Regorafenib pharmacokinetics and its association with real-world treatment outcomes

Purpose Despite the established activity of regorafenib in metastatic colorectal cancer (CRC), gastrointestinal stromal tumor (GIST), and hepatocellular carcinoma (HCC), its toxicity pro�le has limited clinical use. We aimed to evaluate the pharmacokinetics of regorafenib and its active metabolites M-2/M-5, and to clarify the relationships between total drug-related exposure and clinical outcomes in real-world practice. Methods Blood samples were obtained at steady state in Cycle 1 in patients treated with regorafenib. Plasma concentrations of regorafenib and its metabolites were measured by liquid chromatography– tandem mass spectrometry. The e�cacy and safety endpoints were progression-free survival (PFS) and dose-limiting toxicities (DLTs), respectively. The exposure–response relationships were assessed. Results Thirty-four Japanese patients with advanced cancers were enrolled (CRC, n = 26; GIST and HCC, each n = 4). Nine patients started regorafenib treatment at the recommended dose of 160 mg once daily (3 weeks on / 1 week off), while the other patients received a reduced starting dose to minimize toxicities. The median PFS was signi�cantly longer in patients achieving total trough concentrations (C trough ) of regorafenib and M-2/M-5 ≥ 2.9 µg/mL than those who did not (112 vs. 57 days; p = 0.044). Furthermore, the cumulative incidence of DLTs during the �rst 2 cycles was signi�cantly higher in patients with summed C trough levels ≥ 4.3 µg/mL than in others (p = 0.0003). Conclusions Dose titration of regorafenib to achieve drug-related C trough levels in Cycle 1 between 2.9 and 4.3 µg/mL may improve the e�cacy and safety, warranting further investigation in a larger patient population.


Introduction
Regorafenib is an oral multikinase inhibitor that blocks the activity of various protein kinases associated with angiogenesis (vascular endothelial growth factor [VEGF] receptors 1-3 and TIE2), oncogenesis (KIT, RET, RAF-1, and BRAF), and the tumor microenvironment (PDGFR and FGFR) [1].The survival bene t of regorafenib has been demonstrated in metastatic colorectal cancer (CRC), gastrointestinal stromal tumor (GIST), and hepatocellular carcinoma (HCC) [2][3][4][5].The recommended dosage is 160 mg once daily (3 weeks on / 1 week off).However, this standard dosing schedule is associated with signi cant adverse events, including hand-foot skin reaction (HFSR), hypertension, and fatigue, which has limited the clinical use of regorafenib.Hence, there is a need to improve its tolerability and develop an individualized dosing strategy for regorafenib.
Previous studies have suggested that a exible, rst-cycle dose-escalation approach may help some patients with metastatic CRC manage toxicity and continue treatment [6].However, long-term e cacy as measured by progression-free survival (PFS) and overall survival (OS) did not differ signi cantly between the alternative starting dose group and the standard-dose group [6].Regarding pharmacokinetics, regorafenib shows large interindividual variability in its plasma exposure, which may result in altered clinical e cacy and/or safety [7].It has been shown that increased average exposure to regorafenib during Cycle 1 is signi cantly associated with prolonged OS in patients with HCC [8].These data suggest that the titration scheme might bene t from further re nement, including pharmacokinetically guided dosing with therapeutic drug monitoring (TDM) to optimize regorafenib exposure.
In this study, we aimed to (1) evaluate the clinical pharmacokinetics and pharmacogenetics of regorafenib and its metabolites, (2) clarify the relationships between total drug-related exposure and treatment outcomes in real-world practice, and (3) determine a therapeutic window for regorafenib exposure to guide dose titration.

Study design and patients
This study was conducted as a part of a single-center, prospective cohort study in routine clinical practice rather than clinical trials, which was called PRecision dOsing of moLecular-targeted agents based On therapeutic drug monitoriNG (PROLONG) (clinical study identi er, UMIN000036158).The study aimed to optimize targeted therapy with multikinase inhibitors through pharmacokinetically guided dosing with TDM in a real-world setting.The protocol was approved by the institutional ethics committee of Asahikawa Medical University (#15018).No inclusion criteria but being treated with regorafenib was applied.We prospectively enrolled all consecutive patients with metastatic CRC, GIST, and HCC who were receiving regorafenib treatment at our institution.Written informed consent was obtained from each patient.The baseline characteristics of the patients are summarized in Table 1.

