High-throughput in vitro lipolysis-permeation screening compared to in vivo drug exposure for a wide range of oral lipid-based formulations

Abstract

During development of lipid-based formulations (LBFs) for poorly soluble drugs, formulations are often ranked according to in vitro API concentrations in the aqueous phase upon pH stat lipolysis. However, in an attempt to improve the in vivo relevance and to give credit to the free drug hypothesis, a recent study combined in vitro lipolysis with permeation, introducing a material-sparing high-throughput lipolysis-permeation approach on 96 well PermeaPad microtiter sandwich plates. The usefulness of the new approach was demonstrated in comparison to earlier published pH stat data for a set of model type I formulations for which in vivo rat pharmacokinetic (PK) data are available in the literature.

The present study expands the dataset and is the first study to include formulations across the entire lipid formulation classification system (type I, II, IIIa, IIIb, IV) in a head-to-head comparison of in vitro permeation (flux) and in vivo drug exposure from previously published rat PK data, where the impact of supersaturation, lipase inhibition, lipid chain length, and precipitation inhibition was investigated. A statistical modelling approach was applied to identify the impact of the aforementioned formulation parameters (supersaturation, lipase inhibition, lipid chain length, and precipitation inhibition) on in vitro flux and to correlate in vitro results with in vivo drug absorption data.

The analysis revealed that supersaturation increased the flux, in accordance with AUC in vivo for LBFs types I, II, IIIa, and IIIb. Lipase inhibition decreased the flux for type I and IV LBFs in accordance with in vivo results, but not for the other LBF types. The addition of a precipitation inhibitor did not have any significant impact on drug exposure, which was correctly captured in vitro for LBFs types I, II, and IIIa. The impact of lipid chain length was overestimated in vitro, probably due to poor dispersibility of the formulations. Overall, this high-throughput lipolysis-permeation assay was shown to be a promising tool for lipid-based formulation development.

Introduction

Lipid-based formulations (LBFs) composed of oils, surfactants, and cosolvents, are a promising enabling formulation approach capable of making poorly soluble, lipophilic drugs better orally bioavailable. The Lipid Formulation Classification System divides LBFs into five classes: I, II, IIIa, IIIb, and IV. Type I LBFs are composed of oils, type II LBFs in addition contain lipophilic surfactant, type IIIa LBFs are based on hydrophilic surfactants; the lipid content is further decreased in type IIIb LBFs, and type IV LBFs contain only surfactants and cosolvents.1,2 In most LBFs the drug is dissolved in the formulation, and supersaturated LBFs may be used to increase the dose.3

An inherent challenge with supersaturated LBFs is the possibility of precipitation, both after administration and on the shelf, which may be delayed by addition of precipitation inhibitors.4 The addition of precipitation inhibitors to LBFs to prevent in situ precipitation after administration was recently reviewed by Holm and colleagues, whereas there are still gaps on the impact of precipitation inhibitor addition on shelf stability. Components of the LBFs are expected to undergo lipolysis in the gastrointestinal tract, the products of which together with bile components (bile salts, phospholipids and cholesterol) provide additional solubilizing capacity. Therefore, the impact of digestion processes on micellar solubilization may vary with the type of LBFs. It remains controversial, however, whether enhanced micellar solubilization always translates into improved oral bioavailability. To mention just a few examples: during biphasic dissolution testing, an enhanced transfer from aqueous to organic phase induced by bile salt micelles has been postulated to be indicative for enhanced oral absorption and explained by “particle drifting”.5 Enhanced flux of Olmesartan from Soluplus micellar solution across ex vivo rat intestinal loop as compared to its suspension has been postulated to be due to permeation enhancement, yet without mechanistic proof.6 Bile micelles have been demonstrated to induce enhanced absorption of Cinnarizine from a lipidic formulation, but a reduced overall colloidal solubilization has been seen.7

