Challenges in Extrapolating Healthy Volunteer Pharmacokinetics to Oncology Populations: Advocating for a Holistic Perspective

ABSTRACT

Phase I clinical pharmacology (CP) studies in healthy volunteers (HVs) are conducted to assess factors influencing drug pharmacokinetic (PK) properties and guide potential dose adjustments or restrictions for specific subpopulations within the target patient population. However, for oncology drugs, direct extrapolation from HV data may not always yield accurate outcomes. This review examines three published examples—ribociclib, ceritinib, and midostaurin—where substantial changes in drug exposure observed in CP studies conducted in HVs (or non‐cancer subjects) did not align with the more modest effects seen in patient trials. While HV CP studies are crucial for detecting factors that alter drug exposure, their findings may not always translate to the oncology patient population. The underlying reasons are likely multifaceted but can at least be attributed to differences in study populations and study designs. To address this potential discrepancy, the relevance of intrinsic and extrinsic factors identified in HV studies should be validated in patient trials by assessing their impact on drug PK, safety, and efficacy. These examples highlight the importance of a holistic approach that integrates data from both HV and patient studies. Such a framework ensures more informed decisions about the safe and effective use of oncology drugs in the real‐world settings.

Study Highlights.

• What is the current knowledge on the topic?

  • ○
    Phase I clinical pharmacology studies in healthy volunteers are routinely conducted in oncology drug development to assess the effects of intrinsic or extrinsic factors on drug pharmacokinetic properties to inform prescribing decisions.
• What question did this study address?

  • ○
    In oncology, healthy volunteer clinical pharmacology studies and patient trials assessing how a single intrinsic or extrinsic factor affects pharmacokinetics may yield different results. What underlying reasons could explain these discrepancies, and how should they inform prescribing decisions?
• What does this study add to our knowledge?

  • ○
    Drawing on three oncology case studies, we propose a holistic approach that integrates evidence from both healthy volunteer clinical pharmacology studies and patient trials to evaluate how intrinsic or extrinsic factors affect pharmacokinetics, safety, and efficacy, supporting more informed use of these drugs.
• How might this change clinical pharmacology or translational science?

  • ○
    The proposed framework mitigates the limitations of relying solely on healthy volunteer trial results and supports more informed decisions for the safe and effective use of oncology drugs in real‐world practice.

1. Introduction

Phase I clinical pharmacology (CP) studies to identify intrinsic or extrinsic factors (or covariates) that influence the pharmacokinetic (PK) variability of drugs are typically conducted in healthy volunteers (HVs). These factors can include drug–drug interaction (DDI), food effect (FE), renal/hepatic impairment (RI/HI) [1]. Since a drug's PK usually correlates with its safety and efficacy, the results of the HV CP studies provide crucial information for the drug product label. This information helps to understand how these factors may affect the drug's PK and the actions that should be taken to ensure its safe and effective use.

Phase I HV CP studies often aim to assess the effect of a single factor on drug PK, and typically employ a rigorous design. They are adequately powered and controlled to minimize variability, thus ensuring adequate sensitivity to detect any possible effect on the drug's PK. The results are then extrapolated to the intended patient population to determine the need for any dose adjustment or restrictions for the clinical use of the drug. Another approach, often used to gain additional insight into sources of PK variability, involves the use of population PK (popPK) analysis. This involves utilizing PK data from the target patient population, collected from multiple clinical studies, including pivotal Phase III trials, to determine the effect of covariates on the PK properties of a drug.

In oncology drug development, HVs are the standard study population for Phase I CP studies aimed at assessing the effects of PK covariates [2]. Recently, HVs have also been adopted in a few first‐in‐human trials for oncology therapies which have safer and more manageable toxicity profiles [3]. This review focuses on the former, analyzing the comparative utility of HV CP studies versus patient trials in quantifying the impact of intrinsic and extrinsic factors on drug PK.

Using three approved Novartis oncology drugs as examples (Table 1), we illustrate situations where conclusions drawn from HV PK studies did not align with the outcomes from clinical trials in patients with cancer. Subsequently, we propose a practical framework to reconcile this apparent contradiction. This could help assess the clinical relevance and true magnitude of the effect of the covariate on PK in the intended patient population, and guide proper clinical use and prescribing decisions of the drug.

TABLE 1.

Summary of examples showing deviation between data of healthy volunteer clinical pharmacology studies and outcomes of oncology patient trials.

