Exposure‐toxicity relationship of cabozantinib in patients with renal cell cancer and salivary gland cancer

Abstract

Cabozantinib is registered in fixed 60 mg dose. However, 46% to 62% of patients in the registration studies needed a dose reduction due to toxicity. Improved clinical efficacy has been observed in renal cell carcinoma patients (RCC) with a cabozantinib exposure greater than 750 μg/L. In our study we explored the cabozantinib exposure in patients with different tumour types. We included RCC patients from routine care and salivary gland carcinoma (SGC) patients from a phase II study with ≥1 measured C _min at steady‐state. The geometric mean (GM) C _min at the starting dose, at 40 mg and at best tolerated dose (BTD) were compared between both tumour types. Forty‐seven patients were included. All SGC patients (n = 22) started with 60 mg, while 52% of RCC patients started with 40 mg. GM C _min at the start dose was 1456 μg/L (95% CI: 1185‐1789) vs 682 μg/L (95% CI: 572‐812) (P < .001) for SGC and RCC patients, respectively. When dose‐normalised to 40 mg, SGC patients had a significantly higher cabozantinib exposure compared to RCC patients (C _min 971 μg/L [95% CI: 790‐1193] vs 669 μg/L [95% CI: 568‐788]) (P = .005). Dose reductions due to toxicity were needed in 91% and 60% of SGC and RCC patients, respectively. Median BTD was between 20 to 30 mg for SGC and 40 mg for RCC patients. GM C _min at BTD were comparable between the SGC and the RCC group, 694 μg/L (95% CI: 584‐824) vs 583 μg/L (95% CI: 496‐671) (P = .1). The observed cabozantinib exposure at BTD of approximately 600 μg/L is below the previously proposed target. Surprisingly, a comparable exposure at BTD was reached at different dosages of cabozantinib for SGC patients compared to RCC patients Further research is warranted to identify the optimal exposure and starting dose to balance efficacy and toxicity.

Short abstract

What's new?

Cabozantinib, a potent tyrosine kinase inhibitor that targets multiple signaling pathways, is approved for use against advanced renal cell carcinoma (RCC). Variations in cabozantinib clearance, however, warrant further investigation. Here, the authors evaluated cabozantinib exposure in RCC patients and in patients with salivary gland cancer (SGC). SGC patients were found to have significantly higher cabozantinib exposure compared to RCC patients following a 40 mg dose. However, the best‐tolerated cabozantinib exposure was equivalent (~600 μg/L) for both tumor types and was substantially below the previously proposed target. The findings offer insight on exposure, dose, and the balance between efficacy and toxicity for cabozantinib.

Abbreviations

BTD

best tolerated dose

CI

confidence interval

C _max

maximum concentration

C _min

trough concentration

GM

geometric mean

HCC

hepatocellular carcinoma

ICI

checkpoint inhibitors

IMDC

International Metastatic Database Consortium

IRB

Institutional Review Board

MET

hepatocyte growth factor receptor

MTC

medullary thyroid cancer

OD

once daily

ORR

overall response rate

OS

overall survival

PFS

progression free survival

PK

pharmacokinetics

PPI

proton pump inhibitor

RCC

renal cell carcinoma

SGC

salivary gland cancer

TAM

Tyro3, Axl and Mer

TDM

therapeutic drug monitoring

TKI

tyrosine kinase inhibitor

VEGFR

vascular endothelial growth factor receptor

1. INTRODUCTION

Cabozantinib is a potent oral tyrosine kinase inhibitor targeting multiple kinases, including the vascular endothelial growth factor receptor 2 (VEGFR2) and the hepatocyte growth factor receptor (MET). Inhibition of these key signalling pathways affects the tumour vasculature and averts angiogenesis, proliferation and metastasis of cancer cells. ¹ A distinctive feature of cabozantinib over other VEGFR inhibitors is that it also targets the Tyro3, Axl and Mer (TAM) family of receptor kinases, which is thought to contribute to an antitumor immunomodulating effect. ²

