : 30) 124±34 −26±0.3CNP(60 : 40) 123±38 −28±0.3CNP(50 : 50) 106±33 −22±1.1PCL10k–PEG2k 67±22 −15±0.9

Each value was the average of three different experiments±SD.

275A. Layre et al. / Journal of Controlled Release 111 (2006) 271–280

the amphiphilic intrinsic properties of the PCL10k–PEG2k

copolymer, reducing the interfacial tension between the aqueousand the organic phase.

3.2. Nanoparticle morphology

In order to investigate the morphology of the nanoparticles,their cross section was observed by transmission electronmicroscopy after freeze fracture. Photomicrographs of the singlePIBCA, and single PCL10k–PEG2k nanoparticles as well as theCNP(50 : 50) composite nanoparticles are presented in Fig. 1. The

Fig. 1. Transmission electronic microscopy after freeze-fracture images of single PIBrepresents 200 nm.

PIBCA (Fig. 1A) and PCL10k–PEG2k (Fig. 1B) nanoparticlespresented a spherical shape and the latter ones were smaller. TheCNP(50 : 50) typically showed a core-shell structure (Fig. 1C.and 1D). No second population of smaller diameter could bedetected in these samples, as in the PCL10k–PEG2k ones. Theseobservations suggest that both PIBCA and PCL10k–PEG2k

associate to form mixed nanoparticles. We presume that theamphiphilic PCL10k–PEG2k would organize at the surfaceduring particle formation, and finally form a shell around thehydrophobic PIBCA core.

3.3. Drug-free nanoparticle surface characterization

3.3.1. Surface charge of nanoparticlesThe Zeta potential of both single PIBCA and PCL10k–PEG2k

nanoparticles, as well as that of composite nanoparticles (weightratios from 80/20 to 50/50) are reported in Table 1. The singlePIBCAnanoparticles displayed a Zeta potential of around−40mV,whereas the PCL10k–PEG2k nanoparticles had a Zeta potential of−15 mV. The CNP had intermediate Zeta potential values andthe higher the PCL10k–PEG2k content in the polymer mixtures,the higher the Zeta potential values. For example, the CNP(50 : 50)nanoparticles displayed Zeta potential values of −22 mV, closeto the one of the PCL10k–PEG2k nanoparticles. These findings

CA (A); single PCL10k–PEG2k (B) and CNP(50 : 50) (C and D) nanoparticles. Bar

Table 2XPS elemental ratios (C, N, O) and composition in atomic ratios C/N, C/O, andCO /C=O of single PIBCA, single PCL10k–PEG2k and CNP(50 : 50)nanoparticles

Sample XPS elementalratios (%)

Composition in atomic ratios

C N O C/N C/O CO/C=O

Theor XPS Theor XPS Theor XPS

PIBCA 75.0 7.8 17.2 8 9.6 4 4.4 – –PCL10k–PEG2k 75.1 0.0 25.0 0 0 2.8 3.0 2 1.7CNP(50 : 50) 74.8 2.2 23.0 15.9 34 3.3 3.2 – –

276 A. Layre et al. / Journal of Controlled Release 111 (2006) 271–280

also suggested the preferential localization of the PCL10k–PEG2k copolymers at the surface of the composite nanoparticles.

3.4. XPS analysis

The XPS scans of the different dried nanoparticles arepresented in Fig. 2. In the XPS spectra of PCL10k–PEG2k

nanoparticles (Fig. 2B), the main peaks were C1s and O1scentred at 285 and 533 eV, respectively. In PIBCA andCNP(50 : 50) XPS spectra (Fig. 2A and C, respectively), the mainpeaks were C1s, N1s, and O1s centred at 286, 400 and 533 eV,respectively. XPS is of particular interest in our case, as onlyPIBCA displays a N peak. The relative intensity of the N1s peakof the CNP(50 : 50) was lower than that of PIBCA nanoparticles.This is an indication that PIBCA core was covered with

2004006008001000

Binding energy (eV)

2004006008001000

Binding energy (eV)

2004006008001000

Binding energy (eV)

O1s

O1s

O1s

N1s

N1s

C1s

C1s

C1s

H O O OO

x yH

CH2 CHO

C

COOCH2

CHCH3CH3

nH

NA

B

C

Fig. 2. XPS scans of pure PIBCA (A), pure PCL10k–PEG2k (B), and CNP(50 : 50)nanoparticles (C).

