1. Introduction
Systemic treatment is the only available option for hepatocellular carcinoma (HCC) patients with preserved liver function (Child–Pugh class A) presenting with advanced or intermediate HCC stages that are no longer suitable for locoregional procedures [
1].
Since 2007, for such patients, the multikinase inhibitor (MKI) sorafenib has been worldwide considered the standard of care [
2]. After a decade characterized by unsatisfactory results, novel targeted agents with prevalent antiangiogenic profiles have been approved in the context of first-line and second-line settings upon sorafenib discontinuation [
3,
4,
5,
6]. In addition, the advent of immune checkpoint inhibitors (ICIs) either as single agent or in combination has dramatically improved the therapeutic scenario, potentially resulting in a multitude of sequential treatment opportunities [
7,
8,
9,
10,
11,
12,
13,
14,
15,
16,
17].
Today, the increased options for multiple lines of systemic therapies may improve the survival of patients affected by advanced HCC. However, on the other side, reliable factors that can predict treatment benefit as well as post-treatment survival remain unknown and represent an unmet clinical need. Even though new therapeutic strategies became available, sorafenib still is a valid first-line treatment option, especially for patients with absolute contraindications to ICIs [
18].
Therefore, we aimed to investigate prognostic and predictive factors associated with outcomes in a cohort of sorafenib pre-treated patients who were fit for second-line treatments that were delivered in the context of clinical trials. In addition, as the therapeutic landscape of HCC is heading beyond the second line, we also investigated prognostic factors affecting survival after second-line treatment discontinuation.
2. Materials and Methods
2.1. Patient Selection
From January 2012 to June 2018, all patients included in this cohort study received sorafenib as first-line treatment and a subsequent systemic agent in the context of clinical trials at three academic centers (IRCCS Humanitas Research Hospital, Rozzano (Milan); Istituto Oncologico Veneto-IRCCS, Padova; Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori-IRCCS, Meldola) in Italy. They could have experienced disease progression under sorafenib, or they were deemed sorafenib-intolerant. Patients were further stratified according to the subsequent regulatory approval (if granted or not) of the investigational agent they had received during the trial. As a general rule, unless pre-specified, all study protocols for treatment of advanced HCC exclude patients with an Eastern Cooperative Oncology Group Performance Status (ECOG PS) ≥2 and a Child Pugh class B or C. Treatments could be discontinued due to (1) tumor progression; (2) unacceptable treatment toxicity, i.e., grade 2–4 AEs not responding to dose reductions and/or temporary interruption of treatment as per study protocol; (3) liver decompensation defined by the following clinical parameters: jaundice, ascites, gastrointestinal hemorrhage, or encephalopathy; (4) ECOG PS worsening, which was considered cancer progression. AEs were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events, version 4.03. This study was conducted in agreement with Good Clinical Practice guidelines, the ethical principles of the Declaration of Helsinki and local regulations. The protocol and its annexes were subject to review and approval by local Institutional Review Boards at each participating institution (ONC/OSS-01/2018).
2.2. Outcome Measures
Patients’ data were retrospectively analyzed for baseline characteristics; second-line treatments received; and subsequent outcomes in terms of radiological response, OS, time-to-treatment failure (TTF), and post-treatment survival (PTS).
Response was assessed according to RECIST v1.1. OS was measured as the date of enrollment on a second-line trial until death from any cause or date of last follow-up. TTF was measured from the first day of treatment on clinical trial until the patient came off study (for toxicity, disease progression, or death). The decision to discontinue study protocol was made by the treating physician based on patient’s history, AEs, and imaging studies. PTS was the time elapsed between the last day of treatment and death or last follow-up.
We assessed potential predictors of survival including age, sex, Barcelona Clinic Liver Cancer (BCLC) stage, ECOG PS, previous surgery, sorafenib duration, reason for sorafenib discontinuation, time to start of second-line treatment, pattern of disease progression during first-line sorafenib, extra-hepatic spread (EHS), neutrophil-to-lymphocyte ratio (NLR), portal vein thrombosis (PVT), alpha-fetoprotein (AFP) levels, and liver function analyses. The albumin-bilirubin (ALBI) grade was calculated as previously reported [
19].
