Using Polyethyline Oxide Blends to Assess the Effect of Excipient Variability

Published on: 
Pharmaceutical Technology, Pharmaceutical Technology-12-02-2009, Volume 33, Issue 12

The authors discuss a study that demonstrates the use of polyethylene oxide mixtures to investigate the effect of polymer viscosity on formulation robustness.

Robustness in pharmaceutical solid oral dosage forms means consistently providing the same drug-dissolution performance and shelf life from the dosage form when manufactured within the limits of the defined variations of key quality parameters, which include excipients, active pharmaceutical ingredients (APIs), and the manufacturing process. Under ideal circumstances, formulators would have available to them excipients with a broad range of properties with which to assess the impact of excipient variability on the formulation robustness. Realistically, however, the ability of the supplier to manufacture products at the extremes of a product specification is challenging given that manufacturing processes are designed to operate under optimal process conditions to produce a consistent product.

(GARYSLUDDEN/GETTY IMAGES)

For polyethylene oxide (PEO), blends may be the answer. When formulating PEO-based matrix systems, the key polymer attributes that affect drug dissolution are polymer viscosity and particle-size distribution. Because PEO does not have the additional complication of chemical substitution seen with many other excipients, blending is a straightforward approach for producing samples for formulation studies, provided the performance of the blended polymer mixture is representative of the standard product. PEO blends can provide a range of samples across a viscosity specification to achieve the necessary variability in product viscosity more easily than using batches of a single PEO product. These samples can then be used to assess the impact of product viscosity variability on formulation properties. This study used two standard PEO products—Polyox 205 National Formulary (NF) water-soluble resins (WSR) and Polyox N-12K NF WSR (Dow Chemical, Midland, MI)—to develop a series of samples across the viscosity-specification range of another standard product, Polyox 1105 NF WSR, which has a viscosity between these two grades. Polymer viscosity, molecular weight, tablet hardness, and drug dissolution were measured to determine the effect of PEO blends on excipient performance in extended-release matrix tablets.

Experiment

Materials. Excipients used for blending included Polyox 205 NF WSR and Polyox N-12K NF WSR (The Dow Chemical Company) with approximate molecular weights of 6 x 105 and 1 x 106 Da, respectively. Excipients used as standards for comparison were three samples of Polyox 1105 NF WSR (The Dow Chemical Company) with varying solution viscosities with an approximate molecular weight of 9 x 105 Da. The APIs were theophylline and diltiazem hydrochloride (HCl) (Spectrum Chemicals and Laboratory Products, Gardena, CA). Other ingredients included microcrystalline cellulose (MCC, Avicel PH 102, FMC Biopolymer, Newark, DE) and magnesium stearate (Spectrum Chemicals and Laboratory Products).

Methods. Three blends (25/75, 50/50, and 75/25) were prepared with Polyox 205 NF WSR and Polyox N-12K NF WSR using a mixer (Turbula T2F Heavy Duty Shaker–Mixer, Glen Mills, Clifton, NJ). Viscosity determinations for PEO samples were made using a digital viscometer (Brookfield DV-II+ viscometer, Brookfield Engineering Laboratories, Middleboro, MA), 5% aqueous solution, with spindle 2 at 2 rpm, except for the sample of Polyox N-12K NF WSR, which was run at 2% aqueous solution, spindle 1 at 10 rpm.

Table I: Viscosity of polyethylene oxide samples.

Weight-average molecular weight, number–average molecular weight, and polydispersity were determined by aqueous size-exclusion chromatography with a multiangle laser light-scattering (SEC-MALLS) detector. The SEC system consisted of a pump (Waters 2690 pump, Milford, MA), connected with a DAWN (MALLS) detector (Wyatt Technology, Santa Barbara, CA) and a Wyatt differential refractive index detector. The SEC fractionation of analyzed sample was evaluated to ensure an ideal SEC separation was obtained.

Table II: Weight-average and number-average molecular weight and polydispersity of polyethylene oxide (PEO) and PEO blends.

