Journal of Controlled Release 299 (2019) 31–43

Contents lists available at ScienceDirect

Journal of Controlled Release

journal homepage: www.elsevier.com/locate/jconrel

Review article Measuring particle size distribution of nanoparticle enabled medicinal T products, the joint view of EUNCL and NCI-NCL. A step by step approach combining orthogonal measurements with increasing complexity ⁎ ⁎⁎ Caputo F.a, , Clogston J.b, Calzolai L.c, Rösslein M.d, Prina-Mello A.e,f, a Univ. Grenoble Alpes, CEA, LETI, F-38000 Grenoble, France b Nanotechnology Characterization Laboratory, Cancer Research Technology Program, Leidos Biomedical Research, Inc., Frederick National Laboratory for Cancer Research, Frederick, MD 21702, USA c European Commission, Joint Research Centre, Via Enrico Fermi 2749, 21027 Ispra, VA, Italy d Swiss Federal Laboratories for Materials Research and Testing, Laboratory for Particles—Biology Interactions, EMPA, Lerchenfeldstrasse 5, St. Gallen CH-9014, Switzerland e Laboratory for Biological Characterisation of Advanced Materials (LBCAM), Department of Clinical Medicine, Trinity Translational Medicine Institute (TTMI), School of Medicine, Trinity College Dublin, Dublin 8, Ireland f AMBER Centre and CRANN Institute, Trinity College Dublin, Dublin 2, Ireland

ARTICLE INFO ABSTRACT

Keywords: The particle size distribution (PSD) and the stability of nanoparticles enabled medicinal products (NEP) in complex Nanomedicine biological environments are key attributes to assess their quality, safety and efficacy. Despite its low resolution, Nanoparticles enabled medicinal products dynamic light scattering (DLS) is the most common sizing technique since the onset of NEP in pharmaceutical Physico-chemical properties technologies. Considering the limitations of the existing sizing measurements and the challenges posed by complex Particle size distribution NEPs both scientists and regulators encourage the combination of multiple orthogonal high-resolution approaches to Morphology shed light in the NEP sizing space (e.g. dynamic light scattering, electron microscopy, field flow fractionation coupled Standard operating procedures (SOPs) to online sizing detectors, centrifugal techniques, particle tracking analysis and tunable resistive pulse sensing). The pharmaceutical and biotechnology developers are now challenged to find their own pragmatic characterisation ap- proaches, which should be fit for purpose and minimize costs at the same time, in a complicated landscape whereonly a few standards exist. In order to support the community, the European Nanomedicine Characterisation Laboratory (EUNCL) and the US National Cancer Institute Nanotechnology Characterization Laboratory (NCI-NCL) have jointly developed multiple standard operating procedures (SOPs) for NEP assessment, including the measurements of particle size distribution, and are offering wide access to their ‘state of the art’ characterisation platforms, in additionto making SOPs publicly available. This joint perspective article would like to present the NCI-NCL and EUNCL multi- step approach of incremental complexity to measure particle size distribution and size stability of NEPs, consisting of a quick preliminary step to assess sample integrity and stability by low resolution techniques (pre-screening), followed by the combination of complementary high resolution sizing measurements performed both in simple buffers and in complex biological media. Test cases are presented to demonstrate: i) the need for employing at least one high- resolution sizing technique, ii) the importance of selecting the correct sizing techniques for the purpose, and iii) the robustness of utilizing orthogonal sizing techniques to study the physical properties of complex NEP samples.

Abbreviations: AF4, asymmetric flow field flow fractionation; AUC, analytical ultracentrifugation; DCS, Differential centrifugal sedimentation; DLS, Dynamiclight scattering; EMA, European medical agency; EUNCL, European Nanomedicine Characterisation Laboratory; FDA, US Food and Drug Administration; FFF, field flow fractionation; MALS, multi angle light scattering; MNs, multinationals; NCI-NCL, National Cancer Institute Nanotechnology Characterization Lab; NEP, nanoparticle enabled medicinal products; NTA, nanoparticle tracking analysis; PCS, photon correlation spectroscopy; PdI, polydispersity index; PSD, particle size distribution; PTA, Particle tracking analysis; SEC, size exclusion chromatography; SMEs, small and medium enterprises; SOP, standard operating procedure; TEM, Transmission electron microscopy; TRPS, Tunable Resistive Pulse Sensing; UV, ultra-violet. ⁎ Corresponding author. ⁎⁎ Corresponding author at: Laboratory for Biological Characterisation of Advanced Materials (LBCAM), Department of Clinical Medicine, Trinity Translational Medicine Institute (TTMI), School of Medicine, Trinity College Dublin, Dublin 8, Ireland; AMBER Centre and CRANN Institute, Trinity College Dublin, Dublin 2, Ireland. E-mail addresses: [email protected], [email protected] (F. Caputo), [email protected] (A. Prina-Mello). https://doi.org/10.1016/j.jconrel.2019.02.030 Received 20 December 2018; Received in revised form 20 February 2019; Accepted 20 February 2019 Available online 21 February 2019 0168-3659/ © 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/). F. Caputo, et al. Journal of Controlled Release 299 (2019) 31–43

1. Introduction addition, several articles are available where lessons learned and knowledge are shared with the scientific community [3–6,12,13]. The application of nanotechnology in healthcare has a tremendous From this close interaction, it emerged that average particle size, potential to address a variety of medical conditions by providing better particle size distribution (PSD) and aggregation behaviors in complex diagnostics and therapy. A broad range of nanoparticles enabled med- biological media are, among others, critical quality attributes for the icinal products (NEP) have been investigated for medical applications, preclinical characterisation of NEP [3,5,14–17]. In fact, they influence including liposomes, polymeric nanoparticles, lipid-based nano- the body absorption, biodistribution and excretion, during their phar- particles, micelles, nanocrystals, metal colloids, metal oxides, and many macological targeting and off targets effects [1,18,19]. Importantly, the others [1]. Unfortunately, when compared to the extensive research particle size and polydispersity can significantly change due to inter- activities in research laboratories and industries (e.g., SMEs and MNs), actions of plasma proteins with the NEP surface. In biological media the clinically approved nanomedicines are still very limited [2]. Among NEPs may aggregate increasing the average population size and even the extensive number of NEP published in literature, only about 50 sediment or, on the contrary, aggregation may be reduced, or they can candidates have successfully crossed the “valley of death”, and thus start to degrade, producing populations of smaller particles. translated from the research laboratories to scale up approved product Due to the key role in determining the efficacy and safety of NEP for use in clinical setups. Very often NEPs fail to reach late clinical [1], particle size, polydispersity and size stability should be controlled phases due to the lack of pre-clinical characterisation protocols, which in simple and in biological media from the early development stages, are needed to correlate their physico-chemical properties with their ideally from their design. At later stages, e.g. during routine manu- biological effects [1,3,4]. If compared to “classical” small molecule facturing, average size and polydispersity are key attributes to be drugs, the assessment of the quality, safety and efficacy profiles of NEPs controlled for quality purposes. According to the regulators' guidelines, demands the investigation of many additional physico-chemical prop- including EMA [8–10] and FDA [7,11], the NEP developers should not erties, including their chemical composition, average particle size and only characterise the particle size distribution of their products im- polydispersity, dispersion stability, particle shape, surface charge, drug mediately after their synthesis, but also study i) the reproducibility of loading and drug release, surface coating and hydrophobicity [1–3,5]. the manufacturing procedure (batch to batch consistency), ii) the long Moreover, the physico-chemical properties of NEP, including the size, term stability of the formulation during storage, iii) measure the na- physical and chemical stability, surface properties and drug release noparticle size in the final product administered to the patient andfi- profile, need to be investigated, not only in their original formulation, nally iv) analyze the changes of size and shape of the NEP after being in but also when dispersed in biological complex environments, since contact with physiological media (e.g., interactions with serum pro- different pH, high ionic strengths and the interactions with plasmacan teins) [3]. modify the physical properties of the particles, thus influencing their Many techniques are being used to measure the size distribution of safety and efficacy profiles [3,6,7]. To complicate matters even more, nanoparticle formulation in medicinal products, including electron the uniqueness of every NEP imposes a different methodological ap- microscopy (EM), laser scattering techniques (e.g. dynamic and static proach to characterise its physical-chemical properties. In this context, light scattering, laser diffraction), field flow fractionation (FFF) coupled the development of reliable pre-clinical characterisation strategies and to multiple sizing detectors, centrifugal techniques (e.g. analytical ul- of suitable standard protocols is a critical step to determine the safety tracentrifugation and centrifugal particle sedimentation), tunable re- and quality of each emerging NEP to sustain the safe industrial devel- sistive pulse sensing (TRPS) and particle tracking analysis (PTA). opment of nanomedicines. This robust and comprehensive pre-clinical However only a few standards for size measurements of NEP are characterisation has been strongly requested by the public authorities, available, while many are still under development [4]. Based on a re- regulatory bodies and agencies funding R&D, with the aim to translate cent analysis [1] of the dossiers presented to the FDA from 1973 to successful NEP candidates from the laboratories to the clinic [2–4]. 2015, dynamic light scattering (DLS) has been the most recurrently Guidelines and reflection papers describing the pre-clinical charater- used sizing technique by nanomedicine stakeholders (48% of reporting isation needs of NEPs have been published by regulatory agencies, in- applications), followed by laser diffraction and microscopy which are cluding the European Medical Agency (EMA) [8–10] and the US Food adopted in 30% and 14% of the applications. Only approximately 8% of and Drug Administration (FDA) [7,11]. Recently, a review from Hala- the applicants have characterized their products by other com- moda-Kenzaoui et al. has mapped the available standards against the plementary sizing techniques, e.g. gel permeation chromatography, regulatory needs for NEPs pre-clinical characterisation, underlining the asymmetrical flow field-flow fractionation, and centrifugation mea- currents needs, and also presenting the importance and the ongoing surements. DLS is recognized to be a low-resolution method, which is effort to cover the regulatory gaps by developing new standards for only suitable for an initial check of the sample. However, in order to NEPs [4]. avoid misleading results, it is not recommended to apply DLS as the In this context, the US National Cancer Institute Nanotechnology only method for the pre-clinical characterisation of complex NEPs Characterization Laboratory (NCI-NCL) and, since 2015, the European [3,20–22]. On the contrary, the combination of multiple high-resolu- Nanomedicine Characterisation Laboratory (EUNCL) have been estab- tion techniques is often required to understand the PSD of NEPs, lished to unbiasedly support the developers of NEP. EUNCL and NCI- especially when dispersed in complex biological media. In fact, most of NCL are providing a multi-disciplinary infrastructure which provides a the times NEP samples are polydispersed and it is necessary to resolve comprehensive set of preclinical tests (physical, chemical, in-vitro and multiple populations which slightly differ in size and shape, to differ- in-vivo biological testing). The two laboratories work together and also entiate the presence of small aggregates from larger particles, or to in tight collaboration with regulatory bodies and metrology institutes, quantify a small amount of very big aggregates. Unfortunately, the in order to develop SOPs for the assessment of NEP which are relevant article by D'Mello et al. demonstrates that particle size distribution in for regulatory purposes and to promote a harmonized approach be- the nanomedicine field is often still only measured by low resolution tween Europe and the USA. Their testing strategy allows researchers to DLS measurements, possibly due to its low costs and simplicity in use. determine, and fully understand, the immunological effects, biodis- Recently both scientists and regulators have started to encourage the tribution, pharmacokinetics, metabolism and safety profiles of their use of high resolution approaches and suggested to combine orthogonal nanoparticles, in order to detect early fails and to accelerate the measurements in order to better characterise the behavior of NEP sys- translation of promising products to clinical investigation for safety tems, leading to more accurate and realistic assessment for sizing and (Phase I clinical trial). All the SOPs developed by the two laboratories concentrations [3,7]. Despite these efforts, the co-authors believe that are published on their website (https://ncl.cancer.gov/resources/assay- there is still a strong urgency to educate the community on the draw- cascade-protocols, http://www.euncl.eu/about-us/assay-cascade/). In backs of batch mode DLS and to support the development of new

