Abstract
Pharmaceutical products are packaged in containers so that they can be manufactured, distributed, and used. Because extractables from such containers are precursors of leachable impurities in the product, extractables represent potential hazards to user safety. Polypropylene resins are frequently used as materials of construction for packaging of liquid parenteral drug products. Thus, extractables profiling of polypropylene resins may be an effective means of hazard identification associated with the resin's safe use. Twenty-one polypropylene resins were extracted using aqueous and organic extraction solvents, and the resulting extracts were screened for extractables using appropriate general chemistry, chromatographic, and spectroscopic methodologies. The resulting extractables profiles were toxicologically reviewed by a defined process to identify potential hazards given a specified therapeutic application involving long-term use of a large-volume aqueous parenteral drug product. The organic extractables profiles of individual polypropylene resins were variable in terms of the individual extractable identified and their extracted levels, consistent with high variability in polypropylene resin formulations and pharmaceutical product manufacturing. However, the profiles were similar in terms of the groups of extractables measured. Thus, for example, all the resins had extractables associated with antioxidants, as all the resins contained antioxidants but the specific extractables for a given resin depended on the specific antioxidants present in that resin. Few of the targeted extractable elements were present in the extracts at measurable levels, although most resins had measurable levels of extracted aluminum, silicon, and alkali and alkaline earths. A worst-case extractables profile (all the extractables measured in individual resins at their highest levels) was toxicologically reviewed considering an aqueous large-volume parenteral drug product. This review established certain extractables as potential hazards whose actual toxicological safety risk assessment would require more rigorous data and a more rigorous process than those used for hazard identification.
LAY ABSTRACT: Packages used to contain aqueous parenteral drug products may be made of polypropylene. During storage, extractables from the polypropylene may leach out of the container and accumulate in the drug product, where they could affect product quality and/or safety. In this study, 21 polypropylene resins were characterized with respect to their organic and elemental extractables profiles. A worst-case extractables profile (highest level of all the extractables identified) was toxicologically assessed to establish potentially hazardous extractables. Certain extractables were established as potential hazards whose actual toxicological safety risk assessment would require more rigorous data and a more rigorous process than those used for hazard identification.
Introduction
Pharmaceutical products are packaged in containers so that they can be manufactured, distributed, and used. Pharmaceutical packaging serves many purposes, for example, packaging must safely protect a product and deliver that product from its manufacturer to its user. Specifically, packaging must
protect against all adverse external influences that can alter the properties of the product, e.g., moisture, light, oxygen and temperature variations;
protect against biological contamination;
avoid physical damage and leakage;
convey the necessary information related to the product and its effective and appropriate use.
Furthermore, packaging must meet regulatory requirements, as the pharmaceutical industry in general and packaging in particular is highly regulated. Lastly, packaging must meet expectations regarding aesthetics, merchandising, cost, ease of use, environmental impact, and so on.
Packaging and its materials and components of construction must be chosen in such a way that
the packaging does not have an adverse effect on the product (e.g., through chemical interactions such as leaching of packaging materials or absorption of essential drug product ingredients);
the product does not have an adverse effect on the packaging.
Because packaging can add substances to the packaged drug product, the user of the packaged drug product is exposed to those substances during product use. As such substances could have an adverse effect on user health, they are considered to be potential hazards associated with the safe use of packaged drug products.
The presence of packaging-related substances in the packaged drug product (leachables) can be ascertained by testing the drug product for such substances. Once the leachables have been identified and quantified, their potential safety impact can be established by considering the toxicology of the individual leachables and the patient exposure to the leachables (safety risk assessment).
Alternatively, packaging systems can be characterized to establish their extractables profiles, where extractables are substances that can be extracted from the packaging system (or its associated materials and components of construction) under laboratory conditions (which may or may not simulate the drug product's composition and conditions of contact with the packaging). As packaging-related leachables are either the extractables themselves or are derived from the extractables, extractables represent potential hazards to patient safety. Thus, the identification and concentration estimation of extractables is an exercise in hazard identification, that is, individual extractables are assessed to establish their hazard potential. Extractables that are established to be potential hazards may be more rigorously safety risk–assessed by appropriate means (e.g., targeted analysis of drug products for the extractables as leachables followed by safety risk assessment of the targeted substances), while extractables that are established to be non-hazardous would not generally require further assessment.
Clearly, the greatest potential hazard posed by extractables would be that situation where the extractables profile consisted of the greatest number of extractables and each extractable was present in the extract at the highest possible concentration. However, the laboratory conditions required to produce such an absolute worst-case extractables profile might bear little resemblance to the drug product's composition and conditions of use. In such a circumstance, hazard analysis of the absolute worst-case extractables profile could be largely irrelevant to the issue of developing and qualifying safe packaging.
On the other hand, extractables profiles based on laboratory conditions that mimic the drug product's composition and conditions of contact are highly relevant to the issue of developing and qualifying safe packaging. In circumstances where a packaging system is used with a single drug product, designing and implementing an extraction study that simulates the drug product and its conditions of contact is relatively straightforward. However, when a packaging system is used with multiple drug products of widely varying compositions, designing an extraction that simulates the entire population of drug products and their conditions of contact may be more challenging. Rather than perform extraction studies for each individual drug product, a relative worst-case extraction can be performed that seeks to “bracket” the conditions relevant to each individual drug product.
