In-situ forming PLGA implants: Towards less toxic solvents.

In-situ forming poly(lactic-co-glycolic acid) (PLGA) implants offer a great potential for controlled drug delivery for a variety of applications, e.g. periodontitis treatment. The polymer is dissolved in a water-miscible solvent. The drug is dissolved or dispersed in this solution. Upon contact with aqueous body fluids, the solvent diffuses into the surrounding tissue and water penetrates into the formulation. Consequently, PLGA precipitates, trapping the drug. Often, N-methyl-2-pyrrolidine (NMP) is used as a water-miscible solvent. However, parenteral administration of NMP raises toxicity concerns. The aim of this study was to identify less toxic alternative solvent systems for in-situ forming PLGA implants. Various blends of polyethylene glycol 400 (PEG 400), triethyl citrate (TEC) and ethanol were used to prepare liquid formulations containing PLGA, ibuprofen (as an anti-inflammatory drug) and/or chlorhexidine dihydrochloride (as an antiseptic agent). Implant formation and drug release kinetics were monitored upon exposure to phosphate buffer pH 6.8 at 37 °C. Furthermore, the syringeability of the liquids, antimicrobial activity of the implants, and dynamic changes in the latter's wet mass and pH of the release medium were studied. Importantly, 85:10:5 and 60:30:10 PEG 400:TEC:ethanol blends provided good syringeability and allowed for rapid implant formation. The latter controlled ibuprofen and chlorhexidine release over several weeks and assured efficient antimicrobial activity. Interestingly, fundamental differences were observed concerning the underlying release mechanisms of the two drugs: Ibuprofen was dissolved in the solvent mixtures and partially leached out together with the solvents during implant formation, resulting in relatively pronounced burst effects. In contrast, chlorhexidine dihydrochloride was dispersed in the liquids in the form of tiny particles, which were effectively trapped by precipitating PLGA during implant formation, leading to initial lag-phases for drug release.

Drug release from PLGA microparticles can be slowed down by a surrounding hydrogel.

This study aimed to evaluate and better understand the potential impact that a layer of surrounding hydrogel (mimicking living tissue) can have on the drug release from PLGA microparticles. Ibuprofen-loaded microparticles were prepared with an emulsion solvent extraction/evaporation method. The drug loading was about 48%. The surface of the microparticles appeared initially smooth and non-porous. In contrast, the internal microstructure of the particles exhibited a continuous network of tiny pores. Ibuprofen release from single microparticles was measured into agarose gels and well-agitated phosphate buffer pH 7.4. Optical microscopy, scanning electron microscopy, differential scanning calorimetry, X-ray powder diffraction, and X-ray µCT imaging were used to characterize the microparticles before and after exposure to the release media. Importantly, ibuprofen release was much slower in the presence of a surrounding agarose gel, e.g., the complete release took two weeks vs. a few days in well agitated phosphate buffer. This can probably be attributed to the fact that the hydrogel sterically hinders substantial system swelling and, thus, slows down the related increase in drug mobility. In addition, in this particular case, the convective flow in agitated bulk fluid likely damages the thin PLGA layer at the microparticles' surface, giving the outer aqueous phase more rapid access to the inner continuous pore network: Upon contact with water, the drug dissolves and rapidly diffuses out through a continuous network of water-filled channels. Without direct surface access, most of the drug "has to wait" for the onset of substantial system swelling to be released.

Exploration of the physical states of riboflavin (free base) by mechanical milling.

Amorphous riboflavin (free base) could be produced for the first time via high energy ball milling of a commercial crystalline form (Form I). Importantly, this solid state amorphization process allowed to circumvent chemical degradation occurring during melting as well as the lack of suitable solvents, which are required for amorphization via spray- or freeze-drying. The amorphous state of riboflavin was thoroughly characterized, revealing a complex recrystallization pattern upon heating, involving two enantiotropic polymorphic forms (II and III) and a dihydrate. The glass transition temperature (Tg) and heat capacity (Cp) jump of the amorphous form were determined as 144 °C and 0.68 J/g/°C. Moreover, the relative physical stability of the different physical states has been elucidated, e.g., at room temperature: I > II > III. The following rank order was observed for the dissolution rates in water at 37 °C during the first 4 h: amorphous > III ≈ II > I. Afterwards, a dihydrate crystallized from the solutions of amorphous and metastable crystalline riboflavin forms, the solubility of which was well above the solubility of the stable FormI.

