Sage Journals: Discover world-class research

Abstract

Physiologically based pharmacokinetic (PBPK) modeling is increasingly used to anticipate, quantify, and strategically manage drug–drug (DDI) and herb–drug (HDI) interactions that can alter the exposure of chemotherapy agents together with co-administered phytochemicals or nutraceuticals. To evaluate current knowledge, we performed a comprehensive Google Scholar search (2003–2024) and selected studies that employed PBPK platforms, reported quantitative validation, and focused on chemotherapy-related interactions. From these reports, key modeling parameters, validation metrics, and clinically relevant outcomes were extracted, and then the information was synthesized to identify common trends. Collectively, the evidence indicates that unintended changes in drug exposure—most often mediated by CYP3A4 inhibition or induction—may modify efficacy, toxicity, and overall anticancer response; nevertheless, PBPK models reproduce these effects with high accuracy, and emerging AI-enhanced approaches promise even finer precision. Accordingly, our synthesis underscores how PBPK modeling can help clinicians forecast interaction risk, individualize dosing, and avert therapeutic failure, especially in polypharmacy settings. Integrating these models into routine oncology practice therefore offers a proactive path toward safer, more personalized chemotherapy and, ultimately, better patient outcomes within an increasingly complex therapeutic landscape.

Keywords

chemotherapy PBPK modeling drug–drug interactions drug–phytochemical interactions herbal medications herb–drug interactions

Introduction

Drug interactions occur when one substance affects the pharmacokinetic characteristics or pharmacological activity of another, potentially altering efficiency of drug or increasing the risk of toxicity, or even causing a loss of therapeutic effect overall, even though many of them are purposeful and are used to increase desired effects or decrease undesirable ones.^1,2

Numerous studies have combined experimental data with PBPK models to elucidate the mechanisms of drug–drug and herb–drug interactions.^3,4 Although these studies are very useful for examining particular drug interactions, a more holistic viewpoint is essential to bringing similar pathways together and creating thorough, trustworthy models.

Research on DDIs has concentrated on identifying significant interactions, classifying them, and building predictive models.⁵ Key findings emphasize the importance of pharmacokinetic and pharmacodynamic mechanisms such as changes in drug metabolism or cumulative effects, in causing adverse interactions. Digoxin, NSAIDs, warfarin, and antidepressants are among the common medications linked to major DDIs.^2,6

Unplanned DDIs are quite likely to emerge in oncology in which complex treatment strategies often involve chemotherapeutics, immunotherapeutics, targeted therapies, and supportive care medications.¹ This complex treatment landscape is further complicated by the increasing use of phytochemicals, herbal supplements, and nutraceuticals to promote tumor control, boost tolerance, or simply support quality of life, in addition to these pharmacological issues.⁷ Numerous phytochemicals/herbal medications have antiproliferative properties and may alter drug transport and metabolism, offering both opportunities and risks for patient outcomes.

The majority of pharmacokinetic DDIs arise from alterations in absorption, distribution, metabolism, or excretion (ADME).² Primary contributors include transporters like P-gp, whose modulation (via induction or inhibition) can significantly impact plasma levels of medication and its safety/efficacy profile, and cytochrome P450 (CYP) enzymes, of which CYP3A4 is the primary contributor.^1,2 For example, competition for tissue distribution sites or plasma protein binding may change distribution, whereas changes in gastrointestinal pH, motility, or transporter function may affect absorption. Drug exposure and half-life are also altered by changes in renal or biliary excretion, although metabolic interactions frequently rely on modulators of the CYP enzymes themselves.^1,8

Examples of combination therapies include medications such as anti-inflammatory analgesics or phytochemicals or nutraceuticals that can change the metabolic pathways of concurrently delivered chemotherapeutic treatments. Dietary supplements and nutraceuticals such as St. John’s Wort (a CYP3A4 inducer) can quickly lower the plasma concentrations of various anticancer medications, jeopardizing their therapeutic effects.^9-11 A further layer of uncertainty is added by polypharmacy, a condition that is prevalent, particularly with older patients and those with multiple illnesses.⁵ Complex and frequently unpredictable interaction networks result from the interference of different anticancer agents working with various supportive therapy modalities and other additional medications, chemicals, or substances.¹²

Most of the reviews in literature focus on the mechanisms of DDIs at various levels, including ADME to examine the clinical consequences of these interactions, ranging from enhanced toxicity to altered efficacy.^1,13 As also highlighted in review studies^14,15, the growing adoption of PBPK modeling, driven by advancements in computational tools and regulatory acceptance, underscores its potential to enhance drug safety and efficacy. These studies also identify challenges, including the need for more robust physiological data and the harmonization of in vitro experiments to improve predictive accuracy.

Drug interactions are especially critical in oncology due to the narrow therapeutic windows, multiple concurrent treatments, and potential for serious clinical consequences. This review discusses the pharmacokinetic principles underlying drug interactions in oncology, emphasizing PBPK modeling of drug–drug and herb–drug interactions. While recent work has often centered on specific cancer types, our primary goal is to synthesize these findings, evaluate the validity of mechanism-based models, and demonstrate how advanced PBPK tools can guide preventive strategies.

Methods

A literature search covering 2003–2024 was performed in Google Scholar with the keywords chemotherapy, PBPK modeling, drug interactions, and phytochemical interactions. Quantitative PBPK modeling studies on drug interactions using these platforms were included. The search was restricted to studies using a modeling software platform that also provided a quantitative model validation for chemotherapy drug interactions. For each eligible paper, modeling parameters, validation metrics, and clinical outcomes were abstracted to enable a structured synthesis and comparison. Figure 1 illustrates the common workflow observed across these studies, moving from experimental plasma concentration data through PBPK simulation to statistical validation. Taken together, the compiled results demonstrate the robustness and broad clinical applicability of PBPK models.

Figure 1.

General PBPK workflow for evaluating chemotherapy drug–drug and herb–drug interactions.

Drug–Drug Interactions in Oncology

Cancer patients frequently get novel treatments, which, while improving survival results, are susceptible to pharmacokinetic and pharmacodynamic interactions that might impair efficacy or increase toxicity. In those treatments, unknown DDIs present a critical challenge because of the complex pharmacological profiles of anticancer agents and their concurrent use with other medications. The increasing complexity of cancer treatments has therefore brought the management of DDIs to the forefront, significantly impacting therapeutic efficacy, patient safety, and treatment outcomes.¹

DDIs in oncology result from interactions between cancer treatment agents, adjuvant medications, and non-cancer-related pharmaceuticals. These interactions influence the processes of absorption, metabolism, and elimination, which are mostly assisted by cytochrome P450 enzymes and transport proteins such as P-glycoprotein.¹⁶ For example, powerful CYP3A4 inhibitors or inducers significantly alter the pharmacokinetic features of targeted therapeutic drugs, necessitating dose adjustments to reduce the potential for adverse consequences. Table 1 comprises the effect of DDIs on ADME.

Table 1.

Common Mechanisms Underlying Pharmacokinetic DDIs.

Process	Mechanism of DDI	Interaction Example
Absorption	Altered gastric pH, motility, transporter function	PPIs or antacids reducing oral absorption of TKIs (e.g., Dasatinib–Famotidine)²⁰
Distribution	Competition for plasma protein binding or tissue sites	Drugs displacing anticancer agents from binding sites (e.g., Warfarin displaced by NSAIDs, increasing free warfarin levels)²
Metabolism	Induction or inhibition of CYP enzymes (e.g., CYP3A4)	Rifampicin inducing CYP3A4 and reducing drug AUC (e.g., Rifampicin–Dasatinib)²⁰Ketoconazole inhibiting CYP3A4 and increasing drug AUC (e.g., Ketoconazole–Dasatinib)²⁰
Excretion	Changes in renal tubular secretion, biliary excretion, transporter-mediated clearance	Inhibitors of P-gp reducing clearance and raising plasma levels of co-medicated drugs (e.g., Verapamil inhibiting P-gp and increasing digoxin plasma levels)²¹

The complexity of DDIs during cancer treatment is underlined, especially when cytotoxic agents are being incorporated into targeted therapies.¹⁷ The study specifically focuses on dose optimization methodologies developed under the FDA’s Project Optimus initiative that combines pharmacokinetic/pharmacodynamic modeling, quantitative systems pharmacology, and translational data in improving predictions related to clinical outcomes. It further highlights the collaboration with regulatory bodies and the use of iterative modeling techniques in the optimization of dosing strategies. In conclusion, tailored approaches are needed for the unique challenges each drug combination presents, as well as for the diverse patient populations.¹⁷

The vulnerability of geriatric patients to DDIs is emphasized in consideration of physiological changes, polypharmacy, and related comorbidities.¹⁸ During polymedication in geriatric oncology, there are increased chances of adverse drug reactions, thus needing geriatric assessment in order to guide proper dose adjustment in an effort to reduce toxicity.¹⁸ A probabilistic model is developed for managing uncertain DDIs.¹⁹ The model integrated the likelihood of an interaction with the interaction’s severity and avoidability to allow the ranking of patient safety with therapeutic efficacy. Pascal’s Uncertainty Principle underpinned the described approach to enable sound decision-making within clinical oncology.¹⁹

Natural products also present some unique challenges regarding drug interactions. Phytochemicals in products such as grapefruit juice and St. John’s Wort may act as inhibitors or inducers of drug-metabolizing enzymes, which could result in subtherapeutic or toxic levels of anticancer agents.^22,23 Hence, it emphasizes the importance of patient education and strict monitoring while using alternative medications alongside cancer treatments.

Physiologically Based Pharmacokinetic (PBPK) Modeling Strategies

Interest in PBPK modeling has recently increased, suggesting a closer link and predictive power with oncology drugs. The most interesting aspect of software such as PK-Sim and Simcyp is the ease with which in vitro and in vivo data integrate, providing simulations of the main pharmacokinetic processes of ADME through various physiological and pathophysiological states.¹² By modeling complex mechanisms such as enzyme inhibition, induction, competitive binding, and transporter modulation, it is possible to predict changes in drug exposure within clinical scenarios that have not been tested. This allows the clinician and researcher to anticipate potential DDIs, enlighten clinical trial design, and optimize therapeutic strategy.^9,24-26

PBPK model development and validation typically follow four key stages: (1) data preparation, (2) model development, (3) calibration, and (4) validation (Figure 2).²⁷ The initial step often involves an extensive literature review to gather physicochemical properties and ADME parameters from clinical trials (e.g., concentration–time profiles, dosing regimens). Relevant data on interacting drugs, physiological parameters, and population demographics are then compiled, with the overall dataset being split into training and test sets.

