In recent years, much basic and clinical research has focused on defining molecular events that lead to tumor development, growth, and invasion. New blood vessel formation is essential for malignant transformation and the distant spread of cancer cells, and angiogenesis in tumors has been a recent therapeutic target. Several clinical trials have demonstrated a benefit from targeted therapies, including vascular endothelial growth factor (VEGF)-targeted agents, when combined with cytotoxic therapy,1-7 and as single agents.8-18 Furthermore, angiogenesis may be an important target for the secondary prevention of cancer, and biomarkers of angiogenesis may allow hepatocellular carcinoma (HCC) to be diagnosed at a curable stage. This review will therefore focus on the role of VEGF in screening, risk stratification, and treatment monitoring in HCC. It will include a basic review of screening for HCC, VEGF in the biology of HCC, measurement and clinical utility of circulating VEGF, using VEGF measurements to monitor treatment, clinical trials of antiangiogenic agents in HCC, and future directions.
HCC is the most common primary liver malignancy, the sixth most common cancer, and the third most common cause of cancer-related deaths worldwide.19 The incidence of liver cancer in the United States has steadily risen over the last 2 decades, with an estimated 21,370 new cases and 18,410 deaths from this disease in 2008.20 The rising incidence and prevalence of HCC in the United States in the last 2 decades is mainly attributed to the increase in the number of persons living with chronic hepatitis C virus infection.21 Migration from endemic areas and nonalcoholic steatohepatitis also likely play a role. Cirrhosis, irrespective of its etiology, is also an important risk factor for HCC, with patients who have cirrhosis having a 1% to 6% annual risk of developing HCC.22 Because >80% of HCCs originate in cirrhotic livers because of underlying chronic liver disease, surveillance programs have been recommended for patients with known cirrhosis to provide earlier detection of HCC, and improve the chance for cure. Available screening methods include regular imaging of the liver and measurement of serum α-fetoprotein (AFP) blood levels. AFP is the most commonly used serum tumor marker for HCC. However, elevated AFP levels also occur in patients with noncancerous conditions of the liver. A systematic review and critical analysis of the literature, based on a Medline search from 1966 to 2002, indicated that for HCC screening, a serum AFP level >20 μg/L had a sensitivity of 41% to 65%, specificity of 80% to 94%, positive likelihood ratios of 3.1 to 6.8, and negative likelihood ratios of 0.4 to 0.6 for diagnosis of HCC.23 This relatively poor performance, in addition to the unmet need for early HCC diagnosis, led us to a search for more sensitive and specific biomarkers of HCC. Because HCC is a highly vascular tumor, researchers have studied screening patients for markers of increased angiogenesis, which may be detected even before HCC is clinically manifest. However, to the best of our knowledge, most of the studies that discussed the potential role of VEGF in this setting report preliminary results, and none has been validated in a prospective setting.
HCC is potentially curable by surgical resection and liver transplantation. However, the majority of patients present with advanced stage disease, which is most commonly accompanied by severe background liver disease. Therefore, surgery is feasible for only a small fraction of patients with localized disease. Moreover, systemic cytotoxic therapies have demonstrated a very limited impact on the natural history of advanced HCC. In the last decade, molecular characterization of HCC has led to the recognition of defined aberrant signaling pathways, which helped in subsequent development of targeted agents as potential choices for the treatment of this chemoresistant disease. Notably, antiangiogenesis therapies have shown the most promise so far in treatment of advanced HCC.
Angiogenesis in the Biology of HCC
Angiogenesis is a cascade of linked and sequential steps that ultimately lead to the neovascularization of tumors. Significant HCC growth is dependant on angiogenesis, and an increase in tumor dimension beyond 0.5 mm will induce the proliferation of vascular endothelial cells.24 VEGF, also called vascular permeability factor, was initially identified in 1983 as a protein secreted by tumor cells.25 As a potent permeability factor, VEGF promotes extravasation of plasma fibrinogen, leading to the formation of fibrin scaffolding that facilitates cell migration during invasion. As an endothelial growth factor, VEGF stimulates endothelial cell proliferation, thus inducing the budding of new blood vessels around the growing tumor masses.
