Hepatocellular carcinoma (HCC) is a lethal disease in most patients, due to its aggressive course and a lack of effective systemic therapies for advanced disease. Surgical resection and liver transplantation remain the only curative options for a small subset of patients. Few patients with HCC are diagnosed early enough to be eligible for curative treatment. Angiogenesis inhibition is a natural therapeutic target for all solid tumors, but particularly for the highly vascularized HCC tumors. With the approval of the targeted agent sorafenib, there are now additional options for patients with HCC. Although sorafenib does produce some improvement in survival in HCC patients, the responses are not durable. In addition, there are significant dermatologic, gastrointestinal, and metabolic toxicities, and, as importantly, there is still limited knowledge of its usefulness in special subpopulations with HCC. Other angiogenesis inhibitors are in development to treat HCC both in the first-line setting and for use following sorafenib failure; the furthest in development is brivanib, a dual fibroblast growth factor pathway and vascular endothelial growth factor receptor inhibitor. Additional agents with antiangiogenic properties also in phase II and III development for the treatment of patients with HCC include bevacizumab, ramucirumab, ABT-869, everolimus and ARQ 197.

Primary liver cancer, including hepatoma and hepatocellular carcinoma (HCC) is diagnosed in more than 560 000 people worldwide each year[1], including more than 24 000 Americans[2]. HCC accounts for up to 90% of all primary liver cancers[3]. HCC can be treated curatively with surgical resection or liver transplantation if diagnosed at an early stage; however, since most HCC patients present with advanced disease, only 15% are eligible for curative treatments[4]. Even for patients undergoing surgical resection, recurrence rates may be as high as 50% after 2 years and 76% by 10 years[5,6]. Patients meeting the Milan criteria who undergo liver transplantation can achieve a 5-year cancer-free survival rate greater than 60%[7].

As most patients with HCC are diagnosed with advanced disease, they generally have a poor prognosis, with median survival times of less than 1 year[3]. This is due, at least in part, to the absence of effective systemic therapies. Systemic therapies examined in the past, including both cytotoxic and hormonal agents, have provided limited or no benefit for these patients[6]. In late 2007, the angiogenesis inhibitor sorafenib was approved for use in advanced HCC based on an improvement in survival compared with placebo[8,9]. While initial responses are observed, over time a loss of efficacy is apparent that may be due to “resistance”via escape/compensatory mechanisms. Like other angiogenesis inhibitors, sorafenib also has known class side effects, including skin-related toxicities, hypertension, proteinuria, diarrhea, and an increased risk for thromboembolism and bleeding events[10-12]. While most are manageable, certain rare events can be life-threatening (i.e., gastrointestinal perforation, fatal hemoptysis, thromboembolic events). Thus, the balance between risk and benefit in every clinical setting is an integral part of the differentiation and evaluation of targeted agents.

Angiogenesis is a ubiquitous process that is required for tumor growth[13,14]. Angiogenic processes are also indirectly involved in tumor invasion and metastasis through the secretion of matrix-degrading proteinases by vascular endothelial cells[15] and the ability of tumor cells to travel to distant sites via the vascular network[16].

Proangiogenic factors are attractive therapeutic targets because they stimulate cancer formation, growth, and proliferation via angiogenesis using a number of distinct mechanisms. Established proangiogenic factors and their receptor signaling pathways include vascular endothelial growth factor (VEGF), fibroblast growth factor (FGF)-2, platelet-derived growth factor (PDGF), angiogenin, and angiopoietin-2 (Ang-2)[17]. Other mediators, such as c-MET and mTOR, although not directly related to new vessel formation, affect angiogenesis via influences on downstream signaling.

