How to stabilize cilazapril-containing solid dosage forms? The optimization of a final drug formulation

2.1 Chemicals

Cilazapril monohydrate was kindly supplied by Biofarm Ltd., Poland (batch number 1621816). HPR and LAC were purchased from Biofarm Ltd., Poland; MS was derived from Sigma–Aldrich, Germany; TLC was obtained from LG Olsztyn, Poland. Pharmaceutical formulations of CIL 5 mg were purchased from Cefarm, Poland: the original drug further referred as ‘A’ (serial number: E0102B01) and the generic drug further referred as ‘B’ (serial number: 02661). Methanol and acetonitrile (HPLC grade) for the preparation of the mobile phase were obtained from Merck, Germany. Sodium chloride (ACS reagent grade), sodium nitrate (ACS reagent grade), and sodium bromide (ACS reagent grade) were obtained from Sigma–Aldrich, Germany. Potassium phosphate monobasic and benzocaine were purchased from POCh, Poland. The distilled water was used.

2.2 HPLC method

The determination of CIL in the samples was conducted by the previously-developed and validated, stability indicating HPLC method (Paszun et al., 2012). The Shimadzu HPLC equipment consisted of: Shimadzu LC-6A Liquid Chromatograph pump with a 7725 Rheodyne 20 μL fixed loop injector, a Shimadzu SPD-6AV UV–VIS Spectrophotometric Detector and Shimadzu C-R6A Chromatopac integrator. The separation was achieved on the reversed phase LiChroCART® 250-4 HPLC-Cradridge, LiChrospher® 100 RP-18 (5 μm) (Merck, Germany) column with the mobile phase consisting of acetonitrile–methanol–phosphate buffer (pH 2.0) (60:10:30; v/v/v). The mobile phase flow rate was adjusted to 1.0 mL/min and the injection volume was 20 μL. The UV detection wavelength was set at 212 nm. The time of a single separation was approximately 12 min.

2.3 Liquid chromatography/electrospray Ionization–mass Spectrometry System

The analysis of a degradation product of CIL in tablets and binary mixtures was carried out using Waters liquid chromatographic system, interfaced to a 996 detector and a Micromass ZQ 2000 electrospray mass spectrometer. LC separations were performed on LiChrospher 100 RP-18 column (size 10 μm, 250 mm × 4 mm) at 30 °C. The mobile phase consisted of methanol:water:formaldehyde (49:50:0.5 v/v/v) and its flow rate was 0.5 ml min⁻¹. The mobile phase was filtered through a filter (0.22 μm) and degassed by ultrasound prior to analysis. The injection volume was 100 μl. The recorded mass spectrum ranged from 100 to 1000 m/z and both, ES+, ES− ionization modes were applied.

2.4 Sample preparation

2.5 Isothermal stress testing conditions

Isothermal test procedure was applied in order to reduce the time period necessary for chemical reactions to occur. The appropriate thermal conditions were provided in heat chambers with the temperature control accuracy of ±1.0 K. The demanded RH levels were obtained in closed desiccators containing saturated solutions of inorganic salts (RH = 50.9% – sodium bromide, RH = 66.5% – sodium nitrate and RH = 76.4% – sodium chloride). The kinetic test conditions were equilibrated 24 h before samples’ incubation.

2.6 HPLC analysis

After incubation, the content of each sample (i.e. tablet or CIL-excipient model mixture) was quantitatively transferred into 50 mL volumetric flask and mixed with 25.0 mL of methanol. The obtained suspensions were shaken for 15 min and then filtered through a quantitative filter (390, Munktell). 1.0 mL of each clear supernatant was mixed with 0.5 mL of internal standard (a methanolic solution of benzocaine, 0.2 mg/mL) and injected onto the HPLC column.

4.1 Chromatograms’ quality

The analysis of the obtained HPLC chromatograms for both, tablets and model mixtures indicated that in the course of CIL degradation only one degradation product is formed and it corresponds to the signal with the retention time ∼3.9 min (Fig. 2). Based on our previous studies with CIL in pure, we hypothesized that under the experimental conditions CIL decomposes producing its diacidic analog – cilazaprilat (Stanisz and Paszun, 2013) This, in turn, suggests that the process of cilazapril decomposition compromises drug’s bioavailability since cilazaprilat cannot be absorbed from the gastrointestinal tract after oral administration (Williams et al., 1989).