Figure 2.

A stepwise strategy for PBPK model development.

During the model development phase, a middle-out approach combining in vitro, in silico, and clinical data with parameter estimation is employed. Parameter estimation techniques (e.g., Levenberg–Marquardt algorithm) fit the model to training plasma profiles, while key metabolic constants (such as Michaelis–Menten parameters) are derived from the literature.²⁰ Predicted plasma concentration–time profiles are compared to observed values, and performance is evaluated using metrics like mean relative deviations (MRDs) (equation (1)) and geometric mean fold errors (GMFEs) (equation (2)).²⁸

MRD = 10^{x}, where x = \sqrt{\frac{\sum_{i = 1}^{k} {(\log {\hat{c}}_{i} - \log c_{i})}^{2}}{k}}

(1)

GMFE = 10^{y}, where y = \frac{\sum_{i = 1}^{N} | \log (\frac{{\hat{ρ}}_{i}}{ρ_{i}}) |}{N}

(2)

Changes in parameters like C_max and AUC are frequently evaluated by comparing values measured under DDI conditions to results obtained without DDI (equation (3)). This approach applies to both model predictions and clinical observations. A prediction/observation ratio is often used to assess model validity. Prediction success is commonly measured using the two-fold criterion, a widely used metric in pharmacokinetics and drug–drug interaction (DDI) studies to evaluate prediction accuracy.²⁹ According to this criterion, a prediction is considered successful if the predicted value (e.g., the area under the concentration–time curve, or AUC) falls within a range that is double or half the observed experimental value.

\begin{matrix} DDI PBPK model \\ parameter ratio \end{matrix} = \frac{DDI PK paramete r_{DDI}}{\begin{matrix} Without interaction \\ PK paramete r_{Control} \end{matrix}}

(3)

For example, if the observed AUC ratio in the presence of a drug interaction is 1.0, the two-fold range would be between 0.5 and 2.0. This method provides a broad tolerance range for predictions, making it useful for addressing variability in experimental data. However, the two-fold range can sometimes create false impressions of prediction success, particularly when the observed AUC ratio is close to 1 (indicating no interaction). To address this limitation, a new predictive approach is proposed, as illustrated in Figure 3, to refine the success criteria and improve prediction accuracy.²⁹

Figure 3.

Schematic representation of the limits for different predictive methods, comparing the traditional two-fold measure (dashed lines) with the proposed new measure (dotted lines) by Guest et al.

It is challenging to apply physiologically based pharmacokinetic modeling to herbal drugs because the PBPK approach necessitates the physicochemical properties of these phytochemical/food component(s) causing the herb–drug interaction. Moreover, the uniformity of content of these components within and across marketed products and knowledge of the systemic bioavailability and human PK of these component(s) are usually missing.^29-35

Modeling Cancer Drug–Drug Interactions Using PBPK Approaches

The targeted and diverse mechanisms of therapies across different cancer types (chronic myeloid leukemia (CML), chronic lymphocytic leukemia (CLL), small lymphocytic lymphoma (SLL), acute myeloid leukemia (AML), myelofibrosis (bone marrow cancer), follicular lymphoma, HER2-positive metastatic breast cancer, lung, prostate, stomach, ovarian, blood, and head/neck cancers) are investigated in terms of drug–drug and drug–nutraceutical interactions.

Dasatinib

A PBPK model of dasatinib, which exhibits high dependence on CYP3A4-mediated metabolism and pH-dependent solubility, was developed and implemented in PK-Sim and MoBi modeling platforms based on experimental data from various clinical studies using different dosages and dosing intervals.²⁰ The model includes three elimination processes: (i) renal excretion, which accounts for passive glomerular filtration, (ii) metabolic CYP3A4-dependent elimination, which follows Michaelis–Menten kinetics, and (iii) a nonspecific hepatic clearance component, which accounts for metabolic pathways other than CYP3A4.

Plasma concentration–time profiles from clinical DDI studies were used to evaluate the predictive performance of a PBPK model across various interaction scenarios. Table 2 lists all the examined DDI studies, including the perpetrator drugs, interaction types, and their effects on the pharmacokinetics of the affected drugs—dasatinib being one of the evaluated cases.²⁰

Table 2.

Summary of DDI Studies.

Chemotherapy Drug	Role of Drug in Interaction	Interacted Drug(s)	Type of Interaction		Pred. Ratio		Obs. Ratio		Pred/Obs DDI Ratio		Ref
Chemotherapy Drug	Role of Drug in Interaction	Interacted Drug(s)	Type of Interaction		C_max	AUC_last	C_max	AUC_last	C_max	AUC_last	Ref
DASATINIBSecond-generation tyrosine kinase inhibitorFor chronic myeloid and acute lymphoblastic leukemia	Victim	Ketoconazole	CYP3A4 competitive inhibition	Strong inhibitor	2.17	9.14	2.99	7.65	0.73	1.19	Kovar et al, 2023²⁰
		Rifampicin	CYP3A4 induction and competitive inhibition	Strong inducer and weak inhibitor	0.31	0.11	0.17	0.17	1.79	0.64
		Rabeprazole (PPI)	Reduced solubility/absorption	Strong pH elevation impact	0.05	0.1	0.05	0.15	1.16	0.68
		Famotidine (H₂ antagonist)		Moderate impact on pH elevation	0.44	0.24	0.42	0.27	1.05	0.88
		Maalox (antacid)		Moderate impact on pH elevation	0.45	0.97	0.46	0.28	0.97	1.09
		Rabeprazole-BHCl		Lower gastric pH than rabeprazole	0.66	1.02	0.69	0.92	0.95	1.10
	Perpetrator	Simvastatin lactone	CYP3A4 inhibition	Weak inhibition	1.48	1.47	1.38	1.23	1.07	1.20
	Perpetrator	Simvastatin acid	CYP2C8, OATP1B1, and OATP1B3 inhibition	Weak inhibition	1.52	1.45	1.43	1.19	1.06	1.22
	Overall mean GMFE								1.18	1.24
IMATINIBFirst-generation tyrosine kinase inhibitorTargeted cancer therapy chronic myeloid leukemia	Victim	Ketoconazole	Competitive CYP3A4 and P-gp inhibition; non-competitive CYP2C8 inhibition	Strong inhibitor	1.24	1.44	1.39	1.40	0.89	1.02	Loer et al, 2024²⁸
		Rifampicin	Induction of CYP2C8, CYP3A4, and P-gp	Strong inducer	0.55	0.24	0.47	0.26	1.18	0.9
		Gemfibrozil	Mechanism-based inactivation of CYP2C8	Moderate inhibitor	1.09	1.51	0.65	0.93	1.68	1.61
	Perpetrator	Simvastatin (CML patients)	CYP3A4 inhibition	Strong inhibitor	2.84	3.91	2.14	3.22	1.33	1.21

Chemotherapy Drug	Role of Drug in Interaction	Interacted Drug(s)	Type of Interaction			Pred. Ratio		Obs. Ratio		Pred/Obs DDI Ratio		Ref.
Chemotherapy Drug	Role of Drug in Interaction	Interacted Drug(s)	Type of Interaction			C_max	AUC_last	C_max	AUC_last	C_max	AUC_last	Ref.
IMATINIB	Perpetrator	Simvastatin acid (CML patients)	CYP3A4 inhibition		Strong inhibitor	3.84	4.91	2.05	2.99	1.87	1.65	Loer et al, 2024²⁸
		Metoprolol (CML patients)	CYP2D6 competitive inhibition	CYP2D6 NMs	Weak inhibitor	1.12	1.19	1.11	1.26	1.01	0.94
		Metoprolol (CML patients)	CYP2D6 competitive inhibition	CYP2D6 IMs	Moderate inhibitor	1.15	1.38	1.29	1.17	0.89	1.19
		Hydroxymetoprolol (CML patients)	CYP2D6 competitive inhibition	CYP2D6 NMs	Weak inhibitor	1.04	1.12	0.91	1.29	1.14	0.87
		Hydroxymetoprolol (CML patients)	CYP2D6 competitive inhibition	CYP2D6 IMs	Moderate inhibitor	1.06	1.33	1.03	1.42	0.93	1.03
N-DESMETHYL IMATINIB	Victim	Ketoconazole	Competitive CYP3A4 inhibition		Weak inhibitor	0.83	1.04	0.84	0.86	0.99	1.21
		Rifampicin	Induction of CYP2C8 and CYP3A4		Moderate inducer	1.55	0.76	1.90	0.89	0.82	0.85
		Gemfibrozil	Mechanism-based inactivation of CYP2C8		Moderate inhibitor	0.43	0.54	0.43	0.51	1.01	1.06
	Overall mean GMFE									1.23	1.21
PYROTINIBTyrosine kinase inhibitorHER2-positive advanced solid tumors	Victim	Itraconazole	Potent CYP3A4 inhibition			3.28	9.82	3.78	11.4	0.87	0.86	Ni et al, 2023³⁶
		Rifampicin	Potent CYP3A4 induction			0.26	0.12	0.11	0.04	2.36	3.00
		Clarithromycin	Potent time-dependent CYP3A4 inhibitor			3.69	12.9	N/A	N/A	N/A	N/A
		Erythromycin	Moderate CYP3A4 inhibitor			2.74	6.64	N/A	N/A	N/A	N/A
		Fluvoxamine	Moderate to weak CYP3A4 inhibitor			1.37	1.59	N/A	N/A	N/A	N/A
		Cimetidine	Weak CYP3A4 inhibitor			1.12	1.21	N/A	N/A	N/A	N/A
		Efavirenz	Moderate CYP3A4 inducer			0.31	0.18	N/A	N/A	N/A	N/A

