Long term physicochemical stability study of novel ophthalmic formulations combining ceftazidime and vancomycin with and without cyclodextrins

Introduction

Bacterial keratitis is a disease caused by the infiltration of bacteria under the cornea, which can lead to an abscess. The incidence of this illness is estimated to be between 0.03 and 79.9 cases per 10,000 in habitants/year [1]. It has become more common in industrialized countries due to the increase in the number of contact lens wearers, for whom it is approximately of 2–20 cases per 10,000 wearers [2]. This ocular infection is responsible for a decrease in visual acuity, particularly through opacification of the cornea, and in the worst cases it can lead to blindness [3]. In the literature published between 2010 and 2020, bacterial keratitis remained the most frequent type of infectious keratitis, representing between 85 and 100 % of all cases [4]. The treatment of bacterial keratitis with a risk of complications requires the patients to be hospitalized in a hospital care unit for treatment using antibiotic eye drops. The antibiotics used in this indication are mostly ceftazidime and vancomycin, generally at high concentrations (25–50 mg/mL) [5]. Ceftazidime (CZ) is a third-generation cephalosporin; it acts on Gram-positive and Gram-negative bacilli by inhibiting the synthesis of peptidoglycan, thus also allowing it to be active against Pseudomonas aeruginosa. Vancomycin (VM) is a glycopeptide which also acts by inhibiting peptidoglycan synthesis. VM cannot cross the membrane of gram-negative bacteria, hence the importance of using a dual therapy combining these two active substances to cover the entire spectrum. To ensure maximum therapeutic activity, a loading dose is required for the administration of these antibiotics. To achieve this, each medication needs to be instilled every 5–10 min during the first hour of treatment, then once every hour for the next 48 h. This dual antibiotic therapy requires an interval of 5 min between each therapy in order to limit the risk of precipitation, especially in case of co-administration of CZ and VM which are incompatible when mixed in solution together [6]. These close instillations have a direct impact on the nurses who are constrained to respect this incompatibility by performing twice as many administrations. In addition, this two-step administration can lead to a decrease in compliance with the treatment or an increase in errors, for example by forgetting an administration.

CZ is a small (molar mass: 546.57 g mol⁻¹) cephalosporin antibiotic (of the β-lactam super family), bearing pyridinium-1-ylmethyl and {[(2Z)-2-(2-amino-1,3-thiazol-4-yl)-2-{[(2-carboxypropan-2-yl)oxy]imino}acetamido groups at positions 3 and 7, respectively, of the cephem skeleton that is very sensitive to hydrolysis. Its strongest acidic and alkaline pKa are respectively of 2.42 and 4.02 [7]. VM is a branched tricyclic glycosylated nonribosomal peptide, and is much larger than CZ (molar mass: 1,485.72 g mol⁻¹). Its strongest acidic and alkaline pKa are respectively of 3.38 and 9.9 [8]. Both drugs are water soluble in their commercial intravenous medication form (VM as a chloride salt, CZ in a pentahydrate form). These medications suffer however from two limiting factors that must be taken into account: their chemical instability in an aqueous media (especially CZ) and their incompatibility when co-administered together. Indeed, once solubilized, the chemical degradation of CZ is very rapid and causes the apparition of breakdown products, including pyridine (degradation product of CZ), which is toxic [9]. Also, in the absence of commercial specialties available, these eye drops must be prepared by compounding pharmacies and used extemporaneously or stored at −20 °C, to be latter thawed and used within 24 h if stored at room temperature, or within 7 days if stored between 4 and 8 °C [10, 11]. Concerning the physicochemical incompatibility between CZ and VM, it results in the appearance of a white precipitate, visible to the naked eye. Park et al. [12] demonstrated an increase in the precipitate when applying the low temperatures, small volumes and high concentrations required to treat bacterial keratitis. Bouattour et al. [13] partially explored the nature of the incompatibility between these drugs and hypothesised that for a pH between 4 and 7.3 the two drugs (with opposite net charges) combined to form an insoluble complex. They also showed that it was possible to mitigate the precipitation by buffering the mixture at an alkaline pH, with the complete elimination of the precipitation being observed for a pH higher than 8.4. However, this equilibrium remains fragile with notable risks of precipitation under the influence of the contact environment (the eye). This environment could modify the equilibrium between the two molecules and lead to intra-ocular precipitation, as demonstrated in the literature during an intravitreal injection [14].

