Effects of Hepatic or Renal Impairment on Pharmacokinetics of Fruquintinib.

Introduction
Fruquintinib (FRUZAQLATM) is a highly selective, oral, small‐molecule tyrosine kinase inhibitor of all three vascular endothelial growth factor receptors (VEGFRs), ‐1, ‐2, and ‐3.
1
,
2
Fruquintinib was initially approved in China for the treatment of metastatic colorectal cancer (mCRC) in patients who had failed two prior lines of systemic therapy.
3
,
4
It was then approved in the United States in November 2023 for the treatment of adult patients with mCRC who have been previously treated with fluoropyrimidine‐, oxaliplatin‐, and irinotecan‐based chemotherapy, an anti‐VEGF therapy, and, if RAS wild‐type and medically appropriate, an anti‐epidermal growth factor receptor (EGFR) therapy, at a dose of 5 mg once daily (QD) 3 weeks on/1 week off in a 28‐day cycle.
5
,
6
Subsequently, fruquintinib was approved in June 2024 by the European Commission
7
for the treatment of adult patients with mCRC who have been previously treated with available standard therapies, including fluoropyrimidine‐, oxaliplatin‐, and irinotecan‐based chemotherapies, anti‐VEGF and anti‐EGFR therapies, and who have progressed on or are intolerant to treatment with either trifluridine–tipiracil or regorafenib.
Fruquintinib was absorbed quickly after oral administration in patients with cancer; the median time to maximum plasma concentration (Tmax) of fruquintinib was approximately 2 h.
8
Fruquintinib exposure increased proportionally with increasing dose over the dose range of 1‐6 mg.
9
Steady‐state concentrations were reached following 14 days of consecutive QD dosing, and fruquintinib accumulation at steady‐state was approximately 4‐fold relative to a single dose. The elimination half‐life (t1/2) of fruquintinib was approximately 42 h,
9
the weight‐normalized apparent clearance was approximately 0.17 mL/min/kg,
8
and protein binding in human plasma was approximately 95%.
1
In healthy participants, food and the co‐administration of the proton‐pump inhibitor rabeprazole resulted in similar systemic exposure compared with fasted conditions and fruquintinib alone, respectively, with no impact on the pharmacokinetics (PK) of fruquintinib.
10
The metabolism of fruquintinib is mediated by multiple enzymes, including cytochrome P450 (CYP) and non‐CYP450 enzyme systems. In vitro findings indicate that multiple CYP isoforms are involved in the metabolism of fruquintinib, with CYP3A4 demonstrating the most significant contribution compared with other isoforms.
1
The CYP450‐mediated metabolic pathway of fruquintinib to oxidative metabolites contributed approximately 30% of fruquintinib metabolism. M11, an N‐demethylation product of fruquintinib, was identified as a major circulating metabolite, accounting for approximately 17% of the total radioactivity in plasma.
11
Based on the relative potency (one tenth of fruquintinib) and systemic exposure, M11 is not expected to have a clinically meaningful contribution to the total pharmacological activity of fruquintinib.
11

Results from a mass balance study in healthy volunteers showed that after a single oral dose of 5 mg radiolabeled fruquintinib, approximately 60% of the dose was recovered in urine (0.5% unchanged), while 30% of the dose was recovered in feces (5% unchanged). The remainder of the radioactivity excreted in urine and feces consisted of metabolites.
11
These collective results indicate that the primary clearance mechanism for fruquintinib is hepatic metabolism. Therefore, impairment in hepatocellular function may alter the disposition of fruquintinib, which could potentially impact its safety and/or efficacy. Literature suggests that impaired renal function can disrupt drug metabolism and transport pathways in the liver and gut, potentially affecting drugs primarily cleared through nonrenal pathways.
12
In addition, the liver is the dominant site of metastases for patients with mCRC. At least 25% of patients with colorectal cancer develop liver metastases during the course of the disease,
13
leading to impaired hepatic function. Patients with mCRC are likely to experience a decline in renal function due to the side effects of chemotherapy treatments.
14
Results from a recent population PK analysis indicated that mild to moderate renal impairment and mild hepatic impairment had no clinically meaningful impact on fruquintinib PK.
15
However, given the important roles of hepatic and renal function in fruquintinib elimination, two separate studies were undertaken to evaluate the PK and safety of fruquintinib in participants with varying degrees of hepatic or renal function. The objective was to determine whether dosage adjustments were necessary in these specific populations.

Methods

Ethics
The protocol and consent form were approved by an institutional review board (Salus Independent Review Board, Austin, TX) before study initiation, and all participants signed informed consent forms prior to any study procedures. Both studies were performed in accordance with the requirements of the Declaration of Helsinki, the International Council for Harmonisation Guideline for Good Clinical Practice, and other applicable local laws and regulations. Both studies were conducted in the United States at Orlando Clinical Research Center in Florida and at Clinical Pharmacology of Miami in Florida. A third site (Alliance for Multispecialty Research in Knoxville, TN) was added for the renal impairment study. The design for both studies is summarized in Figure S1.