Cabozantinib capsules (Cometriq) have been approved for treatment of progressive, unresectable locally advanced medullary thyroid cancer (MTC) in a fixed dose of 140 mg once daily (OD). Cabozantinib tablets (Cabometyx) in a fixed dose of 60 mg once daily have been approved for treatment of patients with advanced renal cell carcinoma (RCC) and for patients with hepatocellular carcinoma (HCC) who have previously been treated with sorafenib. ³ Recently, cabozantinib in combination with checkpoint inhibitors (ICI) has shown promising results in RCC patients. ⁴ , ⁵ The combination of cabozantinib and nivolumab for patients with metastatic RCC has been approved by the FDA. Furthermore, the combination has been added as a first‐line treatment option for clear cell RCC for all International Metastatic Database Consortium (IMDC) risk groups in the RCC guideline of the European Society for Medical Oncology and in the RCC guideline of the European Association of Urology. ⁶ , ⁷ In addition, multiple trials investigating cabozantinib‐ICI combinations for other solid tumours are ongoing (n = 16). Although the tablets and capsules are not deemed bioequivalent based on a small difference in the maximum concentration (C _max) reached, both the tablets and capsules showed comparable area under the concentration‐time curve. ⁸ Moreover, both formulations exhibited dose proportional pharmacokinetics.

The pharmacokinetics (PK) of cabozantinib has been characterised in a population PK study which combined exposure data from healthy volunteers and cancer patients with different types of solid malignancies. ⁹ , ¹⁰ Cabozantinib shows large interpatient variability of approximately 50%, which is in line with other tyrosine kinase inhibitors. ¹¹ Interestingly, patients with MTC had a two‐fold higher apparent clearance compared to RCC patients. The underlying cause of this difference remains unclear, but it may suggest a difference in cabozantinib clearance due to factors which are affected differently by different tumour types. With cabozantinib currently being studied for multiple tumour types, this finding could be of importance as for most oral tyrosine kinase inhibitors a correlation between the level of systemic exposure and clinical benefit exists. ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶

For cabozantinib, its exposure response relationship has been investigated in patients with RCC in the phase III METEOR study. ¹⁷ , ¹⁸ , ¹⁹ The 60, 40 and 20 mg dose corresponded to an average cabozantinib steady‐state concentration of 1125, 750 and 375 μg/L, respectively. The exposure reached with the 60 mg dose resulted in improved progression free survival (PFS), reduced tumour growth and increased overall response rate compared to the 40 and 20 mg doses. For a dose of 40 mg instead of 60 mg once daily, the model predicted a modest 1.1‐fold increased risk of disease progression. However, this modest loss in efficacy opposes a 1.4‐fold lower risk of developing adverse events. Nevertheless, the starting dose of 60 mg was selected over 40 mg. ¹⁹ In the pivotal trials, adverse events were judged to be manageable with dose interruptions and reductions. The main dose limiting toxicities were fatigue and decreased appetite. ²⁰ However, the proposed cabozantinib target exposure of 1125 μg/L seems poorly tolerated as can be derived from the high percentage of patients necessitating dose reductions in studies (40%‐79%). ²⁰ , ²¹ , ²² In addition, patients in routine patient care often differ from patients treated within trials in terms of increased age, number of comorbidities, disease severity, pretreatment and lower performance status. ²³ , ²⁴ This casts a doubt on the tolerability of the proposed target exposure in these patients. Furthermore, in our clinic the phase II trial of cabozantinib for the treatment of patients with recurrent and/or metastatic salivary gland cancer was terminated prematurely due to severe toxicity, especially severe wound complications within the prior irradiated area of the head and neck. ²⁵ As almost all patients in our study required dose reduction, this may call for a reconsideration of the accurate starting dose of cabozantinib.

In our study, we evaluate cabozantinib exposure in patients with salivary gland cancer and patients with renal cell cancer. Furthermore, potential relationships between cabozantinib exposure, patient characteristics, toxicity and treatment outcomes are explored.

2. METHODS

2.1. Study design and patient population

This observational study was performed using clinical data and cabozantinib trough concentrations (C _min) obtained from patients treated with cabozantinib for RCC and SGC between January 2018 and August 2020. Baseline characteristics were retrospectively retrieved from the electronical health records or retrieved from the case report forms at start of cabozantinib treatment. Missing data at baseline were replaced by the closest value with a time interval up to 21 days prior to start of treatment or if not available, were left missing. For cabozantinib treatment, date, dose, time of intake and use of potent cytochrome P‐450 inhibitors or inducers were collected at start and at each measured steady‐state cabozantinib C _min level. In addition, date and reason for each dose adjustment were collected. Adverse events necessitating dose reduction or treatment discontinuation were scored as clinically relevant toxicity. Best tolerated dose (BTD) was defined as the latest dose level before treatment discontinuation or at time of data cut off (1 August 2020).