Long term behavior of dexamethasone-loaded cochlear implants: In vitro & in vivo.

The aim of this study was to better understand the long term behavior of silicone-based cochlear implants loaded with dexamethasone: in vitro as well as in vivo (gerbils). This type of local controlled drug delivery systems offers an interesting potential for the treatment of hearing loss. Because very long release periods are targeted (several years/decades), product optimization is highly challenging. Up to now, only little is known on the long term behavior of these systems, including their drug release patterns as well as potential swelling or shrinking upon exposure to aqueous media or living tissue. Different types of cylindrical, cochlear implants were prepared by injection molding, varying their dimensions (being suitable for use in humans or gerbils) and initial drug loading (0, 1 or 10%). Dexamethasone release was monitored in vitro upon exposure to artificial perilymph at 37 °C for >3 years. Optical microscopy, X-ray diffraction and Raman imaging were used to characterize the implants before and after exposure to the release medium in vitro, as well as after 2 years implantation in gerbils. Importantly, in all cases dexamethasone release was reliably controlled during the observation periods. Diffusional mass transport and limited drug solubility effects within the silicone matrices seem to play a major role. Initially, the dexamethasone is homogeneously distributed throughout the polymeric matrices in the form of tiny crystals. Upon exposure to aqueous media or living tissue, limited amounts of water penetrate into the implant, dissolve the drug, which subsequently diffuses out. Surface-near regions are depleted first, resulting in an increase in the apparent drug diffusivity with time. No evidence for noteworthy implant swelling or shrinkage was observed in vitro, nor in vivo. A simplified mathematical model can be used to facilitate drug product optimization, allowing the prediction of the resulting drug release rates during decades as a function of the implant's design.

The melting of glucose.

Several sugars are known to undergo a spontaneous liquefaction below their reputed melting point (Tm), but the origin of this apparent melting is not yet clearly understood. In this paper we address this puzzling behavior in the particular case of the crystalline forms of glucose: Gα and Gß, involving respectively the glucose-α and glucose-ß anomers. We show in particular that the spontaneous melting below their reputed melting point Tm (∼151 °C for Gα and ∼156 °C for Gß) corresponds to a horizontal displacement of the system in the eutectic phase diagram of the anomeric mixture glucose-α / glucose-ß. This displacement is associated with mutarotation in the liquid which, in turn, induces additional liquefaction of the remaining crystal. This feedback loop creates a vicious circle which stops when the mixture reaches the liquidus branch, i.e. when the liquefaction is total. It is also shown that this behavior becomes more complex on approaching the eutectic temperature Te (120 °C). Just above Te, the liquefaction process is followed by a recrystallization leading to the crystalline form Gß. On the other hand, just below Te, the spontaneous liquefaction process stops as no melting is expected whatever the anomeric composition.

Evidence of transient amorphization during the polymorphic transformation of sorbitol induced by milling.

In this paper, we show that the polymorphic transformation γ â†’ α of sorbitol upon milling involves a transient amorphization of the material. This could be done by comilling sorbitol with a high Tg amorphous material (Hydrochlorothiazide, Tg = 115 °C) to stabilize any transient amorphous fractions of sorbitol through the formation of a molecular alloy. The results indicate that for large sorbitol concentration (50%), the comilling leads to a heterogeneous mixture made of sorbitol crystallites in the form α embedded into an amorphous molecular alloy sorbitol / HCT. Interestingly, the kinetic investigation of this transformation reveals that these two components are not produced simultaneously. On the contrary, they are produced one after the other, during two distinct consecutive stages. The first stage concerns the formation of the amorphous alloy while the second one concerns the polymorphic transformation γ â†’ α of the fraction of crystalline sorbitol not involved in the alloy. These results clearly indicate that the polymorphic transformation of sorbitol upon milling results from the recrystallization of a transient amorphous state generated by the mechanical shocks. The investigations were mainly performed by calorimetry and powder X-ray diffraction.

Dexamethasone-loaded cochlear implants: How to provide a desired "burst release".

