Artificial neural network as an alternative to multiple regression analysis in optimizing formulation parmaeters of cytarabine liposomes

The objective of the study was to optimize the formulation parameters of cytarabine liposomes by using artificial neural networks (ANN) and multiple regression analysis using 3³ factorial design (FD). As model formulations, 27 formulations were prepared. The formulation variables, drug (cytarabine)/lipid (phosphatidyl choline [PC] and cholesterol [Chol]) molar ratio (X ₁, PC/Chol in percentage ratio of total lipids (X ₂), and the volume of hydration medium, (X ₃) were selected as the independent variables; and the percentage drug entrapment (PDE) was selected as the dependent variable. A set of causal factors was used as tutorial data for ANN and fed into a computer. The optimization was performed by minimizing the generalized distance between the predicted values of each response and the optimized one that was obtained individually. In case of 3³ factorial design, a second-order full-model polynomial equation and a reduced model were established by subjecting the transformed values of independent variables to multiple regression analysis, and contour plots were drawn using the equation. The optimization methods developed by both ANN and FD were validated by preparing another 5 liposomal formulations. The predetermined PDE and the experimental data were compared with predicted data by pairedt test, no statistically significant difference was observed. ANN showed less error compared with multiple regression analysis. These findings demonstrate that ANN provides more accurate prediction and is quite useful in the optimization of pharmaceutical formulations when compared with the multiple regression analysis method.     

Formulation and optimization of piroxicam proniosomes by 3-factor, 3-level box-behnken design

The aim of this study was to investigate the combined influence of 3 independent variables in the preparation of piroxicam proniosomes by the slurry method. A 3-factor, 3-level Box-Behnken design was used to derive a second-order polynomial equation and construct contour plots to predict responses. The independent variables selected were molar ratio of Span 60:cholesterol (X(1)), surfactant loading (X(2)), and amount of drug (X(3)). Fifteen batches were prepared by the slurry method and evaluated for percentage drug entrapment (PDE) and vesicle size. The transformed values of the independent variables and the PDE (dependent variable) were subjected to multiple regression to establish a full-model second-order polynomial equation. F was calculated to confirm the omission of insignificant terms from the full-model equation to derive a reduced-model polynomial equation to predict the PDE of proniosome-derived niosomes. Contour plots were constructed to show the effects of X(1), X(2) and X(3) on the PDE. A model was validated for accurate prediction of the PDE by performing checkpoint analysis. The computer optimization process and contour plots predicted the levels of independent variables X(1), X(2), and X(3) (0, -0.158 and -0.158 respectively), for maximized response of PDE with constraints on vesicle size. The Box-Behnken design demonstrated the role of the derived equation and contour plots in predicting the values of dependent variables for the preparation and optimization of piroxicam proniosomes.     

Liposome/DNA systems: correlation between association, hydrophobicity and cell viability

Small unilamellar vesicles associated with plasmid DNA showed maximum association efficiency for a cationic mixture of egg phosphatidylcholine (EPC):1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE):di-1,2-dioleoyl-3-trimethyl ammonium propane (DOTAP) (16:8:1 molar ratio) [65%], followed by neutral lipids EPC:1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine (DMPE):cholesterol (Chol) (2:2:1 molar ratio) [30%], and a polymerized formulation 1,2-bis(10,12-tricosadiynoyl)sn-glycero-3-phosphocholine (DC8,9PC):DMPE:Chol (2:2:1 molar ratio) [11%]. The hydrophobicity factor (HF) for these formulations followed the trend DC8,9PC:DMPE:CHOL < EPC:DMPE:Chol < EPC:DOPE DOTAP, and DNA association did not alter this trend. Results suggest that the higher the HF value, the more fluid the membrane and the higher the efficiency of DNA association. On the other hand, no differences were observed in cell toxicity with lipids up to 1 mg/ml in VERO cells.     