Drug	Indication	Covariate factor	Study type	Study design	Study data	Regulatory recommendation &Product label
Ribociclib (Kisqali)	Hormone receptor–positive (HR+), human epidermal growth factor receptor–2 negative (HER2–) advanced breast cancer (ABC) or early breast cancer (EBC) in combination with endocrine therapy	Renal impairment	HV CP study	Phase I study of single 400 mg dose in subjects with varying degrees of RI (mild, moderate, severe) compared with matched subjects with normal renal function [5]	PK: Ribociclib exposure increased in subjects with mild (n = 8), moderate (n = 6), and severe RI (n = 7) compared with those with normal renal function (n = 14) by 80%, 79%, and 130%, respectively, for C max, and by 62%, 94%, and 167%, respectively, for AUCinf [5].	No dose adjustment is required in mild or moderate RI patients based on patient data. A reduced dose of 200 mg was recommended in patients with severe RI based on HV CP data [4].
			Patient trials	ABC: a phase Ib/II and three phase III studies in patients with ABC at the dose of 600 mg QD 3 week on/1 week off, and a phase I study in patients with advanced solid tumors or lymphomas [5]	PK: Ribociclib exposure (AUC, Cmax) in patients with mild/moderate RI was comparable with patients with normal renal function after single dose and at steady state based on a Phase I trial in advanced cancer patients and a Phase Ib/II and a Phase III trial in patients with ABC, as well as popPK analysis of pooled data of Phase I‐III trials in ABC or advanced cancer (n = 438, 488 and 113 for normal renal function, mild RI and moderate RI, respectively) [5, 6]; Efficacy: No differences in the estimated relative risk reduction in PFS in patients with mild RI (n = 341) (HR 0.57 [95% CI 0.45–0.72]) or moderate (n = 97) RI (HR 0.64 [95% CI 0.41–1.00]) versus patients with normal renal function (n = 510) (HR 0.58 [95% CI 0.48–0.70]) in patients with ABC; Safety: AESI profiles were generally consistent by renal function [normal renal function (n = 574), mild RI (n = 400), moderate RI (n = 130)] based on three Phase III trials and a Phase Ib/II trial in patients with ABC [5].
				EBC: pivotal Phase III trial in EBC patients at the dose of 400 mg QD 3 week on/1 week off [7]	PK: mild/moderate RI did not noticeably impact the exposure of ribociclib [7].
Ceritinib (Zykadia)	Patients with anaplastic lymphoma kinase (ALK)‐positive metastatic non‐small cell lung cancer (NSCLC) who have progressed on or are intolerant to crizotinib	DDI with PPI	HV CP study	Phase 1 study to assess the effect of steady‐state esomeprazole (Nexium 40 mg capsule) on the single dose PK of 750 mg oral dose of ceritinib in HVs [9]	PK: Co‐administration of a single 750 mg dose of ceritinib with esomeprazole 40 mg for 6 days decreased ceritinib AUC0–∞ by 76% and Cmax by 79% (n = 22) [9].	Use of PPIs and other ARAs is not restricted when coadministered with certinib based on patient data [8].
			Patient trials	Three studies in patients with ALK‐positive NSCLC treated at 750 mg daily provided the PK data to evaluate the effect of PPIs on ceritinib PK exposure along with efficacy data: 1. Global first‐in‐human, open‐label, phase 1 dose‐escalation, and expansion study (ASCEND‐1) includes patients who had previously received crizotinib as well as ALK inhibitor‐naïve patients 2. Global, single‐arm, open‐label phase 2 study (ASCEND‐2) includes patients who had previously received crizotinib 3. Global, single‐arm, open‐label phase 2 study (ASCEND‐3) includes ALK inhibitor‐naïve patients [9]	PK: Coadministration of a single ceritinib dose with PPIs resulted in modest effect on exposure: decrease of AUC by 30% (n = 22 and 48 for with and without PPI, respectively) and Cmax by 25% (n = 43 and 153 for with and without PPI, respectively). Upon daily dosing, there was lack of clinically meaningful effect by PPIs on steady‐state ceritinib Ctrough (n = 106 and 224 for with and without PPI, respectively); Efficacy: Similar ORR observed regardless of concomitant PPI usage [ASCEND‐1: n = 64 (with PPI) and 105 (without PPI), ASCEND‐2: n = 39 (with PPI) and 67 (without PPI); ASCEND‐1: n = 27 (with PPI) and 70 (without PPI)] [9]
Midostaurin (Rydapt)	Patients with newly diagnosed FLT3‐mutated AML in combination with standard induction and consolidation therapy, and with aggressive systemic mastocytosis, systemic mastocytosis with associated hematological neoplasm or mast cell leukemia	DDI with CYP3A4 inhibitor	HV CP study	Phase I parallel group study to evaluate the impact of the strong CYP3A4 inhibitor ketoconazole, administered to steady state, once daily at 400 mg for 10 days, on the PK of a single oral 50 mg dose of midostaurin in HVs [14]	PK: Midostaurin showed 10‐fold higher AUCinf and 1.8‐fold higher Cmax with ketoconazole coadministration (n = 36). AUCinf of CGP6222 increased by 3.5 fold while Cmax changed by 0.56 fold. AUC0‐t of CGP52421 increased only by 1.2 fold while Cmax changed by 0.49 fold [14].	Patients can be treated with midostaurin while on antifungal therapies (with caution). No midostaurin dose adjustment is needed based on patient data [10].
			Patient trials	A Phase I/II trial in patients with relapsed/refractory AML or high‐risk MDS of midostaurin at steady state (50 mg BID) in a small subset of patients without and then with itraconazole (100 mg BID) co‐administration [10]	PK: Cmax and AUCtau increased by only 1.49 and 1.63 fold, respectively, while steady‐state Ctrough increased by only 2.09 fold (n = 7) with itraconazole coadministration after multiple dose of midostaurin in AML patients. Cmax, AUCtau, and Ctrough of CGP6222 showed little changes (by 0.94, 0.86, and 1.16‐fold respectively). Similarly, Cmax, AUCtau, and Ctrough of CGP52421 showed little change (by 1.22, 1.20, and 1.33‐fold respectively) [10].
				A Phase III trial of midostaurin (or placebo) at a dose of 50 mg BID on days 8–21 of each cycle in addition to chemotherapy for standard induction (daunorubicin and cytarabine) and consolidation in patients aged 18–59 years with newly diagnosed FLT3‐mutated AML; note: use of strong CYP3A4 inhibitors was not restricted [12].	PK: No clinical meaningful impact on PK exposure of midostaurin (1.44‐fold increase in Ctrough) based on pooled data (n = 200). Minor changes in the trough exposure of the metabolites were observed (by 0.95‐fold for CGP6222 and by 1.11‐fold for CGP52421); Safety: Grade 3/4 AE frequency and time to first occurrence with (n = 214) or without (n = 131) concomitant strong CYP3A4 inhibitor showed no clinically relevant pattern [13].

