The pH factor: unraveling drug-drug interactions in protein kinase inhibitors through the regulatory lens

Abstract

Drug-drug interactions (DDIs) are a significant concern in clinical practice, DDIs might be relevant when drugs with pH-dependent solubility are co-administered with gastric acid-reducing agents (ARAs) such as histamine, H2-receptor antagonists and proton pump inhibitors. One class prone to such DDI at the absorption phase are the weakly basic protein kinase inhibitors (PKIs). The aim of this work is to review recent Food & Drug Administration (FDA) and European Medicines Agency (EMA) submissions for PKIs and evaluate the various approaches by drug developers to characterize pH-dependent DDI liability potentially affecting efficacy in this class of drugs and assess how this impacts the labelling. For this purpose, 32 FDA New Drug Applications (NDAs) and 25 EMA Market Authorization Applications of PKIs in the last 5 years (2019 through 2024) were reviewed More than two-thirds of the submissions included a dedicated clinical DDI studies with an ARA, which remains the most frequent approach to evaluating gastric pH-dependent DDIs among the PKIs investigated, albeit model-informed drug development approaches are also attempted by applicants in about 20% of the submissions. In cases where no clinical DDI study was submitted and alternative approaches taken, this was accepted by the approving agencies. Only the complete absence of data on the DDI potential triggered the request to provide the information post-marketing. A risk-based approach, considering the drug’s properties and patient population, is crucial for determining the need for a clinical DDI study and should be discussed with the agencies during drug development.

INTRODUCTION

Drug-drug interactions (DDIs) are a significant concern in clinical practice. DDIs can occur when 2 or more drugs are administered together, and one drug affects the pharmacokinetics and potentially the pharmacodynamics of the other drug. Pharmacokinetic DDIs occur when one drug affects the absorption, distribution, metabolism, or excretion of another drug.

Pharmacokinetic DDIs of drugs with pH-dependent solubility may be caused when they are co-administered with gastric acid-reducing agents. Elevation of gastric pH by acid-reducing agents (ARAs) can affect the solubility and dissolution characteristics of orally administered drug products. The Food & Drug Administration (FDA) has provided guidance on evaluating pH-dependent DDIs during drug development [1]. In this guidance the FDA discusses alternative approaches to a clinical study for evaluating pH-dependent DDIs, such as population pharmacokinetics (PopPK) or physiologically based pharmacokinetics (PBPK) simulations to further assess the potential for pH-dependent DDIs or to inform clinical study designs. We searched through protein kinase inhibitor (PKI) products approved by FDA and European Medicines Agency (EMA) between 2019 to 2024, assessed the approach applicants have taken for their submission and in case alternative methods were applied how it was used as part of a New Drug Application (NDA) or a Market Authorization Application (MAA) and what the impact was on the prescribing information.

ARAs

ARAs such as antacids, histamine type 2 receptor antagonists (H2 blockers), and proton pump inhibitors (PPIs) are widely used, also among cancer patients, and many of these drugs are available over the counter [2,3]. DDIs between PPIs and other agents may lead to decreased drug absorption with possible reduced therapeutic benefit, or even increased toxicity [4]. With the exponentially increasing availability of oral cancer drugs on the market, this has gained relevance. In order to be adequately absorbed, many orally administered PKIs rely on pH-dependent solubility to dissolve within the gastric environment [5]. Gastric pH elevation induced by ARAs can, for this reason, impair an oral drug’s ability to reach adequate systemic levels. Although not every drug interaction involving a PPI is of clinical significance, there are prominent examples where the concomitant use of a PPI or H2 blocker is able to significantly reduce drug exposure, e.g. for dasatinib [6], erlotinib or pazopanib [7], leading to significantly reduced area under the curve (AUC). Other mechanisms may also explain DDIs between ARAs and anti-cancer agents, such as altered microbiota induced by ARAs and their metabolism by cytochrome P450 enzymes in the liver, such as CYP3A4 and CYP2C19 [8,9,10,11]. However, the latter interactions affecting the metabolism of the interacting drug are not further discussed in this article, we instead focus on the interactions affecting the absorption though modulation of gastric pH.

In 2023 guidance, FDA recommends the use of a PPI in DDI studies with ARA because PPIs result in prolonged effects on gastric pH and represent a worst-case scenario for pH-dependent DDIs. The dose of the PPI should be chosen to provide a near maximum effect on pH elevation. Another ARA may be selected, if the investigational drug competes on metabolic pathways or if the pharmacokinetics is anticipated to be affected by the investigational drug. Selection of ARAs and associated dosing regimens depends on the purpose, e.g., characterization of a worst-case scenario or identification of an appropriate mitigation strategy such as staggered administration [1].

KINASE INHIBITORS

Kinases phosphorylate specific amino acids on substrate enzymes, which subsequently alter signal transduction leading to downstream changes in cellular biology, such as growth, cell proliferation, differentiation, migration, metabolism and apoptosis, in response to external and internal stimuli [12]. Constitutive activation or inhibition, either by mutations or other means, can lead to dysregulated signal cascades, potentially resulting in malignancy and other pathologies [13,14].

