Unravelling the polysorbate 20 composition: A fusion of UPLC-MS analysis and stochastic modelling

Abstract

Polysorbate 20 (PS20) is one of the most commonly used non-ionic surfactants in cosmetics, pharmaceuticals and food products. Considered as biocompatible and non-irritating, it is further valued for its solubilising and protein stabilising properties. PS20 is manufactured through a multi-stage reaction of sorbitol with various fatty acids and ethylene oxide, resulting in a complex mixture of components with different molecular weights and polarity. Since variations in the distribution of these components can influence its performance, such as the emulsifying or solubilising efficiency, a detailed understanding of the PS20 composition is of importance.

Herein we introduce a combined approach of reversed-phase chromatography with mass detection and automated stochastic modelling that enables the quantitative characterisation of PS20 at the component level. With two straightforward sample preparations and two methods for an ultra-high performance liquid chromatography (UPLC) system coupled to a single quadrupole mass (QDa) detector, this technique ensures efficient data acquisition. Seven PS20 products of different manufacturers, age and qualities were studied using the presented approach. Molar contents and weight percentages were calculated for each of the more than 27′700 components of the PS20, which were fully characterised by i) the substance class (i.e. sorbitan, isosorbide or polyoxyethylene (POE)), ii) the number of esters, iii) the fatty acid combination and iv) the number of OE units. The obtained results allowed not only an accurate prediction of bulk parameters, such as hydroxyl and saponification values, but also a detailed product comparison.

Introduction

Polysorbates belong to the most frequently used excipients in pharmaceutical and biopharmaceutical products. They are administered by various routes, including peroral, subcutaneous, intravenous, ophthalmic and topical [[1], [2], [3], [4]]. In addition, polysorbates are utilised in cosmetics and food products, where polysorbate 20 is assigned the E number E 432 in accordance with European regulations [[5], [6], [7]]. As of April 2025, the FDA’s Inactive Ingredient Database (IID) listed 129 entries for polysorbates, 29 of them specifically for PS20 [8]. In fact, the IID is not exhaustive, as Mieczkowski (2023) listed 40 antibody formulations containing PS20 [9]. Moreover, Wuchner et al. (2022) showed that the use of PS20 in antibody formulations is further increasing [10]. In a survey of 16 globally acting major biotechnology companies, all stated their intention to use PS20 for antibody formulations in the future [10].

This increasing trend is mainly attributable to the multifunctionality of PS20, which serves multiple purposes across various formulations. In biopharmaceutical formulations it acts as stabiliser to protect proteins and peptides from particle formation or denaturation during storage, processing or shipment [11]. Although the exact mechanisms are not fully understood, surfactants likely act stabilising against interfacial stresses at various interfaces encountered during drug product manufacturing (e.g., air–liquid, glass-liquid) [[12], [13], [14]].

Furthermore, PS20 is an important solubiliser for dissolving poorly water-soluble drugs in aqueous media. For this purpose, typical concentrations of 1 % to 5 % are used with a linear correlation of polysorbate concentration and solubility [15]. A widespread, traditional use of PS20 is as hydrophilic emulsifiers for the stabilisation of lotions and creams. An additional role in semi-solid dosage forms is its participation in the formation of liquid crystalline structures. These are formed from hydrophilic surfactants such as polysorbates and consistency agents such as fatty alcohols at a distinct ratio and are responsible for building up the viscosity of creams [3].

The wide-ranging functionality of PS20 stems from its complex composition, which is the result of a multi-step manufacturing process [16,17]. The synthesis begins e.g. with the acid-catalysed dehydration of sorbitol leading in the formation of sorbitan and isosorbide. In the subsequent step, the hydroxyl groups of both compounds are esterified by fatty acids in numerous combinations. In the final oxyethylation step, ethylene oxide (OE) chains of various lengths are incorporated between the fatty acid residues and the sugar backbone by rapid transesterification [[18], [19], [20]]. Thus, several factors such as the ethoxylated core structure (sorbitan, isosorbide or polyoxyethylene (POE)), the type of fatty acids, number of esters and number of OE units contribute to the heterogeneity of PS20 [21]. Further variations introduced by different manufacturers and batch-to-batch variability result in a complex mixture that cannot be linked to a well-defined composition [16]. For these reasons, it is not surprising that polyoxyethylene sorbitan monolaurates, which formally represent the predominant ester components of PS20, were previously found to constitute only 20 % by weight of the overall mixture [22].

Constituents of PS20 cover a broad spectrum of polarity, suggesting that different species may exhibit different physicochemical properties and functions in formulations. Nonetheless, research in this area is still scarce. In a study on protein stabilisation effects of PS20, it was demonstrated that the complete material as well as the isosorbide-POE-monolaurate fraction had superior stabilisation efficiency compared to the sorbitan-POE-dilaurate and sorbitan-POE-trilaurate fractions. However, the data does not yet allow generalised conclusions [23]. Another potential difference between PS20 components is their respective HLB (hydrophilic-lipophilic balance) value, which may result in different solubilisation and emulsifying properties. In the HLB system, first introduced by Griffin (1954) [24], each emulsifier is assigned a dimensionless numerical parameter, with a high number describing a more hydrophilic character and a low number suggesting a more lipophilic nature. Since there is no universally accepted method for determining this empirical parameter, reported HLB values for the same molecule can differ considerably, depending on the underlying approach [25,26]. The Handbook of Pharmaceutical Excipients states an HLB value of 16.7 for PS20 and further lists HLB values of 15.6, 14.9 and 15.0 for palmitate, stearate and oleate polyoxyethylene (20) sorbitan monoesters as well as 10.5 and 11.0 for the stearate and oleate polyoxyethylene (20) sorbitan triesters. While substitution of the fatty acid has little influence on the HLB value, it differs by approx. 4 to 5 units between mono- and triester species of the same fatty acid [27].

