Patient characteristics

A total of 155 patients across 11 sites were included. Patients were predominately male (90.7%) and Caucasian (60.6%) with a median age of 65 years (IQR: 59–75). Most patients had compensated Child–Pugh A (CP A) cirrhosis (78.8%). The main causes of chronic liver disease were hepatitis C infection (HCV) (40.0%), alcohol-related liver disease (34.2%), non-alcoholic fatty liver disease (NAFLD) (25.8%) and hepatitis B infection (HBV) (19.8%). Baseline patient characteristics are summarised in Table 1.

Table 1 Patient characteristics at time of commencement of lenvatinibTumor characteristics

Almost all patients had BCLC stage C (69.7%) or BCLC stage B (27.7%) disease. Tumor thrombus was present in forty-four patients (28.4%). A total of 98 patients (63.2%) had received prior treatment for HCC and 54 patients (34.8%) were treatment naïve. Of those with treatment experience, 61 patients (60.4%) had previously received one treatment modality, 30 patients (29.7%) had received two different treatment modalities and seven patients (6.9%) had received three different treatment modalities. The majority of patients (80.6%) treated with lenvatinib had multiple liver lesions. Patients with only a single liver lesion (n = 30, 19.4%) had extra-hepatic metastatic disease (n = 15), portal vein invasion (n = 8), previous treatment experience leading to treatment stage migration (n = 4) or a very large lesion not suitable for locoregional therapy (n = 3). Tumor characteristics are summarised in Table 2.

Table 2 Tumor characteristics at time of commencement of lenvatinibLenvatinib dosing and duration of use

Patients were prescribed a starting dose of either 12 mg (41.3%), 8 mg (31.6%) or 4 mg (24.5%) daily. The maximum appropriate weight-based daily dose was reached in 65.2% of patients. A temporary interruption to therapy occurred in 36 patients (23.2%) due to intolerance. A total of 67 patients (43.2%) required dose reduction due to development of AEs. The median duration of therapy overall was 5.0 months (IQR 2.3–8.4 months). At time of analysis, therapy had been permanently ceased in 115 patients (74.2%) while 40 remained on therapy (25.8%). A total of 58 patients stopped therapy due to intolerance (50.4%), 44 stopped due to disease progression (38.3%) and 10 patients died while on therapy (8.7%).

Overall survival

Patients were followed for a median period of 9.4 months (IQR 5.8–14.4) and during this period there were 83 deaths (53.5%). The median OS was 7.7 months (IQR 4.8–11.6 months) and the median PFS was 5.3 months (IQR 2.8–9.2 months).

Predictors of overall survival

Kaplan–Meier survival analysis (Fig. 1) revealed the development of new or worsening hypertension was associated with improved OS compared to those who did not develop hypertension (median OS 16.2 vs 9.4 months (p = 0.02)). Patients with treatment emergent diarrhoea also had an improved median OS compared to those without an altered bowel habit (17.5 vs 10.1 months (p = 0.08)). Additionally, a dose reduction due to the development of AEs was associated with improved survival compared to those who maintained stable dosing (19.6 vs 7.8 months respectively (p < 0.01)). Conversely decompensated liver disease was associated with worse OS when CP B/C patients were compared with CP A patients (median OS 5.6 vs 12.5 months (p < 0.01)).

Fig. 1

Kaplan–Meier curves for overall survival stratified by Child Pugh status (a) development of new or worsening hypertension (b) development of diarrhoea (c) and the need for a dose reduction due to AE (d)

