Cabozantinib as a Second-Line Agent in Advanced Hepatocellular Carcinoma

Prof. M. Kudo Editor Liver Cancer

Introduction

The results of the phase III CELESTIAL trial of cabozantinib were presented by Prof. Ghassan Abou-Alfa at the ASCO Gastrointestinal Cancer Symposium held in San Francisco from January 18 to 20, 2018 [1]. Although most of the previous clinical trials of second-line agents, except regorafenib [2], failed [3, 4, 5, 6, 7, 8], the CELESTIAL trial yielded positive results in line with most expectations and produced a fourth molecular-targeted drug option for hepatocellular carcinoma (HCC). Based on this trial, cabozantinib can be added as a second-line option to the first-line drugs sorafenib [9, 10] and lenvatinib [11] and the second-line drug regorafenib [2] (Table 1).

Table 1.

Phase III clinical trials of advanced stage HCC

		Design	Trial name	Result	Presentation	Publication	First author
First line	1	Sorafenib vs. sunitinib	SUN1170	Negative	ASCO 2011	J Clin Oncol 2013	Cheng
	2	Sorafenib ± erlotinib	SEARCH	Negative	ESMO 2012	J Clin Oncol 2015	Zhu
	3	Sorafenib vs. brivanib	BRISK-FL	Negative	AASLD 2012	J Clin Oncol 2013	Johnson
	4	Sorafenib vs. linifanib	LiGHT	Negative	ASCO-GI 2013	J Clin Oncol 2015	Cainap
	5	Sorafenib ± doxorubicin	CALGB 80802	Negative	ASCO-GI 2016
	6	Sorafenib ± HAIC	SILIUS	Negative	EASL 2016	Lancet Gastroenterol Hepatol 2018	Kudo
	7	Sorafenib ± Y90	SARAH	Negative	EASL 2017	Lancet Oncol 2017	Vilgrain
	8	Sorafenib ± Y90	SIRveNIB	Negative	ASCO 2017	J Clin Oncol 2018	Chow
	9	Sorafenib vs. lenvatinib	REFLECT	Positive	ASCO 2017	Lancet 2018	Kudo
	10	Sorafenib vs. nivolumab	CheckMate-459	Ongoing
	11	Sorafenib vs. durvalumab + tremelimumab vs. durvalumab	HIMALAYA	Ongoing
Second line	1	Brivanib vs. placebo	BRISK-PS	Negative	EASL 2012	J Clin Oncol 2013	Llovet
	2	Everolimus vs. placebo	EVOLVE-1	Negative	ASCO-GI 2014	JAMA 2014	Zhu
	3	Ramucirumab vs. placebo	REACH	Negative	ESMO 2014	Lancet Oncol 2015	Zhu
	4	S-1 vs. placebo	S-CUBE	Negative	ASCO 2015	Lancet Gastroenterol Hepatol 2017	Kudo
	5	ADI-PEG 20 vs. placebo	NA	Negative	ASCO 2016
	6	Regorafenib vs. placebo	RESORCE	Positive	WCGC 2016	Lancet 2017	Bruix
	7	Tivantinib vs. placebo	METIV-HCC	Negative	ASCO 2017
	8	Tivantinib vs. placebo	JET-HCC	Negative	ESMO 2017
	9	DT vs. placebo	ReLive	Negative	ILCA 2017
	10	Cabozantinib vs. placebo	CELESTIAL	Positive	ASCO-GI 2018		Ghassan
	11	Ramucirumab vs. placebo	REACH-2	Ongoing
	12	Pembrolizumab vs. placebo	KEYNOTE-240	Ongoing

Red, positive trials; blue, ongoing trials; black, negative trials. HCC, hepatocellular carcinoma; HAIC, hepatic arterial infusion chemotherapy; DT, doxorubicin-loaded nanoparticles.