4.2 Estimation of the reaction kinetic order

The applied stress conditions of kinetic tests, i.e. increased temperature and RH provided the opportunity to accelerate the process of results’ collection and they were in agreement with the approved ICH regulations (ICH Q1A(R2). We exposed the studied samples to the variety of environmental conditions and used different kinetic test procedures in order to achieve the best knowledge of CIL vulnerability to degradation stimulated by various external factors. For practical reasons, the experiment with tablets was intended to calculate the thermodynamic data of CIL degradation which are essential for the estimation of its stability during long-term storage, defined by the shelf-life. On the other hand, the experiment with binary mixtures was designed to enable the selection of the most compatible excipients for CIL formulation, emphasizing the fact that the manufacture process can be performed in different climatic zones.

The kinetic model of CIL degradation in both, binary mixtures and tablets, was established by analyzing the percentage of the remaining drug concentration in the studied samples after their exposure to the predefined stress conditions. As evidenced by the obtained kinetic curves, in each experiment the gradual decrease of CIL concentration with time occurred. Furthermore, each plot c = f(t) took the form of sigmoid with three distinct reaction phases, i.e.: an initial slow phase, an acceleration period and a final termination phase, which is typical of an autocatalytic reaction model (Figs. 3–9). The mathematical procedure of model-fitting confirmed this observation, indicating that under the applied experimental conditions CIL degrades according to Prout–Tompkins kinetics (r > 0.991). Similar mechanism of degradation has been already reported for CIL in pure (Paszun et al., 2012; Stanisz and Paszun, 2013). Hence, the following equation was employed in order to determine the appropriate kinetic parameters: $$\mathit{ln}{c}_t/(c_0-c_t)=C-k·t$$ where c₀ and c_t represent concentration of CIL at time points 0 and t, C is induction period and k stands for degradation rate constant (s⁻¹). As indicated in Figs. 5 and 9, the rate of CIL degradation accelerates along with the increasing stress, demonstrated by increasing k and shortening induction periods (established graphically). This, in turn, suggests that the environmental moisture, high temperatures and the presence of the degradation product negatively influence the stability of CIL.

4.3 Stability of CIL in pharmaceutical formulations

The CIL’s vulnerability to degradation in pharmaceutical formulations was assessed using tablets of the original drug referred as ‘A’ (shelf-life 36 months) and the generic drug referred as ‘B’ (shelf-life 18 months). The samples of whole unpacked tablets and whole blistered tablets (blisters OPA/Al/PVC//Al) were prepared with a view to investigate the impact of excipients and immediate packaging on CIL stability, as well as to detect any possible interactions between formulation and the material of the blister. The following conditions of the isothermal stress testing were applied: T = 318, 323, 333, 338, 343 K and RH = 76.4%. Pure CIL was tested in parallel as a reference.

The obtained results clearly indicate that the studied ACE-I in tablets is less stable than in pure (Figs. 3 and 4), evidenced by higher k for both formulations when compared to the reference. The compromised stability of CIL in dosage forms probably results from the presence of excipients. Alternatively, the observed effect could be explained by the negative impact of tablets’ manufacturing process, which is actually known to involve wet granulation (Niazi, 2004).