PEO samples were formulated into matrix tablets using theophylline and diltiazem HCl as model drugs. The formulations were comprised of the following by weight: 40% API; 30% polymer or polymer blend; 29.5% MCC; and 0.5% magnesium stearate. Tablets weighing 400 mg were prepared by direct compression using a 16-station tablet press (Manesty Beta,Knowsley, Merseyside, UK), equipped with 0.4063-in. (10.2-mm) flat-faced bevel-edge tooling at a compression force of 4000 lb (17.8 kN). Tablet hardness (n = 10) was measured with a hardness tester (Model HT-300, Key International, Englishtown, NJ). Drug dissolution was measured in a USP II apparatus (Varian/VanKel VK 7025, Varian, Palo Alto, CA) dissolution system with an ultraviolet–visible spectrophotometer, using three-prong capsule weights, in deionized water at 37.0 ± 0.5 °C, and a paddle speed of 50 rpm.

Figure 1: Differential weight fraction versus molar mass for Polyox 1105 water-soluble resin (WSR) samples and blend components, Polyox 205 NF WSR and Polyox N-12K WSR. (FIGURE COURTESY OF THE AUTHORS)

Results and discussion

Advertisement

Effect of blending on viscosity. Table I compares the viscosity values for blended PEO samples to individual samples of Polyox 205 NF WSR, Polyox 1105 NF WSR, and Polyox N-12K NF WSR. Samples obtained of commercially available Polyox 1105 NF WSR only covered a narrow viscosity range of 9120–11,520 centipoise (cP) within the broader product-viscosity specification of 8800–17,600 cP. The lack of Polyox 1105 NF WSR samples with viscosities at the upper extremes of the product specification make it difficult to assess the impact of variability in product viscosity on drug- formulation properties. Thus, blends of Polyox 205 NF WSR and Polyox N-12K NF WSR were prepared in varying proportions to achieve a range of sample viscosities representative of Polyox 1105 NF WSR product viscosity. The viscosity results showed that blending of Polyox NF WSR product grades adjacent to the targeted Polyox product grade is a reasonable approach for obtaining representative Polyox NF WSR product viscosities for formulation testing.

Figure 2: Differential weight fraction versus molar mass for Polyox 1105 water-soluble resin samples and polyethylene oxide blends. (FIGURE COURTESY OF THE AUTHORS)

Effect of blending on molecular weight. Several questions can arise concerning blends of materials that contain different molecular weights. Will the blend have a bimodal distribution or show a tailing effect at the high end? Will such differences in molecular weight have the potential to affect other formulation properties?

Table III: Tablet hardness for theophylline formulations containing polyethylene oxide (PEO) compared with formulations containing PEO blends.

Table II shows the molecular-weight data obtained via SEC–MALLS for the Polyox 1105 NF WSR samples, the Polyox 205 NF WSR and Polyox N-12K NF WSR blend components, and the blends. The weight-average molecular-weight values of the blends were between the values for each component, decreasing with increasing amounts of the lower weight-average molecular-weight component (Polyox 205 NF WSR). Standard deviations for the blends were < 1.6%. The weight-average molecular-weight values of the Polyox 1105 NF WSR samples were very similar, with standard deviations of < 0.36%. These standard deviations indicate excellent reproducibility for the SEC-MALLS measurement. A comparison of the approximate molecular weight of standard Polyox WSR samples provided in product literature (viscosity-average molecular weight) to the molecular weight determined by SEC-MALLS (weight-average molecular weight) shows significantly higher values for the SEC measurement. As discussed by Flory (1), viscosity-average molecular weight should be greater than number-average molecular weight and less than weight-average molecular weight for polymers with a polydispersity > 1. This principle explains why molecular-weight data in product literature is not directly comparable to results obtained by SEC-MALLS.

Table: IV: Tablet hardness for formulations of diltiazem hydrogen chloride containing polyethylene oxide (PEO) compared with formulations containing PEO blends.

Molecular-weight distribution obtained from SEC-MALLS showed little difference between Polyox 1105 NF WSR individual samples and both the Polyox 205 NF WSR and POLYOX N-12K NF WSR blend components or the blends themselves. PEO showed a broad molecular-weight distribution regardless of polymer molecular weight (see Figure 1). The molecular-weight distribution can vary to some degree for a specific PEO product as observed for Polyox 1105 NF WSR where PDI ranged from 4.52 to 5.74 (see Table II). The molecular-weight distributions for Polyox 205 NF WSR and Polyox N-12K NF WSR blend components were similar to that of Polyox 1105 NF WSR (see Figure 1). Blends of Polyox 205 NF WSR and Polyox N-12K NF WSR also had molecular-weight distributions similar to Polyox 1105 NF WSR (see Figure 2). Neither bimodality nor high-end tailing was observed in the blends. The SEC-MALLS results showed that blending of Polyox NF WSR product grades adjacent to the targeted Polyox product grade does not adversely impact molecular-weight distribution.

