Impact of Mechanistic Modeling and Simulation Methodologies on Product‐Specific Guidance Development for Non‐Orally Administered Drug Products

ABSTRACT

The U.S. Food and Drug Administration (FDA) publishes product‐specific guidances (PSGs), with bioequivalence (BE) recommendations for prospective generics. Developing BE recommendations for non‐orally administered drug products including long‐acting injectables (LAI), orally inhaled drug products, and drugs applied locally to the skin, ophthalmic, and nasal routes may be challenging because conventional BE methods and considerations for orally administered drug products may not apply. Mechanistic modeling approaches such as physiologically based pharmacokinetic (PBPK) models or computational fluid dynamics (CFD) models are used for the development of BE methods and BE assessment standards in PSGs for non‐orally administered drug products, as evidenced by cases provided here. This manuscript discusses in silico methodologies that support PSG development for non‐orally administered drug products through the model‐integrated evidence paradigm. Specifically for orally inhaled drug products, we highlight modeling and simulation (M&S) recommendations that have occurred in one new PSG that was published for formoterol fumarate; glycopyrrolate inhalation metered aerosol in February 2024 and referred to in 16 other PSGs with different active pharmaceutical ingredients and device types, such that M&S approaches may be optionally used to provide insight on regional drug delivery and support BE assessments. For drug products applied to the skin, we focus on three successful case studies where PBPK models were used to support revised PSG recommendations with reduced need for ex vivo and human studies. Ophthalmic products, LAI, and nasal products illustrated advancements in M&S methodologies that may be potentially used for the development of new or revised PSG recommendations.

1. Introduction

In its 2023 report, the Association for Accessible Medicines estimated that generic drug products provided $398.6 billion in savings in 2023 for the United States and accounted for 91% of dispensed prescriptions, but only 17.5% of spending on prescription drugs while brand name drug products accounted for the bulk of prescription spending [1]. In light of these clear benefits, the U.S. Food and Drug Administration (FDA) is committed to the approval of safe and effective generic drug products. A generic formulation (test) is the product that is compared to a reference listed drug (RLD) or reference standard (RS). The evidence for approval of the generic formulation is bioequivalence (BE), as defined in 21 CFR § 314.3(b) as “the absence of a significant difference in the rate and extent to which the active ingredient or active moiety in pharmaceutical equivalents or pharmaceutical alternatives becomes available at the site of drug action when administered at the same molar dose under similar conditions in an appropriately designed study” [2]. However, the regulations state that “[f]or drug products that are not intended to be absorbed into the bloodstream, bioequivalence may be demonstrated by scientifically valid methods that are expected to detect a significant difference between the drug and the listed drug in safety and therapeutic effect” [3].

The FDA publishes product‐specific guidances (PSGs) that describe the FDA's current thinking on the evidence needed to demonstrate that a generic drug product is bioequivalent to a specific RLD/RS product [4, 5]. PSGs provide recommendations that are meant to assist generic drug developers with identifying the most appropriate methodology for their generic drug development program. To prioritize PSG development, the FDA considers products without published PSGs, pre‐abbreviated new drug application (ANDA) meeting requests, public requests, comments at the PSG docket, controlled correspondences, and citizen petitions. Generic Drug User Fee Amendments (GDUFA) commitments, ANDA submissions, drug availability and access, and public health priorities may reshape the criteria for PSG prioritization [6, 7, 8]. During PSG development of new and revised PSGs, the FDA considers new drug application (NDA) assessments and labeling, previous BE studies and institutional knowledge, pharmacovigilance data, GDUFA‐funded research, and quantitative modeling and simulation approaches [5, 7, 8]. As a result, BE recommendations in PSGs may include in vitro characterization studies, BE studies with PK endpoints, and comparative clinical endpoint BE studies. The FDA is continuously searching for methods to improve BE assessment of generic drugs by supporting PSG development and efficient approaches for establishing BE.