Therefore, the selective PKIs on the target signaling pathways are mainly regulating cellular proliferation and tumor angiogenesis through the involved receptor kinase and/or intracellular kinases, aiming to perturb the cellular pathways that regulate malignant cell growth [15,16]

Imatinib was the first kinase inhibitor approved by the FDA in 2001. It inhibits the Bcr-Abl tyrosine kinase created by the Philadelphia chromosome-positive in chronic myeloid leukaemia [17]. Since the introduction of imatinib, the application of PKIs has been ever-expanding, particularly for cancer treatment, due to their critical role in cellular signaling [18,19]. Over the past 2 decades, more than 100 PKIs have been approved for the treatment of various types of cancer indicating the significant progress achieved in this research area [20]. These PKIs could be generally categorized into macromolecules and small molecule kinase inhibitors [15]. In this work, we focused on the heterogeneous group of small molecules of kinase inhibitors, which potentially possess the risk of gastric pH-dependent DDI.

METHODS

Based on “New Drugs at FDA: CDER’s New Molecular Entities and New Therapeutic Biological Products,” published on the FDA’s official website, PKIs approved by the FDA from 2019 to 2024 were identified. It was checked whether the same drugs were approved by EMA and summarized accordingly. The FDA application review files on Drugs@FDA (including the USPI) and European Public Assessment Reports available from the EMA website then were thoroughly reviewed for their gastric pH-dependent DDIs potential and the approach taken by the applicants and the regulatory agency to investigate this DDI liability. Search terms were: “kinase inhibitors,” “pH-dependent,” “solubility,” “dissolution,” “acid reducing agents,” “drug-drug interactions,” “proton pump inhibitor,” “H2 blocker,” “antacids,” “solubility,” “gastric pH,” and “weak base.” Compiled data included, but not limited to, the generic name, active ingredient, approval date, drug class, indications, type of studies (clinical, preclinical, in vitro, other), clinical study outcomes, pKa values, in vitro solubility assessments, market formulation characteristics, model-informed approaches and post-marketing requirements (PMRs) per the information found in the application review files as well as information contained in the drug’s labels. Data were double-checked by a regulatory intelligence tool (Cortellis by Clarivate, London, UK).

Submissions (NDAs & MAAs) in the last 5 years (2019 through 2024) were evaluated as well as the PMRs issued by the regulatory agencies.

The application of model-informed drug development (MIDD) approaches to answer regulatory questions and PMRs related to gastric pH-dependent DDIs was also checked.

Data were listed and aggregated in summaries using descriptive statistics. Data analysis was executed using Microsoft Excel (for Microsoft 365, Version 16.01.13801.20722).

RESULTS

Thirty-two NDAs to FDA and 25 Market MAAs to EMA of PKIs were thoroughly reviewed and the data is presented in Table 1. More than two-thirds of these drugs were approved in the oncology indication (23 out of 32 [FDA] and 18 out of 25 [EMA]).

Table 1. List of selected PKIs approved by FDA and EMA between 2019 and 2024.