Chemotherapy Drug	Role of Drug in Interaction	Interacted Drug(s)	Type of Interaction	Pred. Ratio		Obs. Ratio		Pred/Obs DDI Ratio		Ref.
Chemotherapy Drug	Role of Drug in Interaction	Interacted Drug(s)	Type of Interaction	C_max	AUC_last	C_max	AUC_last	C_max	AUC_last	Ref.
DOCETAXELBreast, lung, prostate, stomach, and head and neck cancers	Victim	Ketoconazole	Strong inhibitor of CYP3A4	1.12	1.16	1.04	1.06	1.08	1.09	Li et al, 2024³⁷
		Itraconazole	Strong inhibitor of CYP3A4	1.10	1.11	N/A	N/A	N/A	N/A
		Rifampicin	Strong inducer of CYP3A4	0.95	0.84	N/A	N/A	N/A	N/A
		Cyclophosphamide	Weak inhibition of CYP3A4	1.00	1.00	N/A	N/A	N/A	N/A
CYCLOPHOSPHAMIDEBreast, blood and lymph system, and nerve cancers	Victim	Itraconazole	Strong inhibitor of CYP3A4	1.05	1.10	1.04	1.17	1.01	0.94
		Ketoconazole	Strong inhibitor of CYP3A4	1.06	1.61	1.08	1.76	0.98	0.91
		Rifampicin	Induces CYP3A4 metabolism, decreasing drug exposure	0.95	0.71	N/A	N/A	N/A	N/A
EPIRUBICINBreast cancer in patients who have had surgery to remove the tumor	Victim	Itraconazole	Inhibition of CYP3A4	1.03	1.10	N/A	N/A	N/A	N/A
		Ketoconazole	Inhibition of CYP3A4	1.05	1.12	N/A	N/A	N/A	N/A
		Rifampicin	Induction of CYP3A4	0.96	0.82	N/A	N/A	N/A	N/A
		Docetaxel	P-gp-mediated transporter interaction, no significant PK interaction	1.00	1.00	N/A	N/A	N/A	N/A
		Cyclophosphamide	No significant pharmacokinetic interaction observed	0.99	0.98	N/A	N/A	N/A	N/A

Chemotherapy Drug	Role of Drug in Interaction	Interacted Drug(s)	Type of Interaction	Pred. Ratio		Obs. Ratio		Pred/Obs DDI Ratio		Ref.
Chemotherapy Drug	Role of Drug in Interaction	Interacted Drug(s)	Type of Interaction	C_max	AUC_last	C_max	AUC_last	C_max	AUC_last	Ref.
PONATINIBTyrosine kinase inhibitor (TKI)Philadelphia chromosome-positive (Ph+) acute lymphoblastic leukemia (ALL) and chronic myeloid leukemia (CML)	Victim	Ketoconazole	Strong CYP3A4 inhibitor	1.28	1.79	1.47	1.66	0.87	1.08	Morite and Hanada, 2022³⁸
		Rifampicin	Strong CYP3A4 inducers	0.51	0.35	0.57	0.44	0.89	0.79
		Itraconazole	Strong CYP3A4 inhibitor	1.2	1.8	N/A	N/A	N/A	N/A
		Phenytoin	Strong CYP3A4 inducers	0.6	0.5	N/A	N/A	N/A	N/A
		Fluconazole	Moderate CYP3A4 inhibitor	1.1	1.4	N/A	N/A	N/A	N/A
VENETOCLAXChronic lymphocytic leukemia/small lymphocytic lymphoma	Victim	Posaconazole delayed release tablet (DRT)	Inhibition of CYP3A4	5.42	8.07	7.06	8.81	0.76	0.91	Bhatnagar et al, 2021³⁹
NILOTINIBChronic myeloid leukemia (CML)	Victim	Rifampin	Strong CYP3A4 inducers	0.41	0.30	0.35	0.22	1.17	1.36	Liu et al, 2024⁴⁰
NILOTINIBChronic myeloid leukemia (CML)	Victim	Ketoconazole	Strong CYP3A4 inhibitor	1.63	2.17	1.94	3.15	0.84	0.69	Liu et al, 2024⁴⁰

Chemotherapy Drug	Role of Drug in Interaction	Interacted Drug(s)	Type of Interaction	Pred. Ratio		Obs. Ratio		Pred/Obs DDI Ratio		Ref.
Chemotherapy Drug	Role of Drug in Interaction	Interacted Drug(s)	Type of Interaction	C_max	AUC_last	C_max	AUC_last	C_max	AUC_last	Ref.
IVOSIDENIBIDH1 inhibitor acute myeloid leukemia	Perpetrator	Midazolam	CYP3A4 inducer	0.4	0.27	0.42	0.36	0.95	0.75	Bolleddula et al, 2021⁴¹
RUXOLITINIBJAK1/2 tyrosine kinase inhibitor (TKI) inhibitor	Victim	Posaconazole	Inhibition of CYP3A4	1.21	1.59	N/A	N/A	N/A	N/A	Gerner et al, 2022⁴²

AUC_last, area under the plasma concentration–time curve from the first to the last time point of measurement; C_max, maximum plasma concentration; BHCl, betaine hydrochloride; PPI, proton pump inhibitor; Obs., observed, Pred., predicted; NMs, normal metabolizers; IMs, intermediate metabolizers; IDH1, isocitrate dehydrogenase 1.

Ketoconazole and itraconazole are potent CYP3A4 inhibitors that greatly raise dasatinib’s AUC and C_max, whereas rifampicin is a powerful CYP3A4 inducer that decreases dasatinib exposure. On the other hand, increasing stomach pH inhibits dasatinib absorption by around 90%. Thus, it emphasizes the importance of time-dependent medication administration, drug-avoidance measures, and switching techniques.

To forecast probable interactions, simulations of various scenarios were run using obtained PBPK model. Even more complex multi-agent combinations, such as a moderate inhibitor erythromycin and a high inducer carbamazepine, demonstrated balanced effects that reduced the size of the AUC drop, reflecting the increased complexity of induction versus inhibition in the combination.²⁰ This finding emphasizes the complexities of managing DDIs with dasatinib and the need for PBPK in guiding dosage adjustment.

Imatinib

Similarly, a physiologically based pharmacokinetic model is developed for imatinib and its main metabolite, N-desmethyl imatinib, while considering the complex metabolic processes of CYP3A4, CYP2C8, and P-glycoprotein in transport.²⁸ This model accurately predicts the behavior of imatinib not only as a victim, where its exposure was considerably altered by inducers, with a 74% decrease in AUC following co-administration of rifampicin, but also as a perpetrator, affecting the clearance of co-administered drugs such as simvastatin and metoprolol due to inhibition of CYP3A4 and CYP2D6 as given in Table 2. The drug–drug gene interaction (DDGI) studies are also included for metaprolol and hydroxymetaprolol cases by considering CYP2D6 normal and intermediate metabolizer patients. Over 90% of its predictions were within a two-fold margin of error and hence more reliable and clinically significant. However, it still acknowledged a few limitations, like not considering data on genetic polymorphisms.²⁸

Pyrotinib

PBPK modeling approach is applied to assess CYP3A4-mediated DDIs affecting pyrotinib, another TKI used in HER2-positive breast cancer.³⁶ As shown in Table 2, strong inhibitors, such as itraconazole, approximately 10-fold increased pyrotinib AUC, whereas strong inducers, such as rifampicin, comparably decreased AUC. Moderate and weak CYP3A4 inducers/modulators demonstrated graded effects on exposure. Considering these pronounced changes, the authors recommend a general precaution and avoidance of co-administration of pyrotinib with strong and moderate CYP3A4 modulators. Poor inhibitors, however, pose an insignificant threat and can be given together with less concern. These results highlight even more the utility of PBPK modeling in defining safe and efficacious dosing regimens, informing the development of clinical study designs, and minimizing the need for extensive empirical testing in complex polypharmacy scenarios.³⁶ The high predictive performance of the model (GMFE <2.0) enhances its usefulness in informing clinical decisions.

Neoadjuvant Chemotherapies: Docetaxel, Cyclophosphamide, and Epirubicin

By concentrating on three common neoadjuvant chemotherapeutic agents, docetaxel, cyclophosphamide, and epirubicin, a study on the utility of PBPK modeling in DDI prediction is conducted.³⁷ Although CYP3A4 and P-pg activity affected all three of these medications, it is more likely that co-administration of CYP3A4 inducers or inhibitors with docetaxel and cyclophosphamide, but not epirubicin, will produce clinically noticeable changes based on systemic exposures. Predicted values of AUC and C_max changes are very close to observed values (Table 2). Strong inhibitors like ketoconazole and itraconazole intensely increased AUC and thus needed lower doses whereas inducers like rifampicin decreased AUC, which may affect the drug’s efficacy, and thus higher doses may be necessary.³⁷ Epirubicin was the least affected among all these agents; it showed the least variation in exposure, and thus, this drug may be of lesser concern when DDI is considered for this class of drugs. Based on the predicted PBPK model results, docetaxel and cyclophosphamide should be carefully evaluated and their dosages adjusted when taken with strong CYP3A4 inhibitors or modulators, with some assurance that epirubicin will be only little impacted. Once more, this provides an excellent example of how population pharmacokinetic modeling can be used to forecast, measure, and control drug–drug interactions in complex oncologic regimens.³⁷

Ponatinib

Ponatinib is a tyrosine kinase inhibitor (TKI) used in the treatment of Philadelphia chromosome-positive (Ph+) acute lymphoblastic leukemia (ALL) and chronic myeloid leukemia (CML).^38,43 PBPK modeling techniques are employed for predicting possible DDIs with ponatinib (Table 2). Because ponatinib is primarily metabolized via that pathway and exhibits very low renal clearance, the scientists focused on the measurement of the effects of co-administered medications that inhibit or increase CYP3A4.²⁶ According to their validated model, which has given very good agreement with observed clinical data, strong CYP3A4 inducers, such as rifampicin, greatly reduce ponatinib levels, potentially undermining therapeutic efficacy, while strong CYP3A4 inhibitors, such as ketoconazole, may greatly increase ponatinib exposure, increasing the risk of toxicity.³⁸ Changes with both moderate inhibitors and weak inducers were modest, pointing at safer profiles upon co-administration.³⁸