The authors hypothesized that the observed reduced colloidal solubilization induced a higher free drug concentration, yet without direct experimental proof. Permeation of poorly soluble ABT-102 across Caco-2 barriers upon release from amorphous solid dispersions was found unaffected by micellar solubilization: bile salt micelles (fasted intestinal fluid vs. buffer;8) as well as surfactants within the formulation (below vs. above CMC;9), while an experimental proof of correlation between flux and free (molecularly dissolved) drug concentrations was established. In a recent study, when experimentally determining free and overall (free + micellar) drug concentrations during in vitro release from a series of sodium lauryl sulfate-based oral Alectinib formulations, the free drug concentrations employed for PBPK-modelling yielded excellent fits with human plasma AUCs, indicating that non-micellar drug drives absorption in this case.10 A general conclusion on such correlation has been difficult due to lack of systematic in vivo studies.

This deficit has recently been overcome by a systematic in vivo study in Sprague-Dawley rats including cinnarizine-loaded LBFs type I, type II LBFs, type IIIa LBFs, type IIIb LBFs, and type IV LBFs.[11], [12], [13], [14], [15] In these studies, to account for the impact of lipase inhibition, the lipase inhibitor orlistat was pre-dosed, non-supersaturated and supersaturated formulations with different drug loads depending on the drug solubility in the vehicle were dosed, and formulations with and without Soluplus as a precipitation inhibitor were administered. To investigate the impact of lipid chain length, either medium- or long-chain based formulations were used, but not for LBF type IIIb and IV, due to incompatibilities during formulation development. Depending on the formulation and degree of supersaturation, cinnarizine doses ranged from 18.7 to 66.8 mg/kg, with six animals investigated for each parameter combination. Overall, with these studies it could be demonstrated, that with increased hydrophilicity of the cinnarizine-loaded formulations, digestion becomes less significant. It was also shown that the administration of supersaturated LBFs leads to higher drug exposure than that of the respective non-supersaturated LBFs, and that addition of Soluplus had no significant effect on drug absorption for any of the formulations. The lipid chain length (medium- or long-chain fatty acids) had an impact on the dependency on lipid digestion for drug absorption of LBFs type I – IIIa. Formulations composed of long-chain lipids were shown to be more dependent on digestion than formulations based on medium-chain lipids.[11], [12], [13]

In search for appropriate in vitro tools suited to avoid animal testing during formulation performance ranking, the widely established pH-stat lipolysis model has been applied to test the above-mentioned type I LBFs, where the digestion process is mimicked by the addition of lipase to a lipolysis buffer (biomimetic medium which contains calcium, bile salts, and phospholipids). The parameters supersaturation, lipase inhibition, lipid chain length, and precipitation inhibition were tested in vitro using the pH stat approach.16 Unfortunately, in the above-mentioned work, a good correlation with in vivo pharmacokinetic data had been observed for type I medium-chain formulations only. The usefulness of the pH-stat lipolysis model is known to depend on formulations and drug substance properties.17

Various recommendations have been proposed and investigated to improve the in vivo relevance of the pH stat model. One such approach is the inclusion of a gastric digestion step, which significantly impacts overall lipolysis.18,19 While this adaptation aims to better capture gastrointestinal intraluminal drug concentrations, another promising method may be the incorporation of a permeation step to address the free drug fraction which is prone to permeation. Several types of barriers have been used for such combined lipolysis-permeation approaches. One of the first studies combining lipolysis and permeation,20 used the barrier Permeapad which was demonstrated a robust artificial biomimetic barrier that can be used for hours during lipolysis without compromising the barrier.20 However, even Caco-2-cells have been used for short lipolysis-based experiments, as well as organic solutions of lecithin on PVDF filter support (PAMPA).21,22 PVPA (Phospholipid Vesicle-based Permeation Assay)-barriers could withstand the digestion media a bit longer if protected with artificial mucus, while dialysis membranes as non-biomimetic barriers are a more robust alternative.23,24

