High-strength resorbable brushite bone cement with controlled drug-releasing capabilities

Available online at www.sciencedirect.com

Acta Biomaterialia 5 (2009) 43–49 www.elsevier.com/locate/actabiomat

High-strength resorbable brushite bone cement with controlled drug-releasing capabilities M.P. Hofmann a,*, A.R. Mohammed b, Y. Perrie b, U. Gbureck c, J.E. Barralet d a

Biomaterials Unit, School of Dentistry, University of Birmingham, St. Chad’s Queensway, Birmingham B4 6NN, UK b Medicines Research Unit, School of Life and Health Sciences, Aston University, Birmingham B4 7ET, UK c Department for Functional Materials in Medicine and Dentistry, University of Wuerzburg, Am Pleicherwall 2, D-97070 Wuerzburg, Germany d Faculty of Dentistry, McGill University, Montre´al, Que´bec, Canada H3A 2B2 Received 15 February 2008; received in revised form 2 July 2008; accepted 13 August 2008 Available online 26 August 2008

Abstract Brushite cements differ from apatite-forming compositions by consuming a lot of water in their setting reaction whereas apatite-forming cements consume little or no water at all. Only such cement systems that consume water during setting can theoretically produce near-zero porosity ceramics. This study aimed to produce such a brushite ceramic and investigated whether near elimination of porosity would prevent a burst release profile of incorporated antibiotics that is common to prior calcium phosphate cement delivery matrices. Through adjustment of the powder technological properties of the powder reactants, that is particle size and particle size distribution, and by adjusting citric acid concentration of the liquid phase to 800 mM, a relative porosity of as low as 11% of the brushite cement matrix could be achieved (a 60% reduction compared to previous studies), resulting in a wet unprecompacted compressive strength of 52 MPa (representing a more than 100% increase to previously reported results) with a workable setting time of 4.5 min of the cement paste. Up to 2 wt.% of vancomycin and ciprofloxacin could be incorporated into the cement system without loss of wet compressive strength. It was found that drug release rates could be controlled by the adjustable relative porosity of the cement system and burst release could be minimized and an almost linear release achieved, but the solubility of the antibiotic (vancomycin > ciprofloxacin) appeared also to be a crucial factor. Ó 2008 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Calcium phosphate cement; Brushite; Mechanical properties; Controlled drug release; Bone cement

1. Introduction Brushite-forming calcium phosphate bone cements have the advantage of being resorbable in comparison to hydroxyapatite (HA)-forming cements but suffer in application from their fast, water-consuming setting reaction and their low mechanical strength [1,2]. The maximum reported wet compressive strength value for non-compacted, i.e. hand-mixed and applied brushite-forming cements is 24 MPa [3]. Strength improvements for a brushite-forming cement up to 32 MPa could be attained by compaction during setting [4,5]. Apatite cement has been *

Corresponding author. Tel.: +44 121 2372937; fax: +44 121 2372914. E-mail address: [email protected] (M.P. Hofmann).

shown to still be workable whilst retaining high strength of up to 184 MPa for the compacted, and 67 MPa for the uncompacted system [6,7]. The inherent rapid setting of brushite cements, however, leaves insufficient time for both compacting and moulding during surgery, needed for strength improvements [8]. Comparing reported strength values for brushite-forming systems is complicated by variations in testing methods; often dry strength values are given, which are proportionally much higher than wet strength compared with HA cement systems [9], and sometimes the much lower tensile strength is measured which cannot be related to the load-bearing capabilities of the cement [10,11]. Unlike HA cements, which consume little (1 mol per 3 mol of powder reactant in b-TCP systems) or no water

1742-7061/$ - see front matter Ó 2008 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.actbio.2008.08.005

44

M.P. Hofmann et al. / Acta Biomaterialia 5 (2009) 43–49

(TTCP/DCPA systems) during setting, the brushite cement system consumes a lot water during setting reaction (up to 6 mol per 1 mol of powder reactant), theoretically allowing for the formation of cements with low or almost zero porosities. Eq. (1) shows the setting reaction for a brushite system made of b-tricalcium phosphate and monocalcium phosphate monohydrate [12]: Ca3 ðPO4 Þ2 þ CaðH2 PO4 Þ2  H2 O þ 7H2 O ! 4CaHPO4  2H2 O