In view of these polarity differences, a detailed insight into the composition of the raw material appears to be advantageous with regards to its solubilisation and emulsifying properties, especially in the stage of formulation development.

Analytical methods able to characterise different types of PS20 subspecies are essential in the biopharmaceutical industry [16]. Two main objectives are pursued with the analytical methods available, namely the determination of the content of intact PS20 or its components and the analysis of degradation products [10]. A further subdivision can be made according to the type of the sample origin, i.e. either the raw material, or a PS20 containing formulation. The most widely adopted analytical characterisation approach involves (ultra-) high performance liquid chromatography. As PS20 has no significant absorption in the UV–Vis range, detection techniques such as charged aerosol detection (CAD), evaporative light scattering detection (ELSD) or mass detection (MS) have been used [28,29]. The first mentioned detectors typically show a non-linear response and are limited by insufficient specificity for individual PS20 constituents. MS-based methods, on the other hand, allow the detection of in-source fragments common to a particular component group or of selected PS20 components. Characteristic dioxolanylium ions, for example, are generated by collision-induced dissociation (CID) from fatty acid residues of esterified PS20 components and were repeatedly used as diagnostic ions. On this basis, Borisov et al. (2011) presented a semi-quantitative approach capable of profiling the distribution of predominant fatty acids in crude PS20 and PS80 materials, suitable for preliminary batch-to-batch comparisons [30]. Kopf et al. (2023), on the other hand, demonstrated the suitability of these ions for quantifying the content of intact or degraded PS20 or PS80 in protein formulations in relation to the raw material used [31]. A similar field of application for PS20 is addressed by our previous validated method, which in contrast monitors selected species, the so-called markers [21]. Even though these examples illustrate useful analytical tools, no holistic approach has yet been developed to capture the overall composition at the level of individual molecular masses. To date, the heterogeneity of PS20 has not been fully understood, and the differences between PS20 materials have not yet been studied at this detailed level.

In this work, we report on the development of UPLC-MS methods to experimentally access the i) molar ratio of sorbitan, isosorbide and POE components, and the distributions within these substance classes with respect to ii) the number of esters and iii) the number of OE units. By means of stochastic modelling, we demonstrate a high correlation of the latter two with binomial distributions. Based on these findings and the determination of the fatty acid distribution, as described in our previous work [21], the molar composition of PS20 is calculated using an automated tool. This approach is verified by comparison of the predicted hydroxyl and saponification values with those provided in the respective certificates of analysis for seven PS20 products studied.

Download the full article as PDF here Unravelling the polysorbate 20 composition

or read more here

Materials

Two different batches each of polysorbate 20 high purity (PS20 HP) and polysorbate 20 Super Refined (PS20 SR) were obtained from Croda GmbH (Nettetal, Germany). An additional batch of PS20 was sourced from Kolb Distribution Ltd. (Hedingen, Switzerland) and two further batches were procured from Nanjing Well Pharmaceutical Co. Ltd. (Nanjing, China).

Analytical-grade reagents were used throughout the study. Glacial acetic acid and acetone were purchased from Merck KGaA (Darmstadt, Germany), while acetonitrile and methanol were obtained from Schar

lab S.L. (Barcelona, Spain). Formic acid was supplied by Honeywell Fluka (Seelze, Germany), ammonium acetate by VWR International (Radnor, PA, USA), and sodium hydroxide by PanReac AppliChem GmbH (Darmstadt, Germany).

For chromatographic analyses, the following columns were employed: Acquity UPLC™ HSS Cyano (1.8 μm, 2.1 ×50 mm) and XSelect HSS Cyano (3.5 μm, 3.0 ×50 mm), both from Waters GmbH (Eschborn, Germany), as well as the Zorbax 300 SB-C8 column (1.8 μm, 2.1 ×50 mm) from Agilent Technologies, Inc. (Santa Clara, CA, USA).

Dirk-Heinrich Evers, Janek Giebel, Finnya Nienau, Stefan Carle, Sascha Gorissen, Julia Buske, Michael E. Herbig, Patrick Garidel, Elina Hagelskamp, Unravelling the polysorbate 20 composition: A fusion of UPLC-MS analysis and stochastic modelling, European Journal of Pharmaceutics and Biopharmaceutics, Volume 216, 2025, 114854, ISSN 0939-6411, https://doi.org/10.1016/j.ejpb.2025.114854.

Join our next free webinar:

Masterclass in Nanomedicine: Olivia Merkel & Paola Luciani