Study Design
Both studies included a screening phase, an open‐label treatment phase consisting of one single‐dose treatment period, and an end of study/early withdrawal phase. Screening occurred within 28 days of study drug administration. Day 1 was considered as the start of the treatment period. Participants were admitted to the study center at check‐in on Day −1, and discharged after completion of end of study procedures on Day 11, or at early withdrawal.

Hepatic Impairment Study (NCT05216367)
A reduced study design was implemented in the hepatic impairment study. Participants with hepatic impairment were enrolled based on the hepatic function classification determined by Child–Pugh scoring.
16
Initially, eight participants with moderate hepatic impairment (Child–Pugh Class B) were enrolled. A planned interim PK analysis was performed to evaluate the need to enroll participants with mild hepatic impairment (Child–Pugh Class A). Results of the planned interim PK analysis indicated that systemic exposure in participants with moderate hepatic impairment (dose corrected to 5 mg) were not more than 2‐fold higher than historical data from previous studies in participants with normal hepatic function. Thus, participants with mild hepatic impairment were not enrolled. Participants with normal hepatic function were enrolled and matched for sex, age (approximately within ±10 years), and body mass index (BMI, approximately within ±20%) from the mean of the moderate cohort.
The effect of severe hepatic impairment (Child–Pugh Class C) on the PK of fruquintinib was not incorporated into this study design, as it was unlikely that patients with cancer who have severe hepatic impairment would tolerate treatment with fruquintinib.
6

National Cancer Institute Organ Dysfunction Working Group (NCI ODWG) criteria were not utilized to determine eligibility for participants with hepatic impairment. However, the NCI ODWG classification for participants with hepatic impairment
17
was reported for exploratory analysis.

Renal Impairment Study (NCT05216354)
Eight participants with severe renal impairment and eight participants with moderate renal impairment were enrolled based on the renal function classification determined by Cockcroft–Gault equation.
18
Eight participants with normal renal function were enrolled after all other participants (severe and moderate renal impairment) had completed the study, and were matched for sex, as well as age (approximately within ±10 years) and BMI (approximately within ±20%) from the mean of the severe and moderate cohorts combined.
This study did not evaluate mild renal impairment as these patients were participating in the Phase 3 study conducted at the same time.

Dosing
The recently conducted population PK analysis indicated no clinically meaningful impact of mild to moderate renal impairment (creatinine clearance [CrCl] 30‐89 mL/min) and disease state (patients with cancer vs healthy volunteers) on fruquintinib PK.
15
Therefore, participants with normal hepatic function, normal renal function, and moderate renal impairment received a single dose of fruquintinib 5 mg. In order to account for a potential increase in fruquintinib PK exposure, participants with moderate hepatic impairment and severe renal impairment received a single dose of fruquintinib 2 mg. All participants fasted overnight for at least 10 h prior to fruquintinib dosing until 2 h after administration of fruquintinib. At 2 h postdose, participants consumed a low‐fat snack. Lunch was provided 4 h postdose.

Study Participants
In both studies, males and females with a BMI of 18‐40 kg/m2, body weight >50 kg, and 18‐75 years (hepatic impairment study) or 18‐82 years (renal impairment study) at screening were enrolled. Females were of nonchildbearing potential, and males agreed to use highly effective contraception.
In the hepatic impairment study, participants with a Child–Pugh score of 7‐9 were enrolled into the moderate hepatic impairment cohort. In the renal impairment study, participants with CrCl, as calculated by Cockcroft–Gault equation, of 15‐29 or 30‐59 mL/min were enrolled into the severe renal impairment or moderate renal impairment cohorts, respectively. Medical conditions related to organ impairment were allowed into the hepatic and renal impairment cohorts; participants with ascites but no paracentesis within 3 months of screening (moderate hepatic impairment only) and participants with diabetes who had the disease controlled were allowed. Concomitant medications were allowed in the organ impairment cohorts. Participants had to be on a stable dose of medication and/or treatment regimen for at least 2 weeks before dosing, as well as during the study. Participants were excluded if they had used strong CYP3A inhibitors or inducers of CYP3A within 2 weeks prior to Day 1 (or within 3 weeks prior to Day 1 in the case of St. John's wort) or at any time during the study period.
In both studies, participants with organ impairment were excluded if they had a clinically significant change in clinical condition within 30 days before screening. Participants were excluded from the study if they had evidence of clinically significant cardiovascular, gastrointestinal, respiratory, endocrine, hematological, neurological, psychiatric diseases, hepatic impairment (renal impairment study only), or Gilbert's syndrome and renal impairment (hepatic impairment study only). The participants were evaluated per Child–Pugh category classification as follows: encephalopathy (none: 1 point; grades 1 and 2: 2 points; and grades 3 and 4: 3 points); ascites (none: 1 point; slight: 2 points; and moderate: 3 points); bilirubin (under 2 mg/mL: 1 point; 2‐3 mg/mL: 2 points; and over 3 mg/mL: 3 points); albumin (greater than 3.5 mg/mL: 1 point; 2.8‐3.5 mg/mL: 2 points; and less than 2.8 mg/mL: 3 points); prothrombin time (PT; seconds prolonged) (less than 4 s: 1 point; 4‐6 s: 2 points; and over 6 s: 3 points); frequently international normalized ratios (INRs) were used as a substitute for PT (INR under 1.7: 1 point; INR 1.7‐2.2: 2 points; and INR above 2.2: 3 points); Child–Pugh A: 5‐6 points; and Child–Pugh B: 7‐9 points.
Participants with normal organ function were demographically similar to participants with organ impairment and matched approximately within ±10 years, mean BMI (approximately within ±20%), and sex to the mean of the moderate hepatic impairment cohort or the mean of the combined severe and moderate renal impairment cohorts.
Other key exclusion criteria included a known history of any gastrointestinal surgery or any condition possibly affecting drug absorption; diagnosed with acquired immune deficiency syndrome or human immunodeficiency virus, hepatitis B or C (not applicable to moderate hepatic impairment cohort); participation in a clinical study of other drug before screening, and the time since the last use of other study drug was less than five times the t1/2 or 4 weeks, whichever was longer, or the subject was currently enrolled in another clinical study; consumption of grapefruit, starfruit, Seville oranges, herbal preparations/medications including, but not limited to, kava, ephedra (ma huang), and Ginkgo biloba, dehydroepiandrosterone, yohimbe, saw palmetto, and ginseng within 7 days prior to the first dose.