Predictors of worse OS on univariate analysis were baseline CP score (HR 3.00, 95% CI 1.34–2.01, p < 0.01), MELD score (HR 1.10, 95% CI 1.03–1.17, p < 0.01), BCLC stage (HR 1.70, 95% CI 1.01–2.86, p = 0.04), ECOG score (HR 1.58, 95% CI 1.06–2.36, p = 0.02), presence of ascites at baseline (HR 1.86, 95% CI 1.01–3.43, p = 0.04), increased bilirubin (HR 1.02, 95% CI 1.01–1.04, p < 0.01) or presence of tumor thrombus (HR 1.89, 95% CI 1.19–3.01, p < 0.01). Predictors of improved OS were lenvatinib dose reduction (HR 0.38, 95% CI 0.23–0.63, p < 0.01) or withholding therapy (HR 0.44, 95% CI 0.24–0.82, p < 0.01), a higher baseline albumin (HR 0.89, 95% CI 0.85–0.93, p < 0.01) and the development of hypertension (HR 0.56, 95% CI 0.34–0.94, p = 0.02). There was a trend towards improved OS with the development of diarrhoea (HR 0.61, 95% CI 0.34–1.08, p = 0.08). There was no significant co-linearity between factors. Racial background, etiology of liver disease, previous exposure to any HCC therapy including sorafenib or reaching maximum dose of therapy did not impact overall survival.

On multivariate analysis, BCLC stage (HR 2.50 95% CI 1.40–4.45, p < 0.01), baseline albumin (HR 0.89, 95% CI 0.86–0.93, p < 0.01), the development of hypertension (HR 0.42, 95% CI 0.24–0.73, p < 0.01) or diarrhoea (HR 0.47, 95% CI 0.25–0.88, p = 0.01) and dose reduction (HR 0.41, 95% CI 0.24–0.69, p < 0.01) remained independent predictors for OS (Table 3).

Table 3 Univariate and multivariate analysis of factors predictive of mortalityPredictors of progression free survival

Kaplan–Meier survival analysis (Supplementary file 1) revealed either diarrhoea or hypertension as an AE was associated with improved PFS. Patients with diarrhoea had a median PFS of 6.2 months versus 5.6 in those without diarrhoea (p = 0.04). In patients who developed new or worsening hypertension the median PFS was 8.2 months versus a median PFS of 5.5 months in patients who did not develop hypertension (p = 0.01). Additionally, the requirement for a dose reduction was also associated with improved PFS (8.2 vs 4.1 months respectively (p < 0.01)).

Regression analysis

Predictors of improved PFS on univariate analysis were dose reduction (HR 0.49, 95% CI 0.33–0.73, p < 0.01), withholding treatment (HR 0.56, 95% CI 0.34–0.94, p = 0.02), older age (HR 0.98, 95% CI 0.96–0.99, p = 0.02), increased baseline albumin level (HR 0.95, 95% CI 0.92–0.98, p < 0.01), and AEs of either diarrhoea (HR 0.60, 95% 0.40–0.91, p = 0.01) or hypertension (HR 0.83, 95% 0.70–0.99, p = 0.04). A higher baseline CP score was associated with a shorter PFS (HR 1.27, 95% CI 1.02–1.58, p = 0.03).

On multivariate analysis, dose reduction (HR 0.45, 95% CI 0.29–0.68, p < 0.01), older age (HR 0.96, 95% CI 0.94–0.98, p < 0.01) and a higher baseline CP score (HR 1.24, 95% CI 1.01–1.52, p = 0.04) remained as independent and significant predictors of PFS (Table 4).

Table 4 Univariate and multivariate analysis of factors predictive of disease progressionAdverse events

During treatment, 130 patients experienced an AE of any grade (83.9%), of which 48 (28.4%) were grade three or four. Fifty-eight patients required permanent cessation of therapy due to an AE (37.4%). The most common AE was new or worsening hypertension experienced by 55 patients (35.5%), with 43 patients (27.7%) requiring anti-hypertensive medication. The percentage of patients with pre-existing hypertension before commencing lenvatinib was 49%. Sixteen patients (10.3%) required thyroid replacement therapy due to hypothyroidism. The remaining AE, their frequencies and grades are summarised in Table 5. Three patients (1.9%) experienced a grade four AE and these were malignant hypertension, duodenal perforation and HCC necrosis with heart failure. There was one sudden unexplained death while on therapy (day 17 of therapy).

Table 5 Frequency and grade of adverse events post commencement of Lenvatinib