Polypropylene (PP) materials are frequently used as materials of construction for pharmaceutical packaging for liquid parenteral drug products. Thus, extractables profiling of PP resins may be an effective means of hazard identification associated with the safe use of the resins in packaging. Such information could be useful, for example, in the selection of PP resins for use in specific packages.
To illustrate this concept, an extraction process was established for screening PP resins for use in packaging for liquid parenteral products. Twenty-one individual and unique PP resins were extracted using the established extraction process, and the resulting extracts were screened for extractables using appropriate general chemistry, chromatographic, and spectroscopic methodologies. The resulting extractables profiles were toxicologically reviewed to establish potential hazards given a specified therapeutic application.
Experimental
Test Articles
Twenty-one individual and unique PP resins were obtained from various organizations that supply such resins to the pharmaceutical industry (Table I). Publicly available information about the resins that was provided by the suppliers is noted in Table I.
Extraction
The test resins, which were generally received in pellet form, were extracted without processing or modifications (e.g., grinding) as follows. Approximately 7 g of resins were placed in glass extraction vessels containing 200 mL of one of three extracting solvents: a pH 2 salt/acid mixture (0.01 M HCl/0.01 M KCl), a pH 9 phosphate buffer (0.0695 M Na2HPO4/0.0005 M KH2PO4), and a 40/60 (v/v) mixture of ethanol/water. Duplicate or triplicate extraction units were generated for each test article/extraction solvent pair. Extraction blanks were prepared by adding a similar volume of the extracting solutions to otherwise empty glass vessels.
Extraction was accomplished by autoclaving the test units and extraction blanks containing the aqueous extraction solutions at 121 ± 2 °C for 1 h. The test units and extraction blanks containing the ethanol/water solutions were heated at 55 ± 2 °C for 72 h to avoid any potential safety consequence associated with autoclaving such solutions in a closed vessel. These extraction conditions are consistent with national and international standards such as ISO 10993:12 (1). After heating, the extraction units were cooled to ambient temperature, where they remained during chemical testing. Each of the resulting extracts from each individual extraction unit was analyzed independently.
Extractables Profiling
The extracts and extraction blanks were chemically screened for organic semi-volatile and non-volatile extractables using complementary chromatographic methods developed for that purpose. Extracts were not screened for volatile extractables as arguably the best approach to address volatile extractables is to thermally extract the resins directly. Specifically, the extraction solutions were screened for generally semi-volatile organic extractables using gas chromatography with both flame ionization and mass spectrometric detection (GC/FID/MS). The extraction solutions were prepared for analysis by performing individual solvent exchanges with dichloromethane (DCM), after the addition of an internal standard and adjustment of the pH of the extraction solution up (to a pH of 10) and down (to a pH of 2). The DCM solutions obtained after the exchanges at both high and low pH were combined and the combined exchanges were evaporated to near dryness. A portion of the evaporated exchanges was analyzed in this form (underivatized), while a second portion was derivatized with bis(trimethylsilyl)acetamide, producing trimethylsilyl (TMS) derivatives that improve the peak shape and response to certain classes of anticipated extractables (e.g., acids and alcohols).
Extractables were established as those compounds associated with those GC peaks present in the extract chromatograms that were also not present in the extraction blank's chromatograms. The identities of the extractables were obtained by matching the MS spectra obtained for extractable's peaks and those contained in external and internal spectral libraries. The concentrations of the extractables were estimated via use of an internal standard (dimethyl phthalate), which was added to all the tested solutions at a concentration of approximately 100 ppb (μg/L) during sample work-up. The estimated concentrations were not adjusted for response factor variations between extractables.
Additionally, the extracts and extraction blanks were screened for generally non-volatile extractables using high-performance liquid chromatography with UV absorption and mass spectrometric detection (LC/UV/MS). In this case, the extraction solutions were injected directly into the LC system for analysis with no pre-injection processing.
Extractables were associated to those LC peaks present in the extract chromatograms that were also not present in the extraction blank's chromatograms. The identities of the extractables were obtained based on the molecular weight of the compound associated with a peak and its relative retention time. The concentrations of the extractables were not estimated due to the wide compound to compound variation in UV and MS responses. Thus, the primary purpose of the non-volatile analysis was to confirm the identifications that were obtained based on the semi-volatile (GC) analyses.
Operating conditions for the chromatographic methods are summarized in Tables II and III. The reporting thresholds for the organic extractables were generally 1 μg/g (∼0.05 mg/L), where the reporting threshold was established as that concentration above which it was likely that an extractable could be identified with a reasonable degree of certainty.