Cochlear implants containing iridium platinum electrodes are used to transmit electrical signals into the inner ear of patients suffering from severe or profound deafness without valuable benefit from conventional hearing aids. However, their placement is invasive and can cause trauma as well as local inflammation, harming remaining hair cells or other inner ear cells. As foreign bodies, the implants also induce fibrosis, resulting in a less efficient conduction of the electrical signals and, thus, potentially decreased system performance. To overcome these obstacles, dexamethasone has recently been embedded in this type of implants: into the silicone matrices separating the metal electrodes (to avoid short circuits). It has been shown that the resulting drug release can be controlled over several years. Importantly, the dexamethasone does not only act against the immediate consequences of trauma, inflammation and fibrosis, it can also be expected to be beneficial for remaining hair cells in the long term. However, the reported amounts of drug released at "early" time points (during the first days/weeks) are relatively low and the in vivo efficacy in animal models was reported to be non-optimal. The aim of this study was to increase the initial "burst release" from the implants, adding a freely water-soluble salt of a phosphate ester of dexamethasone. The idea was to facilitate water penetration into the highly hydrophobic system and, thus, to promote drug dissolution and diffusion. This approach was efficient: Adding up to 10% dexamethasone sodium phosphate to the silicone matrices substantially increased the resulting drug release rate at early time points. This can be expected to improve drug action and implant functionality. But at elevated dexamethasone sodium phosphate loadings device swelling became important. Since the cochlea is a tiny and sensitive organ, a potential increase in implant dimensions over time must be limited. Hence, a balance has to be found between drug release and implant swelling.

Mechanistic explanation of the (up to) 3 release phases of PLGA microparticles: Diprophylline dispersions.

The aim of this study was to better understand the root causes for the (up to) 3 drug release phases observed with poly (lactic-co-glycolic acid) (PLGA) microparticles containing diprophylline particles: The 1st release phase ("burst release"), 2nd release phase (with an "about constant release rate") and 3rd release phase (which is again rapid and leads to complete drug exhaust). The behavior of single microparticles was monitored upon exposure to phosphate buffer pH 7.4, in particular with respect to their drug release and swelling behaviors. Diprophylline-loaded PLGA microparticles were prepared with a solid-in-oil-in-water solvent extraction/evaporation method. Tiny drug crystals were rather homogeneously distributed throughout the polymer matrix after manufacturing. Batches with "small" (63 µm), "medium-sized" (113 µm) and "large" (296 µm) microparticles with a practical drug loading of 5-7% were prepared. Importantly, each microparticle releases the drug "in its own way", depending on the exact distribution of the tiny drug crystals within the system. During the burst release, drug crystals with direct surface access rapidly dissolve. During the 2nd release phase tiny drug crystals (often) located in surface near regions which undergo swelling, are likely released. During the 3rd release phase, the entire microparticle undergoes substantial swelling. This results in high quantities of water present throughout the system, which becomes "gel-like". Consequently, the drug crystals dissolve, and the dissolved drug molecules rather rapidly diffuse through the highly swollen polymer gel.

Interest of molecular/crystalline dispersions for the determination of solubility curves of drugs into polymers.

We present here a method to increase the dissolution rate of drugs into polymers in order to make easier and faster the determination of the solubility curves of these mixtures. The idea is to prepare molecular/crystalline dispersions (MCD) where the drug is dispersed into the polymer, partly at the molecular level and partly in the form of small crystallites. We show that this particular microstructure greatly increases the dissolution rate of crystallites since: (1) The molecular dispersion has a plasticizing effect which greatly increases the molecular mobility in the amorphous matrix. (2) The fine crystallite dispersion in the matrix strongly reduces the distances over which the drug molecules have to diffuse to invade homogeneously the polymer by dissolution. MCD are here obtained by combining solid-state co-amorphization by high energy mechanical milling and then recrystallization by annealing under a plasticizing atmosphere. We have used MCD to determine the solubility lines of the two polymorphic forms (I and II) of sulindac into PVP. The investigations have been performed mainly by DSC and powder x-ray diffraction.

PLGA implants: How Poloxamer/PEO addition slows down or accelerates polymer degradation and drug release.