Venetoclax with Posaconazole

Posaconazole, a powerful CYP3A4 inhibitor, when dosed in a delayed release tablet (DRT) manner of 300 mg in combination with venetoclax, led to an approximately 8.8- and 7.1-fold increase in venetoclax AUC and C_max, respectively, and the safe dose was reduced from 400 mg to 70 mg daily for safety reasons.³⁹ Treatment with high-dose posaconazole (500 mg DRT) resulted in a 12% increase in venetoclax exposure, well within a concentration range tolerated for safety.³⁹ All of these interactions were predicted when applying the validated PBPK model which further supports its use in guiding venetoclax dose adjustment to ensure safety for individual patients and to limit the number of clinically relevant studies performed. These results emphasize the need for personalized dosing approaches to mitigate CYP3A4-dependent drug–drug interactions in clinic settings.³⁹

Nilotinib and Special Populations

The physiologically based pharmacokinetic model of the drug nilotinib in CML is established and then improved to describe its pharmacokinetic and distribution pattern in functional spaces that comprise drug–drug interaction and select subpopulations such as children, pregnant woman, and lactating woman.⁴⁰ The AUC and C_max of nilotinib increased 3.15-fold and 1.94-fold, respectively, when it was co-administered with the potently CYP3A4 inhibitor ketoconazole making dose reductions necessary when strong CYP3A4 inhibitors are co-administered to avoid toxicity. Conversely, rifampin, an inducer of CYP3A4, was reported to reduce the AUC of nilotinib by as much as 78%, potentially leading to subtherapeutic levels; concomitant use should be avoided or dosage adjusted upwards.⁴⁰

The PBPK model was also used to explore other pharmacokinetic parameters that may be of special interest to certain populations. Considering the slower maturation trend of CYP1A2 in pediatric subjects and the impact of pregnancy on enzyme activity, these factors all play a dramatic impact on nilotinib’s PK along with a requirement for personalized dosing strategies.⁴⁰ Findings underscore the utility of PBPK modeling in forecasting pharmacokinetic changes, optimizing nilotinib therapy, and confirming safety and efficacy in challenging clinical scenarios.⁴⁰ This study further reinforces a growing role of PBPK modeling in clinical precision oncology, acting to facilitate the ability to navigate drug–drug interactions and population-specific issue.⁴⁰

Ivosidenib (IDH1 Inhibitor)

The pharmacokinetics and drug–drug interactions of ivosidenib (AG-120), a novel inhibitor of isocitrate dehydrogenase 1 (IDH1)—a cytosolic enzyme of the tricarboxylic-acid cycle whose gain-of-function mutations drive the oncogenic accumulation of 2-hydroxyglutarate in acute myeloid leukemia (AML)—are studied using the Simcyp platform.⁴¹ Within a physiologically based pharmacokinetic modeling approach, the perpetrator potential of ivosidenib with respect to CYP-mediated and transporter-mediated DDIs is described, capturing inductive and inhibitive mechanisms on key PK parameters such as hepatic clearance, absorption, AUC, C_max, and elimination. The study shows a highly significant induction of CYP3A4, increasing hepatic clearance as high as 371% with moderate induction of CYP2B6, CYP2C8, and CYP2C9. Also, inhibition of transporters, such as P-glycoprotein, OATP1B1/1B3, and OAT3, is observed among others.⁴¹ There is no significant affect observed on the substrates of CYP2C9-like warfarin and substrates of P-gp-like digoxin.⁴¹ Thus, it allows for appropriate predictions of drug–drug interactions and sets requirements for labeling. This research underlines how dosing adjustment of concomitant drugs can play an important role in patients being treated with ivosidenib to effect therapy with reduced possible clinical interactions.

Ruxolitinib and Disease-Specific Factors

PBPK modeling is performed for the estimation of DDI for ruxolitinib (RUX), a JAK1/2 inhibitor, co-administered with posaconazole (POS), a potent CYP3A4 inhibitor.⁴² Since RUX is predominantly metabolized by major metabolic pathways via CYP3A4 and CYP2C9, it was highly susceptible to interact with substantial CYP3A4 inhibitors like POS.⁴² The current investigation utilized clinical data from patients with graft versus host disease (GvHD) to simulate and validate plasma concentration profiles in predicting the change in exposure upon co-administration. Based on the developed PBPK model, POS increased the AUC of RUX by 59%, the C_max by 20.5%, and placed emphasis on dose adjustment for GvHD patients in an attempt to avoid overexposure. GvHD reduces hepatic clearance and triggers changes in the pharmacokinetics of drugs that raise the risk of overexposure to RUX. This effect is potentiated by the receipt of other concomitant CYP inhibitors like atorvastatin and amiodarone, making an overall review of the medication important. The developed PBPK model takes into consideration both pharmacokinetic parameters along with disease-specific factors, providing actionable insights to optimize RUX therapy in GvHD patients receiving POS, thereby ensuring safety and efficacy.⁴²

TQ-B3525 and Multiple Metabolites

A physiologically based pharmacokinetic model is described for TQ-B3525, which is a dual inhibitor of PI3Kα and δ, and its metabolites M3 and M8-3, that would be used to study the emerging pharmacokinetic profile after the co-administration of modulators of CYP3A4 and P-gp. CYP3A4 and FMO3 are major enzymes participating in TQ-B3525 metabolic clearance, and TQ-B3525 is a substrate of P-gp, significantly affected by both its inhibitors and inducers.⁴⁴ Co-administration with rifampicin, the potent inducer of the cytochrome P450 isoenzyme 3A4, resulted in a pronounced decrease of the AUC of TQ-B3525 by 46% and reached only 76.1% of the baseline value in C_max. This interaction significantly reduced the systemic exposure of TQ-B3525. In contrast, M3 and M8-3 experienced significant elevations as a result of the increased metabolism and distribution change. In contrast, concomitant administration of itraconazole, a strong CYP3A4/P-gp inhibitor, significantly increased the AUC (204%) and C_max (131%) of TQ-B3525, and inhibited its formation of M3 and M8-3 (73–80%), indicative of a reduced elimination and increased bioavailability of the parent compound. PBPK results show that modulators of CYP3A4 and P-gp significantly impact exposure to TQ-B3525 and metabolite profiling. These agents significantly modify the pharmacokinetic landscape for TQ-B3525 by accelerating (or delaying) hepatic clearance and altering bioavailability.⁴⁴

Apatinib

Apatinib is a tyrosine kinase inhibitor used in cancer treatment.⁴⁵ To forecast the pharmacokinetics of apatinib in both healthy individuals and cancer patients, as well as to evaluate possible DDIs and drug–disease interactions (DDZIs), Liu et al conducted a PBPK modeling study by Simcyp Simulator and incorporating experimental data and settings from the literature.⁴⁵ Simulations showed that moderate CYP3A4 inducers significantly reduced apatinib exposure, while moderate CYP3A4 inhibitors increased it by 2–4 times (AUC). The exposure to apatinib increased 1.25–2 times when strong CYP2D6 inhibitors were used. For patients with drug–disease interactions, such as hepatic impairment, apatinib’s AUC showed a 2.25-fold increase in Child–Pugh B and a 3.04-fold increase in Child–Pugh C, accompanied by a slight reduction in peak plasma concentration (C_max).⁴⁵ The PBPK model effectively reduced the reliance on resource-intensive in vivo studies by delivering reliable in silico predictions. This approach is especially beneficial for anticancer treatments, where ensuring patient safety is critical and clinical trial populations are often limited.

Another paper on apatinib explores the evaluation of its interactions with the calcium channel blocker nifedipine using PBPK models.⁴⁶ Given that the CYP3A4 enzyme metabolizes both medications, apatinib co-administration considerably raises nifedipine’s C_max and AUC, suggesting extended systemic exposure.⁴⁶ The absorption rate does not alter much, as evidenced by the T_max (time to achieve C_max), which is almost unchanged. Other distribution characteristics, including the volume of distribution (V_d), did not show any significant variations. By blocking CYP3A4-mediated nifedipine metabolism, apatinib raises systemic exposure. This DDI may increase the chance of nifedipine-related side effects, like toxicity or hypotension.⁴⁶

Paclitaxel, Zanubrutinib, Oprozomib, and Fedratinib

One cytotoxic drug that is authorized for the treatment of several cancer types is paclitaxel which has many side effects, including neutropenia (low white blood cell count).⁴⁷ In order to forecast the incidence of neutropenia in cancer patients receiving paclitaxel treatment, a physiologically based pharmacokinetic–pharmacodynamic (PBPK-PD) model is constructed, particularly taking possible DDIs into account and integrating the metabolic routes via the CYP3A4 and CYP2C8 enzymes.⁴⁷ Simulations indicated that the combination of paclitaxel with potent CYP3A4 inhibitors might increase the incidence of neutropenia.⁴⁷ When paclitaxel is co-administered with potent CYP3A4 or CYP2C8 inhibitor ketoconazole, the C_max increases up to two times, as a result of decreased metabolism. C_max is lowered by CYP3A4 inducer rifampin.⁴⁷ Due to decreased clearance when CYP enzymes are inhibited, AUC dramatically rises (by two to three times in the case of CYP3A4 inhibitor ketoconazole and to a lesser amount with CYP2C8 inhibitor gemfibrozil). The toxicity of paclitaxel, particularly the possibility of severe neutropenia, is exacerbated by elevated C_max.⁴⁷ When CYP inducers, like rifampin, are administered concurrently, the AUC is reduced by 50–80%, resulting in subtherapeutic levels. CYP inducers improve clearance, which speeds up the body’s removal of the substance. Changes in clearance affect toxicity as well as efficacy. Because of their slower metabolism and clearance, inhibitors extend the half-life of paclitaxel. On the other hand, inducers shorten it. Extended drug exposure due to a prolonged half-life may result in cumulative toxicity. Since V_d mostly relies on tissue binding and distribution characteristics rather than metabolic pathways, it is less impacted by DDIs.⁴⁷ The dosage of paclitaxel may need to be increased with inducers to maintain therapeutic efficacy or decrease d with CYP inhibitors to avoid toxicity.⁴⁷

