Abstract
The delamination of pharmaceutical glass is a serious issue, as it can cause glass particles to appear in vials, a problem that has forced a number of drug product recalls in recent years. In Type I pharmaceutical glass vials, delamination occurs generally at the bottom and shoulder, where extensive flaming during the conversion process can favor a strong evaporation of alkali and borate species and the formation of heavily enriched silica layers. The contact with parenteral preparations dissolved in an alkaline medium increases the rate of glass corrosion, while the differential hydration of these layers can cause the detachment of flakes. The purpose of this study was to investigate the effect of the pH and the composition of the extraction solutions on the propensity of different glass types to delaminate. Repeated autoclave extractions at 121 °C were carried out on different glass types with different extraction media, including organic extractants like citric and glutaric acid. When vials were in contact with alkaline solutions and similarly aggressive media, an increase in silica extraction values indicated glass corrosion and an increasing risk for further delamination. Under such conditions expansion 33 glass is extensively corroded, showing high silica concentration and heavy flaking as compared to other glass types. Sulfur-treated glass also showed early flaking, even if SiO2 concentration was very low. A similar ranking was observed with extractions carried out with glutaric and citric acids, but at far much higher SiO2 concentration levels. Extractions with 0.9% KCl solution can be used as an accelerated test to highlight the propensity of a glass to delaminate, but in no case it can be taken as a guarantee that the glass will not delaminate when exposed to the pharmaceutical drug, whose extraction ability requires case-by-case study.
LAY ABSTRACT: How can injectable drug manufacturers prevent glass delamination? The issue of delamination is a serious one, as it can cause glass particles to appear in vials, a problem that has forced a number of drug product recalls in recent years. To combat this, pharmaceutical and biopharmaceutical manufacturers need to understand the reasons for glass delamination. The most recent cases of product recall due to the presence of particles in the filling liquid have involved borosilicate glass containers carrying drugs made of active components with known ability to corrode glass and to dissolve the silica matrix. Sometimes these ingredients are dissolved in an alkaline medium that dramatically increases the glass corrosion and potentially causes the issue. As this action is strongly affected by time and temperature, flaking may become visible only after a long incubation during storage and requires systematic monitoring to be detected at its early stage. If the nature of the filling liquid is the driving force of the phenomenon, other factors are of primary importance. The surface morphology created during vial forming is a key issue, being a function of the forming temperature that is higher in the cutting step and the forming of the bottom. Delamination occurs generally on the vial's bottom and shoulder, where extensive flaming can favor a strong evaporation of alkali and borate species and the formation of heavily enriched silica layers. When these layers are in contact with a solution, they are subject to a differential re-hydration that may result in cracking and detachment of scales. The purpose of this investigation is to identify testing conditions and parameters that can be used as indicators of an incipient delamination process. Extractions with 0.9% KCl solution for 1 h at 121 °C can be used to simulate a long-term contact with aggressive pharmaceutical preparations, while SiO2 concentration in the extract solution can be taken as an index of glass dissolution. The conclusions developed by this study can provide pharmaceutical manufacturers with information needed to help prevent glass delamination in their processes.
- Hydrolytic resistance
- Delamination
- Glass corrosion
- ICP-OES
- Accelerated extraction test
- pH
- Organic acid extractants
Introduction
The primary objective of a glass container intended to contain parenteral preparations is to ensure safe usage for the patient. The occurrence of solid particles is an extremely negative event that drastically impairs the suitability of an entire lot—even in the case of a single incidence—and needs to be considered with extreme caution to understand its root cause and nature (precipitation of the active principle? glass fragments? silicate particles? lamellae?). The appearance of shiny, needle-shaped particles in the range of 50 and 200 μm floating in the liquid is generally referred to as a consequence of glass delamination (1). As the phenomenon is triggered by time and temperature, flaking may occur after a long incubation time during storage and requires systematic monitoring to be detected at its early stage. The use of an accelerated test is therefore encouraged to increase the possibility that incipient propensity for delamination is detected well before it may become visible to the naked eye.
Bacon and colleagues (2) followed a similar approach in the early 1940s when they tried to simulate with accelerated tests the behavior of soda-lime glass bottles that were supposed to undergo prolonged storage before the beverage was consumed. A set of containers of different compositions was filled with distilled water and subjected to variable heating conditions between 50 and 121 °C for progressively decreasing time intervals. The appearance of flakes was reported for all sets under progressively stricter test conditions, indicating that flaking was probably related to glass compositions, but an absolute correlation was not found.