Bioanalysis
Blood samples for determination of fruquintinib and metabolite M11 plasma concentrations were collected at pre‐dose and at 0.5, 1, 2, 3, 4, 6, 8, 12, 24, 36, 48, 60, 72, 96, 120, 144, 168, 192, 216, and 240 h after fruquintinib dosing. Plasma samples were analyzed using a validated liquid chromatography with tandem mass spectrometry assay method with an analytical range of 1.00 ng/mL (lower limit of quantitation) to 750 ng/mL (upper limit of quantitation). Quality control (QC) samples at four concentrations (3, 30, 300, and 600 ng/mL) were assayed along with study samples. In the hepatic impairment study, inter‐run variability was ≤8.9% and ≤13.1% for fruquintinib and M11, respectively, and the bias QC sample concentrations deviated by ±2.0% and ±3.0% from the nominal concentrations for fruquintinib and M11, respectively. In the renal impairment study, inter‐run variability was ≤9.4% and ≤11.9% for fruquintinib and M11, respectively, and the bias QC sample concentrations deviated by ±2.7% and ±4.3% from the nominal concentrations for fruquintinib and M11, respectively.
Plasma protein binding is often altered in patients with organ impairment. The unbound drug concentration is generally believed to determine the rate and extent of delivery to the sites of action. At concentrations of 1, 3, and 10 µM, the plasma protein binding of fruquintinib was 95.4%, 95.3%, and 95.3% in human plasma, respectively. The protein binding of fruquintinib is therefore independent of concentration as it remained constant within the range of 1‐10 µM (390‐3900 ng/mL). A blood sample was collected 4 h postdose, around the time of fruquintinib maximum plasma concentration (Cmax) for plasma protein binding assessments, including fraction unbound and concentration levels for human serum albumin (HSA), α‐1 amino acid glycoprotein (AAG), and total protein. The fraction unbound fruquintinib and M11 was determined using high‐throughput equilibrium dialysis coupled with a qualified liquid chromatography with tandem mass spectrometry with an analytical range of 0.100‐75.0 ng/mL. In the hepatic impairment study, QC samples at four concentrations (3, 25, 60, and 600 ng/mL) were assayed along with study samples. Inter‐run variability was ≤6.1% and ≤7.6% for fruquintinib and M11, respectively, and the bias QC sample concentrations deviated by ±2.2% and ±3.8% from the nominal concentrations for fruquintinib and M11, respectively. In the renal impairment study, QC samples at three concentrations (3, 25, and 60 ng/mL) were assayed along with study samples. Inter‐run variability was ≤8.4% and ≤12.5% for fruquintinib and M11, respectively, and the bias QC sample concentrations deviated by ±2.4% and ±7.0% from the nominal concentrations for fruquintinib and M11, respectively.