Molecular Interactions With VEGF
VEGF mediates its angiogenic effects via several different receptors (see Fig. 1). VEGF receptor (VEGFR) 1 and VEGFR2 were originally discovered on endothelial cells as tyrosine kinase receptors. The various members of the VEGF family have different binding affinities for each receptor. VEGFR1 plays an important role in developmental angiogenesis as well as other processes, including monocyte migration, recruiting of endothelial cell progenitors, increasing the adhesive properties of natural killer cells, and inducing growth factors from liver sinusoidal endothelial cells.26 VEGFR2 mediates the majority of the downstream effects of VEGF in angiogenesis, including microvascular permeability, endothelial cell proliferation, tumor cell migration, and survival.
Neuropilin-1 and -2 are transmembrane glycoproteins that interact with several members of the VEGF family ligands, and have been shown to serve as coreceptors for VEGF, suggesting a potential role in angiogenesis.27 They enhance the binding affinity of VEGF family ligands to VEGF receptors and affect subsequent intracellular signaling. VEGF binding to neuropilin-1 and neuropilin -2 leads to increased endothelial mitogenesis and chemotaxis.27, 28
The VEGF family comprises 6 glycoproteins; VEGF-A, VEGF-B, VEGF-C, VEGF-D, VEGF-E, and placental growth factor. Within these major VEGF subtypes are multiple isoforms. The best-characterized VEGF family member is VEGF-A (commonly referred to as VEGF). The human VEGF-A transcript undergoes alternative splicing to yield mature proteins of 121, 145, 165, 183, 189, and 206 amino acids; VEGF-165 is the predominant isoform.29 Notably, VEGF is coded by a single gene that contains 8 exons separated by 7 introns; exons 1, 2, 3, 4, 5, and 8 are common to all isoforms. In addition, VEGF-165 contains exon 7, and VEGF-189 contains exons 6 and 7, which have been shown to encode the extracellular matrix-binding domain of the VEGF. This domain is able to bind heparin sulfate proteoglycans and other matrix proteins. Therefore, after secretion, VEGF-189 and part of VEGF-165 bind to the extracellular matrix and cell membrane. Extracellular enzymes, such as heparinases, plasmin, urokinase-type plasminogen activator, and matrix metalloproteinase, are able to release bound VEGF and regulate the VEGF bioavailability. Thus, circulating VEGF-165 is the most abundant active biologic isoform involved in angiogenesis.
Significance of Tissue VEGF Levels in the Biology of HCC
Studies have shown that VEGF is frequently expressed in HCC. In a quantitative analysis study of 70 HCC cases, VEGF expression as demonstrated by immunohistochemistry (IHC) was observed in 63.9% of encapsulated and 78.3% of nonencapsulated HCCs, and 90.9% of HCCs with extrahepatic metastasis.30 Another study reported VEGF-positive expression, using IHC, in 72 of 105 HCC patients (68.6%).31 Capsular infiltration (P = .005), vascular invasion (P = .035), shorter overall survival (P = .014), and intrahepatic metastasis (P = .008) were observed more frequently in patients with VEGF-positive tumors. Other studies suggested a correlation between the degree of tissue VEGF expression measured by IHC and intensity of both the magnetic resonance signal and the contrast enhancement of the hepatic artery on computed tomography in patients with HCC using IHC.32-34 Furthermore, recent studies demonstrated a definite correlation between tumor VEGF expression using IHC and microvessel density, which is a marker of tumor angiogenesis.35, 36 Similar findings in other studies have laid the groundwork for using VEGF activity as a tool to screen for both premalignant liver lesions and invasive HCC.37, 38 In the first study,37 tissue mRNA level of VEGF-165 was assessed by a semiquantitative reverse-transcriptase polymerase chain reaction in 29 patients with HCC, 26 with cirrhosis, and 15 with chronic hepatitis. The liver expression of mRNA of VEGF-165 was found to be significantly higher in HCC than in chronic liver diseases and in the cirrhotic tissue of HCC patients than in those with HCC-free cirrhosis. The second study38 evaluated the relation between the expression of VEGF and vascular density in the HCC and nonmalignant hepatic parenchyma and reported a strong correlation between the levels of VEGF expression in tissue and the number of vessels, which was significantly higher in the areas of HCC and internodular fibrotic tissue in cirrhotic liver than in non-neoplastic hepatic parenchyma.