The study described the phenomenon but did not explain the source. Further studies carried out with more accurate means using surface analytical techniques (3⇓–5) identified the mechanism responsible for delamination to occur. Storage of the empty bottles under uncontrolled conditions of humidity and temperature seems to be a key factor (6). When empty bottles of soda-lime glass are improperly stored, occurrence of repeated condensation and evaporation cycles of water films on the inner surface may favor delamination. The net result is the progressive alkalinization of the water film at the interface and the corrosion of the outmost layers. At its earliest stage, corrosion is made visible by the formation of a diffused opalescence over the whole body of the bottle; this opalescence, is mostly made of alkali carbonate salts. These salts can easily be removed by rinsing, but they leave an altered glass layer underneath.
Due to different crystallization patterns, opalescence may be more or less visible even for bottles belonging to the same pallet, but infrared reflection spectroscopy (IRRS) (7, 8) can give an absolute estimate of the degree of dealkalinization by observing the shift of the Si-O stretching (S) peak and the silicon-oxygen-alkali (NS) peaks occurring between 900 and 1200 cm−1. The consequence of extensive weathering is hence the formation of an altered, alkali-depleted, silica-rich layer that has an expansion coefficient that is fairly different from the glass substrate underneath. When bottles are filled with any kind of liquid, even water, the substrate and the altered layer are subject to strong rehydration to an extent that depends on their chemical durability and relative thickness. When the thickness and the flexibility of the altered layer become critical as compared to the substrate, the layer begins to crack and a simple shaking is sufficient to start its complete demolition.
Delamination does not affect only soda-lime glasses. In sulfur-treated containers and glass tubing vials, delamination follows a similar route, but the origin is different. The purpose of a sulfur treatment (9) is the formation of an outer layer, which is alkali-depleted and silica-enriched, that acts as a barrier to the extraction of acidic/neutral aqueous solutions. When the silica layer exceeds a given critical thickness, it may become susceptible to flaking according to the same mechanism described above for soda-lime bottles. Figure 1 illustrates the abundant flaking that occurred to a Type II glass container, 500 mL capacity, treated with freon gas, filled with a 0.9% KCl solution and autoclaved for 1 h at 121 °C [Note: According to the European Pharmacopoeia (EP) (10), a Type II glass container is defined as a “soda-lime-silica glass with a high hydrolytic resistance resulting from suitable treatment of the surface.”]
Type II infusion bottle with suspension of brilliant, needle-shaped flakes after extraction with 0.9% KCl solution at 121 °C for 1 h.
In the delamination process of Type I glass tubing vials, the surface morphology created during vial forming is the key point, being a function of the forming temperature, which is higher in the cutting step and shaping of the bottom. [Note: According to EP (10), a Type I glass container is defined as “a borosilicate glass with a high hydrolytic resistance due to the chemical resistance of the glass itself.”]
Delamination occurs generally at the vial's bottom and shoulder, where extensive flaming can favor strong evaporation of alkali and borate species and the formation of heavily enriched silica layers. As in the case of soda-lime containers, the formation of these layers is the first stage of an incipient delamination that develops according to the same mechanism described above.
Severe extraction conditions may favor an increase of the delamination rate. It is well known (11, 12), for example, that silica complexing agents like citric and glutaric acid, EDTA, and phosphates are very aggressive to the glass and promote extensive corrosion. The pH is another key factor because alkaline solutions (pH >8.5) cause the total dissolution of the leached layers (8), contributing to increase the risk of delamination from silica-enriched areas.
Actually, other factors may affect delamination: speed of the transformation process, burner flame temperature, improper annealing stage and occurrence of residual stresses, ionic strength (13) of the filling liquid, chemistry of the buffer supporting the active principle, the terminal sterilization process, storage time, and the type of glass.
Reducing the risk of delamination is therefore a serious problem that requires a systematic approach. The aim of this paper is to highlight the interaction between several glass types in contact with different extractants, including slightly alkaline preparations, and to investigate whether there is a correlation between EP titration values and evidence of delamination.