PCL10k–PEG2k. To have a quantitative estimation about thiscoating, Table 2 presents the atomic composition (C, N, O) andthe composition in atomic ratios (C /N, C /O and CO/C=O)calculated from the data obtained using XPS analysis as well asthe theoretical ones (see Materials and methods section). Thecomparison between the ratios obtained from XPS analysis andfrom the theoretical calculations should provide evidence abouta preferential surface distribution or a uniform distribution ofchemical elements. The experimental and theoretical atomicratios (C /N and C/O) determined with PIBCA nanoparticleswere equal (Table 2), which proves there is a uniform dis-tribution of each chemical element in the nanoparticles. In thecase of the PCL10k–PEG2k nanoparticles, the experimental andthe theoretical CO/C=O ratios were equal too, corresponding toa uniform distribution of the poly(ethylene glycol) block and ofthe poly(ε-caprolactone) block. Therefore, XPS analysis couldnot prove a core-shell organisation of PCL10k–PEG2k polymer.This finding could be due to the XPS analysis depth (about 10nm). In fact, the PEG layer formed by PEG 2000 g/mol has alower thickness than 10 nm [32]. Therefore, the analysis depth(10 nm) is higher than the ‘PEGylated shell’ layer thickness andthe elemental ratio (CO/C=O) obtained by XPS takes intoaccount the shell (PEG2k) and part of the core depth (PCL10k) ofthe nanoparticles. Indeed, only with PEG blocks with molarmasses higher than 5000 g/mol, it was possible using XPS todetect the preferred localization of PEG at the surface [16,33].The theoretical C /N ratio of the CNP(50 : 50) nanoparticles was2-fold lower than the experimental ratio (Table 2). In otherwords, the experimental nitrogen content in the nanoparticleswas lower than the nitrogen content in the polymer blends.These data provide evidence about a preferential surface dis-tribution of the PCL10k–PEG2k polymer. Therefore, CNP(50 : 50)nanoparticles have a ‘core-shell’-type structure, where the“core” is essentially formed by the PIBCA polymer and the“shell” by the PCL10k–PEG2k polymer.

3.5. Complement activation

It is now well established that nanoparticle phagocytosis ismediated by opsonization, and that serum complement is amajor component of the opsonin system [34]. The evaluation ofthe nanoparticles–complement interactions represents a goodpreliminary experiment predictive of the in vivo fate of thecolloidal drug carriers after intravenous administration [35].

0

20

40

60

80

100

0 200 400 600 800 1000 1200

Surface area (cm2/ml)

CH

50 u

nits

con

sum

ptio

n (%

)

Fig. 3. Consumption of CH50 units in the presence of single PIBCAnanoparticles (mean diameter 150 nm) (♦), CNP(80 : 20) nanoparticles (meandiameter 140 nm) (▵), CNP(50 : 50) nanoparticles (mean diameter 120 nm) (●)and single PCL10k–PEG2k nanoparticles (mean diameter 80 nm) (□).

277A. Layre et al. / Journal of Controlled Release 111 (2006) 271–280

Complement consumption, as a function of the PIBCA,PCL10k–PEG2k, CNP(80 : 20) and CNP(50 : 50) nanoparticle sur-face area, is reported in Fig. 3. A dose-dependent consumptionof CH50 units was obtained in all cases. It is well establishedthat a larger amount of nanoparticles increased the contactbetween the nanoparticles and serum opsonins, leading tohigher amounts of adsorbed opsonins [36]. The PIBCAnanoparticles were clearly the best activating suspensions.Indeed, the complement consumption was faster and strongerthan with the other nanoparticle suspensions. The complementconsumption decreased when the PCL10k–PEG2k content in thenanoparticles increased. Therefore, the novel “core-shell”composite nanoparticles were able to diminish the complementconsumption as compared with the control PIBCA nanoparti-cles. For example, for a surface area of 900 cm2/ml, 100%complement consumption occurred for CNP(80 : 20), while only60% and 20% complement consumption were obtained in thecase of CNP(50 : 50) and PCL10k–PEG2k, respectively. Inconclusion, the complement consumption with the “PEGylated”nanoparticles was dependant on their average PEG surfacedensity, as already described by Jeon and Andrade and Jeon etal. [37,38]. Assuming that all the PEG chains migrate at thenanoparticles' surface, the calculated distance between twoterminally attached PEG chains at the surface of CNP(80 : 20),CNP(50 : 50) and PCL10k–PEG2k conventional nanoparticleswere 20, 13 and 11 Å, respectively (see Materials and methods).