The aim of this study was to evaluate the impact of the addition of small amounts of hydrophilic polymers (Poloxamer 188 and PEO 200kDa) to PLGA-based implants loaded with prilocaine. Special emphasis was placed on the importance of the type of preparation technique: direct compression of milled drug-polymer powder blends versus compression of drug loaded microparticles (prepared by spray-drying). The implants were thoroughly characterized before and upon exposure to phosphate buffer pH7.4, e.g. using optical and scanning electron microscopy, X-ray diffraction, DSC and GPC. Interestingly, the addition of Poloxamer/PEO to the PLGA implants had opposite effects on the resulting drug release kinetics, depending on the type of preparation method: in the case of implants prepared by compression of milled drug-polymer powder blends, drug release was accelerated, whereas it was slowed down when the implants were prepared by compression of drug loaded PLGA microparticles. These phenomena could be explained by the swelling/disintegration behavior of the implants upon exposure to the release medium. Systems consisting of compressed microparticles remained intact and autocatalytic effects were of major importance. The presence of a hydrophilic polymer facilitated water penetration into these devices, slowing down PLGA degradation and drug release. In contrast, implants consisting of compressed drug-polymer powder blends rapidly (at least partially) disintegrated and autocatalysis was much less important. In these cases, the addition of a hydrophilic polymer facilitated ester bond cleavage, leading to accelerated PLGA degradation and drug release.

Towards a better understanding of the different release phases from PLGA microparticles: Dexamethasone-loaded systems.

Dexamethasone-loaded, poly(lactic-co-glycolic acid) (PLGA) microparticles were prepared using an oil-in-water solvent extraction/evaporation method. The drug loading was varied from 2.4 to 61.9%, keeping the mean particle size in the range of 52-61µm. In vitro drug release was characterized by up to 3 phases: (1) an (optional) initial burst release, (2) a phase with an about constant drug release rate, and (3) a final, again rapid, drug release phase. The importance and durations of these phases strongly depended on the initial drug loading. To better understand the underlying mass transport mechanisms, the microparticles were thoroughly characterized before and after exposure to the release medium. The initial burst release seems to be mainly due to the dissolution of drug particles with direct access to the microparticles' surface. The extent of the burst was negligible at low drug loadings, whereas it exceeded 60% at high drug loadings. The second release phase seems to be controlled by limited drug solubility effects and drug diffusion through the polymeric systems. The third drug release phase is likely to be a consequence of substantial microparticle swelling, leading to a considerable increase in the systems' water content and, thus, fundamentally increased drug mobility.

Ear Cubes for local controlled drug delivery to the inner ear.

A new type of advanced drug delivery systems is proposed: Miniaturized implants, which can be placed into tiny holes drilled into (or close to) the oval window. They consist of two parts: 1) A cylinder, which is inserted into the hole crossing the oval window. The cylinder (being longer than the depth of the hole) is partly located within the inner ear and surrounded by perilymph. This provides direct access to the target site, and at the same time assures implant fixation. 2) A cuboid, which is located in the middle ear, serving as a drug reservoir. One side of the cuboid is in direct contact with the oval window. Drug release into the cochlea occurs by diffusion through the cylindrical part of the Ear Cubes and by diffusion from the cuboid into and through the oval window. High precision molds were used to prepare two differently sized Ear Cubes by injection molding. The miniaturized implants were based on silicone and loaded with different amounts of dexamethasone (10 to 30 % w/w). The systems were thoroughly characterized before and upon exposure to artificial perilymph at 37°C. Importantly, drug release can effectively be controlled and sustained during long time periods (up to several years). Furthermore, the implants did not swell or erode to a noteworthy extent during the observation period. Drug diffusion through the polymeric matrix, together with limited dexamethasone solubility effects, seem to control the resulting drug release kinetics, which can roughly be estimated using mathematical equations derived from Fick's second law. Importantly, the proposed Ear Cubes are likely to provide much more reliable local long term drug delivery to the inner ear compared to liquid or semi-solid dosage forms administered into the middle ear, due to a more secured fixation. Furthermore, they require less invasive surgeries and can accommodate higher drug amounts compared to intracochlear implants. Thus, they offer the potential to open up new horizons for innovative therapeutic strategies to treat inner ear diseases and disorders.

Perspectives on the amorphisation/milling relationship in pharmaceutical materials.