Materials and Methods
Materials
Analytical grade reagents were used throughout. Grade I deionized water produced by the Millipore ELIX system was exclusively used as a filling liquid. Its quality corresponds to water for injection (WFI) described in the EP. Pharma-grade citric acid powder (Carlo Erba, Rodano, Italy), 99% w/w Glutaric Acid (Sigma Aldrich, Steinheim, Germany), and 99% w/w Potassium Chloride (Sigma Aldrich) were used as extractants. Potassium Hydroxide pellets (Sigma Aldrich) were used for pH adjustment. Calibration curves for Si, Al, and B were prepared from 1000 mg/L commercially available standard stock solutions, (Prolabo, Lutterworth, UK). Rhodium stock solution (1000 mg/L, Prolabo) was used as an internal standard.
Several Type I glass vials, specifically Expansion 51 provided by different suppliers A, B, C (E51/A, 51/B, and 51/C, respectively), Expansion 33 (E33), and sulfur-treated Expansion 51 (E51S) were selected. More particularly, E51/A and E51/C were made of the same base glass, but were produced by a different conversion process. Because migration increases with increasing exposed surface, small vials of 3.2 mL capacity were purposely selected in order to maximize the surface area to volume ratio (SA/V = 4.5 cm−1) and simulate the worst possible conditions of use.
Experimental Design
Vial resistance to delamination was investigated by testing with different extractants under accelerated test conditions. Hydrolytic resistance of each vial set was determined by titration according to the surface test described in the EP (10). Other extractions were carried out with aqueous solutions at pH around 6, with solutions with increased ionic strength (0.9% KCl at pH 6), then the combined effect of ionic strength and alkaline pH was tested with a 0.9% KCl solution adjusted to pH 8. Finally, the effect of organic acids (citric acid and glutaric acid) buffered at pH 8 was investigated.
Autoclave Extractions with Aqueous Solutions (H2O and KCl)
Four different glass vials were tested: E51/A, E51/B, E33, and E51S. A set of 24 vials was taken for each glass; 12 vials from each set were accurately rinsed, filled with high-purity distilled water (WFI according to the EP, pH 5.5), autoclaved for two sequential runs at 121 °C, each run lasting 1 h. After each autoclave run, six vials were taken for visual inspection, three were used for pH measurements, and the remaining three used for inductively coupled plasma–optical emission spectrometry (ICP-OES) analysis (Si, Al, B). Each result is the mean of at least three determinations. The same procedure was followed with another, identical set of 24 vials using a 0.9% KCl solution. Starting pH was around 6.
Autoclave Extractions with Organic Acids (Glutaric and Citric Acids)
The glass vials described above were subjected to the same analytical procedure but using the following filling liquids:
-
a) solution of 3% glutaric acid + 1.5% KCl, pH adjusted to 8
-
b) solution of 3% citric acid + 1.5% KCl, pH adjusted to 8
Extraction at 80 °C for up to 48 Contact Hours with Citric Acid
A set of 36 vials were taken for each of the glass types described above, filled with a solution of 3% citric acid + 1.5% KCl, and the pH was adjusted to 8. Vials were placed in an oven at 80 °C, six samples were taken after 2, 4, 18, 24, 48 contact hours, and measurements and visual inspection performed as above.
Autoclave Extractions with a Slightly Alkaline Solution and High Ionic Strength
The following glass types were selected: E51/A, 51/B, and 51/C, and E51S.
Filling liquid was a 0.9% KCl solution, pH was adjusted to 8.0 with diluted KOH.
The same analytical procedure was followed as above, except that three autoclaving cycles were run to better discriminate between glasses of the same family (E51) on the basis of their silica release.
Scanning Electron Microscopy with Energy Dispersive X-Ray Analysis
Some vials were analyzed by scanning electron microscopy (SEM) to highlight the morphology of the inner surface of containers extracted with 0.9% KCl solution for 1 h at 121 °C, by comparison with as received samples. Energy dispersive x-ray spectroscopy (EDS) was used to confirm that visually observed particles were actually flakes. Samples were observed at five different positions: bottom, 5 mm from the bottom, at half height, 7 mm from the top, and at the shoulder.
Apparatus
A De Lama autoclave was used and the thermal cycle described in the EP was carried out each time. Temperature settings were checked using a properly calibrated external thermocouple; calibration of the whole system was verified using the reference containers CRM 435 (14).
