Study design

The primary objective was to determine the safety and tolerability of GC33 in Japanese patients with advanced HCC, and the secondary objectives were to evaluate the PK, efficacy, and biomarkers. This trial was carried out at three sites across Japan, namely, the National Cancer Center Hospital East (Chiba), the Kanagawa Cancer Center Hospital (Kanagawa), and the National Cancer Center Hospital (Tokyo).

The patients enrolled were treated with GC33 at 5, 10, or 20 mg/kg given every week by i.v. infusion over a period of 30–90 min. Each cycle was defined by approximately 4 weeks of treatment. GC33 treatments were scheduled to continue until progressive disease (PD), occurrence of unacceptable toxicities, or the patient's withdrawal of consent. The starting dose of 5 mg/kg was chosen based on the dose of GC33 estimated from xenograft mouse models that was expected to still have potential for some antitumor activity,(¹³) while taking into account the safety and PK results from the phase I study carried out in the USA.

This study was a conventional 3 + 3 dose-escalation design. At least three patients in each of the 5 and 10 mg/kg cohorts, and at least six patients in the 20 mg/kg cohort were to be treated and monitored from the first dosing of GC33 until prior to the fifth dosing (cycle 1), as described previously.(¹⁴) Dose-limiting toxicity was defined as any grade 3 or higher toxicity occurring during cycle 1 (as categorized by the NCI's Common Terminology Criteria for Adverse Events version 4.03)(¹⁹) other than grade 3 infusion reaction, transient electrolyte abnormalities, grade 3 decreased lymphocyte count, grade 3 decreased platelet count not requiring platelet transfusion, and <10× ULN increase in hepatic enzymes (e.g., increased alanine aminotransferase, increased aspartic aminotransferase, increased alkaline phosphatase), grade 3 hepatic function abnormal (decreased serum albumin, increased total bilirubin, increased prothrombin time-international normalized ratio), and any non-drug-related toxicities. A medical monitor and an independent Data Safety Monitoring Board reviewed the safety data of each cohort and dose escalation.

Study assessment

Types and frequency of adverse events (AEs) and causal relationships between AEs and GC33 were evaluated. Incidence and severity of AEs were collected and graded according to the NCI's Common Terminology Criteria for Adverse Events version 4.03. Particular attention was paid to infusion-associated symptoms possibly as a result of GC33 infusion. Laboratory evaluations during the study included hematology, blood biochemistry, blood coagulation, and natural killer (NK) cell count. Anti-GC33 antibodies were measured at screening and during the study.

Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 was used to assess overall response and progression-free survival (PFS).(²⁰) Tumor burden and response to treatment were evaluated at the baseline and every 8 weeks by imaging (computed tomography or MRI). α-Fetoprotein (AFP) and des-γ-carboxy prothrombin (DCP) were also measured at the baseline and during the study.

Tumoral expression of GPC3 was centrally examined and evaluated in biopsied specimens by two IHC methods, each using mouse anti-human GPC3 mAbs.(²¹) Method 1, which was the method used in the previous FIH study, was carried out at Charles River Laboratories Inc. (Reno, NV, USA) and assessed by three qualified pathologists blinded to clinical outcomes using the scoring criteria shown in Table ​Table11.(¹⁴) Method 2, a fully automated IHC assay, was carried out at Ventana Medical Systems Inc. (Tucson, AZ, USA) using anti-glypican 3 (GC33) mouse monoclonal primary antibody (Catalog number 790–4564; Ventana Medical Systems) according to the manufacturer's instructions. The immunostaining resulting from method 2 was assessed by two qualified pathologists blinded to clinical outcomes. Staining intensity was assessed in the cytoplasm and at the cell membrane separately. Staining intensity was scored on a four-point scale: 0, no staining was observed; 1 (weak), an intensity identified by the pathologist as higher than background and specific to the cells of interest; 2, moderate staining; and 3, maximal (strong) staining. The percentage of cells with specific antibody reactivity was scored for the tumor component of the specimen. The H score was derived by summing the percentage of cell staining at each intensity (weak, moderate, strong) multiplied by the weighted intensity of staining.(²²) Following the evaluation of GPC3 IHC staining intensity, the percentage of tumor cells stained, and the pattern of membrane and/or cytoplasmic positivity, a clinical score for method 2 was assigned according to the criteria described in Table ​Table22.