HCC tumors are generally hypervascularized[18], suggesting that they may be especially vulnerable to angiogenesis inhibition. Several endogenous proangiogenic factors are expressed in HCC[19-22], and evidence indicates they play a role in HCC pathogenesis. For instance, serum VEGF levels increase with advancing HCC stages, being highest in patients with metastatic disease[23]. Elevated VEGF levels after locoregional therapy also are associated with poor prognosis and diminished response to therapy[24,25]. Expression of the proangiogenic factor FGF-2, the target of newer agents, is also elevated in patients with HCC[22] and its expression in HCC correlates with tumor microvessel density[26] and postoperative recurrence rate[27]. Tumor angiogenin expression correlates with microvascular density in patients with HCC, and high serum angiogenin levels are associated with decreased survival at 5 years[28]. Finally, mRNA angiopoietin expression level (via Ang-2/Ang-1 ratio) is positively correlated with tumor portal vein invasion, diameter, microvascular density, and poor prognosis[29]. Taken together, this evidence provides a strong rationale for targeting angiogenesis and related proangiogenic factors to provide more effective therapies for the treatment of HCC.

Sorafenib was the first systemic targeted therapy to be approved by the US Food and Drug Administration for patients with unresectable HCC[30], based on a 2.8-mo survival advantage over best supportive care (BSC) [hazard ratio (HR) 0.69, P = 0.00058] in the Sorafenib Hepatocellular carcinoma Assessment Randomized Protocol (SHARP) trial[31]. Sorafenib, which is also approved for advanced renal cell carcinoma (RCC)[30] inhibits the following receptor tyrosine kinases: VEGFR-2, VEGFR-3, PDGFR-β, c-KIT, and Flt-3[32]. It also binds to the serine-threonine kinases Raf, MEK, and ERK[32,33]. The VEGFR and PDGFR pathways and Raf-1 have all been implicated in the pathogenesis of HCC[34,35], providing a rationale for sorafenib activity in HCC.

Although sorafenib represents a much needed treatment option for patients with HCC, it also produces toxicities that may significantly affect patients’ quality of life. High rates of dermatologic side effects are commonly reported with sorafenib, the most clinically significant being hand-foot skin reaction (HFSR)[36]. HFSR typically develops in the first few weeks of therapy, with painful hyperkeratotic lesions on the palms and soles that are surrounded by a ring of erythema localized on areas of pressure or flexure[37]. A meta-analysis examining the incidence of HFSR in phase II and III trials across solid tumors reported an incidence of 33.8% with sorafenib use, 8.9% of which were grade 3[38]. In the SHARP trial, sorafenib-associated HFSR had an overall incidence of 21% and a grade 3 incidence of 8%[9]. Of particular concern, HSFR appears to be more prevalent in Asian patients, the population most affected worldwide by HCC. A phase III trial conducted solely in Asian patients with HCC reported a doubling of the overall incidence of HFSR compared with the SHARP trial[8,9]. Similarly, a phase III trial of sorafenib monotherapy following transarterial chemoembolization in Asian patients reported an HFSR incidence of 82%[39].

Bleeding events, albeit rare, are also a toxicity of sorafenib and other angiogenesis inhibitors. In a meta-analysis of 2109 patients enrolled in sorafenib clinical trials or expanded-access programs, the relative risk of bleeding events was 1.86 (P < 0.001) and the incidence of grade 3 or higher bleeding events was 2.2%[40]. It is notable that patients with HCC have not demonstrated an increased incidence of bleeding events with sorafenib; for example, in the SHARP trial, grade 3/4 bleeding events were 1% in the sorafenib arm.

By design, SHARP was conducted in the Americas, Europe, and Australasia, and thus generated limited data in Asian patients. Patients included in SHARP had preserved liver function and were of mostly Child-Pugh (C-P) A status[9]. In order to gain additional data, a randomized phase III trial in Asian patients with advanced HCC was completed, but again C-P B and C patients were excluded[8,39]. The overall trend of response to sorafenib from this Asian population was similar to those of SHARP, with time to progression (TTP) and overall survival (OS) improved compared with placebo, and similar treatment effects (HRs of 0.68-0.69). However, median OS appeared to be shorter (6.2 mo) vs SHARP (10.7 mo), most likely due to more advanced disease based on performance status, number of tumor sites, and presence of lung metastases compared with SHARP[8,9]. A subsequent single-arm, phase II trial was conducted in Asian patients with worse prognosis, including C-P B and C status (29%) and portal vein thrombosis (43%)[41]. In this trial, 26% of patients derived clinical benefit from sorafenib. Again, the incidence of toxicities was higher than in SHARP, with grade 3/4 diarrhea occurring in 20% and grade 3/4 HFSR in 16% of patients. As the majority of HCC cases are found in Asia[42], the development of safe and more effective therapies for this population represents a significant unmet need.