ð1Þ

Calcium phosphate cements have been identified as potential drug delivery systems, although these studies mainly focused on non-resorbable hydroxyapatite-forming cements [13,14]. In a study by Bohner et al., in which a brushiteforming cement system was evaluated as a potential drug carrier, it was shown that the release of gentamicin could be controlled by adjusting the porosity of the cement, making encapsulation of the drug within polyacrylic acid unnecessary [15,16]. Recent studies showed the feasibility of drug release from brushite cements for periodontal application [17,18]. As the strength and especially the reaction kinetics (setting time) of brushite cements depend strongly on the particle size of the reactants [19], the aim of this study was to generate a high-strength, low-porosity hand-mixed brushite cement with controlled drug-releasing capabilities by adjusting particle size of the reactants. The drugs investigated were vancomycin, a highly watersoluble antibiotic used to treat severe staphylococcal infections causing osteomyelitis [20], and ciprofloxacin, a slightly water-soluble antibiotic widely used to treat bacterial bone infection [21]. 2. Materials and methods The reactants for the brushite system were phase pure sintered b-TCP, made in house as described previously [22], and commercially available monocalcium phosphate monohydrate (MCPM) powder (Rhodia, UK). The b-TCP sintercake was crushed in a mortar until it passed a 355 lm sieve and afterwards dry-milled for 1 h in a planetary ball mill (PM400 Retsch, Germany) unidirectionally at 200 rpm in 500 ml agate jars with a load of 50 g b-TCP and 4 agate balls (30 mm). For the experiments requiring a b-TCP powder with a much higher or lower particle size, commercial b-TCP (Plasma Biotal, Tideswell, UK) or further wet milled b-TCP (24 h in ethanol) was used, respectively. The MCPM powder was fractioned into four different particle size distributions (<45, 45–63, <63, >63 lm) using a 63 lm and a 45 lm sieve. Particle size distributions were determined by laser diffraction particle size analysis (Mastersizer S, Malvern, UK) of powder suspensed in pure ethanol. To avoid the generation of agglomerates the suspensions in the flow cell were exposed to ultrasound throughout the measurements.

Median particle size d50 and the span (relative width) of the particle size distribution (d90  d10)/d50 were achieved from triplicate measurements. The specific surface area of dry powder reactants was determined using nitrogen absorption (BET) at 77 K (ASAP 2000, Micromeritics, Norcross, GA). The morphology of the powders was examined using scanning electron microscopy (JEOL JSM-5300 LV, UK). Specimens of powder were sprinkled on an adhesive stub and gold-sputter-coated. Images were recorded at 20 kV acceleration voltage. To produce the cement, equimolar amounts of b-TCP and MCPM powders (1.23 g b-TCP per 1.00 g MCPM) were hand-mixed with a spatula in a weighing boat at powder to liquid ratios (PLR) varying from 2.5 to 4.0 g ml1 using 500 and 800 mM citric acid solutions as a retardant. For the drug release measurements 2 wt.% vancomycin or ciprofloxacin (Sigma Aldrich, Steinheim, Germany) were added to the cement paste. The cement slurries were cast without compaction into a PTFE mould to produce 6 mm diameter cylindrical samples with an aspect ratio of 2. After 1 h of drying at 37 °C the samples were immersed in double distilled water for another 23 h at 37 °C for the compressive strength measurements. The wet compressive strength of the samples (n P 6) was measured with a Universal testing machine (Instron 5544, UK) with a 2 kN load cell and a crosshead speed of 1 mm min1. The initial times of the cement slurries were determined by standard Gillmore needles test [23]. The strut densities of the set dried samples were determined by helium pycnometry (Accupyc, Micromeritics, UK). The average densities were calculated on the basis of 10 measurements. By measuring the apparent wet densities (by measuring the dimensions and weight of wet samples) and then calculating the dry densities (derived from weight loss measurements on wet samples until completely dried), the relative porosities (RP) were calculated with the formula RP = 1 – (dry density/strut density). The release studies were carried out on wet cylindrical samples (6  12 mm) which were extracted directly from the moulds after their final setting time (less than 10 min for all compositions used [24]) and immersed in a shaking water bath set at 37 °C. The release medium was phosphate buffered saline (PBS, 0.1 M, pH 7.4, 50 ml) containing 0.02 wt.% sodium azide resulting in a PBS volume to sample surface ratio of around 0.18 ml mm–2. One millilitre of the release media was taken out at each time interval and replenished with fresh buffer immediately. The antibiotic content of the samples (6  12 mm cylinders weighing between 0.632 and 0.763 g) was analysed by measuring absorbance with a Unicam spectrophotometer at 280 and 272 nm for vancomycin and ciprofloxacin, respectively. The amount of drug released was assayed by comparison with a calibration curve for the individual drugs made in PBS.