Table 1

Scoring system of immunohistochemical (IHC) staining of glypican-3 (GPC3) in hepatocellular tumor specimens using method 1

	Score
Positive cell rate (PR)
No staining	0
<20%	1
20–49%	2
≥50%	3
Staining intensity (SI)†
Cytoplasm (SI-Cp)
No staining	0
Minimal staining	1
Cell membrane (SI-Cm)
Weak staining	2
Moderate staining	3
Strong staining	4
Staining pattern of cell membrane (SP-Cm)‡
No cell membrane pattern	0
Incomplete or partly complete (<20%)	1
Focally complete (20–49%)	2
Rather complete (≥50%)	3
GPC3 staining scoring	IHC total = PR + SI-Cp + SI-Cm + SP-Cm

Open in a separate window

^†Scoring method 1, used in a previous first-in-human study, involved assessment by three qualified pathologists blinded to clinical outcomes using the scoring criteria shown here. †GPC3 staining categories: minimal, slight staining recognized at low-power objective, ×10; weak, slight staining recognized at low-power objective, ×4; moderate, staining found easily at ×4, but weaker than that of strong staining; strong, staining apparent strong with low-power objective, ×4.

^‡Complete staining, circular (ring-like) staining of tumor cells; incomplete staining, focal staining of rim of cells.

Blood samples used for serum GC33 measurements were drawn at pre-dose, at the end of the first infusion, and at 1, 8, 24, 48, 72, and 120 h after the end of the first infusion. In addition, trough concentrations of serum GC33 were measured prior to each subsequent infusion until the end of cycle 3. GC33 in the serum was assayed as described previously.(¹⁴) Inter-assay accuracy and precision was within ±20%, and the coefficient of variation was <20%. Pharmacokinetic parameters for GC33 in serum were calculated by non-compartmental methods using data obtained after the first dosing.

Dose and duration of therapy

Of the 13 enrolled patients, four were assigned to cohort 1 (5 mg/kg), three were assigned to cohort 2 (10 mg/kg), and six were assigned to cohort 3 (20 mg/kg). Among the 13 patients, 12 were included in the DLT evaluation, and one patient was excluded as a result of withdrawing due to PD during the DLT evaluation period (in cohort 1). The mean duration of exposure was 99.0 days (range, 29–280 days) and the mean number of doses administered was 14.4 (4–40). Dose intensity was high across all patients in cohorts 1–3, with a mean relative dose intensity of 95.4% (range, 79.5–100%). Good compliance of administration was observed as relative dose intensities were over 90.0% in all three cohorts.

Safety and tolerability

GC33 up to 20 mg/kg/week was considered to be well tolerated as no DLTs were reported in any cohort and a maximum tolerated dose (MTD) was not reached. A summary of the AEs occurring in over 30% of patients and grade 3 AEs by dose group are listed in Tables ​Tables44 and ​and5.5. All patients had at least one AE, and the most common AEs were lymphocyte count decreased (77%), NK cell count decreased (77%), C-reactive protein increased (69%), and pyrexia (62%). Grade 3 AEs were reported in 54% of patients, and grade 3 AEs occurring in two or more patients were blood pressure increased (23%), lymphocyte count decreased (23%), and platelet count decreased (15%). No grade 4 or 5 AEs were reported. No AEs led to death or discontinuation.

Infusion reactions were reported in 62% (8 out of 13 patients) and all of them were medically manageable. The most common infusion-associated symptoms were pyrexia (five patients, 38%), increased C-reactive protein (four patients, 31%), and nausea (three patients, 23%), and grade 3 increased blood pressure was reported in two patients in the 20 mg/kg cohort. The other infusion-associated symptoms were either grade 1 or 2. Almost all of the infusion-associated symptoms occurred on the third day after the first infusion. The investigators or sub-investigators provided premedication to prevent infusion reactions and provided appropriate treatment for any infusion reactions (acetaminophen, dexchlorpheniramine, dexamethasone sodium phosphate, loxoprofen sodium hydrate, prochlorperazine maleate). All infusion-associated symptoms were resolved.

The incidence of AEs seemed not to be dose-dependent. There was no evidence of cumulative toxicity and there was no difference in the incidence of AEs by GPC3 expression level. No anti-GC33 antibodies were detected in any of the patients at pre-treatment or post-treatment.