The FDA recognizes the role of quantitative methods and modeling (QMM) in supporting generic drug development and drug product approval through the model‐integrated evidence (MIE) paradigm. Figure 1 illustrates the integral connections within the MIE paradigm between submitted ANDAs from the generic drug industry, FDA general guidances and PSGs, and the GDUFA regulatory science program. MIE refers to using model‐generated information not only to plan a pivotal study, but to serve as evidence for generic drug product approval [9, 10]. MIE approaches make use of virtual bioequivalence (VBE) assessments supporting a pathway for generic drug approval, through in vitro product characterization, and by reducing the need for the recommended in vivo pivotal studies which, depending on the drug product, may include PK, pharmacodynamics (PD), and/or comparative clinical endpoint studies [9]. The generic drug industry has used modeling and simulation approaches for addressing regulatory issues related to the establishment of BE to occasionally support novel alternative approaches to the recommendations provided in PSGs or complex issues that occur when the recommended BE studies in PSG are conducted [11, 12, 13, 14, 15, 16, 17]. These quantitative methods and modeling approaches include physiologically based pharmacokinetic (PBPK) modeling, computational fluid dynamics (CFD) modeling, population pharmacokinetics (PopPK) modeling, and advanced data analytics methodologies [11, 18, 19]. Applications for MIE, often developed within the scope of the GDUFA regulatory science program, aim to reduce unnecessary human testing and to develop more efficient BE methodologies [9, 12, 13, 19, 20, 21, 22, 23]. As the regulatory acceptance of MIE is increasing, mechanistic QMM are employed to support the development of new and revised PSGs that aim to provide viable options for BE for non‐orally administered products.

FIGURE 1.

Model‐integrated evidence (MIE) paradigm supports ANDA approvals and FDA guidance development and is enhanced by the GDUFA regulatory science program. MIE approaches may be utilized in different stages of ANDA lifecycle such as pre‐ANDA development, ANDA assessment and approval, development of PSGs, and general guidances to industry. The GDUFA‐funded regulatory science research program fosters the development of in silico tools and methodologies that are implemented under the MIE paradigm.

Non‐orally administered drug products include locally acting drug products, such as drugs applied to the skin, ophthalmic, nasal, and orally inhaled drug products, and long‐acting injectable (LAI) and implantable drug products which may act locally or systemically. Locally acting drug products are meant to deliver the active pharmaceutical ingredient (API) at or near the site of action and are not necessarily intended for API delivery into the bloodstream. Establishing BE for these products may be challenging because measuring the drug amount at the local site of action may not be feasible or ethical, and in addition, systemic drug concentrations, when available, may not reflect the drug concentration at the site of action [9, 10]. Among non‐orally administered drug products, LAI products are often administered as a subcutaneous (SC) or intramuscular (IM) injection and provide sustained drug release for a period of time ranging from weeks to months [22]. Generic LAIs face challenges with establishing BE against their RLD/RS because of the long duration of in vivo BE studies, low subject recruitment, high dropout rate, and increased variability with pharmacokinetic (PK) metrics. As a result, there is low availability of generic LAIs in the U.S. market today.

This manuscript summarizes modeling and simulation applications that supported new and revised PSGs for non‐orally administered drug products, namely, orally inhaled drug products and drug products applied to the skin. Additionally, ongoing research with the potential to assist with PSG development for ophthalmic, LAI, and finally, nasal drug products is highlighted here. GDUFA‐funded research to date is discussed to provide context for continued PSG development. The manuscript concludes by discussing the overall impact of modeling and simulation on regulatory decision‐making for generic drugs through PSG development.

2. Orally Inhaled Drug Products

Development of modeling and simulation approaches for orally inhaled drug products has been aided by external research funding provided by the FDA (Additional information on the awarded grants and contracts for orally inhaled drug products is available in Table S1). The outcomes of these research activities, along with other research available in the literature, ongoing internal research activities at the FDA, and the outcomes of a workshop co‐hosted by the Center for Research on Complex Generics (CRCG) and the FDA in April 2023 [24] have directly impacted PSG development for orally inhaled drug products, as illustrated in Figure 2, via the inclusion of specific modeling and simulation recommendations in several current PSGs that are listed in Table 1. For example, in a new PSG that was published for formoterol fumarate; glycopyrrolate inhalation metered aerosol in February 2024 [25], the recommendations include the option for utilizing modeling and simulation approaches to support establishing BE between a generic MDI and its RS, and other new or revised PSGs posted since February 2024 also refer to the detailed language provided in the PSG for formoterol fumarate; glycopyrrolate inhalation metered aerosol [26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41]. The recommendations include language specific to semi‐empirical and CFD regional deposition modeling as well as PBPK modeling. Semi‐empirical or CFD regional deposition models may be used to provide predicted values of total lung deposition as well as central and peripheral lung deposition, while PBPK models may be used to quantify lung absorption for active ingredients that are expected to dissolve and/or permeate slowly. Together, these models can integrate in vitro and in vivo evidence to predict drug delivery directly to the site of action, which provides a tool that can be used to assess the contributions of individual in vitro and in vivo studies to our understanding of BE. To provide further information on the implementation and usefulness of these models, the recommendations include details on model purpose, model credibility, and acceptance criteria for model validation and statistical analyses.