Pharmacokinetic assessment
During hospitalization, the full pharmacokinetic pro le at steady state in Cycle 1 (i.e., at least one week after repetitive administration) was assessed by obtaining serial blood samples (before regorafenib dose and at 2, 4, 8, and 24 h post-dose) at the rst week of treatment, and, if possible, at the second and/or third weeks to check uctuations in regorafenib exposure at a given dose.In outpatient settings, remnant blood specimens after laboratory tests at each visit were used for monitoring trough concentrations (C trough ).
Plasma was separated by centrifugation (1,800 g, 10 min) and stored at -30°C until the analysis.
Additionally, rst-morning or spot urine samples were obtained to evaluate the urinary excretion of regorafenib and its metabolites.Plasma and urine concentrations of regorafenib and its metabolites were measured by validated liquid chromatography-tandem mass spectrometry.Methodological details on pharmacokinetic analysis are provided in the ESM Methods.

Treatment
Regorafenib was administered orally after breakfast at a starting dose of 160 mg once daily (3 weeks on / 1 week off).In patients with poor liver function or co-morbidities and those with poor Eastern Cooperative Oncology Group performance status, the initial dose was reduced at the discretion of each treating physician.If regorafenib was well tolerated with a reduced starting dose, the dosage was titrated stepwise up to 160 mg/day.The treatment was continued until disease progression or patient refusal.When unacceptable toxicities were present, the dose was reduced or interrupted temporarily, followed by resumption at a reduced dose.Regorafenib treatment was discontinued at the discretion of the treating physicians in cases of severe adverse events.
Based on historical data of regorafenib pharmacokinetics after multiple doses of 160 mg once daily (area under the concentration-time curve from 0 to 24 h [AUC 0-24 ] 33,042.8µg*h/L) [15], the average concentration within a 24-h dosing interval (C av ) was estimated to be approximately 1.4 µg/mL.Thus, we utilized regorafenib concentrations higher than this threshold as provisional target concentrations in the present study.Patients' pharmacokinetic pro les at steady state and individual C trough measurements at outpatient visits were provided to physicians (i.e., to indicate whether or not the provisional target was reached) to support the clinical decision regarding the need for dose escalation at the next visit.Finally, physicians decided on regorafenib dose adjustments based on patients' pharmacokinetic assessments in addition to individual tolerability and clinical and laboratory ndings.

Genotyping
Genomic DNA was extracted from the peripheral blood of patients using NucleoSpin Blood QuickPure (Takara Bio, Kusatsu, Japan).Based on previous ndings regarding pharmacogenetic determinants associated with metabolism and disposition of regorafenib and its analogue sorafenib, we examined the CYP3A4*22, CYP3A5*3, UGT1A9*3, and ABCG2 421C>A polymorphisms [12][13][14]16,17].Genotyping was performed using TaqMan SNP genotyping assays (Thermo Fisher Scienti c, Tokyo, Japan).Outcomes E cacy endpoints were PFS and best overall response per Response Evaluation Criteria in Solid Tumors (RECIST) 1.1.For safety assessment, all adverse events were graded according to the Common Terminology Criteria for Adverse Events (CTCAE) version 4.03.The cumulative incidence of dose-limiting toxicities (DLTs), including grade 3/4 adverse events leading to treatment discontinuation and grade 2 HFSR requiring dose interruption, was estimated by adjusting for competing risks (e.g., death or treatment discontinuation due to disease progression) with the Fine and Gray model [18].

