An integrated assessment of morphology, size, and complement activation of the PEGylated liposomal doxorubicin products Doxil®, Caelyx®, DOXOrubicin, and SinaDoxosome

Journal of Controlled Release 221 (2016) 1–8

Contents lists available at ScienceDirect

Journal of Controlled Release journal homepage: www.elsevier.com/locate/jconrel

An integrated assessment of morphology, size, and complement activation of the PEGylated liposomal doxorubicin products Doxil®, Caelyx®, DOXOrubicin, and SinaDoxosome Peter P. Wibroe a, Davoud Ahmadvand b, Mohammad Ali Oghabian c, Anan Yaghmur d, S. Moein Moghimi a,e,⁎ a Nanomedicine Laboratory, Centre for Pharmaceutical Nanotechnology and Nanotoxicology, Department of Pharmacy, Faculty of Health and Medical Sciences, University of Copenhagen, Universitetsparken 2, DK-2100 Copenhagen Ø, Denmark b Department of Medical Laboratory Sciences, School of Allied Medical Sciences, Iran University of Medical Sciences, Tehran, Iran c Department of Medical Physics and Biomedical Engineering, Faculty of Medicine, Tehran University of Medical Sciences, Tehran, Iran d Department of Pharmacy, Faculty of Health and Medical Sciences, University of Copenhagen, Universitetsparken 2, DK-2100 Copenhagen Ø, Denmark e NanoScience Centre, University of Copenhagen, Universitetsparken 5, DK-2100 Copenhagen Ø, Denmark

a r t i c l e

i n f o

Article history: Received 3 October 2015 Received in revised form 12 November 2015 Accepted 18 November 2015 Available online xxxx Keywords: Complement system Cryo-TEM Nanomedicine Nanosimilars Non-biological complex drugs

a b s t r a c t In order to improve patient's benefit and safety, comprehensive regulatory guidelines on specificities of NonBiological Complex Drugs (NBCDs), such as doxorubicin-encapsulated liposomes, and their follow-on versions are needed. Here, we compare Doxil® and its European analog Caelyx® with the two follow-on products DOXOrubicin (approved by the US Food and Drug Administration) and SinaDoxosome (produced in Iran) by cryogenic transmission electron microscopy, dynamic light scattering and Nanoparticle Tracking Analysis, and assess their potential in activating the complement system in human sera. We found subtle physicochemical differences between the tested liposomal products and even between the tested batches of Doxil® and Caelyx®. Notably, these included differences in vesicular population aspect ratios and particle number. Among the tested products, only SinaDoxosome, in addition to the presence of unilamellar vesicles with entrapped doxorubicin crystals, contained empty circular disks. Differences were also found in complement responses, which may be related to some morphological differences. This study has demonstrated an integrated biophysical and immunological toolbox for improved analysis and detection of physical differences among vesicular populations that may modulate their clinical performance. Combined, these approaches may help better product selection for infusion to the patients as well as for improved design and characterization of future vesicular NBCDs with enhanced clinical performance and safety. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Doxil® (or Caelyx® in Canada and Europe) is an established liposomal formulation of doxorubicin originally approved by the US Food and Drug Administration (FDA) for human use in 1995 [1]. It became the first nanomedicine product on the market, and has been a flagship for the development of new nanopharmaceuticals. Doxil®’s patent expired in the US in 2010, but not until 2013 came the first follow-on product Lipodox (DOXOrubicin) on the market. Today, there are also nationally approved follow-on products such as SinaDoxosome in Iran. Together with a large number of emerging nanopharmaceuticals, this has raised dilemmas on what regulatory criteria should be used for assessing and approving future non-biological complex drugs (NBCDs) and their ⁎ Corresponding author at: Nanomedicine Laboratory, Centre for Pharmaceutical Nanotechnology and Nanotoxicology, Department of Pharmacy, Faculty of Health and Medical Sciences, University of Copenhagen, Denmark. E-mail address: [email protected] (S.M. Moghimi).

http://dx.doi.org/10.1016/j.jconrel.2015.11.021 0168-3659/© 2015 Elsevier B.V. All rights reserved.

follow-on products [2–5]. There is a comprehensive regulatory framework for small molecule generics, based on pharmaceutical equivalence and bioequivalence [6,7]. Analytical techniques can be applied to validate identical structure to a reference compound, where preclinical and clinical trials generally are not needed. Biological compounds and their follow-on versions (biosimilars) have, because of their complexity and the difficulties of exact reproduction and characterization, received their own regulatory frameworks [8]. Here, ‘similar’ is a consensus terminology rather than ‘equivalent’ and ‘identical’, and hence the term ‘biosimilars’. It has recently been suggested that NBCDs and their follow-on products (sometimes referred to as ‘nanosimilars’) should mimic the regulatory processes of biologicals, because of their equally complex nature [9,10]. The biological performance of anti-cancer nanopharmaceuticals such as their circulation times, biodistribution (including the extent of extravasation from the blood and retention at solid tumors) and adverse responses (e.g., infusion-related adverse reactions) are often controlled by a complex array of interrelated physicochemical factors such as