Cage molecules, such as cyclodextrins (CD), are known to be able to encapsulate compounds. CD are cyclic oligosaccharides of the α(1,4) D glucopyranose, forming ring polymers (see Supplementary data 1) [15, 16]. The most used shapes are the α, β et γ consisting of 6, 7 and 8 units of glucose. They possess unique advantages by allowing the formation of inclusion complexes with many organic, inorganic, and lipophilic substances. Their use in the field of nanotechnology, especially for drug delivery, is constantly increasing [14]. In addition to their encapsulation capabilities, CD are also widely used for the formulation of eye drops to increase the aqueous solubility of the molecule, and reduce local irritation [17, 18]. It has also been shown that their basket-shaped cavity can protect some molecules from degradation [19]. Misiuk et al. [20] demonstrated by various analyses (including NMR and molecular docking techniques) that CZ and hydroxypropyl-γ-CD formed a complex with a 1:1 stoichiometry, with the thiazole ring with sulphur and nitrogen atoms with C–O and C–N bands of CZ inserted into the cavity of HP-γ-CD being the most likely complexation pathway. Zarif et al. [21], on the other hand, worked on the equimolar complexation of VM with HP-β-CD in order to evaluate the complexes’ ability to prolong the release of VM from a solid form of the complex. Building on the works of these two previous authors, Bouattour et al. [13] also showed that it was possible to successfully mitigate the precipitation by encapsulating CZ and VM in CD hydroxypropyl (HP) derivatives (HP-γ-CD and HP-β-CD, respectively) using ratios of 5:1 and 3:1 for respectively (HP-β-CD/VM and HP-γ-CD/CZ), but they did not investigate the stability of the mixture of these fragile antibiotics.

The objective of the work presented here was therefore to study and compare the physicochemical stability over 168 days (samples frozen for about 6 months) of these two formulations combining VM and CZ: one using CDs (CD formulation) and the other one without CD (CD-free) at an alkaline pH (pH higher than 8.4).

Materials and methods

Results

Validation of the chromatographic method

Application of method and realization of calibration curves

With the method used, ceftazidime and vancomycin showed respectively an average retention times of 6.2 ± 0.1 min and 18.5 ± 0.5 min (mean retention time ± standard deviation). The chromatograms with the Pharmacopoeia analytical standards of the molecules of interest are presented in Figure 1. It can be noted that the area of the peaks obtained from the commercial Viatris^® vials were significantly higher (by approximatively 15 %) than the areas obtained with the European Pharmacopoeia standard.

Figure 1:

Chromatograms of ceftazidime at 256 nm for (A) the European pharmacopoeia standard, (B) the commercial medication (Viatris laboratory) and vancomycin at 220 nm for (C): the European pharmacopoeia standard and (D): the commercial medication (Viatris laboratory).

The chromatograms of CD (HPβCD and HPγCD) were presented in Figure 2. No peaks were visible at the retention time of the molecules of interest.

Figure 2:

Chromatograms of a solution containing hydroxypropyl-β-cyclodextrin and hydroxypropyl-γ-cyclodextrin, at 220 nm (A) and 256 nm (B).

The calibration curves obtained were linear for both molecules for concentrations between 2.5 and 80 μg/mL. The mean linear regression equations for the antibiotics were presented in the form y = ax + b, y being the area under the curve of the antibiotics and x the antibiotic concentration expressed in µg/mL. The mean linear equation for vancomycin was y = 14,116.79x – 5,049.66. The coefficient of determination was 0.999, and no matrix effect was detected with Vancomycin. The relative mean bias coefficients were less than 3 % for the calibration points, except for the 2.5 μg/mL, for which it was of 30 %. The mean repeatability RSD coefficient and mean intermediate precision RSD coefficient were both of less than 5 %. For CZ, the coefficient of determination was also of 0.999. The mean slope of the calibration curves obtained with the Pharmacopoeia reference was significantly different from the one obtained with the prepared formulation (antibiotics originating from commercialized Viatris^® vials + CD), but the intercept was not significantly different. The mean linear equation used to obtain the concentrations was the one obtained from the pharmacopoeia reference (y = 16,710.19x – 2,262.465). The relative mean bias coefficients were of less than 2 % for the calibration points, except for the calibration point of 2.5 μg/mL, for which it was of 3.75 %. The mean repeatability RSD coefficient and mean intermediate precision RSD coefficient were both less than 5 %.

Forced degradations

The retention times, areas under the curve and degradation ratios (Breakdown products divided by the molecule of interest, expressed as a percentage) of all the compounds obtained by forced degradation are visible in Supplementary data 3. Chromatograms representing the degradation products formed upon application of the different types of degradation for are visible in Figure 3. None of the breakdown products possessed a retention time that could interfere with VM or CZ quantification.