M.P. Hofmann et al. / Acta Biomaterialia 5 (2009) 43–49

The release rate from the samples was calculated on the basis of the final stage slope of the cumulative release curve with the assumption that the 2 wt.% antibiotic load in every 700 mg sample was released into 50 ml of release medium. To test the significance of the mean values a one-way ANOVA was performed on the raw data followed by a Tukey’s post hoc test with SPSS 10.0.0 for Windows (SPSS Inc., Chicago, USA). The level of significance p was set at p < 0.05. 3. Results and discussion 3.1. Adjusting the powder physical properties of the reactants All investigated powders had wide but logarithmically symmetrical monomodal particle size distributions (PSD) 20

Distribution density (%)

18 16

MCPM PSD 43 vol% of MCPM particles overlap with β-TCP PSD

14 12 10

β-TCP PSD

8 6 4 2 0 10

100

Particle size (μm)

Fig. 1. Illustrating the concept of overlap. Forty-three vol.% of all MCPM particles of the MCPM particle size distribution with the smallest median particle size (d50 = 32 lm) overlap with the b-TCP PSD and are therefore non-distinguishable from the b-TCP powder, thus representing a less pronounced bimodal system.

45

that were of a shape illustrated in Fig. 1. Thus the particle size distributions of all powders could be characterized by the median particle size d50 (50 vol.% of the particles smaller, 50 vol.% coarser than the median) and the span ((d90 – d10)/d50) of the PSD. Using a median b-TCP particle size of 11 lm and unfractioned MCPM (d50 = 62 lm) an initial setting time of 2.5 min was achieved at a PLR of 3.3 g ml–1 and 500 mM citric acid concentration, see Table 1. The variation of the b-TCP powder component to consist fully or even partially (25 wt.%) of smaller b-TCP powders (d50 = 3 lm) resulted in a rapid setting time of less than 1 min, making sample preparation impossible. As the use of 100 wt.% large particle sized b-TCP (d50 = 38 lm) resulted in a mixture too dry to form an easily workable paste, a median b-TCP particle size of 11 lm was therefore used throughout the study. The particle size and particle size distribution of the MCPM were adjusted by sieve fractioning. By sieving MCPM with d50 = 62 lm and a span of 1.8 through a 45 lm mesh a d50 = 32 lm and a reduced span of 0.9 could be achieved. In a similar manner a mesh of 63 lm gave a powder with d50 = 42 lm and span 1.2 and the powder retained by this mesh size had a d50 = 99 lm and span 0.9. To investigate the effect of particle size distribution width, MCPM that passed through a 63 lm mesh and was retained by a 45 lm mesh, resulting in a d50 = 54 lm and the smallest span of 0.7, see Tables 1 and 2. The compressive strength of the investigated cement system could be increased from 19 MPa for the original cement formulation made with as milled TCP and asreceived MCPM to 52 MPa for a brushite made of fractioned MCPM that was sieved to 42 lm median particle size at PLR 4.0 g ml–1 with a 800 mM citric acid solution retardant. This increase in compressive strength achieved by increasing PLR and citric acid concentration was

Table 1 Properties of brushite cements produced with unfractioned MCPM and different particle size fractions after hand-mixing and pressureless setting at different P/L ratios (n P 5) Cement made with

Molarity of retardant liquid (mM)

PLR (g ml1)

Compressive strength (MPa)

Initial setting time (min)

Apparent dry density (g cm3)

Strut density (g cm3)

Relative porosity (%)

Unfractioned MCPM (d50 = 62 lm)

500 500 800 800

2.5 3.3 3.3 4.0

18.5 ± 2.4 22.8 ± 2.7 21.4 ± 2.4 36.4 ± 3.6

4.5 ± 0.5 2.5 ± 0.5 6.5 ± 0.5 3.5 ± 0.5

1.57 ± 0.03 1.72 ± 0.03 1.73 ± 0.03 1.97 ± 0.03

2.39 ± 0.02 2.42 ± 0.02 2.37 ± 0.02 2.35 ± 0.02

34 ± 1 29 ± 1 27 ± 1 16 ± 1

Sieved <45 lm (d50 = 32 lm)