32 F. Caputo, et al. Journal of Controlled Release 299 (2019) 31–43 reliable high-resolution approaches for the accurate characterisation the NEPs. However, we decided to only focus on the sizing techniques and evaluation of particle size distribution. which are most currently used in NCI-NCL and in EUNCL, in order to It is the intention and objective of this perspective to discuss the provide their direct experience. current state of the art of sizing measurements, comparing the available instrumental capability and their suitability to characterise challenging NEP systems. In this context, we would like to present the NCI-NCL and 1.2. Batch methods: dynamic light scattering EUNCL joint strategy to analyze the particle size distribution of NEPs, which is built in a complexity ladder (e.g., three steps of incremental Dynamic light scattering (DLS) measures the correlation of the time- complexity), as shown in Fig. 1. Practical examples are presented in dependent fluctuations in the light scattered by a suspension ofnano- order to show the suitability of the strategy developed at EUNCL and particles (autocorrelation function), which is determined by their NCI-NCL to measure poly-dispersity, detect dimers and aggregates, Brownian motion. The nanoparticle diffusion coefficient(s) is then measure batch to batch consistency and identify size changes in pre- calculated by fitting the autocorrelation function and used to determine sence of plasma proteins. It is important to point out that this review is the hydrodynamic diameter (or radius) of the particles via the Stokes intended to cover the characterisation of injectable formulations; Einstein equation. The analysis assumes that the object measured is of complex biomaterial or hydrogel embedded nanoparticles with loca- spherical shape and requires that the viscosity of the solution is known. lised delivery are not included. DLS is a very fast and inexpensive analysis that does not require highly specialized personnel. For this reason, it is considered the entry point measurement for particle sizing of NEPs and accepted worldwide 1.1. State of the art techniques for measuring nanoparticle size distribution by the regulators. As stated, DLS is commonly used in the nanomedi- cine, under “batch” mode, which possesses very important limitations Knowledge on the state-of-the-art instrumentations, techniques, and for the analysis of polydispersed samples, and could be considered a capabilities for the characterisation of NEPs, is one of the key aspects low-resolution method [3,20–22]. In fact, as results of the dependence that EUNCL and NCI-NCL have jointly developed and put into practice of scattering intensity on the particle size to the sixth power, a small across the many samples assessed. Table 1 provides a qualitative number of large particles often masks the smaller ones. Therefore, batch comparison of multiple sizing methods applicable to injectable for- mode DLS is a useful user-friendly method to assess sample integrity mulations, their range of applicability and limitation for the measure- and the stability of a nanoformulation when exposed to highly con- ment of PSD and morphology of NEPs (e.g. working size range, re- centrated salts, unsuitable pH or to the presence of plasma [27,28]. solution of polydispersity, detection of small and big aggregates, ability However, when it measures the PSD of polydispersed samples, it often to measure not spherical particles), cost and complexity. produces misleading results. In fact, it is often not able to distinguish Other orthogonal sizing techniques, including small angle X-ray between the PSD of polydispersed samples made of two or three particle scattering (SAXS) [22–24], atomic force microscopy (AFM) [25], and populations in similar size ranges [3,21,22,29], it is unable to distin- single particle ICP-MS (sp-ICP-MS) [26] may also be applicable in guish between small aggregates and larger particles [20] and cannot measuring PSD and particle concentration according to the nature of reliably measure the PSD of NEPs in presence of plasma proteins [3,30].

Fig. 1. EUNCL & NCI-NCL approach to measure particle size distribution. Step by step approach of incremental complexity based on three tiers. The purpose of each step is described and some of the sizing techniques which may be fit for that purpose, according to the specific properties of the NEP under investigation, arelisted.

33 F. Caputo, et al. Journal of Controlled Release 299 (2019) 31–43

For more details on the DLS technique, a recent work published by Maguire and coworkers highlighted the principles, the parameters and reasons influencing the nanoparticle size-measurement size [22]. Moderate Yes Yes Surface charge TRPS 60 nm-10 μm Number-based Yes No Yes Yes Yes Yes but indirect 1.3. Separation particle sizing methods

The resolution of the obtained particle size distribution significantly increases if a fractionation step is introduced prior to sizing measure- ments, in order to separate the particles composing a NEP sample ac- cording to their size. This approach could help to go beyond the limits of batch mode DLS. Very high required No – Cryo-TEM 5–300 nm Number-based Yes No Yes No Yes Yes 1.3.1. Asymmetric field flow fractionation Field-Flow Fractionation (FFF) is a family of flow-based separation methods. Similar to size exclusion chromatography (SEC), in FFF the sample is injected into a liquid stream flowing through a separation channel. However, the FFF channel does not contain packing or sta- tionary phase. Most FFF channels are parabolic flow profile chambers where the retention and separation are generated by an external force, such as a perpendicular liquid flow or centrifugal force [31]. Very high TEM (drying, negative staining) No – 1 nm-10 μm Number-based Yes No Yes Yes Yes Yes In asymmetric flow field flow fractionation4 (AF ) the separation is generated by a cross flow applied perpendicularly to the elution flow, which passes through a porous membrane at the bottom of the channel (accumulation wall). In this gentle fractionation process smaller parti- cles are eluted faster than the larger ones and exit the separation

channel first. AF4 has been successfully applied to many NEP systems including metal oxides, metallic nanoparticles, lipid NPs, liposomes, b virus like particles and polymeric NPs, as reviewed by many authors Trained personnel required Trained personnel required Trained personnel No Molecular weight Drug loading AUC Yes Yes Yes No Yes No [3,32–36]. AF4 NP-sorting coupled to online sizing DLS and/or multi- angle light scattering (MALS) is a very powerful and robust sizing technique which allows measurement of the PSD of polydispersed NEPs -10 μm 1–600 nm

a and the determination of small changes in particle size, e.g. to control b batch to batch consistency and stability of NEPs [3,32–36]. Moreover, Moderate High No Intensity-based Intensity-based Yes Yes Yes Yes Yes No DCS 20 nm the separation of NEPs from plasma proteins allows to study the mod- ifications of PSD in a biologically relevant environment [3]. Interest-

ingly, AF4 can provide additional information than just size measure- ments for NEP systems, by coupling multiple detector combinations

online to the AF4 channel. The ratio between the geometric radius de- rived by the MALS detector and hydrodynamic radius derived by the DLS, can provide indirect indications of the particle shape (spherical vs elongated particles). MALS, in combination with a differential re-