The analysis of the extraction solutions was carried out by ICP-OES using an ICP-OES VARIAN 710 ES spectrometer. Samples and standards were acidified on-line with a 2% HNO3 solution in presence of Rhodium, all calibration curves were accepted within 5% relative error, and a recalibration was run every 10 readings using the reslope function. All readings were processed vs Rh readings, which was used as internal standard (10 mg/L).
An optical device produced by S.P.A.M.I. engineering (Stevanato Group, Piombino Dese, Italy) was used for the visual inspection of flakes in the filled samples. Vials were placed vertically on a sample holder and rotated while illuminated from the bottom with a fiber optic system against a dark background. This contrast allowed easy detection of the particles floating in the filled solution.
The morphological study of the inner surface of vials was conducted using a SEM FEG-ESEM FEI-Quanta 200F, equipped with an EDS microprobe. Acceleration voltage was 25–20 kV, spot size 4.5–4.0. As analysis was carried out in low-vacuum mode, samples were not metal-coated.
Results and Discussion
Extraction Data
Each glass type used in this study was preliminarily tested according to the EP surface test; results are shown in Table I.
Titration Values of the Different Glass Types Used in this Study, Tested According to EP Surface Test. All Data Are Expressed in Milliliters of 0.01N HCl
The results of the extractions with aqueous solutions are reported in Table II in terms of SiO2 concentration (expressed in milligrams per liter) and final pH. It was considered that silica concentration in the extract solutions could be used as an indicator of the extent of dissolution of the glass matrix, while the pH could give some information about the associated corrosion mechanism. Sulfur-treated vials showed the lowest silica migration, which was explained by the low pH reached even after the second autoclave (pH <8 with any kind of extractant). Under this condition extraction is ruled out by leaching, that is, by the exchange between water proton ions and alkaline ions from the glass surface. The net result is the formation of a silica-rich protective layer that is not dissolved by the attack of water and potassium chloride solutions. Conversely, in Expansion 33 glass vials extracted with KCl, which is known to be more aggressive than water because of its ionic strength (13), flakes appeared after the first autoclaving. Extraction pH in water was around 9, while it decreased to about 8.2 in the KCl solution, where silica concentration was much higher (64.2 vs 20.7 mg/L), probably because of the formation of silicic acid (15). These SiO2 concentration values are consistent with a total dissolution mechanism, where OH− ions cause the breakage of the glass network and extensive silica dissolution (8).
Extraction Data (mg/L) for Glasses Extracted with Aqueous Solutions and Subject to Multiple Autoclave Cycles (Starting pH of the KCl Solution was 6)
Figure 2 is a visual representation of the extractions carried out with the 0.9% KCl solution. It is interesting to note that EP titration values (see Table I) correlate well with SiO2 concentrations, as SiO2 increases with increasing EP values. The E33 glass, which in principle is to be considered as the most chemically resistant, is actually the least resistant because its surface integrity had been compromised by the conversion process (EP titration value is the highest among the four glass types, 0.92 mL vs a limit value of 1.30 mL for this capacity).
SiO2 increase in 0.9% KCl extract solutions from different glasses subject to (1 + 1) autoclave cycles at 121 °C.
To confirm whether EP titration values could be used as indicators of delamination in the whole pH range, further extractions were carried out with a 0.9% KCl solution adjusted to pH 8. Glass vials from E51/A, 51/B, and 51/C, and E51S, were tested; the E33 glass was purposely not selected because of its poor resistance to alkali. The results are visualized in Figure 3: the E51S glass gave the best performance in terms of silica concentration, but the presence of flakes was observed right after the first autoclave. This means that the silica layer cannot withstand the alkaline attack and starts flaking. Glasses E51/A and E51/C—which differ because of the conversion process only—do not show a significant difference except for the second extraction. Altogether they showed more or less similar behavior: all produced flakes, and after three autoclave runs the silica concentration was approximately the same for all glasses irrespective of their EP titration values (see Table I). It was concluded that EP values can be a reliable indicator of the glass performance as long as the pH is lower than 7 (i.e., with acidic or neutral solution); with alkaline solutions the observed ranking is reversed. This is consistent with the knowledge that borosilicate glass, when tested for its alkaline resistance, is far less durable than soda-lime glass (16).