Abberaviations: ABC: advanced breast cancer; AE: adverse event, AESI: adverse event of special interest, AML: acute myeloid leukemia; ARA: acid‐reducing agent AUC: area under the curve; AUC_inf: AUC from time zero to infinity; AUC_tau: AUC during a dosing interval; BID: twice daily; C_max: maximum drug concentration; C_trough: trough drug concentration; CYP: cytochrome P450; DDI: drug–drug interaction; HV: healthy volunteer; HR: hazard ratio; MDS: Myeloid Dysplastic Syndrome; NCA: non‐compartmental analysis; ORR: overall response rate; PFS: progression‐free survival; PK: pharmacokinetic(s); PPI: proton pump inhibitor; QD: once daily; RI: renal impairment.

2. Case Studies

2.1. Ribociclib

Ribociclib is approved to treat patients with hormone receptor–positive (HR+), human epidermal growth factor receptor–2 negative (HER2–) advanced or early breast cancer (ABC or EBC). The recommended starting dose is 600 and 400 mg for ABC and EBC patients, respectively, once daily for 21 days followed by 7‐day off treatment [4]. Ribociclib is cleared primarily via hepatic elimination, with renal excretion accounting for only 7% of total excretion. Therefore, renal function is not expected to substantially affect ribociclib PK.

In a Phase I CP study, ribociclib exposure increased in non‐cancer subjects with mild, moderate, and severe RI compared to those with normal renal function (maximum concentration (C_max) by 80%, 79%, and 130%, respectively, and area under the curve (AUC) by 62%, 94%, and 167%, respectively) following a single dose of 400 mg, suggesting a strong impact of RI on ribociclib PK [5]. However, in three separate trials in the patient population of ABC or advanced cancer, ribociclib showed comparable AUC and C_max between patients with mild/moderate RI and those with normal renal function for both single‐dose and steady‐state PK [5]. PopPK analysis based on data from cancer patients also showed similar results [6]. Furthermore, analysis of data from three pivotal Phase III trials indicated no differences in efficacy (progression‐free survival) or safety (adverse events of special interest) across renal functions in ABC patients [5] (Table 1). Similarly, data from the pivotal Phase III trial in EBC patients indicated that mild/moderate RI did not noticeably impact the exposure of ribociclib [7].