Moreover, the role of angiogenesis in the progression of premalignant lesions to invasive cancers prompted investigators to study the role of VEGF in the natural history of HCC. A group of researchers suggested that tissue VEGF expression increased according to the stepwise development of HCC.39 VEGF levels were noted to progressively increase through the successive steps of low-grade dysplasia, high-grade dysplasia, and early HCCs.
Numerous studies, summarized in Table 1, have attempted to determine the impact of tissue VEGF expression on clinicopathologic features, tumor phenotype, and prognosis. VEGF expression in HCCs with microscopic venous invasion was significantly higher than in HCCs without invasion.30, 40, 41 In a study of 105 patients undergoing potential curative resection for HCC,31 patients with VEGF-positive tumors (68.6%) more frequently had capsular infiltration, vascular and intrahepatic metastasis, and a shorter median survival than patients with VEGF-negative tumors. Another study of 20 patients with HCC confirmed that higher VEGF expression was associated with a lower median survival.42 Moreover, other studies have shown that expression of VEGF in HCC was correlated with the probability of recurrence after tumor resection or ablation.43-46 Collectively, these findings suggest that tissue VEGF expression in HCC may indicate more advanced disease and reduced median survival. However, these studies had relatively small patient numbers and used different VEGF detection assays.
Table 1. Clinical Studies of Circulating and Tissue VEGF in HCC
Measurement and Clinical Utility of Circulating VEGF
Peripheral blood offers a readily available means of detecting VEGF levels. However, there has been debate regarding whether the serum VEGF level is indeed a true reflection of tumor angiogenic activity. Serum VEGF can be released from platelets during ex vivo clotting, which does not necessarily have to be related to tumors and therefore may not be truly representative of the tumor milieu. In a study of 60 patients with HCC, serum VEGF-165 isoform levels correlated significantly with platelet counts,47 but when the serum VEGF levels were corrected for the individual platelet counts, the serum VEGF/platelet ratio correlated significantly with tumor cytosolic VEGF concentration. In addition, advancing HCC stage was associated with a significant increase in tumor cytosolic VEGF concentration, serum VEGF/platelet ratio, and serum VEGF level. It was concluded that the serum VEGF/platelet ratio correlated positively with tumor VEGF expression.
Additional reports have suggested that the degree of serum VEGF elevation correlates with tumor stage and more aggressive behavior of HCC.30 One study demonstrated that tumors measuring >2 cm in greatest dimension had higher median serum VEGF levels than those <2 cm.48 Another study of 108 patients indicated that elevated serum VEGF levels were significantly associated with venous invasion and advanced tumor stage. Patients with a serum VEGF level higher than median (>245.0 pg/mL) had significantly worse overall and disease-free survival than those with a lower level.49 Another study confirmed the relation between elevated serum VEGF in HCC patients and tumor size using a cutoff point of 5 cm.50 This study also demonstrated significantly higher serum VEGF levels in those patients with macroscopically evident portal vein invasion and metastasis. In addition, preoperative serum VEGF levels have correlated with clinicopathological features and long-term survival,49 and appeared to reflect the disease's potential for vascular invasion and metastasis.51 In the previously mentioned studies, it was also suggested that higher serum VEGF levels predict a poorer outcome after resection of HCC. Therefore, monitoring serum VEGF levels might not only lead to early diagnosis of HCC, but also provide clinicians with an early indication of disease prognosis.