-DLS-MALS fractive index detector can be used to analyze the molecular weight of 4 Trained personnel required Yes No Yes No Yes No Yes but indirect (MALS) polymeric samples [35,37]. Online ICP-MS detectors can be used to measure the chemical composition and impurities of inorganic parti- cles. Finally, qualitative information of the drug loading of active pharmaceutical ingredients which absorb in the UV-VIS, may be de- rived by coupling also an UV-VIS detector to the system. -600 nm 1–400 nm (DLS) 20–400 nm a ModerateYes Medium Yes Yes Yes Yes No Yes (> 100 nm) Yes No AF4 has also some disadvantages: the instruments are rather com- plex and require experienced operators to perform reliable measure- ments. Measurements conditions cannot be standardized, since each NEP sample may require a specific and often laborious method opti- mization, as described in [31]. One common problem that may arise is the loss of sample in the channel due to the irreversible interactions of Low Yes 1–800 nm 30 No Surface charge Surface charge Molecular weight Drug loading – Intensity-basedNo Number-based Intensity-based No No No No Batch mode DLS PTA (NTA) AF NEP with the semipermeable membrane, which is very common in the case of positively charged samples. However, if an elution method for a

specific NEP is developed and validated, AF4-MALS-DLS is a very robust and powerful tool to monitor the sample stability, for synthesis opti- mization and for quality control purposes [3].

1.3.2. Centrifugation based techniques Differential centrifugal sedimentation (DCS) [14,38] and analytical ultracentrifugation (AUC) [39–42] fractionate the particles due to their mass (determined by their size and density). The analysis is based on

interactions sedimentation experiments, which measures the transport of particles density of the particle is needed in the analysis. Lower sizes detectable for very dense particles.

a b in solution under the influence of a strong centrifugal force. For sphe- Cost Easy to use Size range Particle concentration Additional information PSD type Poly-disperse samples Analysis assuming spherical particlesDimers/small agglomerates Big aggregates (1-10 μm) Size Yes changes due to NP-protein Shape Methods adopted

Table 1 Overview of the particleaggregates, size NP-protein interactions, distribution cost measurement and methods. complexity Qualitative of comparison the measurements. according to the working size range, resolution, suitability to measure poly-dispersity, non-spherical particles, dimers and rical particles of homogeneous (known) density, the particle size (e.g.

34 F. Caputo, et al. Journal of Controlled Release 299 (2019) 31–43 the stoke diameter) is calculated from their sedimentation time. In in the sample. By estimating the absorbance detected at the beginning polydispersed samples, different components sediment at different of the analysis, and the residual absorbance associated to the free API, times, and therefore can be detected separately during a single mea- both the unbound API and NEP-bound API fractions can be estimated in surement. one simple run [3,46]. In DCS instruments, the particles sediment in an optically clear disk, where a density gradient has been generated to stabilize the sedi- 1.4. Indirect counting methods mentation. When the particles reach the outside edge of the disk, they change the scattered light portion of a monochromatic light source. The 1.4.1. Particle tracking analysis change of the scattering light vs time is continuously measured and Particle tracking analysis (PTA) is a single particles size measure- then transformed into an intensity-based particle size distribution. AUC ment, that differently from batch mode DLS, does not overestimate set up is slightly different. During an AUC measurement, the sample is larger aggregates or larger particles, allowing to measure the PSD of left to sediment in an optically clear quartz or Safire cell, which is fixed very complex samples. In PTA small size differences between particles into the AUC rotor. The centrifugation force generates a concentration or populations are measured by analyzing multiple high-resolution profile over the cell radial position c(r), which changes withtime. images generated by the light scattered by individual particles. The During the measurement, the concentration profile is continually Brownian motion of each individual particle (diffusion) is captured by scanned inside the cell in the radial direction and measured by multiple the instrument, and then transformed into a single particle diffusion detectors (e.g. interference and absorbance optics) generating a con- coefficient. Finally, using the Stokes-Einstein equation, the hydro- centration distribution of the sample suspension c(r,t). This distribution dynamic diameter is calculated, generating a high resolution number- is then analyzed by multiple models to determine the sedimentation based PSD. A series of recorded videos of approx. 30–60 s is analyzed by coefficient c(s) distributions of the sample, and finally converted to the software, tracking frame by frame the center of each particles. The mass metric particle size distributions. latest software development for PTA, the Finite Track Length Even if DCS and AUC techniques are based on the same sedi- Adjustment algorithm (FTLA), has significantly improved and simpli- mentation principle, the range of the rotational speed achieved by DCS fied the measurements. The application of this algorithm is essential to and AUC instrumentation is different, thus the two techniques possess achieve high resolution measurements, since it reduces size broadening, different detectable size ranges. Maximum speed achieved by DCSis allowing to detect secondary peaks and to increase the resolution. lower than AUC, limiting the smaller detectable size to 5 nm for dense Nowadays, the instrumentation is able to measure with high resolution particles like metallic NPs, and to size > 20 nm for less dense particles the PSD of organic particles from 30 nm to 600 nm, while in the case of (silica, polymeric particles, liposomes, etc.) [38]. AUC can achieve very inorganic particles with a higher capability to scatter light, smaller sizes high centrifugal forces (> 100,000 g) allowing to analyze very small can be detected. objects, including proteins, polymers, small molecules and very small Interestingly, PTA can simultaneously measure the concentration of nanoparticles with the resolution of 1 nm or less. On the other hand, nanoparticles in addition to their size, as it counts individual particles large objects may sediment too quickly to be analyzed by AUC, while in a field of view associated to a known volume. However the analysis DCS is able to detect large particles up to 10 μm, making it an inter- of particle concentrations can be affected by the reduced field view of esting technique to measure large aggregates. the camera, which limits the duration of the tracking and thus, the In theory both AUC and DCS can measure with high accuracy the number of particles tracked. Regularly introducing fresh particles may particle size distribution of highly polydispersed samples within their help to increase the statistical robustness of the measurement [22]. In working range, however it may be difficult to efficiently fractionate depth analysis of the principles and processes behind PTA and NTA is very small and large particles at the same time by choosing a fixed presented in Maguire et al., where a cause/effect is presented, the dif- rotational speed. Recently, it has been shown that AUC experiments ferences with DLS and the aspects linked to the advanced character- performed by using a rotational ramp composed of many speed steps, isation of polydispersed samples are also presented [22]. can help to overcome this problem, being able to analyze with high resolution very polydispersed samples [43,44]. The main limitation of centrifugal sizing techniques is that, to cal- 1.4.2. Tunable resistive pulse sensing culate the particle size distribution from the measured sedimentation Tunable Resistive Pulse Sensing (TRPS) is an indirect counting coefficient, the density of the particle needs to be known andhomo- method based on the Coulter counter principle, which measures particle geneous over the whole population. This may limit AUC and DCS ap- size, charge and concentration by resistive pulse sensing [47]. By plicability to precisely measure the size of aggregates (or agglomer- combining selected pressure and voltage values, the particles suspended ates), since it will be very difficult to selectively measure their density, in an electrolyte medium (e.g. PBS buffer) are driven through a con- which may be different from the density of the isolated particles. ductive membrane (nanopore), which is subjected to an applied elec- Changes in size caused by NP-protein interactions are difficult to trical potential, and thus to an ionic current. The passage of each single measure due to changes in the density of the NP-protein complex. In particle induces a resistive pulse signal or “blockade” that is detected these situations combining DCS measurements and accurate size mea- and measured by the software of the instrument. Particle volume is surements (for example with AF4-DLS) gives information on both the calculated from the magnitude of the pore blockage and then trans- size of the complex and the thickness of the protein layer [45]. formed into particle size, while surface charge and particle concentra- Modern AUC instruments are equipped with both refractive index tion can be derived from the blockade duration and blockade fre- (RI) and/or absorbance detector(s), allowing to analyze other key quency. The instrument needs to be calibrated before each parameters for NEP systems including drug loading [3,46]. In typical measurement with reference materials (usually polystyrene particles) of NEP, the molecular weight of the active pharmaceutical ingredient known size, concentration and surface charge. Hence, a reliable cali- (API) is much lower than the molecular weight of the nanoparticle bration is essential to provide robust and accurate size measurement, carrier. In this case, the nanoparticles sediment much faster than the and care should be taken that the measurement conditions remain unbound API. If the unbound API possesses a specific UV–Vis absorp- constant during calibration and while performing the sample mea- tion profile, then by choosing a centrifugal force which induce these- surement. The instrumentation covers a wide size range, being able to dimentation of the nanoparticle but not of the unbound API, a time- measure particles from 60 nm to 10 μm. It can be particularly inter- independent and radius-independent absorbance background will be esting to measure big aggregates in the submicron range and TRPS can detected at the end of the analysis, when all the nanoparticles have also be useful for quality control purposes (e.g. batch to batch com- sedimented. The latter signal is associated to the unbound API fraction parison of particle size distribution, concentration and surface charge).