Statistical analysis
The statistical signi cance of differences in non-parametric values between 2 groups was analyzed with the Mann-Whitney U test.A receiver operating characteristic (ROC) curve was constructed to calculate an area under the ROC curve (AUC ROC ) and estimate an optimal cutoff value of total drug-related exposure for predicting development of grade ³2 HFSR.The median PFS was estimated using the Kaplan-Meier method, and the difference between groups was examined using the log-rank test.A 2-sided p < 0.05 was considered statistically signi cant.All statistical analyses were performed using STATA software, version 16 (StataCorp LLC, Texas, USA).

Patients and treatment duration
Between October 2013 and February 2020, 34 Japanese patients with advanced cancers were enrolled (CRC, n = 26; GIST and HCC, each n = 4) (Table 1).Regorafenib treatment was discontinued because of disease progression (n = 26), adverse events (n = 6), or patient refusal (n = 1).Adverse events leading to discontinuation included fatigue (n = 2), HFSR, stula, hyperbilirubinemia, and pruritus (each n = 1).As of the data cutoff date (March 31, 2020), regorafenib therapy was still ongoing in one patient.The median (range) follow-up period was 140 (22-713) days.No pharmacokinetic or pharmacogenetic data were obtained in 8 patients; however, these patients were included in e cacy and safety analyses.
Nine patients started regorafenib treatment at the recommended dose of 160 mg once daily, while 20 patients received a lower starting dose (80 mg, n = 18; 120 mg, n = 2); the lowest starting dose of 40 mg was used in 5 patients to minimize toxicities.The median (range) duration of treatment was 83 (8-520) days in all patients (Fig. 1a).Patients starting at 80 mg tended to continue regorafenib therapy for a longer period than those in the other starting dose groups.In approximately 60% of patients in each starting dose group, the last dose was the same as the initial dose (Fig. 1b); the exception was the 120-mg dose group, in which both patients maintained the original dose.Among patients starting at 40 mg, dose escalation only reached 80 mg.Among those who initially received 80 mg, the dose was nally increased to 120 mg and 160 mg in 22% and 6% of patients, respectively.

Pharmacokinetic pro le
Plasma levels of M-8 were negligible.Full pharmacokinetic pro les and parameters of regorafenib and its metabolites at steady state in Cycle 1 are shown in Fig. 2a and summarized in Table 2, respectively.These results were generally comparable with those in previous reports [15].The sum of the areas under the concentration-time curves (AUCs) of regorafenib and M-2/M-5 within a 24-h dosing interval, which re ects the total pharmacological activity of regorafenib and its two active metabolites, was not correlated with C trough level of regorafenib alone but signi cantly correlated with the sum of their corresponding C trough levels (r 2 = 0.90; p < 0.0001) (ESM Fig. S1).Accordingly, we used the sum of the C trough levels as a surrogate therapeutic marker for subsequent analyses including exposure-response relationships.
Throughout the course of treatment, there was large variability in plasma exposures to regorafenib as well as its metabolites (total no. of observations, 204 each), with a median value of 1,511 ng/mL for regorafenib (Fig. 2b), which was above the provisional target of 1.4 µg/mL.S2).When patients were categorized into 2 groups according to the median value of the M-7/regorafenib metabolic ratio (i.e., 0.52), it was evident that the sum of the dose-adjusted C trough levels was signi cantly lower in the high-ratio group than in the low-ratio group (median 21.2 ng/mL/mg [n = 13] vs. 42.2ng/mL/mg [n = 13]; p = 0.012).
In urine, regorafenib and M-2/M-5 were hardly detectable, as shown in ESM Fig. S3, whereas M-7 and M-8 were detected in signi cant amounts (M-7 > M-8; total no. of observations, 31 each) (Fig. 2c).Furthermore, M-5 was also detected after treatment with b-glucuronidase, suggesting the presence of a previously unidenti ed glucuronide metabolite of M-5 in urine.