Figure 3:

Chromatographs of ceftazidime (A) and vancomycin (B) at different conditions of forced degradation (1): alkaline degradation, (2): acid degradation, (3): UV degradation, (4): oxidative degradation, (5): heat degradation.

Figure 4 represents the degradation products obtained by forced degradation of the molecules of interest (ceftazidime and vancomycin) as a function of relative retention times. The diameter of the disks is expressed as a percentage of the ratio of the area under the curve of the degradation product to the reference area under the curve (before degradation) of the molecule of interest.

Figure 4:

Degradation products of antibiotics applied with conditions of degradation. (A) Degradation products of ceftazidime (CZ) and (B) vancomycin (VM) applied with conditions of degradation. (1): UVA (during 6 h), (2): oxidation degradation (H₂O₂ 3 % 1 h for CZ; H₂O₂ 10 %, 1 h for VM), (3): heat degradation (80 °C, 1 h for CZ, 40 °C, 24 h for VM), (4): alkaline degradation (NaOH 0.002 N 1 h for CZ, NaOH 0.5 N 1 h for VM) and (5) acid degradation (HCl 0.3 N 1 h for CZ; HCl 1 N 1 h for VM).

For CZ, numerous breakdown products where noticed, for all forced degradation conditions except UVA radiations. Contact with a 0.002 N NaOH solution resulted in a degradation of more than 50 % of ceftazidime, indicating a very high sensitivity to alkaline degradation. Heat and acidic degradation also yielded several breakdown products, but to a lesser degree.

For VM, each type of degradation revealed breakdown products. The degradation conditions used allowed to have a degradation of vancomycin ranging from 1 % (UV degradation) to about 25 % (alkaline degradation).

Research and identification of impurities for ceftazidime and vancomycin

Chromatograms of Ceftazidime of peak identification and Vancomycin suitability were studied. For Ceftazidime, impurity G was potentially identified (retention time: 11.26 min and relative retention time compared with ceftazidime: 1.82). The retention times of impurities found with the Vancomycin suitability solution do not interact with the retention time of vancomycin and ceftazidime.

Identification of a degradation compound: pyridine

The RT of pyridine was estimated at 11.6 ± 0.4 min, with a relative retention time (RRT, relative to the ceftazidime retention time) of 1.9. The chromatograms and 3D spectra of pyridine and the alkaline forced degradation of ceftazidime are shown in Figure 5. The peak with a RT of 11.2 min corresponds to pyridine.

Figure 5:

Chromatograms of pyridine and alkaline degradation – spectra of interest molecules. (A) Chromatograms of pyridine (Europeans pharmacopeia reference); (B) ceftazidime alkaline degradation; (C) spectra of degradation product of ceftazidime; (D) spectra of pyridine (reference 270970, Sigma Aldrich).

The limit of detectability of pyridine was estimated at 0.05 μg/mL and the limit of quantification (LOQ) at 0.5 μg/mL.

Visual inspection, chromaticity measurements, particles counting and turbidity

Visual inspection

For the CD-free formulation, a haziness and a precipitate were systematically observed 24 h after thawing. This phenomenon was more intense for the samples analysed on the third day after thawing, resulting in a white opalescent solution. For the formulation with CD, the samples analysed immediately after thawing (M1H0, M3H0, and M6H0) and 12 h post-thawing (M1H12, M3H12 and M6H12) remained limpid, with no visible appearance of particulate matter or haziness. Haziness was present on the third day after thawing, but it remained less intense than the one noticed for the CD-free solution. At the sixth month after thawing (M6H0) time-point, a pinkish colouration was observed for the mixture without cyclodextrins, which was not observed in the formulation containing CD. The visual aspect of the solutions at different time-points is shown in Figure 6.

Figure 6:

Visual aspect of samples 24 and 72 h after thawing (refrigerated storage). CD, cyclodextrins. M0, M1, M3, M6: start of study, then 1, 3 and 6 months of frozen storage. H24 and H72: respectively 24 and 72 h after thawing.

Turbidity

The results of turbidity are presented in Supplementary data 4.

For the CD-free formulation, turbidity results didn’t vary by more than 0.5 FNU only up until M1H24. All other further time points were out of specifications. For the formulation with CD, turbidity measures were within specifications up until M3H24 and M6H12.