Drug name	Year of initial approval (approving agency)	Indication	DDI investigation with ARAs*	Dedicated clinical study details: ARA used, dose and dosing regimen, relevant condition	Effect on exposure parameters in clinical DDI study (% change)	Information in drug label	PMR for assessment of DDI with ARA
Entrectinib	2019 (FDA), 2020 (EMA)	NSCLC	Dedicated clinical study; PBPK	Lansoprazol, 30 mg QD × 8 or 9 days, fasting	Cmax: −24%	No clinically relevant effect on PK	None
					AUC: −26%
Alpelisib	2019 (FDA), 2020 (EMA)	Breast cancer	Dedicated clinical study	Ranitidine, unknown, fasting	Cmax: −51%	No clinically relevant effect on PK when administered with food (recommended)	None
					AUC: −30%
Erdafitinib	2019 (FDA), 2024 (EMA)	Bladder cancer	In vitro	NA	NA	Unlikely effect	None
Fedratinib	2019 (FDA), 2020 (EMA)	Myelofibrosis	Dedicated clinical study	Pantoprazole, 40 mg QD × 7 days, unknown	Cmax: +10%	No clinically relevant effect on PK	None
					AUC: +20%
Upadacitinib	2019 (FDA, EMA)	Rheumatoid arthritis, psoriatic arthritis, atopic dermatitis	In vitro/PopPK	NA	NA	No effect on PK expected	None
Pexidartinib	2019 (FDA)	Tenosynovial giant cell tumour	Dedicated clinical study	Esomeprazole, 40 mg QD × 4 days, fasting	Cmax: −55%	Clinically relevant effect on PK	None
					AUC: −50%
Zanubrutinib	2019 (FDA), 2021 (EMA)	Mantle cell lymphoma	Dedicated clinical study, PBPK	NA	NA	No clinically relevant effect on PK	None
Avapritinib	2020 (FDA, EMA)	GIST	PopPK, PBPK (EMA only)	NA	NA	No clinically relevant effect on PK	None
Selumetinib	2020 (FDA), 2021 (EMA)	Neurofibromatosis type 1	In vitro	NA	NA	No effect on PK expected; study waiver accepted	None
Pemigatinib	2020 (FDA), 2021 (EMA)	Cholangiocarcinoma	Dedicated clinical study	Esomeprazole, 40 mg QD × 6 days, fed	Cmax: −35%	No clinically relevant effect on PK	None
					AUC: −8%
Tucatinib	2020 (FDA, EMA)	Breast cancer	Dedicated clinical study	Omeprazole, 40 mg QD × 5 days, fasting	Cmax: −13%	No clinically relevant effect on PK	None
					AUC: −13%
Ripretinib	2020 (FDA), 2021 (EMA)	GIST	Dedicated clinical study	Pantoprazole, 40 mg, unknown	Cmax: +3%	No clinically relevant effect on PK	None
					AUC: +9%
Selpercatinib	2020 (FDA, EMA)	NSCLC, thyroid cancer	Dedicated clinical study	Omeprazole, unknown, fed	Cmax: −22%	No clinically relevant effect on PK when administered with food (recommended)	None
					AUC: 0%
Capmatinib	2020 (FDA), 2022 (EMA)	NSCLC	Dedicated clinical study	Rabeprazole, 20 mg QD × 4 days, unknown	Cmax: −38%	No clinically relevant effect on PK	None
					AUC: −25%
Pralsetinib	2020 (FDA), 2021 (EMA)†	NSCLC, thyroid cancer	Dedicated clinical study/PopPK	Esomeprazole, 40 mg QD × 6 days, unknown	Cmax: −25%	No clinically relevant effect on PK	None
					AUC: −15%
Asciminib	2021 (FDA), 2022 (EMA)	Chronic myeloid leukemia	PBPK/dedicated clinical study	Rabeprazole, 20 mg QD × 4 days, fasting	Cmax: −9%	Not mentioned	None
					AUC: −1%
Tepotinib	2021 (FDA, EMA)	NSCLC	Dedicated clinical study	Omeprazole, 40 mg QD × 5 days, fasting	Cmax: −7%	No clinically relevant effect on PK	None
					AUC: +2%
Infigratinib	2021 (FDA)	Cholangiocarcinoma	Dedicated clinical study	Lansoprazole, 30 mg QD × 4 days, fasting	Cmax: −49%	Clinically relevant effect on PK	None
					AUC: −45%
Umbralisib	2021 (FDA)	Marginal zone lymphoma and follicular lymphoma	Dedicated clinical study	Omeprazole, 40 mg QD × 4 days, fasting	Cmax: −8%	No clinically relevant effect on PK	None
					AUC: +20%
Mobocertinib	2021 (FDA)	NSCLC	In vitro	NA	NA	Unlikely to have an in vivo effect	None
Abrocitinib	2021 (EMA), 2022 (FDA)	Atopic dermatitis	−	NA	NA	Effect is unknown/not studied	PMR (FDA)
Pacritinib	2022 (FDA)	Myelofibrosis	PopPK	NA	NA	Not mentioned	None
Futibatinib	2022 (FDA), 2023 (EMA)	Cholangiocarcinoma	Dedicated clinical study	Lansoprazole, 60 mg QD × 5 days, unknown	Cmax: +8%	No clinically relevant effect on PK	None
					AUC: +5%
Capivasertib	2023 (FDA), 2024 (EMA)	Breast cancer	Dedicated clinical study	Rabeprazole, unknown, unknown	Cmax: −27%	No clinically relevant effect on PK	None
					AUC: -6%
Deucravacitinib	2022 (FDA), 2023 (EMA)	Psoriasis	Dedicated clinical study	Rabeprazole, 20 mg QD × 7 days, unknown	Cmax: +1%	No clinically relevant effect on PK	None
					AUC: +1%
Ritlecitinib	2023 (FDA), 2023 (EMA)	Severe alopecia areata	In vitro	NA	NA	Not mentioned	None
Fruquintinib	2023 (FDA), 2024 (EMA)	Colorectal cancer	Dedicated clinical study	Rabeprazole, 40 mg QD, unknown	Cmax: +3%	No clinically relevant effect on PK	None
					AUC: +8%
Pirtobrutinib	2023 (FDA), 2024 (EMA)	Mantle cell lymphoma	Dedicated clinical study	Omeprazole, 40 mg QD, unknown	Cmax: −	No clinically relevant effect on PK	None
					AUC: +11%
Leniolisib	2023 (FDA)	Phosphoinositide 3-kinase delta syndrome	Other analysis	NA	NA	No clinically relevant effect on PK	None
Repotrectinib	2023 (FDA)	NSCLC	In vitro	NA	NA	Not mentioned	None
Quizartinib	2023 (FDA, EMA)	Acute myeloid leukemia	Dedicated clinical study	Lansoprazole, 60 mg QD × 5 days, unknown	Cmax: −14%	No clinically relevant effect on PK	None
					AUC: +5%
Momelotinib	2023 (FDA), 2024 (EMA)	Myelofibrosis and post-essential thrombocythemia	Dedicated clinical study	Omeprazole, 20 mg QD × 5 days, fasting	Cmax: −36%	No clinically relevant effect on PK (FDA); Not mentioned (EMA)	None
					AUC: −33%