This paper presents an overview of recent advances in understanding the role of the amorphous state in the physical and chemical transformations of pharmaceutical materials induced by mechanical milling. The following points are addressed: (1) Is milling really able to amorphise crystals?, (2) Conditions for obtaining an amorphisation, (3) Milling of hydrates, (4) Producing amorphous state without changing the chemical nature, (5) Milling induced crystal to crystal transformations: mediation by an amorphous state, (6) Nature of the amorphous state obtained by milling, (7) Milling of amorphous compounds: accelerated aging or rejuvenation, (8) Specific recrystallisation behaviour, and (9) Toward a rationalisation and conceptual framework.

Calendering as a direct shaping tool for the continuous production of fixed-dose combination products via co-extrusion.

In this study calendering is used as a downstream technique to shape monolithic co-extruded fixed-dose combination products in a continuous way. Co-extrudates with a metoprolol tartrate-loaded sustained-release core and a hydrochlorothiazide-loaded immediate-release coat were produced and immediately shaped into a monolithic drug delivery system via calendering, using chilled rolls with tablet-shaped cavities. In vitro metoprolol tartrate release from the ethylcellulose core of the calendered tablets was prolonged in comparison with the sustained release of a multiparticulate dosage form, prepared manually by cutting co-extrudates into mini-matrices. Analysis of the dosage forms using X-ray micro-computed tomography only detected small differences between the pore structure of the core of the calendered tablet and the mini-matrices. Diffusion path length was shown to be the main mechanism behind the release kinetics. Terahertz pulsed imaging visualized that adhesion between the core and coat of the calendered tablet was not complete and a gradient in coat thickness (varying from 200 to 600µm) was observed. Modulated differential scanning calorimetry and X-ray diffraction indicated that the solid-state properties of both drugs were not affected by the calendering procedure.

Controlled delivery of a new broad spectrum antibacterial agent against colitis: In vitro and in vivo performance.

Coated pellets and mini-tablets were prepared containing a new broad spectrum antibacterial agent: CIN-102, a well-defined, synergistic blend of trans-cinnamaldehyde, trans-2-methoxycinnamaldehyde, cinnamyl acetate, linalool, ß-caryophyllene, cineol and benzyl benzoate. The aim was to provide a new treatment method for colitis, especially for Inflammatory Bowel Disease (IBD) patients. Since the simple oral gavage of CIN-102 was not able to reduce the pathogenic bacteria involved in colitis (rat model), the drug was incorporated into multiparticulates. The idea was to minimize undesired drug release in the upper gastrointestinal tract and to control CIN-102 release in the colon, in order to optimize the resulting antibiotic concentration at the site of action. A particular challenge was the fact that CIN-102 is a volatile hydrophobic liquid. Pellet cores were prepared by extrusion-spheronization and coated with polymer blends, which are sensitive to colonic bacterial enzymes. Mini-tablets were prepared by direct compression. The release of the main compound of CIN-102 (cinnamaldehyde, 86.7% w/w) was monitored in vitro. Optimized coated pellets and mini-tablets were also tested in vivo: in seven-week-old, male mice suffering from dextran sodium sulfate induced colitis. Importantly, both types of multiparticulates were able: (i) to significantly reduce the number of luminal and mucosal enterobacteria in the mice (the levels of which are increased in the disease state), and (ii) to improve the clinical course of the intestinal inflammation (decrease in the percentages of mice with bloody stools and diarrhea). Thus, the proposed coated pellets and matrix mini-tablets allowing for controlled CIN-102 release show a promising potential for new treatment methods of colitis.

Dynamic nuclear polarization of a glassy matrix prepared by solid state mechanochemical amorphization of crystalline substances.

A mechanochemical "solvent-free" route is presented for the preparation of solid samples ready to be employed in the Dynamic Nuclear Polarization (DNP). (1)H-DNP build-up curves at 3.46 T as a function of temperature and radical concentration show steady state nuclear polarization of 10% (0.5% TEMPO concentration at 1.75 K).

Accelerated ketoprofen release from polymeric matrices: importance of the homogeneity/heterogeneity of excipient distribution.