800

4.0

46.7 ± 6.6

3.0 ± 0.5

2.01 ± 0.03

2.33 ± 0.02

14 ± 1

Sieved <63 lm (d50 = 42 lm)

500 800

4.0 4.0

45.7 ± 9.8 51.6 ± 6.9

3.5 ± 0.5 4.5 ± 0.5

2.00 ± 0.03 2.02 ± 0.03

2.28 ± 0.02 2.26 ± 0.02

12 ± 1 11 ± 1

Sieved 45–63 lm (d50 = 54 lm)

800

4.0

37.7 ± 8.9

5.0 ± 0.5

1.95 ± 0.03

2.33 ± 0.02

16 ± 1

Sieved >63 lm (d50 = 99 lm)

800

4.0

20.0 ± 2.3

4.5 ± 0.5

1.89 ± 0.03

2.41 ± 0.02

22 ± 1

All cements were made with b-TCP of median particle size 11 lm. Either the standard deviation or the estimated minimum error of the method is given as errors.

46

M.P. Hofmann et al. / Acta Biomaterialia 5 (2009) 43–49

Table 2 Median particle size d50 and span, i.e. relative width (d90  d10)/d50, of the sieved and fractioned MCPM in comparison to unfractioned MCPM and the b-TCP component. The specific surface area is also given Median particle size d50 (lm)

Span of PSD

Specific surface area (m2 g1)

Overlap of MCPM-PSD with b-TCP PSD (vol.%)

MCPM fraction <45 lm (grey curve in Fig. 2)

32 ± 1

0.9 ± 0.1

0.505 ± 0.002

43 ± 2

MCPM fraction <63 lm (light grey curve Fig. 2)

42 ± 2

1.2 ± 0.1

0.485 ± 0.001

27 ± 2

MCPM fraction 45–63 lm (grey dotted curve)

54 ± 2

0.7 ± 0.1

0.485 ± 0.001

7±1

Unfractioned MCPM (right black curve) MCPM fraction >63 lm (black dotted curve)

62 ± 2 99 ± 2

1.8 ± 0.1 0.9 ± 0.1

0.464 ± 0.001 0.437 ± 0.001

20 ± 2 4±1

b-TCP (left black curve)

11 ± 1

1.4 ± 0.2

0.661 ± 0.001

N/A

0.7) and one of the smallest overlaps (7%), thus representing the system with the most pronounced bimodal character, see Table 2. A median particle size of 42 lm, however, seemed to be near the optimum compromise of low particle size (causing higher strength) and setting time (workability), having the highest strength and second highest setting time of all PLR 4.0 g ml–1 systems. Geometrically the specific surface area of a powder composed of spherical monosized particles with the same particle shape doubles when the particle size is halved. However, when the MCPM median particle size was reduced from 62 to 32 lm (MCPM 45–63 lm fraction compared to the <63 lm fraction), the specific surface area increased from 0.46 m2 g–1 to just 0.51 m2 g–1, Table 2. This indicated that the MCPM particles contained internal porosity, which was corroborated by scanning electron microscopy, see Fig. 3. That also suggested that the effect of MCPM particle size variation must therefore be mainly due to a ‘‘liquefying” effect of the bimodal distribution and not due to the changes in dissolution kinetics. Takahashi et al. reported the validity of a model where strength of a set cement matrix is dependent on the porosity of the system and the number of flaws within the cement matrix and proved that there is an inverse linear relationship between the natural logarithm of compressive strength (CS) and relative porosity (RP) [26], a relationship later

Distribution density (%)