Given the closeness of the patient data to real‐world settings, the robustness of patient PK data, and the favorable benefit–risk profiles established by the efficacy and safety data in patients, the same ribociclib starting dose as used for patients with normal renal function was recommended for patients with mild/moderate RI. Patients with severe RI were not included in the clinical trials. To be on the conservative side, a lower starting dose of 200 mg was recommended for patients with severe RI, based on the increase in exposure (130% for C_max and 167% for AUC) in the CP study and the comorbidity of patients with both cancer and severe RI [4, 5, 7] (Table 1).

2.2. Ceritinib

Ceritinib is approved to treat patients with anaplastic lymphoma kinase (ALK)‐positive metastatic non‐small cell lung cancer (NSCLC) who have progressed on or are intolerant to crizotinib [8]. Ceritinib demonstrates pH‐dependent solubility in vitro, with its solubility decreasing as pH increases [9]. Therefore, acid reducing agents like proton pump inhibitors (PPIs), which cause an increase in gastric pH, could reduce its bioavailability.

A Phase I HV DDI study demonstrated a substantial decrease in ceritinib exposure (C_max by 79% and AUC by 76%) with concomitant dosing of esomeprazole, a prototypical PPI [9] (Table 1). This substantial reduction was not replicated in patient trials, which showed similar ceritinib exposure at steady state with or without concomitant PPIs or a more modest decrease (25% for C_max to 30% for AUC_0–24h) following its single dose administration with concomitant PPIs [9] (Table 1). The efficacy (overall response rate) of ceritinib in the three cancer trials remained unaffected by PPI use [9] (Table 1). Considering all the evidence, there was no clinically meaningful DDI between PPIs and ceritinib, and the use of acid reducing agents is not restricted with ceritinib [8].

2.3. Midostaurin

Midostaurin is approved to treat patients with newly diagnosed FLT3‐mutated acute myeloid leukemia (AML) at 50 mg twice daily (BID) on days 8–21 of each cycle, and patients with aggressive systemic mastocytosis, systemic mastocytosis with associated hematological neoplasm, or mast cell leukemia at 100 mg BID [10]. Midostaurin is metabolized by cytochrome P450 (CYP)3A4 to form two active metabolites, CGP52421 and CGP6222. Midostaurin and both metabolites are substrates, reversible and time‐dependent inhibitors, and inducers of CYP3A4, displaying non‐linear PK [11].

In a small subset of patients with AML (n = 7), coadministration of the strong CYP3A4 inhibitor itraconazole with multiple doses of midostaurin showed that the trough concentration (C_trough) of midostaurin, CGP62221 and CGP52421 only increased by 2.1‐, 1.2‐ and 1.3‐fold, respectively, compared to midostaurin alone [10] (Table 1). Based on these results, coadministration of strong CYP3A inhibitors were not excluded in the pivotal Phase III study [12]. From the PK data collected in this study (n = 200), a modest 1.44‐ and 1.11‐fold increase was observed in the exposure of midostaurin and CGP52421, respectively, with concomitant usage of strong CYP3A4 inhibitors, and no exposure increase was observed for CGP62221 [13] (Table 1). In contrast, a Phase I DDI study in HVs showed that co‐administration of the strong CYP3A4 inhibitor ketoconazole with a single dose of 50 mg midostaurin resulted in approximately 10‐ and 3.5‐fold increase in midostaurin and CGP62221 exposure (AUC_inf), respectively, while AUC_0‐t of CGP52421 increased only 1.2 fold [14] (Table 1). Overall, as the effect of strong CYP3A4 inhibitors is considered modest in the pivotal Phase III trial, no midostaurin dose adjustment is needed, but caution/alternative therapies to strong CYP3A4 inhibitors is advised [10].