For evaluation of the time on treatment, the time from start until the last day of cabozantinib treatment or data cut off was recorded. Also, the reason for treatment discontinuation was documented. Patient weight at baseline and at the end of cabozantinib treatment was collected.

2.2. Cabozantinib pharmacokinetics

Patients had plasma cabozantinib C _min levels measured as part of routine patient care (RCC) or as part of the CABO‐ASAP study (SGC) (NCT03729297). All blood samples were collected in ethylenediamine tetra acetic acid plasma tubes and were subsequently centrifuged on the same day. Samples were stored at 4°C to 8°C until the day of analysis for a maximum of 14 days. A validated high‐performance liquid chromatography coupled with tandem mass spectrometry detection assay was used to determine total cabozantinib concentrations in plasma. ²⁶ Only samples measured at steady‐state were taken into account and patients needed at least one cabozantinib C _min level at steady‐state to be included. Steady‐state was defined as cabozantinib treatment at a fixed dose for more than 17 consecutive days based on four times the half‐life of cabozantinib.

No predefined sampling moments were set for measuring cabozantinib plasma concentrations. However, therapeutic drug monitoring is standard of care in our clinic and the first measurement is usually performed approximately 4 weeks after treatment initiation, at the moment of dose reduction and to evaluate the effect of dose adjustment. Patients were instructed to postpone cabozantinib intake till after sample collection on hospital visits. For each sample, the date and time of last intake of cabozantinib and the date and time of the plasma sample collection were recorded. In case the sample was not collected 24 hours after intake, the trough concentration was calculated by the approach described by Wang et al with a minimum of 5 and a maximum of 72 hours between blood collection and last pill intake. ²⁷

Cabozantinib exposures at start, 40 mg and BTD, respectively, were determined using extrapolation of the dose‐normalised average exposure based on all available cabozantinib C _min measurements per patient. Cabozantinib shows dose‐proportional exposure over the range of 20 to 140 mg which supports the use of this approach. For each patient, the average dose‐normalised cabozantinib C _min exposure per milligram of cabozantinib was calculated as follows: First, each available cabozantinib C _min measurement at steady‐state was dose‐normalised by dividing it by the administered dose of cabozantinib used at the moment of measurement (cabozantinib exposure/administered dose of cabozantinib in mg). Subsequently, the average of these dose‐normalised exposures was calculated per patient. This average concentration was thereafter multiplied by the specific dosages of interest for example, starting dose, 40 mg and best tolerated dose. The average cabozantinib exposure over the course of treatment was estimated based on steady‐state exposures. This estimate was calculated by multiplying the number of days at each dose level of cabozantinib and with the corresponding milligrams of cabozantinib. The sum of the amount of cabozantinib was subsequently divided by the duration of treatment in days to yield the average amount of cabozantinib in milligram per day. The average cabozantinib exposure was then calculated by multiplying this average amount of cabozantinib with the dose‐normalised exposure. Correlations between average calculated and measured cabozantinib C _min levels at start, 40 mg and BTD were assessed to confirm dose‐proportionality in our cohort.

2.3. Statistical analysis

For the primary outcome, geometric mean cabozantinib C _min levels at start, at 40 mg and at BTD were calculated for both SGC and RCC patients and the log‐transformed data were compared to an independent samples T‐test. Baseline patient and cabozantinib treatment characteristics were described using descriptive statistics. Associations between previously identified baseline patient characteristics (albumin, gender and body mass index) and cabozantinib exposure were first explored visually followed by linear regression (continuous variables) or with an independent samples T‐test (categorical variables) if deemed of interest. The exposure‐toxicity relationship was explored by comparing log‐transformed cabozantinib exposure at start of treatment and the occurrence of dose reduction (yes/no) with an independent samples T‐test. The relationship between average cabozantinib exposure and relative weight loss was explored visually. The preliminary efficacy was assessed by comparing progression free survival (PFS) and overall survival (OS) for patients with RCC with cabozantinib exposure below and above the median average exposure per day and for patients requiring a dose reduction (yes/no) relative to the starting dose, through Kaplan‐Meier methods using the unstratified log‐rank test.

