Evaluation of Comparative Performance of Orally Inhaled Drug Products in View of the Classical Bioequivalence Paradigms: An Analysis of the Current Scientific and Regulatory Dilemmas of Inhaler Evaluation

Introduction: Development of the BE Toolbox

Physical, Chemical and In Vitro Components. The scientific origins of bioequivalence (BE) can be traced to events that began to draw attention in the medical and pharmaceutical communities in the late 1950s, a period when the number of available drug products and manufacturers began to grow significantly. There were reports, mostly anecdotal, of variable treatment results when different batches of the same product from the same manufacturer were prescribed, loss of effectiveness when switching from one manufacturer to another, or oppositely, an increase in side effects or, in some cases, fatalities when switching from one manufacturer to another. Some of the earliest evidence for an explanation was offered by Levy and Hayes⁽¹⁾ in 1960 who, using a simple stirred beaker system, found significant differences in the in vitro dissolution rates of different brands of aspirin tablets. One year later⁽²⁾ they demonstrated a rank order relationship between the amount of salicylate excreted in urine and the dissolution profile of aspirin in their in vitro system. Further work and refinement led to a 1965 publication⁽³⁾ that showed a direct linear correlation between in vitro dissolution rate and the rate and extent of in vivo absorption. Among the significant contributions of this work was the demonstration that features of the human gastrointestinal tract could be duplicated in the laboratory using a simple continuous stirred tank reactor (CSTR) model. This fostered the idea that other physiological systems involved in drug absorption or transport via any administration route could be similarly modeled or replicated.

With this in vitro tool, other workers began investigations into factors that could play a role in drug dissolution from solid oral dosage forms. A rather large body of literature began to accumulate showing that drug substance dissolution could be affected by drug solubility, particle size, polymorphic form, salt form, hydrate or solvate presence, formulation composition, excipient quality, type of manufacturing equipment, manufacturing site, manufacturing parameters, and chemical and physical changes as a result of temperature or environmental exposure during product storage. In addition to identifying these risk factors for dissolution and absorption, the research also showed that they were controllable factors, meaning that they could be manipulated and monitored using appropriate quality metrics to reduce their risk of interfering with optimal and consistent bioavailability. In parallel, alternates to Dr. Levy's beaker configuration were developed. Serious investigations into factors that influenced test method variability and approaches to method validation were also begun. The first official pharmacopeial monographs in which dissolution testing was specifically stated as a requirement appeared in USP in 1970. Continued reports of widespread and serious clinical consequences of bioequivalence linked to variability in dissolution^(4,5) brought pressure to expand the number of monographs and institute other control measures.

Pharmacokinetic Components

The rapid development and access to a variety of analytical methodologies in the 1950s allowed other researchers to map the kinetic profile of drug and metabolite concentrations in body fluids after administration of drug substances. This led to the observation that relatively simple zero–order and first–order chemical kinetic or enzyme kinetic mathematical models could be used to describe appearance and elimination behaviors. Three simple and understandable kinetic parameters, namely peak concentration, time to peak concentration, and area under the concentration versus time curve (AUC), became the basis for comparing rates and extents of in vivo absorption between reference and test articles. Different statistical approaches, both model-dependent and model-independent, were developed and introduced as ways to make unbiased, quantitative determinations of equivalence. The number of published bioavailability studies increased and the noncontrollable components affecting pharmacokinetics (i.e., inter- and intra-subject variability) became better understood, leading to consensus on statistical power requirements and the minimum number of subjects to enroll in comparative studies to achieve reliable bioequivalence assessments.⁽⁶⁾ Normal healthy volunteers were recommended to reduce interferences and decrease experimental variability.⁽⁷⁾

Pharmacodynamic Components

Work in the field of antibiotics and hypnotic/sedative agents in the 1950s and 1960s uncovered the importance of both the amount of drug reaching pharmacologic receptors, and the temporal pattern of receptor exposure. The emerging field of pharmacodynamics showed that clinically desirable effects, as well as undesirable side effects, could be correlated with blood or tissue levels. This led to the proposition that blood or plasma levels could be used as a surrogate for actual activity at the receptor level once the core pharmacology of the substance had been mapped out and minimum effective concentrations established. Also accepted was the premise that surrogate blood levels could be used to extrapolate to situations when direct pharmacological effects were not easily measured or took a long time to become apparent. More sophisticated mathematical modeling of response curves and pharmacokinetic profiles expanded the understanding of the relationships between systemic exposure and pharmacologic effects. This brought about justification of the argument that if blood level profiles for two administrations differed by no more than±20%, they would be indistinguishable from one another with respect to both efficacy and safety. Further refinement of statistical tests eventually led to the establishment of the current equivalence boundaries of 80%–125% applied to blood/plasma AUC and C_max.