3. Factors Leading to Differences in PK Outcomes Between HV Studies and Patient Trials

Phase 1 CP studies in HVs have an established role in identifying factors that could alter a drug's PK properties. These factors often lead to actions such as adjusting dosages or placing contraindications to minimize the risks of toxic or subtherapeutic exposure in the target patient population [15]. The question then arises whether the data from these HV CP studies should be the sole basis for guidelines on drug administration in cancer patients, particularly regarding how the drug might be administered under the influence of same extrinsic or intrinsic factors. In the three examples presented, the substantial change in PK exposure seen in HV CP studies due to DDI or RI was absent in cancer patients. What could cause such deviations? Several potential explanations exist, but likely be ascribed to two key factors: differences in the study population and study design (Table 2).

TABLE 2.

Comparison of healthy volunteer clinical pharmacology studies and pivotal patient clinical trials for oncology drugs.

		Healthy volunteer clinical pharmacology studies	Pivotal patient clinical trials
Study population	Population type	Healthy volunteers or non‐cancer subjects with standardized demographics	Cancer patients of the target indication
	Health condition	Clean medical history (except liver and renal impairment for dedicated organ impairment studies)	Heavily pretreated, presence of comorbidities, chronic inflammation due to tumor burden, and potential impairment of renal and hepatic functions
	Concomitant medications	Prohibited	Not prohibited except those predefined in the protocol, concomitant medication profiles (including comedication types, dose regimens and treatment duration) vary between patients
Study design	Phase and primary objective	Phase I study to evaluate the effect of a single intrinsic or extrinsic covariate factor on drug PK in HVs or non‐cancer subjects	Phase II or Phase III study to evaluate the efficacy and safety of a drug in the target cancer patient population
	Design	Cross‐over, or randomized parallel in subjects matched by demographic factors to study the covariate effect	Patients randomized to treatment arms to study efficacy and safety
	Dose and duration of treatment	Dose level may be lower than therapeutic dose; Single‐dose or short‐term treatment with multiple doses	Therapeutical dose level; Chronic dosing until disease progression, death or censoring
	Enrollment criteria	Strict restrictions and exclusions	Less restriction for covariate factors, representative of real‐world setting
	Sample size	Small, powered to assess the covariate effect	Large, powered to assess efficacy and safety
	PK collection	Intensive PK sampling	Sparse PK sampling
Data analysis	Dataset	Single‐study data	Single‐trial data or pooled data from multiple trials
	Methodology	Dense PK data analyzed using NCA to generate PK parameters and statistical assessment of covariate effects, aligning with the study objectives	Population PK analysis of mostly sparse PK data to determine significance of covariate effect; if available, intensive PK data from a subset of patients analyzed by NCA to assess covariate effect

Abbreviations: HV: healthy volunteer; NCA: non‐compartmental analysis; PK, pharmacokinetics.

3.1. Study Population

Phase 1 CP studies are usually conducted in HVs due to their relatively clean medical history and lack of co‐morbidities. The demanding nature of assessments, such as frequent blood sampling, requirement of strict restrictions (e.g., overnight fasting before dose administration, prohibition of concomitant medications) and complex study designs (e.g., crossover), also make HVs more suitable for the CP studies. These attributes aim to minimize the inter‐ and intra‐subject PK variability so that the effect of the factor to be investigated in the HV studies are not obscured by PK variabilities from other sources. In contrast, patient trials recruit population that will benefit from the drug's intended use and thus would likely have participants suffering from multiple co‐morbidities such as RI or HI, especially in oncology indication. For ethical considerations, the use of concomitant medication in patient trials will be less restrictive compared to HV trials (Table 2).

Cancer patients often suffer from chronic inflammation due to tumor burden and may have multiple co‐morbidities [16], which in turn can modulate plasma proteins, metabolizing enzymes and transporters responsible for the drug's absorption, distribution, metabolism and elimination [17, 18, 19]. As a result, cancer patients exhibit higher PK variability compared to HVs. In the case of ribociclib, RI was not expected to have a substantial effect on its PK as it is eliminated primarily via hepatic pathway. Nonetheless, a standalone RI study showed that ribociclib exposure (AUC) increased in non‐cancer subjects with mild, moderate, and severe RI compared with those with normal renal function, by 62%, 94%, and 167%, respectively. This finding is perhaps consistent with emerging evidence suggesting that RI can modify metabolic enzymes, transporters, and plasma proteins, thereby affecting drug PK exposure [20, 21]. However, no such increase in ribociclib exposure was noted in patients with mild or moderate RI in the Phase I to Phase 3 trials of ribociclib in breast cancer patients [5]. This suggests that in cancer patients, the disease itself may already reduce drug clearance, leaving less room for further inhibitory effects from RI.