Formulation design and optimization of mouth dissolve tablets of nimesulide using vacuum drying technique

The purpose of this research was to develop mouth dissolve tablets of nimesulide. Granules containing nimesulide, camphor, crospovidone, and lactose were prepared by wet granulation technique. Camphor was sublimed from the dried granules by exposure to vacuum. The porous granules were then compressed. Alternatively, tablets were first prepared percentage friability, wetting time, and disintegration time. In the investigation, a 3² full factorial design was used to investigate the joint influence of 2 formulation variables: amount of camphor and crospovidone. The results of multiple linear regression analysis revealed that for obtaining a rapidly disintegrating dosage form, tablets should be prepared using an optimum concentration of camphor and a higher percentage of crospovidone. A contour plot is also presented to graphically represent the effect of the independent variables on the disintegration time and percentage friability. A checkpoint batch was also prepared to prove the validity of the evolved mathematical model. Sublimation of camphor from tablets resulted in superior tablets as compared with the tablets prepared from granules that were exposed to vacuum. The systematic formulation approach helped in understanding the effect of formulation processing variables.     

A comparison of the in vitro antimicrobial activity of liposomes containing meropenem and gentamicin

The antimicrobial activity of eight cationic, two neutral and three anionic liposome compositions containing meropenem and gentamicin was tested in vitro in broth and serum medium. The cationic formulations showed better antibacterial efficacy against both Gram-positive and Gram-negative bacteria than the anionic and neutral ones, regardless of the encapsulated drug. The most effective formulations were the cationic PC/DOPE/DOTAP 3:4:3 and PC/Chol/DOTAP 3:4:3, as the MICs with meropenem were 2 to 4 times lower than those of the free drug.     

Investigation of Formulation Variables and Excipient Interaction on the Production of Niosomes

The aim of this study was to investigate the effects of formulation and process variables on the properties of niosomes formed from Span 40 as nonionic surfactant. A variety of formulations encapsulating Paclitaxel, a hydrophobic model drug, were prepared using different dicetyl phosphate (DCP) and Span 40-cholesterol (1:1) amounts. Formulations were optimized by multiple regression analysis to evaluate the changes on niosome characteristics such as entrapment efficiency, particle size, polydispersity index, zeta potential and in vitro drug release. Multiple regression analysis revealed that as Span 40-cholesterol amounts in the formulations were increased, zeta potential and percent of drug released at 24th hour were decreased. Besides, DCP was found to be effective on increasing niosome size. As a process variable, the effect of sonication was observed and findings revealed an irreversible size reduction on Span 40 niosomes after probe sonication. Monodisperse small sized (133 ± 6.01 nm) Span 40 niosomes entrapping 98.2% of Paclitaxel with a weight percentage of 3.64% were successfully prepared. The drug–excipient interactions in niosomes were observed by differential scanning calorimetry and X-ray powder diffraction analysis. Both techniques suggest the conversion of PCTs’ crystal structure to amorphous form. The thermal analyses demonstrate the high interaction between drug and surfactant that explains high entrapment efficiency. After 3-month storage, niosomes preserved their stability in terms of drug amount and particle size. Overall, this study showed that Span 40 niosomes with desired properties can be prepared by changing the content and production variables.Key words: drug delivery systems, drug release, multiple regression, niosomes, paclitaxel     

A hybrid of back propagation neural network and genetic algorithm for optimization of collagen extraction from Malaysian cultured catfish (Hybrid Clarias sp.)

Fractional factorial design (FFD) was applied to evaluate the effects of various process parameters in influencing the extraction efficiency of pepsin soluble collagen (PSC) from muscles of cultured catfish (Clarias gariepinus×C. macrocephalus). Result of the first order factorial design showed that acetic acid concentration, acid extraction time, acetic acid to muscles ratio, and stirring speed posed significant effect (P<0.05) on the yield of PSC obtained at the end of the extraction process. Two different artificial intelligence techniques namely artificial neural network (ANN) and genetic algorithm (GA) were then integrated for optimizing the extraction conditions to obtain the highest yield of PSC. The ANN was trained using the back propagation algorithm. A model was successfully generated with R ² value of 0.9527 and MSE value of 0.1672 for unseen data set, implying a good generalization of the network. Input parameters of the established ANN model were subsequently optimized using GA. The hybrid of ANN-GA model predicted a maximum extraction yield of PSC at 238.25 mg/g under the following conditions: an acetic acid concentration of 0.70 M, the acetic acid to muscles ratio of 25.78 mL/g, and the stirring speed of 432.50 rpm. Verification of the optimization showed the percentage error differences between the experimental and predicted values were less than 5%, indicating excellent modeling, predicting ability and optimization by the ANN-GA model.     

