Understanding and Managing the Impact of HPMC Variability on Drug Release from Controlled Release Formulations

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Understanding and Managing the Impact of HPMC Variability on Drug Release from Controlled Release Formulations DELIANG ZHOU, DEVALINA LAW, JUDIE REYNOLDS, LYNN DAVIS, CLIFFORD SMITH, JOSE L. TORRES, VIRAJ DAVE, NISHANTH GOPINATHAN, DANIEL T. HERNANDEZ, MARY KAY SPRINGMAN, CASEY CHUN ZHOU AbbVie Inc., North Chicago, Illinois 60064 Received 20 December 2013; revised 26 February 2014; accepted 4 March 2014 Published online 20 March 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.23953 ABSTRACT: The purpose of this study is to identify critical physicochemical properties of hydroxypropyl methylcellulose (HPMC) that impact the dissolution of a controlled release tablet and develop a strategy to mitigate the HPMC lot-to-lot and vendor-to-vendor variability. A screening experiment was performed to evaluate the impacts of methoxy/hydroxypropyl substitutions, and viscosity on drug release. The chemical diversity of HPMC was explored by nuclear magnetic resonance (NMR), and the erosion rate of HPMC was investigated using various dissolution apparatuses. Statistical evaluation suggested that the hydroxypropyl content was the primary factor impacting the drug release. However, the statistical model prediction was not robust. NMR experiments suggested the existence of structural diversity of HPMC between lots and more significantly between vendors. Review of drug release from hydrophilic matrices indicated that erosion is a key aspect for both poorly soluble and soluble drugs. An erosion rate method was then developed, which enabled the establishment of a robust model and a meaningful HPMC specification. The study revealed that the overall substitution level is not the unique parameter that dictates its release-controlling properties. Fundamental principles of polymer chemistry and dissolution mechanisms are important in the C 2014 Wiley Periodicals, Inc. and the development and manufacturing of hydrophilic matrices with consistent dissolution performance.  American Pharmacists Association J Pharm Sci 103:1664–1672, 2014 Keywords: controlled release; HPMC variability; hydrophilic matrices; erosion; diffusion; dissolution; NMR

INTRODUCTION Extended-release (ER) dosage forms may provide a number of benefits including reduced dosing frequency, improved efficacy, reduced adverse events, and improved patient compliance, and therefore have competitive advantages over their immediate release counterparts. Among the various types of ER dosage forms, the hydrophilic matrix is the most widely used platform of drug delivery. It is versatile and can accommodate both low and high drug loading, and drug molecules with a wide range of physicochemical properties. In addition, a hydrophilic matrix is generally more straightforward to manufacture, scale up, and more cost effective than other types of more complicated ER dosage forms, such as reservoir and osmotic pumps. Hydroxypropyl methylcellulose (HPMC) is a commonly used release-controlling polymer in hydrophilic matrix tablets.1,2 A plethora of information exists in the pharmaceutical literature on drug release from hydrophilic matrices, such as the general role of drug diffusion versus polymer dissolution/matrix erosion,3–5 gel swelling and drug diffusion,6,7 the general release mechanisms,8–11 the fundamental HPMC properties12,13 relevant to controlled release, and a variety of specific drug formulations. The basic premise of the polymer’s controlled release functionality is that when it comes into contact with dissolution medium in vitro or in vivo, the polymer at the surface of the tablet quickly hydrates and forms a surrounding gel layer that serves as a physical barrier to the tablet core and regulates the release of drug from the matrix.

Correspondence to: Deliang Zhou (Telephone: +847-938-2823; Fax: +847-9360095; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 103, 1664–1672 (2014)

 C 2014 Wiley Periodicals, Inc. and the American Pharmacists Association

1664

The consensus of the scientific community is that drug release from a hydrophilic matrix occurs primarily via two routes: (1) diffusion of drug molecules through the gel layer, and (2) movement of drug molecules and/or undissolved drug particles with the erosion of gel layer of the tablet matrix, caused by polymer erosion at the solvent edge of the gel layer. It is also believed that soluble drugs are primarily released by a diffusion mechanism, whereas insoluble drugs are released by an erosion mechanism. Given the important role of the gel layer in controlling drug release from hydrophilic matrices, the development of the gel layer (i.e., its time course) during dissolution is considered an important aspect that may impact the overall drug release. For HPMC gels, the influx of water into the polymer decreases the effective polymer concentration, and therefore lowers its effective glass transition temperature, Tg . The resulting abrupt glassy-rubbery transition leads to the formation of a gel layer.5 While the influx of water causes the gel layer to expand inward and increase with time, the concentration of the polymer in the gel layer decreases as more water influxes in. At the surface of the gel layer adjacent into the dissolution medium, the polymer concentration is reduced to the critical disentanglement concentration so that disentanglement of the polymer chains takes place and polymer dissolves, causing the gel layer to shrink. Hence, the gel layer thickness is governed by two concurrent, but opposite, processes. In the early stage, the influx of water dominates and the gel layer thickness increases with time. However, with the increase in the gel layer thickness, water needs to travel over an increasingly longer distance in order to reach deeper inside the tablet, and therefore, the overall water penetration rate decreases and eventually approaches the rate of gel erosion in a later stage and a plateau may be observed where the gel layer thickness remains constant (i.e.,