The role of serum VEGF as a biomarker for HCC screening has been investigated. In a study of 63 patients, patients with HCC had significantly higher serum VEGF levels (median, 245 pg/mL) than healthy volunteers (median, 180 pg/mL) (P = .042).49 Serum VEGF was also combined with serum AFP in attempts to improve screening for HCC. A recent report of 77 individuals including controls demonstrated that the combination of a serum AFP value of >19.8 ng/mL and a serum VEGF value of >355.2 pg/mL increased HCC screening sensitivity to 95.5% compared with the individual sensitivities of 68.2% and 86.4%, respectively.52 However, test for a significant difference was not provided by the authors.
Furthermore, noncancerous hepatic conditions associated with elevated serum VEGF have been found to be associated with lower serum VEGF values than the previously mentioned cancerous conditions; serum VEGF levels (as measured by enzyme-linked immunoadsorbent assay [ELISA]) were 58.0 pg/mL, 44.1 pg/mL, and 172.7 pg/mL in patients with chronic hepatitis (n = 40), cirrhosis (n = 34), and acute hepatitis (n = 21), respectively.53
Thus, serum VEGF level in HCC patients has shown promise for HCC screening whether used alone or combined with serum AFP. However, these studies report preliminary results, and none has been validated in a prospective setting.
The confounding effect of a platelet's storage and release of VEGF on the serum levels of VEGF may be overcome by measuring the plasma levels of VEGF. Increased plasma VEGF levels were found to be associated with the presence of extrahepatic metastases and portal vein involvement in 45 patients with HCC undergoing transarterial chemoembolization (TACE).43 In addition, significant differences were noted when plasma VEGF was categorized by tumor size, distant metastasis, and TNM stage. In another study of 30 patients with HCC, increased plasma VEGF expression was associated with the development of HCC metastases after TACE.54 Another study reported that plasma VEGF was a promising predictor for distant metastases from HCC.55 The study reported that the levels were significantly elevated in the HCC group, compared with the control, chronic hepatitis, and cirrhosis groups. Furthermore, stage IV patients with distant metastases demonstrated significantly marked elevation of plasma VEGF compared with the patients at the other stages. Notably, several other studies reported consistently lower degrees of VEGF elevation in noncancerous liver conditions, with mean plasma VEGF levels ranging from 12.1 pg/mL to 89.9 pg/mL as measured by ELISA.56-58
Serum VEGF versus plasma VEGF measurements
Current evidence supports the use of serum VEGF level as an indirect measure of tumor VEGF expression.47 However, serum VEGF concentrations may fluctuate based on clotting that is related to sample processing.59 Conversely, in plasma collections, platelet degranulation is minimized by adding anticoagulants to the blood samples. Consequently, plasma VEGF concentrations are up to 20× lower than the matched serum VEGF concentrations.60 Moreover, plasma VEGF levels are not affected by the amount of time between blood sampling and centrifugation.59 In light of these considerations, plasma VEGF is likely a more reproducible and reliable estimate of HCC VEGF activity.
VEGF protein versus mRNA detection
We made several attempts to evaluate the feasibility of measuring of tissue and circulating mRNA levels. In 1 report, the preoperative level of circulating serum VEGF mRNA, especially isoform VEGF-165, was predictive for disease recurrence in a study of 50 patients with HCC.61 However, mRNA changes, overall, had surprisingly minimal correlation with changes in the coding proteins, a discrepancy that reflects post-translational modifications that regulate protein structure, function, activation, and degradation.62 In addition, the unbound VEGF form remains the biologically active entity as previously mentioned. Therefore, detecting VEGF-165 by ELISA may be the optimal method of assay.
Using VEGF Levels to Monitor Treatment Response
Changes in VEGF levels in tissue and peripheral blood during therapy may help establish VEGF as a biomarker for treatment effect and improve our understanding of the influence of antiangiogenic therapies in patients with HCC.