The relationships shown in Fig. 5 suggest that CIL in tablets is sensitive to temperature changes. Furthermore, the logarithm of k in each experiment and the corresponding temperature reciprocal fulfill the Arrhenius equation: $$\ln k=\ln A-E_a/(R·T)$$ where E_a is activation energy, A – frequency coefficient, R – gas constant and T – temperature (Pawełczyk and Hermann, 1982). Based on this the thermodynamic parameters of CIL degradation in tablets A and B (Table 1) were established using the following relationships: $$\mathit{Ea}=-a·R$$ $$E_a=\mathrm{\Delta }{H}^{\ne }+\mathit{RT}$$ $$\mathrm{\Delta }{S}^{\ne }=R·\ln A-\ln \mathit{KT}/h$$ where a is the slope of ln k = f(1/T) straight-line, A is a frequency coefficient, E_a is the activation energy (J mol⁻¹), R is the universal gas constant (8.3144 J K⁻¹ mol⁻¹), T is the temperature (K), ΔS^≠ is the entropy of activation (J K⁻¹ mol⁻¹), ΔH^≠ is the enthalpy of activation (J mol⁻¹), K is the Boltzmann constant (1.3806488(13) × 10⁻²³ J K⁻¹), and h is the Planck’s constant (6.62606957(29) 10^–34 J s). Here, E_a describes the strength of the cleaved bond during CIL degradation and it defines the minimum of energy required for the reaction to occur. Thus, according to our results, CIL in the formulation A seems to be slightly more sensitive to temperature changes than in formulation B since the activation of its decay requires less energetic input (187.4 ± 173.5 kJ/mol in drug A vs. 199.6 ± 348.6 kJ/mol in drug B). However, the formulation A in blister was found to protect CIL from degradation more efficiently than the formulation B in blister, as evidenced by respective E_a (343.5 ± 387.9 kJ/mol for blistered drug A vs. 182.0 ± 38.1 kJ/mol for blistered drug B) and the corresponding reaction rate constants (Fig. 5). Our data also indicate that the immediate packaging in formulation A stabilizes CIL considerably, evidenced by higher E_a and lower k for blistered drug A when compared to unpacked drug A, which further impinges on its longer shelf-life (Fig. 5, Table 1). Surprising results were, however, achieved in the experiment with blistered formulation B in which CIL unexpectedly turned out to be less stable than in unpacked formulation B as indicated by higher reaction rate constants and lower E_a. The differences between these parameters for blistered and non-blistered drug B proved, however, to be statistically insignificant (F-Snedecor test at α = 0.05 and df = 12; p < 0.05). Nonetheless, our observations suggest that the immediate packaging in formulation B serves no additional protective function as far as CIL stability is concerned and hence an additional investigation of its quality is advisable.

According to the summaries of product characteristics available for the formulation A and B, both products share the same type of immediate packaging (i.e. blister OPA/Al/PVC//Al) and they have analogous qualitative composition of tablet core. They differ however, in the qualitative composition of tablet coating and probably in the quantities of the excipients used (Table 3). Thus, we hypothesized that the heterogeneous stability profiles of the tested samples might result from chemical interactions between CIL and other tablet ingredients, and therefore we investigated the impact of the selected excipients present in both formulations on CIL stability.

4.4 Stability of CIL in model mixtures with the selected excipients

The analysis of CIL stability in model mixtures with the excipients was conducted in order to detect any possible interactions affecting the process of its degradation in solid state. HPR, LAC, MS and TLC were mixed with pure CIL at 1:1 weight ratio and then subjected to the conditions of increased RH: 50.9%, 66.5%, and 76.4%. at 333 K. The applied thermal conditions allowed for the maintenance of physiochemical properties of the studied compounds over the test duration. Furthermore, pure CIL was analyzed in parallel as a reference. Our results (Table 2) clearly indicate that all the investigated excipients, except for MS, accelerate CIL degradation. As evidenced, under the conditions of T = 333 K and RH = 76.4% the content of pure CIL in the studied samples after 174 days of incubation equaled 92.4%, while in the binary mixtures it did not exceed 71.6% (in CIL/HPR it was 58.8%, in CIL/LAC – 55.2%, and in CIL/TLC – 71.6%). Only in case of CIL/MS the remaining drug concentration after 174 days of the kinetic study was above 94%, and after 350 days it equaled 89.2%. This indicates that in CIL/MS experiment only the induction period of the autocatalytic reaction was observed and for this reason no kinetic equation could be calculated. It is also important to emphasize that the content of pure CIL after 350 days of heating under the above RH conditions equaled 65%.

According to the theory of the autocatalytic reaction kinetics, the degradation rate constants for CIL decay were calculated only if the acceleration period occurred. As evidenced in Table 2, for CIL/MS even after 350 days of heating under RH = 76.4% the reaction did not enter into its rapid phase suggesting good drug stability. Hence, k values for this process could not be provided. Our results also confirm that the increasing RH stimulates CIL degradation, evidenced by the gradually shortening induction periods and increasing autocatalytic reaction rate constants (Fig. 9), supporting the need for the special protection of CIL from moisture. For this reason the process of wet granulation should be avoided in case of CIL and the procedure of direct compressing should be followed instead, similar to that of enalapril (Bibi et al., 2011). It must be, however, strongly emphasized that this alternative method of tableting may be associated with the occurrence of additional incompatibilities (Bharate et al., 2010), and hence its applicability for CIL-containing tablets should be thoroughly pre-examined. Furthermore, in order to improve the stability of CIL in its solid dosage forms, the excipients exhibiting the lowest moisture sorption capacity should be preferred for its final formulation. Also desiccants such as calcium chloride, silica gel or molecular sieve could be considered as an additional, primary package component.