Zhou et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:1664–1672, 2014

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

steady state). In the terminal stage, when the entire tablet core becomes a gel, the dimension of the gel will then start to shrink because polymer erosion is the only remaining process. The gel layer thickness development may be simulated using established mass transport principles, such as the work reported by Harland et al.3 and Lee and Peppas.4 Please note that polymer erosion and polymer dissolution are used synonymously in this text. However, they do not mean exactly the same as partial chain entanglement may remain during erosion but not dissolution. Nevertheless, the two processes are highly correlated and polymer erosion can be estimated from polymer dissolution. The molecular weight of HPMC impacts the diffusion of water into the gel layer differently than it impacts the dissolution of the polymer into the dissolution medium. Diffusivity of water and a number of small molecules in HPMC gels has been studied using nuclear magnetic resonance (NMR) and other techniques and has been found largely independent of the molecular size of HPMC/viscosity grade.7,14 This finding is plausible because the polymer is much larger than the small molecule probes; therefore, the environment around the probes is essentially identical even though the size of the polymer may be different. However, the rate of polymer dissolution is dependent on its size because of the differences in the effective disentanglement concentrations, which increases with the size of a polymer. Ju et al.5,15 proposed a power law model to account for the dependence of polymer dissolution on its molecular weight/viscosity grade. Despite the extensive use of HPMC for controlled release drug delivery and extensive studies devoted to understanding drug release mechanism, reproducibly controlling drug release from HPMC matrices has not always been straightforward. Significant challenges to obtaining reproducible dissolution performance may occur in the manufacture of hydrophilic matrix tablets, as evidenced by some limited reports,16,17 and perhaps far more unreported cases exist across the pharmaceutical industry. When it occurs, commercial production may get disrupted, potentially resulting in drug shortages and adversely impacting patients. Conventional thinking is that properties of HPMC, namely the overall substitution,13,18 that is, methoxy (MeO) and hydroxypropyl (HP), and the viscosity/molecular weight,5,13,15,19 are important factors that impact drug release from HPMC matrices. Indeed, the various HPMC types are defined in the compendia, for example, United States Pharmacopeia (USP) or European Pharmacopeia (EP), by the level of substitution, and further divided by their viscosity grades. However, each HPMC substitution type as defined in the compendia covers a broad range of levels. For example, hypromellose type 2208 (e.g., the type K chemistry) allows MeO substitution ranges of 19.0%–24.0%, and HP ranges of 4.0%–12.0%. Given this broad chemical space, it is hardly conceivable that the HPMC properties will be equivalent from one boundary to the other. In addition, the average level of substitution may be too simplistic, as evidenced by more recent findings20–23 on the chemical heterogeneity of HPMC substitution and the potential impact on its solution and gelling properties and thus, on drug release characteristics. This investigation attempts to elucidate the reasons behind the variation observed in the in vitro release profile for a watersoluble molecule, niacin, from an HPMC-based controlled release formulation. These observed variations were found to beDOI 10.1002/jps.23953

1665

come significant when evaluating alternative vendors. A fundamental investigation into polymer chemistry revealed that comparability assessment based on compendial specifications of the functional excipient HPMC is not sufficient to ensure comparable drug release, as markedly different chemistries can meet compendial requirements while impacting the mechanism involved during the drug release process; therefore, a new test for HPMC was developed that can meaningfully predict the drug release from the product.

EXPERIMENTAL Materials Batches of HPMC, substituent grade of USP hypromellose 2208, and viscosity grade of 15,000 mPa s were obtained from Dow (Dow Chemicals, Michigan), Shin-Etsu (Shin-Etsu Chemical Company Ltd., Tokyo, Japan), and Hercules (Hercules Doel BVBA, Doel, Belgium), which are commercially designated as K15M, METOLOSE 90SH-15000, and Benecel K15M PH, respectively. Niacin was obtained from Lonza AG (Visp, Switzerland), stearic acid was obtained from PMC Biogenix (Memphis, Tennessee), and Povidone K90D was obtained from ISP Technologies (Calvert City, Kentucky). Ammonium acetate, crystal, and glacial acetic acid were purchased from J.T. Baker (Phillipsburg, New Jersey). Purified water was obtained through a MilliQ system (Millipore, Billerica, Massachusetts). R

Nuclear Magnetic Resonance Nuclear magnetic resonance spectra were recorded at 25◦ C on a Varian Inova 500 MHz spectrometer equipped with a cold probe using the VNMR 6.1 C software package. The standard 1 H spectra were collected with a 45◦ pulse flip angle, 3.25 s acquisition time, and total 8.25 s relaxation delay. A known quantity of trimethylsilyl propionate (TSP)d4 was dissolved in D2 O solvent and TSP-d4 was used as an internal standard for quantification in proton spectra. About 10 mg HPMC was added to 1.5 mL D2 O containing TSP-d4 and mixed overnight to ensure that the HPMC was well dissolved. The pulse programs of 2D experiments were taken from the Varian software library. Sucrose, 1-methoxy-2-propanol, and 1,2-dimethoxy propane were used to test 2D pulse programs such as gHSQC, gHMQC, HSQC, and HMQC for quantification and to define appropriate parameters. HMQC experiments were selected for the quantification with a 7 s delay. Preparation of Tablets Niacin (81%, w/w), povidone (2.8%, w/w), HPMC (15.7%, w/w), and stearic acid (0.5%, w/w) were blended in a laboratory V-blender for 6 min at 26 rpm. The blend batch size ranged from 100 to 300 g. All raw materials were passed through 18– 20 mesh screens before use. The blends were compressed into 19 × 9.6 mm2 oval tablets of approximately 1.23 g on a compaction simulator (PressterTM ; Metropolitan Computing Corporation, East Hanover, New Jersey), at a precompression force of approximately 5 kN, a main compression force of approximately 25 kN, and a target hardness of 196 N (range 176–225 N). The simulator duplicates a compression speed of 30 rpm on a Fette 2090 tablet press with 30 stations (or 64,800 tablets per hour). Zhou et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:1664–1672, 2014