For midostaurin, the divergence from HV DDI to patient outcome was more dramatic. It showed that in HVs there was a 10‐fold increase in midostaurin exposure due to the inhibitory effect of ketoconazole, whereas there was a modest increase in exposure of up to 2‐fold in AML patients due to concomitant administration of strong CYP3A4 inhibitors [13, 14]. It should be noted that both the parent drug (midostaurin) and its two active metabolites are substrates, inducers and reversible inhibitors of CYP3A4 [11]. Due to the complex metabolic profiles, with non‐linear and time‐dependent PK of the parent drug and active metabolites, the resulting net effect on clearance of midostaurin could be different in cancer patients upon chronic dosing versus HVs upon single dosing. The inhibited state of the drug‐metabolizing enzymes or transporters in cancer patients could in turn reduce the susceptibility of the drug to further reduction of its clearance, resulting in decreased sensitivity to DDI effect.

3.2. Study Design

The key difference in study design between Phase 1 HV CP studies and patient trials with PK assessments include the differences of study condition, dose level, dosing frequency, and the rigors of PK data collection (Table 2). HV CP studies are conducted under restricted and well‐controlled condition, which may not reflect the actual clinical setting. One restriction commonly used in HV CP studies is the requirement of overnight fasting of around 10 h before and 4 h after drug administration [22]. In patient trials as well as real‐world practice, if fasting is required, it would be much shorter duration (i.e., modified fasting where drug could be administered 1 or 2 h before and 2 h after a meal). For drugs like ceritinib, whose bioavailability is increased with meal [23], the shorter duration of fasting could have diminished the effect of PPIs in reducing ceritinib bioavailability in patient trials, whereas the much longer period of fasting (10 + 4 h) in the HV study provided conditions to study the maximum effect of esomeprazole in reducing ceritinib exposure.

3.2.1. Dosing

Difference in treatment duration and dose level could be another contributing factor. Unlike patient trials, HV CP studies do not employ chronic dosing and sometimes, especially for oncology drugs, use dose lower than therapeutic dose, to protect HVs from any untoward side effects. In the case of midostaurin, the net effect of auto‐inhibition of CYP3A4 enzymes over time resulted in its reduced clearance upon chronic dosing. This may explain the more moderate impact of strong CYP3A inhibitors in reducing clearance of midostaurin and its two active metabolites in the pivotal trial in AML patients, compared to the single‐dose HV DDI study.

For drugs with nonlinear PK, the question is open if the magnitude of a certain covariate effect (e.g., DDI or organ impairment) measured at single dose in HV would be similar to chronic dosing. The deviations in PK outcomes between single and multiple dosing have been observed previously for FE of lonafarnib where PK of lonafarnib was affected by food at single‐dose in HVs but not after multiple‐dose administration in patients with advanced cancer [24]. Similarly, FE on the PK of abiraterone, an oral drug for the treatment of metastatic castration‐resistant prostate cancer, was more pronounced in HVs following single dose compared to the effect in the target patients after repeat dosing [22].

3.2.2. Data Analysis

Due to the study design difference, methodology applied for PK data analysis could also be different between HV PK studies and patient trials (Table 2). In CP studies, dense PK samples are collected to capture the drug's PK profile allowing use of non‐compartmental analysis (NCA) to calculate the PK parameters of the observed data. The mean ratio of the PK parameters (often AUC and C_max) are then computed to determine the impact of the potential factor on the drug's PK. In patient trial though, PK samples are usually collected in a sparse manner, and popPK modeling is usually employed to describe the PK profile and evaluate the effect a potential covariate on PK parameters (e.g., clearance) based on a certain likelihood test. However, it should be noted that the three examples implemented dense sampling in the patient trials, allowing for the traditional NCA to be applied to derive the PK parameters.

3.2.3. Sample Size

The sample size of HV CP studies is much smaller than that of later‐phase patient trials, which also include populations that are more heterogenous. On the other hand, HV CP studies are statistically powered to measure the effect of a single variable using either cross‐over or parallel study design. The sample size difference, in combination with the methodology difference in data analysis, could also contribute to the difference in outcomes. For example, standalone DDI studies typically use a randomized, fixed sequence, multi‐period design as was the case in the ceritinib and midostaurin DDI studies. In contrast, for patient data, PK information from large sized patient trials is derived from existing treatment arms and analyzed by the presence or absence of the factor that is under investigation (often by popPK analysis) without consideration of randomization of the treatment arms for PK comparison. As a result, unexplained variabilities may influence the outcome in the patient trials.