associated with a pronounced and significant (p < 0.05) decrease in relative porosity from 34% to 11%, without a decrease in initial setting time and therefore the working time. Interestingly, when comparing cement compositions at the highest PLR of 4.0 g ml–1 and 800 mM citric acid, led MCPM particle size reduction alone from a d50 of 62 or 99 lm to a d50 of 42 lm to significant (p < 0.05) decreases in porosity from 16% or 22% to 11%, significant (p < 0.05) increases in apparent dry density from 1.95–1.89 to 2.02 g cm–3 and also a significant (p < 0.05) increase in strength from 38/20 to 52 MPa (Table 1). With a further reduction of particle size to a d50 of 32 lm at the highest PLR and citric acid concentration, strength decreased noticeably (but not significantly) from 52 to 47 MPa and setting time decreased significantly (p < 0.05) from 4.5 to 3.0 min. The particle size distributions (PSD) of the sieved and fractioned MCPM, the unsieved MCPM and, as comparison, the b-TCP component are shown in Fig. 2. Median particle sizes d50 and spans, i.e. relative width of PSD (d90 – d10)/d50, specific surface area of the powders and also the overlap, that is the volume percentage of MCPM particles having a size similar to the particle size covered by the b-TCP PSD (see Fig. 1), are given in Table 2. The b-TCP-MCPM powder mixture had a pronounced bimodal character as indicated by an overlap of b-TCP and MCPM-PSD between 4% and 27% for MCPM powders of 42 lm and larger, see Fig. 2 and Table 2. Bimodal particle size distributions have been shown to decrease water demand in a single powder component HA-forming system where the addition of an inert filler of much smaller particle size enabled higher workable PLR [25]. In the case of the two-component brushite-forming system reported here, the smaller b-TCP intrinsically acted as a reactive filler for the larger MCPM, leading to an increase in setting time and improved strength. However, when the MCPM particle size was reduced to 32 lm, resulting in an overlap of the b-TCP and MCPM PSD of 43%, the system was no longer strictly bimodal and displayed decreased strength, workability and setting time when compared to optimum particle size of 42 lm, where the overlap was only 27%. Maximum initial setting time of 5.0 min for a highstrength brushite cement (PLR 4.0 g ml–1) was achieved for a cement paste made with the median particle size of the MCPM of 54 lm and the narrowest distribution (span

28 26 24 22 20 18 16 14 12 10 8 6 4 2 0

from left to right: β-TCP MCPM < 45 μm MCPM < 63 μm MCPM 45-63 μm MC PM unfractioned MCPM >63 μm

10

100

Particle size (μm) Fig. 2. Particle size distributions of unsieved and fractioned (sieved) MCPM and b-TCP.

M.P. Hofmann et al. / Acta Biomaterialia 5 (2009) 43–49

proven to be valid for high-strength hydroxyapatite cements [7]: lnðCSÞ / RP

ð2Þ

In this study this relationship was confirmed for all PLRs and all MCPM particle sizes except the MCPM with the largest median particle size of 99 lm as illustrated in Fig. 4. The established relationship indicated that the highest theoretically achievable wet strength for brushite cement was 83 MPa (at zero porosity), which is higher than the reported strength for spongeous bone in humid conditions (13 MPa [27]) and almost in the range of cortical bone (90–209 MPa [28]). The departure from the linear porosity–ln(strength) relationship for the largest MCPM particle size can be explained by changes in the degree of reaction leading to lower strength. The strut densities of all but this PLR 4.0 g ml–1 system, see Table 2, were in the brushite specific

47

region of 2.3 g cm–3, indicating an almost complete conversion of the cement powder mixture (q = 2.71 g cm–3) to brushite. This corresponded with results from an earlier study on the cement system where Rietveld analysis of XRD pattern showed that the use of 500 or 800 mM citric acid as a setting retardant led to a degree of conversion of more than 90% [3]. However, the higher strut density of the system with the largest MCPM particle size (q = 2.41 g cm–3) indicated either a lower degree of conversion of the reactants or the generation of dicalcium phosphate anhydrous, DCPA, (q  2.9 g cm–3) instead of brushite, with both mechanisms known to increase porosity and decrease strength [3,29–31]. Previous reported strength for resorbable brushite systems are 24 MPa for a non-precompacted cement [3] and 49 MPa with a clinically unachievable 9 MPa precompaction [5]. The achieved wet strength of 52 MPa thereby represents the highest ever reported wet strength for a resorbable brushite cement (compacted or uncompacted) and is comparable to even the highest reported strength for an unprecompacted non-resorbable HA-forming cement, 79 MPa, achieved by using a filler to achieve bimodality of the cement powder mixture [25]. 3.2. Drug release profiles

Fig. 3. High-magnification scanning electron micrograph of MCPM illustrating that the specific surface area of the MCPM stems from its surface porosity.

4.2

MCPM d50 = 99 μm all other MCPM d50 ln(CS) = 4.42 - 0.046 ∗ RP,

ln (compressive strength)

4.0 3.8

2

R = 0.97, p < 0.0001

3.6 3.4 3.2 3.0 2.8 2.6 0

5

10

15

20

25

30

35

Relative porosity (%) Fig. 4. Correlation between the porosity of the brushite cement formulations and the natural logarithm of wet compressive strength.