1666

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Potency Assay and Drug Release Potency of the tablets was determined using HPLC with a Waters :Bondapak C18 RP, 15 cm × 3.9 mm column. The mobile phase consisted of water/methanol (44:56, v/v) with 2.5% Waters PIC B-7 (or equivalent) as ion pairing agent, at a flow rate of 1 mL/min. UV detection at 254 nm was employed. Dissolution was performed with a USP Apparatus I in 900 mL purified water with basket rotation of 100 rpm and bath temperature of 37.0 ± 0.5◦ C. Dissolution samples were pulled at 1, 3, 6, 9, 12, and 20 h through automated dissolution sampling fitted with a 10 :m full flow filter (Quality Lab Accessories QLA, Bridgewater, New Jersey), and analyzed by the same HPLC assay method. Replicates from six tablets are performed. R

Preparation of HPMC Compacts and Measurements of HPMC Dissolution Rate Compacts were prepared by transferring approximately 300 mg of HPMC to a rotating disk intrinsic dissolution die (Varian Inc., Cary, North Carolina) with a 0.8 cm diameter cavity. The powder was compressed using a hydraulic bench top Carver press and controller (Carver, Inc., Wabash, Indiana) set at a compression force of 1000 lbs (∼4.45 kN) at a pump speed of 25% and a dwell time of 30 s. Dissolution of the HPMC compacts was performed in a Varian VK7000 dissolution bath with a VK8000 autosampler (Varian Inc.). The die assembly was affixed to the intrinsic dissolution spindle assembly and the height of the compact was set at 3.8 cm from the bottom of the vessel. Dissolution was conducted in 900 mL of degassed purified water maintained at 37.0 ± 0.5◦ C, at a rotation speed of 100 rpm. Samples were collected at 10, 15, 20, 25, 30, and 35 h and filtered with a 10 :m full flow filter (Quality Lab Accessories QLA, Bridgewater, New Jersey). Assay of HPMC A size-exclusion chromatography (SEC) method with evaporative light scattering detector (ELSD) was used to assay the concentration of HPMC in the HPMC dissolution samples. The HPLC system was composed of: two LC-10AD VP pumps, a SIL-20ACHT auto-sampler with cooler, a DGU-14A online degasser, a SCL-10AVP system controller, and an ELSD-LT evaporative light scattering detector (Shimadzu Scientific Instruments, Colombia, Maryland). Atlas software was used for data acquisition. A Shodex OHpak SB-804 HQ column, 10 :m, 8.0 × 300 mm2 (Showa Denko K.K., Kawasaki, Japan) was employed, along with a Guard column (Shodex OH pak SB-G), at ambient temperature. The mobile phase was 0.1 M ammonium acetate pH 6.0 buffer at a flow rate of 1.0 mL/min. Injection volume was 50 :L. ELSD settings was as follows: nitrogen flow of 350 kPa, drift tube temperature of 80◦ C, and gain of 10. The run time was 15 min. A calibration curve was used to calculate the concentration of HPMC in the dissolution samples. Working HPMC standards were prepared from the stock solution in the concentrations ranges of 5–200 ppm. Dissolution samples were injected neat, with no dilution.

RESULTS AND DISCUSSION Niacin ER tablet is an ER hydrophilic matrix tablet based on HPMC. It has a high drug loading and approximately 16% Zhou et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:1664–1672, 2014

Figure 1. Control chart of niacin release at 20 h, segmented by HPMC lot.

Table 1.

Batches of HPMC and Resulted Variability on Drug Release

HPMC Lot Lot 1 Lot 2 Lot 3 Lot 4 Lot 5 Lot 6 Lot 11 Lot 13 Lot 14 Lot 15 Lot 16 Lot 17 a

Vendor

MeOa (%)

HPa (%)

Viscositya (mPa s)

D20h Actual (%)

A A A A A A A B B C C C

21.8 22.6 22.7 22.8 22.9 22.8 22.8 23.1 23.3 23.9 23.9 23.9

9.1 8.8 9.4 9.8 9.2 9.6 9.8 9.7 9.5 8.3 8.6 8.4

16,696 21,903 17,218 14,342 17,645 23,724 18,064 14,400 14,700 18,063 20,362 18,496

71 76 82 91 77 73 79 94 92 92 93 92

Values from certificate of analysis (COA).

HPMC. Figure 1 is a control chart of niacin release at 20 h, hereafter designated as D20h, segmented by the HPMC lot used in the manufacture. Although all these HPMC lots were obtained from a single vendor, it is apparent that the drug release was impacted by the lot of HPMC. Some of the tablet lots barely passed the lower specification limit of not less than 75% and several lots failed. The low dissolution process capability of this product presented a significant challenge to uninterrupted commercial production and posed potential quality risks. The properties of a few selected HPMC batches from vendor A and the resulted tablet dissolution results are listed in Table 1. There were noticeable differences in the 20 h dissolution, ranging from approximately 70% to 90%. Additional tablets were manufactured with HPMC from two alternative vendors (referred to as B and C). The properties of these HPMC batches and the D20h results are also included in Table 1. All HPMC batches from vendor B or vendor C resulted in much higher drug release at 20 h (92%–94%), despite some significant differences in their properties (such as HP content and viscosity). The observed lot-to-lot variability and vendor-to-vendor variability cannot be fully explained by the MeO substitution, HP content or viscosity, and warrants further investigation. DOI 10.1002/jps.23953