Additionally, the extracts and extraction blanks were screened for their levels of targeted metals and trace elements via inductively coupled plasma atomic emission spectroscopy (ICP-AES). Targeted elements included Ag, Al, As, B, Ba, Be, Bi, Ca, Cd, Co, Cr, Cu, Fe, Ge, Ir, Li, Mg, Mn, Mo, Na, Ni, P, Pb, Pd, Pt, Rh, Ru, S, Sb, Se, Si, Sn, Sr, Ti, Tl, V, Zn, and Zr. Extracted quantities of the targeted elements were calculated as the difference in the element's concentration in the extracts and the extraction blanks. The reporting threshold for the ICP-AES analyses was generally 1 μg/g (∼0.05 mg/L), except for P, Se, and S, which had reporting thresholds of 3 μg/g.
Lastly, certain of the extracts and extraction blanks were characterized for certain general chemical properties such as pH (ethanol/water extracts), total organic carbon (TOC, aqueous extracts), and UV absorbance (all extracts). Although the absolute values obtained for the measured pH in the ethanol/water extracts may have a limited utility, the difference in pH (delta pH, extract versus extraction blank) is a clear indicator of acidic or basic extractables.
Hazard Identification
Supporting toxicological information for the extractables was obtained from published literature, with a preference for either no observed effects (NOEL) or no observed adverse effects (NOAEL) data. If NOEL or NOAEL values were not found, other available toxicological endpoints were used—TDLO (lowest published toxic dose), LD50 (amount of an ingested substance that kills 50% of a test sample), etc. Once identified, the toxicological endpoints, which were generally based on oral dosing, were converted into hazard thresholds (HTs), in mg/kg (milligrams of substance per kilogram of body weight), after applying relevant uncertainty factors (UFs) described as follows and considering a body weight of 50 kg (eq 1). where
Uncertainty factor UF1 = factor that considers inter-individual variations among humans. UF1 = 10 except when UF1 is based on a residual solvent PDE (permissible daily exposure) or a U.S. Environmental Protection Agency (EPA) reference dose (RfD). In these cases UF1 = 1, as inter-individual variations have been built into the PDE and RfD.
Uncertainty factor UF2 = factor that considers the relevance of data derived from species other than humans. UF2 = 2 for dog, 2.5 for rabbit, 5 for rat, 10 for guinea pig (and animals not specifically listed here), and 12 for mouse.
Uncertainty factor UF3 = factor that considers the quality and relevance of the toxicological data used. UF3 = 1 for a PDE or RfD. UF3 = 25 for an LD50. UF3 = 15 for a TDLo or LDLo (lowest dose of a toxic material at which the death of the exposed test animal occurs), although a value of 25 was applied to a few extractables for which severe toxicity was indicated. For NOAEL, NOEL, and LOAEL (lowest observed adverse effect level), the value of UF3 depended on the duration of the study supporting the toxicological endpoint; UF3 = 1 for chronic or multi-generational studies of longer than 1 year, UF3 = 2 for studies of between 6 and 12 months, UF3 = 5 for studies of between 3 and 6 months, and UF3 = 10 for studies of less than 3 months (90 days).
Uncertainty factor UF4 = factors that account for bioavailability. For data obtained for an intravenous route, UF4 = 1. For data obtained for an intraperitoneal, dermal, or subcutaneous route, UF4 = 2. The value of UF4 used for the oral route depended on the availability of oral bioavailability data. UF4 = 1 if the oral bioavailability was 90% or greater. UF4 = 2 if the oral bioavailability was between 50% and 90%. UF4 = 5 if the oral bioavailability was between 10% and 50%. UF4 = 10 if the oral bioavailability was less than 10% or if the oral bioavailability was unknown.
Information used in the calculation of the HT is contained in Table IV. This exact process for calculating the HT was adopted for the purpose of this study and is not, to the author's knowledge, specifically employed by individual pharmaceutical companies for the purpose of either hazard identification or safety risk assessment of extractables or leachables.
To illustrate the hazard review process, the extractables data was processed considering the following therapeutic application (case study):
Container for an aqueous large-volume parenteral (LVP) drug product consisting of 5 g of an individual PP resin and which contains 100 mL of the drug product.
Daily dosing of the drug product is 600 mL/day (6 containers daily exposure).
Acute therapy (drug product administered as above for a period of less than 14 days).
In this situation, the daily patient exposure (DE) to an individual extractable becomes: where Ce = highest concentration of an extractable measured in any extracting solution (μg/mL)
Ve = extraction solution volume (200 mL)
We = weight of resin extracted (7 g)
The margin of concern (MC) is calculated as the ratio of the daily exposure (DE) and the hazard threshold (HT):
Collection of Additional Toxicological Safety Data
Individual extractables were assigned to a Cramer class (2), using the Cramer rules, with extensions. Based on quantitative structure–activity relationships (QSARs), the Cramer classification is a rules-based process that sorts compounds into three classes; Class 1 (low risk of toxicity), Class 2 (intermediate between 1 and 3), and Class 3 (either no basis to presume safety or positive indication of toxicity).
Additionally, individual extractables were assessed by in silico QSAR analysis for their mutagenic/carcinogenic potential using the Benigni/Bossa rule base (3). The Cramer classifications and mutagenicity assessments were performed using the appropriate modules of Toxtree software (4).