2

P.P. Wibroe et al. / Journal of Controlled Release 221 (2016) 1–8

particulate size, shape, number and surface chemistry [11–13]. Small differences in nanopharmaceutical production, however, may lead to subtle changes in these parameters. Consequentially, this may generate different subpopulations of particles within a typical batch, thereby modulating the overall biological performance of the product [12]. The complex nature of nanoparticle engineering together with the need for exact reproducibility puts strong requirements not only on the instrumental setup, but also on the quality and batch variation of the raw materials. In addition, it is difficult to unambiguously evaluate parameters such as nanoparticle size distribution, morphology and aspect ratios, surface density and conformation of projected polymers and drug crystallization [14]. Thus, with difficulties in ensuring exact reproducibility in nanoparticle production as well as challenges in identification and characterization of the vital physical and chemical parameters, the regulatory authorities face a challenge in defining tests and requirements for approving future follow-on products and in identifying the required extent of similarity [2,15]. With respect to adverse reactions, inadvertent activation of the complement system is believed to be a causal factor for liposomal-mediated infusion-related reactions in human subjects [16–19]. Minor differences in liposome size and surface representation can incite complement differently, for instance through binding of anti-phospholipid and anticholesterol antibodies, and C3b as well as a plethora of pattern recognition molecules including C1q, mannose binding lectin, ficolins, collectin 11 and properdin [16,20]. Further complexity may emerge from the presence of yet undefined complement activating aggregated contaminants or unencapsulated free drug crystals that could elicit cardiopulmonary distress [21]. It is also plausible that infusion-related reactions to some extent are related to a particular population of vesicles with distinct physicochemical parameters. In addition to size and surface characteristics, shape also affects the biological performance of nanoparticulate matters [12,22,23]. This includes the flow properties of nanoparticles within the blood vessels and at bifurcations in vascular and capillary systems, nanoparticle extravasation from the blood and accumulation at the pathological sites as well as the mode and kinetics of nanoparticle macrophage recognition and associated responses (e.g., release of proinflammatory cytokines and modulators) [24–26]. A recent study, with the aid of small angle X-ray scattering (SAXS), studied and compared the membrane electron density, the thickness and density of poly(ethylene glycol) (PEG) layers and the structure of entrapped doxorubicin between certain batches of Doxil®, Caelyx® and Lipodox, and concluded that these products were structurally similar [27]. SAXS, however, is not a sensitive method for assessing vesicular shape. Accordingly, here we compare shape and size distribution characteristics of a number of liposomal doxorubicin products by different and complementary modalities, and assess vesicular-mediated complement activation in sera of human subjects. We highlight subtle physicochemical differences among the available batches of tested products as well as complement responses, and discuss these observations in relation to regulatory “non-binding” recommendations, which are in place for evaluation and production of Doxil® follow-on products [28]. 2. Methods 2.1. Liposomal products Four versions of liposomal doxorubicin were tested and our study is limited to a few batches available to us. These included Caelyx® (European trademark of Doxil®, Janssen-Cilag, BHZ0V00), Doxil® (NDC#59676-960-01, lot 1211158), “DOXOrubicin hydrochloride liposome” (DOXOrubicin, NDC# 47335-049-40, lot JKM0622A, Sun Pharma Global FZE, India), and an Iranian product SinaDoxosome [Iran Registration Code (IRC): 1228184499, lots 9005 and DO00209307, Exir Nano Sina Co., Iran]. Samples of Doxil® and

DOXOrubicin were kindly supplied by the Nanotechnology Characterization Laboratory (Frederick, MD, USA). Caelyx® was from a local pharmacy in Budapest and provided by Dr. Janos Szebeni (Semmelweis University, Budapest, Hungary). Throughout, all liposomal products were handled, shipped and stored in accordance with the manufacturer's recommendations. The manufacturer's stated doxorubicin HCl concentration in all formulations was 2 mg/mL. Phospholipid and doxorubicin concentrations of the products were not evaluated further. Control empty liposomes resembling Doxil® consisting of a mixture of N-(carbonyl-methoxypolyethylene glycol 2000)-1,2-distearoyl-snglycero-3-phosphoethanolamine sodium salt (Avanti Polar Lipids, AL, USA), fully hydrogenated soy phosphatidylcholine (Avanti Polar Lipids, AL, USA), and cholesterol (Sigma, Poole, UK) was prepared in a weight ratio of 1:3:1 as previously reported [29]. Briefly, the lipid mixture was dissolved in a chloroform/methanol (2:1 v/v) solution, dried to form a thin dry lipid film by rotary evaporation and desiccated overnight. The dry lipid film was then rehydrated with 300 mM sucrose in sterile water for injection, warmed to 60 °C and lyophilized overnight. The dried lipid–sucrose mixture was rehydrated with 240 mM ammonium sulfate in sterile water for injection and subjected to 6 freeze–thaw cycles at 60 °C followed by extrusion through a series of polycarbonate filters (400, 200 and 100 nm). The extruded liposome solution was stored at 4 °C.