All statistical analyses were performed using IBM SPSS statistics for Windows version 25.0 (IBM Corp, Armonk, New York). Geometric mean values are reported with 95% confidence intervals (CI). All P values were calculated with two‐sided statistical testing. Outcomes with P values less than .05 were considered to be statistically significant.

3. RESULTS

3.1. Patients characteristics

In total 47 patients were included, 22 patients with SGC and 25 patients with RCC. Baseline characteristics are shown in Table 1. For patients with RCC, most patients received cabozantinib as third line treatment. All patients with RCC received prior anti‐VEGFR therapy and the majority received prior treatment with an immune checkpoint inhibitor. For patients with SGC, the majority underwent surgery and received postoperative radiotherapy. Seven SGC patients (31.8%) received prior systematic therapy in the palliative setting.

TABLE 1.

Baseline characteristics at start of cabozantinib treatment

	RCC patients (n = 25)	SGC patients (n = 22)
Age (years)	67 (37‐81)	57 (49‐72)
Gender
Male	21 (84)	10 (45)
KPS
90‐100	7 (28)	9 (41)
80‐90	9 (36)	1 (4.5)
70‐80	8 (32)	12 (55)
60‐70	1 (4)	0
Histology
Clear cell	20 (80)	NA
Papillary	3 (12)	NA
Other RCCs	2 (8)	NA
ACC	NA	16 (72)
SDC	NA	4 (18)
Other SGCs	NA	2 (9)
IMDC risk score
Intermediate	13 (52)	NA
Poor	12 (48)	NA
Prior treatment
Surgery	14 (56) a	18 (82)
Postoperative radiotherapy	NA	18 (82)
Palliative radiotherapy	DNC	10 (45)
Adjuvant systemic therapy	NA	2 (9)
Palliative systemic therapy	25 (100)	7 (32)
Line of systemic therapy (median, range)	3 (2‐6)	2 (1‐5)
Prior VEGFR‐I	25 (100)	1 (5)
Prior ICI	19 (76)	0 (0)
Weight (kg)	82 (51‐121)	76 (54‐111)
BMI (kg/m2)	25.5 (17.8‐36.3)	25.1 (21.4‐39.9)
Haemoglobin (μmol/L)	7.1 (4.8‐8.7)	8.2 (6.5‐9.7)
Leukocytes (109/L)	7.9 (4.5‐14.2)	6.7 (4.1‐11.3)
Thrombocytes (109/L)	337 (210‐984)	270 (131‐454)
Neutrophils (109/L)	5.9 (3.7‐11.5)	4.6 (2.3‐8.9)
Calcium (mmol/L)	2.41 (2.24‐2.73)	2.41 (2.24‐2.65)
Serum creatinine (μmol/L)	121 (52‐228)	64 (46‐108)
ALT (IU/L)	22 (9‐107)	23 (12‐76)
AST (IU/L)	24 (18‐49)	27 (14‐68)
Albumin (g/L)	31.5 (20‐38)	38.0 (28‐42)

Note: Data are presented as n (%) for categorical variables and median (range) for continuous variables.

Abbreviations: ACC, adenoid cystic carcinoma; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; DNC, data not collected; ICI, immune checkpoint inhibitor; IMDC, International Metastatic RCC Database Consortium; NA, not applicable; RCC, renal cell carcinoma; SDC, salivary duct carcinoma; SGC, salivary gland cancer; VEGFR‐I, vascular endothelial growth factor inhibitor.

^a Nephrectomy.

3.2. Cabozantinib treatment

Patients with SGC all received the 60 mg starting dose as was mandated by the study protocol. For patients with RCC, the selection of the starting dose was at the physicians' discretion. The majority of patients with RCC (48%) started with 40 mg. Cabozantinib treatment characteristics are depicted in Table 2. Median duration of treatment was 22 (4‐151) weeks for RCC and 29.5 (3‐55) weeks for SGC. The SGC phase II study was terminated prematurely due to the occurrence of severe wound complications in previously irradiated areas. Dose reductions relative to the starting dose were needed in 20 (91%) patients with SGC and 15 (60%) patients with RCC. All dose reductions were related to toxicity. Median best tolerated dose, defined as the latest dose level before treatment discontinuation, was between 20 and 30 mg in patients with SGC and 40 mg in patients with RCC. Median average cabozantinib dose intensity over the duration of treatment was approximately 40 mg in both tumour types. In the RCC groups, 16 patients (64%) discontinued treatment due to progressive disease and 5 (20%) due to toxicity. One patient had liver test elevations, one patient had a nonhealing wound on the lower extremity, one patient had a gastrointestinal mucosal lesion and two patients had a gastrointestinal perforation.