35 F. Caputo, et al. Journal of Controlled Release 299 (2019) 31–43

1.5. Direct counting methods not only on NEP size, but also on their shape and internal structure (morphology). Visualization of the particle shape is critical to com- 1.5.1. Transmission electron microscopy plement and interpret the results obtained by other indirect sizing Electron microscopy (EM) techniques, e.g. transmission electron method, including light scattering and centrifugal measurements, microscopy (TEM) and scanning electron microscopy (SEM), are in- which assume a spherical shape to calculate the particle size distribu- dispensable tools for the characterisation of colloidal systems, including tion (DLS, DCS & AUC) or need to define a shape to process the acquired NEPs. EM is a direct counting method, which measures a size value for data (MALS). Therefore, EM observations should always be included in each particle selected for the analysis, thus providing a number-based the integrated analytical approach to characterise NEP systems. particle size distribution. In the basic mode, the analysis of the images Unfortunately, EM measurements use complex and expensive in- acquired by EM are based on a 2D projection of 3D particles. The most strumentation which requires highly trained personnel. For this reason, often used size value calculated by EM images is the diameter of a they may not be used in routine analysis (e.g. for batch to batch com- sphere having the same area as the 2D projected area associated to the parison). 3D particle (e.g. diameter of an equivalent sphere possessing the same Among the EM techniques, TEM is the most widely used in the field 2D projected area). Three dimensional reconstitution and 3D tomo- of nanomedicine: it has been used to image inorganic NEP, liposomes, graphy methods also exist to create 3D images, combining the analysis polymeric micelles, nano-emulsions, lipidic nanoparticles and nano- of the same 2D objects taken by analysis the same slide from many vesicles [50–53]. An appropriate sample preparation method should be different angles [48]. Specialized software help to analyze the images carefully chosen and optimized in order to avoid the artificial mod- and to automate the very laborious counting process to determine the ification of the sample prior to imaging. In presence of artifacts, the PSD, allowing to count a relatively large number of particles for a final images may show a completely modified structure which maynot meaningful statistical analysis (ideally n > 500) [49]. Automated be representative of the NEP sample. For this reason, care should al- procedures have the potential to be significantly more efficient, accu- ways be taken during the interpretation of the TEM images. Different rate, and repeatable than human analysis, reducing the risk of user bias techniques for sample preparation are available including drying, ne- that can be introduced by manual particle selection and counting. gative staining, freeze-fracture and cryo-TEM [50,51,53]. EM allows visualization of the NEPs, providing direct information

Fig. 2. Batch-mode and flow-mode diameters. Reprinted by permission from Hansen, M.; Smith, M. C.; Crist, R. M.; Clogston, J. D.; McNeil, S. E.,Anal.Bioanal. Chem. 2015, 407(29), 8661–8672. A) Batch-mode diameter by batch mode DLS of 2-kDa (red), 5-kDa (green), 10-kDa (blue), and 20-kDa (purple) PEGylated gold nanoparticles and of their mixture (dotted black trace). B) AF4-MALS-DLS chromatogram of the mixed sample is reported together with the hydrodynamic diameter (Z-ave, blue line) calculated by batch mode DLS.

36 F. Caputo, et al. Journal of Controlled Release 299 (2019) 31–43

2. The EUNCL/NCI-NCL step by step approach leading to different pharmacological outcome. An educational example was published by Hansen et al. [56] where the authors compared batch

In our opinion, each measurement technique has its advantages and mode DLS and AF4-DLS resolution power by measuring the mixture of disadvantages and no single technique is capable to measure the par- four 30-nm gold nanoparticles covered by various PEG moieties, thus ticle size distribution of all the potential NEP systems under all the differing in their hydrodynamic diameters. The batch-mode measure- conditions which are relevant for pre-clinical studies. Therefore, the ment was incapable to distinguish the single populations in the mixed solution adopted, and here presented, by EUNCL and NCI-NCL is a sample and detected a single peak of a moderate with PdI < 0.17, multi-step approach, composed of three steps of incremental com- which could be misinterpreted as a rather monodispersed sample plexity, and based on the integration of multiple orthogonal measure- composed by only one population, while by AF4-DLS the four popula- ments, as shown in Fig. 1. A pragmatic and robust combination of the tions were all resolved (Fig. 2). The fact that batch mode DLS detected a available assays should be selected on a case by case basis, by con- single unresolved population demonstrated the limited resolution sidering the nature of the NEPs, the expected size range and by bal- achievable by batch mode DLS even in cases of low polydispersity ancing benefits and costs of the analysis. (PdI < 0.2), and thus emphasizes the importance of introducing a se- paration step (e.g. by AF4) that should be combined with DLS to in- dividuate and characterise individual populations within a multimodal 2.1. Step1: Pre-screening sample [56].

As a first step, a fast and simple measurement is performed toverify sample integrity and to detect major failures before engaging in more 2.2. Step 2: Analysis of PSD using high resolution techniques complex and time-consuming analysis. This pre-screening is particu- larly useful also to check sample stability in relevant clinical conditions As a second step, we strongly suggest combining two or more high- (e.g. stability in the clinical buffers, stability vs time, major aggregation resolution orthogonal techniques aiming at (i) resolving the PSD of the in presence of plasma proteins), and eventually to identify the condi- sample (ii) visualizing the particle morphology by EM (e.g. shape, tions that induce a significant aggregation or degradation of NEPs, complex structures etc) and (iii) eventually measure particle con- dramatically changing their PSD [27]. Batch mode DLS, or alternative centration (e.g. by NTA). Our view in this instance is to strongly re- fast and inexpensive techniques (e.g. PTA), may be suitable for the commend performing the analysis of PSD in relevant clinical condi- purpose since they quickly allow you to measure the size range and tions. Multiple high resolution techniques may be suitable for the polydispersity of the NEP population in the sample [27,28]. purpose (see Table 1) and should be selected according to the nature of If the measurement is performed according to a robust SOP the sample (size range, surface charge, known density, particle shape) [27,28,54,55], this approach can be widely adopted by all nanomedi- [31,54,57]. TRPS and NTA can be used to measure particle con- cine developers, in order to verify at every stage of their development centration in addition to PSD but they are only applicable if size is the integrity of their product, its stability and to identify early failures. larger than 30 nm. AF4 separation coupled with downstream DLS and However, especially if batch DLS is used, even if the measured average MALS detectors is a very powerful alternative to measure the PSD size is within the expected range and polydispersity is low (PdI < 0.2) distribution with high resolution over a wide size range, as described in we suggest to proceed to a more complex analysis beyond this pre- [31]. liminary step. In fact, even the simplest case of what appears to be a To illustrate the need for high resolution techniques, two nano- stable NEP population which seems rather monodispersed by batch formulations were subjected to both batch-mode DLS and flow-mode mode DLS can hide a more complex particle size distribution composed DLS (AF4 separation coupled with downstream DLS) for the determi- by multiple populations that could differs in size and shape and thus nation of their PSDs, as illustrated in Fig. 3. The batch-mode DLS results

Fig. 3. Comparison of batch-mode versus flow-mode (AF4 separation with in-line DLS) DLS for dual drug-loaded liposomes (top panel) and loaded micelles (bottom panel). Batch-mode DLS measurements were made at 25 °C after a 100-fold and 10-fold dilution in PBS for the dual drug-loaded liposomes and drug-loaded micelles, respectively. For flow-mode DLS, the z-average was measured across the peaks (green squares) using the Malvern Zetasizer and was based on an intensitythreshold of > 240 kcps and > 200 kcps for the dual drug-loaded liposomes and drug-loaded micelles, respectively. The size ranges are given in the figure for each peak. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