2.2. Liposome morphology and size determinations Vesicular morphology was assessed by cryogenic transmission electron microscopy (cryo-TEM). The liposomal products were diluted 5× in HEPES buffer (133 mM NaCl, 4.5 mM KCl, 10 mM HEPES, pH 7.4) and vitrified using an automated FEI Vitrobot Mark 5. Blotting parameters (e.g., blotting speed) were kept constant for all samples, and similar blotting qualities (vitrification, film thickness) were observed between all samples and blotting replicates. Images were acquired with a FEI Tecnai G2 200 kV cryo-TEM microscope and a FEI 4x4k CCD Eagle camera (FEI, Eindhoven, The Netherlands) at 29,000× and 50,000× magnification. Particle identification and size/morphology determination were performed by fully automated image processing using Matlab 2014b (Mathworks, Massachusetts, US) applying functions of the Image Processing Toolbox. Briefly, a series of random images of the same magnification was imported and adjusted by adaptive contrast enhancement, from which a smoothened binary representation with no holes or objects on the edges was constructed. Closely overlapped objects were separated by watershed segmentation using an eroded binary image to identify individual objects. The identified objects were analyzed for major and minor axes, circumference, area and equivalent spherical diameter using the ‘regionprops’ function. The minor and major axes for each object were used to calculate the expected ellipsoidal circumference and area, which were compared with the measured area and perimeter. Only objects with correlation between calculated and measured geometries were included. All images were batchprocessed with identical thresholds and settings, accumulating over 5 to 9 images for each sample. Dynamic light scattering (DLS) was performed on samples diluted 100-fold in 10 mM NaCl and measured on a Brookhaven ZetaPALS (Brookhaven Instruments Corporation, NY, USA) at 25 °C and 90° scattering angle. Zeta potential measurements were done in 10 mM NaCl. Nanoparticle Tracking Analysis (NTA) was also used for vesicular size and concentration evaluation. Briefly, samples were diluted 106 times with purified water (18.2 MΩ cm) and monitored with an LM20 NanoSight mounted with a blue (405 nm) laser (Malvern Instruments, UK) using the Nanosight 2.3 software for analysis [30,31]. Particle size and scattered light intensity of individual particles were recorded and pooled from at least five different videos.

P.P. Wibroe et al. / Journal of Controlled Release 221 (2016) 1–8

2.3. Small angle X-ray scattering (SAXS) X-ray measurements on liposomal products were performed at the beamline I911-SAXS (MAX II storage ring, MAX-lab synchrotron facility, Lund University, Sweden) at operating electron energy of 1.5 GeV and a wavelength of 0.91 Å. The scattering patterns were recorded with Pilatus 1 M (Dectris Ltd., Baden, Switzerland) using collection times of 120–240 s. The camera was kept under vacuum during data collection to minimize background scattering. The detector covered the q-range (q = 4π sinθ/λ, where λ is the wavelength and 2θ is the scattering angle) of interest from about 0.01 to 5.0 nm−1. Silver behenate [CH3-(CH2)20-COOAg with a d spacing value of 5.84 nm] was used as a standard to calibrate the angular scale of the measured intensity. The samples were measured in custom-made glass capillaries at 25 °C (±0.1 °C) by the aid of a Peltier element. 2.4. Complement activation in human serum Human blood was collected from healthy donors and serum was prepared and stored at − 80 °C. Complement activation was initiated by mixing the liposomal products with undiluted serum to reach 80% v/v serum and a doxorubicin concentration of 0.4 mg/mL. Sterile, endotoxin-free PBS (Sigma, UK) was used to substitute for the liposomal products for background measurements. As a positive control, zymosan (Sigma, UK) was introduced at 0.1 mg/mL to assess the overall

3

functionality of each donor's complement system. After incubation for 30 min at 37 °C, the complement activity was stopped by diluting serum with a cold stop solution supplied with the ELISA kit containing 25 mM EDTA. Soluble activation products remaining in the serum were quantified with sandwich ELISA kits targeting neo-antigens on cleaved complement proteins C4d, Bb and C5a as well as the complement activation product SC5b-9 (Quidel, USA) according to the manufacturer's protocol and as described in detail elsewhere [32–34]. A multiple comparison two-way ANOVA was applied to test for significant differences in the responses from the highest-activating product of each marker with the remaining product-induced responses. 3. Results and discussion 3.1. Morphological assessment Liposome morphology was assessed by cryo-TEM. The main advantage of this modality is the preservation of vesicles, and other structures that may co-exist, and the direct visualization of morphology and presence of other impurities and abnormalities [35]. The images in Fig. 1 show similarities and differences among the product batches. Doxil®, Caelyx® and DOXOrubicin micrographs show frequent presence of a wide range of intact spherical and prolate ellipsoidal shaped unilamellar vesicles, encapsulating electron dense needle-shaped structures, putatively representing stacked doxorubicin molecules [36,37] (Fig. 1A–F).

Fig. 1. Cryo-TEM analysis of four liposomal doxorubicin formulations. Cryo-TEM images of Doxil® (A, B), Caelyx® (C, D), DOXOrubicin (E, F) and SinaDoxosome (G, H). Scale bars: 200 nm. Black arrows indicate empty liposomes (D), an oligolamellar vesicle (E) and disks (G). White arrows represent face on view of disks (G, H). I–L: Liposome size distribution from image processing. Equivalent diameter corresponds to the diameter of a circle occupying the same area as the detected vesicle, but this does not consider contribution by surface projected mPEG chains.