TABLE 2.

Cabozantinib treatment details

	RCC patients (n = 25)	SGC patients (n = 22)
Dose at start of treatment
60	8 (32)	22 (100)
40	13 (52)	0
20	4 (16)	0
Median treatment duration (weeks)	22 (4‐151)	29.5 (3‐55)
Dose reduction during treatment	15 (60)	20 (91)
Reason of treatment discontinuation	21 (84) a	22 (100)
Toxicity	5 (20)	3 (14)
Progressive disease	16 (64)	6 (27)
Death	0	0
Other b	0	13 (59)
Best tolerated cabozantinib dose (mg)
20	5 (20)	11 (50)
27 c	1 (4)	0
30 c	3 (12)	2 (9)
40	13 (52)	7 (32)
50 c	1 (4)	0
60	2 (8)	2 (9)
Median average dose intensity (mg)	39 (20‐60)	38 (24‐60)
Total number of samples at steady state	42	37
Median number of samples per patient	1 (1‐5)	1 (1‐3)

^a Four patients still on treatment at data cut‐off.

^b Stop of study due to safety concerns.

^c Cabozantinib dose reached via an alternative dosing schedule (eg, 20 and 40 mg used alternately).

3.3. Cabozantinib exposure at start, 40 mg and at BTD

In total 79 cabozantinib samples at steady‐state were available, 37 of patients with SGC and 42 of patients with mRCC. The number of samples per patient ranged from 1 to 5. Cabozantinib C _min levels were measured for 13 (59%) patients with SGC at the start dose level of 60 mg, 9 (41%) at 40 mg and 16 (73%) at best tolerated dose. For patients with RCC, these percentages were 68%, 76% and 80% at start, 40 mg and at BTD, respectively. The average dose‐normalised cabozantinib C _min based on all available measurement was used to calculate the cabozantinib C _min level at a specific dose level for each individual patient. There was a good correlation between measured and calculated cabozantinib C _min levels (R ² > 0.89, Figure S1A‐C).

At the start dose, patients with SGC had approximately two‐fold higher GM cabozantinib exposure compared to patient with RCC (1456 μg/L [95% CI 1185‐1789] vs 682 μg/L [95% CI 572‐812], P < .001, for patients with SGC and RCC, respectively) (Figure 1). After dose normalisation to 40 mg, patients with SGC still had a significantly higher cabozantinib exposure compared to patients with RCC (GM C _min 971 μg/L [95% CI 790‐1193] vs 669 μg/L [95% CI 568‐788] (P = .005). Cabozantinib GM C _min levels at BTD were comparable between the SGC group and the RCC group, 694 μg/L (95% CI 584‐824) versus 583 μg/L (95% CI 496‐671) (P = .1) as depicted in Figure 1. At BTD, 64% of SGC patients and 80% of RCC patients had an exposure below the proposed target of >750 μg/L.

FIGURE 1.

Scatter plots of the individual cabozantinib C _min levels at start, 40 mg and at BTD for patients with SGC and RCC. Data are presented with geometric mean (95% CI). BTD, best tolerated dose; CI, confidence interval; C _min, trough concentration, RCC, metastatic renal cell carcinoma; SGC, salivary gland cancer

3.4. Cabozantinib exposure and patient characteristics

To explore the potential relationship between patient baseline characteristics and cabozantinib exposure, previously identified demographic variables that differed in our cohorts at baseline (albumin and gender) were visually assessed via scatter plots (Figure S2). No apparent correlation between cabozantinib exposure and albumin levels was observed. A higher cabozantinib exposure was observed for female patients compared to their male counterparts. In female patients with SGC GM cabozantinib C _min at 40 mg was 1254 μg/L (95% CI 996‐1577) compared to 714 μg/L (95% CI 540‐945) in male patients (P = .002). For patients with RCC, no difference could be observed.