Pharmacokinetic and Statistical Assessments
PK parameters were determined by noncompartmental analysis using Phoenix® WinNonlin® version 8.3 (Certara, L.P. Princeton, NJ) and included Cmax, Tmax, area under the plasma concentration (AUC) from 0 to time of the last measurable concentration (AUC0‐t), AUC from 0 to infinity (AUC0‐inf), half‐life (t1/2), and the metabolite‐to‐parent ratios for Cmax (MPR Cmax), AUC0‐t (MPR AUC0‐t), and AUC0‐inf (MPR AUC0‐inf). Apparent oral clearance (CL/F) and apparent volume distribution (Vz/F) were also estimated for fruquintinib. Fruquintinib demonstrates linear PK over the dose range of 1‐6 mg and low inter‐individual variability (26%) in apparent oral clearance,
15
which supported dose normalization of PK parameters therefore to 5 mg for participants with organ impairment who received a single dose of fruquintinib at 2 mg. Following determination of fraction unbound, Cmax unbound (Cmax_unb), AUC0−t unbound (AUC0−t_unb), AUC0−inf unbound (AUC0‐inf_unb), and CL/F unbound were also estimated.
In both studies, the sample size of eight participants per cohort complies with Food and Drug Administration guidelines for hepatic (FDA 2003)
19
and renal (FDA 2024) impairment.
18
Based on data from previous single‐dose studies, the between subject CV% is assumed to be 27% and the true geometric mean ratio (GMR) of participants with organ impairment to participants with normal organ function is assumed to be 1. Given a sample size of eight evaluable participants per cohort, it is estimated that the probability for the 90% confidence intervals (CIs) of GMR to be within the no‐effect boundaries of 60%‐140% for the GMR of participants with organ impairment to participants with normal organ function is 75.6%. The power calculations are based on two 1‐sided tests at the 0.05 level of significance, equivalent to a 90% CI. Anticipating the need for log transformation of PK parameters, evaluation of acceptance criteria that are symmetrical on the log‐scale, 60%‐167%, were performed. These demonstrated that a sample size of eight participants per cohort would provide approximately 0.89‐1 probability that the 90% CI for the GMR would not exceed the 60%‐167% limits if the organ impairment cohort conferred no more than a minimal increase in drug exposure.
All statistical analyses were performed using SAS® version 9.4 or higher (SAS Institute, Cary, NC). The ln‐transformation PK parameters AUC0‐t, AUC0‐inf, and Cmax of fruquintinib and M11 were analyzed using a one‐way ANOVA model with cohort as the fixed effect. Geometric means were calculated by back transformation (exponent) of means, and the GMRs were calculated by back transformation (exponent) of the means difference. The 90% CI for the GMR was calculated by back transformation of the 90% CI for the mean difference. The GMR between each organ impairment test cohort and their matched normal organ function cohort was calculated, and 90% CIs of the GMRs, consistent with both 1‐sided tests, were provided for each comparison.
The comparison of interest was PK in participants with hepatic impairment, based on Child–Pugh classification (test) versus PK in participants with normal hepatic function (reference), or PK in participants with either severe or moderate renal impairment (test) versus PK in participants with normal renal function (reference). Similar analysis was conducted based on the unbound PK parameters of fruquintinib for exploratory purposes.

Graphical Analysis of Hepatic and Renal Function
Correlations between the PK parameters (dose normalized and Cmax and AUC0−inf) of fruquintinib and M11 versus hepatic function based on Child–Pugh Criteria or on NCI ODWG classification were evaluated graphically via scatterplots. Similarly, correlations between the PK parameters (dose normalized and Cmax and AUC0‐inf) of fruquintinib and M11 versus renal function, based on the Cockcroft–Gault equation, were evaluated graphically via scatterplots.

Results

Demographics
The demographics and baseline characteristics for both studies are summarized in Table 1. In both studies, the age, BMI, and sex of participants with normal organ function were similar to the mean of the organ impairment cohorts as a result of matching requirements implemented. In the hepatic impairment study, most participants were White, half were male, and less than half (∼44%) were Hispanic or Latino. In the renal study, most participants were White males and half were Hispanic or Latino.

Hepatic Impairment Study

Fruquintinib Pharmacokinetics
A single oral dose of fruquintinib 2 mg was administered to eight participants with moderate hepatic impairment and a single oral dose of fruquintinib 5 mg was administered to eight participants with normal hepatic function. For comparison purposes, all data were dose normalized to 5 mg. Fruquintinib plasma concentration–time profiles are shown in Figure 1a, and plasma PK parameters are shown in Table 2.
Peak concentrations were achieved slightly earlier in the moderate hepatic impairment cohort (1.5 h) compared with the normal hepatic function cohort (3 h); however, hepatic impairment did not affect dose‐normalized peak concentrations levels and the linear decline (Figure 1a). Mean t1/2 of fruquintinib was comparable with 47.3 h in moderate hepatic impairment versus 45.1 h in normal hepatic function (Table 2).
Systemic exposure to fruquintinib, as assessed by Cmax, AUC0‐t, and AUC0‐inf, was similar between participants with moderate hepatic impairment and participants with normal hepatic function. The GMRs of participants with moderate hepatic impairment to normal hepatic function for Cmax, AUC0‐t, and AUC0‐inf were 1.04, 0.89, and 0.91, respectively. The 90% CIs (lower to upper) contained unity for Cmax (0.87‐1.24), AUC0‐t (0.64‐1.23), and AUC0‐inf (0.66‐1.26). CL/F in participants with moderate hepatic impairment was 17.4 mL/min, compared with 15.3 mL/min in participants with normal hepatic function. The Vz/F of fruquintinib was 64.8 L in participants with moderate hepatic impairment and 56.1 L in participants with normal hepatic function (Table 2).
Plasma protein binding of fruquintinib was similar in participants with moderate hepatic impairment and matched participants with normal hepatic function. The mean fraction unbound of fruquintinib was 6.36% and 5.91% in participants with moderate hepatic impairment and participants with normal hepatic function, respectively. Dose‐normalized PK parameters corrected by fraction unbound (Cmax_unb, AUC0‐t_unb, and AUC0‐inf_unb) were comparable across both cohorts. Measured protein concentrations (AAG, HSA, and total protein) were also comparable between the two cohorts (Table 3).