Lastly, two carcinogenicity/mutagenicity databases, the Carcinogenic Potency Database (CPDB) and the Chemical Carcinogenesis Research Information System (CCRIS), were queried for relevant information. The CPDB (5) is a widely used international resource containing the results of 6540 chronic, long-term animal cancer tests on 1547 chemicals. The CCRIS (6) is a database sponsored by the National Cancer Institute containing primarily in vitro carcinogenicity and mutagenicity test results for over 8000 chemicals.
Results
General Chemistry Testing
The general chemistry test results are summarized in Table V. In general, the pH of the ethanol/water extracts was fairly constant across all 21 test resins and was relatively unchanged as a result of the extraction (delta pH values less than ±0.5 at a nominal pH of approximately 6.0). Delta pH was not measured on the aqueous extracts because they were either buffered or at a pH where large amounts of acidic or basic extractables were required to produce a measurable pH change.
TOC and UV absorbance provide roughly comparable data in terms of the total amount of organic extractables, as the extracts with the highest TOC values also generally had the highest UV absorbances. However, TOC and UV may or not be closely correlated depending on the UV absorption properties of specific extractables. In general, the TOC and UV data provided a means of categorizing the resins in terms of their relative levels of organic extractables. Specifically, several resins could be classified as lower extractables, with TOC values generally less than 10 μg/g and UV absorbances (at 210 nm) of less than 0.010 in the aqueous extracts and less than 0.025 in the ethanol/water extracts. At the other extreme, a few resins could be classified as higher extractables, with TOC values above 50 μg/g and UV absorbances of 0.080 (at 210 nm) or greater. However, a majority of the tested resins had intermediate TOC and UV values, between 10 and 40 μg/g for TOC and between 0.010 and 0.050 for UV absorbances (at 210 nm).
Organic Extractables
Organic extractables that surfaced as a result of the GC and LC analyses are summarized in Table VI. The extractables surfaced by both methods were consistent in the sense that many extractables were revealed by both methods. This is to be expected as the methods' capabilities overlap to a certain extent. However, the LC method was more suited to the detection of higher molecular weight extractables (such as the antioxidants), which were generally present in the ethanol/water extracts while the GC method was more suited to the detection of lower molecular weight extractables such as smaller chain fatty acids.
The most commonly encountered organic extractables were palmitic acid, stearic acid (acid scavengers/lubricants), and erucamide (processing aid), which were present in measurable quantities in approximately 80% of the resins. Most of the resins contained extractables associated with various antioxidants, although the exact extractable and its associated antioxidant varied from resin to resin. Extractables related to other additives such as antiblocking and nucleating agents were also observed in the GC and LC chromatograms. The chromatograms of many of the resins contained peaks whose associated extractables were not identifiable, although the number and amount of such unknown extractables varied widely across resins.
Extractables Expressed as Elemental Impurities
Those extractables that contained elements that were targeted by the ICP-AES analyses and that were present in the extracts at levels above the reporting threshold of 1 μg/g are reported in Table VII. The only “true” elemental impurity [reported in ICH Q3D (33)] was barium, which was extracted from several resins at low levels. Other alkaline and alkali earth metals, such as sodium and calcium, were extracted from most resins at low levels. Aluminum was measured at low levels in many (but not all) of the tested resins, primarily in the low-pH extracts. Extracted silicon was reported in many resins. Extractable sulfur was reported in only one resin and is likely a false positive (analytical artifact).
Hazard identification
The hazard review of the organic extractables is summarized in Table VIII, which includes the margin of concern (MC) and supplementary toxicological data for individual extractables as well as groups of extractables such as total Irganox antioxidant-related degradation products, total antiblock-related extractables, total extractables related to Atmer 163 as a nucleating agent, and total unknowns.
Discussion
Sources of Extractables in Polypropylene
A polymeric material's extractables profile is closely related to the polymer's ingredients, as the ingredients are either potential extractables themselves or are the source of extractables (e.g., extractables as their decomposition or reaction products). Ingredients are added to polymers to perform specific functional roles. As the performance of polymers is adversely affected if degradation occurs during the various stages of polymer manufacture, fabrication, and subsequent use, polymers are protected via the addition of sacrificial antioxidants. Polypropylene specifically is sensitive to oxidation during high-temperature processing and generally cannot be processed without adequate stabilization via a primary antioxidant (e.g., hindered phenols). Moreover, PP is typically stabilized for use with a secondary preventative antioxidant (e.g., peroxide decomposers).
In addition to antioxidants, many modern PP resins are formulated with processing aids (lubricants) that facilitate the conversion of the resin into a finished component, acid scavengers whose purpose is intuitive, and nucleating agents that provide the proper polymer morphology.
Moreover, virtually all modern PPs are produced with the help of a Ziegler-Natta (ZN) catalyst. Generally, the current ZN catalysts are prepared in situ starting from a pre-catalyst mixture, consisting of magnesium and titanium (IV) chloride, and an internal donor, which is essential for the control of the stereoregularity (isotacticity) of the final polymer, thereby affecting the mechanical properties of the final material. Before or during the polymerization process, the pre-catalyst mixture is added to the reactor, where, together with an aluminum alkyl and an external donor, the activated catalyst forms in situ. The activated catalyst polymerizes propylene and its co-monomers into the various types of PPs and thus is consumed or rapidly decomposed. Moreover, the production process may include steps whose purpose is to remove residual catalysts and degradation products. Nevertheless, such residues and degradation products can end up as catalyst residues in the PP resin.