The tyrosine kinase inhibitor zanubrutinib is used in the treatment of B-cell malignancies.⁴⁸ The primary enzyme responsible for the metabolism of zanubrutinib is cytochrome P450 3A (CYP3A). Based on its interactions with cytochrome P450 (CYP) enzymes, specifically CYP3A, CYP2C9, and CYP2C19, zanubrutinib exhibits significant changes in important pharmacokinetic (PK) parameters when used in combination with other medications.⁴⁸ Optimizing therapeutic approaches requires an understanding of its dual roles as a victim and a perpetrator in DDIs. Physicochemical characteristics, in vitro data, and clinical pharmacokinetic data from patients with B-cell malignancies and healthy volunteers are used to build the PBPK model for zanubrutinib.⁴⁸ Strong CYP3A inhibitor (itraconazole), as well as mild inhibitors (erythromycin, diltiazem), raises C_max if zanubrutinib is a victim (affected by CYP modulators) because they decrease first-pass metabolism and slow systemic clearance. The impairment of metabolism results in a considerable increase in AUC (2–4 times). Because CYP3A-mediated elimination is hindered, clearance is drastically reduced. The longer the drug is in the systemic circulation, the longer its half-life (t_1/2) is. Clinically, these could raise the chance of zanubrutinib-related side events such hematologic toxicity, tiredness, or diarrhea. On the other side, CYP3A inducer rifampin raises metabolism, which lowers C_max. Subtherapeutic exposure results from a large (up to 80%) reduction in AUC.⁴⁸ As zanubrutinib is digested more quickly and clearance increases,⁴⁸ t_1/2 is shortened, indicating a quicker rate of drug removal. Zanubrutinib’s effectiveness is decreased in these circumstances; a dose increase or other treatment may be required.⁴⁸

The oral proteasome inhibitor oprozomib was created to treat advanced cancers, such as multiple myeloma, among other cancers.⁴⁹ The Jak2 inhibitor fedratinib is used in the treatment of myelofibrosis, a kind of blood cancer.⁵⁰ Oprozomib and fedratinib’s drug–drug interaction risks are evaluated using physiologically based pharmacokinetic (PBPK) modeling by Simcyp population-based simulator (Simcyp, Sheffield, UK), with an emphasis on their effects on cytochrome P450 enzymes, to guarantee safe use in patients receiving cancer therapy.^49,50 The pharmacokinetic (PK) parameters of fedratinib and oprozomib were simulated under various drug–drug interaction scenarios in the PBPK modeling experiments. When oprozomib and fedratinib are taken with potent CYP3A4 inhibitors (midazolam and ketoconazole, respectively), key PK outcomes include elevated AUC and C_max, indicating enhanced systemic exposure. Additionally, when co-administered with CYP3A4 inducers like rifampin, simulations revealed a lower AUC.^49,50 The fedratinib model also evaluated the effect of CYP2D6 inhibition. These findings inform dose optimization for minimizing toxicity and ensuring efficacy in cancer patients.

Performance of PBPK Modeling in Chemotherapy DDIs

Drug-interaction modeling is only as useful as the criteria against which its predictions are judged. The acceptance limits presented in Figure 3—the traditional two-fold window and the concentration-dependent limits proposed by Guest et al²⁹—are used as quantitative yardsticks for deciding whether a simulated change in exposure (e.g., C_max or AUC_last) lies “close enough” to its clinical counterpart for dosing, labeling, or further study. Because chemotherapy regimens are characterized by narrow therapeutic indices and complex polypharmacy, both sets of limits were applied in the present work to provide a balanced assessment of model performance: the two-fold range is regarded as representing the regulatory standard for PBPK verification, whereas the tighter Guest boundaries are intended to reduce false indications of success when the observed DDI ratio is near unity.²⁹

Predicted-to-observed ratios for C_max (panel a) and AUC_last (panel b) obtained from all chemotherapy DDI studies in Table 2 are presented in Figure 4.

Figure 4.

Predicted versus observed DDI C_max (a) and DDI AUC_last (b) ratios of given chemotherapy drugs in DDI studies in Table 2.

Close alignment with the identity line is evident for most simulations, and interaction magnitudes between 0.1 and 10 are spanned. Specifically, 79 % (n = 34) of the C_max ratios and 86 % (n = 35) of the AUC_last ratios fall within the conventional two-fold acceptance limits, whereas 54% and 61%, respectively, fall within the stricter Guest boundaries.²⁹ Deviations beyond two-fold are symmetrically distributed above and below unity, so no systematic prediction bias is apparent. Accordingly, the PBPK framework is judged to capture both peak- and exposure-based interaction magnitudes across a broad spectrum of oncology DDIs and is considered suitable for prospective clinical DDI risk assessment.

PBPK-Based Evaluation of Herb–Drug and Food–Drug Interactions in Oncology

Phytochemicals and herbal products have gained increasing popularity in oncology practice for their perceived health benefits and antiproliferative properties. Unfortunately, their abilities to modulate critical drug-metabolizing enzymes and transporters add yet another layer of complexity to cancer pharmacotherapy. Concomitant administration of herbal medications (e.g., St. John’s Wort, Curcuma longa, Schisandra species) with anticancer agents can precipitate clinically relevant herb–drug interactions. These HDIs share mechanistic drivers with classical drug–drug interactions—notably CYP3A4 modulation and P-gp transport—yet are complicated by variable phytochemical content and sparse pharmacokinetic data.^23,51 PBPK modeling provides a mechanism-based quantitative methodology to understand how the agents influence the disposition of anticancer therapies and allows forecasting of clinically significant herb–drug and food–drug interactions without exhaustive empiricism.³¹

Schisandra Lignans and Targeted Therapy (Tyrosine Kinase Inhibitors)

Interactions between the hepatoprotective agents of Schisandra sphenanthera lignans and TKIs (bosutinib-used to treat certain type of chronic myeloid, imatinib-used to treat different types of cancer or bone marrow conditions) were studied with the developed PBPK models. In vitro experiments are combined with PBPK simulations to investigate how CYP3A4 and CYP2C8 are modulated by Schisandra lignans: schisantherin A, schisandrin A, and schisandrol B.^24,52,53 Schisandra lignans strongly inhibited CYP3A4 and CYP2C8; this significantly altered the pharmacokinetics of bosutinib but had very minimal clinical influence on imatinib (Table 3).

Table 3.

Pharmacokinetic Effects of Various Phytochemicals/Herbal Medications on Chemotherapy Drugs.

Phytochemical/Substance	Drug	Mechanism	Effect on Pharmacokinetics	Clinical Implications	Ref
Schisantherin A (Schisandra)	Bosutinib, imatinib	CYP3A4/CYP2C8 inhibition	AUC ↑ 3-fold (bosutinib); minimal effect (imatinib)	Dose reduction needed for bosutinib to prevent toxicity	Adiwidjaja et al, 2020⁵²
Schisandrin A (Schisandra)	Imatinib	CYP3A4 induction/inhibition	Slight increase in imatinib bioavailability	No dose adjustment needed for imatinib
Schisandrol B (Schisandra)	Bosutinib	CYP3A4 reversible inhibition	Significant impact on bosutinib metabolism	Potential for enhanced drug levels; monitor closely
Hydrastine (goldenseal)	Bosutinib	CYP3A4 MBI and P-gp inhibition	AUC ↑ 2.0-fold	Requires dose reduction to avoid toxicity	Adiwidjaja et al, 2022⁵³
Berberine (goldenseal)	Bosutinib	CYP3A4 MBI	AUC ↑ 1.3-fold	Monitor plasma levels; possible dose adjustment	Adiwidjaja et al, 2022⁵³
Hydrastine (goldenseal)	Imatinib	Weak reversible CYP3A4 inhibition	No significant change in AUC	No dose adjustment necessary	Adiwidjaja et al, 2022⁵³
Berberine (goldenseal)	Imatinib	Weak reversible CYP3A4 inhibition	No significant change in AUC	Safe for co-administration at standard doses	Adiwidjaja et al, 2022⁵³
Bergamottin (from grapefruit)	Osimertinib	CYP3A4 inhibition	AUC ↑ 2.48-fold, C_max ↑ 1.59-fold	Increased risk of toxicity; consider dose reduction	Fuhr et al, 2021⁵⁴
Bergamottin (from grapefruit)	Olaparib	CYP3A4 inhibition	AUC ↑ 1.11-fold, minimal impact on C_max	No dose adjustment required	Fuhr et al, 2021⁵⁴
Curcumin (turmeric)	Acalabrutinib	CYP3A4 inhibition	AUC ↑ 1.57-fold, C_max ↑ 1.34-fold	Dose adjustment recommended	Pilla Reddy et al, 2021⁵⁵
Hyperforin (St. John’s Wort)	Acalabrutinib	CYP3A4/P-gp induction	AUC ↓ 60%, C_max ↓ 59%	Risk of subtherapeutic exposure; avoid combination	Adiwidjaja et al, 2019⁵⁶
Hyperforin (St. John’s Wort)	Osimertinib	CYP3A4/P-gp induction	AUC ↓ 62%, C_max ↓ 59%	Therapeutic efficacy compromised; avoid combination
Hyperforin (St. John’s Wort)	Midazolam (oral)	Intestinal CYP3A4 induction	AUC ↓ 80%, C_max ↓ 75%	Reduced efficacy; avoid SJW co-administration
Hyperforin (St. John’s Wort)	Docetaxel (IV)	Hepatic CYP3A4 induction	AUC ↓ 12%, minimal impact on C_max	Clinically insignificant interaction
Hyperforin (St. John’s Wort)	Imatinib	CYP3A4 induction and auto-inhibition	AUC ↓ 5%, clinically insignificant	No significant dose adjustment required
Hyperforin (St. John’s Wort)	Clopidogrel	CYP3A4 induction	Enhanced active metabolite (AUC ↑)	Potentially beneficial for platelet aggregation

MBI, mechanism-based inhibition; AUC, area under the plasma concentration–time curve; C_max, maximum plasma concentration; P-gp, P-glycoprotein (transporter); CYP, cytochrome P450 enzymes.