2.3. Statistical Analysis
Demographical and clinical characteristics were summarized as number and percentages or as median and interquartile range (IQ range). The number of patients at the three participating institutions during the study period determined the sample size. Differences in distribution were estimated using the chi-square or the Fisher exact test (when appropriate) and the Wilcoxon’s t-test. Survival curves were generated using the Kaplan–Meier method. Differences between groups were evaluated using the log-rank test. A Cox proportional hazards regression model was used to calculate the hazard ratios (HR) and their 95% confidence intervals (95% CI). A multivariable model was built considering factors statistically significant in the univariate model which confirmed their effect. The relationship between baseline clinical characteristics and second-line treatment effect was evaluated using a Cox proportional hazards model with an interaction term. Statistical significance was set at p < 0.050. All reported p values were two-sided. All analyses were carried out with the SAS software v9.4.
4. Discussion
In this retrospective study, we investigated outcomes and prognostic factors for a cohort of patients with preserved liver function and an ECOG PS ≤ 1 being enrolled in second-line trials at three academic Italian Institutions after first-line sorafenib. Whereas according to published data nearly 25% of patients discontinue sorafenib due to liver function deterioration which precludes additional treatments, we sought to identify prognostic factors in the remainder who are eligible for clinical trial enrolment [
20,
21]. For OS, multivariate analyses indicate that treatment with OT and presence of PVT are negative prognostic factors. In addition, presence of EHS at the beginning of second-line treatments is both prognostic of shorter OS and predictive of increased AT benefit over OT.
The survival outcomes are overall consistent with those reported across pivotal trials investigating targeted agents after first-line sorafenib [
4,
5,
6] and those from a recent meta-analysis on the role of regorafenib as a valuable second-line treatment option after sorafenib [
22]. Similarly, the survival figures with combinations of anti-PD1 antibodies plus targeted agents or anti-PD1 plus anti-CTLA4 are in line with the results of the most recent studies exploring different ICIs combinations [
14,
15,
16,
17]. Notably, in our cohort, the use of anti-PD1 antibodies + targeted agents +/– anti-CTLA4 antibodies resulted in numerically longer OS compared with approved agents (14.7 months vs. 10.9 months), even though this comparison, which was performed regardless of the etiology, relied on low numbers. Moreover, following the publication by Pfister et al suggesting a reduced efficacy of ICIs in patients with nonalcoholic steatohepatitis (NASH), we performed an exploratory survival analysis in a subset of patients with known nonviral etiology [
23]. However, no significant survival difference was observed comparing ICIs-based treatment and targeted agents. Excepting the RESORCE trial, which enrolled only sorafenib-tolerant patients, the other second-line trials herein considered at each participating center had broadly overlapping inclusion criteria [
4]. As such, all patients were assumed to represent a relatively homogeneous cohort, bearing preserved liver function (Child–Pugh A), ECOG PS ≤ 1, in addition to clinical and prognostic factors that allowed for subsequent trial enrolment. In contrast, patients who were given only frontline sorafenib were excluded from this analysis, as they represent a markedly different population bearing poor prognosis, driven by a deteriorating liver function [
24,
25]. The Child–Pugh status remains indeed a major prognostic determinant and it is a critical inclusion criterion for most clinical studies in HCC.
During the past five years, the approval of numerous systemic treatment options has widened and changed the therapeutic scenario of advanced HCC. Even though the combination of atezolizumab plus bevacizumab is now considered the new first-line standard of care, sorafenib still represents a mainstay of advanced HCC treatment, since all second and further-line options have been evaluated in patients who were refractory or intolerant to sorafenib. Moreover, the optimal treatment sequencing after failure of the first line still represents a major topic, even considering the lack of determinants of treatment benefit and prognosis.
We observed in univariate analysis a negative prognostic impact dictated by higher NLR. Interestingly, we also detected an increased benefit from AT over OT in patients with low NLR. These findings are consistent with prior analyses in both first- and second-line contexts [
26,
27,
28]. In addition, in the immunotherapy era, a higher NLR has been also associated with worse survival in HCC patients treated with nivolumab monotherapy [
29]. Similarly, lymphocyte-to-monocyte ratio (LMR) has been evaluated as inflammation-based biomarker. According to a meta-analysis, a higher LMR was associated with increased OS and disease-free survival (DFS) after liver resection [
30]. Furthermore, both NLR and LMR were suggested to be independent prognostic factors for DFS after hepatectomy [
31].
