Comparison of the outcomes between sorafenib and lenvatinib as the first-line systemic treatment for HBV-associated hepatocellular carcinoma: a propensity score matching analysis

This study was a single-center study in Korea. Patients treated with either sorafenib or lenvatinib at a tertiary referral center (Seoul National University Hospital, Seoul, Korea) from January 1, 2018 to April 30, 2020, as a first-line systemic therapy for HBV-related advanced HCC and patients who were not eligible for surgical resection were screened for potential inclusion in the study. For a first-line systemic therapy of advanced HCC, selection of sorafenib or lenvatinib was decided by the clinician because both drugs were suggested in Korean practice guidelines [20]. Patients who received previous local anti-HCC treatments such as radiofrequency ablation (RFA), transcatheter arterial chemoembolization (TACE), transarterial radioembolization (TARE), or hepatic resection, and also liver transplantation prior to sorafenib or lenvatinib treatment were included. Among the 376 eligible patients, patients were excluded if they had received previous systemic treatment (n = 34), discontinued lenvatinib or sorafenib within 20 days after initiation of treatment (n = 49), lost to follow-up before the first tumor response assessment (n = 74), or had insufficient information (n = 13). Finally, 44 patients treated with lenvatinib and 162 patients treated with sorafenib remained after exclusion. Consecutive PSM was conducted by 1:2 (lenvatinib:sorafenib) matching, and finally 44 patients treated with lenvatinib and 88 patients treated with sorafenib were included for statistical analysis as matched cohort (Fig. 1).

The following clinical parameters were collected from all 206 patients: (1) demographics including age, sex, Eastern Cooperative Oncology Group (ECOG) performance status, Child–Pugh classification; (2) serum biochemical parameters at baseline including total bilirubin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), prothrombin time (PT), alpha-fetoprotein (AFP), protein induced by vitamin K absence or antagonists-II (PIVKA-II); (3) parameters on tumors including Barcelona Clinic Liver Cancer (BCLC) stage, initial and last follow-up tumor size measured according to the modified Response Evaluation Criteria in Solid Tumor (mRECIST), the presence of extrahepatic metastasis, portal vein invasion, biliary involvement, and previous local treatment of HCC; (4) date of treatment initiation; (5) date of tumor progression; (6) date of death or last follow-up; (7) antiviral drug history; (8) viral status.

Primary outcome was OS, which was defined as the time interval from the date of initiation of sorafenib or lenvatinib to the date of death or last follow-up. Secondary outcomes were PFS, TTP, and tumor response. PFS was defined as the time from the date of initiation of sorafenib or lenvatinib to the date of radiologic or clinical progression or death from any cause. TTP was defined as the time from the date of initiation of sorafenib or lenvatinib to the date of radiologic or clinical progression. The best tumor response was assessed using the mRECIST. The best tumor response was defined as the best response recorded from the start of sorafenib or lenvatinib treatment until disease progression. Objective response rate (ORR) was defined as complete response (CR) or partial response (PR). Disease control rate (DCR) was defined as CR, PR, or stable disease (SD). Predictive factors associated with OS, PFS and TTP were evaluated, and subgroup analysis was conducted.

Continuous variables were denoted by the median with interquartile range (IQR), and categorical variables were denoted by counts with percentages. Categorical variables were compared using the χ² test. Kaplan–Meier curves were used to estimate OS, PFS and TTP, and the log-rank test was used for group comparisons. The log-rank test was performed with the matched cohort. Hazard ratios (HRs) were estimated using the Cox proportional hazard model. Univariable and multivariable Cox proportional hazard analyses were performed with the unmatched cohort to assess the effects of variables on each outcome of interest. The variables with p < 0.05 in the univariable analysis as well as the variable of interest (i.e., lenvatinib treatment (vs. sorafenib)) were used in the multivariable analysis.

To reduce confounding, propensity score was used to match patients treated with lenvatinib to patients treated with sorafenib. The following six covariates expected to be associated with the prognosis of uHCC were considered: age, sex, Child–Pugh classification, BCLC stage, AFP, and tumor size. AFP was stratified into two groups: AFP > 200 ng/mL and AFP ≤ 200 ng/mL. PSM was performed using the nearest-neighbor 1:2 (lenvatinib:sorafenib) matching method with a caliper width of 0.1.

Standardized mean differences (SMDs) were used to evaluate the balance of the observed covariates across groups before and after matching. SMDs were the absolute value of the difference in the mean or proportions divided by the square root of the mean variance [21]. SMDs < 0.25 indicated an acceptable balance of a baseline covariate between the two groups [22].

Before PSM, a significant imbalance was observed for PT, AFP, Child–Pugh class, ECOG, and subsequent anti-HCC treatment as follows: higher PT international normalized ratio (INR), less frequent AFP > 200 ng/mL, more patients in Child–Pugh class B or C, more patients in ECOG 0, and more patients received subsequent anti-HCC treatment in the lenvatinib group compared to the sorafenib group (Table 1). The final analyses for the primary and secondary outcomes were performed on 44 patients treated with lenvatinib and 88 patients treated with sorafenib (Fig. 1). After PSM, the SMDs of most of the variables were < 0.25, which indicated an acceptable balance between patients treated with lenvatinib or sorafenib after matching (Table 1). Significantly different baseline characteristics were not observed between patients treated with lenvatinib or sorafenib except for subsequent anti-HCC treatment (34.1% vs 52.3%, SMD = − 0.439).

After PSM, the median patient age was 58 years. Most patients (90.9%) were male in both groups. The median PTs were 1.16 INR (IQR, 1.06–1.24 INR) in the lenvatinib group and 1.10 INR (IQR, 1.02–1.20 INR) in the sorafenib group. The median AFPs were 132.6 ng/mL (IQR, 14.6–5980.0 ng/mL) in the lenvatinib group and 86.8 ng/mL (IQR, 6.9–3600.0 ng/mL) in the sorafenib group. The proportions of patients with liver function in Child–Pugh classes A, B, and C were 65.9%, 29.5%, and 4.5%, respectively, in the lenvatinib group and 71.6%, 21.6%, and 6.8%, respectively, in the sorafenib group. The percentages of patients with ECOG 0 and 1 were 71.9% and 28.1%, respectively, in the lenvatinib group and 62.3% and 37.7%, respectively, in the sorafenib group. Fifteen (34.1%) and 46 (52.3%) patients received subsequent anti-HCC treatment including systemic treatment and non-systemic treatment. A total of 35 (75%) and 67 (76.1%) patients received antiviral therapy before initiating lenvatinib or sorafenib, respectively.

Most patients were in the immune-inactive phase of HBV infection at the initiation of either lenvatinib or sorafenib, characterized by hepatitis B e antigen (HBeAg) negativity, hepatitis B e antibody (HBeAb) positivity, persistent normal ALT levels, and HBV DNA levels below 2000 IU/mL [5]. Specifically, 35 (79.5%) patients in the lenvatinib group and 63 (71.6%) in the sorafenib group were in the immune-inactive phase. Patients in the HBeAg-negative immune-active phase made up the second highest number of patients, including 6 (13.6%) patients who received lenvatinib and 8 (9.1%) patients who received sorafenib. The HBeAg-negative immune-active phase is characterized by HBV DNA levels ≥ 2000 IU/mL, increased ALT levels, and HBeAg negativity. In sorafenib group, also 8 (9.1%) patients were in HBsAg loss phase.

Previous anti-HCC treatment included TACE, TARE, RFA, and surgical resection. Previous anti-HCC treatment was given to 107 patients (81.1%): lenvatinib to 35 patients (79.5%) and sorafenib to 72 patients (81.8%). Subsequent anti-HCC treatment with lenvatinib or sorafenib is presented in Table 2. The patients treated with lenvatinib had less frequent subsequent anti-HCC treatment than patients treated with sorafenib (34.1% vs 52.3%, p = 0.048). Most frequent systemic treatment for HCC after first-line lenvatinib was sorafenib (75.0%) and after first-line sorafenib was regorafenib (79.1%).

We analyzed the common characteristics of 61 patients who received subsequent treatment after first-line systemic therapy. Forty-eight (78.7%) patients were Child–Pugh class A, 42 (68.9%) patients were ECOG 0 or 1, and 39 (63.9%) patients had a tumor size less than 50% of the total liver volume. Thirty-one patients (50.8%) had a liver function of Child–Pugh class A, performance status of ECOG 0 or 1, and tumor size less than 50% of the total liver volume.

At the time of data cut-off, there were 71 deaths (53.8%): 24 in patients treated with lenvatinib (54.5%) and 47 in patients treated with sorafenib (53.4%). Median follow-up was 6.7 months (IQR, 3.0–11.6 months): 4.7 months (IQR, 3.0–9.8 months) for patients treated with lenvatinib and 8.3 months (IQR, 3.0–12.9 months) for patients treated with sorafenib.

Lenvatinib did not show a survival advantage over sorafenib (HR = 1.46; 95% confidence interval, CI 0.97–2.22, log-rank test p = 0.070) in patients with HBV-related advanced HCC (Fig. 2). The median survival was 7.0 months in patients treated with lenvatinib (95% CI 4.6–9.3 months) and 9.2 months in patients treated with sorafenib (95% CI 6.2–12.2 months).

To identify prognostic factors of OS, univariable and multivariable analyses were performed with the unmatched cohort (Table 3). The univariable analysis showed that PIVKA-II > 250 mAU/mL, AFP > 200 ng/mL, higher total bilirubin levels, lower serum albumin levels, higher ALT levels, higher AST levels, higher PT INR, the presence of ascites, ECOG 1 (vs 0), Child B or C (vs A), BCLC stage C or D (vs B), larger tumour size, presence of portal vein thrombosis, and previous anti-HCC treatment were significantly associated with poor OS (all p < 0.05). From the multivariable analysis for OS, AFP > 200 ng/mL (adjusted HR [aHR] = 1.73, 95% CI 1.23–2.45, p = 0.002), higher total bilirubin (aHR = 1.18, 95% CI 1.09–1.29, p < 0.001), higher ALT levels (aHR = 1.00, 95% CI 1.00–1.01, p = 0.039), ECOG 1 (vs 0) (aHR = 1.98, 95% CI 1.41–2.77, p < 0.001), BCLC stage C or D (vs B) (aHR = 1.77, 95% CI 1.01–3.10, p = 0.045), and larger tumor size (aHR = 1.01, 95% CI 1.00–1.01, p < 0.001) were significant independent prognostic factors associated with the probability of mortality. Based on multivariable analysis, lenvatinib (vs sorafenib) treatment was not an independent risk factor for OS.

Subgroup analysis was performed to investigate characteristics of patients who benefited from lenvatinib or sorafenib treatment. Each variable was stratified as shown in Fig. 5. Sex, BCLC stage, and biliary invasion were excluded in subgroup analysis because the number of patients in the subdivided group was too small to conduct Kaplan–Meier estimate: female (n = 12); BCLC B (n = 12); presence of biliary invasion (n = 8). Sorafenib provided longer survival than lenvatinib in patients> 60 years of age (HR = 1.84, 95% CI 1.01–3.36, p = 0.044), AFP ≤ 200 ng/mL (HR = 2.09, 95% CI 1.16–3.76, p = 0.012), and tumor size ≤ 5 cm (HR = 2.39, 95% CI 1.13–5.04, p = 0.019; Fig. 5).

In contrast, in subgroup analysis for PFS, lenvatinib showed longer median PFS in patients ≤ 60 years of age (HR = 0.49, 95% CI 0.29–0.83, p = 0.006) and AFP > 200 ng/mL (HR = 0.52, 95% CI 0.29–0.93, p = 0.021; Fig. 6).

Lenvatinib showed longer median TTP than sorafenib in patients ≤ 60 years of age (HR = 0.36, 95% CI 0.19–0.67, p = 0.001), without portal vein thrombosis (HR = 0.54, 95% CI 0.30–0.97, p = 0.034), with previous anti-HCC treatment (HR = 0.60, 95% CI 0.37–0.98, p = 0.038), AFP > 200 ng/mL (HR = 0.27, 95% CI 0.12–0.58, p < 0.001), and tumor size > 5 cm (HR = 0.46, 95% CI 0.26–0.83, p = 0.008; Fig. 7).

Additional analysis was performed by dividing the lenvatinib and sorafenib groups into subgroups (with or without subsequent treatment; Additional file 1: Figure S1). Patients who received subsequent treatment after first-line sorafenib (Group 1) had longer OS than patients without subsequent treatment after first-line sorafenib treatment (Group 3) (12.4 vs 5.2 months, p = 0.004), and than those without subsequent treatment after first-line lenvatinib (Group 4) (12.4 vs 4.5 months, p < 0.001). Patients who received subsequent treatment after first-line lenvatinib (Group 2) had longer OS than patients without subsequent treatment after first-line levatinib treatment (Group 4) (13.4 vs 4.5 months, p = 0.001), and than those without subsequent treatment after first-line sorafenib (Group 3) (13.4 vs 5.2 months, p = 0.029). There is no significant difference in OS between groups with subsequent treatment after sorafenib (Group 1) and lenvatinib (Group 2) (12.4 vs 13.4 months, p = 0.882), and between groups without subsequent treatment after sorafenib (Group 3) and lenvatinib (Group 4) (5.2 vs 4.5 months, p = 0.176).

In short, The OS of the subgroups receiving subsequent treatment was longer than those without subsequent treatment regardless of the first-line medication. The sorafenib and lenvatinib subgroups that received subsequent treatment had comparable OS, and both subgroups without subsequent treatment had similar OS.