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

1667

Empirical Dissolution Model Based on HPMC Substitution and Viscosity Conventionally, the degree of MeO substitution, the HP content, and viscosity have been considered to be the critical properties of HPMC that generally determine its applications.1,13,18 This is reflected in the various pharmaceutical compendia (e.g., USP and EP), where the types of HPMC chemistry (A, E, F, and K) are defined by substitution type based on the levels of MeO and HP. However, the substitution ranges are broad (e.g., MeO of 19.0%–24.0% and HP of 4.0%–12.0% for hypromellose 2208) and the properties of HPMC could be different from one boundary to the other. Therefore, a specific ER formulation may require HPMC with much narrower substitution ranges, depending on the physicochemical properties of the drug molecule and the dosage design. In an attempt to predict tablet dissolution, experiments were conducted with HPMC batches of varying MeO/HP levels and viscosity. A conventional design of experiment (DOE) was found difficult in this case due to limitations in obtaining HPMC batches with desired properties in all aspects. Instead, attempts were made to spread the MeO/HP substitutions over typical ranges as much as possible (Figure 2), then the viscosity values were screened in order to reasonably randomize its values along with the substitution levels. Therefore, this is not an orthogonal design. In this experiment, a large number of HPMC lots were used because of the concerns that there might be some errors in the determination of the HPMC substitution levels as well as initial observation of some “outliers.” The primary purpose of this study was to screen which HPMC properties might be correlated to the dissolution of niacin ER tablets. To minimize the potential differences in chemistry, all HPMC lots used in the study came from vendor A. Tablet dissolution results from this above study were analyzed using JMP statistical software (version 8; SAS Institute Inc., Cary, North Carolina) using multiple linear regression (standard least square). The particle sizes information (fraction passing through a 230-mesh screen) was also included in the analysis. All factors were treated as continuous variables. A statistical analysis of the DOE results indicated that the MeO substitution (p = 0.79) and the particle size (p = 0.48) had no significant impact on the dissolution of the product. Therefore, a final model was developed based on the level of HP

Figure 2. Design space for HPMC substitutions. DOI 10.1002/jps.23953

Figure 3. A dissolution model based on HPMC substitution and viscosity.

substitution (p < 0.0001) and viscosity (p = 0.07). Two of the HPMC lots were suspected to be outliers and were cautiously removed from the final analysis in order not to skew the model. Figure 3 illustrates good consistency between the actual and predicted dissolution based on the final model: D20h(%) = 7.948 + 9.214 × HP% − 0.0004671 × viscosity (mPa s)

(R2 = 0.795)

The finding that the tablet dissolution is primarily driven by the HP content but not by the MeO substitution is not surprising. The thermal gelation properties of HPMC that underlie its release controlling and other applications are a result of the balance between the polymer–solvent and polymer– polymer interactions, which are modulated by the MeO and HP substitutions.13 Although it has been shown that the MeO substitution provides the primary hydrophobe–hydrophobe interaction for HPMC to gel in solution, it is still possible that the range of MeO substitution in this study is not broad enough to cause a meaningful change to affect the dissolution of niacin ER tablets. Our observation was similar to the experiences reported by Dahl et al.16 and Lucisano et al.17 when evaluating HPMC from different sources. Increasing the HP substitution has been shown to increase the incipient gelation temperature in 2% HPMC solution and decrease the gel strength,13 and both effects are consistent with weakened overall hydrophobic interactions. Because the HP substitution, similarly to the MeO substitution, makes the cellulose more hydrophobic, this observed effect of HP on drug release must be largely attributable to the stronger steric effect of the HP group, which tends to push adjacent polymer chains apart, disrupt the hydrophobic interactions between MeO units so hinders the polymer–polymer interaction, resulting in weakened gels and increasing the gelation temperature. This explanation is consistent with our observation that the dissolution of niacin ER tablets generally increases with increasing HP content for HPMC obtained from vendor A. Further evaluation revealed that a number of HPMC lots from vendor A did not conform to the dissolution model derived Zhou et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:1664–1672, 2014

1668

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Table 2. Summary of HPMC Lots That do not Conform to the Substitution/Viscosity Model HPMC Viscositya D20h D20h Pred– a Lot Vendor HP (%) (mPa s) Pred (%) Actual (%) Actual (%) Lot 1 Lot 5 Lot 6 Lot 11 Lot 15 Lot 16 Lot 17 a

A A A A C C C

9.1 9.2 9.6 9.8 8.3 8.6 8.4

16,696 17,645 23,724 18,064 18,063 20,362 18,496

84 84 85 90 76 78 77

71 77 73 79 92 93 92

13 7 12 11 −16 −15 −15

Values from COA.

based on the HP substitution level and the viscosity, as shown in Table 2. In the evaluation of alternative sources, dissolution results for tablets manufactured with a limited number of HPMC lots from vendor B were comparable to predictions from the empirical substitution/viscosity model, albeit their HP substitutions are all high (9.5%–9.7%). However, all of the HPMC lots from vendor C generated tablet dissolution that were much higher than the empirical model prediction (i.e., negative prediction error), in contrast to the positive prediction errors for those exception lots from vendor A (Table 2). The inconsistency of the substitution/viscosity prediction model suggested that there might be differences in the HPMC structural features that were not captured by the average substitution levels or viscosity. Therefore, additional studies were conducted to better understand the potential structural differences of HPMC. NMR Investigation One possible explanation for the inconsistency of the empirical prediction model was the potential experimental error in measuring the substitution levels of HPMC. It has been known that the current compendia method used for determining HPMC substitutions, Zeissel-GC, may have relatively high variability due to the volatile reagent (hydriodic acid) and elevated temperature used. Hence, an alternative approach to HP determination was undertaken to evaluate this possibility. A 1 H NMR method was developed to quantify the HP substitution of HPMC, based on the HP signal at the chemical shift of 1.1–1.2 ppm. TSP was used as an internal standard for the purpose of quantification. In general, the results were comparable to the HP contents reported on the COA, suggesting that the experimental errors were not a primary cause of the differences between actual and predicted dissolution results. Subsequently, a heteronuclear multiple quantum coherence (gHMQC) was developed to further probe the potential chemical differences of the various types of HPMC substitutions. Figure 4 is an example gHMQC spectrum. The HP group is shown to split at slightly different 13 C chemical shifts, representing two different sub-groups. Because each HP has two hydroxyl groups, the substitution reaction can take place on one of the three hydroxyl groups in the anhydroglucose unit (AGU), or it can occur at the free hydroxyl group located on an existing HP substituent. Let HP1 represents the native HP whose hydroxyl is free and HP2 represents the group whose hydroxyl group is further substituted by either a MeO group or another HP Zhou et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:1664–1672, 2014