Extractables Profiles for the PP Resins
At first glance, Table VI suggests that the organic extractables profiles of PP resins are quite variable, consistent with high variability in PP resin formulations and PP manufacturing. While this may be true from a compound perspective, there is a clear consistency in terms of compound classes. Thus, for example, all the PPs tested are stabilized with an antioxidant package and virtually all of the PPs had extractables that are antioxidant-related. Polypropylenes with comparable antioxidant packages had comparable antioxidant-related extractables, although the levels of these extractables varied from resin to resin based on the resin's formulation and heat history. In fact, certain of the PP resins that were tested had no measurable antioxidant-related extractables, even though the resins were formulated to contain antioxidants.
The pervasive presence of fatty acids such as stearic and palmitic acids in the extractables profiles is consistent with common use of stearate salts as acid scavengers or processing aids. The presence of the lower molecular weight fatty acids reflects either impurities in or degradation products of the specific stearate salts used. The resin-to-resin differences in the levels of extracted fatty acids reflects the fact that PP resins vary significantly in terms of their levels of intentionally and unintentionally added stearate salts and that the level of stearic acids in some resins are tightly controlled, as extracted stearates have been linked to the formation of particulate matter (precipitated stearate salts), especially in high-pH solutions (34).
Catalyst-related extractables were present in nearly all the PP resins tested in detectable levels. However, as was the case with antioxidant-related extractables, the specific catalyst-related extractables present in each resin and the levels extracted from the resin varied from resin to resin. While the variation in the specific extractables reflects the widely different catalysts systems that are used with the tested resins, the resin-to-resin differences in the levels of catalyst-related extractables may reflect the degree to which individual resin suppliers control their level of residual catalyst.
Most, but not all, of the resins tested had low levels of extracted erucamide, consistent with its general use as a processing aid (slip agent). However, two of the resins tested had significantly higher levels of extractable erucamide, consistent with the fact that these specific resins were intentionally formulated to contain this substance in higher quantities.
Considering other extractables that could be linked to substances intentionally added to the PP resin or used in the resin manufacturing, various dissolved silicate species were extracted from those resins which were known to contain amorphous silica as an antiblocking agent. Additionally, the presence of ethoxylated amine extractables in the profiles for certain resins suggests that Atmer 163 (or a similar substance) was used as a nucleating agent during the production of these resins.
The extractables profiles of the tested resins include “orphan” extractables, that is, extractables that were measured in only a few of the resins, generally at low levels. When an extractable is an orphan, it is possible that the extractable may not be legitimate and that it reflects either an analytical artifact or “wishful thinking” as opposed being to a true extractable. In some cases, orphan extractables can be justified in that they can be linked to groups or sources of legitimate extractables. For example, while 2,6-di-tert-butyl-4-methylenecyclohexa-2,5-dien-1-one was measured as an extractable in only one resin in relatively small quantities, it can be justified as a legitimate extractable as it can be linked to either antioxidants or peroxide additives as a degradation product. Similarly, although several amides were measured in only a few resins, they logically can be linked to erucamide as impurities and in fact they were reported only in those resins that had the highest levels of extractable erucamide.
Beyond these extractables, Table VII contains several substances that are true orphans in the sense that they were observed infrequently at low levels and could not readily be justified based on available compositional and process knowledge.
Considering the extracted elements results (Table VII), a vast majority of the targeted metals were not extracted from any of the resins in measurable quantities. This is the expected outcome as there is no logical source of these metals as they are not intentionally added to the resins and they are not derived from the resins' manufacturing process. Those elements which were extracted from the resins in measurable quantities can be justified based on their intentional use with PP resins. Thus, for example, the extracted Al is associated with the catalyst, extracted Si is either associated with the catalyst (organic extractables) or the antiblock (inorganic extractables), B is associated with additives used to improve the resins' thermal conductivity and flame resistance, and sodium, calcium and barium are counterions associated with various formulation components (e.g., acid scavengers, antiblock, etc.).
It is noted that the organic extractables and extractables elements results reported herein are generally consistent with previously reported information for PP materials (35⇓–37).
Reconciliation of the Extractables Profiles with the General Chemistry Results
The general chemistry tests are useful as they reflect generic characteristics of the extractables. For example, TOC is a measure of the total amount of organic extractables and thus it is reasonable to expect that there is a correlation between the measured TOC and number and/or amount of all the organic extractables revealed in the chromatographic analysis (38), with the highest measured TOC corresponding to those resins with the largest number of organic extractables at the highest concentrations. This general relationship held true for the 21 resins investigated in this study, as the resins with the highest TOC levels had the highest levels of individual aqueous organic extractables. Thus, for example, two resins had the highest TOC values and contained relatively large quantities of extractable erucamide and related substances. The relatively higher TOC in the aqueous extracts of another resin correlated with the presence of larger quantities of butylated hydroxytoluene (BHT) in these extracts. Higher levels of catalyst-related extractables in resins 1 and 2 correlated to their relatively high TOC levels. The relatively lower levels of organic extractables associated with two additional resins was consistent with the lower TOC of their aqueous extracts.