Hence, while the sorafenib trials have provided valuable information in patients with preserved liver function[8,9,41], determining efficacy and safety in the substantial portion of patients with advanced HCC remains a challenge. Other recent phase II trials are beginning to provide preliminary evidence of sorafenib safety and efficacy in patients with more advanced disease. A trial of sorafenib monotherapy in 59 patients with unresectable HCC, including 39% with C-P B status and 17% of C-P C status, showed promising activity regardless of disease stage and liver function[43]. Responses for patients of C-P B status were similar to those from SHARP. Median OS, however, declined with more advanced C-P status, most likely due to underlying cirrhosis, and because drug toxicities are more prevalent with compromised liver function, causing liver-related or systemic complications that lead to early treatment discontinuations. In a second phase II trial, pharmacokinetic profiles of sorafenib were similar in both C-P A and B subgroups, while median TTP and OS appeared shorter and adverse events related to poorer liver function were more frequent in C-P B patients[44]. Finally, another single-arm, phase II trial in 51 Asian patients, including 15 (29%) with C-P B/C status, found no significant differences between C-P B/C status patients and C-P A patients in disease control rate, median OS, grade 3/4 hematologic toxicities, or grade 3/4 nonhematologic toxicities[41].

Despite initial responses to sorafenib, and similar to other targeted agents, most HCC patients experience a loss of efficacy. Furthermore, across clinical trials, 20%-38% of patients discontinued sorafenib due to adverse events[8,9,41]. Similar to what has been reported with bevacizumab, some data indicate that patients who discontinue sorafenib therapy may experience “rebound,” whereby symptom and tumor progression develops rapidly upon discontinuation[45]. While this accelerated growth effect appears to be temporarily curtailed with re-initiation of therapy, insensitivity to treatment returns quickly. No effective second-line treatment options currently exist outside of clinical trials for patients who are resistant or refractory to and/or intolerant of sorafenib.

Next to sorafenib, sunitinib is the most studied multitargeted tyrosine kinase inhibitor. Like sorafenib, sunitinib is an inhibitor of VEGFR and PDGFR and is currently indicated for the treatment of RCC, as well as for gastrointestinal stromal tumors[46]. While early indications were that sunitinib would have efficacy in HCC, the phase III SUN 1170 trial comparing sunitinib with sorafenib in patients with advanced HCC was discontinued due to increased serious sunitinib-related adverse events and the improbability of achieving noninferior efficacy[47]. As a result, sunitinib is no longer in development for the treatment of HCC.

The 2 traditional imaging criteria widely used for measuring tumor responses to treatment are Response Evaluation Criteria In Solid Tumors (RECIST)[48], which are used primarily in the United States, and the World Health Organization (WHO) criteria[49], which are used internationally. With the advent of targeted therapies that are cytostatic, and in particular regarding HCC tumors, these traditional criteria are limited in their usefulness to assess treatment response because (1) cirrhotic livers may not remodel around a necrotic tumor; (2) HCC tumors are occasionally diffuse and infiltrative in cirrhotic livers; (3) arterial phase enhancement of premalignant dysplastic nodules can be mistaken for progression; and (4) cytostatic targeted agents alter tumor vascularity without affecting tumor size[50]. Both criteria have been updated in the past decade to account for HCC tumor viability in order to appropriately evaluate the extent of tumor necrosis and/or viability to quantify treatment response. The European Association for the Study of the Liver updated the WHO criteria in 2000[51], and in 2008 the American Association for the Study of Liver Diseases updated RECIST[52] and then further clarified the modified criteria in 2010[53]. Specific differences between these current criteria-modified WHO (mWHO) and modified RECIST (mRECIST)-for treatment response in HCC are listed in Table 1.