Most recently, a study of an anti-VEGF antibody, bevacizumab, in advanced HCC reported preliminary results that indicated a possible correlation between response to therapy and plasma VEGF levels.12 Plasma was available in 8 HCC patients, before therapy, and after an initial 8-week treatment period. Plasma was also available at time of disease progression for 6 of these 8 patients. Plasma VEGF levels decreased from baseline in all patients after 8 weeks of bevacizumab therapy (baseline: mean VEGF-A level, 746 pg/mL; 8 weeks: mean VEGF-A level, 249 pg/mL [P = .003]). At the time of disease progression, VEGF-A increased from 8-week values to near baseline levels in 5 of 6 patients for whom plasma was available (mean VEGF-A level at progression, 653 pg/mL; P = .013). Although this is an interesting observation, the supporting data are preliminary and merit more vigorous investigation before suggesting the use of VEGF as a surrogate marker in this setting. Interestingly, 2 previous reports indicated increase in circulating total VEGF level after antiangiogenic therapy.63, 64 However, the authors described a reduction of free serum VEGF in cancer patients treated with escalating doses of bevacizumab when compared with basal serum concentrations in the first study.63 The second study measured plasma VEGF levels, before and 12 days after a single-dose infusion of bevacizumab, and found a significant increase in the levels of plasma VEGF; the authors commented that it will be crucial to establish how much of the VEGF in the plasma represents free protein (unbound to bevacizumab) at Day 12.64 Therefore, future studies should be directed toward obtaining serial samples at more time points, including mid–cycles, in addition to evaluating free and total VEGF to shed more light on the biology of VEGF signaling in this setting.
Furthermore, plasma VEGF levels were shown to be increased shortly after treatment with TACE.65, 66 The peak plasma VEGF occurred anywhere from Day 1 to Day 7 after TACE, with a slow decline thereafter. These increased plasma VEGF levels are thought to be related to the induction of tissue hypoxia, which leads to VEGF production via increased HIF-1 transcription. One study did not find a significant difference between serum VEGF levels before and after radiofrequency ablation.67 This finding may have been related to the use of serum VEGF levels, less hypoxia induced by radiofrequency ablation than by TACE, or timing of the samples. Alternatively, the impact of treatment-induced hypoxia may be limited to the tissue VEGF levels, which would not be detected in the blood. Higher tissue levels of VEGF and basic fibroblast growth factor were also found in resected HCC specimens from patients who received preresection TACE than in specimens from patients who did not receive preresection TACE.68 Another study demonstrated that the microvessel density significantly increased after TACE in patients with HCC.69 Collectively, these findings suggest that the short-term treatment-induced increase of plasma VEGF reflected the up-regulated tissue VEGF levels and new blood vessel formation in the remaining viable tumor and/or normal hepatic parenchyma.
Studies also explored the prognostic value of plasma VEGF levels after TACE. Risk of disease progression has been shown to increase in patients whose plasma VEGF levels after TACE were increased above baseline.54 In a 6-month follow-up of 30 patients after TACE, all 22 patients who had progression had elevated plasma VEGF levels after TACE, whereas none of the patients with decreased plasma VEGF levels after TACE exhibited progressive disease. Similar findings were reported in a recent study of patients who underwent radiofrequency ablation.70 The rise in VEGF levels after local therapy for patients with HCC indicates a potential for biomarker-guided antiangiogenic therapy in this setting. Furthermore, circulating VEGF levels may be predictive of the outcomes after local therapies or a biomarker-guided adjunctive antiangiogenic approach. However, serial measurements of circulating VEGF before and after local therapies in future clinical trials are warranted to further evaluate changes in angiogenesis pathway in this setting.
Clinical Trials of Antiangiogenic Agents in Patients With HCC
Several targeted agents have recently been tested in patients with advanced HCC, including agents targeting the VEGF pathway, either alone8-18, 71, 72 or in combination with other systemic therapies (Table 2).1, 7, 73-78
Table 2. Summary of Selected Clinical Trials in Patients With Advanced HCC
Median Survival, mo
HCC indicates hepatocellular carcinoma; NR, not reported; PFS, progression-free survival; OS, overall survival; GEMOX, gemcitabine and oxaliplatin; TTP, time to disease progression; CLIP, Cancer of the Liver Italian Program.