Pharmacokinetics

The PK data were evaluated for all 13 patients in all cohorts. The PK parameters of GC33 are shown in Table ​Table6.6. At doses of 5, 10, and 20 mg/kg, the mean half-life (t_1/2) was 4.17, 7.01, and 6.13 days and the mean total clearance was 0.566, 0.373, and 0.510 L/day, respectively. The PK exposure (C_max and AUC_inf) after the first dose increased as the dose increased. The AUC_inf showed a dose-proportional increase from the 5 mg/kg dose group to the 20 mg/kg dose group. The trough concentrations of GC33 appeared to reach a steady state after the fourth to the sixth dose. The trough concentrations after reaching a steady state in all patients in all cohorts were over 30 μg/mL, which was the predicted effective concentration in the xenograft models.

Table 6

Pharmacokinetics parameters of GC33, a humanized antibody against glypican-3

	t1/2 (day)	CL (L/day)	Cmax (μg/mL)	AUCinf (day μg/mL)	Vd,ss (L)	Ctrough (week 4; μg/mL)	Ctrough (week 6; μg/mL)
5 mg/kg
No. of patients	3	3	4	3	3	3	2
Mean (min.–max.)	4.17 (2.90–5.39)	0.566 (0.452–0.746)	141 (114–186)	584 (419–759)	3.13 (2.56–3.80)	61.7 (38.9–73.9)	61.2 (33.9–88.4)
CV (%)	29.9	27.9	22.3	29.2	20.0	32.0	63.0
10 mg/kg
No. of patients	3	3	3	3	3	3	3
Mean (min.–max.)	7.01 (6.40–7.66)	0.373 (0.327–0.432)	258 (237–270)	2020 (1790–2390)	3.70 (3.53–3.97)	259 (221–299)	288 (246–353)
CV (%)	9.02	14.3	7.16	16.0	6.55	15.1	19.9
20 mg/kg
No. of patients	6	6	6	6	6	5	5
Mean (min.–max.)	6.13 (3.80–8.02)	0.510 (0.308–0.816)	395 (326–505)	2670 (1480–3580)	4.12 (3.23–5.31)	406 (325–532)	477 (379–616)
CV (%)	28.9	33.9	17.2	25.7	17.6	19.5	23.8

Open in a separate window

PK parameters for GC33 in serum were calculated by non-compartmental methods. Pharmacokinetic parameters estimated following the first GC33 infusion are: t_1/2, elimination half-life; CL, serum clearance; C_max, maximum drug concentration; AUC_inf, area under the concentration time curve from time 0 to infinity; and Vd,ss, volume of distribution at steady state. In addition, C_trough (predose concentrations) at week 4 and 6 are summarized. CV, coefficient of variation.

Tumoral GPC3 expression

All 13 unique core-needle biopsied specimens taken from tumor lesions in the liver were stained by both methods and were evaluated by using their respective scoring criteria. The representative GPC3-IHC features for both staining methods are shown in Figure ​Figure1(a)1(a) and all cases are shown in Figure S1. Twelve patients had a total GPC3 staining score of 7 or more by method 1, and 12 patients had positive clinical scores by method 2 (Fig. ​(Fig.1b).1b). Both staining methods produced a similar staining pattern in the majority of specimens. The GPC3 staining score of method 1 was highly correlated with the clinical score of method 2 (Spearman's correlation coefficient, 0.76; P = 0.00255). Higher clinical scores appeared to be associated with increased H scores for both the membrane and cytoplasm in this limited number of samples (Fig. ​(Fig.11c,d).

Open in a separate window

Fig. 1

Immunohistochemistry (IHC) of glypican-3 (GPC3) in hepatocellular carcinoma. (a) Representative GPC3 staining features observed in matched specimens evaluated using two IHC methods. Method 1 was used in a previous first-in-human study; method 2 was a fully automated IHC assay. Scores are indicated under each photograph. Bar = 50 μm (b) Comparison of total GPC3 staining score by method 1 (0–14) and clinical score by method 2 (0–3). Spearman's correlation was 0.76 (P = 0.00255). (c) Comparison of clinical score and membrane H score. (d) Comparison of clinical score and cytoplasm H score. There was a positive association between the H scores and the GPC3 clinical scores, with R² values of 0.75 and 0.33, respectively.

We thank all of the participating patients and their families, as well as the investigators, research nurses, study coordinators, and operations staff. The authors thank the pathologists Drs. Janine Feng and Bharathi Vennapusa (Ventana Medical Systems Inc., USA), Dr. Kevin S. McDorman (Charles River Laboratories Inc., USA), Dr. Vikram Deshpande (Massachusetts General Hospital, USA), and Dr. Hiroaki Kataoka (University of Miyazaki, Japan), Norihisa Ohishi for the PK analysis and discussion, and Toshihiko Ohtomo, Atsuhiko Kato, and Zhaodi Gu for their assistance.