4

P.P. Wibroe et al. / Journal of Controlled Release 221 (2016) 1–8

Some vesicles were elongated accommodating longer electron dense needles, but we did not observe vesicles containing doxorubicin bundles in other morphologies such as curved, U-shaped or circular as reported elsewhere [38]. For SinaDoxosome, cryo-TEM images (Fig. 1G– H), revealed co-existence of flat circular disks with unilamellar vesicles [35,39]. Depending on the disk orientation, they appear either as small rods with high contrast (edge on view) or as circular structures with low uniform contrast (face on view); these may represent disk-shaped mixed micelles [35]. These observations are only indicative of the tested samples and by no means a global representation of the SinaDoxosome product. Accordingly, our analysis may suggest differences in synthesis parameters or in storage conditions, or in thermal pre-treatment (prior to our purchase and handling) of SinaDoxosome. The presence of unilamellar vesicles in SinaDoxosome, however, was confirmed by synchrotron SAXS and compared with Caelyx®, where the patterns were similar to recent findings [27] and mainly dominated by diffuse scattering with a detected peak at q ~ 2.3 nm−1, indeed, indicating the presence of entrapped doxorubicin in crystalline state (Fig. 2). The assignment of this crystalline structure was recently determined based on the characterization of a pellet formed from a solution of 25 mg/mL doxorubicin dispersed in 125 mM ammonium sulfate [27]. It was found that the free doxorubicin-sulfate crystalline state has a 3D hexagonal lattice with unit cell dimensions a = b = 3.18 nm and c = 2.02 nm and the first strongest scattering reflection (1,0,0) appears at q = 2.3 nm− 1, in good agreement with previous work [36]. The appearance of a reflection at this q value in the present work therefore indicates that the entrapped drug crystalline state in vesicles of Caelyx® and SinaDoxosome has similar structure to that of free doxorubicin-sulfate crystalline state. Using an automated image processing approach, a series of several cryo-TEM images was objectively analyzed for estimation of vesicular size (Fig. 1I–L) and apparent aspect ratio (AR) between the longest and shortest visualized diameters (Fig. 3), counting between 90 and 1423 vesicles depending on the formulation. This approach addresses the user interference and lack of statistical power that otherwise challenges semi-quantitative analysis from cryo-TEM image handling, however, vesicles that were overlapping or not clearly visible in the images were excluded from the analysis. It should also be emphasized that the orientation of non-spherical vesicles, and therefore the apparent AR, can be affected by sample layer thickness. The latter, in turn is affected by the blotting speed and product viscosity. Overlapping vesicles and vesicles of all orientations were observed in all samples, indicating that the film is thicker than the dimensions of the vesicles [35]. In a twodimensional projection of prolate spheroids in free rotation, the major axes may not be viewed in their full extent. Consequently, the apparent

Fig. 2. SAXS patterns of the two liposomal formulations Caelyx® and SinaDoxosome. The experiments were performed at 25 °C.

Fig. 3. Cumulative particle distribution as a function of liposome aspect ratio (AR) as estimated from cryo-TEM image processing. The AR is the ratio between the largest and smallest diameters in each detected vesicle.

AR may be an underestimation of the actual AR. This underestimation however, applies to all investigated particles, when assuming the same level of free rotation between samples. Considering these limitations, we show that Doxil® contained a high proportion (corresponding to 44% of the population) of near-spherical vesicles (AR b 1.05) compared with Caelyx® (14%) and DOXOrubicin (10%) (Table 1). On the other hand, Caelyx® and DOXOrubicin had comparable AR profiles throughout, but having higher proportions of vesicles with ARs between 1.05 and 1.15 compared with Doxil® (64% and 66%, respectively compared with 33% in the case of Doxil®). The results further show that Doxil®, Caelyx® and DOXOrubicin contain similar proportions of highly elongated vesicles (AR N 1.15) with encapsulated long needles. In the case of SinaDoxosome, deformed circular disks were not included in the analysis and the size and apparent ARs are derived explicitly from intact vesicles (Table 1). However, a direct quantitative morphological comparison of this batch with the other three products is challenged by the low overall quality and vesicle count among the processed images and is therefore avoided.

3.2. Zeta potential assessment Further subtle differences between Doxil®, Caelyx® and DOXOrubicin were reflected in zeta potential values derived from electrophoretic mobility measurements under identical experimental conditions. Indeed, the zeta potential of Doxil® was less negative than both Caelyx® and DOXOrubicin (Table 1). On the basis of cryo-TEM analysis, Doxil® contains a high proportion of vesicles with AR b 1.05 compared with Caelyx® and DOXOrubicin. It is therefore tempting to speculate that the differences in the average zeta potential values may be a net reflection of vesicular population curvature differences that in turn modulate methoxyPEG (mPEG) shielding efficacy and therefore the electrical double layer. For instance, with near-spherical vesicles (AR b 1.05) the projected mPEG chains may assume a brush-like configuration and hence better surface protection, whereas in prolate ellipsoidal vesicles the projected mPEGs may display a mushroom-like architecture and moderate shielding. Accordingly, future experiments should evaluate these possibilities with vesicles loaded with lower concentrations of doxorubicin (leading to formation of shorter needles) or utilizing ammonium methanesulfonate (rather than ammonium sulfate) for doxorubicin remote loading [27], since these may yield vesicles with more defined ARs. In contrast to the other products, SinaDoxosome showed the largest average negative zeta potential, and this presumably arises from the contribution of other structures (Fig. 1) present in this product. Some

P.P. Wibroe et al. / Journal of Controlled Release 221 (2016) 1–8

5

Table 1 Selected measured physical parameters of liposomal products. Doxil®

Caelyx®

DOXOrubicin

SinaDoxosomea

Cryo-TEM Vesicle count Mean AR AR b 1.05 (% of total) 1.05 b AR b 1.15 (% of total) AR N 1.15 (% of total) Mean vesicle size ± SD (nm) Median vesicle size (nm)