Figure 4. gHMQC spectrum of HPMC. Table 3. Distribution of the Hydroxypropyl Substitution for Selected HPMC Lots HPMC HP1 :HP2 Lot Vendor (NMR) Lot 1 Lot 3 Lot 4 Lot 5 Lot 6 Lot 7 Lot 8 Lot 9 Lot 10 Lot 11 Lot 12 Lot 13 Lot 17 Lot 18 a

A A A A A A A A A A A B C C

3.0:1 3.2:1 3.3:1 3.2:1 3.1:1 3.5:1 3.3:1 3.6:1 3.2:1 3.3:1 3.0:1 1.9:1 2.0:1 2.3:1

MeOa (%) 21.8 22.7 22.8 22.9 22.8 23.8 23.2 24.0 23.4 22.8 23.0 23.1 23.9 23.9

D20h HPa (%) D20h (%) Pred (%) 9.1 9.4 9.8 9.2 9.6 8.7 8.0 8.3 10.0 9.8 9.5 9.7 8.4 8.4

71 82 91 77 73 79 72 73 90 79 85 94 92 87

84 87 92 84 85 79 72 75 91 90 86 91 77 77

Values from COA.

group, then the ratio between these two groups represents the chemical makeup of the HP substitution. The ratio between these two types of HP groups is listed in Table 3 for a number of HPMC lots. Although some variability exists for this ratio even within the same vendor, the most striking difference is that the HPMC chemistry for vendor A differs significantly from that for vendors B and C. HPMC from vendor A has more native HP groups (HP1 :HP2 ∼ 3.3:1), whereas HPMC from vendors B and C have more HP groups that are further substituted (HP1 :HP2 ∼ 2:1). No significant difference was observed for HPMC from B and C. The structural differences in the distribution of the HP groups may impact the gel properties and, therefore, drug release. The HP2 (i.e., further substituted) group is bulkier than the HP1 (i.e., native) group, so that more HP2 will weaken the gel due to the increased steric effect, while keeping the total HP content the same. On the contrary, the HP2 group may exist as oligomers (although they are not distinguished with our current NMR technique), and the HP oligomer may be long enough to impart increased hydrophobic interactions due to their capability to overlap with the hydrophobic MeO and HP DOI 10.1002/jps.23953

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

groups on the neighboring polymer, thus strengthens the gel. Based on our findings on drug release, it appears that the enhanced steric hindrance plays a more dominant role, consistent with the findings in the substitution model. Hence, HPMC lots with more HP2 population (such as vendor B and vendor C) resulted in dissolution that was on the fast side, regardless of whether the overall HP content is high (e.g., 9.5–9.7 for vendor B) or lower (e.g., 8.3–8.6 for vendor C). It has also come to our attention that the chemical makeup of HP groups in hydroxypropyl cellulose (HPC) was previously reported to impact dissolution of hydrochlorothiazide tablets where HPC was used as a binder.24 The substitution/viscosity modeling study using HPMC lots from vendor A indicated that drug release generally increased when the HP content was higher. However, this model does not differentiate the differences between the two different types of HP shown in the above discussion. Therefore, the empirical model is not expected to be applicable to HPMC lots manufactured by a different vendor if the chemical makeup of the substitution is different (i.e., vendors C). The agreement of dissolution of HPMC from vendor B with this model may simply be a coincidence because the evaluated vendor B HPMC lots all have high levels of HP (9.5–9.7). Based on the NMR investigation, HPMC from vendor C is similar to those from B, except the overall HP level is lower (HP ∼8.3%–8.6%). Hence, HPMC from B and C may be considered as complement to each other with regard to the overall HP contents within the same HP1 :HP2 chemistry. It is then clear that the HPMC chemistry of B or C (HP1 :HP2 ∼ 2:1) does not fit into the model that was developed based on HPMC chemistry of vendor A (HP1 :HP2 = 3.3:1). For example, HP level significantly impacts drug release for HPMC from vendor A chemistry, while drug release is high and is largely independent on the HP level for vendor B and C chemistry, at least in the range studied. The underlying reasons for the nonconformity of some lots from vendor A (lots 1, 5, 6, and 11) are not clear. However, it may be related to the interlot variability in the HPMC chemistry. Vastly different possibilities exist when the substitution reaction of the cellulose is conducted during the manufacturing of HPMC. In the total of three free hydroxyl groups in each of the AGUs of cellulose, the C-2 hydroxyl is the most reactive based on its greatest acidity,20 followed by the C-6, while C-3 hydroxyl is the least acidic. The reactivity of the C-3 hydroxyl group is further reduced due to its engagement in the intramolecular hydrogen bonding in crystalline cellulose.25 However, the C-3 hydroxyl reactivity may be enhanced upon substitution of the neighboring C-2 hydroxyl, probably because of the disruption of the intramolecular hydrogen bonding. Therefore, during manufacturing, the substitution reaction may not occur at a random hydroxyl group because of the interplay of substitution and the subsequent modification of reactivity. In addition to the difference in reactivity among the three substitutable hydroxyl groups in the same AGU, large difference may also exist in the distribution of the substitution along the polymer chain. The starting material cellulose is generally semi-crystalline or microcrystalline, and there exists a significant amount of disordered or amorphous regions. Because of the heterogeneous nature of the reaction where cellulose only swells in the alkaline environment, the initial substitution is likely occurring on the crystal surface or in the amorphous/disordered regions where those hydroxyl groups are more accessible and more reactive. However, initiation of a substitution opens up the celDOI 10.1002/jps.23953