Subvisible particles counting

The result of particles counting (mean and standard deviation) are visible in Figure 7. The number of particles larger than 10 µm remained very high from the first day of sampling for the CD-free formulation, and not measurable at M6H24. For the formulation with CD, the number of particles remained quite low on the first day of thawing but became unquantifiable on the third day. Particle numbers were not measurable for all samples after 72 h of refrigeration after thawing (M1H72, M3H72 and M6H72) (beyond instrumental measuring possibilities).

Figure 7:

Particles counting of two formulations of eye drops. CD, cyclodextrins; HQLT, high quantity limit tolerance. M0, M1, M3, M6: start of study, then 1, 3 and 6 months of frozen storage. H24 and H72: respectively 24 and 72 h after thawing.

These results were remained coherent with the visual inspections, notably the appearance of a precipitate.

Chromaticity and lightness

The chromaticity and luminance results are presented in Table 3. For the formulation with CD, the lightness L* was stable until the third day after thawing, after which a clear decrease was observed (respectively of about 30 and 94 % for M1H72 and M3H72). The increase in ∆E was notable (>1) 72 h after defrosting (M1H72 and M3H72). After 6 months of storage, the change was notable 24 h after thawing (M6H24).

Table 3:

Evolution of lightness and chromaticity for two formulations (pooled samples, n=1).

	With CD				CD free
	L*	a*	b*	ΔE	L*	a*	b*	ΔE
M0	98.36	−0.99	5.97		99.05	−1.22	4.44
MlH0	98.63	−0.99	5.95	0.15	98.58	−0.26	4.63	2.36
M1H12	98.31	−0.98	5.76	0.09	98.48	−0.28	4.82	2.71
M1H24	97.69	−1.02	6.02	0.90	58.83	4.81	26.05	4,242.00
M1H72	67.53	2.62	2 1.75	2,425.06	0.2	0.01	0.12	19,583.00
M3H0	98.46	−0.96	5.92	0.03	97.13	0.52	5.74	16.81
M3H12	98.5	−0.89	5.69	0.22	97.79	0.88	5.06	12.76
M3H24	98.76	−1.01	5.94	0.32	98.61	−0.86	6.22	6.98
M3H72	6.18	0.86	3.96	17,009.23	0.17	0.03	0.12	19,594.96
M6H0	98.5	−0.8	5.81	0.17	96.67	2.38	5.45	39.29
M6H12	98.41	−0.92	5.68	0.18	66.13	6.06	27.77	3,362.03
M6H24	99.58	−0.97	5.61	3.24	0.85	0.46	0.8	19,318.62
M6H72	2.36	0.4	1.72	18,471.99	0.09	0.03	0.09	19,627.13

1. CD, cyclodextrins. M0, M1, M3, M6: start of study, then 1, 3 and 6 months of frozen storage. H0, H12, H24 and H72: storage hours after thawing.

For the CD-free formulation, the decrease in L* was observed from the first day after thawing after the first month of storage and was of the order of 40 %. The increase in a* at M6H0 (2.38 vs. 0.52 at M3H0 and 0.26 at M1H0) was correlated with the evolution of the colour observed during the visual inspection. The increase in ∆E was notable at all time-points, but even more so at M6 after thawing and refrigerated storage.

pH

The initial pH was of 8.46 ± 0.01 for the CD-free formulation and 8.13 ± 0.01 for formulation with CD. A small decrease in pH was observed over time for both formulations, the minimum observed pH was 7.62 ± 0.01 for the formulation with CD and 8.08 ± 0.03 for the CD-free formulation (at M1H72). The complete results for pH are presented in the Supplementary data 4. The variations in pH measurements at the different sampling times of the two formulations are of less than 1 pH unit, making them compliant with the expected specifications.

Osmolality

The initial mean osmolality was of 282.25 ± 7.04 mOsm/Kg for formulation with CD, and 75.5 ± 0.92 mOsm/Kg for the CD-free solution. All the results are visible in Supplementary data 4. The variation of the osmolality of the samples remained below 10 % for both formulations. These results are therefore consistent with the specifications.

Vancomycin, ceftazidime and breakdown products analysis

At the beginning of this study (M0H0), the concentrations (mean ± 95 % confidence interval) of ceftazidime and vancomycin were respectively of 27.50 ± 1.08 mg/mL and 23.25 ± 2.17 mg/mL for the CD-Free formulation, and of 26.47 ± 1.58 mg/mL and 23.23 ± 0.64 mg/mL for the formulation with CD. Whereas VM concentrations remained relatively stable, CZ concentrations decreased rapidly after thawing for both formulations (Figure 8). For example, for the CD formulation, they were only above the inferior specification limit up until time-points M1H72, M3H72, and from M6H24.