Metabolite M11 Pharmacokinetics
The formation of metabolite M11 was comparable, albeit slightly lower, in the participants with moderate hepatic impairment compared with participants with normal hepatic function (Figure 1b). As shown in Table 4, Tmax was slightly longer in participants with moderate hepatic impairment (72 h) as compared with matched participants with normal hepatic function (66 h). Mean t1/2 of M11 was longer in participants with moderate hepatic impairment (124 h) versus participants with normal hepatic function (73.1 h) (Table 4).
Dose‐normalized exposure parameters of M11 Cmax, AUC0‐t, and AUC0‐inf were similar with moderate hepatic impairment to normal hepatic function ratios for Cmax, AUC0‐t, and AUC0‐inf were 0.83, 0.76, and 0.99, respectively. The lower to upper 90% CIs contained unity for Cmax (0.61‐1.14), AUC0‐t (0.46‐1.27), and AUC0‐inf (0.59‐1.65). MPRs were similar; although MPR Cmax was slightly lower in the moderate hepatic impairment cohort, the MPR AUCs were comparable across both cohorts (Table 4).
Following equilibrium dialysis, receiver sample M11 concentration values were below the limit of quantification for most samples. As such, the M11 fraction unbound was not reportable in participants with moderate hepatic impairment, but was evaluable in three of the eight participants with normal hepatic function, with a mean value of 5.27% (Table 4).

Graphical Analysis of Hepatic Function
No notable relationship was observed between fruquintinib PK parameters (dose‐normalized Cmax, AUC0‐t, and AUC0‐inf) and hepatic function as assessed by Child–Pugh score, albumin, PT, INR, and bilirubin according to the Child–Pugh classification and as assessed by bilirubin and aspartate aminotransferase (AST) according to the NCI ODWG criteria.

Renal Impairment Study

Fruquintinib Pharmacokinetics
The dose‐normalized concentration–time profiles were comparable across all cohorts (Figure 2a) following a single oral dose of fruquintinib 2 mg in eight participants with severe renal impairment and a single oral dose of fruquintinib 5 mg in eight participants with moderate renal impairment or eight participants with normal renal function. Fruquintinib absorption was rapid, with a median Tmax of 2 and 2.5 h in the renal impairment cohorts and the normal renal function cohort, respectively, followed by a linear decline. The mean t1/2 of fruquintinib was slightly longer in the renally impaired cohorts (55.3 h in severe renal impairment and 49.1 h in moderate renal impairment) compared with the normal renal function cohort (43.5 h, Table 2).
Systemic exposure to fruquintinib was similar between participants with severe and moderate renal impairment compared to participants with normal renal function. The GMR of participants with severe renal impairment compared with normal renal function for Cmax, AUC0‐t, and AUC0‐inf were 0.89, 0.97, and 1.01, respectively. The 90% CIs (lower and upper) contained unity for Cmax (0.78‐1.03), AUC0‐t (0.83‐1.14), and AUC0‐inf (0.85‐1.19). The GMR of participants with moderate renal impairment to normal renal function for Cmax, AUC0‐t, and AUC0‐inf were 0.95, 1.06, and 1.07, respectively. The 90% CIs (lower to upper) contained unity for Cmax (0.78‐1.15), AUC0‐t (0.89‐1.26), and AUC0‐inf (0.89‐1.28). The CL/F was comparable in participants with severe renal impairment (15.0 mL/min) and in participants with moderate renal impairment (14.2 mL/min) compared with participants with normal renal function (15.0 mL/min). The Vz/F of fruquintinib decreased slightly with increasing renal function across the cohorts (Vz/F of 69.8 L in participants with severe renal impairment, 59.4 L in participants with moderate renal impairment, and 55.9 L in participants with normal renal function as shown in Table 2).
Plasma protein binding of fruquintinib was similar in participants with severe renal impairment and moderate renal impairment compared with cohort‐matched participants with normal renal function. The mean fraction unbound of fruquintinib was 7.36%, 7.19%, and 7.24% in participants with severe renal impairment, participants with moderate renal impairment, and participants with normal renal function, respectively. Dose‐normalized PK parameters of fruquintinib corrected by fraction unbound (Cmax_unb, AUC0‐t_unb, and AUC0‐inf_unb) were similar across all cohorts. Measured protein concentrations (AAG, HSA, and total protein) were also comparable across the three cohorts (Table 3).