Dose/exposure-response
Overall, only one patient with CRC who started regorafenib treatment at 120 mg/day achieved a partial response (3%; 4% in CRC), whereas 6 had stable disease (18%), 26 had progressive disease (76%), and one was not evaluated for tumor response.As shown in Figure 3a, the median PFS was numerically shorter in the lowest-dose group compared with the other starting dose groups, although the difference was not statistically signi cant (log-rank p = 0.62).When patients were divided into 2 groups according to the median value of the sum of the C trough levels of regorafenib and M-2/M-5 at steady state in Cycle 1, patients with higher exposure (³2.9 µg/mL) demonstrated signi cantly longer PFS than those with a level <2.9 µg/mL (112 vs. 57 days; log-rank p = 0.044) (Fig. 3b).
The most frequently observed adverse event of any grade was HFSR (68%) (ESM Fig. S4).No grade 5 toxicity was reported in this study.The most common grade 3/4 adverse events were elevated gammaglutamyltransferase levels (12%), followed by HFSR, hypertension, and anorexia (9% each), proteinuria, fatigue, and elevated total bilirubin levels (6% each), and decreased platelet counts and elevated alanine aminotransferase levels (3% each).As shown in Figure 4, the sum of the C trough levels among patients who experienced grade 2/3 HFSR was signi cantly higher than that in patients with grade 0/1 HFSR (median 5.2 vs. 2.2 µg/mL; p = 0.021).Furthermore, the optimal cutoff value for the sum of the C trough levels to predict development of grade ³2 HFSR was 4.3 µg/mL, with an AUC ROC of 0.78 (95% con dence interval [CI] 0.57-0.98).
Starting regorafenib treatment at a reduced dose was associated with a lower cumulative incidence of DLTs during the rst 2 cycles, although the difference was not statistically signi cant (p = 0.87) (Fig. 5a).Notably, the cumulative incidence of DLTs was signi cantly higher in patients in whom the sum of the C trough levels was greater than or equal to the aforementioned cutoff of 4.3 µg/mL, as compared with others (sub-hazard ratio 8.84; 95% CI 2.71-28.8;p = 0.0003) (Fig. 5b).