The cancer cell has been the only target of anticancer therapy for more than 50 years. However, the cancer cell is genetically unstable, and mutations accumulate. Conversely, antiangiogenic therapy targets endothelial cells that are genetically stable. Thalidomide's mechanism of action was believed to be partly based on its antiangiogenic effects. Nevertheless, several clinical trials of thalidomide showed rare responses ranging from 0% to 6.3%.71, 72 Sunitinib is an oral multikinase inhibitor, which exerts an antiangiogenic effect by targeting VEGFR and platelet-derived growth factor receptor tyrosine kinases. Sorafenib is an oral multikinase inhibitor that exerts antitumor effect through targeting Raf/MEK/ERK signaling at the level of Raf kinase, and exerts an antiangiogenic effect by targeting VEGFR-2/-3. Most recently, 2 phase 3 trials of sorafenib were reported.8, 18 The pivotal randomized, placebo-controlled phase 3 trial of sorafenib in patients with advanced HCC, the SHARP trial, reported a 2.8-month improvement in median overall survival rate (P = .0006), along with increased time to disease progression and disease control rate, and a 2.3% response rate.8 This has led to the US Food and Drug Administration approval of sorafenib for advanced HCC. Bevacizumab is a recombinant, humanized monoclonal antibody that targets VEGF and may augment chemotherapy administration by making tumor vasculature less permeable and decreasing the elevated tumor interstitial pressure. Erlotinib is an oral tyrosine kinase inhibitor that blocks phosphorylation at the intracellular domain of the epithelial growth factor receptor. Most recently, we reported a phase 2, single-arm, open-label trial of bevacizumab and erlotinib that demonstrated an improved response rate, median overall survival, and progression-free survival.73
Moreover, there are several ongoing clinical trials of different vascularly targeted agents in patients with HCC, as shown in Table 3.
Table 3. Selected Ongoing Clinical Trials of Systemic Antiangiogenic Therapy in HCC
HCC indicates hepatocellular carcinoma; NCI, National Cancer Institute.
Collectively, application of antiangiogenesis to patients with advanced HCC leads to improvement in survival despite surprisingly low response rates. There is poor correlation between survival benefit and conventional methods of response assessment, such as Response Evaluation Criteria in Solid Tumors (RECIST), which poses questions of how best to quantify efficacy of antiangiogenic agents. Despite tumors increasing in size, the observation of tumor necrosis in many studies is intriguing. Therefore, in 2000, a panel of experts recommended that the response criteria be amended to take into account tumor necrosis induced by treatment.79 Although the usefulness of tumor necrosis in assessing efficacy of anticancer agents in HCC remains to be established, it is a potentially significant clinical endpoint that warrants further investigation.
Presently, there are major deficiencies in all aspects of HCC management, including poor screening methods and lack of biomarkers to guide therapy. We believe that ELISA detection of circulating VEGF-165 as a biomarker should be investigated in these challenging areas. In screening for HCC, early evidence suggested that both serum VEGF and plasma VEGF levels may improve test characteristics when combined with the current practice of testing for serum AFP and imaging. We suggest that further study in HCC screening should examine the utility of plasma VEGF, because unlike serum VEGF, plasma VEGF is not affected by blood coagulation. The utility of plasma VEGF in this setting could be assessed in parallel with the most commonly used HCC screening strategy of combining serum AFP measurement every 3 months with liver ultrasound every 6 months. Understanding the biologic role of VEGF and its influence on the clinicopathologic features of HCC highlights the potential role of targeting VEGF in cancer therapies. The availability of drugs that target VEGF itself (bevacizumab) or VEGF receptors (sunitinib, sorafenib) provides opportunities to develop newer approaches of combining those agents with the current treatment options available for this disease (eg, TACE) and other targeted therapies. The limited benefit of conventional response monitoring by RECIST in patients treated with antiangiogenesis emphasizes the importance of the development of biomarkers to monitor treatment effects. To our knowledge, there currently are no studies that adequately address the biomarker role of VEGF in patients treated with systemic antiangiogenic agents. VEGF measurements should be investigated as potential biomarkers to optimize antiangiogenic therapy, in addition to monitoring the patient's response to therapy.