Metabolite M11 Pharmacokinetics
The formation of metabolite M11 was slightly higher in participants with normal renal function compared with participants with moderate and severe renal impairment (Figure 2b). Median Tmax of M11 was longer in participants with severe or moderate renal impairment (72 h for each cohort) compared with matched participants with normal renal function (54 h). Mean t1/2 of M11 was longer in participants with severe renal impairment (127 h) and moderate renal impairment (80.1 h) compared with those with normal renal function (70 h, Table 4).
Formation of M11 decreased with decreasing renal function. The severe renal impairment to normal renal function ratios (GMR) for dose‐normalized Cmax, AUC0‐t, and AUC0‐inf were 0.53, 0.45, and 0.78, respectively. The lower to upper 90% CIs did not contain unity for Cmax (0.38‐0.74), AUC0‐t (0.27‐0.76), and AUC0‐inf (0.62‐0.98). However, MPRs were similar across both cohorts. The moderate renal impairment to normal renal function ratios (GMR) for Cmax, AUC0‐t, and AUC0‐inf were 0.68, 0.72, and 0.75, respectively. The lower to upper 90% CIs did not contain unity for Cmax (0.54‐0.86), AUC0‐t (0.56‐0.91), and AUC0‐inf (0.59‐0.95). Accordingly, MPRs were lower with decreasing renal function. MPR Cmax and AUC values for the severe and moderate renal impairment cohorts were lower compared with normal renal function (Table 4).
Fraction unbound of M11 was not reportable in any of the participants with severe renal impairment. The mean fraction unbound of M11 based on the results from four out of the eight participants with moderate renal impairment was 8.94%. Fraction unbound of M11 was evaluable in one of the eight participants with normal renal function, with a fraction unbound of 6.24%. In general, M11 fraction unbound data were limited (Table 4).

Graphical Analysis of Renal Function
There was no notable relationship between fruquintinib PK parameters (dose‐normalized Cmax, AUC0‐t, and AUC0‐inf) and renal function as assessed by CrCl (mL/min) when comparing severe renal impairment or moderate renal impairment to normal renal function (Figure S2).

Safety and Tolerability
In the hepatic impairment study, a total of five treatment emergent adverse events (TEAEs) were reported in 4 of the 16 participants (25.0%). All five TEAEs were reported in participants with moderate hepatic impairment. Three (18.8%) participants experienced treatment‐related TEAEs. One of the 8 (12.5%) participants with moderate hepatic impairment experienced dizziness, which was assessed as grade 1 (mild) in intensity, nonserious, and related to fruquintinib treatment. Three (37.5%) participants experienced headache, which was assessed as grade 1 (mild) in intensity, nonserious; two were considered related, and one was considered not related to the fruquintinib treatment. One (12.5%) subject experienced urinary tract infection, which was assessed as grade 2 (moderate) in intensity, nonserious, and considered not related to the fruquintinib treatment, and experienced headache, which was assessed as grade 1 (mild) in intensity, nonserious, and considered related to the fruquintinib treatment.
In the renal impairment study, a total of four TEAEs were reported in 3 of the 24 participants (12.5%). One of the eight participants (12.5%) with moderate renal impairment experienced hypoglycemia on two occasions, which was assessed as grade 3 (severe) in intensity, nonserious, and not related to fruquintinib treatment. One (12.5%) participant with moderate renal impairment experienced headache, which was assessed as grade 1 (mild) in intensity, nonserious, and not related to fruquintinib treatment. One of the eight participants (12.5%) with normal renal function experienced eye pruritis and periorbital edema, which was assessed as grade 1 (mild) in intensity, nonserious, and not related to fruquintinib treatment.
In both studies, all TEAEs resolved/recovered prior to completion of the study. There were no clinically significant changes in chemistry, hematology, urinalysis, and coagulation laboratory test results over time during the study. There were no clinically significant changes in postdose vital signs, postdose electrocardiogram results, or postdose physical examination findings compared with baseline that were attributable to fruquintinib.