Phase II Trial of Cabozantinib

The structural formula of cabozantinib is relatively similar to that of regorafenib [12, 13] (Fig. 1), although the kinase inhibitory activity (IC₅₀) of cabozantinib is different. Cabozantinib was originally identified as a dual inhibitor of VEGFR-2 and c-MET [14, 15], whereas current data suggest that it is a more potent inhibitor of MET, AXL, RET, FLT3, and TIE-2 than regorafenib (Tables 2, 3). VEGF, MET, and AXL are involved in tumor proliferation and angiogenesis, and MET and AXL are involved in the acquisition of resistance to antiangiogenic drugs [14, 15, 16, 17, 18]. VEGF, MET, or AXL expression is considered a poor prognostic factor [14, 15, 16, 17, 18].

Fig. 1.

Chemical structure of cabozantinib and regorafenib.

Table 2.

Cabozantinib targets VEGFR-2, c-MET, RET, AXL, TIE2, and FLT3

Biochemical activity	IC50, nmol/L
VEGFR-2	14
c-MET	2
c-KIT	752
RET	8
AXL	8
TIE2	13
FLT3	21
PDGFR-β	575

Table 3.

Mode of action: cabozantinib and regorafenib

Biochemical activity	Cabozantinib IC50, nM	Regorafenib IC50 ± SD, nM
MET	2	NA
AXL	8	NA
VEGFR-2	14	4.2±1.6
VEGFR-1	NA	13±0.4
VEGFR-3	NA	46±10
BRAF	NA	28±10
TIE-2	13	311±46
PDGFR-β	575	22±3
FGFR1	NA	202±18
c-Kit	752	7±2
RET	8	1.5±0.7
Flt-3	21	NA

Modified from [12, 13].

A waterfall plot from the phase II trial showed tumor shrinkage in a large proportion of patients. Progression-free survival (PFS) was 4.2 months in sorafenib-naïve patients and 5.5 months in sorafenib-pretreated patients, and overall survival (OS) was 11.5 months [15] (Table 4). The overall response rate (ORR) was 5%, the disease control rate was 81%, and PFS was 5.2 months (Table 4). Considering that some patients in the cabozantinib trial received first-line therapy, the results were not very good compared with the results of the phase II trial of regorafenib [19] (Table 4). Cabozantinib was also associated with a higher incidence of adverse events (AEs) than regorafenib (Table 5).

Table 4.

Comparison of efficacy (phase II): cabozantinib and regorafenib

	Cabozantinib (n = 41)	Regorafenib (n = 36)
ORR, %	5	3
DCR, %	81	72
PFS/TTP, months	5.2 (5.5)	4.3
OS, months	11.5	13.8

ORR, objective response rate; DCR, disease control rate; PFS, progression-free survival; TTP, time to progression; OS, overall survival.

Table 5.

Cabozantinib vs. regorafenib: comparison of adverse events (phase II)

	Cabozantinib (n = 41)		Regorafenib (n = 36)
	all grades	grade 3–4	all grades	grade 3–4
Hand foot skin reaction	56	15	53	14
Fatigue	56	2	53	17
Hypertension	24	10	36	3
Appetite loss	29	0	36	0
Nausea	37	2	33	0
Vomiting	37	2	14	0
Diarrhea	63	20	53	6
Body weight loss	22	2	19	0
Constipation	22	0	25	0

Values are shown as percentages. Modified from [15, 19].

Phase III CELESTIAL Trial

In light of these results, a phase III trial of cabozantinib was conducted (Fig. 2). The trial design was not as sophisticated as that of the RESORCE trial [2, 20]. For example, vascular invasion and/or extrahepatic spread was included as a stratification factor, which may result in an unfavorable imbalance regarding patients with vascular invasion. In fact, this unfavorable imbalance was present in the BRISK-PS trial and resulted in negative results [3]. The BRISK-PS trial did not include alpha-fetoprotein as a stratification factor, which caused an unfavorable balance against the trial drug similar to that seen in the REFLECT trial [11]. The RESORCE trial led to the inclusion of vascular invasion as an independent stratification factor and alpha-fetoprotein as a stratification factor in the design of trials of second-line drugs [21]. However, the CELESTIAL trial had a conventional design with few strategic elements (Table 6) and did not even exclude sorafenib-intolerant patients as in the RESORCE trial [2, 20, 21]. The only inclusion criteria regarding prior treatment were (a) prior sorafenib treatment, (b) progression following at least 1 prior systemic treatment for HCC, and (c) up to 2 prior systemic regimens for advanced HCC; the exact number of sorafenib-intolerant patients enrolled remains unclear.

