The present retrospective study was performed in five institutions: Kurume University Hospital (Kurume, Japan), Yokokura Hospital (Miyama, Japan), Omuta City Hospital (Omuta, Japan), Yanagawa Hospital (Yanagawa, Japan) and Iwamoto Internal Medical Clinic (Kitakyushu, Japan). The current protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki, as reflected by the prior approval of the Ethical Committees of Kurume University School of Medicine. An opt-out approach was employed to obtain informed consent from patients, and personal information was protected during data collection.

The patient inclusion criteria for the present study were as follows: i) Diagnosis of intermediate-stage HCC refractory to TACE according to the Japan Society of Hepatology (8,21); ii) age >18 years; iii) Eastern Cooperative Oncology Group performance status 0 (22); and iv) complete follow-up until death or study cessation (June 15, 2019). The patient exclusion criteria were as follows: i) A history of malignant tumors other than HCC in the 5 years preceding the present study; ii) participation in any clinical trial; iii) history of pretreatment with TKIs; iv) Child-Pugh class B or C (23); v) creatinine levels >1.5 mg/dl (normal level 0.5–1.3 mg/dl); vi) infiltrative HCC, defined as a true infiltration of tumor cells into the liver parenchyma, a confluence of tiny nodules or both in computed tomography (CT) or magnetic resonance imaging (MRI) scans (24); vii) the presence of portal vein thrombosis or extrahepatic metastasis; viii) the presence of ascites; ix) esophageal varices with high risk of rupture; and x) a history of choledochojejunostomy or liver transplantation.

A total of 641 consecutive patients with HCC who underwent TACE between Jan 1, 2009, and Feb 31, 2017, were registered; data cut-off for this analysis was June 15, 2019. Intermediate-stage HCC refractory to TACE was observed in 210 patients, and 39 patients with Child-Pugh class B or C were excluded from the analysis. Therefore, a total of 171 patients were enrolled in the present study. These patients were classified into three groups according to their treatment for HCC: The lenvatinib (Eisai Co., Ltd, Tokyo; n=45), sorafenib (Bayer Yakuhin, Ltd., Osaka, Japan; n=53) and TACE (n=73) groups (Fig. 1).

HCC was diagnosed using a combination of following serum markers and imaging: α-fetoprotein (AFP) and des-γ-carboxy prothrombin (DCP), and imaging procedures, including ultrasonography, CT and MRI scans. HCC was classified using the Barcelona Clinic Liver Cancer (BCLC) staging system (4).

HCC refractory to TACE was defined as previously described (8). Briefly, in the present study, HCC refractory to TACE was defined if any of following conditions were fulfilled: i) ≥2 consecutive ineffective responses of treated tumors (viable lesions >50%); ii) ≥2 consecutive progressive increases in total tumor count; or iii) continuous elevation in AFP or DCP levels after TACE.

NLR was calculated by the ratio of neutrophil count to lymphocyte count: By dividing neutrophil count by lymphocyte which were the fraction of white blood cell (25).

The treatment strategy for HCC refractory to TACE was based on previous reports (10,26). Sorafenib was administrated for patients with refractory to TACE; however, patients who refused sorafenib were treated with TACE. After lenvatinib was approved in Japan, 45 patients were treated with lenvatinib instead of sorafenib. After obtaining written informed consent from each patient, lenvatinib was orally administered at 12 mg/day in patients with bodyweight ≥60 kg or 8 mg/day in patients with bodyweight <60 kg. Sorafenib was orally administered at 400 mg twice daily, regardless of bodyweight. Discontinuation and dose reduction of TKIs were based on the manufacturer's protocol (lenvatinib; Eisai Co., Ltd; sorafenib; Bayer Yakuhin, Ltd.).

TACE was performed by doctors with >10 years of experience in interventional therapy at the start of the present study (27). A catheter was inserted into the tumor-feeding artery; subsequently, anticancer drugs, such as epirubicin (20–50 mg) or cisplatin (20–50 mg) (depending on the size and number of tumors), were manually emulsified with lipiodol (Guerbet Japan K.K.) and administered, followed by embolization with 1-mm absorbable gelatin sponge particles (Nippon Kayaku Co., Ltd.).

HCC was evaluated using CT or MRI scans 4–6 weeks after the initiation of treatment, and thereafter every 2–3 months until death or study cessation. The therapeutic response was evaluated using the Modified Response Evaluation Criteria in Solid Tumors (28): Complete response was defined as the disappearance of any intratumoral arterial enhancement in all target lesions; partial response (PR) was defined as a ≥30% decrease in the sum of the diameters of viable (contrast enhancement in the arterial phase) target lesions; progressive disease (PD) was defined as an increase of ≥20% in the sum of the diameters of viable (enhancing) target lesions, taking as reference the smallest sum of the diameters of viable (enhancing) target lesions recorded since the treatment started; and stable disease was defined as any other case that did not qualify for either PR or PD. Upon HCC recurrence, additional treatment was selected based on the evidence-based clinical practice guidelines of the BCLC staging system and treatment strategy (4).

Adverse events (AEs) and serious adverse events (SAEs) were monitored and recorded. AEs were assessed according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) v4.0 (29). In the lenvatinib and sorafenib groups, when a patient developed any grade ≥3 SAE (according to CTCAE), or if any unacceptable grade 2 drug-related AE occurred, the drug dose was reduced or discontinued. Additionally, the effects of the treatments on liver function were evaluated by measuring changes in the albumin-bilirubin (ALBI) score. ALBI score was calculated as previously described (30) based on serum albumin and total bilirubin levels; ALBI-score=[log₁₀ bilirubin (µmol/l) ×0.66]+[albumin (g/l) x-0.085], and was graded as following: ≤-2.60=ALBI grade 1, >-2.60 to ≤-1.39=ALBI grade 2, >-1.39=ALBI grade 3).