3.2.4. Other Factors

As explained earlier, HV PK studies are conducted under well‐controlled conditions, minimizing multiple sources of variability in order to focus on a single intrinsic or extrinsic factor that could impact the drug's PK. In the example of ceritinib and midostaurin, the effect of a single perpetrator drug—esomeprazole or ketoconazole (for ceritinib or midostaurin DDI, respectively) was used to study the effect of a PPI or strong CYP3A4 inhibitor. In contrast, PPIs or CYP3A4 inhibitors of different types at different dose levels were used to assess their effect in the patient trials and may be more representative of a real‐world setting.

Other confounding factors such as demographic variability and less restrictive use of concomitant medications could further impact PK outcomes from patient trials. However, to what extent these multiple variables can influence PK and explain the discrepancy between the standalone HV PK studies and cancer patient trials remains unknown. Nevertheless, the PK information collected from patients, who are representative of the target population to be treated with the drug, cannot be disregarded.

4. Discussion and Conclusion

In small molecule drug development, early CP studies in HVs provide essential data to inform patient trials and prescribing labels. These studies provide an efficient means to evaluate the impact of individual factors on a drug's PK profile, contributing to informed decision‐making in drug development. It should be noted that for oncology drugs, the use of HVs in Phase 1 CP studies continues to be an important and valuable strategy to characterize intrinsic and extrinsic covariate effects on drug PK. Under certain situations, however, the findings of HV studies may not always translate correctly to specific patient populations, such as those with advanced cancer, potentially leading to unnecessary actions to adjust drug exposure. Therefore, for oncology drug development, a holistic framework integrating both HV and patient data could provide valuable insights into the magnitude of the intrinsic or extrinsic covariate effects on drug PK in the intended patient population, ultimately aiding informed prescribing decisions, as illustrated in Figure 1.

FIGURE 1.

A Holistic Framework for Evaluating Intrinsic and Extrinsic Covariate Effects on Drug PK in Oncology Drug Development.

In patient trials, it would be beneficial to conduct robust PK assessments to identify the influence of multiple factors on drug exposure in clinically relevant settings. We also recommend adopting less restrictive enrollment criteria, such as permitting broader use of concomitant medications or including patients with mild to moderate organ impairment, provided the absence of significant risk of substantially altered drug exposure. The FDA also advocated for broader eligibility criteria in early‐stage dose‐finding trials in oncology [25]. When HV and patient data show significant differences in the impact of PK covariates, a thorough analysis of patient safety and efficacy data becomes crucial. Sufficient PK assessments and less restrictive enrollment criteria were employed in the three examples. In the case of ribociclib, approximately 60% of patients in the Phase 3 trials had mild or moderate RI, and with robust PK collection implemented in one of the Phase 3 trials [5], the data allowed thorough assessment of the impact of RI on ribociclib PK and benefit–risk profile in the patient population. If sufficient representation of patients with varying renal function and adequate PK data in clinical studies are available, analysis of patient data may eliminate the need for standalone RI studies. For ceritinib, patient trials included robust PK data collection without restricting PPI use [9], enabling evaluation of PPI effects on ceritinib PK and efficacy within the target population. Similarly, for midostaurin, antifungal azoles (strong CYP3A4 inhibitors) were allowed in the Phase 3 AML trial, with 60.8% of patients receiving these drugs alongside midostaurin [13]. This data was crucial for guiding its clinical use, as antifungal treatments are the standard of care for patients at high risk of fungal infections from chemotherapy.

In all the three examples, patient data overrode conclusions from HV trials, which would have suggested unnecessarily restrictive measures. A holistic approach, integrating covariate effects from both HV CP studies and patient trials, is crucial to address limitations of HV data alone and ensure informed decisions for the safe and effective use of oncology drugs in real‐world settings.

Conflicts of Interest

All authors are employees and own shares of Novartis.

Ji Y., Sechaud R., and Chakraborty A., “Challenges in Extrapolating Healthy Volunteer Pharmacokinetics to Oncology Populations: Advocating for a Holistic Perspective,” Clinical and Translational Science 18, no. 11 (2025): e70374, 10.1111/cts.70374.