1669

lulose chain, creates more disordered cellulose, and therefore enhances further reactions around the vicinity of the existing substitution and may cause significant heterogeneity on the substituent distribution along the polymer chains. Therefore, the substituent may get clustered along the polymer chain instead of a true random distribution across it. The above structural differences of HPMC substitution (deviation from a statistical distribution on the AGU level and along the polymer chain) may be called the spatial heterogeneity, and together with the differences in chemical makeup of the substituent (e.g., HP1 and HP2 ), they may be collectively termed as the chemical heterogeneity of HPMC substitution. The reality is, HPMC is intrinsically heterogeneous depending on the quality of the wood pulp raw material and the specific controls applied in the chemical reaction during the manufacturing process. Attention to the spatial heterogeneity of HPMC substitution has increased in recent years, mostly due to improved analytical techniques.20,26,27 A number of enzymes can selectively break down the (1→4)-$-D-glycoside of the cellulose chain. Because of the steric hindrance presented by the substituent groups on the AGU, the densely substituted area of HPMC are not hydrolyzed by these enzymes, whereas the less substituted areas are more prone to enzymatic break down. Hence the extent to which an HPMC lot is hydrolyzed can be used as an indication of its substitution distribution along the polymer chain. For example, more randomly substituted HPMC will be less hydrolyzed than HPMC with substituent groups that are more clustered around certain regions of the HPMC chain. In conjunction with SEC, matrix-assisted laser desorption–ionization time-of-flight mass spectrometry, gas chromatography with flame ionization detector, high-performance anion exchange chromatography with pulsed amperometric detection, and other detecting techniques such as NMR and FTIR, useful information may be deduced on the substitution patterns of HPMC. These above techniques are helpful because they provide structural information on the cellulose segments that are broken down during the enzymatic treatments, thus can assist pinpointing structural characteristics of the original cellulosic backbones. The substitution heterogeneity of HPMC has recently been shown to impact important properties of HPMC such as cloud point, swelling, and drug release.21–23,28,29 In summary, the chemical heterogeneity of HPMC is a very important aspect that may impact the key release-controlling properties of this polymer. However, characterization of the heterogeneity is analytically challenging, time-consuming, and at most qualitative in nature. It is also possible that a number of different HPMC chemistries may result in equivalent controlled release properties. Therefore, although characterizing the chemical heterogeneity of HPMC lots provides great insight into the potential impact on drug release, it is not yet suitable to be used as a primary quality tool and an alternative approach is needed. Drug Release from Hydrophilic Matrices Significant difference in the molecular weight of HPMC of the same chemistry grade (such as different viscosity grades) is known to impact polymer erosion as indicated previously.15 In addition, the erosion rate may be impacted by the chemistry of HPMC, such as the chemical makeup of the substituent group (e.g., the native HP or further substituted HP as characterized in our NMR investigation), and/or the spatial distribution of the Zhou et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:1664–1672, 2014

1670

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

substituent groups at the AGU level and at the polymer chain level. The gel may have different characteristics (i.e., swelling, cloud point)21,22,28 when it is formed with HPMC that is more homogeneously substituted versus HPMC that is significantly heterogeneous. Therefore, differences in the erosion rate of HPMC lots may account for the observed differences in the dissolution of niacin ER tablets at 20 h. However, this hypothesis has to be examined at a more fundamental level because of its apparent conflict with conventional wisdom for water-soluble drugs. It has been contended in literature that soluble drugs are primarily released by a diffusion mechanism and insoluble drugs by an erosion mechanism.30–32 Therefore, based on this notion, release of water soluble drugs from hydrophilic matrices is unlikely to be significantly impacted by the erosion mechanism. However, the erosion was shown to play a significant role in our discussion on the development of gel layer thickness. A slower polymer erosion rate leads to an increased gel layer thickness, keeping other factors the same, which reduces the flux of drug molecules via the diffusion route. Of course, slower polymer erosion will also reduce drug release via the erosion mechanism. Hence, the resulted decreases in both diffusion and erosion mechanisms lead to a slower overall drug release rate. An empirical approach may be used to derive a “controlling” mechanism based on fitting of the release data to the simple exponential equation below10,11 : Mt = ktn M∞ where M is the amount of drug released, t is time, n is the diffusional exponent, k is a fitting constant, and the subscripts t and ∞ designate quantities at time t and infinity, respectively. This leads to n of 0.5 (i.e., square root of time) when drug releases via the diffusional mechanism and n of 1.0 when the erosion mechanism operates, assuming a film/slab geometry, whereas n of 0.5–1.0 indicates a combination thereof. The critical values of the exponents are somewhat different for other geometries.10,11 For example, the critical diffusional exponents are 0.45 and 0.43 for Fickian diffusion from cylinders and spheres, respectively. The corresponding limiting values for erosion-based release are 0.89 and 0.85, respectively. The critical values are not available for geometries other than the cylinder and sphere (such as the oval tablet in our case). However, they are expected to lie in between those values obtained from cylinder and spheres, given their intermediate aspect ratios. The dissolution profiles of niacin ER tablets prepared using four different HPMC lots are presented in Figure 5. The dissolution data up to 60% were fitted to the above equation. The values of the diffusional exponent, n, obtained from fitting to the equation ranged from 0.69 to 0.76, indicating combined drug release mechanisms of diffusion and erosion, or the so-called anomalous transport mechanism. Although the interpretation of the exact numerical value of the diffusion exponent is not straightforward, it still provides a qualitative picture on the relative contribution from the two mechanisms, depending on whether the value is closer to which end of the two extremes. It appears that the rate of drug release decreases as the contribution via the diffusion mechanism increases (n = 0.66 vs. n = 0.74), which is consistent with the consideration that the larger Zhou et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:1664–1672, 2014