Table 3Characteristics of freeze-dried composite nanoparticles (weight ratios from 80:20 to 5polymer : cryoprotectant mass ratios

Polymer : cryoprotectant mass ratio CNP(80 : 20) CNP(70 : 3

P188 Treh AS TE Sf /Si AS

1 :2 1 :5 − + 1.5 −1 :2 1 :10 − + 1.7 −1 :10 1 :5 − + 1.2 −1 :10 1 :10 − + 1.1 −

(AS) Aggregation scale: (−) absent; (+) scarce.(TE) Tyndall effect: (−) absent, (+) present.Sf /Si: ratio between the mean diameters, after and before lyophilization.

In Jeon's model, the minimum average distance between twoPEG chains needed for repelling proteins was estimatedaround 20 Å. So, our data are in the same range as thesetheoretical predictions. Similar distances were also found inthe case of nanoparticles prepared using diblock PLA–PEGcopolymers, to reduce complement consumption [28], plas-ma protein adsorption and uptake by polymorphonuclearcells [25].

3.6. Freeze-drying of nanoparticles

The characteristics of the freeze-dried composite nanoparti-cles (weight ratios 80 :20 to 50 :50) are presented in Table 3.The CNP preserved the Tyndall effect upon redispersionwhatever the freeze-drying conditions. The ratio between thenanoparticles' diameter after and before freeze-drying (Sf /Si) inthe presence of a mixture of Poloxamer 188 (weight ratio 1 :2)and trehalose (weight ratios 1 :5 and 1 :10) increased when thePCL10k–PEG2k proportion increased. However, the use of amixture of Poloxamer 188 (weight ratio 1 :10) and trehalose(weight ratios 1 :10 and 1 :5), whatever the nanoparticlecomposite suspension, resulted in a good redispersion with aSf /Si ratio close to one. Thus, freeze-drying could besuccessfully used to dry composite nanoparticle suspensions(weight ratios 80 :20 to 50 :50) after the addition of a mixture ofPoloxamer 188 (weight ratio 1 :10) and trehalose (weight ratio1 :10 or 1 :5).

3.7. Busulfan loaded nanoparticle characterization

Table 4 shows the mean diameter of the nanoparticles beforeand after purification (see materials and methods), and thevalues of busulfan loading into PIBCA and PCL10k–PEG2k

pure nanoparticles as well as into composite nanoparticles(weight ratios from 80 :20 to 50 : 50). In all cases, thenanoparticle size slightly decreased after purification. Asspecified in the Materials and methods section, the nanoparticlesuspensions were filtrated on the 0.45 μm filter. Indeed, thisfilter enables to remove the busulfan crystals in the suspensionmedium, but also retains some of the largest nanoparticles,which accounts for the observed size reduction after purifica-tion. The nanoparticle size was dependent on the PIBCA:PCL10k–PEG2k ratio. It increased when the PCL10k–PEG2k

decreased, except for the CNP(50 : 50). These data could be

0 :50) in the presence of poloxamer 188 (P188) and trehalose (treh) for different

0) CNP(60 : 40) CNP(50 : 50)

TE Sf /Si AS TE Sf /Si AS TE Sf /Si

+ 1.4 − + 1.5 − + 2.3+ 2.3 − + 1.6 − + 1.9+ 1.2 − + 1.3 − + 1.1+ 1.3 − + 1.2 − + 1.2

40%

50%

60%

70%

80%

90%

100%

0 100 200 300 400

time (min)

rele

ased

bus

ulfa

n (%

)

Fig. 5. Busulfan release profiles from CNP(50 : 50) at various total busulfanconcentration in the medium (busulfan both in nanoparticles and releasemedium). The experiments were performed in water at 37 °C. The busulfanconcentration in the medium was: ♦ 7 μg/ml; ▪ 13 μg/ml; ▴ 27 μg/ml; •36 μg/ml.