Fig. 2.

CELESTIAL trial: study design.

Table 6.

Phase III clinical trials: advanced stage second line versus placebo

	BRISK-PS Brivanib	EVOLVE-1 Everolimus	REACH Ramucirumab	S-CUBE S1	RESORCE Regorafenib	METIV-HVV Tivantinib	KEYNOTE-240 Pembrolizumab	CELESTIAL Cabozantinib
Intolerance of sorafenib, %	12–13	18.5–20	13–15	30.6–33.8	0	17–21	-	N/A
Stratification factor	Reason for sorafenib discontinuation ECOG-PS score Extrahepatic spread, and/or vascular invasion	Region MVI	Region Cause of liver disease (HBV, HCV, other)	Medical institutions Extrahepatic metastasis and/or vascular invasion	Region ECOG-PS score Extrahepatic spread Vascular invasion AFP	Extrahepatic spread Vascular invasion AFP	Region Vascular invasion AFP	Region Disease etiology (HBV, HCV, other) Extrahepatic metastasis and/or vascular invasion

After the BRISK-PS trial, where there was an imbalance of AFP and MVI in the testing arm, AFP and MVI started to be included as independent stratification factors in most trials, but not in the CELESTIAL trial. Modified from [1, 2, 3, 4, 5, 6]. AFP, alpha-fetoprotein; MVI, macrovascular invasion.

Between September 2013 and September 2017, the trial enrolled 773 patients with unresectable HCC showing disease progression after at least 1 prior systemic chemotherapy regimen containing sorafenib. The second interim analysis performed in January 2016 demonstrated the superiority of cabozantinib in terms of the primary endpoint of OS. There was an imbalance in baseline patient characteristics between the cabozantinib and placebo groups caused by the failure to include vascular invasion and extrahepatic spread as independent stratification factors; namely, the rate of vascular invasion was only 27% in the cabozantinib group compared with 34% in the placebo group, which favored the cabozantinib group (Table 7). This resulted in significantly better OS in the cabozantinib group (10.2 months, 95% CI: 9.1–12.0) than in the placebo group (8.2 months, 95% CI: 9.1–12.0) and consequently in a positive result for the clinical trial. PFS, the secondary endpoint, was also better in the cabozantinib group (5.2 months, 95% CI: 4.0–5.5) than in the placebo group (1.9 months, 95% CI: 1.9–1.9) (Table 8). PFS of 1.9 months in the placebo arm in the CELESTIAL trial was similar to that of 1.5 months in the placebo arm in the RESORCE trial (Fig. 3). Moreover, ORR was superior in the cabozantinib group (4 vs. 0.4%; p = 0.0086) (Table 9). Post-trial treatment was performed in a comparably low proportion of patients in the cabozantinib and placebo groups (25 vs. 30%), demonstrating the poor condition of the patient population. In summary, although the relative number of sorafenib-intolerant patients in the trial was not reported, it can be inferred from the trial results that the proportion was relatively low (Table 10).

Table 7.

Baseline characteristics

	Cabozantinib (n = 470)	Placebo (n=237)
Median (range) age, years	64 (22–86)	64 (24–86)
Male, %	81	85
ECOG performance status 0/1, %	52/48	55/45
AFP ≥400 ng/mL, %	41	43
Enrollment region, %
Asia/Europe/North America/Pacific	25/49/23/3	25/46/25/5
Etiology of HCC, %
HBV	38	38
HCV	22	22
Other	40	41
Extrahepatic spread of disease, %	79	77
Macrovascular invasion, %	27	34
Extrahepatic spread and/or macrovascular invasion, %	85	84

Asia: Hong Kong, South Korea, Singapore, Taiwan; Pacific: Australia and New Zealand. Cited from [1]. AFP, alpha-fetoprotein; HCC, hepatocellular carcinoma.