1423 1.12 44.0 33.2 22.8 64 ± 13 63.9

705 1.12 13.9 63.7 22.4 59 ± 12 58.9

580 1.13 10.0 65.9 24.1 56 ± 16 55.0

90 1.08 37.8 51.1 11.1 62 ± 14 62.5

DLS Zeta potential (mV) Mean size ± SEM (nm) Polydispersity index

−9.17 ± 2.40 85.9 ± 0.6 0.026

−15.54 ± 0.22 84.3 ± 1.4 0.079

−15.86 ± 1.53 85.2 ± 2.0 0.047

−20.75 ± 0.96 101.5 ± 2.2 0.12

NTA Particle count Particle conc (mL−1) Mean size ± SD (nm) Median size (nm)

3612 4.46·1013 88 ± 27 85

2893 3.53·1013 79 ± 25 83

2902 3.19·1013 94 ± 25 92

1459 1.62·1013 98 ± 37 91

a

Cryo-TEM parameters for SinaDoxosome are restricted to intact vesicles and exclude the presence of deformed circular disks.

degree of phospholipid hydrolysis may also explain these observations, but this was not investigated. 3.3. Size analysis On the basis of the cryo-TEM image processing, we calculated an equivalent spherical diameter for the vesicles, which due to the aforementioned free rotation is given with an underestimation of the actual diameter (Table 1). PEG chains are not visible in cryo-TEM images due to their low electron density, where the contribution of PEG adlayer thickness may correspond to 4.8 nm [27]. Therefore, size measurements based on cryo-TEM is not a true representation of the vesicular hydrodynamic diameter. Hydrodynamic size determination was therefore followed by both DLS and NTA, where both modalities also assume equivalent spherical sizes. DLS observes the time-dependent fluctuations in scattering intensity caused by constructive and destructive interference resulting from the relative Brownian motion of the particles in a suspension. The average hydrodynamic diameters obtained by DLS, which measures total scattered light from an ensemble of vesicles, is heavily weighted towards large particles, leading to significant errors, if particle populations are polydisperse. The results in Table 1 show that Doxil®, Caelyx® and DOXOrubicin all exhibit low polydispersity indices and comparable mean hydrodynamic equivalent spherical diameters. As expected, DLS derived diameters are considerably higher than the cryo-TEM derived estimates. SinaDoxosome, however, was more polydisperse and exhibited a slightly larger mean diameter than the other products (Table 1). DLS size estimates are presumably more accurate for the current Doxil® batch, since this contains a high proportion of vesicles with AR b 1.05 compared with Caelyx® and DOXOrubicin. Unlike DLS, NTA measures the Brownian motion through video analysis, tracking the movement of individual particles, thus overcoming some of the inherent problems associated with DLS, particularly with heterogeneous samples [30,31]. The NTA results revealed similar mean equivalent spherical sizes to DLS for all four products (Fig. 4, Table 1). The plots of light scatter intensity versus particle size allow differences in particle refractive indices to be explored, which may represent differences in particle composition. This analysis showed a broad scattering for the four products, where a small proportion of particles scattered with high intensities. These may represent the presence of aggregates, micelles, etc. within a typical size bracket, or simply random fluctuations in vesicle position and orientation relative to the illuminated and focal volume. The scattering patterns, however, are similar for all four products, as indicated by their mean scatter values and population sizes on arbitrary scattering/size gating (Fig. 4). Interestingly, the gating patterns remain similar to empty PEGylated

liposomes of similar size (mean hydrodynamic size 98 ± 28 nm; median size 93 nm) and bilayer composition. This presumably indicates that the entrapped needle-shaped doxorubicin crystals play a minor modulatory role in the vesicular scattering. NTA further allows the particle concentration to be estimated directly. The results reveal higher particle concentration in Doxil® compared with Caelyx® and DOXOrubicin (Table 1). This is in line with cryo-TEM observations (Fig. 3) as Doxil® contains a high proportion of vesicles with AR b 1.05. Since the manufacturer's stated final doxorubicin concentration in the product is 2 mg/mL, more vesicles must be present to entrap doxorubicin. For Caelyx® and DOXOrubicin, particle concentrations are rather similar and correlating well with their similar vesicular AR distribution. In contrast to these observations, SinaDoxosome had the lowest measured particle concentration in line with cryo-TEM observations. Based on this low particle count, which for NTA includes empty disks, we speculate that the encapsulated doxorubicin concentration in this product is below 2 mg/mL. This assumption does not exclude possible presence of free doxorubicin in solution, resulting in a total doxorubicin concentration of 2 mg/mL in this batch. 3.4. Complement activation Since PEGylated liposomes, and in particular Doxil®, is known to activate the complement system [16,19,40] we compared the complement activation from all 4 products in sera of five healthy individuals. The four established complement markers C4d (indicative of activation of classical and lectin pathways), Bb (demonstrating alternative pathway (AP) activation and general response amplification), C5a (complement anaphylatoxin and indicative of C5 convertase activity), and SC5b9 (activation marker of the terminal complement pathway and a measure of complement activation as a whole) were measured [32,33]. In relation to clinical dosing, complement activation was compared on the basis of equivalent stated doxorubicin levels. The results presented in Fig. 5A show that all four products considerably raise serum levels of C4d with some variability among the tested donors. Notably, Caelyx®, compared with other products, causes significant increases in C4d levels in the sera of subjects A, D and E. Earlier, we suggested that PEGylated liposomes activate calcium-sensitive pathways of the human complement system through C1q binding either directly, or via anti-phospholipid antibody deposition [32]. Therefore, a plausible explanation for variations in Caelyx® responses is attributed to the presence of a population of vesicles that exhibit improved C1q and/or anti-phospholipid antibody binding, where the observed donor variability could represent individual differences in antibody titer.