Figure 8:

Concentration of ceftazidime and vancomycin at the different sampling points (expressed in % of the initial quantity, mean ± 95 % confidence interval). CD, cyclodextrins. M0, M1, M3, M6: start of study, then 1, 3 and 6 months of frozen storage.

The evolution in pyridine peak area is presented in Figure 9. The areas under the curve were lower than the LOQ of pyridine, not allowing the expression of the results in pyridine concentration. The formation of pyridine remained low but was more important for the formulation with CD than for the CD-free formulation.

Figure 9:

Pyridine area formed (AUC, area under curve). Comparatively: AUC of ceftazidime at M0 CD-free: 369909 and with CD: 356068. CD, cyclodextrins. M1, M3, M6: respectively 1, 3 and 6 months of frozen storage. x-axis: number of days after thawing.

Discussion

In this study, we proved that the two formulations that we tested of eye drops made of CZ and VM, formulated so that the antibiotics were compatible in a mixture together, did not possess extended stability. Overall, our results showed a very limited physicochemical stability for both formulations when stored at refrigerated temperature after thawing, with the appearance of a physical instability (precipitate) from the first day for the CD-free formulation and at the third day for the formulation with CD, as well as a decrease in ceftazidime concentrations for both formulations.

The first formulation did not contain CD but was formulated at a target pH of 8.5: this pH was determined in order to avoid the phenomenon of precipitation when ceftazidime and vancomycin are mixed [6], and whilst respecting as much as possible the physiological pH of the eye [38]. In the second formulation, we tested the protective power of the cage molecules over a long period of time, following the work published by Bouattour et al. [13] that demonstrated an absence of precipitation of this formulation during 48 h for the optimized molar rations that were used here (5:1 and 3:1 for respectively HP-β-CD/VM and HP-γ-CD/CZ). Regarding the feasibility of the preparation, the major difficulty lay in the dissolution of the CD in the buffer solution (3.33 g/L for HPβCD and 5.8 g/L for HPγCD), where the formation of large clusters delayed the dissolution, and this despite the concentrations applied here being far from the theoretical maximum solubility concentrations for these CD [39]. As the pH of this formulation was set at pH 8.0, thus close to the limit of efficiency of the CD on the incompatibility, a 30 mM phosphate buffer was added, to stabilize the pH [13]. To take into account the risk of corneal calcification described by the European medical agency (EMA) [40], its molarity was set at 30 mM, thus far from the concentrations of 50–100 mM described as being at risk [41, 42]. In regard of the quick degradation of ceftazidime [43], freezing was the method of choice to delay as much as possible any physicochemical reaction during storage.

To adequately perform the stability study, a SIM was developed to analyse both antibiotics simultaneously. The method allowed us to quantify the antibiotics and research breakdown products. The analysis of the validation data highlighted a decreased precision when the excipients were used, very likely linked to the increased experimental difficulty they induced. The comparison of the calibration curves of reference ceftazidime (Pharmacopoeia standard) with the curves of matrix ceftazidime solutions (commercial specialty with sodium carbonate, HPβCD and HPγCD) also highlighted a significantly different value between the slopes. All the calibration points performed with the solution with matrix had values higher by about 10–15 % on average compared to the calibration points performed with the pharmacopoeia reference. No difference in calibration slopes were found for VM. Moreover, the chromatographic analysis of the cyclodextrins alone did not show any difference compared to a classical blank sample. This isolated increase in CZ peak area was also noted when validation points were performed at 20 μg/mL without cyclodextrins (using the commercial Viatris vials^®, with several different batches). Two hypotheses can therefore be considered: a potential matrix effect of the excipient of the commercial ceftazidime product (sodium carbonate) or an overfilling of the commercial ceftazidime vials corresponding to the tolerance limits of industrial production, such as the one found for Zavicefta^® (Ceftazidime/avibactam), as the EMA report for that product states that “An overfill is used in order to ensure that the entire contents of the reconstituted vial can be accurately removed” [44]. For all the experiments, the reconstitution of ceftazidime was carried out after removal of the crimp from the vial (metal tray and rubber stopper), which reduced the loss of active ingredient compared to conventional sampling with a needle and syringe through the rubber stopper carried out by the nurses [45]. In addition, a terminal rinse was performed on all vials to minimize the loss of active ingredients. It is therefore highly likely that the excess in ceftazidime found during our study resulted from the overfill of the ceftazidime Viatris^® vials.