The cumulative release of two antibiotics, vancomycin and ciprofloxacin, from brushite cement systems of PLR 2.5 and 4.0 g ml–1 (2 wt.% antibiotic addition to the powder phase of the cement) is shown in Fig. 5a and b. For cement containing 2 wt.% vancomycin, 80% of the drug was released within 24 h for a PLR 2.5 g ml–1 system (relative porosity 34%), whereas only 60% were released in 24 h for PLR 4.0 g ml–1 (relative porosity 16%), Fig. 5a. For cement with 2 wt.% ciprofloxacin also 60% of the drug were released for PLR 2.5 g ml–1 within 24 h. However, for the low-porosity PLR 4.0 g ml–1 cement less than 30% of the drug were released in 24 h with another 30% being released almost linear over the rest of the 2 weeks, indicating that a burst release could be minimized by the lowporosity system, Fig. 5b. These findings corresponded to a study by Bohner et al. on a gentamicin-releasing brushite cement system where drug release rate could be trebled by increasing porosity from 38% to 69%. However, a linear release profile and a minimizing of initial burst release could not be achieved in this study, probably due to the high minimum porosity of 38% achievable for that cement system [16]. In another study by Tamimi et al. on doxycycline release from a brushite cement burst release in the first 12 h was also evident, possibly also due to the low PLRs of 1.75–2.5 g ml–1, used causing a high porosity of the set cement [17]. The release profile of the two antibiotics from the cements suggested that only the low-porosity cement matrix could provide a sustained linear release. However, the results also indicated that the controlling factors for

48

Cumulative drug release / %

a

M.P. Hofmann et al. / Acta Biomaterialia 5 (2009) 43–49

100

PLR 2.5g/ml

PLR 4.0g/ml

80

60

40

20

0 0

20

40

60

80

100

Time / h

Cumulative drug release / %

b

100

PLR 2.5g/ml

PLR 4.0 g/ml

80

60

40

20

0 0

50

100

150

200

250

300

350

Time / h Fig. 5. (a) Cumulative release of vancomycin (2 wt.% addition to the paste) for up to 96 h from brushite cement for two different PLRs. (b) Release of ciprofloxacin (2 wt.% addition) for up to 14 days (336 h). BSpline fits are included to illustrate the overall trend in drug release.

the release profile were both the solubility of the drug candidate, with vancomycin having a higher solubility than ciprofloxacin, and the porosity of the matrix. Cements with a PLR of 4.0 g ml–1 provided a compact and less porous matrix as indicated by the decrease in the relative porosity to 16% compared with 34% for PLR 2.5 g ml–1 (standard system with unmodified MCPM), possibly decreasing the volume of release media which can enter the matrix. This trend was evident for ciprofloxacin, where cumulative release was lower than 50% for the low-porosity system compared with the high-porosity cement for up to 7 days, Fig. 5b. For vancomycin this trend was less pronounced but still noticeable for up to 4 days, Fig. 5a. The drug release rate for the low-porosity and high-porosity vancomycin-containing system was at least 1 mg l–1 h–1 (milligram per litre and hour) up to 1 day and 4 days, respectively, thus releasing the minimum inhibitory concentration (MIC) of vancomycin for MRSA (methicillinresistant Staphylococcus aureus) every 4 h [32]. For the lower solubility ciprofloxacin the drug release rate was at