37 F. Caputo, et al. Journal of Controlled Release 299 (2019) 31–43 for dual drug-loaded liposomes (top panel, Fig. 3) shows a poly- also confirmed by TEM analysis. Again, batch mode DLS provided dispersed multimodal size distribution. The flow-mode DLS, after4 AF misleading results, since it was not able to identify the presence of elution optimization, mimics the batch-mode DLS in the sense that two multiple populations in the peptide-NP mixture and only showed one major peaks are observed. However, with AF4-DLS, the size distribu- single population characterized by a broader PSD shifted to larger sizes. tions are determined for each of these peaks. The first peak contains two The capability to synthetize multiple batches with consistent phy- size populations within this peak; a constant size population ~31 nm on sico-chemical properties, including the PSD, is crucial for the successful average (27–35 nm size range) and an increasing size population up to translation of NEPs into the clinic. The NEP developers need to control 100 nm. The second peak contains an increasing size population from the batch to batch variability of their products from the very early 110 to 274 nm. The AF4-DLS results correspond very well with the stages, since significant variation of the size and polydispersity between batch-mode DLS results. However, the size distribution associated with batches can affect their biological effects. The stability of the stockNEP the two peaks are now also known. In the second example (bottom during storage should also be investigated during the early develop- panel, Fig. 3), the batch-mode DLS of the drug-loaded micelles dis- ment phases. In order to detect small changes in PSD, AF4-MALS-DLS, played a single broad peak with multiple size populations contained NTA or TRPS may be suitable options to be used in routine. In mea- within. With flow-mode DLS, these broad overlapping size distributions suring the batch-to-batch variability of NEPs, it is extremely important observed in the batch-mode DLS measurements are now resolved. Two to choose the correct sizing technique, one that is sensitive enough to distinct size populations were observed; the first peak corresponded toa detect small size variations. For example, three lots of PEGylated metal fairly constant size population of 55 nm whereas the second peak oxide nanoparticles were initially assessed for size and morphology by contained an increasing size population up to 235 nm. The results de- TEM. TEM was chosen over DLS in this case as TEM is typically used for monstrate that AF4, as a separation technique, is capable of resolving metallic nanoparticle sizing as well as the technique being able to broad and polydispersed size populations, and when coupled with DLS measure the morphology at the same time. Representative TEM images detection, the size distribution can be measured as well. along with their size summary are shown in Fig. 4. As can be seen by Regulatory agencies recommend to always combine an indirect both the TEM images and size analysis, all three lots look similar. If one method with EM observations [7]. An illustrative example of the were to only perform TEM as the sizing technique, all three lots would powerful combination of AF4-MALS-DLS and TEM is the characterisa- have no batch-to-batch variation. However, further analysis utilizing tion of anionic liposomal formulations reported by Iavicoli et al. [58]. AF4 coupled with MALS detection (an example of a high-resolution Their work demonstrated how the combination of AF4-DLS-MALS and technique) revealed different conclusions. The resulting fractogram is TEM can reveal fine changes in the PSD and morphology of afor- given in Fig. 4 and shows that all three lots consist of a wide range of mulation of anionic liposomes, e.g. when interacting with positively sizes. Moreover, two lots were identical while lot 1 was smaller. Thus, charged antimicrobial peptides. The AF4-MALS analysis of the complex in this case, using AF4-MALS, all three lots were not identical. This formed between the peptides and the liposomes revealed the presence example highlights three significant issues: the importance of selecting of additional populations of particles shifted to larger size values, in the correct sizing technique to control batch to batch variability, the addition to the pristine liposomes (composed by unilamellar vesicles). importance of utilizing orthogonal sizing techniques, and the need for Moreover, the value of the shape ratio (Rg/Rh), suggested the forma- employing at least one high-resolution sizing technique. tion of a multilamellar phase in the larger population of particles, as In another example, three lots of PEGylated metal core-shell

Fig. 4. Assessment of batch to batch variability for three lots of PEGylated metal oxide nanoparticles. Sizing was measured by TEM (top panel) and AF4-MALS (bottom panel).

38 F. Caputo, et al. Journal of Controlled Release 299 (2019) 31–43 nanoparticles were examined by TRPS for batch to batch variability. In size populations above a micron exist, DLS cannot accurately measure this case, both size and zeta potential were measured for each particle sizes on the micron scale as they generally settle to the bottom of a and the results shown in Fig. 5. As shown in the figure, lot 1 had a more cuvette over time, making those particles inaccessible to the laser uniform size and zeta potential distribution compared to the other two without agitation. DLS is based on Brownian motion and particles that lots which were similar. Such information (i.e. per particle) would not are settling cannot be interpreted correctly. With laser diffraction, this be possible by batch-mode DLS and zeta potential measurements. is not the case and agitation is possible. Laser diffraction analysis When there are discrepancies between results obtained by ortho- (middle panel, Fig. 6) showed two size populations; one at approxi- gonal techniques, when it is difficult to resolve the PSD of the sample mately 100 nm, and a second size population at approximately 7.7 μm. (e.g. to differentiate between larger particles and multimers), orifa The laser diffraction size of the smaller population is in good agreement very large size range needs to be covered (e.g. if few very large ag- with the DLS data. Furthermore, laser diffraction was able to confirm gregates are detected), the combination of multiple high resolution the presence of a larger size population, and provide a measured size for techniques (three or more) may be advisable to get a more accurate this population, which DLS could not. A third technique, AF4-DLS- picture. MALS-UV, was used to further asses the presence of aggregates. As this In heterogeneous samples, it could be difficult to understand the represents a high-resolution technique, more information was obtained origin of sample polydispersity. In particular, it may be challenging to compared to the previous two techniques. The AF4 fractogram (bottom understand if population of larger sizes are composed by larger single panel, Fig. 6) showed three populations based on the UV, MALS, and particles or by small aggregates of the main population. The capability DLS signals, namely free protein, lipid-protein nano-vesicles, and lipid of different techniques to recognize the presence of dimer and multi- aggregates. As demonstrated by this example, multiple orthogonal mers was investigated by Mehn et al. [20], by considering the potential techniques are needed to better assess the particle size distribution of a of single and combined complementary sizing techniques. In this work, NEP. Moreover, the higher the resolution of the technique, the more a bimodal sample of polystyrene particles was analyzed by AUC, DCS, information and understanding of the NEP can be obtained.

NTA, AF4-MALS-DLS, TRPS and cryo-TEM. The presence of dimers and multimers was unquestionably proven by the combination of TEM 2.3. Step 3: Stability of particle size in biological media imaging after FFF separation. Interestingly, AF4-MALS-DLS, and cen- trifugal methods including AUC and DCS were able to give indirect As recently reported by the FDA [7], after entering into the systemic information on particle shape, suggesting the presence of dimers made circulation, the bound plasma proteins may endow nanomaterials with of rigid spherical nanoparticles instead of larger particles. new biological properties, including the modification of the PSD, which The detection of a small amounts of very large aggregates (> 1 μm) may affect NEP safety and efficacy profile and, thus, should becon- in a NEP mixture is another nontrivial task, where the use of multiple sidered a quality attribute. Interaction of the NEPs with plasma proteins orthogonal techniques is highly recommended in order to cover a wide may induce various effects on the particle size, depending onthe size range. To illustrate this point, multiple orthogonal sizing techni- properties of the NEP system, including the increase of the particle size ques were employed to detect the presence of large aggregates in a of a few nanometres due to the formation of a protein corona on the lipid-protein nanovesicle formulation, as reported by Grossman et al. particle surface, or more drastic size changes due to aggregation, dis- [12]. Results for batch-mode DLS, laser diffraction, and AF coupled 4 aggregation or destabilization of the particles. Therefore, as a third with MALS, UV, and DLS, are presented in Fig. 6. By batch-mode DLS step, our suggestion is the investigation of the effects of the interaction (top panel, Fig. 6), a very broad size distribution ranging from ap- of NEP with plasma proteins on particle size. proximately 20 nm to over a micron, was observed, with 87 nm being The reader should be aware that, since every NEP is different, the the major peak size. Although the broad size distribution indicates that range of PSD for safety and efficacy of each formulation will vary

Fig. 5. Assessment of batch to batch variability for three lots of PEGylated metal core-shell nanoparticles. Sizing and zeta potential were measured by TRPS.

39 F. Caputo, et al. Journal of Controlled Release 299 (2019) 31–43

Fig. 6. Summary of the orthogonal sizing techniques used to assess the presence of large aggregates in a lipid-protein nanovesicle formulation. Adapted by J.H.

Grossman, R.M. Crist, J.D. Clogston, AAPS Journal. 19 (2017) 92–102 [48]. The techniques employed were dynamic light scattering (DLS), laser diffraction, and AF4 coupled with MALS, UV, and inline DLS detectors. Note that the aggregates were clearly observed in the laser diffraction and AF4 results and not in the batch-mode DLS results as it is not suitable for > 1 μm particles.