Table 4PIBCA, PCL10k–PEG2k pure nanoparticles and composite nanoparticles (weightratios from 80 :20 to 50 :50) characteristics: Z-average mean diameter beforeand after purification, and drug loading efficiency

Particle type Particle diameter (nm) Drug loading(% (w/w))

Before purification After purification

PIBCA 176±40 169±39 5.9±0.2CNP(80 : 20) 168±37 152±36 3.4±0.3CNP(70 : 30) 142±37 136±31 2.3±0.1CNP(60 : 40) 135±32 129±37 1.6±0.2CNP(50 : 50) 151±39 147±34 1.7±0.2PCL10k–PEG2k 85±28 87±23 0.8±0.2

Each value is the average value from six different experiments±SD.

278 A. Layre et al. / Journal of Controlled Release 111 (2006) 271–280

explained by the amphiphilic characteristics of the PCL10k–PEG2k copolymer, reducing the interfacial tension between theaqueous and organic phases.

The drug loading of the CNP was dependent on the PIBCA:PCL10k–PEG2k ratio (Table 4). For example, the drug loadingof PCL10k–PEG2k nanoparticles was only 0.8±0.2 (% (w/w))whereas it increased to 3.4±0.2 (% (w/w)) in the case of theCNP(80 : 20). The higher the amount of PIBCA in the PIBCA:PCL10k–PEG2k mixtures, the higher the drug loading. Thus,the busulfan loading of the composite nanoparticles wasincreased up to 4 folds in comparison to the drug loading of thePCL10k–PEG2k nanoparticles. Therefore, the use of PIBCA as“core” of the composite nanoparticles provided higherencapsulation efficiency. For example, the busulfan encapsu-lation efficiency of PCL10k–PEG2k nanoparticles was only4.0±0.2% whereas it was increased to 17.0±0.2% (w/w) inthe case of CNP(80 : 20).

3.8. In vitro release

The in vitro release studies were performed in water or ratplasma (Fig. 4), under “sink” conditions with the CNP(50 : 50)suspensions containing 1.7±0.2% (% (w/w)) busulfan. After afast release of 70% (water) and 78% (plasma) of the entrappeddrug during the first ten minutes, the remaining drug was

40%

50%

60%

70%

80%

90%

100%

0 50 100 150 200

time (min)

rel

ease

d bu

sulf

an (

%)

Fig. 4. Busulfan release profiles from CNP(50 : 50) containing initially 1.7±0.2%(w/w) drug were performed at 37 °C under “sink condition” in water (opensquare) or rat plasma (close diamond). Each value is derived from three differentexperiments±SD.

released slower over six hours. This profile may be explained bythe fact that busulfan is a semi-polar drug (log P=−0.59) [39],which rapidly partitions in favour of the dispersion medium,accounting for the immediate release.

To confirm this point, four additional experiments wereperformed in water at 37 °C with CNP(50 : 50) suspensions byvarying the total busulfan content in the medium (nanoparticlesand supernatant) (Fig. 5). The CNP(50 : 50) suspension containing1.7±0.2% (w/w) busulfan were appropriately diluted withwater to achieve busulfan concentration in the medium from 7to 36 μg/ml. All these concentrations are below the solubilityvalue of busulfan in water (260 μg/ml). Under these conditions,when the busulfan concentration in the medium increased, thedrug was released more slowly. Indeed, after a 2 h incubation,all the busulfan was released when the busulfan concentration inthe release medium was 7 μg/ml, whereas only 75% of busulfancame out of the nanoparticles when the concentration in therelease medium was 36 μg/ml (Fig. 5). It has been reported inthe literature that busulfan is mainly taken up by liver in the first10 to 30 min after intravenous or oral administration [40–42].We therefore expect that despite the rapid busulfan releasefrom the CNP, these carriers, potentially able to reduce liveruptake, would improve busulfan pharmacokinetics and allowthe distribution of this drug towards other organs than theliver.