As for some specific CIL – excipient interactions observed in the course of our study, out of all the tested compounds it was HPR that exerted the most disadvantageous effect on the stability of the investigated ACE-I. In fact, for this binary mixture the calculated degradation rate constants reached the highest values, equaling k_CIL/HPR = (3.26 ± 0.358) · 10⁻⁷ s⁻¹ for T = 333 K, RH = 76.4% and k_CIL/HPR = (1.14 ± 0.15) · 10⁻⁷ s⁻¹ for T = 333 K, RH = 66.5%, and they were significantly higher than those, obtained for pure CIL (analyzed by statistical F-Snedecor test at α = 0.05 and df = 13) (Fig. 6). The negative impact of HPR on CIL stability could be probably attributed to its highly hygroscopic properties after drying (Rowe et al., 2006), as well as to its nucleophilic nature that may be responsible for the promotion of CIL hydrolysis (Narang et al., 2012). Interestingly, analogous observations have been already reported in the pre-formulation studies with other ACE-I – quinapril hydrochloride (Stanisz, 2005) and also with trichlormethiazide (Teraoka et al., 2009). Hence, in CIL-containing solid dosage forms the substitution of HPR by another excipient, e.g. copovidone, which possesses moisture-scavenging properties, seems reasonable (Fox, 2007).

A similar effect of compromised drug stability (yet to a lesser extent) has been observed in CIL/LAC binary mixture, confirmed by F-Snedecor test at α = 0.05 and df = 16 (Fig. 7). Here, the obtained degradation rate constants were the following: k_CIL/LAC = (2.18· ± 0.16) · 10⁻⁷ s⁻¹ for T = 333 K, RH = 76.4% and k_CIL/LAC = (2.42 ± 0.659) · 10⁻⁸ s⁻¹ for T = 333 K, RH = 66.5%. LAC being a reducing sugar, could interact with the nucleophilic secondary amine function of CIL via Millard reaction with the production of colored products (Gamble et al., 2010). In fact, this type of incompatibility has been previously reported for lisinopril (Eyjolfsson, 1998) yet here a primary amine of ACE-I participated in the chemical process. Our results clearly evidence that LAC does not stimulate Millard reaction with CIL since no color change of the stressed samples occurred and no additional signals on the chromatograms were obtained. Irrespective of this lactose contains approximately 5% w/w water of crystallization (Rowe et al., 2006) and under certain conditions it also could be hygroscopic (Listiohadi et al., 2008), which might compromise CIL stability.

In analogy to the previously discussed excipients, also TLC deteriorated the stability of CIL (k_CIL/TLC = (1.86·10 ± 0.271) · 10⁻⁷ s⁻¹ for T = 333 K, RH = 76.4%) in the statistically significant manner when compared to the pure substance (confirmed by F-Snedecor test at α = 0.05 and df = 12) (Fig. 8). However, its negative impact was less pronounced that that of LAC and HPR.

The only excipient that exerted a positive effect on CIL stability was MS since under the experimental conditions only 10.8% loss of active substance occurred for as much as 360 days of stressing (compared to 35% loss of pure CIL under similar environmental conditions) and this effect could be attributed to its moisture-scavenging properties.

4.5 Mechanism of CIL degradation in tablets and binary mixtures

In order to establish the mechanism of CIL decay the determination of CIL degradation products in tablets A and B, both with and without immediate packaging and in binary mixtures with HPR, TLC, LAC and MS was performed by LC-MS method using soft ionization technique ESI. Thanks to this, we obtained simple mass spectra which enabled an unambiguous verification of the pseudo-molecular ions type [M+H]+ or [M−H]− as the most abundant signals. The mass spectra (Fig. 10a) of non-stressed samples exhibit one peak at m/z = 436 for ES+ and m/z = 434 for ES−, which matches with the molecular mass of CIL (435 u), confirming its identity. However, the mass spectra of the stressed samples (Fig. 10b) under the conditions of T = 333 K, RH = 76.4% for the time period necessary to induce a complete CIL degradation display one signal at m/z = 408 (ES+) and m/z = 406 (ES−) which corresponds to the molecular mass of diacidic derivative of CIL – cilazaprilat (407 u) formed by the ester bond hydrolysis. Based on these findings it can be summarized that CIL in the form of tablets as well as in the investigated binary mixtures exposed to the stress conditions of increased T and RH is susceptible to ester bond hydrolysis with the formation of one impurity – cilazaprilat (Fig. 11).