A three-fold increase of the AUC of bosutinib is observed (Table 3) when the extract was co-administered because of the dual effect involving the reversible inhibition and mechanism-based inactivation of the CYP3A enzymes.⁵² However, the lignans only slightly altered the pharmacokinetics of imatinib at its steady state, with AUCs only 5% higher. Schisandra lignans may act both as inhibitors and inducers of CYP3A enzymes, depending on the type of interaction mechanism. Such duality is of importance in the context of cancer therapy, since TKIs, such as bosutinib, are highly sensitive to CYP3A modulation.⁵² Schisandra might increase bosutinib exposure to potentially toxic levels and thus may require bosutinib dose reductions. On the other hand, the negligible effect on imatinib would indicate selective herbal medicine interactions with drugs driven by metabolic pathways.⁵²

This study demonstrates that PBPK modeling has the potential to foresee clinically relevant herb–drug interactions of Schisandra sphenanthera when combined with cancer therapeutics. Besides being a very encouraging hepatoprotective agent, its potent change in bosutinib pharmacokinetics may lead to adverse effects. Thus, combination therapy of Schisandra with CYP3A4-metabolized cancer drugs requires dose adjustment with careful monitoring; however, the drugs like imatinib are not affected significantly.

Goldenseal Alkaloids and Targeted Therapy (Tyrosine Kinase Inhibitors)

Berberine, hydrastine, palmatine, and smaller levels of canadine and hydrastinine are among the alkaloids found in the dried root or rhizome of goldenseal (Hydrastis canadensis L.). Goldenseal-based preparations have been used to treat conditions affecting the skin and eyes. PBPK models of two key goldenseal alkaloids, hydrastine and berberine, were created with great accuracy to predict the degree of CYP3A- and ABCB1-mediated drug interactions with goldenseal extracts and high-dose berberine.^24,25,53 Clinically relevant dosages of goldenseal extract and berberine supplements were anticipated to have no effect on imatinib pharmacokinetics while increasing systemic exposure to bosutinib by up to 2.0 and 1.3 times, respectively.⁵³ The interactions were attributed to a reduction in hepatic clearance and alterations in transporter-mediated drug efflux. However, imatinib’s exposure was unaffected due to its lower dependency on CYP3A4 and auto-inhibition of its own metabolism. Bosutinib dose reductions may be required to maintain exposure within the defined therapeutic range, although more clinical trials to confirm these interactions and their consequences for clinical outcomes are warranted.⁵³

PBPK model results underline the clinical importance of goldenseal alkaloids in altering tyrosine kinase inhibitor pharmacokinetics, especially for CYP3A4-dependent drugs such as bosutinib.⁵³ These results signify that dose adjustments of bosutinib are essential when combined with goldenseal products to evade its adverse effects. For imatinib, the absence of significant interactions seems to present a relatively safer profile for concurrent use with goldenseal extracts. Therefore, this study represents an important way to understand how herb–drug interactions can be managed in cancer therapy using PBPK modeling.

Food-Derived Phytochemicals: Bergamottin, Curcumin, and Hyperforin

PBPK modeling is used to investigate the interactions of common phytochemicals derived from food with other oncology drugs. These included bergamottin, curcumin, and hyperforin (found in grapefruit juice, turmeric, and St. John’s Wort, respectively) acting on anticancer drugs.^9,55 These agents act by inhibiting hepatic drug-metabolizing enzymes and transporters such as P-gp and BCRP (Table 3).

According to the results of PBPK models, bergamottin increases the AUC of osimertinib (used to treat non-small cell lung cancer (NSCLC)) approximately 2.5-fold, and curcumin increases acalabrutinib (used to treat mantle cell lymphoma (MCL), chronic lymphocytic leukemia (CLL), and small lymphocytic lymphoma (SLL)) AUC by 1.57-fold.⁵⁵ On the other hand, the induction of both CYP3A4 and P-gp by hyperforin lowers the acalabrutinib and osimertinib AUC by 60%–62% which can lead to subtherapeutic drug levels.

St. John’s Wort (SJW) interactions with pharmacological drugs are investigated via its main ingredient, hyperforin, using physiologically based pharmacokinetic modeling and in silico analysis.³⁵ Researchers demonstrated that hyperforin induces intestinal CYP enzymes such as CYP3A4, CYP2C9, and CYP2C19, as well as ABCB1 transporters (Table 3). This method significantly decreases bioavailability and, as a result, the therapeutic efficacy of any given medicines metabolized by these pathways. This was shown to be effective against a variety of CYP substrates, including imatinib, midazolam, and docetaxel that are used to treat certain types of cancers (breast, lung, prostate, stomach, head, and neck). Furthermore, with ratios of 15.5 to 1.1, respectively, it demonstrated a considerable preference for intestinal induction over hepatic induction. Imatinib, a CYP3A substrate, had a lower impact due to its auto-inhibitory action, resulting in a clinically insignificant steady-state effect.⁵⁶

Concerns about possible interactions between SJW and cancer treatments that rely on CYP3A4 metabolism or P-gp transport are brought to light by literature studies. However, with prodrugs such as clopidogrel, increased metabolic activity may result in a superior therapeutic response, even if such pathway activation restricts drug intake, as demonstrated with midazolam and docetaxel.⁵⁶

Artificial Intelligence (AI) and Drug–Drug Interactions

The fusion of artificial intelligence (AI) with physiologically based pharmacokinetic (PBPK) modeling is rapidly redefining how oncology teams anticipate, quantify, and ultimately manage DDIs. Classic PBPK workflows—built on in vitro and in vivo parameter estimation—are now being augmented by machine-learning (ML) algorithms that uncover latent relationships, tighten parameter uncertainty, and drive more reliable forward simulations.⁵⁷ Comparative work by Gill et al shows that combining ML with PBPK or population-PK (PopPK) frameworks improves both interpretability and clinical utility, while Chou and Lin demonstrate how ML aids feature selection, parameter fitting, and virtual-population generation, thereby reducing reliance on labor-intensive experimental datasets.^57-60

Across representative studies, ML models built with random forests, support-vector machines, or deep neural networks routinely predict key ADME parameters with high fidelity (R² >0.8, RMSE <0.5) and, once embedded in PBPK simulators, yield exposure metrics (AUC, C_max, V_d) that match—or surpass—traditional mechanistic efforts.⁵⁹ Particularly noteworthy is the Neural-ODE architecture, which can learn irregular time-series PK trajectories directly from data, opening a path toward fully data-driven PBPK simulations that require minimal prior structural assumptions.⁵⁹

A prominent application is ML-assisted PBPK hybridization for whole-body human PK prediction. Independent investigations by research groups confirm that such hybrids accurately reproduce absorption, distribution, metabolism, and excretion across diverse chemotypes and dosing routes, outperforming standalone statistical or mechanistic models in both scalability and adaptability.^61-63

Complementing these technical advances, a wider AI-DDI field has recently been catalogued.⁶⁴ After quantifying the clinical burden of interaction-related harm, the researchers mapped the data landscape—DrugBank, ChEMBL, PubChem, TWOSIDES, DDInter, and others—that now feeds multimodal features (structures, targets, pathways, side effects, real-world reports) into AI pipelines.⁶⁴ Methods fall into three tiers: (i) undirected (“interact or not”) models that rely on similarity scores or matrix propagation; (ii) event-level systems (e.g., DeepDDI, MDF-SA-DDI) that use transformers or GNNs to label tens of PK/PD outcomes; and (iii) asymmetric approaches (e.g., DGAT-DDI) that capture perpetrator to victim directionality.⁶⁴ On public benchmarks, the best binary tools exceed AUC 0.90, and leading multiclass models approach F ≈ 0.8 despite much larger label spaces.⁶⁵

Although momentum is strong, several obstacles must be overcome before AI-DDI tools can be deployed with confidence at the bedside. First, most deep models lack interpretability, preventing clinicians from understanding which molecular or clinical features drive an interaction alert. Second, the data are exceedingly imbalanced: confirmed DDIs represent a tiny fraction of all possible drug pairs, and many presumed “negatives” may simply be unreported positives, biasing model training. Third, no harmonized, gold-standard corpus exists—research groups curate different slices of DrugBank, FAERS, and other sources—so results are difficult to compare or validate. Finally, current algorithms incorporate little patient-specific context such as genotype, organ function, comorbidities or dosing schedules, even though these factors critically influence interaction risk. Addressing these four gaps is essential for translating AI-driven insights into reliable, personalized pharmacotherapy.⁶⁶

Future Perspectives: Potential for Personalized Medicine

DDIs that occur during cancer treatment remain a significant challenge, particularly as treatment protocols increasingly integrate multiple oncologic agents, adjuvant medications, and various phytochemical or nutritional supplements. In fact, PBPK modeling has demonstrated remarkable utility across a broad range of clinical scenarios—from tyrosine kinase inhibitors and conventional chemotherapeutics to novel targeted therapies—by predicting the magnitude and directionality (induction vs inhibition) of potential interactions. Such predictions, derived by combining in vitro and in vivo data, enable evidence-based guidance on avoiding potent enzyme inducers or inhibitors, timing administration around gastric pH modifiers, and adjusting dosages when herb–drug or food–drug interactions threaten safety or efficacy (Table 3).

PBPK modeling can be used to anticipate and address issues that reduce therapy effectiveness. For example, when cancer cells or the surrounding tumor microenvironment adapt and avoid immunosurveillance through mechanisms that reduce the therapeutic impact of PD-L1 inhibition, PD-L1 treatment resistance develops. PBPK-guided dose modifications may lessen specific resistance patterns associated with insufficient drug exposure in the context of PD-L1 inhibitors. However, complex immunobiological pathways that necessitate a wider range of approaches—from combination regimens to biomarker-based treatment optimization—are frequently involved in immune checkpoint therapy resistance. As a result, whereas PBPK-driven dose modifications can be very helpful in treating some aspects of PD-L1 inhibitor resistance, coordinated, mechanism-based treatment approaches yield the best outcomes.