Discussion

Hepatic Impairment
The liver is involved in the clearance of many drugs through a variety of metabolic pathways; alterations in the metabolic activities as a result of hepatic impairment can lead to drug accumulation.
20
Fruquintinib is primarily metabolized in the liver and mostly eliminated renally.
21
A previous drug–drug interaction study reported that co‐administration of fruquintinib with itraconazole, a strong CYP3A inhibitor, did not result in clinically meaningful change in fruquintinib exposure, while concurrent use of rifampin, a strong metabolic enzyme inducer, reduced fruquintinib exposure by approximately 65%.
22
Contributions of CYP3A4 and CYP2C9 were assumed to be 20% and 10%, respectively, based on a sensitivity analysis of preliminary PBPK modeling using data from this prior study.
Here we describe two studies, which were conducted to determine if participants with impaired liver or renal function may experience increased systemic exposure to fruquintinib compared to participants with normal organ function. Fruquintinib is not recommended for use in patients with severe hepatic impairment due to hepatotoxicity. These two studies employed a single‐dose design, as fruquintinib demonstrated linear and dose‐proportional PK over the dose range of 1‐6 mg after single‐ and multiple‐dose administrations in patients with solid tumors.
9
Population PK analysis revealed no meaningful difference in fruquintinib PK between healthy participants and patients with mCRC.
15
Hence, the outcomes from the single‐dose studies in healthy volunteers can be extrapolated to multiple‐dose scenarios in patients with mCRC.
When dose normalized to 5 mg, exposure parameters of fruquintinib Cmax, AUC0‐t, and AUC0‐inf were similar in the moderate hepatic impairment participants compared with participants with normal hepatic function, with a difference of less than 5% in Cmax. AUCs were approximately 10% lower in participants with moderate hepatic impairment compared with normal hepatic function. However, this is not considered clinically significant. Similarly, formation of metabolite M11 was comparable across both cohorts. While dose‐normalized M11 Cmax was 17% lower in the moderate hepatic impairment cohort, dose‐normalized M11 AUC0‐inf remained nearly the same (1% lower in moderate hepatic impairment) across both cohorts. The mean t1/2 of M11 appears to be longer in participants with moderate hepatic impairment. This may be due to moderate hepatic impairment‐reduced elimination of M11, leading to prolonged t1/2 of the circulating metabolite.
Plasma protein binding of fruquintinib was similar in participants with moderate hepatic impairment and matched participants with normal hepatic function. The mean fraction unbound of fruquintinib was 6.36% and 5.91% in participants with moderate hepatic impairment and participants with normal hepatic function, respectively. Dose‐normalized PK parameters (Cmax, AUC0‐t, and AUC0‐inf) of fruquintinib corrected by fraction unbound were comparable across both cohorts.
Understanding whether HSA or AAG is the predominant binding protein for a new molecular entity can be of fundamental importance when determining exposure at the intended target site.
3
,
23
Changes in these proteins in patients with cancer may require additional monitoring, diagnostic tests, or co‐medications to address possible changes in PK.
24
,
25
Mean alpha‐1‐acid glycoprotein (AAG) has been shown to be nearly 2.1‐fold higher in patients with cancer (patients with advanced or refractory solid tumors or lymphoma) compared with healthy participants while mean albumin was 10% lower among the patients with cancer compared with healthy participants.
26
To evaluate if fruquintinib binds predominantly to AAG or HSA, a separate in vitro study using equilibrium dialysis was conducted. The results indicated that fruquintinib binds to both AAG and HSA in vitro. At 0.1, 1, and 10 µM, the fraction unbound of fruquintinib in AAG was 35.6%, 35.8%, and 41.3%, respectively (data on file). At 0.1, 1, and 10 µM, the fraction unbound of fruquintinib in HSA was 5.65%, 6.35%, and 6.27%, respectively. Thus, fruquintinib binds primarily to HSA, and there was no concentration dependency for either protein tested. In this clinical study, measured protein concentrations (AAG, HSA, and total protein) were comparable between participants with moderate hepatic impairment and matched participants with normal hepatic function (Table 3). Measured protein concentrations were also comparable in the severe and moderate renal impairment cohorts compared to the matched normal renal function cohort.
In line with the absence of significant changes in PK parameters, there was no discernible correlation observed between fruquintinib dose‐normalized PK parameters and hepatic function as assessed by Child–Pugh score, albumin, PT, INR, and bilirubin, when comparing moderate hepatic impairment to normal hepatic function based on Child–Pugh classification. There was also no discernable correlation observed between the dose‐normalized PK parameters and hepatic functions as assessed by bilirubin and AST based on NCI ODWG classification.
Overall, PK results based on total and unbound parameters indicate that moderate hepatic impairment (Child–Pugh B) does not result in a clinically meaningful change in systemic exposure of fruquintinib and M11.