While UV absorption data also provides an indication of the level of organic extractables, the interpretation of UV data is somewhat more complicated than the interpretation of the TOC data given the widely variable spectral properties of the organic extractables. In point of fact, trends in UV data are typically useful when they are interpreted in the context of the relationship between the UV and the TOC data and in the context of the absorbances as a function of wavelength as such an interpretation provides insight into the chemical nature of the extractables. For example, while the aqueous extracts of one resin had only modest TOC levels (30 μg/g), they had the highest measured UV absorbance in the low pH and ethanol/water extracts. Furthermore, the higher UV absorbances persist at the higher detection wavelengths. This behavior of this resin is consistent with the observation that its major extractable was BHT. On the other hand, two other resins had TOC values slightly higher than the previous resin but much lower UV absorbances, consistent with the fact that their major extractables are silanediols (associated with the resin's catalyst). Lastly, the resin with the highest levels of extracted erucamide both had high TOC values and among the highest UV absorbances.
In general, the pH of the ethanol/water extracts was relatively unchanged as a result of the extraction (delta pH values less than ± 0.5 at a nominal pH of approximately 6.0), indicating that neither strong acids or strong bases were extracted from any of the PP resins tested. Small pH changes associated with these extracts are consistent with the weak fatty acids that were present in many extracts.
Hazard Identification
The question of whether the PP resins that were tested in this study are potential hazards with respect to patient safety when they are used in parenteral packaging can be addressed considering two dimensions; the potential hazard posed by the materials themselves and the potential hazard posed by substances which would leach from the resins during their clinical use. Considering the first dimension, Table I provides evidence that the resins themselves are safe for use in parenteral product packaging. As noted in Table I, the vendors of many of the tested resins provide documentation and test results which establish that their resins meet the compendial specifications of the United States and Europe. Furthermore, most of the resins have been certified by their vendors as complying with international indirect food additive regulations, as meeting the biological reactivity specifications consistent with a USP Class VI designation and as complying with REACH regulations with respect to the absence of restricted substances. While none of these pieces of information individually definitively establishes that the resins are not hazardous, taken in composite the preponderance of evidence suggests that a majority of the resins would meet basic safety expectations.
The second dimension considers the hazard posed by the specific substances that have been identified as extractables, as extractables may become leachables in the packaged drug product. Because the testing performed was extractables profiling and because the testing was performed on resins (and not the packaging system), it is inappropriate for the test results to be used for the purpose of rigorous toxicological risk assessment as it is scientifically difficult to justify extrapolating resin extractables data to packaging system leachables data. Nevertheless, the resin's extractables profile can be toxicologically reviewed to establish whether any of the extractables, individually or as a composite, present a potential safety hazard. Extractables which are toxicologically established to pose a negligible hazard would be deemed to be suited for use without a more rigorous risk assessment. Extractables which are toxicologically reviewed to be a potential hazard would be more rigorously assessed both from a testing perspective (e.g., as targeted leachables) and from a toxicological perspective (more rigorous safety risk assessment process than that used for hazard identification). Resins which contained extractables which were clearly established as likely hazards could be eliminated as candidates for use in packaging systems. In this way, extractables profiling of resins followed by toxicological hazard review serves as a means of focusing the more rigorous risk assessment on those extractables that truly present a potential hazard.
The hazard assessment process used herein compares a patient's worst case exposure to leachables, established by the highest levels of extractables measured in the extraction study (DE), to a Hazard Threshold (HT) that has been calculated by a defined process for evaluating the toxicological safety data that is available for the extractables. The comparison is made “quantitative” in the sense that it is reflected in the Margin of Concern (MC), which is the ratio of the HT to the DE. The interpretation of the MC is straightforward; if the MC is greater than 1 (the exposure is less than the threshold), then it is unlikely that the extractable is a patient safety hazard. Alternatively, if the MC is less than 1 (the exposure exceeds the threshold), then the extractable is established to be a potential hazard. In fact, if the MC is much less than 1, it might be concluded that the extractable is a sufficient hazard that use of its associated resin is not recommended.
However, while the MC can be rigorously calculated from available data (thus making it “quantitative”), it is inherently imprecise given the imprecision of the data upon which it is based, specifically the analytical uncertainty in the means by which extractables' concentrations are estimated and the evaluation uncertainty that is inherent in the hazard identification process. Therefore, the MC is more properly interpreted in terms of certain bands or regions within the continuum defined by possible MC values. Thus, MC values in the band from 3 to 0.2 are established as being linked to extractables that are potential hazards. Extractables with MC values greater than or equal to 3 are established as presenting a negligible hazard while extractables with a MC less than 0.2 are established as likely hazards.