PKI, protein kinase inhibitor; FDA, Food & Drug Administration; EMA, European Medicines Agency; DDI, drug-drug interaction; ARA, acid-reducing agent; PMR, post-marketing requirement; NSCLC, non-small cell lung cancer; PBPK, physiologically based pharmacokinetics; QD, once daily; Cmax, peak concentration; AUC, area under the curve; PK, pharmacokinetics; NA, not applicable; PopPK, population pharmacokinetics; GIST, gastrointestinal stromal tumor.

^*Categories used: dedicated clinical study, In vitro; modeling, PBPK modeling or PopPK modeling; other analysis, dash indicates none.

^†EMA withdrawn by applicant for use in thyroid cancer.

As weak bases, most of the PKIs exhibited pH-dependent solubility in in vitro dissolution testing except for 6 compounds which were not or only very slightly soluble in gastric or intestinal fluids or did not demonstrate pH-dependent solubility (erdafitinib, selumetinib, ripretinib, ritlecitinib, repotrectinib, and tepotinib). Generally, agencies—based on the FDA’s framework to assess clinical DDI risk with ARA for immediate-release products of weak-base drugs [1]—agreed on study waivers in these cases and no clinical DDI study with an ARA was required.

A total of 22 (68.8%, FDA) and 19 (76.0%, EMA) submissions included dedicated clinical DDI studies with ARAs. Sometimes DDI studies with ARAs were combined with other DDI studies.

Modeling approaches (both empirical and mechanistic modeling) have been attempted for 7 (21.9%, FDA)/5 (20.0%, EMA) of these submissions. For 4 (12.5%, FDA) and 2 (8.0%, EMA) PKIs, empirical modeling (i.e., PopPK modeling) was used for DDI assessment (for alpelisib and pralsetinib combined with clinical data). For 3 (9.4%, FDA) and 3 (12.0%, EMA) submissions, physiologically-based pharmacokinetic (PBPK) modeling was included (for entrectinib, upadacitinib and zanubrutinib). However, since the applicants also submitted negative data from a dedicated clinical DDI study, PBPK models were not always reviewed by the agencies.

For 6 (18.8%, FDA) and 4 (16.0%, EMA) PKIs, the applicant provided justification based on in vitro assessments. In 1 case (upadacitinib) in vitro data was supplemented by a modeling approach based on late-phase clinical data (PopPK analysis).

In cases where a clinical DDI study was conducted the percent changes in peak concentration (Cmax) and AUC were broad ranging from +10% to −55% for Cmax and +20% to −50% for AUC, respectively (Table 1, Fig. 1). Only 2 PKIs (pexidartinib and infigratinib) presented with a substantial %change in AUC by −50% and −45%, respectively and were considered a study with a positive outcome presenting respective language in the label (Table 1).

Figure 1. Clinical DDI study outcome evaluation as accepted by the approving agency (positive [light blue] or negative [dark blue]) versus the percentage change in AUC taken from the calculated point estimates reported in these studies.

DDI, drug-drug interaction; AUC, area under the curve.

If no dedicated clinical DDI study was conducted and data were provided by modeling approaches only or a justification was given based on in vitro data, agencies did not find any gaps resulting in any PMRs. Only the absence of data on the DDI potential triggered the request to provide the information post-marketing (FDA only). Interestingly, for leniolisib which exhibits pH-dependent solubility, no dedicated studies evaluating the effects of gastric ARAs on leniolisib pharmacokinetics (PK) have been conducted as noted by the FDA reviewer. However, a post-hoc analysis of interim data by statistical subgroup analysis showed nearly comparable plasma leniolisib concentration-time profiles between study participants who reported taking PPIs and those who did not. This analysis was deemed acceptable, and it was concluded that leniolisib may be used concomitantly with PPIs or H2 receptor antagonists with no dose adjustment.

Only few of the clinical DDI investigations had a significant clinical impact. Indeed, no implications were seen on their usage in the drug labels in 72% (FDA and EMA) as the DDIs were deemed not clinically relevant and did not require any dosage adjustments, specific therapeutic monitoring or recommendations for use. For 2 (6.3%, FDA) and 1 (4.0%, EMA) compounds a clinically relevant effect was observed resulting in recommendations for use (such as avoidance of PPIs or staggering administration e.g. drug intake 2 hours before or 10 hours after an H2 blocker) of the products for concomitant use with ARAs.