4. Conclusion

The encapsulation of busulfan represents a challenge dueto the physico-chemical characteristics of this molecule whichcrystallises spontaneously in water solution. In this paper, wehave shown that novel core-shell composite nanoparticles canbe prepared, making it possible to get significant drug loadingratio values. In addition, they may combine the advantages ofthe poly(isobutylcyanoacrylate) core for its encapsulation ef-ficiency with the steric repulsive effect of the poly(ε-capro-lactone)–poly(ethylene glycol)-surface layer, preventing rapidelimination from the blood stream.

279A. Layre et al. / Journal of Controlled Release 111 (2006) 271–280

Acknowledgements

The authors would like to acknowledge the French nationalresearch center (CNRS) and Ethypharm for their financialsupport, G. Frebourg and J.P. Lechaire from the ServiceMicroscopie Electronique de l'IFR de Biologie Interactive(PARIS V) for the freeze-fracture observations, Dr. M. Chehimifrom Interfaces, Traitements, Organisation et Dynamique desSystèmes (PARIS VII) for the XPS analysis and Dr AlainChevailler from the Laboratoire d'Immunologie et Allergolo-gie, Centre Hospitalo–Universitaire d'Angers for the supply ofhuman serum.

References

[1] C.D.R. Dunn, The chemical and biological properties of busulfan(‘Myleran’), Hemat 2 (1974) 101–117.

[2] D.A.G. Galton, Myleran in chronic myeloid leukemia: results of treatment,Lancet 1 (1953) 208–213.

[3] B.R. Blazar, N.K.C. Ramsay, J.H. Kersey, W. Krivit, D.C. Arthur, A.H.Filipovich, Pretransplant conditioning with busulfan (Myleran) andcyclophosphamide for nonmalignant diseases, Transplantation 39 (1985)597–603.

[4] M. Hassan, G. Oberg, A.N. Bekassy, J. Aschan, H. Ehrsson, P. Ljungman,G. Lonnerholm, B. Smedmyr, A. Taube, I. Wallin, et al., Pharmacokineticsof high-dose busulphan in relation to age and chronopharmacology, CancerChemother. Pharmacol. 28 (2) (1991) 130–134.

[5] L.B. Grochow, R.J. Jones, R.B. Brundrett, H.G. Braine, T.L. Chen, R. Saral,G.W. Santos, O.M. Colvin, Pharmacokinetics of busulfan: correlation withveino-occlusive disease in patients undergoing bone marrow transplanta-tion, Cancer Chemother. Pharmacol. 25 (1) (1989) 55–61.

[6] H.P. Bhagwatwar, S. Phadungpojna, D.S.L. Chow, B.S. Andersson,Formulation and stability of busulfan for intravenous administration inhigh-dose chemotherapy, Cancer Chemother. Pharmacol. 37 (1996)401–408.

[7] U.S. Schuler, U.D. Renner, F. Kroschinsky, C. Johne, A. Jenke, R. Naumann,M. Bornhaüser, H.J. Deeg, G. Ehninger, Intravenous busulfan forconditioning before autologous or allogenic human blood stem celltransplantation, Br. J. Haematol. 114 (2001) 944–950.

[8] G.L.J. Kennedy, H. Sherman, Acute and subchronic toxicity ofdimethylformamide and dimethylacetamide following various routes ofadministration, Drug Chem. Toxicol. 9 (2) (1986) 147–170.

[9] P. Yellowlees, C. Greenfield, N. McIntyre, Dimethylsulphoxide-inducedtoxicity, Lancet 2 (8202) (1980) 1004–1006.

[10] Z. Hassan, C. Nilsson, M. Hassan, Liposomal busulfan: bioavailability andeffect on bone marrow in mice, Bone Marrow Transplant. 22 (1998)913–918.

[11] J. Bouligand, Développement de Nanosphères Furtives de Busulfan,Université Paris V, Paris, 2001 (master report).

[12] A.M. Layre, R. Gref, J. Richard, D. Requier, H. Chacun, M. Appel, A.J.Domb, P. Couvreur, Nanoencapsulation of a crystalline drug, Int. J. Pharm.298 (2) (2005) 323–327.

[13] A.M. Layre, P. Couvreur, H. Chacun, C. Aymes-Chodur, N.-E. Ghermani,J. Poupaert, J. Richard, D. Requier, R. Gref, Busulfan loading into poly(alkyl cyanoacrylate) nanoparticles: physico-chemistry and molecularmodelling, J Biomed Mater Res Part B (Accepted for publication).