SiO2 increase after extraction in 0.9% KCl adjusted to pH 8 from different glasses subject to multiple autoclave cycles at 121 °C for 1 h.
The effect of organic acids was also tested. The extraction solution was a mixture of 3% organic acid + 1.5% KCl, adjusted to pH 8 given that it was supposed to simulate the composition of a given pharmaceutical formulation. Extraction data are reported in Table III. All glasses were strongly attacked: flakes appeared in all glass vials after just the first autoclave run. Citric acid was about 3 times more aggressive than glutaric acid; the pH was almost buffered at about 8.5. Silica concentration was very high, especially for the E33 glass which confirmed to be the less durable under conditions of alkaline attack. It is important to point out that the steep increase of silica concentration in such small volumes may cause the precipitation of silicate compounds because of exceeded solubility limits. Only the filtration of the particulate and a SEM analysis, which in this study was carried out for pH 6 KCl extract solutions only, could be able to discriminate between real flakes and undissolved silica particles. As flakes were observed first with aqueous solutions, it was assumed that the particles that appeared under the most severe conditions were also flakes.
Extraction Data (mg/L) for Glasses Extracted with Organic Acid Solutions Adjusted to pH 8 and Subject to Multiple Autoclave Cycles
Extraction with the citric acid mixture was replicated under less severe conditions. The extraction temperature was kept at 80 °C, and extract solutions were analyzed after specified contact times. Again, the concentration of silica in E33 vials increased very steeply and flakes appeared after the first 2 h. The behavior of sulfur-treated glass vials was similar to the one observed with alkaline solution: silica concentration was rather low, but flakes appeared after 2 h extraction. The best performance was seen with the E51 glass vials, where flakes appeared only after 24 h (see Figure 4). Once again it was confirmed that sulfur-treated and Expansion 33 glasses are the least resistant to flaking with strongly alkaline extractants.
SiO2 concentration after extraction with 3% citric acid + 1.5% KCl, pH 8, at 80 °C for up to 48 contact hours.
To confirm the data and for mechanism interpretation, the ratio SiO2/B2O3 for the different kind of glasses under different extract conditions was investigated. When a glass is subject to a total dissolution process, the ratio between the main elements remains constant and the composition of the extract is similar to the composition of the base glass (8). If the ratio changes, other mechanisms are involved. For the E51/A glass extracted with KCl at pH 8, the ratio remains constant over all autoclave cycles (see Table IV). This means that total dissolution occurs starting from the first autoclave run; flakes appear soon but they do not alter the dissolution equilibrium. For the E51S glass, extraction with KCl at pH 6, the ratio increases progressively with increasing autoclave numbers. During the first run no flakes are formed, the protective layer remains intact, and extraction occurs according to a leaching mechanism. During autoclaves numbers 2 and 3, leaching is accompanied by partial dissolution of the silica-rich layer, flakes are formed, and silica concentration increases (ratio R = 1.81). With KCl at pH 8, flakes have already formed during the first autoclave run (R = 1.62). The protective layer is dissolved early and silica concentration increases according to the increased dissolution ability due to the higher pH (R = 2.32). With glutaric acid, the dissolution is more pronounced (R = 2.54 after the first autoclave), an extended dissolution of the protective layer occurs, and flakes are formed early. During the second autoclave the process continues and silica concentration increases due to further dissolution of the base glass (R = 3.54). It was concluded that any increase of the SiO2/B2O3 ratio correlates well with the occurrence of flakes and may be used to describe the delamination propensity of a glass, even at low SiO2 concentration values, as for sulfur-treated glasses.
SIO2/B2O3 Ratio in Extract Solutions from Glasses Subject to Multiple Autoclave Cycles with Different Extraction Media
SEM-EDS Results
The EDS results obtained on the glass types extracted with a pH6 KCl solution for 1 h at 121 °C indicated that the composition of the particles removed from the inner surface contained the same elements of the glass of origin. SEM micrographs (a) and (b) in Figure 5 are representative of the inner bottom surface of E33 glass vials before extraction. Micrographs (c) and (d) were obtained from the same glass vials after the KCl extraction treatment and show important signs of corrosion. Similar micrographs taken 5 mm from the bottom showed the appearance of pitting and incipient delamination, indicating that the process formation together with the corrosion ability of the extractant had a significant influence on the delamination route.