3.5. Cabozantinib exposure and toxicity

An explorative analysis was performed between the occurrences of clinically relevant toxicity, defined as an adverse event necessitating dose reduction or treatment discontinuation, and cabozantinib exposure in patients which RCC. Since nearly all patients with SGC (91%) required dose reductions, no comparison could be made between patients with and without dose reduction for this group. In patients with RCC, a trend was observed for a higher cabozantinib C _min level for patients who needed a dose reduction relative to the start dose, that is, 769 μg/L (95% CI 663‐893) versus 568 μg/L (95% CI 384‐842), respectively (P = .079) (Figure S3).

Patient weight was collected at start of cabozantinib treatment and at discontinuation or last moment of follow‐up. Median percentage weight loss from baseline was 6.2% (range: −2.6 to 34) and 7.2% (range: −0.5 to 18.7) in patients with SGC and RCC, respectively. No correlation was observed between cabozantinib exposure and weight loss (Figure S4).

3.6. Cabozantinib exposure and response

Median estimated average cabozantinib exposure was 573 μg/L (range: 225‐1088) in patients with RCC. Median PFS was 34.0 weeks (95% CI 24.2‐43.8). In the preliminary efficacy analysis, no difference in PFS was observed between patients with an average cabozantinib exposure above and below the median average cabozantinib exposure (19.0 weeks, 95% CI 0‐45.7 vs 34 weeks, 95% CI 32.6‐35.5, respectively). Interestingly, a longer PFS was observed for patients with a dose reduction relative to the start dose compared to patients without a dose reduction (41.0 weeks, 95% CI 13.5‐68.5 vs 20.0 weeks, 95% CI 9.6‐30.4, P = .023, respectively). In the exploratory OS analysis, no difference could be observed for patients with an average cabozantinib exposure above and below the median and start dose of 60 mg or below 60 mg (Figures S5 to S8).

4. DISCUSSION

In this observational study, we observed a significantly higher cabozantinib exposure in patients with SGC compared to those with RCC, also after correcting for differences in the dose administered. Dose reductions due to toxicity were needed in a large percentage of patients. The cabozantinib exposure reached at the best tolerated dose level was comparable between both groups of patients. However, this exposure lies well below the previously proposed target of >750 μg/L for which improved treatment outcomes have been observed.

Cabozantinib pharmacokinetics have previously been characterised in patients with various solid tumours. ⁹ , ¹⁷ Female gender has been associated with a 21% decrease in cabozantinib clearance compared to male gender. ¹⁷ This difference was however not deemed clinically significant given the high interpatient variability of cabozantinib. In our study, the female gender was more evenly distributed in patients with SGC compared to patients with RCC (55% vs, 16%, respectively) and this may partly explain the higher geometric mean cabozantinib C _min levels observed in the SGC group. This potential gender effect warrants further research in larger cohorts as indications for cabozantinib are extending and females are often underrepresented in clinical trials. ²⁸ Another possible explanation for the observed difference in cabozantinib concentrations in SGC and RCC patients may be the difference in serum albumin concentrations. Previously, lower albumin concentrations have been associated with lower maximal cabozantinib concentrations. ²⁹ Therefore, the higher albumin concentrations in patients with SGC may be an explanation for the higher cabozantinib concentrations observed in these patients. Interestingly, a two‐fold lower cabozantinib exposure was previously reported in patients with MTC, which possibly indicates that specific shared characteristics within a tumour type may affect cabozantinib exposure. However, for other registered tyrosine kinase inhibitors targeting VEGFR, no apparent differences have been reported in pharmacokinetics between groups of patients with different tumour types, excluding cancers affecting organs known to be involved in drug metabolism, such as hepatocellular carcinoma and thyroid cancer. ³⁰ , ³¹ , ³² Cabozantinib exposure is not affected by renal impairment, so the difference in serum creatinine between both groups is unlikely to be of relevance. ²⁹ Prior treatment with VEGFR inhibitors in RCC patients could potentially have contributed to a lower absorption from the gastrointestinal tract. Theoretically, reduced salivary function and changes in salivary composition after irradiation may lead to changes in the digestive process and hence increased drug absorption and/or reduced metabolism in SGC patients. ³³ , ³⁴ However, prior studies of TKIs in SGC have not yet reported pharmacokinetic data to support this hypothesis. ³⁵ , ³⁶