Discussion
In this real-world study that included all patients who were consecutively treated with regorafenib at our institution, the safety pro le was generally consistent with that reported in phase 3 trials, except for proteinuria, which was more common in the current study [2][3][4][5].Overall, in approximately 60% of patients, the initial dose was the same as the dose in the last cycle of treatment or at last follow-up, a nding comparable to that of a previous study under routine clinical practice (CORRELATE) [19].Although the present study was limited by a small sample size, lack of randomization, and patient heterogeneity (i.e., tumor types and prior systemic therapies), we observed clear relationships of e cacy and safety with early exposure to regorafenib and its active metabolites M-2 and M-5, but not with the starting dose itself (Figs. 3   and 5).
To our best knowledge, this study provides the rst evidence regarding a potential therapeutic window of regorafenib exposure, which might range from 2.9 to 4.3 µg/mL in terms of the sum of the C trough levels at steady state in Cycle 1.Recently, Keunecke et al. reported a population pharmacokinetic model for regorafenib and M-2/M-5, incorporating enterohepatic circulation [20].The model-informed precision dosing with Bayesian estimation to optimize regorafenib therapy could be used to rapidly achieve the target therapeutic exposure in individual patients.Further investigation is warranted to determine the bene t of this approach in routine clinical practice.
This study identi ed no genetic determinants of interindividual variability regarding total exposure to regorafenib and M-2/M-5, a result that is comparable with the ndings of previous studies [12,13].Recently, we demonstrated that poor absorption of axitinib, a multikinase inhibitor used to treat patients with metastatic renal cell carcinoma, was linked to higher glucuronidation of the drug, primarily by UGT1A1 in the liver [21].The present study revealed that like axitinib, greater glucuronidation activity of UGT1A9 toward regorafenib (i.e., higher M-7/regorafenib metabolic ratio) was associated with decrease in the sum of the dose-adjusted C trough levels of regorafenib and M-2/M-5, probably due to a reduction in the relative contribution of CYP3A4-mediated regorafenib metabolism into M-2 (ESM Fig. S2c).Therefore, rapid dose titration should be considered in patients with increased glucuronidation activity toward regorafenib to prevent subtherapeutic exposure to regorafenib and M-2/M-5.
Interestingly, M-8 was not detectable in plasma but was observed in urine (Fig. 2).Furthermore, M-7 was more abundant in urine than in plasma.These ndings suggest that renal UGT1A9 plays an important role in the glucuronidation of regorafenib, M-2, and possibly M-5, thus facilitating the urinary excretion of their metabolites [22].Further investigation is necessary to clarify whether the interindividual variability in urinary concentrations of M-7 and M-8 is associated with the clinical outcomes of regorafenib.
In a randomized, open-label, phase 2 study in patients with refractory metastatic CRC (ReDOS), a weekly dose-escalation strategy starting from 80 mg to 160 mg based on individual tolerability has been demonstrated as an alternative approach for optimizing regorafenib dosing, allowing more patients to initiate Cycle 3 [23].Although this dose titration schedule is not included in the labeling, the National Comprehensive Cancer Network Colon Cancer Guidelines recently included it as an option to treat patients with metastatic CRC.Furthermore, a similar randomized phase 2 trial (REARRANGE) showed that exible initial dosing (i.e., 120 mg/day 3 weeks on / 1 week off [reduced dose] or 160 mg/day 1 week on / 1 week off [intermittent dose] in Cycle 1) alleviated toxicities without compromising e cacy [24].Despite the improved outcomes of regorafenib with these modi ed dosing approaches, there was no long-term survival bene t [23,24].In outpatient settings, even if TDM is applied, initiating regorafenib at the standard dose might lead to early termination due to intolerable toxicity before dose is modi ed based on drug overexposure detected at next clinical visit.Taking our ndings on regorafenib dose/exposure-response relationships into consideration, a dose titration strategy starting from a reduced dose (e.g., 80 mg or 120 mg) might be bene cial to prevent severe toxicities in certain patients, such as those with poor performance status, and patients may be able to continue regorafenib treatment while maintaining disease control.In particular, the "start low and go slow" approach integrating pharmacokinetically guided dosing may de ne a safer, individualized regorafenib regimen for each patient and thus prolong survival.In future clinical studies, toxicity-based, pharmacokinetically guided dose titration of regorafenib should be validated for monotherapy as well as for combination treatment with immunotherapy such as nivolumab (REGONIVO) [25].This should be done in selected groups of patients with various types of cancer to develop a more re ned dosing approach for regorafenib.

Conclusions
The present study suggests that the potential therapeutic window for the sum of the C trough levels of regorafenib and M-2/M-5 at steady state in Cycle 1 is between 2.9 and 4.3 µg/mL.In addition to individual tolerability, TDM of regorafenib that evaluates total drug-related exposure will help guide dose titration and improve treatment outcomes in routine clinical practice.Larger, randomized studies are warranted to con rm this approach.
Declarations is associated with drug exposure and UGTIA9 genetic polymorphism in patients with solid tumors: a preliminary study.PLoS One 7:e42875 AUC 0-24 area under the concentration-time curve from 0 to 24 h, C av average concentration within a 24-h dosing interval (= AUC 0-24 /24), CL/F apparent oral clearance, C max maximum concentration, C trough trough concentration, T max time to C max a Serial blood samples were obtained at the following time points: Week 1 (n = 5) for 40 mg Weeks 1, 2, and 3 (n = 5, 3, and 2, respectively) for 80 mg; Weeks 1 and 3 (n = 1 and 2 respectively) for 120 mg; and Weeks 1 and 2 (n = 5 and 3, respectively) for 160 mg b Data are expressed as geometric mean (geometric CV%) or median (range) No CYP3A4*22 or UGT1A9*3 alleles were found in our patients.Genetic polymorphisms of CYP3A5*3 (*1/*1 and *1/*3 [n = 15] vs. *3/*3 [n = 11]) and ABCG2 421C>A (C/C [n = 19] vs. C/A and A/A [n = 7]) did not in uence the sum of the dose-adjusted C trough levels of regorafenib and M-2/M-5 at steady state in Cycle 1 (ESM Fig.

Figure 1 a
Figure 1