6

P.P. Wibroe et al. / Journal of Controlled Release 221 (2016) 1–8

Fig. 4. Size determination by Nanoparticle Tracking Analysis (NTA). Top: Mean size distribution (black line) ± SD (gray area) of the 4 liposomal products and an empty (doxorubicin-free) liposome of same lipid composition. Middle: Light scatter intensity as a function of particle diameter. Red lines indicate gating, where the vertical gates are set at 40 and 100 nm, and the horizontal gates at the scatter mean and 1.5× scatter mean. Mean scatter values of the four liposomal products were 1.008 ± 0.017 a.u. Bottom: Population of the different gates, as indicated by the red areas. For the measurements, all samples were diluted in water to reach 1.6 × 108 to 4.5 × 108 particles/mL.

Next, we examined serum elevation of liposome-mediated Bb levels. SinaDoxosome was the most potent in accelerating the alternative pathway turnover as evident from dramatic rises of Bb levels above background in all tested sera, except for donor B who showed modest responses for all products (Fig. 5B). Although PEGylated liposomes can also directly increase AP turnover, or through the AP amplification loop [32], the dramatic increase in SinaDoxosome-mediated Bb levels may be related to the presence of large quantities of disk-shaped mixed-micelles in this product. SinaDoxosome was also the most potent in raising serum levels of C5a and SC5b-9 (Fig. 5C–D), Another interesting trend in complement activation studies is that in some individuals the initial liposome-mediated activation of calciumsensitive pathways (and to some extent alternative pathway) may not necessarily translate to strong activation of the terminal pathway of the complement system (Fig. 5C–D). This is particularly clear for Doxil® (e.g., sera A, B, D & E)and Caelyx® (sera B & D), whose increase in C4d levels does not necessarily elevate C5a and SC5b-9 levels considerably. The reasons for these observations are not clear, but they could be related to vesicular population differences. For instance, lateral mobility of lipid constituents may induce curvature-dependent domains in the bilayer that could have different complement-activating potentials and this is presumably amplified by ARs. 4. Conclusions Here, we presented direct morphological and immunological comparison between Doxil®, Caelyx® DOXOrubicin and SinaDoxosome. Our attempt, despite having examined limited samples, highlighted

some potential differences among the tested batches. Through a panintegrated morphology, particle number and size characterization, we not only observed some morphological population differences between Doxil® and the selected follow-on products, but also in two productions, which have been perceived to be identical (Doxil® and Caeylx®). These differences may have contributed to fluctuations in vesicularmediated complement responses in the sera of tested subjects. Complement studies further suggested that there could be additional parameters, such as a population of vesicles with appropriate pro-complement activating domains (e.g., in the case of Caelyx®), which are not easily distinguishable with current analytical tools. However, larger sample sets are necessary to examine differences within and between batches, and between follow-on products to statistically assess significant variations, which may translate into differences in clinical performances. Accordingly, our observations may not necessarily represent the global condition/quality of any of the liposomal products, but are representative of the investigated batches. Based on our findings, we encourage a systematic study to investigate the role of particle shape (and ARs) in complement and other clinical adverse responses (e.g., palmar-plantar erythrodysesthesia). For instance, in a typical patient complement reactivity may be examined in vitro with different lots of the liposomal products. This may lead to identification of the least reactive (or non-reactive) batches for infusion. Such profiling may further reveal correlations between complement activation (and perhaps other safety issues) with vesicular ARs and help for better and safer vesicular product design that could be applicable to broader aspects of non-biologic complex drug development, thus leading to improved regulatory guidelines. On this line, it is tempting

P.P. Wibroe et al. / Journal of Controlled Release 221 (2016) 1–8

7

Fig. 5. Complement activation by the four liposomal products. Four soluble complement markers were measured in human serum obtained from five healthy donors: C4d (A), Bb (B), C5a (C) and SC5b-9 (D). Insets: Zymosan responses show high and comparable complement functionality for all 5 tested donors. The most reactive product from each complement marker was tested for significance against other products in respective donors (*p b 0.05; **p b 0.01; ***p b 0.001). The complement activation patterns were similar between two different batches of SinaDoxosome (only one example is shown).

to speculate that vesicular populations with high ARs (e.g., N1.15) may be contributing to the palmar-plantar erythrodysesthesia (‘hand-foot’ syndrome), which has an increased incidence for PEGylated liposomal doxorubicin compared with conventional doxorubicin [41]. These vesicular populations display a prolate ellipsoidal shape and are presumably less deformable than their spherical counterparts, thereby inducing ‘hand-foot’ syndrome with local pressure as a contributor causing doxorubicin release and excretion in sweat [37]. Finally, we further recommend that, as part of the analytical characterization portfolio of intravenous liposomal products, complement activation tests as well as comprehensive morphology analysis by cryoTEM should be considered and introduced. Competing interests The authors declare no competing financial interest. The authors further declare that they have no collaboration and/or consultancy agreement with the manufacturers of the investigated liposomal products. Acknowledgments This work was supported by the Danish Agency for Science, Technology and Innovation, reference 09-065736 (Det Strategiske Forskningsråd) (SMM). PPW is a recipient of a PhD Scholarship Award from the Faculty of Health and Medical Sciences, University of Copenhagen. We also acknowledge Dr. Marina A. Dobrovolskaia (Nanotechnology Characterization Laboratory, Cancer Research Technology Program, Frederick National Laboratory for Cancer Research, Ledios Biomedical Research, Inc., Frederick, MD, USA) for providing samples of Doxil® and DOXOrubicin, as well as for the fruitful discussion throughout this

work. We are also grateful to the Core Facility for Integrated Microscopy, Faculty of Health and Medical Sciences, University of Copenhagen for technical support in sample preparation and acquisition of cryotransmission electron microscopy images.