Additionally, we found a significant interaction between treatment and disease extent, that translated into more pronounced OS benefits from AT in patients with EHS. These results are in keeping with subgroup analyses of the RESORCE and CELESTIAL trials reporting greater OS benefits from regorafenib and cabozantinib in patients with EHS compared with patients with intrahepatic disease [
4,
5]. In contrast, the magnitude of benefit deriving from sorafenib over placebo was previously reported to be greater in patients without evidence of EHS [
27]. In all, these findings may be helpful to gauge the benefit of approved second-line treatments in specific subgroups of patients.
Previous investigations showed that duration of first-line sorafenib does not affect time to progression nor OS after regorafenib [
32,
33]. However, available data are not consistent, as retrospective analyses from CELESTIAL suggest that OS in patients receiving cabozantinib or placebo could be related to prior sorafenib exposure [
34].
In our experience, patients treated with sorafenib followed by AT had a median OS of 22.8 months (95% CI 18.4–28.1) from start of sorafenib, while the OS achieved sequencing sorafenib and OT was 21.9 months (95% CI 17.5–24.6). These findings suggest a better prognosis for patients potentially eligible for clinical trials [
25]. Consistent with post hoc analyses of RESORCE and CELESTIAL, in patients receiving AT we estimated higher survival rates than those of patients on OT [
33,
34]. In particular, these were markedly diverging at 36 months from start of sorafenib, with a survival probability of 28% for patients treated with AT and 17% for patients treated with OT. This is in line with the hypothesis that survival outcomes are mostly driven by second-line treatments that are sequenced after sorafenib [
24]. In this clinical scenario, OT represents a heterogenous group of treatments, including a placebo arm. However, results from previous negative trials showed similar outcomes for experimental arms and placebo arms [
35,
36,
37,
38]. Although exploratory in nature, the significant interactions we detected indicate that specific subgroups of patients, namely those with EHS and lower NLR, can derive greater survival benefit from AT over OT and deserve to be further investigated.
Most patients fail second-line treatments because of disease progression and AEs. Of note, across different second-line treatments considered, we observed similar rates of liver decompensation and PS worsening leading to treatment discontinuations. Regardless of the type of second-line agent, the overall PTS of 4 months we detected after second-line therapy align with the rates reported by Iavarone et al and by Shao et al. after first-line treatments [
21,
25]. To the best of our knowledge, the present study is also the first to investigate course and predictors of survival in patients who discontinue second-line AT or OT. To this end, we considered a previous model suggesting a strong prognostic impact linked to the reason of sorafenib discontinuation [
21]. In this series, we observed a significantly longer OS in those patients who fail second-line treatment because of AEs or tumor progression, as compared to those who experience liver decompensation or PS worsening. Irrespective of treatment line, these latter patients still represent a hard-to-treat population better managed by best supportive care approaches. Eventually, low AFP levels, ALBI grade 1, and enrolment onto third-line trials (which recapitulates well-preserved liver function and absence of constitutional symptoms) emerge as independent prognostic factors for longer PTS. Despite the relatively small number of patients who were enrolled onto further studies, these findings clearly speak to the potential of third-line systemic treatments in selected patients.
This study has several limitations related to its retrospective nature. Firstly, our study is based on data collected before the approval of ICI-based systemic treatment for advanced HCC, thus limiting the applicability of our results. Secondly, beyond the physicians’ expertise, some degree of heterogeneity in terms of disease management is possible across the tertiary referral centers involved. Thirdly, the pattern of disease progression after second-line treatment was not captured. Of note, the appearance of new extrahepatic lesions was reported to be linked to post-progression survival in post hoc analysis of REACH and REACH-2 studies [
39]. Lastly, while we demonstrated that OS with placebo roughly aligns OS with agents not granted subsequent approval for HCC, among the latter, some could have had clinical activity and cannot be assumed to replace a placebo control.