Table 8.

Time to event: CELESTIAL (second and third line) versus RESORCE

	CELESTIAL trial (second and third line)			RESORCE trial (SOR → REG)
	cabozantinib (n = 470)		placebo (n = 237)	regorafeni (n=379)		placebo (n = 194)
TTP, months	N/A		N/A	3.2		1.5
HR		N/A			0.44
p value					<0.0001
PFS, months	5.2		1.9	3.1		1.5
HR		0.44			0.46
p value		<0.0001			<0.0001
OS, months	10.2		8.2	10.6		7.8
HR		0.76			0.63
p value		0.0049			<0.0001

TTP, time to progression; PFS, progression-free survival; OS, overall survival. Modified from [1, 2].

Fig. 3.

Phase III trial: second line. Time to progression (TTP) was shorter in the placebo arm in the CELESTIAL trial, similar to placebo arm in the RESORCE trial, suggesting that there were fewer sorafenib-intolerant patients in the CELESTIAL trial. Anticancer activity may be higher than that of other agents, as indicated by the longer TTP. MKI, multikinase inhibitor.

Table 9.

Tumor response: CELESTIAL vs. RESORCE

	CELESTIAL trial			RESORCE trial
	cabozantinib (n = 470)		placebo (n = 237)	regorafenib (n = 379)		placebo (n = 194)
Response criteria		RECIST 1.1			RECIST 1.1
ORR, %	4		0.4	6.6		2.6
p value		0.0086			0.02
DCR, %	64		33.4	65.7		34.5
p value		N/A			<0.0001

ORR, objective response rate; DCR, disease control rate. Modified from [1, 2].

Table 10.

Time to event: CELESTIAL (SOR → CAB) vs. RESORCE

	CELESTIAL trial (SOR→CAB)			RESORCE trial (SOR → REG)
	cabozantinib (n = 331)		placebo (n = 164)	regorafenib (n = 379)		placebo (n = 194)
TTP, months	N/A		N/A	3.2		1.5
HR		N/A			0.44
p value					<0.0001
PFS, months	5.5		1.9	3.1		1.5
HR					0.46
p value		0.40			<0.0001
OS, months	11.3		7.2	10.6		7.8
HR		0.70			0.63
p value					<0.0001

TTP, time to progression; PFS, progression-free survival; OS, overall survival. Modified from [1, 2].

Comparison between Regorafenib and Cabozantinib: Efficacy and Safety

Cabozantinib and regorafenib had comparable efficacy in terms of OS, ORR, and PFS (Tables 8, 9). Patients who received prior treatment with sorafenib alone showed slightly better outcomes (Table 10), which were comparable to those of regorafenib.

The duration of treatment with cabozantinib was 3.8 months, which was comparable to the 3.6 months for regorafenib and indicates acceptable tolerability, similar to that of regorafenib.

Dose reduction or discontinuation because of treatment-related AEs was more common with cabozantinib than with regorafenib. Specific AEs such as palmar-plantar erythrodysesthesia, diarrhea, and asthenia were more common with cabozantinib than with regorafenib, indicating that cabozantinib may have a slightly higher toxicity than regorafenib (Table 11).

Table 11.

Safety analysis: CELESTIAL vs. RESORCE

	Cabozantinib (n = 467)	Regorafenib (n = 374)
Treatment duration, months	3.8	3.6
Dose reduction due to adverse event, %	62	48
Discontinuation due to TRAE, %	16	10
Grade 3/4
Any grade 3 or 4 adverse event, %	68	66
Palmar-plantar erythrodysesthesia, %	17	13
Fatigue, %	10	9
Hypertension, %	16	15
Diarrhea, %	10	3
Asthenia, %	7	NA
Bilirubin increased, %	NA	10
AST increased, %	12	11
Ascites, %	NA	4
Anemia, %	4	5
Hypophosphatemia, %	NA	9

AST, aspartate aminotransferase; NA, not applicable. TRAE, treatment-related adverse event. NCI-CTCAE v4.03.