least 0.25 mg l–1 h–1, releasing the MIC for MRSA every 2 h [33], up to 2 days for the low PLR and up to 14 days for the high PLR cement. However, the local concentration of drugs is likely to depend on the unknown in vivo replenishing rate of body fluid in and around the cement implant side. Around 15% of the antibiotic load in the cement was not released in the low PLR system during the release study; this fraction is thought to be released slowly over larger time periods or upon degradation of the cement structure. The water solubility of these drugs seemed to be a crucial factor and must be considered too. Vancomycin is around 10 times more water-soluble than ciprofloxacin and the release profile showed that higher concentrations of vancomycin were released compared with ciprofloxacin for both investigated PLRs (2.5 and 4.0 g ml–1). This suggested that water solubility also dictated drug release from a matrix system as is the case for many, but not all, matrix systems. Also, electrostatic interactions between the drug and the matrix components may influence the release profile [34]. The acidity of the citric-acid-containing cement paste is thought to have no effect on the antibacterial activity of the antibiotics, with vancomycin even showing an increased antibacterial activity at acidic pH values between 3 and 6 [20]. Importantly, the addition of up to 2 wt.% ciprofloxacin to the cement paste had no significant effect on the wet compressive strength and also on the setting time (not shown) of the standard brushite cement system (unmodified MCPM) for lower (2.5 g ml–1) and higher PLR (4.0 g ml–1), as wet compressive strength was 17 and 37 MPa for the drug containing high- and low-porosity systems compared to 19 and 36 MPa for the non-drug containing standard system, thus indicating that the incorporation of the drug into the cement matrix did not generate flaws or cause significant interaction with the setting reaction which would have weakened the cement structure. 4. Conclusions By adjusting particle size and size distribution of the bTCP and the MCPM powder reactants and also the citric acid concentration the PLR could be raised to as high as 4.0 g ml–1, resulting in a decrease of the cement porosity to 11%, giving an unprecedented non-precompacted wet compressive strength of 52 MPa. It was further demonstrated that drug release rates could be controlled by minimizing the relative porosity of the cement system. The drug release of ciprofloxacin could be increased from 2 to 14 days by using the near-zero porosity system with the MIC for MRSA being released every 2 h. Thus we have demonstrated the successful creation of a hand-mixable resorbable inorganic matrix capable of sustained release of therapeutic compounds that has mechanical properties exceeding commercial and even

M.P. Hofmann et al. / Acta Biomaterialia 5 (2009) 43–49

reported experimental calcium phosphate cements. This will enable greater application of this biomaterial in orthopedic and craniofacial applications. Acknowledgements We acknowledge the financial support of the EU by the FP 6 Mobility 5 Marie Curie Fellowship 500694 (M.P. Hofmann) and the financial support of the Canada Research Chair program (J.E. Barralet). The authors also thank Pamela Habibovic for her helpful comments. References [1] Bohner M, Gbureck U, Barralet JE. Technological issues for the development of more efficient calcium phosphate bone cements: a critical assessment. Biomaterials 2005;26:6423–9. [2] Bohner M. Calcium orthophosphates in medicine: from ceramics to calcium phosphate cements. Injury 2000;31:37–47. [3] Hofmann MP, Young AM, Gbureck U, Nazhat SN, Barralet JE. FTIR-monitoring of a fast setting brushite bone cement: effect of intermediate phases. J Mater Chem 2006;16:3199–206. [4] Grover LM, Gbureck U, Hutton A, Farrar DF, Ansell C, Barralet JE. Low porosity CaHPO42H2O cement. Key Eng Mater 2004;254– 256:205–8. [5] Gbureck U, Dembski S, Thull R, Barralet JE. Factors influencing calcium phosphate shelf-life. Biomaterials 2005;26:3691–7. [6] Gbureck U, Barralet JE, Spatz K, Grover LM, Thull R. Ionic modification of calcium phosphate cement viscosity. Part I: hypodermic injection and strength improvement of apatite cement. Biomaterials 2004;25:2187–95. [7] Barralet JE, Hofmann M, Grover LM, Gbureck U. High-strength apatitic cement by modification with a-hydroxy acid salts. Adv Mater 2003;15(24):2091–4. [8] Barralet JE, Grover LM, Gbureck U. Ionic modification of calcium phosphate cement viscosity. Part II: hypodermic injection and strength improvement of brushite cement. Biomaterials 2004;25:2197–203. [9] Pittet C, Lemaitre J. Mechanical characterisation of brushite cements: a Mohr circles approach. J Biomed Mater Res 2000;53:769–80. [10] Bohner M, Merkle HP, van Landuyt P, Trophardy G, Lemaitre J. Effect of several additives and their admixtures on the physicochemical properties of a calcium phosphate cement. J Mater Sci Mater Med 2000;11:111–6. [11] Charriere E, Terrazzone S, Pittet C, Mordasini P, Dutoit M, Lemaitre J, et al. Mechanical characterization of brushite and hydroxyapatite. Biomaterials 2001;21:2937–45. [12] Mirtchi AA, Lemaitre J, Terao N. Calcium phosphate cements: study of the beta-tricalcium phosphate–monocalcium phosphate system. Biomaterials 1989;10:475–80. [13] Ginebra MP, Traykova T, Planell JA. Calcium phosphate cements as bone drug delivery systems: a review. J Controlled Release 2006;113:102–10. [14] Ginebra MP, Traykova T, Planell JA. Calcium phosphate cements: competitive drug carriers for the musculoskeletal system? Biomaterials 2006;27:2171–7.