FIGURE 2.

Flowchart that describes process for including modeling and simulation language in products specific guidances (PSGs) for orally inhaled drug products.

TABLE 1.

Published PSGs for generic orally inhaled drug products that include recommendations on the use of MIE to support determination of BE. The first publication date for PSGs is noted as “Recommendation/Recommended” followed by revision dates, noted as “Revision/Revised”, when applicable [5].

Active ingredient(s)	Route of administration	Dosage form	RLD/RS number	Recommendation and revision dates
Formoterol fumarate; glycopyrrolate	Inhalation	Aerosol, Metered	NDA 208294	Recommended February 2024
Mannitol	Inhalation	Powder	NDA 022368	Recommended February 2024
Mannitol	Inhalation	Powder	NDA 202049	Recommended February 2024
Zanamivir	Inhalation	Powder	NDA 021036	Recommended February 2024
Budesonide; formoterol fumarate; glycopyrrolate	Inhalation	Aerosol, Metered	NDA 212122	Recommended February 2024
Albuterol sulfate	Inhalation	Powder	NDA 020503, NDA 020983, NDA 021457	Recommended April 2013; Revised June 2013; December 2016; March 2020; November 2023; August 2024
Fluticasone propionate	Inhalation	Aerosol, Metered	NDA 021433	Recommended October 2017; Revised July 2018; November 2023; August 2024
Fluticasone propionate; salmeterol xinafoate	Inhalation	Aerosol, Metered	NDA 021254	Recommended May 2019; Revised June 2020; November 2023; August 2024
Fluticasone propionate	Inhalation	Powder	NDA 020833	Recommended October 2017; Revised February 2024; August 2024; November 2024
Fluticasone propionate; salmeterol xinafoate	Inhalation	Powder	NDA 021077	Recommended September 2013; Revised February 2024; August 2024; November 2024
Formoterol fumarate	Inhalation	Powder	NDA 020831	Recommended September 2015; Revised May 2023; November 2024
Formoterol fumarate; mometasone furoate	Inhalation	Aerosol, Metered	NDA 022518	Recommended January 2016; Revised May 2023; November 2024
Mometasone furoate	Inhalation	Aerosol, Metered	NDA 205641	Recommended April 2016; Revised November 2023; November 2024
Salmeterol xinafoate	Inhalation	Powder	NDA 020692	Recommended October 2017; Revised February 2024; November 2024
Budesonide; formoterol fumarate	Inhalation	Aerosol, Metered	NDA 021929	Recommended June 2015; Revised November 2023; November 2024
Tiotropium bromide	Inhalation	Powder	NDA 021395	Recommended October 2017; Revised November 2023; November 2024
Treprostinil	Inhalation	Powder	NDA 214324	Recommended November 2024

Abbreviation: NDA, new drug application.

With respect to model purpose, language in PSGs recommends that holistic modeling and simulation approaches may support either the establishment of biorelevant BE limits for recommended in vitro and/or in vivo studies, or the conduct of VBE trials to serve as an additional assurance of BE besides other recommended in vitro and in vivo studies. For both in vitro studies (e.g., aerodynamic particle size distribution, plume geometry) and in vivo PK studies, these models could aid in establishing biorelevant BE limits by correlating the metric(s) of interest with regional lung delivery. If a generic developer was to utilize a BE approach that did not include the FDA's recommended BE studies, VBE trials could be a way to further support the establishment of BE between a test product and its RS, which may alleviate the need for additional in vitro and/or in vivo studies. For either of these model purposes, appropriate model credibility exercises are needed, which may be executed using a risk‐based modeling approach such as the American Society of Mechanical Engineers (ASME) Verification & Validation (V&V) 40 standard [42]. Altogether, the use of credible modeling and simulation approaches is expected to facilitate faster approval of generic orally inhaled drug products by either establishing biorelevant BE limits or by reducing the number of in vivo BE studies.