Figure 5. Example dissolution profiles of niacin ER tablets manufactured with different HPMC lots.

distance of the diffusive path is a result from slower polymer erosion. Erosion Rate of HPMC and Dissolution Model Based on the above discussion, a likely cause of the variability in the dissolution performance of niacin ER tablets was the variability in the erosion rate of the gel matrix resulted from different HPMC lots. Therefore, the dissolution rate of pure HPMC tablets was explored using USP I and II dissolution apparatuses, with various settings and conditions, such as basket mesh size, rotation speed, type of sinker, and so on. One challenge was the significant intertablet variability. After extensive trials, a reproducible method was developed using the USP intrinsic dissolution – rotating disc apparatus (Figure 6). The HPMC dissolution samples were assayed using SEC with ELSD detection. Representative dissolution profiles of four different HPMC lots are also presented in Figure 6. Polymer erosion rates ranging from approximately 0.8 to 2.0 mg/h were observed for 11 batches of HPMC that covered the D20h range from approximately 72% to 92% when used in niacin ER tablets. The HPMC batches were supplied from three different vendors. Visual observations during the dissolution test confirmed that the gel layer was larger for the HPMC lots with slower dissolution rates. A niacin release prediction model was established based on the strong correlation (R2 ≈ 0.97) between the HPMC dissolution rate and the niacin ER tablet dissolution (Figure 7). The conformance to the model of several production lots is also illustrated in Figure 7. A specification for the HPMC raw material was established using the dissolution prediction model based on polymer erosion rate. Considerations for establishing the specification limits were given to the analytical variability of the erosion rate test, the analytical variability of the tablet dissolution test, the general variability introduced by the manufacturing process, and other potential impacts such as variability in other components of the formulation. The previous empirical model was based on the average substitution levels that discount the different HP chemistry. It may not be appropriate for all vendors, nor can it account for potential drifting in the chemistry of the HPMC from the same vendor. In contrast, the dissolution prediction model with the DOI 10.1002/jps.23953

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

1671

Figure 6. Typical experimental setup and representative profiles for measuring the HPMC erosion rate.

Figure 7. Dissolution model based on HPMC erosion rate.

polymer erosion rate is based on a property that directly impacts the release controlling functionality of HPMC in matrix tablet and is independent of the HPMC vendor or the vendor lot. Indeed, this erosion test is a more relevant functional test for HPMC when used in manufacturing controlled release hydrophilic matrices. Compared with the substitution levels, the advantage of this erosion test is its independence of HPMC chemistry or vendors. Since equivalent dissolution performance may be achieved with two or more different HPMC chemistries, the erosion test is useful when qualifying alternative HPMC sources or changes to the HPMC manufacturing process.

CONCLUSIONS Variability in the dissolution performance of an ER product was investigated by focusing on a basic understanding of HPMC, the release-controlling excipient. Although the conventional DOEs based on the overall substitution levels, viscosity, and particle size did not fully mitigate the variability, investigations DOI 10.1002/jps.23953

of the structural aspect of HPMC revealed significant chemical and spatial heterogeneity between vendors and potentially within the same vendor. Extensive review of drug release mechanisms indicated that polymer erosion plays an important role in modulating drug release from hydrophilic matrices. A reproducible method was developed to measure the erosion rate of HPMC and the wide range of erosion rates within the same substitution type/viscosity grade of HPMC confirmed its chemical heterogeneity. A robust dissolution prediction model was established based on the correlation between the HPMC erosion rate and the product dissolution. The model facilitated the establishment of a raw material specification to ensure consistent dissolution performance and uninterrupted supply of the drug product. A fundamental understanding of polymer chemistry and drug release mechanisms from the dosage form is essential to meet the challenges of developing and manufacturing ER pharmaceutical products. In perspective, we have found that the compendial specifications are inadequate ensuring HPMC properties for manufacturing niacin ER tablets with consistent dissolution performance. Further probing the HPMC variability and developing meaningful functional tests may be warranted. Although broader applicability of the erosion test needs to be further evaluated, it is proved useful in controlling of drug release of the niacin ER tablet.

ACKNOWLEDGMENT Conflict of interest: This study was funded by AbbVie Inc. AbbVie participated in the study design, research, data collection, analyses, and interpretation of data, as well as writing, reviewing, and approving the publication. All authors are either current or former AbbVie employees and may own AbbVie stock/options.