The use of artificial neural networks to predict the presence of small-bodied fish in a river

1. Discriminant factorial analysis (DFA) and artificial neural networks (ANN) were used to develop models of presence/absence for three species of small-bodied fish (minnow, Phoxinus phoxinus , gudgeon, Gobio gobio , and stone loach, Barbatula barbatula ).
2. Fish and ten environmental variables were sampled using point abundance sampling by electrofishing in the Ariège River (France) at 464 sampling points.
3. Using DFA, the percentage of correct assignments, expressed as the percentage of individuals correctly classified over the total number of examined individuals, was 62.5% for stone loach, 66.6% for gudgeon and 78% for minnow. With back-propagation of ANN, the recognition performance obtained after 500 iterations was: 82.1% for stone loach, 87.7% for gudgeon and 90.1% for minnow.
4. The better predictive performance of the artificial neural networks holds promise for other situations with non-linearly related variables.     

Cubic phases in phosphatidylcholine-cholesterol mixtures: cholesterol as membrane "fusogen"

X-ray diffraction reveals that mixtures of some unsaturated phosphatidylcholines (PCs) with cholesterol (Chol) readily form inverted bicontinuous cubic phases that are stable under physiological conditions. This effect was studied in most detail for dioleoyl PC/Chol mixtures with molar ratios of 1:1 and 3:7. Facile formation of Im3m and Pn3m phases with lattice constants of 30-50 nm and 25-30 nm, respectively, took place in phosphate-buffered saline, in sucrose solution, and in water near the temperature of the Lalpha-HII transition of the mixtures, as well as during cooling of the HII phase. Once formed, the cubic phases displayed an ability to supercool and replace the initial Lalpha phase over a broad range of physiological temperatures. Conversion into stable cubic phases was also observed for mixtures of Chol with dilinoleoyl PC but not for mixtures with palmitoyl-linoleoyl PC or palmitoyl-oleoyl PC, for which only transient cubic traces were recorded at elevated temperatures. A saturated, branched-chain PC, diphytanoyl PC, also displayed a cubic phase in mixture with Chol. Unlike the PEs, the membrane PCs are intrinsically nonfusogenic lipids: in excess water they only form lamellar phases and not any of the inverted phases on their own. Thus, the finding that Chol induces cubic phases in mixtures with unsaturated PCs may have important implications for its role in fusion. In ternary mixtures, saturated PCs and sphingomyelin are known to separate into liquid-ordered domains along with Chol. Our results thus suggest that unsaturated PCs, which are excluded from these domains, could form fusogenic domains with Chol. Such a dual role of Chol may explain the seemingly paradoxical ability of cell membranes to simultaneously form rigid, low-curvature raft-like patches while still being able to undergo facile membrane fusion.     

Cholesterol modulates interaction between an amphipathic class A peptide, Ac-18A-NH2, and phosphatidylcholine bilayers

Cholesterol (Chol) in phosphatidylcholine large unilamellar vesicles (PC LUV) modulated interaction of the bilayers with a class A amphipathic peptide, Ac-18A-NH2: Chol increased the peptide binding capacity and reduced the affinity together with the peptide-induced leakage of calcein from LUV. Similar effects of Chol have been observed on the interaction of LUV with apoA-I [Saito, H., Miyako, Y., Handa, T., and Miyajima, K. (1997) J. Lipid Res. 38, 287-294]. Circular dichroism (CD) spectra of the peptide indicated a similar helical structure formation in LUV with and without Chol. The fluorescence spectral shift, quantum yield, anisotropy, and acrylamide-quenching of the peptide Trp indicated that in PC:Chol (3:2) LUV, Ac-18A-NH2 was located in a more polar membrane environment with increased motional freedom and greater accessibility to the aqueous medium. Fluorescence energy transfer from the Trp indole ring to acceptors situated at different depths in the bilayers revealed that the amphipathic peptide penetrated the hydrophobic interior of PC bilayers, while the peptide was located at the polar zwitterionic surface in PC:Chol LUV. The inclusion of Chol causes the headgroup separation of PC at the surface of LUV and increases the binding maximum of the wedge-shaped amphipathic peptide without disrupting the membrane structure. In addition, the rigidifying effect of Chol on PC acyl chains prevents the penetration of the peptide into the bilayer interior. These findings imply that Chol in membranes affects the binding and motional freedom of exchangeable plasma apolipoproteins containing class A amphipathic sequences, e.g., apoA-I and apoCs.     