3. Drug Products Applied to the Skin

Topical dermatological drug products are products applied on the outer skin surface to treat dermatological diseases. A major challenge faced by developers of generic versions of these products relates to resource‐intensive comparative clinical endpoint BE studies performed between the RS and a prospective generic [43]. To improve patient accessibility to generic topicals, the FDA is developing efficient in vitro characterization approaches for establishing BE for certain topical drug products [43]. These approaches may be coupled with mechanistic modeling and simulation applications under the MIE paradigm to facilitate the development of complex topical drug products and decrease uninformative in vivo human testing [9, 14, 15, 16].

Through the GDUFA regulatory science program, the FDA has funded the advancement of mechanistic in silico models, such as PBPK and CFD models, for generic drug products that are applied on the skin (Table S2). These mechanistic models provide a mechanistic link between drug product quality attributes, characterized in vitro, and the in vivo performance of the drug product. As such, they can be successfully applied to support drug product development and regulatory approval by informing BE recommendations in draft PSGs. Additionally, they can inform decisions on lifecycle management for drug products by supporting risk‐based BE assessments and by identifying the “safe space” for drug product quality attributes impacting product in vivo performance [14, 16, 44, 45, 46]. Three noteworthy case studies outlining the applications of PBPK models toward PSG development of topical dermatological drug products are presented below.

The PSGs for ketoconazole cream, 1% [47] and clindamycin phosphate lotion, equivalent 1% base [48] were revised in 2022 and 2023, respectively, to provide the FDA's most current thinking with regard to in vitro characterization approaches available within these PSGs. In the case of ketoconazole cream, 1%, in silico modeling provided insights on the ketoconazole skin permeation profile. Model predictions of limited skin permeation for ketoconazole aligned well with the primary site of action being the skin surface for this product and the revised BE recommendations [47]. The model developed for clindamycin phosphate lotion mechanistically described the API distribution between the two product phases and improved the FDA's understanding of the formulation complexities for this locally acting drug product [49]. These findings reinforced the FDA's thinking on the appropriateness of the current BE recommendations for this product.

The PSGs for dapsone topical gels were revised in February 2024 [50, 51, 52], to remove in vitro permeation test (IVPT) and in vivo PK BE studies under Option 1. Currently, Option 1 is for a prospective generic that has no difference relative to the RS in inactive ingredient components or composition, or in other aspects of the drug product (e.g., physicochemical and structural [Q3] attributes [53]) that may significantly affect the local or systemic bioavailability (BA). Under this Option, the described characterization‐based BE approach involves an in vitro release test (IVRT) BE study and other product characterization tests. Option 1 no longer includes an IVPT BE study or a BE study with PK endpoints. Option 2 includes a comparative clinical endpoint BE study in the revised PSG. The PSG revision was supported by mechanistic dermal PBPK models that characterized the link between Q3 attributes of the single phase dapsone gels 7.5% and 5%, where dapsone is partially dispersed, and the product in vivo performance [46]. These dermal PBPK models were developed using drug product Q3 attributes such as formulation pH, apparent viscosity, and particle size distribution and validated against in vivo PK data from approved ANDAs. Sensitivity analyses showed that apparent viscosity and particle size distribution for the undissolved dapsone may impact product in vivo performance. The application of these models toward VBE assessments in skin and plasma demonstrated that for the studied ANDAs when the prospective generic product and the RS are Q3 the same, especially with respect to apparent viscosity and particle size distribution, they are found to be bioequivalent in the studied matrices, in accordance with the outcome of the in vivo BE study with PK endpoints for the studied ANDAs. The model‐based approach supported the conclusion that direct assessments of local and systemic BA from IVPT and in vivo PK BE studies within the scope of BE recommendations are not necessary for generic dapsone topical gels, 7.5% and 5% that meet the criteria for no significant difference in components or composition with the RS as captured in the PSG revision for these products.

