QUALITY BY DESIGN
In a quality-by-design (QbD) approach to manufacturing, the goal is to design in the quality of the final product by understanding
all critical parameters and implementing robust manufacturing processes to control those parameters, as opposed to attempting
to test in the quality from an unstable, poorly understood manufacturing process. The importance of QbD in extractables and
leachables risk assessments, particularly in the OINDP application, was recently discussed (23).
In the risk assessment of leachables, the critical QbD goal is to understand and control the safety of the tool in the application.
The author's preferred process for achieving this safety is shown in Figure 1. The base of the pyramid is the responsibility
of the tool manufacturer and is where most of the safety is built in, as indicated by its size. Knowledge of the technical
literature could, for example, be used to understand and predict the impact of gamma sterilization on physical properties
and the amount and type of gamma-induced leachables.
Figure 1: Strategies for mimimizing the risks of leachables. (ALL FIGURES ARE COURTESY OF THE AUTHOR)
The green levels in the figure represent steps only the user of the tool can perform because they are highly application specific.
The brown level represents steps that both the manufacturer and user of the tool can perform. The manufacturer of the tool
tends to perform generic analytical testing, whereas the end user is more likely to perform analytical testing closely aligned
with the application of the tool. The size of each level reflects the degree to which it helps lower the risk of leachables
that affect safety. The key point in the graphic is to not be overly reliant on analytical chemistry and subsequent toxicological
assessment of the analytical data, but to understand, robustly design in, and control the safety of leachables, rather than
to test in the quality in the final application.
When Fawley published his milestone paper on the threshold approach to toxicology, the phrase "common sense" was prominent
in the title (24). While it took many years to gain legal acceptance, the threshold strategy is now well entrenched and is
being expanded on a global basis to a multilevel threshold strategy using the TTC approach. The FDA CFSAN still has only the
single-level TOR, which individual scientists at FDA have described as too inflexible (25).
The pharmaceutical arena has seen some well-publicized examples of leachables that potentially might affect patient health;
virtually all were from container closures. Examples in the past few decades have included polycyclic aromatic hydrocarbons
from carbon black fillers in elastomers, N-nitrosoamines or mercaptothiazole in rubbers, and diethylhexylphthalates from plasticized polyvinyl chloride blood and intravenous
bags and tubing (26, 27). Even permeation of leachables from labels and their adhesives through a low-density polyethylene
film into a drug-containing vial has been observed (28).
In the biopharmaceutical industry, the published leachable examples are fewer due to the relatively short time that biologics
have been manufactured. The issues in biopharmaceuticals seem more centered on API interactions with leachables and less about
potential direct toxicological issues, undoubtedly due to the greater inherent instability of biologicals relative to traditional
small-molecule pharmaceuticals (29). Nevertheless, a rubber leachable after a formulation change apparently caused an increased
risk of red-cell aplasia in European patients receiving EPO therapy (30).
Case histories of leachable problems present several clear trends in risks due to leachables. Because of their complex formulations
and manufacturing processes, cured elastomers often have a much greater chance of having leachables with direct health risks
than thermoplastics, and drug-leachable instability interactions are much more prevalent problems than direct leachable toxicity
concerns. The higher risk of cured elastomer issues should be addressed by minimizing contact area and time, or selecting
noncured (i.e., TPE) elastomers or over-molded elastomers (31). Drug-stability studies should be performed early in the material
evaluation process, and analytical-leachables studies done to characterize the performance of acceptable materials or establish
root cause for materials that reduce drug stability.
THE KNOWLEDGE APPROACH IN RISK ASSESSMENT
The goal of any risk assessment should be to promote a rational resource allocation to address potential problems, with the
highest risk areas receiving the highest scrutiny. To assess the toxicological risk of leachables from product-contact surfaces,
one must understand material science, solubility parameters, the effects of sterilization procedures such as gamma irradiation,
application-specific parameters (i.e., contact time, temperature, surface area and volume, solution properties, and proximity
to the final formulation), and relevant toxicology to assess the value of extractables and leachables testing.
This scientific assessment must be combined with information from the material supplier. Supplier information should substantiate
that the raw materials have appropriate 21 CFR clearance for the application, the proper controls are in place for cGMP manufacturing, and whether available generic extractables
or leachables data can help in the risk assessment. Often the risk assessment using the combination of the manufacturer's
generic leachables data with the end-use application-specific parameters and a TTC approach will conclude that further leachables
studies are not necessary to establish the safety of the leachables in terms of direct toxicity.
Table III shows the analysis of the toxicology risk using a series of potentially important variables when using three devices
in three applications, roughly based on the protocol suggested by the Biopharmaceutical Process Extractables Core Team (17).
Other possible risks from leachables, such as product formulation instability or assay interferences, would be assessed separately.
Table III: Toxicological risk assessment of leachables for three devices/applications. OINDP is orally inhaled and nasal drug
The first section of the table contains estimations of six variables that could affect the concentration of observed leachables.
The second section contains estimations of two variables related to the potential toxicological risk of the leachables. Rather
than assign numerical values to each risk level, such as the 1–10 scale previously suggested, the overall risk is estimated
with high, medium, or low categories. Rather than sum up the numerical risk levels to achieve an overall risk assessment,
the relative risk of toxicology of the leachables and the relative risk of the amount of leachables are evaluated separately.
The two risks are viewed as multiplicative, in line with the normal definition of risk as equal to the degree of the hazard
times the level of the exposure. This separate evaluation allows for the possibility that if the toxicology is estimated to
be low risk, then the concentrations of the leachable are not as important, much as in the TTC approach.