References [1] Y. Barenholz, Doxil® — the first FDA-approved nano-drug: lessons learned, J. Control. Release 160 (2012) 117–134. [2] S. Tinkle, S.E. McNeil, S. Mühlebach, R. Bawa, G. Borchard, Y. Barenholz, L. Tamarkin, N. Desai, Nanomedicines: addressing the scientific and regulatory gap, Ann. N.Y. Acad. Sci. 1313 (2014) 35–56. [3] S.M. Moghimi, Z.S. Farhangrazi, Defining and characterizing nonbiological complex drugs (NBCDs) — is size enough? The case for liposomal doxorubicin generics (‘liposomal nanosimilars’) for injection, GaBI J. 3 (2014) 56–57. [4] S. Gaspani, B. Milani, Access to liposomal generic formulations: beyond AmBisome and Doxil/Caelyx, GaBI J. 2 (2013) 60–62. [5] S.M. Moghimi, D. Peer, R. Langer, Reshaping the future of nanopharmaceuticals: ad iudicium, ACS Nano 5 (2011) 8454–8458. [6] M.L. Chen, V. Shah, R. Patnaik, W. Adams, A. Hussain, D. Conner, M. Mehta, H. Malinowski, J. Lazor, S.M. Huang, D. Hare, L. Lesko, D. Sporn, R. Williams, Bioavailability and bioequivalence: an FDA regulatory overview, Pharm. Res. 18 (2001) 1645–1650. [7] S.E. McNeil, Characterization of Nanoparticles Intended for Drug Delivery, Humana Press/Springer, New York, 2011. [8] EMA guideline on biosimilar medicinal products, http://www.ema.europa.eu/docs/ en_GB/document_library/Scientific_guideline/2014/10/WC500176768.pdf. 2014 (accessed October 2015). [9] D.J.A. Crommelin, V.P. Shah, I. Klebovich, S.E. McNeil, V. Weinstein, B. Fluthmann, S. Muhlebach, J.S.B. de Vlieger, The similarity question for biologicals and nonbiological complex drugs, Eur. J. Pharm. Sci. 76 (2015) 10–17. [10] H. Schellekens, S. Stegemann, V. Weinstein, J.S.B. de Vlieger, B. Fluhmann, S. Muhlebach, R. Gaspar, V.P. Shah, D.J.A. Crommelin, How to regulate nonbiological complex drugs (NBCD) and their follow-on versions: points to consider, AAPS J. 16 (2014) 15–21. [11] S.M. Moghimi, A.C. Hunter, J.C. Murray, Nanomedicine: current status and future prospects, FASEB J. 19 (2005) 311–330.

8

P.P. Wibroe et al. / Journal of Controlled Release 221 (2016) 1–8

[12] S.M. Moghimi, A.C. Hunter, T.L. Andresen, Factors controlling nanoparticle pharmacokinetics: an integrated analysis and perspective, Annu. Rev. Pharmacol. Toxicol. 52 (2012) 481–503. [13] I. Hamad, S.M. Moghimi, Critical issues in site-specific targeting of solid tumours: the carrier, the tumour barriers and the bioavailable drug, Expert Opin. Drug Deliv. 5 (2008) 205–219. [14] R.I. MacCuspie, K. Rogers, M. Patra, Z.Y. Suo, A.J. Allen, M.N. Martin, V.A. Hackley, Challenges for physical characterization of silver nanoparticles under pristine and environmentally relevant conditions, J. Environ. Monit. 13 (2011) 1212–1226. [15] M.A. Hamburg, FDA's approach to regulation of products of nanotechnology, Science 336 (2012) 299–300. [16] S.M. Moghimi, A.J. Andersen, D. Ahmadvand, P.P. Wibroe, T.L. Andresen, A.C. Hunter, Material properties in complement activation, Adv. Drug Deliv. Rev. 63 (2011) 1000–1007. [17] S.M. Moghimi, P.P. Wibroe, S.Y. Helvig, Z.S. Farhangrazi, A.C. Hunter, Genomic perspectives in inter-individual adverse responses following nanomedicine administration: the way forward, Adv. Drug Deliv. Rev. 64 (2012) 1385–1393. [18] J. Szebeni, F. Muggia, A. Gabizon, Y. Barenholz, Activation of complement by therapeutic liposomes and other lipid excipient-based therapeutic products: prediction and prevention, Adv. Drug Deliv. Rev. 63 (2011) 1020–1030. [19] A. Chanan-Khan, J. Szebeni, S. Savay, L. Liebes, N.M. Rafique, C.R. Alving, F.M. Muggia, Complement activation following first exposure to pegylated liposomal doxorubicin (Doxil): possible role in hypersensitivity reactions, Ann. Oncol. 14 (2003) 1430–1437. [20] S.M. Moghimi, I. Hamad, Liposome-mediated triggering of complement cascade, J. Liposome Res. 18 (2008) 195–209. [21] G. Milosevits, Z. Rozsnyay, G.T. Kozma, J. Milosevits, G. Tomory, H. Robotka, L. Rosivall, J. Szebeni, Flow cytometric analysis of supravesicular structures in doxorubicin-containing pegylated liposomes, Chem. Phys. Lipid. 165 (2012) 482–487. [22] S. Mitragotri, J. Lahann, Physical approaches to biomaterial design, Nat. Mater. 8 (2009) 15–23. [23] J.A. Champion, Y.K. Katare, S. Mitragotri, Particle shape: a new design parameter for micro- and nanoscale drug delivery carriers, J. Control. Release 121 (2007) 3–9. [24] Y. Geng, P. Dalhaimer, S.S. Cai, R. Tsai, M. Tewari, T. Minko, D.E. Discher, Shape effects of filaments versus spherical particles in flow and drug delivery, Nat. Nanotechnol. 2 (2007) 249–255. [25] J.A. Champion, S. Mitragotri, Role of target geometry in phagocytosis, Proc. Natl. Acad. Sci. U. S. A. 103 (2006) 4930–4934. [26] D.A. Christian, S.S. Cai, O.B. Garbuzenko, T. Harada, A.L. Zajac, T. Minko, D.E. Discher, Flexible filaments for in vivo imaging and delivery: persistent circulation of filomicelles opens the dosage window for sustained tumor shrinkage, Mol. Pharm. 6 (2009) 1343–1352. [27] Y. Schilt, T. Berman, X. Wei, Y. Barenholz, U. Raviv, Using solution X-ray scattering to determine the high-resolution structure and morphology of PEGylated liposomal doxorubicin nanodrugs, Biochim. Biophys. Acta-Gen. Subj. 1860 (2016) 108–119. [28] Non-binding recommendations from FDA for liposomal Doxorubicin hydrochloride evaluation, http://www.fda.gov/downloads/Drugs/