4. Ophthalmic Drug Products

Ophthalmic drug products are sterile products intended for drug delivery to the eye to treat various ocular diseases such as glaucoma, inflammation, and infection in human patients. The development of generic complex ophthalmic drug products is challenging due to the complexity of the human ocular system and a lack of sensitive testing methods to evaluate the complex interplay of human physiology and ophthalmic drug products. Ophthalmic drugs are locally acting, giving rise to challenges in measuring detectable local drug levels, as measurements in human eye tissues are often impractical, unethical, and cost‐prohibitive. To tackle these challenges, the FDA has sponsored external scientific research to advance generic ophthalmic drug development through the GDUFA regulatory science program, including mechanistic modeling focused grants and contracts (Table S3) [17].

One focus of model development in this area has been to mechanistically understand the complex ocular system and its interplay with complex ophthalmic formulations. Related areas of scientific interest and noteworthy case studies are briefly highlighted below.

The development of a combined CFD and PBPK approach that integrates in vitro dissolution data and fluid dynamic kinetics in the eye is one area of research focus. A CFD tear film model was developed to simulate the dynamic response of the tear film volume to the sudden addition of an eyedrop to the eye surface. The model also addresses how the volume is restored through tear secretion by the lacrimal gland and drainage via the superior and inferior lacrimal puncta [54]. This CFD‐PBPK framework was implemented and validated based on in vivo/in vitro data, and it supported the mechanistic modeling of drug retention time on the eye surface and resulting ocular exposure after topical administration of ophthalmic formulations.

Another area of research was to develop and validate various ocular PBPK models for multiple dosage forms and formulations including ophthalmic solutions [55], suspensions [56], and ointments [57]. These models account for information on formulation critical quality attributes (CQAs), such as particle size distribution and viscosity, spreadability, viscoelastic behavior, and drug release performance. These are powerful tools used to predict the impact of formulation deviations on the in vivo performance of complex ophthalmic products. By understanding and implementing the differences in anatomy and physiology between preclinical species and humans in PBPK models, preclinical data were leveraged to successfully predict ocular drug exposure in humans for ophthalmic solutions [55] and suspensions [58] based on interspecies extrapolations.

Within the scope of supporting BE assessments, a mechanistic modeling approach for a generic ophthalmic product may integrate evidence specific to a drug product, including data obtained from in vitro, ex vivo, and in vivo studies. Validated ocular mechanistic models may be utilized to execute in silico VBE analyses and risk‐based assessments for the proposed generic products. Ultimately, the MIE approaches may complement the BE recommendations outlined in draft PSGs for generic ophthalmic drug products, particularly in situations that contain lengthy or insufficiently sensitive comparative clinical endpoint BE studies.

5. Long‐Acting Injectable (LAI) Drug Products

The development of generic LAI drugs is often difficult due to their complex formulations, the limited knowledge about how critical drug product attributes influence the in vivo release and behavior of the drug product, and the challenges associated with conducting in vivo BE studies. To promote the development of generic LAIs, the FDA has funded research projects aiming to develop mechanistic PBPK models of LAI crystalline suspension drug products and polymer‐based implants (Table S4). The primary objective of developing LAI PBPK models is to mechanistically describe the in vitro and in vivo drug release mechanism. To achieve this goal, comprehensive in vitro characterization of LAI drug products has been conducted through these research projects to identify the product CQAs that can impact the product performance. The knowledge leveraged from in vitro characterization studies is then accounted for during PBPK model development to delineate the influence of product CQAs on the in vivo drug release mechanism. These projects also aim to describe the intricate relationships between product properties and injection site physiology through PBPK modeling, as such interplay may substantially influence the in vivo drug release mechanism. Such understanding is expected to improve our ability for regulatory decision‐making and support PSG recommendations on comparative assessment of product CQAs.

