A design of experiment approach to identify the most stable composition of a ternary co-amorphous system
Abstract
Ternary co-amorphous systems, comprising a drug, a low molecular weight co-former, and a polymer, are a promising approach to address the solubility and stability challenges of poorly water-soluble drugs. However, it is unclear how the addition of a third component influences the stability of the binary system and how to identify the optimal composition of a ternary system. In previous studies we calculated weight percentages of the components via a modified Gordon-Taylor equation assuming the measured glass transition temperature reflected the composition of the ternary system. In this study, the underlying assumptions for these calculations are experimentally verified using a range of ternary mixtures of the system carvedilol-tryptophan- hydroxypropyl methylcellulose. Samples were prepared either by ball-milling all three components simultaneously or by establishing different binary systems and subsequently adding the third component. Design of experiments combined with multivariate analysis of differential scanning calorimetry and X-ray powder diffraction results was used to investigate the influence of preparation time and pathway on thermal and diffractometric properties of the systems, as well as their physical stability. It was hypothesized that the composition with least dependence on the input variables (i.e. the most robust composition) would be the most stable one. The study confirmed this hypothesis. The calculation method proposed in previous studies was verified and the most stable composition found in this study matched the calculated composition.
Highlights
- The optimal composition is independent of the preparation pathway.
- Relevant descriptors were identified via a principal component analysis.
Introduction
The pharmaceutical development challenge of the high number of drug candidates and marketed drugs that exhibit poor aqueous solubility is well known [[1], [2], [3], [4]]. As low aqueous solubility is often linked to low or suboptimal oral bioavailability [5,6], strategies have been developed to overcome this issue and thus to (potentially) enhance the bioavailability of those drugs which belong to the biopharmaceutics classification system (BCS) classes II and IV [[6], [7], [8]] More recently, co-amorphous systems (CAMs), i.e., systems composed of the drug and a low molecular weight co-former, have gained increasing attention in this context [6,[9], [10], [11], [12]]. For binary CAMs, it has been established that a 1:1 M ratio approach does not necessarily lead to CAMs with the best critical quality attributes such as the highest physical stability [[13], [14], [15], [16]]. Ternary CAMs consists of a binary (low molecular weight drug + co-former) system combined with a third component, frequently a polymer. This can result in higher solubility, prolonged supersaturation less prone to precipitation, and no changes in the physical stability of the ternary system compared to binary systems [[17], [18], [19]]. These improvements have, for example, been shown when using the CAM carvedilol-aspartic acid and HPMC [20] or ezetimibe-lovastatine with either Soluplus®, PVP K30, PVP VA64 or HPMC [21]. However, adding a third component to the binary CAMs can affect the pre-established system, for example, due to new interactions with the added polymer, which can replace existing interactions in the CAM. Alternatively, due to limited miscibility of the drug and co-former in the polymer, phase separation and recrystallisation can occur [19,20,22,23].
In a recent review, an overview of the advances in binary and ternary CAMs as well as ternary solid dispersions encouraged the use of Design of Experiments (DoE) as a tool to implement Quality by Design (QbD) in developing CAMs [18]. Multivariate data analysis has been applied as a tool to investigate the optimal molar ratio of binary CAMs [24]. However, the third component is usually simply added to an already existing binary system, and little is known about the best component ratio for ternary CAMs [13,25]. To address the complex interplay between preparation pathway and the various compositions, a DoE approach in combination with multivariate data analysis may help to better understand ternary CAMs.
In our previous study [26], we investigated the behaviour of ternary CAMs prepared via ball-milling where hydroxypropyl-methyl cellulose (HPMC) was combined with a binary CAM composed of carvedilol (CAR) and tryptophan (TRP). It was discovered that the addition of polymer increased the initial dissolution rate and yielded prolonged supersaturation of the drug. However, the glass transition temperature (Tg) decreased with increased ball-milling times towards the Tg of the binary CAR-HPMC system. Furthermore, X-ray powder diffraction (XRPD) showed reflections between 17 and 19 °2θ, attributed to either recrystallised CAR or TRP. To further study the phase evolution during milling and to answer the question of whether different starting conditions and kinetic pathways would lead to the same or to a different end result, the mixing of the components was performed in three different sequences [26]. Each pathway began with a binary System (“A – B”) which was ball-milled for 60 min followed by adding the third component (“+ C”) and continued ball-milling for a total of 180 min. The Tgs obtained for the various milling times were used to calculate the weight ratio in the amorphous systems by using a modified Gordon-Taylor equation. The Tg was therefore used to calculate the composition. The concentration ratio in the saturated ternary co-amorphous phase could be determined, however, under the condition that no specific interactions exist the ternary system, beyond those that were not already accounted for in the respective binary CAR-TRP and CAR-HPMC systems. Independent of the preparation pathways, the Tgs for the three prepared ternary systems after 180 min of ball-milling were rather similar, indicating that the finally established single phase amorphous systems possessed similar weight ratios, independent of the starting material. On the molecular level, it was found that the addition of a polymer reduced the likelihood for interaction of TRP with CAR, as the formation of hydrogen bonds between CAR and HPMC became the prevailing interaction pattern.
The first aim of the current study was thus to verify experimentally whether the compositions of the ternary CAMs can be calculated using the Tgs by the above-mentioned method. To do this, a DoE setup with 13 different compositions was chosen and the three components were ball-milled together directly. The thermal behaviour was investigated and the Tg was taken as an output parameter of the composition, rather than the other way as in the previous study. To gain further insight into these ternary CAMs, the 13 samples were also subjected to stability testing.
The second aim was to investigate whether the same final amorphous system would be achieved, independent of the preparation pathway. As all samples were produced using ball-milling the term “pathway” here refers to sequence of preparation: either an initial ternary system (with the two low molecular weight components initially being crystalline) or one of three binary amorphous systems where the third component (crystalline in case of the low molecular weight components) was added afterwards.
The final aim was to investigate if there is a link between the ease of formation of an amorphous system and its stability, i.e., to link uniformity and robustness across varying preparation pathways of the various ternary systems with the physical stability of the resulting composition. In this study, we define robust compositions as compositions with quality attributes (thermal and diffractometric behaviour) independent of process parameters (preparation pathway and milling time). It is hypothesized that the physically most stable composition is equal to the composition which shows the least susceptibility to changes in the preparation of the ternary system (i.e., the four different pathways) and which thus reaches a state of equilibrium in the shortest ball-milling time. To evaluate this, principal component analysis (PCA) was applied on thermal and diffractometric responses.
Download the full article as PDF here: A design of experiment approach to identify the most stable composition of a ternary co-amorphous system
or read it here
Materials
Carvedilol (CAR, molecular weight: 406.47 g/mol) was purchased from Cipla Ltd. (Mumbai, India). The two co-formers L-tryptophan (TRP, molecular weight: 204.23 g/mol) and hydroxypropylmethyl cellulose (HPMC, Pharmacoat 606, substituition type 2910, viscosity 6 mPas) were purchased from Sigma-Aldrich (St. Louis, MO, USA) and Harke Pharma (Mühlheim an der Ruhr, Germany), respectively. All material were used as received, without further purification.
Wiebke Traichel, Thomas Rades, Holger Grohganz, A design of experiment approach to identify the most stable composition of a ternary co-amorphous system, Journal of Drug Delivery Science and Technology, Volume 107, 2025, 106834, ISSN 1773-2247, https://doi.org/10.1016/j.jddst.2025.106834.
Read also our introduction article on Quality by Design (QbD) here:
