Programmed cell death protein‐1 (PD‐1)‐targeted immunotherapy in advanced hepatocellular carcinoma: efficacy and safety data from an international multicentre real‐world cohort

Summary Background Programmed cell death protein‐1‐targeted immunotherapy has shown promising results in phase II studies of hepatocellular carcinoma. Aim To evaluate safety and efficacy of nivolumab and pembrolizumab in an international, multicentre, real‐world cohort of patients with advanced hepatocellular carcinoma. Methods Sixty‐five patients treated with nivolumab (n = 34) or pembrolizumab (n = 31) between July 10, 2015 and December 31, 2018 (data cut‐off) across six centres in Austria and Germany were retrospectively analysed. Results Child‐Pugh class A/B/C was 32 (49%)/28 (43%)/5 (8%). Immunotherapy was used as systemic first‐/second‐/third‐/fourth‐line treatment in 9 (14%)/27 (42%)/26 (40%)/3 (5%) patients. Fifty‐four patients had at least one follow‐up imaging and were, therefore, available for radiological response assessment. The overall response and disease control rates were 12% and 49% respectively. Of 52 evaluable patients, four (8%) had hyperprogressive disease. Median time to progression was 5.5 (95% CI, 3.5‐7.4) months, median progression‐free survival was 4.6 (95% CI, 3.0‐6.2) months, and median overall survival was 11.0 (95% CI, 8.2‐13.8) months. Most common adverse events were infections (n = 7), rash (n = 6), pruritus (n = 3), fatigue (n = 3), diarrhoea (n = 3) and hepatitis (n = 3). Efficacy and safety results were comparable between Child‐Pugh A and B patients; however, median overall survival (OS) was shorter in Child‐Pugh B patients (16.7 vs 8.6 months; P = 0.065). There was no difference in terms of efficacy and adverse events between patients who received immunotherapy as first‐/second‐line and third‐/fourth‐line respectively. Conclusions Programmed cell death protein‐1‐targeted immunotherapy with nivolumab or pembrolizumab showed promising efficacy and safety in patients with advanced hepatocellular carcinoma, including subjects with Child‐Pugh stage B and patients with intensive pretreatment.


| INTRODUC TI ON
Hepatocellular carcinoma (HCC) represents the most common primary liver cancer and usually develops in patients suffering from underlying chronic liver disease. [1][2][3][4][5] Despite recommendations for surveillance of patients at risk, HCC is often diagnosed at an advanced stage where only systemic treatment can be offered. Many patients develop recurrence or disease progression after initial surgical or loco-regional treatment and then become candidates for palliative systemic therapy. 1,5,6 For the last decade, the tyrosine kinase inhibitor sorafenib was the only effective drug available for HCC, 6 with two randomised controlled phase III trials showing a survival benefit compared to placebo. 7,8 Only recently, three more tyrosine kinase inhibitors were approved for HCC, lenvatinib in first-line and regorafenib and cabozantinib in second-line drug treatment. [9][10][11] Ramucirumab, a monoclonal antibody against vascular endothelial growth factor receptor (VEGFR)-2, improved survival in a secondline phase III study of patients with advanced HCC and elevated alpha-fetoprotein, 12 and thus will likely be included in the treatment algorithm shortly.
Immunotherapy with checkpoint blockers demonstrated encouraging efficacy in certain cancer types, particularly in melanoma and lung cancer. 13 HCC may also be an attractive candidate for immunotherapy, as it represents an immunogenic tumour, but also fosters an immunosuppressive microenvironment (eg, by up-regulation of immune checkpoint molecules). This may be further supported by the tolerogenic liver milieu and chronic inflammation due to the underlying liver disease. 6,[14][15][16] Notably, overexpression of the checkpoint molecules programmed cell death-ligand 1 (PD-L1) and programmed cell death protein-1 (PD-1) was associated with tumour aggressiveness and postoperative recurrence in HCC. 17,18 Nivolumab and pembrolizumab, two monoclonal antibodies against PD-1, have shown promising efficacy and safety results in noncomparative, open-label phase II studies of advanced HCC, 19,20 and the United States Food and Drug Administration (FDA) already granted accelerated conditional approval to both agents for sorafenib-experienced patients with HCC. Both nivolumab and pembrolizumab are currently being investigated in ongoing phase III trials.
In the present study, we aimed to analyse the safety and efficacy of anti-PD-1 targeted therapy with nivolumab or pembrolizumab in an international, multicentre, real-life cohort of patients with advanced HCC. In contrast to the phase II studies of nivolumab and pembrolizumab, 19,20 our cohort also includes patients with more advanced liver cirrhosis (Child-Pugh B/C) as well as patients who received immunotherapy as third or even fourth line of systemic therapy. Thus, this cohort reflects the treatment reality in advanced HCC outside of clinical trial programs.

| Study design and patients
This was a retrospective study of patients treated with nivolumab or pembrolizumab across six centres in Austria and Germany. Patients with histologically or radiologically confirmed HCC 1 who received PD-1-targeted immunotherapy with nivolumab or pembrolizumab were eligible. All data, including patient history, laboratory results and radiological information were collected retrospectively. The retrospective analysis was approved by local Ethics Committees.

| Dosing of nivolumab and pembrolizumab
Nivolumab was administered at 1-3 mg/kg body weight or at a fixed dose of 240 mg every 2 weeks intravenously. Pembrolizumab was given at 2 mg/kg body weight or at a fixed dose of 200 mg every 3 weeks intravenously. Dose delays were made based on toxicity.

| Assessments
Radiological response was recorded by computed tomography (CT) or magnetic resonance imaging (MRI) at baseline, 6-12 weeks after treatment initiation, and about every 3 months thereafter.
Tumour response was assessed according to the modified Response Evaluation Criteria in Solid Tumours (mRECIST). 21 Patients with progressive disease of target lesions (increase of at least 20%) at the first radiological evaluation were assessed for hyperprogressive disease.
Hyperprogression was defined as a progressive disease (RECIST version 1.1 22 ) on the first radiological evaluation during immunotherapy with a delta tumour growth rate of > 50%, corresponding to an absolute increase in tumour growth rate exceeding 50% per month. 23 diarrhoea (n = 3) and hepatitis (n = 3). Efficacy and safety results were comparable between Child-Pugh A and B patients; however, median overall survival (OS) was shorter in Child-Pugh B patients (16.7 vs 8.6 months; P = 0.065). There was no difference in terms of efficacy and adverse events between patients who received immunotherapy as first-/second-line and third-/fourth-line respectively.

Conclusions:
Programmed cell death protein-1-targeted immunotherapy with nivolumab or pembrolizumab showed promising efficacy and safety in patients with advanced hepatocellular carcinoma, including subjects with Child-Pugh stage B and patients with intensive pretreatment.
Tumour growth rate was calculated as described previously, 23 and delta tumour growth rate (tumour growth rate during immunotherapy minus tumour growth rate before immunotherapy) was then used to assess the association of immunotherapy with tumour growth. Tumour growth rate was only quantified for target lesions. 23 To calculate tumour growth rate before and during immunotherapy, images of CT/MRI scans were required from three different time points: (a) before baseline (while patient was receiving prior therapy/ no therapy), (b) at baseline (before initiation of immunotherapy), (c) at first evaluation during immunotherapy. Hence, hyperprogressive disease could only be calculated in patients of whom radiographic images were available from all three above mentioned time points.
Side effects were recorded at every visit and graded according to the Common Terminology Criteria for Adverse Events (CTCAE) version 4.

| Statistics
Data on baseline characteristics, radiological tumour response and side effects were summarised using descriptive statistics. Chi square test or Fisher's exact test were used to compare nominal data. We avoided statistical comparison between nivolumab-and pembrolizumab-treated patients as this would be unreliable due to the retrospective study design. Median duration of treatment was defined as time from the date of the first administration until the date of last infusion. Patients who were still receiving immunotherapy at data cut-off were censored. Patients who had at least one follow-up imaging assessment were evaluable for radiological response and time to progression. Time to progression (TTP) was defined as the time from the date of first checkpoint inhibitor administration until the date of first radiologically confirmed tumour progression. Data from patients who died without radiologically confirmed tumour progression were censored at the date of last radiological assessment. Progression-free survival (PFS) was defined as the time from the date of first checkpoint inhibitor administration until radiological disease progression or death, whatever came first. Patients who were still alive and without radiologically confirmed progression were censored at the date of last contact or data cut-off. Overall survival (OS) was defined as the time from start of immunotherapy until the date of death. Patients who were still alive were censored at the date of last contact or data cut-off. Survival curves were calculated using the Kaplan-Meier method and compared by means of the log rank test. Statistical analyses were performed using IBM SPSS Statistics version 25 (SPSS Inc., Chicago, IL). A P < 0.05 was considered significant.