By incorporating key physiological parameters such as local blood flow rate, tissue composition, pH, and enzymatic activity, PBPK models can mechanistically illustrate how a drug product behaves once injected, highlighting factors that govern its rate and extent of release. This is especially relevant for complex injectable products, where release profiles may be impacted by multiple concurrent processes, including dissolution and diffusion, all of which can vary depending on the specific administration route (e.g., subcutaneous vs. intramuscular) and patient‐specific attributes. Pertaining to LAI suspension drug products, particle size distribution remains one of the CQAs that may substantially influence the in vivo drug release mechanism by different means. A published PBPK model of Depo‐SubQ Provera 104, a medroxyprogesterone acetate (MPA) subcutaneous injectable suspension, suggested that the effective in vivo particle size may be potentially larger than in vitro measured particle size as a result of particle aggregation at the injection site. Simulation work also suggested that local tissue inflammation is likely to cause a transient increase in depot volume, which has substantial influence in the disposition of MPA SC injection. Based on further investigations to evaluate the impact of the immune cell layer (ICL), which is formed due to tissue reaction to LAI crystalline particles, the previously implemented mathematical model describing the temporal changes in the ICL thickness was improved in the LAI PBPK model structure [22].

External research on the mechanistic modeling of generic LAIs provides scientific insights regarding the impact of Q3 attributes of LAI drug products on the in vivo drug release by accounting for product‐physiology interplay in the model. Such understanding may be leveraged to refine or support PSG's recommendations of LAI drug products. Additionally, mechanistic modeling of LAIs may support the development of efficient BE approaches. As assessing the impact of CQA difference (i.e., difference in Q3 values) on the systemic and local BA remains one of the strong traits of mechanistic PBPK model of LAIs, such a model may be used to justify the inclusion of partial AUC (partial area under the plasma concentration vs. time curve or pAUC) or recommended study types (i.e., single vs. multidose, parallel vs. crossover BE study) in BE recommendations for LAIs. These insights help streamline the development of generic products, as modeling‐based assessments can guide applicants in formulating or modifying designs to better match the RS product's in vivo performance before conducting in‐depth clinical studies. This holistic approach not only accelerates product development timelines but also reinforces patient safety by ensuring consistent therapeutic outcomes across different drug products.

6. Nasal Drug Products

Nasal drug products target topical delivery of administered droplets or particles to the nasal mucosa, which are then absorbed and can exert local or systemic therapeutic effects. For locally acting drug products, such as glucocorticoids used to treat allergic rhinitis and nasal congestion, the site of action is considered to reside in the nasal cavity. From a generic drug product perspective, demonstrating BE for solution‐based nasal spray drug products intended for local drug delivery may be considered relatively straightforward because demonstration of formulation sameness, device comparability, and a series of in vitro tests is expected. However, for locally acting suspension‐based nasal spray drug products, one of the options for establishing BE includes in vivo studies along with the in vitro studies recommended for solution‐based products [59, 60]. These additional in vivo studies can present challenges for generic drug developers, and an improved understanding of the critical factors for drug delivery for locally acting suspension‐based nasal drug products can aid in the development of more efficient BE methods.

To address these challenges, the FDA has sponsored external scientific research projects through the GDUFA regulatory science program (Table S5). A key research focus has been to develop in silico modeling tools for suspension‐based nasal spray drug products. These efforts involve in vitro, in vivo, and in silico studies, including CFD and PBPK modeling studies, focused on the influence of device performance and formulation differences on nasal regional deposition and prediction of mucociliary clearance. Outcomes of these projects include development of a CFD modeling approach that included three‐dimensional deposition predictions and a quasi‐two‐dimensional mucus layer dissolution, absorption, and clearance model [61, 62, 63], CFD modeling to investigate the impact of variations in spray characteristics and nasal anatomy on regional nasal deposition [64], and a CFD‐PK modeling approach to predict PK following administration of triamcinolone acetonide. Additionally, research has focused on improving our understanding of the emerging class of nasal drug products that may target nose‐to‐brain drug delivery, where a CFD model has been used to predict olfactory region absorption [65] and a nose‐to‐brain PBPK model is in development.

The CFD‐PBPK models for nasal drug products are valuable tools for answering research and regulatory questions because they provide the capability to correlate in vitro performance metrics with in vivo performance. From the generic drug development perspective, the models provide useful tools that may be incorporated in MIE approaches used to evaluate the nasal spray performance based on in vitro data collected early in the development phase, as well as to evaluate different testing scenarios prior to the nasal spray undergoing in vivo testing. Taken together, these models have the potential to support FDA's ongoing efforts to refine PSG recommendations and ensure the most sensitive and efficient BE approaches are utilized.