Application of artificial neural network model for the development of optimized complex medium for phenol degradation using <Emphasis Type="Italic">Pseudomonas pictorum</Emphasis> (NICM 2074)

Biodegradation of phenol using Pseudomonas pictorum (NICM 2074) a potential biodegradant of phenol was investigated for its degrading potential under different operating conditions. The neural network input parameter set consisted of the same set of four levels of maltose (0.025, 0.05, 0.075 g/l), phosphate (3, 12.5, 22 g/l), pH (7, 8, 9) and temperature (30°C, 32°C, 34°C) on phenol degradation was investigated and a Artificial Neural Network (ANN) model was developed to predict the extent of degradation. The learning, recall and generalization characteristic of neural networks was studied using phenol degradation system data. The efficiency of the model generated by the ANN, was tested and compared with the results obtained from an established second order polynomial multiple regression analysis (MRA). Further, the two models (ANN and MRA) were used to predict the percentage of degradation of phenol for blind test data. Performance of both the models were validated in the cases of training and test data, ANN was recommended based on the following higher coefficient of determination R ²; lower standard error of residuals and lower mean absolute percentage deviation.     

Liposomes induce chromosome aberrations in human cultured cells

The genotoxic effect of multilamellar lipid vesicles (MLV) was analysed on cultured heteroploid and diploid human cells. Dose-dependent reduction of cell survival and mitotic rate as well as induction of chromosome aberrations were observed. Chromatid and chromosome breaks and chromatid exchanges were found in 24-h culture after liposome treatment, whereas chromosome rearrangements were prevalent at 48 h. Neutral (PC/Chol) and positive (PC/SA) MLV showed a greater damage than negative (PS/PC; PS) MLV. Fibroblasts were the most sensitive cell type. In the case of PC/Chol MLV vesicles, control experiments with PC and Chol of controlled purity ruled out the possibility that the observed chromosome aberrations were caused by toxic oxidation products present in commercial preparations.     

Effect of reconstituted discoidal high-density lipoproteins on lipid mobilization in RAW 264.7 and CHOK1 cells

Reconstituted discoidal high‐density lipoproteins (rHDL) resemble nascent HDL, which are formed at the early reverse cholesterol transport steps, and constitute the initial cholesterol (Chol) acceptors from cell membranes. We have used different sized rHDL containing or not Chol, to test their abilities to promote cholesterol and phospholipid efflux from two different cell lines: Raw 264.7 macrophages and CHOK1 cells. All rHDL and lipid‐free apolipoprotein A‐I (apoA‐I) were found to be bound to CHO and RAW cells. In RAW cells, a positive correlation between cellular binding and Chol removal was found for 78 and 96 Å rHDL. Chol‐free rHDL were more effective than Chol‐containing ones in binding to RAW cells and promoting Chol removal. These results were more evident in the 96 Å rHDL. On the other hand, rHDL binding to CHO cells was relatively independent of disc size and Chol content. In spite of the fact that apoA‐I and rHDL promoted Chol efflux from both cellular lines, only in CHOK1 cells this result was also associated to decrease Chol esterification. Among choline‐containing phospholipids, only phosphatidylcholine (PC) (but not sphingomyelin) was detected to be effuxed from both cellular lines. With the only exception of Chol‐free 96 Å discs, the other rHDL as well as apoA‐I promoted PC efflux from RAW cells. Chol‐containing rHDL were more active than Chol‐free ones of comparable size to promote PC efflux from RAW macrophages. Regarding CHO cells, only apoA‐I and Chol‐free 78 Å rHDL were active enough to remove PC. J. Cell. Biochem. 113: 1208–1216, 2012. © 2011 Wiley Periodicals, Inc.     