Similar to the registration trials, the majority of patients required dose reduction due to toxicity. Median best tolerated dose was between 20 and 30 mg for SGC patients and 40 mg for RCC patients. However, at best tolerated dose, cabozantinib exposure was comparable between patients with SGC and RCC (694 vs 583 μg/L, respectively). The slightly higher exposure in SGC patients is most likely the result of the fixed starting dose of 60 mg once daily. In contrast to real world patients, use of a lower starting dose was not allowed. In addition, as the study was terminated early, not all patients may have received dose reduction and therefore the best tolerated dose and corresponding exposure could have been overestimated. Nevertheless, our finding in RCC patients is line with a previous study of Cerbone et al, who observed a significantly higher median cabozantinib C _min in routinely treated RCC patients with Grade 3 to 4 toxicity versus those without toxicity (623 vs 452 μg/L, respectively). ³⁷ Furthermore, they reported a median C _min of 405 and 521 μg/L in patients with and without progressive disease, respectively. This indicates a narrow range in which efficacy and tolerability need to be balanced.

The observed percentage of dose reductions of 60% in our study is in line with the 57% reported in a cohort of 410 RCC patients treated with cabozantinib in an early access programme in France (CABOREAL). ³⁸ In CABOREAL study, a significantly longer OS was observed for patients who initiated cabozantinib treatment with 60 mg. The CABOREAL study did not report PFS. However, in our preliminary analysis on efficacy, we did not observe an effect of the 60 mg starting dose or a higher average cabozantinib exposure on PFS or OS, but it is important to note that our analysis a based on a limited number of heterogenous patients in terms of pretreatment and tumour histology. The median age in our cohort was 67 years. The observed median PFS of approximately 7.8 months in our study is in line with the reported median of 8.1 months for patients with clear cell RCC aged 65 to 74 in the METEOR phase 3 trial and just above the observed median of 7.0 months in a cohort of nonclear cell RCC patients. ³⁹ , ⁴⁰ Interestingly, we observed a longer PFS for patients with a dose reduction relative to the start dose, which may suggest a tipping point between optimal exposure versus long‐term tolerable exposure. However, this finding needs to be evaluated in a larger group of patients.

The phase II study of cabozantinib in SGC patients was terminated prematurely due to severe wound complications in six (24%) patients, mainly in areas which received prior radiotherapy. ²⁵ In our RCC cohort, five patients discontinued cabozantinib treatment due to toxicity. Four out of these five patients developed moderate to severe wound (healing) complications likely related to cabozantinib treatment. These percentages exceed the incidence of 1% to 3% reported in the registration studies. No clear associations could be made between cabozantinib exposure and the occurrence of these complications. Therefore, these complications may be a result of individual sensitivity and the presence of risk factors associated with wound complications or inclusion of patients that were not eligible for the registration studies. Cabozantinib is a strong inhibitor of VEGFR‐2, but the additional inhibition of C‐MET and TAM kinases may be an explanation for the high percentage of wound complications reported, as these kinases also play an important role in tissue repair and maintaining vascular integrity. ⁴¹ , ⁴² Clarification of these risk factors in future studies is of major importance to be able to identify patients at risk and ideally prevent these patients from severe toxicity.

Our study has some potential limitations. First, the CABO‐ASAP study was a phase II study whereas the RCC patients were collected from routine patient care. In general, trials participants are fitter and medication adherence is higher compared to patients in routine care and this may have contributed to a higher observed cabozantinib exposure in SGC patients. ⁴³ , ⁴⁴ However, this is likely of limited influence as the exposure at best tolerated dose was comparable between both groups. Furthermore, the exposure in RCC patients at 40 mg was just slightly lower than the reported average cabozantinib exposure in the phase III trial. ¹⁷ Another limitation is the small number of cabozantinib blood samples per patient in our analysis. The pharmacokinetic analysis in the phase II CABO‐ASAP was not an objective of the study. Consequently, we did not have a C _min level at each dose level of every patient. Cabozantinib samples of RCC patients were collected from routine care and although therapeutic drug monitoring is standard of care in our clinic, C _min levels were not available for all received dose levels, as patients sometimes required a rapid dose adjustment before a visit for PK evaluation could be scheduled. To be able to compare cabozantinib exposure between dose levels, we performed extrapolation based on the dose‐normalised average exposure of all available C _min levels. This was allowed as cabozantinib has been reported to exhibit dose proportional pharmacokinetics in the range of 20 to 140 mg. In addition, we showed an adequate correlation between the measured and calculated average cabozantinib C _min levels which supports the legitimacy of our approach. However, for some TKIs a decrease in exposure over time has been reported, possibly due to a decrease in absorption. ⁴⁵ , ⁴⁶ Cabozantinib samples taken at a later time during treatment may therefore distort this dose proportionality. Based on our data we could not detect a decrease in cabozantinib exposure over time and therefore believe that this does not affect our conclusions. Finally, even though the observations are based on limited number of patients, they show large similarity with the reported percentage of dose reductions and cabozantinib exposure in the registration studies. ¹⁸ , ²² Therefore, the pharmacology of cabozantinib in this relatively small group of patient appears to be representative for clinical practice.