REFERENCES 1. Dow Chemicals Publication. 2000. Using methocel cellulose ethers for controlled release of drugs in hydrophilic matrix systems, The Dow Chemical Company, Midland, Michigan. 2. Dow Chemicals. 2002. Methocel cellulose ethers technical handbook, The Dow Chemical Company, Midland, Michigan. Zhou et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:1664–1672, 2014

1672

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

3. Harland RS, Gazzaniga A, Sangalli ME, Colombo P, Peppas NA. 1988. Drug/polymer matrix swelling and dissolution. Pharm Res 5:488– 494. 4. Lee PI, Peppas NA. 1987. Prediction of polymer dissolution in swellable controlled-release systems. J Control Release 6:207–215. 5. Ju TTC, Nixon PR, Pael MV. 1995. Drug release from hydrophilic matrices. 1. New scaling laws for predicting polymer and drug release based on the polymer disentanglement concentration and the diffusion layer. J Pharm Sci 84:1455–1463. 6. Gao P, Meury RH. 1996. Swelling of hydroxypropyl methylcellulose matrix tablets. 1. Characterization of swelling using a novel optical imaging method. J Pharm Sci 85:725–731. 7. Brown W, Johnsen RM. 1981. Transport from an aqueous phase through cellulosic gels and membranes. J Appl Polym Sci 26:4135– 4148. 8. Lee PI. 1980. Diffusional release of a solute from a polymeric matrix—Approximate analytical solutions. J Membr Sci 7:255–275. 9. Lee PI, Kim CJ. 1991. Probing the mechanisms of drug release from hydrogels. J Control Release 16:229–236. 10. Ritger PL, Peppas NA. 1987. A simple equation for description of solute release. II. Fickian and anomalous release from swellable devices. J Control Release 5:37–42. 11. Ritger PL, Peppas NA. 1987. A simple equation for description of solute release. I. Fickian and non-Fickian release from non-swellable devices in the form of slabs, spheres, cylinders or disks. J Control Release 5:23–36. 12. Klug ED. 1971. Properties of water-soluble hydroxyalkyl celluloses and their derivatives. J Polym Sci Polym Symp 36:491–508. 13. Sarkar N. 1979. Thermal gelation properties of methyl and hydroxypropyl methylcellulose. J Appl Polym Sci 24:1073–1087. 14. Gao P, Fagerness PE. 1995. Diffusion in HPMC gels. I. Determination of drug and water diffusivity by pulsed-field-gradient spin-echo NMR. Pharm Res 12:955–964. 15. Ju TTC, Nixon PR, Patel MV, Tong DM. 1995. Drug release from hydrophilic matrices. 2. A mathematical model based on the polymer disentanglement concentration and the diffusion layer. J Pharm Sci 84:1464–1477. 16. Dahl TC, Calderwood T, Bormeth A, Trimble K, Piepmeier E. 1990. Influence of physicochemical properties of hydroxypropyl methyl cellulose on naproxen release from sustained release matrix tablets. J Control Release 14:1–10. 17. Lucisano LJ, Breech JA, Angel LA, Franz RM. 1989. Evaluation of an alternate source of hydroxypropyl methylcellulose for use in a substained-release tablet matrix. Pharm Tech 13:88–98. 18. Bonferoni MC, Rossi S, Ferrari F, Bertoni M, Sinistri R, Carmella C. 1995. Characterization of three hydroxypropyl methyl cellulose substi-

Zhou et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:1664–1672, 2014

tution types. Rheological properties and dissolution behavior. Eur J Pharm Biopharm 41:242–246. 19. Reynolds TD, Gehrke SH, Hussain AS, Shenouda LS. 1998. Polymer erosion and drug release characterization of hydroxypropyl hethylcellulose matrices. J Pharm Sci 87:1115–1123. 20. Richardson S, Gorton L. 2003. Characterisation of the substituent distribution in starch and cellulose derivatives. Anal Chim Acta 497:27–65. 21. Viriden A, Wittgren B, Andersson T, Abrahmsen-Alami S, Larsson A. 2009. Influence of substitution pattern on solution behavior of hydroxypropyl methylcellulose. Biomacromolecules 10:522–529. 22. Viriden A, Wittgren B, Larsson A. 2009. Investigation of critical polymer properties for polymer release and swelling of HPMC matrix tablets. Eur J Pharm Sci 36:297–309. 23. Virid´en A, Wittgren B, Andersson T, Larsson A. 2009. The effect of chemical heterogeneity of HPMC on polymer release from matrix tablets. Eur J Pharm Sci 36:392–400. 24. Desai D, Rinaldi F, Kothari S, Paruchuri S, Li D, Lai M, Fung S, Both D. 2006. Effect of hydroxypropyl cellulose (HPC) on dissolution rate of hydrochlorothiazide tablets. Int J Pharm 308:40–45. 25. O’Sullivan AC. 1997. Cellulose: The structure slowly unravels. Cellulose 4:173–207. 26. Mischnick P. 1997. New developments in the analysis of the substitution pattern of polysaccharide derivatives. Macromol Symp 120:281– 290. 27. Mischnick P. 2001. Challenges in structure analysis of polysaccharide derivatives. Cellulose (Dordrecht, Netherlands) 8:245–257. 28. Fitzpatrick F, Schagerloef H, Andersson T, Richardson S, Tjerneld F, Wahlund K-G, Wittgren B. 2006. NMR, cloud-Point measurements and enzymatic depolymerization: Complementary tools to investigate substituent patterns in modified celluloses. Biomacromolecules 7:2909– 2917. 29. Viriden A, Larsson A, Schagerloef H, Wittgren B. 2010. Model drug release from matrix tablets composed of HPMC with different substituent heterogeneity. Int J Pharm 401:60–67. 30. Bettini R, Catellani PL, Santi P, Massimo G, Peppas NA, Colombo P. 2001. Translocation of drug particles in HPMC matrix gel layer: Effect of drug solubility and influence on release rate. J Control Release 70:383–391. 31. Colombo P, Bettini R, Santi P, De Ascentiis A, Peppas NA. 1996. Analysis of the swelling and release mechanisms from drug delivery systems with emphasis on drug solubility and water transport. J Control Release 39:231–237. 32. Pham AT, Lee PI. 1994. Probing the mechanisms of drug release from hydroxypropyl methyl cellulose matrixes. Pharm Res 11:1379– 1384.

DOI 10.1002/jps.23953