In 4 (12.5%, FDA) and 5 (20.0%, EMA) the results of DDIs with ARA investigated by any method were not mentioned in the label at all. In vitro outcomes were generally not mentioned on the label.

All PKIs were tested in vitro for solubility over the physiological pH range and not surprisingly most of them demonstrated a pH-dependent solubility (Table 2). In some of these cases (4 out of 32) the data were redacted in the reviewer files or not presented and are thus considered unknown for this research. A poor solubility at the intestinal pH was assumed when the dose/solubility ratio exceeded the standard volume of 250 mL. However not all PKIs showed poor sink conditions and demonstrated moderate or even high solubility in some cases. In these conditions the precipitation risk when transitioning from acidic to neutral pH was considered low to moderate. Nevertheless, the majority of drugs had a high precipitation risk inherent to the physicochemical nature of the molecules (Table 2). As two-thirds of the PKIs were investigated in a clinical DDI study with an ARA the precipitation risk was compared to the clinical outcome of PK AUC (Fig. 2). Drugs with a low to moderate risk for precipitation tended to have lower %changes in AUC or even percent increases, while compounds with a high precipitation risk more frequently showed a pronounced decrease in AUC values (Fig. 2), including the 2 drugs with a clinically positive outcome in the DDI study (see above). Further, PKIs with a low to moderate precipitation risk less frequently underwent a clinical DDI investigation as their solubility behavior could justify waiving a dedicated clinical study. When looking at trends over the five years we could not identify any strong movements within the categories of dedicated clinical study, modeling approaches or in vitro assessment. The latter showed a slight upward trend over the years in both regions from 13% in 2020 to 25% in 2023 (FDA) and from 17% in 2021 to 25% in 2023 (EMA).

Table 2. List of formulation features and in vitro characteristics listed by PKI.