The results of such a hazard assessment for the extractables measured in this study and for the clinical therapy example used, are listed in Table VIII. When considering these results, the reader is reminded that the results are specific to the extraction study performed and the clinical situation specified in this manuscript. Any conclusions drawn about the hazard potential of individual extractables or any inferences associated with the safe use of an individual resin are specific to the situation considered in this manuscript and thus cannot be extrapolated to all possible pharmaceutical uses of the resins. A majority of the extractables have MC values well above 3 and thus are concluded to present a negligible safety hazard. No extractable had MC values less than 0.2 and thus no extractables have been identified as likely hazards. However, as a whole, “unknown” extractables have been established as presenting a likely hazard. A number of extractables have MCs between 0.2 and 3 and thus are established to be potential hazards, including catalyst-related extractables such as methylcyclohexylsilanediol and 1,1-dicyclopentylsilanediol, a specific antioxidant–related extractable (2,6-di-tert-4-methylene-cyclohexa-2,5-diene-1-one) and a set of extractables related to nucleating agents (multiple ethoxylated amines which were hazard reviewed as a group).
As the hazard review was based on the highest concentration for an extractable reported over the three extractions solvents used in this study and over all 21 resins tested in this study, it is instructive to consider whether the potential hazard presented by potentially hazardous extractables reflects PP resins in general or specific PP resins in particular. For example, in the case of 1,1-dicyclopentyl-silanediol, 4 of the 9 resins which had measurable levels of this extractable had levels that were high enough to produce an MC of less than 3. In the case of methylcyclohexylsilanediol, 1 of the 7 resins which had measurable levels of this group of extractables had levels high enough to produce an MC of less than 3. In the case of the ethoxylated amines, 1 of the 3 resins which had measurable levels of this group of extractables had levels high enough to produce an MC of less than 3. Thus, it is concluded that the potential hazard associated with these extractables is limited to specific resins and not to the resins in general. Nevertheless, investigators seeking to perform extractables profiling of PP resins would be well advised to pay specific attention to catalyst-related substances and residual nucleating agents.
In the case of total unknowns, the hazard review is so rigorous due (a) to the fact that the unknowns are assessed collectively and not individually and (b) the increased degree of caution required in the toxicological hazard review of unidentified substances. This is the case as the unknown extractables must be reviewed assuming that they are in fact a potential hazard. Given the low hazard threshold, the total levels of extractables in all 10 resins which had reportable unknowns are sufficiently high that the MC would be 3 or lower. Unless identities can be obtained for the unknowns or unless the hazard potential of the resins can be established by another means, the resins containing these unknowns would have to be classified as potential hazards. This outcome reflects the high daily dose volumes used in the case study and illustrates the situation in which high daily dose volumes requires extractables identifications for substances that are present in the extract at such low levels that securing the identities is a considerable analytical challenge.
In additional to the numerical hazard review, the hazard potential of extractables can be inferred based on others means of evaluation. For example, the safety hazard associated with extractables can be addressed by in silico methods that use structural characteristics of the extractables to “predict” their mutagenic potential. Additionally, certain in vitro tests, such as the Ames test, are well established as a means of establishing the carcinogenic potential of substances. The use of both in silico assessment and in vitro testing in the safety assessment of impurities such as extractables is well-established (32, 39).
Relevant in silico and in vitro information for the PP extractables, contained in Table VIII, was not universally available for all the extractables reported. In general, however, Cramer classifications for the extractables were poorly correlated with their MC values and thus provide a limited insight into the safety hazard associated with these extractables. Although one extractable, urea, triggered an in vitro alert, urea's relatively high MC and its lack of in silico alerts suggests that in the case study used herein it does not present a safety hazard. In silico alerts for several antioxidant-related extractables were consistent with the relatively low MCs obtained for these extractables and thus support the conclusion that these extractables pose a potential safety hazard. In silico alerts obtained for extractable phthalates were not confirmed by in vitro test results; this outcome, in additional to the relatively high MCs for these compounds, supports the conclusion that the phthalates do not present a safety hazard.
Considering the extracted elemental impurities, barium is the only reported EI that is listed in the ICH Q3D document. The DE for Ba is 51 μg/day, well below the parenteral permissible daily exposure (PDE) of 700 μg/g (34). With an MC of approximately 14, Ba is not considered to be a potential safety hazard. While silicon is reported as an extracted elemental impurity, it does not exist in the extract as silicon but rather as the organic and inorganic substances that were revealed by the chromatographic analyses. Thus, Si is hazard reviewed via the hazard reviews performed for the individual Si-containing extractables. The alkali and alkaline earths (Na, Ca) are generally not regarded as safety hazards. Considering boron, the U.S. EPA chronic oral reference dose (RfD) for boron is 0.2 mg/kg/day (40). As absorption of boron (as borate) is virtually complete (95% in humans and rats) and it appears rapidly in the blood and tissue of several mammalian species (41), no correction for oral bioavailability is necessary. With an MC of approximately 21 (DE of 480 μg/day, HT of 10000 μg/day), B is not considered to be a potential safety hazard. The hazard assessment of aluminum is sufficiently complex that it falls outside the scope of this manuscript.