| Patients
Sixty-five patients in whom PD-1 targeted immunotherapy was initiated between July 10, 2015 and April 27, 2018 were included.

| Efficacy
Fifty-four patients had at least one follow-up imaging and were therefore available for radiological tumour response assessment (nivolumab, n = 30; pembrolizumab, n = 24). Of the 11 subjects not available for response assessment, 9 patients died before the first radiological evaluation and 2 patients were lost to follow-up. In the nivolumab group, no patient had complete response (CR) and 5 (15%) participants achieved partial response (PR), resulting in an overall response rate (ORR) of 15%. Ten (29%) patients showed stable disease (SD) and 15 (44%) subjects had progressive disease at first radiological evaluation. The disease control rate (DCR) was 44%. In the pembrolizumab-treated patients, 0 and 3 (10%) participants achieved complete response and partial response, respectively. Fourteen (45%) patients had stable disease and 7 (23%) individuals showed progressive disease. The overall response rate and disease control rate were 10% and 55% respectively.
Median OS for patients with partial response or stable disease was 16.2 (95% CI, 9.1-23.3) months and was significantly longer compared to that of patients with progressive disease, which was 7.4 (95% CI, 6.3-8.5; P = 0.039) months ( Figure 8).

| Efficacy and safety according to Child-Pugh stage
As the number of patients with Child-Pugh stage C was too low (n = 5) for meaningful analysis, we excluded these patients and only  Table S3.

| Efficacy and safety according to systemic line of immunotherapy
As the number of patients who received immunotherapy as first (n = 9) or fourth (n = 3) line of systemic treatment was low, we grouped patients who received immunotherapy as first-or secondline (group 1) and those in whom immunotherapy was used as thirdor fourth-line of systemic treatment (group 2).
In terms of safety, the number of patients who developed any grade (Group 1 vs 2, n = 12 (33%) vs n = 13 (45%); P = 0.344) or highgrade (Group 1 vs 2, n = 6 (17%) vs n = 5 (17%); P = 1.000) adverse events was similar between group 1 and 2.  Monoclonal antibodies are not metabolized by the liver but eliminated predominantly via uptake and catabolism by the reticuloendothelial system and target tissue. 26 This could make the pharmacokinetic profile of immune checkpoint inhibitors more predictable even in patients with advanced liver cirrhosis. 15 However,   inter alia by an upregulation of immune checkpoint molecules. 31,32 Indeed, sorafenib intensified tumour hypoxia and increased tumoural PD-L1 expression in experimental models of HCC. 33,34 Thus, immunotherapy may be particularly attractive following or combined with anti-vascular endothelial growth factor (VEGF)-targeted therapies.

| D ISCUSS I ON
In line, preliminary data of pilot studies testing the combination of lenvatinib plus pembrolizumab (n = 26) and bevacizumab combined with atezolizumab (n = 68) showed encouraging response rates of 42% and 34% respectively. 35,36 Hyperprogressive disease-an increased tumour growth rate during treatment-is a new pattern of progression that was recently reported for patients treated with PD-1-/PD-L1-targeted immunotherapy. 23,37,38 Four (8%) patients in our cohort had hyperprogression during immunotherapy. This is in line with a previous study that reported hyperprogressive disease in 9% of patients with advanced cancers, 37 but lower compared to recurrent and/or metastatic head and neck cancer (29%) 38 and advanced non-small-cell lung cancer (13.8%) treated with PD-1/PD-L1 blockers. 23 Notably, hyperprogressive disease was defined differently in these studies as there is currently no consensus on the optimal definition. 23