49

[15] Bohner M, Lemaitre J, Merkle HP, Gander B. Control of gentamicin release from a calcium phosphate cement by admixed poly(acrylic acid). J Pharm Sci 2000;89:1262–70. [16] Bohner M, Lemaitre J, van Landuyt P, Zambelli PY, Merkle HP, Gander B. Gentamicin-loaded calcium phosphate bone cement as antibiotic delivery system. J Pharm Sci 1997;86:565–72. [17] Tamimi F, Torres J, Bettini R, Ruggera F, Rueda C, Lopez-Ponce M, et al. Doxycycline sustained release from brushite cements for the treatment of periodontal diseases. J Biomed Mater Res A 2008;85:707–14. [18] Young AM, Ng PYJ, Gbureck U, Nazhat SN, Barralet JE, Hofmann MP. Characterization of chlorohexidine-releasing, fast setting, brushite bone cement. Acta Biomater 2008;4:1081–8. [19] Bohner M. Reactivity of calcium phosphate cements. J Mater Chem 2007;17:3980–6. [20] Armstrong DW, Rundlett KL, Chen JR. Evaluation of the macrocyclic antibiotic vancomycin as a chiral selector for capillary electrophoresis. Chirality 1994;6:496–509. [21] Castro C, Evora C, Baro M, Soriano I, Sanchez E. Two-month ciprofloxacin implants for multibacterial bone infections. Eur J Pharm Biopharm 2005;60:401–6. [22] Gbureck U, Grolms O, Barralet JE, Grover LM, Thull R. Mechanical activation of b-TCP. Biomaterials 2003;24:4123–31. [23] ASTM-Standard C266-99: standard test method for time of setting of hydraulic cement paste by Gilmore needles. ASTM International; 2002. [24] Hofmann MP, Nazhat SN, Gbureck U, Barralet JE. Real-time monitoring of the setting reaction of brushite-forming cement using isothermal differential scanning calorimetry. J Biomed Mater Res B Appl Biomater 2006;79:360–4. [25] Gbureck U, Spatz K, Thull R, Barralet JE. Rheological enhancement of mechanically activated alpha-tricalcium phosphate cements. J Biomed Mater Res B Appl Biomater 2005;73:1–6. [26] Takahashi T, Yamamoto M, Ioku K, Goto S. Relationship between compressive strength and pore structure of hardened cement pastes. Adv Cem Res 1997;9:25–30. [27] Tamimi F, Kumarasami B, Doillon C, Gbureck U, Le Nihouannen D, Lopez-Cabarcos E, et al. Brushite-collagen composites for bone regeneration. Acta Biomater 2008;4:1315–21. [28] Reilly DT, Burstein AH. The elastic and ultimate properties of compact bone tissue. J Biomech 1975;8:393–405. [29] Sarda S, Fernandez E, Nilsson M, Balcells M, Planell JA. Kinetic study of citric acid influence on calcium phosphate bone cements as water-reducing agent. J Biomed Mater Res 2002;61:653–9. [30] Bohner M, van Landuyt P, Merkle HP, Lemaitre J. Composition effects on the pH of a hydraulic calcium phosphate cement. J Mater Sci Mater Med 1997;8:675–81. [31] Ginebra MP, Driessens FCM, Planell JA. Effect of the particle size on the micro and nanostructural features of a calcium phosphate cement: a kinetic analysis. Biomaterials 2004;25:3453–62. [32] Assadullah S, Kakru DK, Thoker MA, Bhat FA, Hussain N, Shah A. Emergence of low-level vancomycin resistance in MRSA. Indian J Med Microbiol 2003;21:196–8. [33] Al-Nawas B, Shah PM. Intracellular activity of ciprofloxacin and moxifloxacin, a new 8-methoxyquinolone, against methicillin-resistant Staphylococcus aureus. J Antimicrob Chemother 1998;41:655–8. [34] Boomareddy GS, Paker-Leggs S, Saripella KK, Neau SH. Extruded and spheronized beads containing carbopolo(R) 974P to deliver nonelectrolytes and salts of weakly basic drugs. Int J Pharm 2006;321:62–71.