SEM micrographs of the inner bottom surface of Expansion 33 glass vials before (a, b) and after (c, d) extraction with 0.9% KCl for 1 h at 121 °C.
Conclusions
Several types of borosilicate glasses, both sulfur-treated and untreated, were subjected to treatment with different extractants for repeated autoclave cycles of 1 h at 121 °C. The propensity for delamination was observed measuring the increase of SiO2 concentration in the extraction solutions by ICP-OES, while the presence of particles was monitored by optically assisted visual inspection. Results obtained with neutral aqueous solutions (H2O and 0.9% KCl solutions) indicate that SiO2 concentration correlates with hydrolytic resistance, increasing according to the same order. Under these conditions the sulfur-treated glass showed the best performance in terms of dissolved silica; no flakes were observed in the examined glass types.
When slightly alkaline solutions were used as extractants (0.9% KCl solution adjusted to pH 8), SiO2 concentrations in the extracts increased very steeply, irrespective of their EP values. In this case the glass ranking found with neutral solution was reversed: in the sulfur-treated glass, flakes occurred very early even if the SiO2 concentration was low, while E51 glasses were the best performing.
It was concluded that under alkaline attack, EP values did not respond as performance indicators and that SiO2 concentration alone was not sufficient to predict flaking. Visual inspection was an essential support in detection.
When extractions were carried out with organic acids like glutaric acid and citric acid, dissolved silica was very high and early flaking was observed in all tested glasses, more particularly in the E33 glass.
This study suggests that EP titration values can be used as indicators of the chemical durability of the glass against neutral aqueous solutions only. When vials are in contact with alkaline solutions and similarly aggressive solutions, the glass performance is better represented by the concentration of the extracted silica. An increase in silica concentrations indicates glass corrosion and an increasing risk for further delamination.
It was concluded that extractions with 0.9% KCl solutions can be used as an accelerated test to highlight the propensity of a glass to delaminate, but in no case can be taken as a guarantee that the glass will not delaminate when exposed to the pharmaceutical drug, whose extraction ability needs to be studied case by case.
Conflict of Interest Declaration
The Authors declare that they do not have any financial or non-financial competing interests related to the content of the manuscript.
Acknowledgments
The authors gratefully acknowledge the contribution of Monica Favaro from Consiglio Nazionale delle Ricerche (CNR) for the analyses by SEM-EDS and valuable comments.
- © PDA, Inc. 2012
Jump to section
Related Articles
Cited By...
- Characterization of a Novel Particle in a Pharmaceutical Drug Product
- Investigating the Effects of the Chemical Composition on Glass Corrosion: A Case Study for Type I Vials
- A Contribution to an Understanding of Complexing Agents and Salt Solutions Reaction Mechanism on Chemical Attack of Type I Molded Glass Containers
- Introducing the Alba(R) Primary Packaging Platform. Part 2: Inorganic Extractables Evaluation
- Historical Review of Glasses Used for Parenteral Packaging
- Complexing Agents and pH Influence on Chemical Durability of Type I Molded Glass Containers
- Development of Conductivity Method as an Alternative to Titration for Hydrolytic Resistance Testing Used for Evaluation of Glass Vials Used in Pharmaceutical Industry
- Comparative Delamination Study to Demonstrate the Impact of Container Quality and Nature of Buffer System
- A Biopharmaceutical Industry Perspective on the Control of Visible Particles in Biotechnology-Derived Injectable Drug Products
- Comparison of Acid Titration, Conductivity, Flame Photometry, ICP-MS, and Accelerated Lamellae Formation Techniques in Determining Glass Vial Quality
- Particulate Study on NeoProfen, a Neonatal Injectable Product
- Filtration of Glass Delamination Particles with West Pharmaceutical Vial Adapters
- Implementation of a High-Throughput Ion Chromatographic Assay To Assess Glass Degradation in Drug Product Formulations
- Materials in Manufacturing and Packaging Systems as Sources of Elemental Impurities in Packaged Drug Products: A Literature Review
- A Quick Test To Monitor the Delamination Propensity of Glass Containers
- Novel Mechanism of Glass Delamination in Type 1A Borosilicate Vials Containing Frozen Protein Formulations
- Determining the Delamination Propensity of Pharmaceutical Glass Vials Using a Direct Stress Method