Beyond the traditional applications, PBPK models also have the potential to reveal rare or underreported interactions caused by newly approved drugs, natural supplements, and off-label therapy, thus helping clinicians avoid such unanticipated and potentially lethal risks. Moreover, when combined with AI and machine learning, models can identify hidden patterns in pharmacokinetic data because its predictability increases. The synergy of PBPK modeling and advanced analytics with proactive clinical planning might well help shift treatment strategies from a reactive problem-solving approach to a more preventive, individualized one.

PBPK modeling has the potential to become a key component of precision oncology in the future. PBPK-informed decisions can improve clinical outcomes, optimize therapeutic efficacy, and strengthen patient safety by empowering clinicians to make data-driven changes instead of depending on indirect or sparse evidence. This ensures that all possible advantages are realized in the continuous endeavor to improve cancer care.

Conclusion

The possibility of pharmacokinetic and pharmacodynamic interactions is growing as polypharmacy and new therapeutic drugs are used more frequently. These interactions have the potential to worsen toxicity, impair patient outcomes, and have a substantial effect on therapeutic efficacy. Therefore, improving cancer therapy requires a thorough understanding of DDIs. To reduce hazards and improve therapeutic success, future research should concentrate on developing predictive tools, improving medication monitoring procedures, and incorporating personalized medicine approaches. To reduce the negative effects of DDIs and enhance the safety and effectiveness of cancer treatment, cooperation between pharmacologists, researchers, and clinicians is essential.

Footnotes

Author Contributions

E.E.T.: Conceptualization and design, Data acquisition, analysis, interpretation, Writing - drafting, Final approval, Accountability for integrity and accuracy; K.O.U.: Conceptualization and design, Data acquisition, analysis, interpretation, Writing - review and editing, Final approval, Accountability for integrity and accuracy.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors gratefully acknowledge the support from the Scientific Research Projects (BAP) Coordination Unit (Project No 20161).

ORCID iDs

Enes Emre Taş

Kutlu O. Ulgen

References

Beijnen

Schellens

. Drug interactions in oncology. Lancet Oncol. 2004;5(8):489-496. doi:10.1016/S1470-2045(04)01528-1

Cascorbi

. Drug interactions--principles, examples and clinical consequences. Dtsch Arztebl Int. 2012;109(33-34):546-555. doi:10.3238/arztebl.2012.0546

Foti

. Utility of physiologically based pharmacokinetic modeling in predicting and characterizing clinical drug interactions. Drug Metab Dispos. 2024;53:100021. doi:10.1124/dmd.123.001384

Abouir

Samer

Gloor

Desmeules

Daali

. Reviewing data integrated for PBPK model development to predict metabolic drug-drug interactions: shifting perspectives and emerging trends. Front Pharmacol. 2021;12:708299. doi:10.3389/fphar.2021.708299

Seymour

Routledge

. Important drug-drug interactions in the elderly. Drugs Aging. 1998;12(6):485-494. doi:10.2165/00002512-199812060-00006

Rai

Rozario

. Mechanisms of drug interactions II: pharmacokinetics and pharmacodynamics. Anaesth Intensive Care Med. 2023;24(4):217-220. doi:10.1016/j.mpaic.2023.01.001

Efferth

Saeed

MEM

Mirghani

, et al. Integration of phytochemicals and phytotherapy into cancer precision medicine. Oncotarget. 2017;8(30):50284-50304. doi:10.18632/oncotarget.17466

Bathaei

Imenshahidi

Hosseinzadeh

. Effects of berberis vulgaris, and its active constituent berberine on cytochrome P450: a review. Naunyn-Schmiedebergs Arch Pharmacol. 2024;398:179-202. doi:10.1007/s00210-024-03326-x

Adiwidjaja

Boddy

McLachlan

. Physiologically based pharmacokinetic modelling of hyperforin to predict drug interactions with St John’s Wort. Clin Pharmacokinet. 2019;58(7):911-926. doi:10.1007/s40262-019-00736-6

10.

Olsen

Sletvold

, eds. Medication Safety in Municipal Health and Care Services. Oslo, Norway: Cappelen Damm Akademisk/NOASP (Nordic Open Access Scholarly Publishing); 2022. doi:10.23865/noasp.172

11.

Souza-Peres

Flores

Umloff

Heinan

Herscu

Babos

. Everyday evaluation of herb/dietary supplement–drug interaction: a pilot study. Medicine (Baltim). 2023;10(3):20. doi:10.3390/medicines10030020

12.

Frechen

Solodenko

Wendl

, et al. A generic framework for the physiologically-based pharmacokinetic platform qualification of PK-Sim and its application to predicting cytochrome P450 3A4-mediated drug-drug interactions. CPT Pharmacometrics Syst Pharmacol. 2021;10(6):633-644. doi:10.1002/psp4.12636

13.

Scripture

Figg

. Drug interactions in cancer therapy. Nat Rev Cancer. 2006;6(7):546-558. doi:10.1038/nrc1887

14.

Min

Bae

. Prediction of drug-drug interaction potential using physiologically based pharmacokinetic modeling. Arch Pharm Res (Seoul). 2017;40(12):1356-1379. doi:10.1007/s12272-017-0976-0

15.

Calzetta

Page

Matera

Cazzola

Rogliani

. Drug-drug interactions and synergy: from pharmacological models to clinical application. Pharmacol Rev. 2024;76(6):1159-1220. doi:10.1124/pharmrev.124.000951

16.

Santamaria

Roberto

Buccilli

, et al. Clinical implications of the drug-drug interaction in cancer patients treated with innovative oncological treatments. Crit Rev Oncol Hematol. 2024;200:104405. doi:10.1016/j.critrevonc.2024.104405

17.

Yap

Bullock

Chong

, et al. Oncology combination drug development strategies for Project Optimus. Clin Cancer Res. 2024;30(23):5237-5241. doi:10.1158/1078-0432.CCR-24-1836

18.

Rousseau

Géraud

Geiss

, et al. Safety of solid oncology drugs in older patients: a narrative review. ESMO Open. 2024;9(11):103965. doi:10.1016/j.esmoop.2024.103965

19.

Hansten

Gomez-Lumbreras

Villa-Zapata

Malone

. Pascal’s uncertainty principle: managing drug–drug interactions when the risks are unclear. J Am Coll Clin Pharm. 2023;7(12):1197-1206. doi:10.1002/jac5.2031

20.

Kovar

Loer

HLH

Rüdesheim

, et al. A physiologically-based pharmacokinetic precision dosing approach to manage dasatinib drug-drug interactions. CPT Pharmacometrics Syst Pharmacol. 2024;13(7):1144-1159. doi:10.1002/psp4.13146

21.

Verschraagen

Koks

Schellens

Beijnen

. P-glycoprotein system as a determinant of drug interactions: the case of digoxin-verapamil. Pharmacol Res. 1999;40(4):301-306. doi:10.1006/phrs.1999.0535

22.

Chan

Adiwidjaja

McLachlan

Boddy

Harnett

. Interactions between natural products and cancer treatments: underlying mechanisms and clinical importance. Cancer Chemother Pharmacol. 2023;91(2):103-119. doi:10.1007/s00280-023-04504-z

23.

Guedon

Le Bozec

Brugel

, et al. Cannabidiol-drug interaction in cancer patients: a retrospective study in a real-life setting. Br J Clin Pharmacol. 2023;89(7):2322-2328. doi:10.1111/bcp.15701

24.

Embuldeniya

Goralski

. Unlocking the goldenseal reveals the complexities of natural product-drug interactions. J Pharmacol Exp Therapeut. 2023;387(3):249-251. doi:10.1124/jpet.123.001863

25.

Wang

Chen

Guo

, et al. Application of physiologically based pharmacokinetics modeling in the research of small-molecule targeted anti-cancer drugs. Cancer Chemother Pharmacol. 2023;92(4):253-270. doi:10.1007/s00280-023-04566-z

26.

Jia

Yao

, et al. Utilization of physiologically based pharmacokinetic modeling in pharmacokinetic study of natural medicine: an overview. Molecules. 2022;27(24):8670. doi:10.3390/molecules27248670

27.

Shebley

Sandhu

Emami Riedmaier

, et al. Physiologically based pharmacokinetic model qualification and reporting procedures for regulatory submissions: a consortium perspective. Clin Pharmacol Ther. 2018;104(1):88-110. doi:10.1002/cpt.1013

28.

Loer

HLH

Kovar

Rüdesheim

, et al. Physiologically based pharmacokinetic modeling of imatinib and N-desmethyl imatinib for drug–drug interaction predictions. CPT Pharmacometrics Syst Pharmacol. 2024;13(6):926-940. doi:10.1002/psp4.13127

29.

Guest

Aarons

Houston

Rostami-Hodjegan

Galetin

. Critique of the two-fold measure of prediction success for ratios: application for the assessment of drug-drug interactions. Drug Metab Dispos. 2011;39(2):170-173. doi:10.1124/dmd.110.036103

30.

Hermann

von Richter

. Clinical evidence of herbal drugs as perpetrators of pharmacokinetic drug interactions. Planta Med. 2012;78(13):1458-1477. doi:10.1055/s-0032-1315117

31.

Cox

Tian

Clarke

, et al. Modeling pharmacokinetic natural product-drug interactions for decision-making: a NaPDI center recommended approach. Pharmacol Rev. 2021;73(2):847-859. doi:10.1124/pharmrev.120.000106

32.

Oyanna

Clarke

. Mechanisms of intestinal pharmacokinetic natural product-drug interactions. Drug Metab Rev. 2024;56(3):285-301. doi:10.1080/03602532.2024.2386597

33.

Shannar

Chou

Peter

, et al. Pharmacodynamics (PD), pharmacokinetics (PK) and PK-PD modeling of NRF2 activating dietary phytochemicals in cancer prevention and in health. Curr Pharmacol Rep. 2025;11(1):6. doi:10.1007/s40495-024-00388-6

34.

Sivelli

Rossi

Baccetti

Di Stefano

Gallo

Firenzuoli

. Herb-drug interactions in oncology: a clinical up-date. Obm Icm. 2019;4(4):1. doi:10.21926/obm.icm.1904060

35.