Formulation and optimization of mouth dissolve tablets containing rofecoxib solid dispersion

The purpose of the present investigation was to increase the solubility and dissolution rate of rofecoxib by the preparation of its solid dispersion with polyvinyl pyrrolidone K30 (PVP K30) using solvent evaporation method. Drug-polymer interactions were investigated using differential scanning calorimetry (DSC), x-ray diffraction (XRD), and Fourier transform infrared spectroscopy (FTIR). For the preparation of rofecoxib mouth dissolve tablets, its 1∶9 solid dispersion with PVP K30 was used with various disintegrants and sublimable materials. In an attempt to construct a statistical model for the prediction of disintegration time and percentage friability, a 3² randomized full and reduced factorial design was used to optimize the influence of the amounts of superdisintegrant and subliming agent. The obtained results showed that dispersion of the drug in the polymer considerably enhanced the dissolution rate. The drug-to-carrier ratio was the controlling factor for dissolution improvement. FTIR spectra revealed no chemical incompatibility between the drug and PVP K30. As indicated from XRD and DSC data, rofecoxib was in the amorphous form, which explains the better dissolution rate of the drug from its solid dispersions. Concerning the optimization study, the multiple regression analysis revealed that an optimum concentration of camphor and a higher percentage of crospovidone are required for obtaining rapidly disintegrating tablets. In conclusion, this investigation demonstrated the potential of experimental design in understanding the effect of the formulation variables on the quality of mouth dissolve tablets containing solid dispersion of a hydrophobic drug.     

Characterization of the drug retention and pharmacokinetic properties of liposomal nanoparticles containing dihydrosphingomyelin

The drug retention and circulation lifetime properties of liposomal nanoparticles (LN) containing dihydrosphingomyelin (DHSM) have been investigated. It is shown that replacement of egg sphingomyelin (ESM) by DHSM in sphingomyelin/cholesterol (Chol) (55/45; mol/mol) LN results in substantially improved drug retention properties both in vitro and in vivo. In the case of liposomal formulations of vincristine, for example, the half-times for drug release (T(1/2)) were approximately 3-fold longer for DHSM/Chol LN as compared to ESM/Chol LN, both in vitro and in vivo. Further increases in T(1/2) could be achieved by increasing the drug-to-lipid ratio of the liposomal vincristine formulations. In addition, DHSM/Chol LN also exhibit improved circulation lifetimes in vivo as compared to ESM/Chol LN. For example, the half-time for LN clearance (Tc(1/2)) at a low lipid dose (15 micromol lipid/kg, corresponding to 8 mg lipid/kg body weight) in mice was 3.8 h for ESM/Chol LN compared to 6 h for DHSM/Chol LN. In addition, it is also shown that DHSM/Chol LN exhibit much longer half-times for vincristine release as compared to LN with the "Stealth" lipid composition. It is anticipated that DHSM/Chol LN will prove useful as drug delivery vehicles due to their excellent drug retention and circulation lifetime properties.     