[14] V. Lenaerts, J.F. Nagelkerke, T.J. Van Berkel, P. Couvreur, L. Grislain,M. Roland, P. Speiser, In vivo uptake of polyisobutyl cyanoacrylatenanoparticles by rat liver Kupffer, endothelial, and parenchymal cells,J. Pharm. Sci. 73 (7) (1984) 980–982.

[15] M.T. Peracchia, C. Vauthier, C. Passirani, P. Couvreur, D. Labarre,Complement consumption by poly(ethylene glycol) in different conforma-tions chemically coupled to poly(isobutyl 2-cyanoacrylate) nanoparticles,Life Sci. 61 (7) (1997) 749–761.

[16] R. Gref, Y. Minamitake, M.T. Peracchia, V. Trubetskoy, V. Torchilin,R. Langer, Biodegradable long-circulating polymeric nanospheres,Science 263 (1994) 1600–1603.

[17] S.J. Douglas, S.S. Davis, L. Illum, Biodistribution of poly(butyl 2-cya-noarcrylate) nanoparticles in rabbits, Int. J. Pharm. 34 (1–2) (1986)145–152.

[18] V. Lenaerts, P. Couvreur, D. Christiaens-Leyh, E. Joiris, M. Roland,B. Rollman, P. Speiser, Degradation of poly(isobutyl cyanoacrylate)nanoparticles, Biomaterials 5 (2) (1984) 65–68.

[19] M.T. Peracchia, C. Vauthier, F. Puisieux, P. Couvreur, Development ofsterically stabilized poly(isobutyl 2-cyanoacrylate) nanoparticles bychemical coupling of poly(ethylene glycol), J. Biomed. Mater. Res. 34(3) (1997) 317–326.

[20] J.L. Grangier, M. Puygrenier, J.C. Gautier, P. Couvreur, Nanoparticles ascarriers for growth hormone releasing factor, J. Control. Release 15 (1)(1991) 3–13.

[21] M. Simeonova, R. Velichkova, G. Ivanova, V. Enchev, I. Abrahams, Poly(butylcyanoacrylate) nanoparticles for topical delivery of 5-fluorouracil,Int. J. Pharm. 263 (1–2) (2003) 133–140.

[22] M.T. Peracchia, D. Desmaële, P. Couvreur, J. d'Angelo, Synthesis of novelpoly(MePEG cyanoacrylate–co-alkyl cyanoacrylate) amphiphilic copoly-mer for nanoparticle technology, Macromolecules 30 (4) (1997) 846–851.

[23] P. Calvo, B. Gouritin, H. Chacun, D. Desmaele, J. D'Angelo, J.P. Noel,D. Georgin, E. Fattal, J.P. Andreux, P. Couvreur, Long-circulatingPEGylated polycyanoacrylate nanoparticles as new drug carrier for braindelivery, Pharm. Res. 18 (8) (2001) 1157–1166.

[24] D. Bazile, C. Prud'homme, M.-T. Bassoullet, M. Marlard, G. Spenle-hauer, M. Veillard, Stealth Me.PEG–PLA nanoparticles avoid uptake bythe mononuclear phagocytes system, J. Pharm. Sci. 84 (4) (1995)493–498.

[25] R. Gref, M. Luck, P. Quellec, M. Marchand, E. Dellacherie, S. Harnisch,T. Blunk, R.H. Muller, ‘Stealth’ corona-core nanoparticles surfacemodified by polyethylene glycol (PEG): influences of the corona (PEGchain length and surface density) and of the core composition onphagocytic uptake and plasma protein adsorption, Colloids Surf., B. Bio-interfaces 18 (3–4) (2000) 301–313.

[26] A. Layre, R. Gref, P. Couvreur, J. Richard, Nanoparticules polymériquescomposites, FR 2872703, July 7, 2004.

[27] H.C. Fessi, J.P. Devissaguet, Process for the preparation of dispersiblecolloidal systems of a substance in the form of nanoparticles, US Patent5,118,528, June 2, 1992.

[28] M. Vittaz, D. Bazile, G. Spenlehauer, T. Verrecchia, M. Veillard, F. Puisieux,D. Labarre, Effect of PEO surface density on long-circulating PLA–PEOnanoparticles which are very low complement activators, Biomaterials 17(16) (1996) 1575–1581.