Drug	Therapeutic dosing regimen	Quantitative solubility at different pH	Sink condition assessment	Precipitation risk when transitioning from acidic to neutral pH	pKa value	Final formulation used for market approval
Entrectinib	600 mg orally once daily	pH 1.2: 40 mg/mL; pH 5.3: 0.03 mg/mL; pH 6.4: 0.002 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	2.5, 7.5	HPMC hard capsule
Alpelisib	300 mg orally once daily	pH 1.2: 3.64 mg/mL; pH 4.0: 0.03 mg/mL; pH 6.8: 0.02 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	3.3, 9.4	Film-coated tablets (bile increases intestinal solubility)
Erdafitinib	8 mg orally once daily (may increase to 9 mg)	pH 1.1: 0.23 mg/mL; pH 5.3: 0.037 mg/mL; pH 7.4: 0.65 mg/mL	Moderate solubility at intestinal pH; DSR < 250 mL	Low	1.9, 9.2	Film-coated tablet
Fedratinib	400 mg orally once daily	pH 1.0: 112 mg/mL; pH 4.5: > 30 mg/mL; pH 6.8: 0.02 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	6.3, 9.5	Hard gelatin capsule
Upadacitinib	15 mg orally once daily; UC induction 45 mg	pH 1.0: 38.4 mg/mL; pH 4.0: 1.0 mg/mL; pH 6.8: < 0.2 mg/mL	High solubility in acidic environment; DSR < 250 mL	Low	4.7	Extended-release tablet
Pexidartinib	250 mg orally twice daily	pH 1.1: 2.4 mg/mL; pH 4.0: 0.02 mg/mL; pH 7.5: < 0.001 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	2.6, 5.4	HPMC hard capsule
Zanubrutinib	160 mg twice daily or 320 mg once daily	pH 1.2: > 0.24 mg/mL; pH 4.5: 0.07 mg/mL; pH 6.8: 0.05 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	Not reported	Hard gelatin capsule
Avapritinib	300 mg once daily (GIST); 200 mg (AdvSM)	pH 1.0: 3.6 mg/mL; pH 4.0: 0.14 mg/mL; pH 7.0: < 0.001 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	1.5, 3.5, 4.6, 7.0	Film-coated tablet
Selumetinib	25 mg/m2 twice daily	pH < 1.5: freely soluble; pH 1.5–3.0: sparingly soluble; pH > 3.0: slightly soluble	Unknown	Unknown	2.8, 8.4	HPMC hard capsule
Pemigatinib	13.5 mg once daily (14 days on, 7 off)	pH 1.2: > 0.71 mg/mL; pH 4.3: 0.03 mg/mL; pH 6.5: < 0.001 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	3.1, 5.7	Uncoated tablet
Tucatinib	300 mg twice daily	pH < 2.9: 18.7 mg/mL; > pH 4.6: < 0.4 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	Moderate	2.1, 4.2, 6.2	Film-coated tablet
Ripretinib	150 mg once daily	pH 2.0: 0.001 mg/mL; pH 6.5: < 0.001 mg/mL	Very poor solubility across physiological pH range	Low	4.5	Uncoated tablet
Selpercatinib	160 mg twice daily	No values provided	Unknown	Unknown	3.2, 7.3	Hard gelatin capsule, film-coated tablet
Capmatinib	400 mg twice daily	pH 1.0: 12.0 mg/mL; pH 4.0:< 0.02 mg/mL; pH 6.8: < 0.02 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	0.9, 4.5	Film-coated tablet
Pralsetinib	400 mg once daily	pH 2.0: 0.9 mg/mL; pH 3.9: 0.4 mg/mL; pH 7.6: < 0.01 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	Not reported	HPMC hard capsule
Asciminib	40 mg twice daily or 80 mg once daily	pH 1.0: 0.11 mg/mL; pH 5.0: < 0.001 mg/mL; pH 6.8: < 0.001 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	3.9	Film-coated tablet
Tepotinib	450 mg once daily	No values provided	Unknown	Unknown	9.5	Film-coated tablet
Infigratinib	125 mg once daily (21 days on, 7 off)	pH 1.0: < 1 mg/mL; pH 4.5: 1.7 mg/mL; pH 6.8: < 0.001 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	Not reported	Hard gelatin capsule
Umbralisib	800 mg once daily	pH 1.2: 0.03 mg/mL; pH 4.5: 0.02 mg/mL; pH7.4: 0.003 mg/mL	Very poor solubility across physiological pH range	Moderate	2.7	Film-coated tablet
Mobocertinib	160 mg once daily	pH 1.0: 152 mg/mL; pH 4.5: 0.1 mg/mL; pH 6.8: > 17.6 mg/mL	High solubility across physiological pH range; DSR < 250 ml	Low	2.5, 9.1	Gelatin capsule
Abrocitinib	100 mg or 200 mg once daily	pH 2.9: 12.7 mg/mL; pH 4.8: 0.13 mg/mL; pH 7.0: 0.02 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	5.3	Film-coated tablet
Pacritinib	200 mg twice daily	No values provided	Unknown	Unknown	Not reported	Hard gelatin capsule
Futibatinib	20 mg once daily	pH 1.2: 0.69 mg/mL; pH 4.5: 0.004 mg/mL; pH 6.8: 0.003 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	3.2	Film-coated tablet
Capivasertib	400 mg twice daily (4 days on, 3 off)	pH 1.2: and 4.5: higher solubility; pH 6.8: 0.19 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	Not reported	Film-coated tablet
Deucravacitinib	6 mg once daily	pH 1.05: > 3 mg/mL; pH 6.5: 0.009	Moderate solubility at intestinal pH; DSR < 250 mL	Moderate	3.4	Film-coated tablet
Ritlecitinib	50 mg daily	pH 0.8: 28.1 mg/mL; pH 4.5: 22.0 mg/mL; pH 6.8: 6.80 mg/mL	High solubility across physiological pH range; DSR < 250 mL	Low	Not reported	HPMC hard capsule
Fruquintinib	5 mg once daily (21 days on, 7 off)	pH 1.2: 0.13 mg/mL; pH 4.5: < 0.001 mg/mL; pH 6.8: < 0.001 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	2.9	Hard gelatin capsule
Pirtobrutinib	200 mg once daily	No values provided	Unknown	Unknown	Not reported	Film-coated tablet
Leniolisib	70 mg twice daily	pH 1.0: 8.88 mg/mL; pH 4.2: 0.7 mg/mL; pH 6.8: 0.04 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	Not reported	Film-coated tablet
Repotrectinib	160 mg twice daily	pH 1.2: 0.008 mg/mL; pH 6.5: 0.02 mg/mL; pH7.4: 0.006 mg/mL	Neutral compound with pH-independent solubility	Low	Not reported	Hard gelatin capsule
Quizartinib	35.4 mg once daily	pH 1.0: 0.93 mg/mL; pH 2–3: < 0.006 mg/mL; pH 4.5–8: < 0.0004 mg/mL	Poor solubility at intestinal pH; DSR > 250mL	High	4.8, 3.2	Film-coated tablet
Momelotinib	200 mg once daily	pH 1.0: 0.41 mg/mL; pH 4.0: 0.12 mg/mL; pH 5.0: 0.001 mg/mL	Poor solubility at intestinal pH; DSR > 250 mL	High	3.7	Film-coated tablet

PKI, protein kinase inhibitor; DSR, dose-solubility ratio; HPMC, hydroxypropyl methylcellulose; GIST, gastrointestinal stromal tumor; AdvSM, advanced systemic mastocytosis.

Figure 2. Clustered column plot of the reported percentage change in AUC from the clinical DDI studies versus the theoretical precipitation risk when transitioning the drug from an acidic to a neutral pH environment. Dark blue, high precipitation risk; violet, high precipitation risk and clinical studies with a positive DDI outcome; light blue, low to moderate precipitation risk; grey, risk unknown.

AUC, area under the curve; DDI, drug-drug interaction.