As an example of the variability in response assessment that can arise according to which of the 2 criteria is used, Finn et al[54] retrospectively compared response assessment by the necrosis-adjusted mRECIST (mRECIST has not yet been validated) for HCC with the prospective use of mWHO criteria in 101 patients with HCC who received brivanib, an antiangiogenic targeted agent. They found that mWHO criteria underreported treatment benefit compared with mRECIST for HCC, with 31 patients classified as having progressive disease under the mWHO criteria but stable disease or partial response under mRECIST. This suggests that these patients may have been prematurely discontinued from treatment while still deriving benefit. These discrepancies can be explained by differences in the methodologies of mRECIST and mWHO, which employ different measurements and calculations of tumor size, thereby producing distinct interpretations of treatment response. In addition, there are differences in how progression is defined, with cytologic confirmation of ascites required with mRECIST and small new lesions (< 1 cm) unlikely representing progression with the mRECIST. While not conclusive, these results suggest that treatment response based on assessment criteria vary widely and indicate the need for further clinical validation.

Clinical trials of antiangiogenic agents have shown that most patients with advanced tumors eventually experience progression, including those who initially respond to treatment[8,9,55-57]. Recent evidence suggests that relapse during treatment with antiangiogenic agents occurs due to VEGF inhibition-driven hypoxia, which induces upregulation of alternate proangiogenic signals such as FGF, which overrides the VEGFR inhibition[58]. This was demonstrated preclinically in murine tumors that initially responded to treatment with an anti-VEGFR-2 antibody, but relapsed after 2 wk, showing higher levels of other proangiogenic signals, including FGF-1 and FGF-2, than untreated tumors[59]. Moreover, blockage of FGF signaling in this model slowed tumor growth and attenuated its revascularization during the relapse phase. Clinically, this has been observed in patients with HCC[26,27,60] and, more recently, in glioblastoma patients treated with the pan-VEGFR inhibitor AZD2171, in whom increased plasma levels of FGF were detected upon relapse[61]. Another study showed that approximately half of patients with metastatic colorectal cancer who received bevacizumab plus chemotherapy had more than a 5-fold increase in either placental growth factor or FGF prior to progression[62]. Patients with late-stage breast cancer have been reported to express a large number of proangiogenic factors, including FGF-2, in contrast to earlier stage lesions, which primarily express VEGF[58,63]. Taken together, these data support the hypothesis that tumor progression during inhibition of angiogenesis may be facilitated via activation of compensatory proangiogenic and tumorigenic mechanisms.

Several compounds in development stand to address the challenges and limitations of targeted therapy in the treatment of HCC. These are discussed in the following sections and summarized in Table 2.

Unmet needs for HCC remain, despite the availability of sorafenib. Indeed, sorafenib has some significant limitations, including modest, transient benefits, and toxicity challenges; and its use in patients with more advanced liver disease and in Asian patients has not yet been fully defined. The development of newer targeted therapies that inhibit angiogenesis simultaneously with inhibition of other key proangiogenic factors in HCC, such as FGFR or c-MET signaling, is providing further insights into the underlying pathogenesis of HCC tumors. Compounds that directly block angiogenesis and tumorigenesis via dual inhibition of FGFR and VEGFR, such as brivanib, and other compounds that indirectly modulate angiogenesis, such as mTOR inhibitors, are providing novel mechanisms that exploit critical pathways in HCC tumor progression and may have the potential to improve clinical outcomes both as monotherapy and in the case of escape from sorafenib. In the coming years, a number of phase III clinical trials examining these angiogenesis inhibitors will be mature, providing a better picture of the clinical utility and treatment options for patients with HCC.

We thank Lena Prisco, PhD, and Mark English, PhD, of PAREXEL and Bristol-Myers Squibb for their editorial assistance with this paper.