Key Factors Contributing to the Success of the CELESTIAL Trial

The following 5 factors may have contributed to the success of the CELESTIAL trial of cabozantinib despite the unsophisticated trial design compared with that of the RESORCE trial and the drug's slightly higher toxicity (Table 12).

Table 12.

CELESTIAL trial: key factors of the success

• Cabozantinib has good anticancer activity

• Acceptable toxicity and tolerability

• Imbalance of vascular invasion favoring cabozantinib

• Small number of sorafenib-intolerant patients (short time to progression in placebo)

• Extremely high numbers of enrolled patients (n = 470 vs. 379, 362, 283, 263)

→ Higher power to detect the small difference and eliminate the effect of tiny imbalance

1. Cabozantinib has a sufficiently potent antitumor activity.

2. Toxicity and tolerability were clinically acceptable.

3. An imbalance in vascular invasion favored cabozantinib.

4. The short time to progression in the placebo arm and low proportion of patients having post-trial treatment indicate low enrollment of sorafenib-intolerant patients, which was similar to no enrollment of sorafenib-intolerant patients in the RESORCE trial.

5. The sample size of 470 patients was considerably higher than that of other second-line trials and provided sufficient power to eliminate the effect of the small imbalance and detect small differences as significant (Table 13).

Table 13.

Phase III trials in a second line setting

	CELESTIAL cabozantinib arm (n = 470)	RESORCE regorafenib arm (n = 379)	BRISK-PS brivanib arm (n = 263)	EVOLVE-1 everolimus arm (n = 362)	REACH ramucirumab arm (n = 283)
Male, %	81	88	82	84	83
Median age (range), years	64 (22–86)	64 (19–85)	64 (19–89)	67 (21–86)	64 (28–87)
Asian race, %	25	41	48	38	46
ECOG PS 0/1, %	52/48	65/35	57/39	59/36	56/44
Child Pugh A, %	NA	98	92	98	98
BCLC stage, B/C, %	NA	14/86	9/87	14/87	12/88
AFP ≥400 ng/mL, %	41	43	50a	47a	42
MVI, %	27	29	31	33	29
EHS, %	79	70	65	74	73
Etiology, %
Alcohol	NA	24	23	18	-
HBV	38	38	39	25	35
HCV	22	21	28	26	27
NASH	NA	7	-	4	-
Intolerance of sorafenib, %	-	0	13	19	13
Median total duration of prior sorafenib, months	5.3	7.8	-	-	-
Median time from disease progression to randomization, months	1.6	0.9	-	-	-

AFP, alpha-fetoprotein; MVI, macrovascular invasion; EHS, extrahepatic spread; NASH, non-alcoholic steatohepatitis.

^aAFP ≥200 ng/mL.

Paradigm Shift in the Treatment Strategy for HCC

Sorafenib was the only HCC drug available between 2007 and 2016. Between 2017 and 2018, 5 drugs, sorafenib, lenvatinib, regorafenib, cabozantinib, and nivolumab, became available. Therefore, it is necessary to establish how these drugs should be used in clinical practice (Fig. 4). Combinations of immune checkpoint inhibitors and molecular-targeted drugs or molecular-targeted drugs and established locoregional therapies [22] are particularly likely to produce a paradigm shift in the treatment of HCC. The treatment landscape for HCC will soon undergo major changes as systemic therapy is integrated into the treatment for all stages, from early to intermediate to advanced, which could drastically improve the prognosis of patients with HCC.

Fig. 4.

New treatment landscape in HCC associated with the emergence of multiple molecular-targeted agents. Identification of the subgroup that easily develops to TACE failure/refractories may be important. BSC, best supportive care.

Conclusion

The success of the clinical trial of cabozantinib increased the treatment options for HCC, and combination treatment with immunotherapy may soon improve the prognosis of patients with HCC.