[29]

[30]

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

GuidanceComplianceRegulatorInformation/Guidances/UCM199635.pdf. 2014 (accessed October 2015). A. Soundararajan, A. Bao, W.T. Phillips, R. Perez, B.A. Goins, [Re-186] Liposomal doxorubicin (Doxil): in vitro stability, pharmacokinetics, imaging and biodistribution in a head and neck squamous cell carcinoma xenograft model, Nucl. Med. Biol. 36 (2009) 515–524. M. Wright, Nanoparticle tracking analysis for the multiparameter characterization and counting of nanoparticle suspensions, in: M. Soloviev (Ed.) Nanoparticles in Biology and MedicineHumana Press 2012, pp. 511–524. P. Van der Meeren, M. Kasinos, H. Saveyn, Relevance of two-dimensional Brownian motion dynamics in applying nanoparticle tracking analysis, in: M. Soloviev (Ed.), Nanoparticles in Biology and Medicine, Humana Press 2012, pp. 525–534. S.M. Moghimi, I. Hamad, T.L. Andresen, K. Jorgensen, J. Szebeni, Methylation of the phosphate oxygen moiety of phospholipid-methoxy(polyethylene glycol) conjugate prevents PEGylated liposome-mediated complement activation and anaphylatoxin production, FASEB J. 20 (2006) 2591–2593. A.J. Andersen, J.T. Robinson, H.J. Dai, A.C. Hunter, T.L. Andresen, S.M. Moghimi, Single-walled carbon nanotube surface control of complement recognition and activation, ACS Nano 7 (2013) 1108–1119. N.K. Banda, G. Mehta, Y. Chao, G.K. Wang, S. Inturi, L. Fossati-Jimack, M. Botto, L.P. Wu, S.M. Moghimi, D. Simberg, Mechanisms of complement activation by dextran-coated superparamagnetic iron oxide (SPIO) nanoworms in mouse versus human serum, Part. Fibre Toxicol. 11 (2014) 64, http://dx.doi.org/10.1186/s12989014-0064-2. J. Kuntsche, J.C. Horst, H. Bunjes, Cryogenic transmission electron microscopy (cryoTEM) for studying the morphology of colloidal drug delivery systems, Int. J. Pharm. 417 (2011) 120–137. D.D. Lasic, P.M. Frederik, M.C.A. Stuart, Y. Barenholz, T.J. Mcintosh, Gelation of liposome interior — a novel method for drug encapsulation, FEBS Lett. 312 (1992) 255–258. D.D. Lasic, B. Ceh, M.C.A. Stuart, L. Guo, P.M. Frederik, Y. Barenholz, Transmembrane gradient driven phase-transitions within vesicles — lessons for drug-delivery, Biochim. Biophys. Acta-Biomembr. 1239 (1995) 145–156. X.G. Li, D.J. Hirsh, D. Cabral-Lilly, A. Zirkel, S.M. Gruner, A.S. Janoff, W.R. Perkins, Doxorubicin physical state in solution and inside liposomes loaded via a pH gradient, Biochim. Biophys. Acta-Biomembr. 1415 (1998) 23–40. M. Almgren, K. Edwards, G. Karlsson, Cryo transmission electron microscopy of liposomes and related structures, Colloid. Surf. Physicochem. Eng. Aspects 174 (2000) 3–21. J. Szebeni, L. Baranyi, S. Savay, H.U. Lutz, E. Jelezarova, R. Bunger, C.R. Alving, The role of complement activation in hypersensitivity to pegylated liposomal doxorubicin (DOXIL®), J. Liposome Res. 10 (2000) 467–481. D. Lorusso, A. Di Stefano, V. Carone, A. Fagotti, S. Pisconti, G. Scambia, Pegylated liposomal doxorubicin-related palmar-plantar erythrodysesthesia (‘hand-foot’ syndrome), Ann. Oncol. 18 (2007) 1159–1164.