40 F. Caputo, et al. Journal of Controlled Release 299 (2019) 31–43 depending on the nature of the NEP and on its specific clinical appli- specific cases to study aggregation and drug release. However, inthe cation. Is not the scope of this article trying to specify the appropriate/ presence of plasma adsorbed to the particle surface the density of the allowable PSD range for each NEP system. The main aim of this section particles may change (especially in the case of inorganic particles) and is to discuss the EUNCL and NCI-NCL strategy to assess fine or major needs to be measured independently. changes in PSD of NEPs in complex media. The reader should also be aware that other critical quality attributes should be assessed in com- 2.4. SOPs and quality control plex media, including the chemical stability of the particles, their de- gradation products and, in presence of an API, the drug release profile. In order to perform robust characterisation in line with the reg- To further explore those aspects, the reader could refer to the following ulatory needs [7] the methodologies and techniques to adopt should be references [3, 6, 7, 12, 12, 15, 46, 59–61]. selected considering their overall applicability and suitability, in order The sizing analysis in complex media requires the use of high-re- to provide robust and reliable data. The techniques chosen should be solution techniques due to the complexity of the nanoparticle and able to measure the size range of interest and the sample should not be protein mixture. Two illustrative examples that discourage the use of unintentionally altered/modified/transformed by the measurement the low-resolution batch mode DLS, and propose the AF -DLS-MALS as 4 (see Table 1). Thus, the sampling of NEP should be performed in a a suitable alternative are reported below. representative state of the process stage being evaluated, and the Gioria et al. [3] measured the PSD of a Doxorubicin liposomal for- sample preparation should be adequately controlled to ensure that the mulation before and after incubation with plasma by batch mode DLS process do not substantially change its properties [7]. Importantly, and AF -DLS-MALS for comparison purposes. According to AF -DLS- 4 4 suitable quality controls should be applied to verify the working con- MALS the average particle size of the liposomal formulation was not dition of the instruments, the calibration status and also the detection modified by the presence of plasma proteins, confirming theknown limits of the selected techniques. For example, in case of samples stability of PEGylated liposomal samples in physiological conditions. composed of multiple populations, specific pseudo-polydispersed con- On the contrary, the batch mode DLS analysis was biased toward trol samples (e.g. artificial mixtures of polystyrene control particles in smaller size simply because the instrument was not able to resolve the the same size range) could be tested to evaluate capacity of the method signal of the serum proteins (smaller size) from the contribution of the of choice to resolve multiple populations in a specific size range. The liposomal population. The study showed that AF -MALS DLS, due to the 4 use of validated methods and of SOPs with a defined specific mea- fractionation step performed before the detection, can often separate surement purpose is strongly recommend, and to this end the SOPs from serum proteins from the NEP populations, and therefore is an inter- EUNCL & NCI-NCL are freely available to be adopted and used at esting option to detect and quantify the small changes in the PSD after http://www.euncl.eu/about-us/assay-cascade/ & at https://ncl.cancer. the incubation of NEPs with plasma. gov/resources/assay-cascade-protocols). The protein layer absorbed on the particle surface cannot be de- tected by TEM. However, if major changes in PSD are detected by in- direct measurements after incubation with serum proteins, e.g. due to 3. Conclusions and perspectives particle degradation, the changes in the particle physical properties may be visualized by electron microscopy. Caputo et al. have recently To the best of our knowledge, there is not a single sizing technique compared the potential of batch mode DLS, TEM and AF4- DLS to detect which can resolve and provide all the relevant information needed for fine modification of lipid-based nanoparticle size in the presence of pre-clinical studies. Batch mode DLS is suitable to detect major issues plasma proteins [62]. Interestingly, by AF4- MALS and the TEM ana- with sample integrity and stability at early stages, however it is hardly lyses the authors were able to measure a fine change in particle size applicable as the only method to characterise the size and polydispersity (PSD narrower and shifted to smaller sizes), while batch mode DLS was of complex NEP. EM is the only direct technique that allows to visualize not able to resolve any difference (Fig. 7). polydispersed samples over a wide size range, and thus providing direct Other high resolution techniques, including centrifugation techni- shape information. However, limitations do exist since the sample pre- ques, NTA or TRPS may be useful to investigate the stability of NEP in paration is often critical and may introduce artifacts and cannot prevent biological media that contains plasma proteins. For example, AUC and sample aggregation. Furthermore, imaging observation, recording and DCS may give very informative results on both major changes of the data analysis can be prolonged and user-biased. This makes EM difficult sedimentation coefficient (and thus indirectly on particle size) andof to be adopted as a technique for routine checks. NTA, TRPS, DCS and drug release in presence of plasma proteins with one single measure- AUC are cost-effective and easy to use alternatives which can provide ment [3]. Therefore, they may be interesting techniques in some satisfactory resolution for polydispersed samples e.g. for batch to batch variability and stability checks. However, the centrifugal techniques are

Fig. 7. Particle size distribution changes of lipid-based nanoparticles, without and with 10% plasma. Adapted from [49]: A) Batch mode DLS B) AF4-DLS and C) TEM via negative staining of Lipidots™ in phosphate buffer with and without 10% plasma (red and black curves respectively). (For interpretation of the referencesto colour in this figure legend, the reader is referred to the web version of this article.)

41 F. Caputo, et al. Journal of Controlled Release 299 (2019) 31–43 only applicable when the density of the particles is known, while NTA is whole or in part with Federal funds from the National Cancer Institute, not suitable for small organic particles below 30 nm. AF4-DLS-MALS is National Institutes of Health, under Contract No. another interesting alternative to analyze NEP in routine and to detect HHSN261200800001E. The content of this publication does not ne- changes in PSD caused by interactions with plasma. Unfortunately, cessarily reflect the views or policies of the Department of Health and method development may be time consuming, and some samples may Human Services, nor does mention of trade names, commercial pro- suffer from irreversible loss on the membrane in the fractionation ducts, or organizations imply endorsement by the U.S. Government. channel. The solution presented by EUNCL and NCI-NCL is a incremental multi-step approach which consists of a quick preliminary step to verify Conflict of interest disclosure sample integrity and stability by low resolution techniques (pre- screening), followed by the combination of complementary high re- The authors declare that the research was conducted in the absence solution sizing measurements performed in simple buffers and complex of any commercial or financial relationships that could be construed as biological environments, which mimics the conditions encountered by a potential conflict of interest. NEPs during their clinical administration (e.g. reconstitution buffer, be- havior in plasma). We have demonstrated the need for employing at least Authors' contributions and disclaimer one high-resolution sizing technique, the importance of selecting the correct sizing techniques for the purpose, and the robustness of utilizing All the authors participated in the development of the step by step orthogonal sizing techniques to study the physical properties of the characterisation approach described, contributed to the scientific dis- complex samples. The use of orthogonal techniques allows to measure cussion, contributed to the writing of the article, and approved the the particle size distribution of complex polydispersed materials, which manuscript. JC and FC performed the experiments reported in the case are very common in the nanomedicine and pharmaceutical industry. In studies and performed the data analysis. The views expressed in this this context, we have demonstrated that the combination of EM, AF4- article are those of the individual co-authors and do not necessarily DLS-MALS, AUC, TRPS is able to (i) discriminate between populations of represent official views of the European Commission. particles characterized by different sizes and shapes (ii) distinguish larger particles from small aggregates, (iii) resolve small changes of PSD caused References by a not reproducible synthetic procedures (batch to batch variability) and/or by NEP instability upon storage and (iv) study the influence of [1] S.R. D'Mello, C.N. Cruz, M.-L. Chen, M. Kapoor, S.L. Lee, K.M. Tyner, The evolving biological environments to average particle size and polydispersity, landscape of drug products containing nanomaterials in the United States, Nat. Nanotechnol. 12 (2017) 523–529, https://doi.org/10.1038/nnano.2017.67. which are crucial to understand the biological effects and physiological [2] J.-B. Coty, C. Vauthier, Characterization of nanomedicines: a reflection on a field stability of NEP in vitro and in vivo. Other orthogonal sizing techniques, under construction needed for clinical translation success, J. Control. Release 275 including SAXS/WAXS, AFM, sp-ICP-MS may also be applicable ac- (2018) 254–268, https://doi.org/10.1016/j.jconrel.2018.02.013. [3] S. Gioria, F. Caputo, P. Urbán, C.M. Maguire, S. Bremer-Hoffmann, A. Prina-Mello, cording to the nature of the NEPs. The choice of the specific techniques L. Calzolai, D. Mehn, Are existing standard methods suitable for the evaluation of to be adopted for each NEP should be defined on a case by case basis nanomedicines: some case studies, Nanomedicine 13 (2018) 539–554, https://doi. according to the nature of the sample and of the complexity of its PSD. org/10.2217/nnm-2017-0338. However, according to the authors, by adopting a characterisation ap- [4] B. Halamoda-Kenzaoui, U. Holzwarth, G. Roebben, A. Bogni, S. Bremer-Hoffmann, Mapping of the available standards against the regulatory needs for nanomedicines, proach of incremental complexity and by using orthogonal methods from Wiley Interdisc. Rev. 0 (n.d.) e1531. doi: https://doi.org/10.1002/wnan.1531. the early phases of product development, a robust sizing characterisation [5] S. Bremer-Hoffmann, B. Halamoda-Kenzaoui, S.E. Borgos, Identification ofreg- helps to identify the physical property of the product and eventually to ulatory needs for nanomedicines: regulatory needs for nanomedicines, J. Interdisc. Nanomedicine 3 (2018) 4–15, https://doi.org/10.1002/jin2.34. detect early fails, thus reducing the cost of development and speeding up [6] R.M. Crist, J.H. Grossman, A.K. Patri, S.T. Stern, M.A. Dobrovolskaia, the translation of the promising NEP into the clinic. It is encouraging to P.P. Adiseshaiah, J.D. Clogston, S.E. McNeil, Common pitfalls in nanotechnology: see a mind-change from the experts and regulators which are starting to lessons learned from NCI's nanotechnology characterization laboratory, Integr. Biol. 5 (2013) 66–73, https://doi.org/10.1039/C2IB20117H. support the orthogonal approach for the physico-chemical characterisa- [7] A. FDA/CDER draft guidance, Drug Products, Including Biological Products, that tion of NEPs. In the recent years, thanks to commitment, efforts and tight Contain Nanomaterials - Guidance for Industry, (2017) https://www.fda.gov/ collaboration of experts, regulatory agencies and metrology institutes, downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ UCM588857.pdf accessed February 14, 2019 https://doi.org/10.5694/mja17. there has been an increase in the number of orthogonal techniques, in- 00712. cluding NTA, cryo-TEM and AF4 which have been standardized by ISO [8] Joint MHLW-EMA Reflection Paper on the Development of Block Copolymer and ASTM International, and thus introduced for quality controls (QC) Micelle Medicinal Products. Reference Number EMA/CHMP/130299/2013, https://www.ema.europa.eu/development-block-copolymer-micelle-medicinal- and acceptance (QA) across the stakeholders [4]. Nonetheless, these are products , Accessed date: 14 February 2019. impacting the R&D in terms of manufacturing and QC/QA costs and thus [9] Reflection Paper on the Data Requirements for Intravenous Iron-Based Nano- forcing industry to often adjust to their limited financial levels and thus Colloidal Products Developed with Reference to an Innovator Medicinal Product. to rely, or adopt, limited characterisation steps. This has been, for in- EMA Reference Number EMA/CHMP/SWP/620008/2012, https://www.ema. europa.eu/en/data-requirements-intravenous-iron-based-nano-colloidal-products- stance, the case for the use of batch mode DLS as the only sizing mea- developed-reference-innovator , Accessed date: 14 February 2019. surement. The extensive and comprehensive work carried out by EUNCL [10] Reflection Paper on the Data Requirements for Intravenous Liposomal Products and NCI-NCL, not only support knowledge sharing and open access for Developed with Reference to an Innovator Liposomal Product. EMA/CHMP/ 806058/2009/Rev. 02, https://www.ema.europa.eu/documents/scientific- SOPs and protocols, but also provides a platform for educating the guideline/reflection-paper-data-requirements-intravenous-liposomal-products- community and offering free of charge and unbiased access to their state developed-reference-innovator_en-0.pdf , Accessed date: 14 February 2019. of the art characterisation services. In this matter, we believe, the use of [11] FDA Guidance for Industry, Liposome Drug Products: Chemistry, Manufacturing, and Controls; Human Pharmacokinetics and Bioavailability; and Labeling incremental complexity orthogonal techniques for the characterisation of Documentation, https://www.fda.gov/downloads/drugs/guidances/ucm070570. NEPs will have a positive impact in terms of scale-up effectiveness, in- pdf, (2018) , Accessed date: 14 February 2019. dustrialization cost, and time to market for any current and future na- [12] J.H. Grossman, R.M. Crist, J.D. Clogston, Early development challenges for drug products containing nanomaterials, AAPS J. 19 (2017) 92–102, https://doi.org/10. notechnology-based medicinal products. 1208/s12248-016-9980-4. [13] J.H. Grossman, S.E. McNeil, Nanotechnology in cancer medicine, Phys. Today 65 Acknowledgments (2012) 38–42, https://doi.org/10.1063/PT.3.1678. [14] T.P.J. Linsinger, G. Roebben, D. Gilliland, L. Calzolai, F. Rossi, N. Gibson, C. Klein, Institute for reference materials and measurements, requirements on measurements This research was in part funded by the project EUNCL 2015, a for the implementation of the European Commission definition of the term “nano- European Union's Horizon 2020 Research and Innovation Program material,”, Publications Office, Luxembourg, 2012 http://dx.publications.europa. under grant agreement No 654190. This project has been funded in eu/10.2787/63490 accessed August 17, 2018.