Further Expansion of the BE Risk Management Toolbox

The Biopharmaceutics Classification System (BCS) was first described in 1995⁽⁸⁾ and helped explain the relationships between solubility and permeability as it relates to drug absorption from the gastrointestinal tract. It allowed more precise estimation of which drug substances or products presented higher risks for absorption variability or failure due to dissolution rate limitations. It also improved the ability to identify situations where the likelihood of developing predictive in vitro–in vivo correlations (IVIVC) was higher. Finally, the BCS permitted better identification of situations where GI absorption was primarily driven by uncontrollable physiological factors, such as pH, motility, and transit times, from those situations where controllable physical factors like dissolution rate took over and became interferences.

Also by this point in the mid-1990s, the number of in vitro dissolution models had expanded with seven different apparati recognized in the pharmacopeia. Three different levels of IVIVC's were acknowledged. Where a predictive one-to-one correlation could be established for a particular drug product, bioequivalence trials could be eliminated (so-called biowaiver) in some circumstances. The three levels of IVIVC were crucial elements of the Scale-up and Post Approval Changes (SUPAC) guidance⁽⁹⁾ that defined ranges for inconsequential changes in formulation and manufacturing equipment, thus allowing a reduced regulatory filing burden.

Last, accumulated empirical evidence showed that 3 to 6 months of accelerated (40°C/75% RH) stability data comfortably assured a practical shelf-life of 24 months at reference room temperature storage. This is important in the bioequivalence discussion since ultimately product performance is not a one-time issue but something that must be assured consistently for any product batch until the assigned expiry is reached.

Extension of the Oral BE Toolbox

Anecdotal reports of differences in the degree of local skin vasoconstriction following application of topical corticosteroids from different manufacturers also began to appear and draw concern, although the therapeutic risks are somewhat lower for this class of products.⁽¹⁰⁾ This concern spawned research into different techniques for quantitating the degree of blanching (whitening) as a surrogate for topical anti-inflammatory response. This was necessary because the amount of steroid ultimately making its way into the systemic circulation was incredibly small and could not be reliably quantitated. Using different forms of visual imaging, a kinetic profile reflecting the extent and duration of steroid transport into the skin could thus be mapped to which statistical metrics analogous to those used for oral BE could be applied.⁽¹¹⁾

One can also see the extension of the risk management toolbox to the BE assessment of locally acting topical nasal spray products. With the common opinion that this class of products poses limited risks with respect to therapeutic failure or excessive systemic exposure, extensive physicochemical and in vitro performance characterizations were put forward as appropriate surrogates for establishing BE.⁽¹²⁾

Current Oral BE Risk Management Tools

The elements of our contemporary BE risk management toolbox are shown in Table 1. These elements to one degree or another have been fairly uniformly codified worldwide into guidances and regulations that in principle assure that drug products with the same physical and chemical composition will produce comparable therapeutic effects. In other words, once judged equivalent, the substitution of a generic product for its innovator reference or a generic for any other generic is unrestricted and trusted.

It is important to note that reaching this state of risk management took almost 40 years if we use Dr. Levy's 1960 experiments as the starting point and culminate (in this author's view) with the 1999 issuance of the SUPAC Equipment Addendum⁽¹³⁾ by FDA. Refinement of the toolbox continues even today, but the framework is for all practical purposes complete.

How Does the BE Risk Management Toolbox Look when Applied to Locally Acting Inhalation Drug Products?

With such a well-equipped toolbox and effective extension to other routes of administration, why is it that the topic of bioequivalence for OIP's continues to be a source of debate and regulatory disharmony? A partial answer lies in a comparative assessment (Fig. 1) of the level of scientific understanding for each of the BE risk management components when applied to pressurized metered dose inhalers (pMDI) and dry powder inhalers (DPI) versus the reference case of oral dosage forms.

OPEN IN VIEWER

FIG. 1.

Comparison of the levels of scientific understanding of current oral BE risk management tools against the current status for orally inhaled locally acting drug products.

Each risk management component has been graphically categorized to reflect one of three states:

1. scientific understanding and tools have been developed/validated to a relatively high degree (green);

Because there can be multiple issues or factors connected with each of the broader risk management components, the status assessment has been refined in a few instances to reflect that some aspects may be better understood than others at this time. The list has also been expanded to include a new risk factor “Device-User Interface.” This is a special feature of inhalation drug products that is generally not of concern with their orally administered counterparts. While Figure 1 is a qualitative ranking based on the author's personal evaluation of current scientific literature, summaries of key workshops in recent years,^(14–16) and accumulated experience, it clearly shows that there are key elements in the toolbox where the science is still incomplete. The incomplete scientific picture means that there are unresolved risks and therefore uncertainty when it comes to judging equivalence between two products.