5. CONCLUSION

In our study, we observed an unexpected higher cabozantinib exposure in patients with SGC compared to those with RCC even after correcting for dose differences between both groups. However, the cabozantinib exposure at BTD were comparable between both groups. At BTD the majority of patients did not reach the proposed target exposure of >750 μg/L. For most patients with RCC, the cabozantinib level at BTD corresponded to a dose of 40 mg. Therefore, a 40 mg start dose followed by dose‐adjustment based on exposure and tolerability may be preferred over the 60 mg dose to avoid unnecessary toxicity yet the effect of this approach on efficacy requires more research. Future studies should focus on identifying an optimal and tolerable exposure‐response target value for cabozantinib and elucidate factors that contribute to the differences in exposure, in order to individualise and improve treatment outcomes.

CONFLICT OF INTEREST

All mentioned relationships are outside the submitted work. FGAJ has been on an advisory board for Amgen and Genzyme. IMED received a research grant from Novartis. NPvE has received research grants from Novartis, Astellas, Janssen‐Cilag, Pfizer, Ipsen, has been on an advisory board for Pfizer, and received honoraria from Bayer and Sanofi. CMLvH has received research grants form AstraZeneca, Bristol Meyers Squibb, Merck Sharp and Dohme, Merck, Ipsen, Sanofi, and Novartis, has been on an advisory board for Bayer, Bristol‐Meyers Squibb, Ipsen, Merck Sharp and Dohme and Regeneron. SFM received honoraria from Pfizer, Roche and Merck Sharp and Dohme and Bristol Meyers Squibb. The other authors declare no conflicts of interest. (SDK, MU, WvB).

ETHICS STATEMENT

The current study has been performed in accordance with the Declaration of Helsinki (October 2013). The CABO‐ASAP (NCT03729297) phase II study has been approved by the medical ethical comity and the Institutional Review Board of the Radboudumc (IRC) and all patients gave written informed consent. The analysis in patients with RCC has been approved by the IRB of the Radboudumc (December 2018; 2018‐4617) and with their consent no informed consent was required, as we used data from routine care.

Supporting information

Figure S1 Correlation between measured and calculated average cabozantinib C _min levels for (A) the 40 mg dose level, (B) the dose level at start and (C) BTD level

Figure S2 Scatter plot of gender (A) and baseline albumin levels (B) vs calculated cabozantinib C _min levels at 40 mg

Figure S3 Scatter plot of the calculated individual cabozantinib C _min levels at the start dose for RCC patients who did and did not receive a dose reduction relative to the starting dose

Figure S4 Scatter plot of percentage weight loss from baseline vs average cabozantinib C _min exposure

Figure S5 Kaplan‐Meier curve of progression free survival for patients with advanced RCC with an exposure above and below the median average exposure of 573 μg/L

Figure S6 Kaplan‐Meier curve of progression free survival for patients with advanced RCC with an without dose reductions relative to the starting dose level

Figure S7 Kaplan‐Meier curve of overall survival for patients with advanced RCC with an exposure above and below the median average exposure of 573 μg/L

Figure S8 Kaplan‐Meier curve of overall survival for patients with advanced RCC with a 60 mg starting dose and those with a lower than 60 mg stating dose

Krens SD, van Boxtel W, Uijen MJM, et al. Exposure‐toxicity relationship of cabozantinib in patients with renal cell cancer and salivary gland cancer. Int. J. Cancer. 2022;150(2):308-316. doi: 10.1002/ijc.33797

DATA AVAILABILITY STATEMENT

The datasets generated and analysed for the current study are available from the corresponding author on reasonable request.