42 F. Caputo, et al. Journal of Controlled Release 299 (2019) 31–43

[15] Importance of physicochemical characterization prior to immunological studies | [38] L. Calzolai, D. Gilliland, F. Rossi, Measuring nanoparticles size distribution in food Handbook of Immunological Properties of Engineered Nanomaterials, (n.d.). and consumer products: a review, Food Addit. Contam. 29 (2012) 1183–1193, https://www.worldscientific.com/doi/abs/10.1142/9789813140431_0002 (ac- https://doi.org/10.1080/19440049.2012.689777. cessed February 14, 2019). [39] D.J. Scott, S.E. Harding, A.J. Rowe, Royal Society of Chemistry (Great Britain), [16] Physicochemical Characterization of Polymer Nanoparticles: Challenges and Analytical Ultracentrifugation: Techniques and Methods, RSC Pub, Cambridge, UK, Present Limitations | SpringerLink, (n.d.). https://link.springer.com/chapter/10. 2005, https://doi.org/10.1080/13577140500349717. 1007/978-3-319-41421-8_7 (accessed February 14, 2019). [40] P. Schuck, H. Zhao, C. Brautigam, R. Ghirlando, Basic Principles of Analytical [17] Measuring the Hydrodynamic Size of Nanoparticles in Aqueous Media Using Batch- Ultracentrifugation, CRC Press, 2015, https://doi.org/10.1201/b19028. Mode Dynamic Light Scattering, SpringerLink, 2010, https://link.springer.com/ [41] K.L. Planken, H. Cölfen, Analytical ultracentrifugation of colloids, Nanoscale 2 protocol/10.1007%2F978-1-60327-198-1_4 , Accessed date: 14 February 2019. (2010) 1849, https://doi.org/10.1039/c0nr00215a. [18] W.H. De Jong, W.I. Hagens, P. Krystek, M.C. Burger, A.J.A.M. Sips, R.E. Geertsma, [42] P. Schuck, Sedimentation Velocity Analytical Ultracentrifugation, CRC Press, 2016, Particle size-dependent organ distribution of gold nanoparticles after intravenous https://doi.org/10.1590/0001-376520168833. administration, Biomaterials 29 (2008) 1912–1919, https://doi.org/10.1016/j. [43] D. Mehn, I.M. Rio-Echevarria, D. Gilliland, M. Kaiser, K. Vilsmeier, P. Schuck, biomaterials.2007.12.037. W. Wohlleben, Identification of nanomaterials: a validation report of two labora- [19] S. Mitragotri, J. Lahann, Physical approaches to biomaterial design, Nat. Mater. 8 tories using analytical ultracentrifugation with fixed and ramped speed options, (2009) 15–23, https://doi.org/10.1038/nmat2344. NanoImpact 10 (2018) 87–96, https://doi.org/10.1016/j.impact.2017.12.005. [20] D. Mehn, F. Caputo, M. Rösslein, L. Calzolai, F. Saint-Antonin, T. Courant, P. Wick, [44] J. Ma, H. Zhao, J. Sandmaier, J. Alexander Liddle, P. Schuck, Variable field ana- D. Gilliland, Larger or more? Nanoparticle characterisation methods for recognition lytical ultracentrifugation: II. gravitational sweep sedimentation velocity, Biophys. of dimers, RSC Adv. 7 (2017) 27747–27754, https://doi.org/10.1039/ J. 110 (2016) 103–112, https://doi.org/10.1016/j.bpj.2015.11.027. C7RA02432K. [45] R. Capomaccio, I. Ojea Jimenez, P. Colpo, D. Gilliland, G. Ceccone, F. Rossi, [21] F. Varenne, A. Makky, M. Gaucher-Delmas, F. Violleau, C. Vauthier, Multimodal L. Calzolai, Determination of the structure and morphology of gold dispersion of nanoparticles: a comprehensive evaluation of size distribution with 9 nanoparticle–HSA protein complexes, Nanoscale 7 (2015) 17653–17657, https:// size measurement methods, Pharm. Res. 33 (2016) 1220–1234, https://doi.org/10. doi.org/10.1039/C5NR05147A. 1007/s11095-016-1867-7. [46] D. Mehn, P. Iavicoli, N. Cabaleiro, S.E. Borgos, F. Caputo, O. Geiss, L. Calzolai, [22] C.M. Maguire, M. Rösslein, P. Wick, A. Prina-Mello, Characterisation of particles in F. Rossi, D. Gilliland, Analytical ultracentrifugation for analysis of doxorubicin solution – a perspective on light scattering and comparative technologies, Sci. loaded liposomes, Int. J. Pharm. 523 (2017) 320–326, https://doi.org/10.1016/j. Technol. Adv. Mater. 19 (2018) 732–745, https://doi.org/10.1080/14686996. ijpharm.2017.03.046. 2018.1517587. [47] E. Weatherall, G.R. Willmott, Applications of tunable resistive pulse sensing, [23] B.R. Pauw, C. Kästner, A.F. Thünemann, Nanoparticle size distribution quantifica- Analyst 140 (2015) 3318–3334, https://doi.org/10.1039/C4AN02270J. tion: results of a small-angle X-ray scattering inter-laboratory comparison, J. Appl. [48] P.L. Stewart, Cryo-electron microscopy and cryo-electron tomography of nano- Crystallogr. 50 (2017) 1280–1288, https://doi.org/10.1107/S160057671701010X. particles: Cryo-electron microscopy and cryo-electron tomography of nanoparticles, [24] Small Angle X-Ray Scattering for Nanoparticle Research - Chemical Reviews (ACS Wiley Interdisc. Rev. 9 (2017) e1417, https://doi.org/10.1002/wnan.1417. Publications), (n.d.). https://pubs.acs.org/doi/10.1021/acs.chemrev.5b00690 (ac- [49] S.B. Rice, C. Chan, S.C. Brown, P. Eschbach, L. Han, D.S. Ensor, A.B. Stefaniak, cessed February 14, 2019). J. Bonevich, A.E. Vladár, A.R.H. Walker, J. Zheng, C. Starnes, A. Stromberg, J. Ye, [25] J. Grobelny, F.W. DelRio, N. Pradeep, D.-I. Kim, V.A. Hackley, R.F. Cook, Size E.A. Grulke, Particle size distributions by transmission electron microscopy: an measurement of nanoparticles using atomic force microscopy, in: S.E. McNeil (Ed.), interlaboratory comparison case study, Metrologia 50 (2013) 663–678, https://doi. Characterization of Nanoparticles Intended for Drug Delivery, Humana Press, org/10.1088/0026-1394/50/6/663. Totowa, NJ, 2011, pp. 71–82, https://doi.org/10.1007/978-1-60327-198-1_7. [50] V. Klang, C. Valenta, N.B. Matsko, Electron microscopy of pharmaceutical systems, [26] M.D. Montaño, J.W. Olesik, A.G. Barber, K. Challis, J.F. Ranville, Single particle Micron 44 (2013) 45–74, https://doi.org/10.1016/j.micron.2012.07.008. ICP-MS: advances toward routine analysis of nanomaterials, Anal. Bioanal. Chem. [51] J. Kuntsche, J.C. Horst, H. Bunjes, Cryogenic transmission electron microscopy 408 (2016) 5053–5074, https://doi.org/10.1007/s00216-016-9676-8. (cryo-TEM) for studying the morphology of colloidal drug delivery systems, Int. J. [27] L. Calzolai, EUNCL SOP PCC-021 measuring the NP stability against aggregation as Pharm. 