All data are presented as the number or median (range). All statistical analyses were performed using JMP Pro v13 (SAS Institute, Inc.). Categorical variables were compared using the χ² test, and continuous variables were compared using one-way ANOVA with Scheffe's post hoc test. OS and PFS rates were compared between the groups using the log-rank test or Bonferroni method. Variable factors associated with PFS were analyzed using decision-tree analyses. Univariate and multivariate analyses were conducted using the Cox proportional hazards model to identify risk factors associated with PFS. P<0.05 (two-tailed) was considered to indicate a statistically significant difference.

The characteristics of the enrolled patients are shown in Table I. The median age was 72 years, and 18.1% (31/171) of the patients were female. The median BMI was 22.7 kg/m². ALBI grade 1 was observed in 29.2% (50/171) of patients, and the median NLR was 2.25. The median tumor size was 27 mm, and 84.2% (144/171) of patients were beyond the up-to-seven criteria, beyond 7 being the sum of the maximum size and number of tumors for any given HCC case (17) (Table I). In the TACE group, the female-to-male ratio was significantly higher than that in the lenvatinib and sorafenib groups (Table I). No significant differences were seen for age, BMI, etiology of liver disease, AFP level or NLR among the three groups. Furthermore, no significant differences were observed in the prevalence of beyond the up-to-seven criteria or ALBI grade among the three groups.

The overall objective response rate (ORR) was 42.2% (19/45), 1.9% (1/53) and 6.8% (5/73) in the lenvatinib, sorafenib and TACE groups, respectively, 1 month after treatment (Table II). The disease control rate (DCR) was 88.9% (40/45), 56.6% (31/53) and 19% (14/73) in the lenvatinib, sorafenib and TACE groups, respectively, 1 month after treatment (Table II).

In the present study, the RDI was 80.7 and 73.3% at 4 and 8 weeks, respectively. There was a significant difference in ORR between patients in the 8 week-RDI ≥75% group and the 8 week-RDI <75% group [ORR, 10/24 (41.7%) vs. 5/21 (23.8%), respectively; P=0.02] (date not shown).

The median PFS time was 5.8, 3.2 and 2.8 months in the lenvatinib, sorafenib and TACE groups, respectively (Fig. 2). Among the three groups, the longest median PFS time was observed in the lenvatinib group. The median PFS time in the lenvatinib group was significantly longer than that in the sorafenib and TACE groups [lenvatinib vs. sorafenib: Hazard ratio (HR), 0.56; 95% confidence interval (CI), 0.36–0.88; P=0.01; lenvatinib vs. TACE: HR, 0.23; 95% CI, 0.15–0.36; P<0.001].

Treatment with lenvatinib, being within the up-to-seven criteria and ALBI grade 1 were selected as variables via univariate analysis. In the multivariate analysis, treatment with lenvatinib and being within up-to-seven were identified as independent factors for PFS (Table III).

The period of PFS was 130±114.5 days at the study censor time. To determine the profile for PFS, a decision-tree analysis was performed. Treatment for HCC was selected as the variable for the initial split. PFS time was 80.3±44.7 days in patients treated with TACE (Fig. 3); in patients receiving treatments other than TACE and beyond the up-to-seven criteria, the PFS time was 148.8±104.1 days. In patients beyond the up-to-seven criteria, lenvatinib treatment was selected as the third split. The PFS time was 184.8±101.8 and 117.6±96.8 days in patients with and without lenvatinib treatment, respectively (Fig. 3). In addition, the ALBI grade was selected as the fourth split in patients treated with lenvatinib. The PFS time was 245.2±107.9 and 147.1±78.6 days in patients with ALBI grades 1 and 2, respectively.

In the Lenvatinib group, an estimated median OS was not observed within the present study (Fig. S1).

The ORRs were 66.7% (12/18) and 25.9% (7/27) in patients with ALBI grade 1 and 2, respectively (data not shown). The median medication period with lenvatinib in patients with ALBI grades 1 and 2 was 11.2 and 5.9 months, respectively (P=0.002; Fig. S2).

In the TACE group, 44 patients were treated with repeated TACE, 10 patients were treated with repeated TACE followed by sorafenib after reaching BCLC stage C, 9 patients were treated with hepatic arterial infusion chemotherapy (HAIC) and 10 patients received palliative care after obtaining informed consent, as hepatic function deteriorated to Child-Pugh class C. In the sorafenib group, 29 patients were treated with TACE, 13 patients with HAIC, 3 patients with regorafenib and 8 patients received palliative care. In the lenvatinib group, 16 patients were treated with TACE, 8 patients were treated with HAIC, 7 patients were continuously treated with lenvatinib even after disease progression, 3 patients were treated with sorafenib, 1 patient was treated with radiation therapy and 8 patients were treated with lenvatinib alone. Furthermore, in patients whose hepatic function deteriorated to Child-Pugh class B, TACE was performed after obtaining informed consent; in patients whose hepatic function deteriorated to Child-Pugh class C, palliative care was selected after obtaining informed consent.

AEs (grade ≥3) were determined by the attending physician for the lenvatinib, sorafenib and TACE groups, and are shown in Table IV. In the lenvatinib group, the most common AEs were hypertension and fatigue, which occurred in 15.6% (7/45) and 13.3% (6/45) of the patients, respectively. In the sorafenib group, hand-foot-skin-reactions occurred in 17.0% (9/53) of the patients. In the TACE group, ascites occurred in 9.6% (7/73) of the patients.