Dealing with Unresolved Risks

Different approaches to dealing with these unresolved risks have evolved in the US and in the European Union, which seems somewhat surprising since both regulatory groups are viewing the same scientific and clinical literature. The two systems are compared schematically in Figure 2. Both systems share similar core features, namely, in vitro evaluations, systemic PK comparison, and pharmacodynamic comparisons.

OPEN IN VIEWER

FIG. 2.

Comparison of FDA and EMA approaches in the regulatory assessment of bioequivalence for orally inhaled locally acting drug products.

They differ in the construction of the decision pathways. The current FDA view represents a very conservative approach in that it requires simultaneously in vitro similarity, systemic PK similarity, and pharmacodynamic similarity in order to obtain marketing approval. In the EU, the risk-benefit ratios have been weighted differently, leading to a three-tier decision pathway. The first requirement is a demonstration of in vitro similarity focusing heavily on analysis of aerodynamic particle size distribution (APSD). The requirements for demonstration of similarity are admittedly stringent, but in the EU the possibility exists to obtain Marketing Authorization for a generic product without having to perform any in vivo studies. If in vitro similarity cannot be shown, the next step is to demonstrate systemic pharmacokinetic similarity following the more or less standard criteria for oral bioequivalence. In the last tier, in the event that pharmacokinetic similarity cannot be shown, then pharmacodynamic studies are invoked to prove therapeutic equivalence principally using measures of lung function. In the EU, if in vitro similarity cannot be demonstrated, applicants have two other opportunities to demonstrate therapeutic equivalence. While EMA allows for approval based on one of three criteria (or a combination), the FDA requires all three to be proven.

The Device–User Interface

This is probably the least understood risk area but is critical if we are to realize the goal of unrestricted substitutability of multisource inhalation products. Both FDA and EMA guidances are relatively silent on this aspect. As stated previously, this interface is a nearly nonexistent risk factor for orally administered solid dosage forms with the possible exception of packaging design as it affects ease of access to the medication. When it comes to pMDI's, there are some well-recognized flaws concerning the coordination of breathing with actuation of the device, which seem to be offset by the ubiquitous presence of these pressurized aerosol products that fundamentally do not differ very much in their design and use. DPI's present a more serious situation because of the wide variety of designs that are available, overlaid with the fact that the user supplies all of the energy necessary for dose delivery, powder de-agglomeration/de-aggregation, and transport of the aerosolized particles into the pulmonary airways. At the moment there are no standardized protocols to compare usage errors for two different devices, although recently FDA has introduced some qualitative similarity criteria for fluticasone/salmeterol DPI's.⁽¹⁷⁾

Nonscientific Influence Factors

The final implementation of the BE toolbox that brought about the trusted interchangeability of equivalent products was not realized entirely through the application of scientific logic. The need to reduce the cost of access to medications whose patents had expired has been a constant driver that prompted, at times, legislative intervention. In the US, there were two significant events in this arena that can be cited. The first of these was the 1974 report from the Drug Bioequivalence Study Panel to the Congressional Office of Technology Assessment.⁽¹⁸⁾ This panel of medical and pharmaceutical experts was convened to try to understand why many of the scientifically identified risks to bioequivalence had not been incorporated into federal regulations or compendial standards. The recommendations of this panel led to significant increases in FDA authority to request data and information on bioequivalence from sponsors. It also prompted a major revision in regulations governing drug substance and drug product manufacturers that became the framework for current Good Manufacturing Practices (cGMP's). The second significant event was the passage of the 1984 Drug Price Competition and Patent Term Restoration Act, also known as the Hatch–Waxman Act. This legislation acknowledged that the BE risk management tools had reached a suitable level of development and that they could be reliably used as a basis for approval of generic products without resorting to extensive clinical trials. This legislation was a primary enabler leading to the formation of the modern generic pharmaceutical industry.

Closing the Gaps

In the arena of pharmaceutical inhalation products, increasing health care costs continue to add political pressure to increase the availability of quality generic products or allow innovators to introduce improvements and cost-saving changes more readily. As Figure 1 illustrates, there are some scientific and informational gaps that still need to be closed. Some of these gaps have persisted for quite a while, possibly due to the fact that inhalation products are a niche area compared to oral dosage forms, which translates to more restricted availability of research funding. The situation may change slightly as a result of the US Generic Drug User Fee Act (GDUFA), which requires sponsorship of scientific studies that are aimed at reducing regulatory barriers and facilitating the approval of multisource products.

The difference in risk-benefit assessment between the US and EU offers another opportunity to gather critical information through post-market surveillance. Tracking adverse event reports or product complaint reports in both regions can help determine which regulatory paradigm offers the optimum balance of efficacy, safety, and accessibility.

Last, as the number of agents intended for systemic delivery using the inhalation route for administration increases, we should expect some important clues for the establishment of IVIVC's to come from this arena.