417 (2011) 120–137, https://doi.org/10.1016/j.ijpharm.2011.02.001. a function of time and buffers with batch-mode DLS, http://www.euncl.eu/about- [52] V. Klang, N.B. Matsko, C. Valenta, F. Hofer, Electron microscopy of nanoemulsions: us/assay-cascade/PDFs/PCC/EUNCL-PCC-021.pdf?m=1468937870 , Accessed an essential tool for characterisation and stability assessment, Micron 43 (2012) date: 14 February 2019. 85–103, https://doi.org/10.1016/j.micron.2011.07.014. [28] L. Calzolai, et al., EUNCL SOP PCC-001. Measuring the Size of Nanoparticles Using [53] L.E. Franken, E.J. Boekema, M.C.A. Stuart, Transmission electron microscopy as a Batch Mode Dynamic Light Scattering, http://www.euncl.eu/about-us/assay- tool for the characterization of soft materials: application and interpretation, Adv. cascade/PDFs/Prescreening/EUNCL-PCC-001.pdf?m=1468937875& , Accessed Sci. 4 (2017) 1600476, https://doi.org/10.1002/advs.201600476. date: 14 February 2019. [54] C. Maguire et al. EUNCLPCC-023. Measuring the Size and Concentration of [29] L. Calzolai, D. Gilliland, C.P. Garcìa, F. Rossi, Separation and characterization of Nanoparticles Using Particle Tracking Analysis (PTA). http://www.euncl.eu/about- gold nanoparticle mixtures by flow-field-flow fractionation, J. Chromatogr. A1218 us/assay-cascade/PDFs/PCC/EUNCL_PCC_023.pdf?m=1526712237&. (accessed (2011) 4234–4239, https://doi.org/10.1016/j.chroma.2011.01.017. February 14, 2019), DOI: https://doi.org/10.1007/s11136-019-02122-y. [30] N. Hondow, R. Brydson, P. Wang, M.D. Holton, M.R. Brown, P. Rees, H.D. Summers, [55] V.A. Hackley, Measuring the Size of Nanoparticles in Aqueous Media Using Batch- A. Brown, Quantitative characterization of nanoparticle agglomeration within Mode Dynamic Light Scattering, National Institute of Standards and Technology, biological media, J. Nanopart. Res. 14 (2012), https://doi.org/10.1007/s11051- (2015), https://doi.org/10.6028/NIST.SP.1200-6. 012-0977-3. [56] M. Hansen, M.C. Smith, R.M. Crist, J.D. Clogston, S.E. McNeil, Analyzing the in- [31] D. Mehn et al. EUNCL SOP PCC-22. Measurement of Particle Size Distribution of fluence of PEG molecular weight on the separation of PEGylated gold nanoparticles Protein Binding, of Mean Molecular Weight of Polymeric NP Components, Study of by asymmetric-flow field-flow fractionation, Anal. Bioanal. Chem. 407 (2015) Batch to Batch Reproducibility, and Study of Release of Free Coating from NP 8661–8672, https://doi.org/10.1007/s00216-015-9056-9. Surface by FFF- MALS. http://www.euncl.eu/about-us/assay-cascade/PDFs/PCC/ [57] F. Caputo, EUNCL SOP PCC-020: Cryo-TEM. http://www.euncl.eu/about-us/assay- EUNCL-PCC-022.pdf?m=1468937868&. (accessed February 14, 2019), DOI: cascade/PDFs/PCC/EUNCL-PCC-020.pdf?m=1468937869&. (accessed February https://doi.org/10.1016/j.ijpharm.2017.03.046. 14, 2019), DOI: https://doi.org/10.1002/hep.30574. [32] M. Wagner, S. Holzschuh, A. Traeger, A. Fahr, U.S. Schubert, Asymmetric flow field- [58] P. Iavicoli, P. Urbán, A. Bella, M.G. Ryadnov, F. Rossi, L. Calzolai, Application of flow fractionation in the field of nanomedicine, Anal. Chem. 86 (2014) 5201–5210, asymmetric flow field-flow fractionation hyphenations for liposome–antimicrobial https://doi.org/10.1021/ac501664t. peptide interaction, J. Chromatogr. A 1422 (2015) 260–269, https://doi.org/10. [33] G. Yohannes, M. Jussila, K. Hartonen, M.-L. Riekkola, Asymmetrical flow field-flow 1016/j.chroma.2015.10.029. fractionation technique for separation and characterization of biopolymers and [59] S.T. Stern, M.N. Martinez, D.M. Stevens, When is it important to measure unbound bioparticles, J. Chromatogr. A 1218 (2011) 4104–4116, https://doi.org/10.1016/j. drug in evaluating nanomedicine pharmacokinetics? Drug Metab. Dispos. 44 (2016) chroma.2010.12.110. 1934–1939, https://doi.org/10.1124/dmd.116.073148. [34] C. Contado, Field flow fractionation techniques to explore the “nano-world”, Anal. [60] S. Skoczen, S.E. McNeil, S.T. Stern, Stable isotope method to measure drug release Bioanal. Chem. 409 (2017) 2501–2518, https://doi.org/10.1007/s00216-017- from nanomedicines, J. Control. Release 220 (2015) 169–174, https://doi.org/10. 0180-6. 1016/j.jconrel.2015.10.042. [35] S.K.R. Williams, K. Caldwell, Field-Flow Fractionation in Biopolymer Analysis, [61] Characterization Methods: Physical and Chemical Characterization Techniques - Springer, Vienna; London, 2012. Pharmaceutical Nanotechnology: Innovation and Production - Wiley Online [36] A. Zattoni, B. Roda, F. Borghi, V. Marassi, P. Reschiglian, Flow field-flow fractio- Library, (n.d.). https://onlinelibrary.wiley.com/doi/pdf/10.1002/ nation for the analysis of nanoparticles used in drug delivery, J. Pharm. Biomed. 9783527800681.ch7 (accessed February 14, 2019). Anal. 87 (2014) 53–61, https://doi.org/10.1016/j.jpba.2013.08.018. [62] F. Caputo, A. Arnould, M. Bacia, W.L. Ling, E. Rustique, I. Texier, A. Prina-Mello, A.- [37] S. Podzimek, Light Scattering, Size Exclusion Chromatography and Asymmetric C. Couffin, Measuring particle size distribution by asymmetric flow fieldflow Flow Field Flow Fractionation: Powerful Tools for the Characterization of Polymers, fractionation: a powerful method for the pre-clinical characterisation of lipid-based Proteins and Nanoparticles, John Wiley & Sons, Inc., Hoboken, NJ, USA, 2011, nanoparticles, Mol. Pharm. (2019), https://doi.org/10.1021/acs.molpharmaceut. https://doi.org/10.1002/9780470877975. 8b01033.

43