Surendran

Dhurjad

Nanjappan

. Phytotherapeutics: the rising role of drug transporters in herb-drug interactions with botanical supplements. In: Mandal

Chakraborty

Sen

, eds. Evidence-Based Validation of Traditional Medicines. Singapore: Springer; 2021. doi:10.1007/978-981-15-8127-4_23

36.

Zheng

Liu

, et al. Physiologically based pharmacokinetic modeling to simulate CYP3A4-mediated drug-drug interactions for pyrotinib. Adv Ther. 2023;40(10):4310-4320. doi:10.1007/s12325-023-02602-1

37.

Zhou

Wang

Zhao

Wang

Shao

. Docetaxel, cyclophosphamide, and epirubicin: application of PBPK modeling to gain new insights for drug-drug interactions. J Pharmacokinet Pharmacodyn. 2024;51(4):367-384. doi:10.1007/s10928-024-09912-z

38.

Morita

Hanada

. Physiologically based pharmacokinetic modeling of ponatinib to describe drug-drug interactions in patients with cancer. Cancer Chemother Pharmacol. 2022;90(4):315-323. doi:10.1007/s00280-022-04466-8

39.

Bhatnagar

Mukherjee

Salem

Miles

Menon

Gibbs

. Dose adjustment of venetoclax when co-administered with posaconazole: clinical drug-drug interaction predictions using a PBPK approach. Cancer Chemother Pharmacol. 2021;87(4):465-474. doi:10.1007/s00280-020-04179-w

40.

Liu

Leong

Burckart

Dallmann

. Physiologically based pharmacokinetic modeling of nilotinib for drug–drug interactions, pediatric patients, and pregnancy and lactation. J Clin Pharmacol. 2024;64(3):323-333. doi:10.1002/jcph.2379

41.

Bolleddula

Yang

Prakash

. PBPK modeling to predict drug-drug interactions of ivosidenib as a perpetrator in cancer patients and qualification of the Simcyp platform for CYP3A4 induction. CPT Pharmacometrics Syst Pharmacol. 2021;10(6):577-588. doi:10.1002/psp4.12619

42.

Gerner

Aghai-Trommeschlaeger

Kraus

, et al. A physiologically-based pharmacokinetic model of ruxolitinib and posaconazole to predict CYP3A4-mediated drug-drug interaction frequently observed in graft versus host disease patients. Pharmaceutics. 2022;14(12):2556. doi:10.3390/pharmaceutics14122556

43.

Cheng

Wang

Zhang

. Clinical pharmacokinetics and drug-drug interactions of tyrosine-kinase inhibitors in chronic myeloid leukemia: a clinical perspective. Crit Rev Oncol Hematol. 2024;195:104258. doi:10.1016/j.critrevonc.2024.104258

44.

Zhu

Wang

. Predict the drug-drug interaction of a novel PI3Kα/δ inhibitor, TQ-B3525, and its two metabolites using physiologically based pharmacokinetic modeling. J Clin Pharmacol. 2024;64(12):1517-1527. doi:10.1002/jcph.6111

45.

Liu

Guo

Wang

Han

Xiang

. Application of physiologically based pharmacokinetic modeling to evaluate the drug-drug and drug-disease interactions of apatinib. Front Pharmacol. 2021;12:780937. doi:10.3389/fphar.2021.780937

46.

Liu

, et al. Application of physiologically based pharmacokinetic/pharmacodynamic models to evaluate the interaction between nifedipine and apatinib. Front Pharmacol. 2022;13:970539. doi:10.3389/fphar.2022.970539

47.

Mendes

MDS

Hatley

Gill

Yeo

. A physiologically based pharmacokinetic - pharmacodynamic modelling approach to predict incidence of neutropenia as a result of drug-drug interactions of paclitaxel in cancer patients. Eur J Pharmaceut Sci. 2020;150:105355. doi:10.1016/j.ejps.2020.105355

48.

Wang

Yao

Zhang

, et al. Comprehensive PBPK model to predict drug interaction potential of zanubrutinib as a victim or perpetrator. CPT Pharmacometrics Syst Pharmacol. 2021;10(5):441-454. doi:10.1002/psp4.12605

49.

Gore

, et al. Physiologically-based pharmacokinetic modelling to predict oprozomib CYP3A drug-drug interaction potential in patients with advanced malignancies. Br J Clin Pharmacol. 2019;85(3):530-539. doi:10.1111/bcp.13817

50.

Krishna

Surapaneni

. Physiologically based pharmacokinetic modeling to assess metabolic drug-drug interaction risks and inform the drug label for fedratinib. Cancer Chemother Pharmacol. 2020;86(4):461-473. doi:10.1007/s00280-020-04131-y

51.

Rizeq

Gupta

Ilesanmi

AlSafran

Rahman

Ouhtit

. The power of phytochemicals combination in cancer chemoprevention. J Cancer. 2020;11(15):4521-4533. doi:10.7150/jca.34374

52.

Adiwidjaja

Boddy

McLachlan

. Potential for pharmacokinetic interactions between Schisandra sphenanthera and bosutinib, but not imatinib: in vitro metabolism study combined with a physiologically-based pharmacokinetic modelling approach. Br J Clin Pharmacol. 2020;86(10):2080-2094. doi:10.1111/bcp.14303

53.

Adiwidjaja

Boddy

McLachlan

. Physiologically based pharmacokinetic model predictions of natural product-drug interactions between goldenseal, berberine, imatinib and bosutinib. Eur J Clin Pharmacol. 2022;78(4):597-611. doi:10.1007/s00228-021-03266-y

54.

Fuhr

Marok

Fuhr

Selzer

Lehr

. Physiologically based pharmacokinetic modeling of bergamottin and 6,7-dihydroxybergamottin to describe CYP3A4 mediated grapefruit-drug interactions. Clin Pharmacol Ther. 2023;114(2):470-482. doi:10.1002/cpt.2968

55.

Pilla Reddy

Neuhoff

. Food constituent- and herb-drug interactions in oncology: influence of quantitative modelling on drug labelling. Br J Clin Pharmacol. 2021;87(10):3988-4000. doi:10.1111/bcp.14822

56.

Zlatnik

Sagakyants

Nepomnyashchaya

Zakharova

Ulyanova

. Possibilities of using secondary plant metabolites as antitumor agents. Kazan Med J. 2024;105(5):813-824. doi:10.17816/KMJ634368

57.

Gill

Moullet

Martinsson

, et al. Comparing the applications of machine learning, PBPK, and population pharmacokinetic models in pharmacokinetic drug-drug interaction prediction. CPT Pharmacometrics Syst Pharmacol. 2022;11(12):1560-1568. doi:10.1002/psp4.12870

58.

Danishuddin

Kumar

Faheem

Woo Lee

. A decade of machine learning-based predictive models for human pharmacokinetics: advances and challenges. Drug Discov Today. 2022;27(2):529-537. doi:10.1016/j.drudis.2021.09.013

59.

Chou

Lin

. Machine learning and artificial intelligence in physiologically based pharmacokinetic modeling. Toxicol Sci. 2023;191(1):1-14. doi:10.1093/toxsci/kfac101

60.

Zhou

, et al. Predicting pharmacodynamic effects through early drug discovery with artificial intelligence-physiologically based pharmacokinetic (AI-PBPK) modelling. Front Pharmacol. 2024;15:1330855. doi:10.3389/fphar.2024.1330855

61.

Wang

, et al. A combination of machine learning and PBPK modeling approach for pharmacokinetics prediction of small molecules in humans. Pharm Res. 2024;41(7):1369-1379. doi:10.1007/s11095-024-03725-y

62.

Arav

. Advances in modeling approaches for oral drug delivery: artificial intelligence, physiologically-based pharmacokinetics, and first-principles models. Pharmaceutics. 2024;16(8):978. doi:10.3390/pharmaceutics16080978

63.

Jia

Teutonico

Dhakal

, et al. Application of machine learning and mechanistic modeling to predict intravenous pharmacokinetic profiles in humans. J Med Chem. 2025;68(7):7737-7750. doi:10.1021/acs.jmedchem.5c00340

64.

Zhang

Deng

Feng

Junliang

. Application of artificial intelligence in drug-drug interactions prediction: a review. J Chem Inf Model. 2024;64(7):2158-2173. doi:10.1021/acs.jcim.3c00582

65.

Nguyen

NTK

Kha

NQK

. On the road to explainable AI in drug-drug interactions prediction: a systematic review. Comput Struct Biotechnol J. 2022;20:2112-2123. doi:10.1016/j.csbj.2022.04.021

66.

Paliwal

Jain

Kumar

, et al. Predictive Modelling in pharmacokinetics: from in-silico simulations to personalized medicine. Expert Opin Drug Metab Toxicol. 2024;20(4):181-195. doi:10.1080/17425255.2024.2330666

Predictive Modeling of Pharmacokinetic Drug–Drug and Herb–Drug Interactions in Oncology: Insights From PBPK Studies

Abstract

Keywords

Introduction

Methods

Drug–Drug Interactions in Oncology

Physiologically Based Pharmacokinetic (PBPK) Modeling Strategies

Modeling Cancer Drug–Drug Interactions Using PBPK Approaches

Dasatinib

Imatinib

Pyrotinib

Neoadjuvant Chemotherapies: Docetaxel, Cyclophosphamide, and Epirubicin

Ponatinib

Venetoclax with Posaconazole

Nilotinib and Special Populations

Ivosidenib (IDH1 Inhibitor)

Ruxolitinib and Disease-Specific Factors

TQ-B3525 and Multiple Metabolites

Apatinib

Paclitaxel, Zanubrutinib, Oprozomib, and Fedratinib

Performance of PBPK Modeling in Chemotherapy DDIs

PBPK-Based Evaluation of Herb–Drug and Food–Drug Interactions in Oncology

Schisandra Lignans and Targeted Therapy (Tyrosine Kinase Inhibitors)

Goldenseal Alkaloids and Targeted Therapy (Tyrosine Kinase Inhibitors)

Food-Derived Phytochemicals: Bergamottin, Curcumin, and Hyperforin

Artificial Intelligence (AI) and Drug–Drug Interactions

Future Perspectives: Potential for Personalized Medicine

Conclusion

Footnotes

Author Contributions

Declaration of Conflicting Interests

Funding

ORCID iDs

References