Pitfalls and problems in studies on the methylation of phosphatidylethanolamine

Recent articles have confused the steady state concentration of radioactivity in N-methylphosphatidylethanolamine (PME) and N,N-dimethylphosphatidylethanolamine (PDE) with the amount of these products formed during the conversion of phosphatidylethanolamine (PE) to phosphatidylcholine (PC). This paper clarifies this problem and reports the apparent Km values for AdoMet and pH optima for the conversion of PE to PME, PDE, and PC by rat liver microsomes. We purified AdoMet and [methyl-3H]AdoMet and measured the transfer of tritium to PME, PDE, and PC as a function of time. There was an initial lag in the formation of [3H]PC followed by linear incorporation of isotope. In contrast, labeled PME and PDE reached and maintained steady state levels within 1 to 2 min. Hence, calculations of the rate of formation of PME, PDE, and PC must take into account the subsequent conversion of PME and PDE to PC. The PE N-methyltransferase was assayed at pH 6.6, 9.2, and 10.25 and the apparent Km for AdoMet for the three methylation reactions was calculated. The formation of PME was best estimated by the dpm in PME + 1/2 dpm in PDE + 1/3 dpm in PC. The synthesis of PDE from PME was estimated from 1/2 dpm in PDE and 1/3 dpm in PC, and the formation of PC from PDE estimated by 1/3 dpm in PC. The apparent Km for AdoMet at pH 10.25 for the conversion of PE to PME was 58 microM, PME to PDE was 65 microM, and PDE to PC was 96 microM. The pH optimum for each of these methylation reactions was 10.25. This high value was not due to alkaline degradation of AdoMet or denaturation of the enzyme. The apparent Km for AdoMet was also estimated for the conversion of exogenous PME to PDE (50 microM) and exogenous PDE to PC (45 microM). Since recent studies on the methylation of PE have not taken into account the conversion of newly formed PME and PDE to PC, the results and conclusions about apparent Km values for AdoMet, pH optima, and the number of enzymes involved must be re-evaluated.     

Proniosomal Oral Tablets for Controlled Delivery and Enhanced Pharmacokinetic Properties of Acemetacin

Free-flowing proniosomal powders of acemetacin (AC) were prepared using the slurry method and maltodextrin as carrier. Positively charged proniosomes composed of 70:20:10 of Span 60/cholesterol (Chol)/stearylamine (SA), respectively, were successively compressed into tablets using direct compression method. The tablets were characterized for weight variability, friability, hardness, drug content uniformity, and dissolution properties. The in vivo evaluation of the prepared proniosomes (powder or tablet forms) after oral administration was investigated by the determination of AC and its active metabolite indomethacin (IND) in the blood of albino rabbits. Results indicated that the increase of Chol from 10% to 20% markedly reduced the efflux of the drug. Further Chol addition from 30% to 50% led to increased AC release rates. The proniosome tablets of AC showed greater hardness and disintegration time and less friability than AC plain tablets. The dissolution of proniosomal tablets indicated a lower drug release percentage compared to powdered proniosomes and AC plain tablets. The mean pharmacokinetic parameters of AC and IND from different formulations indicated increased t_1/2 and area under the curve (AUC) of both AC and IND for proniosomal tablets compared with both proniosomal powders and AC plain tablets. This study suggested the formulation of AC proniosomal powder into tablets to control and extend its pharmacologic effects.KEY WORDS: acemetacin, proniosomes, sustained-release tablet, pharmacokinetics     

Rapid optimization of protein freeze‐drying formulations using ultra scale‐down and factorial design of experiment in microplates

Retaining biopharmaceutical proteins in a stable form is critical to their safety and efficacy, and is a major factor for optimizing the final product. Freeze‐dried formulations offer one route for improved stability. Currently the optimization of formulations for freeze‐drying is an empirical process that requires many time‐consuming experiments and also uses large quantities of product material. Here we describe a generic framework for the rapid identification and optimization of formulation excipients to prevent loss of protein activity during a lyophilization process. Using factorial design of experiment (DOE) methods combined with lyophilization in microplates a range of optimum formulations were rapidly identified that stabilized lactose dehydrogenase (derived from Lactobacillus leichmanii) during freeze‐drying. The procedure outlined herein involves two rounds of factorially designed experiments—an initial screen to identify key excipients and potential interactions followed by a central composite face designed optimization experiment. Polyethylene glycol (PEG) and lactose were shown to have significant effects on maintaining protein stability at the screening stage and optimization resulted in an accurate model that was used to plot a window of operation. The variation of freezing temperatures and rates of sublimation that occur across a microplate during freeze‐drying have been characterized also. The optimum formulation was then freeze‐dried in stoppered vials to verify that the microscale data was relevant to the effects observed at larger pilot scales. This work provides a generic approach to biopharmaceutical formulation screening where possible excipients can be screened for single and interactive effects thereby increasing throughput while reducing costs in terms of time and materials. Biotechnol. Bioeng. 2009; 104: 957–964. © 2009 Wiley Periodicals, Inc.