Resin Screening in the Context of Safety Qualification of Pharmaceutical Packaging
It is possible to read this article and conclude that the author is advocating routine extractables profiling of resins as part of a defined and strategic process for qualifying a pharmaceutical packaging system for its suitability for use. This is not the case as the purpose of the article was to examine resins for potential safety hazards by performing a controlled extraction study.
Nevertheless, a three-phase approach which includes resin (material) characterizations has been advocated as being the appropriate means for qualifying a packaging system with respect to its suitability for use (42, 43). The justification for performing some degree of resin characterization is the simple truth that the single most significant determinant of a packaging resin's ability to impart leachables to a packaged drug product is its composition. Thus, when resins are selected for use in a packaging system, some knowledge about their composition and their behavior under defined extraction conditions provides the basis for rational and science-based selection (or rejection) of resins. Furthermore, composition information for resins may facilitate post-commercialization packaging system change control.
Whether the compositional information required for material selection and justification includes extractables profiling or whether it can be more focused on establishing composition is largely at the discretion of the material's user, based on the user's hazard identification and risk management systems and strategies.
In performing its hazard identification, this article focused on a specific pharmaceutical situation. Although the exact outcome of the hazard assessment performed in this article may or may not be relevant in other situations, the evaluation process used in this article can be used to assess these other situations.
Conclusions
Extractables profiles have been established for 21 PP resins that can be used for aqueous parenteral drug product packaging. While the extractables profiles of the resins are distinct at the compound and concentration level (consistent with the resin's varying formulations and processing), they are similar at the compound class level. Thus, for example, resins stabilized with antioxidants will have extractables profiles which include the degradation products of the antioxidants (which are typically Irganox 1010, Irganox 1076 and/or Irgafos 168). Similarly, as most resins contain acid scavengers/processing aids (which are typically a salt of stearic acid), their extractables profiles will include stearic acid and its related lower molecular weight acidic impurities, particularly palmitic acid. Many resins will container erucamide as a processing aid (slip agent) and their extractables profiles will include erucamide and related amides. As all resins require a catalyst during its production, the extractables profiles will include catalyst-related extractables including silanediols, phthalates and aluminum. Those resins that contain silica as an antiblock will have extractables that include inorganic silicates. If the resin contains a nucleating agent, then the extractables profile will contain the agent and/or its related substances.
As PP resins are not formulated with metals, they contain few extracted metals. As noted previously, extracted aluminum is derived from the catalyst while extracted silicon is derived from both the catalyst and the antiblock agent. Boron is extracted from PP resins that are formulated with agents for improving the resin's thermal conductivity and/or flame resistance.
As PP resins in general comply with the relevant compendial standards, meet relevant biological reactivity specifications, are approved for application in food packaging and meet relevant “bad actor” compositional standards, they are predisposed to have a negligible adverse impact on patient safety when used in container closure systems for aqueous parenteral drug products. Considering the safety hazard posed by resin extractables, it is noted that the organic extractables associated with the tested resins rarely trigger either in vitro or in silico alerts for carcinogenicity. More quantitatively, the extractables were reviewed for toxicological hazard potential considering a specific test case involving the use of a fictitious “worst case” PP resin (which possess all the extractables measured for each of the 21 individual resins studied at their highest measured levels) in the container for an aqueous parenteral drug product. Such a worse case is an exaggeration of reality as no individual resin had such a worse case extractables profile. In general, most of the inorganic and organic extractables from PP resins were established as negligible safety hazards, based on a review of their toxicological safety information. However, certain extractables associated with the resin's catalyst and extractables related to nucleating agents were established to be potential safety hazards. Proper risk assessment of these potential safety hazards requires (1) a more rigorous process than that used herein for hazard identification and (2) that the more rigorous process be applied to resins individually based on their own extractables profile.
Thus, it is concluded that the potential hazard associated with PP extractables is limited to specific resins and not to PP resins in general as many of the resins examined contained no extractable whose level was so high that it presented a potential safety hazard in the specific circumstance considered in this manuscript.
It is noted that such conclusions drawn from this manuscript and the study documented herein are only relevant to the specific circumstances (extraction study design and clinical use conditions) considered in this manuscript. The assessment performed in this manuscript is relevant only to the circumstances considered herein and cannot be used for, and is not relevant to, other potential pharmaceutical applications of the resins that were studied.
Extractables profiling of resins followed by toxicological hazard review serves as a means of focusing the more rigorous toxicological safety risk assessment on those extractables that truly present a potential hazard. Extractables which are toxicologically reviewed and established to pose a negligible hazard are deemed to be suited for use without a more rigorous risk assessment. Resins which contained extractables which were clearly established as likely hazards could be eliminated as candidates for use in certain packaging systems under certain packaging conditions. Extractables which are toxicologically reviewed and concluded to be a potential hazard would be more rigorously assessed both from a testing perspective (e.g., as targeted leachables with validated methods) and from a toxicological perspective (more rigorous safety risk assessment process than that used for hazard identification). So doing would provide a more rigorous and accurate toxicological safety assessment as the DE would be based on highly accurate product data and the tolerable intake (TI) would be based on a through and complete review of all available toxicological data.
Conflict of Interest Declaration
The author declares that he has no competing or conflicting interests.
- © PDA, Inc. 2017