[29] M.D. Kazatchkine, G. Hauptmann, U.E. Nydegger, Technique ducomplement, INSERM (1986) 22–23.

[30] C. Passirani, G. Barratt, J.-P. Devissaguet, D. Labarre, Interactions ofnanoparticles bearing heparin or dextran covalently bound to poly(methylmethacrylate) with the complement system, Life Sci. 62 (8) (1998)775–785.

[31] C. Vauthier, C. Schimdt, P. Couvreur, Measurement of the density ofpolymeric nanoparticulate drug carriers by isopycnic centrifugation,J. Nanopart. Res. 1 (3) (1999) 411–418.

[32] J. Groll, Z. Ademovic, T. Ameringer, D. Klee, M. Moeller, Comparison ofcoatings from reactive star shaped PEG-stat-PPG prepolymers and graftedlinear PEG for biological and medical applications, Biomacromolecules 6(2) (2005) 956–962.

[33] P. Quellec, R. Gref, L. Perrin, E. Dellacherie, F. Sommer, J.M. Verbavatz,M.J. Alonso, Protein encapsulation within polyethylene glycol-coatednanospheres. I. Physicochemical characterization, J. Biomed. Mater. Res.42 (1) (1998) 45–54.

[34] E. Allemann, P. Gravel, J.C. Leroux, L. Balant, R. Gurny, Kinetics of bloodcomponent adsorption on poly(D,L-lactic acid) nanoparticles: evidence ofcomplement C3 component involvement, J. Biomed. Mater. Res. 37 (2)(1997) 229–234.

[35] C. Passirani, J.P. Benoit, in: R.I. Mahato (Ed.), Biomaterials for Deliveryand Targeting of Proteins and Nucleic Acids, 2005, pp. 187–229.

280 A. Layre et al. / Journal of Controlled Release 111 (2006) 271–280

[36] M.F. Zambaux, F. Bonneaux, R. Gref, E. Dellacherie, C. Vigneron,MPEO–PLA nanoparticles: effect of MPEO content on some of theirsurface properties, J. Biomed. Mater. Res. 44 (1) (1999) 109–115.

[37] S.I. Jeon, J.D. Andrade, Protein–surface interactions in the presence ofpolyethylene oxide: II. Effect of protein size, J. Colloid Interface Sci. 142(1) (1991) 159–166.

[38] S.I. Jeon, J.H. Lee, J.D. Andrade, P.G. De Gennes, Protein–surfaceinteractions in the presence of polyethylene oxide: I. Simplified theory,J. Colloid Interface Sci. 142 (1) (1991) 149–158.

[39] G.R. Westerhof, R.E. Ploemacher, A. Boudewijn, I. Blokland, J.H.Dillingh, A.T. McGown, J.A. Hadfield, M.J. Dawson, J.D. Down, Com-parison of different busulfan analogues for depletion of hematopoietic stemcells and promotion of donor-type chimerism in murine bone marrowtransplant recipients, Cancer Res. 60 (19) (2000) 5470–5478.

[40] M. Hassan, G. Oberg, K. Ericson, H. Ehrsson, L. Eriksson, M. Ingvar,S. Stone-Elander, J.O. Thorell, B. Smedmyr, N. Warne, et al., In vivodistribution of [11C]-busulfan in cynomolgus monkey and in the brainof a human patient, Cancer Chemother. Pharmacol. 30 (2) (1992) 81–85.

[41] J.T. Slattery, Letter, Biol. BloodMarrow Transplant. 9 (4) (2003) 282–284.[42] A. Kashyap, J.Wingard, P. Cagnoni, J. Roy, S. Tarantolo,W. Hu, K. Blume,

J. Niland, J.M. Palmer, W. Vaughan, H. Fernandez, R. Champlin, S.Forman, B.S. Andersson, Intravenous versus oral busulfan as part of abusulfan/cyclophosphamide preparative regimen for allogeneic hemato-poietic stem cell transplantation: decreased incidence of hepatic venooc-clusive disease (HVOD), HVOD-related mortality, and overall 100-daymortality, Biol. Blood Marrow Transplant. 8 (9) (2002) 493–500.