Polymeric matrices loaded with 10-50% ketoprofen were prepared by hot-melt extrusion or spray-drying. Eudragit E, PVP, PVPVA and HPMC were studied as matrix formers. Binary "drug-Eudragit E" as well as ternary "drug-Eudragit E-PVP", "drug-Eudragit E-PVPVA" and "drug-Eudragit E-HPMC" combinations were investigated and characterized by optical macro/microscopy, SEM, particle size measurements, mDSC, X-ray diffraction and in vitro drug release studies in 0.1 M HCl. In all cases ketoprofen release was much faster compared to a commercially available product and the dissolution of the drug powder (as received). Super-saturated solutions were obtained, which were stable during at least 2 h. Importantly, not only the composition of the systems, but also their inner structure potentially significantly affected the resulting ketoprofen release kinetics: For instance, spray-drying ternary ketoprofen:Eudragit E:HPMC combinations led to a more homogenous HPMC distribution within the systems than hot-melt extrusion, as revealed by mDSC and X-ray diffraction. This more homogenous HPMC distribution resulted in more pronounced hindrance for water and drug diffusion and, thus, slower drug release from spray-dried powder compared to hot-melt extrudates of identical composition. This "homogeneity/heterogeneity effect" even overcompensated the "system size effect": the surface exposed to the release medium was much larger in the case of the spray-dried powder. All formulations were stable during storage at ambient conditions in open vials.

Mechanism of solid state amorphization of glucose upon milling.

Crystalline α-glucose is known to amorphize upon milling at -15 °C while it remains structurally invariant upon milling at room temperature. We have taken advantage of this behavior to compare the microstructural evolutions of the material in both conditions in order to identify the essential microstructural features which drive the amorphization process upon milling. The investigations have been performed by differential scanning calorimetry and by powder X-ray diffraction. The results indicate that two different amorphization mechanisms occur upon milling: an amorphization at the surface of crystallites due to the mechanical shocks and a spontaneous amorphization of the crystallites as they reach a critical size, which is close to 200 Å in the particular case of α-glucose.

Controlled release implants based on cast lipid blends.

The aim of this study was to use lipid:lipid blends as matrix formers in controlled release implants. The systems were prepared by melting and casting and thoroughly characterized before and after exposure to the release medium. Based on the experimental results, a mechanistic realistic mathematical model was used to get further insight into the underlying drug release mechanisms. Importantly, broad spectra of drug release patterns could be obtained by simply varying the lipid:lipid blend ratio in implants based on Precirol ATO 5 (glyceryl palmitostearate):Dynasan 120 (hardened soybean oil) mixtures loaded with propranolol hydrochloride. Release periods ranging from a few days up to several months could be provided. Interestingly, the drug release rate monotonically decreased with increasing Dynasan 120 content, except for implants containing about 20-25% Precirol, which exhibited surprisingly high release rates. This could be attributed to the incomplete miscibility of the two lipids at these blend ratios: DSC thermograms showed phase separation in these systems. This is likely to cause differences in the implants' microstructure, which determines the mobility of water and dissolved drug as well as the mechanical stability of the systems. Purely diffusion controlled drug release was only observed at Precirol ATO 5 contents around 5-10%. In all other cases, limited drug solubility effects or matrix former erosion are also expected to play a major role. Thus, lipid:lipid blends are very interesting matrix formers in controlled release implants. However, care must be taken with respect to the mutual miscibility of the compounds: in case of phase separation, surprisingly high drug release rates might be observed.

Drug release mechanisms of cast lipid implants.

The aim of this work was to better understand which physicochemical processes are involved in the control of drug release from lipid implants prepared by melting and casting. Lipid implants gain steadily in importance as controlled parenteral drug delivery systems: In contrast to PLGA-based devices, no acidic microclimates are created, which can inactivate incorporated drugs. The melting and casting method offers various advantages over the commonly used direct compression technique. For example, powder de-mixing during manufacturing and highly challenging scale-up due to poor powder flowability are avoided. Importantly, broad spectra of drug release patterns can be easily provided by varying the type of lipid. The resulting drug release rates are generally lower than those of implants prepared by direct compression. This is probably due to the differences in the microstructure of the pore network of the systems. Drug or water diffusion plays a dominant role for the control of drug release, potentially combined with limited drug solubility effects, caused by the low amounts of water available within the implants. In the case of pure diffusion control, a mechanistic realistic mathematical theory is proposed, which allows for quantitative predictions of the effects of formulation parameters on the resulting drug release kinetics. Importantly, these theoretical predictions could be successfully confirmed by independent experiments. Thus, the obtained new insight into the underlying drug release mechanisms can significantly facilitate the optimization of this type of advanced drug delivery systems. This is particularly helpful if long release periods are targeted, requiring time-consuming experimental studies.