During the stability study, the visual examinations revealed the presence of a cloudiness as early as 24 h after thawing for the CD-free mixture; the turbidity found on these samples was within specifications only for points M1H0 and M1H12. For the solutions containing CDs, the turbidity was visible to the naked eye only for the H72 post-thaw points, where the turbidity was also found to be outside the specifications for these points as well as for all the samples taken at month 6 (M6). These results are quite consistent with particle count data where the number of particles increased with thawing time regardless of the month of thawing. The particulate counting limits were reached for all the H72 points whatever the formulation as well as the M6H24 for the CD-free formulation.

None of the colour difference (∆E) measured for CD-free formulation complied with the specifications. For the formulation with CD, ∆E was above specifications for all H72 points as well as the M6H24. These results are therefore consistent with the visual inspection and the turbidity results found. pH and osmolality were within specifications and within tolerance with the physiological pH of the eye, so these two parameters were not considered as limiting factors in the stability study.

The quantification of the antibiotics did not show a significant loss of VM at any of the sampling points: the values obtained always remain between 90 and 110 % of the target value for both formulations despite the appearance of precipitate. A more pronounced loss of CZ in both formulations was noticed, especially after thawing for all the H72 samples and for the M6H24 time-point, thus suggesting a higher degree of instability for CZ than for VM, which is consistent with previously published data. Indeed, an experimental study by Moreno et al. showed a loss of 85 % of ceftazidime after an exposure to a temperature of 45 °C for 24 h [46] and the work of Nguyen et al. reported a 10 % loss of ceftazidime after 19 h at room temperature [47]. Our work supports the notion that frozen storage inhibits most if not all of the CZ loss, but that refrigerated storage only slows it down. In our study, the decrease in CZ concentrations is more pronounced in the formulation with CD than the CD-free formulation. This could be caused either by CZ precipitation (possibly more the case in the CD-free formulation) or by chemical degradation. NMR modelling showed good inclusion of CZ in CD [13], however the encapsulation does not seem to confer adequate protection against CZ degradation, perhaps because the fragile part of the molecule may be outside the hydrophobic core of the CD. Another hypothesis could be an interaction of the CD with CZ, thus being able to create a destabilization of the molecule leading to its degradation [48]. This destabilization could be caused by an intrinsic catalytic activity of the CDs mediated by their free hydroxyl groups, as previously illustrated for example by Wouessidjewe et al. [49] or Popielec et al. [50], who in their excellent review of the effects of cyclodextrins on the chemical stability of drugs summarised that CDs could catalyse β-lactam degradation under basic conditions, possibly by ionization/deionization of carboxyl group of β-lactam antibiotic. This degradation may also lead to the appearance of degradation products outside the CD, possibly creating an interaction with VM and the formation of a precipitate. Moreover, the risk of intra-ocular precipitation must be considered: the moderate buffering capacity of tears could decrease the pH initially selected, which could lead to an even greater precipitation [38]. A lacrimation phenomenon could also cause a dilution of the eye drops, and the balance of H⁺ and OH⁻ ions could re-equilibrate the pH and create the precipitation phenomenon [51].

The peak area of pyridine was also investigated, as it is a known CZ breakdown product; this was notably found in a study from Stendal et al. [52] which showed the formation of 0.4 % pyridine in mass after reconstitution and dilution of ceftazidime to 60 mg/mL stored for 7 days at 4 °C. However, in view of the low pyridine levels (pyridine peak areas were not high enough to allow an accurate quantification), we cannot assert a significant difference between the two formulations. The ocular toxicities mentioned report the absence of toxicity for concentrations of 0.08 M pyridine, a value which we are well below in this study [53]. The systemic passage by the ocular route would also remain quite low [54]. The toxicity found in the literature is only applicable to the IV route and with much higher doses.

Overall, in view of these initial results, neither of the formulations that were tested were good candidates for further more advanced studies, such as a pre-clinical study/safety usage investigation, as they have proven to be too unstable physiochemically.

Conclusions

In this study, which lasted 168 days, we showed that both formulations (with and without CD) showed very limited physicochemical stability after thawing for the storage conditions that we tested (3 months frozen and 24 h after defrosting for the solution with cyclodextrins). The cyclodextrins-free for formulation proved too instable after freezing and defrosting. In view of the complexity of the formulation, and the instabilities of both formulations, the theoretical advantages of combining VM and CZ seem somewhat nullified.