7. Conclusions

For non‐orally administered drug products, PSGs are often challenging to develop because drug delivery to the site of action may not be easily quantified or due to other challenges associated with conducting in vivo BE studies. Mechanistic modeling and simulation approaches are currently facilitating generic drug development through the MIE paradigm. Modeling and simulation outcomes from GDUFA‐funded research studies, as well as related outcomes supporting the development of general guidance and PSGs, are provided in the annual research reports published each fiscal year [66]. As supported by GDUFA‐funded research, internal FDA efforts, and industry–academia–regulatory agencies interactions during FDA public workshops, mechanistic in silico models, such as PBPK and CFD models, have contributed to PSG development and revision for non‐orally administered drug products [24, 67, 68]. These impactful and successful cases were discussed here, highlighting the beneficial role these innovative in silico methodologies can play in PSG development. As the regulatory use of mechanistic modeling is increasing and credible in silico tools are developed, it is expected that additional PSGs will be impacted in the future.

For the orally inhaled drug products area, FDA's research activities have positively impacted PSG development through the inclusion of specific modeling and simulation recommendations in several current PSGs, where these recommendations are expected to facilitate approvals via either establishment of biorelevant BE limits for the recommended in vitro and in vivo studies or through the conduct of VBE trials to provide additional assurance in concert with in vitro and in vivo studies. VBE trials may support the establishment of BE between a test and RS product and minimize the burden of human testing such as a PK BE study or a comparative clinical endpoint BE study. In the area of drug products applied to the skin, three case studies outlined the applications of PBPK models toward PSG revisions that removed IVPT and in vivo BE studies with PK endpoints from the BE recommendations. Overall, these PSG revisions have the potential to lead to cost savings for generic developers and limit unnecessary human testing.

The ophthalmic products area illustrated the use of CFD simulations and PBPK modeling of the eye with a focus on the effect of drug product quality attributes on in vivo exposure and BA. Human ocular PBPK models of ophthalmic solutions and suspensions have been successfully extrapolated from the preclinical species. These ocular PBPK models may support regulatory decision‐making and provide evidence for potentially refining PSG recommendations. For the area of LAIs, PBPK modeling has been utilized to establish mechanistic IVIVC/IVIVR (in vitro‐in vivo correlation/relationship) by incorporating formulation CQAs into PBPK models. A PBPK model of LAI drug products may be potentially utilized for developing/revising PSG to support in vitro characterization based BE approach. For nasal products, two hybrid CFD‐PBPK models have been developed for suspension‐based nasal sprays, where drug regional deposition was predicted by CFD models and local tissue concentration was predicted by PBPK models. These nasal models are expected to be useful tools for regulatory decision‐making purposes and could provide evidence toward updating PSG recommendations for in vitro studies.

To further tackle issues related to the development of BE recommendations for generic versions of non‐orally administered drug products, refinement of the current in silico tools may be necessary to adopt drug product specific aspects and patient‐related conditions that impact model‐based predictions on the in vivo performance of drug products under real‐life scenarios. Enrichment of currently considered in silico tools that include mechanistic PBPK and CFD models and their combinations with sophisticated machine learning and advanced data analytics methodologies may increase the efficiency of these combined in silico tools when applied for decision making during generic development and virtual BE assessments. Finally, leveraging real‐world data for model development and validation could improve the robustness of model informed BE recommendations. In general, improving standardization and harmonization across regulatory agencies of guidelines on establishing credibility of modeling and simulation approaches could increase acceptability for these methodologies and tools for BE recommendations in PSGs and their implementation toward generic drug product development and approval.

Overall, PBPK modeling, CFD modeling, and other mechanistic in silico methodologies will continue to play an important role in the development of BE recommendations for non‐orally administered drug products. The FDA is dedicated to continued efforts integrating the MIE paradigm into the process of PSG development by leveraging the mechanistic nature of QMMs for these products. This much‐needed integration aims to offer generic applicants efficient and viable BE approaches supporting generic drug product development and regulatory approval.

Disclosure

The article reflects the views of the authors and should not be construed to represent the U.S. Food and Drug Administration's views or policies.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1

Table S2

Table S3

Table S4

Table S5

Table S6