Most formulations used for market approval were film-coated tablets or conventional hard gelatin capsules. Five used hydroxypropyl methylcellulose (HPMC) as part of the shell for the hard capsules.

DISCUSSION

This work provides an overview of how drug developers and regulatory agencies approach and evaluate potential pH-dependent DDI liability in the class of PKIs of which a majority exhibit weak base properties and pH-dependent solubility. Hence, PKIs might have been impacted by gastric ARA if co-administered with PKIs that would impair their bioavailability and potentially their efficacy at the therapeutic dose. Despite this risk, FDA is currently the only regulatory agency who has issued a specific and detailed guidance for pH-dependent interactions with investigational drugs. The International Council for Harmonisation (ICH) harmonized guideline on drug interaction studies (ICH M12) specifically mentions that pharmacokinetic interactions, such as the impact on absorption (e.g., gastric pH change) are not within scope of this guideline and refer to regional guidelines [21]. EMA has recently adopted ICH M12 guideline and acknowledged the gap by issuing a concept paper on a guideline on investigation of drug interactions in the gastrointestinal tract as well as a question & answer document on the need for bioequivalence studies with ARA [22,23]. However, also the previous EMA guidance on DDI investigations [24] mentioned only briefly that if the solubility of the drug or the dissolution of the formulation is markedly pH-dependent in the physiological pH range, the potential effect of drugs which increase gastric pH, such as PPIs, H2-receptor antagonists or antacids, should be investigated in vitro, without providing any details on the design, or the data analysis, or any references to modelling approaches. In the 2023 FDA guidance [1] the framework for potential clinical DDI risk is based on a drug’s pH-dependent solubility at the relevant physiological pH range and in particular at pH 6.0–6.8. If the conditions of low solubility (less than dose divided by 250 mL) are met, then the agency recommends a clinical DDI study [1].

We categorized the investigation of DDI liabilities with ARAs into 3 major categories of in vitro assessments, dedicated clinical study and modeling approaches. In drug development there is usually a step-wise approach to characterize a compound starting with in vitro testing. Not surprisingly for most of the compounds included in this research there is a pH-dependent solubility with a high risk of precipitation when transitioning from acidic to neutral pH environment, but there were several cases where this risk was moderate to low based on the in vitro data.

In line with current guidelines, clinical DDI studies with ARAs (standalone or combined design) are currently the mainstream approach to evaluate gastric pH-dependent DDIs for PKIs in these FDA & EMA submissions. However, there is a growing interest in pharmaceutical industry in reducing the need for clinical studies while ensuring drug safety and efficacy. As a result, in vitro and in silico methods are becoming more crucial in pharmaceutical development [25,26].

We found that justification for a low precipitation risk can be provided based on in vitro data only, e.g., solubility assessment or dissolution testing across the pH range, to waive a dedicated clinical DDI study. This was done successfully e.g. with selumetinib or ritlecitinib for which a study waiver was accepted by the agencies based on the in vitro dissolution testing. We recommend discussing such approach with the regulatory agency during drug development, e.g., at a milestone meeting, to de-risk any issues during review once the application is submitted.

Although our data is not able to indicate any recent trend, around 20% of the drugs showed solubility independent of pH and thus were not critical in terms of potential interactions with ARAs. Early optimization and pre-selection of drug candidates based on favorable in vitro characteristics appears to be another way of successfully coping with later challenges on the drug development path.

As the sink conditions indicated a high risk for precipitation for the majority of PKIs, most applicants chose to conduct a dedicated clinical DDI study as suggested by the FDA guidance. This approach was taken for approximately two-thirds of the submissions we analyzed for this work. Overall, these clinical investigations revealed that for the vast majority no dosage adjustments were required, because the DDI studies did not show a clinically relevant effect on the drugs’ PK, except for 2 cases (pexidartinib and infigratinib). We recognized that the threshold for the agencies for a positive outcome was about 30% reduction in AUC, which was still deemed acceptable, and those studies were considered negative.

Model-informed drug development approaches nowadays are extensively used to assess and predict effects of various extrinsic and intrinsic factors on drug exposure [27]. Both empirical (PopPK) and mechanistic (PBPK) modeling have been used by the companies in the respective submissions. PopPK was used with co-medications, including ARAs as a covariate, while PBPK modeling by virtue of its bottom-up model building approach was conducted to answer questions regarding absorption DDI and ‘what if’ scenarios as well as generating hypotheses or informing study design. So PBPK was a tool for drug developers used for optimization during drug development but not applied for regulatory purposes, i.e., to submit data in lieu of a clinical study in this particular application. Modelled data were either confirmed by clinical data with other techniques, e.g. by PopPK, or by a dedicated clinical study, as shown for entrectinib [28,29] and Zanubrutinib [30,31]. FDA guidance (2023) [1] on gastric pH-dependent DDI with acid-reducing agents states that “In conjunction with the assessment framework outlined, PBPK simulations can sometimes be used to further assess the potential for pH-dependent DDIs. PBPK approaches can also be useful to inform clinical study designs. The application of PBPK is still evolving and is continuously being evaluated by the FDA.” Also, the FDA draft guidance on PBPK (2020) [32] to support biopharmaceutics applications for oral drug product development, manufacturing changes, and controls during the clinical process and post-approval changes, provides suggestions and recommendations for sponsors on how to use PBPK as a tool for formulation development and optimization.