Figure 3: Drug dissolution for theophylline formulations containing polyethylene oxide compared with formulations containing polyethylene oxide blends. (FIGURE COURTESY OF THE AUTHORS)

Effect of blending on tablet hardness. Tables III and IV compare tablet hardness for PEO formulations containing theophylline and diltiazem HCl, respectively. Hardness was significantly higher (ANOVA, P < 0.05) for both theophylline and diltiazem HCl formulations containing PEO blends when compared to formulations using standard Polyox 1105 NF WSR products. The variation in tablet hardness was relatively small for theophylline formulations (0.5%) as opposed to diltiazem HCl formulations (13%) when comparing formulations with PEO blends to the formulations where standard Polyox 1105 NF WSR was used. The variation may be a result of particle-size effects in the PEO blends. Further investigation is required to understand the impact of particle properties on tablet-hardness properties in formulations containing PEO blends.

Figure 4: Drug dissolution for diltiazem hydrogen chloride (HCl) formulations containing polyethylene oxide compared with formulations containing polyethylene oxide blends. (FIGURE COURTESY OF THE AUTHORS)

Effect of blending on drug dissolution. A comparison of drug dissolution profiles was made using the similarity factor, f2, which measures the closeness between two profiles (2, 3). The US Food and Drug Administration has set a public standard of f2 value between 50–100 to indicate similarity between two dissolution profiles. Drug-dissolution profiles for both theophylline (see Figure 3) and diltiazem HCl (see Figure 4) formulations showed acceptable similarity for all formulations (f2 > 75 and f2 > 82, respectively). This indicates that the viscosity of PEO, whether as an individual product or a blend, did not significantly affect drug dissolution within the range evaluated. The rate of drug dissolution for formulations with blended PEO showed more variation in the first several hours of dissolution with the formulation containing theophylline (see Figure 5) than with the formulation containing diltiazem HCl (see Figure 6), suggesting that viscosity of PEO blends is slightly sensitive to drug solubility.

Figure 5: Rate of drug dissolution for theophylline formulations containing polyethylene oxide compared with formulations containing polyethylene oxide blends. (FIGURE COURTESY OF THE AUTHORS)

Conclusions

Blends of PEO polymers produced viscosities across the range of a standard PEO product specification and remained consistent with the typical PEO product to which it was compared. Molecular weight and polydispersity for PEO blends were consistent with standard PEO products, which showed a typical unimodal distribution. When formulated in a matrix tablet, the PEO blends showed good f2 similarity to typical PEO products when comparing drug-dissolution profiles. The rate of drug dissolution showed more variation in the first several hours of dissolution with theophylline for formulations containing PEO blends, suggesting that variability in the viscosity of PEO blends is slightly sensitive to drug solubility. This information demonstrates a reasonable approach to assess the impact of excipient variability in formulation development by blending standard products, and has practical implications in the design of robust extended-release, PEO-based formulations.

Figure 6: Rate of drug dissolution for diltiazem hydrogen chloride formulations containing polyethylene oxide (PE0) compared with formulations containing PEO blends. (FIGURE COURTESY OF THE AUTHORS)

Acknowledgments

The authors would like to thank Ali Rajabi-Siahboomi, PhD, director of scientific affairs; Marina Levina, PhD, senior manager of product development; and Xiaguang (Guang) Wen, PhD, formulation technologies manager, all of Colorcon (West Point, PA) for their input during this joint project.

Jennifer L'Hote-Gaston* is an application development specialist at Dow Wolff Cellulosics for The Dow Chemical Company, Midland, MI 48642, tel. 989.638.0794, fax 989.638.9836, jlhote@dow.com. Robert Schmitt, PhD, is a scientist, and Cheryl Karas is a laboratory technologist, both at Dow Wolff Cellulosics. Yongfu Li is a senior analytical chemist in the analytical sciences laboratory at The Dow Chemical Company.

*To whom all correspondence should be addressed.

Submitted: July 8. 2009; Accepted Aug. 17, 2009.

References

1. P.J. Flory, Principles of Polymer Chemistry (Cornell University Press, Ithaca, NY), Chapter VII-4c, pp. 311–314 (1953).

2. J.W. Moore and H.H. Flanner, Pharm. Tech. 20 (6), 64–74 (1996).

3. V. Shah, Y. Tsong, and P. Sathe, Pharm. Res. 15 (6) 889–896 (1998).