Renal Impairment
In the renal impairment study, participants with severe renal impairment had dose‐normalized fruquintinib Cmax that was approximately 11% lower compared with participants with normal renal function, however, AUCs were similar, with less than 3% difference. In participants with moderate renal impairment, fruquintinib systemic exposure was comparable to participants with normal renal function, with a difference less than 8% across Cmax, AUC0‐t, and AUC0‐inf.
M11 Cmax, AUC0‐t, and AUC0‐inf were approximately 47%, 55%, and 22% lower, respectively, in participants with severe renal impairment and approximately 32%, 28%, and 25% lower, respectively, in participants with moderate renal impairment compared with participants with normal renal function. Fruquintinib was extensively metabolized prior to excretion with approximately 60% of the dose recovered in urine as metabolites. The lower systemic exposure of M11 in participants with moderate to severe renal impairment compared to healthy participants may be attributed to uremia‐induced reduction in CYP3A4 activity, which resulted in decreased formation of metabolites.
27
This hypothesis is supported by a similar observation that co‐administration of fruquintinib with itraconazole, a CYP3A4 inhibitor, reduced M11 exposure by approximately 44% to 55% without affecting fruquintinib exposure.
22
Because M11 only has 10% of the pharmacological potency as fruquintinib, the extent of reduction in M11 exposure is not expected to be clinically significant.
Plasma protein binding of fruquintinib was similar in participants with severe renal impairment and moderate renal impairment compared with cohort‐matched participants with normal renal function. The mean fraction unbound of fruquintinib was 7.36%, 7.19%, and 7.24% in participants with severe renal impairment, participants with moderate renal impairment, and participants with normal renal function, respectively. Dose‐normalized PK parameters of fruquintinib corrected by fraction unbound (Cmax_unb, AUC0‐t_unb, and AUC0‐inf_unb) were comparable across all cohorts. Measured protein concentrations (AAG, HSA, and total protein) were also comparable across the three cohorts.
In line with the absence of significant changes in PK parameters, there was no notable relationship observed between fruquintinib PK parameters (dose‐normalized Cmax, AUC0‐t, and AUC0‐inf) and renal function as assessed by CrCl (mL/min) when comparing severe renal impairment or moderate renal impairment to normal renal function (Figure S2).
Overall, PK results based on total and unbound parameters indicate that severe renal impairment and moderate renal impairment did not result in a clinically meaningful change in systemic exposure of fruquintinib and M11.

Statistical Comparison of Cohorts and Analysis of Hepatic Function Using Child–Pugh Classification Scoring and NCI ODWG Criteria
The hepatic impairment study was designed according to the FDA and European Medicines Agency (EMA) guidance on the evaluation of the PK of medicinal products in patients with impaired hepatic function, which recommend using Child–Pugh classification as the most appropriate evaluation for hepatic impairment, but acknowledge that the Child–Pugh score is not specifically designed for assessing drug elimination.
28
However, until an optimal classification system is available, the FDA and EMA recognize the Child–Pugh classification as the most appropriate method of hepatic impairment classification and it is adequate to continue using the Child–Pugh classification in a dedicated hepatic impairment study.
29
As such, Child–Pugh scoring was utilized to classify participants with hepatic impairment and to determine eligibility for this study. However, in the clinical setting, the NCI ODWG criteria is also utilized to determine hepatic impairment in patients with cancer, yet enrollment into a hepatic impairment study using both criteria proves to be difficult as there is discordance between both methods. A concordance analysis between Child–Pugh and NCI ODWG classifications of hepatic impairment was conducted for exploratory purposes (Table S1). Similar to the observations reported in the literature,
30
severity of hepatic impairment was generally assessed as less when using NCI ODWG criteria compared with Child–Pugh scoring. Only one subject was classified to have moderate hepatic impairment and three participants were classified to have mild hepatic impairment based on the NCI criteria in the present study. An exploratory comparison between the mild (3 participants) and normal cohorts (12 participants) based on NCI ODWG classification showed no meaningful differences in the dose‐normalized exposures for fruquintinib and M11. This finding is consistent with the results based on population PK analysis.
15
No notable relationship was observed between fruquintinib PK parameters and hepatic function as assessed by bilirubin and AST when comparing mild hepatic impairment to normal hepatic function based on NCI ODWG classification.

Conclusion
Moderate hepatic impairment (Child–Pugh B) does not result in a clinically meaningful change in systemic exposure of fruquintinib and M11. Severe renal impairment (CrCl 15‐29 mL/min) or moderate renal impairment (CrCl 30‐59 mL/min) did not result in a clinically meaningful change in systemic exposure of fruquintinib and M11. In summary, our results support the fruquintinib dose of 5 mg daily, 3 weeks on and 1 week off and no adjustments are required in patients with moderate hepatic impairment or in patients with moderate to severe renal impairment.

Funding
This work was supported by HUTCHMED International Corporation.

Conflicts of Interest
Martha Gonzalez was employed by and owned stock in HUTCHMED at the time of the study. William Schelman and Caly Chien are employed by and own stock in HUTCHMED.elman, and Caly Chien are employed by and own stock in HUTCHMED. Thomas C. Marbury is an employee and equity owner of Orlando Clinical Research Center. William Smith is an employee of Alliance for Multispecialty Research, LLC—a clinical research site. Xiaofei Zhou is employed by Takeda Development Center Americas, Inc. (TDCA). Neeraj Gupta is an employee of and reports ownership of stocks/shares in Takeda. Zhao Yang and Juan C. Rondon have no conflicts of interest to report.

Supporting information
Supporting Information