In a commentary published by Zhang et al. [33] from FDA the authors summarized the PBPK analyses submitted to the FDA’s Office of Clinical Pharmacology in support of original NDAs between 2018 and 2019 and their impact on prescribing information. DDI-ARA investigations were found in 4% of the PBPK analyses in NDA submissions to the FDA (data from 2018–2019). This is lower than the numbers we have seen in the data covering the years 2019 to 2024 (~9–12%) and may indicate that sponsors feel more confident in applying these techniques. Zhang et al. [33] acknowledged that, while great progress has been made in applying PBPK analysis for regulatory decision making, there were cases in which models were considered inadequate for the intended purposes and additional refinement or another approach (e.g., a clinical study) might be needed. Even if not used for regulatory purposes PBPK simulations can be useful to inform clinical study designs and establishing the relationship of critical quality attributes, including formulation variables [34]. To facilitate its further use, sponsors planning to submit an NDA are advised to consult the appropriate FDA review division if they pursue a PBPK simulation approach to evaluate pH-dependent DDIs, as mentioned in the FDA guidance [1]. To further validate PBPK modeling, several approaches (in vitro, in silico, PBPK modeling, clinical trial) may still be needed to study DDIs of weakly basic drugs with ARAs to allow for a “cross-check” among methods as well as against in vitro clinical data, building further confidence in their use before they are ready to substitute a clinical study. This will require more time and examples from this heterogenous class of drugs because dissolution/precipitation behavior across the class of weakly basic drugs is wide currently hampering predictability [35]. Some of the applicants took an empirical modeling approach using collected clinical data on the coadministration with ARAs in conjunction with PK data. Our results show that PopPK (as a top-down approach) was as frequently applied as PBPK modeling (as a bottom-up approach), but with some regional differences. The biggest hurdle for the application of PopPK is data integrity and quality as in a phase 3 study PK sampling time points may not be optimal, and the records of dosing information may be incomplete. It is important that the pH-dependent DDI effect is sensitive to the time of administration of the investigational drug relative to the ARA (e.g., H2 blockers or antacids) and can also be affected by the dose of ARAs and the intake of food. Thus, it is critical for sponsors to have a prospective plan to ensure that relevant information such as dose, timing, and duration of administration of the investigational drug and ARAs as well as food intake and content are accurately captured. For PK sampling it is most relevant to obtain sufficient blood sampling during the absorption phase of the investigational drug to better capture the potential DDI effect.

As example, for avapritinib the effect of concomitant PPI use on absorption PK was evaluated in a PopPK model. During PK sample collection, the 77 patients who had 5 full days of PPI use, had the majority of PK samples (433 out of 718) collected in the absorption phase within 4 hours of last dose. The data did not point to any clinically significant interactions between avapritinib and ARAs in patients with gastrointestinal stromal tumor [36,37]. This illustrates that an adequate PK sampling strategy is key to the acceptance of the data and in this case the applicant did not provide data from a dedicated clinical study. Also, other applications in our dataset used late phase clinical patient data on ARA coadministration and could demonstrate in the PopPK analysis that such concomitant use had no clinically relevant impact on the PK of the investigational drugs. As no PMR was issued by the regulators in these cases we can assume that the data quality was deemed sufficient by the assessors to not request further clinical investigations.

Somewhat surprisingly most formulations of the final market image are conventional coated tablets or hard gelatin capsules, while few use HPMC in the shell to control release to some extent and enhance solubility. For entrectinib for example, the solubility could be enhanced by changing the clinical formulation from hard gelatin capsule to HPMC capsule shell for the market formulation.

CONCLUSION AND FUTURE PERSPECTIVES

Our work has shown that during the observation period from 2019 to 2024 most companies submitted clinical DDI studies in their packages to assess DDI liabilities with ARAs. However, several companies also have taken a different approach in this heterogenous class of weakly basic drugs and waived a clinical study by justifying with alternative data, mostly by in vitro data, which was accepted by the regulating bodies. Modeling approaches have been attempted less frequently for approval purposes but also served to optimize development such as dosing recommendations and formulation selection. More model validation work is needed to increase confidence in their use as a regulatory tool. It is promising that high-quality PK data used in PopPK analysis resulted in regulatory acceptance and did not require additional clinical studies to assess the interaction potential. A risk-based approach, considering the drug’s properties and patient population, is crucial for determining the need for a clinical DDI study and should be discussed with the agencies during drug development. A science-based approach is required to justify the absence of a dedicated clinical DDI study, on a case-by-case basis.