Most of the lipolysis/permeation studies in literature are focusing on type IIIa LBFs, which makes it difficult to judge the general suitability of these approaches for all kinds of lipid-based formulations. In these studies, the impact of lipid chain length or degree of supersaturation on permeation is reported.21,25,26 Furthermore, these setups are hardly suited for high-throughput screening. This is likely also the reason why only a limited number of formulations were tested. Falavigna and coworkers investigated a modified PVPA barrier with artificial mucus to evaluate the IVIVR (in vitro-in vivo relationship) of fenofibrate-loaded type IIIa LBFs for which rat pharmacokinetic data was available and found a good correlation (R2 > 0.9), in contrast to pH-stat lipolysis data.27 The setup was also modified to increase the throughput by using a strongly buffered lipolysis medium to avoid the need for titration during the experiment. The in vitro model was able to capture the in vivo observed impact of supersaturation with and without lipolysis.28 However, the PVPA approach can still not be considered high-throughput because the number of samples per plate is limited and the barrier is not readily available and needs to be prepared prior to the experiment, which is cumbersome and time-consuming.

Recently, Paulus et al.29 successfully developed a high-throughput lipolysis-permeation assay based on 96 well sandwich plates with a biomimetic barrier (PermeaPad), which, originally developed for drug permeability screening,30,31 recently has come into widespread use for combined dissolution/permeation of various enabling formulations.[32], [33], [34] In Paulus’ study IVIVR was seen for long-chain based type I LBFs giving the correct ranking compared to in vivo AUC.29 In particular the supersaturated systems were correctly ranked, which are promising to increase drug exposure in vivo. It was concluded that for the limited fraction of formulations studied, an improvement in IVIVR was observed compared to results obtained with the pH-stat model.

The aim of the present work was therefore to apply the recently introduced high-throughput, combined lipolysis-permeation approach to screen LBFs across the entire LFCS (type I – IV), for which in vivo reference PK data are available, under various conditions. The aim was to evaluate the in vivo relevance of this approach. By doing so, the 3Rs principles – Replacement, Reduction, and Refinement – were also considered as a framework for promoting more ethical and sustainable research.

Download the full article as PDF here High-throughput in vitro lipolysis-permeation screening compared to in vivo drug exposure for a wide range of oral lipid-based formulations

or continue reading here

Materials

Labrafac Lipophile WL 1349 (medium-chain triglycerides (MCT)), Transcutol HP, and Maisine CC (long-chain mixed glycerides (LCM)) were kindly provided from Gattefossé (Lyon, France). Soluplus was a gift from BASF (Ludwigshafen, Germany). Imwitor 742 (medium-chain mixed glycerides (MCM)) was gifted from IOI Oleo (Hamburg, Germany). Cinnarizine and flunarizine (internal standard for analysis of permeation samples) were from Janssen Pharmaceutica NV (Beerse, Belgium). Sesame oil (long-chain triglycerides (LCT)), Tween 85, Cremophor EL, and orlistat, were obtained from Sigma-Aldrich (Bornem, Belgium). Tetrahydrofuran, acetonitrile, ammonia solution 25 % (LC-MS-grade) and trifluoroacetic acid were purchased from VWR (Leuven, Belgium). Pancreatin from porcine pancreas (8 * USP), Bovine Serum Albumin (BSA), glycerol, calcium chloride dihydrate, and BIS-TRIS were from Sigma-Aldrich (Søborg, Denmark). FaSSIF (3F)-powder was obtained from Biorelevant (London, UK). All the chemicals used throughout this work were of analytical grade. Ultra-purified water for the experiments was from a Milli-Q Advantage A10 Water Purification System (Burlington, USA).

Felix Paulus, René Holm, Yimer Wasihun Kifle, Laure Cougnaud, Nicolas Sauwen, Tatsiana Khamiakova, Jef Stappaerts, Martin Brandl, Annette Bauer-Brandl, High-throughput in vitro lipolysis-permeation screening compared to in vivo drug exposure for a wide range of oral lipid-based formulations, Journal of Pharmaceutical Sciences, 2026, 104557, ISSN 0022-3549, https://doi.org/10.1016/j.xphs.2026.104557.

See our next webinar:

Elevate process, maximize flexibility: advanced QbD and Continuous Manufacturing for moisture-protective coatings